| ![]() | |||||||||||
![]() |
| |||||||||||
Services Research Outcomes Study (SROS) |
HISTORY OF THE SROS SAMPLE
Two data collections frame the Services Research Outcomes Study (SROS): the Uniform Facility Data Set/National Drug and Alcoholism Treatment Unit Survey (UFDS/NDATUS) represents the first data collection effort, and the Drug Services Research Survey (DSRS) the second.
A detailed description of the DSRS, subsequent SROS facility universe, the facility sample, and client sample is presented in Tables 2-1 and 2-2. The April 1990, 10,649 facilities constituted the known substance abuse treatment facilities in the continental United States that served as the sampling universe. DSRS first sampled 1,803 facilities for its Phase I survey of facilities; 1,442 proved eligible (as active treatment facilities in the continental United States), and 1,183 of the eligible facilities responded, comprising 138 hospital inpatient facilities, 185 residential, 80 outpatient methadone detoxification/maintenance, 372 outpatient drug-free, 91 alcohol-only, and 317 whose facility type was unknown.
Since DSRS focus was on clients discharged from drug treatment facilities, the alcohol-only and treatment-type unknown facilities were excluded from the Phase II facility sample frame.
DSRS Phase II randomly selected 146 facilities from the DSRS Phase I facility respondents and abstracted 2,222 client records from 120 facilities in the four strata of: hospital inpatient, residential, outpatient detoxification/maintenance, and outpatient drug-free treatment.
The DSRS sample of 120 facilities was the base from which the SROS sample was contacted. When SROS staff approached the 120 facilities to seek enrollment in the SROS protocol of record reabstraction, sample supplementation, and client interviews, a dozen facilities were no longer operational and a larger number had changed ownership or leadership in the interim. SROS gained the cooperation of 99 of the 120 DSRS facilities. SROS supplemented the DSRS client sample target from these 99, increasing the sample from 2,222 to 3,047 clients who had been discharged in the 12 months ending August 31, 1990.
Of the 3,047 clients whose records were abstracted during 1994, 2,489 clients (82 percent) in the sample were located during the nine month 19951996 field period. A detailed description of field data collection is presented in the Appendix A. Nine percent (277 clients) of the sample of 3,047 had died between discharge from drug treatment and the SROS field period. Twelve percent (351 clients) refused or were unavailable for interview before the end of the interview period, less than one percent (14 clients) proved ineligible for the study, and 1,799 were successfully interviewed, comprising 59 percent of the total sample, 65 percent of those alive and eligible for the study, and 82 percent of those alive and located during the field period. The remaining 558 clients (18 percent) were not located before time and resources for fieldwork expired.
There is a difference between simple response rates and cumulative response rates. The overall completion rate was 65 percent when those who died before the field period are excluded. The completion rate for subgroups of the sample are: 63 percent for males, 70 percent for females, 65 percent for white non-Hispanics, 66 percent for black non-Hispanics, and 54 percent for Hispanics.
When those who died before the field period are counted as completions, the overall completion rate was 68 percent. The completion rate for subgroups of the sample are: 67 percent for males, 72 percent for females, 68 percent for white non-Hispanics, 69 percent for black non-Hispanics, and 59 percent for Hispanics.
When SROS is viewed as part of a longitudinal study, following DSRS Phase I, DSRS Phase II, the recapture of DSRS facilities for SROS, and the completion of cases for SROS, the cumulative response rate would be the product of each of the four individual response rates. [NOTE: The computation is (.82)(.82)(.83)(.68) = 38 percent as shown on Table 2-1.]
Whenever respondents to a study are not 100 percent of those eligible for inclusion, the respondents representativeness [The question centers on whether the respondents who were not included were lost to followup for essentially random and unconnected reasons. If so, the loss of those data would not introduce any distortions or biases into the data, but would only result in some loss in precision due to a smaller sample size than might otherwise have been achieved .] is always an important issue. The following section analyzes the extent of bias introduced by nonresponse, comparing the clients interviewed with those not interviewed.
![]() 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 |