We read with interest the recent Movement Disorders Task Force recommendations for scales to assess psychosis in Parkinson's disease (PD).1 They defined several recommended scales, cautioning that no scale captures the full range of PD psychosis and highlighting that treatment responsiveness has not been established. We report a secondary data analysis evaluating the treatment responsiveness of the Scale for Assessment of Positive Symptoms (SAPS).
The SAPS fulfills criteria as a recommended scale. It is a structured patient interview developed for schizophrenia,2 containing five domains: hallucinations, delusions, bizarre behavior, positive formal thought disorder, and inappropriate affect, with higher scores indicating more severe psychosis. We analyzed data from the psychosis and clozapine in Parkinson's disease trial conducted by the Parkinson Study Group.3 Fifty-four subjects completed the 4-week randomized trial of clozapine versus placebo. Data collected for psychiatric assessment included the Mini-Mental Status Examination (MMSE), Clinical Global Impression–Change, and SAPS.
We assessed the distribution of SAPS values by domain for baseline and final visits, both overall and divided into MMSE > 24 (nondemented) or MMSE ≤ 24 (demented). To measure the responsiveness, we calculated effect sizes, defined as the change from baseline divided by the standard deviation (SD) of the change. We constructed two linear regression models. In one model, the independent variable of interest was MMSE, and the model adjusted for age, gender, baseline SAPS scores, and total UPDRS scores. In the other, CGI change was the variable of interest with adjustment for age, gender, and baseline SAPS, MMSE, and total UPDRS scores.
Demographic characteristics for this population have been reported.3 The mean total score for the baseline SAPS was 21.6 (SD, 12.6); for hallucinations, 9.3 (4.2); delusions, 5.8 (5.1); and positive formal thought disorder, 4.4 (5.6). The mean was 1.7 (3.2) for bizarre behavior and 0.4 (1.1) for inappropriate affect. The demented cohort had a higher total SAPS score of 24 (vs. 18), driven primarily by the thought disorder domain (6.1 vs. 2.4). Demented and nondemented cohorts had similar means for hallucinations and delusions (9.8 vs. 8.7 and 6.0 vs. 5.5, respectively).
Total score improved 39%, and both the SAPS total and individual domain scores improved. Changes in hallucinations and delusions accounted for 83% of the total score change. The demented cohort showed less change (total score change −6.16 vs. −10.04 for nondemented cohort), although MMSE scores were not significantly associated with responsiveness to treatment. The effect size of the SAPS was −0.73 overall and −0.91 for hallucinations. In the regression analysis, each point increase on baseline SAPS was associated with a 0.54 point decrease in SAPS change (95% CI [−0.76, −0.31], P < 0.0001).
Improvement in CGI correlated with improvement in the SAPS, driven primarily by the hallucinations and delusions domains (Fig. 1). In 7 cases (13%), the SAPS improved but CGI rating was worse. In these cases, hallucinations and delusions improved, but scores for bizarre behavior worsened. In the regression analysis, each unit change in CGI was associated with a 4.53 point change in the total SAPS (95% CI [2.37, 6.69], P = 0.0001).
These results confirm that the SAPS is responsive to change in PD psychosis. Change in SAPS was largely driven by the hallucination and delusion domains, which are the most common manifestations of PD psychosis.4 Subjects with worse psychosis were less responsive to treatment change. Therefore, the apparent effect of cognitive impairment on treatment change may be a result of confounding, as subjects with cognitive impairment had higher total SAPS scores. Other medical or psychiatric comorbidities that were not included in our analyses may mediate treatment response, and these findings should be examined in other data sets. However, our findings challenge the notion that the severity of cognitive impairment independently limits treatment response in PD psychosis. Overall, the regression analysis suggests that CGI is strongly associated with SAPS scores. Based on this analysis, a change in total SAPS score of 4 to 5 points appears clinically meaningful.