Gonadotropin‐releasing hormone agonist protocols for pituitary suppression in assisted reproduction

Abstract

Background

Gonadotropin‐releasing hormone agonists (GnRHa) are commonly used in assisted reproduction technology (ART) cycles to prevent a luteinising hormone (LH) surge during controlled ovarian hyperstimulation (COH) prior to planned oocyte retrieval, thus optimising the chances of live birth. We compared the benefits and risks of the different GnRHa protocols used.

Objectives

To evaluate the effectiveness and safety of different GnRHa protocols used as adjuncts to COH in women undergoing ART.

Search methods

We searched the following databases in December 2022: the Cochrane Gynaecology and Fertility Group's Specialised Register, CENTRAL, MEDLINE, Embase, and registries of ongoing trials. We also searched the reference lists of relevant articles and contacted experts in the field for any additional trials.

Selection criteria

We included randomised controlled trials (RCTs) comparing any two protocols of GnRHa, or variations of the protocol in terms of different doses or duration, used in in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI) cycles in subfertile women.

Data collection and analysis

We used standard methodological procedures recommended by Cochrane. Our primary outcome measures were number of live births or ongoing pregnancies and incidence of ovarian hyperstimulation syndrome (OHSS) per woman/couple randomised. Our secondary outcome measures included number of clinical pregnancies, pregnancy losses, number of oocytes retrieved, amount of gonadotropins used, and cost and acceptability of the treatment protocols.

Main results

We included 40 RCTs (4148 women). The trials evaluated 10 different comparisons between protocols. The evidence is current to December 2022. Only half of the studies reported the primary outcome of live birth or ongoing pregnancy rates. We restricted the primary analysis of live birth and ongoing pregnancy to trials with low risk of selection and reporting bias.

Nineteen studies compared long and short protocols. The primary analysis restricted to trials with low risk of bias included five studies reporting on live birth or ongoing pregnancy rates. Results showed little or no difference when the long protocol was compared with a short protocol (odds ratio (OR) 1.45, 95% confidence interval (CI) 0.83 to 2.52; I² = 0%; 5 studies, 381 women; low‐certainty evidence). For the same comparison, there was evidence that the long protocol may improve clinical pregnancy rates when compared to the short protocol (OR 1.56, 95% CI 1.01 to 2.40; I² = 23%; 8 studies, 552 women; low‐certainty evidence). No study in this comparison reported on OHSS.

We are uncertain if there is a difference between groups in terms of live birth and ongoing pregnancy rates when the following GnRHa protocols were compared: long versus ultrashort (OR 1.78, 95% CI 0.72 to 4.36; 1 study, 150 women; very low‐certainty evidence); long luteal versus long follicular phase (OR 1.89, 95% CI 0.87 to 4.10; 1 study, 223 women; very low‐certainty evidence); GnRHa reduced‐dose versus GnRHa same‐dose continued in the long protocol (OR 1.59, 95% CI 0.66 to 3.87; 1 study, 96 women; very low‐certainty evidence); GnRHa administration for two versus three weeks before stimulation (OR 0.88, 95% CI 0.37 to 2.05; 1 study, 85 women; very low‐certainty evidence); GnRHa continued versus discontinued after human chorionic gonadotropin (HCG) administration in the long protocol (OR 0.89, 95% CI 0.49 to 1.64; 1 study, 181 women; very low‐certainty evidence); and 500 µg dose versus 80 µg dose in the short protocol (OR 0.31, 95% CI 0.10 to 0.98; 1 study, 200 women; very low‐certainty evidence). Clinical pregnancy rates may improve with a 100 µg dose compared to a 25 µg dose in the short protocol (OR 2.30, 95% CI 1.06 to 5.00; 2 studies, 133 women; low‐certainty evidence).

Only four of the 40 included studies reported adverse events. We are uncertain of any difference in OHSS rate in the GnRHa reduced‐dose versus GnRHa same‐dose regimen in the long protocol (OR 0.47, 95% CI 0.04 to 5.35; 1 study, 96 women; very low‐certainty evidence) or when administration of GnRHa lasted for two versus three weeks before stimulation (OR 0.93, 95% CI 0.06 to 15.37; 1 study, 85 women; very low‐certainty evidence). Regarding miscarriage rates, we are uncertain of any difference when the GnRHa long protocol was administered for two versus three weeks before stimulation (OR 0.93, 95% CI 0.18 to 4.87; 1 study, 85 women; very low‐certainty evidence) and when a 500 µg dose was compared with an 80 µg dose in the short protocol (OR 3.15, 95% CI 0.32 to 31.05; 1 study, 131 women; very low‐certainty evidence). No studies reported on cost‐effectiveness or acceptability of the different treatment protocols.

The certainty of the evidence ranged from low to very low. The main limitations were failure to report live birth or ongoing pregnancy rates, poor reporting of methods in the primary studies, imprecise findings due to lack of data, and insufficient data regarding adverse events. Only eight of the 40 included studies were conducted within the last 10 years.

Authors' conclusions

When comparing long and short GnRHa protocols, we found little or no difference in live birth and ongoing pregnancy rates, but there was evidence that the long protocol may improve clinical pregnancy rates overall. We were uncertain of any difference in OHSS and miscarriage rates for all comparisons, which were reported by only two studies each. There was insufficient evidence to draw any conclusions regarding other adverse effects or the cost‐effectiveness and acceptability of the different treatment protocols.

Plain language summary

Which hormone (GnRHa) medications work best with ovarian stimulation hormones in assisted conception cycles?

Key messages

• When comparing long versus short gonadotropin‐releasing hormone agonists (GnRHa) treatment, there was little to no difference between groups for live birth and ongoing pregnancy rates, but the long treatment may result in a higher clinical pregnancy rate (where the foetus can be seen or heard).

• We are uncertain if there is any difference in live birth and clinical pregnancy rates for the other comparisons studied, except for the comparison 100 µg dose versus 25 µg dose in a short treatment, which showed that clinical pregnancy may improve with a 100 µg dose.

• Further research is needed to look at the cost‐effectiveness and acceptability of the different treatments.

What did we want to find out?

Gonadotropin‐releasing hormone agonists (GnRHa) are given along with hormone injections that stimulate the ovaries to try to prevent the premature release of eggs before they can be harvested in a planned way by means of a surgical procedure. GnRHa have been shown to improve pregnancy rates. Various ways of giving GnRHa are described in the literature. We wanted to find out the most effective way of giving GnRHa to increase the number of babies born and ongoing pregnancy rates, in addition to reducing the rates of pregnancy loss and ovarian hyperstimulation syndrome (OHSS) (an unwanted effect of treatment with fertility medicines).

What did we do?

We reviewed the evidence on which medications (in the form of GnRHa) are best given together with hormones to stimulate ovaries for women trying to become pregnant through assisted conception.

What did we find?

We found 40 studies of 4148 women comparing the use of GnRHa in different ways during assisted conception treatment. Nineteen of these studies (1582 women) compared a long treatment (where GnRHa is started at least two weeks prior to hormone stimulation) with a short treatment (where GnRHa is started with the hormone stimulation).

Main results

When comparing long versus short GnRHa treatment, there was little to no difference between groups for live birth and ongoing pregnancy rates. Our findings suggest that in a population in which 14% of women achieve a live birth or ongoing pregnancy using a short treatment, between 12% and 30% will achieve this with a long treatment. There is evidence that the long treatment may result in a higher clinical pregnancy rate (where the foetus can be seen or heard) when compared to the short treatment. Our findings suggest that, in a population where 16% of women will achieve a clinical pregnancy with the short treatment, 17% to 32% will achieve this with a long treatment.

For other comparisons of GnRHa treatments, we are uncertain if there is any difference in terms of live birth and clinical pregnancy rate, except for the comparison of 100 µg dose versus 25 µg dose in a short treatment, which showed that clinical pregnancy may improve with a 100 µg dose.

We are uncertain if there is any difference in OHSS rates and pregnancy loss, which were reported by only two studies each.

There was not enough evidence to reach any conclusions regarding other harmful effects. Further research is needed to look at the cost‐effectiveness and acceptability of the different treatments.

What are the limitations of the evidence?

We have low to very low confidence in the evidence. The main limitations in the evidence were non‐reporting of live birth or ongoing pregnancy in half of the studies, poor reporting of study methods, unclear findings, very few studies reporting on unwanted effects such as OHSS, and lack of data regarding other harmful events, cost‐effectiveness, and acceptability of the treatments. Only eight of the 40 included studies were conducted within the last 10 years.

How up‐to‐date is this evidence?

The evidence is current to December 2022.

Summary of findings

Summary of findings 1. Long protocol compared with short protocol for pituitary suppression in assisted reproduction.

Long protocol compared with short protocol for pituitary suppression in assisted reproduction
Population: women undergoing pituitary suppression in assisted reproduction Intervention: long protocol Comparison: short protocol
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Short protocol	Long protocol
Live birth or ongoing pregnancies per woman randomised	143 per 1000	195 per 1000 (122 to 296)	OR 1.45 (0.83 to 2.52)	381 (5 studies)	⊕⊕⊝⊝ Lowa	No difference between groups
OHSS per woman randomised	Not reported by any study in this comparison
Clinical pregnancies per woman randomised	165 per 1000	235 per 1000 (166 to 321)	OR 1.56 (1.01 to 2.40)	552 (8 studies)	⊕⊕⊝⊝ Lowa	Favours long protocol
*The basis for the assumed risk is the median control group risk across studies. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; OHSS: ovarian hyperstimulation syndrome; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded two levels for serious risk of bias, poor reporting of methods in primary studies.

Summary of findings 2. Long protocol compared with ultrashort protocol for pituitary suppression in assisted reproduction.

Long protocol compared with ultrashort protocol for pituitary suppression in assisted reproduction
Population: women undergoing pituitary suppression in assisted reproduction Intervention: long protocol Comparison: ultrashort protocol
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Ultrashort protocol	Long protocol
Live birth and ongoing pregnancies per woman randomised	122 per 1000	198 per 1000 (91 to 376)	OR 1.78 (0.72 to 4.36)	150 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
OHSS per woman randomised	Not reported by any study in this comparison
Clinical pregnancies per woman randomised	161 per 1000	230 per 1000 (133 to 370)	OR 1.56 (0.8 to 3.06)	230 (2 studies)	⊕⊝⊝⊝ Very lowb	Uncertain
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; OHSS: ovarian hyperstimulation syndrome; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded three levels for very serious imprecision, serious risk of bias, single study (indirectness), low number of participants.
^bDowngraded three levels for very serious risk of bias, serious imprecision, poor reporting of methods in the primary studies.

Summary of findings 3. Long luteal phase protocol compared with long follicular phase protocol for pituitary suppression in assisted reproduction.

Long luteal phase protocol compared with long follicular phase protocol for pituitary suppression in assisted reproduction
Population: women undergoing pituitary suppression in assisted reproduction Intervention: long luteal phase protocol Comparison: long follicular phase protocol
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Long follicular phase protocol	Long luteal phase protocol
Live birth and ongoing pregnancies per woman randomised	102 per 1000	177 per 1000 (90 to 319)	OR 1.89 (0.87 to 4.1)	223 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
OHSS per woman randomised	Not reported by any study in this comparison
Clinical pregnancies per woman randomised	269 per 1000	281 per 1000 (219 to 351)	OR 1.06 (0.76 to 1.47)	750 (5 studies)	⊕⊝⊝⊝ Very lowb	Uncertain
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; OHSS: ovarian hyperstimulation syndrome; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded three levels for very serious imprecision, serious risk of bias, single study (indirectness), low number of participants.
^bDowngraded three levels for very serious risk of bias, serious imprecision, poor reporting of methods in the primary studies.

Summary of findings 4. Long protocol: continue GnRHa versus stop GnRHa for pituitary suppression in assisted reproduction.

Long protocol: continue GnRHa versus stop GnRHa for pituitary suppression in assisted reproduction
Population: women undergoing pituitary suppression in assisted reproduction Intervention: long protocol continue GnRHa Comparison: long protocol stop GnRHa
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Long protocol stop GnRHa	Long protocol continue GnRHa
Live birth and ongoing pregnancies Number of live births or ongoing pregnancies per woman randomised	276 per 1000	206 per 1000 (99 to 380)	OR 0.68 (0.29 to 1.61)	116 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
OHSS per woman randomised	Not reported by any study in this comparison
Clinical pregnancies Number of clinical pregnancies per woman randomised	197 per 1000	157 per 1000 (89 to 261)	OR 0.76 (0.40 to 1.44)	264 (3 studies)	⊕⊕⊝⊝ Lowb	No difference between groups
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; GnRHa: gonadotropin‐releasing hormone agonist; OHSS: ovarian hyperstimulation syndrome; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded three levels for single study (indirectness), low number of participants.
^bDowngraded two levels for very serious risk of bias associated with poor reporting of methods in the primary studies.

Summary of findings 5. Long protocol: continued same‐dose versus reduced‐dose GnRHa for pituitary suppression in assisted reproduction.

Long protocol: continued same‐dose versus reduced‐dose GnRHa for pituitary suppression in assisted reproduction
Population: women undergoing pituitary suppression in assisted reproduction Intervention: long protocol continued same‐dose GnRHa Comparison: long protocol reduced‐dose GnRHa
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Long protocol, reduced‐dose GnRHa	Long protocol, continued same‐dose GnRHa
Live birth and ongoing pregnancies Number of live births or ongoing pregnancies per woman randomised	660 per 1000	755 per 1000 (561 to 882)	OR 1.59 (0.66 to 3.87)	96 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
OHSS incidence per woman randomised	226 per 1000	534 per 1000 (117 to 1430)	OR 0.47 (0.04 to 5.35)	96 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
Clinical pregnancies per woman randomised	377 per 1000	382 per 1000 (292 to 479)	OR 1.02 (0.68 to 1.52)	407 (4 studies)	⊕⊝⊝⊝ Very lowb	Uncertain
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; GnRHa: gonadotropin‐releasing hormone agonists; OHSS: ovarian hyperstimulation syndrome; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded three levels for very serious imprecision, single study (indirectness), low number of participants.
^bDowngraded three levels for very serious risk of bias, wide CIs, poor reporting of methods in the primary studies.

Summary of findings 6. Long protocol: administration of GnRHa for 2 versus 3 weeks before stimulation for pituitary suppression in assisted reproduction.

Long protocol: administration of GnRHa for 2 versus 3 weeks before stimulation for pituitary suppression in assisted reproduction
Population: women undergoing pituitary suppression in assisted reproduction Intervention: long protocol: administration of GnRHa for 2 weeks before stimulation Comparison: long protocol: administration of GnRHa for 3 weeks before stimulation
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Long protocol: administration of GnRHa for 3 weeks before stimulation	Long protocol: administration of GnRHa for 2 weeks before stimulation
Live birth and ongoing pregnancies per woman randomised	488 per 1000	456 per 1000 (261 to 661)	OR 0.88 (0.37 to 2.05)	85 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
OHSS incidence per woman randomised	124 per 1000	1029 per 1000 (442 to 2345)	OR 0.93 (0.06 to 15.37)	85 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
Clinical pregnancies per woman randomised	585 per 1000	568 per 1000 (355 to 757)	OR 0.93 (0.39 to 2.21)	85 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; GnRHa: gonadotropin‐releasing hormone agonists; OHSS: ovarian hyperstimulation syndrome; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded three levels for very serious imprecision, high risk of bias for blinding, single study (indirectness), low number of participants.

Summary of findings 7. Short protocol: different doses of GnRHa for pituitary suppression in assisted reproduction.

Short protocol: different doses of GnRHa for pituitary suppression in assisted reproduction
Population: women undergoing pituitary suppression in assisted reproduction Intervention: short protocol higher dose Comparison: short protocol lower dose
Outcomes	Illustrative comparative risks* (95% CI)		Relative effect (95% CI)	Number of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Short protocol lower dose	Short protocol higher dose
Live birth and ongoing pregnancies (500 µg versus 80 µg) per woman randomised	13 per 1000 (4 to 39)	40 per 1000	OR 0.31 (0.10 to 0.98)	200 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
OHSS per woman randomised	Not reported by any study in this comparison
Clinical pregnancies (500 µg versus 80 µg) per woman randomised	36 per 1000 (14 to 90)	70 per 1000	OR 0.50 (0.19 to 1.32)	200 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
Clinical pregnancies (100 µg versus 50 µg) per woman randomised	353 per 1000 (168 to 596)	353 per 1000	OR 1.00 (0.37 to 2.70)	68 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
Clinical pregnancies (100 µg versus 25 µg) per woman randomised	673 per 1000 (487 to 817)	472 per 1000	OR 2.30 (1.06 to 5.00)	133 (2 studies)	⊕⊕⊝⊝ Lowb	Favours dose of 100 µg
Clinical pregnancies (50 µg versus 25 µg) per woman randomised	357 per 1000 (156 to 628)	353 per 1000	OR 1.02 (0.34 to 3.10)	57 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
Miscarriage rate (500 µg versus 80 µg) per woman randomised	46 per 1000 (5 to 323)	15 per 1000	OR 3.15 (0.32 to 31.05)	131 (1 study)	⊕⊝⊝⊝ Very lowa	Uncertain
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; GnRHa: gonadotropin‐releasing hormone agonists; OHSS: ovarian hyperstimulation syndrome; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aDowngraded three levels for serious imprecision, serious risk of bias, single study (indirectness), low number of participants.
^bDowngraded two levels for serious imprecision, serious risk of bias, low number of participants.

Background

Description of the condition

Subfertility affects one in six couples, and an increasing proportion are seeking assisted reproduction technology (ART) to achieve their goal of having a healthy live birth (Farquhar 2018). A single oocyte is normally produced in a natural cycle. In contrast, an in vitro fertilisation (IVF) cycle usually aims to produce more than one oocyte to improve the chances of having a sufficient number of embryos from which to choose for transfer to the uterus. At the same time, it is crucial to prevent an excessive ovarian response, resulting in ovarian hyperstimulation syndrome (OHSS). In order to produce more oocytes, the ovaries are stimulated with high doses of gonadotropins. However, there is a risk of a premature surge of luteinising hormone (LH), which could disrupt both normal follicle and oocyte development. Gonadotropin‐releasing hormone agonists (GnRHa) have been used in controlled ovarian hyperstimulation (COH) protocols to reversibly suppress the pituitary function and prevent a premature LH surge. The use of GnRHa has resulted in significant improvements in treatment, including decreased cancellation of started treatment cycles prior to oocyte recovery and higher pregnancy rates (Fields 2013).

Description of the intervention

Different GnRHa drugs, routes of administration (nasal or systemic), and GnRHa protocols have been used in ART. There are three main protocols involving GnRHa administration, namely long, short, and ultrashort protocols, as described below.

• Long protocol: GnRHa is administered at least two weeks before starting ovarian stimulation (to achieve pituitary suppression) and continued up until human chorionic gonadotropin (HCG) is given, starting from either the second day of the menstrual cycle (long follicular protocol) or the mid‐luteal phase (usually 21st day) of the previous cycle (long luteal protocol).

• Short protocol: GnRHa is administered from day one or two of the cycle (day one being the start of the menstrual bleed) and continued with ovarian stimulation until the day of HCG administration.

• Ultrashort protocol: GnRHa is given for three days, from day two of the cycle.

We excluded the ultralong protocol for pituitary suppression in IVF, used in women with endometriosis, as it is the subject of another Cochrane review (Georgiou 2019). Of note, there is a proposed network meta‐analysis on the clinical effectiveness and safety profile of currently applied COH protocols, published at the protocol stage (Gallos 2017).

How the intervention might work

The administration of multiple doses of GnRHa causes a reversible blockade of the pituitary function after an initial stimulatory phase, the so‐called flare‐up effect. The GnRHa suppresses the GnRH receptors and causes inhibition of postreceptor events. The resulting reduction in bioactive serum LH levels allows multiple follicular development to continue (before oocyte recovery) (Meldrum 1984; Regan 1990), avoiding the risk of an LH surge, and hence premature ovulation (Barlow 1998).

GnRHa protocols are commonly used in ART (IVF Worldwide 2022). Traditionally, the long protocol involves GnRHa use during the entire ovarian stimulation phase until HCG administration. Reports have shown that low endogenous LH concentrations persist until 10 to 14 days after discontinuation of GnRHa (Donderwinkel 1993; Sungurtekin 1995). Earlier studies argued that the continuation of GnRHa during the stimulation phase can also lead to a profound suppression of mid‐follicular LH, which might be associated with early pregnancy loss (Westergaard 2000). GnRHa could therefore be stopped earlier in the long‐protocol stimulation cycle (Simons 2005), allowing the pituitary to recover in time for the luteal phase without risking a premature LH surge. This could reduce both cost and inconvenience, as fewer injections would be needed.

Why it is important to do this review

The original Cochrane reviews on this topic, published in 1998 and updated in 2009, showed superiority of the long protocols compared with the short or ultrashort protocols. Of note, long protocols have traditionally been used in ART, whereas most of the newer alternatives (e.g. GnRH antagonists or mild ovarian stimulation protocols) have been compared with them (Mancini 2011; Mohsen 2013; Schimberni 2016). A subsequent 2015 update of this review showed that the long protocol was superior in terms of clinical pregnancy rates (based on moderate‐quality evidence), but did not show any conclusive evidence of a difference in live birth or ongoing pregnancy rates; in the remaining comparisons, there was no difference in live birth or other pregnancy outcomes, and there was insufficient evidence regarding adverse effects. The current update aims to identify any new evidence published since 2015 on the relative effectiveness of the different GnRHa protocols used as adjuncts to COH for ART, and to determine whether there is any further evidence on their adverse effects, cost‐effectiveness, and acceptability.

Objectives

To evaluate the effectiveness and safety of different gonadotropin‐releasing hormone agonist (GnRHa) protocols used as adjuncts to controlled ovarian hyperstimulation (COH) in women undergoing assisted reproduction technology (ART).

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials (RCTs) comparing various GnRHa protocols in ART. We included IVF or intracytoplasmic sperm injection (ICSI) treatment cycles. We excluded trials with non‐random allocation, as such trials are associated with a high risk of bias. We also excluded cross‐over trials, as the design is not suitable for this review. We excluded quasi‐randomised trials, even if they had been included in the original review.

Types of participants

We included women and couples with all types of infertility, undergoing ART and using GnRHa for pituitary suppression.

Types of interventions

Inclusion criteria

We included studies comparing any two protocols using GnRHa for pituitary suppression in an ART programme. We included ultrashort, short, and long protocols or any variations of these protocols in terms of different durations or doses of drugs administered.

We used the following definitions for the various protocols in this review.

• Long protocol: GnRHa commenced at least two weeks before starting ovarian stimulation and continued up until HCG was administered.

• Short protocol: GnRHa commenced at the same time as starting ovarian stimulation and continued up until the day of HCG administration.

• Ultrashort protocol: ovarian stimulation was commenced one to two days after starting GnRHa (and given only for three days).

We considered the following comparisons of variations of the above protocols for inclusion.

• Follicular or luteal phase start of the GnRHa in a long protocol

• Stopping the GnRHa at the start of ovarian stimulation ('long stop')

• Reducing the dose of GnRHa or continuing with the same dose after downregulation was achieved in a long protocol

• Two weeks versus three weeks of downregulation in a long protocol

• Discontinuing versus continuing GnRHa after HCG administration in a long protocol

• Stopping the GnRHa early in a short protocol ('short stop')

• Different doses of the GnRHa given daily in a short protocol

Exclusion criteria

We excluded women receiving donor oocytes.

We also excluded the following study comparisons.

• GnRH agonist versus GnRH antagonist protocols

• Different routes of administration of GnRHa

• GnRHa versus placebo protocols (Hughes 1992)

• Depot versus daily administration of GnRHa, as this is the topic of another Cochrane review (Albuquerque 2013)

• Addition of any drug in GnRHa protocols

• GnRH agonists for IVF in women with endometriosis, as this is the topic of another Cochrane review (Georgiou 2019)

• GnRH use in frozen embryo transfer (FET) cycles

• GnRH use for luteal phase support, as this is the topic of another Cochrane review (Van der Linden 2015)

Types of outcome measures

We measured the following primary and secondary outcome measures.

Primary outcomes

• Number of live births or ongoing pregnancies per woman/couple randomised. We defined live birth as the delivery of a live foetus after 20 completed weeks of gestational age. We defined ongoing pregnancy as evidence of a gestational sac with foetal heart activity at 12 weeks or later, confirmed by ultrasound. We decided to combine the two outcomes, as very few studies reported live births, and ongoing pregnancy comprises a meaningful clinical measure. Where there were multiple live births (e.g. twins or triplets), we counted these as one live birth event.

• Number of OHSS events per woman/couple randomised. We defined OHSS as an exaggerated systemic response to ovarian stimulation, characterised by a wide spectrum of clinical and laboratory manifestations. OHSS is classified as mild, moderate, or severe according to the degree of abdominal distension, ovarian enlargement, and respiratory, haemodynamic, and metabolic complications (Zegers‐Hochschild 2017).

Secondary outcomes

• Number of clinical pregnancies per woman/couple randomised, defined as evidence of a gestational sac with foetal heart activity after six weeks of gestation, confirmed by ultrasound. Where there were multiple gestational sacs in one woman, we counted these as one clinical pregnancy (Griffin 2002).

• Number of oocytes retrieved per woman randomised.

• Amount of gonadotropins administered per woman randomised.

Adverse outcomes

• Cycle cancellation (defined as cancelled cycle before oocyte retrieval).

• Number of pregnancy losses, defined as the sum of the number of miscarriages (pregnancy loss before 20 completed weeks of gestation) and the number of stillbirths (pregnancy loss after 20 completed weeks of gestation) (Griffin 2002).

• Number of premature LH surges.

Other outcomes

• Cost‐effectiveness of treatment.

• Acceptability of the intervention measure (Weiner 2017).

Search methods for identification of studies

We searched for all published and unpublished RCTs comparing the various regimens for pituitary suppression using GnRHa in ART, without language restriction and in consultation with the Cochrane Gynaecology and Fertility Information Specialist.

Electronic searches

We searched the following databases on 19 December 2022, using the search strategies developed by the Cochrane Gynaecology and Fertility Group:

• Cochrane Gynaecology and Fertility Group's Specialised Register, searched 19 December 2022, ProCite platform (Appendix 1);

• Cochrane Central Register of Controlled Trials (CENTRAL, Issue 11, 2022; the Cochrane Library), searched 19 December 2022, Ovid platform (Appendix 2);

• MEDLINE, searched from 1946 to 19 December 2022, Ovid platform (Appendix 3);

• Embase, searched from 1980 to 19 December 2022, Ovid platform (Appendix 4);

• CINAHL (Cumulative Index to Nursing and Allied Health Literature), searched from 1961 to 8 January 2020 (CINAHL references from 8 January to 9 December 2020 were included in the CENTRAL December 2020 output), EBSCO platform (Appendix 5).

Searching other resources

We searched the citation lists of relevant publications and review articles.

In liaison with the Cochrane Gynaecology and Fertility Information Specialist, we searched the abstracts of scientific meetings, including the European Society of Human Reproduction and Embryology (ESHRE) annual meeting of 2020, not covered in the Cochrane Gynaecology and Fertility Group Specialised Register.

We searched OpenGrey, the database for grey literature produced in Europe, including research reports, doctoral dissertations, and conference papers (www.opengrey.eu/).

We contacted experts in the field for any additional trials.

We searched the following trials registries for ongoing or registered trials on 19 December 2022:

• ISRCTN registry (www.isrctn.com);

• US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (clinicaltrials.gov);

• World Health Organization International Clinical Trials Registry Platform (trialsearch.who.int/Default.aspx).

Data collection and analysis

Selection of studies

Two review authors (SRCB, CS) independently selected trials for inclusion using forms designed according to Cochrane guidelines. In cases where trial methodology was unclear, or trial data were in a form unsuitable for meta‐analysis, we attempted to contact trial authors via email to obtain the additional information. Any differences of opinion about trial eligibility were resolved through discussion with a third review author (AM).

We documented the study selection process with a PRISMA flowchart (Figure 1).

1.

Study flow diagram.

We constructed Characteristics of included studies tables for eligible trials. We constructed Characteristics of excluded studies tables listing the excluded studies along with the reasons for their exclusion.

Data extraction and management

Two review authors (SRCB, SC) independently extracted data from eligible studies using a data extraction form that we had designed and pilot tested. Any disagreements were resolved through discussion with a third review author (AM). We extracted data on study characteristics and outcomes, as described in Appendix 6. Where studies had multiple publications, we collated the multiple reports of the same study so that each study, rather than each report, was the unit of interest in the review. Such studies have a single study identification with multiple references. We corresponded with study investigators via email to obtain further data on methods, results, or both, as required.

The data extraction forms included risk of bias criteria and methodological details, which we have presented in the Characteristics of included studies tables. We managed the data using RevMan software (RevMan 2024).

Assessment of risk of bias in included studies

Two review authors (SRCB, CS) independently assessed risk of bias in the included studies using Cochrane's RoB 1 tool (Higgins 2011). The tool assesses the following domains:

• selection bias (random sequence generation and allocation concealment);

• performance bias (blinding of participants and personnel);

• detection bias (blinding of outcome assessors);

• attrition bias (incomplete outcome data);

• reporting bias (selective reporting); and

• other bias.

Any disagreements between review authors were resolved by discussion with a third review author (AM). We described all judgements fully and presented them in the Characteristics of included studies tables, including commentary about each of the domains, which led to an overall assessment of risk of bias in the included studies (Figure 2; Figure 3).

2.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

3.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

We searched for within‐trial selective reporting, such as trials failing to report obvious outcomes or reporting them in insufficient detail to allow inclusion. We sought published protocols and compared the outcomes between the protocol and the final published study.

Measures of treatment effect

For dichotomous data (e.g. live birth and ongoing pregnancy rates), we used the numbers of events in the control and intervention groups of each study to calculate Mantel–Haenszel odds ratios (ORs). For continuous data (e.g. number of oocytes retrieved), we calculated the mean difference (MD) between treatment groups. We presented 95% confidence intervals (CIs) for all outcomes. Where data to calculate ORs or MDs were not available, our intention was to utilise the most detailed numerical data available that might facilitate similar analyses of included studies (e.g. test statistics, P values). We compared the magnitude and direction of effect reported by studies with how they were presented in the review, taking account of legitimate differences.

Unit of analysis issues

The primary analysis was per woman randomised; we included per‐pregnancy data for some outcomes (e.g. miscarriage). We counted multiple live births (e.g. twins or triplets) as one live birth event.

Dealing with missing data

In the case of missing data, we contacted the original investigators by email to request the relevant information. We sent a reminder if we had received no reply after 30 days. We reported the data according to the intention‐to‐treat principle wherever possible. We assumed that live births had not occurred in participants without a reported outcome. For other outcomes, we analysed only the available data.

If studies reported sufficient detail to calculate MDs but provided no information on the associated standard deviation (SD), we assumed the outcome to have an SD equal to the highest SD from other studies within the same analysis.

Assessment of heterogeneity

Before conducting meta‐analysis, we judged whether there was sufficient similarity between studies in terms of their design and clinical characteristics to ensure that pooling was valid. We assessed statistical heterogeneity in the results of trials by using the Chi² test. A low P value (or a large Chi² statistic relative to its degree of freedom) provides potential evidence of heterogeneity of intervention effects and shows that results are not influenced by chance alone (Higgins 2011). We used the I² statistic to assess the impact of the heterogeneity on the meta‐analysis, interpreting an I² statistic greater than 50% as marked heterogeneity (Higgins 2011).

Assessment of reporting biases

In view of the difficulty of detecting and correcting for publication bias and other reporting biases, we aimed to minimise their potential impact by ensuring a comprehensive search for eligible studies and by being alert to duplication of data. In the presence of 10 or more studies in an analysis, we used a funnel plot to explore the possibility of small‐study effects to determine whether the difference was due to publication or reporting bias. We are aware that there are other sources of asymmetry in funnel plots (Stuck 1998).

Data synthesis

For data synthesis, we distinguished the following 10 comparisons.

• Any long protocol versus any short protocol

• Any long protocol versus ultrashort protocol

• Any short protocol versus ultrashort protocol

• Long protocol: luteal phase versus follicular phase protocol

• Long protocol: continuing GnRHa versus stopping GnRHa ('long stop') at the start of stimulation

• Long protocol: continuation of same‐dose GnRHa versus reduced‐dose GnRHa until HCG administration

• Long protocol: discontinuing versus continuing GnRHa after HCG administration

• Long protocol: administration of GnRHa for two weeks versus three weeks before stimulation

• Short protocol: continuing GnRHa versus stopping GnRHa ('short stop')

• Short protocol: comparison of different doses

We performed analysis using RevMan software (RevMan 2024). For binary (or dichotomous) outcomes, we expressed the results for each study as OR with 95% CI, and combined them for meta‐analysis where appropriate. For continuous outcome data, we expressed the results from each study as a difference in means with 95% CI, and combined them for meta‐analysis using the MD.

An increase in the odds of a particular outcome, which may be beneficial (e.g. live birth) or detrimental (e.g. adverse effects), is displayed graphically in the meta‐analyses to the right of the centre‐line, and a decrease in the odds of an outcome to the left of the centre‐line.

Subgroup analysis and investigation of heterogeneity

Where data were available, we planned to conduct subgroup analyses for the primary outcomes and the secondary outcome of clinical pregnancy to determine separate evidence within the following subgroups: normal or poor responders, number of embryos transferred, previous failed cycles, maternal age, and duration of infertility. In cases of substantial heterogeneity, our aim was to explore possible explanations in sensitivity analyses. We took any statistical heterogeneity into account when interpreting the results, especially if there was any variation in the direction of effect. We used a fixed‐effect model.

Sensitivity analysis

The primary analysis was restricted to studies at low risk of selection bias and reporting bias for primary outcomes as well as the secondary outcome of clinical pregnancy. We conducted sensitivity analyses including all the studies to determine whether review conclusions were robust to arbitrary decisions made regarding eligibility and analysis. These analyses included consideration of whether the review conclusions would have differed if:

• we had included all studies in the analysis (i.e. no restriction to studies considered to be at low risk of selection and reporting bias);

• we had adopted a random‐effects model;

• the summary effect measure used was risk ratio (RR) rather than OR.

There were insufficient studies to support meta‐regression or other formal considerations of prognostic factors.

Summary of findings and assessment of the certainty of the evidence

Two review authors (SRCB, CS) prepared the summary of findings tables using GRADEpro GDT software (GRADEpro GDT). These tables evaluate the overall quality of the body of evidence for the main review outcomes (live birth/ongoing pregnancy, OHSS, clinical pregnancy) using GRADE criteria (study limitations (i.e. risk of bias), consistency of effect, imprecision, indirectness, and publication bias). We judged the certainty of the evidence as high, moderate, low, or very low. We justified our ratings and documented and incorporated these into the reporting of results for each outcome.

We presented summary of findings tables for seven comparison groups (long versus short protocol; long versus ultrashort protocol; long luteal versus follicular phase protocol; long continued versus long stop; long same‐dose versus reduced‐dose protocol; long two weeks versus three weeks agonist before stimulation protocol; and short protocol different doses) to summarise the findings and present the certainty of evidence.

Results

Description of studies

See Characteristics of included studies, Characteristics of excluded studies, Characteristics of ongoing studies, and the PRISMA flowchart (Figure 1).

Results of the search

Our December 2022 searches of the electronic databases yielded a total of 1893 records: 370 in the Cochrane Gynaecology and Fertility Group's Specialised Register; 529 in CENTRAL; 211 in MEDLINE; 691 in Embase; and 92 in CINAHL. After removal of duplicates, 1343 records remained. Two review authors (SRCB, CS) independently screened the titles and abstracts of these records, classifying them as 'retrieve' or 'do not retrieve'. We retrieved 20 full‐text publications, of which 15 were excluded. One study is awaiting classification, and one study is ongoing. We identified three new eligible studies, for a total of 40 included studies.

We sent two emails (initial and reminder) to trial authors for further information, and received responses from 10 (Chatillon‐Boissier 2012; Corson 1992; Isikoglu 2007; Lin 2013; NCT00436319; Sarhan 2013; Sarhan 2016; Sunkara 2014; Tanaka 2014).

Included studies

Design

We included 40 studies (4148 women). All studies were parallel‐group RCTs. There were 10 different comparisons (see Data synthesis).

Participants

1. Long versus short protocol

Inclusion criteria for the studies varied widely. Some studies included women with all causes of infertility (Acharya 1992; Tan 1992; Tasdemir 1995), while others included only women with tubal factor infertility (Fenichel 1988; Frydman 1988; Loumaye 1989; Van de‐Helder 1990a; Zhang 2009), or tubal and unexplained infertility (Hazout 1993; Hedon 1988). Some studies excluded women with polycystic ovary syndrome (PCOS) (Foulot 1988; Yang 1996).

The age of the women included varied across studies. Some studies included only women under 38 years (Fenichel 1988; Hazout 1993; Zhang 2009), while others included women up to the age of 40 years (Chatillon‐Boissier 2012; Loumaye 1989; Sunkara 2014; Van de‐Helder 1990a).

Some studies included only women undergoing their first IVF cycle (San Roman 1992; Tasdemir 1995), while others included all IVF cycles (Hazout 1993). Some studies included only previous low or poor responders (Chatillon‐Boissier 2012; Dirnfeld 1991; Sunkara 2014; Weissman 2003), whereas others excluded previous poor responders (Frydman 1988; Van de‐Helder 1990a).

2. Long versus ultrashort protocol

Two studies included couples with all causes of infertility: Kingsland 1992 only included women undergoing their first IVF cycle, whereas Chen 1992 included women with an average age of 33 years and various types of infertility (tubal factor (70%), male factor (10%), endometriosis (18%), and oocyte donation (2%)).

3. Short versus ultrashort protocol

A total of 82 poor responders who underwent ICSI were included in this comparison. Criteria included at least one of the following: day three serum follicle‐stimulating hormone (FSH) level below 10 milli‐international units (mIU)/mL; fewer than six total antral follicles; prior cycle cancellation; prior poor response to controlled ovarian hyperstimulation (COH) (either peak oestradiol (E2) < 500 pg/mL, fewer than six oocytes retrieved, or both); and aged less than 41 years (Berker 2010).

4. Long protocol: luteal versus follicular start of GnRHa

One study included women undergoing their first IVF cycle only (Kondaveeti‐Gordon 1996). Ron‐El 1990 included 216 women undergoing IVF, but the inclusion criteria were not specified, whereas Pellicer 1989 included women with normal ovarian function; Urbancsek 1996 included women with tubal and unexplained infertility; and Sarhan 2013 included women with all types of infertility.

5. Long protocol: continuing GnRHa versus stopping GnRHa at start of stimulation

Dirnfeld 1999 included only women with a previous poor response or high FSH and up to the age of 42 years and excluded those with irregular menstrual cycles. Simons 2005 included only women under 39 years of age and excluded women with PCOS or poor ovarian reserve, while Garcia‐Velasco 2000 reported no exclusion criterion for age. Some studies included only previous low responders (Garcia‐Velasco 2000; Simons 2005).

6. Long protocol: continuation of same‐dose GnRHa versus reduced‐dose GnRHa until HCG administration

Inclusion criteria for the studies varied widely. Simon 1994 restricted inclusion to only tubal factor infertility participants, while Dal Prato 2001 included tubal and unexplained infertility participants. Dal Prato 2001 excluded women with a risk of hyperstimulation or with poor ovarian reserve, while Ding 2013 included women with high response to gonadotropin stimulation, that is "women with eight or more subcapsular follicles of 2 to 8 mm in diameter in one plane in either ovary". The upper age limit of included women varied across studies: under 35 years (Ding 2013), under 38 years (Dal Prato 2001), and under 39 years (Simon 1994). Fábregues 2005 and Ding 2013 included women undergoing their first IVF cycle.

7. Long protocol: discontinuing versus continuing GnRHa after HCG administration

In the sole included study for this comparison (Isikoglu 2007), 181 women undergoing 181 ICSI cycles were included with a mean age of 30 years. We identified further studies, but we have not attempted to extend the analysis as this comparison is now a topic of another Cochrane review (Van der Linden 2015).

8. Long protocol: administration of GnRHa for two versus three weeks before stimulation

In the sole included study for this comparison (Lin 2013), 100 participants undergoing IVF/ICSI cycle were included with a mean age of 29 years. The inclusion criteria in Lin 2013 were: (a) subfertile participants undergoing first IVF/ICSI with tubal factor, male factor, or unexplained factor; (b) undertaking a luteal long protocol; (c) basal FSH levels 10 IU/L; and (d) age ≤ 35 years. The study excluded women with endometriosis, adenomyosis, and PCOS.

9. Short protocol: continuing versus stopping GnRHa

Cedrin‐Durnerin 2000 excluded women older than 43 years and those with anovulation.

10. Short protocol: comparison of different doses of GnRHa

The Sarhan 2016 study included a total of 91 participants divided into three arms to compare three different doses of GnRH agonist, which we included in our review. Inclusion and exclusion criteria were not specified.

The Ghaffari 2020 study had 200 participants meeting at least two of the following criteria: advanced maternal age (≥ 38 years); day two to three serum FSH higher than 12 mIU/mL; antral follicle count of four or less; prior poor response to COH (< 3 oocytes retrieved); and prior cycle cancellation. Exclusion criteria were: one ovary; myoma of 6 cm or larger; follicles larger than 13 mm; age older than 42 years; or severe male factor infertility (azoospermia).

We collected data from the first two arms of the Al‐Jeborry 2020 study, which encompassed 76 participants (38 in each group). This study included women with poor ovarian reserve according to Patient‐Oriented Strategies Encompassing IndividualiseD Oocyte Number (POSEIDON) criteria, and excluded women under 18 or over 43 years of age, with a body mass index (BMI) above 30, severe endometriosis, premature ovarian failure, thyroid dysfunction, or hyperprolactinaemia.

Interventions

1. Long versus short protocol

Nineteen trials compared a long protocol with a short protocol. GnRHa was commenced in the follicular phase in six of these studies (Acharya 1992; Foulot 1988; Frydman 1988; Hazout 1993; Hedon 1988; Tan 1992), and in the luteal phase in the remaining studies (Chatillon‐Boissier 2012; De Placido 1991; Fenichel 1988; Loumaye 1989; San Roman 1992; Sunkara 2014; Tasdemir 1995; Van de‐Helder 1990a; Weissman 2003; Ye 2001; Zhang 2009). In two studies (Dirnfeld 1991; Yang 1996), it was unclear whether a follicular or luteal start was employed.

There were wide variations in the dose, type, and route of GnRHa used for downregulation in long protocols. Buserelin was administered either by nasal spray or subcutaneous injections: 1000 µg twice a day (Dirnfeld 1991); 200 µg five times a day (Acharya 1992); 900 µg/day (Loumaye 1989; Tasdemir 1995; Ye 2001); 300 µg twice a day (De Placido 1991; Frydman 1988; Hedon 1988); 200 µg three times daily (Van de‐Helder 1990a); 0.3 mL daily (Foulot 1988); 200 µg daily (Tan 1992); and 100 µg/day (Weissman 2003). Triptorelin (Decapeptyl) was used either as a short‐acting (100 µg) (Chatillon‐Boissier 2012) or long‐acting single intramuscular injection (3.75 mg) (Fenichel 1988), or 1.88 mg of intramuscular Diphereline (Zhang 2009). Other studies used leuprorelin (1 mg/day) (San Roman 1992; Yang 1996). Hazout 1993 repeated the Decapeptyl injection twice, which may explain the much higher requirement of gonadotropins. One study, Sunkara 2014, used nafarelin nasal spray 400 mg twice daily.

In studies comparing a long protocol versus a short protocol, GnRHa was continued at the same dose until HCG administration, except in five studies that reduced the dose at confirmation of downregulation: Dirnfeld 1991 and San Roman 1992 reduced the dose from 1000 µg to 600 µg and from 1 mg to 0.5 mg/day, respectively; Weissman 2003 and Chatillon‐Boissier 2012 halved the agonist dose; and Sunkara 2014 continued with a reduced dose of nafarelin 200 mg twice daily until the administration of HCG injection.

Similarly, the dose of GnRHa for short protocols varied. Weissman 2003 applied a modified short protocol using the flare effect initially (500 µg/day for the initial four days followed by 100 µg until the day of HCG). Yang 1996 used another modification of the short protocol, where GnRHa was stopped after seven days.

Dose, regimen, and drugs used for stimulation also varied in all studies, as did the inclusion criteria of the population studied (see Characteristics of included studies).

2. Long versus ultrashort protocol

Of the two studies included in this comparison, Kingsland 1992 used 200 µg daily of buserelin, whereas Chen 1992 used 1 mg daily of subcutaneous triptorelin (Decapeptyl) for the long protocol. Both studies discontinued GnRHa after confirmation of downregulation.

The dose of GnRHa for the ultrashort protocol differed across studies as well. Chen 1992 used leuprorelin 1 mg daily, whereas Kingsland 1992 used 500 µg/day of buserelin on days two, three, and four of the cycle.

Chen 1992 used FSH + human menopausal gonadotropin (HMG) for stimulation, whereas Kingsland 1992 used HMG alone.

3. Short versus ultrashort protocol

Participants in the Berker 2010 study were randomised into two groups, as follows.

• Participants in the ultrashort GnRH agonist/GnRH antagonist group (n = 41) were administered leuprorelin at 40 µg subcutaneously twice daily, started on day two of menses and continued for three consecutive days, followed by gonadotropins, and the GnRH antagonist cetrorelix at 0.25 mg/day when the leading follicle was more than 14 mm, which was continued up to HCG injection.

• Participants in the microdose group (n = 41) began use of leuprorelin at 40 µg subcutaneously twice daily on day two of menses, and two days after initiation of GnRHa, gonadotropin stimulation was initiated and continued until HCG day.

The starting dose of recombinant FSH depended on age, BMI, and ovarian response to the previous cycle and was increased to a maximum of 450 IU/day depending on the ovarian response; it was then individualised after day five (Berker 2010).

4. Long protocol: luteal versus follicular start of GnRHa

Two out of five studies included in this comparison used the same dose of GnRHa for suppression (1200 µg/day) (Kondaveeti‐Gordon 1996; Urbancsek 1996). Sarhan 2013 used 0.1 mg of triptorelin subcutaneously daily; Ron‐El 1990 used a long‐acting preparation (3.2 mg triptorelin (Decapeptyl)), whereas Pellicer 1989 used 600 µg/day of buserelin in two divided doses. In Pellicer 1989, the day for luteal start varied, ranging from 4 to 10 days after ovulation compared with the day 21 to 22 start in the other included studies. This could have impacted the outcomes of the luteal phase results. Urbancsek 1996 considered more than one cycle per woman, whereas the remaining four studies evaluated only the first cycle. All studies except Pellicer 1989 and Sarhan 2013 used HMG for ovarian stimulation; the former used HMG plus FSH, and the latter administered either HMG or FSH.

5. Long protocol: continuing GnRHa versus stopping GnRHa at start of stimulation

Of the three studies included in this comparison, one used buserelin (1000 µg/day) (Dirnfeld 1999); one used leuprorelin (1 mg/day) (Garcia‐Velasco 2000); and the third used triptorelin (0.1 mg/day) for downregulation (Simons 2005). All studies stopped GnRHa at confirmation of downregulation in the control group. Apart from one study (Garcia‐Velasco 2000), which used FSH plus HMG, all studies used HMG alone for COH.

6. Long protocol: continuation of same‐dose GnRHa versus reduced‐dose GnRHa until HCG administration

Across the four studies included in this comparison, there were variations in type and dose of GnRHa and the reduction in dose after downregulation was confirmed: luteinising hormone‐releasing hormone agonist (LHRHa) commenced at 0.5 mg/day and was reduced to 0.1 mg/day (Simon 1994); triptorelin acetate commenced at 0.1 mg/day and was reduced to 0.05 mg/day (Fábregues 2005); GnRHa commenced at 100 µg/day and was reduced to 50 µg/day (Dal Prato 2001); and triptorelin was initiated during the luteal phase, 0.1 mg/day for 10 days followed by 0.05 mg/day until the concentration of serum oestradiol was 40 pg/mL or less, then ovarian stimulation was started, and when the diameter of one or more follicles was 14 mm, triptorelin (0.05 mg/day) was withdrawn for two (15/47) or three (32/47) days (Ding 2013).

The COH regimen varied across studies: Simon 1994 used HMG; Fábregues 2005 and Ding 2013 used recombinant FSH; and Dal Prato 2001 used urofollitropin (Metrodin).

7. Long protocol: discontinuing versus continuing GnRHa after HCG administration

In the only study included in this comparison (Isikoglu 2007), GnRHa was administered from the 21st day of the preceding cycle. Participants were divided into two groups: 90 participants were continuously administered GnRHa for 12 days after embryo transfer versus 90 participants whose GnRHa was discontinued on the day of HCG administration.

8. Long protocol: administration of GnRHa for two versus three weeks before stimulation

In the only study included in this comparison (Lin 2013), a single dose of long‐acting GnRHa (triptorelin (Diphereline), 1.25 mg) was administered in the mid‐luteal phase, in both groups. Participants were divided into two groups according to the initiation of gonadotropins (14 versus 21 days after GnRHa administration). Either recombinant follicle‐stimulating hormone (rFSH) or HMG was used for COH.

9. Short protocol: continuing versus discontinuing GnRHa

There was only one study in this comparison (Cedrin‐Durnerin 2000). A short protocol was compared with stopping GnRHa halfway through COH rather than continuing until the day of HCG.

10. Short protocol: comparison of different doses of GnRHa, microdose versus standard dose

In the three studies included in this comparison, the type and dose of GnRHa used in the short protocol varied: 80 μg versus 500 μg buserelin daily (Ghaffari 2020); 100 μg, 50 μg, or 25 μg of triptorelin daily (Sarhan 2016); and 25 μg versus 100 μg of triptorelin daily (Al‐Jeborry 2020).

Outcomes

Twenty studies reported on either live birth rate or ongoing pregnancy rates (Acharya 1992; Chatillon‐Boissier 2012; Ding 2013; Dirnfeld 1991; Dirnfeld 1999; Foulot 1988; Frydman 1988; Ghaffari 2020; Hedon 1988; Isikoglu 2007; Kingsland 1992; Lin 2013; Loumaye 1989; San Roman 1992; Simons 2005; Sunkara 2014; Urbancsek 1996; Van de‐Helder 1990a; Ye 2001; Zhang 2009). Regarding adverse outcomes, 24 studies reported on cycle cancellation rates (Acharya 1992; Berker 2010; Cedrin‐Durnerin 2000; Chatillon‐Boissier 2012; Dal Prato 2001; Ding 2013; Dirnfeld 1991; Dirnfeld 1999; Foulot 1988; Frydman 1988; Garcia‐Velasco 2000; Ghaffari 2020; Hazout 1993; Isikoglu 2007; Kingsland 1992; Kondaveeti‐Gordon 1996; San Roman 1992; Sarhan 2013; Sarhan 2016; Simons 2005; Sunkara 2014; Van de‐Helder 1990a; Weissman 2003; Zhang 2009), while two trials reported on OHSS (Ding 2013; Lin 2013), and two studies reported on miscarriage rates (Ghaffari 2020; Lin 2013); no studies reported stillbirths or premature LH surge. Acceptability of the regimens was not reported in any of the included studies.

Excluded studies

A list of the 77 excluded studies along with the reasons for their exclusion is provided in Characteristics of excluded studies. We excluded studies for the following reasons:

• ineligible study design (non‐randomised) (36 studies);

• ineligible intervention (32 studies);

• incomplete data (4 studies);

• different study populations (4 studies);

• different primary outcome (1 study).

Risk of bias in included studies

A complete overview of our classification of risk of bias domains can be found in Characteristics of included studies; risk of bias summaries are shown in Figure 2 and Figure 3. The risk of bias domains for the various comparison groups are as follows.

1. Long versus short protocol

This comparison included a total of 19 studies. Only nine of these studies reported on random sequence generation (Chatillon‐Boissier 2012; Dirnfeld 1991; Fenichel 1988; Foulot 1988; Hazout 1993; Sunkara 2014; Tan 1992; Weissman 2003; Ye 2001). Two studies reported on allocation concealment (Sunkara 2014; Tan 1992). Blinding was not undertaken for the majority of studies. The funnel plot did not suggest any publication bias (Figure 4).

4.

Funnel plot of comparison 1: long versus short protocol; outcome 1.2: clinical pregnancies.

2. Long versus ultrashort protocol

This comparison included two studies (Chen 1992; Kingsland 1992). Kingsland 1992 reported on random sequence generation and allocation concealment.

3. Short versus ultrashort protocol

This comparison included one study (Berker 2010). The paper described random sequence generation and allocation concealment, but no blinding.

4. Long protocol: luteal versus follicular start of GnRHa

This comparison included five studies. Kondaveeti‐Gordon 1996 and Sarhan 2013 reported on random sequence generation and allocation concealment. Although blinding until objective outcome assessment was planned for one study (Kondaveeti‐Gordon 1996), this was not followed after the study was started. Ron‐El 1990, Pellicer 1989, and Urbancsek 1996 did not mention random sequence generation, allocation concealment, or blinding. Urbancsek 1996 reported more than one cycle per participant.

5. Long protocol: continuing GnRHa versus stopping GnRHa at start of stimulation

This comparison included three studies (Dirnfeld 1999; Garcia‐Velasco 2000; Simons 2005). Two of these studies reported random sequence generation (Garcia‐Velasco 2000; Simons 2005), and only one study reported allocation concealment and was double‐blinded (Simons 2005).

6. Long protocol: continuation of same‐dose GnRHa versus reduced‐dose GnRHa until HCG administration

This comparison included four studies (Dal Prato 2001; Ding 2013; Fábregues 2005; Simon 1994). Three of these studies reported on random sequence generation and allocation concealment, but they were not blinded (Dal Prato 2001; Ding 2013; Fábregues 2005).

7. Long protocol: discontinuing versus continuing GnRHa after HCG administration

This comparison included one study (Isikoglu 2007), which reported adequate random sequence generation.

8. Long protocol: administration of GnRHa for two versus three weeks before stimulation

This comparison included one study (Lin 2013). The study reported on random sequence generation (computer‐generated random numbers two weeks after GnRHa administration), but allocation concealment was unclear, and there was no blinding.

9. Short protocol: continuing versus stopping GnRHa

This comparison included one study (Cedrin‐Durnerin 2000). The study reported random sequence generation, but allocation concealment was unclear, and there was no blinding.

10. Short protocol: comparison of different doses of GnRHa

This comparison included three studies (Al‐Jeborry 2020; Ghaffari 2020; Sarhan 2016). Ghaffari 2020 reported computer‐based random sequence generation, but no allocation concealment or blinding. The other two studies did not report on randomisation method, allocation concealment, or blinding of participants (Al‐Jeborry 2020; Sarhan 2016).

Allocation

Random sequence generation

Twenty‐two studies reported adequate sequence generation and were assessed as having a low risk of selection bias. The remaining 18 studies gave no clear description of the method of randomisation (Acharya 1992; Al‐Jeborry 2020; Chen 1992; De Placido 1991; Dirnfeld 1999; Frydman 1988; Hedon 1988; Loumaye 1989; Pellicer 1989; Ron‐El 1990; San Roman 1992; Sarhan 2016; Simon 1994; Tasdemir 1995; Urbancsek 1996; Van de‐Helder 1990a; Yang 1996; Zhang 2009), so we judged these to be at unclear risk of bias.

Allocation concealment

Eight studies used adequate methods for concealment of the random sequence (sealed envelopes) and were judged as at low risk of selection bias (Berker 2010; Dal Prato 2001; Ding 2013; Fábregues 2005; Kingsland 1992; Simons 2005; Sunkara 2014; Tan 1992). Twenty‐seven studies did not report allocation concealment and were judged to be at unclear risk of bias. We assessed five studies as at high risk of selection bias, as the authors reported no concealment of allocation (Dirnfeld 1991; Fenichel 1988; Hazout 1993; Kondaveeti‐Gordon 1996; Ye 2001).

Blinding

Although our outcomes of interest were objective, we believe that blinding of clinicians and participants is important in order to avoid performance and detection biases. Twenty‐seven studies did not blind the clinician or participants and were therefore considered to be at high risk of bias. We judged only two studies to be at low risk of bias in this domain (Simons 2005; Sunkara 2014). We assessed 11 studies as at unclear risk of bias, as blinding was not reported (Al‐Jeborry 2020; Berker 2010; De Placido 1991; Ding 2013; Foulot 1988; Frydman 1988; Garcia‐Velasco 2000; Sarhan 2013; Sarhan 2016; Yang 1996; Zhang 2009).

Incomplete outcome data

We assessed nine studies as at high risk (Ding 2013; Dirnfeld 1999; Fábregues 2005; Ghaffari 2020; Hazout 1993; Hedon 1988; Simon 1994; Tasdemir 1995; Urbancsek 1996), three studies as at unclear risk (Chen 1992; De Placido 1991; Pellicer 1989), and the remaining 28 studies as at low risk of attrition bias (Acharya 1992; Al‐Jeborry 2020; Berker 2010; Cedrin‐Durnerin 2000; Chatillon‐Boissier 2012; Dal Prato 2001; Dirnfeld 1991; Fenichel 1988; Foulot 1988; Frydman 1988; Garcia‐Velasco 2000; Isikoglu 2007; Kingsland 1992; Kondaveeti‐Gordon 1996; Lin 2013; Loumaye 1989; Ron‐El 1990; San Roman 1992; Sarhan 2013; Sarhan 2016; Simons 2005; Sunkara 2014; Tan 1992; Van de‐Helder 1990a; Weissman 2003; Yang 1996; Ye 2001; Zhang 2009).

Selective reporting

Twenty studies reported our primary outcome: live birth or ongoing pregnancy rate. We judged 16 studies to be at low risk of reporting bias (Chatillon‐Boissier 2012; Ding 2013; Dirnfeld 1991; Foulot 1988; Frydman 1988; Ghaffari 2020; Hedon 1988; Isikoglu 2007; Kingsland 1992; Lin 2013; Loumaye 1989; San Roman 1992; Simons 2005; Sunkara 2014; Urbancsek 1996; Van de‐Helder 1990a). Twenty studies did not report the primary outcome for this review, of which 18 were judged to be at unclear risk of reporting bias. We judged two trials to be at high risk of reporting bias (Al‐Jeborry 2020; De Placido 1991). The De Placido 1991 study only reported clinical pregnancy. In addition, the study did not mention numbers of participants undergoing IVF, nor any other outcomes, and we received no reply to our requests for information (De Placido 1991). In the Al‐Jeborry 2020 study, clinical pregnancy rates were reported as percentages, while actual numbers were not presented. We received no reply to our requests for information.

Other potential sources of bias

In over half of the included studies (n = 25), there was insufficient information to assess whether another potential source of bias existed; we judged these studies to be at unclear risk of other bias.

We judged six trials as being at high risk of other bias for different reasons, as these might have changed the magnitude of the effect estimate (Cedrin‐Durnerin 2000; De Placido 1991; Dirnfeld 1991; Pellicer 1989; Sarhan 2016; Tasdemir 1995). In Tasdemir 1995, the median number of embryos transferred was significantly different between the intervention and the control group, and the study did not describe the exact number of participants in each group. In another study (Sarhan 2016), inclusion and exclusion criteria were not specified apart from reporting that 91 normal ovulatory women undergoing ICSI had participated. In two other trials, data regarding the number of participants and other inclusion criteria were lacking (De Placido 1991), and participants with chronic anovulation were excluded (Cedrin‐Durnerin 2000). In one study, the intervention and the control group commenced GnRHa on different days (Pellicer 1989). In Dirnfeld 1991, the long GnRH protocol was commenced in either the luteal or follicular phase.

We judged the remaining nine trials as at low risk of other bias.

Effects of interventions

See: Table 1; Table 2; Table 3; Table 4; Table 5; Table 6; Table 7

1. Long versus short protocol

We included 19 RCTs in this comparison (Acharya 1992; Chatillon‐Boissier 2012; De Placido 1991; Dirnfeld 1991; Fenichel 1988; Foulot 1988; Frydman 1988; Hazout 1993; Hedon 1988; Loumaye 1989; San Roman 1992; Sunkara 2014; Tan 1992; Tasdemir 1995; Van de‐Helder 1990a; Weissman 2003; Yang 1996; Ye 2001; Zhang 2009). See Table 1.

Primary outcomes

1.1 Live birth and ongoing pregnancy rates

A total of 12 studies reported on live birth and ongoing pregnancy rates.

1.1.1 Primary analysis

We included five studies with low risk of selection and attrition bias (Chatillon‐Boissier 2012; Dirnfeld 1991; Foulot 1988; Sunkara 2014; Ye 2001), and found that there may be little or no difference in live birth and ongoing pregnancy rates between the two protocols (odds ratio (OR) 1.45, 95% confidence interval (CI) 0.83 to 2.52; I² = 0%; 5 studies, 381 women; low‐certainty evidence; Analysis 1.1; Figure 5).

1.1. Analysis.

Comparison 1: Long versus short protocol, Outcome 1: Live birth/ongoing pregnancies primary analysis

5.

Forest plot of comparison 1: long versus short protocol; outcome 1.1: live birth/ongoing pregnancies.

1.1.2 Sensitivity analysis

We undertook our pre‐planned sensitivity analysis, including all studies reporting on the primary outcome of live birth/ongoing pregnancy rate (Analysis 1.2). Results showed that there may be no difference in the live birth/ongoing pregnancy rate between the two groups (OR 1.30, 95% CI 0.94 to 1.81; I² = 0%; 12 studies, 976 women; low‐certainty evidence). Results were similar if analysed using a random‐effects model (OR 1.25, 95% CI 0.89 to 1.76; I² = 0%; 12 studies, 976 women; low‐certainty evidence) or presented as risk ratio (RR 1.22, 95% CI 0.95 to 1.57; 12 studies, 976 women; low‐certainty evidence). There was also no difference when analysing one study, Sunkara 2014, with low risk of bias in all domains (OR 1.00, 95% CI 0.19 to 5.31; 1 study, 74 women; low‐certainty evidence).

1.2. Analysis.

Comparison 1: Long versus short protocol, Outcome 2: Live birth/ongoing pregnancies all studies

1.2 Ovarian hyperstimulation syndrome (OHSS) rate

None of the studies reported on this outcome.

Secondary outcomes

1.3 Clinical pregnancy rate

1.3.1 Primary analysis

We included eight studies with low risk of selection bias and attrition bias. The long protocol may improve clinical pregnancy rate when compared to the short protocol (OR 1.56, 95% CI 1.01 to 2.40; I² = 23%; 8 studies, 552 women; low‐certainty evidence; Analysis 1.3). Subgroup analysis of four studies each in the unselected group and poor responder group showed that, for the subgroup of studies including poor responders only, the long protocol may improve the clinical pregnancy rate compared to the short protocol (OR 3.12, 95% CI 1.39 to 7.02; I² = 0%; 4 studies, 232 women; low‐certainty evidence), but there was no difference between the two protocols in the unselected group (OR 1.13, 95% CI 0.67 to 1.91; I² = 14%; 4 studies, 320 women; low‐certainty evidence; Analysis 1.3; Figure 6).

1.3. Analysis.

Comparison 1: Long versus short protocol, Outcome 3: Clinical pregnancies primary analysis

6.

Forest plot of comparison 1: long versus short protocol; outcome 1.2: clinical pregnancies.

1.3.2 Sensitivity analysis

We undertook sensitivity analysis testing, including all the studies reporting on the primary outcome of clinical pregnancies (Analysis 1.4). The long protocol may improve clinical pregnancy rate overall when compared to the short protocol (OR 1.50, 95% CI 1.18 to 1.92; I² = 27%; 19 studies, 1582 women; low‐certainty evidence). Results were similar if analysed using a random‐effects model (OR 1.50, 95% CI 1.10 to 2.05; I² = 27%; 19 studies, 1582 women; low‐certainty evidence) or presented as risk ratio (RR 1.36, 95% CI 1.13 to 1.64; I² = 22%; 19 studies, 1582 women; low‐certainty evidence) and when one study with low risk of bias in all domains was analysed (Sunkara 2014: OR 2.19, 95% CI 0.50 to 9.53; 74 women; low‐certainty evidence).

1.4. Analysis.

Comparison 1: Long versus short protocol, Outcome 4: Clinical pregnancies all studies

1.4 Number of oocytes retrieved

Due to the high heterogeneity of the pooled analysis (10 studies, 739 women; I² = 91%), we did not pool data. The heterogeneity was among the six studies of unselected women. Two of these studies showed a significant difference in favour of the long protocol. Subgroup analysis of the four studies including poor responders showed that the long protocol may improve the number of oocytes when compared with the short protocol (mean difference (MD) 1.40, 95% CI 0.75 to 2.06; I² = 0%; 4 studies, 227 women; low‐certainty evidence; Analysis 1.5).

1.5. Analysis.

Comparison 1: Long versus short protocol, Outcome 5: Number of oocytes

1.5 Amount of gonadotropins administered

Due to the high heterogeneity of the pooled analysis (8 studies, 666 women; I² = 97%), we did not pool data. The heterogeneity was among the four studies of unselected women. All of these studies showed a significant difference in favour of the long protocol. Subgroup analysis of the studies including poor responders showed evidence of a substantial increase in the requirement of gonadotropins in the long protocol compared with the short protocol (MD 7.07, 95% CI 3.06 to 11.08; I² = 0%; 4 studies, 227 women; low‐certainty evidence; Analysis 1.6).

1.6. Analysis.

Comparison 1: Long versus short protocol, Outcome 6: Amount of gonadotropins administered

1.6 Cycle cancellation rate

There may be no difference between groups in cycle cancellation rate (OR 0.95, 95% CI 0.59 to 1.55; I² = 42%; 11 studies, 1026 women; low‐certainty evidence; Analysis 1.7). Subgroup analysis of the four studies including poor responders showed fewer cancellations in the long protocol compared with the short protocol (OR 0.31, 95% CI 0.12 to 0.76; I² = 0%; 4 studies, 227 women; low‐certainty evidence; Analysis 1.7).

1.7. Analysis.

Comparison 1: Long versus short protocol, Outcome 7: Cycle cancellation

1.7 Pregnancy loss and other outcomes

No studies for this comparison reported on pregnancy loss (miscarriage and stillbirth), premature LH surges, cost‐effectiveness, or acceptability of these two protocols.

2. Long versus ultrashort protocol

We included two RCTs in this comparison (Chen 1992; Kingsland 1992). See Table 2.

Primary outcomes

2.1 Live birth and ongoing pregnancy rates

2.1.1 Primary analysis

We analysed one study with low risk of bias for random sequence generation and outcome reporting (Kingsland 1992). We are uncertain if a long protocol improves live birth and ongoing pregnancy rates when compared with an ultrashort protocol (OR 1.78, 95% CI 0.72 to 4.36; 1 study, 150 women; very low‐certainty evidence; Analysis 2.1).

2.1. Analysis.

Comparison 2: Long protocol versus ultrashort protocol, Outcome 1: Live birth and ongoing pregnancies

2.1.1 Sensitivity analysis

Results were similar when analysed using a random‐effects model (OR 1.78, 95% CI 0.72 to 4.36; 1 study, 150 women; very low‐certainty evidence) or presented as risk ratio (RR 1.62, 95% CI 0.76 to 3.48; 1 study, 150 women; very low‐certainty evidence).

2.2 OHSS

Neither of the two studies reported on this outcome.

Secondary outcomes

2.3 Clinical pregnancy rate

We are uncertain if a long protocol improves clinical pregnancy rate when compared to an ultrashort protocol (OR 1.56, 95% CI 0.80 to 3.06; I² = 0%; 2 studies, 230 women; very low‐certainty evidence; Analysis 2.2).

2.2. Analysis.

Comparison 2: Long protocol versus ultrashort protocol, Outcome 2: Clinical pregnancies

2.4 Number of oocytes retrieved

We are uncertain if there is a difference in the number of oocytes retrieved between the long protocol and the ultrashort protocol (MD 0.53, 95% CI −0.61 to 1.66; I² = 67%; 2 studies, 230 women; very low‐certainty evidence; Analysis 2.3).

2.3. Analysis.

Comparison 2: Long protocol versus ultrashort protocol, Outcome 3: Number of oocytes

2.5 Amount of gonadotropins administered

We are uncertain if there is a difference in the amount of gonadotropins used between the long protocol and the ultrashort protocol (MD 1.10, 95% CI −1.81 to 4.01; 1 study, 80 women; very low‐certainty evidence; Analysis 2.4).

2.4. Analysis.

Comparison 2: Long protocol versus ultrashort protocol, Outcome 4: Amount of gonadotropins administered

2.6 Cycle cancellation rate

We are uncertain if a long protocol compared to an ultrashort protocol improves cycle cancellation rate (OR 1.11, 95% CI 0.40 to 3.05; 1 study, 150 women; very low‐certainty evidence; Analysis 2.5).

2.5. Analysis.

Comparison 2: Long protocol versus ultrashort protocol, Outcome 5: Cycle cancellation

2.7 Pregnancy loss and other outcomes

Neither study for this comparison reported on pregnancy loss (miscarriage and stillbirth), premature LH surges, cost‐effectiveness, or acceptability of these two protocols.

3. Short versus ultrashort protocol

We found only one RCT for this comparison (Berker 2010).

Primary outcomes

None of the primary outcomes were reported in this study.

Secondary outcome measures

3.1 Clinical pregnancy rate

We are uncertain if there is any difference in clinical pregnancy rate between the short protocol and the ultrashort protocol (OR 1.33, 95% CI 0.47 to 3.81; 1 study, 82 women; very low‐certainty evidence; Analysis 3.1).

3.1. Analysis.

Comparison 3: Short versus ultrashort protocol, Outcome 1: Clinical pregnancies

3.2 Number of oocytes retrieved

We are uncertain if there is any difference in the number of oocytes retrieved between the short protocol and the ultrashort protocol (MD 0.70, 95% CI −1.83 to 3.23; 1 study, 82 women; very low‐certainty evidence; Analysis 3.2).

3.2. Analysis.

Comparison 3: Short versus ultrashort protocol, Outcome 2: Number of oocytes

3.3 Amount of gonadotropins administered

We are uncertain if there is any difference in the amount of gonadotropins used in the short protocol when compared with the ultrashort protocol (MD −13.85, 95% CI −21.49 to −6.21; 1 study, 82 women; very low‐certainty evidence; Analysis 3.3).

3.3. Analysis.

Comparison 3: Short versus ultrashort protocol, Outcome 3: Amount of gonadotropins administered

3.4 Cycle cancellation rate

We are uncertain if there is any difference in cycle cancellation rate between the short protocol and the ultrashort protocol (OR 1.00, 95% CI 0.13 to 7.46; 1 study, 82 women; very low‐certainty evidence; Analysis 3.4).

3.4. Analysis.

Comparison 3: Short versus ultrashort protocol, Outcome 4: Cycle cancellation

3.5 Pregnancy loss and other outcomes

Berker 2010 did not report on pregnancy loss (miscarriage and stillbirth), premature LH surges, cost‐effectiveness, or acceptability of either of these protocols.

4. Long luteal versus long follicular phase protocol

We included five RCTs in this comparison (Kondaveeti‐Gordon 1996; Pellicer 1989; Ron‐El 1990; Sarhan 2013; Urbancsek 1996). See Table 3.

Primary outcomes

4.1 Live birth and ongoing pregnancy rates

4.1.1 Primary analysis

One study reported this outcome (Urbancsek 1996). We are uncertain if there is any difference in ongoing pregnancy rates with a long luteal versus a long follicular phase protocol (OR 1.89, 95% CI 0.87 to 4.10; 1 study, 223 women; very low‐certainty evidence; Analysis 4.1).

4.1. Analysis.

Comparison 4: Long protocol: luteal versus follicular phase, Outcome 1: Live birth and ongoing pregnancies

4.1.2 Sensitivity analysis

Results were similar when analysed using a random‐effects model (OR 1.89, 95% CI 0.87 to 4.10; 1 study, 223 women; very low‐certainty evidence) or presented as risk ratio (RR 1.73, 95% CI 0.88 to 3.39; 1 study, 223 women; very low‐certainty evidence).

4.2 OHSS

None of the studies reported on this outcome.

Secondary outcomes

4.3 Clinical pregnancy rate

We are uncertain if there is any difference in pregnancy rate between the long luteal and the long follicular phase protocols (OR 1.06, 95% CI 0.76 to 1.47; I² = 52%; 5 studies, 750 women; very low‐certainty evidence; Analysis 4.2).

4.2. Analysis.

Comparison 4: Long protocol: luteal versus follicular phase, Outcome 2: Clinical pregnancies

4.4 Number of oocytes retrieved

We are uncertain if there is any difference between groups in the number of oocytes retrieved (MD −1.29, 95% CI −1.86 to −0.71; I² = 74%; 4 studies, 527 women; very low‐certainty evidence; Analysis 4.3).

4.3. Analysis.

Comparison 4: Long protocol: luteal versus follicular phase, Outcome 3: Number of oocytes

4.5 Amount of gonadotropins administered

We are uncertain if there is any difference in the amount of gonadotropins required between the long luteal and the long follicular phase protocols (MD 1.12, 95% CI −0.73 to 2.97; I² = 51%; 4 studies, 527 women; very low‐certainty evidence; Analysis 4.4).

4.4. Analysis.

Comparison 4: Long protocol: luteal versus follicular phase, Outcome 4: Amount of gonadotropins administered

4.6 Cycle cancellation

We are uncertain if there is any difference in cycle cancellation rates between the long luteal and the long follicular phase protocols (OR 1.45, 95% CI 0.35 to 6.01; I² = 0%; 2 studies, 267 women; very low‐certainty evidence; Analysis 4.5).

4.5. Analysis.

Comparison 4: Long protocol: luteal versus follicular phase, Outcome 5: Cycle cancellation

4.7 Pregnancy loss and other outcomes

None of the studies for this comparison reported on pregnancy loss (miscarriage and stillbirth), premature LH surges, cost‐effectiveness, or acceptability of either of these protocols.

5. Long protocol (continue GnRHa versus stop GnRHa) ('long stop')

We included three RCTs in this comparison (Dirnfeld 1999; Garcia‐Velasco 2000; Simons 2005). See Table 4.

Primary outcomes

5.1 Live birth and ongoing pregnancy rates

5.1.1 Primary analysis

We analysed one study with low risk of bias in random sequence generation and outcome reporting (Simons 2005). We are uncertain of any difference in the number of live births and ongoing pregnancies when GnRHa was stopped compared to continuing GnRHa in a long protocol (OR 0.68, 95% CI 0.29 to 1.61; 1 study, 116 women; very low‐certainty evidence; Analysis 5.1).

5.1. Analysis.

Comparison 5: Long protocol: continuing GnRHa versus stopping GnRHa ('long stop'), Outcome 1: Ongoing pregnancy rate primary analysis

5.1.2 Sensitivity analysis

We undertook sensitivity analysis including two studies that reported ongoing pregnancy rates. Results were no different between fixed‐effect (OR 0.66, 95% CI 0.30 to 1.49; I² = 0%; 2 studies, 194 women; very low‐certainty evidence) (Analysis 5.2) and random‐effects models (OR 0.66, 95% CI 0.30 to 1.49; I² = 0%; 2 studies, 194 women; very low‐certainty evidence), and this finding persisted if we presented the results as risk ratio (RR 0.73, 95% CI 0.39 to 1.36; 2 studies, 194; very low‐certainty evidence). Analysis of Simons 2005, with low risk of bias in all domains, did not show any difference in the results (OR 0.68, 95% CI 0.29 to 1.61; 1 study, 116 women; very low‐certainty evidence).

5.2. Analysis.

Comparison 5: Long protocol: continuing GnRHa versus stopping GnRHa ('long stop'), Outcome 2: Ongoing pregnancy rate all studies

5.2 OHSS

None of the studies reported on this outcome.

Secondary outcomes

5.3 Clinical pregnancies

There may be no difference in the clinical pregnancy rate whether GnRHa is continued or stopped (OR 0.76, 95% CI 0.40 to 1.44; I² = 0%; 3 studies, 264 women; low‐certainty evidence; Analysis 5.3).

5.3. Analysis.

Comparison 5: Long protocol: continuing GnRHa versus stopping GnRHa ('long stop'), Outcome 3: Clinical pregnancies

5.4 Number of oocytes

There may be no difference in the number of oocytes retrieved between groups (MD −1.12, 95% CI −2.40 to 0.15; I² = 66%; 3 studies, 264 women; low‐certainty evidence; Analysis 5.4).

5.4. Analysis.

Comparison 5: Long protocol: continuing GnRHa versus stopping GnRHa ('long stop'), Outcome 4: Number of oocytes

5.5 Amount of gonadotropins administered

There may be no difference in the amount of gonadotropins required between groups (MD −0.30, 95% CI −3.61 to 3.01; I² = 76%; 3 studies, 264 women; low‐certainty evidence; Analysis 5.5).

5.5. Analysis.

Comparison 5: Long protocol: continuing GnRHa versus stopping GnRHa ('long stop'), Outcome 5: Amount of gonadotropins administered

5.6 Cycle cancellation rate

There may be little or no difference in cycle cancellation rate between groups (OR 1.41, 95% CI 0.56 to 3.56; I² = 69%; 3 studies, 264 women; low‐certainty evidence; Analysis 5.6).

5.6. Analysis.

Comparison 5: Long protocol: continuing GnRHa versus stopping GnRHa ('long stop'), Outcome 6: Cycle cancellation

5.7 Pregnancy loss and other outcomes

None of the studies for this comparison reported on pregnancy loss (miscarriage and stillbirth), premature LH surges, cost‐effectiveness, or acceptability of either of these protocols.

6. Long protocol (continued same‐dose GnRHa versus reduced‐dose GnRHa)

We included four RCTs in this comparison (Dal Prato 2001; Ding 2013; Fábregues 2005; Simon 1994). See Table 5.

Primary outcomes

6.1 Live birth and ongoing pregnancy rates

6.1.1 Primary analysis

One study reported this outcome (Ding 2013). We are uncertain if there is any difference in live birth rate when the dose of GnRHa is reduced versus when the same dose is continued in a long protocol (OR 1.59, 95% CI 0.66 to 3.87; 1 study, 96 women; very low‐certainty evidence; Analysis 6.1).

6.1. Analysis.

Comparison 6: Long protocol: continued same‐dose GnRHa versus reduced‐dose GnRHa, Outcome 1: Ongoing pregnancy rate

6.1.2 Sensitivity analysis

Results were similar when analysed using a random‐effects model (OR 1.59, 95% CI 0.66 to 3.87; 1 study, 96 women; very low‐certainty evidence) or presented as risk ratio (RR 1.14, 95% CI 0.88 to 1.48; 1 study, 96 women; very low‐certainty evidence).

6.2 OHSS

One study reported this outcome (Ding 2013). We are uncertain if there is any difference in OHSS incidence when the dose of GnRHa is reduced versus when the same dose is continued in a long protocol (OR 0.47, 95% CI 0.04 to 5.35, 1 study, 96 women; very low‐certainty evidence; Analysis 6.2).

6.2. Analysis.

Comparison 6: Long protocol: continued same‐dose GnRHa versus reduced‐dose GnRHa, Outcome 2: OHSS

Secondary outcomes

6.3 Clinical pregnancy rate

We are uncertain if there is any difference in clinical pregnancy rate with a reduced dose of GnRHa compared with continuing on the same dose (OR 1.02, 95% CI 0.68 to 1.52; I² = 0%; 4 studies, 407 women; very low‐certainty evidence; Analysis 6.3).

6.3. Analysis.

Comparison 6: Long protocol: continued same‐dose GnRHa versus reduced‐dose GnRHa, Outcome 3: Clinical pregnancies

6.4 Number of oocytes

We are uncertain if there is any difference in the number of oocytes retrieved with a reduced dose of GnRHa compared with continuing on the same dose (MD 1.03, 95% CI −0.04 to 2.10; I² = 0%; 3 studies, 275 women; very low‐certainty evidence; Analysis 6.4).

6.4. Analysis.

Comparison 6: Long protocol: continued same‐dose GnRHa versus reduced‐dose GnRHa, Outcome 4: Number of oocytes

6.5 Amount of gonadotropins administered

We are uncertain if there is any difference in the amount of gonadotropins required between groups (MD 0.98, 95% CI −1.72 to 3.69; I² = 58%; 2 studies, 228 women; very low‐certainty evidence; Analysis 6.5).

6.5. Analysis.

Comparison 6: Long protocol: continued same‐dose GnRHa versus reduced‐dose GnRHa, Outcome 5: Amount of gonadotropins administered

6.6 Cycle cancellation rate

We are uncertain if there is any difference in cycle cancellation rate between groups (OR 1.00, 95% CI 0.14 to 7.32; I² = not applicable; 2 studies, 228 women; very low‐certainty evidence; Analysis 6.6).

6.6. Analysis.

Comparison 6: Long protocol: continued same‐dose GnRHa versus reduced‐dose GnRHa, Outcome 6: Cycle cancellation

6.7 Pregnancy loss and other outcomes

None of the studies for this comparison reported on pregnancy loss (miscarriage and stillbirth), premature LH surges, cost‐effectiveness, or acceptability of either of these protocols.

7. Long protocol: discontinuing versus continuing GnRHa after HCG administration

We included only one RCT in this comparison (Isikoglu 2007).

Primary outcomes

7.1 Live birth and ongoing pregnancy rates

7.1.1 Primary analysis

We are uncertain if there is any difference in live birth and ongoing pregnancy rates between discontinuing and continuing GnRHa after HCG administration (OR 0.89, 95% CI 0.49 to 1.64; 1 study, 181 women; very low‐certainty evidence; Analysis 7.1).

7.1. Analysis.

Comparison 7: Long protocol: discontinuing versus continuing GnRHa after HCG administration, Outcome 1: Live birth and ongoing pregnancies

7.1.2 Sensitivity analysis

Results were similar when analysed using a random‐effects model (OR 1.89, 95% CI 0.49 to 1.64; 1 study, 181 women; very low‐certainty evidence) or presented as risk ratio (RR 0.93, 95% CI 0.63 to 1.37; 1 study, 181; very low‐certainty evidence).

7.2 OHSS

This outcome was not reported.

Secondary outcome measures

7.3 Clinical pregnancy rate

We are uncertain if there is any difference in clinical pregnancy rate between discontinuing and continuing GnRHa after HCG administration (OR 1.02, 95% CI 0.57 to 1.83; 1 study, 181 women; very low‐certainty evidence; Analysis 7.2).

7.2. Analysis.

Comparison 7: Long protocol: discontinuing versus continuing GnRHa after HCG administration, Outcome 2: Clinical pregnancies

7.4 Number of oocytes retrieved

We are uncertain if there is any difference between groups in the number of oocytes retrieved (MD −0.90, 95% CI −3.04 to 1.24; 1 study, 181 women; very low‐certainty evidence; Analysis 7.3).

7.3. Analysis.

Comparison 7: Long protocol: discontinuing versus continuing GnRHa after HCG administration, Outcome 3: Number of oocytes

7.5 Amount of gonadotropins administered

We are uncertain if there is any difference in the amount of gonadotropins required between groups (MD 2.80, 95% CI −0.55 to 6.15; 1 study, 181 women; very low‐certainty evidence; Analysis 7.4).

7.4. Analysis.

Comparison 7: Long protocol: discontinuing versus continuing GnRHa after HCG administration, Outcome 4: Amount of gonadotropins administered

7.6 Cycle cancellation

We are uncertain if there is any difference in cycle cancellation rate between groups (OR 1.50, 95% CI 0.24 to 9.20; 1 study, 181 women; very low‐certainty evidence; Analysis 7.5).

7.5. Analysis.

Comparison 7: Long protocol: discontinuing versus continuing GnRHa after HCG administration, Outcome 5: Cycle cancellation

7.7 Pregnancy loss and other outcomes

Isikoglu 2007 did not report on pregnancy loss (miscarriage and stillbirth), premature LH surges, cost‐effectiveness, or acceptability of either of these protocols.

8. Long protocol: administration of GnRHa for two versus three weeks before stimulation

We included one RCT in this comparison (Lin 2013). See Table 6.

Primary outcomes

8.1 Live birth and ongoing pregnancy rates

8.1.1 Primary analysis

We are uncertain if there is any difference in live birth and ongoing pregnancy rates between administration of GnRH for two and three weeks before stimulation (OR 0.88, 95% CI 0.37 to 2.05; 1 study, 85 women; very low‐certainty evidence; Analysis 8.1).

8.1. Analysis.

Comparison 8: Long protocol: GnRHa for 2 weeks versus 3 weeks before stimulation, Outcome 1: Live birth and ongoing pregnancies

8.1.2 Sensitivity analysis

Results were similar when analysed using a random‐effects model (OR 0.88, 95% CI 0.37 to 2.05; 1 study, 85 women; very low‐certainty evidence) or presented as risk ratio (RR 0.93, 95% CI 0.59 to 1.46; 1 study, 85 women; very low‐certainty evidence).

8.2 OHSS

We are uncertain if there is any difference in OHSS rate between groups (OR 0.93, 95% CI 0.06 to 15.37; 1 study, 85 women; very low‐certainty evidence; Analysis 8.5).

8.5. Analysis.

Comparison 8: Long protocol: GnRHa for 2 weeks versus 3 weeks before stimulation, Outcome 5: Adverse outcomes: OHSS and pregnancy loss

Secondary outcomes

8.3 Clinical pregnancy rate

We are uncertain if there is a difference in clinical pregnancy rate between administration of GnRH for two and three weeks before stimulation (OR 0.93, 95% CI 0.39 to 2.21; 1 study, 85 women; very low‐certainty evidence; Analysis 8.2).

8.2. Analysis.

Comparison 8: Long protocol: GnRHa for 2 weeks versus 3 weeks before stimulation, Outcome 2: Clinical pregnancies

8.4 Total number of oocytes retrieved

We are uncertain if there is a difference between groups in the number of oocytes retrieved (MD 0.12, 95% CI −1.90 to 2.14; 1 study, 85 women; very low‐certainty evidence; Analysis 8.3).

8.3. Analysis.

Comparison 8: Long protocol: GnRHa for 2 weeks versus 3 weeks before stimulation, Outcome 3: Number of oocytes

8.5 Amount of gonadotropins administered

We are uncertain if there is a difference in amount of gonadotropins required between administration of GnRH for two and three weeks before stimulation (MD 207.00, 95% CI −44.65 to 458.65; 1 study, 85 women; very low‐certainty evidence; Analysis 8.4).

8.4. Analysis.

Comparison 8: Long protocol: GnRHa for 2 weeks versus 3 weeks before stimulation, Outcome 4: Amount of gonadotropins administered

8.6 Pregnancy loss

We are uncertain if there is a difference in miscarriage rates between administration of GnRH for two and three weeks before stimulation (OR 0.93, 95% CI 0.18 to 4.87; 1 study, 85 women; very low‐certainty evidence; Analysis 8.5). Stillbirths were not reported.

8.7 Other outcomes

Lin 2013 did not report on premature LH surges, cost‐effectiveness, or acceptability of either of these protocols.

9. Short protocol: continuing GnRHa versus stopping GnRHa ('short stop')

We included one RCT in this comparison (Cedrin‐Durnerin 2000).

Primary outcomes

9.1 Live birth and ongoing pregnancy rates

This outcome was not reported.

9.2 OHSS

This outcome was not reported.

Secondary outcomes

9.3 Clinical pregnancy rate

We are uncertain if there is a difference in clinical pregnancy rate between a short protocol and a short stop protocol (OR 0.59, 95% CI 0.30 to 1.17; 1 study, 230 women; very low‐certainty evidence; Analysis 9.1).

9.1. Analysis.

Comparison 9: Short protocol: continuing GnRHa versus stopping GnRha ('short stop'), Outcome 1: Clinical pregnancies

9.4 Total number of oocytes retrieved

We are uncertain if there is a difference in the number of oocytes retrieved between groups (MD −0.90, 95% CI −1.03 to −0.77; 1 study, 200 women; very low‐certainty evidence; Analysis 9.2).

9.2. Analysis.

Comparison 9: Short protocol: continuing GnRHa versus stopping GnRha ('short stop'), Outcome 2: Number of oocytes

9.5 Amount of gonadotropins administered

We are uncertain if the short protocol required fewer ampoules of gonadotropins compared to the short stop protocol (MD −5.20, 95% CI −8.11 to −2.29; 1 study, 230 women; very low‐certainty evidence; Analysis 9.3).

9.3. Analysis.

Comparison 9: Short protocol: continuing GnRHa versus stopping GnRha ('short stop'), Outcome 3: Amount of gonadotropins administered

9.6 Cycle cancellation

We are uncertain if there is a difference in cycle cancellation rate between groups (OR 0.73, 95% CI 0.34 to 1.59; 1 study, 230 women; very low‐certainty evidence; Analysis 9.4).

9.4. Analysis.

Comparison 9: Short protocol: continuing GnRHa versus stopping GnRha ('short stop'), Outcome 4: Cycle cancellation

9.7 Pregnancy loss and other outcomes

Cedrin‐Durnerin 2000 did not report on pregnancy loss (miscarriage or stillbirth), premature LH surges, cost‐effectiveness, or acceptability of either of these protocols.

10. Short protocol: comparison of different doses of GnRHa

We included three RCTs in this comparison (Al‐Jeborry 2020; Ghaffari 2020; Sarhan 2016). See Table 7.

Primary outcomes

10.1 Live birth and ongoing pregnancy rates

10.1.1 Primary analysis

One study reported this outcome (Ghaffari 2020). We are uncertain if there is a difference in live birth rate between 500 µg and 80 µg doses (OR 0.31, 95% CI 0.10 to 0.98; 1 study, 200 women; very low‐certainty evidence; Analysis 10.1).

10.1. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 1: Live birth 500 µg versus 80 µg

10.1.2 Sensitivity analysis

Results were similar when analysed using a random‐effects model (OR 0.31, 95% CI 0.10 to 0.98; 1 study, 200 women; very low‐certainty evidence) or presented as risk ratio (RR 0.33, 95% CI 0.11 to 1.00; 1 study, 200 women; very low‐certainty evidence).

10.2 OHSS

This outcome was not reported.

Secondary outcomes

10.3 to 10.6 Clinical pregnancy rate

We are uncertain if there is a difference in clinical pregnancy rate for the following comparisons: 500 µg versus 80 µg (OR 0.50, 95% CI 0.19 to 1.32; 1 study, 200 women; very low‐certainty evidence; Analysis 10.2); 100 µg versus 50 µg (OR 1.00, 95% CI 0.37 to 2.70; 1 study, 68 women; very low‐certainty evidence; Analysis 10.3); or 50 µg versus 25 µg (OR 1.02, 95% CI 0.34 to 3.10; 1 study, 57 women; very low‐certainty evidence; Analysis 10.5). A dose of 100 µg may improve clinical pregnancy rate when compared to a dose of 25 µg (OR 2.30, 95% CI 1.06 to 5.00; 2 studies, 133 women; low‐certainty evidence; Analysis 10.4).

10.2. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 2: Clinical pregnancy 500 µg vs 80 µg

10.3. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 3: Clinical pregnancy 100 µg vs 50 µg

10.5. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 5: Clinical pregnancy 50 µg vs 25 µg

10.4. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 4: Clinical pregnancy 100 µg vs 25 µg

10.7 to 10.10 Total number of oocytes retrieved

We are uncertain if there is a difference in the number of oocytes retrieved for the following comparisons: 500 µg versus 80 µg (MD 0.50, 95% CI −0.42 to 1.42; 1 study, 131 women; very low‐certainty evidence; Analysis 10.6); 100 µg versus 50 µg (MD −1.30, 95% CI −1.63 to −0.97; 1 study, 68 women; very low‐certainty evidence; Analysis 10.7); and 50 µg versus 25 µg (MD 0.60, 95% CI 0.09 to 1.11; 1 study, 57 women; very low‐certainty evidence; Analysis 10.9). There may be little or no improvement with a dose of 100 µg compared to a dose of 25 µg (MD −0.66, 95% CI −1.11 to −0.20; 2 studies, 133 women; low‐certainty evidence; Analysis 10.8).

10.6. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 6: Number of oocytes 500 µg vs 80 µg

10.7. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 7: Number of oocytes 100 µg vs 50 µg

10.9. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 9: Number of oocytes 50 µg vs 25 µg

10.8. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 8: Number of oocytes 100 µg vs 25 µg

10.11 Amount of gonadotropins administered

Two comparisons are presented as subgroups 10.10.1 and 10.10.2. We are uncertain if there is a difference in amount of gonadotropins used between groups when comparing 500 µg versus 80 µg (MD −290.40, 95% CI −607.75 to 26.95; 1 study, 131 women; very low‐certainty evidence) or 100 µg versus 25 µg (MD −65.13, 95% CI −578.83 to 448.57; 1 study, 76 women; very low‐certainty evidence; Analysis 10.10).

10.10. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 10: Amount of gonadotropins administered

10.12 Cycle cancellation rates

Two comparisons are presented as subgroups 10.11.1 and 10.11.2. We are uncertain if there is a difference in cycle cancellation rates between groups when comparing 500 µg versus 80 µg (OR 0.84, 95% CI 0.37 to 1.91; 1 study, 200 women; very low‐certainty evidence) or 100 µg versus 25 µg (OR 1.00, 95% CI 0.19 to 5.34; 1 study, 68 women; very low‐certainty evidence; Analysis 10.11).

10.11. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 11: Cycle cancellation

10.13 Pregnancy loss and other outcomes

We are uncertain if there is a difference in miscarriage rate between 500 µg and 80 µg doses (OR 3.15, 95% CI 0.32, 31.05; 1 study, 131 women; very low‐certainty evidence; Analysis 10.12).

10.12. Analysis.

Comparison 10: Short protocol variations: comparison of different doses of GnRHa, Outcome 12: Other outcomes: pregnancy loss

None of the three studies for this comparison reported on premature LH surges, stillbirths, cost‐effectiveness, or acceptability of either of these protocols.

Discussion

Summary of main results

The aim of this review was to evaluate the effectiveness and safety of different GnRHa protocols as adjuncts to COH in women undergoing assisted reproduction. We included 40 trials involving a total of 4148 women in this review. The studies evaluated 10 different comparison groups. Nineteen trials compared a long versus short GnRHa protocol. We included trials with a low risk of selection and reporting bias in the primary analysis. We undertook a sensitivity analysis involving all studies for the primary outcomes of live birth and ongoing pregnancy rate, as well as for the secondary outcome of clinical pregnancy.

Based on the primary analysis, a long GnRHa protocol was not associated with an increase in live birth and ongoing pregnancy rates in comparison with a short GnRHa protocol. In contrast, we found low‐certainty evidence that the long protocol may improve clinical pregnancy rates overall, as well as in studies including poor responders only. The findings remained unchanged after performing sensitivity analyses: results were similar using a random‐effects model, when presented as risk ratio, and when studies with low risk of bias in all domains were analysed. A subgroup analysis including four trials on poor responders only showed a difference in clinical pregnancy rates, number of oocytes retrieved, and cancellation rates, favouring the long GnRHa protocol, although the amount of gonadotropins needed was higher.

Regarding the remaining comparisons between GnRHa protocols used for pituitary suppression in assisted reproduction technology, we are uncertain if there is any difference in terms of live birth and ongoing pregnancy rates due to the very low certainty of the evidence. Our findings remained unchanged after performing sensitivity analyses.

Incidence of OHSS was reported in two long GnRHa protocol comparisons (continued same‐dose GnRHa versus reduced‐dose GnRHa and administration of GnRHa for two versus three weeks before ovarian stimulation). We are uncertain if there is any difference between the groups studied. Similarly, miscarriage rates were reported in two studies only in the long GnRHa protocol (administration of GnRHa for two versus three weeks before ovarian stimulation) and the short GnRHa protocol (500 µg versus 80 µg). For both comparisons, we are uncertain if there is any difference between groups.

For the remaining comparisons between GnRHa protocols, we are uncertain if there is any difference between groups for our secondary outcomes due to the very low certainty of the evidence, except for the comparison of 100 µg versus 25 µg in the short protocol, which showed that clinical pregnancy may be improved by the former dose, based on low‐certainty evidence.

Of note, there was very poor reporting of any adverse events among studies in all comparisons. Cycle cancellation rates were reported in all except one comparison group. We found that there may be fewer cancellations in poor responders receiving the long rather than the short GnRHa protocol. We are uncertain of any difference in the remaining comparison groups. None of the studies included in this review reported on premature LH surge, cost‐effectiveness, or acceptability of alternative protocols.

Overall completeness and applicability of evidence

The included trials were relevant to the review question and were analysed in 10 different comparison groups. Of these, three compared the standard GnRHa protocols (long GnRHa, short GnRHa, and ultrashort GnRHa protocol); five compared variations of the long protocol; and two compared variations of the short protocol. The first three comparisons of the standard protocols are likely to be more relevant to current practice. The long versus short protocol group had the highest number of studies (19). Only two studies evaluated the long versus the ultrashort protocol, and one study evaluated the short versus the ultrashort protocol. Similarly, the number of studies comparing various ways of administering GnRHa in the long or the short protocol were small. Of note, the studies included in the primary analysis were even fewer in number, as we decided to include only those with adequate randomisation and complete outcome data reporting.

There was evidence of clinical heterogeneity, with wide variation in participant age, dose regimens, and the preparation of GnRH used, with some studies including all types of infertility, and some limited to specific types or subgroups of poor responders only.

The primary outcome of live birth and ongoing pregnancy rates was reported by half of the included studies, whereas clinical pregnancy rate was reported by the majority of studies. Adverse outcomes were poorly reported; only two studies reported the primary adverse outcome of OHSS, while two studies reported miscarriage rates. More than half of the included studies evaluated the most commonly reported adverse event, cycle cancellation rate. None of the five studies comparing the time of starting a long GnRHa protocol (i.e. follicular or luteal phase start) commented on either the risk of cyst formation, which can be associated with the follicular phase start (Jenkins 1996), or the risk of inadvertently exposing a pregnancy to GnRHa during a luteal phase start (Ron‐El 1990). Furthermore, none of the studies reported on the acceptability of the regimens or cost‐effectiveness.

In the comparison between long and short GnRHa protocols, 12 of 19 studies reported on the primary outcome of live birth and ongoing pregnancy rate. Of these, we included five studies with adequate randomisation and primary outcome data (i.e. live birth and ongoing pregnancy rates; I² = 0%). Our results suggest that the primary outcome was not different between the two groups. However, in the same comparison group, an analysis of eight studies with low risk of selection bias and attrition bias (I² = 23%) suggests that the long protocol may result in a higher clinical pregnancy rate than the short protocol. A subgroup analysis including four studies on poor responders showed improved pregnancy rates following the long protocol, but this was not apparent in an unselected group of participants. Therefore, the long protocol could be more effective in poor responders, although this finding was based on low‐certainty evidence.

Quality of the evidence

The certainty of the evidence ranged from low to very low. We downgraded the certainty of the evidence due to failure to report live birth or ongoing pregnancy rates, poor reporting of methods in the primary studies, imprecise findings due to lack of data, and insufficient data regarding adverse events. Only eight of the 40 included studies were conducted within the last 10 years. For further details, see Table 1; Table 2; Table 3; Table 4; Table 5; Table 6; Table 7.

The included trials were analysed in 10 different comparison groups. Of these, three compared the standard GnRHa protocols (long GnRHa, short GnRHa, and ultrashort GnRHa protocol); five compared variations of the long protocol; and two compared variations of the short protocol. The long versus short protocol group had the highest number of studies (19). There were only two studies in the long versus the ultrashort protocol, and one study in the short versus the ultrashort group. Similarly, the number of studies comparing various ways of administering GnRHa in the long or the short protocol were small. Of note, the studies included in the primary analysis were even fewer in number, as we decided to include only those with adequate randomisation and complete outcome data reporting.

There was evidence of clinical heterogeneity, with wide variation in the participants' age, dose regimens, and the preparation of GnRH used, with some studies including all types of infertility and some limited to specific types or subgroups of poor responders only.

The primary outcome of live birth and ongoing pregnancy rates was reported by half of the included studies, whereas clinical pregnancy rate was reported by the majority of them. Adverse outcomes were poorly reported; only two studies reported the primary adverse outcome of OHSS, while two reported miscarriage rates. The most commonly reported adverse event, cycle cancellation rate, was included in over half of the studies. None of the five studies comparing the time of starting a long GnRHa protocol (i.e. follicular or luteal phase start) commented on either the risk of cyst formation, which can be associated with the follicular phase start (Jenkins 1996), or on the risk of inadvertently exposing a pregnancy to GnRHa during a luteal phase start (Ron‐El 1990). Also, none of the studies reported on the acceptability of the regimens or cost‐effectiveness.

Potential biases in the review process

We assessed risk of bias in the included studies using the Cochrane RoB 1 tool (Higgins 2011).

Nineteen studies featured in the comparison of long versus short GnRHa protocols. Of these, only nine studies reported random sequence generation, and two reported allocation concealment. Blinding was not undertaken in the majority of the studies. The funnel plot did not suggest any publication bias.

The overall assessment of all 40 trials showed that 22 of them were at low risk of bias for randomisation, while the remaining 18 were classified as having unclear risk of bias. Allocation concealment was unclear in 27 studies, and blinding was not undertaken in a similar number of studies. Therefore, selection bias was apparent. Regarding attrition and reporting bias, we assessed 28 and 20 studies, respectively, as having a low risk of bias, with the rest judged as having an unclear or high risk of bias. In 25 studies, there was insufficient information to assess whether another potential source of bias existed.

Finally, while almost all the included studies reported on clinical pregnancies, very few (including the most recent ones) reported on live birth rates, ongoing pregnancy rates, or both, and OHSS rates. Live birth is considered to be the most reliable outcome for the effectiveness of an intervention in assisted reproduction technology (Duffy 2021), and the lack of such reporting undermines the robustness of the results found in this update.

Agreements and disagreements with other studies or reviews

The results of the current update are similar to those of the previous versions of this review (Daya 2000; Maheshwari 2011; Siristatidis 2015). This updated review includes three new studies and one further comparison not part of the previous update, namely the comparison of different doses of GnRHa in a short protocol.

Of note, there are no non‐Cochrane reviews on this topic.

Authors' conclusions

Implications for practice.

We found no difference in live birth and ongoing pregnancy rates between the long gonadotropin‐releasing hormone agonist (GnRHa) protocol and the short GnRHa protocol. However, there was low‐certainty evidence of higher clinical pregnancy rates in the long protocol group, mainly in the poor responder subgroup. Specifically, the comparison of these two protocols in poor responders showed a higher number of oocytes retrieved, fewer cycle cancellations, and higher gonadotropin use with the long protocol; however, the certainty of the evidence was low.

Regarding the remaining comparisons between GnRHa protocols, we are uncertain if there is any difference between groups in terms of live birth or ongoing pregnancy rates and our secondary outcomes, findings that are based on very low‐certainty evidence. Adverse events were generally poorly reported, with only two studies each reporting on ovarian hyperstimulation syndrome (OHSS) rates and miscarriage rates. There was insufficient evidence to draw any conclusions regarding the cost‐effectiveness or acceptability of regimens.

Consequently, the long protocol has an edge over the short protocol in terms of clinical pregnancy rate, specifically in poor responders, where there may be a benefit in terms of a higher number of oocytes retrieved and fewer cycle cancellations, although the certainty of evidence is low.

Implications for research.

As GnRHa are widely used in in vitro fertilisation (IVF) protocols, further research with high‐quality trials is needed to determine the optimal protocol for women in terms of dose, preparation, and duration of gonadotropin used. Such trials should include acceptability of the different protocols and cost‐effectiveness as outcomes.

We propose comparisons of GnRHa protocols, stratified by women's age and prognosis, specifically looking at the long protocol and variations of the long protocol, as this is the most preferred GnRHa regimen worldwide (IVF Worldwide 2022). Trials based on specific subpopulations, such as poor responders, defined as per Patient‐Oriented Strategies Encompassing IndividualiseD Oocyte Number (POSEIDON) criteria (Alviggi 2016), or Bologna criteria (Ferraretti 2011), will enable pooling of trial results by reducing clinical heterogeneity.

Most importantly, for all comparisons, live birth rates or ongoing pregnancy rates, or both, should be the primary outcome reported, along with adverse events, as GnRHa protocols have been associated with a high incidence of OHSS and miscarriage.

Finally, future research should consider the increasing popularity of GnRH antagonist protocols, which may diminish the need and funding for trials focused exclusively on GnRH agonists.

What's new

Date	Event	Description
9 January 2025	New search has been performed	A total of 3 studies were added in this update, and 1 co‐author was added. The text was changed comprehensively according to current Gynaecology and Fertility Group guidelines. The 3 new studies formed a new comparison group (number 10) for presenting data comparing different doses of GnRH agonist in a short protocol. Our conclusions regarding the long versus short protocol did not change as no new studies were added.
9 January 2025	New citation required but conclusions have not changed	The addition of 3 new studies did not change our conclusions regarding the primary outcome.

History

Protocol first published: Issue 1, 2008
Review first published: Issue 8, 2011

Date	Event	Description
11 January 2009	Amended	The original review was withdrawn and a new protocol published. The review title was changed back from 'Long versus short gonadotropin releasing hormone agonist protocols for pituitary desensitization in assisted reproduction cycles' to 'Gonadotrophin‐releasing hormone agonist protocols for pituitary suppression in assisted reproductive treatment'. 11 December 2008: Title changed from 'Gonadotrophin‐releasing hormone agonist protocols for pituitary down regulation in assisted reproductive treatment' to 'Long versus short gonadotropin releasing hormone agonist protocols for pituitary desensitization in assisted reproduction cycles'.
12 November 2007	New citation required and major changes	Substantive amendment

Appendices

Appendix 1. Cochrane Gynaecology and Fertility Specialised Register search strategy

ProCite platform

Searched 19 December 2022

Keywords CONTAINS "IVF" or "ICSI" or "in‐vitro fertilisation " or "in‐vitro fertilisation procedure" or "in vitro fertilization" or "intracytoplasmic sperm injection" or "intracytoplasmic morphologically selected sperm injection" or "controlled ovarian hyperstimulation" or "controlled ovarian stimulation" or "COH" or "embryo transfer" or "ovarian hyperstimulation" or "ovarian stimulation" or Title CONTAINS "IVF" or "ICSI" or "in‐vitro fertilisation " or "in‐vitro fertilisation procedure" or "in vitro fertilization" or "intracytoplasmic sperm injection" or "intracytoplasmic morphologically selected sperm injection" or "controlled ovarian hyperstimulation" or "controlled ovarian stimulation" or "COH" or "embryo transfer" or "ovarian hyperstimulation" or "ovarian stimulation"

AND

Keywords CONTAINS "Gonadorelin" or "gonadotropin releasing agonist" or "gonadotropin releasing hormone agonist" or "Goserelin" or "goserelin acetate" or "goserelin pretreatment" or "Gosereline "or "buserelin" or "busereline" or "leuprolide "or"leuprolin"or"leuprorelin"or"nafarelin"or"triptorelin"or"Lupron"or "Zoladex" or "deslorelin" or "GnRH agonist" or "GnRH a" or "GnRH agonists" or "GnRHa" or "GnRH analog" or "GnRH analogue" or "GnRH analogues" or "Luteinising hormone releasing hormone" or "luteinizing hormone supplementation" or "Lutenising hormone releasing hormone" or "menotropin" or "menotrophin" or "human menopausal gonadotropin" or "human menopausal gonadotropins" or "human menopausal gonadotropins" or Title CONTAINS "Gonadorelin" or "gonadotropin releasing agonist" or "gonadotropin releasing hormones" or "gonadotropin releasing hormone agonist"

(1341 records)

Appendix 2. CENTRAL search strategy

OVID platform

Searched 9 December 2022

1 exp embryo transfer/ or exp fertilization in vitro/ or exp sperm injections, intracytoplasmic/ or exp gamete intrafallopian transfer/ or exp in vitro oocyte maturation techniques/ (2514)
2 embryo transfer.tw. (3792)
3 in vitro fertili?ation.tw. (3553)
4 intracytoplasmic sperm injection$.tw. (1332)
5 (ivf or icsi).tw. (7829)
6 exp Infertility, Female/ (1558)
7 exp Primary Ovarian Insufficiency/ (164)
8 exp Infertility/ (3651)
9 (ovar$ adj2 stimulat$).tw. (2788)
10 (ovar$ adj2 hyperstimulat$).tw. (1613)
11 COH.tw. (458)
12 or/1‐11 (13224)
13 exp gonadotropin‐releasing hormone/ or exp buserelin/ or exp goserelin/ or exp leuprolide/ or exp nafarelin/ or exp triptorelin pamoate/ (2735)
14 gonadotropin‐releasing hormone$.tw. (1701)
15 (buserelin or goserelin or leuprolide or nafarelin or triptorelin).tw. (2430)
16 (Lupron or Suprefact or Suprecor).tw. (116)
17 (histrelin or Supprelin).tw. (10)
18 (Zoladex or deslorelin).tw. (347)
19 (Suprelorin or Ovuplant).tw. (2)
20 Synarel.tw. (10)
21 GnRHa.tw. (560)
22 GnRH‐a.tw. (429)
23 GnRH agonist$.tw. (1919)
24 GnRH analog$.tw. (543)
25 luteinizing hormone releasing agonist$.tw. (1)
26 exp Menotropins/ (451)
27 human menopausal gonadotropin$.tw. (389)
28 or/13‐27 (6534)
29 desensiti?ation.tw. (2041)
30 (long adj2 protocol).tw. (659)
31 (short adj2 protocol).tw. (249)
32 (ultra short adj2 protocol).tw. (3)
33 (long adj2 follicular).tw. (14)
34 (ultrashort adj2 protocol).tw. (3)
35 reduced dos$.tw. (1681)
36 down regulat$.tw. (2270)
37 downregulat$.tw. (2003)
38 (follicular adj5 luteal).tw. (434)
39 high dose$.tw. (28338)
40 stop versus non stop.tw. (1)
41 prolonged protocol.tw. (4)
42 7 day.tw. (8853)
43 continu$ versus stop$.tw. (17)
44 short acting.tw. (3076)
45 early cessation.tw. (106)
46 early follicular.tw. (421)
47 different phase$.tw. (455)
48 daily.tw. (206061)
49 long acting.tw. (8928)
50 long luteal.tw. (35)
51 desensiti?e.tw. (84)
52 suppression.tw. (15775)
53 suppress.tw. (4076)
54 (inhibition or inhibit).tw. (36072)
55 (long adj2 protocol$).tw. (701)
56 (short adj2 protocol$).tw. (292)
57 or/29‐56 (292420)
58 12 and 28 and 57 (1385)

Appendix 3. MEDLINE search strategy

OVID platform

Searched from 1946 to 19 December 2022

1 exp embryo transfer/ or exp fertilization in vitro/ or exp sperm injections, intracytoplasmic/ or exp gamete intrafallopian transfer/ or exp in vitro oocyte maturation techniques/ (48389)
2 embryo transfer.tw. (13777)
3 in vitro fertili?ation.tw. (26620)
4 intracytoplasmic sperm injection$.tw. (8498)
5 (ivf or icsi).tw. (31836)
6 exp Infertility, Female/ (30666)
7 exp Primary Ovarian Insufficiency/ (3320)
8 exp Infertility/ (72542)
9 (ovar$ adj2 stimulat$).tw. (8451)
10 (ovar$ adj2 hyperstimulat$).tw. (5669)
11 COH.tw. (2075)
12 or/1‐11 (130815)
13 exp gonadotropin‐releasing hormone/ or exp buserelin/ or exp goserelin/ or exp leuprolide/ or exp nafarelin/ or exp triptorelin pamoate/ (34028)
14 gonadotropin‐releasing hormone$.tw. (15641)
15 (buserelin or goserelin or leuprolide or nafarelin or triptorelin).tw. (5226)
16 (Lupron or Suprefact or Suprecor).tw. (209)
17 (histrelin or Supprelin).tw. (79)
18 (Zoladex or deslorelin).tw. (707)
19 (Suprelorin or Ovuplant).tw. (54)
20 Synarel.tw. (13)
21 GnRHa.tw. (1830)
22 GnRH‐a.tw. (1198)
23 GnRH agonist$.tw. (4871)
24 GnRH analog$.tw. (2623)
25 luteinizing hormone releasing agonist$.tw. (5)
26 exp Menotropins/ (3185)
27 human menopausal gonadotropin$.tw. (1573)
28 or/13‐27 (44859)
29 desensiti?ation.tw. (24900)
30 (long adj2 protocol).tw. (1349)
31 (short adj2 protocol).tw. (775)
32 (ultra short adj2 protocol).tw. (11)
33 (long adj2 follicular).tw. (76)
34 (ultrashort adj2 protocol).tw. (16)
35 reduced dos$.tw. (5562)
36 down regulat$.tw. (160620)
37 downregulat$.tw. (191440)
38 (follicular adj5 luteal).tw. (3735)
39 high dose$.tw. (149873)
40 stop versus non stop.tw. (1)
41 prolonged protocol.tw. (20)
42 7 day.tw. (25300)
43 continu$ versus stop$.tw. (18)
44 short acting.tw. (8541)
45 early cessation.tw. (562)
46 early follicular.tw. (2192)
47 different phase$.tw. (13800)
48 daily.tw. (606652)
49 long acting.tw. (28051)
50 long luteal.tw. (67)
51 desensiti?e.tw. (1762)
52 suppression.tw. (270177)
53 suppress.tw. (116378)
54 (inhibition or inhibit).tw. (1209537)
55 (long adj2 protocol$).tw. (1723)
56 (short adj2 protocol$).tw. (1010)
57 or/29‐56 (2523824)
58 12 and 28 and 57 (2936)
59 randomized controlled trial.pt. (582930)
60 controlled clinical trial.pt. (95137)
61 randomized.ab. (585732)
62 randomised.ab. (115874)
63 placebo.tw. (240336)
64 clinical trials as topic.sh. (200654)
65 randomly.ab. (397608)
66 trial.ti. (275642)
67 (crossover or cross‐over or cross over).tw. (96054)
68 or/59‐67 (1569425)
69 exp animals/ not humans.sh. (5074784)
70 68 not 69 (1444815)
71 58 and 70 (867)

Appendix 4. Embase search strategy

OVID platform

Searched from 1980 to 19 December 2022

1 exp embryo transfer/ or exp infertility therapy/ (130126)
2 exp female infertility/ or exp fertilization in vitro/ or exp intracytoplasmic sperm injection/ (124344)
3 embryo transfer.tw. (23749)
4 in vitro fertili?ation.tw. (36181)
5 intracytoplasmic sperm injection$.tw. (11756)
6 (ivf or icsi).tw. (58622)
7 exp premature ovarian failure/ (6012)
8 (ovar$ adj2 stimulat$).tw. (14082)
9 (ovar$ adj2 hyperstimulat$).tw. (8599)
10 COH.tw. (2966)
11 or/1‐10 (188419)
12 exp gonadorelin/ or exp gonadorelin agonist/ (50330)
13 exp buserelin acetate/ or exp buserelin/ (5349)
14 exp goserelin/ (7816)
15 exp leuprorelin/ (12773)
16 exp nafarelin acetate/ or exp nafarelin/ (1485)
17 exp triptorelin/ (6284)
18 gonadotrop?in‐releasing hormone$.tw. (21201)
19 (buserelin or goserelin or leuprolide or nafarelin or triptorelin).tw. (7792)
20 (Lupron or Suprefact or Suprecor).tw. (3020)
21 (histrelin or Supprelin).tw. (194)
22 (Zoladex or deslorelin).tw. (2571)
23 (Suprelorin or Ovuplant).tw. (104)
24 Synarel.tw. (354)
25 GnRHa.tw. (2769)
26 GnRH‐a.tw. (1543)
27 GnRH agonist$.tw. (7669)
28 GnRH analog$.tw. (3937)
29 luteinizing hormone releasing agonist$.tw. (8)
30 exp human menopausal gonadotropin/ (9752)
31 human menopausal gonadotrop?in$.tw. (2404)
32 or/12‐31 (80514)
33 11 and 32 (21353)
34 desensiti?ation.tw. (30396)
35 (long adj2 protocol).tw. (2194)
36 (short adj2 protocol).tw. (1259)
37 (ultra short adj2 protocol).tw. (18)
38 (long adj2 follicular).tw. (102)
39 (ultrashort adj2 protocol).tw. (22)
40 reduced dos$.tw. (9330)
41 down regulat$.tw. (213267)
42 downregulat$.tw. (245131)
43 (follicular adj5 luteal).tw. (4361)
44 high dose$.tw. (217454)
45 stop versus non stop.tw. (1)
46 prolonged protocol.tw. (27)
47 7 day.tw. (35308)
48 continu$ versus stop$.tw. (28)
49 short acting.tw. (12407)
50 early cessation.tw. (806)
51 early follicular.tw. (2782)
52 different phase$.tw. (16147)
53 daily.tw. (869318)
54 long acting.tw. (40894)
55 long luteal.tw. (109)
56 desensiti?e.tw. (2201)
57 suppression.tw. (326745)
58 suppress.tw. (140845)
59 (inhibition or inhibit).tw. (1435962)
60 (long adj2 protocol$).tw. (2730)
61 (short adj2 protocol$).tw. (1594)
62 or/34‐61 (3200465)
63 33 and 62 (5653)
64 Clinical Trial/ (1039669)
65 Randomized Controlled Trial/ (736575)
66 exp randomization/ (95813)
67 Single Blind Procedure/ (48500)
68 Double Blind Procedure/ (198511)
69 Crossover Procedure/ (72218)
70 Placebo/ (375048)
71 Randomi?ed controlled trial$.tw. (302029)
72 Rct.tw. (49762)
73 random allocation.tw. (2405)
74 randomly allocated.tw. (42909)
75 allocated randomly.tw. (2831)
76 (allocated adj2 random).tw. (854)
77 Single blind$.tw. (29718)
78 Double blind$.tw. (229040)
79 ((treble or triple) adj blind$).tw. (1689)
80 placebo$.tw. (346574)
81 prospective study/ (814176)
82 or/64‐81 (2612946)
83 case study/ (90165)
84 case report.tw. (497427)
85 abstract report/ or letter/ (1214388)
86 or/83‐85 (1788047)
87 82 not 86 (2550726)
88 63 and 87 (1857)

Appendix 5. CINAHL search strategy

EBSCO platform

Searched from 1961 to 8 January 2020 (CINAHL records from the 9 Decemeber 2020 search were included in the CENTRAL output)

Results
S33	S18 AND S32	191
S32	S19 OR S20 or S21 or S22 OR S23 OR S24 OR S25 OR S26 OR S27 OR S28 OR S29 OR S30 OR S31	1,372,404
S31	TX allocat* random*	11,308
S30	(MH "Quantitative Studies")	24,115
S29	(MH "Placebos")	11,559
S28	TX placebo*	60,526
S27	TX random* allocat*	11,308
S26	(MH "Random Assignment")	56,787
S25	TX randomi* control* trial*	180,912
S24	TX ( (singl* n1 blind*) or (singl* n1 mask*) ) or TX ( (doubl* n1 blind*) or (doubl* n1 mask*) ) or TX ( (tripl* n1 blind*) or (tripl* n1 mask*) ) or TX ( (trebl* n1 blind*) or (trebl* n1 mask*) )	1,044,734
S23	TX ( (trebl* n1 blind*) or (trebl* n1 mask*) )	249
S22	TX ( (trebl* n1 blind*) or (trebl* n1 mask*) )	249
S21	TX clinic* n1 trial*	256,178
S20	PT Clinical trial	86,321
S19	(MH "Clinical Trials+")	271,788
S18	S8 AND S17	600
S17	S9 OR S10 OR S11 OR S12 OR S13 OR S14 OR S15 OR S16	2,821
S16	TX (gonadotropin releasing hormone agonist*)	473
S15	TX (Luteinising hormone releasing hormone)	47
S14	TX GnRH a	1,031
S13	TX buserelin or TX leuprolin or TX leuprorelin or TX nafarelin or TX triptorelin or TX Lupron or TX Zoladex or TX deslorelin	227
S12	TX (GnRH agonist*)	521
S11	TX Gonadorelin OR TX Leuprolide	1,900
S10	TX Goserelin	331
S9	(MM "Gonadorelin") OR (MM "Leuprolide") OR (MM "Goserelin")	1,079
S8	S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7	10,591
S7	TX embryo* N3 transfer*	3,097
S6	TX ovar* N3 hyperstimulat*	837
S5	TX ovari* N3 stimulat*	1,003
S4	TX IVF or TX ICSI	5,015
S3	(MM "Fertilization in Vitro")	3,435
S2	TX vitro fertilization	6,994
S1	TX vitro fertilisation	6,994

Appendix 6. Information from the studies selected for the review

Trial characteristics

(1) Method and timing of randomisation:

• randomisation was adequate (e.g. by computer, random number tables, or drawing lots); or

• not clear (e.g. stated but not further described, or did not fall into one of the randomisation categories).

(2) Allocation concealment.

(3) Duration, timing, and location of the trial (single‐centre or multicentre trial), duration of follow‐up, and:

• outcome data used for primary analysis were complete (follow‐up to live birth), all randomised women were accounted for with an intention‐to‐treat analysis;

• completeness of data uncertain; or

• outcome data incomplete, with 5% of the cycles commenced missing some outcome data.

(4) Co‐intervention:

• other care provided with the intervention under study was equivalent in the treatment and control groups;

• issue of co‐intervention was not considered; or

• co‐intervention variations definitely existed.

(5) The presence of a power calculation:

(a) yes (prospective and valid or not valid); or
(b) no.

Baseline characteristics of the studied groups

(a) Cause and duration of pre‐existing subfertility
(b) Age of women and parity
(c) Investigative work‐up prior to in vitro fertilisation (IVF)
(d) Previously administered treatment(s)

Intervention

(a) Type of intervention and control comparator
(b) Dose and type of regimen
(c) We differentiated between whether the studied population included all women undergoing assisted reproduction technology (ART) or was limited to women who had responded poorly in a previous attempt or were expected to have a diminished response. As different drug regimens of ovarian stimulation can lead to a variable ovarian response, data on the drugs employed were also extracted.

Outcomes

(a) Outcomes reported
(b) How outcomes were defined
(c) Timing of outcome measurement

Data and analyses

Comparison 1. Long versus short protocol.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Live birth/ongoing pregnancies primary analysis	5	381	Odds Ratio (M‐H, Fixed, 95% CI)	1.45 [0.83, 2.52]
1.2 Live birth/ongoing pregnancies all studies	12	976	Odds Ratio (M‐H, Fixed, 95% CI)	1.30 [0.94, 1.81]
1.3 Clinical pregnancies primary analysis	8	552	Odds Ratio (M‐H, Fixed, 95% CI)	1.56 [1.01, 2.40]
1.3.1 Non‐selected group	4	320	Odds Ratio (M‐H, Fixed, 95% CI)	1.13 [0.67, 1.91]
1.3.2 Poor responder group	4	232	Odds Ratio (M‐H, Fixed, 95% CI)	3.12 [1.39, 7.02]
1.4 Clinical pregnancies all studies	19	1582	Odds Ratio (M‐H, Fixed, 95% CI)	1.50 [1.18, 1.92]
1.5 Number of oocytes	10		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.5.1 Non‐selected group	6	512	Mean Difference (IV, Fixed, 95% CI)	2.47 [2.21, 2.72]
1.5.2 Poor responders	4	227	Mean Difference (IV, Fixed, 95% CI)	1.40 [0.75, 2.06]
1.6 Amount of gonadotropins administered	8		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.6.1 Non‐selected group	4	439	Mean Difference (IV, Fixed, 95% CI)	15.64 [14.05, 17.22]
1.6.2 Poor responders	4	227	Mean Difference (IV, Fixed, 95% CI)	7.07 [3.06, 11.08]
1.7 Cycle cancellation	11	1253	Odds Ratio (M‐H, Fixed, 95% CI)	0.73 [0.48, 1.10]
1.7.1 Non selected	11	1026	Odds Ratio (M‐H, Fixed, 95% CI)	0.95 [0.59, 1.55]
1.7.2 Poor responders	4	227	Odds Ratio (M‐H, Fixed, 95% CI)	0.31 [0.12, 0.76]

Comparison 2. Long protocol versus ultrashort protocol.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Live birth and ongoing pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.2 Clinical pregnancies	2	230	Odds Ratio (M‐H, Fixed, 95% CI)	1.56 [0.80, 3.06]
2.3 Number of oocytes	2	230	Mean Difference (IV, Fixed, 95% CI)	0.53 [‐0.61, 1.66]
2.4 Amount of gonadotropins administered	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
2.5 Cycle cancellation	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only

Comparison 3. Short versus ultrashort protocol.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Clinical pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.2 Number of oocytes	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.3 Amount of gonadotropins administered	1	82	Mean Difference (IV, Fixed, 95% CI)	‐13.85 [‐21.49, ‐6.21]
3.4 Cycle cancellation	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 4. Long protocol: luteal versus follicular phase.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
4.1 Live birth and ongoing pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4.2 Clinical pregnancies	5	750	Odds Ratio (M‐H, Fixed, 95% CI)	1.06 [0.76, 1.47]
4.3 Number of oocytes	4	527	Mean Difference (IV, Fixed, 95% CI)	‐1.29 [‐1.86, ‐0.71]
4.4 Amount of gonadotropins administered	4	527	Mean Difference (IV, Fixed, 95% CI)	1.12 [‐0.73, 2.97]
4.5 Cycle cancellation	2	267	Odds Ratio (M‐H, Fixed, 95% CI)	1.45 [0.35, 6.01]

Comparison 5. Long protocol: continuing GnRHa versus stopping GnRHa ('long stop').

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
5.1 Ongoing pregnancy rate primary analysis	1	116	Odds Ratio (M‐H, Fixed, 95% CI)	0.68 [0.29, 1.61]
5.2 Ongoing pregnancy rate all studies	2	194	Odds Ratio (M‐H, Fixed, 95% CI)	0.66 [0.30, 1.49]
5.3 Clinical pregnancies	3	264	Odds Ratio (M‐H, Fixed, 95% CI)	0.76 [0.40, 1.44]
5.4 Number of oocytes	3	264	Mean Difference (IV, Fixed, 95% CI)	‐1.12 [‐2.40, 0.15]
5.5 Amount of gonadotropins administered	3	264	Mean Difference (IV, Fixed, 95% CI)	‐0.30 [‐3.61, 3.01]
5.6 Cycle cancellation	3	264	Odds Ratio (M‐H, Fixed, 95% CI)	1.41 [0.56, 3.56]

Comparison 6. Long protocol: continued same‐dose GnRHa versus reduced‐dose GnRHa.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
6.1 Ongoing pregnancy rate	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
6.2 OHSS	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
6.3 Clinical pregnancies	4	407	Odds Ratio (M‐H, Fixed, 95% CI)	1.02 [0.68, 1.52]
6.4 Number of oocytes	3	275	Mean Difference (IV, Fixed, 95% CI)	1.03 [‐0.04, 2.10]
6.5 Amount of gonadotropins administered	2	228	Mean Difference (IV, Fixed, 95% CI)	0.98 [‐1.72, 3.69]
6.6 Cycle cancellation	2	228	Odds Ratio (M‐H, Fixed, 95% CI)	1.00 [0.14, 7.32]

Comparison 7. Long protocol: discontinuing versus continuing GnRHa after HCG administration.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
7.1 Live birth and ongoing pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
7.2 Clinical pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
7.3 Number of oocytes	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
7.4 Amount of gonadotropins administered	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
7.5 Cycle cancellation	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only

Comparison 8. Long protocol: GnRHa for 2 weeks versus 3 weeks before stimulation.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
8.1 Live birth and ongoing pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
8.2 Clinical pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
8.3 Number of oocytes	1	85	Mean Difference (IV, Fixed, 95% CI)	0.12 [‐1.90, 2.14]
8.4 Amount of gonadotropins administered	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8.5 Adverse outcomes: OHSS and pregnancy loss	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
8.5.1 Miscarriages	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected
8.5.2 OHSS	1		Odds Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 9. Short protocol: continuing GnRHa versus stopping GnRha ('short stop').

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
9.1 Clinical pregnancies	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
9.2 Number of oocytes	1	200	Mean Difference (IV, Fixed, 95% CI)	‐0.90 [‐1.03, ‐0.77]
9.3 Amount of gonadotropins administered	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
9.4 Cycle cancellation	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only

Comparison 10. Short protocol variations: comparison of different doses of GnRHa.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
10.1 Live birth 500 µg versus 80 µg	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.2 Clinical pregnancy 500 µg vs 80 µg	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.3 Clinical pregnancy 100 µg vs 50 µg	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.4 Clinical pregnancy 100 µg vs 25 µg	2	133	Odds Ratio (M‐H, Fixed, 95% CI)	2.30 [1.06, 5.00]
10.5 Clinical pregnancy 50 µg vs 25 µg	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.6 Number of oocytes 500 µg vs 80 µg	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
10.7 Number of oocytes 100 µg vs 50 µg	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
10.8 Number of oocytes 100 µg vs 25 µg	2	133	Mean Difference (IV, Fixed, 95% CI)	‐0.66 [‐1.11, ‐0.20]
10.9 Number of oocytes 50 µg vs 25 µg	1	57	Mean Difference (IV, Fixed, 95% CI)	0.60 [0.09, 1.11]
10.10 Amount of gonadotropins administered	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
10.10.1 500 micrograms buserelin versus 80 micrograms buserelin	1	131	Mean Difference (IV, Fixed, 95% CI)	‐290.40 [‐607.75, 26.95]
10.10.2 100 micrograms triptorelin versus 25 micrograms triptorelin	1	76	Mean Difference (IV, Fixed, 95% CI)	‐65.13 [‐578.83, 448.57]
10.11 Cycle cancellation	2		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.11.1 500 micrograms versus 80micrograms of buserelin	1	200	Odds Ratio (M‐H, Fixed, 95% CI)	0.84 [0.37, 1.91]
10.11.2 100 micrograms triptorelin versus 50 micrograms triptorelin	1	68	Odds Ratio (M‐H, Fixed, 95% CI)	1.00 [0.19, 5.34]
10.12 Other outcomes: pregnancy loss	1		Odds Ratio (M‐H, Fixed, 95% CI)	Subtotals only

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Acharya 1992.

Study characteristics
Methods	Randomised trial The method of allocation was not described. The trial was not blinded.
Participants	Couples with all causes of infertility (unexplained: 20%, male factor: 7%, endometriosis: 18%, tubal factor: 55%)
Interventions	Long follicular GnRHa protocol with buserelin acetate 200 μg I/M x 5 daily from day 2 for at least 13 days until ovarian suppression, then 4 ampoules of HMG daily x 3, then 3 ampoules x 1 day, then 2 ampoules daily thereafter and adjusted based on the response versus short GnRHa protocol with buserelin acetate (dose as above) and HMG (dose as above) commencing 1 day later
Outcomes	Clinical pregnancy per started cycle Multiple pregnancy Number of oocytes retrieved (median and range) Median number of ampoules of gonadotropins used
Notes	Participants in the short protocol group received norethisterone 5 mg twice daily from day 21 of the previous cycle for 7 to 14 days to ensure ovarian suppression and to schedule the cycle start in such a way that the oocyte retrieval was more likely to occur on a weekday. 60% of participants in both groups had 3 embryo transfers. There was 1 cycle per woman.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Participants were randomly allocated to one or the other protocol using a predetermined schedule.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	87 participants were randomised; all participants received treatment and were analysed.
Selective reporting (reporting bias)	Unclear risk	Outcomes other than primary outcomes (live birth/ongoing pregnancy) reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Al‐Jeborry 2020.

Study characteristics
Methods	Randomised trial Randomisation method not described
Participants	POR women were selected according to POSEIDON (Patient‐Oriented Strategies Encompassing IndividualizeD Oocyte Number) criteria and those who underwent ICSI‐ET cycles for primary or secondary infertility. We considered participants in the first 2 arms only (n = 76) and excluded the third arm, which compared GnRH agonist with antagonist protocol. Exclusion criteria: less than 18 years or more than 43 years old, BMI > 30, severe endometriosis, premature ovarian failure, thyroid dysfunction, or hyperprolactinaemia
Interventions	Group A: microdose flare, 25 μg of triptorelin subcutaneously on cycle day 2 or 3 and high‐dose gonadotropins in the form of 300 IU FSH (Gonal–F), and 150 IU HMG (Menopur) given daily until the day of HCG administration Group B: standard flare, 100 μg triptorelin on cycle day 2 or 3 and high‐dose gonadotropins in the form of 300 IU FSH (Gonal–F), and 150 IU HMG (Menopur) given daily until the day of HCG administration
Outcomes	Clinical pregnancy rate Dose of gonadotropin Number of mature oocytes
Notes	Author contacted twice to seek clarification of data reporting, as gross numbers not provided for some outcomes. No response received. Clinical pregnancy data presented as percentages rather than numbers.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation method not described
Allocation concealment (selection bias)	Unclear risk	No information provided regarding allocation concealment
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Blinding not mentioned
Incomplete outcome data (attrition bias) All outcomes	Low risk	All participant data presented
Selective reporting (reporting bias)	High risk	Clinical pregnancy rates reported as percentages; gross numbers not presented. Correspondence sent to authors x2; authors did not respond. For purposes of analysis, percentages converted to numbers, therefore might not be fully accurate.
Other bias	Unclear risk	Prospective trial registration not found

Berker 2010.

Study characteristics
Methods	Randomised trial The trial used computer‐generated block randomisation with sealed envelopes.
Participants	82 poor responder participants who underwent ICSI Inclusion criteria: At least 1 of: day 3 serum FSH level > 10 mIU/mL, < 6 total antral follicles, prior cycle cancellation, prior poor response to COH (peak E2 < 500 pg/mL, < 6 oocytes retrieved, or both) Age > 41 Exclusion criteria: Participants with only 1 ovary, BMI > 30, PCOS, endometriosis stage III to IV, endocrine or metabolic disease, chromosomal disorders, and participants whose partners were azoospermic
Interventions	Participants were randomised into 2 groups: participants in the ultrashort GnRH agonist/GnRH antagonist group (n = 41) were administered leuprolide acetate at 40 μg SC twice daily, started on day 2 of menses and continued for 3 consecutive days, followed by gonadotropins, which were initiated on the last day of leuprolide administration with maximal doses continuing until HCG day. Once the leading follicle had reached a size of 14 mm, co‐treatment was initiated with the GnRH antagonist cetrorelix at 0.25 mg/day, which was continued up to HCG injection; participants in the microdose group (n = 41) began using 40 μg SC twice daily leuprolide acetate on day 2 of menses, and 2 days after initiation of GnRHa, gonadotropin stimulation was initiated and continued until HCG day. The starting dose of rFSH depended on the woman's age, BMI, and ovarian response to the previous cycle and was increased to a maximum of 450 IU/day depending on ovarian response. Dosage of rFSH was individualised after day 5 according to ultrasonographic and hormonal follow‐up. Luteal support was initiated on the day of oocyte retrieval and continued until the day of pregnancy testing with vaginal progesterone.
Outcomes	Number of mature oocytes Clinical pregnancy rate Fertilisation rate Implantation rate Grade A embryo rate Cycle cancellation rate
Notes	Cycle cancellation rates were similar across groups. There was 1 cycle per woman. The population was a selective group (poor responders).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Block randomisation was computer‐generated using sealed envelopes.
Allocation concealment (selection bias)	Low risk	On the day of stimulation initiation, a nurse who assigned participants to their groups opened sealed envelopes with treatment allocation instructions.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not mention blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	A total of 82 poor responder participants underwent 78 COH‐ICSI cycles. Of these participants, 41 received the ultrashort GnRH agonist/GnRH antagonist protocol, and 41 received the microdose flare‐up protocol. Cycle cancellation was carried out for 2 participants in ultrashort GnRH agonist/GnRH antagonist protocol group.
Selective reporting (reporting bias)	Low risk	Most of the outcomes of interest except live birth were reported.
Other bias	Low risk	We suspected no other bias.

Cedrin‐Durnerin 2000.

Study characteristics
Methods	Randomised trial
Participants	230 infertile women undergoing new or repeated IVF cycles Exclusion criteria: women aged 43 or older and those who had chronic anovulation
Interventions	Daily subcutaneous injection of triptorelin (Decapeptyl, 100 μg/day) from day 1 of IVF cycle followed by ovarian stimulation with exogenous gonadotropins 150 IU I/M, with the dose adjusted according to response Women were randomised into 2 groups: GnRHa injected daily from day 1 of IVF cycle to the time of HCG administration; GnRHa administration stopped on the 7th day of the IVF cycle.
Outcomes	Number of HMG ampoules Number of oocytes Pregnancy rate per started cycle Miscarriage rate
Notes	2 variants of the short protocol were used. There was 1 cycle per woman.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	230 women were randomised and received therapy. 30 cycles were cancelled, and analysis was presented for 200 women. The paper thoroughly presented reasons for cancellation.
Selective reporting (reporting bias)	Unclear risk	The published report did not include our primary outcome (live birth/ongoing pregnancy).
Other bias	High risk	An important group of IVF participants (participants with chronic anovulation) were excluded from participation.

Chatillon‐Boissier 2012.

Study characteristics
Methods	Prospective randomised trial
Participants	44 "poor responder" participants undergoing an IVF cycle
Interventions	Participants were randomised into 2 groups: long agonist half‐dose group (20 participants); short agonist group (19 participants). COH with rFSH 300 to 450 IU/d
Outcomes	Number of retrieved oocytes Total number of embryos Pregnancy rate per cycle Pregnancy rate per retrieval Live birth rate
Notes	There was 1 cycle per woman. There was no pretreatment prior to initiation of GnRHa in both groups. This was a special category of participants (poor responders).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were randomised from a computer‐generated list of pseudo‐random permutation of blocks of variable size.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	44 participants were randomised; 39 participants received treatment (reasons specifically mentioned)
Selective reporting (reporting bias)	Low risk	The paper reported most outcomes of interest.
Other bias	Low risk	We suspected no other bias.

Chen 1992.

Study characteristics
Methods	Randomised trial The method of allocation was not described.
Participants	Infertile couples with tubal factor (70%), male factor (10%), endometriosis (18%), and oocyte donation (2%) Average female age: 33 years
Interventions	2 groups: Long follicular GnRHa protocol with leuprolide acetate 1 mg SC daily from day 2 or 3 until ovarian suppression, then FSH 2 ampoules and HMG 2 to 4 ampoules daily in divided doses adjusted depending on the response Ultrashort GnRHa protocol with leuprolide acetate (as above) from day 3. Luteal support with HCG: 1500 IU X 3 and progesterone in oil 50 mg I/M daily
Outcomes	Clinical pregnancy rate Number of oocytes retrieved Number of ampoules of gonadotropins
Notes	‐
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The random sequence generation was not described.
Allocation concealment (selection bias)	Unclear risk	The paper did not describe allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There was insufficient reporting of attrition/exclusions to permit a judgement of 'low risk' or 'high risk'.
Selective reporting (reporting bias)	Unclear risk	The trial reported planned outcome measures but did not report the primary outcomes of interest, live birth and ongoing pregnancy rate.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Dal Prato 2001.

Study characteristics
Methods	Prospective randomised trial
Participants	132 women undergoing COH for IVF/ICSI, aged between 25 and 38 years, with infertility caused by tubal idiopathic and male factor infertility Exclusion criteria: Cases with active endometriosis or only 1 ovary, or with FSH concentration > 15 IU/L on day 3 of menstrual cycles Women with previous COH requiring high doses of gonadotropins in a long GnRHa protocol, or conversely a known history of risk of severe hyperstimulation
Interventions	2 groups: In group 1 (66 women), pituitary desensitisation was performed with single I/M injection of triptorelin, 3.75 mg starting from day 21 of the cycle preceding treatment. In group 2, 66 women received daily SC injections of 100 μg triptorelin starting from day 21 of the preceding cycle. At the onset of menses (start time for FSH stimulation), the dose was reduced to 50 μg SC daily until the day of HCG administration. Luteal support ‐ natural progesterone in oil
Outcomes	Pregnancy rate per woman Number of oocytes Number of ampoules of gonadotropins Miscarriage rates
Notes	Pregnancy was defined as the presence of gestational sac on ultrasound scan performed 4 weeks after embryo transfer. There was 1 cycle per woman.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Allocation was done using sealed envelopes containing the name of 1 of the 2 groups.
Allocation concealment (selection bias)	Low risk	Sealed envelopes were used.
Blinding (performance bias and detection bias) All outcomes	High risk	Participants were not blinded to the treatment.
Incomplete outcome data (attrition bias) All outcomes	Low risk	132 women were randomised; all women received treatment as allocated.
Selective reporting (reporting bias)	Unclear risk	The published report did not include our primary outcome of live birth and ongoing pregnancy rate.
Other bias	Low risk	We suspected no other bias.

De Placido 1991.

Study characteristics
Methods	Randomised trial
Participants	Information not provided.
Interventions	2 groups: Long luteal GnRHa protocol with SC buserelin acetate (0.3 mg x 2 daily from the luteal phase) Short GnRHa protocol with the same dose of buserelin acetate using a short protocol
Outcomes	Clinical pregnancy rate per started cycle
Notes	This trial was a randomised comparison of depot versus daily GnRHa formulation; it was assumed that allocation to the long or short GnRHa protocol was also randomised. No data were provided on the number of participants undergoing oocyte retrieval and ET. Gonadotropin administration, method of oocyte retrieval, and luteal phase management were not described. Most of the information in the risk of bias table is incomplete as this was an abstract. We wrote to the authors but did not receive a reply.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The method of randomisation was not described.
Allocation concealment (selection bias)	Unclear risk	The paper did not describe allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not describe blinding.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	There was insufficient reporting of attrition/exclusions to permit a judgement of 'low risk' or 'high risk'.
Selective reporting (reporting bias)	High risk	Only the clinical pregnancy rate per started cycle was reported. Participant number undergoing oocyte retrieval and embryo transfer not reported. Data taken from abstract. Authors did not reply to correspondence
Other bias	High risk	Data regarding the number of participants and other inclusion criteria were lacking.

Ding 2013.

Study characteristics
Methods	Prospective randomised trial
Participants	96 participants with high response to gonadotropin stimulation compared with reference concentrations undergoing IVF/ICSI cycle Inclusion criteria: Infertile participants with 8 or more subcapsular follicles of 2 to 8 mm in diameter in 1 plane in either ovary undergoing IVF treatments Exclusion criteria: Basal FSH > 10 IU/L Age > 35 years BMI > 30 kg/m² Ovarian surgery, radiotherapy, or chemotherapy Ovarian dysfunction Endometriosis Hyperprolactinaemia thyroid dysfunction Presence of organic pelvic diseases
Interventions	96 participants were allocated to 2 independent groups: GnRH agonist withdrawal group (47 participants): triptorelin was initiated during the luteal phase of the previous cycle (day 21), 0.1 mg/day for 10 days followed by 0.05 mg/day until the concentration of serum E2 ≤ 40 pg/mL. Once the serum E2 concentration was 40 pg/mL, the stimulation of the ovaries was initiated using rFSH (doses ranging from 150 to 250 IU/day). When the diameter of 1 or more follicles was 14 mm, triptorelin (0.05 mg/day) was withdrawn for 2 (15/47) or 3 (32/47) days; control group (49 participants): triptorelin was administered as in group 1, but administration of triptorelin (0.05 mg/day) was continued to the day of triggering ovulation. rFSH administration was administered until the triggering of ovulation in both groups.
Outcomes	Implantation rate per transferred embryo Clinical pregnancy rate per transfer cycle Ongoing pregnancy rate per transfer cycle Multiple pregnancy rate per pregnancy OHSS (moderate/severe)
Notes	Clinical pregnancy was determined by observing a gestational sac by means of echographic screening at 7 weeks of pregnancy. Ongoing pregnancy was defined as a conception cycle with at least 1 foetal sac with a positive heartbeat reaching beyond 12 weeks of amenorrhoea. There was 1 cycle per participant. ET in 29 out of 47 participants in group 1 ET in 26 out of 49 participants in group 2
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was done by computer software.
Allocation concealment (selection bias)	Low risk	The trial used closed envelopes.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not mention blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	This study enrolled 96 participants. Oocyte retrieval cycles: 47/47 and 49/49. The number of retrieved oocytes was reported, but only 54 out of 96 reached ET because on day 3 (18 cycles in the GnRH agonist withdrawal group and 23 cycles in the control group), all embryos were cryopreserved. The criteria for this choice were not mentioned.
Selective reporting (reporting bias)	Low risk	ET occurred in 29 out of 47 participants in group 1, and for 26 out of 49 participants in group 2, ET was not reported. However, the ongoing pregnancy rate was reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Dirnfeld 1991.

Study characteristics
Methods	Randomised controlled trial
Participants	Infertile couples with a previously cancelled or unsuccessful IVF cycle owing to inadequate response Mean female age: 33.5 (range = 26 to 40)
Interventions	2 groups: Long GnRHa protocol with buserelin acetate 1000 μg intranasal daily for 15 to 30 days until ovarian suppression, then reduced to 600 μg daily and HMG 2 to 3 ampoules daily Short GnRHa protocol with buserelin acetate 600 μg intranasal daily from day 1 and HMG 2 to 3 ampoules from day 3 Luteal support from day of oocyte retrieval with progesterone oil 100 mg I/M daily
Outcomes	Number of oocytes Number of ampoules of gonadotropins Clinical pregnancy rate per cycle
Notes	We contacted the author. Long GnRH protocol was commenced in either luteal or follicular phase, although no explanation was given regarding how this decision was made.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	High risk	Allocation was not concealed.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	54 participants were randomised and received treatment.
Selective reporting (reporting bias)	Low risk	The paper reported most of our prespecified relevant outcomes.
Other bias	High risk	We contacted the authors. Long GnRH protocol was commenced in either the luteal or follicular phase

Dirnfeld 1999.

Study characteristics
Methods	Prospective randomised controlled trial
Participants	63 participants with previous poor response to COH, high basal FSH (> 8 mIU/mL), or both, undergoing 78 IVF‐ET cycles All causes of infertility were included. Exclusion criteria: Participants > 42 years of age Participants with irregular menstrual cycles (> 42 or < 21 days)
Interventions	2 groups: Group 1 received 1000 μg/day of nasal spray or 0.1 mg/day of SC triptorelin (Decapeptyl). Treatment with GnRHa was started in the mid‐luteal phase and ended at downregulation. In group 2, ovarian downregulation was performed in an identical manner and was continued through the follicular phase until HCG administration.
Outcomes	Number of oocytes Number of ampoules of gonadotropins Clinical pregnancy rate per cycle Ongoing pregnancy rate per cycle
Notes	There was more than 1 cycle per participant. Outcomes were described as per cycle. Clinical pregnancy was defined as the presence of intrauterine gestational sac on first trimester USG, and ongoing pregnancy was defined as one that progressed beyond 20 weeks' gestation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Although the authors mentioned that a random number table was used to generate a random sequence, it was not clear how the table was created.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	63 women agreed to participate in the trial, but 78 were included in analysis (78 cycles). It was not clear if 63 or 78 participants were randomised.
Selective reporting (reporting bias)	Unclear risk	Live birth/ongoing pregnancy and adverse outcomes not reported
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Fábregues 2005.

Study characteristics
Methods	Prospective randomised study
Participants	150 consecutive infertile women undergoing their first IVF/ICSI cycle Inclusion criteria: Regularly menstruating (26 to 33 days) Aged 26 to 40 years with a normal BMI (19.5 to 28.0) All women had normal ovaries and no previous surgery; none of them had occult ovarian failure on the basis of their basal FSH < 12 IU/L.
Interventions	2 groups: Group 1: pituitary desensitisation was achieved by SC administration of triptorelin acetate (Decapeptyl) 0.1 mg/day started in the mid‐luteal phase of the previous cycle and continued until administration of HCG Group 2: standard daily dose of triptorelin acetate was reduced to 0.05 mg once the ovarian suppression was confirmed and stimulation with recombinant FSH was commenced
Outcomes	Clinical pregnancy Number of oocytes
Notes	Intention‐to‐treat analysis was not done. There was 1 cycle per woman. A total dose of gonadotropin with variance was given rather than number of ampoules.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was computer‐generated.
Allocation concealment (selection bias)	Low risk	Sealed envelopes were used.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	150 women were randomised; all received treatment (13 cycles were cancelled due to low response ‐ there were analyses for 137 women). Although there were 75 women in each group, full data were reported only for 68 and 69 women, respectively.
Selective reporting (reporting bias)	Low risk	Two relevant outcomes reported.
Other bias	Unclear risk	Although the paper gave an a priori sample size calculation, there was insufficient information to assess whether an important risk of bias existed.

Fenichel 1988.

Study characteristics
Methods	Randomised trial Allocation was done by drawing lots.
Participants	Inclusion criteria: Women with tubal factor infertility No more than 3 previous IVF cycles Female aged < 38 years (mean: 31 years) Partner with normal semen analysis
Interventions	2 groups: Long luteal GnRHa protocol with depot triptorelin 3.75 mg I/M then 15 days later HMG 4 ampoules daily x 4 days, then dose adjusted according to response Short protocol with triptorelin 0.1 mg SC daily from cycle day 2 with HMG starting the same day 2 to 4 ampoules x 2 days, then dose adjusted according to response
Outcomes	Clinical pregnancy rate per started cycle Number of oocytes retrieved Number of ampoules of gonadotropins used
Notes	The study also included an arm treated with clomiphene citrate and HMG without GnRHa.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generation was achieved through drawing lots.
Allocation concealment (selection bias)	High risk	The paper did not conceal allocation.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	30 women were randomised; all received treatment.
Selective reporting (reporting bias)	Unclear risk	Our primary outcomes of interest were not reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Foulot 1988.

Study characteristics
Methods	Randomised trial Allocation was done by drawing lots. The trial was not blinded.
Participants	Infertile couples Mean female age: 32 years Exclusion criteria: women with polycystic ovaries
Interventions	2 groups: Long follicular GnRHa protocol with buserelin 0.3 mL SC daily for 14 days, then HMG 2 to 4 ampoules daily Short GnRHa protocol with buserelin (same dose) from day 2, and HMG 1 ampoule on days 2 and 3, then 2 ampoules daily from day 4 Luteal phase support with progesterone (Utrogestan) from day of oocyte retrieval
Outcomes	Clinical and ongoing pregnancy rate per started cycle/per oocyte retrieval/per ET Number of oocytes retrieved Number of ampoules of gonadotropins
Notes	A measure of variance was not given for the number of oocytes and ampoules of gonadotropins.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Allocation was done by drawing lots.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not report blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	100 participants were randomised; all received treatment.
Selective reporting (reporting bias)	Low risk	The ongoing pregnancy rate was reported, but adverse outcomes were not.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Frydman 1988.

Study characteristics
Methods	Randomised controlled trial
Participants	186 infertile couples with predominantly tubal factor (90%) Exclusion criteria: poor responders in previous IVF cycles (defined by cancelled cycles because of low E2)
Interventions	2 groups: In group 1, pituitary desensitisation was obtained by SC injection of buserelin (300 μg twice daily) from day 2 followed by HMG or FSH 2 ampoules twice daily for 7 days, then adjusted based on response In group 2: short GnRHa protocol: triptorelin 0.1 mg SC daily from day 1 or 2 followed by HMG or FSH 2 ampoules daily from day 3 Luteal phase support with dydrogesterone 30 mg daily
Outcomes	Clinical and ongoing pregnancy rate Number of oocytes retrieved
Notes	Participants were randomised to receive HMG and FSH in both protocols (2 interventions). We included the article as current evidence has shown that both gonadotropins are equally effective. Although the number of oocytes retrieved was one of the outcomes measured in the study, we did not include it in the meta‐analysis, as the results varied according to the gonadotropin used.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The method of allocation was not described; we wrote to the trial authors but did not receive any reply.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The trial did not report blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	186 participants were randomised; all received treatment. Our outcomes were reported.
Selective reporting (reporting bias)	Low risk	Most relevant outcomes were reported (ongoing pregnancy rate was reported, but adverse outcomes were not).
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Garcia‐Velasco 2000.

Study characteristics
Methods	Prospective randomised controlled trial
Participants	70 women who were undergoing stimulation for IVF/ICSI cycles and were previous low responders Inclusion criteria: women had to have at least 1 previous cycle cancelled due to poor response and FSH < 12 IU/mL Exclusion criteria: there were no exclusion criteria or age limit
Interventions	2 groups: GnRHa was started in the luteal phase of the previous cycle (leuprolide acetate 1 mg/day SC) on day 21 and was continued in group 1 up until the day of HCG. In group 2, GnRHa was stopped as soon as gonadotropins were commenced. On day 1 and 2 of stimulation, 3 ampoules of HMG were administered together with 5 ampoules of FSH. On days 3, 4, and 5 of ovarian stimulation, 2 ampoules of HMG and 3 ampoules of FSH were administered.
Outcomes	Pregnancy per cycle/per woman Pregnancy per transfer Number of cancellations due to poor response Number of oocytes obtained
Notes	There was 1 cycle per woman. This was a special category of participants (poor responders).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A computerised random number list was used.
Allocation concealment (selection bias)	Unclear risk	The paper did not describe the method of allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not report blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	70 women were randomised; all women were included in analysis.
Selective reporting (reporting bias)	Unclear risk	Primary outcomes of interest were not reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Ghaffari 2020.

Study characteristics
Methods	Randomised controlled trial Assigned in 2 groups, 1:1 ratio, blocks of 4. Randomisation by a third party using computer‐generated random numbers (SPSS version 18.0; IBM, Armonk, NY, USA) prepared by the statistician.
Participants	200 women randomised; poor responders undergoing ICSI Inclusion criteria: presence of at least 2 of the following: Advanced maternal age (≥ 38 years) Day 2 to 3 serum FSH higher than 12 mIU/mL Antral follicle count of 4 or less Prior poor response to COH (< 3 oocytes retrieved) Prior cycle cancellation Exclusion criteria: Only 1 ovary Myoma of 6 cm or larger Follicles larger than 13 mm Older than 42 years Severe male factor infertility (azoospermia)
Interventions	Group A: microdose flare‐up with administration of 80 μg of buserelin (Superfact; Hoechst, Frankfurt, Germany) SC on day 2 or 3 of the menstrual cycle, followed by gonadotropin stimulation 1 day later; both drugs were continued daily until the day of HCG administration. Gonadotropin stimulation was initiated at a dose of 225 IU of Gonal‐F (Serono Laboratories Ltd, Geneva, Switzerland) and 75 IU of Menopur (Ferring GmbH, Kiel, Germany); the dose was subsequently adjusted (maximum 450 IU/day) depending on ovarian response. Group B: standard flare‐up protocol with buserelin administered at a daily dose of 500 μg; the rest similar.
Outcomes	Live birth rate Clinical pregnancy rate Number of oocytes retrieved (median and range) Median number of ampoules of gonadotropins used Miscarriage rate Cycle cancellation rate
Notes	Ethical approval obtained and funding source stated.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated randomisation
Allocation concealment (selection bias)	Unclear risk	Allocation concealment not reported
Blinding (performance bias and detection bias) All outcomes	High risk	Trial not blinded
Incomplete outcome data (attrition bias) All outcomes	High risk	High dropout from number randomised to number analysed. Numbers/reasons presented for those that discontinued intervention and those not included in analysis (not reaching OR or ET) but higher than 30% which was calculated for numbers randomised.
Selective reporting (reporting bias)	Low risk	Live birth rate and miscarriage reported. All outcomes from the numbers analysed as well as numbers per cycle were reported.
Other bias	Low risk	No other bias apparent

Hazout 1993.

Study characteristics
Methods	Randomised trial Allocation was done by using permutation blocks of 8.
Participants	Inclusion criteria: New or repeat IVF participants with either unexplained infertility (31%) or tubal factor (69%) Females aged less than 38 years and duration of infertility < 4 years
Interventions	2 groups: Long follicular GnRHa protocol with triptorelin (Decapeptyl), 3.75 mg depot, administered cycle day 2 then 18 days later or when E2 suppression was achieved, HMG at 4 ampoules daily (or dose based on participant's response in previous cycles) for 5 days with dose adjusted thereafter depending on the response Short protocol with triptorelin (Decapeptyl) 0.1 mg daily for 7 days starting cycle day 2, then starting cycle day 4, HMG 3 ampoules daily for 5 days, with dose adjusted thereafter depending on the response
Outcomes	Clinical pregnancy rate per cycle started, per oocyte retrieval, per ET Number of mature oocytes retrieved Number of ampoules of gonadotropins used/cycle
Notes	The number of pregnancies in the short protocol group was estimated from the pregnancy rates given.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Allocation was done by using a permutation block of 8.
Allocation concealment (selection bias)	High risk	Allocation was not concealed.
Blinding (performance bias and detection bias) All outcomes	High risk	The paper did not report blinding.
Incomplete outcome data (attrition bias) All outcomes	High risk	182 women were randomised. 96 received the long protocol. 84 reported in the text versus 86 in the table received the 7‐day protocol. There were no cancellations mentioned for the 7‐day group.
Selective reporting (reporting bias)	Unclear risk	There was no mention of our primary outcomes.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Hedon 1988.

Study characteristics
Methods	Randomised trial
Participants	Infertile couples (tubal factor: 53%, unexplained: 19%, endometriosis: 7%, combined cause: 22%), excluding those with male factor infertility and ovulation disorders
Interventions	2 groups: Long follicular GnRHa protocol with buserelin 0.3 mL SC x 2 daily on days 2 to 14, then HMG 4 ampoules x 3 days, 2 ampoules x 2 days, then dose adjusted based on response Short GnRHa protocol with buserelin (as above) from day 2 together with HMG 1 ampoule x 2 days, 1.67 ampoules x 3 days, then dose adjusted based on response Luteal support HCG 1500 IU x 2
Outcomes	Clinical and ongoing pregnancy rate per started cycle/per oocyte retrieval/per ET Number of oocytes retrieved Number of ampoules of gonadotropins used
Notes	A measure of variance was not given for the number of oocytes retrieved and number of gonadotropin ampoules.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation was not described.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Although 120 women were randomised, data were available for only 112 women; we wrote to the authors but did not receive any reply. 120 participants were randomised, but 56 participants received treatment in each group = 8 participants were not included due to the reasons mentioned.
Selective reporting (reporting bias)	Low risk	Most relevant outcomes, including 1 of the primary outcomes in this review, were reported except adverse outcomes.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Isikoglu 2007.

Study characteristics
Methods	Prospective randomised trial
Participants	181 women undergoing IVF/ICSI
Interventions	GnRHa was administered from the 21st day of the preceding cycle. Participants were divided into 2 groups: (n = 90): participants were continuously administered GnRHa for 12 days after ET; (n = 91): GnRHa was stopped on the day of HCG administration.
Outcomes	Number of gonadotropin ampoules used Number of mature oocytes recovered Rates of testicular sperm usage Number of embryos transferred Cycle and transfer cancellation rates Clinical pregnancy rate Implantation rate Live birth rate
Notes	Participants were randomised by a computer‐generated list.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was achieved by a computer‐generated list.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	Only embryologists were reported to be blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	181 participants were randomised; all participants were included and mentioned in the analysis.
Selective reporting (reporting bias)	Low risk	All our outcomes were mentioned.
Other bias	Low risk	We suspected no other bias.

Kingsland 1992.

Study characteristics
Methods	Randomised trial Allocation was concealed using sealed envelopes.
Participants	Couples with all causes of infertility (tubal factor: 50%, unexplained: 29%, male factor: 14%, endometriosis: 5%) undergoing their first IVF attempt
Interventions	Women were randomised into 4 groups: A and B without GnRHa, and C and D with GnRHa: group A: 2 ampoules per day of HMG were administered by I/M injection (starting from day 2 of the cycle, 3 ampoules were administered if the woman was over 35 years of age); group B: in addition to regimen in group A, participants were given clomiphene citrate 100 mg/day from day 2 to 6 of the menstrual cycle; group C: ultrashort GnRHa protocol with buserelin 500 μg SC on days 2, 3, and 4 and HMG from day 3; group D: long follicular GnRHa protocol with buserelin 200 μg SC daily from day 1 until pituitary desensitisation, then HMG 3, 4, or 5 ampoules daily (for participants ≤ 35 years, > 35 years, and > 40 years, respectively).
Outcomes	Clinical pregnancy and live birth rate per cycle/per ET Number of oocytes Number of ampoules of gonadotropins
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A random number table was used.
Allocation concealment (selection bias)	Low risk	Sealed envelopes were used.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	308 women were randomised into 4 groups; all participants received treatment. The number of cancelled cycles was reported.
Selective reporting (reporting bias)	Low risk	Most of our outcomes were reported.
Other bias	Low risk	We suspected no other bias.

Kondaveeti‐Gordon 1996.

Study characteristics
Methods	Randomised prospective study
Participants	Women undergoing IVF/ICSI (first cycle only)
Interventions	Downregulation (buserelin acetate intranasal spray 6 times daily for a total daily dose of 1200 μg) commenced on day 1 or day 21 of the cycle
Outcomes	Pregnancy rate Number of oocytes obtained
Notes	There was 1 cycle per woman. Although an a priori power calculation was done, the study was powered only to detect difference in the use of gonadotropins.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was computer‐generated with a permuted block.
Allocation concealment (selection bias)	High risk	Allocation was not concealed.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	86 participants were randomised; all participants received treatment and were analysed.
Selective reporting (reporting bias)	Unclear risk	There was no mention of our primary outcomes.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Lin 2013.

Study characteristics
Methods	Prospective randomised controlled trial
Participants	100 participants undergoing IVF/ICSI cycle Inclusion criteria: Subfertile participants undergoing first IVF/(ICSI) with tubal factor, male factor, or unexplained factor Undertaking a luteal long protocol Basal FSH levels 10 IU/L Aged ≤ 35 years Exclusion criteria: Endometriosis Adenomyosis PCOS
Interventions	In both groups, a single dose of long‐acting GnRHa (triptorelin (Diphereline), 1.25 mg, 3.75 mg/ampoule) was administered on days 20 to 22 of the mid‐luteal phase. Participants were divided into 2 groups: group A: initiation of gonadotropins occurred on the 21st day; group B: initiation of gonadotropins on the 14th day after GnRHa administration. Ovarian stimulation was performed with an initial gonadotropin dose of 75 to 300 IU (rFSH or HMG).
Outcomes	Clinical pregnancy rate Implantation rate Live birth rate Miscarriage rate Moderate OHSS rate
Notes	Clinical pregnancy was defined as a positive serum HCG result, with US evidence of a gestational sac and foetal heartbeat. Μiscarriage rate was defined as the proportion of participants with an initially positive pregnancy test and US evidence of a gestational sac with a foetal pole where pregnancy failed to develop by 12 weeks of gestation. Live birth rate was defined as pregnancies over 28 weeks per treatment cycle of ET. Luteal phase support was started immediately after oocyte retrieval.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was achieved by computer‐generated random numbers 2 weeks after GnRHa administration.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	100 participants from random visits who met the inclusion criteria were recruited. 85 participants were included in analysis. However, all reasons and numbers mentioned for the 15 participants were missing (6 cycles were cancelled due to low response or privacy reasons; ET was cancelled in 6 cycles due to no useable embryos or high risk OHSS ‐ there was no extra justification).
Selective reporting (reporting bias)	Low risk	Relevant outcomes were reported.
Other bias	Low risk	We suspected no other bias.

Loumaye 1989.

Study characteristics
Methods	Randomised trial The method of allocation was not described.
Participants	Inclusion criteria: Couples with tubal factor infertility Females aged < 40 years
Interventions	2 groups: Long luteal GnRHa protocol with buserelin 300 µg intranasally, 3 times daily, then HMG 3 ampoules daily from day 3 of subsequent menses Short GnRHa protocol with buserelin (as above) from day 1 followed by 3 ampoules of HMG daily from day 3 Luteal phase support with HCG 1500 IU I/M on days 6 and 9 after retrieval
Outcomes	Clinical pregnancy rate Ongoing pregnancy rate Number of oocytes obtained
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation was not described.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	18 participants were randomised; all received therapy. Cancellation was not mentioned; 17 out of 18 transferred.
Selective reporting (reporting bias)	Low risk	Our prespecified relevant outcomes were reported, including ongoing pregnancy.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Pellicer 1989.

Study characteristics
Methods	Randomised trial The method of randomisation was not reported.
Participants	Women undergoing IVF between 15 January and 31 May 1998 Inclusion criteria: women who had both ovaries and normal ovarian function prior to IVF
Interventions	Pituitary desensitisation was achieved with 300 μg of buserelin twice a day. Groups 1 and 2 commenced GnRHa in luteal phase (group 1: 4 to 7 days after ovulation, whereas group 2 commenced 8 to 10 days after ovulation). Group 3 commenced GnRHa in the follicular phase. HMG + FSH were used for stimulation. Standard dose was used up to day 5, which was then modified according to individual response.
Outcomes	Number of oocytes Clinical pregnancy
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Participants were randomly allocated into groups, but the method was not described.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	The total number of participants randomised was not mentioned in the methods. In the results section, 44 participants were mentioned as receiving treatment after randomisation.
Selective reporting (reporting bias)	Unclear risk	Our primary outcomes of interest were not reported.
Other bias	High risk	Groups 1 and 2 commenced GnRHa on different days, although both were in the luteal phase.

Ron‐El 1990.

Study characteristics
Methods	Random allocation into 2 groups
Participants	216 consecutive women undergoing IVF/ICSI
Interventions	GnRHa (3.2 mg triptorelin (Decapeptyl) single I/M injection) was given either on day 1 to 3 of the menstrual cycle (group A) or on day 22 (group B). HMG was used for stimulation with a standard dose for the first 4 days followed by individual adjustment of doses.
Outcomes	Number of oocytes retrieved Ampoules of gonadotropins required Clinical pregnancy
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Although the paper reported random allocation, it did not describe the exact method.
Allocation concealment (selection bias)	Unclear risk	The paper did not describe allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	216 women were randomised; all were reported to have received treatment.
Selective reporting (reporting bias)	Unclear risk	Our primary outcomes were not reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

San Roman 1992.

Study characteristics
Methods	Randomised trial
Participants	55 women undergoing IVF‐ET regardless of previous cycle response or number of previous cycles undertaken
Interventions	Group 1 received GnRHa (leuprolide (Lupron)) 1 mg/day SC for 10 days commencing on cycle day 21. After 10 days, if serum E2 was ≤ 184 pmol/L, the GnRHa dose was reduced to 0.5 mg/day SC, and HMG was administered 225 IU I/M. GnRHa was continued until HCG (long protocol). Group 2 commenced concurrent therapy with GnRHa 0.5 mg/day SC and HMG 225 IU I/M beginning on menstrual cycle day 3. Concurrent treatment with GnRHa + HMG was continued for 5 days. GnRHa was continued until HCG (short protocol).
Outcomes	Cycle cancellation Number of ampoules of gonadotropins Clinical pregnancy Live birth
Notes	A clinical pregnancy was defined as USG visualisation of gestational sac or pathological evidence of trophoblast. GnRHa dose was reduced at the start of stimulation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation was not described.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment was not mentioned.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	55 women were recruited and randomised. All women received treatment; outcomes were not reported for 5 of them (low response).
Selective reporting (reporting bias)	Low risk	Our prespecified relevant outcomes were reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Sarhan 2013.

Study characteristics
Methods	Prospective randomised controlled trial
Participants	181 infertile participants undergoing ICSI cycles
Interventions	All participants started treatment with SC daily injections of GnRHa (triptorelin). Participants were divided into 2 groups: group A (66 participants): treatment with the agonist was started on the first or second day of the menstrual period; group B (115 participants): treatment with the agonist was started on day 20 to 22 of the cycle. In both groups, agonist treatment was continued until the day of HCG administration.
Outcomes	Days of stimulation Number of ampoules of gonadotropins used Number of oocytes retrieved per cycle Number of embryos per cycle Fertilisation rate Cleavage rate Pregnancy rate Clinical pregnancy rate per cycle and per ET
Notes	Clinical pregnancy was defined as the presence of intrauterine gestational sac(s) with pulsating heartbeats on transvaginal ultrasound scan at 5 to 6 weeks' gestation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was achieved using closed envelopes.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not report blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	181 participants were randomised. All participants were reported as included in analysis.
Selective reporting (reporting bias)	Unclear risk	Our primary outcomes were not reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Sarhan 2016.

Study characteristics
Methods	Randomised trial Method of randomisation not specified.
Participants	91 women undergoing ICSI cycles Exclusion criteria not specified.
Interventions	All participants had short GnRH agonist protocol with urinary gonadotropins, 2 ampoules per day (HMG (Merional, IBSA, Italy) or purified urinary FSH (Fostimon, IBSA, Italy)) and GnRHa (triptorelin) started on the first day of the menstrual period and continued until the day of HCG administration. The daily dose of triptorelin was different in the 3 groups for comparison: 0.1 mg (100 µg), 50 µg, and 25 µg in groups A, B, and C, respectively.
Outcomes	Clinical pregnancy rate per cycle started Number of oocytes retrieved (median and range) Cycle cancellation rate
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation method not mentioned and there were unequal numbers in group C (23) compared to groups A and B (34 in each). When the corresponding author was contacted, he was unable to recall the reason for this variation. It was, however, not deliberately planned and may have been due to dropouts.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Blinding was not mentioned.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data were presented for all participants.
Selective reporting (reporting bias)	Unclear risk	Our primary outcomes of interest were not reported. Prospective trial registration not found
Other bias	High risk	Inclusion and exclusion criteria not specified. Reported as 91 normal ovulatory women undergoing ICSI.

Simon 1994.

Study characteristics
Methods	Prospective randomised trial
Participants	42 women undergoing a fresh cycle of IVF due to tubal obstruction Inclusion criteria: women under 39 years old who had 2 ovaries and normal ovarian function Exclusion criteria: suspected male factor
Interventions	After pituitary downregulation (serum E2 < 30 pg/mL, serum progesterone < 0.5 ng/mL, and the absence of any ovarian follicle > 10 mm in size), participants were allocated into 2 groups: group A continued to receive the standard dose of 0.5 mg/day LHRHa; group B were given a reduced dose of 0.1 mg/day of LHRHa. Luteal support was provided with I/M progesterone injection in oil.
Outcomes	Number of oocytes retrieved per woman Clinical pregnancies/woman Implantation rate Pregnancy per ET
Notes	There was 1 cycle per woman.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation was not described.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	In the text, it was reported that 43 women were randomised and received treatment, while in the abstract and tables, it was reported that 42 women received treatment.
Selective reporting (reporting bias)	Unclear risk	There was no mention of our primary outcomes.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed

Simons 2005.

Study characteristics
Methods	Double‐blind, randomised, multicentre study
Participants	178 women undergoing IVF/ICSI treatment, history of spontaneous regular cycle between 24 and 35 days Inclusion criteria: Aged 18 to 38 years at the time of screening BMI < 33 Exclusion criteria: Women with either a history of PCOS or incipient ovarian failure Ovulation induction treatment or an IVF/ICSI attempt in the 2 months before the study Poor response to stimulation in previous cycle
Interventions	Group L received the traditional long protocol, i.e. mid‐luteally started triptorelin was continued until the day of HCG injection. In group M, triptorelin was continued up to and including the fourth day of HMG treatment. In group S, triptorelin was stopped at the first day of HMG treatment. Groups L and S were included in the review.
Outcomes	Occurrence of premature LH surge Number of oocytes, implantation rate, clinical and ongoing pregnancy, dose of triptorelin
Notes	Comparison groups for this review: group L versus group S
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was computer‐generated.
Allocation concealment (selection bias)	Low risk	Allocation was concealed in a sealed envelope in a central locker.
Blinding (performance bias and detection bias) All outcomes	Low risk	Participants and personnel were blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	178 participants were randomised; 18 participants were not included in the analysis. (Reasons and numbers were thoroughly reported: discontinuation during the stimulation phase or missing luteinising hormone (LH) data.)
Selective reporting (reporting bias)	Low risk	All our prespecified outcomes were reported.
Other bias	Low risk	There was insufficient information to assess whether an important risk of bias existed.

Sunkara 2014.

Study characteristics
Methods	Prospective randomised controlled trial. Participants were allocated to 1 of the 3 study groups by a third‐party, distant, internet‐based block randomisation to ensure complete allocation concealment. The clinician performing the OR and the embryologist involved were blinded to the treatment allocation.
Participants	111 women with previous poor ovarian response undergoing IVF Exclusion criteria: women aged > 40 years and women with a single ovary
Interventions	Women were allocated to 3 groups: long GnRH agonist group: pituitary downregulation with nafarelin nasal spray 400 mg twice daily was commenced in the mid‐luteal phase of the menstrual cycle and continued for 2 weeks. Οvarian stimulation was commenced with gonadotropin injections at a dose of 450 IU/day and continued with a reduced dose of nafarelin 200 mg twice daily until the administration of HCG injection; short GnRH agonist group: nafarelin nasal spray was commenced on day 2 or 3 of the cycle. Nafarelin was administered at a dose of 200 mg twice daily followed by gonadotropin injections at a dose of 450 IU/day commenced 1 day later. Both nafarelin and gonadotropin injections were continued until the administration of HCG; GnRH antagonist group. Only the first and second groups were included in the review.
Outcomes	Number of oocytes retrieved Mature oocytes retrieved Clinical pregnancy rates Ongoing pregnancy rates Gonadotropin consumption Duration of stimulation Cycle cancellation rate Fertilisation rate Cycles reaching ET
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was computerised.
Allocation concealment (selection bias)	Low risk	Sealed envelopes were used.
Blinding (performance bias and detection bias) All outcomes	Low risk	The clinician performing the OR and the embryologist involved were blinded to the treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	111 women were randomised. 19 women did not receive the allocated intervention (reasons mentioned: 3 conceived spontaneously; 16 decided to postpone IVF treatment).
Selective reporting (reporting bias)	Low risk	All our prespecified outcomes were reported.
Other bias	Low risk	The study appeared to be free of other sources of bias.

Tan 1992.

Study characteristics
Methods	Randomised trial
Participants	Couples with all causes of infertility (unexplained: 25%, male factor: 11%, endometriosis: 5%, tubal factor: 58%) undergoing their first IVF cycle
Interventions	2 groups: Long follicular GnRHa protocol with buserelin acetate 200 μg SC daily from day 1 for at least 14 days until ovarian suppression, then HMG 3, 4, or 5 ampoules daily based on age Short protocol GnRHa protocol with buserelin (as above) from day 2 and HMG from day 3 Luteal support with HCG 2000 IU x 2
Outcomes	Clinical pregnancy rate Number of oocytes retrieved Number of ampoules of gonadotropins
Notes	Significantly more cleaved embryos were available for transfer in participants on the long versus the short protocol. A measure of variance was not given for the number of oocytes retrieved and number of ampoules of gonadotropins.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random tables were used.
Allocation concealment (selection bias)	Low risk	Sealed envelopes were used.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	91 women were randomised and received treatment.
Selective reporting (reporting bias)	Unclear risk	Our primary outcomes were not reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Tasdemir 1995.

Study characteristics
Methods	Randomised trial
Participants	Couples with all causes of infertility (tubal factor: 40%, male factor: 29%, unexplained: 19%, endometriosis: 10%) undergoing their first IVF cycle
Interventions	2 groups: Long luteal protocol with buserelin acetate 900 μg intranasal daily then start HMG cycle day 2 Short GnRHa protocol with buserelin (as above) from cycle day 1 and HMG from cycle day 2
Outcomes	Clinical pregnancy rate Number of oocytes retrieved
Notes	The trial author confirmed that the study was randomised. We assumed 45 participants in each group.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The exact method of randomisation is not known.
Allocation concealment (selection bias)	Unclear risk	The paper did not describe allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	90 participants were randomised. The number of participants allocated to each group was not mentioned in the text or tables. The number of participants receiving treatment and analysed was not mentioned.
Selective reporting (reporting bias)	Low risk	Live birth rate was reported.
Other bias	High risk	The median number of embryos transferred was 4 with the long GnRHa protocol and 1 with the short protocol. We obtained confirmation of randomisation in the original review. We did not receive any reply to further queries. There was no power calculation and no mention of the exact number of participants in each group.

Urbancsek 1996.

Study characteristics
Methods	Prospective randomised trial
Participants	124 women undergoing IVF due to tubal factor or unexplained infertility
Interventions	Buserelin acetate (intranasally 300 μg 4 times a day) starting on day 1 of cycle or in the mid‐luteal phase for pituitary downregulation
Outcomes	Live birth Clinical pregnancy
Notes	A measure of variance for the number of oocytes was not given. Only unexplained infertility and tubal factor participants were included. There was more than 1 cycle per woman; data for only 1 cycle were not available separately.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation was centrally prepared.
Allocation concealment (selection bias)	Unclear risk	The paper did not report allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	High risk	Intention‐to‐treat analysis was not done.
Selective reporting (reporting bias)	Low risk	Live birth and other prespecified outcomes were reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Van de‐Helder 1990a.

Study characteristics
Methods	Randomised trial
Participants	Inclusion criteria: Infertile women with blocked tubes and regular cycles Female aged < 41 years (mean age: 32; range = 23 to 40 years) Partners with normal semen analysis
Interventions	2 intervention groups: Long luteal GnRHa protocol with buserelin 200 μg intranasally daily (3 times) from the mid‐luteal phase until ovarian suppression was confirmed (after which stimulation with HMG was started) Short GnRHa protocol with buserelin from day 1 at the same dose (with start of stimulation with HMG from day 4) Buserelin was continued until the day of HCG administration.
Outcomes	Clinical and ongoing pregnancy rates, per started cycle, per oocyte recovery, per ET Number of oocytes retrieved Number of ampoules of gonadotropins used
Notes	The trial included a third group that was randomised to not receive GnRHa. Clinical pregnancy was defined as foetal heart activity seen on ultrasound. A measure of variance was not provided for the average number of gonadotropins ampoules and average number of oocytes retrieved.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation was not described.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	152 participants were randomised; 152 participants received treatment ‐ there were 23 cancellations, all due to low response.
Selective reporting (reporting bias)	Low risk	Our prespecified outcomes, including ongoing pregnancy rate, were reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Weissman 2003.

Study characteristics
Methods	Randomised prospective study
Participants	60 low responders (from previous cycle) who were undergoing IVF Poor responders were defined as fewer than 5 oocytes retrieved, 3 or fewer follicles 16 mm or larger on the day of cancellation or serum E2 less than 500 pg/mL on the day of HCG administration. Only participants with FSH less than 20 IU/L were included.
Interventions	2 groups: Short protocol: high‐dose GnRHa (500 μg/day) was administered for the first 4 days followed by a standard agonist dose (100 μg/day) Long protocol: standard GnRHa dose (100 μg/day) used until pituitary downregulation, after which the agonist dose was halved during stimulation
Outcomes	Number of oocytes retrieved Number of cancellations Implantation rate Clinical pregnancy
Notes	Both short and long protocols were modified protocols. 1 cycle per woman
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation was computer‐generated.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	60 participants were randomised; treatment was allocated to all of them.
Selective reporting (reporting bias)	Low risk	Most outcomes of interest were reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Yang 1996.

Study characteristics
Methods	Randomised trial
Participants	Couples with all causes of infertility except severe male factor and PCOS (tubal factor: 52%, unexplained: 28%, endometriosis: 17%, male factor: 3%)
Interventions	2 groups: Long GnRHa protocol with leuprolide acetate 1 mg SC daily until ovarian suppression, then dose reduced to 0.5 mg daily together with HMG 3 to 6 ampoules I/M daily x 5 days then HMG dose reduced according to response Modified short protocol with triptorelin (Decapeptyl) 0.1 mg SC daily from cycle day 1 to 7 with HMG (as above) starting cycle day 3 Luteal support with progesterone vaginal suppositories 200 mg x 2 daily with HCG 1500 IU I/M x 4
Outcomes	Clinical pregnancy rate per cycle started Number of oocytes retrieved Number of ampoules of gonadotropins required
Notes	Long GnRHa was commenced in either the luteal or follicular phase. There was 1 cycle per woman.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation was not described; we wrote to the trial authors but received no reply.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not report blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	60 participants were randomised; all received treatment, and no cancellations were reported.
Selective reporting (reporting bias)	Unclear risk	Primary outcomes of interest were not reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Ye 2001.

Study characteristics
Methods	Prospective randomised trial
Participants	109 infertile couples undergoing IVF
Interventions	2 groups: GnRHa long protocol (GnRHa taken by nasal spray 0.9 mg/day starting on day 21 of previous menstrual cycle; gonadotropins were started once pituitary suppression was achieved) Short protocol (GnRH agonist 0.45 mg per day commenced on day 2 of the menstrual cycle; gonadotropins were commenced on the same day)
Outcomes	Ampoules of gonadotropins required Number of oocytes retrieved Pregnancy rate
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐based randomisation was used.
Allocation concealment (selection bias)	High risk	Allocation was not concealed.
Blinding (performance bias and detection bias) All outcomes	High risk	The trial was not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	109 participants were randomised; all received therapy as shown in the tables.
Selective reporting (reporting bias)	Unclear risk	Primary outcomes were not reported.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

Zhang 2009.

Study characteristics
Methods	Prospective randomised trial
Participants	88 participants with infertility due to tubal factor Inclusion criteria: Aged < 35 years of age, BMI 18 to 29 kg/m², duration of menstrual cycle (25 to 35 days), spontaneous ovulation Existence of both ovaries and uterus Adequate male sperm quality feasible for IVF fertilisation Exclusion criteria: PCOS, endometriosis, or severe male factor Systemic, endocrine, or metabolic disease Undergoing radiotherapy and chemotherapy Smokers Those taking narcotics
Interventions	Participants were divided into 2 groups: short GnRHa group (44 participants); long GnRHa group (44 participants).
Outcomes	Total dose of gonadotropins Number of oocytes retrieved, cleavage and fertilisation rates Clinical pregnancy and miscarriage rates Concentrations of IGF‐II and IGFBP‐4 in the follicular fluid
Notes	The article was in Chinese.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random sequence generation was not mentioned.
Allocation concealment (selection bias)	Unclear risk	The paper did not mention allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The paper did not mention blinding.
Incomplete outcome data (attrition bias) All outcomes	Low risk	88 participants were randomised; all participants received treatment. Analyses were mentioned for 88 participants (data derived from tables).
Selective reporting (reporting bias)	Unclear risk	There was no mention of primary outcomes.
Other bias	Unclear risk	There was insufficient information to assess whether an important risk of bias existed.

BMI: body mass index
COH: controlled ovarian hyperstimulation
E2: oestradiol
ET: embryo transfer
FSH: follicle‐stimulating hormone
GnRH: gonadotropin‐releasing hormone
GnRHa: gonadotropin‐releasing hormone agonists
HCG: human chorionic gonadotropin
HMG: human menopausal gonadotropin
ICSI: intracytoplasmic sperm injection
IGFBP‐4: insulin‐like growth factor binding protein‐4
IGF‐II: insulin‐like growth factor II
I/M: intramuscular(ly)
IU: international unit(s)
IVF: in vitro fertilisation
IVF‐ET: in vitro fertilisation pre‐embryo transfer
LH: luteinising hormone
LHRHa: luteinising hormone‐releasing hormone analogue
OHSS: ovarian hyperstimulation syndrome
OR: oocyte retrieval
mIU: milli‐international unit
PCOS: polycystic ovary syndrome
POR: poor ovarian response
rFSH: recombinant follicle‐stimulating hormone
SC: subcutaneous(ly)
US/USG: ultrasound/ultrasonography

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Abd Rabo 2012	Ineligible intervention (letrozole) and specific study population (women with endometriosis using a long agonist protocol)
Aflatoonian 2012	Ineligible intervention: low‐dose HCG in the late follicular phase in controlled ovarian stimulation using a GnRH agonist protocol
Albuquerque 2013	Ineligible study design: this was a Cochrane review
Antoine 1990	Placebo controlled: comparison of GnRHa with no GnRHa
Awaad 2022	Ineligible intervention: GnRH used as a trigger
Azem 2010	Only the abstract was available. There were no data for comparison after repeated attempts to contact the authors.
Beckers 2000	Ineligible intervention: 2 interventions were compared including stopping GnRHa on day 3 of stimulation versus continuing, with or without luteal support
Bloch 2011	This paper used the same study population as in Azem 2010, assessing psychological outcomes.
Braendle 1989	Ineligible study design: allocation to a short or long protocol was sequential and not random
Buvat 1993	Ineligible study design: quasi‐randomisation was used (randomised by year of birth)
Cambiaghi 2011	Ineligible intervention: comparison of GnRH daily or alternate day in a long protocol
Cao 2020	Ineligible study design and population: meta‐analysis of RCTs and non‐RCTs comparing different GnRH agonist protocols in women with endometriosis
Check 1992	Ineligible study design: this was a randomised trial (allocation was based on the last digit of the participant's social security number) comparing long versus ultrashort protocol, but it was excluded due to cross‐over design
Chen 2018	Ineligible study design: retrospective study comparing GnRH agonist in luteal versus follicular phase in 2 of the 3 arms
Cheon 2008	Ineligible intervention: depot versus daily administration of GnRH in long protocol
CN‐02329751	Combined oral contraceptive pill used as add‐on with triptorelin during downregulation.
Corson 1992	This study compared 3 protocols: (a) stopping GnRHa at start of stimulation; (b) reducing GnRHa at start of stimulation; (c) no GnRHa at all for both IVF and GIFT cycles. We could not extract data on IVF cycles separately. We contacted the authors, but separate data were not available, as the study was very old.
Dakhly 2016	Ineligible intervention: growth hormone as an added intervention in 3 GnRHa protocols in poor responders
Dessolle 2011	Ineligible study design: this was a prospective non‐randomised study
Devroey 1994	Ineligible study design: this was a non‐randomised pilot study
Dor 1992	Placebo controlled: this study compared GnRHa with no GnRHa
Eftekhar 2013	Ineligible intervention: daily versus depot dose of GnRHa in a long protocol
Elgendy 1998	Ineligible study design: this paper reported quasi‐randomisation (alternate IVF numbers)
Faber 1998	Ineligible study design: this was a non‐randomised study
Ferraretti 1996	Ineligible study design: this was a retrospective data analysis
Fujii 1997	Ineligible study design: this paper reported quasi‐randomisation (group allocated based on day of visit to the unit)
Garcia 1990	Study design could not be confirmed: the method of allocation to short or long luteal GnRHa protocol was stated to be prospective, but no information was provided on whether randomisation was used. We attempted to contact the authors, but received no reply.
Gersak 1994	Placebo controlled: the paper compared GnRHa with no GnRHa
Gianaroli 1994	Ineligible intervention: this study compared 3 different long protocols: (a) buserelin 0.5 mg SC twice a day 15 days prior to ovarian stimulation; (b) a single dose of long‐acting triptorelin (3.75 mg) 15 days before ovarian stimulation; (c) long‐acting triptorelin 4 weeks prior to ovarian stimulation followed by daily administration of 0.1 mg agonist until HCG injection. None of these comparisons were defined in our protocol.
Gizzo 2014	Ineligible intervention: luteal phase supplementation in GnRH protocols
Harrison 1994	Placebo controlled: this paper compared GnRHa with no GnRHa
Huang 2012	Ineligible study design: this was a retrospective study
IRCT2016022326711N1	Ineligible intervention: antagonist as an added intervention in group 2
Jinno 1996	Ineligible intervention: bromocryptine as an intervention
Jinno 2009	Ineligible intervention: 2 groups received standard long protocol with same dose of GnRHa throughout, but in the experimental group a couple of extra doses of GnRH were given a few hours before and at the time of trigger
Ku 2005	Ineligible study design: this was a retrospective study
Kubik 1990	Placebo controlled: the paper compared GnRHa with no GnRHa
Kuć 2011	Ineligible study design: this was a retrospective study
Li 2012	Ineligible intervention: this paper compared 2 different doses of leuprorelin (Lupron Depot) in GnRH analogues in a long 21 protocol
Liu 2012	Ineligible study design: this was a non‐RCT
Lorusso 2004	Ineligible study design: this was a non‐randomised study
Loutradis 1998	Ineligible intervention: this paper compared 2 regimens in long protocols that did not meet our specified regimens
Lukaszuk 2015	Ineligible intervention: added pretreatment with oestradiol valerate in short protocol and OCP in long protocol
Maged 2018	Comparing ultralong versus standard luteal phase downregulation in endometriosis patients (having unilateral single endometrioma of less than 5 cm in diameter); this is a topic of another systematic review
Marcus 1993	Ineligible study design: this was a randomised trial (allocation was by the last digit of the medical file number) comparing long versus ultrashort protocol. We excluded it because of its cross‐over design.
Maroulis 1991	Ineligible study design: 192 women who were referred for IVF. Randomly allocated to group A (protocol with pure FSH‐HMG); group B (received GnRHa in the luteal phase); or group C (received GnRHa in the follicular phase). During the first 9 months, participants were randomly allocated between protocol A and B (in 2:1 ratio), whereas for the last 11 months between protocols A, B, and C.
McKenna 1989	Ineligible study design: allocation was not random
Mochtar 2011	Ineligible intervention: comparison within 1 protocol (long protocol) only
NCT00436319	Trial not completed as per personal communication with authors.
NCT02342197	Data not available/not published.
NCT05103228	Ineligible intervention: gonadotropin dose compared
Neuspiller 1998	Ineligible population: this was a study on oocyte donors
Norman 1991	Ineligible study design: allocation to a short or long luteal GnRHa protocol was not random, but based on clinical grounds
Padilla 1991	Ineligible intervention: participants were allocated to 5 different protocols based on the results of the leuprorelin (Lupron) screening test. Those with pattern C were randomised into 1 of 3 protocols in phase 1: (1) no GnRHa, (2) double dose GnRHa with flare protocol (not clear whether this was short or ultrashort protocol), or (3) luteal phase GnRHa. In phase 2, they were all given luteal phase GnRHa.
Pantos 1994	Ineligible study design: this study was quasi‐random (alternate)
Polyzos 2015	Ineligible study design: prospective uncontrolled feasibility trial
Remorgida 1989	Ineligible study population: only GIFT cycles were included
Rezaeian 2016	Ineligible intervention: short protocol different to our study definition of the short protocol. Discrepancies in data presented. We were unsuccessful in reaching authors via the contact email provided.
Rodrigues 2014	Ineligible study design: this was a non‐RCT
Ron‐El 1992	Ineligible study design: allocation to ultrashort GnRHa protocol was based on the ability of the participant to attend the clinic on day 1 or 2. These participants were matched by age and indication for IVF to participants having the long GnRHa protocol.
Sarhan 2012	Ineligible intervention: this paper compared 3 GnRH analogues in long protocols
Sathanandan 1989	Ineligible study design: this study comprised a long luteal GnRHa protocol with leuprorelin in participants identified as having poor or abnormal response in a previous stimulation cycle versus short GnRHa protocol with leuprorelin in participants undergoing their first cycle of treatment or who had had a satisfactory response in a previous cycle. Allocation was not random, and participant groups were not similar.
Smitz 1992a	Ineligible study design: quasi‐randomisation (allocated to groups according to year of birth)
Smitz 1992b	Ineligible study design: the method of allocation to short or long GnRHa protocol was not stated. Pregnancy was not the outcome in this study because none of the participants had embryo transfer owing to complete failure of fertilisation.
Stenbaek 2013	Ineligible intervention: this trial compared a short antagonist versus long agonist protocol
Suganuma 1996	Ineligible study design: this paper reported pseudo‐randomisation (alternate participants were allocated to groups). Some participants crossed over between groups.
Tanaka 2014	Ineligible study design: not an RCT
Tarlatzis 1993	Ineligible study design: although the study was designed to have random allocation, in practice the randomisation was incomplete as it was done according to the stimulation protocol, scheduling convenience, and cost of the analogue used.
Tarlatzis 1994	Ineligible study design: although the study was designed to have random allocation, in practice the randomisation was incomplete as it was done according to the stimulation protocol, scheduling convenience, and cost of the analogue used.
Tehraninejad 2010	Ineligible intervention: the paper compared daily doses versus leuprorelin (Lupron Depot) of GnRH in a long 21 protocol
Tehraninejad 2021	Ineligible intervention: the paper compares different doses of GnRHa
Umemmuo 2020	Ineligible study design: not an RCT
Van de‐Helder 1990b	Placebo controlled: the paper compared GnRHa with no GnRHa
Wu 2012	Ineligible study design: participants were assigned to 4 groups according to serum progesterone and oestradiol concentrations on the day of HCG administration
Xu 2020	Ineligible intervention: depot versus standard long GnRH agonist protocol, which is the subject of another Cochrane review (Albuquerque 2013)
Yang 1991	Placebo controlled: the paper compared GnRHa with no GnRHa
Ying 2019	Ineligible study design: retrospective cohort study comparing GnRH in luteal versus follicular phase

FSH‐HMG: follicle‐stimulating hormone‐human menopausal gonadotropin
GIFT: gamete intra‐fallopian transfer
GnRHa: gonadotropin‐releasing hormone agonists
HCG: human chorionic gonadotropin
HMG: human menopausal gonadotropin
ICSI/ET: intracytoplasmic sperm injection/embryo transfer
IVF: in vitro fertilisation
OCP: oral contraceptive pill
RCT: randomised controlled trial
SC: subcutaneous(ly)

Characteristics of studies awaiting classification [ordered by study ID]

Kabodmehri 2021.

Methods	Randomised controlled trial
Participants	394 participants who were candidates for intracytoplasmic sperm injection/embryo transfer were included in this study between April 2019 and January 2020.
Interventions	Long‐acting GnRHa compared to short‐acting GnRHa
Outcomes	Ovarian hyperstimulation syndrome rate, progesterone and oestradiol levels at human chorionic gonadotropin administration day
Notes	Abstract only

GnRHa: gonadotropin‐releasing hormone agonist

Characteristics of ongoing studies [ordered by study ID]

NCT02681536.

Study name	Miniflare versus long protocol in poor responders
Methods	Randomised trial
Participants	200
Interventions	Mini‐dose long protocol versus microdose flare protocol
Outcomes	Number of oocytes retrieved
Starting date	February 2016
Contact information	Dina M Dakhly, MD; 01003498919 ext 002; Dinadakhly@gmail.com
Notes	Correspondence sent to author; no response received.

NCT02940535.

Study name	Low dose GnRHa early luteal phase down regulation versus GnRHa ultra‐short protocol for poor ovarian response
Methods	Randomised trial
Participants	200
Interventions	Early luteal phase downregulation versus ultrashort protocol
Outcomes	Live birth rate, clinical pregnancy rate, number of oocytes, miscarriage, ovarian hyperstimulation syndrome
Starting date	December 2016
Contact information	Wei Shang; shang.wei@163.com
Notes	Correspondence sent to contact person, Dr Wei Shang, in April 2020, who confirmed the study had not started recruiting yet but is happy to share data when available.

NCT05567731.

Study name	Effect of long, short, and ultrashort GnRH agonist treatment protocols in intracytoplasmic sperm injection candidates on ovarian reserve
Methods	Randomised controlled trial
Participants	30 participants 18 to 35 years old undergoing intracytoplasmic sperm injection
Interventions	Ultrashort, short, and long GnRHa protocols
Outcomes	Primary outcome: number of ongoing pregnancies per woman randomised, defined as evidence of a gestational sac with foetal heart motion at 12 weeks or later, confirmed with ultrasound Secondary outcomes: number of oocytes retrieved, oestradiol, follicle‐stimulating hormone and luteinising hormone level on day 2 of menstruation
Starting date	15 October 2022
Contact information	Ahmed Ossman, Tanta University
Notes

GnRHa: gonadotropin‐releasing hormone agonist

Differences between protocol and review

2024

In the current review update, we introduced one new comparison group for the short protocol, where the use of different doses has been compared. This was not a prespecified comparison for the review, but was added as a result of the included studies.

In accordance with suggestions from peer reviewers and editors, in this third update, we decided to restrict our primary analysis strategy for live birth and clinical pregnancy to include only studies at low risk of selection and reporting bias. We acknowledge that while this strategy means fewer studies are included, the final conclusion is more robust. For the long versus short protocol comparison, which had a higher number of studies, we carried out sensitivity analysis including all studies for the primary outcomes and the secondary outcome of clinical pregnancies, as we felt this would be clinically relevant.

2014

We added two comparisons:

• long protocol: discontinuing versus continuing gonadotropin‐releasing hormone agonists (GnRHa) after human chorionic gonadotropin (HCG) administration; and

• long protocol: administration of GnRHa for two versus three weeks before stimulation.

Our protocol mentioned one of the comparison groups as GnRHa versus placebo. However, there is a review of randomised controlled trials on this topic (Fields 2013), which suggests that the use of GnRHa is associated with a better outcome in assisted reproduction technology. The current review intended to explore which protocol of GnRHa was better.

A short protocol versus a 'short stop' protocol was not listed in the initial comparison groups. However, given that we were looking at all protocols for GnRHa for pituitary downregulation, we felt it was appropriate to include studies comparing these groups.

2011

A further title change agreed upon from 'Gonadotropin‐releasing hormone agonist protocols for pituitary down‐regulation in assisted reproductive treatment' to 'Gonadotropin‐releasing hormone agonist protocols for pituitary suppression in assisted reproduction'.

2009

The original review was withdrawn and a new protocol published. Title changed back from 'Long versus short gonadotropin releasing hormone agonist protocols for pituitary desensitization in assisted reproduction cycles' to 'Gonadotrophin‐releasing hormone agonist protocols for pituitary suppression in assisted reproductive treatment'.

2008

Title changed from 'Gonadotrophin‐releasing hormone agonist protocols for pituitary down regulation in assisted reproductive treatment' to 'Long versus short gonadotropin releasing hormone agonist protocols for pituitary desensitization in assisted reproduction cycles'.

Original protocol published in 2008.

Contributions of authors

CS: co‐drafted the initial protocol, selected studies for inclusion, confirmed data extraction, analysed the data, wrote the second update, and edited the current update.

LNY: screened the database searches and approved the current update.

AM: initiated and conceptualised the protocol, edited the second update, provided overall supervision and editing of the review.

SRCB: undertook data searching, selected studies for inclusion, performed data extraction and analysis, and wrote the current update.

CS and AM are members of the Cochrane Gynaecology and Fertility Group editorial board. They took no part in editorial decisions and related activities (including selection of and contact with peer reviewers, liaison with editors, sign‐off for publication).

Sources of support

Internal sources

• Assisted Reproduction Unit, University of Aberdeen, UK

  Institution Research

• Assisted Reproduction Unit, 3rd Department of Obstetrics and Gynecology, National and Kapodistrian University of Athens, Greece

  Institution research

External sources

• Cochrane Gynaecology and Fertility (CGF) Group, New Zealand

  Institution research

Declarations of interest

Charalampos S Siristatidis: nothing to declare concerning this review. I run my private practice office. In my new Unit (2nd Dept of Obstet‐Gynecol, University of Athens, Greece) I have no funding from companies. The only funding that I use comes from E‐learning programs through the National and Kapodistrian University of Athens.

Li Ning Yong: nothing to declare.

Abha Maheshwari: declares receipt of travel support from Ferring Pharmaceuticals, Gedeon Richter, and Merck, and consultancy/independent contractor fees from Ferring Pharmaceuticals and Cook Medical.

Smriti Ray Chaudhuri Bhatta: nothing to declare.

These authors should be considered joint first author

New search for studies and content updated (no change to conclusions)