Published Online: April 15, 2013
Katy B. Kozhimannil, PhD, MPA; Tetyana P. Shippee, PhD; Olusola Adegoke, MPH, MBBS; and Beth A. Virnig, PhD, MPH
Objectives: Childbirth is the leading reason for hospitalization in the United States, and maternityrelated expenditures are substantial for many health insurance programs, including Medicaid. We studied the relationship between primary payer and trends in hospital-based childbirth care.
Study Designs: Retrospective analysis of hospital discharge data from the Nationwide Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project, a 20% stratified sample of US hospitals.
Methods: Data on 6,717,486 hospital-based births for the years 2002 through 2009 came from the NIS. We used generalized estimating equations to measure associations over time between primary payer (Medicaid, private insurance, or self) and cesarean delivery, vaginal birth after cesarean (VBAC), labor induction, and episiotomy.
Results: Controlling for clinical, demographic, and hospital factors, births covered by Medicaid had lower odds of cesarean delivery (adjusted odds ratio [AOR], 0.91), labor induction (AOR, 0.73), and episiotomy (AOR, 0.62) and higher odds of VBAC (AOR, 1.20; P <.001 for all AORs) compared with privately insured births. Cesarean rates increased 6% annually among births paid by private insurance (AOR, 1.06; P <.001) and less rapidly (5% annually) among those covered by Medicaid.
Conclusions: US hospital-based births covered by private insurance were associated with higher rates of obstetric intervention than births paid for by Medicaid. After controlling for clinical, demographic, and hospital factors, cesarean delivery rates increased more rapidly among births covered by private insurance, compared with Medicaid. Changes in insurance coverage associated with healthcare reform may impact costs and quality of care for women giving birth in US ospitals.
Am J Manag Care. 2013;19(4):e125-e132
We studied the relationship between primary payer and trends in hospital-based childbirth care from 2002 to 2009 and found the following:
Obstetric intervention rates increased over time for all births, regardless of payer.
Presence and type of health insurance affected hospital-based childbirth care; births covered by private insurance had higher rates of obstetric intervention compared with Medicaid-funded births.
Controlling for individual and hospital-level factors, the cesarean delivery rate increased more rapidly for births covered by private insurance, compared with Medicaid.
Changes in health insurance coverage associated with the Affordable Care Act may have important health implications for women giving birth in US hospitals.
Childbirth is the leading reason for hospitalization of women in the United States,1 and hospital-based maternity care has undergone marked changes in recent years. Cesarean delivery rates have increased from 20.7% in 1996 to 32.9% in 2009..2-4 Rates of induction of labor have also increased, from 9.5% in 1990 to 23.1% in 2008.2,5 Meanwhile, rates of vaginal birth after cesarean (VBAC) have decreased from 28.3% in 1996 to 8.5% a decade later.4,6,7 These changes have catalyzed interprofessional dialogue on rising rates of obstetric intervention, especially because these changes have occurred alongside increases in adverse birth outcomes and persistent racial/ethnic disparities in maternal and neonatal health.2,8-11 Although advances in obstetric care have historically improved outcomes for mothers and babies, higher-than-expected risk-adjusted cesarean delivery rates are not associated with health gains, and high procedure intensity comes at a cost to women, infants, and the healthcare system.12,13
Maternity and newborn care is the top expenditure category for hospital payments by Medicaid and private insurers alike,14 and costs of childbirth care have been increasing. The average facility charge for a vaginal delivery in 2004 was $7772, and this increased to $9617 by 2009. Facilities charge more for a cesarean delivery, on average, $15,779 in 2009, which was up from $12,223 in 2004.15,16
Health insurance may impact childbirth-related healthcare such as cesarean delivery through benefits coverage, payment structures, and provider networks.17,18 State-level analyses have indicated that privately insured women experience higher rates of cesarean delivery, including elective cesarean delivery, compared with uninsured and publicly insured women.19,20 In 2009, private insurers paid $3.8 billion for 52% of all cesarean births in US hospitals.21 The public-sector role is also large. State Medicaid programs finance 45% of all US births.22 Thus, as public payers, states have significant policy leverage over obstetric practice via Medicaid coverage and benefits structures; they also have an important fiscal stake in improving the quality and value of childbirth care and associated health outcomes.23
Although research and vital statistics reports have documented changes in obstetric practice,2,24,25 the role of health insurance in facilitating or mitigating these longitudinal trends remains largely unexplored. The potential payer impact on trends in maternity care has important policy implications, given the volume of births in the United States per year (approximately 4 million), the public sector stake in terms of costs and health outcomes, and the role of private payers and self-insured employers in financing childbirth care.2 The goal of this study was to characterize differences in hospital-based obstetric care by primary payer, while controlling for relevant clinical, demographic, and hospital characteristics. Our analysis focused on changes over time (2002-2009) in delivery mode and obstetric procedures, and it measured whether these trends were changing more or less rapidly for women who gave birth without insurance coverage and for births financed by public versus private payers in a nationally representative sample of US hospitals.
Data and Study Populations
This study used data from the Nationwide Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project (HCUP), from 2002 through 2009. The NIS includes hospital discharge records for all payers for inpatient care for an approximate 20% stratified sample of US community hospitals and is the only national hospital database with discharge records for all patients, regardless of payer,26 which makes it well suited for this analysis. The quality and validity of NIS data have been previously reported, and these data have been widely used in health services research, including studies of maternity and obstetric care. 27-29
We analyzed discharge summaries for 6,717,486 maternal hospitalizations for childbirth in US hospitals in 44 states during the period 2002 to 2009 in which the primary payer was self (N = 241,503), Medicaid (N = 2,832,321), or private insurance (N = 3,643,643). We relied on a previously validated and published method of using NIS data to identify obstetric deliveries.27
Outcome variables are defined using both International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes and Clinical Classifications Software (CCS) codes, which were developed by HCUP for use with ICD-9-CM codes.26 Delivery mode outcomes were (1) cesarean delivery (ICD- 9-CM codes 740xx, 741xx, 742xx, 744xx, and 7499) and (2) VBAC, which was calculated as a vaginal birth conditional on indication of prior cesarean delivery (CCS code 189). The low-risk cesarean delivery rate (shown in unadjusted comparisons in Table 1) is calculated as the rate of cesarean delivery among women with term, singleton, vertex pregnancies, and no prior cesarean deliveries, following the definition employed by the American Congress of Obstetricians and Gynecologists (ACOG) as closely as the data allow.30 Secondary obstetric care outcomes included (3) induction of labor (either medical or surgical induction; ICD-9-CM codes 731 and 734) and (4) episiotomy (ICD-9-CM code 736). Insurance status was indicated by the primary payer for the childbirth hospitalization and categorized as Medicaid, private insurance, or self.
Our analysis controlled for patient-level demographics, including maternal age and race as recorded on hospital discharge records. Inclusion of clinical control variables is based on recommendations from ACOG and published analyses of risk-adjustment methodologies in obstetric care30,31; these variables include the following: diabetes (ICD-9-CM codes 6488x and 250xx), hypertension (ICD-9-CM codes 6420x, 6421x, 6422x, 6423x, 6424x, 6425x, and 6426x), preeclampsia (ICD-9-CM codes 6424x and 6425x), eclampsia (ICD-9-CM code 6426x), multiple gestation (ICD-9-CM code 651xx), post date pregnancy (pregnancy exceeding 40 weeks of gestation; ICD-9-CM codes 645, 64510, 64511, 64513, and 6453), placenta complications (ICD-9-CM codes 640 and 641), malpresentation (CCS code 187), fetal disproportion (CCS code 188), preterm delivery (delivery prior to 37 weeks of gestation; ICD-9-CM codes 6442, 64420, and 64421), and prior cesarean delivery (CCS code 189). We also controlled for hospital characteristics including bed size, teaching status, and rural versus urban location. We used the hospital bedsize categories defined by AHRQ.26 Hospital teaching status is based on information from the American Hospital Association’s Annual Survey of Hospitals. Finally, classification of hospitals as either urban or rural was based on Core Based Statistical Area codes from Census 2000 data.26
We constructed a series of multivariable regression models, with childbirth-related maternal hospitalization as the unit of analysis, to assess the effect of health insurance status on obstetric care from 2002 through 2009. Because units (hospitalizations) were clustered within hospitals, we used generalized estimating equations (GEE) models with clustered standard errors.32 We built GEE models with a log link for each of the dichotomous outcomes: cesarean delivery, VBAC, labor induction, and episiotomy. Final GEE regression models controlled for all of the individual demographic and clinical covariates and hospital characteristics described above and also included interaction terms between year and insurance status to test for differential annual trends in study outcomes. Our primary goal in this context was to examine whether trends (eg, rising rates of cesarean delivery) changed more or less rapidly depending on the primary payer.
In order to display time trends by insurance status, we also present results as predicted probabilities. Calculation of predicted probabilities required that we specify particular covarinate values. We based these specifications on mean covariate values (Table 1) to represent a typical childbirth-related hospitalization.The only factors that varied in the calculation of predicted probabilities were primary payer and year of childbirth. Predicted probabilities are calculated by using these covariate values and coefficients generated by the GEE regression models described above. All analyses were performed using SAS, version 9.2. The study was granted exemption from review by the University of Minnesota Institutional Review Board.
PDF is available on the last page.