Objective
Because the diagnosis of postpartum hemorrhage (PPH) depends on the accoucheur’s subjective estimate of blood loss and varies according to mode of delivery, we examined temporal trends in severe PPH, defined as PPH plus receipt of a blood transfusion, hysterectomy, and/or surgical repair of the uterus.
Study Design
We analyzed 8.5 million hospital deliveries in the US Nationwide Inpatient Sample from 1999 to 2008 for temporal trends in, and risk factors for, severe PPH, based on International Classification of Diseases, 9th revision, clinical modification diagnosis and procedure codes. Sequential logistic regression models that account for the stratified random sampling design were used to assess the extent to which changes in risk factors explain the trend in severe PPH.
Results
Of the total 8,571,209 deliveries, 25,906 (3.0 per 1000) were complicated by severe PPH. The rate rose from 1.9 to 4.2 per 1000 from 1999 to 2008 ( P for yearly trend < .0001), with increases in severe atonic and nonatonic PPH, due especially to PPH with transfusion, but also PPH with hysterectomy. Significant risk factors included maternal age ≥35 years (adjusted odds ratio [aOR], 1.5; 95% confidence interval [CI], 1.5–1.6), multiple pregnancy (aOR, 2.8; 95% CI, 2.6–3.0), fibroids (aOR, 2.0; 95% CI, 1.8–2.2), preeclampsia (aOR, 3.1; 95% CI, 2.9–3.3), amnionitis (aOR, 2.9; 95% CI, 2.5–3.4), placenta previa or abruption (aOR, 7.0; 95% CI, 6.6–7.3), cervical laceration (aOR, 94.0; 95% CI, 87.3–101.2), uterine rupture (aOR, 11.6; 95% CI, 9.7–13.8), instrumental vaginal delivery (aOR, 1.5; 95% CI, 1.4–1.6), and cesarean delivery (aOR, 1.4; 95% CI, 1.3–1.5). Changes in risk factors, however, accounted for only 5.6% of the increase in severe PPH.
Conclusion
A doubling in incidence of severe PPH over 10 years was not explained by contemporaneous changes in studied risk factors.
Postpartum hemorrhage (PPH) is widely recognized as a major cause of maternal mortality and severe morbidity, particularly in low-income countries. Recent evidence from a number of developed countries indicates a rising incidence of PPH, especially atonic PPH. One of the problems with the diagnosis of PPH is its inherent subjectivity, which depends on the accoucheur’s estimate of blood loss. As a result, the clinical importance of the reported rise in total PPH could be questioned, especially because no temporal increase in maternal mortality due to PPH has been reported. Increase in blood transfusion for PPH has been observed in the US and Australia, but not in Canada or France. Joseph et al reported increases in hysterectomy with atonic PPH in Canada from 1991 to 2004. Few studies have had sufficient numbers of cases with severe morbidity to ensure adequate statistical power to detect a rise in such cases over time, and thus to study temporal trends and risk factors that could explain any observed trend in severe PPH.
To provide new insight into severe PPH, we analyzed a very large and nationally representative sample of hospital-based deliveries in the United States’ Nationwide Inpatient Sample, which includes 8.5 million births from 1999-2008. In this study, we examine the temporal trend in severe PPH, which we define as PPH accompanied by blood transfusion, hysterectomy, or surgical repair of the uterus, and attempt to explain any observed temporal trend based on contemporaneous trends in risk factors.
Materials and Methods
This population-based cohort study uses data from the Healthcare Cost and Utilisation Project Nationwide Inpatient Sample (NIS) from 1999 to 2008, inclusively. Hospital admissions from this database include approximately 20% of admissions to hospitals in the United States, including rural hospitals, urban teaching hospitals, and urban nonteaching hospitals. The database contains individual-level clinical diagnostic discharge codes classified according to the International Classification of Diseases, 9th revision , clinical modification (ICD-9-CM), ICD-9-CM procedure codes, hospital length of stay, and whether death occurred during the admission. The NIS database also includes a sampling weight that reflects the number of hospital discharges from each hospital unit, based on the stratified 20% random sampling design. The first author completed the HCUP Data Use Agreement Training Course (Certification Code HCUP-6HYR48K91). The Centers for Disease Control and Prevention (CDC) classified this research as not involving human subjects because the public use dataset does not include any personal identifying information, and no additional Institutional Review Board approval was sought.
We identified deliveries within the NIS database using any of the following ICD-9-CM codes: V27, 640-649 with fifth digit = 1 or 2, 650, or 651-676 with fifth digit = 1 or 2. To focus on cases of severe PPH, we developed 3 exhaustive and mutually exclusive subcategories of severe PPH: (1) PPH (ICD-9-CM diagnosis codes 666.0, 666.1, 666.2, and/or 666.3) plus blood transfusion alone (ICD-9-CM procedure codes 990, 990.3 and/or 990.4), (2) PPH plus hysterectomy (ICD-9-CM procedure codes 683.9, 684, and/or 689), with or without blood transfusion or uterine repair, and (3) PPH plus surgical repair of the uterus (ICD-9-CM procedure code 755), without hysterectomy and with or without blood transfusion. No change in these ICD-9-CM diagnosis or procedure codes occurred during the study period.
We also subdivided cases of severe PPH into 2 clinical subtypes: atonic PPH (code 666.1, immediate atonic PPH) and nonatonic PPH (codes 666.0, retained placenta; 666.2, delayed or secondary PPH; and/or 666.3, postpartum coagulation defects). To reduce the reported overcoding of atonic PPH and render this subdivision exhaustive and mutually exclusive, we classified cases of PPH as nonatonic if any of the 3 nonatonic codes were recorded, even if code 666.1 was also recorded. We also coded all cases of PPH with cervical laceration and/or uterine rupture as nonatonic, even if code 666.1 was recorded.
Statistical methods included overall descriptive statistics (in weighted %), highlighting the temporal trend in overall severe PPH, as well as the 3 severity subcategories. Risk factors for severe PPH were analyzed using cross-tabulations and logistic regression analysis to estimate crude and adjusted odds ratios (aORs) and their 95% confidence intervals (CIs). These regression analyses accounted for the sampling design (hospital discharge weights) in the model, so that the results can be considered representative for the United States. Covariates adjusted for in the multivariable logistic regression models included maternal age (≤19, 20-34, ≥35 years) and the following diagnoses and procedures identified by ICD-9-CM codes: elderly primigravidity, grand multiparity, medical and surgical induction of labor, previous cesarean delivery, presentation (cephalic vs noncephalic), placenta previa or abruption, hypertension (none, preeclampsia, eclampsia, or other [pregestational hypertension or gestational hypertension without proteinuria]), diabetes (gestational or pregestational), fetal macrosomia, fetal growth restriction, postdates pregnancy, prelabor rupture of membranes, cervical laceration, uterine rupture, fetal distress, dystocia, and year of delivery. The multivariable logistic regression analyses grouped the potential risk factors and covariates into temporal categories defined as those existing prior to the index pregnancy, those occurring during pregnancy but before labor and delivery, and those occurring during labor and delivery. This temporal grouping was used to ensure that appropriate adjustment for potential confounding variables did not modify (attenuate or amplify) their associations with severe PPH, which can occur when adjusting for causal intermediates. We next analyzed sequential logistic models, including a term for year, to examine the extent to which temporal changes in risk factors could explain (ie, reduce) the effect of the year term in the logistic models.
Additional outcomes studied in women experiencing severe PPH included stillbirth, maternal death, and prolonged (≥8 days) length of hospital stay. aORs for these outcomes controlled for the same covariates as for severe PPH.
All statistical analyses were carried out using SAS software, version 9.2 (SAS Institute, Cary, NC).
Results
Of the 8,571,209 deliveries recorded in the database, a total of 240,472 (2.8%) had the ICD-9-CM code for PPH, of which 180,763 (2.1% of total deliveries) were coded as atonic PPH and 59,709 (0.7%) were classified as nonatonic PPH. The total rate of PPH (ie, not restricted to severe PPH) rose from 2.7 to 3.0% from 1999 to 2005, then fell to 2.5% in 2007 before increasing again to 2.8% in 2008. The overall rate for atonic PPH rose slightly from 2.0% in 1999 to 2.3% in 2005 and 2006, then fell to 1.9% in 2007 before rising again to 2.1% in 2008. Overall rates for nonatonic PPH showed no trend over the study period.
A total of 25,906 (3.0 per 1,000 of all deliveries) of the total PPH cases met one or more of our criteria for severe PPH. For the 3 severity subcategories, the breakdown was as follows: 18,021 for PPH plus transfusion, 4124 for PPH plus hysterectomy, and 3761 for PPH plus surgical repair of the uterus. The overall weighted transfusion rate for the 8.5 million births was 0.6%: 0.4% among those without PPH and 8.8% among those with PPH. Figure 1 shows the yearly rate per 1000 deliveries for total severe PPH and for severe atonic and nonatonic PPH. Both of these clinical subtypes showed a temporal increase. As shown in Figure 2 for both subtypes, no appreciable increase was observed for PPH plus uterine repair, but the rate of PPH plus blood transfusion rose between 2- and 3-fold over the 10-year study period. The rate of PPH plus hysterectomy, albeit much lower, doubled over the same period.
Table 1 contains a description of the cohort according to the potential risk factors and covariates studied, grouped according to their temporal sequence, along with the number (and weighted rate per 1000) experiencing severe PPH (any clinical subtype or severity subcategory) for each risk factor and the crude and aORs with their 95% CIs. Substantial increases in risk were associated with older maternal age, grand multiparity, previous cesarean delivery, preeclampsia/eclampsia and other types of maternal hypertension, amnionitis, placenta previa or abruption, multiple pregnancy, fetal macrosomia, uterine fibroids, instrumental vaginal delivery, cervical laceration, and uterine rupture.
Characteristic | % of total cohort | Confirmed cases, n (rate per 1000) | Crude OR (95% CI) | Adjusted OR (95% CI) |
|---|---|---|---|---|
| BEFORE INDEX PREGNANCY | ||||
| Maternal age | ||||
| ≤19 | 10.7 | 2973 (3.2) | 1.2 (1.1–1.3) | 1.2 (1.2–1.3) |
| 20-34 | 75.0 | 17,375 (2.7) | 1.0 (Reference) | 1.0 (Reference) |
| ≥35 | 14.3 | 5134 (4.2) | 1.6 (1.5–1.6) | 1.5 (1.5–1.6) |
| Eldery primigravidity | 1.3 | 555 (5.1) | 1.7 (1.6–1.9) | 1.3 (1.2–1.4) |
| Grand multiparity | 0.5 | 216 (4.7) | 1.6 (1.4–1.8) | 1.4 (1.2–1.7) |
| Prior cesarean delivery | 13.8 | 4273 (3.6) | 1.3 (1.2–1.3) | 1.3 (1.2–1.3) |
| DURING PREGNANCY BUT BEFORE LABOR AND DELIVERY | ||||
| Hypertension | ||||
| None | 93.3 | 21,306 (2.7) | 1.0 (Reference) | 1.0 (Reference) |
| Preeclampsia | 3.2 | 2680 (9.7) | 3.7 (3.4–3.9) | 3.1 (2.9–3.3) |
| Eclampsia | 0.1 | 113 (15.6) | 6.0 (5.0–7.2) | 5.1 (4.3–6.2) |
| Other | 3.4 | 1385 (4.7) | 1.7 (1.6–1.9) | 1.7 (1.6–1.8) |
| Diabetes | 6.4 | 1989 (3.6) | 1.3 (1.2–1.3) | 1.0 (1.0–1.1) |
| Polyhydramnios | 0.6 | 249 (4.7) | 1.6 (1.4–1.9) | 1.3 (1.2–1.5) |
| Amnionitis | 1.8 | 1415 (9.3) | 3.3 (2.9–3.8) | 2.9 (2.5–3.4) |
| Placenta previa or abruption | 1.8 | 3099 (19.9) | 7.6 (7.3–8.0) | 7.0 (6.6–7.3) |
| Multiple pregnancy | 1.2 | 1095 (11.1) | 3.9 (3.6–4.2) | 2.8 (2.6–3.0) |
| Fetal macrosomia | 2.9 | 929 (3.7) | 1.3 (1.2–1.4) | 1.4 (1.3–1.5) |
| Noncephalic presentation | 7.4 | 2815 (4.4) | 1.6 (1.5–1.6) | 1.2 (1.1–1.2) |
| Fibroids | 0.9 | 626 (8.3) | 2.9 (2.6–3.2) | 2.0 (1.8–2.2) |
| DURING LABOR AND DELIVERY | ||||
| Induction of labor | ||||
| Medical | 15.4 | 4329 (3.3) | 1.1 (1.1–1.2) | 1.1 (1.04–1.1) |
| Surgical | 3.7 | 859 (2.7) | 0.9 (0.8–1.0) | 0.9 (0.8–0.99) |
| Mode of delivery | ||||
| Spontaneous vaginal | 64.5 | 12,471 (2.3) | 1.0 (Reference) | 1.0 (Reference) |
| Instrumental vaginal | 7.1 | 2472 (4.1) | 1.8 (1.7–1.9) | 1.5 (1.4–1.6) |
| Cesarean | 28.4 | 10,541 (4.3) | 1.9 (1.8– 2.0) | 1.4 (1.3–1.5) |
| Cervical laceration | 0.2 | 3309 (187.8) | 88.4 (82.4–94.7) | 94.0 (87.3–101.2) |
| Uterine rupture | 0.1 | 381 (64.5) | 23.1 (20.4–26.2) | 11.6 (9.7–13.8) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
