Implementation of Mandatory Medicaid Managed Care in Missouri: Impacts for Pregnant Women

July 1, 2005
Anna Stauber Sommers, PhD
Anna Stauber Sommers, PhD

,
Genevieve M. Kenney, PhD
Genevieve M. Kenney, PhD

,
Lisa Dubay, ScM
Lisa Dubay, ScM

Volume 11, Issue 7

Objectives: To assess the impact of mandatory Medicaid managedcare in Missouri on prenatal care, maternal behavior, and lowbirth weight among pregnant women enrolled in Medicaid.

Study Design: Pre-post design using a comparison group withbirth certificate and Medicaid enrollment data in 1995 and 2000.

Methods: Pregnant women delivering in 38 counties that implementedmanaged care in Medicaid were compared preimplementationand postimplementation with pregnant womendelivering under Medicaid in 78 counties that remained fee-for-service(FFS) for separate samples of white (37 561) and black(13 640) non-Hispanic women. We calculated difference-in-differenceestimates using linear probability regression models thatcontrolled for maternal characteristics and time-invariant countydifferences. Analyses were stratified based on Medicaid enrollmentbefore and after conception, managed care region, andmarital status.

Results: Both managed care and FFS counties showed largeimprovements in prenatal care measures over time for both whiteand black women. Managed care was associated with a smallerpercentage point increase relative to FFS counties in adequate careamong whites of 1.9 and among blacks of 8.5, and a largerdecrease in smoking of 4.8 and in Women, Infants, and ChildrenProgram enrollment of 2.3 among white women. No pattern acrossmanaged care regions was found with respect to timing of implementation.Smaller effects were evident among black and singlewomen.

Conclusions: Although women experienced significantimprovements statewide in prenatal care under Medicaid,improvements were smaller for managed care counties. Managedcare may have a positive impact on smoking cessation, but otherpolicy changes may be needed to improve birth outcomes.

(Am J Manag Care. 2005;11:433-442)

Medicaid managed care enrollment has growndramatically since 1990. By 2000, more than50% of all Medicaid beneficiaries were enrolledin managed care, compared to less than 10% in 1991.1Understanding the implications of managed care onaccess to care and health behaviors for pregnant womenis of particular interest given that Medicaid now coversup to 50% of all births in some states.2

The movement to managed care under Medicaid hasthe potential to alter the timeliness and intensity of careprovided to pregnant women. For women enrolled inMedicaid after conception, the need to be assigned to ahealth maintenance organization (HMO) or primarycare physician may delay access to prenatal care relativeto a fee-for-service (FFS) setting. However, managedcare plans may provide assistance in finding aphysician and transport to appointments soon afterenrollment. Thus, how prenatal care use may changeunder managed care is difficult to anticipate.

Previous evaluations of effects of mandatoryMedicaid managed care on prenatal care use have beeninconclusive because of important methodologicalweaknesses, including small sample sizes,3,4 uncontrolleddifferences in risk between treated and untreatedgroups,5 endogenous independent predictors,6 andlack of comparison groups.7 Studies with the strongestmethodological approaches have yielded mixed results.

Two such studies relied on a small subset of countiesin single states.8,9 Findings varied based on the managedcare county examined. Levinson and Ullman foundHMO enrollment was associated with significantimprovements in adequacy of prenatal care,8 and Tai-Seale and colleagues found that longer enrollment timeprior to delivery was associated with more prenatal visitsin 2 California counties.9 Two other studies evaluatedTennessee's managed care program, but arrived atdifferent conclusions.10,11 Although both studies used apre-post design, they relied on different comparisongroups and methods.

In yet another study, researchers used the NationalNatality Files for the period 1990 through 1996 to estimatethe effect of the presence in a woman's county ofresidence of different types of Medicaid managed care.12A pre-post research design controlled for time-invariantcounty effects and time-varying state trends. The sample was stratified based on education and marital statusas proxies for Medicaid eligibility. The results suggestedthat among white non-Hispanic women, most types ofMedicaid managed care programs were associated witha 2% decrease in the number of prenatal visits and a 3%to 5% increase in the incidence of inadequate prenatalcare.12 A critical limitation of this study due to thenational sample was the inability to link specific implementationsettings with impacts.

In the only study using a cohort design, Howell et al13examined changes in outcomes between 2 pregnanciesto the same mothers in 10 Ohio counties during the period1993 through 1998. Outcomes for women in countiesimplementing mandatory HMO enrollment were comparedwith those for women in counties that retainedvoluntary enrollment across the same period. Authorsreported significant findings that were restricted toOhio's largest county, Cuyahoga. Mandatory enrollmentwas associated with significant worsening in the timingof initiation of care between pregnancies, but anincrease in the number of prenatal visits, declines inmaternal smoking, and no effect on infant birth weight.13Another Ohio study using a cross-sectional pre-postdesign for non-Hispanic white women also showedmandatory enrollment was associated with reductions inmaternal smoking and no impact on infant birth weight,but found improvements in prenatal care.14

Findings from these studies suggest that the effect ofmandatory managed care may be multifaceted. Time,place, and systems preceding mandatory enrollment, aswell as the managed care system itself, may all beimportant factors that limit the generalizability of anyevaluation. Because Medicaid policy is implemented atthe state level, multiple state-based studies that allowevaluation of specific policy choices and their impacton specific settings (eg, rural vs urban) and populations(single versus married women) may provide the mosteffective means for applying lessons to other comparablesettings and populations.

For the present study we focused on Missouri, a statethat moved directly from FFS to mandatory enrollmentand implemented Medicaid managed care in metropolitanand rural counties, which sets it apart from priorstudies. Missouri implemented mandatory managedcare in only some regions of the state, allowing otherregions of the state that retained the FFS coverage toserve as a comparison group.

BACKGROUND

Prior to the adoption of managed care, access toobstetric providers in Medicaid had been a concernthroughout Missouri. Reportedly, provider reimbursementunder Medicaid was low, little assistance wasavailable to help enrollees find providers, and fewwomen were referred to case management services. In1993, the state expanded eligibility for pregnant womenfrom 133% to 185% of poverty, and implemented aglobal rate structure that paid higher rates for deliveryif the woman had a minimum of 5 prenatal care visits.Such changes reportedly increased provider participationacross the state, and Medicaid caseloads subsequentlyrose.

Under a 1915(b) waiver, Missouri implementedmandatory HMO enrollment across 3 regions. Implementationin the Eastern, Central, and Western regionsoccurred in September 1995, March 1996, and January1997, respectively. The Eastern region included the cityof St. Louis and 4 surrounding counties. The Centralregion included 2 minor metro areas and 17 rural counties.The Western region included Kansas City and surroundingcounties. The remaining counties in the statehave FFS payment. They are primarily rural, butinclude 2 minor metro areas and 1 neonatal intensivecare unit. Payments to providers varied considerablyacross plans within regions, ranging from FFS paymentsto full capitation, suggesting no systematic pattern inthe incentives placed on providers. Given differences inurbanicity and plan financing, we would expect somevariation in managed care effects across regions.

Interviews with key informants suggested that HMOsmight have difficulty reaching pregnant women duringthe first trimester of pregnancy due to administrativeprocedures, especially for women who become Medicaideligible because of pregnancy. Under both managed careand FFS, these women must first show proof of pregnancyfor Medicaid eligibility determination. In additionto the eligibility determination process, managed carecoverage requires the additional step of assignment toan HMO. Enrollment in a managed care plan can bedelayed for other reasons: reportedly about one third ofenrollees are assigned to an HMO because they did notrespond to requests to choose one, and 1 plan reporteda 20% returned mail rate of plan enrollment packages.As a result of these delays, the opportunity for HMOs toencourage early initiation of prenatal care and minimizehigh-risk behaviors may be diminished for womenenrolled in Medicaid later in pregnancy. However, managedcare could lead to reductions in smoking: Statereports have indicated that 10 of 26 plans operating in1999 paid for smoking cessation programs and that, onaverage, 62% of smokers in HMOs were advised to quitby their doctors.15

Our research objective was to determine whether theimplementation of mandatory managed care enrollmentin Missouri's Medicaid managed care program affectedprenatal care use, maternal health behaviors, or lowbirth weight among pregnant women.

METHODS

We used a quasi-experimental pre-post design thatincluded multivariate analysis with county fixed effectsto assess the impact of movement to mandatoryMedicaid managed care in Missouri. Medicaid-coveredbirths in counties that implemented managed carewere the treatment group and their counterparts incounties that remained FFS Medicaid were the comparisongroup. We constructed baseline measures froma sample of live singleton Medicaid-covered births in1995 and postimplementation measures using births in2000. A site visit was conducted to collect informationon implementation.

Outcomes included prenatal care, maternal smoking,enrollment in the Women, Infants, and Children (WIC)Program, and low birth weight. Prenatal care was measured3 ways. Early initiation of care was a dichotomousmeasure equal to 1 if the first prenatal visit occurredduring the first trimester. Number of prenatal visits wasmeasured as a continuous variable. Because the totalnumber of visits was dependent on the length of gestation,we also included a measure of adequacy of careusing the Kotelchuck Adequacy of Prenatal CareUtilization (APNCU) Index. The APNCU Index incorporatestiming of care initiation, number of prenatal carevisits, and gestational age to classify the level of adequacyusing recommended guidelines for the expectednumber of visits based on gestational age.16,17 We set ourmeasure of adequate care equal to 1 if the woman initiatedcare during the first 4 months of pregnancy andreceived the expected number of visits given gestationalage, equivalent to the Adequate and Adequate Pluscategories in the APNCU Index.

All women were asked at the time of delivery if theysmoked during pregnancy, and the answer was recordedon the birth certificate. No other data were collectedon smoking behavior in Missouri. Smoking was measuredas a dichotomous variable equal to 1 if the motherreported smoking during pregnancy. All pregnantwomen on Medicaid are eligible for the WIC nutritionalprogram, so health plans may try to facilitate enrollmentin order to improve the nutritional status ofenrollees. Participation in the WIC Program was adichotomous variable equal to 1 if the mother reportedenrolling in WIC prior to delivery. Low birth weight wasmeasured as dichotomous and equal to 1 if birth weightwas less than 2500 g.

Our estimation strategy for all models was to includeonly factors that were exogenous to or predated prenatalcare initiation in the model. We included health risksthat are chronic and assumed to be known to thewoman prior to pregnancy and that may determine howearly she initiates care and the number of visits sought.We also included factors that were proxies for income,including education, marital status, and Medicaid eligibilitygroup. We excluded conditions that were likely tobe pregnancy-related, such as anemia. We used countylevelfixed effects to control for unmeasured variationacross counties constant over time, such as time costsfor obtaining prenatal care.

Our variable of interest was the managed care effect.Managed care status was defined for each woman basedon whether mandatory managed care was implementedin her county prior to pregnancy. We did not measureindividual managed care status because ofconcerns about selection bias. Women in the mandatorymanaged care counties could have received careunder FFS for their entire pregnancy, because exceptionswere made, for example, for women who enrolledin Medicaid late in pregnancy. Enrollment data suggestedthat between 8% and 28% of women in mandatorycounties were exempt from HMO enrollment in2000, depending on the region and eligibility category.This means that our estimates reflected the averageeffect of the policy intervention in Missouri for allwomen in managed care counties.

Managed care status for an individual woman in managedcare counties was dependent on the woman's timingof Medicaid enrollment relative to pregnancy onset.In turn, managed care effects on prenatal care andmaternal behaviors would depend on length of timeenrolled in an HMO. We expected managed care effectsto be weaker for women who had been enrolled in managedcare for shorter periods. Women enrolled inMedicaid before pregnancy (preconception) shouldalready be enrolled in a health plan in the firsttrimester, allowing enhanced services to be deliveredearly in pregnancy, and allowing managed care to havethe greatest impact. In contrast, women enrolled inMedicaid postconception would enroll in an HMO laterduring pregnancy, resulting in weaker managed careinterventions. Therefore, we stratified analysis bywomen enrolled in Medicaid preconception and postconceptionas a sensitivity test for the managed careeffect. One caveat is that only poor women eligible forMedicaid through the Aid to Families with DependentChildren or Temporary Assistance to Needy Families(AFDC/TANF) programs can enroll before conception,whereas near-poor women who are eligible only duringpregnancy can enroll only after conception. About threeMandatory Medicaid Managed Care and Pregnancyfourths of pregnant women in our sample were enrolledafter conception.

All models were estimated using linear probabilityregression with county fixed effects adjusted for clusteringby county in Stata 7.0 (StataCorp LP, CollegeStation, Tex). The Huber-White sandwich estimate ofvariance was applied to account for heteroscedasticity.This study was approved in advance by the institutionalreview board of The Urban Institute.

Data

We used Missouri birth certificate data linked withMedicaid enrollment data to obtain a census of all livesingleton births by women enrolled in Medicaid at thetime of delivery. We restricted our analysis to womenenrolled on date of delivery through eligibility related toAFDC/TANF, or Transitional Medicaid Assistance; andstate expansions to pregnant women at higher incomes.The sample was analyzed separately for non-Hispanicwhite and black women (referred to as white and blackwomen). When interpreting findings for black women,the small sample in FFS areas must be considered. Asmall number of Hispanic women and women of otherminorities (N = 2788) were excluded due to the lack ofsufficient sample for stratified analysis. The analysissample of 51 201 is described in Table 1.

Independent Variables

A complete list of variables, sample means, andchanges in means are presented in Table 2 for whitewomen and Table 3 for black women. We included thefollowing independent variables in regression: interactionsof maternal marital status and educationlevel; Medicaid enrollment history during pregnancy;maternal age; born in the United States; maternalchronic risk factors; parity risk level; and missingdata indicators.

 We combined data on Medicaid enrollment timing relativeto pregnancy onset and eligibility route to construct4 categories: enrolled preconception and eligiblethrough AFDC/TANF at time of delivery; enrolled postconceptionand eligible through AFDC/TANF at delivery;enrolled preconception and eligible through state povertyexpansion at time of delivery; and enrolled postconceptionand eligible through state poverty expansion(reference). This construction was used to control fordifferences between poor women continuously enrolledfor long periods unrelated to pregnancy and near-poorwomen who enrolled during pregnancy. Parity risk levelwas defined as low if this delivery was a first birth, mediumif a second birth at any age or a third birth at age 25years or older, and high parity if a third birth at age lessthan 25 years or a fourth birth at any age.

Estimation Issues

The validity of the pre-post design hinges on theidentification of a comparison group that serves as acounterfactual for what would have happened to thetreatment group in the absence of mandatory managedcare. Because we have controlled for county-specificcross-sectional differences through fixed county effects,the threat to validity of our estimates rested in differencesacross counties in time-varying unobserved factorsthat were correlated with study outcomes. Ourprimary concern was that Medicaid managed careregions were more urban than regions that remainedFFS, and that changes occurring during our study periodrelated to the expanding economy and welfarereform may have affected these areas differently.

Significant changes in the composition of the groupsaccording to marital/education and Medicaid eligibilitystatus suggested that differential changes occurred inmanaged care and FFS counties (Tables 2 and 3). Weaddressed this potential threat to validity by conductingsensitivity analyses. We estimated separate effects forthe 3 managed care regions and stratified results bymarital status for the white and black samples.

RESULTS

P

Between 1995 and 2000, the percentage of whitewomen initiating prenatal care during their firsttrimester increased 4.0 percentage points in managedcare counties from 78% in 1995 to 82% in 2000, and 3.8percentage points in FFS counties from 79% in 1995 to82.8% in 2000 (data not shown). Likewise, the percentageof white women with adequate prenatal careincreased 4.4 percentage points in managed care areasand 5.5 percentage points in FFS areas (< .05). Themean number of visits for FFS areas increased significantly,albeit slightly. We found similar trends amongblack women, who experienced lower prenatal care useMandatory Medicaid Managed Care and Pregnancyat baseline, but larger increases in prenatal care overalland greater differences between managed care and FFSareas.

We found a significant decline in smoking from 1995to 2000 for white women in managed care counties of1.5 percentage points from 40% in 1995, compared to anincrease of 3.2 percentage points in smoking in FFScounties from 35%. Participation in WIC remained stablein managed care areas and dropped in FFS areas by2.4 percentage points. Among black women, we foundsimilar trends but no decrease in smoking in managedcare counties. Trends in low birth weight were not significantfor either sample.

Multivariate results are presented for white and blackwomen, including stratification by enrollment inMedicaid preconception and postconception. In theregression context, the period effect ("Year 2000" inTables 4-6) captures the average effect of time-varyingfactors between 1995 and 2000 across women in allcounties. The managed care effect ("MC × 2000") is thedifferential impact of implementation on the women inmanaged care counties relative to FFS counties.

P

There was a strong positive period effect of .048among white women for early initiation of care (< .01),meaning an average increase of 4.8 percentage pointsstatewide (Table 4). But the managed care effect wasnot significant. This finding suggests a statewide trendtoward earlier prenatal care initiation in Medicaidunrelated to managed care implementation.

Models also showed substantial increases in adequatecare and number of visits across both managedcare and FFS areas, but these increases were smaller inmanaged care areas. Among white women, managedcare was associated with a 1.9 percentage point smallerincrease in adequate care (based on a managed careeffect of -.019), and a smaller increase in visits by-0.16 visits relative to FFS. The managed care effect forvisits is not interpreted as a percentage point changebecause the dependent variable is continous. Prenatalcare models among black women showed large increasesin adequate care in both FFS and managed careregions, but an 8.5 percentage point smaller increase inmanaged care counties (Table 5).

For women enrolled in Medicaid before and afterconception (Tables 4 and 5), there were no differentialeffects between managed care and FFS on early initiationor visits, but significant differences in the effect onadequate care. Managed care showed a significantlysmaller increase in adequate care for white womenenrolling postconception (-.028), and for black womenenrolled preconception (-.089) relative to FFS. Furtheranalysis showed that FFS counties experiencedincreases twice as large as managed care counties inthe proportion of white and black women who receivedthe recommended number of visits among those initiatingcare early in pregnancy (data not shown). In summary,managed care does not appear to give women anadvantage in obtaining adequate care or receiving theMandatory Medicaid Managed Care and Pregnancyrecommended number of visits in Missouri. These findingsare consistent with those found nationally byKaestner et al12 but are not consistent with positive prenatalcare findings for managed care found by Kenney etal14 in Ohio.

As shown in Table 4, managed care was associatedwith a 4.8 percentage point relative reduction in smokingduring pregnancy among Medicaid-covered whitewomen. The smoking result must be treated cautiously.First, women may have underreported smoking. Second,we do not know if women who responded "no," stoppedsmoking or never smoked. Finally, it is uncertainwhether women were reporting their smoking behaviorthroughout pregnancy or simply at the time of delivery.HMOs cannot reduce the number of women enrolledpostconception who smoked at any point during pregnancyif they were smoking prior to enrollment.Therefore, we would expect women enrolled preconceptionto account for most of the managed care effect, buteffects among women enrolled postconception could stillexist, under the assumption that some mothers reportedtheir smoking status at the time of delivery. In stratifiedmodels for white women, reductions in smoking in managedcare areas relative to FFS areas were evident forboth preconception and postconception groups, but theeffect was almost twice as large (-.080 versus -.042) forthe preconception group (Table 4). Thus, managed careappears to have reduced maternal smoking among whitewomen, particularly those enrolled preconception. Theseresults are consistent with findings by Kenney et al.14 andHowell et al.13 in Ohio. Among black women, the managedcare effect was significant and large only for thepostconception group (-.011 versus -.062), a findingthat runs counter to our hypothesis (Table 5). Becausesmoking is half as prevalent among black women, interventionsmay have smaller effects relative to whitewomen.

P

P

We found significant increases in WIC enrollment formanaged care counties compared to FFS counties forwhite women. On average, managed care was associatedwith a 2.3 percentage point increase in the proportionenrolling in WIC (Table 4; < .01). Since WICenrollment could occuranytime prior to delivery,health plans could havean impact for womenenrolling both preconceptionand postconception.We found significant andpositive managed careeffects on WIC for thepostconception group,and an effect of similarmagnitude for the preconceptiongroup (< .10).No effect was observedamong black women, butWIC enrollment was higheramong blacks thanwhites at baseline. Weconclude that the managedcare effect on WICenrollment was positive,but note that it appearsmore driven by declinesin FFS areas than byincreases in managedcare areas.

No managed careeffects were found on lowbirth weight for the whiteor black sample. We concludethat managed carehad an insufficient effecton smoking to reduce thenumber of low birthweightbirths, and aninsufficient effect on prenatalcare to increasethem. These findings areconsistent with bothOhio studies.13,14

To test the sensitivityof findings to differentpopulations, we conducted additional stratified analysis.Models stratified by marital status were consistentwith findings for white and black women overall (datanot shown). Single white women in managed carecounties showed smaller increases in both adequatecare and visits relative to FFS. Positive managed careimpacts on smoking and WIC were found for both singleand married white women, but the WIC findingswere not significant at conventional levels.

Models estimating separate effects for the 3 managedcare regions showed no consistent pattern with respectto the timing of implementation (Table 6). The metropolitanEastern managed care region, which implementedmanaged care first, showed significantly smallerincreases in visits and larger reductions in smoking forwhite women compared to FFS. The primarily ruralCentral managed care region showed smaller increasesrelative to FFS in adequate care and number of visitsand larger increases in WIC enrollment. Black womenare heavily concentrated in the East and West, and wefound large decreases in adequate care for black womenin the East and in adequate care and visits in the West.Mandatory Medicaid Managed Care and PregnancyNo managed care region showed significant positivemanaged care effects for prenatal care or negative effectson smoking or WIC enrollment. Only the East showed asignificant increase in low birth weight, despite largerreductions in smoking. These findings suggest that variationsin delivery systems, populations, and capitationmethods yielded some variation in specific managed careeffects, but little difference in the overall impact.

CONCLUSIONS

Managed care in Missouri's Medicaid program mayhave reduced the prevalence of maternal smoking andincreased WIC enrollment among white women, butefforts did not lead to reductions in the prevalence oflow birth weight infants. This finding suggests thatother policy changes are needed to promote betterbirth outcomes.

Statewide, both black and white Medicaid womenexperienced large and positive trends in prenatal careuse between 1995 and 2000, but improvements weresmaller in managed care than in FFS areas. Delays inenrollment into HMOs were reported for some pregnantwomen in both our Missouri and Ohio site visits.18 Suchdelays could lead to barriers to care at worst, and lostopportunities for intervention at best. Further improvementsin prenatal care may be hindered by eligibilitypolicies that limit Medicaid enrollment until pregnancyfor many women.

Overall, this study indicates that black women inmandatory managed care counties may have experiencedsmaller prenatal care gains relative to those inFFS counties, and no gains with respect to smoking cessationor WIC enrollment. As indicated earlier, thesmall sample sizes of black women in FFS areas meansthat these findings need to be treated very cautiously.Enrollment in specific managed care plans and planquality may vary by race, however. Larger samples ofblack women linked to more detailed enrollment datacould elucidate this issue.

Finally, among individuals covered by Missouri'sMedicaid program in 2000, one third of black womenstill did not receive adequate prenatal care based onrecommended guidelines, and almost 40% of whitewomen reported that they smoked during their pregnancy.It may be necessary for policy makers to implementother interventions to improve the health oflow-income women and their infants.

Acknowledgments

We acknowledge Robert Kaestner, PhD, and Embry Howell, PhD, fortheir input; Ruth Almeida, MPP, for her assistance with the site visit and fileconstruction; state officials for providing data for the study and for sharingtheir expertise and insights, and other key informants in Missouri.

From The Urban Institute, Washington, DC.

This research was funded by The Robert Wood Johnson Foundation's Health CareFinancing and Organization Initiative, grant #032105. All opinions expressed herein are theopinions of the authors and should not be attributed to The Robert Wood JohnsonFoundation, The Urban Institute, or its funders.

Address correspondence to: Anna Sommers, PhD, The Urban Institute, 2100 M StreetNW, Washington, DC 20037. E-mail: asommers@ui.urban.org.

A Profile of Medicaid: Chartbook 2000.

1. Health Care Financing Administration. Washington, DC: US Department of Health and Human Services; 2000.

Medicaid Coverage of Perinatal Services: Results of a National Survey.

2. Schwalberg R, Mathis SA, Giffen M, Mohamadi L, Zimmerman B, Sines E.Menlo Park,Calif: Henry J. Kaiser Family Foundation; November 2001.

Health Serv Res.

3. Carey TS, Weis K, Homer C. Prepaid versus traditional Medicaid plans: lack ofeffect on pregnancy outcomes and prenatal care. 1991;26:165-181.

Med Care.

4. Goldfarb NI, Hillman AL, Eisenberg JM, Kelley MA, Cohen AV, Dellheim M.Impact of a mandatory Medicaid case management program on prenatal care andbirth outcomes. A retrospective analysis. 1991;29:64-71.

Am J Public Health.

5. Schulman ED, Sheriff DJ, Momany ET. Primary care case management and birthoutcomes in the Iowa Medicaid program. 1997;87:80-84.

Health Serv Res.

6. Oleske DM, Branca ML, Schmidt JB, Ferguson R, Linn ES. A comparison ofcapitated and fee-for-service Medicaid reimbursement methods on pregnancy outcomes.1998;33:55-73.

Am J Public Health.

7. Griffin JF, Hogan JW, Buechner JS, Leddy TM. The effect of a Medicaid managedcare program on the adequacy of prenatal care utilization in Rhode Island.1999;89:497-501.

J Health Econ.

8. Levinson A, Ullman F. Medicaid managed care and infant health. 1998;17:351-368.

Health Serv

Res.

9. Tai-Seale M, LoSasso AT, Freund DA, Gerber SE. The long-term effects ofMedicaid managed care on obstetric care in three California counties. 2001;36:751-771.

The Influence of TennCare on Perinatal Outcomes. Perspectives on

Medicaid Managed Care.

10. Moreno L. Princeton, NJ: Mathematica Policy Research, Inc; June1999.

J Health Polit Policy Law.

11. Conover CJ, Rankin PJ, Sloan FA. Effects of Tennessee Medicaid managed careon obstetrical care and birth outcomes. 2001;26:1291-1324.

Soc Sci Med.

12. Kaestner R, Dubay L, Kenney G. Managed care and infant health: an evaluationof Medicaid in the US. 2005;60:1815-1833.

Health Serv Res.

13. Howell EM, Dubay L, Kenney G, Sommers AS. The impact of Medicaid managedcare on pregnant women in Ohio: a cohort analysis. 2004;39:825-846.

Med Care.

14. Kenney GM, Sommers AS, Dubay L. Moving to mandatory Medicaid managedcare in Ohio: impacts for pregnant women and infants. In press.

"Show

Me . . ." Buyer's Guide: Commercial Managed Care Plans &#8216;99.

15. Missouri Department of Health, Missouri Center for Health Statistics. Jefferson City, Mo:Missouri Department of Health; 1999. Publication no. 215.

Am J Public Health.

16. Kotelchuck M. An evaluation of the Kessner Adequacy of Prenatal Care Indexand a proposed Adequacy of Prenatal Care Utilization Index. 1994;84:1414-1420.

Adequacy of Prenatal Care Utilization Index

17. Kotelchuck M. [SAS computerprogram, version 3]. Chapel Hill, NC: Department of Maternal and Child Health,The University of North Carolina at Chapel Hill; September 1994. Available at:http://www.mchlibrary.info/databases/HSNRCPDFs/APNCU994_20SAS.pdf.Accessed March 8, 2005.

Manag Care Interface.

18. Sommers AS, Howell EM. The implementation of Medicaid managed care forpregnant women in Ohio: a case study. 2003;16(10):27-31,34.