Alcohol & Drug Abuse: The Impact of Terminating Disability Benefits for Substance Abusers on Substance Use and Treatment Participation

We selected a random sample of 400 SSI recipients in Los Angeles County who were on the Social Security Administration's October 1996 roll and who were classified as disabled primarily by drug addiction and alcoholism. Although the new law affected both SSI and SSDI beneficiaries, we expected that the effects of the legislation would be most visible in the SSI population, given their lack of economic resources. In addition, 57 percent of the drug- and alcohol-disabled population received SSI alone (written communication, Social Security Administration, Nov 27, 1996). We focused on Los Angeles because one-fifth of all SSI recipients nationwide who received benefits because of disability related to substance abuse lived in California, and 15 percent of those lived in Los Angeles.

A letter was sent to each of the 400 selected recipients inviting them to participate in the study. The number of participants was 302 at baseline, a 77.4 percent participation rate. Seventy-eight persons either refused to participate or could not be located, nine were confirmed deceased at baseline, and one was misclassified. The demographic characteristics, work history, and appeal status of the persons we interviewed were not significantly different from those we were unable to interview. Of the participants, 196 (65 percent) were male, and 160 (53 percent) were African American. The mean± SD age was 44±9.4 years. As of March 1997, a total of 94 participants (31 percent) had been reclassified and continued to receive benefits, according to information obtained from the Social Security Administration. At one-year follow-up we were able to interview 273 (90 percent) of the original 302 participants.

Each participant was interviewed face to face between December 1996 and April 1997 and again 12 months after the original interview. In addition to demographic information, participants were asked about their substance use and treatment participation, mental and physical health, health care utilization, housing, and individual and family resources, including employment and income. Participants were paid $25 for each interview.

Substance use was measured by self-report. A person who reported any drug or alcohol use in the preceding six months was considered a "user." Although occasional substance use may not indicate abuse or dependence in some populations, it is important to note that all the participants in this study were considered disabled by substance abuse; thus any use was likely to be problematic. Treatment participation was defined as being in substance abuse treatment—excluding self-help groups such as Alcoholics Anonymous and Narcotics Anonymous—for at least one week during the previous six months.

We developed a conceptual model based on the Behavioral Model of Health Services Use of the effects of the legislation on substance use and treatment participation (2,3,4). Because individuals whose benefits were continued had more severe comorbid conditions (5), we used logistic regression to adjust for measured differences between the two groups at baseline. Using the conceptual model, we selected covariates and included them in the model if they were associated with substance use at 12 months at a p level below .2 after adjusting for baseline substance use and treatment participation. Covariates included in the final model were baseline substance use and treatment participation, demographic characteristics, mental health status, and whether or not the person was in jail. Mental health status was assessed by the number of bed-days or restricted-activity days in the past month due to an emotional problem.

At 12-month follow-up, substance use had decreased from 75 percent at baseline to 63 percent among respondents whose benefits were continued, but was essentially unchanged among those whose benefits were terminated—76 percent at baseline and 75 percent at follow-up. The difference between the groups was significant (odds ratio=1.9, 95% CI=1.01 to 3.02), adjusting for covariates. Among all respondents, baseline substance abuse was also associated with substance use at 12 months (OR=1.7, CI=1.3 to 2). Being in jail at the time of the initial interview was negatively associated with use at 12 months (OR=.3, CI=.1 to .9). Treatment participation decreased equally among respondents whose benefits were continued and respondents whose benefits were discontinued, from 47 percent to 21 percent.

Among the factors underpinning the passage of P.L. 104-121 was the widespread perception that providing cash assistance to substance abusers contributed to ongoing substance use (6). Our results are not consistent with this view. On the contrary, continuation of SSI benefits was associated with less substance use, and this was the case even though recipients were no longer required to have a representative payee or to be in treatment. Loss of benefits was not associated with a significant change in use.

The decrease in use among those whose benefits were continued occurred despite similar baseline substance use between the two groups and worse comorbid health conditions among the continued-benefits group (5). Among persons with a substance abuse disorder, co-occurring mental health problems are associated with worse substance abuse and other outcomes (7).

The data do not tell us what drove the observed decrease in substance use among those whose benefits were continued. Possible explanations include the delayed effect of treatment or the natural history of substance use among this group. Another possibility is that use decreased when individuals learned that the uncertainty about whether they would continue to receive SSI benefits had been resolved in their favor. However, these explanations do not preclude the possibility that ongoing receipt of benefits may contribute to decreased use.

Treatment participation decreased for both groups, perhaps in part because of the discontinuation of mandated treatment. Although we do not know what would have happened in the absence of the policy change, the drop in treatment participation is striking. However, despite the decrease in treatment rates, substance abuse did not increase, even in the group that lost benefits. Several explanations are possible. Among this population, mandated treatment may not reduce substance use, or the type of treatment received may not have been of sufficient quality or intensity to have an impact on use.

This study had several limitations. We relied on brief self-report measures, which may have underestimated the true prevalence of substance use (8). Individuals whose benefits were continued might have been less likely to admit to use for fear that their benefits would be terminated again. A nonresponse bias is a possibility: 78 individuals either refused to participate in the study or could not be located, nine were confirmed dead at baseline, and an additional four died by the one-year follow-up. However, as noted, the characteristics of individuals we were unable to interview were not significantly different from those of the participants. Finally, although we adjusted for measured differences between the two groups, other unmeasured differences may have contributed to a selection effect.

Our study has several public policy and clinical implications. Policies that provide income and health insurance to substance abusers may not contribute to ongoing substance abuse. Most studies that have suggested a relationship between disability income and ongoing substance abuse (9,10) have been done with seriously mentally ill participants, and their results may not apply to individuals without comorbid mental illness.

Although treatment participation was already low—despite the mandate—it fell even further when treatment was no longer a condition for receiving benefits. Clinicians and policy makers can continue to advocate for mandated treatment for persons with a substance use disorder who are receiving federal or state income support, and they can encourage ongoing monitoring of treatment participation.

The authors acknowledge the role of the Supplemental Security Income study group of the Center for Substance Abuse Treatment in the design of the study and the valuable contributions of Emilia Lombardi, Ph.D., John Light, Ph.D., Audrey Burnam, Ph.D., Ken Wells, M.D., Ph.D., and Douglas Anglin, Ph.D.

Dr. Watkins is a research psychiatrist at Rand, 1700 Main Street, Santa Monica, California 90407-2138 (e-mail, [email protected]). Dr. Podus is an assistant research sociologist at the Drug Abuse Research Center, University of California, Los Angeles. Richard J. Frances, M.D., is editor of this column.