Trigger medications and patient-related risk factors for Parkinson disease psychosis requiring anti-psychotic drugs: a retrospective cohort study

BACKGROUND Psychoses such as hallucinations are a frequent non-motor problem in patients with Parkinson disease (PD) and serious psychosis requires anti-psychotic medications that worsen Parkinsonism. Although psychosis could be associated with patient-related or biological factors such as cognition, age, and severity of PD, it can also be associated with medications.Therefore we aimed to investigate patient-related and medication-related risks of psychosis requiring anti-psychotic medications (serious psychosis). METHODS A retrospective cohort of 331 PD patients was followed for 2 years. Patient-related factors associated with risk of psychosis were identified by a survival time analysis. In patients who developed psychosis, medications during the hazard period (1-14 days before psychosis) were contrasted with those during the control periods (1 and 3 months before psychosis) using a case-crossover analysis to identify medication-related risks of psychosis. RESULTS Serious psychosis was detected in 52 patients and the incidence was estimated to be 116 (95% confidence interval [CI], 85-148) per 1,000 person-years. Analyses of baseline characteristics revealed the risk to be higher in patients with a modified Hoehn-Yahr stage of ≥4 (hazard ratio [HR], 2.22; 95% CI, 1.11-4.40), those with a longer duration of PD (HR, 1.25; 95% CI, 1.00-1.55, per 5 years) and those with Mini-Mental State Examination scores of ≤24 (HR, 2.66; 95% CI, 1.37-5.16). The case-crossover analysis revealed that anti-cholinergics use (HR, 19.7; 95% CI, 2.39-162) elevated the risk, while donepezil use reduced it (HR, 0.48; 95% CI, 0.27-0.85). CONCLUSIONS Risk of psychosis was elevated by increasing severity of PD, cognitive dysfunction and duration of the disease. It was elevated by use of anti-cholinergic drugs and reduced by use of donepezil. The medication-related risk was higher in patients aged ≥ 70 years. In contrast, there was no significant medication-related risk in younger patients, suggesting different pathomechanisms between young and old patients.