J CUmE#dmbl Vol. 43, No. 11, pp. 12134220, 1990 printedin GreatBritah. Al1right~merved

0895-4356/90 s3.00+ 0.00 Copyright@J1990PergamonRem pk

PREPAID VERSUS TRADITIONAL MEDICAID PLANS: EFFECTS ON PREVENTIVE HEALTH CARE* TIM CAEEy,*t KATHI Wnrs2 and CHARLES HOMER~ Departments of ’General Medicine and Clinical Epidemiology and *Healtb Policy and Administration, University of North Carogna at Chapel Hill, Chapel Hill, NC 27599-7110aad )Department of Pediatrics, Massachusetts General Hospital, Boston, Mass., U.S.A. (Receiued in reuisedform 15 March 1990)

Abstrac-Prepaid, case managed systems have been proposed as a method of controlling costs in Medicaid populations. We investigated the utilixation of preventive services in two prepaid Medicaid Competition Demonstration programs in Santa Barbara County, Calif., and Jackson County, Mo. (containing the city of Kansas City). Care in the demonstration sites was compared with care given in similar counties functioning under a traditional fee-for-service Medicaid system-Ventura County, Calif., and St Louis, Mo. We tested the hypothesis that preventive care would be less in the capitated demonstrations. 2735 Children’s and 3389 adult’s charts were abstracted for care received during the calendar year 1985, after the prepaid demonstration had been in place for more than 1 year. NO significant differences were found between the demonstration and comparison counties in the proportion of children with complete DPT or OPV immunixations at 1 year of age, with 56% complete in both Califomia counties and 69 and 65% complete in Jackson Coanty and St Louis, respectively. Regression analysis demonstrated a slight, but statistically significant trend towards more immunixations in the demonstration counties. Pap smear use in women of 15-44 years of age was little different in the Califomia counties, but significantly greater in the Jackson County demonstration in Missouri (64 VS 45%). Physician breast examinations were somewhat more likely to occur in the prepaid, case managed demonstration counties. Capitated, case managed systems for the AFDC Medicaid population appear to result in no diminution of preventive services. Substantial problems exist in this, as in ether

poor populations, in childhood immunizations. Prepaid VS traditional Medicaid Preventive organization Capitation Medicaid

INTRODUCTION

Total federal health care expenditures quadrupled between 1970 and 1980, and total state and local ‘health outlays more than tripled over this period. The Medicaid outlays rose from *Supported through funding by the Health Care Financing Administration, Contract No. HCFA-50-83-500. The opinions expres& in this article represent those of the authon and not neuwarily these of the Health Care Financing Administration. tAuthor for cmrespondence.

health care

Health maintenance

$12.2 billion in 1975 to $37.5 billion in 1985 [ll. One approach chosen by govemment to address this tost increase has ‘been to enroll Medicaid recipients in managed care systems, utilizing the mechanisms of case management and prepaid, capitated payment to providers [2]. The health care costs of patients enrolkd in health maintenante organizations are substantially lower compared with fee-for-service systems [3]. The reduced costs are largely due to decreased hospitalixation. Quality of care and health outcomes

1213

TIM CAREYet al.

1214

in prepaid systems have generally been viewed as equivalent to the fee-for-service sector [4]. Most research has, however, studied middleclass populations, and concern has been expressed that such studies may not be generalizable to poor populations. One study in the U.S.A. demonstrated a trend for worsened health outcomes among individuals in lower income groups [5]. To address these concerns, the Health Care Financing Administration conducted an evaluation of the Medicaid Competition Demonstrations. The demonstrations consisted of 7 programs in 6 states that combined elements of capitated payment to primary care providers, case management and mandatory enrollment in the prepaid system [6]. The evaluation included financial, patient satisfaction, administrative and quality of care components. Here we present results from consumer survey and quality of care chart abstraction components of the evaluation. Preventive care is of particular interest in low socioeconomic status populations. Poor children are at greater risk for problems with growth, development and incomplete infectious disease immunizations [7]. We specifically tested the hypothesis that less preventive care would be provided in the capitated, case managed demonstration projects, in which the provider has no monetary incentive to see the patient on a regular basis. METHODS

A cross-sectional comparison of care given to the iargest Medicaid group, AFDC mothers and children, was examined in the two most mature Medicaid Competition Demonstrations. Care in the demonstration site of Santa Barbara County, Calif., was compared with care given in the adjacent Ventura County, which was

operating its Medicaid system under a traditional fee-for-service system. The demonstration county in Missouri was Jackson County, containing the city of Kansas City. Jackson County was compared with the city of St Louis. Al1 AFDC recipients in both prepaid demonstration counties were enrolled in the prepaid, case managed demonstration, although they were given a choice as to which participating provider was their case manager. The providers of care in the demonstration plans in both California and Missouri were essentially the same as the providers of care prior to the institution of the demonstrations in 1983. Almost al1 providers at al1 four sites were board eligible or certified in their specialities. NO differente was found in physician age between demonstration and comparison counties. Providers in the demonstrations received a fixed fee on a monthly basis to provide primary care and coordinate referred care. Some element of financial risk was present for the demonstration providers. The calendar year 1985 was used for the quality of care study. Both demonstrations had been in operation for at least 1 year prior to the evaluation year of 1985. The Medicaid populations and medical systems in the two California counties were similar, with the exception that Ventura County had a county hospital. Both Santa Barbara and Ventura Counties have large networks of county clinics, which provide both primary and some specialty care. As demonstrated in Table 1, the population of Medicaid recipients in Santra Barbara was smaller than in Ventura County; the numbers of providers sampled was also smaller (68 VS 96). Care given in the demonstration site of Jackson County, Mo. (containing the city of Kansas City) was compared with fee-for-service care in St Louis city, the other major urban area in Missouri. While Jackson

Table 1. Sampling strategy Santa Barbara (prepaid) Stratum Child outpatient Wel1 child care Otitis media Al1 other care Adult outpatient Hypertension Prenatal care-no claim submitted Prenatal care-claim submitted Urinary tract infection Pelvic inflarnmatory disease FFS = fee-for-service.

Ventura (FFS)

Population

Sample

Population

Sample

105 511 2813

105 285 421

684 1729 5827

250 250 250

45 291 422 267 349

45 143 286 267 349

188 510 496 481 840

125 125 250 250 375

Prepaid VSTraditional MdicaidPlans

County does contain some iural areas, essentially all of the AFDC population is contained within urban Kansas City. Health care for Medicaid recipients in both Missouri cities relied heavily on public clinics and neighborhood health centers. Care in St Louis was distributed among a larger number of sampled Medicaid providers than in Jackson County (59 VS 21) reflecting a larger population base (Table 2). Jackson County did have two very large neighborhood health centers; care in St Louis was distributed among smaller centers and a number of public clinics. Two data sources were used. Primary data was collected through personal interviews of a stratified random sample of Medicaid demonstration enrollees in Santa Barbara County, Calif., and Jackson County, Mo., and a similarly defìned sample of nondemonstration Medicaid enrollees in the comparison sites of Ventura County, Calif., and, St Louis city, Mo. The interviews were conducted in spring 1986. The overall response rate was 91.7% in the demonstration sites and 92.$% in the comparison site. A second portion of the study involved abstraction of outpatient medical charts. Patients were randomly sampled from Medicaid claims files, which were used to find diagnoses of interest. Multiple conditions were sampled. Sample sizes for each diagnostic stratum are presented in Tables 1 and 2. Claims were then linked with Medicaid eligibility files to assure the appropriate eligibility category. Length of eligibility for the sample difIered between the two study states. Due to the small population in the Califomia sites, individuals with any AFDC cash assistance benefits were sampled. Details of the sampling and abstraction techniques used are presented elsewhere [8].Eighty-three percent of outpatient providers approached cooperated

1215

with the study. Al1 charts sampled were reviewed for evidente of preventive care obtained or documented in 1985. Chart abstractions were performed on site by trained personnel, either registered nurses or medical record technicians hired for the study. Abstracters were not capable of being blinded as to the study intervention, although they were blinded as to the major hypothesized study outcomes. Immunizations or other preventive measures were considered to have been given if there was documentation of their having been performed at the site of the abstraction or if there was a chart notation that they had been administered elsewhere. Completed forms were mailed to the Research Triangle Institute for editing and coding. Since Mediid eligibility requirements varied from state to state, al1 analyses were between the demonstration and comparison counties within each state (California or Missouri). Reliability of the chart abstraction instruments was 90% for medical information for the outpatiênt abstraction forms. Weights were applied to the sample to account for sampling effect, since stratum size was unequal between demonstration and comparison sites in many instances. The analysis was performed using a statistical package (RATIOEST, RTI-SAS) specifically modXed for analysis of weighted survey data. t-Tests were used to test diíferences between means. Linear and logistic regression were used for control of multiple potential confounding variables. RESULTS

Overall, 71% of adult and 82% of childrens’ sampled charts Were abstracted, with a range from 65 to 88% (Table 3). Reasons for inability to complete an abstraction included provider

Table2. Samplbqstrategy Jackson (prepaid) Stratum Child outpatient Wel1 child care Otitis media Otitis media + well child care All other care Adult outpatient Hypertension Prenatal care-no claim submitted Prenatal claim submitted UrinaryMammatoty Pelvic tract infection discase FFS = fee-for-service.

St Louis (FFS)

Population

Sample

Population

Sample

1300 858 823 1995

261 172 164 299

4871 4525 1661 15,608

256 239 89 292

195 305 829

195 149 299

926 1364 4285

292 142 293

435 509

448 157

1756 1788

::

TIM CAmY er al.

1216

Table 3. Sample sizes for preventive care chart abstraction studies Site

Children’s charts located (% sampled)

Adult charts located (% sampled)

115 (88) 615 (82) 763 (85) 642 (72)

822 (75.4) 817 (72.6) 900 (72.1) 850 (64.8)

Santa Barbara (prepaid) Ventura (FFS) Jaclcson (prepaid) St Louis (FFS) FFS = fee-for-service.

noncooperation, inability to locate the provider, inability to locate the chart at the office or clinic and dates of service falling out of the sampling frame.

and 42% in the Califomia sites and 58 and 52% at the Missouri prepaid and fee-for-service sites, respectively (p = 0.04). Logistic regression was used to control for some of the multiple potential confounding factors present in this type of cross-sectional study (Table 6). After adjustment, children in the demonstration sites were slightly more likely to have completed immunizations, although the odds ratios reflecting increased likelihood were relatively modest, 1.08 and 1.05. Children were more likely to have completed immunizations if they had more outpatient visits, and less likely if the chart abstracted was from an emergency room or outpatient clinic. “Outpatient clinic” designates hospita1 ambulatory clinics or neighborhood health centers. The reference category was solo or group physician offices. Measles, mumps and rubella (MMR) immunizations were documented in a smaller proportion of children. Such immunizations were recorded more commonly in Jackson County compared with St Louis. Hematocrit screening for anemia was recorded in only a minority of children in this high-risk population, occurring more frequently in the demonstration county of Santa Barbara, Calif., and the comparison site of St Louis, Mo. Growth parameters (height, weight and head circumference) are used to detect failure to thrive and neurologie conditions of childhood. Such parameters were performed more com-

Childhood preventive care The characteristics of the children whose care was abstracted are presented in Table 4. The mean age was 2 years in Califomia and 1.5 years in Missouri. Due to the smaller population under study in California, the age range studied was 0-4 years, as compared to 0-3 years in the Missouri sites. Patients in California were predominantly white, those in Missouri black. Patients in Santa Barbara, a demonstration site, were somewhat more likely to have incomplete Medicaid eligibility for the year. Approximately one-fourth of childrens’ records were from emergency rooms, distributed equally between prepaid and fee-for-service counties. NO significant differences were noted in immunization rates (Table 5), with 56% of children having had three DPT and OPV immunizations by their first birthday at the California sites, 69% in Jackson County (prepaid) and 66% in St Louis (FFS). Emergency room charts were excluded from this assessment. Immunizations were rarely abstracted or immunization status referred to in emergency room charts. When care delivered in emergency rooms was included in the analysis, rates of immunization completion at 12 months of age dropped to 43

Table 4. Children’s sample characteristics and utilization

Mean age (yrs) Percent black Percent hispanic Percent with less than 6 months Medicaid eligibility Percent of records abstracted from ER Mean number of 1985 visits used in abstraction Mean number of 1984 visits for children from claims files

Santa Barbara (prepaid)

Ventura (FFS)

Jackson (prepaid)

St Louis (FFS)

2.0 3.2 27.1

1.9 3.6 33.8

1.4 83.5 0.5

1.5 80.5 0.2

28

21.8

9.2

9.2

24.8

23.7

24.4

23.0

3.06

3.60*

5.13

3.34.

3.65

5.42

3.65

3.41

*p < 0.05 for comparison between demonstration FFS = fee-for-service,

and comparison counties.

Prepaid vs Traditional Medicaid Plans Table 5, Preventive care for chiWn:.results SantaBarbara (pnpaid) Prop childmn 1 yr or older with basic immtmixation series complete (DPT, OPV x 3), excluding ER care Prop children b 15 montbs with MMR Prop children 2 12 months with het recorded Rop visits 0-1 yrs with wt recorded Prop visits 0-1 yrs with height recorded Prop visits 0-1 yrs with head circ. recordcd

1217

from chart abstraction Venrura Jackson CFFS) (pnpaid)

st Louis (FFS)

0.565

0.567

0.695

0.658

0.454

0.431

0.628

0.52.

0.478

0.338’

0.391

0.465*

0.793

0.906.

0.896

0.875

0.527

0.622”

0.726

0.481.

0.474

0.511

0.432

0.321*

*p < 0.05 for comparison between demonstration and comparison counties. FFS = fee-for-service. Prop = proportion of abstractions meeting criterion. DPT = diptheria, pertussis, tetanus vaccine. OPV = oral polio vaccine. MMR =measles, mumps, rubella vaccine.

monly on a per-visit basis in the comparison county of Ventura, Calif., and in the demonstration county of Jackson, Mo. Adult health maintenance

Women in the age range enrolled in AFDC programs have several recommended periodic health maintenance interventions. Pap smears are recommended yearly for women in their early 2Os,and should continue yearly for.women with a history of abnormal pap smears or who fa11into a high-risk category with early age of íìrst intercourse or history of sexually transmitted diseases Less than half of women had a pap smear documented in 1985 at three out of the four &es (Table 7). Women in Jackson County were more likely to have pap smears than women in St Louis. Pregnant women were more likely to receive pap smears. When stratiíìed by pregnancy status, women in the demonstration

county of Jackson remained more likely to receive pap smears, with mixed results in the CaliEornia sites. Logistic regression was used to control for several confounding variables (Table 8). After control for variables of number of 1985 visits, patient age and type of provider, patients at the demonstration site of Santa Barbara were less likely to have had a pap smear, at a borderlîne leve1 of statistical significante. Patients in the Missouri demonstration of Jackson County were 50% more likely to have had a pap smear. Performance of a pap smear was associated with increasing utilization, decreasing patient age and sites of care in ambulatory clinics and Gynecology specialists’offices. Nongynecology physicians’ offices were used as the reference group. The directions of differences seen in the univariate and multivariate regression analyses are similar, with fewer pap smears in

Table 6. Regression analysis of DPT and OPV immtmixations. Dependent variable: completion of three DPT, OPV immtmimtions after 1 yr of age Califomia Variable Demonstration site Number of 1985 visits ER chart abstracted Clinic chart abstracted Std farms usually or fîequently used

OR estimate 1.082 1.18 0.001 0.355 1.025

Missouri

p-Valse

Prepaid versus traditional Medicaid plans: effects on preventive health care.

Prepaid, case managed systems have been proposed as a method of controlling costs in Medicaid populations. We investigated the utilization of preventi...
881KB Sizes 0 Downloads 0 Views