Therapeutic advances in the clinical management of AIDS have created unprecedented optimism for improved patient outcomes and enhanced survival. Although the incidence of AIDS cases is declining, the number of persons living with AIDS continues to rise. With appropriate and timely clinical care, patients may experience at least a transient return to premorbid states of health. However, optimal outcomes may be disproportionately available to patients with medical insurance and access to state-of-the-art AIDS care.1
As a consequence of steeply increasing Medicaid budgets, many state Medicaid programs have adopted managed care programs as a means of stabilizing medical costs. In 1996, the Maryland General Assembly authorized a Medicaid managed care program. Following the passage of this legislation, and approval by the Centers for Medicare and Medicaid Services, Maryland's Medicaid program changed from a fee-for-service based program to a capitated managed care program. Under this program, Maryland's 330,000 Medicaid recipients were subsequently enrolled in managed care organizations (MCOs).
Maryland's managed care system, known as Health-Choice, began enrolling clients in July 1997. Medicaid recipients receive care within MCOs that are responsible for the provision, coordination, and fiscal management of Medicaid benefits for recipients assigned to them. Health maintenance organizations, community-based organizations, county health departments, hospitals, and individual providers may contract with MCOs for provision of services to recipients. All Medicaid-eligible individuals were included in the waiver with the exception of recipients who were also receiving Medicare; those in spend-down eligibility categories; individuals who were inpatients in either nursing homes, chronic hospitals, psychiatric hospitals, or intermediate care facilities; and individuals in other waiver programs. Children with HIV or AIDS were also excluded.
The objective of this study was to evaluate the care received by people with AIDS in a statewide Medicaid managed care system and to examine factors that may influence the provision of quality care; specifically, to examine whether differences in volume of patients seen within a practice are associated with variations in quality of care.
MCOs receive an enhanced AIDS capitation rate ($2161 per month for adult patients residing in the Baltimore area and $1812 for patients residing in the rest of the state). Each MCO submits an eligibility form to Medicaid to receive the enhanced rate. The eligibility forms were cross-checked against AIDS Administration records to verify the AIDS diagnosis according to Centers for Disease Control and Prevention criteria. The matched list of Medicaid members with AIDS and the MCO applications were used to create the study population. Additional information on the AIDS patients' mode of exposure and date of AIDS diagnosis was obtained from the state's HIV/AIDS surveillance database. Vital status was obtained from chart review, review of HIV and AIDS registries, and a death record search.
A sampling strategy was developed to obtain a representative sample of Maryland Medicaid patients with AIDS that took into account the large volume of charts to be reviewed and the importance of reviewing charts at all the participating Medicaid MCOs, most of whom provided care at multiple clinical sites. To ensure an adequate sample, the patient charts of the 3 small MCOs were all reviewed. All charts at the 1 staff model MCO were reviewed. At the 4 larger MCOs, all of the clinical sites that were associated with these MCOs and provided care for ≥10 patients were visited and all available charts were reviewed. The remaining sites associated with these 4 larger MCOs that had <10 patients each were stratified by geographic location and randomly sampled.
A standardized chart review instrument was used to examine compliance with Public Health Service guidelines. Medical charts were reviewed for specific notation indicating major psychiatric illness or indication of substance or alcohol abuse. Psychiatric illness and substance or alcohol abuse were categorized separately. Current use of antiretroviral medications was assessed. Dose and frequency were not included in the analysis due to the high volume of missing data. The clinical measures that were used and the periods during which they were assessed are presented in Table 1. For all clinical measures, patients were included if they were in the MCO a sufficient period to allow adequate time for care to be delivered. Three periods were used: 90 days, 6 months, and 1 year. Appropriate CD4 and viral load testing was defined as testing that occurred within 6 months of entry into the MCO. In addition to these quality indicators, we assessed the provision of case management services.
The analysis examined clinical indicators of quality stratified by a site volume variable that consisted of the total number of Medicaid AIDS patients seen at a clinical site. The site volume variable was created as a surrogate measure of the intensity of a provider' s experience of caring for AIDS patients. The data did not permit direct calculation of the number of AIDS patients seen by each provider. Instead, we examined the number of Medicaid AIDS patients cared for by providers within a given site-a clinic or office, which might include several or many physicians-and established 3 categories of patient volume. Sites were classified as low-, medium-, or high-volume sites. Low-volume sites provided care for 1-15 patients, medium-volume sites provided care for 16-100 AIDS patients, and high-volume sites were those where 101-500 AIDS patients were seen.
Cross-tabulations of patient and provider characteristics were examined using χ2 and Fisher exact tests and analysis of variance with a P value of 0.05 as the determination of statistical significance. Logistic regression analysis was performed to determine the statistically significant predictors of outcome. The Kaplan-Meier method was used to explore differences in survival. Cox proportional hazards regression was performed to assess the relative hazard of death using patient volume as the main unit of analysis.
Patient Demographics and Health Care Utilization Stratified by Patient Volume
Table 2 presents sociodemographic variables stratified by patient volume. The low-volume sites included 43 sites with 1-15 patients each, the medium-volume sites included 5 sites with 16-100 patients each, and the high-volume sites included 3 sites with 101-500 or more patients each.
Gender (60% male), race (87% African American), age (mean 41 years), and vital status at 6-month follow-up (14% dead) did not significantly vary by the patient volume of the provider sites. One hundred percent of high-volume sites were located in Baltimore City, vs. 49% of low and 84% of medium volume sites (P < 0.001).
Injection drug use was the predominant mode of exposure across all patient volume levels, although higher in the medium-volume sites (74%) as compared with high- (68%) and low- (54%) volume sites; in addition, the low-volume sites had a higher proportion of patients reporting male-male homosexual contact (16%) and heterosexual contact (20%) as compared with the medium- (8 and 12%) and high- (12 and 14%) volume sites (P < 0.001). Alcohol abuse was significantly different across patient volume categories (P < 0.001). The high-volume sites had more patients with a history of psychiatric conditions (61 vs. 48% at low and 49% at medium volume sites, P < 0.01). Patients in all categories had similar mean CD4 counts (177.0 cells/mm3), and similar percents in all categories had high viral loads (49% with viral load >20,000 copies). Patients in the medium-volume sites had a slightly longer length of enrollment in the study (357 vs. 313 days in low- and 282 days in high-volume sites, P < 0.001).
Health Care Utilization
Figure 1 presents health care measures stratified by patient volume. For health service utilization indices among patients enrolled at least 90 days (n = 924), the high-volume sites had the greatest proportion of patients with PPD testing and reading (65 vs. 50% at low- and 45% at medium-volume sites, P < 0.001). Sites at all patient volume levels had high coverage of Pneumocystis carinii pneumonia (PCP) prophylaxis for patients whose CD4 levels had ever fallen to <200 (96% of 850 patients), as well as for patients with a history of PCP (99% of 227 patients).
The high-volume sites had a substantially greater proportion of patients with pneumococcal vaccination (among 485 patients, 77 vs. 55% in low- and 39% in medium-volume sites, P < 0.001), history of hepatitis B testing (among 924 patients, 94 vs. 60% in low- and 66% in medium-volume sites, P < 0.001), the offer of hepatitis B vaccine if there was no positive test before the study (among 518 patients, 35 vs. 20% in low- and 10% in medium-volume sites, P < 0.001), hepatitis C screening (among 924 patients, 82 vs. 51% in low- and 50% in medium-volume sites, P < 0.001), and syphilis testing (among 924 patients, 57 vs. 22% in low- and medium-volume sites, P < 0.001).
For health utilization indices among patients enrolled at least 6 months, the high-volume sites had the greatest proportion of patients with appropriate CD4 testing (among 772 patients, 74 vs. 47% in low- and 56% in medium-volume sites, P < 0.001) and appropriate viral load monitoring (among 772 patients, 74 vs. 49% in low- and 51% in medium-volume sites, P < 0.001). Over 96% of patients with CD4 ≤200 or viral load >20,000 have a history of antiretroviral use regardless of patient volume of their provider site (97% of 561 patients).
Forty-seven percent of women who had been enrolled for at least 12 months (n = 182) had a Papanicolaou test performed in the previous year; this did not differ significantly by patient volume. High-volume sites reported a substantial proportion of patients with case management and a documented plan of care (among 924 patients, 63.2 vs. 26.3% in low- and 21.7% in medium-volume sites, P < 0.001).
Antiretroviral medication had been prescribed for most patients during the study period (813 of 1052 patients, 77%). However, patients seen at high-volume sites were significantly less likely to be current users of antiretroviral therapy than patients seen in low- or medium-volume sites (20.3 vs. 13.1% in low and 13.5% in medium sites) (data not shown).
Compared with low-volume sites, the relative hazard of death in medium-volume sites was 0.76 (95% CI: 0.36-1.63, P = 0.761), and the relative hazard of death in high-volume sites was 1.45 (95% CI: 0.85-2.49, P = 0.175). Because survival curves for low- and medium-volume sites overlapped and because there was no statistically significant difference between them, low- and medium-volume sites were combined as the reference group. To reduce the potential confounding effects of other factors, the final survival model adjusted for antiretroviral use, time from AIDS diagnosis, appropriateness of viral load testing, inpatient hospitalizations, and patient age (Table 3). Compared with current users of antiretroviral therapy, former users had an increased relative hazard of death (P < 0.05), as did patients who had never used antiretroviral therapy (P < 0.001). Having inpatient hospitalizations was also associated with an increased relative hazard of death (P < 0.001), as was older age (P < 0.001), while having appropriate viral load testing was associated with improved survival (P < 0.001). Time from AIDS diagnosis was not a statistically significant predictor of death in this study (P = 0.108). Adjusting for potential confounders only strengthened the association between high-volume sites and increased hazard of death compared with low- and medium-volume sites (relative hazard = 2.14, P < 0.01).
Medicaid is an important source of payment for health care services among people living with HIV and AIDS throughout the United States. Although estimates of the proportion of AIDS patients covered by Medicaid vary, it is believed to be the largest single publicly funded program for AIDS health care services. Many states grappling with increasing costs of providing health care for people in Medicaid have implemented managed care programs as a way of stabilizing rapidly escalating costs. Currently 42 states report enrolling some or all people with HIV or AIDS into risk-based managed care.2 As the number of HIV/AIDS patients in Medicaid MCOs grows, it is critical to evaluate the quality of care provided for individuals with such serious illnesses.
The care provided to AIDS patients was analyzed by creating 3 levels of patient volume. Low-volume sites were primarily individual providers with small Medicaid AIDS practices. Medium-volume sites consisted of individual providers with larger practices and included 1 staff model HMO. Among the 3 high-volume provider sites, 2 were specialized HIV/AIDS academic medicine facilities that were each part of an infectious disease program. The 3rd site was a specialized HIV/AIDS clinic that provided a full range of HIV/AIDS and ancillary services within a self-contained and full service clinic.
For the majority of health service utilization measures, the high-volume sites did substantially better in providing services than the low- and medium-volume sites. These included such important clinical assessments such as PPD testing and reading, pneumococcal vaccination, hepatitis B testing, hepatitis B immunization, hepatitis C screening, syphilis testing, and appropriate CD4 and viral load monitoring. All these measures are considered to be critical elements in the medical management of individuals with HIV/AIDS.3 In addition to the lack of compliance with these Public Health Service clinical guidelines, the low- and medium-volume sites were significantly less likely to provide important ancillary services such as case management services.
Case management services have long been an integral part of HIV/AIDS services and are described as services that augment and coordinate existing care services4,5 and improve the quality of life of people with HIV/AIDS by ensuring appropriate individualized medical care.6 In 1991 the National Commission on AIDS described case management services as having important benefits for cost savings, reducing the number and length of hospitalizations, and bringing coherence to the service delivery system, as well as enhancing satisfaction and quality of life.7 Our findings indicate that low- and medium-volume sites are frequently not providing these services, which may significantly limit the coordinating and supportive services that their clients may receive.
Overall we did not find improved survival among patients seen in the high-volume sites. This differs from earlier studies in which increased experience was associated with improved survival.8,9 However, this study precludes observation of any possible long-term effects of experience on outcome. An 18-month period was exhaustively reviewed for measures of quality and health care utilization measures. Deaths occurring during the study period or in the 6 months following the end of the study were reviewed. This approach differed from the method used in the study by Kitahata et al,8 in which a retrospective analysis was performed that included patients from time of diagnosis to death and was limited to patients who had the same care provider during that period. Our study's design could not adjust for the quality of care preceding the study period, nor could it adequately adjust for the differences in severity of illness indicated by factors such as decreased current use of antiretrovirals or viral resistance.
States have taken a variety of approaches to ensure that persons living with HIV/AIDS receive appropriate health care in managed care settings. Several states have required that HIV/AIDS patients be seen by physicians experienced in providing care for persons living with HIV/AIDS. The New York Medicaid special needs program requires physicians treating persons living with HIV/AIDS to have experience treating 20 patients per year. HIV/AIDS patients seen in the Tennessee Medicaid managed care program (Tenn Care) receive care in an AIDS Center of Excellence that is defined as an actively managed base of at least 50 HIV-positive patients. California requires that all patients with HIV/AIDS (regardless of payer) have standing referrals for physicians with HIV/AIDS experience as defined by experience caring for 20 patients and 30 hours of HIV-specific continuing medical education or successful completion of the HIV competency examination. Arizona offers a supplemental reimbursement for members of its managed care system if protease inhibitors are prescribed and requires that such patients must be under the care of a qualified HIV/AIDS treatment professional defined as “having a special interest, knowledge and experience with the treatment of HIV/AIDS.”10 Maryland has recently revised its regulations to require MCOs to have HIV/AIDS specialists available to patients. An HIV/AIDS specialist is defined as a board-certified infectious disease physician or a health care provider who has provided care for at least 20 patients within the past 2 years and has either 30 hours of HIV-related category I credits (within the past 2 years) or has successfully completed the American Academy of HIV Medicine credentialing examination.
Before firm conclusions are drawn, several limitations of the study must be acknowledged. Because the data did not permit direct calculation of the number of AIDS patients seen by each provider, the study used site volume as a surrogate measure of the intensity of the providers' experience in caring for AIDS patients. This approach may have obscured variations in quality of care among providers at high-volume sites.
In addition, because our study focused on chart reviews of Medicaid patients only, our characterization of low-, medium-, and high-volume sites may have misclassified sites that provide care for large numbers of non-Medicaid patients. To ensure the correct classification of patient volume sites, we used a supplemental data source to test our patient volume classification. Maryland law requires all laboratories licensed to perform tests in the state of Maryland to report CD4 counts of <200 to the state AIDS Administration using a unique identifier. For the period from July 1997 to December 1998, the 4318 CD4 laboratory tests (CD4 <200) with complete unique identifiers and that matched to the Maryland AIDS Registry were used to evaluate our classification of low-, medium-, and high-volume sites. Using the 3-level site variable created for the original analysis, the providers were grouped according to the total number of patients who had CD4 <200 test results in the study period. A comparison of site volume, as determined by CD4 tests and by the method used in this study, revealed only 5 of 51 sites whose classification of site size changed. Four of the 5 changes were moves between low- and medium-volume sites, which would have no impact on the finding of this analysis of differences between the high-volume sites relative to the low- and medium-volume sites. The Kendall Tau b correlation coefficient (0.73975; P < 0.0001) indicated no significant differences between the 3 levels of patient volume observed in the Medicaid chart classification and the classification based on number of patients with CD4 counts of <200 (data not included).
Finally, this study was conducted with data collected during the first 18 months of the program. The initiation of any new program is accompanied by challenges and obstacles that may correct themselves with time. The problems observed may have been corrected as clinicians and systems adjusted to the new medical care environment.
HIV care has evolved substantially over the past 18 years. Although there was initial interest in having primary care physicians provide HIV care, rapidly evolving clinical standards and expanding understanding of the complex, mul-tisystem complications of HIV disease have placed heavy demands on clinicians. The pace of new developments in HIV/AIDS care is ever-increasing, thereby requiring physicians to devote substantial resources to understand antiretroviral therapy and viral resistance and to be able to follow rapidly changing clinical guidelines. Research over the past 6 years has generated increasing documentation of the importance of physician experience in reducing morbidity and mortality among HIV/AIDS patients. We observed substantial differences in adherence to Public Health Service guidelines among the low-/medium- and high-volume sites studied. The evidence available to date supports the conclusion that provider experience significantly improves clinical outcomes among AIDS patients. These findings and the substantial literature suggests that policies that require that HIV/AIDS care be provided by high-experience/high-volume providers may offer the best opportunity to ensure that the care received by Medicaid HIV/AIDS patients meet established clinical standards.
The authors would like to acknowledge the following individuals for their valuable contributions to this project: Lois Eldred for shepherding the project through its early phases; Laura Smith for her work as Project Director; Christine Weston for development of the chart abstraction instrument; Abbey Small for her assistance with data analysis; Nancy McCaslin, Marilyn Walkins, and Mohammed Haque for performing the more than 1000 chart abstractions; and Monique Cheatham for production of this manuscript.