Objective
We sought to identify risk factors for uterine atony or hemorrhage.
Study Design
We conducted a secondary analysis of a 3-arm double-blind randomized trial of different dose regimens of oxytocin to prevent uterine atony after vaginal delivery. The primary outcome was uterine atony or hemorrhage requiring treatment. In all, 21 potential risk factors were evaluated. Logistic regression was used to identify independent risk factors using 2 complementary predefined model selection strategies.
Results
Among 1798 women randomized to 10, 40, or 80 U of prophylactic oxytocin after vaginal delivery, treated uterine atony occurred in 7%. Hispanic (odds ratio [OR], 2.1; 95% confidence interval [CI], 1.3–3.4), non-Hispanic white (OR, 1.6; 95% CI, 1.0–2.5), preeclampsia (OR, 3.2; 95% CI, 2.0–4.9), and chorioamnionitis (OR, 2.8; 95% CI, 1.6–5.0) were consistent independent risk factors. Other risk factors based on the specified selection strategies were obesity, induction/augmentation of labor, twins, hydramnios, anemia, and arrest of descent. Amnioinfusion appeared to be protective against uterine atony (OR, 0.53; 95% CI, 0.29–0.98).
Conclusion
Independent risk factors for uterine atony requiring treatment include Hispanic and non-Hispanic white ethnicity, preeclampsia, and chorioamnionitis.
The incidence of postpartum hemorrhage in developed countries is increasing. In the United States, one estimate of the overall rate of postpartum hemorrhage increased approximately 26%, from 2.3% in 1994 to 2.9% in 2006. Uterine atony may specifically account for up to 80% of the cases of postpartum hemorrhage. Unlike other causes of obstetric hemorrhage such as placental abnormalities that may be detected prenatally, uterine atony is difficult to predict. Many risk factors for uterine atony and postpartum hemorrhage have been reported. The specific risk factors examined and the magnitude of risk attributable to each of them vary across reports. Therefore confounding may be a major reason for the discrepancies. Furthermore, reports of true risk factors may be missed in some studies because of limited power to demonstrate statistical significance.
A greater understanding of well-defined independent risk factors may improve our ability to determine which women may be at risk for postpartum hemorrhage. This is important since obstetrical hemorrhage, primarily postpartum, is a significant cause of maternal morbidity and mortality worldwide. Therefore the purpose of this study was to conduct a more comprehensive multivariable analysis to identify independent risk factors for uterine atony or postpartum hemorrhage. We applied multivariable statistical models that allowed us to identify probable independent risk factors that may be confirmed in larger studies.
Materials and Methods
We conducted a secondary analysis of a 3-arm double-blind randomized clinical trial of different doses of oxytocin. The primary aim of the trial was to evaluate higher doses of oxytocin compared to a standard low-dose oxytocin regimen used for prophylaxis among women undergoing vaginal delivery. Women were randomized to a 10- (standard), 40-, or 80-U dose regimen of oxytocin at vaginal delivery. Women were excluded if they were <24 weeks’ gestation, underwent cesarean delivery, had fetal demise, had pulmonary edema, or had coagulopathy or cardiomyopathy. Each regimen comprised the specified dose in 500 mL of a crystalloid solution administered rapidly over 1 hour after delivery of the placenta (ie, at a rate of 500 mL/h). The protocol was approved by the Institutional Review Board at the University of Alabama at Birmingham. All participating women granted informed consent. Information concerning patient demographic and clinical characteristics as well as outcomes of interest, including postpartum hemorrhage or atony requiring therapy, was abstracted by trained research nurses. For this secondary analysis, we retained the same primary outcome: uterine atony or hemorrhage requiring treatment. Treatment included the use any use of uterotonics, or the need for transfusion, balloon tamponade, surgery, or interventional radiology procedure for uterine or arterial embolization. Transfusion was based on the need for whole or packed red cells prior to discharge home from the hospital. The diagnosis of uterine atony was made based on the discretion of the treating obstetrical team.
The study variables or exposures of interest consisted of a large set of 21 potential risk factors (including demographic characteristics) for uterine atony/hemorrhage ( Table 1 ) identified from the published literature. The risk factors were defined accordingly: overweight was defined as a body mass index (BMI) of 25-29.9 kg/m 2 and obesity was defined as a BMI of ≥30 kg/m 2 . Ethnicity was self-reported as Hispanic, black, white, or other. Chorioamnionitis was defined as the presence of clinical signs (primarily intrapartum fever) leading to a clinical diagnosis and antibiotic treatment of chorioamnionitis. Anemia was determined by a hemoglobin of <9 g/dL. Hydramnios was defined as an amniotic fluid volume of >25 cm or a greatest vertical pocket of >8 cm. Protracted second stage of labor was >1 hour from complete cervical dilation to delivery if multiparous and >2 hours between complete cervical dilation and delivery if nulliparous. A protracted third stage was defined as being >30 minutes from the times of delivery of the infant to delivery of the placenta.
| Characteristic or factor | Prevalence, n (%) |
|---|---|
| Maternal age, a y | 24.1 ± 5.4 |
| Oxytocin dose, U | |
| 10 | 659 (37) |
| 40 | 481 (27) |
| 80 | 658 (37) |
| Nulliparity | 673 (37) |
| Hispanic | 327 (18) |
| White and other | 410 (23) |
| Black | 1061 (59) |
| Obese (BMI ≥30) | 1036 (58) |
| Overweight (BMI 25-30) | 523 (29) |
| Normal, underweight | 239 (13) |
| Augmentation | 955 (53) |
| Induction | 572 (32) |
| Spontaneous | 271 (15) |
| Preeclampsia/eclampsia | 218 (12) |
| MgSO4 use | 188 (10) |
| Twins | 13 (<1) |
| Chorioamnionitis | 122 (7) |
| Hydramnios | 43 (2) |
| Amnioinfusion | 299 (17) |
| Epidural anesthesia | 1504 (84) |
| Breast-feeding | 1000 (57) |
| Spontaneous ROM | 584 (33) |
| Prior cesarean | 86 (5) |
| Operative deliveries | 147 (8) |
| Anemia | 44 (2) |
| Protracted second stage | 108 (6) |
| Prolonged third stage | 35 (2) |
| GA at delivery, a wk | 38.8 ± 2.1 |
| Total birthweight, g | |
| <2500 | 190 (11) |
| 2500-3999 | 1499 (83) |
| ≥4000 | 109 (6) |
| GA at delivery, wk | |
| <37 | 223 (12) |
| 37-41 | 1395 (78) |
| ≥41 | 180 (10) |
Logistic regression analyses were conducted to individually evaluate each of the risk factors as predictors of the primary study composite. Risk factors identified as significant, either by statistical significance at a .05 level or by large effect (odds ratio [OR], >1.5 or <0.7) at the univariate level, were considered in multivariable logistic regression models. A parsimonious regression model of independent risk factors was derived using traditional backward model selection strategy whereby only factors significant at the .05 level at each stage were retained for further consideration. Risk factors were progressively eliminated from the model until a parsimonious model comprising only factors statistically satisfying the specific selection criteria was obtained.
To identify additional risk factors that may not be statistically significant within the limitations of our sample size, we utilized a previously described modified backward selection strategy that gave premium to the magnitude of risk as well as to statistical significance. A priori, factors associated with a minimally important change in the outcome (defined as OR, ≥1.5 or ≤0.67) or a P value < .05 were retained in the model. The minimally important difference was selected as we judged this level to be of public health significance. Variables with adjusted ORs (aORs) <1.5 and >0.67 were then progressively removed, starting with the variable with the highest P value until the final parsimonious models were obtained. Software (SAS, version 9.2; SAS Institute Inc, Cary, NC) was used for all statistical analyses.
Results
Our study sample included all 1798 women randomized and analyzed in the primary trial. In all, 658 women were randomized to 80 U of oxytocin, 481 women to 40 U (this arm was terminated at interim review), and 659 to 10 U of oxytocin. The distribution of the population according to characteristics under study is presented in Table 1 . Of note, the dose of prophylactic oxytocin did not influence the outcome in the primary trial. The study population consisted of women who were predominantly obese, black, underwent labor induction, and received an epidural.
The frequency of the primary outcome, treated uterine atony or hemorrhage, was 7% overall (118 women) and did not differ by study group. The prevalence of this outcome by categories of each of the 21 potential risk factors for uterine atony/hemorrhage and the corresponding unadjusted OR (95% confidence interval [CI]) are presented in Table 2 . BMI, race/ethnicity, labor induction, twins, preeclampsia, breast-feeding, anemia, protracted second stage, and chorioamnionitis were significantly associated with postpartum atony or hemorrhage in univariate analyses.
| Factor | Incidence of uterine atony, n (%) | No uterine atony, n (%) | OR (95% CI) |
|---|---|---|---|
| Oxytocin dose, U | |||
| 10 (n = 659) | 45 (7) | 614 (93) | Referent |
| 40 (n = 481) | 31 (6) | 450 (94) | 0.9 (0.6–1.5) |
| 80 (n = 658) | 42 (6) | 616 (94) | 0.9 (0.6–1.4) |
| Obese (BMI ≥30) (n = 1036) | 75 (7) | 961 (93) | 2.6 (1.2–5.7) |
| Overweight (BMI 25-30) (n = 523) | 36 (7) | 487 (93) | 2.5 (1.1–5.6) |
| Normal, underweight (n = 239) | 7 (3) | 232 (97) | Referent |
| Hispanic (n = 327) | 33 (10) | 294 (89) | 2.1 (1.3–3.3) |
| White and other (n = 410) | 31 (8) | 379 (92) | 1.5 (1.0–2.4) |
| Black (n = 1061) | 54 (5) | 1007 (95) | Referent |
| Augmentation (n = 955) | 52 (5) | 903 (95) | 1.2 (0.7–2.4) |
| Induction (n = 572) | 54 (9) | 518 (91) | 2.3 (1.2–4.3) |
| Spontaneous (n = 271) | 12 (4) | 259 (96) | Referent |
| Birthweight, g | |||
| <2500 (n = 190) | 10 (5) | 180 (95) | 0.8 (0.4–1.5) |
| 2500-3999 (n = 1499) | 100 (7) | 1399 (93) | Referent |
| ≥4000 (n = 109) | 8 (7) | 101 (93) | 1.1 (0.5–2.3) |
| Nulliparity (n = 673) | 54 (8) | 619 (92) | 1.4 (1.0–2.1) |
| Parity (n = 1125) | 64 (6) | 1061 (94) | |
| Preeclampsia (n = 218) | 31 (14) | 187 (86) | 2.8 (1.8–4.4) |
| No preeclampsia (n = 1580) | 87 (6) | 1493 (94) | |
| MgSO4 use (n = 188) | 27 (14) | 161 (86) | 2.8 (1.8–4.4) |
| No MgSO4 use (n = 1610) | 91 (6) | 1519 (94) | |
| Twins (n = 13) | 2 (15) | 11 (85) | 2.6 (0.6–12.0) |
| Singletons (n = 1785) | 116 (7) | 1669 (93) | |
| Chorioamnionitis (n = 122) | 17 (14) | 105 (86) | 2.5 (1.5–4.4) |
| No chorioamnionitis (n = 1676) | 101 (6) | 1575 (94) | |
| Hydramnios (n = 43) | 5 (12) | 38 (88) | 1.9 (0.7–5.0) |
| No hydramnios (n = 1755) | 113 (6) | 1642 (94) | |
| Amnioinfusion (n = 299) | 13 (4) | 286 (96) | 0.6 (0.3–1.1) |
| No amnioinfusion (n = 1499) | 105 (7) | 1394 (93) | |
| Epidural anesthesia (n = 1504) | 97 (6) | 1407 (94) | 0.9 (0.5–1.5) |
| No epidural (n = 294) | 21 (7) | 273 (93) | |
| Breast-feeding (n = 1000) | 76 (8) | 924 (92) | 1.5 (1.0–2.2) |
| No breast-feeding (n = 751) | 39 (5) | 712 (95) | |
| Spontaneous ROM (n = 584) | 36 (6) | 548 (94) | 0.9 (0.6–1.4) |
| Artificial ROM (n = 1212) | 82 (7) | 1130 (93) | |
| Prior cesarean (n = 86) | 7 (8) | 79 (92) | 1.3 (0.6–2.8) |
| No prior cesarean (n = 1712) | 111 (6) | 1601 (94) | |
| Operative deliveries (n = 147) | 12 (8) | 135 (92) | 1.3 (0.7–2.4) |
| Nonoperative delivery (n = 1651) | 106 (6) | 1545 (94) | |
| Anemia (hemoglobin <9 ) (n = 44) | 5 (11) | 39 (89) | 1.9 (0.7–4.8) |
| No anemia (n = 1754) | 113 (6) | 1641 (94) | |
| Protracted second stage (n = 108) | 14 (13) | 94 (87) | 2.3 (1.3–4.1) |
| No protracted second stage (n = 1690) | 104 (6) | 1586 (94) | |
| Prolonged third stage (n = 35) | 4 (11) | 31 (89) | 1.9 (0.6–5.4) |
| No prolonged third stage (n = 1763) | 114 (6) | 1649 (94) | |
| Maternal age, y | 24.7 ± 5.7 | 24.7 ± 5.3 | 1.0 (1.0–1.1) |
| GA at delivery, wk | 38.9 ± 2.1 | 38.9 ± 2.1 | 1.0 (0.9–1.1) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
