Effect of an Oxytocin Protocol on Secondary Uterotonic Use in Women Undergoing Cesarean Delivery

Abstract

Purpose:

Protocol-driven oxytocin regimens can reduce oxytocin administration compared with a nonprotocol free-flow continuous infusion. The aim of this before-and-after study was to retrospectively compare secondary uterotonic use between a modified “rule of 3s” oxytocin protocol and a free-flow continuous oxytocin infusion after cesarean delivery.

Methods:

Women who had cesarean delivery between January 1, 2010, and December 31, 2013 (preprotocol), were compared with women who had cesarean delivery between January 1, 2015, and August 31, 2017 (postprotocol). The preprotocol group received free-flow oxytocin administration; the postprotocol group received oxytocin in a modified rule of 3s algorithm. Primary outcome was secondary uterotonic use. Secondary outcomes included blood transfusion, hemoglobin value less than 8 g/dL, and estimated blood loss.

Results:

In total, 4,010 cesarean deliveries were performed for 3,637 women (2,262 preprotocol and 1,748 postprotocol). The odds of receiving secondary uterotonic drugs were increased in the postprotocol group (odds ratio [OR] [95% CI], 1.33 [1.04 to 1.70]; P=.02). Women in the postprotocol group were less likely to receive a blood transfusion. Nevertheless, the 2 groups were similar for the composite end point of transfusion or hemoglobin less than 8 g/dL (OR [95% CI], 0.86 [0.66 to 1.11]; P=.25). The odds of an estimated blood loss greater than 1,000 mL were reduced postprotocol (OR [95% CI], 0.64 [0.50 to 0.84]; P=.001).

Conclusions:

Women in the modified rule of 3s oxytocin protocol group were more likely to receive a secondary uterotonic than women in the preprotocol group. Estimated blood loss and transfusion outcomes were similar.

Introduction

Postpartum hemorrhage (PPH) complicates 2.9% of all deliveries¹ and is the most common cause of maternal death worldwide.² Previous work has shown that PPH was associated with a 10% hospital readmission rate within 2 months of delivery and with 19.1% of all in-hospital deaths after delivery.^3,4 Possible causes of PPH include physical trauma, retained placental products, and coagulation defects; however, the most common cause of PPH is uterine atony, occurring in 79% of cases.^1,3

Conventional therapy for uterine atony after vaginal delivery or cesarean delivery (CD) involves uterine massage and uterotonic medications. Oxytocin is the most commonly used medication for the prevention and treatment of uterine atony. Second-line (or rescue) agents include methylergonovine, carboprost tromethamine, and misoprostol. The likelihood of oxytocin adverse events, such as severe hypotension, cardiac arrhythmias, and death,^5,6 may increase as the administered dose increases. Although the evidence continues to support oxytocin use prophylactically after CD,^7–10 the optimal administration and dosing regimen has yet to be determined.

Historically, oxytocin was administered immediately after birth during a CD by placing it into the intravenous fluid reservoir and administering it as a free-flow infusion, resulting in varying doses from patient to patient. This often resulted in inaccurately charted doses, and patients potentially received greater doses of oxytocin than necessary.

Subsequently, many institutions have developed standardized oxytocin protocols involving less oxytocin administration, with no demonstrated change in PPH incidence.^11,12 More specifically, Kovacheva et al¹³ looked at a so-called rule of 3s oxytocin protocol. This rule is a series of low-dose oxytocin boluses of 3 IU administered every 3 minutes, depending on obstetrician assessment of uterine tone. The rule of 3s was compared with a free-flow infusion of oxytocin of 30 IU in 500 mL 0.9% saline. The investigators determined that the rule of 3s resulted in patients receiving an overall lesser amount of postpartum oxytocin.

This rule is based on the work of Carvalho et al,¹⁴ Balki et al,¹⁵ and Butwick et al,¹⁶ favoring the minimum effective dose (ED₉₀) of oxytocin required to produce adequate uterine contraction in 90% of laboring women who subsequently undergo CD (2.99 IU) vs the ED₉₀ for nonlaboring patients (0.35 IU). However, the protocol evaluated by Kovacheva et al¹³ looked only at nonlaboring patients undergoing an elective CD. At our institution, we adopted a modified rule of 3s protocol in March 2014 for all patients having a CD.

We hypothesized that the modified rule of 3s protocol would decrease the use of secondary uterotonic agents and decrease the time to secondary uterotonic administration when it was needed. The aim of the present study was to retrospectively compare secondary uterotonic use between our modified rule of 3s oxytocin administration protocol and our previous free-flow continuous infusion of oxytocin to achieve adequate uterine tone after a CD for both nonlaboring and laboring women.

Methods

The Mayo Clinic Institutional Review Board approved this single-center, controlled, before-and-after study on October 2017. We compared a modified rule of 3s oxytocin protocol (called postprotocol) to our previous free-flow oxytocin administration practice (called preprotocol) of patients undergoing a CD. All pregnant women who had a CD between January 1, 2010, and December 31, 2013, and between January 1, 2015, and August 31, 2017, were queried after identification from the electronic health record database.

The preprotocol group had a CD in the earlier study period. The group received oxytocin (approximately 20 IU placed in 1 L) in a freely flowing infusion of 1 L lactated Ringer solution infused until completion and then oxytocin administered at a rate of 15 IU per hour over 2 hours in the postanesthesia care unit. The postprotocol group had a CD in the later period and received 1 to 3 separate 3-IU microboluses of oxytocin over 30 seconds at 3 minutes apart. When adequate uterine tone was achieved, an oxytocin infusion of 6 IU per hour was initiated for 5 hours. Thus, our oxytocin protocol represented a modified rule of 3s protocol, given that the original rule of 3s protocol used an oxytocin infusion of 3 IU per hour for 6 hours.

The modified rule of 3s protocol was implemented in March 2014. Therefore, all CDs after that year were excluded from the study to allow time for protocol implementation into our practice. Patients were excluded from the study if they denied research consent, had placenta accreta, or had an intrauterine fetal demise.

Patient data were collected from the electronic health record. Demographic and obstetric information included maternal age, gravidity, parity, body mass index, gestational age in weeks, and birth weight in grams. The collected data on maternal risk factors for postpartum hemorrhage included abnormal placentation, chorioamnionitis, multiple gestations, and polyhydramnios. Delivery information included magnesium administration, anesthesia type, trial of labor, and indications for CD.

Our primary outcome measure was secondary uterotonic use (methylergonovine, carboprost tromethamine, or misoprostol). Secondary outcomes included time to secondary uterotonic administration from time of delivery, additional hemostatic surgical interventions, packed red blood cell (pRBC) transfusion, recorded hemoglobin less than 8 g/dL, estimated blood loss (EBL) greater than 1,000 mL, and median EBL. EBL was obtained from operative notes. Data were entered in Research Electronic Data Capture software (version 9.1.15; Vanderbilt University).

Statistical Analyses

Maternal, fetal, and delivery characteristics were reported according to preprotocol and postprotocol groups as number and percentage for categorical variables and as median (IQR) for continuous variables. The variables were compared without accounting for multiple observations per patient. Categorical variables were compared with use of Pearson χ² tests, and continuous variables were compared with Kruskal-Wallis rank sum tests. Outcomes were summarized as median (IQR) or number (percent) as appropriate.

Outcomes were compared between preprotocol and postprotocol groups with use of multivariable generalized linear models, accounting for multiple observations per patient using generalized estimating equations with robust covariance estimates. Categorical outcomes were modeled on the log-odds scale; estimates for odds ratio (OR) with 95% CI are presented. To satisfy model assumptions, continuous outcomes were modeled after log transformation using linear regression; estimates for the multiplicative increase in geometric mean are presented with 95% CI. Minutes to secondary uterotonic administration result were compared in a subset of patients who required secondary uterotonic use.

To assess the outcome of measured hemoglobin less than 8 g/dL, patients with no hemoglobin measurements were categorized as having no measured hemoglobin level below 8 g/dL. Patients with missing EBL data were assumed to have an EBL less than 1,000 mL. The composite outcome of any blood product transfusion or measured hemoglobin less than 8 g/dL was defined and was compared between groups. We chose this composite outcome to identify patients with a substantial hemorrhage and not solely rely on blood transfusion, as the trigger point for transfusion may have varied over the study period. Interaction effects were assessed to determine whether the effect of oxytocin protocol differed according to the initial delivery plan. When interaction P values were not statistically significant, the final models were adjusted only for the main effect of the initial delivery plan. In a post hoc analysis, association between oxytocin protocol and the 3-level outcome for location of the secondary uterotonic use (i.e., intraoperative, postoperative, or not needed) was assessed with use of multinomial logistic regression.

All analyses were done with statistical software (SAS version 9.4; SAS Institute Inc). P values less than .05 were considered statistically significant. The reporting of this study is in compliance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement.

Results

During the study time frame, 4,393 CDs were identified. Of these, 383 deliveries were excluded for meeting 1 or more of the exclusion criteria. In total, 4,010 CDs (2,262 in the preprotocol group and 1,748 in the postprotocol group) were performed among 3,637 women, with 3,287 (90%), 327 (9%), and 23 (1%) women undergoing 1, 2, or 3 CDs, respectively (Figure).

Figure.

Patient Inclusion and Exclusion for the Present Study. IUFD indicates intrauterine fetal demise.

Maternal, fetal, and delivery characteristics were similar between the 2 groups, with some exceptions. Women in the postprotocol group were more likely to have the diagnosis of polyhydramnios (1% in preprotocol vs 5% in postprotocol; P<.001) (Table 1). They also were more likely to be older, to have risk factors for PPH, and to receive predelivery magnesium, and they were less likely to be allowed to labor. Women in the preprotocol and postprotocol groups had similar intraoperative EBL (median [IQR], 800 [800–900] mL postprotocol vs 800 [800–1,000] mL preprotocol). There was no meaningful interaction between being allowed to labor and being in the preprotocol or postprotocol group for any of the primary or secondary outcomes (P>.05 in all cases).

Table 1.

Maternal, Fetal, and Course of Delivery Characteristics^a

	Oxytocin administration groupb
Characteristic	Preprotocol (n=2,262)	Postprotocol (n=1,748)	P valuec
Maternal and fetal characteristics
Age (yr)	31 [27–34]	31 [28–35]	.004
Gravida	2 [1–3]	2 [1–3]	.77
Parity	1 [0–1]	1 [0–2]	.86
BMI (n=2,158 and 1,743)	27.9 [23.8–33.2]	28.0 [24.0–33.4]	.14
Gestational age (wk)	39.0 [37.0–39.5]	39.0 [37.0–39.4]	.13
Birth weight (g) (n=2,255 and 1,747)	3,320 [2,780–3,730]	3,300 [2,760–3,710]	.25
Maternal risk factor
Any risk factord	630 (28)	585 (33)	<.001
Abnormal placentation	303 (13)	269 (15)	.07
Chorioamnionitis	249 (11)	223 (13)	.09
Multiple gestation	136 (6)	101 (6)	.75
Polyhydramnios	21 (1)	93 (5)	<.001
Course of delivery
Oxytocin	392 (17)	281 (16)	.29
Magnesium	110 (5)	124 (7)	.003
Anesthesia plan			.16
Neuraxial	2,055 (91)	1,610 (92)
General	207 (9)	138 (8)
Trial of labor			.04
No	1,111 (49)	917 (52)
Yes	1,151 (51)	831 (48)
Indication for CD			.56
Elective	1,367 (60)	1,085 (62)
Failure of labor	384 (17)	288 (16)
Emergency	511 (23)	375 (21)

Abbreviations: BMI, body mass index; CD, cesarean delivery.

^aIn total, 3,637 women underwent 4,010 deliveries from January 1, 2010, through August 31, 2017, with 3,287 (90%), 327 (9%), and 23 (1%) women undergoing 1, 2, or 3 deliveries, respectively.

^bContinuous variables are presented as median [IQR] and compared with Kruskal-Wallis rank sum test. Categorical variables are presented as number (percent) and compared with Pearson χ² test. When not all data were available, numbers of observations with available data are presented.

^cP values do not account for multiple observations per patient.

^dPresence of 1 or more of the listed maternal risk factors.

In adjusted analysis, the odds of a secondary uterotonic administration within the first 24 hours after delivery were greater in the postprotocol group (OR [95% CI], 1.33 [1.04 to 1.70]; P=.02) (Table 2). In addition, the time from delivery to secondary uterotonic use was greater in the postprotocol group (median, 9 minutes preprotocol vs 33 minutes postprotocol; estimated multiplicative increase in geometric mean [95% CI], 2.74 [2.02 to 3.74]; P<.001). The odds of intraoperative EBL exceeding 1,000 mL were reduced in the postprotocol compared to the preprotocol group (OR [95% CI], 0.64 [0.50 to 0.84]; P=.001). Intraoperative EBL was less in the postprotocol group (estimated multiplicative increase in geometric mean [95% CI], 0.97 [0.96 to 1.00]; P=.01). The odds of receiving a pRBC transfusion during the first 24 hours after delivery were also reduced in the postprotocol group (OR [95% CI], 0.41 [0.26 to 0.64]; P<.001). However, the groups were similar in comparison of composite end point of transfusion or measured hemoglobin less than 8 g/dL (OR [95% CI], 0.86 [0.66 to 1.11]; P=.25).

Table 2.

Univariable Comparison of Outcomes Between Time Periods^a

	Oxytocin administration group
Outcome	Preprotocol (n=2,262)	Postprotocol (n=1,748)	Estimate,b OR (95% CI)	P value
Any secondary uterotonic use	141 (6.2)	139 (8.0)	1.33 (1.04 to 1.70)	.02
Time to secondary uterotonic (min) (n=141 and 138)c	9 [5–44]	33 [17–120]	2.74 (2.02 to 3.74)	<.001
Location of initial secondary uterotonic used
No secondary uterotonic	2,121 (93.8)	1,609 (92.0)	Referent	NA
Operating room	114 (5.0)	81 (4.6)	0.97 (0.72 to 1.30)	.82
Postoperative	27 (1.2)	58 (3.3)	2.84 (1.78 to 4.51)	<.001
EBL >1,000 mLe	184 (8.1)	92 (5.3)	0.64 (0.50 to 0.84)	.001
EBL (mL) (n=2,250 and 1,746)c	800 [800–1,000]	800 [800–900]	0.97 (0.96 to 1.00)	.01
RBC transfusion	79 (3.5)	25 (1.4)	0.41 (0.26 to 0.64)	<.001
Measured Hb <8 g/dLf	123 (5.4)	94 (5.4)	1.01 (0.76 to 1.34)	.94
Transfusion or Hb <8 g/dLf	154 (6.8)	101 (5.8)	0.86 (0.66 to 1.11)	.25
Any obstetric intervention	227 (10.0)	182 (10.4)	1.01 (0.82 to 1.25)	.92

Abbreviations: EBL, estimated blood loss; Hb, hemoglobin; NA, not applicable; OR, odds ratio; RBC, red blood cell.

^aData are presented as number (percent) for categorical variables and median [IQR] for continuous variables. Estimates and P values are from multivariable linear regression with generalized estimating equations to account for multiple observations.

^bEstimates represent the increased odds of the given event associated with oxytocin protocol. When not all data were available, numbers of observations with available data are given. Models are adjusted for initial delivery plan (vaginal vs cesarean).

^cData were modeled on log scale. Estimates represent the multiplicative increase in geometric mean associated with oxytocin protocol.

^dPost hoc analysis. Estimates correspond to the multiplicative increase in odds associated with oxytocin protocol for receipt of secondary uterotonic agent in the given location compared with no secondary uterotonic use.

^ePatients with missing values (n=14) were assumed to have EBL <1,000 mL.

^fThe majority of patients did not have a measured Hb (n=1,536, 68% preprotocol; n=1,100, 63% postprotocol). Patients with no Hb measurement were categorized as having no measured Hb <8 g/dL.

A post hoc analysis looked at the 3-level outcome of timing of initial secondary uterotonic use (i.e., no use, initial use intraoperatively, and initial use postoperatively). In the postprotocol group, the odds of a secondary uterotonic use intraoperatively versus no use did not differ from the preprotocol group (5.0% preprotocol vs 4.6% postprotocol; OR [95% CI], 0.97 [0.72 to 1.30]; P=.82). In the postprotocol group, among patients receiving boluses of 3, 6, or 9 IU oxytocin, 0.2%, 2.7%, and 18.0% of patients, respectively, required a secondary uterotonic in the operating room; the relatively low rates among patients receiving 3 and 6 IU reflect the somewhat rare circumstances where clinicians opted for a secondary uterotonic before completion of the maximal dose of oxytocin. However, the odds of an initial secondary uterotonic use postoperatively versus no use increased compared with the preprotocol group (1.2% vs 3.3%; OR [95% CI], 2.84 [1.78 to 4.51]; P<.001). Among postprotocol patients who did not receive a secondary uterotonic in the operating room, 2.2%, 3.2%, and 6.9% of patients receiving oxytocin boluses of 3, 6, or 9 IU, respectively, received a secondary uterotonic postoperatively.

Discussion

In this retrospective before-and-after study, we compared a modified rule of 3s oxytocin protocol (postprotocol) to our previous free-flow oxytocin administration practice (preprotocol) in women undergoing CD. We concluded that the odds of receiving a secondary uterotonic agent were increased in the postprotocol group. However, intraoperative bleeding outcomes in the first 24 hours after delivery did not increase, and outcomes did not differ according to whether patients were allowed to labor before CD.

Contrary to our hypothesis, overall, a small and statistically significant increase was observed in total incidence of secondary uterotonic use (6.2% vs 8.0%, P=.02) in the postprotocol group and a longer median time to the first dose (9 minutes vs 33 minutes, P<.001).

The overall results may be suggestive that the rule of 3s oxytocin dosing was not as effective at maintaining uterine tone as the free-flow infusion. On further evaluation of the timing of secondary uterotonic use, we determined that 1.2% of the preprotocol group versus 3.3% of the postprotocol group received a secondary uterotonic agent for the first time in the postoperative period (P<.001). This evaluation suggests that perhaps more patients required secondary uterotonic use after adequate uterine tone had been established intraoperatively in the rule of 3s group than with the free-flow infusion group. Possibly, this was due to the lower maintenance oxytocin infusion (6 IU per hour vs 15 IU per hour) in the rule of 3s group. A systematic review found inconclusive results on the need for additional uterotonic use and mean blood loss when infusion-only regimens were compared with bolus-plus-infusion regimens.¹⁷

Practice changes over time could have caused greater uterotonic administration after the patient left the operating room. One such change was the 2015 publication of the National Partnership for Maternal Safety bundle, which encouraged several actions to respond to hemorrhage, including the ubiquitous use of hemorrhage carts that made uterotonics more readily available and likely more used.¹⁸ However, another compelling theory for this observation is a wearing-off effect of oxytocin when it is administered through the rule of 3s protocol. The preprotocol group received a minimum of 20 IU of oxytocin slowly administered during the intraoperative phase and 30 IU in the first 2 hours postoperatively. By comparison, the postprotocol group received a more quickly administered 3 to 9 IU intraoperatively and an additional 30 IU over the first 5 hours after the initial oxytocin administration. It is possible that the more rapid rate of oxytocin infusion in the first 2 postoperative hours (15 IU per hour preprotocol vs 6 IU per hour postprotocol) more effectively avoided the need for additional uterotonics after the patient left the operating room.

A limitation of the present study is that the free-flow oxytocin practice was not made into a protocol, and we therefore believe the amount of oxytocin documented in the preprotocol group is less reliable. For example, the custom of some clinicians in the preprotocol practice could have been to add more oxytocin to the free-flowing infusion bag, and this addition may have been poorly documented.

The secondary outcome of blood loss was assessed through 2 measures—total EBL and occurrence of any blood product transfusion. Total EBL was less in the postprotocol group, as was the incidence of EBL greater than 1,000 mL. An EBL cutoff that exceeds 1,000 mL was used conservatively to avoid the risk of overestimating the PPH incidence, as opposed to the PPH definition provided by the American College of Obstetricians and Gynecologists of 1,000 mL or more.¹⁹

Use of EBL instead of quantitative blood loss (QBL) is a limitation of this study, but our institution had not yet implemented QBL during the study time frame. Furthermore, we saw a decrease in blood product transfusion in the postprotocol group (3.5% vs 1.4%, P<.001). Yet, around the same time as our change in oxytocin protocol, an institutional change also occurred with the implementation of stricter criteria for blood transfusions. Before this change, most patients did not receive a transfusion until their hemoglobin was less than 8 g/dL; after the change, many patients did not have a transfusion until hemoglobin was less than 7 g/dL. To account for this institutional change, we included the outcomes of incidence of any measured hemoglobin less than 8 g/dL and the incidence of a combined end point, defined as either blood transfusion or a measured hemoglobin less than 8 g/dL. Neither of these outcomes differed significantly between the study groups. Considering everything, there was no evidence to indicate that the postprotocol group experienced worse bleeding outcomes compared to the preprotocol group, and there was some evidence to indicate a reduction in the proportion of patients experiencing elevated EBL.

Previous work has shown that women exposed to more oxytocin predelivery were significantly more likely to have severe PPH²⁰ and required higher oxytocin infusion rates to maintain adequate uterine tone postpartum.²¹ We assessed this effect by comparing women who were allowed to labor before their CD with those who were not in labor before their CD, and we found no statistically significant interactions. This result was unexpected because we anticipated a difference in outcomes if women had labored before their CD. However, no difference was observed between the preprotocol and postprotocol groups among women exposed to oxytocin before CD.

The present study has the natural limitations of a retrospective study, including health charting error, uncertainty about why and when certain treatments were or were not administered, and inability to have a guaranteed standard treatment algorithm for each patient; in addition, aspects of management of PPH prevention, diagnosis, and treatment may have evolved during this period. Most notably, this study is limited by the subjective nature of uterine tone assessment, use of EBL rather than QBL, and the clinical threshold of each physician of when to administer secondary uterotonic agents and blood products. We attempted to account for these limitations by collecting a sample that spanned nearly 7 years of practice and 4,010 deliveries and a wide range of obstetricians and anesthesiologists and to select conservative outcome thresholds when able.

In conclusion, this study reports the comparison of desired clinical outcomes of a modified rule of 3s oxytocin administration protocol to a historical free-flow continuous infusion of oxytocin following CD for nonlaboring and laboring women. The proposed modified rule of 3s protocol led to an overall increase in secondary uterotonic use but no difference in intraoperative use. It also resulted in a decrease in intraoperative EBL, intraoperative PPH rates, and overall blood product transfusion requirements. Moreover, while not tracked in this study, an oxytocin protocol such as the rule of 3s may still provide benefit by reducing the incidence of oxytocin-related adverse effects (e.g., intraoperative hypotension). The observed outcomes are encouraging for use of an initial 3-IU microbolus of oxytocin to achieve initial adequate uterine tone without requiring a high total dose. However, as seen by the increased postoperative secondary uterotonic use, additional studies may be needed to evaluate whether the rule of 3s protocol adequately maintains uterine tone after it is achieved intraoperatively, and further prospective studies are needed to fully evaluate the hemodynamic and side-effect profile of this protocol.

Implication Statement.

This before-and-after study found that women who received a modified “rule of 3s” oxytocin protocol after cesarean delivery, compared to a free-flow continuous oxytocin infusion, were more likely to receive a secondary uterotonic administration; however, estimated blood loss and blood transfusions were reduced.

Abbreviations

CD

cesarean delivery

EBL

estimated blood loss

ED₉₀

minimum effective dose of oxytocin required to produce adequate uterine tone for 90% of patients

OR

odds ratio

PPH

postpartum hemorrhage

pRBC

packed red blood cell

QBL

quantitative blood loss

STROBE

Strengthening the Reporting of Observational Studies in Epidemiology

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.