Secondary Logo

Journal Logo

Contents: Health Services Research: Original Research

Effects of Delivery Volume and High-Risk Condition Volume on Maternal Morbidity Among High-Risk Obstetric Patients

Bozzuto, Laura MD, MS; Passarella, Molly MS; Lorch, Scott MD, MSCE; Srinivas, Sindhu MD, MSCE

Author Information
doi: 10.1097/AOG.0000000000003080
  • Free

Maternal morbidity and mortality are increasing in the United States, with a 66% increase in postpartum maternal mortality and 75% increase in maternal morbidity between 1998 and 2009.1–3 As part of the efforts to improve care and outcomes, some policy makers have proposed regionalizing obstetric services based on levels of maternal specialty care, similar to trauma and neonatal networks, for obstetric patients with recognized high-risk conditions.4,5 These efforts are based on the presumption that increased annual volume will lead to better systems of care and improved outcomes.6

Although volume is a key element in the rationale for the development of specialized regional care facilities in obstetrics, investigation into the relationship between volume and maternal outcomes has focused on all patients or low-risk obstetric patients.7,8 Studies have shown mixed results with increased rates of obstetric complications at both very low and very high-volume hospitals.7,9–11 The importance and interaction between overall delivery volume compared with volume of high-risk patients and outcomes has not previously been investigated.

Our objective was to examine the relationship between hospital delivery volume, high-risk condition volume, and the combined effect of both types of volume on maternal outcomes. The first aim was to examine the effect of total annual obstetric delivery volume on maternal morbidity for high-risk surgical and medical patients respectively. Second, we examined the effect of high-risk condition volume on morbidity for both high-risk cohorts. Finally, we examined the interaction between annual delivery volume and high-risk condition volume on maternal morbidity for patients with high-risk obstetric conditions (Table 1).

Table 1.
Table 1.:
Study Objectives


To investigate this question, a retrospective cohort of more than 10 million deliveries in three states was examined. All linked state birth and hospital discharge records were collected from California, Missouri, and Pennsylvania from 1995 to 2009. Data were obtained through a linked record of vital statistics, entered by each state's vital statistics program under the purview of the national and hospital administrative data programs.12 Two high-risk patient cohorts were identified. Patients with one of three surgical conditions were included in the surgical high-risk cohort, and patients with one of fourteen high-risk medical diagnoses were included in the medical high-risk cohort (Table 2). These were based on previously studied prenatally identifiable high-risk pregnancy conditions.2,13 Diagnoses were identified by International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) coding from hospital and birth records.14

Table 2.
Table 2.:
High-Risk Diagnoses

Hospitals were excluded if they had an average annual total delivery volume of less than 50, because this was likely to represent a hospital without an obstetrics delivery unit. Patients were included if they had a hospital admission leading to a delivery and, as such, could have only one delivery admission per pregnancy. Women could be represented more than one time if they had more than one pregnancy during the 15-year time period. Patients were excluded if they had a delivery at less than 20 weeks of gestational age or were transferred during their hospitalization, resulting in delivery of the neonate, because, based on this data set, there was no ability to determine the chronology of complications to attribute to each hospital. Transfers were determined by comparing the hospital of record at admission and discharge and were excluded. Patients with missing data were less than 2% of the cohort and were excluded from the analysis. ICD-9-CM coding data were preferentially used, and where discrepancies in birth certificate and ICD-9-CM coding occurred, ICD-9-CM codes were used.15

The primary outcome was a composite maternal outcome consisting of the previously defined diagnoses established as markers of severe maternal obstetric morbidity by the Centers for Disease Control and Prevention (Table 3).2,16

Table 3.
Table 3.:
Severe Maternal Morbidity by Hospital Delivery Volume

For both the surgical and medical high-risk cohorts, delivering hospitals were divided into volume categories based separately on: 1) annual total delivery volume in quartiles and 2) annual high-risk patient delivery volume in tertiles. Quartiles and tertiles were chosen, respectively, based on the size and range of the population for analysis to create equivalent groupings rather than based on predefined volume cutoffs, because there are have been no preexisting thresholds established in the literature. The association between the primary composite outcome and overall delivery volume as well as the association between the primary outcome and high-risk condition volume were evaluated. Finally, the combined effect of both total obstetric volume and high-risk volume was evaluated in the surgical patient and medical patient groups.

Unadjusted comparisons were performed with χ2 tests for categorical data and analysis of variance for continuous data. Adjustment was performed for age, sex, race, level of education, insurance status, prior cesarean delivery, multiple gestation, number of hospital beds, and state.9,17 Logistic regression models determined the association between hospital volume and each outcome, adjusting for the confounding variables above. Standard errors were calculated after clustering by hospital using the method of White to account for the nonindependence of patients delivering in a specific hospital.18–20 Statistical significance was indicated by a two-sided P<.05 or 95% CIs. Statistical analysis was performed using STATA 14. This study was approved by the Children's Hospital of Philadelphia Institutional Review Board.


The analyses divided patients into two cohorts: 1) those with a high-risk surgical diagnosis and 2) those with a high-risk medical diagnosis. There was a total of 142,194 high-risk surgical patients and 1,322,276 high-risk medical patients included in the two cohorts. The characteristics of the surgical and medical patients stratified by hospital high-risk delivery volume tertile are displayed in Tables 4 and 5. Annual hospital high-risk delivery volumes varied in the surgical high-risk cohort from a mean of 15 in the low-volume group to 96 in the high-volume group. In the medical high-risk cohort, the annual delivery volume in the low-volume tertile hospitals was 130 deliveries and 821 in the high-volume tertile hospitals (Tables 4 and 5). In both the surgical and medical cohorts, women at higher high-risk volume centers were older, less likely to have private insurance, and less likely to be white compared with the lowest-volume centers. California contributed the most deliveries for women at high risk for both medical and surgical outcomes. Patients with at least one complication were found in 10–12.2% of deliveries in the surgical population and 4.7–6.2% of medical high-risk deliveries (Table 3).

Table 4.
Table 4.:
High-Risk Surgical Patient Population by Hospital Tertile
Table 5.
Table 5.:
High-Risk Medical Patient Population by Hospital Tertile

To first assess the effect of total obstetric delivery volume on maternal morbidity among high-risk patients, hospitals were divided into quartiles of total annual obstetric delivery volume (Table 6). Among the surgical high-risk cohort, in unadjusted analyses, as total annual delivery volume increased, there was a significant increased risk of maternal morbidity, with a 26% increased risk in the highest quartile for total volume compared with low volume (surgical 4th quartile odds ratio [OR] 1.26, 95% CI 1.20–1.32) (Table 6). The same relationship was observed for the medical high-risk cohort in unadjusted analyses, with an 18% higher risk of adverse outcomes found in quartile three of overall hospital delivery volume compared with low-volume centers (Table 6). However, after adjusting for confounders, the relationship reversed with surgical patients delivering in the highest overall delivery volume quartile hospitals, having a risk that was more than 20% lower than that of patients delivering in hospitals in the lowest delivery volume quartile (4th quartile OR 0.78, 95% CI 0.64–0.94). The same relationship was found in the medical high-risk cohort with the significantly lowest risk of complications in the highest overall delivery volume quartile after adjustment (4th quartile OR 0.72, 95% CI 0.59–0.86).

Table 6.
Table 6.:
Risk of Severe Maternal Morbidity by Overall Hospital Delivery Volume

For the second analysis, we examined the association between high-risk surgical and medical volume on severe maternal morbidity for high-risk surgical and medical patients. Hospitals were divided into groupings based on the hospital volume of high-risk patients. Annual high-risk volume ranged from 1 to 497 high-risk deliveries per hospital for the surgical high-risk cohort and 1–2,222 high-risk deliveries per hospital in the medical high-risk cohort (Table 7).

Table 7.
Table 7.:
Risk of Severe Maternal Morbidity by Hospital High-Risk Patient Volume

The unadjusted analyses demonstrated increased risk of severe maternal morbidity in the highest medical and surgical volume hospitals for both the surgical and medical high-risk cohorts (surgical high-volume category OR 1.24, 95% CI 1.19–1.30; medical high-volume category OR 1.40, 95% CI 1.38–1.43). After adjusting for confounders, there was no significant difference between the volume tertiles in the surgical high-risk cohort (Table 7). In the medical high-risk cohort, after adjustment, the highest risk of severe maternal morbidity continued to be seen in the high-volume tertile of hospitals, with a 27% increased risk of an adverse outcome compared with the lowest medical volume centers.

Finally, the combined effect of overall delivery volume and high-risk condition volume was examined for both the surgical and medical patient cohorts. Hospitals were assigned to categories based on the quartile of total volume and tertile of high-risk condition volume.

For the surgical high-risk cohort, the combined effect of the two adjusted volume types is seen in Figure 1. The observed interaction effect between overall obstetric delivery volume and surgical high-risk patient volume was significant (P=.006). At the highest total obstetric delivery volume hospitals, there were the lowest odds of severe maternal morbidity across all surgical high-risk volumes. This risk was lowest in the low, high-risk volume hospitals.

Fig. 1.
Fig. 1.:
Severe maternal morbidity risk by total obstetric volume and surgical high-risk volume. *P<.05; P<.01; P<.001.Bozzuto. Hospital High-Risk OB Volume and Maternal Morbidity. Obstet Gynecol 2019.

For the medical high-risk cohort (Fig. 2), there was a significant interaction effect of the adjusted overall obstetric delivery volume and high-risk patient volumes (P<.001). In quartiles 2 through 4, as high-risk patient volume increased, complication rates also increased. The lowest odds of severe maternal morbidity were seen in the high total obstetric volume, low high-risk volume centers.

Fig. 2.
Fig. 2.:
Severe maternal morbidity risk by total obstetric volume and medical high-risk volume. *P<.05; P<.001.Bozzuto. Hospital High-Risk OB Volume and Maternal Morbidity. Obstet Gynecol 2019.

For both surgical and medical high-risk cohorts, the lowest risk hospitals were those with high total obstetric delivery volumes but low volumes of high-risk patients (surgical OR 0.47, P=.04; medical OR 0.56, P<.001).


This study demonstrates the importance of both overall delivery volume as well as high-risk condition volume on maternal morbidity among high-risk surgical and medical obstetric patients. First, we found significantly lower rates of complications among surgical and medical high-risk women as annual overall obstetric delivery volume increased. Second, we saw increased rates of complications for high-risk surgical and medical patients in hospitals with a higher volume of these patients. Third was the interplay of both types of volume on maternal morbidity. The combined effect of both overall delivery volume and high-risk patient volume showed the lowest risk of complications at high total obstetric volume and low high-risk patient volume centers. This interaction was statistically significant in both the medical and surgical high-risk cohorts.

Our findings were consistent with results previously seen in low-risk obstetrics.7,10 For low-risk patients, analyses have shown a higher rate of adverse outcomes for low-volume centers for cesarean delivery complications, neonatal outcomes, and maternal morbidity.8,9 Other studies have shown no relationship or a bimodal relationship between hospital volume and outcomes.21,22

Based on this analysis, high-risk patients had the lowest risk of morbidity at centers with high annual overall delivery volume. Increased volumes of high-risk patients were associated with higher rates of maternal morbidity in nearly all hospital volume categories for medical patients, with a relatively stable risk across surgical volumes. This finding was surprising, as increased volume was hypothesized to drive improved outcomes as exposure to these patients increased. The most likely explanation for this finding is the inability for administrative coding data to accurately assess illness severity. Despite having the same diagnoses, patients with more severe disease are likely referred to or present at centers that care for a high volume of high-risk surgical and medical condition patients. This in turn leads to the appearance of increased maternal morbidity at high-volume, high-risk condition centers, whereas, in fact, it is the lack of our ability to discern disease severity among patients with the same diagnosis codes. This inability to accurately assess disease severity is a well-known limitation of research involving administrative coding data, and has been shown in studies of the effect of neonatal intensive care level and volume on patient outcomes.23,24 Health services researchers have developed comorbidity scores for use in medical populations, but because of the differences in patient characteristics and physiology of pregnancy, these have shown poor performance among obstetric patients.25

Additional explanations for these findings include: 1) a potential artificial lowering of risk at some centers if there were disproportionate transfers of more severely ill patients, because transfers were excluded; or 2) a potential volume threshold of high-risk–condition patients that hospitals, despite having a high total obstetric volume, are not as able to accommodate owing to resources and systems factors. However, we hypothesize that the most likely explanation is unaccounted for disease severity through discharge data.

The exception to the observed higher rates of complications at higher high-risk condition volume hospitals was seen at hospitals with low total obstetric volume and the highest volume of high-risk patients. These hospitals may represent specialty centers with specific referral patterns, targeted patient populations, or increased staffing of specialists. Additionally, these may represent referral centers with a low volume of local deliveries but a high volume of high-risk transfers.

As with all analyses of large hospital discharge data based on ICD-9-CM coding, there are some limitations. Prior investigations have shown the validity of ICD-9-CM coding in assessing maternal morbidity.14 However, as has been mentioned, it is not possible to determine the level of severity of that diagnosis. Additionally, it is possible that this cohort may not be generalizable to all regions of the country where populations, practice, and referral patterns may differ. However, the three states selected in this analysis have a representation from different areas of the country and a diversity of population. Although one of the strengths of this study is the size of the data set enabling this type of analysis, we acknowledge that this allows for statistical significance to be seen with adjusted odds in the 0.7–1.3 range, consistent with only a modest association.

Patients who were transferred during their hospitalization were excluded from this analysis owing to an inability to determine timing of transfer in relationship to development of complications for hospital attribution. This exclusion may result in an underestimation of complications in certain hospitals, particularly lower-volume centers that may transfer out more severely ill patients. Further investigation in this particular population to elucidate patient outcomes is critical as we look to evaluate the guidelines, timing, and effect of regionalized care.

The current analysis looks only at the effect of two volume types on the maternal outcomes of high-risk surgical and medical patients. It does not examine the effect on low-risk patients delivering at the same facilities. Future analyses will examine how the volume of high-risk patients at a hospital center may affect outcomes for low-risk patients.

As policy makers attempt to move regionalization forward, guidelines will need to be refined as to the characteristics and qualifications of regional centers.5 The recent document on levels of maternal care by the American College of Obstetricians and Gynecologists and the Society for Maternal-Fetal Medicine proposes a list of hospital and system attributes associated with higher levels of care.5 Total delivery volume and high-risk surgical condition volume may need to be considered as we evaluate the importance of different hospital level factors in regionalization and optimizing maternal outcomes.

To aid in policy development, additional work attempting to determine the thresholds of volume at which hospital system performance is optimized in obstetrics will be needed. This analysis importantly shows that this determination must consider more than simply total obstetric delivery volume or even volume of high-morbidity conditions.

In conclusion, we found that overall higher delivery volume obstetric centers were associated with lower risk of severe maternal morbidity for patients with high-risk pregnancies, and higher volumes of high-risk patients may increase the risk of adverse outcomes. To better track and analyze this important relationship, registry data that enables assessment of transfer data and illness severity will be essential to determine both which high-risk patients would most benefit from regionalized care centers as well as how those centers truly perform.


1. Callaghan WM, Creanga AA, Kuklina EV. Severe maternal morbidity among delivery and postpartum hospitalizations in the United States. Obstet Gynecol 2012;120:1029–36.
2. Callaghan WM, MacKay AP, Berg CJ. Identification of severe maternal morbidity during delivery hospitalizations, United States, 1991-2003. Am J Obstet Gynecol 2008;199:133.e1–8.
3. Creanga AA, Berg CJ, Ko JY, Farr SL, Tong VT, Bruce FC, et al. Maternal mortality and morbidity in the United States: where are we now? J Women's Health 2014;23:3–9.
4. Levels of maternal care. Obstetric Care Consensus No. 2. American College of Obstetrician and Gynecologists. Obstet Gynecol 2015;125:502–15.
5. Menard MK, Kilpatrick S, Saade G, Hollier LM, Joseph GF Jr, Barfield W, et al. Levels of maternal care. Am J Obstet Gynecol 2015;212:259–70.
6. Ozimek JA, Eddins RM, Greene N, Karagyozyan D, Pak S, Wong M, et al. Opportunities for improvement in care among women with severe maternal morbidity. Am J Obstet Gynecol 2016;215:509.e1–6.
7. Snowden JM, Cheng YW, Emeis CL, Caughey AB. The impact of hospital obstetric volume on maternal outcomes in term, non-low-birthweight pregnancies. Am J Obstet Gynecol 2015;212:380.e1–9.
8. Chung JH, Phibbs CS, Boscardin WJ, Kominski GF, Ortega AN, Gregory KD, et al. Examining the effect of hospital-level factors on mortality of very low birth weight infants using multilevel modeling. J Perinatol 2011;31:770–5.
9. Glancel LG, Dick AW, Glantz JC, Wissler RN, Qian F, Marroquin BM, et al. Rates of major obstetrical complications vary almost fivefold among US hospitals. Health Aff 2014;33:1330–6.
10. Kyser KL, Lu X, Santillan DA, Santillan MK, Hunter SK, Cahill AG, et al. The association between hospital obstetrical volume and maternal postpartum complications. Am J Obstet Gynecol 2012;207:42.e1–17.
11. Chang JJ, Stamilio DM, Macones GA. Effect of hospital volume on maternal outcomes in women with prior cesarean delivery undergoing trial of labor. Am J Epidemiol 2007;167:711–8.
12. Murthy K, Macheras M, Grobman WA, Lorch SA. Hospital of delivery and the racial differences in late preterm and early-term labor induction. Am J Perinatol 2015;32:952–9.
13. Korst LM, Fridman M, Lu MC, Mitchell C, Lawton E, Griffin F, et al. Monitoring childbirth morbidity using hospital discharge data: further development and application of a composite measure. Am J Obstet Gynecol 2014;211:268.e1–16.
14. Goff SL, Pekow PS, Markenson G, Knee A, Chasan-Taber L, Lindenauer PK. Validity of using ICD-9-CM codes to identify selected categories of obstetric complications, procedures and co-morbidities. Paediatr Perinat Epidemiol 2012;26:421–9.
15. Li Q, Jenkins DD, Kinsman SL. Birth settings and the validation of neonatal seizures recorded in birth certificates compared to medicaid claims and hospital discharge abstracts among live births in South Carolina, 1996–2013. Matern Child Health J 2017;21:1047–54.
16. Main EK, Abreo A, McNulty J, Gilbert W, McNally C, Poeltler D, et al. Measuring severe maternal morbidity: validation of potential measures. Am J Obstet Gynecol 2016;214:643e1–10.
17. Kozhimannil KB, Arcaya MC, Subramanian SV. Maternal clinical diagnoses and hospital variation in the risk of cesarean delivery: analyses of a national US hospital discharge database. PLoS Med 2014;11:e1001745.
18. Williams RL. A note on robust variance estimation for cluster-correlated data. Biometrics 2000;56:645–6.
19. Huber PJ. The behavior of maximum likelihood estimates under nonstandard conditions. In: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. Berkeley (CA): University of California Press; 1967. Vol. 1, 221–233.
20. White H. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 1980;48:817–30.
21. Janakiraman V, Lazar J, Joynt KE, Jha AK. Hospital volume, provider volume, and complications after childbirth in U.S. Hospitals. Obstet Gynecol 2011;118:521–7.
22. Phibbs CS, Bronstein JM, Buxton E, Phibbs RH. The effects of patient volume and level of care at the hospital of birth on neonatal mortality. JAMA 1996;276:1054–9.
23. Schneeweiss S, Maclure M. Use of comorbidity scores for control of confounding in studies using administrative databases. Int J Epidemiol 2000;29:891–8.
24. Lorch SA, Baiocchi M, Ahlberg CE, Small DS. The differential impact of delivery hospital on the outcomes of premature infants. Pediatrics 2012;130:270–8.
25. Aoyama K, D'Souza R, Inada E, Lapinsky SE, Fowler RA. Measurement properties of comorbidity indices in maternal health research: a systematic review. BMC Pregnancy Childbirth 2017;17:372.

Supplemental Digital Content

© 2019 by the American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.