Welfare Reform and Substance Abuse

Abstract

The 1996 Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) changed the nature, purpose, and financing of public aid. Researchers, administrators, and policymakers expressed special concern about the act's impact on low-income mothers with substance use disorders. Before PRWORA's passage, however, little was known about the true prevalence of these disorders among welfare recipients or about the likely effectiveness of substance abuse treatment interventions for welfare recipients. Subsequent research documented that substance abuse disorders are less widespread among welfare recipients than was originally thought and are less common than other serious barriers to self-sufficiency. This research also showed significant administrative barriers to the screening, assessment, and referral of drug-dependent welfare recipients. This article summarizes current research findings and examines implications for welfare reform reauthorization.

Since the 1970s, federal policymakers have enacted major tax reforms and major social policy initiatives ranging from the establishment of Supplemental Security Income (SSI) to the Americans with Disabilities Act and Medicare prescription drug coverage. The most radical and controversial social policy initiative was the 1996 Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) (Danziger 1999; Danziger and Haveman 2001; Danziger et al. 2002; Jayakody, Danziger, and Pollack 2000; Weaver 2000).

By changing the basic pattern of social obligations between public aid recipients and other citizens, PRWORA ended a 60-year-old federal entitlement, Aid to Families with Dependent Children (AFDC). In place of AFDC, PRWORA established the avowedly transitional program, Temporary Assistance to Needy Families (TANF). Survey data indicate that welfare reform has become one of the most popular social policy innovations of the last 30 years (Weaver 2000).

PRWORA converted the financing of public aid from an entitlement to a system of block grants that shifts many financial incentives and risks from the federal government to the 50 states and, in many cases, ultimately to the recipients themselves. TANF block grants accord states broad discretion to determine who is eligible for TANF and for how long.

States are given broad discretion to sanction recipients who do not comply with program rules (Edelman 1997). As long as the states comply with due process requirements, they are free to impose a wide range of penalties, ranging from small and temporary benefit reductions to the removal of recipient families from the TANF rolls. Such sanctions are widely applied. A study by the General Accounting Office found that the benefits of an average of 113,000 families per month (4.5% of TANF recipients) were reduced because of sanctions. Equally significant, in 1999 seven states reported that sanctions accounted for at least 20 percent of their case closures (Goldberg and Schott 2000; Pavetti and Bloom 2001).

At least 80 percent of TANF recipients face a five-year lifetime limit on the receipt of federally funded cash aid. PRWORA and related legislation restricted the ability of documented and undocumented immigrants to receive public aid. Concomitant legislation restricted the ability of children with behavioral disorders and adults with substance use disorders to receive federal disability benefits.

The number of families receiving public cash aid sharply declined in the years following welfare reform. Between August 1996 and March 2003, the number of TANF recipients fell by 59 percent, from 12.2 million to 4.96 million people (U.S. DHHS 2002, 2003). By 2003, welfare caseloads were at their lowest levels since 1969. The participation of unmarried mothers in the labor force reached record levels, a pattern that reflected the changing opportunities and incentives facing low-income parents balancing home obligations and paid work (Danziger et al. 2002). Researchers continue to debate the relative impact of welfare reform, the booming 1990s economy, and other public policy changes on the reduction in the number of welfare recipients (Blank and Schoeni 2000).

When PRWORA was passed, the prevalence of substance use, abuse, and dependence among welfare recipients was largely unknown. Many advocates, administrators, and researchers believed that substance use disorders among public aid recipients were widespread and severe, and some feared that such disorders would prevent many recipients from complying with the TANF requirements.

The likely impact of PRWORA provisions on low-income mothers who use or misuse alcohol, licit, or illicit drugs was also unknown. Although work requirements, sanctions, and lifetime time limits were not specifically related to substance use disorders, many participants in the welfare reform debate believed that substance-using, -abusing, or -dependent recipients would be disproportionately affected.

Joseph Califano, director of Columbia University's National Center on Addiction and Substance Abuse, spoke for many advocates, researchers, and policymakers when he commented, “Today the bulk of mothers on welfare—perhaps most—are drug and alcohol abusers and addicts, often suffering from serious mental illness and other ailments” (Califano 2002, A29). Writing from a liberal perspective, Califano suggested that expanded access to high-quality substance abuse treatment is essential in order for TANF recipients to achieve economic self-sufficiency. Writing from a very different perspective, conservatives advanced similar claims about the prevalence of substance use disorders among recipients of public aid. For example, New York's former welfare commissioner Jason Turner proposed more aggressive drug testing of applicants for public aid and argued that “over 10 percent of welfare applicants needed some form of help for substance abuse” (Turner 2003, 1).

Many participants in the welfare policy debate also believed that substance users and misusers would be disproportionately affected by the stringent new regulations, such as the Gramm Amendment, sponsored by Senator Phil Gramm of Texas in 1996, which imposed a lifetime ban on food stamps and TANF aid to individuals with drug-related felony convictions.

Despite widespread concerns, the policy and research communities lacked specific information about key issues:

• The prevalence of substance use and substance use disorders among TANF recipients.

• The extent to which substance use and substance use disorders discourage work and self-sufficiency.

• The extent and effectiveness of screening and assessment practices in TANF offices.

• The effectiveness of interventions and substance abuse treatment services in reducing TANF recipients’ substance use and substance use disorders and improving self-sufficiency.

These issues form the core of our review. We consider the implementation of policies and interventions pertinent to substance use and substance use disorders and then summarize the current published literature and identify gaps in the research. Last, we consider the implications of the current research findings for the reauthorization of welfare reform.

Prevalence and Severity of Substance Use and Substance Use Disorders

How widespread is substance use by TANF recipients? What are the consequences of such use? How many substance users satisfy the screening criteria for abuse or dependence? How many TANF recipients are currently receiving treatment? Does the receipt of TANF facilitate or hinder treatment? The existing research suggests four patterns in the TANF population:

1. About 20 percent of TANF recipients reported using an illicit drug in the last year. Substance use and misuse are covert, stigmatized behaviors, and so it is difficult to estimate their prevalence. Published estimates are as high as 37 percent and as low as 6 percent of the welfare rolls (Jayakody, Danziger, and Pollack 2000). The variation reflects great differences in the subgroups of welfare recipients being investigated and in the nature of the substance use, misuse, or dependence posited to indicate a disorder.

   Available data are generally drawn from respondents’ self-reports in population surveys such as the National Household Survey of Drug Abuse (NHSDA), fielded periodically from 1972 and annually from 1991 to 2001, and its successor, the National Survey of Drug Use and Health (NSDUH), beginning in 2002. These data provide a useful perspective on the prevalence of substance use and substance use disorders in well-defined, representative populations. In all survey waves, researchers (Jayakody, Danziger, and Pollack 2000; Pollack et al. 2002a) found that illicit substance use was about twice as common among welfare recipients as among other women with dependent children who did not receive public cash aid.

   In the 2002 NSDUH, the most recent data available for this analysis, Pollack and colleagues (2002a) found that 22.3 percent of women TANF recipients aged 18 to 49 reported recently using illicit drugs. In contrast, 12.8 percent of women with dependent children who did not receive TANF reported recently using illicit drugs.

   Some commentators are concerned that by removing the most employable recipients from the welfare rolls, PRWORA would introduce compositional changes that would result in a higher prevalence of substance use disorders (along with other difficulties) among those remaining on the TANF rolls. But the existing data do not bear out these predictions. The self-reported prevalence of illicit substance use by NHSDA respondents receiving TANF aid dropped by 10 percentage points between 1990 and 2001. The reported prevalence of illicit substance use was no higher in 2001 than in the 1996 NHSDA survey wave, and broad trends of substance use among TANF recipients appeared to parallel trends in the general population (Pollack, Danziger, Jayakody, et al. 2002).

   Similar results were obtained in the Women's Employment Study (WES), a longitudinal survey of current and former TANF recipients in one urban Michigan county. According to the fall 1999 WES survey, 22 percent of continuing TANF recipients (43 out of 194) and16.6 percent of former TANF recipients (72 out of 432) reported having recently used drugs (Pollack et al. 2002a).

   At least three analyses examined the correlation between substance use and the probability of welfare receipt. Kaestner (1998) used the National Longitudinal Survey of Youth (NLSY) to explore the association between substance use and the subsequent receipt of welfare. Although drug use during the previous year was associated with a higher probability of current welfare receipt, it accounted for only a little of the observed variation in welfare receipt. The elimination of drug use was predicted to reduce welfare participation by 3 to 5 percent. Jayakody, Danziger, and Pollack (2000) obtained similar results in logistic regression analyses of 1994/95 NHSDA data.

   Dooley and Prause (2002) also analyzed NLSY data. They found that entering welfare was associated with increased depression and alcohol consumption but that depression and alcohol did not prevent recipients from leaving welfare.

   Schmidt, Weisner, and Wiley (1998) found even weaker results for California AFDC recipients between 1989 and 1995. During a six-year period, drug dependence was not associated with long welfare stays, repeat welfare use, or total time spent on welfare. All these studies indicate that widespread substance use is not a major cause of TANF recipients’ continued economic dependence.

2. Only a minority of substance users receiving public aid meet diagnostic screening criteria for serious drug use disorders, although many substance users have other difficulties, including psychiatric disorders. Although approximately one-fifth of TANF recipients report having recently used illicit substances, the impact of such use on recipients, their children, and others is not clear. Based on both NHSDA and NSDUH data, about half the TANF recipients who reported having used illicit substances in the past year used marijuana only, and not other illicit substances (Pollack et al. 2002a). Only 5 percent of TANF recipients—one-fourth of self-reported substance users—satisfied the standard dependence screening criteria used to refer individuals to substance abuse treatment (Jayakody, Danziger, and Pollack 2000).

   Alcohol use poses special challenges when distinguishing the use and misuse of intoxicating substances. Although alcohol consumption is both legal and widely accepted in American society, heavy drinking, alcohol abuse, and dependence threaten well-being and social performance (Dooley, Catalano, and Hough 1992; Gill 1994). Research on alcohol highlights the difficult issues of causality in policy research, and research on job loss reveals the causal impact of unemployment on alcohol abuse and depression (Dooley, Catalano, and Hough 1992; Dooley, Catalano, and Wilson 1994; Dooley and Prause 1998).

   Grant and Dawson (1996) provided the most extensive analysis of substance abuse and dependence among welfare recipients. Using 1992 data, they found that 3.3 percent of welfare recipients satisfied the DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, 4th ed.) criteria for drug abuse or dependence and that 7.3 percent satisfied its criteria for alcohol abuse or dependence. Jayakody, Danziger, and Pollack (2000) found similar results in 1994/95 NHSDA data; 9 percent of welfare mothers were alcohol dependent, compared with 5 percent of nonrecipient single mothers.

3. Substance abuse or dependence should be considered as one of many barriers threatening the well-being and social performance of TANF recipients. Although substance use disorders attract widespread attention, they appear to be no more common, and are no more important to employment and welfare receipt, than are such concerns as poor physical health, poor academic skills, psychiatric disorders, transportation difficulties, and more general concerns such as racial minority status, language barriers, and immigration concerns. Even before welfare was reformed, researchers documented many barriers impeding welfare recipients’ economic self-sufficiency (Bane and Ellwood 1994; Goldberg 2002; Harris 1993; Merrill et al. 2001; Metsch et al. 1998; Olson and Pavetti 1996; Pavetti 1993; Schmidt et al. 2002; Speiglman et al. 1999; Taylor and Barusch 2004; Zedlewski 1999).

   Several studies show that psychiatric disorders, especially major depression and posttraumatic stress disorder (PTSD), are more common than drug and alcohol dependence among welfare recipients (Jayakody, Danziger, and Pollack 2000; Jayakody et al. 2004; Pollack et al. 2002b). Pre-reform NHSDA data indicate that about 19 percent of recipients had at least one of the four psychiatric disorders examined (Jayakody, Danziger, and Pollack 2000).

   PTSD is an especially common comorbidity among low-income women with substance use disorders (e.g., Bromet, Sonnega, and Kessler 1998; Haller and Miles 2004). According to the Michigan WES data, approximately 15 percent of TANF recipients satisfied the PTSD criteria, making this diagnosis alone more prevalent than substance dependence among WES respondents (Danziger et al. 2000; Lichter and Jayakody 2002).

   Although substance disorders affect only a small minority of TANF recipients, those affected often face a range of significant difficulties. Drug-dependent or drug-abusing recipients often have other physical and mental health disorders and face other complex barriers, such as domestic violence, that may bar employment.

   Gutman and colleagues (2003) assessed the prevalence of potential employment barriers and their relationship to later employment for 366 substance-using women receiving TANF who were enrolled in the CASAWORKS for Families demonstration program (discussed in a later section). This intervention was implemented in ten locations across the United States. Substance-using TANF recipients who participated in CASAWORKS exhibited multiple potential employment barriers at enrollment, averaging six of 14 barriers assessed. The most common barriers were transportation problems (88%), little work experience (81%), and few job skills (65%). These women reported significantly more obstacles than did a general welfare sample of women (n= 157) recruited from the same localities. Examining 12-month employment outcomes of TANF recipients, CASAWORKS researchers found that the total number of barriers (rather than any specific barrier) was the most powerful predictor of whether a given CASAWORKS recipient would be employed.

   Other studies have also found that substance users faced more employment barriers on average than did nonsubstance-user comparison groups. Morgenstern and colleagues (2003) explored the characteristics of 214 women who screened positive for substance use and were subsequently diagnosed with substance dependence in welfare settings in an urban New Jersey county (Morgenstern, McCrady, et al. 2003). They compared these women's characteristics with those of 69 nonsubstance-affected women also recruited from welfare settings. Overall, the substance-dependent women faced many more barriers (mean = 4.1) than did the nonaffected sample (mean = 1.9). In a second study of 248 TANF recipients in Houston, Texas, Montoya and colleagues (Atkinson et al. 2001; Montoya, Atkinson, and Struse 2001) found that no substance users were working at the time of the interview, compared with 9.7 percent of the nonusers.

   Schmidt and colleagues examined the role of drug use disorders in welfare exits and subsequent returns to welfare between 1989 and 1995 in northern California (Schmidt et al. 2002). They found that dependence was actually associated with shorter periods on welfare, although these shorter periods did not appear to indicate better outcomes. In a multivariate analysis, alcohol and drug dependence were not statistically significant predictors of work-related welfare exits, although substance dependence was significantly associated with welfare exits due to family or household changes.

   Schmidt and colleagues suggested that these family exits were likely to be caused by unstable living arrangements (including domestic violence) or the removal of children by child protective services. That is, the recipients’ relationships with drug-using partners, relatives, and friends sometimes led them to relocate or to separate from their children, and some of these changes in family and household composition jeopardized their welfare eligibility. Substance-dependent recipients were also more likely to leave for administrative reasons, including the denial of benefits owing to failure to file paperwork or to participate in required work programs. Qualitative analyses of the same study indicated that preoccupation with drug use and with drug procurement were key issues in violations of program requirements (Mulia and Schmidt 2003).

4. Substance use disorders appear to be more common among sanctioned and nonworking TANF recipients than among the general population receiving TANF aid. Substance use disorders appeared to be the most prevalent in the most disadvantaged segments of the TANF population: nonworking recipients and those who have been sanctioned. This finding underscores the importance of substance use treatment interventions to address the needs of highly disadvantaged recipients. But it also suggests that the experiences of very disadvantaged recipients provide a misleading picture of the broader population of families receiving public aid.

   Morgenstern and colleagues examined substance use disorders among sanctioned recipients (Morgenstern, Riordan, Dephilippis, et al. 2001; Morgenstern, Riordan, et al. 2001b) and found significantly elevated patterns of substance dependence. In a specialized screening program of sanctioned recipients in one New Jersey county, 49 percent of the sanctioned clients interviewed met the criteria for a substance use disorder.

   Those persons who failed to comply with the program's requirements or who failed to find employment within specified time limits were more likely to have such disorders. Furthermore, recipients who left welfare because of sanctions were less likely to be working than were those who left welfare for other reasons (Pavetti and Bloom 2001). The use of alcohol and other substances is especially common among families who come to the attention of child welfare and child protective services (Morgenstern 1999; Morgenstern, Riordan, Dephilippis, et al. 2001; Morgenstern, Riordan, et al. 2001a; Waldfogel 2000).

   Using WES data, Pollack and collaborators (2002b) found that work status was the most powerful predictor of substance dependence. Current and former TANF recipients who worked less than 20 hours per week were more likely than other WES respondents to satisfy the criteria for drug or alcohol dependence. When controlling for work status (Michigan's comparatively high benefits allow many recipients to combine TANF with paid work), continued TANF receipt was not associated with an increased prevalence of substance use or dependence.

   Some evidence correlates illicit substance use with the duration of welfare receipt. WES researchers found that participants who reported using illicit substances over several years were more likely to accumulate additional months on TANF (Seefeldt and Orzol 2003). This raises the question of whether self-reported substance users would eventually receive welfare for longer periods. A study of African-American women in one Chicago community indicated that women who received cash aid for five or more years were more likely to report recent marijuana or cocaine use than were nonrecipients or women who received cash aid for less than five years (Williams, Juon, and Ensminger 2004).

Gaps in Prevalence/Severity Research

Capturing the prevalence and severity of substance use and substance use disorders among the welfare population is difficult. Relying on self-reports may be the most serious shortcoming of available data. WES, NHSDA, NLSY, and other prominent data sets used in policy research all rely on self-reports, making deceptive or inaccurate responses important concerns (Gfroerer, Wright, and Kopstein 1997; National Institute on Drug Abuse 1985). In addition, self-reports are likely to understate true prevalence. Although the extent and correlates of underreporting are not fully known, several studies have shed light on these concerns.

In the case of alcohol, Midanik and colleagues (Midanik 1982; Midanik and Greenfield 2003) provided the most extensive analysis, whose comparisons of telephone and in-person interview modalities are particularly useful. Perhaps for reasons of social desirability, telephone interviews yield higher self-reported prevalence of alcohol-related harm (Midanik, Greenfield, and Rogers 2001).

Fendrich and colleagues used chemical testing as an adjunct to a general population household survey in Chicago (Fendrich et al. 2004a; Fendrich et al. 2004b). They found heroin and cocaine use to be widely underreported, as indicated by hair and urine tests. Of those who tested positive for cocaine or heroin, less than 20 percent reported that they had used these substances in the last 30 days. Respondents offered far more candid responses in the case of marijuana use. A multivariate analysis of detected cases of cocaine, marijuana, and heroin use indicated that African Americans and women were less likely than others to have revealed their use. More than 90 percent of the survey respondents agreed to participate in at least one form of chemical testing, with oral fluid tests proving the modality most acceptable to them. Such findings highlight the potential feasibility of chemical testing to complement traditional household surveys.

Two groups of researchers explored self-reporting by TANF recipients. In an urban New Jersey sample of TANF recipients, Kline and colleagues found that between half and two-thirds of cocaine users identified through chemical testing did not report their use (Kline et al. 1998). Podus and colleagues found similar underreporting patterns in a southern California urban sample (n= 511). Of those TANF recipients who tested positive for recent opiate or stimulant use, 92.3 percent reported that they had not used in the last three days. Based on all respondents, whether or not they volunteered to be tested, 0.6 percent reported using cocaine, amphetamines, or opiates. Survey data were more complete in the case of marijuana use. Of those respondents who tested positive for marijuana use, 72 percent reported that they had used this substance within the past 30 days (Podus et al. 2002).

To our knowledge, Michigan is the only state that has tested TANF recipients not otherwise suspected of substance use. In October 1999, Michigan began mandatory testing in three welfare offices. All new recipients were required to provide urine tests as a condition of eligibility for aid. Testing was halted by judicial order in November 1999 (for further citations and discussion, see Pollack et al. 2002b). Out of 258 tested recipients, 21 tested positive for illicit drug use. All but three of these 21 recipients tested positive for marijuana alone.

Michigan's urine test technology was capable of detecting marijuana residue that occurred several weeks before samples were drawn. The technology would not have detected heroin or cocaine used more than 72 hours earlier (Fendrich et al. 2004a). Despite this limitation, the low prevalence of detected drug use supports findings from household survey data that illicit drug use (and use disorders) is uncommon among TANF recipients.

Lack of Consistent Trend Data

Another limitation is the trend data. No single data set allows researchers to construct a consistent historical time series for the prevalence of abuse and dependence among welfare recipients. NHSDA/NSDUH—the only annual, nationally representative survey covering recent years—has used different questions in different years to capture substance use disorders.

Need for More Longitudinal Studies

Researchers must also address the issue of causality in the relationship among substance use, welfare receipt, and employment. Few longitudinal studies have followed welfare recipients (and potential recipients) with substance use disorders over time to observe welfare and work patterns in the post-TANF era. Existing longitudinal studies of welfare recipients could address some of these difficulties by including additional measures of substance use and use disorders.

Because welfare participation is in itself an endogenous variable reflecting unobserved individual preferences and circumstances, as well as characteristics of local welfare systems, longitudinal surveys should include data from low-income families who are not currently receiving TANF. Such longitudinal surveys could help address complex selection issues in administrative data sets. For example, persons with substance use disorders may be discouraged from receiving TANF because of the possibility of increased monitoring, screening, and assessment. TANF also provides disposable income, which could be used to buy such substances.

Substance Use Disorders

PRWORA has removed many low-income mothers and families from the TANF program and may have deterred others from applying in the first place. Because TANF includes fewer low-income families with dependent children, the needs of low-income nonrecipients may require greater research and policy attention.

Of those low-income women who satisfy the criteria for substance dependence or abuse, Reuter, Pollack, and Ebener (2001) found that receiving TANF was associated with a greater probability of receiving treatment. This pattern may reflect improved access to treatment associated with TANF recipients’ entitlement to Medicaid. Among WES respondents, mothers who neither worked nor received cash assistance were more apt to be substance users than were any other subgroup in the WES sample. Low-income substance users outside the TANF system may have significant needs and may have limited access to treatment services. Because WES was based on a sample of women who received AFDC or TANF, these data do not cover low-income mothers who use illicit substances but have not received cash aid. At present, little is known about this population (Reuter, Pollack, and Ebener 2001).

The needs of nonrecipients and of hard-to-serve recipients are especially pressing when one considers marginalized populations who make up a small proportion of low-income individuals but who face especially high alcohol- or drug-related risks. Such populations include persons under the supervision of the criminal justice system, homeless individuals, commercial sex workers, and others at high risk of contracting HIV and other infectious diseases.

The Impact of PRWORA on Criminal Offenders

PRWORA enacted strict penalties such as the Gramm Amendment against individuals convicted of drug-related felonies. Some women barred from TANF owing to felony convictions for the sale or distribution of illicit drugs also had substance use disorders. Few studies documented substance use disorders among pertinent felony drug offenders (Sherman and Latkin 2002).

Twenty Percent Exclusion

States may exempt up to 20 percent of the TANF caseload from time limits on the receipt of federally funded aid. The epidemiological foundations of this threshold remain obscure, particularly given large reductions in AFDC/TANF caseloads. From the beginning, researchers have questioned whether this proportion is large enough to cover all those who might need additional assistance (Danziger and Seefeldt 2002). The challenges facing drug users and other potentially exempt individuals remain unknown.

Administration and Implementation of Screening, Assessment, and Treatment Services

Most persons with substance use disorders do not seek treatment of their own accord (Danziger and Seefeldt 2002; McCrady and Langenbucher 1996; Morgenstern, McCrady, et al. 2003; Morgenstern, Riordan, Dephilippis, et al. 2001; Weisner and Schmidt 1995). Screening and assessment for substance use disorders are therefore especially important. In one survey of TANF recipients in Alameda County, California, 12 percent of respondents reported having taken illicit drugs at least weekly during the past 12 months. Only 4.2 percent reported needing assistance with an alcohol or drug problem in the past 12 months. Only 1.8 percent had received treatment, and only 2.1 percent reported that they currently needed treatment (Speiglman et al. 1999). This trend was consistent in follow-up studies of welfare recipients in Alameda County (Dasinger, Speiglman, and Norris 2002; Speiglman, Dasinger, and Norris 2003).

Local welfare offices may not provide sufficient screening and assessment to reliably identify substance users and to assess the full range of accompanying personal and family concerns. In addition, this process should ideally lead to the provision of effective and appropriate services in order to improve the lives of TANF recipients and their children.

During the winter of 2001/02, the Legal Action Center surveyed TANF agencies in the 50 states and the District of Columbia to explore how they were addressing alcohol and drug problems as barriers to self-sufficiency in their welfare caseloads (LAC 2002). From the 86 percent response rate of these agencies, 59 percent of the states said that they were screening TANF recipients for alcohol and drug problems. Sixteen percent reported that screening decisions were up to the individual counties, and 25 percent stated that they were not screening for any substance use.

Typical screening and assessment methods included a brief questionnaire such as the CAGE (Cut, Annoy, Guilty, Eye-opener), which asks four to ten questions. This was the most common screening method used by 58 percent of the states as presented in the Legal Action Center study. Other frequently used measures were the Substance Abuse Subtle Screening Inventory (SASSI) and the Addiction Severity Index (ASI). Widely used screening measures such as the SASSI (Miller, Roberts, and Brooks 1997) and the ASI (McLellan et al. 1980; McLellan et al. 1992) are short measures used to identify individuals with a high probability of having a substance use disorder. Notably, 35 percent of the states indicated that they had developed their own tool. Eighty percent relied on TANF workers to conduct the screening, although they provided little, if any, training on how to identify alcohol and substance use disorders. Almost all states (96%) that reported screening TANF recipients for alcohol and drug problems indicated that those who screened positive were referred for a comprehensive assessment. Thus, the states vary widely in their handling of TANF recipients’ alcohol and drug problems.

We should note that this study focused on practices identified at the state level and did not survey welfare offices at the county level, where many of the decisions regarding screening and assessment are made and where public policy is actually implemented.

A GAO study of welfare offices in 600 counties examined the ways in which welfare programs identified recipients who faced various barriers to employment and who might need specialized services. While most counties did some screening, more than three-quarters of welfare offices relied on recipients’ self-disclosure of barriers instead of using standardized screening instruments and assessment tools (GAO 2001). This study also found that many recipients in these welfare offices did not receive services to address identified barriers. Sixty-three percent of the counties completely exempted from work participation requirements those TANF recipients with physical or mental impairments. In these counties, exempted recipients were much less likely than other recipients to receive services to address both their impairments to improve self-sufficiency and other concerns (GAO 2001).

Morgenstern and colleagues conducted a preliminary evaluation of two approaches to screening in New Jersey. The first approach, which they termed the “generic approach,” had welfare caseworkers administer to all those being interviewed for initial or redetermination of TANF benefits a short paper-and-pencil measure designed to screen for alcohol and other drug use problems. Those persons who responded positively to two or more items were referred to a trained substance use counselor for further evaluation. This approach was implemented statewide in New Jersey beginning in 1998. Using this approach, an analysis of state records indicated that approximately 1 percent of welfare recipients received a referral for substance abuse assessment during the first six months of screening and that this number increased in the succeeding 12 months to 4.4 percent. Morgenstern (1999) suggested that the low rates of identification were partly due to the lack of training of front-line caseworkers, the recipients’ reluctance to disclose their substance use, and the inadequacy of the paper-and-pencil measures.

The second approach evaluated by Morgenstern and colleagues was “specialized screening,” which contained three components (Morgenstern, Riordan, Dephilippis, et al. 2001). First, an interview format was used to establish rapport with recipients. Second, specially trained staff (not necessarily addiction professionals) conducted the interviews. Third, an additional, more intensive screening procedure targeted recipients considered to be at high risk for substance use disorders.

Two specialized screening programs were implemented in New Jersey. In the first program, all sanctioned recipients in one New Jersey county (n= 352) were required to be assessed by a substance abuse counselor as part of the process of restoring benefit eligibility. Letters were sent to these clients, and 86 (24%) responded to the letter and agreed to be interviewed. Of these, 49 percent (n= 42) who were interviewed met the criteria for a substance abuse disorder.

In the second program, two welfare caseworkers with a special interest in assisting substance abusers were assigned to conduct specialized screening in the New Jersey county with the highest caseload. These caseworkers interviewed high-risk clients (n= 853), such as those requesting emergency assistance or persons identified as having a potential disorder through a short screening questionnaire. Of the screened individuals, 36.5 percent (312 of 853) were referred for further evaluation. Thirty-two of the 312 referred women, approximately 3.75 percent of the original screened sample, were identified as having a substance abuse disorder and were referred for treatment. These results suggest that specialized screening programs may help increase identification rates.

As noted by Morgenstern and colleagues, future experimental studies could establish the effectiveness of these approaches. This screening program was undertaken in one New Jersey county and may not provide a representative sample of TANF recipients. It is nonetheless noteworthy that the proportion of recipients referred for treatment was similar to the estimated prevalence of illicit drug dependence in the WES and NSDUH survey data we described earlier.

Few Interventions Conducted to Address Multiple Barriers

Given the many barriers faced by substance-using welfare recipients and their families, best-practices programs should address multiple barriers and also recognize that substance abuse is a chronic condition (McLellan 2002; McLellan et al. 2000). Few comprehensive treatment programs or interventions have been rigorously evaluated for identifying substance-using welfare recipients and their children.

Morgenstern and colleagues conducted a randomized intervention trial to evaluate two intervention approaches targeting welfare recipients who were identified as having substance use disorders (Morgenstern, Riordan, et al. 2001a; Morgenstern, Riordan, et al. 2001b; Wilkins 2003). These approaches were care coordination and intensive case management.

For the care coordination intervention, TANF caseworkers referred clients with substance abuse problems to addiction counselors, who arranged for treatment and follow-up. In addition, counselors reviewed the state's welfare work requirements and time limit statutes. According to these regulations, women who did not attend treatment or engage in a work activity could lose their benefits.

In the intensive case management intervention, welfare recipients with substance use disorders were assigned to a pair of case managers who worked with them to identify and address barriers to entering and remaining in treatment. Case managers also addressed the recipients’ resistance to entering drug treatment and used motivational counseling to encourage their acceptance of treatment. In some instances, case managers used extensive outreach, including home visits and contact with family members. Once the recipients entered treatment, case managers worked with the treatment programs to help coordinate the needed services and continued to meet with the recipients each week for up to two years. The participants in this intensive arm also received vouchers for attending drug treatment.

Preliminary outcome findings indicate that those recipients who received intensive case management were more likely to enter drug treatment and to continue outpatient care (Morgenstern, Nakashian, et al. 2003; Wilkins 2003). Eighty-eight percent of an initial cohort of 155 recipients who were assigned to the intensive case management arm entered treatment, compared with 65 percent of the recipients assigned to the care coordination arm. Eighty-six percent of those in the intensive case management arm also participated in outpatient treatment, compared with 53 percent of those in the care coordination arm.

Differences in the rates of entry into inpatient treatment were suggestive but not statistically significant (78% versus 69%). Although those recipients in the intensive case management arm were more likely to remain in drug treatment, retention remained a challenge in both groups. Case management participants attended 42 percent of the days they were assigned to treatment, compared with 22 percent of days for participants in the care coordination arm. This difference was particularly pronounced for the outpatient treatment. Case management participants, on average, attended more than four times the outpatient sessions (M = 29.7, SD = 30) than were attended by participants in the case coordination arm (M = 6.6, SD = 13).

Participants were assessed nine months after their recruitment, in order to document their employment as well as cessation of drug and alcohol use (Morgenstern et al. 2002). At this follow-up, 50 percent of the intensive case management participants and 40 percent of the care coordination participants reported completely abstaining from drugs and alcohol. But the rates of participation (from baseline to follow-up) in work and training activities for both groups did not rise. On average, participants in both intervention groups reported working 1.5 to two days per month in the first three months after recruitment and three to four days per month at the nine-month assessment.

Although these findings are preliminary, the authors (Morgenstern, Riordan, Dephilippis, et al. 2001) concluded that both care coordination and intensive case management could be useful to introducing substance abuse treatment into welfare programming. Most of the participants in each group (88% and 65%) entered drug treatment after receiving an evaluation and referral. Case management, however, appeared more successful in attracting substance-using welfare recipients to outpatient care and retaining them once they entered treatment. This pattern matches previous findings that case management could improve substance abuse treatment outcomes (Shwartz et al. 1997; Siegal et al. 1997).

The CASAWORKS for Families program is an integrated, multiservice intervention designed for women on TANF combining substance abuse treatment, work-related services (including work readiness, vocational training, and basic education), mental health care, and parenting services, with case management serving as the primary integrator of all these services (Morgenstern, Nakashian, et al. 2003; Woolis 1998). Additional services to be delivered as needed include medical care, child care, and assistance with transportation, shelter, and clothing. CASAWORKS was designed to address the many barriers and problems facing substance-using women on welfare by integrating numerous social service agencies to deliver comprehensive, concurrent services that would be required to alleviate these barriers and problems.

An initial pilot demonstration was evaluated at ten treatment sites in nine states. The pilot included a field evaluation with repeated measures at six and 12 months on a sample of 366 women who were receiving TANF, who had not been working in the past month, and whose screening had shown presumptive evidence of a substance use disorder in the preceding month (McKay et al. 2003; McLellan et al. 2003; Morgenstern, Nakashian, et al. 2003; Woolis 1998).

Significant improvements were demonstrated in employment, substance use, and family and social functioning over a six-month interval. After 12 months, participants had significantly reduced their substance use, with almost half (46%) reporting complete abstinence from all substances over the previous six months. The proportion of clients who had worked at least half-time also greatly increased. At baseline, 16 percent of CASAWORKS participants reported having worked at least half-time in the month before entering the intervention. This percentage rose to 41 percent by the 12-month assessment. No significant improvements, however, were demonstrated in medical or psychiatric status.

This initial demonstration did not have a comparison or control group. A random assignment experimental study of CASAWORKS is now under way in New York City. One of the original pilot sites—North Carolina—built the CASAWORKS for Families model into its request for proposals (RFP) for treatment services for women (Wilkins 2003).

Few traditional outpatient drug treatment programs have been developed and evaluated to address the most important barriers to TANF recipients. Traditional treatment programs are often designed to meet the needs of male patients and clients, and many do not provide services such as employment counseling, mental health services, and support in their performance of family roles (CSAT 2000; Nakashian 2002; Schottenfeld, Pascale, and Sokolowski 1992).

Studies have shown that residential treatment is difficult or unfeasible without affordable, appropriate, and attractive child care services (Hughes et al. 1995; Szuster et al. 1996; Wobie et al. 1997). Moreover, TANF recipients often require or request help with parenting skills and other child welfare services during their substance abuse treatment (Volpicelli et al. 2000).

Vocational services, in particular, have been linked to positive employment outcomes for drug treatment participants. Luchansky and colleagues (2000) followed clients over a 4.5-year follow-up period in administrative databases. Controlling for background characteristics and pretreatment earnings of those clients who completed their treatment, they showed that those clients who received vocational services in addition to substance abuse treatment earned an average of $138 per month more than did otherwise similar clients who had received treatment alone. This is a substantial increase for low-income mothers. (Working current and former TANF recipients in the WES study reported mean 1999 monthly earnings of $955 per month; see Danziger et al. 2002.) The Luchansky analysis, however, excluded welfare (AFDC) clients.

Time spent in treatment is the best single predictor of treatment success (Howell, Heiser, and Harrington 1999). Time spent in treatment also has been shown to have a modest but positive and statistically significant impact on labor market outcomes such as posttreatment weeks worked and earnings (Condelli and Hubbard 1994; French et al. 1991). Analyzing data from the Treatment Outcome Prospective Study (TOPS), French and colleagues (1991) showed that an average of one additional week in treatment raised annual earnings for methadone clients by $21 and for residential clients by $44 during the year following drug treatment. Hubbard and colleagues (Hubbard et al. 1989) found that drug treatment clients needed at least 12 months of treatment to expand their full-time employment and to lower their drug use.

Such findings suggest that services allowing or encouraging women to remain in treatment produce better outcomes. Recent studies have questioned length of stay as the primary predictor of successful treatment outcomes. These studies argue that completing the treatment is important (in addition to length of stay) to explaining subsequent outcomes (Arria and Group 2003; Zarkin et al. 2002). In a randomized study examining three-month and six-month doses of outpatient treatment, Kamara and Van der Hyde (1998) found no major differences in posttreatment drug use or employment. They did, however, find significant duration-related differences among clients who subsequently found employment, with clients in the longer-dose treatment having fewer work-related problems.

Studies of welfare recipients before and after PRWORA demonstrate favorable work outcomes for those who were treated for substance abuse. These improvements reflect both increased employment and increased earnings when employed (Kirby and Anderson 2000; Metsch et al. 2003; Wickizer et al. 2000).

Statewide administrative database studies also indicate that participation in substance abuse treatment was associated with more probable employment and less receipt of welfare (Metsch et al. 2003; Wickizer et al. 2000). Examining data from Washington State, Wickizer and colleagues investigated the effects of substance abuse treatment on employment outcomes among AFDC recipients admitted to treatment during a two-year period beginning in 1994 (Wickizer et al. 2000). Their findings demonstrated that exposure to drug treatment was associated with a greater probability of employment and higher earnings for those who were hired. Sixty-six percent of AFDC clients who received in-patient treatment (n= 629) had some positive earnings after treatment. Only 50 percent of a comparison group of participants (n= 260) who received only detox and no other treatment reported positive earnings. Wickizer and colleagues also showed that clients who received outpatient treatment (compared with other treatment modalities) had the highest annualized earnings over a two-year follow-up period.

Metsch and colleagues (2003) conducted a similar administrative database study of 4,236 drug treatment participants in Florida, finding that length of stay and treatment completion were associated with increased employment and decreased welfare receipt. Participants who completed drug treatment were 1.4 times more likely to be working and to have left welfare at six months after their discharge from treatment than were those participants who did not complete their treatment. The odds of working and being off welfare increased with each month of drug treatment.

In both studies, exposure to drug treatment, time spent in treatment, and treatment completion were associated with favorable employment outcomes. These observational studies did not address unobserved patient characteristics correlated with treatment duration or completion that also may be correlated with posttreatment outcomes. Moreover, little is known about the treatment components that are most helpful to welfare recipients moving from welfare to work. The paucity of randomized trial data prevents researchers from definitively examining these links.

Gaps in Research on Administration, Implementation, and Treatment

Little systematic research, with the exception of that by Morgenstern and the Legal Action Center (2002), has explored the quality and range of screening and assessment services, or whether individuals deemed in need of specific services actually receive them. Many questions remain regarding best-practice approaches to screening, assessment, and referral regarding substance use issues in the welfare setting. Data are needed, as well, regarding effective approaches to training welfare personnel on how to screen and assess for substance use and substance use disorders. Policymakers, researchers, and administrators must determine whether TANF recipients are linked to appropriate services and must document the barriers that recipients face in treatment engagement, retention, and completion (LAC 2002).

The impact of subclinical illicit substance use remains unclear, and it is uncertain whether typical welfare office personnel are appropriately trained to address these issues. In both the NHSDA/NSDUH and WES samples, many women who reported recent illicit substance use also said that they experienced some adverse impact of substance use on their economic well-being, family function, or relationships with others. Few of these respondents fully satisfied the DSM-IV criteria for dependence or abuse. Thus, few appeared to satisfy the entry criteria for substance abuse treatment, in which the full impact of substance use can be more carefully assessed and explored. The impact of such subclinical symptoms remains unclear, although the literature on alcohol “diagnostic orphans” suggests that such individuals face some elevated risks (Eng, Schuckit, and Smith 2003). “Diagnostic orphans” satisfy some of the DSM–IV abuse or dependence criteria but do not meet the full criteria for substance use disorders.

Further research is also needed regarding the impact, cost effectiveness, and proper implementation of ancillary services for TANF recipients. Substance abuse treatment is not a single product. It is a bundle of related health and social services, including some services that may continue after an individual has completed the main elements of standard inpatient and outpatient treatment interventions. Overall, further research is needed regarding how to best integrate services that meet the multiple needs of substance-using welfare recipients and also are cost effective (Hilton et al. 2003; Schmidt and McCarty 2000). Research is especially needed on the impact of sanctions to explore what happens when persons who experience substance use disorders are removed from welfare support.

Another gap in current research concerns the number of substance users and persons with substance use disorders who may never be seen in the welfare offices or who leave welfare without achieving self-sufficiency. Nonworking former TANF recipients are a particularly important group, about whom little is said in the policy literature (Zedlewski et al. 2003). Research is required to explore how these persons support themselves and their families financially when not working, and whether they have other, more serious concerns, including severe poverty, criminal involvement, child neglect and abuse, domestic violence, or other family problems.

Chandler and colleagues (2004) showed that many substance-using TANF recipients in two California counties who left welfare had no employer or self-reported earnings. Such populations are likely to include the most seriously affected drug users who may be homeless, have a severe mental illness, or have an elevated risk of contracting HIV. Sanctioned recipients may be more likely to engage in illegal activities to earn money or to buy illicit substances.

Criminal offenses may also lead to incarceration and the subsequent loss of welfare benefits (Mulia and Schmidt 2003). A recent longitudinal study of TANF recipients (both substance users and nonsubstance users) conducted in Houston showed that involvement in criminal activities fell over time, although substance users were more likely to commit crimes than were their nonusing counterparts (Brown et al. 2004).

Recipients with substance use disorders may have difficulty adhering to the welfare system's stringent requirements. Schmidt and colleagues (2002) showed that substance dependence was a strong predictor of persons exiting welfare for administrative reasons. In a qualitative focus group study, substance-using women reported that they felt stigmatized and disrespected when interacting with welfare officials (Bush and Kraft 2001; Schmidt et al. 2002). It is unknown whether issues of stigma and disrespect are more common among substance-using women than among other subgroups of TANF recipients.

One promising arena concerns intensive outreach and retention strategies for substance-using women under criminal justice supervision. Hammett, Gaiter, and Crawford (1998) offered one overview of pertinent concerns and strategies. Welfare offices might consider strategies to meet the basic needs of TANF recipients who persist in substance use, abuse, or dependence and yet may not be ready to become abstinent or to enter drug treatment (Kirby et al. 1999).

Dunlap, Golub, and Johnson (2003) conducted an ethnographic study between 1995 and 2001 to examine the impact of welfare reform on substance-using households in New York City. They found that nearly all the surveyed households had lost their welfare benefits, had difficulty with work programs, and reported additional difficulties covering basic expenses.

Conclusion

Substance use disorders affect the well-being and self-sufficiency of some low-income mothers. Like other difficulties, such as physical health ailments, psychiatric disorders, poor academic skills, and transportation difficulties, substance use disorders hinder efforts to find and keep work, and such disorders require effective screening, assessment, and intervention.

Although much remains unknown about patterns of substance use among TANF recipients, the existing data indicate that such disorders affect a small minority of TANF recipients. These disorders are less common than many policymakers and advocates originally assumed and are easily overstated. Eliminating substance use disorders among TANF recipients would have only a small impact on the size and composition of the population of families receiving public aid. There is little direct evidence that substance use or substance use disorders are more common among welfare recipients today than they were before welfare reform. Measured prevalences of substance use were lower among TANF recipients in 2001 and 2002 than they were ten years earlier.

Although these disorders are not common among TANF recipients, substance abuse and dependence do raise policy concerns. Most substance users receiving public aid do not satisfy the clinical criteria for abuse or dependence. Many, however, face a higher risk for other personal and familial concerns. The high prevalence of psychiatric disorders among substance users is particularly worrisome. Effective services for these disorders can improve outcomes for an important, difficult-to-serve subgroup of families dependent on public aid.

Substance use disorders are likely to be more prevalent in specific populations of hard-to-serve TANF recipients. Data regarding hard-to-serve populations are still emerging in the changed environment after welfare reform. Early evidence suggests that sanctioned recipients and nonworking recipients are more likely to have substance use disorders. Screening and assessment services for these populations should pay special attention to substance use disorders.

TANF recipients who experience substance use disorders often require a mix of pertinent services. Despite many program evaluations and proposed best-practice interventions, questions remain about how best to promote favorable long-term outcomes for adult recipients and their children. TANF recipients often require vocational assistance and assistance with child care during their substance abuse treatment. Given the correlation between substance use disorders and child abuse or neglect (Young and Gardner 1997), effective interventions for TANF recipients should also include help with parenting skills, careful assessment of their children's well-being, and appropriate provision of subsequent services.

Some substance users—and some other recipients of public aid—face multiple and complex challenges that fit uneasily into the TANF vision of time-limited public aid. Welfare recipients with substance use disorders often face other barriers to finding employment. Although work and treatment are sometimes considered in opposition to each other in policy discussion, the development of work skills is an important goal of many treatment interventions. Some treatment interventions also provide support and monitoring to help clients succeed in jobs that they take during the latter stages of their treatment. Such services may be especially important to TANF recipients, although the peer-review research base is limited regarding the effectiveness and cost-effectiveness of such efforts.

It also is important to recognize that PRWORA was enacted during a period of national prosperity and during a period of declining general drug use in the U.S. population. Whether the welfare system will be successful in more difficult economic times is not known. Substance abuse policies may also face greater strains if the prevalence or severity of substance use increases in the TANF population, although according to recent data, this does not does not appear to be occurring (Danziger and Seefeldt 2002; Pollack, Danziger, Seefeldt, et al. 2002). But we cannot rule out subsequent changes in drug markets or in the epidemiology of substance use that would place greater strain on TANF recipients and on the system of substance abuse treatment.

Welfare reform is a political success because it achieved its main stated goals: converting welfare from an open-ended entitlement to a transitional, work-oriented program, discouraging dependence on welfare, and reducing the number of families receiving public cash aid. Whether welfare reform has improved the well-being of low-income families is less clear. The 1996 reforms did not end, or seek to end, poverty and unemployment in America. Welfare reform has spawned new efforts to address the complex challenges associated with substance use. The success of these efforts remains unclear, although clearer findings are emerging.

Given apparent public consensus in favor of time-limited aid, the central challenge facing TANF echoes challenges confronted in other social policy arenas: to execute interventions that are broadly supported in concept but are difficult to implement well (Mashaw 1983; Nathan 1993). Local welfare offices need the resources and expertise to screen and assess individuals who may have substance use disorders. States and localities need appropriate treatment interventions to meet the many needs of substance users who receive TANF aid. Given the declining number of caseloads, states and localities must strengthen outreach and referral services for low-income mothers who do not receive TANF aid. Best-practice models are emerging to address these challenges.

Matters of implementation and organizational capacity attract less attention than did the large political debate that resulted in the 1996 reforms. Yet these quieter issues shape the reality of TANF experienced by program recipients and so are central to welfare reform's ultimate success.