 |
 |
 |  |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Services Research Outcomes Study (SROS)
| | |||||||||||
| |
| |||||||||||
|
Services Research Outcomes Study (SROS) | ||||||||||||
CHANGES IN CLIENT BEHAVIOR AND CIRCUMSTANCES IN THE FIVE YEARS BEFORE AND AFTER TREATMENT
The SROS interview led respondents through a series of explicit comparisons between the five-year period preceding the index treatment episode and the five years after the treatment episode, using identically worded items to compare the clients’ drug and alcohol use, criminal behavior, housing characteristics, physical and mental health, and employment and other sources of income across these two periods. This section presents an analysis, using methods described in Chapter II, of changes on these measures, weighted to reflect the 967,000 clients represented by SROS as a whole as well as subgroups divided on a variety of dimensions. The findings generally reflect appreciable changes across these periods, particularly in the drug and alcohol use and criminal activity measures, and these changes correlate in important ways with factors such as the length of stay in treatment. The observation that statistically significant changes occurred from the period before treatment to the period after treatment does not necessarily mean that treatment was a sole cause (nor, strictly speaking, does it prove that treatment caused any) of these changes, but it does mean that real changes occurred, that something caused these changes, that treatment is associated with these changes, and that further analysis is warranted to isolate and assess the contributions of the potential causes.
Changes in Drug and Alcohol Use Among Clients as a Whole
There were substantial decreases in the percentage of respondents who reported using any illicit drugs after treatment, compared with the period before treatment (see Table 3-9). Those using any illicit drug decreased from 75 percent before treatment to 59 percent after treatment, which is a statistically significant difference of minus 16 points, and a percentage difference (that is, 100*[{59.3/75.4}-1.0]) of 21 percent. This decrease was also found for almost every individual drug, from the most prevalent (marijuana and cocaine, down 28 and 45 percent) to those rarely used, such as PCP. The only listed drug for which the decrease in use was not statistically significant was illegal methadone.
Table 3-10 presents a similar analysis of the change in drug use among the client population, but measures change in a different way—namely, the average number of days that each listed drug was used before and after treatment among those who had used the drug before treatment. Within this more restricted sample, there were still substantial decreases in the frequency of use for every listed drug. For example, clients who used marijuana before treatment used it an average (mean) of 15 days per month; after treatment, the same group used marijuana half as often, or fewer than seven days per month. The use of cocaine by pretreatment cocaine users fell by 63 percent, from 12 days to five days per month, and the use of crack by pretreatment crack users fell by 49 percent, from 16 days to eight days per month.
Finally, Table 3-11 presents the results of an additionally restricted analysis, examining the mean number of days that the listed drugs were used in the 5 years before and 5 years after treatment among clients who had used each drug both before treatment and after treatment. In essence, this table shows the change in use for those who did not stop using specific drugs. Even in this group, statistically significant reductions were found for the four drugs with the largest number of cases:
Alcohol, marijuana, cocaine, and crack (the number of days used per month decreased by 15, 28, 21, and 11 percent, respectively). [ In contrast, clients who used illegal methadone both before and after treatment increased their rate of use from around six days to eight days per month on average. Although this finding was statistically significant, the number of cases in the sample was small, and additional analyses in later research will be needed to interpret this finding correctly.]
To summarize, among the client population as a whole, the percentage of clients using all drugs and any specific drug was lower after treatment than before; users of specific drugs before treatment used them less after treatment; and even clients who continued to use the most prevalent drugs after treatment used them less frequently than before. In the following discussion of changes in drug use by sociodemographic and treatment subgroups, attention is focused on only the first of these measures (i.e., the percentage of clients using drugs before versus after treatment).
Changes in Drug Use by Sociodemographic and Treatment Subgroups
There were distinct differences in the level of change in drug use depending on sex, age, and racial/ethnic group. Both male and female clients were less likely to use drugs after treatment than before (see Figure 3-4 and Appendix Tables B-16 to B-21). However, for any illicit drug overall and for each of the most frequently used illicit drugs (marijuana, cocaine, crack, and heroin), the decrease was greater among female clients than male. Adults and adolescents differed as well (see Figure 3-5): No statistically significant decreases were detected for adolescents in overall use of any illicit drug during the five years after treatment or in the specific use of marijuana, cocaine, and heroin. However, adolescents significantly increased their use of crack after treatment, albeit from a low pretreatment base of five percent (see Appendix Table B-20), because the age of first use of crack is typically more than 20 years. Among adults, there were gradients in the extent of change by age, with older age groups generally reducing their drug use to a greater degree (percentage change) after treatment than younger age groups did. Finally, among black, white, and Hispanic clients, there were similar decreases in the overall measure of any illicit drug use (see Appendix Tables B-16 to B-21). However, only the black clients reduced their crack and heroin use to a statistically significant extent (23- and 18-percent declines); in both instances, black clients had been more likely to use these drugs before treatment than the other two racial/ethnic groups.
Figure 3-6 (see also Appendix Tables B-16 to B-21) shows changes in drug use rates by the type of treatment that the client received. Clients in methadone treatment facilities composed the only group showing a significant decrease in heroin use (27-percent decline); methadone treatment had lesser effects on cocaine and crack use, but this finding corresponds well to the predominance of heroin use in the methadone treatment group’s pretreatment profile. By far the largest decrease in crack use was among residential clients (32-percent decline). Cocaine use also decreased most among residential clients (55-percent decline), although the contrast with other treatment types was less pronounced in this case.
Figure 3-4. Percentage change* in drug use by sex
Figure 3-5. Percentage change* in drug use by age
*The percentage change is the difference between (a) the percentage using five or more times across the five years after treatment and (b) the percentage using five or more times across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. "NS" means that the difference was not significant. Data are from Appendix Tables B-16 to B-21.
**"Any illicit drug" includes marijuana, cocaine, crack, heroin, inhalants, PCP, other hallucinogens, illegal methadone, narcotics, methamphetamines, downers, and other illicit drugs.
Figure 3-6. Percentage change* in drug use by type of treatment
*The percentage change is the difference between (a) the percentage using five or more times across the five years after treatment and (b) the percentage using five or more times across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. "NS" means that the difference was not significant. Data are from Appendix Tables B-1 to B-6.
**"Any illicit drug" includes marijuana, cocaine, crack, heroin, inhalants, PCP, other hallucinogens, illegal methadone, narcotics, methamphetamines, downers, and other illicit drugs.
Length of Stay and Change in Drug Use after Treatment:
In addition to the type of treatment, the length of stay in treatment (Figure 3-7) was consistently associated with the extent of change in drug use. Although the pattern does not appear strictly linear in Figure 3-7, this is a consequence of grouping clients in types of treatment with shorter (generally the 24-hour facilities) and longer (outpatient) planned terms. By and large, the "less than one week" and "less than six month" groups are more heavily weighted with clients with suboptimal lengths of stay. The multivariate regression results reported below make it more clear that, after controlling for type of treatment (and therefore planned length of stay), length of stay is associated with the extent of decrease in drug use. Clients who completed their treatment plan were more likely to reduce their pretreatment drug use than noncompleters for every principal drug except crack.
Figure 3-7. Percentage change* in drug use by length of stay
*The
Percentage change is the difference between (a) the percentage using five or more times across the five years after treatment and (b) the percentage using five or more times across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. "NS" means that the difference was not significant. Data are from Appendix Tables B-16 to B-21.
**"Any illicit drug" includes marijuana, cocaine, crack, heroin, inhalants, PCP, other hallucinogens, illegal methadone, narcotics, methamphetamines, downers, and other illicit drugs.
Changes in Criminal Behavior
Just as clients reported overall reductions in drug use in the five years after treatment, they reported overall reductions in criminal activities. Drawn together in Table 3-12 and Figure 3-8 are percentage changes in a series of individual types of criminal activities as well as some key summary measures of contact with the criminal justice system. Virtually all of the measures of individual types of income-producing crimes (such as breaking and entering, larceny, prostitution, and drug sales), as well as more violent and disorderly offenses (such as driving under the influence and weapons use), declined by roughly one-third (the range shown in the figure is 23 to 38 percent; a few of the rarer offenses in the client population, such as auto theft, rape, and murder, are outside this range). However, self-reported arrests declined by a smaller proportion (17 percent) than crimes, incarceration actually increased by 17 percent, and violations of probation or parole conditions rose by 26 percent. These findings seem paradoxical. It is possible that the same circumstances that led clients to seek treatment also placed them under more stringent supervision in the form of probation, parole, and incarceration, and this increase in supervision reduced the commission of primary offenses. Another possibility is that treatment had the effect of making clients both less prone to committing offenses and more visible or easily collared when they did. The evidence in SROS cannot further discriminate among these or other possible explanations.
Figure 3-8. Percentage change* in criminal behavior
*The percentage change is the difference between (a) the percentage reporting criminal activity across the five years after treatment and (b) the percentage reporting criminal activity across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. Data are from Table 3-12.
Changes in Criminal Behavior by Sociodemographic and Treatment Subgroups
Appendix Tables B-31 to B-37 analyze the changes of criminal behavior within subgroups defined by selected demographic and treatment characteristics; some of these findings are illustrated in Figures 3-9 to 3-13. Percentage changes in specific criminal activities before and after treatment differed somewhat by sex (see Figure 3-9) and race/ethnicity (see Appendix Tables B-31 to B-37) but with no unifying pattern of consistent differences. In contrast, changes by age group showed a strong positive association between age and the extent of change (see Figure 3-10): Older age groups were much more inclined to reduce criminal activities after treatment than younger ones. This is particularly dramatic with respect to driving under the influence or while intoxicated, selling drugs, and committing acquisitive crimes, some of which increased in the five years after treatment among the youngest age group. Increase in criminal activity from the early teenage years into young adulthood is an extremely common observation in cohort studies (Blumstein et al., 1986), and in order to isolate more precisely the intervention effects, investigative efforts must adopt the assumption of a steeply rising growth curve (Johnson et al., 1996, 1997). Although the consistent finding of greater reductions in post-treatment criminal activity with increasing age is striking, this relationship continued to hold only for drug selling in the multivariate analyses discussed in the next section.
Changes in Criminal Behavior by Length of Stay:
There were no consistent patterns of before/after change in criminal activity across type of treatment (see Figure 3-11 and Appendix Tables B-31 to B-37), but lengths of stay greater than six months (see Figure 3-12) and completion of the treatment plan (see Figure 3-13) were clearly associated with greater reductions in crime than were shorter lengths of stay and noncompletion of the treatment plan. The decreases in crime were about one-half larger for completers and clients who had at least a six-month stay; these results were consistent with the findings for drug use. In addition, for most of the crimes listed, the decline in criminal activity after treatment for those discharged from their first treatment experience was larger, and in some cases twice as large, as the decline for those who had one or more treatment experiences. For example, rates of theft/larceny and breaking and entering declined by 52 and 62 percent, respectively, among clients receiving their first lifetime treatment, but declined by only about 30 percent for those receiving a second or greater treatment.
A parallel finding was that post-treatment reduction in crime for those who took no further drug treatment after 1989–1990 was two to four times as large as the corresponding reductions among those who did reenter treatment after 1989–1990 (see Appendix Tables B-31 to B-37). For example, those who took no additional treatment decreased drug selling by 43 percent; for clients who later took additional treatment, the corresponding decline was 21 percent. Although about 50 percent of both groups reported driving under the influence before treatment, there was a 40-percent drop after treatment for those receiving no further treatment, and a much smaller drop of 10 percent for those receiving additional treatment.
Figure 3-9. Percentage change* in criminal behavior by sex
Figure 3-10. Percentage change* in criminal behavior by age
*The percentage change is the difference between (a) the percentage reporting criminal activity across the five years after treatment and (b) the percentage reporting criminal activity across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. "NS" means that the difference was not significant. Data are from Appendix Tables B-31 to B-37.
Figure 3-11. Percentage change* in criminal behavior by type of treatment
Figure 3-12. Percentage change* in criminal behavior by length of stay
*The percentage change is the difference between (a) the percentage reporting criminal activity across the five years after treatment and (b) the percentage reporting criminal activity across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. "NS" means that the difference was not significant. Data are from Appendix Tables B-31 to B-37.
Figure 3-13. Percentage change* in criminal behavior by treatment completion
*The percentage change is the difference between (a) the percentage reporting criminal activity across the five years after treatment and (b) the percentage reporting criminal activity across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. Data are from Appendix Tables B-31 to B-37.
These results suggest a dynamic relationship between treatment activity and criminal activity, in which clients who completed their treatment plans and stayed the full course were less likely to resume crime and also less likely to reenter treatment at a later time. In contrast, those who stayed for shorter periods and did not complete their treatment plans were more likely to commit offenses and seek treatment again. Although the analyses necessary to evaluate these characteristics thoroughly are beyond the scope of the present report, the likelihood that this kind of dynamic operates across careers in drugs, crime, and treatment is reinforced by observations such as those presented in Figure 3-14 (see also Appendix Table B-40). Clients with many episodes before 1989–1990 were the most likely to reenter treatment after 1989–1990; clients with the shortest lengths of stay during 1989–1990 were the most likely to enter treatment again later; and clients in outpatient nonmethadone facilities—the type of treatment with the longest length of stay (see Table 3-3)—were the least likely to reenter treatment after 1989–1990.
Changes in Lifestyle Characteristics
Appendix Tables B-38 and B-39 present data on changes in client lifestyle characteristics, such as retaining custody of children, homelessness, physical abuse, mental health characteristics, and employment before and after treatment in 1989–1990. Most of these characteristics changed significantly after treatment; the changes are summarized in Figures 3-15 and 3-16.
Figure 3-14. Percentage receiving additional treatment after the index episode
Note: Data are from Appendix Table B-40.
Figure 3-15. Percentage change* in lifestyle characteristics
*The percentage change is the difference between (a) the percentage reporting characteristic across the five years after treatment and (b) the percentage reporting characteristic across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. "NS" means that the difference was not significant. "Was attacked/beaten" means having been attacked with a weapon or seriously hit or beaten. Data are from Appendix Table B-38.
Figure 3-16. Percentage change* in full-time employment
*The percentage change is the difference between (a) the percentage employed full time across the five years after treatment and (b) the percentage employed full time across the five years before treatment, divided by (b). All percentages shown are significant at the 0.05 level. "NS" means that the difference was not significant. Data are from Appendix Table B-39.
After treatment there was a shift toward retaining and regaining child custody, having more reliable housing, [ Homelessness was fairly common; more than one out of every five clients (22 percent) had been homeless for at least two nights prior to treatment. One-half of these of these people (10 percent of the total population) spent that time on the streets, while the other one-half spent time both on the streets and in shelters. However, after treatment, the percentage who spent at least two nights homeless had decreased by 34 percent. The percentage who spent this time on the streets decreased by 40 percent.] avoiding physical abuse, and not attempting suicide, which dropped from 15 percent to 9 percent—despite a continuation of the earlier prevalence of sadness in the clients.
No change was found in the overall rate of full-time employment: About three-quarters of clients had been employed full time for some period before and after treatment. However, this stability masked significant negative trends: Clients were less likely to be employed in a full-time job after treatment than before if they were Hispanic (12-percent reduction), were black (17-percent), were 30 to 39 or 40 years or older (10- and 24-percent), had 9 or fewer years of education (10-percent), were in methadone facilities (25-percent), or reported heroin as their main drug (25-percent).
National Estimates: Converting Percentage Change to Net Differences
This section develops estimates based on the net differences for the entire discharged client population—that is, changes in the behavior of treatment clients when the differences in the before/after behavior are statistically significant. Estimates of how many individuals changed their behavior after treatment were developed by multiplying the raw differences in before/after behavior by the size of the relevant population. For example, using Table 3-9, the number of individuals who both had used any illicit drug in the five years before the 1989–1990 treatment episode and did not use any illicit drug in the five years after the episode would be computed by multiplying the 967,012 population by the 16.1-percent raw difference (i.e., 967,012 × -0.161 = 155,689). As a result, there were 155,689 fewer individuals (which will be rounded to the nearest 1,000; i.e., 156,000) using any illicit drug in the five years after the SROS episode than in the five years before the treatment episode. Additional computations lead to the following national estimates for drug use and crime:
Drug Use
·156,000 fewer users of any illicit drug in the five years after treatment than before;
·152,000 fewer marijuana users in the five years after treatment than before;
·187,000 fewer cocaine users in the five years after treatment than before;
·46,000 fewer crack users in the five years after treatment than before;
·18,000 fewer heroin users in the five years after treatment than before;
·22,000 fewer inhalant users in the five years after treatment than before;
·26,000 fewer PCP users in the five years after treatment than before;
·71,000 fewer hallucinogen users in the five years after treatment than before;
·26,000 fewer narcotics users in the five years after treatment than before; and
·126,000 fewer alcohol users in the five years after treatment than before.
Crime (using the naive assumption that each individual only commits these crimes once in five years [ In "Does Crime Pay?, " James Q. Wilson and Allan Abhahamse ( Justice Quarterly , 9(3):359 –376, 1992) write that "mid - rate " burglars/thieves commit crimes at the following rates: theft/larceny, 11.1 per year; fraud/forgery, 0.6 per year; drug selling, 98.6 per year; and vehicle theft, 0.9 per year. ] ):
·48,000 fewer vehicle thefts in the five years after treatment than before;
·165,000 fewer driving-while-intoxicated cases in the five years after treatment than before;
·131,000 fewer driving-under-the-influence cases in the five years after treatment than before;
·82,000 fewer cases of public disorder in the five years after treatment than before;
·101,000 fewer drug sales in the five years after treatment than before;
·31,000 fewer cases of prostitution/procurement in the five years after treatment than before;
·34,000 fewer cases of fraud/forgery in the five years after treatment than before;
·96,000 fewer cases of theft/larceny in the five years after treatment than before;
·50,000 fewer cases of breaking and entering in the five years after treatment than before; and
·28,000 fewer cases of threat/attack in the five years after treatment than before.
These estimates represent the pool of behavior change within which, if treatment had any success in meeting its goals, the extent of that success is to be found. As discussed previously, the association in time between an episode of treatment and a variety of changes in behavior is not conclusive, but it encourages a search for greater precision and certainty. The next section of this chapter uses some conventional statistical analytic approaches to test the extent to which elements of treatment might plausibly be viewed as contributing to these estimated outcomes.
This page was last updated on June 03, 2008. |
* Adobe™ PDF and MS Office™ formatted files require software viewer programs to properly read them.
Click here to download these FREE programs now
| Highlights | Topics | Data | Drugs | Pubs | Short Reports | Treatment | Help | OAS |
