Maternal characteristics associated with adequacy of prenatal care utilization on the birth file were age, race, education, marital status, parity, insurance status, ...
The Effect of a Medicaid Managed Care Program on the Adequacy of Prenatal Care Utilization in Rhode Island
Jane E Griffin, MPH, Joseph W Hogan, ScD, Jav S. Bluechne,; PhD, and Tricia M. LeddA, MS In the past 10 years in the United States eligibility for the Medicaid program has dramatically increased for pregnant women. It is estimated that from 1979 to 1992 the percentage of pregnant women aged 15 through 44 years who were eligible for Medicaid increased from 12.4% to 43.3%.1 The major purpose of these expansions was to improve prenatal care for low-income women. However, early evaluation studies of the effect of Medicaid expansions and Medicaid managed care have shown little improvement in prenatal care for pregnant women on Medicaid. Of 5 statewide studies that measured the effect of Medicaid expansions on improvements in prenatal care,5- 3 failed to show that Medicaid expansions improved initiation into prenatal care as measured by entry into care in the first trimester.)- Two others failed to show improvement in adequacy of prenatal care, as measured by the Kessner Index, after implementation of Medicaid managed care.8"o Kreiger et al., in their evaluation of a statewide program in Washington, found no significant difference in adequacy of prenatal care between pregnant women enrolled in a managed care health plan and those enrolled in fee-forservice plans.8 Schulman et al., in their evaluation of Iowa's state Medicaid program, found that pregnant women enrolled in a fee-for-service Medicaid program received better prenatal care than pregnant women enrolled in a primary care case management program.9 There was concurrence among these studies that simply expanding Medicaid coverage to more low-income women was not enough to improve prenatal care. The Rite Care program, Rhode Island's managed care program for Medicaid families and uninsured pregnant women and children, was implemented in August 1994 under a Health Care Financing Administration section II15 research and demonstration
waiver." Beginning in August 1994, through state legislation and the waiver mechanism, RIte Care enrolled all Medicaid families (excluding aged participants and participants with disabilities) into their choice of 5 health plans. The waiver also allowed Rhode Island to expand coverage through RIte Care to uninsured pregnant women and children up to age 6 years with annual family incomes up to 250% of the poverty level (this amount is currently $40 080 for a family of 4 in Rhode Island).
The state changed the health care financing system under Rlte Care from a fee-for-service payment system, in which families had to find providers who would accept Medicaid, to a system in which the state contracts for a set of services and pays health plans a monthly capitation payment. Rite Care beneficiaries enroll in their choice of 5 licensed health plans. These health plans are responsible for ensuring that each program participant has a "medical home," which is a primary care physician who coordinates all of the member's health care. RIte Care's primary care physicians practice in private office settings, community health centers, hospital clinics, and a staff model health maintenance organization. The purpose of this study was to determine whether the adequacy of prenatal care
Jane F. Griffin is with MCH Exaluation, Inc, Barrington, RI. Joseph W Hogan is Nvith the Center for Statistical Sciences. Brown Unixersityv Prox idence, RI. Jay S. Buechner is with the Office of Health Statistics, Rhode Island Department of Health, Providence. Tricia M. Leddv is with the Center for Child and Family Health, Rhode Island Department of Human Services, Cranston. Requests for reprints should be sent to Jane F. Griffin, MPH, MCH Evaluation, Inc, 14 Bullock Ave. Barrington, RI 02806 (e-mail:
jane_griffin(ci brow-n.edu).
This paper wxas accepted October 1, 1998.
American Journal of Public Health 497
Griffin et al.
utilization improved in 3 groups of patients after the implementation of the RIte Care Medicaid managed care program. The 3 groups were Medicaid patients; Medicaid patients receiving care in private practice settings, at community health centers, and at hospital clinics; and privately insured patients.
Methods Sample The sample included data for 39065 Rhode Island resident births occurring in Rhode Island for calendar years 1993, 1994, and 1995. Birth certificate data were provided by the Office of Vital Records, Rhode Island Department of Health. Each variable used in this study was self-reported on the birth worksheet by the mother in the hospital after delivery. The number of missing items is low, ranging from 0% to 6.9% ofthe total. Excluded from the original sample of 39065 were 980 births in which insurance status was missing and 439 births to women with no insurance; in addition, 625 multiple births (i.e., twins and triplets) were identified, and these were counted only once because the outcome of interest, adequacy of prenatal care utilization, related to one pregnancy. After these exclusions, the analytic database contained 37 021 birth records.
Explanatory Variables Maternal characteristics associated with adequacy of prenatal care utilization on the birth file were age, race, education, marital status, parity, insurance status, and type of practice setting. 2-16 These characterisfics are important in analysis to rule out alternative explanations for changes in adequacy of prenatal care utilization. For example, if prenatal care improves, it is important to control for the effects of demographic changes within the population during the time period observed. Explanatory variables were categorized as follows. Four categories of race and ethnicity were used: White, non-Hispanic; Black, non-Hispanic; Hispanic; and other. Hispanic women make up the largest minority group enrolled in Medicaid in Rhode Island. Age was dichotomized as younger than 18 years vs 18 years and older; education as less than 12 years vs 12 or more years; marital status as unmarried vs married; and parity as fewer than 3 children vs 3 or more children. Insurance status was categorized as Medicaid or private. Categories for type of practice setting were private 498 American Journal of Public Health
physician's office (primarily obstetricians in private practice), community health center, hospital clinic, and staff model health maintenance organization.
CharacterizingAdequacy ofPrenatal Care Utilization The major dependent variable or outcome for this study was adequacy ofprenatal care utilization. Kotelchuck's Adequacy of Prenatal Care Utilization Index'7 was used for this analysis; the index is calculated on the basis of month prenatal care began, number of prenatal care visits, and gestational age. Prenatal care utilization was dichotomized as inadequate vs a for analysis in this study. The category inadequate care combined Kotelchuck's inadequate and intermediate categories and was defined as (1) prenatal care begun after the fourth month of pregnancy; (2) prenatal care with fewer than 50% of recommended visits completed, regardless of when care began; or (3) prenatal care begun by the fourth month of pregnancy and only 50% to 79% of recommended visits completed. The category adequate care combined Kotelchuck's adequate and adequate-plus categories and was defined as prenatal care begun by the fourth month of pregnancy and 80% or more of recommended visits completed. Kotelchuck's index improves on other measures of prenatal care adequacy because it controls for timing of entry into care and gestational age at delivery. Adequacy of prenatal care was selected as the outcome variable because it is a specific performance measure of the RIte Care program, it has long been recognized as a way to identify health and social problems among pregnant women, and it has been used as a mechanism for identifying women at risk for poor pregnancy outcomes.18 Prenatal care provides a standard measure of access to care by using indicators collected routinely on all birth certificates in the United States-month of initiation of care and number of prenatal visits.19 These 2 indicators, along with gestational age, create an adequacy of prenatal care utilization index that measures the adequacy of prenatal care a woman receives.
StatisticalAnalysis We used logistic regression to model the probability of receiving adequate prenatal care, and we used the model to assess whether the odds of receiving adequate prenatal care had improved for Medicaid patients after implementation of the program. Relative change in adequacy was measured
by an odds ratio (OR; odds post-RIte Care/odds pre-RIte Care), adjusted for the following categorical variables (categories defined above): race/ethnicity, marital status, age, education, parity, and type of practice setting. Temporal variations were modeled with categorical variables corresponding to each quarter year during the study period (1993-1995). Relative change in adequacy was estimated separately for Medicaid and private insurance patients by means of an interaction term. The private insurance patients are included in the model so that the effect of the RIte Care program can be separated from any underlying temporal changes in adequacy of prenatal care (after controlling for the effects of the explanatory variables listed above). Pairwise associations between covariates were examined to identify potential collinearities, and none were found. Using a likelihood ratio test, we found no evidence of underlying temporal changes over and above the RIte Care effect. Thus, our inferences are based on a model that assumes that within insurance type (and conditional on covariates), the percentage of adequate prenatal care utilization is constant across time, with a shift after implementation of the program. To assess the impact of RIte Care, we used a likelihood ratio test to compare the adjusted relative change (odds ratio) for pregnant women on Medicaid with that for pregnant women who had private insurance. The RIte Care effect was tested overall and separately by type of practice setting. All analyses were carried out with SAS version 6.11.20 Logistic regression models were fit with Proc GENMOD (SAS, Cary, NC).
Results Characteristics ofBirths to Medicaid Patients Table 1 shows the characteristics of pregnant women who gave birth in Rhode Island from 1993 through 1995 by insurance status. The characteristics shown in Table 1 are the study covariates and show the distribution of these explanatory variables. Pregnant women on Medicaid represented 32.4% of all Rhode Island women giving birth during the study period. Pregnant women receiving Medicaid were statistically significantly more likely than pregnant women with private insurance to be non-White, to be unmarried, to be younger than 18 years, to have less than a high school education, and to have 3 or more children. Table 2 shows the distribution of types of practice setting in which prenatal care was April 1999, Vol. 89, No. 4
Prenatal Care Utilization
received before and after the implementation of RIte Care. The proportion of Medicaid patients who went to private physicians increased after the implementation of RIte Care (P