Postpartum hemorrhage (PPH) is a leading cause of maternal morbidity and mortality worldwide. Its incidence has continued to rise in developed countries, including in Canada and the United States.1–3 The rise in PPH from 2003–2010 has been attributed largely to an increase in postpartum uterine atony despite the use of prophylactic agents such as oxytocin to improve uterine tone.3,4

Oxytocin is a neurohypophyseal hormone that induces uterine contractions. It is the mainstay uterotonic drug for PPH prophylaxis and treatment as well as for labour induction and augmentation.5,6 Use of prophylactic oxytocin during active management of the third stage of labour reduces the incidence of PPH by up to 60%.6,7 Conversely, the use of high-dose or prolonged infusions of oxytocin for labour augmentation is associated with an increased risk of PPH secondary to uterine atony.8,9 Exposure of myometrial oxytocin receptors (OTRs) to oxytocin during labour augmentation results in receptor desensitization, which then reduces the responsiveness of OTRs to subsequent oxytocin administered for the prevention of primary PPH.10–13

There is some evidence in the literature suggesting resensitization of OTRs following cessation of oxytocin stimulation in a previously desensitized myometrium; however, this has not been consistently demonstrated in vitro.14–16 Moreover, in the clinical setting, it is unclear if discontinuing oxytocin after labour augmentation will allow for the myometrium to respond effectively to subsequent oxytocin stimulation.

The primary objective of our study was to assess the association between the oxytocin recovery interval (defined as the time from the cessation of oxytocin infusion during labour to delivery) and postpartum blood loss in women undergoing a Cesarean delivery (CD) for labour arrest after oxytocin-augmented labour. We hypothesized that an increase in the oxytocin recovery interval would reduce postpartum blood loss in these patients, possibly as a function of the recovery of myometrial OTRs.

Methods

Study design and population

After obtaining Research Ethics Board approval (REB 15-0181-C; August 10, 2015) at Mount Sinai Hospital (Toronto, Ontario, Canada), we conducted this retrospective chart review spanning July 1, 2013 to July 19, 2015. Patients who received oxytocin for labour augmentation and underwent CD for labour arrest under epidural anesthesia were identified by hospital Health Record Services. The charts were then screened to include patients with a gestational age of 37-41 weeks who received at least two hours of oxytocin infusion for labour augmentation. The two-hour threshold was based on prior studies indicating that this length of exposure induced OTR desensitization.10,14,15 Exclusion criteria included patients with PPH causes other than atony (such as genital tract or uterine trauma, coagulopathies and retained placental products), those with missing information on the outcome variables, or those who received general anesthesia. Both electronic and paper records were reviewed.

The data extracted from the charts included patient demographics, labour and delivery characteristics, oxytocin recovery interval, and known PPH risk factors8,9,16,17 (such as preeclampsia, chorioamnionitis, morbid obesity, macrosomia, multiple pregnancy, polyhydramnios, oxytocin induction, and oxytocin augmentation). The oxytocin protocol for labour augmentation at our institution consists of an initial rate of 2 mU·min−1 and an increase of 2 mU·min−1 approximately every 30 min as required to a maximum of 32 mU·min−1. The attending nurses document the rate and timing of oxytocin administration in the electronic recording system in real time. The oxytocin infusion is typically stopped when the decision is made to proceed with CD after failure to progress in labour. Fluid administration during labour is standardized at the rate of 125 mL·hr−1 iv via infusion pump.

At our hospital, anesthesia for CD is normally achieved by a top-up via the epidural catheter using 2% lidocaine with 1:200,000 epinephrine along with fentanyl 50 µg and epimorphine 2.5 mg. Immediately following delivery, all patients receive oxytocin as part of the active management of the third stage of labour for the prevention of PPH. Oxytocin 20 U·L−1 is administered as an initial rapid infusion of 150-250 mL (i.e., 3-5 units), followed by 150 mL·hr−1 for approximately six hours. Patients with uterine atony are managed with additional bolus doses of oxytocin, or an increase in its concentration (40 U·L−1), prior to receiving other uterotonic medications such as ergometrine, carboprost, and misoprostol.

Study outcomes

The primary outcome of the study was the amount of postpartum blood loss, expressed as the estimated blood loss (EBL). The EBL was determined using the hematocrit variation method, which assesses the change between the hematocrit level measured on admission and the level measured 24 hr after the delivery (approximately 48 hr apart) using the following formula: estimated blood volume x (preoperative hematocrit – postoperative hematocrit / preoperative hematocrit), where the estimated blood volume is the patient’s weight (kg) x 85 (mL·kg−1).18–21

Additional interventions used to manage blood loss in the postpartum period were analyzed as the secondary outcomes. These interventions included administration of additional uterotonic drugs (bolus doses of oxytocin, ergometrine, and/or prostaglandins) or medications (tranexamic acid and calcium), transfusion of packed red blood cells (PRBCs) or other blood products, and surgical maneuvers (B-Lynch uterine compression sutures,22 uterine massage, Bakri balloon,23 hysterectomy, and uterine arterial clamping, ligation, and embolization).

Sample size estimation and statistical analysis

The sample size calculation was based on achieving a power of 80% and a global alpha of 0.05 for the multiple linear regression model. For each independent predictor variable in the regression model, it is recommended to have an absolute minimum of ten or up to 30 participants per predictor to ensure better power for detecting small differences.24 With nine independent variables in the regression model for this study, the selected sample size of 490 ensured that there was sufficient power for the study. The independent variables included in the analysis were: oxytocin recovery interval, preeclampsia, chorioamnionitis (confirmed by pathology), morbid obesity (body mass index [BMI] at delivery ≥ 40 kg·m−2), macrosomia (birth weight > 4,500 g), multiple gestation, polyhydramnios, oxytocin induction, and the dose/duration of oxytocin for augmentation.

A multiple linear regression analysis was performed to assess the association between the oxytocin recovery interval and EBL. A multiple logistic regression model assessed the association between oxytocin recovery interval and the secondary outcome (i.e., need for additional interventions to manage postpartum blood loss). The two statistical models incorporated potential risk factors for PPH as covariates to account for possible confounding variables. The covariates were tested for collinearity prior to building a regression model in order to prevent overadjustment of the data. Since we found a high correlation between the total dose and duration of oxytocin for labour augmentation, only the total amount of oxytocin infused was included in the statistical models. Data were presented as mean and (standard deviation [SD]), median with range or [interquartile range (IQR)], and frequencies, as appropriate. We used SAS® v9.3 (Cary, NC, USA) for the statistical analyses.

Results

The medical records of 490 women who received oxytocin for labour augmentation and underwent CD for labour arrest were reviewed. The process used to select the study population is described in Fig. 1. Patient demographics and labour and delivery characteristics are listed in Table 1. Cesarean delivery was performed for failure of dilation in 309 (63%) cases, failure of descent in 106 (22%) cases, and failure of both dilation and descent in 75 (15%) cases. Potential risk factors for PPH are shown in Table 2.

Fig. 1
figure 1

Selection of the study population (EBL = estimated blood loss)

Full size image
Table 1 Demographics and labour and delivery characteristics of the study population
Full size table
Table 2 Risk factors for postpartum hemorrhage8,9,16,17
Full size table

The mean (SD) oxytocin recovery interval was 99 (65) min, and the mean (SD) EBL in the study population was 1,341 (577) mL. Eighty (16%) patients required at least one additional intervention to manage their postpartum blood loss (Table 3). Of the 75 (15%) patients receiving additional uterotonics or medications to manage uterine atony and blood loss, 17 received only additional oxytocin, whereas 43 patients received ergometrine, 28 received carboprost, 23 received misoprostol, and two received tranexamic acid. The remaining treatments, in descending order of frequency, were uterine massage, B-Lynch suture, transfusions of PRBCs or other blood products, and Bakri balloon (Table 3). None of the patients required a hysterectomy or uterine artery clamping, ligation, or embolization.

Table 3 Clinical interventions to manage postpartum blood loss
Full size table

After controlling for PPH risk factors, there was a significant inverse correlation between the oxytocin recovery interval and EBL (Fig. 2). Every ten-minute increase in the recovery period was associated with a 10-mL decrease in EBL (95% CI, −18 to −3; P = 0.009) (Table 4). Thus, in parturients waiting the average oxytocin recovery interval of 99 min, the expected reduction in EBL would be 102 mL. The mean (SD) EBL in women with a recovery interval < one hour was higher than those with a recovery interval ≥ one hour [1,438 (586) mL vs 1,298 (569) mL, respectively; mean difference, 140 mL; 95% CI, 29 to 248; P = 0.01]. The length of the recovery period was not associated with the need for additional interventions to manage postpartum blood loss (odds ratio, 0.997; 95% CI, 0.959 to 1.036; P = 0.87) (Table 4).

Fig. 2
figure 2

Correlation between oxytocin recovery interval and estimated blood loss in patients undergoing Cesarean delivery for labour arrest. Line of best fit showing the correlation in women with body mass index ≥ 40 kg·m−2 (black line) and < 40 kg·m−2 (grey line)

Full size image
Table 4 Multiple linear and logistic regression models using EBL and need for additional interventions to manage postpartum blood loss, respectively, as the outcomes
Full size table

Approximately 210 (43%) patients had one or more PPH risk factors. Except for morbid obesity, other risk factors were not found to have a notable effect on the amount of EBL or on the need for additional interventions. Patients who were considered morbidly obese (i.e., BMI ≥ 40 kg·m−2) had a significantly higher mean (SD) EBL than those with a BMI < 40 kg·m−2 [1,871 (882) mL vs 1,299 (525) mL, respectively; mean difference, 572 mL; 95% CI, 382 to 762; P < 0.001]. Nevertheless, morbid obesity was not associated with a significant increase in the need for additional interventions (odds ratio, 1.33; 95% CI, 0.56 to 3.2; P = 0.51).

We also tested an interaction between oxytocin recovery interval and EBL in morbidly obese women and found that an interaction exists between the two variables (P value for the interaction = 0.004). The effect of the oxytocin recovery interval on EBL was found to be more profound in morbidly obese women than in non-obese women. For morbidly obese women, every ten-minute increase in the recovery period was associated with a 45-mL decrease in EBL (95% CI, −71 to −20; P < 0.001). For women with a BMI < 40 kg·m−2, although statistically not significant, the inverse association between the oxytocin recovery interval and EBL was still present; every ten-minute increase in the recovery period was associated with a 7-mL decrease in EBL (95% CI, −15 to 1; P = 0.10).

Oxytocin was used in 138 (28%) patients to induce labour. All patients received oxytocin to augment labour. The mean (SD) duration and amount of oxytocin used for labour augmentation were 619 (355) min and 6,447 (6868) mU, respectively. The maximum oxytocin infusion rate during labour augmentation was 13.9 (7.9) mU·min−1. Sixteen (3.3%) women reached the maximum oxytocin infusion rate of 32 mU·min−1 for a mean (SD) duration of 197 (237) min. Both the total amount and duration of oxytocin infusion during labour augmentation were associated with the need for additional interventions (P ≤ 0.01); however, they were not associated with an increase in EBL. For every 100-mU increase in the total amount of oxytocin infused, the odds of requiring additional interventions for control of bleeding increased by 1.004 (95% CI, 1.001 to 1.008; P = 0.005). Since the amount and duration of oxytocin infusion were highly correlated (Spearman correlation coefficient = 0.92; P < 0.001), only the total amount of oxytocin infused was included in the two statistical models. The results were similar when the duration of oxytocin infusion was used for analysis.

Twenty-three (4.7%) patients required additional interventions beyond uterotonics (blood transfusion, uterine massage, B-Lynch suture, and/or Bakri balloon). These patients had a greater mean EBL of 578 mL (95% CI, 342 to 815; P < 0.001) than those who did not receive additional uterotonics. These patients also received a greater mean dose of oxytocin during labour augmentation by 3,505 mU (95% CI, 637 to 6,373; P = 0.02) and for a longer duration by 178 min (95% CI, 30 to 326; P = 0.02). No significant difference was noted in the recovery interval or BMI between the two groups.

Discussion

Our study showed that an increase in the oxytocin recovery interval is associated with a reduction in postpartum blood loss in parturients undergoing a CD for labour arrest following oxytocin augmented labour. Amongst the presumed risk factors for PPH, we found morbid obesity to be associated with an increase in blood loss at delivery. Moreover, the total amount and duration of oxytocin administered for labour augmentation were associated with the higher use of additional interventions to control postpartum bleeding.

At the study institution, oxytocin is typically discontinued after a CD is indicated for labour arrest. There is variability in the duration of the oxytocin-free period prior to the CD depending on the availability of an operating room and personnel and prioritization of cases. These conditions provided us with an opportunity to investigate the clinical implications of a varying oxytocin-free period following augmented labour. We included women who received oxytocin for labour augmentation for more than two hours to select those who received sufficient oxytocin to induce OTR desensitization. The length of this inclusion criterion was based on studies of pregnant human myometrium in vitro showing that two hours of exposure to oxytocin resulted in a substantial decline in subsequent oxytocin-induced contractions.10,14 We also based our parameter on a clinical study in which higher doses of oxytocin were required to achieve adequate uterine tone in women exposed to oxytocin.15 Clinically, OTR desensitization from augmented labour is associated with an increased risk of PPH depending on the dose and duration of oxytocin administered during labour.8–10,13,16,17,24 Although we did not observe more blood loss with increasing oxytocin administration during labour, an increased use of additional interventions suggests that there was poor uterine tone at delivery.

Various studies in the literature have described OTR desensitization10–13; however, little is known about the potential implications of OTR resensitization. Cell-based studies have shown that OTR resensitization following oxytocin pretreatment occurs in a time-dependent manner following discontinuation of oxytocin. An oxytocin recovery interval of four hours markedly enhanced oxytocin-OTR binding as compared with a 45-min duration,25 while an increase in the interval from 5-15 min led to an increase in effective oxytocin responsiveness.26 A more recent study found that increasing the recovery period from 30-90 min did not improve the contractile response to oxytocin in myometrial samples in vitro.27 The discrepancies between these results may be a consequence of differing cell types, culture conditions, and oxytocin pretreatment regimens. Furthermore, the recycling or synthesis of OTRs may depend, at least in part, on cell signals mediated by in vivo factors not reproduced in the in vitro studies. These issues highlight the difficulty of studying OTR resensitization in vitro and extrapolating results to the clinical setting.

Following OTR desensitization during labour augmentation, the oxytocin recovery interval in our study represented the anticipated time interval during which resensitization can occur. The time-dependent reduction in postpartum blood loss may have resulted from OTR resensitization, which restores the myometrial responsiveness to oxytocin administered for PPH prophylaxis. These clinical findings parallel the time-dependent restoration of OTR responsiveness to subsequent oxytocin challenges following oxytocin pretreatment in in vitro studies.25,26

In addition to the oxytocin recovery interval, morbid obesity was associated with EBL. The role of obesity as a risk factor for PPH remains debated.28–30 The contrasting results may be due to differences in the covariates, such as ethnicity, mode of delivery and oxytocin exposure in the statistical models, or the outcomes used to define PPH. We observed that the blood loss in morbidly obese women was about 600 mL greater than that in non-morbidly obese mothers; however, this was not associated with the need for more interventions to manage the blood loss. Interestingly, studies finding an association between PPH and obesity used blood loss (≥ 1,000 mL) to define PPH,28,29 whereas the study with an opposing finding used PPH requiring blood transfusion as their outcome.30 Obese women have different signalling pathways modulating OTR function and ligand binding as well as alterations in OTR membrane viscosity and fluidity due to high cholesterol levels. Hence, they are likely to have impaired oxytocin-induced myometrial contractility, predisposing them to dysfunctional labour and PPH.31

Other PPH risk factors in our study were not found to be associated with the study outcomes. Prior studies examining the association of chorioamnionitis with PPH have been conflicting.8,17,32 We did not find an association despite the fact that our cases were confirmed by pathology reports. The lack of an association with some other previously identified risk factors may be the result of the low number of cases in our study with macrosomia, preeclampsia, multiple pregnancy, and polyhydramnios. Consideration of other factors contributing to blood loss is also needed. Patients who waited a shorter time for their CD may have been at higher risk of adverse obstetric outcomes that increase the likelihood of PPH. To this end, multivariable models and selection criteria were used.

There were some limitations to our study. For example, we cannot eliminate residual confounders completely, particularly those contributing to blood loss to a subclinical extent without any notable impact on PPH risk. These are not clearly defined in the literature but may convolute the association examined, since we measured the quantity of blood loss rather than the occurrence of PPH. Nevertheless, we carefully controlled for PPH risk factors, including the use of oxytocin as a continuous variable to account for its contribution to blood loss more accurately. We excluded potential confounders such as early or late gestation, the use of general anesthesia, and uterine rupture. Hence, our multivariable model and selection criteria gave us the incentive to examine the association between oxytocin recovery interval and blood loss.

The hematocrit variation method to estimate blood loss was used because it provides more accurate results than visual estimations. Specifically, visual estimations have been shown to be subjective and dependent on the experience of the observer, and they tend to underestimate actual blood loss.21,33–35 Nevertheless, the hematocrit variation method is not without its limitations. It is possible that estimating blood loss using this method may lead to overestimation in morbidly obese women. Although it is known that indexed blood volume decreases in a non-linear manner with increasing body weight,36 it is unclear how this relates to the morbidly obese and pregnant population. Secondly, following the decrease in estrogen in the immediate postpartum period, there is diuresis of fluids accumulated during pregnancy and a relative elevation of the hematocrit.37 Since the EBL relies on the difference between pre- and post-CD hematocrit levels, a postpartum elevation in hematocrit from this physiologic phenomenon can artificially lower the EBL. Additional fluids can also alter patients’ hematocrit level. A final limitation to our study is that, although the quantified reduction in blood loss was statistically significant, the clinical consequences of this blood loss were uncertain based on our secondary outcomes (i.e., little need for blood transfusion or additional uterotonics).

In summary, the study showed that an increase in the oxytocin recovery interval is associated with a reduction in postpartum blood loss. This suggests that, once a CD is considered for labour arrest, cessation of oxytocin for augmented labour may reduce the amount of postpartum blood loss. The extent of the reduction in blood loss found in this study likely does not warrant a mandatory recovery interval beyond what typically occurs. Nevertheless, particular attention must be given to morbidly obese women, as they are at a significantly higher risk of bleeding than non-obese parturients. Further investigations may be needed to assess the impact of the recovery interval on the need for additional interventions to manage blood loss as well as the implications of a recovery interval on patients at higher risk for PPH, such as morbidly obese parturients.