Spontaneous vaginal birth varies significantly across US hospitals

Abstract

Background:

Birth is the most common reason for hospitalization in the United States. Hospital variation in maternal outcomes is an important indicator of health care quality. Spontaneous vaginal birth (SVB) is the most optimal birth outcome for the majority of mothers and newborns. The purpose of this study was to examine hospital-level variation in SVB overall and among low-risk women in a four-state sample representing 25% of births in the United States in 2016.

Methods:

Women giving birth in California, Pennsylvania, New Jersey, and Florida were identified in 2016 state discharge abstracts. Patient data were merged with hospital data from the American Hospital Association’s (AHA) 2016 Annual Survey. Overall and low-risk SVB rates were calculated for each hospital in the sample and stratified by bed size, teaching status, rurality, birth volume, and state.

Results:

Our final sample included 869 681 women who gave birth in 494 hospitals. The mean overall SVB rate in the sample was 61.1%, ranging from 16.8% to 79.9%. The mean low-risk SVB rate was 78% and ranged from 34.6% to 93.3%. Variation in SVB rates cut across all the hospital structural characteristic strata.

Discussion:

The wide variation in SVB rates indicates significant room for improvement in this maternal quality metric. Our finding, that hospitals of all types and locations had both low and high SVB rates, suggests that excellent maternal outcomes are possible in all hospital settings. The variation in SVB rates across hospitals warrants research into modifiable hospital factors that may be influencing SVB rates.

1 |. INTRODUCTION

Birth is the most common reason for hospitalization in the United States, resulting in four times more admissions than the next most common reason for hospitalization.^1,2 Efforts to improve maternal and neonatal outcomes typically focus on decreasing or minimizing poor outcomes, instead of increasing and maximizing positive outcomes.^3,4 Spontaneous vaginal birth (SVB), defined as vaginal birth without vacuum or forceps, is an example of a positive maternal and neonatal outcome. Evidence demonstrates that SVB is associated with better short- and long-term maternal and neonatal outcomes, compared to cesarean birth.^5,6 Also, most women prefer an SVB.⁷ Much of the maternal outcomes research related to quality and safety improvement focuses on cesarean birth, vaginal birth after cesarean, or maternal morbidity and mortality. SVB, despite its desirability, has received comparatively little research attention.

As a maternal outcome metric, SVB is related to, but distinct from, cesarean birth. While SVB is close to being the inverse of cesarean (operative vaginal births comprise less than 5% of births annually), there are at least two important distinctions between the two metrics. First, while interventions to increase SVB or decrease cesareans are similar, efforts to decrease cesarean rates are likely to focus on a subset of these interventions, such as increasing the time allowed for labor progression, administration of oxytocin, or decreasing the unnecessary use of continuous fetal monitoring. Efforts to increase SVB rates might include increased nurse staffing to allow one-to-one care, a greater use of doulas, and a focus on prenatal education. In other words, while decreasing cesarean birth rates works within our current maternity care paradigm, working to increase the rate of SVB represents a culture change. Second, decades of attention given to cesarean births have failed to reduce the national rate significantly. Experts in behavior change have noted that it is more effective to replace a negative habit with a positive one, rather than try to combat the negative habit. Based on this logic, if we want to decrease the cesarean birth rate, we may be more successful if we focus on supporting and increasing the SVB rate, instead of trying to decrease the rate of unnecessary cesareans.

Spontaneous vaginal birth is an important outcome for policy. In 2017, a typical SVB admission costs $15 000, compared to a cesarean admission, which cost $11 200.^8,9 These costs may be a particular concern for Medicaid which pays for 43% of the births in the United States,¹⁰ as well as for hospitals that find that labor and delivery are a “money-losing” service.^11,12 There is also the cost of interventions that do not improve outcomes and might decrease the chance of an SVB, such as continuous fetal heart monitoring.¹³ By contrast, the interventions that support and encourage SVB are low-cost and high-touch, including labor support, ambulation, intermittent auscultation, and watchful waiting.^3,13,14

Recent guidance from the National Partnership for Women and Families and the American College of Obstetricians and Gynecologists recommends increasing the rate of SVB and limiting unnecessary intervention, respectively.^13,14 The World Health Organization (WHO) determined that a cesarean rate greater than 10% is not associated with reductions in maternal and newborn mortality at the population level, based on a systematic review to identify the ideal cesarean rate in a given country or population, and an international country-level analysis with the most current data.¹⁵ If, as the WHO maintains, a cesarean rate greater than 10% is not associated with a decrease in maternal or neonatal mortality, it is possible that the SVB rate could be as high as 90%.^15,16 Current data suggest that the US SVB rate is around 65%, as the cesarean birth rate is around 31.7% and the operative vaginal birth rate is around 3%.^2,17

Hospital variation in SVB may be an important indicator of health care quality. Research on cesarean birth and vaginal birth after cesarean indicates that hospital factors may be a main driver of variation.^18–20 Factors such as hospital guidelines, midwifery-led care, and nurse-to-patient staffing ratios have been identified as potential drivers of variation in maternal outcomes.^21–24 We were not able to identify research on hospital-level variation in SVB. The purpose of this study is to examine hospital-level variations in SVB overall and among low-risk women in a four-state sample representing 25% of births in the United States in 2016.

2 |. STUDY DATA AND METHODS

Deidentified patient data were obtained from 2016 state discharge abstracts that were made available to the researchers from the RN4CAST study (PI Aiken R01NR014855). These annual inpatient discharge summaries were obtained from the Office of Statewide Health Planning and Development in California, the Healthcare Cost and Utilization Project’s State Inpatient Database for Florida, New Jersey Department of Health and Senior Services, and the Pennsylvania Health Care Cost Containment Council.²⁵ The subset of inpatients with diagnosis-related group codes 765 to 780 from Major Diagnostic Category 14 (Pregnancy, Childbirth & the Puerperium) were provided to the authors. These diagnostic related groups included delivery and postpartum, false labor, abortion, and ectopic pregnancy coding and excluded antepartum diagnoses and patients whose principal diagnosis was invalid as a discharge diagnosis.²⁶ From this initial sample, women giving birth were identified using a validated methodology that we crosswalked to the International Classification of Diseases, Tenth Revision (ICD-10).²⁷ Hospital characteristics data were obtained from the American Hospital Association’s (AHA) 2016 Annual Survey data and merged with the patient data using the AHA hospital identifier. Hospitals were excluded if they had less than 100 births. This exclusion criterion is consistent with other research in which a minimum of 100 births is considered enough births to allow meaningful comparisons across hospitals.^19,28

We had two hospital-level outcomes: overall SVB rate and the SVB rate among low-risk women. While validated methods exist for identifying cesarean births, we were not aware of a validated methodology to identify operative vaginal deliveries or SVB. We identified SVB using ICD-10 codes and diagnosis-related group codes O80, 767, 768, 774, 775, 796, 797, 798, 805, 806, 807, and 10E0XZZ. To evaluate potentially misclassified cases, we then identified cesarean births (765, 766, O82, 10D00Z0, 10D00Z1, 10D00Z)²⁹ and operative vaginal deliveries (10D07Z, O81) and cross-tabulated them with SVB. A total of 2994 women were listed as having both an operative and a spontaneous vaginal birth and were recategorized as having an operative vaginal birth. A total of 718 women were listed as having both a cesarean and an SVB and were recategorized as having a cesarean. Low risk was defined as a term, singleton, vertex pregnancy, following AHRQ’s definition.²⁹ Nulliparity is sometimes included in this low risk definition, but it was not included in this study since this variable is not available in administrative hospital discharge abstracts. The decision to not include nulliparity in the definition of low risk accords with prior research.¹⁹ In comparison with more specific risk adjustment models, using term, singleton, and vertex has been found to not vary significantly from the more specific models.³⁰

A hospital’s SVB rate was calculated as the percentage of all births that were spontaneous vaginal. Low-risk SVB was calculated as the percentage of all low-risk births that were spontaneous vaginal. We identified women with low-risk births by excluding the following characteristics (followed by the corresponding ICD-10 procedure or diagnosis codes), as has been done previously: preterm birth (O601); multiple birth (O30, O31, O632, O661, O666, Z372, Z373, Z374, Z3750, Z3751, Z3752, Z3753, Z3754, Z3759, Z3760, Z3761, Z3762, Z3763, Z3764, Z3769, Z377); breech or other malposition (O32, O64); and prior cesarean delivery (O3421, O6641).¹⁹

Hospital structural characteristics were derived from the AHA’s Annual Survey and included bed size, teaching status, and rurality.³¹ Bed size categories followed the Healthcare Cost and Utilization Project’s definition, which incorporate region, teaching status, and rurality.³² Forty-three hospitals were missing a bed size because they were missing information either on bed size (29 hospitals), teaching (29 hospitals), or rural status (19 hospitals). These hospitals were included in the overall totals, but not the stratified. Rurality was based on the Core-Based Statistical Area codes, where metropolitan was classified as urban, and micropolitan was combined with rural. Birth volume was defined as a hospital’s number of births per year, and categories were created in accord with prior research.³³ Categories ranged from ≥100-<500; 500–999; 1000–2499; and ≥2500.

Rates of overall and low-risk SVB were calculated across all the hospitals in the sample. These rates were then stratified by bed size, teaching status, rurality, and state and were presented as the mean, minimum, maximum, and interquartile range (the difference between the seventy-fifth and twenty-fifth percentile). In order to graphically represent the distributions of SVB both overall and for low-risk women, hospitals were grouped into 1 percent increments by their percent of spontaneous vaginal birth. Since the data for this study were deidentified, it met the criteria for non-human subjects research and was granted exemption from review by the Institutional Review Board.

3 |. STUDY RESULTS

Our final sample included 869 681 women who gave birth in 494 hospitals in 2016. Thirty-two hospitals had been excluded previously because they had less than 100 births. There was considerable variation in both overall and low-risk SVB rates. The mean overall SVB rate in the sample was 61.1%, ranging from 16.8% to 79.9%, nearly a fivefold difference. Of the women in the sample, 70.8% (622,522) were at low risk for cesarean birth. The mean low-risk SVB rate was 78% and ranged from 34.6% to 93.3%, a nearly threefold difference. Figures 1 and 2 represent the distribution of hospital SVB rates visually.

FIGURE 1.

Distribution of overall SVB in US hospitals in 2016

FIGURE 2.

Distribution of SVB in US hospitals among low-risk pregnancies in 2016

Across the 494 hospitals in the sample, the mean number of deliveries was 3101, ranging from 107 to 12467. Large hospitals, which averaged 3164 births per year, made up nearly half the sample (44.5%). The rest of the hospitals were divided fairly evenly between small and medium hospitals, which averaged 2484 and 2760 births per year, respectively. A little less than half of the hospitals were teaching hospitals, and only 8.7% of the hospitals were rural. Of the 494 hospitals in the sample, one-fifth each had between 100–500 and 500–999 births per year, 184 had 1000–2499, and 122 had ≥2500. Average rates of overall vaginal and low-risk vaginal birth were fairly consistent across birth volume categories. Further details on the hospitals are available in Table 1.

Table 1.

Delivery volume and SVB rates in US hospitals, overall and by size, teaching status, and location, 2016

	All hospitals	Hospital HCUP bed category			Teaching hospitals	Rural hospitals	Birth volume categoriesa
		Small	Medium	Large			I	II	III	IV
Number of Hospitals (n = 494)	494	113	118	220	211	43	94	94	184	122
Number of obstetric deliveries (n = 879,681)
Mean	3,101.2	2,485.5	2,760.5	3,164	3,478.6	1,042.2	324.5	752.3	1655.1	3885
Min	107	126	108	113	107	108	107	505	1000	2509
Max	12,473	8,841	9,072	7,015	12,473	3,467	496	996	2496	12473
Total vaginal birth rate (%) (n = 541,160)
Mean	61.1	61.9	60.7	61	60.7	63.2	60.6	60.9	60.5	61.6
Min	16.8	33.1	30.9	34.2	16.8	40.3	16.8	38.6	33.1	37.1
Max	79.9	79.6	74.9	75	77.1	74.9	79.9	73.4	79.6	77.1
Low-risk vaginal birth rate (%) (n = 486,958)
Mean	78	78.8	78	77.6	77.2	77.5	77.6	77.9	77.8	78.5
Min	34.6	54.5	50.6	45.6	34.6	48.7	34.6	56.1	50.6	57.4
Max	93.3	90.5	93.3	89.2	87.9	89.2	89.2	87.9	93.3	87.3

^aBirth volume categories: I (<500 births per year); II (500-999); III (1000-2499); and IV (≥2500).

The average overall SVB rate was similar across bed size and teaching status, ranging from 60.7% to 61.9%, though it was slightly higher in rural hospitals at 63.2%. Small and teaching hospitals had slightly higher maximum rates of overall SVB rate. Rural hospitals had the highest minimum overall SVB rate (40.3%), which was more than twice the minimum for teaching hospitals (16.8%). While the overall IQR for all hospitals was 8.5, there was slightly more variability in medium and rural hospitals. Table 1 provides further descriptions of the overall SVB rate. While the overall SVB rate was fairly similar across hospitals with different birth volumes (60.5% to 61.6%), hospitals with the highest birth volume had the least variation and, thus, the lowest interquartile range (7.2).

The average rate of low-risk SVB was higher, at 78%, than the overall rate. The minimum and maximum rates of low-risk SVB were also higher, with the exception of teaching hospitals where the lowest rate of low-risk SVB rate was 34.6%. Rural and small hospitals had the highest minimum rates of low-risk SVB at 56.1% and 54.5%, respectively. Large and teaching hospitals had the lowest minimums at 45.6% and 34.6%, respectively. As with the overall SVB rate across hospitals with different birth volumes, the average low-risk SVB rate was similar across birth volume categories, but hospitals with the highest birth volume had the least variation (IQR 6.4).

While the overall SVB rate was fairly consistent across states, Florida and New Jersey both had averages less than the overall mean of 60.4%. There were more marked differences among low-risk SVB rates with California having a much higher average (79.8%) and minimum (56.1%) low-risk SVB rate than the other states. By contrast, there were hospitals in the other three states in which only a little over a third of low-risk women had an SVB. Further variation between the states is presented in Table 2.

Table 2.

Delivery volume and SVD rates in US hospitals, by states in the sample, 2016

	All hospitals (n=494)	CA (n = 234)	FL (n = 116)	NJ (n = 50)	PA (n = 94)
Number of obstetric deliveries (n = 879,681)
Mean	3,101.2	3,214.5	3,049.4	3,136.1	2,745.8
Min	107	113	107	173	108
Max	12,473	9,072	12,473	6,233	8,841
Total vaginal birth rate (%) (n = 541,160)
Mean	61.1	62.7	57.1	59	62.5
Min	16.8	36.4	16.8	37.9	30.9
Max	79.9	79.9	73.4	77.1	75.5
Low-risk vaginal birth rate (%) (n = 486,958)
Mean	78	80.1	75.1	76.6	77.4
Min	34.6	56.1	34.6	58.9	45.6
Max	93.3	93.3	87.4	87.3	88.5

Hospitals where less than 60% of low-risk women had an SVB were found in all four states. These hospitals were found in both rural and metropolitan areas, were both teaching and non-teaching, and were represented among all bed size and birth volume categories. The same wide array was seen among hospitals where greater than or equal to 90% of low-risk women had an SVB.

4 |. DISCUSSION

We were interested in understanding hospital-level variation in SVB rates in order to develop foundational knowledge about SVB rates in the United States. Hospital variation in SVB rates indicates an opportunity for improving delivery outcomes for women nationally. We found an almost fivefold variation in overall SVB rates between hospitals, ranging from 16.8% to 79.9%. While low-risk SVB rates varied less, with only around a threefold variation between the hospital with the lowest rate (34.6%) and the highest (93.3%), the variation was still surprisingly wide.

Current national data suggest that the overall SVB rate in the United States is around 65%. In our sample, 67.6% of the hospitals had an overall SVB rate at or below 65%, indicating marked room for improvement. Only 8.5% of hospitals had an overall SVB rate greater than or equal to 70%. Even fewer (1.2%) had an overall SVB rate greater than or equal to 75%. One reason for the low rate of overall SVB is repeat cesarean. Safely increasing the rates of vaginal birth after cesarean is a critical quality initiative that may help to improve rates of SVB and decrease unnecessary cesareans.^34,35 A recent study looked at hospital-level variation in vaginal birth after cesarean and concluded that the hospital where a woman gave birth explained a large proportion of the variation in these rates.²⁰ These findings, taken in conjunction with our own, suggest that there are modifiable hospital characteristics that can improve birth outcomes for all women.

The low-risk SVB rate ranged across hospital from 1 in 3 women to 9 out of 10 women. In fact, three in ten hospitals in the sample had a low-risk SVB rate of 90% or greater. The hospitals with a 90% or greater SVB among low-risk women were found in all four states, in both rural and metropolitan areas. Teaching and non-teaching hospitals, with small, medium and large bed sizes, and with annual birth volume ranging from a few hundred to thousands were all represented. Location, teaching status, bed size, and birth volume are all structural characteristics of hospitals that cannot easily be changed. The wide variation in hospitals that were able to achieve such a high rate of SVB among low-risk women, however, suggests that such results might be attainable no matter the structural characteristics of a hospital. The 90% low-risk SVB rate complements the 10% necessary cesarean birth rate reported in the literature, especially among low-risk women.^15,16,36,37 The HealthyPeople 2020 cesarean rate goal for low-risk nulliparous women is 24.7%.³⁸ This is the first study we are aware of that shows that a hospital SVB rate over 90% is possible, indicating that a hospital cesarean rate of 10%, well below the HealthyPeople 2020 goal, is attainable.

To the best of our knowledge, this is the first study looking at hospital-level variation in SVB rates in the United States. The study which served as inspiration for this one looked at hospital-level variation in cesarean birth using 2009 data and also found significant hospital-level variation.¹⁹ The authors concluded that this variation was not entirely attributable to variations in patient populations. They performed a further study which found that individual maternal risk factors did not explain hospital variation in cesarean rates.^18,19 This study found greater variation in cesarean rates among low-risk women, as compared to the overall cesarean rate.¹⁹ By contrast, we found less variation in SVB among low-risk women, than for SVB overall.

We found that hospitals of all types and in all locations were represented among hospitals with both the lowest and the highest SVB rates. This suggests excellent maternal outcomes are possible in all hospital settings. It also suggests poor maternal outcomes in settings where excellence would be expected due to plentiful resources. Since high or low SVB rates were not consistently associated with any structural hospital characteristics in this sample, and based on the findings of other researchers,^19–21 it is likely that there are modifiable factors influencing the wide variation in this outcome. Other research suggests that where a woman gives birth is the primary driver of what kind of birth she will have, regardless of her medical risk factors.^18–20 If the hospital where a woman gives birth explains a significant portion of what kind of birth she has, and if hospital structural characteristics do not explain variation in the type of birth a woman has, then the question is: what is it about a hospital that explains variation in birth outcomes? Using SVB as a metric will assist researchers, policy-makers, and those working on quality improvement in their hospitals to assess the effectiveness of different interventions intended to improve birth outcomes.

Research in other populations had indicated that a hospital’s nursing resources, that is, the working environment, nurse staffing and nurse education, are all significantly associated with variation in patient outcomes across hospitals.^39,40 Qualitative research suggests that hospital nursing resources may explain some of the variation in maternal outcomes.^24,41,42 Specifically, in situations where there is insufficient nurse staffing, maternity nurses reported increased incidences of missed patient care and failure to rescue.⁴¹ According to a different group of nurses, delayed or missed maternity care and insufficient staffing led to cesarean birth, hemorrhage, depressed newborns at birth, difficulties establishing breastfeeding, and negative effects on patient satisfaction and experience.²⁴ Finally, a study involving nurses, physicians, and new mothers found that all three groups believed nursing care was critical to maternal and newborn outcomes, including caesarean and breastfeeding rates.⁴² Other studies have reported that changes in hospital policy, especially regarding labor definitions, are associated with decreased rates of cesarean birth.^21,43–46 For example, defining active labor as starting at 6 cm, rather than 4 cm, and allowing longer lengths of labor (especially for nulliparous women and women with an epidural) are associated with increased rates of SVB.^45,46 Another body of research indicates that midwife-led care and labor support are associated with increased rates of SVB.^22,23

Future research should determine which modifiable hospital characteristics are most strongly associated with SVB rates. Characteristics to consider, based on prior research, would include hospital guidelines, nursing resources, and midwife-led care. From there, researchers might also consider which of these characteristics are implicated in racially disparate maternal outcomes. Finally, interventions addressing modifiable hospital characteristics in order to increase SVB rates are likely warranted.

This study has some limitations. It is not possible to determine parity in the data set, so we were unable to identify nulliparous women. While having that aspect of the definition of “low-risk for cesarean” would have been helpful, the more limited definition of a term, singleton, and vertex pregnancy is also accepted and used in comparable research.^19,29 More importantly, we believe the addition of nulliparity to the definition is unlikely to explain the variation in low-risk SVB rates between hospitals. While the low risk for cesarean birth definition we used is standard, it does not exclude women with risk factors for cesarean such as gestational diabetes, pre-eclampsia, and obesity, among others. Other research, however, has found that accounting for these maternal medical risk factors does not explain hospital variation in cesarean birth rates.¹⁹ While maternal medical risk factors are unlikely to explain hospital variation in SVB rates, empirical attention to confirm this is warranted. We excluded hospitals with fewer than 100 births, so our findings may not apply to these hospitals. Relatively few hospitals with maternity services, however, have fewer than 100 births per year (6% of hospitals with any births in the original data). Finally, while the data are from 2016, the cesarean birth rate in the United States has remained relatively stable in the intervening years. We believe, therefore, that the observed relationships have also remained stable.

5 |. Conclusions

We found substantial hospital-level variation in SVB rates both overall and among low-risk women. This variation remained sizeable across stratification categories, including bed size, teaching status, rurality, and state. The extent of the variation and the fact that it cuts across hospital structural characteristics suggest that other factors are at play. For women desiring a SVB, our results are simultaneously encouraging and disheartening. They are encouraging because many hospitals have the capacity to support SVB, but they are disheartening because a large portion of hospitals have poor SVB rates.

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no new data were created or analyzed in this study.