Uptake of the core outcome set on polycystic ovary syndrome before and after its publication

Abstract

STUDY QUESTION

Does the core outcome set (COS) on polycystic ovary syndrome (PCOS) impact the selection of research outcomes?

SUMMARY ANSWER

Following the publication of the COS on PCOS, an increasing number of trials are reporting both the generic domain and body mass index; however, the uptake of this COS has not been as extensive as expected.

WHAT IS KNOWN ALREADY

The COS on PCOS included 33 core outcomes in the following seven domains: the generic (3), metabolic (8), reproductive (7), pregnancy (10), psychological (3), oncological (1), and long-term (1). This was done to improve consistency in outcome selection and definition. However, thus far, no studies have investigated the effectiveness of this COS in the above-mentioned tasks.

STUDY DESIGN, SIZE, DURATION

A methodological study based on the trial registries, including 395 eligible clinical trials registered between 1 January 2018 and 21 September 2022.

PARTICIPANTS/MATERIALS, SETTING, METHODS

A total of 1258 registered clinical studies on PCOS were retrieved from the World Health Organization International Clinical Trials Registry Platform. Of those, 395 were selected according to the inclusion and exclusion criteria, and divided into two groups based on the publication date of the COS on PCOS (4 February 2020): pre-publication and post-publication. The practical uptake of this COS was explored after data collation, assessment, comparison of the uptake of core outcomes or domains before and after the publication of this COS, and correlation analysis between the domains.

MAIN RESULTS AND THE ROLE OF CHANCE

There were 26 out of 33 core outcomes and five out of seven domains reported in the 395 trials. The highest uptake was observed for the reproductive domain and the reproductive hormonal profile (63.0% and 38.7%, respectively). After the publication of the COS on PCOS, the uptake of the generic domain and body mass index increased from 24.1% to 35.8% (P = 0.011) and 17.8% to 26.5% (P = 0.039), respectively. The total number of reported core outcomes in the generic domain met statistical significance (P = 0.012). Moreover, multivariable analyses still supported the above finding in the generic domain. Correlation analysis showed that most of the domains were positively correlated with each other. However, the pregnancy domain was negatively correlated with the metabolic domain. Reasons responsible for the unsatisfactory uptake may be the absence of specific definitions of core outcomes, as well as the lack of awareness among researchers regarding this COS.

LIMITATIONS, REASONS FOR CAUTION

Due to the lack of standardized definition of outcomes, it was difficult to avoid some subjectivity in the process of consistency assessment.

WIDER IMPLICATIONS OF THE FINDINGS

Two years after its publication, there was no substantial improvement in the uptake of the COS on PCOS. This suggests that this COS may require further revision, refinement, and promotion to improve the comparability of PCOS studies.

STUDY FUNDING/COMPETING INTEREST(S)

This work was funded by Beijing Municipal Health Science and Technology Achievements and Appropriate Technology Promotion Project (BHTPP2022069), and the special fund of Beijing Key Clinical Specialty Construction Project. The authors do not have conflicts of interest to declare.

TRIAL REGISTRATION NUMBER

N/A.

Introduction

Minimizing bias when designing clinical trials is necessary for the direct comparison of the effects of different interventions. Therefore, the selection of appropriate outcomes is crucial (Williamson et al., 2012; Smith et al., 2019a). However, reporting outcomes in clinical studies is generally inconsistent, unimportant, or incomplete, thereby resulting in a waste of resources (Thornley and Adams, 1998; Chalmers and Glasziou, 2009; Williamson et al., 2020). A core outcome set (COS) refers to an agreed standard set of outcomes that should be measured and reported, as a minimum, in all clinical trials for specific areas of health or health care (Williamson et al., 2012; Clarke and Williamson, 2016; WHO Working Group on the Clinical Characterisation and Management of COVID-19 Infection, 2020). The main function of COSs was to reduce risk of reporting bias, improve consistency between similar studies, facilitate integration and comparison of data, and simplify the selection of study outcomes (Kirkham et al., 2010; Dwan et al., 2013; Smith et al., 2015; Clarke and Williamson, 2016; Topjian et al., 2020).

Polycystic ovary syndrome (PCOS) is a key disorder of concern and an active topic of scientific research in the reproductive, metabolic, and endocrine disciplines (Escobar-Morreale, 2018). The COS on PCOS comprises 33 core outcomes in the following seven domains: 3 generic, 8 metabolic, 7 reproductive, 10 pregnancy, 3 psychological, 1 oncological, and 1 long-term outcomes. This COS was developed to assist researchers in the selection of outcomes in practice. The developers of this COS expected researchers to tailor the reporting of outcomes according to their research questions. The objective was to cover all relevant outcomes in this COS while justifying the lack of reporting for any remaining outcomes (Al Wattar et al., 2020). The World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) provides information on PCOS clinical trials collected from various registries, with data reflecting the design, planned completion time and outcomes selection of clinical trials. It can help in efficient evaluation of the uptake of COSs and thus has been used in some studies (Kirkham et al., 2017; Smith et al., 2019a,b).

In recent years, an increasing number of researchers have been focusing on COSs. A substantial increase in the development of COSs has been recorded (Williamson et al., 2020). However, thus far, studies on the uptake of COSs are lacking (Hughes et al., 2021). Previous studies showed that despite the rapid development of COSs, the clarity of reporting of core outcomes was suboptimal and numerous COSs still require improvement (Kirkham et al., 2016; Gargon et al., 2019; Goren et al., 2023). Moreover, the low utilization rate causes a waste of resources; this contradicts the original intention of COSs development. Therefore, it is necessary to ensure the representativeness, universality, and practicability of COSs. Based on systematic assessment, this study investigated the uptake of the COS on PCOS research before and after its publication. The assessment can offer developers the opportunity to review their research results, and avoid situations in which a COS was developed but not effectively utilized.

Materials and methods

Research data acquisition

Using the keyword ‘polycystic ovary syndrome’, clinical studies registered between 18 October 1999 and 28 August 2022 were retrieved from the WHO ICTRP Search Portal.

The inclusion criteria were (i) registered clinical studies involving patients with PCOS, (ii) interventional clinical trials, and (iii) clinical trials registered from 1 January 2018 to 21 September 2022 (date of retrieval). The exclusion criteria were (i) studies which included non-PCOS populations and (ii) studies missing information on key outcomes.

Finally, eligible clinical trial registrations were ultimately included in this analysis. We collected relevant information (trial ID, public title, time of data registration, web links, study type, study design, country or region, primary outcomes, secondary outcomes, etc.). The purpose was to establish a database through Research Electronic Data Capture (REDCap: Vanderbilt University, Nashville, TN, USA), as well as to retrieve and extract the original registration data in registries through the web links to update the information directly provided by the WHO ICTRP.

Assessment for the uptake of the COS on PCOS

The outcomes of these clinical trial registry entries were evaluated in REDCap to determine whether they were recommended by the COS on PCOS (including three options, namely consistent, inconsistent, and derivable outcomes). Derivable outcomes refer to some outcomes in trial registrations that are not exactly consistent with the outcomes of this COS, but can be associated with them through a certain conceptual extension. For example, only reporting ‘blood pressure’ does not conform to ‘hypertension’ in this COS; however, it can be included in the derivable outcomes of hypertension according to the above criteria. Independent evaluations by two assessors were conducted to explore the consistency between the reported all outcomes in PCOS clinical trial registrations and this COS. Disagreements were resolved through group discussion; for unresolved disagreements and some undefined outcomes, a consensus was reached through consultation of experts.

According to the publication date of the COS on PCOS (4 February 2020), the eligible trials were divided into two groups, namely the pre-publication (Pre-Pub) group and post-publication (Post-Pub) group. Among the three options of all core outcomes, consistent outcomes were regarded as the reporting of core outcomes, while the others were regarded as not reporting outcomes. As randomization, blinding and prospective registration may affect the uptake of the COS, we also included these three factors in the analysis. To explore the adoption of this COS, the number of clinical trials reporting each core outcome or domain, and the total number of reported core outcomes were calculated. Clinical trials that reported at least one core outcome from a core domain were considered to have adopted the domain. The total number of reported core domains was categorized into four categories (i.e. 0, 1, 2, ≥3); subsequently, the number of trials for each category was calculated. In addition, the pairwise correlation between domains was analyzed to determine the domains that researchers prefer to report simultaneously.

Statistical analysis

The study database was established using REDCap. Statistical analysis was conducted using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Measurement data were presented as the median with upper and lower quartiles. The Mann–Whitney U test was used for comparisons between the groups in terms of the total number of reported core outcomes in all domains and each of them. Enumeration data were expressed as numbers and percentages. The chi-squared test was used for comparisons between two groups in terms of the number of trials reporting each core outcome or domain, and the number of trials classified into the four aforementioned categories. Multivariable logistic regression and a generalized linear model were used to explore the influence of four factors on the uptake of this COS. Spearman correlation was used to explore the pairwise correlation between the domains. Figures were plotted using OriginPro 2022 (OriginLab Corp., Northampton, MA, USA). Two-tailed P-values ≤0.05 indicated statistically significant differences.

To evaluate the robustness of our findings, we conducted sensitivity analysis by adjusting the evaluation criteria and the time of trial registration. The former was achieved by selecting the consistent and derivable outcomes as the reporting of core outcomes and the inconsistent outcomes as not reporting outcomes. The latter was achieved through exploring annual variations in the distribution of the total number of reported core outcomes in all domains and each of them. The procedures described above were repeated once.

Results

The screening process for PCOS registered clinical studies is shown in Fig. 1. A total of 1258 PCOS-related studies were retrieved from the WHO ICTRP. Finally, 395 trials were selected and divided into the Pre-Pub and Post-Pub groups (191 and 204 interventional trials, respectively).

Figure 1.

Flowchart of the screening process for PCOS clinical trial registrations retrieved from WHO ICTRP. PCOS, polycystic ovary syndrome; WHO, World Health Organization; ICTRP, International Clinical Trials Registry Platform.

The characteristics of the 395 eligible clinical trials are presented in Table 1. Of those, 281 (71.1%) were randomized and 114 (28.9%) were non-randomized controlled trials. The median sample size was 78 (Q₁, Q₃: 52, 128). According to the data, Iran had the largest number of clinical trial registrations (n = 115, 29.1%). Notably, the vast majority of clinical trials were single-center studies (n = 386, 97.7%). The minimum and the maximum age of the study population were 18 years (Q₁, Q₃: 18, 20) and 40 years (Q₁, Q₃: 36, 44), respectively. There were no statistically significant differences between two groups in any of the characteristics.

Table 1.

Characteristics of the 395 clinical trial registrations on PCOS.

Characteristics	Pre-Pub n = 191	Post-Pub n = 204	Total n = 395	P-value
Study design, n (%)*
RCT	143 (74.9)	138 (67.6)	281 (71.1)	0.113
Non-RCT	48 (25.1)	66 (32.4)	114 (28.9)
Median sample size (Q1, Q3) †	80 (60, 120)	75 (50, 140)	78 (52, 128)	0.970
Country or region, n (%)*
Iran	62 (32.5)	53 (26.0)	115 (29.1)	0.069
China	40 (20.9)	57 (27.9)	97 (24.6)
Egypt	19 (9.9)	16 (7.8)	35 (8.9)
India	16 (8.4)	19 (9.3)	35 (8.9)
USA	18 (9.4)	8 (3.9)	26 (6.6)
Othersa	36 (18.8)	51 (25.0)	87 (22.0)
Median age of participants (Q1, Q3), years †
Minimum (n = 323)b	18 (18, 20)	18 (18, 19)	18 (18, 20)	0.461
Maximum (n = 320)c	40 (38, 44)	40 (35, 44)	40 (36, 44)	0.540

^aCountries or regions with five or fewer registered trials were classified as others. There were three and six international multicenter studies in the Pre-Pub and Post-Pub groups, respectively.

^b30 and 42 trials had missing data in the Pre-Pub and Post-Pub groups, respectively.

^c31 and 44 trials had missing data in the Pre-Pub and Post-Pub groups, respectively.

^*χ² test was used.

^†Mann–Whitney U test was used.

PCOS, polycystic ovary syndrome; Pre-Pub, pre-publication; Post-Pub, post-publication; Q₁, first quartile; Q₃, third quartile; RCT, randomized controlled trial.

Overall, 26 of the 33 core outcomes of the COS on PCOS were reported. Table 2 lists the uptake of this COS in included trials. The reporting percentages of core outcomes ranged from 0.3% to 38.7%. Nine outcomes had an uptake percentage >10%, including four reproductive outcomes, three metabolic outcomes, and two generic outcomes. The reproductive hormonal profile exhibited the highest uptake percentage (38.7%). The uptake of 12 outcomes increased following the publication of this COS. BMI was more frequently used in the Post-Pub group compared with the Pre-Pub group (17.8% versus 26.5%, respectively, P = 0.039). No significant intergroup differences were noted in terms of the uptake of the remaining core outcomes.

Table 2.

Uptake of the COS on PCOS in the clinical trial registrations before versus after publication (outcome level).^†

Outcomes	Pre-Pub n = 191 n (%)	Post-Pub n = 204 n (%)	Total n = 395 n (%)	P-value
Generic outcomes
BMI	34 (17.8)	54 (26.5)	88 (22.3)	0.039
Quality of life	20 (10.5)	29 (14.2)	49 (12.4)	0.259
Treatment satisfaction	1 (0.5)	5 (2.5)	6 (1.5)	0.217*
Metabolic outcomes
Waist circumference	18 (9.4)	28 (13.7)	46 (11.6)	0.183
Insulin resistance	36 (18.8)	42 (20.6)	78 (19.7)	0.664
Impaired glucose tolerance	17 (8.9)	18 (8.8)	35 (8.9)	0.979
Hypertension	1 (0.5)	0 (0)	1 (0.3)	0.484*
Coronary heart disease	0 (0)	1 (0.5)	1 (0.3)	>0.999*
Lipid profile	50 (26.2)	47 (23.0)	97 (24.6)	0.469
Venous thromboembolic disease	1 (0.5)	0 (0)	1 (0.3)	0.484*
Reproductive outcomes
Viable pregnancy	24 (12.6)	30 (14.7)	54 (13.7)	0.536
Hyperandrogenism	75 (39.3)	71 (34.8)	146 (37.0)	0.358
Menstrual regularity	38 (19.9)	46 (22.5)	84 (21.3)	0.519
Reproductive hormonal profile	75 (39.3)	78 (38.2)	153 (38.7)	0.833
Ovulation stimulation success and number of stimulated follicles ≥12 mm	4 (2.1)	5 (2.5)	9 (2.3)	>0.999*
Incidence and severity of ovarian hyperstimulation syndrome	7 (3.7)	11 (5.4)	18 (4.6)	0.411
Pregnancy outcomes
Live birth	14 (7.3)	16 (7.8)	30 (7.6)	0.847
Miscarriage	14 (7.3)	9 (4.4)	23 (5.8)	0.216
Stillbirth	1 (0.5)	0 (0)	1 (0.3)	0.484*
Preterm birth	5 (2.6)	6 (2.9)	11 (2.8)	0.845
Gestational weight gain	1 (0.5)	0 (0)	1 (0.3)	0.484*
Gestational diabetes	4 (2.1)	2 (1.0)	6 (1.5)	0.436*
Hypertensive disease in pregnancy	4 (2.1)	2 (1.0)	6 (1.5)	0.436*
Baby birthweight	5 (2.6)	2 (1.0)	7 (1.8)	0.271*
Psychological outcomes
Depression	17 (8.9)	18 (8.8)	35 (8.9)	0.979
Anxiety	13 (6.8)	11 (5.4)	24 (6.1)	0.557

^†χ² test was used.

^*Fisher’s exact test was used.

PCOS, polycystic ovary syndrome; COS, core outcome set; Pre-Pub, pre-publication; Post-Pub, post-publication; BMI, body mass index.

Figure 2A shows that the total number of reported core outcomes in all domains was slightly higher after than before the publication of the COS on PCOS (2 (Q₁, Q₃: 1, 3) versus 2 (Q₁, Q₃: 1, 4), P = 0.742). There were significant increases in the generic domain after the publication of this COS (P = 0.012), as well as in the metabolic domain between the randomization versus non-randomization group (P = 0.016) and between the blinding versus non-blinding group (P = 0.015) (Fig. 2; Supplementary Figs S1, S2, S3, and S4).

Figure 2.

Distribution of the total number of reported core outcomes in all domains, the generic domain and the metabolic domain. The range of whiskers in the boxplot is ±1.5*IQR. There were statistically significant differences in the generic domain between the Post- versus Pre-Pub group (P = 0.012), in the metabolic domain between the randomization versus non-randomization group (P = 0.016), and in the metabolic domain between the blinding versus non-blinding group (P = 0.015). The effect of differences in the number of trials between the groups on the shape of distribution has been eliminated. Pre-Pub, pre-publication; Post-Pub, post-publication; IQR, interquartile range.

Among the included trials, five of the seven domains of the COS on PCOS were reported (Table 3). The reproductive domain was the most commonly reported (63.0%). Fewer trials reported the pregnancy and psychological domains (10.6% and 9.4%). The oncological and long-term domains were not reported. The uptake percentage of the generic domain in the Pre-Pub and Post-Pub groups was 24.1% and 35.8%, respectively, reaching statistical significance (P = 0.011). Significant increases were observed in the metabolic domain between the randomization versus non-randomization group (P = 0.017) and between the blinding versus non-blinding group (P = 0.023) (Supplementary Table S1).

Table 3.

Uptake of the COS on PCOS in the clinical trial registrations before versus after publication (domain level) and multivariable logistic regression analysis for trials reporting each domain or not.

Domains	Pre-Pub n = 191 n (%)	Post-Pub n = 204 n (%)	Total n = 395 n (%)	P-value*	aOR (95% CI) #
					Post- versus Pre-Pub	Randomization	Blinding	Prospective registration
Generic outcomes	46 (24.1)	73 (35.8)	119 (30.1)	0.011	1.78 (1.14–2.78)	1.34 (0.77–2.34)	1.30 (0.80–2.13)	1.51 (0.97–2.35)
Metabolic outcomes	72 (37.7)	75 (36.8)	147 (37.2)	0.848	1.00 (0.66–1.51)	1.55 (0.91–2.64)	1.37 (0.86–2.17)	1.21 (0.79–1.83)
Reproductive outcomes	123 (64.4)	126 (61.8)	249 (63.0)	0.588	0.87 (0.57–1.31)	1.07 (0.64–1.79)	0.81 (0.51–1.29)	1.29 (0.85–1.95)
Pregnancy outcomes	22 (11.5)	20 (9.8)	42 (10.6)	0.581	0.80 (0.42–1.52)	0.75 (0.35–1.62)	0.89 (0.43–1.84)	1.24 (0.64–2.37)
Psychological outcomes	18 (9.4)	19 (9.3)	37 (9.4)	0.970	1.01 (0.51–2.00)	1.17 (0.48–2.84)	1.21 (0.56–2.59)	1.00 (0.51–1.99)

^*χ² test was used. The oncology and long-term domains were not reported.

^#Multivariable logistic regression model was used by adjusting for factors involving Post- versus Pre-Pub, randomization, blinding, and prospective registration.

PCOS, polycystic ovary syndrome; COS, core outcome set; Pre-Pub, pre-publication; Post-Pub, post-publication; aOR, adjusted odds ratio; CI, confidence interval.

Nonetheless, multivariable logistic regression analysis (adjusted odds ratio = 1.78 (95% confidence interval: 1.14–2.78), P = 0.011) and the generalized linear model (β = 0.41, P = 0.018) only supported the statistical significance in the generic domain between the Post- versus Pre-Pub group (Table 3; Supplementary Table S2). Moreover, the practice of prospective registration may affect the total number of reported core outcomes in all domains (β = 0.25, P < 0.001).

The number and percentage of trials classified into each category are presented in Fig. 3. Among the four categories, the highest number was recorded for trials reporting only one domain (n = 128, 32.4%). Moreover, 19 (4.8%) trials and 1 (0.3%) trial reported four and five domains, respectively. The percentage of trials reporting three or more domains increased from 17.13% to 22.5%, indicating an improvement in the consistency of outcome reporting. However, there was no statistically significant difference between the two groups (P = 0.578).

Figure 3.

Number of trials for totals of reported core domains before versus after the publication of the COS on PCOS. The categories have been adjusted. Since few trials reported four or more domains, they were merged with the category reporting three domains. There was no statistically significant difference between the two groups (χ² = 1.974, P = 0.578). PCOS, polycystic ovary syndrome; COS, core outcome set; Pre-Pub, pre-publication; Post-Pub, post-publication.

As shown in Fig. 4, most domains were positively correlated with each other before and after the publication of the COS on PCOS. The generic and metabolic domains showed the strongest correlation in the Pre-Pub and Post-Pub groups (r = 0.345, P < 0.001 and r = 0.406, P < 0.001, respectively). This suggested that these domains were more likely to be simultaneously selected by researchers as the outcomes of clinical trials than other domains. However, we also found that the pregnancy domain was negatively correlated with the metabolic domain before and after the publication of this COS (P = 0.003 and 0.034, respectively). But these correlations were weak.

Figure 4.

Pairwise correlation between the five domains before versus after the publication of the COS on PCOS. ‘*’ represented that there was a statistical significance in the correlation between the pairwise of five domains by Spearman correlation (P-values ≤0.05). PCOS, polycystic ovary syndrome; COS, core outcome set; Pre-Pub, pre-publication; Post-Pub, post-publication.

Sensitivity analysis

For sensitivity analysis, we altered the grading standard for the uptake of the COS on PCOS. There was no statistically significant evidence that the overall results of this study were substantially changed by the two adjudication standards and the time of trial registration (Supplementary Tables S3 and S4 and Figs S5, S6, S7, and S8).

Discussion

In the present study, the reproductive domain and reproductive hormonal profile (63.0% and 38.7%, respectively) were recorded as having the highest uptake. After the publication of the COS on PCOS, except for the generic domain and its included body mass index (P = 0.011 and 0.039, respectively), the number of trials reporting other outcomes or domains did not significantly change. The total number of reported core outcomes in the generic domain met statistical significance (P = 0.012). Additionally, multivariable analyses supported the above finding in the generic domain. Studies conducting prospective registration may have more comprehensive research designs that took into account this COS when selecting study outcomes. In addition, the generic and metabolic domains showed a relatively strong correlation. In contrast, the pregnancy and metabolic domains were negatively correlated.

The results of this study showed that the uptake of some core domains and outcomes increased to a certain extent after the publication of the COS on PCOS. A comparison of Table 3 and Supplementary Table S4 shows that the percentage of derivable outcomes reported in the generic and reproductive domains decreased in the Post-Pub group compared with the Pre-Pub group (by 8.9% versus 6.8% and by 13.1% versus 7.8%, respectively). These results indicated that there are fewer ambiguous outcomes in these domains after the publication of this COS, and that this COS improved consistency in reporting of these outcomes. However, the percentage of total uptake of each outcome or domain was low. In addition, no significant improvement was observed in terms of the consistency. Several similar studies also found low uptake rates for other COSs (Araújo et al., 2015; Smith et al., 2015, 2019a,b; Boric et al., 2019; Farag et al., 2019; Krsticevic et al., 2020). Therefore, developers of COSs should further publicize these domains and outcomes to improve uptake.

The following reasons may be responsible for the unsatisfactory uptake of COSs. Firstly, researchers may have been unaware of the need to seek and use COSs at the stage of study design, or had a limited understanding of COSs. Therefore, more aggressive promotion of COSs may be needed (Kirkham et al., 2013; Krsticevic et al., 2020). Secondly, the COS on PCOS did not include recommendations for effective measurement tools, time frames and guidelines for defining outcomes; these aspects should be further investigated and updated (Palominos et al., 2012; Farag et al., 2019). In addition, the number of domains and outcomes recommended by this COS was high, as PCOS has a long disease cycle and involves a wide range of fields. However, no intervention could affect all domains of the PCOS. A survey showed that the inclusion of six domains in a COS is excessive (Boric et al., 2018). Patients have perceived the completion of these measures of outcomes as a burden, which led to a reluctance to implement them (Mulla et al., 2015). Finally, some studies emphasized the lack of representation of stakeholders in the development of COSs (Smith et al., 2019a; Hughes et al., 2021). In short, longer observation may be necessary to identify the effects of this COS (Al Wattar et al., 2021). In this study, the main difficulty encountered during the assessment of consistency was the lack of clear definition of outcomes in this COS and clinical trial registries, which complicated the matching of the outcomes. For example, it was difficult to determine whether insulin resistance was consistent with fasting blood glucose or fasting insulin in trial registries. Therefore, it is preferable to have a detailed definition of outcomes in accordance with standards recognized by academia.

In this study, psychological outcomes were rarely reported, and oncological and long-term outcomes were not reported (Dokras et al., 2018; Al Wattar et al., 2021; Kiconco et al., 2022a). The following reasons may have hindered researchers from selecting these outcomes and conducting the evidence synthesis. The standardized definition of eating disorders remained inconclusive; there is a paucity of scales specifically designed to assess depression, anxiety and eating disorders in PCOS populations (Al Wattar et al., 2021), and longitudinal studies are confronted with challenges in conducting long-term follow-up (Gibson-Helm et al., 2017; Kiconco et al., 2022a). Furthermore, taking into consideration the aforementioned outcomes, it was proved to result in a significant increase in healthcare-related economic burdens (Riestenberg et al., 2022). Hence these core outcomes remained highly significant and warranted attention. In addition to the international efforts made toward common PCOS features, Azziz et al. has continued to investigate long-term complications of PCOS, including psychiatric and neurological disorders and malignancy, and particularly explored the mechanisms by which the intrauterine environment and genetics impacted long-term offspring outcomes (Goodarzi et al., 2011; Azziz et al., 2016, 2019; Azziz, 2018; Joham et al., 2022). The academic community should continuously recommend the rarely evaluated outcomes (Teede et al., 2011; Dokras et al., 2018), facilitate the development of long-term prospective longitudinal studies (Kiconco et al., 2022a,b), and attract the attention of experts from diverse fields, including but not limited to gynecologists and endocrinologists. Developing the targeted outcome assessment scales and standardized outcome definitions may have been a significant promoting effect.

The correlation analysis showed that most of the domains were positively correlated with each other before and after the publication of the COS on PCOS. For example, researchers who reported the metabolic domain were more likely to focus on the generic domain. The pregnancy domain was negatively correlated with the metabolic domain, implying that researchers who paid attention to the pregnancy domain were less likely to focus on the metabolic domain. Furthermore, a certain competitive relationship may exist between these two domains. During pregnancy, some outcomes in the metabolic domain (e.g. hypertension and diabetes) were included in the pregnancy domain, which ultimately led to a negative correlation. However, PCOS is complex and shows discrepant subtypes and clinical manifestations (Teede et al., 2018; Palomba et al., 2021; Dapas and Dunaif, 2022). Therefore, concurrently focusing on the pregnancy domain and reporting the outcomes in metabolic domain may help in providing more comprehensive evidence or clues for the in-depth understanding of the mechanism of PCOS and more accurate treatment for PCOS.

The strength of this study is that we were able to obtain representative registration data from various clinical trial registries through the WHO ICRTP. A previous study showed that the uptake rate obtained for outcomes in trial registries can reasonably reflect the practical uptake of COSs in research publications (Kirkham et al., 2017). Moreover, the method based on registration information was also preferable to citation analysis (Barnes et al., 2017); it shortens the period required for conducting and publishing clinical trials, and offers first-hand information regarding the uptake of COSs.

Limitations

Firstly, some results in this study may have been influenced by the subjective judgment of the evaluators. However, the opinions of clinical experts in related fields were used to assess the consistency of outcomes, and independent assessments were used to enhance the objectivity of the assessment. In case, it was difficult to assess consistency with the COS on PCOS, the outcomes were categorized as derivable outcomes, and a sensitivity analysis was conducted. Secondly, this study evaluated registration information only years after the publication of this COS. This period may be insufficient to estimate the impact of this COS on researchers. Nonetheless, improvements in the uptake of certain outcomes and domains were observed, indicating that this COS plays a role in the consistency of outcome reporting.

Conclusion

Based on systematic assessment, the results of this study showed an increasing trend in the consistency between outcomes in PCOS trial registry entries and the COS on PCOS. However, the overall uptake of this COS was not ideal. The pertinent domains of PCOS deserve a more in-depth exploration by researchers. Moreover, developers of the COS on PCOS should clarify the specific definition of core outcomes, actively update this COS and promote its uptake in clinical trials. These measures may improve the comparability of research outcomes on PCOS to provide more meaningful clinical evidence for PCOS in the future.

Data availability

The data underlying this article are available from the corresponding author.

Authors’ roles

L.Z. conceived the original idea and study design. W.Q.L. and G.L.L. participated in the assessment of consistency, analysis and interpretation of data, manuscript drafting, and critical discussion and revision for important intellectual content. L.Z., H.B.C., and H.N.W. contributed to the resolving the disagreements, revising the manuscript, and critically discussing and approving the final version of the paper.

Funding

This work was funded by Beijing Municipal Health Science and Technology Achievements and Appropriate Technology Promotion Project (BHTPP2022069), and the special fund of Beijing Key Clinical Specialty Construction Project.

Conflict of interest

The authors have no conflicts of interest to declare.