Methods and overall conclusions
This chapter analyzes the effects of substance abuse treatment on post-treatment outcomes, including drug use, income sources, employment, criminal activity, health, and living arrangements. The principal focus throughout this chapter is on female treatment clients with children, especially those who received welfare income. As in the original CALDATA report, the estimates of treatment effects in this chapter use the "before-after" or "pre/post" research design, comparing the same behaviors/characteristics of the same treatment clients before and after treatment. As in the CALDATA report, we depend on the retrospective reports of sample clients at the time of the post-discharge interview to measure the presence or absence and the levels of behaviors/characteristics during the before-treatment and after-treatment reference periods. While this chapter's findings are consistent with the hypothesis of beneficial treatment effects, there are other possible explanations for these patterns of results including nonresponse biases (perhaps fewer of those who did poorly after treatment participated in the research), under reporting of negative outcomes in the post-treatment period by those who did participate, and simple regression to more typical patterns of behavior after treatment than the behavior that happened to occur immediately before. The CALDATA report includes a detailed discussion of the assumptions underlying CALDATA methods and of the approaches to estimation and measurement that are applied in this chapter, including the reasons why these alternative explanations are less persuasive than the hypothesis of beneficial treatment effects.
The main conclusions of this chapter are as follows:
- Among women with children who received welfare income, the number of drug users (those who used a specified drug more than five times per annum) fell after treatment by about 39 percent in the case of crack cocaine, 42 percent in the case of cocaine powder, 14 percent in the case of heroin, 48 percent in the case of amphetamines, and 26 percent in the case of alcohol. (For example, 32.1 percent of this subgroup used crack cocaine before treatment, while only 19.6 percent used crack cocaine after treatment, for a percentage decline of about -38.9 percent, i.e.,
(19.6-32.1) / 32.1= -38.9%.
- Among women with children who received welfare income, the percentage who sold or helped to sell drugs fell by about 60 percent after treatment, and the percentage who were arrested, booked, or taken into custody dropped by about 54 percent after treatment.
- Among women with children who received welfare income, the percentage hospitalized during the course of one year fell by about 58 percent after treatment.
- Among women with children who received welfare income, the percentage who were homeless for two or more days during the year declined by about 61 percent after treatment.
- The proportion of the whole treatment population on welfare stayed about the same before and after treatment among men and women, with attrition from the welfare-receiving group balancing accession to that group. Of the minority of women who received welfare prior to treatment, 30 percent no longer received welfare during the year after treatment; however, of the majority who did not receive welfare prior to treatment, 16 percent were receiving welfare after treatment.
- The proportion working for wage or salary income declined by about 20 percent after treatment. We believe these results were strongly influenced by the overall doubling of the unemployment rate in California across the period covered by the study.
- In general, women with children, including those receiving welfare income, appeared as likely to benefit from drug abuse treatment as any other treatment clients. With very limited exceptions, the positive outcomes of treatment were similar among women with children who received welfare, women with children who did not receive welfare, women without children, and men.
The following three sections present a detailed analysis of the effects of treatment on the behavior and well-being of men, women, and key subgroups of women, including those living with children who received welfare income and who did not receive welfare income. These sections present representative data (and describe other data that we have analyzed) on several classes of outcome variables: drug use, employment, welfare receipt, crime, health, and homelessness.
This analysis focuses on the five "main drugs," i.e., the five drugs that were most commonly mentioned by treatment clients as reasons for entering treatment: 1) crack cocaine (a crystalline or "free-base" form of cocaine best suited for smoking), 2) cocaine powder (the hydrochloride salt of cocaine, which is water-soluble and can be injected or snorted), 3) heroin, 4) amphetamines, and 5) alcohol. Both among treated women with children and among other clients, close to 50 percent of clients identified heroin as a reason for their entering treatment, while close to 25 percent identified alcohol, and crack, cocaine powder, and amphetamines (see Chapter 3) each were identified by more than 10 percent of clients.
For each of the five drugs, treatment significantly reduced drug use from before to after treatment (Figures 4.1- 4.5). The measure of drug use in each of the figures below was the percentage of treatment clients who used the drug five or more times in the previous year, which we call "prevalence." For crack (Figure 4.1), prevalence in the total treatment population declined from 28 percent before treatment to 16 percent after treatment, a percentage change of -42 percent, i.e., (16 - 28)/28. For cocaine powder, heroin, amphetamines, and alcohol, the percentage changes of prevalence in the total treatment population were -46 percent, -23 percent, -47 percent, and -29 percent (Figures 4.2 4.5), respectively.
Significant declines in drug use following treatment occurred regardless of whether the treatment client was a woman, had children, received welfare income, or had lost custody of a child or identified parenting issues as an important reason for entering treatment.6 In the subgroup of women with children who received welfare income, the percentage changes in prevalence for crack, cocaine, heroin, amphetamines, and alcohol were -39 percent, -42 percent, -14 percent, -48 percent, and -26 percent, respectively. These percentage changes are closely similar to, and not significantly different from, the corresponding percentage changes for each drug in the total treatment population. Drug use declined significantly across the board. For each of the five drugs, the prevalence among women with children who received welfare income declined by about the same percentage as the corresponding percentage among treatment clients as a whole.
Figures 4.1-4.5 suggest that, in general, the magnitudes of the effects of treatment on drug use were similar whether or not the treatment client was a woman, had children in the household, or received welfare income. (In additional data analyses, we determined that outcomes also did not vary depending on whether the client had lost custody of a child or named parenting concerns as important reasons for their entering treatment.) There is little indication in any of these data that women with children who received welfare income differed in their response to treatment from women with children who did not receive welfare income or from other treatment clients. For each of the five drugs, the prevalence among women with children who received welfare income declined by about the same percentage as the corresponding percentage among treatment clients as a whole.
We tested a variety of alternative measure of drug use before and after treatment, such as the mean number of different drugs used during the previous year, the number of months specific drugs were used, the intensity of use during the period of highest use, and the number of drugs used across the period. These measures all led to very similar conclusions about the effects of treatment on drug use. For example, in the overall treatment population, the total number of drugs used by each client declined from 2.8 before treatment to 1.6 after treatment. In the subgroup of women with children who received welfare income, the decline was almost exactly the same.
Income sources and employment
CALDATA clients reported receiving income from a variety of sources, including full-time and part-time employment, welfare, illegal activities, and disability income. Figures 4.6 and 4.7 present estimates for two of six before-after comparisons we completed regarding income and employment (all figures were adjusted to represent an annual basis). The six measures we analyzed were:
- percentage who received employment income (Figure 4.6);
- percentage who received welfare income (Figure 4.7);
- number of months worked full-time;
- percentage who worked full-time;
- number of months worked part-time;
- percentage who worked part-time.
In the total treatment population, the percentage receiving employment income declined from about 55 percent before treatment to about 45 percent after treatment, a relative decline of about 17 percent in job-holding. Men were much more likely to have worked before treatment than women (63% of men versus 41% of women), but for both groups, employment declined by a similar proportion. In every subgroup of clients, the percentage receiving employment income (that is, having had a job during the year) appears lower after treatment than before, and it was significantly lower for all women, all men, and for women (as well as men) with children who had not collected welfare. Women who did collect welfare reported the lowest rate of employment before treatment (25%) and a nonsignificant change to 23 percent after treatment. The percentages working full-time, the average number of months they worked full-time, and the part-time statistics are all consistent with the results in Figure 4.6.
Source: NORC CALDATA
Despite the overall reduction in employment during the time period analyzed in this report, there was no significant change in the percentage receiving welfare income in the CALDATA population as a whole or among men and women considered separately: women were twice as likely as men to have collected welfare in the period after treatment as well as before treatment (Figure 4.7). There were significant reductions in welfare receipt in the subgroups where all members were initially welfare recipients for example, among women with children who received welfare income before treatment, only 78 percent did so after treatment but these decreases were balanced by the onset of welfare receipt treatment among the subgroups selected to exclude welfare before treatment.
Both the welfare and employment results are consistent with the generally austere economic trend in California over the period 1991-1993 (see below, Table 5.2), which makes interpretation of treatment effects difficult in the area of employment and income. The problem was that the downward secular trends in the California economy during the period of the CALDATA follow- up—a doubling of the unemployment rate (see Chapter 5) and immense fiscal pressure on the state government probably obscured any independent positive effects that treatment might have had on treatment clients' prospects for employment and income.
Source: NORC CALDATA
Crime, health, and living arrangements
Substance abuse treatment can affect the prevalence and incidence of criminal activity in at least two ways. Treatment can directly affect criminal activity by providing new reference groups and new moral and ethical standards to substitute for reference groups and standards that helped to engender criminal activities in the past. Treatment can also affect criminal activity indirectly by reducing the economic motivations for crime. For example, to the extent that treatment reduces drug and alcohol use, it may also reduce crimes committed to obtain money to buy drugs or alcohol.
Source: NORC CALDATA
Figures 4.8 and 4.9 report on two major measures of criminal activity: the percentage who sold or helped to sell drugs (the most prevalent form of illegal activity, outside of drug possession per se, in this population) and the percentage who were arrested, booked, or taken into custody. In each analysis, the "before" measurement pertains to the 12 months preceding the sample treatment episode, and the "after" measurement to the interval between discharge from the sample episode and the CALDATA interview. To enhance comparability, after-treatment measurements were adjusted to a per-annum basis. As with previous domains, we analyzed numerous other measures of criminal activity, such as crimes against property and persons, and other criminal justice sanctions, such as conviction, parole, and incarceration. The two items displayed here are among the most conservative indicators of change.
Source: NORC CALDATA
Each measure shows substantial before-after declines overall and among every subgroup. The overall decline was a 68 percent reduction in drug selling and a 60 percent reduction in arrests. Men as a whole and in each corresponding subgroup generated more criminal activity and sanctions than women, both before and after treatment. For neither measure does the change differ significantly among subgroups of women or men. The fact that potential reductions in criminal activity are lower for female clients than for male clients has an important bearing on Chapter 5's finding that the overall benefit-cost ratio of substance abuse treatment was lower for female clients with children who received welfare income than for the total treatment population.
Figure 4.10 presents before-after comparisons for the most costly single indicator of clients' health and health care utilization, the percentage hospitalized for any reason, including physical conditions, mental conditions, or drug-related conditions (including overdose). The percentage hospitalized changed significantly from before-treatment to after-treatment, declining by about 58 percent, from 31 percent to 13 percent. These declines were evenly balanced across physical, mental, and drug-related causes of hospital stays. All of the subgroup changes were significant but none were significantly different in size from each other.
Source: NORC CALDATA
Finally, Figure 4.11 presents before-after comparisons of the percentage who were home-less for two or more days (per-annum basis). On the whole, men were more likely to have been homeless than women in each period. Every subgroup experienced less homelessness after treatment, with one exception: women with children who did not receive welfare income. This group had the lowest level of homelessness before treatment and the least proportionate reduction across the CALDATA period. In the subgroup of women with children who received welfare income, the percentage who were homeless for two or more days declined from 22 percent before treatment to 9 percent after treatment.
Source: NORC CALDATA
On the dimensions of drug use, crime, health care, and homelessness, women in CALDATA as a group were appreciably better off after treatment than before, improving to the same extent as men; and this result held for women with children and women receiving welfare as well as for other treatment subgroups. On the economic dimensions of employment and separation from welfare dependence, the CALDATA groups did not improve, but stayed even in terms of welfare and lost ground on employment—California's severe economic downturn during the study period appears to have swamped whatever positive economic effect the period of treatment might have generated. From the standpoint of public policy, we believe these results encourage the support of treatment options for welfare clients.