ORIGINAL CONTRIBUTION Association of Dopamine Agonist Use With Impulse Control Disorders in Parkinson Disease Daniel Weintraub, MD; Andrew D. Siderowf, MD, MSCE; Marc N. Potenza, MD, PhD; Joseph Goveas, MD; Knashawn H. Morales, ScD; John E. Duda, MD; Paul J. Moberg, PhD; Matthew B. Stern, MD Objective: To determine the frequency and correlates of cluding 11 (4.0%) with an active ICD. Compulsive gam- impulse control disorders (ICDs) in Parkinson disease (PD). bling and compulsive sexual behavior were equally com- mon. In a multivariate model, treatment with a dopamine Design: An unstructured screening interview for ICDs agonist (P=.01) and a history of ICD symptoms prior to (compulsive gambling, buying, and sexual behavior) fol- PD onset (P=.02) predicted current ICD. There were no lowed by a telephone-administered structured inter- differences between the dopamine agonists in their as- view for screen-positive patients. sociation with ICDs (P=.21), and daily doses of dopa- Setting: Two university-affiliated movement disorders mine agonists were higher in patients with an ICD than centers. in dopamine agonist–treated patients without an ICD (PϽ.001). Participants: A convenience sample of 272 patients with idiopathic PD who were screened for psychiatric com- Conclusions: Patients with PD treated with a dopa- plications. mine agonist should be made aware of the risk of devel- oping an ICD and monitored clinically. Because dopa- Main Outcome Measures: Presence of compulsive mine agonists are increasingly being used for other gambling, buying, or sexual behavior as assessed by the indications, future research should assess the dopamine Minnesota Impulsive Disorders Interview. agonist–associated risk for ICDs in other populations. Results: Eighteen patients (6.6%) with PD met criteria for an ICD at some point during the course of PD, in- Arch Neurol. 2006;63:969-973 ECENT OBSERVATIONAL case series, all 11 patients with PD identi- studies suggest that im- fied as meeting DSM-IV criteria for patho- pulse control disorders logical gambling were taking a dopamine (ICDs), particularly patho- agonist, 9 of whom were taking pramipex- logical gambling, may have ole and 2, ropinirole hydrochloride.3 increased frequency in Parkinson disease Regarding other ICDs in PD, in a series R1-3 (PD). Impulse control disorders consti- of 15 patients with either PD or multiple sys- Author Affiliations: tute a group of psychiatric disorders in tem atrophy and compulsive hypersexual- Departments of Psychiatry DSM-IV-TR,4 their essential feature being ity, dopamine agonist treatment was impli- (Drs Weintraub, Goveas, and a failure to resist an impulse, drive, or cated as the cause of the behavior in 14 Moberg), Neurology 7 (Drs Weintraub, Siderowf, temptation to perform an act that is harm- cases. There have also been anecdotal re- Duda, Moberg, and Stern), and ful to the person or to others. Other ICDs ports of compulsive buying in association 8 Biostatistics and Epidemiology without formal diagnostic criteria in DSM- with dopamine replacement therapies. (Dr Morales), University of IV-TR include compulsive sexual behav- We report the results of a screening and Pennsylvania, and Parkinson’s ior and compulsive buying.5 assessment study of ICDs in PD investi- Disease Research, Education, Although there are case reports of le- gating the: (1) frequencies of compulsive and Clinical Center vodopa-induced ICDs in PD,6 recent case buying, gambling, and sexual behaviors; (Drs Weintraub, Duda, Moberg, series have implicated treatment with (2) demographic and clinical correlates of and Stern) and Mental Illness dopamine agonists as a more frequent the aforementioned ICDs; and (3) asso- Research, Education, and cause of pathological gambling. Driver- ciation between ICDs and dopamine ago- Clinical Center (Dr Weintraub), 2 Philadelphia Veterans Affairs Dunckley et al identified 9 patients (0.5% nist use. We hypothesized that ICDs in PD Medical Center, Philadelphia; of clinic sample) with a documentation of are associated with dopamine agonist treat- Department of Psychiatry, pathological gambling, 8 of whom were ment and that this association is dose de- Yale University, New Haven, treated with pramipexole dihydrochloride pendent and similar across the entire class Conn (Dr Potenza). and 1 with pergolide mesylate. In another of dopamine agonists. (REPRINTED) ARCH NEUROL / VOL 63, JULY 2006 WWW.ARCHNEUROL.COM 969 ©2006 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/01/2021 METHODS culated both for dopamine agonists only (dopamine agonist LEDD) and for dopamine agonistsϩlevodopa (total LEDD). To probe for possible risk factors for the development of ICDs PARTICIPANTS in PD, data were obtained for factors that have been reported to be associated with ICDs in PD (type and dose of dopamine The study population was outpatients diagnosed with idio- replacement therapy, disease duration, age, and sex)1-3 or that pathic PD, predominantly of mild to moderate severity, con- were factors of interest (history of ICD behavior, global cog- firmed by a movement disorders specialist. Subjects were es- nition, educational level, and marital status). For the pur- tablished patients at 1 of 2 movement disorders centers (either poses of this study, dopamine agonists were considered to be the University of Pennsylvania or the Philadelphia Veterans Af- pramipexole, ropinirole, and pergolide. Amantadine hydro- fairs Medical Center) and were thought to represent a cross- chloride, which has an unclear mechanism of action but has section of the clinics’ populations, save the exclusion of pa- some dopamine agonist properties,15 was considered sepa- tients unable to provide informed consent because of cognitive rately, and no patient was prescribed bromocriptine mesylate impairment. Participants completed a psychiatric screening in- or apomorphine hydrochloride. Reliable data were not avail- terview as part of a study of the frequency and correlates of de- able for the duration of treatment with dopamine replacement pression in PD. The institutional review boards at the 2 insti- therapies. All clinical and demographic data were obtained di- tutions approved the study, and written informed consent was rectly from the patient during the screening interview and, when obtained from all subjects. possible, verified by medical record review. DATA COLLECTION AND MEASURES ANALYSIS Patients were screened between July 2004 and June 2005. Movement disorders professionals were instructed to refer Demographic and clinical characteristics of ICD and non-ICD any willing patient with PD, without regard for their psychi- subjects were compared using a Fisher exact test for categori- atric status (eg, no patient was referred for having an ICD), cal variables and the Wilcoxon 2-sample test for continuous for the screening interview at the conclusion of his or her variables. A comparison of the frequency of ICDs on different clinic appointment. dopamine agonists was made using a Fisher exact test. The as- Two trained research assistants administered the screen- sociation between dopamine agonist dosage and ICDs was de- ing battery, which included open-ended questions about the termined using the Wilcoxon 2-sample test. Variables signifi- existence (lifetime, anytime during PD, and currently) of re- cant at the 0.1 significance level (uncorrected for multiple current compulsive buying, gambling, or sexual behaviors. Sub- comparisons) in the univariate analysis were entered as inde- jects were also administered the 15-item Geriatric Depression pendent variables in an exact logistic regression model, with Scale9 and the Mini-Mental State Examination10 as part of the presence of an active ICD as the dependent variable. For the screening process. logistic regression model, all continuous measures were di- Those who screened positive for an ICD during the chotomized at the median. Results of the exact logistic regres- course of PD were contacted by telephone in August or Sep- sion model are presented in terms of odds ratios, 95% confi- tember 2005 by 1 of us (D.W. or J.G.) and administered a dence intervals, and P values. Analyses were performed with modified Minnesota Impulsive Disorders Interview (MIDI),11 SPSS version 13.0 (SPSS Inc, Chicago, Ill) and SAS version 9.1 which includes queries for the presence of clinically signifi- (SAS Institute, Inc, Cary, NC). cant compulsive gambling, sexual, and buying behaviors. Patients were instructed to answer the questions based on RESULTS their state at the time they were symptomatic. Impulse con- trol disorders were defined as answering in the affirmative to 1 (compulsive sexual behavior and compulsive shopping) or PATIENT CHARACTERISTICS 2 (compulsive gambling) gateway questions plus an affirma- tive answer to 1 or more of the remaining questions of the Two hundred seventy-two patients, ranging in age from relevant ICD module of the MIDI. The same threshold has 35 to 91 years, completed the screening process. The use been used to define problem gambling in other studies.12 of the Philadelphia Veterans Affairs Medical Center as a The MIDI was administered to confirm the presence of ICDs site led to a preponderance of men in the study popula- during the course of PD only (ie, not applied to pre-PD– tion. One half of subjects (137 [50.4%] of 272) were tak- onset ICD symptoms). ing a dopamine agonist at screening. For patients taking To verify data accuracy, the study primary investigator (D.W.) reviewed the medical records