Chapter 2
Discharge Data Description and Data Overview for All Types of Service: 2004

Data Definitions
Comparison of Discharge and Admission Data
State Data
Data Overview
Client Characteristics Associated with Completion of Treatment or Transfer to Further Treatment

Chapter 2 describes the discharge data and provides an overview of the 1,046,522 linked admission/discharge records for clients discharged in 2004 in 28 reporting States [Table 2.2a]. Discharge data linked to admissions data represent treatment episodes that enable analyses of treatment completion and length of stay (LOS) in treatment.

Data Definitions

Reason for Discharge

The different reasons for discharge are listed below. Because both treatment completion and transfer to further treatment represent positive conclusions to a treatment episode, the rates for treatment completion and transfer are combined in some of the analyses.

Length of Stay (LOS) in Treatment

The length of stay (LOS) in days was calculated for each record by subtracting the date of admission from the earlier of either the date of last contact (if available) or the date of discharge; stays of under one day were rounded to one day. Two measures of LOS are included: the median—the number of days at which half of all clients had been discharged—and the average (or mean). The average LOS was calculated by summing the number of days in treatment for all Year 2004 discharges and dividing the total by the number of discharges in 2004.

For each type of service, the average LOS was longer, often much longer, than the median LOS, indicating that the distribution of LOS was skewed. In other words, while half of all clients had been discharged by the median LOS, some of the remainder had exceptionally long stays, ranging up to the maximum possible in this analysis, 4 years (about 1,400 days) [Appendix C]. Many of these exceptionally long periods may be the result of administrative procedures in which records for clients who had dropped out of the system were periodically “purged.” In this case, if the date of last contact with these clients was unknown, the records were assigned a discharge date that represented the date of the administrative procedure.

Type of Service

Data are presented for each of eight discrete types of service. Treatment completion rates, median LOS, and average LOS were not aggregated across service types. There was significant State-to-State variation in types of services available [Tables 2.3a and 2.3b]. Treatment completion rates, median LOS in treatment, and average LOS in treatment varied considerably among the service types [Tables 2.5 and 2.6].

Six service types exclude records where use of methadone was planned as part of treatment:

a) Free-standing residential—24-hour per day services in non-hospital setting providing for safe withdrawal and transition to ongoing treatment (89 percent of discharges from
detoxification);
b) Ambulatory—Outpatient treatment services providing for safe withdrawal in an ambulatory setting (7 percent of discharges from detoxification); and
c) Hospital—24-hour per day medical acute care services in hospital setting for detoxification for persons with severe medical complications associated with withdrawal (4 percent of discharges from detoxification).

Methadone treatment can occur in any service type [Table 2.6]. In 2004, 62 percent of discharges from methadone treatment were from an outpatient setting and 35 percent of discharges from methadone treatment were from detoxification. Data are presented in separate chapters for:

Comparison of Discharge and Admission Data

Tables 2.1a and 2.1b present the distribution of key variables for all Year 2004 admissions, for Year 2004 admissions in the 28 States submitting usable discharge data, and for linked and non-linked Year 2004 discharge records in the 28 States reporting discharge data. These tables indicate that admission characteristics for the discharges submitted by the 28 States were similar to the admission characteristics in all States.

State Data

Data Linkage

Tables 2.2a and 2.2b. A total of 1,105,644 records for clients discharged in Year 2004 were submitted by 28 States. Ninety-five percent of these discharge records (n = 1,046,522) could be linked to a TEDS admission record from 2001, 2002, 2003, or 2004. No matching admission record was found in those years for 59,122 of the Year 2004 discharges.

Type of Service

Tables 2.3a and 2.3b present type of service at discharge by State for the 28 States that submitted discharge data in 2004. There was considerable State-to-State variability in the combination of service types available and in the proportions of clients in each of these service types.

Figure 2.1. Overall, 44 percent of discharges were from outpatient treatment, 21 percent were from detoxification, 11 percent were from intensive outpatient treatment, 9 percent were from long-term residential treatment, 8 percent were from short-term residential treatment, 6 percent were from methadone treatment, and 1 percent were from hospital residential treatment.

Figure 2.1
Type of service at discharge: TEDS 2004

Pie chart comparing Type of service in TEDS Year 2004 Discharges

* Methadone discharges may be any service type.
SOURCE: Office of Applied Studies, Substance Abuse and Mental Health Services Administration, Treatment Episode Data Set (TEDS). Data received through 2.1.2006.

Reason for Discharge

Table 2.4 presents reason for discharge by State. Overall, 40 percent of the Year 2004 discharges completed treatment, 12 percent were transferred to further treatment, and 22 percent dropped out of treatment. Eight percent had treatment terminated by the facility, 1 percent had treatment terminated because of incarceration, less than 1 percent died, and 8 percent failed to complete treatment for other reasons.* The reason for discharge was unknown for 9 percent of discharges.


* Percentages do not sum to 100 percent because of rounding.

Data Overview

Primary Substance

Tables 2.1a and 2.1b. More than a third (36 percent) of all Year 2004 linked discharges reported primary alcohol abuse at admission. Nineteen percent reported opiate abuse, 16 percent reported marijuana abuse, 14 percent reported cocaine abuse, 10 percent reported stimulant abuse, and 4 percent reported abuse of other substances at admission.*


* Percentages do not sum to 100 percent because of rounding.

Reason for Discharge by Type of Service

Table 2.5 and Figure 2.2 present reason for discharge by type of service. Of the 957,491 Year 2004 discharges who reported a reason for discharge, 44 percent completed treatment and 13 percent transferred to another treatment program. About one quarter (24 percent) dropped out of treatment, while the treatment of an additional 8 percent was terminated by the facility. Ten percent left treatment for other reasons.*

The treatment completion rate was greater than 60 percent among discharges from hospital residential treatment (69 percent), detoxification (62 percent), and short-term residential treatment (61 percent). Completion rates were lower in longer-term and less structured settings: 43 percent for long-term residential treatment, 38 percent for intensive outpatient treatment, and 36 percent for outpatient treatment.

The completion rate was lowest among discharges from methadone treatment (17 percent). Of those who were discharged from methadone treatment, 47 percent dropped out of treatment. Methadone treatment could be in any type of service; for more detail on methadone discharges, see Table 2.6 and Chapters 9 and 10.

Discharges from intensive outpatient treatment were the most likely to have been transferred to further substance abuse treatment (18 percent). In each of the other service types, 12 percent to 16 percent were transferred to further treatment.


* Percentages do not sum to 100 percent because of rounding.
Figure 2.2
Reason for discharge, by type of service: TEDS 2004

Stacked bar chart comparing Reason for discharge, by type of service in TEDS 2004

* Methadone discharges may be any service type.
SOURCE: Office of Applied Studies, Substance Abuse and Mental Health Services Administration, Treatment Episode Data Set (TEDS). Data received through 2.1.2006.

Median LOS by Reason for Discharge and Type of Service

Table 2.5 and Figure 2.3. The median LOS among all non-methadone discharges was longest for discharges from outpatient treatment (69 days), followed by long-term residential treatment (46 days) and intensive outpatient treatment (42 days). The median LOS for all discharges from short-term residential treatment was 20 days; from hospital residential treatment, 11 days; and from detoxification, 3 days.

The median LOS among non-methadone treatment completers was longest for outpatient treatment (104 days), followed by long-term residential treatment (79 days) and intensive outpatient treatment (52 days). The median LOS for treatment completers in short-term residential treatment was 25 days; for hospital residential treatment, 14 days; and for detoxification, 3 days.

In most service types, the median LOS was longer for discharges who completed treatment than for those who transferred to further treatment. Detoxification and methadone discharges were exceptions. In turn, the median LOS among those who did not complete treatment was generally shorter than among either treatment completers or transfers.

Figure 2.3
Median length of stay, by reason for discharge and type of service: TEDS 2004

Bar chart comparing Median length of stay, by reason for discharge and type of service in TEDS 2004

* Methadone discharges may be any service type.
SOURCE: Office of Applied Studies, Substance Abuse and Mental Health Services Administration, Treatment Episode Data Set (TEDS). Data received through 2.1.2006.

Average LOS by Reason for Discharge and Type of Service

Table 2.5 and Figure 2.4. The average LOS among all non-methadone discharges was longest for discharges from outpatient treatment (107 days), followed by long-term residential treatment (74 days) and intensive outpatient treatment (72 days). The average LOS for all discharges from hospital residential treatment was 29 days; from short-term residential treatment, 26 days; and from detoxification, 11 days.

The average LOS among non-methadone treatment completers was longest for outpatient treatment (138 days), followed by long-term residential treatment (97 days), and intensive outpatient treatment (82 days). The average LOS for treatment completers in hospital residential treatment was 30 days; for short-term residential treatment, 27 days; and for detoxification, 8 days.

The average LOS for treatment completers was generally longer than for clients who transferred to further treatment, similar to the pattern seen for median LOS.

Figure 2.4
Average length of stay, by reason for discharge and type of service: TEDS 2004

Bar chart comparing Average length of stay, by reason for discharge
and type of service in TEDS 2004

* Methadone discharges may be any service type.
SOURCE: Office of Applied Studies, Substance Abuse and Mental Health Services Administration, Treatment Episode Data Set (TEDS). Data received through 2.1.2006.

Discharges from Methadone Treatment

Table 2.6. There were 51,263 discharges where records showed that methadone use was planned at the time of admission and where data could be linked to admission records from 2001 to 2004, and where data indicated reason for discharge. These records were limited to methadone treatment that began and ended within a maximum period of 4 years and, because methadone treatment can extend for a much longer period, may not be representative of all methadone discharges. Sixty-two percent of methadone discharges were from outpatient treatment (including both outpatient and intensive outpatient treatment), 35 percent from detoxification, 2 percent from long-term residential treatment, 1 percent from short-term residential treatment, and less than 1 percent from hospital residential treatment.*

The treatment completion rate for discharges from methadone treatment was 17 percent overall, but varied by type of service. It was 12 percent among discharges from methadone outpatient treatment and 22 percent among discharges from methadone detoxification. The completion rate was higher among the few discharges from more structured settings, at 64 percent for hospital residential treatment, 44 percent for short-term residential treatment, and 40 percent for long-term residential treatment.

The median LOS for all clients discharged from methadone treatment was 42 days. It was longest (115 days) among methadone outpatient discharges. The median LOS was 16 days among methadone detoxification discharges, and it was between 7 days and 46 days for methadone discharges from the other service types. Overall, the median LOS for methadone discharges completing treatment was 28 days. It was longest (159 days) among methadone outpatient completers. The median LOS was 20 days among methadone detoxification completers, and it was between 8 days and 63 days for methadone completers from the other service types.

The average LOS for all clients discharged from methadone treatment was 154 days. It was longest (224 days) among methadone outpatient discharges. The average LOS was 37 days among methadone detoxification discharges, and it was between 29 days and 98 days for methadone discharges from the other service types. Overall, the average LOS for methadone discharges completing treatment was 144 days. It was longest (270 days) among methadone outpatient completers. The average LOS was 35 days among methadone detoxification completers, and it was between 24 days and 97 days for methadone completers from the other service types.


* Percentages do not sum to 100 percent because of rounding.

Client Characteristics Associated with Treatment Completion or Transfer to Further Treatment

Two of the Reason for discharge categories, treatment completion and transfer to further substance abuse treatment, represent positive conclusions to a treatment episode. A statistical analysis of the client characteristics associated with treatment completion or transfer to further treatment was conducted and is presented below.

To examine the client characteristics associated with treatment completion or transfer to further treatment, univariate logistic regression was conducted for all service types combined and for each service type separately. Then, a multivariate conditional logistic regression model was used to identify, for all service types combined and for each service type separately, client characteristics that remained associated with treatment completion or transfer to further treatment when controlled for the other client characteristics.1

Univariate Analyses of Discharges from All Service Types Combined

Table 2.7a presents the odds ratio, 95 percent confidence interval, and p value from the univariate analyses of each client characteristic for all services types combined and for each service type separately.

In univariate analyses of discharges from all service types combined, all client characteristics were associated with treatment completion or transfer to further treatment:

Univariate Analyses of Discharges from Individual Service Types

Among the individual service types, univariate analyses demonstrated that, for discharges from outpatient treatment, intensive outpatient treatment, short-term residential treatment, long-term residential treatment, and detoxification, each client characteristic was significantly associated with treatment completion or transfer to further treatment.

For discharges from hospital residential treatment, methadone outpatient treatment, and methadone detoxification, the majority of the client characteristics were significantly associated with treatment completion or transfer to further treatment.

Multivariate Analysis of Discharges from All Service Types Combined

Table 2.7b presents the odds ratio, 95 percent confidence interval, and p value from the multivariate analyses of each client characteristic for all services types combined and for each service type separately.

Among all discharges combined, all of the client characteristics remained significantly associated with treatment completion or transfer to further treatment. However, the association was weakened for most client characteristics when adjusted for the presence of the other client characteristics in the model:

The association of two client characteristics, prior treatment and referral source, were strengthened slightly after adjustment for the other client characteristics in the model.

Multivariate Analyses of Discharges from Individual Service Types

Table 2.7b and Figure 2.5. For the multiple conditional logistic regression models, a different set of predictors of treatment completion or transfer to further treatment was evident for each of the different service types. A number of client characteristics that displayed significant univariate associations with treatment completion or transfer to further treatment were no longer significant when the other client characteristics in the model were taken into account. For example, race/ethnicity was significantly associated with treatment completion or transfer to further treatment in all of the eight service types in univariate analysis. When other client characteristics were included in the model, it was significant in only six of the eight service types. All client characteristics lost significance as predictors of treatment completion or transfer to further treatment in some service types.

Figure 2.5
Odds ratios for client characteristics associated with treatment completion or transfer to further treatment, by type of service: TEDS 2004
Multiple conditional logistic regression1

Bar and scatter chart comparing Average length of stay, by reason for discharge
and type of service in TEDS 2004

1 Bars show client characteristics with significant effects at the p < 0.05 level or less. See Table 2.7b.

Figure 2.5 (cont.)
Odds ratios for client characteristics associated with treatment completion or transfer to further treatment, by type of service: TEDS 2004
Multiple conditional logistic regression1

Bar and scatter chart comparing Average length of stay, by reason for discharge
and type of service in TEDS 2004

1 Bars show client characteristics with significant effects at the p < 0.05 level or less. See Table 2.7b.
SOURCE: Office of Applied Studies, Substance Abuse and Mental Health Services Administration, Treatment Episode Data Set (TEDS). Data received through 2.1.2006.

Overall model fit was good for most of the multivariate analyses (p < 0.0001). It was weaker for hospital residential treatment (p = 0.0006) and methadone detoxification (p = 0.0199), where the numbers of discharges were smallest (n = 8,080 and n = 17,976, respectively).

Primary substance. The use of alcohol rather than other drugs as a primary substance was a significant predictor in the multivariate model for each service type.2 For all service types, discharges reporting alcohol as their primary substance were more likely to complete treatment or transfer to further treatment by factors ranging from 27 percent (for long-term residential discharges) to 99 percent (for detoxification discharges).

Education. A higher level of education was also a consistent predictor of treatment completion or transfer to further treatment for six of the eight service types. (Hospital residential treatment and methadone outpatient treatment were the exceptions.) Discharges who had 12 years or more of education were more likely to complete treatment or transfer to further treatment by factors ranging from 7 percent for outpatient treatment to 29 percent for short-term residential treatment.

Age at admission. Age at admission was a significant predictor in the multivariate model for six of the eight service types. (Hospital residential treatment and methadone detoxification were the exceptions.) Discharges who were 30 years of age and older were more likely than those who were under 30 years of age to complete treatment or transfer to further treatment by factors ranging from 3 percent for outpatient treatment to 40 percent for short-term residential treatment.

Race/ethnicity. Race/ethnicity was significantly associated in multivariate analysis with treatment completion or transfer to further treatment in six of the eight service types. (Short-term residential treatment and long-term residential treatment were the exceptions.) In most service types, being non-Hispanic White (vs. other racial/ethnic groups) was a predictor of treatment completion or transfer by factors ranging from 8 percent to 49 percent. In detoxification and methadone outpatient treatment, however, discharges who were non-Hispanic White were less likely to complete treatment or transfer to further treatment than discharges belonging to other racial/ethnic groups by factors of 22 percent and 7 percent, respectively.

Employment status. Employment status was a consistent predictor of treatment completion or transfer to further treatment for seven of the eight service types. (Methadone detoxification was the exception.) For all service types except methadone outpatient treatment, discharges who were employed full or part time were more likely to complete treatment or transfer to further treatment by factors ranging from 19 percent for intensive outpatient treatment to 72 percent for long-term residential treatment. For methadone outpatient treatment, discharges who were employed were 12 percent less likely to complete treatment or transfer to further treatment than discharges who were unemployed or not in the labor force.

Gender. Gender was significantly associated in multivariate analysis with treatment completion or transfer to further treatment in six of the eight service types. (Short-term residential treatment and hospital residential treatment were the exceptions.) Being female was associated with treatment completion or transfer to further treatment in five of the six service types by factors of 9 percent to 42 percent. Intensive outpatient treatment was the only service type where being male was associated with treatment completion or transfer to further treatment, and this was by a factor of 19 percent.

Frequency of use. No use of the primary substance (vs. any use) in the month before treatment entry was associated with treatment completion or transfer to further treatment for six of the eight service types by factors ranging from 16 percent to 137 percent. (Hospital residential treatment and detoxification were the exceptions.)

Prior treatment. Having had no prior treatment (vs. some prior treatment) was significantly associated with treatment completion or transfer to further treatment for seven of the eight service types by factors ranging from 16 percent to 49 percent. (Short-term residential treatment was the exception.)

Referral source. Referral to treatment through the criminal justice system was a significant predictor in the multivariate model for seven of the eight service types. (Hospital residential treatment was the exception.) For all service types except methadone outpatient treatment, discharges referred to treatment through the criminal justice system were more likely to complete treatment or transfer to further treatment by factors ranging from 20 percent (for intensive outpatient residential discharges) to 60 percent (for discharges from methadone detoxification) than discharges referred to treatment by other sources. For methadone outpatient treatment, discharges referred to treatment through the criminal justice system were 17 percent less likely to complete treatment or transfer to further treatment than discharges referred to treatment by other sources.


1 To examine the client characteristics associated with substance abuse treatment completion or transfer to further treatment, the variables representing these characteristics were dichotomized. (See Tables 2.7a and 2.7b.) Univariate logistic regression was conducted for all service types combined and for each service type separately to test whether the client characteristic was related to completion of treatment or transfer to further treatment for that service type. Logistic regression yields an odds ratio, that is, the odds of one group completing treatment or transferring to further treatment over the odds of the other group completing treatment or transferring. For example, the odds ratio among all discharges combined for males completing treatment or transferring to further treatment versus females completing treatment or transferring is 1.150 (Table 2.7a). This can be expressed as a percentage; that is, male discharges were 15 percent more likely than female discharges to complete treatment or transfer to further treatment.

Many of the client characteristics are related to each other, and the univariate odds ratio can reflect that interrelatedness. For example, if both age and gender are related to treatment completion or transfer, and age is related to gender, then calculation of the univariate odds ratios for age and gender will effectively count some treatment completers twice (i.e., once in the age calculation and once in the gender calculation). To produce an odds ratio for age that is independent of (or adjusted for) the odds ratio for gender, and vice versa, a multivariate conditional logistic regression model is used. In this analysis, for each service type and for all service types combined, client characteristics that were associated with treatment completion or transfer to further treatment at the significance level of p < 0.05 were used in SAS software stepwise regression procedures (alpha = 0.001 to enter or remove).

Note: Age at first use was reported for 76 percent of all discharges, and for less than 85 percent of discharges from outpatient treatment, long-term residential treatment, detoxification, methadone outpatient treatment, and methadone detoxification. It was excluded from the logistic regression analyses.

2Alcohol vs. other substances was not included in the logistic regression analysis of methadone outpatient or methadone detoxification services, as 95 percent and 97 percent, respectively, of discharges from these service types reported their primary substance as opiates.

To Tables

To table of contents
To Table of Contents

This page was last updated on  .