Approximately 50,000 women per year in the United States experience severe maternal morbidity or life-threatening conditions during delivery, and this number has been increasing over the past decade.1 To facilitate population-based monitoring of severe maternal morbidity, the Centers for Disease Control and Prevention (CDC) developed a standard definition for severe maternal morbidity in 2012 that includes 25 indicators of the presence and management of severe complications (Table 1).2 Nationally, the leading indicators of severe maternal morbidity are blood transfusion, hysterectomy, disseminated intravascular coagulation, adult respiratory distress syndrome, and acute renal failure.1
In response to increasing rates of both maternal mortality and morbidity, there has been renewed calls to focus on maternal health.3,4 National strategies include increased collaboration among maternal health stakeholders such as maternity providers from multiple professions, facility-based review of severe maternal morbidity cases, and the development of national guidelines or bundles around the most common underlying causes.5 For instance, the bundle for obstetric hemorrhage developed by the National Partnership for Maternal Safety recommends that hospitals develop action plans, increase access to medical supplies and blood components, and conduct education and drills with staff.6
Despite growing attention and interventions to address severe maternal morbidity, minimal information exists on the economic burden associated with these events. To address this gap, study objectives were to quantify the average and total hospital delivery costs associated with severe maternal morbidity in excess of nonsevere maternal morbidity deliveries over a 5-year period within the five boroughs of New York City adjusting for sociodemographic and clinical factors. Although the interest and motivation to promote maternal health are clear, these findings can be used to demonstrate the magnitude of the problem and to evaluate the cost-effectiveness of interventions to reduce severe maternal morbidity.
MATERIALS AND METHODS
We conducted a population-based cross-sectional study using birth certificate records linked to hospital discharge data from the Statewide Planning and Resource Cooperative for all deliveries that resulted in a live birth in New York City from 2008 to 2012. A longer description of the data sources, linkage, and match rate is reported elsewhere.7 Additionally, we excluded 14 records from three nonobstetric hospitals that had fewer than five births in a given year and did not represent the standard of care, resulting in a final sample size of 588,218 (Fig. 1). The study protocol was approved by the New York City Department of Health and Mental Hygiene institutional review board (#14-052).
We defined severe maternal morbidity using the CDC's definition, which contains 25 indicators of severe complications and life-saving procedures identified by International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes (Table 1). Delivery hospitalizations with at least one indicator were categorized as complicated by severe maternal morbidity. In cases in which only diagnosis-related severe maternal morbidity codes were present, an additional severity adjustment was conducted in line with the CDC methodology.2
Hospital delivery costs were estimated by converting hospital charges from the Statewide Planning and Resource Cooperative to costs using year- and hospital-specific cost-to-charge ratios and diagnosis-specific adjustment factors (Fig. 1). Total charges contained in the Statewide Planning and Resource Cooperative are the sum of all accommodation and ancillary services and reflect the amount hospitals bill for services (excluding professional fees) rather than the amount it costs to provide services. To account for this difference, we multiplied charges by a cost-to-charge ratio, which were created by the Healthcare Cost and Utilization Project by comparing reported charges with hospital accounting reports collected by the Centers for Medicare and Medicaid.8 Year- and hospital-specific cost-to-charge ratios were linked with Statewide Planning and Resource Cooperative data using American Hospital Association Linkage Files, also available from the Healthcare Cost and Utilization Project.9 New York City hospital cost-to-charge ratios from 2008 to 2012 ranged from 0.348 to 0.495 across 42 hospitals, meaning costs were 34.8–49.5% of what hospitals charged during this time. We imputed the cost-to-charge ratio for one hospital–year combination that was not available using the New York City average cost-to-charge ratio for the same year.
Furthermore, to account for known variation in department-level markup (eg, higher markup for operating room services), we multiplied charges by adjustment factors specific to the diagnosis-related group (DRG) code, which groups hospitalizations by similar clinical and demographic characteristics for the purposes of reimbursement.10,11 The most common DRG adjustment factors were 1.383 for vaginal delivery without complication (DRG=775) and 1.189 for cesarean delivery without complication (DRG=766). Finally, we accounted for inflation using the Consumer Price Index for Medical Care from the Bureau of Labor Statistics by multiplying adjusted charges by a year-specific inflation factor (Fig. 1).12 In the sample, costs increased by 12% in 2008, 11% in 2009, 8% in 2010, and 5% in 2011.
Additional variables of interest included maternal age (19 or younger, 20–24, 25–29, 30–34, 35–39, and 40 years or older), race and ethnicity (white non-Latina, black non-Latina, Latina, Asian-Pacific Islander, and other or multiple non-Latina ethnicities), and primary payer (Medicaid, private, and other). Neighborhood poverty level was assigned based on data from the American Community Survey collected by the U.S. Census Bureau and linked with the delivery record based on the reported census tract of residence. Neighborhood poverty was defined as the percentage of residents whose family income is below the federal poverty level categorized based on Department of Health and Mental Hygiene's standards as low (less than 10%), medium (10–20%), high (20–30%), and very high poverty (greater than 30%).13 The neighborhood poverty of women who delivered in a New York City facility but reside outside city limits could not be determined and were categorized as non-New York City residents.
Clinical variables included method of delivery (primary cesarean, repeat cesarean, and vaginal), plurality (singleton and multiple), and comorbidity (present and absent). We used a published maternal comorbidity index to assess a broad range of chronic conditions using codes available in the Statewide Planning and Resource Cooperative .14 The index includes 20 conditions relevant to obstetric patients such as gestational diabetes, placenta previa, and human immunodeficiency virus. We excluded two conditions (multiparous pregnancy and previous cesarean delivery) from the index because they were included as separate covariates and we removed specific ICD-9-CM codes (282.6 and 642.6) from two other conditions (sickle cell anemia and preeclampsia) that overlapped with the definition for severe maternal morbidity (Table 2). Women were categorized as having a comorbidity if they had one or more conditions. Additionally, we categorized hospital according to the level of perinatal care (level II, level III, and regional perinatal center) based on information from the New York State Health Profiles15 and type of hospital (nonprofit or public).16 Finally, we used the Statewide Planning and Resource Cooperative to estimate the total length of stay in the hospital and category of services provided (eg, laboratory, intensive care, or respiratory services) based on the reported revenue code, which captures both the accommodation and ancillary services provided and corresponds with standards produced by the National Uniform Bill Committee.17 Because there are approximately 100 service categories, only the most common categories were presented, and these services are not mutually exclusive (meaning women may receive multiple services as part of one delivery).
We tested for significant differences between deliveries with and without severe maternal morbidity by maternal sociodemographic, clinical, and hospital characteristic using χ2 tests. We calculated unadjusted mean costs and 95% CIs for deliveries with and without severe maternal morbidity and by select demographic, clinical, and hospital characteristics and tested for significant differences using Kruskal-Wallis tests, a nonparametric test of rank. Multivariable regression models were used to estimate costs associated with severe maternal morbidity, controlling for age, race and ethnicity, neighborhood poverty, primary insurance payer, number of deliveries, method of delivery, comorbidity, and year of birth. To account for the skewed and clustered nature of the data at the hospital level, we used a generalized linear model with a log link, gamma distribution, and robust standard errors. Hospital fixed effects were included in the model to address potential biases attributable to systematic sorting of patients into different hospitals based on observable and unobservable patient characteristics and variation in hospital pricing, two-well documented phenomenon in New York City.18,19 To calculate adjusted mean costs from the model, we took the exponent of the β coefficient for the intercept plus the β coefficient for each of the parameters of interest. Observations with missing data on any of the variables included in the multivariable model (less than 1%) were dropped, resulting in a sample size of 582,668. Using the difference in adjusted mean cost for deliveries with and without severe maternal morbidity and the total number of severe maternal morbidity cases, we estimated the average and total excess costs from 2008 to 2012. All analyses were performed using SAS 9.2.
From 2008 to 2012 there were 588,218 deliveries in New York City birthing hospitals, 2.3% of which met the criteria for severe maternal morbidity (Table 3). Severe maternal morbidity deliveries were more likely to be among women of color, women who were older, and women with Medicaid insurance. Severe maternal morbidity deliveries were also more likely to be cesarean, multiple births, and have clinical comorbidities present.More than 40% of all hospitalizations complicated by severe maternal morbidity lasted 5 days or longer compared with 6.4% of deliveries without severe maternal morbidity. Severe maternal morbidity deliveries required higher than normal service use, including coronary care (0.9% vs 0.01%), intensive care (9.1% vs 0.2%), nuclear medicine (0.7% vs 0.0%), computed tomography scans (12.9% vs 0.6%), and the administration, processing, and storage of blood components (46.8% vs 2.3%).
The average cost of delivery from 2008 to 2012 was $7,454 (95% CI $7,439–7,469) (Table 4). Severe maternal morbidity deliveries cost an average of $14,442 (95% CI $14,128–14,756) compared with $7,289 (95% CI $7,276–7,302) among deliveries without severe maternal morbidity. Some of the most expensive severe maternal morbidity indicators were sepsis, shock, and cardiac arrest (Table 1). Unadjusted costs for all deliveries were also significantly (P<.001) different by age, race and ethnicity, and neighborhood poverty with the highest cost among women 40 years or older ($9,036, 95% CI $8,943–9,128), white non-Latina ($8,021, 95% CI $7,993–8,049), living in low-poverty neighborhoods ($8,045, 95% CI $8,014–8,079) or outside New York City ($9,484, 95% CI $9,412–9,557), or privately insured ($8,744, 95% CI $8,717–8,770). Some of the highest unadjusted costs were among deliveries with a multiple birth ($13,452, 95% CI $13,149–13,755), primary cesarean ($10,069, 95% CI $10,023–10,115), or clinical comorbidities ($9,717, 95% CI $9,652–9,781). Additionally, delivery costs were high overall when codes for occupational therapy ($60,560, 95% CI $54,483–66,636), physical therapy ($34,452, 95% CI $32,148–36,755), coronary care ($29,431, 95% CI $25,359–33,504), and intensive care ($27,821, 95% CI $26,609–29,032) were present.
After adjusting for maternal sociodemographic, clinical, and hospital-level factors, deliveries complicated by severe maternal morbidity were still significantly more costly compared with deliveries without severe maternal morbidity (P<.001) (Table 5). The adjusted mean cost of a delivery complicated by severe maternal morbidity was $14,816 (95% CI $14,173–15,488) compared with $8,691 (95% CI $8,847–8,537) for deliveries without severe maternal morbidity, for a difference of $6,126. Adjusted mean costs remained significantly higher among deliveries to women at older ages, but were attenuated for deliveries to women who were white non-Latina, privately insured, or living in low-poverty neighborhoods or outside New York City. Adjusted mean costs were also significantly higher among deliveries to women younger than 25 years of age and those with a primary cesarean or repeat cesarean delivery, multiple birth, and clinical comorbidity. Over 5 years, deliveries complicated by severe maternal morbidity cost approximately $200 million in total (13,502×$14,817), representing 4% of all delivery costs. Of this, $83 million, or 41%, was in excess of nonsevere maternal morbidity delivery costs (13,502×$6,126).
In this large 5-year cross-sectional study, we found that severe maternal morbidity nearly doubled the costs associated with delivery in New York City, even after adjusting for other sociodemographic and clinical factors. As a result, tens of millions of dollars were spent in excess of normal delivery costs from 2008 to 2012. Using the same adjusted difference in cost, we might expect that $306 million was spent in excess nationally in 2014.1
Our findings suggest that the higher cost of care for severe maternal morbidity is associated with the complexity of the delivery, including the presence of clinical comorbidities, multiple birth, and cesarean delivery, all of which have been shown previously to increase the cost of childbirth.20,21 We also found that severe maternal morbidity deliveries required a longer length of stay and more services, including greater imaging, blood transfusion, intensive care, and rehabilitation. Other significant drivers of overall delivery cost included age, race and ethnicity, neighborhood, and insurance, although some of these associations were attenuated or changed direction in adjusted analysis, likely a result of clustering of patients at certain hospitals and variation in pricing.
Although substantial, reported costs likely represent a small proportion of the total physical, emotional, and economic burden associated with the severe maternal morbidity before and after delivery. For instance, women in our sample with severe maternal morbidity were two times more likely to be hospitalized before their delivery than those women without severe maternal morbidity (8.4% vs 4.1%, P<.001). Although not specific to severe maternal morbidity, research has also shown that women with pregnancy complications (such as placental abruption or preeclampsia) are at greater risk for future cardiovascular and metabolic diseases later in life, a significant potential cost to the health care system.22,23 Future studies should explore the costs related to additional hospitalizations before and after delivery, adverse newborn outcomes, long-term rehabilitation, and lost productivity.
These findings are relevant to health care providers and hospital administrators, many of whom have committed to implementing maternal safety bundles as part of the Alliance for Innovation on Maternal Health, for which the American College of Obstetricians and Gynecologists is a lead partner.24 Although the bundles require significant time, buy-in, and financial resources, our study suggests that they could reduce health care costs. For instance, using the cost estimates from this study we might expect the Alliance for Innovation on Maternal Health project to save $635 million excess costs nationally if the goal of eliminating 100,000 cases of severe maternal morbidity over the course of 4 years is reached. These savings may be particularly important as more payers move to value-based payment models, which are designed to incentivize the quality of care over the volume of services by providing a “bundled” payment for all aspects of maternity care.25 Our analysis can be used to demonstrate the magnitude of severe maternal morbidity and evaluate the cost-effectiveness of potential interventions. The specific estimates by sociodemographic and clinical characteristics may also help in designing targeted interventions.
This study has several limitations that should be considered. First, our data set was limited to deliveries that resulted in a live birth and their reported outcomes and charges in the time until discharge; therefore, the costs related to other types of pregnancy outcomes (eg, stillbirths) or pregnancy-related hospitalizations (eg, antepartum) could not be estimated. Second, there are known limitations with the charges reported in hospital discharge data.26 Although the Healthcare Cost and Utilization Project methodology attempts to correct for both hospital- and department-specific variation in markup, the precision of our estimates is limited. Recognizing differences in institutional coding expertise may exist, this method is still widely used in health services research, is readily accessible to researchers, and comparable across the United States. Nevertheless, by using this method, we were only able to analyze total charges, not those associated with specific revenue codes, which would have been useful for isolating services driving costs. Reported charges also do not include additional costs such as professional fees (eg, work performed by physicians), which are billed separately for a delivery; nor do they represent costs incurred by payers or individuals as a result of insurance practices. Third, the CDC definition for severe maternal morbidity is based on ICD-9-CM codes and has a reasonably high sensitivity (77%), but may result in false-positive cases.27 Finally, cost variation by hospital was treated as a fixed rather than a random effect in the model, potentially overlooking useful information from a cross-hospital difference. To assess the robustness of our estimates, we estimated a model including hospital as a random effect and obtained a similar, statistically significant estimate for severe maternal morbidity.
In summary, interventions to address severe maternal morbidity are needed to reduce the burden of disease on women and on the health care system. By providing estimates on the hospital delivery costs associated with severe maternal morbidity, the current study can be used to demonstrate the magnitude of the problem and to evaluate the cost-effectiveness of interventions to improve maternal health.
1. Centers for Disease Control and Prevention. Severe maternal morbidity in the United States. Available at: https://www.cdc.gov/reproductivehealth/maternalinfanthealth/severematernalmorbidity.html
. Retrieved January 12, 2017.
2. Callaghan WM, Creanga AA, Kuklina EV. Severe maternal morbidity among delivery and postpartum hospitalizations in the United States. Obstet Gynecol 2012;120:1029–36.
3. Lu MC, Highsmith K, de la Cruz D, Atrash HK. Putting the “M” back in the Maternal and Child Health Bureau: reducing maternal mortality and morbidity. Matern Child Health J 2015;19:1435–9.
4. D'Alton ME, Bonanno CA, Berkowitz RL, Brown HL, Copel JA, Cunningham FG, et al. Putting the “M” back in maternal-fetal medicine. Am J Obstet Gynecol 2013;208:442–8.
5. Kilpatrick SJ. Next steps to reduce maternal morbidity and mortality in the USA. Womens Health (Lond) 2015;11:193–9.
6. D'Alton ME, Main EK, Menard MK, Levy BS. The national partnership for maternal safety. Obstet Gynecol 2014;123:973–7.
7. Severe maternal morbidity New York City, 2008-2012. New York (NY): The New York City Department of Health and Mental Hygiene; 2016.
8. Healthcare Cost and Utilization Project. Cost-to-charge ratio files. Available at: https://www.hcup-us.ahrq.gov/db/state/costtocharge.jsp
. Retrieved July 27, 2015.
9. Healthcare Cost and Utilization Project. American Hospital Association linkage files. Available at: https://www.hcup-us.ahrq.gov/db/state/ahalinkage/aha_linkage.jsp
. Retrieved December 5, 2017.
10. Salemi JL, Comins MM, Chandler K, Mogos MF, Salihu HM. A practical approach for calculating reliable cost estimates from observational data: application to cost analyses in maternal and child health. Appl Health Econ Health Policy 2013;11:343–57.
11. Sun Y, Friedman B. Tools for more accurate inpatient cost estimates with HCUP databases, 2009. Errata added October 25, 2012. Methods series report #2011-04. Available at: https://www.hcup-us.ahrq.gov/reports/methods/2011_04.pdf
. Retrieved Retrieved October 25, 2015
12. U.S. Department of Labor. CPI—all urban consumers. Available at: https://data.bls.gov/cgi-bin/surveymost?cu
. Retrieved August 5, 2015.
13. Toprani A, Hadler JL. Selecting and applying a standard area-based socioeconomic status measure for public health data: analysis for New York City. New York (NY): New York City Department of Health and Mental Hygiene; 2013.
14. Bateman BT, Mhyre JM, Hernandez-Diaz S, Huybrechts KF, Fischer MA, Creanga AA, et al. Development of a comorbidity index for use in obstetric patients. Obstet Gynecol 2013;122:957–65.
15. New York State Department of Health. NYS health profiles. Available at: https://profiles.health.ny.gov/
. Retrieved February 28, 2017.
16. NYC Health+Hospitals. Health care locations. Available at: http://www.nychealthandhospitals.org/hospitals/
. Retrieved December 5, 2017.
17. New York State Department of Health. SPARCS inpatient output data dictionary version 1.0. Available at: https://www.health.ny.gov/statistics/sparcs/sysdoc/inpatientoutputdd.pdf
. Retrieved December 5, 2017.
18. Howell EA, Egorova N, Balbierz A, Zeitlin J, Hebert PL. Black-white differences in severe maternal morbidity and site of care. Am J Obstet Gynecol 2016;214:122.e1–7.
19. Xu X, Gariepy A, Lundsberg LS, Sheth SS, Pettker CM, Krumholz HM, et al. Wide variation found in hospital facility costs for maternity stays involving low-risk childbirth. Health Aff (Millwood) 2015;34:1212–9.
20. Huynh L, McCoy M, Law A, Tran KN, Knuth S, Lefebvre P, et al. Systematic literature review of the costs of pregnancy in the US. Pharmacoeconomics 2013;31:1005–30.
21. Law A, McCoy M, Lynen R, Curkendall SM, Gatwood J, Juneau PL, et al. The additional cost burden of preexisting medical conditions during pregnancy and childbirth. J Womens Health (Larchmt) 2015;24:924–32.
22. Neiger R. Long-term effects of pregnancy complications on maternal health: a review. J Clin Med 2017;6:pii: E76.
23. Park K, Wei J, Minissian M, Bairey Merz CN, Pepine CJ. Adverse pregnancy conditions, Infertility, and future cardiovascular risk: implications for mother and child. Cardiovasc Drugs Ther 2015;29:391–401.
24. Alliance for Innovation on Women's Health. AIM program overview. Available at: http://safehealthcareforeverywoman.org/wp-content/uploads/2016/10/AIM-Program-Description.pdf
. Retrieved December 5, 2017.
25. Whitman E. Bundles of joy? How new payment models for maternal care could deliver lower costs. Modern Healthcare; 2016.
26. Reinhardt UE. The pricing of U.S. hospital services: chaos behind a veil of secrecy. Health Aff (Millwood) 2006;25:57–69.
27. 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;214:643.e641–10.