General Resources / Legal Resources / Medical Resources / Briefing Papers / State Activity    
Hospital Closures / Preventable Tragedies / Press Room / Search Our Site / Home

The Efficacy of Involuntary Outpatient Treatment in Massachusetts.

Geller, J., Grudzinskas, A.J. Jr., McDermeit, M., Fisher, W.H., and Lawlor, T. (1998).

Administration and Policy in Mental Health 25:3. 271- 285.

COMPARISON OF ADMISSIONS DATA FOR IOT AND CTS GROUPS

The purpose of this Massachusetts study was to learn whether IOT was effective in reducing the number of psychiatric admissions and inpatient days for patients who in the past had an excessive use of inpatient treatment. Nineteen patients with court orders for IOT were matched with CTS data from the Massachusetts Department of Mental Health that provided information on the demographic, clinical, and utilization of services for clients statewide. First the researchers matched the IOT patients using demographic data. Researchers also matched the 19 IOT patients with other clients according to inpatient service use (ISU). The following comparisons were made:

    1. Pre-treatment data with post-treatment data for IOT;
    2. Pre-treatment data with post-treatment data for CTS Demographic and ISU patients (All and Best Matches); and
    3. Post-treatment data for IOT with CTS Demographic and ISU (All and Best Matches))

Based on these comparisons, the researchers evaluated whether the IOT group after treatment showed a greater decrease in number of admissions and inpatient days as compared to the CTS and ISU study groups. Results indicated that the IOT group showed significantly greater decreases in number of admissions and days than did either the All/Best Demographic groups. The IOT patients went from 1.63 to .58 admissions, a decrease of 1.05 and from 112.73 to 44.3 days, a decrease of 68.43. The All Demographic matched group went from .49 to .21 admissions, a decrease of .28, and from 19.42 to 22.1 days, a gain of 2.68 days. The Best Demographic group demonstrated a .1 decrease in admissions (.21 to .11) and a 3.7 decrease in days (15.1 to 11.4). However, when IOT was matched according to inpatient use, the groups were statistically similar in their decreases in number of admissions and days following treatment. The IOT decrease in number of admissions was 1.05 compared to a .85 decrease for ISU All and a 1.05 decrease for ISU Best; IOT decrease in number of days was 68.43 compared to 50 for ISU All and 47.8 for ISU Best.

Researchers concluded, "The results of this analysis are unfortunately equivocal. The trends observed in our data seem to suggest that IOT has had an effect, particularly with respect to reduced use of hospital days that was less pronounced for patients not under an IOT order. But these differences did not achieve statistical significance" (p. 282).

Researchers indicated that the small sample size of this study would yield low statistical power. Observed trends indicating differences in the groups might achieve statistical significance with a larger sample size (p. 282).

SUMMARY

Nineteen patients in IOT were matched with CTS clients on gender, diagnosis, age, and the quarter of the year they first appeared in CTS. The number of admissions and hospital days were compared for the same two 6-month periods both before and after treatment. The number of admissions and number of inpatient days for the IOT clients were compared with all matching CTS patients and also with the best single match. In addition to demographic matching, the researchers matched the 19 IOT patients with other patients according to inpatient service use (ISU) during the pre-treatment period using the all and best match strategy. Figure 1 displays the recorded data.

Figure 1: IOT Data Compared to Demographic (CTS) and Inpatient Service Usage (ISU) Matched Groups

1. Comparison 1: Pre and post-treatment data of IOT indicated a significant decrease in number of admissions (from 1.63 to .58; p=.008)) and number of days (from 112.73 to 44.3; p = .002).
   
2. Comparison 2: Demographic (CTS) All Matches showed a significant decrease in number of admissions, (from .49 to .21; p = .012) but not in number of days (from 19.42 to 22.1). Changes for the Demographic Best Matches group following treatment in both number of admissions (from .21 to .11) and number of days (from 15.1 to 11.4) were not significant. In the post-treatment Inpatient Service Usage (ISU) Matched group, researchers observed significant decreases in average number of admissions (1.53 to .68 for All; 1.63 to .58 for Best) and in average number of days (119.9 to 69.9 for All; 111.9 to 64.1 for Best).
   
3. Comparison 3: Although post-treatment data of IOT and CTS All Matches showed an actual difference in number of admissions (.58 to IOT; .21 for CTS All Matches) and number of days (44.3 for IOT; 22.1 for CTS All Matches), the researchers reported "no statistically significant difference for either days or admissions" (p. 279). Post treatment data indicated that although the IOT group had both a higher number of admissions (.58 compared to .11 for CTS Best) and of days (44.3 to 11.4 for CTS Best), these two groups were not statistically different on either factor (p. 280). Comparing post treatment data for IOT to the All ISU Matches where the original numbers for admissions and days were similar, the IOT patients ended up with lower numbers of admissions (.58 to .68) and of days (44.3 to 69.9), but the differences were not significant. Comparing IOT to ISU Best Matches, where the number of admissions was identical in the pre-treatment (1.63), the number of post treatment admissions was also identical (.58). The number of days for the two groups was not statistically significant.
   
   The key comparison in this research, as previously discussed in Comparison of Admission Data for IOT and CTS Groups, was whether the IOT group showed a greater decrease in number of admissions and number of days compared to the CTS and ISU study groups. Although post-treatment data for IOT showed a greater decrease in number of days compared to both ISU groups, the differences were not significant.