Revised Cost Estimates of Medicaid Recipients With Serious Mental Illness and HIV-AIDS

Persons selected for this study were a subset of the previous study cohort who could be linked to the CMS Medicaid files ( 1 ) and met the selection criteria for the study presented here: aged 18–64 years, eligible to receive Medicaid services in Philadelphia, continuously enrolled in the Medicaid program in 1996, and having a claim with a serious mental illness or HIV-AIDS diagnosis between 1985 and 1996. Individuals dually eligible for Medicaid and Medicare were excluded, as were those enrolled in voluntary health maintenance organizations, the majority of whom had no serious physical or mental health problems. All of those with no service claims for serious mental illness or HIV-AIDS service during 1996 were selected as the comparison group.

Persons were deemed to have a serious mental illness on the basis of a claim for treatment that included a diagnosis of schizophrenia ( ICD-9 code 295) or major affective disorder ( ICD-9 code 296) and at least one inpatient or two outpatient treatment contacts with these diagnoses, following a protocol developed by Lurie and colleagues ( 5 ) to ensure accurate diagnoses. ICD-9 codes of 042 (HIV infection with specified conditions), 043 (HIV infection causing other specified conditions), and 044 (other HIV infections) were used to identify an HIV-AIDS-related diagnosis. Psychiatric drugs were classified using therapeutic class codes 74–77 and 80–84 (anticonvulsants; anti-Parkinsonian agents; antidepressants; barbiturates; cerebral stimulants; nonbarbiturates, sedatives, or hypnotics; other psychotherapeutics; and tranquilizers). No direct client contact was involved in the study. This research was approved by the University of Pennsylvania's institutional review board through the Office of Regulatory Affairs, which also provided ongoing oversight for the protection of human subjects.

The 1996 Pennsylvania Medicaid Analytic eXtract data, formerly known as the State Medicaid Research Files, were used for the analysis. The data files from the CMS contain person-level sociodemographic information and service-level information, such as type of service, date of service, provider, diagnostic codes, and cost for the Medicaid recipient ( 6 ). Medicaid claims records were used both to identify diagnostic groups and to construct reimbursement costs by type of service. The Medicaid claims files represent an excellent resource for health services research on vulnerable population groups, given that one in five Americans is enrolled in the Medicaid program, Medicaid is a major payer for mental and substance use disorders, and cost estimates are comprehensive because of the extensive nature of the benefit coverage ( 7 ).

Annual costs were aggregated for the four groups (serious mental illness and HIV-AIDS, serious mental illness only, HIV-AIDS only, and neither condition) using 1996 Medicaid-reimbursed service claims with separate categories for health and psychiatric services including drug and alcohol treatment. Service records were aggregated at the individual level to reflect the total reimbursement costs. A descriptive analysis was employed to compare characteristics and mean expenditures per group using chi square and analysis of variance tests of significance.

The study participants included 23,729 adult Medicaid recipients—196 individuals had a diagnosis of co-occurring serious mental illness and HIV-AIDS, 5,076 had a diagnosis of serious mental illness only, 482 had a diagnosis of HIV-AIDS only, and 17,975 were in the comparison group. Statistically significant differences between groups were found in gender, race, age, and Medicaid eligibility status. Persons with co-occurring serious mental illness and HIV-AIDS constituted 1% of the study population. Compared with the group with only HIV-AIDS, the group with co-occurring illnesses was older and was more likely to be disabled (as measured by receipt of Supplemental Security Income [SSI]), black, and female. Compared with the group with serious mental illness only, the group with co-occurring illnesses was older, more likely to be disabled and male and less likely to be black. The group with co-occurring illnesses was older and more likely than the comparison group to be male and disabled. Because the comparison group for this analysis represented individuals who did not voluntarily enroll in a Medicaid managed care program before it was mandated, they may have been those who were relatively more disabled than the general Medicaid population, as evidenced by the high number of individuals in our sample receiving SSI disability benefits (60%).

Persons selected for this study were a subset of the previous study cohort (N=23,729 of 60,503, or 39%). Only 39% of the original sample was able to be analyzed in the study presented here because of exclusion criteria and matching issues between the CMS data files and the original Medicaid data files that the original sample was drawn from. Compared with the group with comorbid illnesses in the original study, the subset with comorbid illnesses in this analysis had the same median age of 37, was less likely to be male (N=116 of 196, or 59%, versus N=173 of 268, or 65%), was less likely to be black (N=45 of 196, or 23% versus N=136 of 268, or 51%), and was more likely to be disabled (N=184 of 196, or 94% versus N=182 of 268, or 68%). Compared with the group with only HIV-AIDS in the original study, the subset with only HIV-AIDS in this updated study had a slightly younger median age (36 years versus 39 years), was less likely to be black (N=83 of 482, 17%, versus N=265 of 410, 65%), was less likely to be male (N=318 of 482, 66%, versus N=307 of 410, 75%), and was more likely to be disabled (N=414 of 482, 86%, versus N=258 of 410, 63%).

Table 1 shows overall cost patterns for the groups in our sample. The group with co-occurring serious mental illness and HIV-AIDS had the highest annual expenditures, totaling $20,038 per person. In comparison, those with only HIV-AIDS used $14,714 of services per person, and those with only serious mental illness used $9,039 of services per person. The comparison group had a cost of $6,613 per person.

Table 1 also provides a detailed breakdown of costs by service category for each study group. Inpatient and other institutional care accounted for 65% of the costs for the group with co-occurring illnesses, 57% of the costs for the group with only serious mental illness, and 65% of the costs for the comparison group. Pharmacy costs were 25% of the costs for the group with HIV-AIDS alone, 17% of the costs for the group with co-occurring illnesses, 16% of the costs for the group with serious mental illness alone, and 14% of the costs for the comparison group. In ranking the costs within each group, psychiatric costs were highest in the group with co-occurring illnesses and costs for medical services were the highest in the group with co-occurring illnesses.

This study improved on our previous 1996 cost estimates of treating persons with serious mental illness and HIV-AIDS and provides a baseline for longitudinal comparisons. Results are consistent with our previous study, which also reported that the group with co-occurring serious mental illness and HIV-AIDS was the most costly population and required significant levels of care, compared with the comparison group and the group with serious mental illness alone. Little difference was seen in the medication and cost for treating HIV-AIDS between the group with co-occurring illnesses and the group with HIV-AIDS only. Even though the group with only HIV-AIDS did not have co-occurring serious mental illness, the group still had substantial psychiatric costs for mental health problems resulting from anxiety and other diagnoses. Using the consumer price index to estimate the 1996 cost of HIV-AIDS in 2008 dollars, the mean cost for the group with only HIV-AIDS would be $20,520 and for the group with comorbid illnesses it would be $27,945.

Our findings for the costs for HIV-AIDS care are consistent with the estimates in the Hopkins study of Medicaid AIDS patients ( 3 ), and our costs of care for the population with co-occurring serious mental illness and HIV-AIDS are also similar to those of a four-state Medicaid study that looked at a similar population ( 8 ). When we look at the costs for the HIV-AIDS population groups, regardless of the presence of serious mental illness, the cost was $16,253 per study participant, which is similar to the 1998 costs of $16,008 found by Roberts and colleagues ( 9 ). We found that 7% of the expenditures for those with HIV-HIDS were for nursing home costs. We would expect even smaller percentages in the future with the newer treatment technologies, such as highly active antiretroviral therapy.

It should be noted that a limitation to these estimates is the lack of cost data on Medicaid clients who are dually eligible for Medicare and Medicaid. Now that Medicare Part D is the primary reimbursement source for pharmacy costs, as well as inpatient hospital care, studies of the dually eligible group will need to link Medicaid and Medicare data sets to get comprehensive information. Additionally, it is possible that the severity of illness for HIV-AIDS differs in the population with only HIV-AIDS and the population with comorbid serious mental illness and HIV-AIDS, which might explain the higher medical costs for the participants with comorbid illnesses. Evidence of this was found by study authors showing a higher prevalence of opportunistic infections in the population with comorbid serious mental illness and HIV-AIDS than in the HIV-AIDS-only population ( 10 ).

Although the absolute costs of care are much higher than we reported earlier, these results are consistent with our previous study in that both studies showed that individuals with co-occurring serious mental illness and HIV-AIDS had the highest costs. These improved estimates provide an excellent baseline for follow-up studies examining changes in treatment and cost of care since the advent of newer antipsychotic and antiretroviral drug combinations in the mid-1990s. We expect annual as well as lifetime costs to be higher in the future both for those with HIV-AIDS alone and for those with comorbid serious mental illness and HIV-AIDS, because of potential medical complications associated with aging and some of the side effects of treatment.

This research was supported by grant 5-RO1-DA-015627-05 from the National Institute on Drug Abuse (HIV Prevention Program among Substance Abusing SMI) and by grant 5-RO1-NR-008851-05 from the National Institute of Nursing Research (Nursing Intervention for HIV Regimen: Adherence among SMI).

The authors report no competing interests.

Dr. Rothbard and Dr. Blank are affiliated with the Department of Psychiatry, University of Pennsylvania, 3535 Market St., Philadelphia, PA 19104 (e-mail: [email protected]). Ms. Miller is with Thomson Reuters Healthcare, Santa Barbara, California. Dr. Lee is with the Department of Social Welfare, Pusan National University, Busan, Korea.