Comparative Utility of Atypical Antipsychotics for the Treatment of Psychosis in Parkinson’s Disease: A Systematic Review and Bayesian Network Meta-analysis

Abstract

We performed a systematic review and Bayesian network meta-analysis to determine atypical antipsychotics that are effective and safe for the treatment of psychosis in Parkinson’s disease (PD). We conducted a comprehensive literature search using PubMed/MEDLINE, Cochrane Library, and Japana Centra Revuo Medicina (Ichu-shi Web). We used randomized controlled trials evaluating the utility of atypical antipsychotics for the treatment of psychosis in PD using the Brief Psychiatric Rating Scale (BPRS) and the Unified PD rating Scale parts III (UPDRS-III) as the endpoints. Posterior distributions of mean differences between each treatment and placebo were estimated using Bayesian network meta-analysis. The distributions describing each treatment effect were expressed as means (95% credible intervals). Ten trials involving any two treatment arms using clozapine (64 subjects in four trials), olanzapine (99 subjects in three trials), quetiapine (79 subjects in five trials), risperidone (five subjects in one trial), or placebo (156 subjects in seven trials) were finally included in the present study. Pooled estimates of each posterior distribution based on the BPRS were as follows: clozapine, −2.0 (−6.7 to 2.7); olanzapine, 0.5 (−2.3 to 3.4); quetiapine, 0.3 (−3.9 to 4.5); and risperidone, −4.7 (−57.4 to 53.3). Based on the UPDRS-III, the pooled estimates were clozapine, 0.7 (−3.8 to 4.3); olanzapine, 2.8 (0.8 to 5.1); quetiapine, 3.3 (−0.7 to 5.8); and risperidone, 4.5 (−57.7 to 63.4). Although clozapine had an effective and relatively safe profile, all atypical antipsychotics included in the present study may be unsafe, as they may worsen motor function when compared to placebo.

It is estimated that there are more than 10 million people with Parkinson’s disease (PD) worldwide.¹⁾ Given our aging society, it is predicted that the number of patients with PD older than 50 years will double by 2030.²⁾ PD is a progressive neurodegenerative disorder characterized by motor dysfunction and non-motor symptoms, such as sleep disorder, cognitive change, and psychosis.³⁾ Psychosis is in fact a major non-motor symptom in PD. Previous studies have shown that 40 to 60% of patients with PD experience psychosis at some time.^4,5) Occasionally, severe psychosis leads to hospitalizations and is related to mortality.⁶⁾ Therefore, psychosis should be treated and its deterioration prevented. However, this symptom is often caused by PD medications essential to managing motor dysfunction.⁷⁾ It is suggested to reduce or discontinue PD medications when drug-induced psychosis occurs, although these measures may deteriorate motor function.⁸⁾ Thus, other medications are used to treat psychosis.

The use of atypical antipsychotics (APs) is one option to treat psychosis in PD, as they lead to less deterioration of motor function when compared to typical APs.^8,9) These drugs have somewhat different effects on dopamine and serotonin receptors. Several studies investigating the efficacy of each atypical AP have been conducted. Clozapine has demonstrated clear efficacy without deteriorating motor function.¹⁰⁾ However, it is difficult to use in clinical practice due to the necessity of blood monitoring. Quetiapine is another candidate for the treatment of psychosis, as proposed in relevant reviews and guidelines.^8,9,11) This agent has also been shown to have a safe profile in previous small-scale studies.^12–14) However, these previous studies assessed the drug’s safety based on the absence of statistically significant deterioration of motor function. This may indicate that these studies were too small to detect statistically significant changes in motor function. However, it may be difficult to address this issue using clinical trials in a frequentist framework, as it is difficult to retain statistically sufficient sample sizes due to low participation rates and high dropout rates in this field.¹⁵⁾ Therefore, the safety and efficacy of these drugs have not been established conclusively.

Network meta-analysis is a statistical methodology synthesizing direct and indirect comparisons between competing treatments.^16–18) The results of this kind of analysis can be used to determine the best treatment included in the network. Additionally, performing network meta-analysis in a Bayesian framework enables us to easily interpret the obtained results. Bayesian analysis calculates the direct probability of interesting hypotheses or questions.¹⁸⁾ Thus, using Bayesian network meta-analysis to identify effective and safe atypical APs may provide us with useful information for clinical decision-making. We thus conducted a systematic review and Bayesian network meta-analysis to evaluate the efficacy and safety of each atypical AP used for the treatment of psychosis in PD.

METHODS

Data Sources and Search Strategies

PubMed/MEDLINE, Cochrane Library, and Japana Centra Revuo Medicina (Ichu-shi Web) were independently searched by two reviewers (RI and YK) for articles potentially related to our topic published between 1966 and June 2017. The following search terms were used: Parkinson’s disease AND psychosis AND antipsychotic. The WHO’s International Clinical Trials Registry Platform, ClinicalTrials.gov, and the Japanese National University Hospital Medical Information Network-Clinical Trials Registry were also explored. In addition, we performed manual searches on the reference lists of relevant reviews.^8,9,11)

Study Selection and Data Extraction

We included randomized controlled trials that evaluated the utility of atypical APs for the treatment of psychosis in PD by comparing any atypical AP to placebo or other active treatment. The articles were required to report both of the following endpoints for the evaluation of the efficacy and safety of atypical APs: 1) the Brief Psychiatric Rating Scale (BPRS) by Overall and Gorham¹⁹⁾; and 2) the Unified Parkinson’s Disease Rating Scale part III (UPDRS-III). There was no language restriction for the articles.

Based on these requirements, two reviewers (RI and YK) independently selected studies and extracted data regarding the study characteristics and outcomes from the selected studies. The extracted data included first author, publication year, study design, interventions, comparisons, participant characteristics, eligibility criteria, psychosis definition, and changes in the BPRS total score and UPDRS-III subtotal score. Quality assessments for each included study were also conducted using the Cochrane Collaboration’s tool for assessing risk of bias.²⁰⁾ Disagreements at any step were resolved through discussion.

Statistical Analysis

The characteristics of the included studies are summarized using means and standard deviations (S.D.s) for continuous variables, and frequencies and percentages (%) for categorical variables.

Change in the BPRS total score was used as the efficacy endpoint. The score ranges from 18 to 126, and lower values indicate more normal psychiatric condition. Change in the UPDRS-III subtotal score was used as the safety endpoint. The score ranges from 0 to 56, and lower values indicate more normal motor function. The mean differences (MDs) between each treatment and a mutual comparator were estimated by synthesizing eligible studies. The treatment used most among the included studies was set as the comparator and its treatment effect was set to zero. Uniform distributions (mean, 0; variance, 10000) were adopted as non-informative priors for MDs.

Posteriors for MDs were computed using Markov–Chain Monte Carlo (MCMC) simulation, which is based on the Metropolis–Hastings algorithm. The first 50000 iterations were discarded as the burn-in period. After the burn-in period, 500000 iterations with a thinning interval of 100 were used to estimate posteriors. MCMC convergence was evaluated based on visual inspection of the iteration plot. Inconsistency between direct and indirect comparisons was evaluated using the node-splitting method.²¹⁾ The computed posteriors for MDs are summarized using means and 95% credible intervals (CrIs). The probability (p) indicating each treatment superiority to the mutual comparator was also calculated.

Rankograms were produced by calculating ranking probabilities indicating the best treatment, the second best treatment, and so on, based on the posterior probabilities for the MDs.²²⁾ The surface under the cumulative ranking (SUCRA) (%) was also calculated.²¹⁾ If the SUCRA is equal to 100%, the treatment is the best among the treatments included in the network, and if it is equal to 0%, the treatment is the worst among those included in the network.

Graphical outputting of the network and inconsistency evaluations were conducted using STATA/SE 13.0 software (StataCorp LP; College Station, TX, U.S.A.). Other statistical analyses were performed using SAS software, version 9.4 for Windows (SAS Institute Inc.; Cary, NC, U.S.A.).

RESULTS

Study Identification

Fifty articles were identified based on the database search, and three articles were added to these candidates based on the manual search (Fig. 1). After removing duplicated articles, 34 articles were screened by title or abstract. At this step, 12 articles were excluded as six were not related to this study, four did not meet the study design criteria, one reported the results of an interim analysis, and one was an abstract for a meeting. When assessing the eligibilities of the full-text articles, 12 articles were excluded. Five of these articles did not use the BPRS or UPDRS-III as the endpoint,^23–27) five were lacked necessary information,^12,28–31) one was a study including not patients with PD,¹³⁾ and one was an extension of an open-label study of another appropriate study.³²⁾ Therefore, ten articles were potentially appropriate for the present study. Although one article did not report results required for the present study, we included it in the study because raw data from the study were available.³³⁾ We calculated the required results using data from participants who had completed the study. One article comparing clozapine and olanzapine was excluded because it led to a serious inconsistency in the comparison network.³⁴⁾ Additionally, one article reported the results of two randomized controlled trials.³⁵⁾ Ten studies were thus included in the present study.^{10,14,15,33,35–39)} The comparison network generated based on these studies is presented in Fig. 2. Placebo treatment was set as the mutual comparator, and its effect was set to zero, as it was the most frequently used intervention.

Fig. 1. Flow Diagram for the Study

Fig. 2. Comparison Network Generated from the Included Studies

The size of each node represents the total number of subjects. Line weight reflects the number of studies.

Study Characteristics

The characteristics of included studies are summarized in Table 1 and the qualities determined using the Cochrane Collaboration’s tool for assessing risk of bias are presented in Fig. 3. The study duration ranged from 4 to 56 weeks. Six-tenth of the studies clearly targeted psychosis associated with PD treatment, and others targeted psychosis often recognized in patients with PD.

Table 1. Characteristics of the Studies Included in the Bayesian Network Meta-analysis

Study	Design	Psychosis definition	Arm 1	Arm 2
Breier A (1), 2002	RCT double-blind 4 weeks	Treatment-associated psychosis defined according to the DSM-IV	Olanzapine	Placebo
			n=41	n=42
			age, 73.5 (8.7)	age, 68.0 (11.0)
			% male, 63.4	% male, 76.2
			BPRS, −2.7	BPRS, −3.1
			UPDRS-III, 2.6	UPDRS-III, −0.2
Breier A (2), 2002	RCT double-blind 4 weeks	Treatment-associated psychosis defined according to the DSM-IV	Olanzapine	Placebo
			n=49	n=28
			age, 70.9 (6.3)	age, 70.5 (8.2)
			% male, 67.3	% male, 64.3
			BPRS, −4.3	BPRS, −5.5
			UPDRS-III, 2.7	UPDRS-III, −0.3
Ellis T, 2000	RCT double-blind 6 weeks	Levodopa-induced psychosis not satisfactorily managed by levodopa dose reduction	Clozapine	Risperidone
			n=4	n=5
			age, 74.0 (5.9)	age, 74.4 (5.9)
			% male, NA	% male, NA
			BPRS, −3.0	BPRS, −6.0
			UPDRS-III, −1.8	UPDRS-III, 2.4
Fernadez HH, 2009	RCT double-blind 4 weeks	Nocturnal visual hallucination on stable doses of PD medications	Quetiapine	Placebo
			n=8	n=8
			age, 64.6 (7.5)	age, 71.5 (7.5)
			% male, NA	% male, NA
			BPRS, 1.0	BPRS, −0.3
			UPDRS-III, −5.7	UPDRS-III, 2.8
Morgante L, 2002	RCT open-label rater-blinded 12 weeks	Antiparkinsonian drug-induced psychosis defined according to the DSM-IV	Clozapine	Quetiapine
			n=10	n=10
			age, 68.0 (11.0)	age, 69.0 (12.0)
			% male, 50.0	% male, 50.0
			BPRS, −10.7	BPRS, −9.3
			UPDRS-III, −2.1	UPDRS-III, 1.6
Morgante L, 2004	RCT open-label rater-blinded 12 weeks	Antiparkinsonian drug-induced psychosis	Clozapine	Quetiapine
			n=20	n=20
			age, 69.0 (10.7)	age, 70.0 (10.1)
			% male, 50.0	% male, 50.0
			BPRS, −10.7	BPRS, −8.4
			UPDRS-III, −1.3	UPDRS-III, 1.0
Nichols MJ, 2013	RCT double-blind 4 weeks	Clinically significant hallucinations or delusions, as judged by the treating neurologist or psychiatrist and the investigator	Olanzapine	Placebo
			n=9	n=7
			age, 69.4 (4.9)	age, 69.7 (6.0)
			% male, 44.4	% male, 85.7
			BPRS, −0.1	BPRS, −0.1
			UPDRS-III, 3.2	UPDRS-III, 1.6
Rabey JM, 2007	RCT double-blind 12 weeks	Visual or auditory hallucinations and/or delusions that significantly affected the subject’s quality of life	Quetiapine	Placebo
			n=30	n=28
			age, 75.5 (8.1)	age, 74.5 (8.7)
			% male, 56.7	% male, 57.1
			BPRS, −0.2	BPRS, −4.1
			UPDRS-III, 2.2	UPDRS-III, −1.9
Shotbolt P, 2009	RCT double-blind 12 weeks	Hallucinations, suspiciousness, or unusual thought content (delusions) of a severity >3/7 on the BPRS	Quetiapine	Placebo
			n=11	n=13
			age, 74.0 (8.0)	age, 70.0 (8.0)
			% male, 63.6	% male, 69.2
			BPRS, −4.2	BPRS, −2.5
			UPDRS-III, −3.0	UPDRS-III, 1.1
The Parkinson Study Group, 1999	RCT double-blind 56 weeks	Antiparkinsonian drug-induced psychosis	Clozapine	Placebo
			n=30	n=30
			age, 70.8 (8.6)	age, 71.9 (8.1)
			% male, 46.7	% male, 66.7
			BPRS, −9.3	BPRS, −2.6
			UPDRS-III, −3.6	UPDRS-III, −1.8

Abbreviations: BPRS, Brief Psychiatric Rating Scale; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders 4th edition; RCT, randomized controlled trial; UPDRS-III, Unified Parkinson’s Disease Rating Scale part III. Age is reported as mean (standard deviation). BPRS and UPDRS-III are reported as mean score changes from baseline in each treatment arm.

Fig. 3. Quality Assessments of Each Included Study Using the Cochrane Collaboration’s Tool for Assessing Risk of Bias

A plus symbol indicates that the study report is judged as low-risk against the question. A minus symbol indicates that the study report is judged as high-risk against the question. A question symbol indicates that the study report has unclear risk against the question.

Bayesian Network Meta-analysis

MCMC simulations were well-converged without serious autocorrelations for both endpoints. No inconsistencies were detected in a closed loop of the network using the node-splitting method. Figure 4 shows the posterior distributions for MDs of the BPRS. The estimates for each treatment were as follows: clozapine (mean, −2.0; 95% CrI, −6.7 to 2.7; p, 0.70); olanzapine (mean, 0.5; 95% CrI, −2.3 to 3.4; p, 0.33); quetiapine (mean, 0.3; 95% CrI, −3.9 to 4.5; p, 0.51); and risperidone (mean, −4.7; 95% CrI, −57.4 to 53.3; p, 0.56). Figure 5 shows the posterior distributions for MDs of the UPDRS-III. The estimates for each treatment were as follows: clozapine (mean, 0.7; 95% CrI, −3.8 to 4.3; p, 0.29); olanzapine (mean, 2.8; 95% CrI, 0.8 to 5.1; p, 0.01); quetiapine (mean, 3.3; 95% CrI, −0.7 to 5.8; p, 0.05); and risperidone (mean, 4.5; 95% CrI, −57.7 to 63.4; p, 0.44).

Fig. 4. Posterior Distributions for Each Treatment When Compared to Placebo for the Brief Psychiatric Rating Scale

Abbreviations: BPRS, Brief Psychiatric Rating Scale; CrI, credible interval.

Fig. 5. Posterior Distributions for Each Treatment When Compared to Placebo for the Unified Parkinson’s Disease Rating Scale Part III

Abbreviations: CrI, credible interval; UPDRS-III, Unified Parkinson’s Disease Rating Scale part III.

Rankograms for both endpoints are illustrated in Fig. 6. SUCRAs for the BPRS were as follows: clozapine (74.0%), olanzapine (36.7%), quetiapine (33.2%), and risperidone (56.1%). SUCRAs for the UPDRS-III were as follows: clozapine (79.4%), olanzapine (45.2%), quetiapine (28.4%), and risperidone (47.0%).

Fig. 6. Rankograms for the Brief Psychiatric Rating Scale (A) and Unified Parkinson’s Disease Rating Scale Part III (B)

Abbreviations: BPRS, Brief Psychiatric Rating Scale; UPDRS-III, Unified Parkinson’s Disease Rating Scale part III.

DISCUSSION

We directly and indirectly synthesized relevant studies using Bayesian network meta-analysis to identify effective and safe atypical APs for the treatment of psychosis in PD. Point estimates from posteriors for the BPRS indicate that risperidone and clozapine improve psychosis when compared to placebo. Similarly, point estimates for the UPDRS-III suggest that clozapine leads to the smallest deterioration of motor function, although it is inferior to placebo in this respect. Olanzapine and quetiapine may not only deteriorate motor function, but also impair psychosis. The superior probabilities of these treatments when compared to placebo and their SUCRAs also support these findings. However, the comparison network moderately depended on indirect comparisons. Therefore, the obtained results should be interpreted with caution.

Assessments of olanzapine and risperidone strongly depended on the indirect comparison for estimating posteriors, as these treatments formed single-standing nodes within the comparison network. Thus, their estimates are less accurate than those obtained using the closed loop. Furthermore, risperidone was not directly compared to placebo and was only investigated in a small study.³⁶⁾ Its estimates may thus be considerably imprecise. In fact, the posteriors for risperidone for both endpoints were very vague. Although the probabilities and SUCRAs indicate that risperidone is a moderately useful treatment among the atypical APs, these findings are not comparable to those for other treatments. The closed loop of the network was formed between placebo, clozapine, and quetiapine. Since we investigated whether the posterior estimations were appropriately conducted, no inconsistencies within the loop were detected using the node-splitting method. Additionally, we confirmed the robustness of estimates within the loop by conducting sensitivity analyses excluding olanzapine and/or risperidone from the network. We are thus able to compare and discuss the results obtained using the loop.

The present study suggests that clozapine is a useful treatment for psychosis in PD. Although this finding is consistent with those of previous reports, clozapine is rarely used in clinical practice due to its severe side effects.⁸⁾ Further investigations regarding the safety of clozapine are necessary. It was reported that quetiapine is also useful for the treatment of psychosis in PD because of its acceptable tolerability.^8,9) However, our study indicates that it determinately deteriorated motor function, while its probability for improving psychiatric symptoms when compared to placebo was approximately one-half. It appears that the utility of this drug for the treatment of psychosis in PD is inferior to that of placebo. A retrospective cohort study reported that the use of quetiapine for the treatment of psychosis during hospitalization increased mortality in patients with PD when compared to no use of APs.⁴⁰⁾ Therefore, quetiapine use for the treatment of psychosis in PD is unfavorable. Taken together, our data indicate that there may not be clinically safe conventional atypical APs for the treatment of psychosis unresponsive to reductions in the dose of PD medications. This interpretation is subjective, although it is based on findings obtained using not a frequentist framework, but a Bayesian framework. Aripiprazole, which we did not include in this study, is clinically used for psychosis in PD because its pharmacological function is distinct from other APs.⁴¹⁾ One single arm, open-label study previously evaluated the utility of aripiprazole.⁴²⁾ The study reported that aripiprazole significantly improved psychosis based on the BPRS, but caused deterioration of motor function leading to the study’s discontinuation. The safety of aripiprazole may need to be evaluated in future studies, similarly to the APs analyzed in this study. Pimavanserin, which had a safety profile equivalent to that of placebo in a phase III study, was recently introduced as the first agent for the treatment of psychosis in PD.^23,24) We were also unable to include the study of pimavanserin, as it did not use the BPRS as an endpoint. Its efficacy and safety should thus be evaluated in future clinical trials or Bayesian network meta-analyses comparing conventional atypical APs in order to establish its utility.

The limitations of the present study are as follows. The major limitation was that the numbers of included studies and the participants in those studies were limited. We synthesized data from a diverse set of studies regardless of their qualities or designs and only excluded studies when there were inconsistencies that were statistically detected. Despite the application of the random-effect model, the estimations may not have been appropriately conducted. The fact that the posterior distributions of clozapine and quetiapine for the BPRS were distorted raises this concern. Additionally, it would be difficult to detect potential inconsistencies because of estimation dispersions due to the small number of participants. Therefore, we should also pay attention to estimation accuracies and precisions in the closed loop. The number of databases used for the systematic search was also a limitation. Although we referred to previous reviews and guidelines, we may have failed to detect potentially relevant studies, as recently published studies were not covered within those references. Finally, the differences in the BPRS and UPDRS-III observed between each treatment and placebo may not be clinically meaningful. Regarding the BPRS, an approximately 10-point change is needed to consider that the severity level of the psychosis has been improved according to the clinical global impression severity score.⁴³⁾ Such clinically significant differences were not observed in this study. Nevertheless, there were differences regarding the UPDRS-III indicating that active treatment deteriorated motor function even if the differences were clinically small. Therefore, our study would not actively support the use of atypical APs, including clozapine, for psychosis in PD.

In summary, the utility of clozapine, as suggested in our study, was consistent with the findings of previous reports. However, our results also suggested that the use of quetiapine for the treatment of psychosis in PD may be inadequate, as it was likely to lead to deterioration of motor function despite its limited efficacy. Additionally, any of the atypical APs included in the present study would be unsafe for patients with PD. Therefore, they should be used with sufficient caution or should not be used in this patient population.

Conflict of Interest

The authors declare no conflict of interest.

REFERENCES

• 1) Parkinson’s Disease Foundation. “Statistics on Parkinson’s.”: ‹http://www.pdf.org/parkinson_statistics›, cited 24 July, 2017.
• 2) Dorsey ER, Constantinescu R, Thompson JP, Biglan BA, Holloway RG, Kieburtz K, Marshall FJ, Ravina BM, Schifitto G, Siderowf A, Tanner CM. Projected number of people with Parkinson’s disease in the most populous nations, 2005 through 2030. Neurology, 68, 384–386 (2007).
• 3) Parkinson’s Disease Foundation. “Symptoms.”: ‹http://www.pdf.org/symptoms›, cited 24 July, 2017.
• 4) Fénelon G, Soulas T, Zenasni F, de Langavant LC. The changing face of Parkinson’s disease-associated psychosis: a cross-sectional study based on the new NINDS-NIMH criteria. Mov. Disord., 25, 763–766 (2010).
• 5) Starkstein SE, Brockman S, Hayhow BD. Psychiatric syndromes in Parkinson’s disease. Curr. Opin. Psychiatry, 25, 468–472 (2012).
• 6) Forsaa EB, Larsen JP, Wentzel-Larsen T, Alves G. What predicts mortality in Parkinson disease?: a prospective population-based long-term study. Neurology, 75, 1270–1276 (2010).
• 7) Sawada H, Oeda T, Yamamoto K, Umemura A, Tomita S, Hayashi R, Kohsaka M, Kawamura T. Trigger medications and patient-related risk factors for Parkinson disease psychosis requiring anti-psychotic drugs: a retrospective cohort study. BMC Neurol., 13, 145 (2013).
• 8) Cooney JW, Stacy M. Neuropsychiatric issues in Parkinson’s disease. Curr. Neurol. Neurosci. Rep., 16, 49 (2016).
• 9) Japanese Society of Neurology. “Treatment guidelines for Parkinson’s disease 2011.”: ‹https://www.neurology-jp.org/guidelinem/parkinson.html›, cited 24 July, 2017.
• 10) The Parkinson Study Group. Low-dose clozapine for the treatment of drug-induced psychosis in Parkinson’s disease. N. Engl. J. Med., 340, 757–763 (1999).
• 11) Friedman JH. Atypical antipsychotic drugs in the treatment of Parkinson’s disease. J. Pharm. Pract., 24, 534–540 (2011).
• 12) Ondo WG, Tintner R, Dat Voung K, Lai D, Ringholz G. Double-blind, placebo-controlled, unforced titration parallel trial of quetiapine for dopaminergic-induced hallucinations in Parkinson’s disease. Mov. Disord., 20, 958–963 (2005).
• 13) Kurlan R, Cummings J, Raman R, Thal L; Alzheimer’s Disease Cooperative Study Group. Quetiapine for agitation or psychosis in patients with dementia and parkinsonism. Neurology, 68, 1356–1363 (2007).
• 14) Rabey JM, Prokhorov T, Miniovitz A, Dobronevsky E, Klein C. Effect of quetiapine in psychotic Parkinson’s disease patients: a double-blind labeled study of 3 months’ duration. Mov. Disord., 22, 313–318 (2007).
• 15) Shotbolt P, Samuel M, Fox C, David AD. A randomized controlled trial of quetiapine for psychosis in Parkinson’s disease. Neuropsychiatr. Dis. Treat., 5, 327–332 (2009).
• 16) Higgins JP, Whitehead A. Borrowing strength from external trials in a meta-analysis. Stat. Med., 15, 2733–2749 (1996).
• 17) Salanti G. Indirect and mixed-treatment comparison, network, or multiple-treatments meta-analysis: many names, many benefits, many concerns for the next generation evidence synthesis tool. Res. Synth. Methods, 3, 80–97 (2012).
• 18) Tonin FS, Rotta I, Mendes AM, Pontarolo R. Network meta-analysis: a technique to gather evidence from direct and indirect comparisons. Pharm. Pract (Granada), 15, 943 (2017).
• 19) Overall JE, Gorham DR. The brief psychiatric rating scale. Psychol. Rep., 10, 799–812 (1962).
• 20) Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, Savovic J, Schulz KF, Weeks L, Sterne JA, Cochrane Bias Methods Group, Cochrane Statistical Methods Group. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ, 343 (oct18 2), d5928 (2011).
• 21) Dias S, Welton NJ, Caldwell DM, Ades AE. Checking consistency in mixed treatment comparison meta-analysis. Stat. Med., 29, 932–944 (2010).
• 22) Salanti G, Ades A, Ioannidis JP. Graphical methods and numerical summaries for presenting results from multiple-treatment meta-analysis: an overview and tutorial. J. Clin. Epidemiol., 64, 163–171 (2011).
• 23) Cummings J, Isaacson S, Mills R, Williams H, Chi-Burris K, Corbett A, Dhall R, Ballard C. Pimavanserin for patients with Parkinson’s disease psychosis: a randomised, placebo-controlled phase 3 trial. Lancet, 383, 533–540 (2014).
• 24) Meltzer HY, Mills R, Revell S, Williams H, Johnson A, Bahr D, Friedman JH. Pimavanserin, a serotonin(2A) receptor inverse agonist, for the treatment of Parkinson’s disease psychosis. Neuropsychopharmacology, 35, 881–892 (2010).
• 25) Merims D, Balas M, Peretz C, Shabtai H, Giladi N. Rater-blinded, prospective comparison: quetiapine versus clozapine for Parkinson’s disease psychosis. Clin. Neuropharmacol., 29, 331–337 (2006).
• 26) Pollak P, Tison F, Rascol O, Destée A, Péré JJ, Senard JM, Durif F, Bourdeix I. Clozapine in drug induced psychosis in Parkinson’s disease: a randomised, placebo controlled study with open follow up. J. Neurol. Neurosurg. Psychiatry, 75, 689–695 (2004).
• 27) Chacón JR, Durán E, Durán JA, Alvarez M. Usefulness of olanzapine in the levodopa-induced psychosis in patients with Parkinson’s disease. Neurologia, 17, 7–11 (2002).
• 28) Friedman JH. Melperone is ineffective in treating Parkinson’s disease psychosis. Mov. Disord., 27, 803–804 (2012).
• 29) Tariot PN, Schneider L, Katz IR, Mintzer JE, Street J, Copenhaver M, Williams-Hughes C. Quetiapine treatment of psychosis associated with dementia: a double-blind, randomized, placebo-controlled clinical trial. Am. J. Geriatr. Psychiatry, 14, 767–776 (2006).
• 30) The French Clozapine Parkinson Study Group. Clozapine in drug-induced psychosis in Parkinson’s disease. Lancet, 353, 2041–2042 (1999).
• 31) Wolters EC, Hurwitz TA, Mak E, Teal P, Peppard FR, Remick R, Calne S, Calne DB. Clozapine in the treatment of parkinsonian patients with dopaminomimetic psychosis. Neurology, 40, 832–834 (1990).
• 32) Factor SA, Friedman JH, Lannon MC, Oakes D, Bourgeois K; Parkinson Study Group. Clozapine for the treatment of drug-induced psychosis in Parkinson’s disease: results of the 12 week open label extension in the PSYCLOPS trial. Mov. Disord., 16, 135–139 (2001).
• 33) Nichols MJ, Hartlein JM, Eicken MG, Racette BA, Black KJ. A fixed-dose randomized controlled trial of olanzapine for psychosis in Parkinson’s disease. F1000Res., 2, 150 (2013).
• 34) Goetz CG, Blasucci LM, Leurgans S, Pappert EJ. Olanzapine and clozapine: comparative effects on motor function in hallucinating PD patients. Neurology, 55, 789–794 (2000).
• 35) Breier A, Sutton VK, Feldman PD, Kadam DL, Ferchland I, Wright P, Friedman JH. Olanzapine in the treatment of dopamimetic-induced psychosis in patients with Parkinson’s disease. Biol. Psychiatry, 52, 438–445 (2002).
• 36) Ellis T, Cudkowicz ME, Sexton PM, Growdon JH. Clozapine and risperidone treatment of psychosis in Parkinson’s disease. J. Neuropsychiatry Clin. Neurosci., 12, 364–369 (2000).
• 37) Morgante L, Epifanio A, Spina E, Di Rosa AE, Zappia M, Basile G, La Spina P, Quattrone A. Quetiapine versus clozapine: a preliminary report of comparative effects on dopaminergic psychosis in patients with Parkinson’s disease. Neurol. Sci., 23 (Suppl. 2), s89–s90 (2002).
• 38) Morgante L, Epifanio A, Spina E, Zappia M, Di Rosa AE, Marconi R, Basile G, Di Raimondo G, La Spina P, Quattrone A. Quetiapine and clozapine in parkinsonian patients with dopaminergic psychosis. Clin. Neuropharmacol., 27, 153–156 (2004).
• 39) Fernandez HH, Okun MS, Rodriguez RL, Malaty IA, Romrell J, Sun A, Wu SS, Pillarisetty S, Nyathappa A, Eisenschenk S. Quetiapine improves visual hallucinations in Parkinson’s disease but not through normalization of sleep architecture: results from a double-blind clinical-polysomnography study. Int. J. Neurosci., 119, 2196–2205 (2009).
• 40) Weintraub D, Chiang C, Kim HM, Wilkinson J, Marras C, Stanislawski B, Mamikonyan E, Kales HC. Association of antipsychotic use with mortality risk in patients with Parkinson’s disease. JAMA Neurol., 73, 535–541 (2016).
• 41) Weintraub D, Chen P, Ignacio RV, Mamikonyan E, Kales HC. Patterns and trends in antipsychotic prescribing for Parkinson’s disease psychosis. Arch. Neurol., 68, 899–904 (2011).
• 42) Friedman JH, Berman RM, Goetz CG, Factor SA, Ondo WG, Wojcieszek J, Carson WH, Marcus RN. Open-label flexible-dose pilot study to evaluate the safety and tolerability of aripiprazole in patients with psychosis associated with Parkinson’s disease. Mov. Disord., 21, 2078–2081 (2006).
• 43) Leucht S, Kane JM, Kissling W, Hamann J, Etschel E, Engel R. Clinical implications of Brief Psychiatric Rating Scale scores. Br. J. Psychiatry, 187, 366–371 (2005).