Triggers and Risk Factors for Parkinson Disease Psychosis
We conducted a retrospective cohort study to investigate the associations between patient-related factors and psychosis, and then analyzed the risk associated with different medications. First, a survival time analysis was adopted to identify patient-related factors. Then, with adjustment for these factors, a case-crossover design study (self-matched comparison that compared hazard and control periods in each patient with psychosis) was performed to evaluate the risk associated with medications.
This study was approved by the Bioethics Committee of Utano National Hospital (No.18–16). According to the Bioethics Committee, informed patient consent was not needed because the study was a retrospective review and data were analyzed anonymously.
The medical records of consecutive patients with PD who were treated in the Department of Neurology, the National Regional Center for Neurological Disorders and Utano National Hospital from March 2004 to November 2007 were retrospectively reviewed over a period of two years. Inclusion criteria were as follows. Patients who were diagnosed with Parkinson disease according to the United Kingdom Parkinson's disease Brain Bank Diagnostic Criteria (step 1 and 2), and who were prescribed dopaminergic replacement therapy. Irrespective of age, sex and severity of PD, both outpatients and hospitalized patients were included. Patients with a past history of psychosis were also included. Exclusion criteria were as follows; patients prescribed anti-psychotic drugs during the preceding 1 month; patients with a history of schizophrenia; and patients who met the DLB consensus criteria for probable or possible dementia with Lewy bodies.
The primary study outcome was serious psychoses, defined as those requiring a prescription of anti-psychotic drugs. According to the guidelines of PD psychosis, we assumed that prescription of anti-psychotic drugs was justified principally when psychosis was not improved by other treatments and patient suffering and behavioral changes were severe enough to endanger the patients or others. The diagnosis of psychosis was made based on the presence of psychotic symptoms (illusions, false sense of presence, hallucinations, or delusions) that were recurrent or prolonged for ≥1 month according to the provisional diagnostic criteria for PD psychosis. Anti-psychotic drugs included all anti-psychotic drugs available in Japan; tiapride, sulpiride, risperidone, chlorpromazine, thioridazine, fluphenazine, propericiazine, levomepromazine, haloperidol, quetiapine, olanzapine, perospirone, and aripiprazole. Prescriptions were collected longitudinally from study enrollment to the endpoint, which was defined as the occurrence of psychosis that required anti-psychotic drugs or the end of the 730 days of the study period.
In a survival time analysis, the observation period, censoring, and endpoint were defined as follows. The observation period was from the time of study enrollment to the endpoint, which was defined as the occurrence of any psychosis that required anti-psychotic drugs or the end of the 730-day study period. Observation was censored if patients were lost to follow-up or experienced an alternative outcome. Patients were censored when they were transferred to other hospitals because data could not be obtained (lost to follow-up). Because anti-psychotic medications were often prescribed prophylactically by surgeons to avoid surgery-associated delirium, patients who underwent surgery were also censored just before surgery (alternative outcome).
In patients with motor fluctuations modified Hoehn-Yahr (mH-Y) stage was evaluated in "ON period." Although age and duration of PD increased during the study period, and mH–Y stage and Mini-Mental State Examination (MMSE) scores might deteriorate, these variables were collected at the time of enrollment on the assumption that changes in them would not have much influence on the results of the survival time analysis.
For the case-crossover analysis, the prescriptions in the hazard period and those in the control periods were compared. Medications were prescribed every 14 days or 28 days in most cases, and drugs that were taken 1 day before the endpoint (the start of anti-psychotic medications) had been taken for 14 or 28 days before the endpoint; therefore, medications taken 1 day before the endpoint represent those taken for 14 days before the endpoint. In this context, the prescription 1 day before the endpoint was regarded as that in the hazard period. Medications in the hazard period were compared with those in the control periods (30 and 90 days before psychosis) (Additional file 1: Figure S1).
Prescription information was collected consecutively because it was changeable weekly or monthly. All prescriptions were collected to investigate the prescription of antipsychotic drugs throughout the observation period (of up to 2 years). Patients who had never been prescribed antipsychotic drugs within the observation period were regarded as controls or censored. In cases where antipsychotic drugs were prescribed, the reason why the antipsychotic drugs were prescribed was confirmed based on medical records.
To identify trigger medications, the doses of dopaminergic agents including L-Dopa, entacapone, dopamine agonists, amantadine and selegiline were collected. L-Dopa dose was calculated using the following formula: 1.0 × regular levodopa dose or 1.25 × regular levodopa dose if taking entacapone. Dopamine agonist dose was calculated as the L-Dopa equivalent dose (LDED) according to the following formula:
Because psychosis could be related to anticholinergic drugs, records of prescription of central anticholinergic drugs (trihexyphenidyl, piroheptine, biperiden, and promethazine) and that of donepezil (a central cholinergic drug) were collected. Records of prescription of rivastigmine were not collected because this drug was unavailable in Japan during the study period.
The incidence of psychosis was estimated as the number of patients with psychosis divided by the corresponding person-years at risk. The survival time was defined as the duration from enrollment to the first occurrence of psychosis. The risk of psychosis owing to patient-related factors was estimated using the Cox proportional hazard model incorporating age, sex, duration of PD, mH–Y stage (1.0–3.0 versus 4–5), and MMSE scores (≤ 24 versus > 24) at study enrollment as predictable variables. Age and sex were entered into the model and the other factors were selected according to a backward stepwise likelihood ratio test (Analysis I). Considering the patient-related factors identified in Analysis I, medications were analyzed in a case-crossover study design. Doses of L-Dopa, dopamine agonists, amantadine, and selegiline, as well as the use of anticholinergic drugs, were analyzed. Donepezil hydrochloride, an inhibitor of brain acetylcholine esterase, was also included in the analysis. The relative risk was then estimated using a generalized estimating equation that is the most appropriate technique for case-crossover designs. Link function was logistic and an autoregressive working correlation matrix was adopted because the data were collected longitudinally and it was assumed that the correlation of medication dose depends on the intervals between data collection time points. A variance-covariance matrix was estimated using the Huber-White sandwich estimator. All two-way interactions between predictable factors or covariates were examined and interactions were considered and they were incorporated in the analysis if statistically significant (Analysis II). Because risk of psychosis owing to medications may be enhanced in elderly patients, the case-crossover analysis was also performed in the subgroup of patients aged ≥ 70 years.
P values of less than 0.05 were considered statistically significant. Statistical analyses were performed using the statistical software program IBM SPSS version 21.
In the multiple variable analysis patients with missing variables were excluded. Dose of L-Dopa with adjustment for entacapone and dopamine agonists, selegiline and amantadine were regarded as scale variables. Anticholinergic drugs and donepezil hydrochloride were regarded as dichotomous (use versus no use).
On the basis of previous reports, incidence of psychosis was estimated 80 per 1,000 person-years 18]. The sample size was calculated to be 375 because 60 psychosis events were required within the 2-year period (multiple variable logistic analysis of six kinds of medications).
Methods
Study Design
We conducted a retrospective cohort study to investigate the associations between patient-related factors and psychosis, and then analyzed the risk associated with different medications. First, a survival time analysis was adopted to identify patient-related factors. Then, with adjustment for these factors, a case-crossover design study (self-matched comparison that compared hazard and control periods in each patient with psychosis) was performed to evaluate the risk associated with medications.
This study was approved by the Bioethics Committee of Utano National Hospital (No.18–16). According to the Bioethics Committee, informed patient consent was not needed because the study was a retrospective review and data were analyzed anonymously.
Patients
The medical records of consecutive patients with PD who were treated in the Department of Neurology, the National Regional Center for Neurological Disorders and Utano National Hospital from March 2004 to November 2007 were retrospectively reviewed over a period of two years. Inclusion criteria were as follows. Patients who were diagnosed with Parkinson disease according to the United Kingdom Parkinson's disease Brain Bank Diagnostic Criteria (step 1 and 2), and who were prescribed dopaminergic replacement therapy. Irrespective of age, sex and severity of PD, both outpatients and hospitalized patients were included. Patients with a past history of psychosis were also included. Exclusion criteria were as follows; patients prescribed anti-psychotic drugs during the preceding 1 month; patients with a history of schizophrenia; and patients who met the DLB consensus criteria for probable or possible dementia with Lewy bodies.
Definition of Psychosis
The primary study outcome was serious psychoses, defined as those requiring a prescription of anti-psychotic drugs. According to the guidelines of PD psychosis, we assumed that prescription of anti-psychotic drugs was justified principally when psychosis was not improved by other treatments and patient suffering and behavioral changes were severe enough to endanger the patients or others. The diagnosis of psychosis was made based on the presence of psychotic symptoms (illusions, false sense of presence, hallucinations, or delusions) that were recurrent or prolonged for ≥1 month according to the provisional diagnostic criteria for PD psychosis. Anti-psychotic drugs included all anti-psychotic drugs available in Japan; tiapride, sulpiride, risperidone, chlorpromazine, thioridazine, fluphenazine, propericiazine, levomepromazine, haloperidol, quetiapine, olanzapine, perospirone, and aripiprazole. Prescriptions were collected longitudinally from study enrollment to the endpoint, which was defined as the occurrence of psychosis that required anti-psychotic drugs or the end of the 730 days of the study period.
Survival Time Analysis
In a survival time analysis, the observation period, censoring, and endpoint were defined as follows. The observation period was from the time of study enrollment to the endpoint, which was defined as the occurrence of any psychosis that required anti-psychotic drugs or the end of the 730-day study period. Observation was censored if patients were lost to follow-up or experienced an alternative outcome. Patients were censored when they were transferred to other hospitals because data could not be obtained (lost to follow-up). Because anti-psychotic medications were often prescribed prophylactically by surgeons to avoid surgery-associated delirium, patients who underwent surgery were also censored just before surgery (alternative outcome).
In patients with motor fluctuations modified Hoehn-Yahr (mH-Y) stage was evaluated in "ON period." Although age and duration of PD increased during the study period, and mH–Y stage and Mini-Mental State Examination (MMSE) scores might deteriorate, these variables were collected at the time of enrollment on the assumption that changes in them would not have much influence on the results of the survival time analysis.
Case-crossover Analysis
For the case-crossover analysis, the prescriptions in the hazard period and those in the control periods were compared. Medications were prescribed every 14 days or 28 days in most cases, and drugs that were taken 1 day before the endpoint (the start of anti-psychotic medications) had been taken for 14 or 28 days before the endpoint; therefore, medications taken 1 day before the endpoint represent those taken for 14 days before the endpoint. In this context, the prescription 1 day before the endpoint was regarded as that in the hazard period. Medications in the hazard period were compared with those in the control periods (30 and 90 days before psychosis) (Additional file 1: Figure S1).
Prescription information was collected consecutively because it was changeable weekly or monthly. All prescriptions were collected to investigate the prescription of antipsychotic drugs throughout the observation period (of up to 2 years). Patients who had never been prescribed antipsychotic drugs within the observation period were regarded as controls or censored. In cases where antipsychotic drugs were prescribed, the reason why the antipsychotic drugs were prescribed was confirmed based on medical records.
To identify trigger medications, the doses of dopaminergic agents including L-Dopa, entacapone, dopamine agonists, amantadine and selegiline were collected. L-Dopa dose was calculated using the following formula: 1.0 × regular levodopa dose or 1.25 × regular levodopa dose if taking entacapone. Dopamine agonist dose was calculated as the L-Dopa equivalent dose (LDED) according to the following formula:
Because psychosis could be related to anticholinergic drugs, records of prescription of central anticholinergic drugs (trihexyphenidyl, piroheptine, biperiden, and promethazine) and that of donepezil (a central cholinergic drug) were collected. Records of prescription of rivastigmine were not collected because this drug was unavailable in Japan during the study period.
Statistical Analysis
The incidence of psychosis was estimated as the number of patients with psychosis divided by the corresponding person-years at risk. The survival time was defined as the duration from enrollment to the first occurrence of psychosis. The risk of psychosis owing to patient-related factors was estimated using the Cox proportional hazard model incorporating age, sex, duration of PD, mH–Y stage (1.0–3.0 versus 4–5), and MMSE scores (≤ 24 versus > 24) at study enrollment as predictable variables. Age and sex were entered into the model and the other factors were selected according to a backward stepwise likelihood ratio test (Analysis I). Considering the patient-related factors identified in Analysis I, medications were analyzed in a case-crossover study design. Doses of L-Dopa, dopamine agonists, amantadine, and selegiline, as well as the use of anticholinergic drugs, were analyzed. Donepezil hydrochloride, an inhibitor of brain acetylcholine esterase, was also included in the analysis. The relative risk was then estimated using a generalized estimating equation that is the most appropriate technique for case-crossover designs. Link function was logistic and an autoregressive working correlation matrix was adopted because the data were collected longitudinally and it was assumed that the correlation of medication dose depends on the intervals between data collection time points. A variance-covariance matrix was estimated using the Huber-White sandwich estimator. All two-way interactions between predictable factors or covariates were examined and interactions were considered and they were incorporated in the analysis if statistically significant (Analysis II). Because risk of psychosis owing to medications may be enhanced in elderly patients, the case-crossover analysis was also performed in the subgroup of patients aged ≥ 70 years.
P values of less than 0.05 were considered statistically significant. Statistical analyses were performed using the statistical software program IBM SPSS version 21.
Variables
In the multiple variable analysis patients with missing variables were excluded. Dose of L-Dopa with adjustment for entacapone and dopamine agonists, selegiline and amantadine were regarded as scale variables. Anticholinergic drugs and donepezil hydrochloride were regarded as dichotomous (use versus no use).
Study Size
On the basis of previous reports, incidence of psychosis was estimated 80 per 1,000 person-years 18]. The sample size was calculated to be 375 because 60 psychosis events were required within the 2-year period (multiple variable logistic analysis of six kinds of medications).
Source...