Syncope and Its Consequences in Patients with Dementia Receiving Cholinesterase Inhibitors a Population-Based Cohort Study

Home , Diuretic

ORIGINAL INVESTIGATION Syncope and Its Consequences in Patients With Dementia Receiving Cholinesterase Inhibitors A Population-Based Cohort Study

Sudeep S. Gill, MD, MSc; Geoffrey M. Anderson, MD, PhD; Hadas D. Fischer, MD; Chaim M. Bell, MD, PhD; Ping Li, PhD; Sharon-Lise T. Normand, PhD; Paula A. Rochon, MD, MPH

Background: Cholinesterase inhibitors are commonly val [CI], 1.57-1.98). Other syncope-related events were prescribed to treat dementia, but their adverse effect pro- also more common among people receiving cholinester- file has received little attention. These drugs can pro- ase inhibitors compared with controls: hospital visits for voke symptomatic bradycardia and syncope, which may bradycardia (6.9 vs 4.4 events per 1000 person-years; HR, lead to permanent pacemaker insertion. Drug-induced 1.69; 95% CI, 1.32-2.15), permanent pacemaker inser- syncope may also precipitate fall-related injuries, includ- tion (4.7 vs 3.3 events per 1000 person-years; HR, 1.49; ing hip fracture. 95% CI, 1.12-2.00), and hip fracture (22.4 vs 19.8 events per 1000 person-years; HR, 1.18; 95% CI, 1.04-1.34). Re- Methods: In a population-based cohort study, we in- sults were consistent in additional analyses in which sub- vestigated the relationship between cholinesterase in- jects were either matched on their baseline comorbidity hibitor use and syncope-related outcomes using health status or matched using propensity scores. care databases from Ontario, Canada, with accrual from April 1, 2002, to March 31, 2004. We identified 19 803 Conclusions: Use of cholinesterase inhibitors is asso- community-dwelling older adults with dementia who were ciated with increased rates of syncope, bradycardia, pace- prescribed cholinesterase inhibitors and 61 499 controls who were not. maker insertion, and hip fracture in older adults with dementia. The risk of these previously underrecognized Results: Hospital visits for syncope were more fre- serious adverse events must be weighed carefully against quent in people receiving cholinesterase inhibitors than the drugs’ generally modest benefits. in controls (31.5 vs 18.6 events per 1000 person-years; adjusted hazard ratio [HR], 1.76; 95% confidence inter- Arch Intern Med. 2009;169(9):867-873

ORE THAN 24 MILLION sodes of syncope can involve multiple po- people worldwide tential contributors8; in some cases, no have dementia, and es- clear cause is identified despite extensive timates suggest this and costly diagnostic evaluation.9 number will rise above The cardiovascular effects of cholines- 81M million by the year 2040.1 Cholines- terase inhibitors are complex, but they gen- terase inhibitors, such as donepezil hy- erally augment vagal influences on the drochloride, galantamine hydrobromide, heart and promote bradycardia that can re- Author Affiliations: Institute and rivastigmine, are commonly pre- sult in neurocardiogenic syncope.10-12 In- for Clinical Evaluative Sciences, scribed to treat the symptoms of Alzhei- frequent reporting of syncope in random- Toronto, Ontario, Canada mer disease and related dementias.2 De- ized controlled trials (RCTs) and reviews (Drs Gill, Anderson, Fischer, spite increasing rates of use,3 debate of cholinesterase inhibitors may explain Bell, Li, and Rochon); Department of Medicine, persists about the clinical effectiveness and why this adverse effect is underrecog- 4-6 Queen’s University, Kingston, cost-effectiveness of these drugs. Less at- nized. Similarly, clinical practice guide- Ontario (Dr Gill); Departments tention has been paid to their adverse effect lines on both syncope13,14 and dementia of Medicine and Health Policy, profile. In particular, many clinicians are treatment6 do not mention this impor- Management, and Evaluation, unaware of the potential for cholinester- tant adverse drug effect. University of Toronto, Toronto ase inhibitors to provoke syncope. Drug-induced syncope can have seri- (Drs Anderson, Bell, and Rochon); and Harvard Medical Syncope is a common presenting symp- ous consequences. For example, clini- School and Harvard School of tom that involves brief loss of conscious- cians may consider permanent pace- Public Health, Boston, ness with spontaneous recovery, and it is maker insertion for patients receiving Massachusetts (Dr Normand). usually accompanied by falling.7 Epi- cholinesterase inhibitors who present with

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 9), MAY 11, 2009 WWW.ARCHINTERNMED.COM 867

Downloaded From: https://jamanetwork.com/ on 09/23/2021 symptomatic bradycardia and syncope if these drugs are had not received any cholinesterase inhibitor prescriptions in not recognized as potential precipitants of the episode. the year prior to cohort entry. Cohort entry was defined as the Furthermore, syncope can lead to fall-related injuries, in- date of the first dispensed cholinesterase inhibitor. Controls were cluding hip fractures.15 Patients with dementia who sus- required to have contact with a physician within the 3 months tain hip fractures are at high risk of subsequent func- prior to cohort entry to ensure that they had access to health 16 services. In addition, the control cohort was matched to the drug tional decline, institutionalization, and death. cohort by year and quarter of cohort entry to ensure that the It remains unclear how often drug-induced syncope cohorts were contemporaneous. We included only new users and its consequences occur in older adults with demen- of cholinesterase inhibitors to reduce the potential for selec- tia who are seen in routine clinical practice; many of these tion bias.21 Three drugs in this class are available through the people were ineligible to participate in the RCTs evalu- ODB (donepezil, galantamine, and rivastigmine). Other de- ating cholinesterase inhibitors.17 The few postmarket- mentia medications, including tacrine hydrochloride and ing studies available on this topic are limited by small memantine hydrochloride, were not licensed for use in Canada sample size.18,19 We therefore undertook a large, popu- during the period of this study. Health Canada approved meman- lation-based cohort study to determine the relationship tine in November 2004, and tacrine never received approval. between cholinesterase inhibitor use and a spectrum of To improve comparability of the 2 cohorts, subjects had to meet the following 3 restriction criteria22,23 at the time of co- interrelated events, including hospital visits for (1) syn- hort entry: (1) evidence of a dementia diagnosis recorded in cope, (2) bradycardia, (3) permanent pacemaker inser- OHIP or DAD within the past 5 years,24 (2) community- tion, and (4) hip fracture. To demonstrate the consis- dwelling status (ie, not residing in a long-term care facility at tency of our findings, we conducted secondary analyses baseline), and (3) no hospitalizations for syncope within the in which drug users and control subjects were either past year. We excluded long-term care residents because we matched according to the number of comorbid condi- felt they might have different patterns of health services utili- tions they possessed or matched using propensity scores. zation (such as receiving on-site care for episodes of syncope To demonstrate the specificity of our findings for syncope- rather than visiting a hospital). We excluded subjects with epi- related outcomes, we also performed analyses to ensure sodes of syncope in the past year to increase the likelihood that the absence of any significant association between cho- cholinesterase inhibitor use was the primary precipitant of outcomes following cohort entry. linesterase inhibitor use and 2 outcomes that were un- likely to be affected by exposure to these drugs (ie, pulmonary embolism and cataract extraction). DRUG EXPOSURE AND DISCONTINUATION We assumed that treatment with a cholinesterase inhibitor was METHODS discontinued and censored follow-up if the patient did not re- fill their prescription within 120 days of the prescription’s dis- DATA SOURCES pensation date. The ODB plan permits a maximum drug dis- pensation of 100 days’ supply, and prescriptions written for During the period of this study, Ontario, Canada, had a popu- longer periods have the remaining days converted by the dis- lation of approximately 12 million people, of whom 1.4 mil- pensing pharmacist into refills. Few patients were observed to lion were 65 years or older. A universally funded health care switch from one cholinesterase inhibitor to another during the program in Ontario covers nearly all physician services, medi- period of this study, and we did not examine outcomes for in- cations, and hospital services for these seniors. Five adminis- dividual cholinesterase inhibitors. trative health care databases that track this population were linked to develop the study cohort. These databases included OUTCOMES pharmacy records from the Ontario Drug Benefit program (ODB), emergency department records from the National Am- We examined first hospital visits for syncope (International Clas- bulatory Care Reporting System (NACRS), hospitalization rec- sification of Diseases and Related Health Problems,10th Revision, ords from the Canadian Institute for Health Information Dis- Canada [ICD-10], code R55), as recorded in either the NACRS charge Abstract Database (DAD), physician billing information (ie, emergency department visits) or DAD (ie, hospital admis- for inpatient and outpatient services from the Ontario Health sions). Preliminary data suggested that approximately 60% of Insurance Plan (OHIP), and basic demographic information and patients assessed in emergency departments for syncope were vital statistics from the Registered Persons Database (RPDB). not subsequently admitted to hospital, suggesting that reli- Encrypted unique identifiers that are common between data- ance on hospital admission data alone would lead to an under- bases were used to link anonymous information on demo- recognition of the true event rate.25 Other investigators have graphics and health services utilization for patients in our study. used similar approaches to identify episodes of syncope.26,27 Only There is little basic information on patients missing in these the first hospital visit for syncope was counted as an event for databases because of administrative requirements. For ex- patients who experienced recurrent syncope. ample, the coding accuracy and completeness of drug claims 20 We also examined several related outcomes (Figure): hos- in the ODB database is excellent, with an error rate of only 0.7%. pital visits for bradycardia or complete atrioventricular block The study was approved by ethics review boards at Sunny- (ICD-10 codes R00.1 and I44.2), permanent pacemaker inser- brook Health Sciences Centre and Queen’s University. tion (identified using DAD Canadian Classification of Health In- terventions [CCI] codes), and hip fracture (ICD-10 codes S72.0, COHORT ASSEMBLY S72.1, S72.2, and S72.9). Hip fractures were excluded if they were pathological, associated with trauma, or associated with We identified all Ontario residents aged 66 years or older with epilepsy. Bradycardia and atrioventricular block are potential a prior diagnosis of dementia from April 1, 2002, through March mechanisms by which cholinesterase inhibitors can precipi- 31, 2004. From this pool of subjects, we defined 2 cohorts: those tate syncope, whereas pacemaker insertion and hip fracture are who were new users of cholinesterase inhibitors and those who potential consequences of syncope. Other studies have used simi-

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 9), MAY 11, 2009 WWW.ARCHINTERNMED.COM 868

Downloaded From: https://jamanetwork.com/ on 09/23/2021 lar methods to examine the relationship between drug use and pacemaker insertion.28 The ICD-10 coding for hip fracture and Bradycardia Syncope the CCI coding for medical procedures in the DAD have been shown to be reliable sources of information with very high posi- tive predictive values when validated against reabstracted health Syncope-related outcomes records.29

STATISTICAL ANALYSIS Pacemaker insertion Fall-related injuries (eg, hip fracture) We first calculated event rates for the cohorts, using the number of events per 1000 person-years for the 4 outcomes (syncope, bradycardia, permanent pacemaker insertion, and hip frac- Figure. Conceptualizing syncope-related outcomes. ture). We conducted time-to-event analyses using Cox proportional hazards models to derive hazard ratios (HRs) and 95% confidence intervals (CIs) for each of the outcomes in cho- We developed a third set of results by matching subjects using linesterase inhibitor users vs control subjects. With each of these a propensity score. The propensity score is the probability of re- outcomes examined in a separate analysis, patients were ob- ceiving treatment for an individual with specific prognostic fac- served until they experienced the outcome event of interest, tors. It is a scalar summary of all observed confounders. The ra- discontinued drug therapy (for those in the drug cohort), ini- tionale and methods underlying the use of propensity scores are tiated drug therapy (for those in the control cohort), died, or detailed elsewhere.31,32 We computed a propensity score for new reached the end of the follow-up period (March 31, 2004). The receipt of a cholinesterase inhibitor by developing a logistic re- proportional hazards assumption was confirmed in each model gression model with 34 covariates describing patient character- using an interaction term between the independent variable and istics. Covariates were selected based on guidance from recent time. Analyses were performed using SAS statistical software studies.33,34 Table 1 lists many of the characteristics included in for UNIX (version 9.1; SAS Institute, Cary, North Carolina). the propensity score. After an assessment of the balance of measured covariates between cholinesterase inhibitor users and non- PRIMARY ANALYSIS user controls, we used the resulting predicted probabilities as propensity scores and matched each drug user with up to 3 controls. For each of the 4 outcomes we used risk adjustment to ac- Matching involved a caliper width (ie, interval for successful match) count for differences in baseline risk. The covariates in our mod- of 0.6 of the standard deviation of the log odds of the propensity els included factors that would influence the development or score. This method has been demonstrated to remove approxi- 35 recognition of syncope or the related outcomes. For the syn- mately 90% of the bias from measured confounders. cope and bradycardia outcomes, the covariates included demo- Finally, to assess the specificity of our findings we exam- graphic factors (such as age and sex), use of antiarrhythmic drugs ined the associations between cholinesterase inhibitor use and or other drugs with negative chronotropic effects, presence of either subsequent hospitalization for pulmonary embolism coronary artery disease, aortic stenosis, atrial fibrillation or other (ICD-10 codes I26.x) or subsequent cataract extraction (OHIP cardiac conduction disorders, previous insertion of a perma- ophthalmologist procedure code E140). The goal was to con- nent pacemaker or implantable cardioverter defibrillator, and firm the absence of associations where none would be plausi- the Charlson comorbidity index score.30 For the pacemaker out- bly expected. Confirming negative findings in these analyses come, covariates included demographic factors, use of antiar- provides support for the notion that the cholinesterase inhibi- rhythmic drugs or other drugs with negative chronotropic ef- tor users and controls in our study were similar in terms of their fects, and the presence of atrial fibrillation or other cardiac access to care (eg, cataract extraction) and vulnerability to un- conduction disorders. For the pacemaker outcome, we ex- related diseases (eg, pulmonary embolism). cluded subjects who had previously received a permanent pacemaker or implantable cardioverter defibrillator. For the hip frac- RESULTS ture outcome, covariates included demographic factors, the Charlson comorbidity index score, history of hip fracture, and use of drugs that influence fracture risk (hormone replace- For the primary analysis, we identified 19 803 new us- ment therapy; bisphosphonates; raloxifene hydrochloride; thia- ers of cholinesterase inhibitors (13 641 receiving donep- zide diuretic agents; steroids; benzodiazepines; and anticon- ezil; 3448, galantamine; and 2714, rivastigmine) and vulsant, antidepressant, antipsychotic, and antiparkinsonian 61 499 control subjects. The 2 cohorts had comparable agents). baseline characteristics (Table 1), including similar pro- To assess the possibility of biased ascertainment of out- portions with pacemaker insertions in the 5 years prior comes, we compared the number of emergency department vis- to cohort entry (1.6% of drug cohort vs 1.8% of con- its in the year prior to cohort entry made by new users of cho- trols). Emergency department utilization in the year prior linesterase inhibitors with those made by controls. to cohort entry was also similar in the cholinesterase inhibitor and control cohorts (Table 2). ADDITIONAL ANALYSES Hospital visits for syncope were more frequent in people receiving cholinesterase inhibitors than in con- To demonstrate the consistency of our findings, we controls (31.5 vs 18.6 events per 1000 person-years; ad- ducted several complementary analyses. We reanalyzed our re- justed HR, 1.76; 95% CI, 1.57-1.98). Syncope-related out- sults by matching subjects from the drug cohort with up to 3 individuals from the control cohort. Matching was based on comes were also more common among people receiving the burden of comorbid disease as measured by a modified ver- cholinesterase inhibitors compared with controls: hos- sion of the Charlson comorbidity index that excluded the point pital visits for bradycardia (6.9 vs 4.4 events per 1000 for presence of dementia.30 Subjects were matched on having person-years; HR, 1.69; 95% CI, 1.32-2.15), permanent 0, 1, or 2 or more points on this modified comorbidity index. pacemaker insertion (4.7 vs 3.3 events per 1000 person-

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 9), MAY 11, 2009 WWW.ARCHINTERNMED.COM 869

Downloaded From: https://jamanetwork.com/ on 09/23/2021 Table 1. Baseline Characteristics in the Cholinesterase Table 2. Subjects With 0, 1, or 2 or More Emergency Inhibitor and Control Cohortsa Department (ED) Visits in the Year Prior to Cohort Entrya

Cholinesterase Cholinesterase Inhibitor Control Inhibitor Cohort Control Cohort Cohort Cohort ED Visits, No. (n=19 803) (n=61 499) Characteristic (n=19 803) (n=61 499) 0 10 847 (54.8) 33 774 (54.9) Age, mean (SD), y 80.4 (6.3) 80.4 (7.4) 1 4603 (23.2) 13 467 (21.9) Age, median (IQR), y 80 (76-85) 80 (75-86) Ն2 4353 (22.0) 14 258 (23.2) Sex, male 7433 (37.5) 23 879 (38.8) Low-income status 5782 (29.2) 19 551 (31.8) a Data are given as number (percentage). Medical historyb Coronary artery disease 8363 (42.2) 28 916 (47.0) Pulmonary embolism 64 (0.3) 273 (0.4) Aortic valve stenosis 56 (0.3) 236 (0.4) Table 3. Event Rates and Hazard Ratios (HRs) Atrial fibrillation 4827 (24.4) 17 217 (28.0) for Syncope and Related Outcomes Conduction disorder 1086 (5.5) 4346 (7.1) Seizure disorder 608 (3.1) 3578 (5.8) Cholinesterase Control Permanent pacemaker 314 (1.6) 1091 (1.8) Outcome Inhibitor Cohort Cohort Implantable cardioverter 15 (0.1) 65 (0.1) Syncope (n=19 803) (n = 61 499) defibrillator Hospital visits, No. (%) 428 (2.2) 944 (1.5) Myocardial infarction 1280 (6.5) 5670 (9.2) Follow-up time, mean (SD), d 251 (185) 302 (209) Congestive heart failure 1255 (6.3) 6921 (11.3) Person-time, d 4 960 953 18 540 121 Peripheral vascular disease 440 (2.2) 2212 (3.6) Event rate, events per 1000 31.5 18.6 Cerebrovascular disease 1729 (8.7) 8706 (14.2) person-years Chronic pulmonary disease 1341 (6.8) 6882 (11.2) HR (95% CI) Connective tissue disease 198 (1.0) 906 (1.5) Unadjusted 1.67 (1.49-1.87) 1 [Reference] Ulcer disease 268 (1.4) 1332 (2.2) Adjusteda 1.76 (1.57-1.98) 1 [Reference] Mild liver disease 19 (0.1) 212 (0.3) Bradycardia (n=19 803) (n=61 499) Diabetes mellitus 1581 (8.0) 6782 (11.0) Hospital visits, No. (%) 95 (0.5) 224 (0.4) Diabetes mellitus with 177 (0.9) 1123 (1.8) Follow-up time, mean (SD), d 254 (186) 304 (210) end-organ damage Person-time, d 5 022 969 18 693 345 Hemiplegia or paraplegia 161 (0.8) 1295 (2.1) Event rate, events per 1000 6.9 4.4 Moderate or severe renal 415 (2.1) 2529 (4.1) person-years disease HR (95% CI) Primary cancer 819 (4.1) 3370 (5.5) Unadjusted 1.58 (1.24-2.01) 1 [Reference] Moderate or severe liver 16 (0.1) 222 (0.4) Adjusteda 1.69 (1.32-2.15) 1 [Reference] disease Permanent pacemaker (n=19 489) (n=60 408) Metastatic cancer 137 (0.7) 864 (1.4) b c insertion Medications Pacemaker insertions, 64 (0.3) 166 (0.3) ␤ -Blockers 4252 (21.5) 13 693 (22.3) No. (%) Non–dihydropyridine calcium 1493 (7.5) 4675 (7.6) Follow-up time, mean (SD), d 254 (186) 304 (210) channel blockers Person-time, d 4 956 824 18 382 659 Digoxin 1479 (7.5) 5610 (9.1) Event rate, events per 1000 4.7 3.3 Conventional antipsychotics 225 (1.1) 1117 (1.8) person-years Atypical antipsychotics 2352 (11.9) 6101 (9.9) HR (95% CI) Antiarrhythmics 495 (2.5) 1890 (3.1) Unadjusted 1.40 (1.05-1.87) 1 [Reference] Anticonvulsants 530 (2.7) 3022 (4.9) Adjusteda 1.49 (1.12-2.00) 1 [Reference] Antidepressants 5177 (26.1) 14 850 (24.1) Hip fracture (n=19 803) (n=61 499) Antimanic agents 78 (0.4) 438 (0.7) Hospitalizations for hip 306 (1.6) 1008 (1.6) Benzodiazepines 3624 (18.3) 14 127 (23.0) fracture, No. (%) Index year (quarter) Follow-up time, mean (SD), d 252 (185) 302 (209) 2002 (2) 2536 (12.8) 7662 (12.5) Person-time, d 4 991 987 18 557 006 2002 (3) 2377 (12.0) 6986 (11.4) Event rate events per 1000 22.4 19.8 2002 (4) 2441 (12.3) 7371 (12.0) person-years 2003 (1) 2323 (11.7) 7542 (12.3) HR (95% CI) 2003 (2) 2607 (13.2) 8164 (13.3) Unadjusted 1.13 (0.99-1.28) 1 [Reference] 2003 (3) 2380 (12.0) 7359 (12.0) Adjusteda 1.18 (1.04-1.34) 1 [Reference] 2003 (4) 2555 (12.9) 8243 (13.4) 2004 (1) 2584 (13.0) 8172 (13.3) Abbreviation: CI, confidence interval. a Covariates for risk adjustment for each outcome detailed in the ”Primary Abbreviation: IQR, interquartile range. Analysis” subsection in the “Methods” section. a Data are presented as number (percentage) unless otherwise noted. b Patients with previously inserted pacemakers were excluded from this b Conditions recorded within the 5 years prior to cohort entry. analysis, leaving 19 803 − 314 = 19 489 drug users and 61 499 − 1091 = 60 408 c Medications dispensed in the 120 days before cohort entry. controls who were eligible for the pacemaker outcome.

years; HR, 1.49; 95% CI, 1.12-2.00), and hip fracture (22.4 In the comorbidity-matched analyses, 14 324 drug vs 19.8 events per 1000 person-years; HR, 1.18; 95% CI, users (72.3%) were each matched to 3 control subjects, 1.04-1.34) (Table 3). and the remaining 5479 drug users were each matched

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 9), MAY 11, 2009 WWW.ARCHINTERNMED.COM 870

Downloaded From: https://jamanetwork.com/ on 09/23/2021 to 2 controls. In these analyses conducted with subjects matched on their baseline burden of comorbid disease, Table 4. Outcomes From Additional Analyses results were consistent with those in the primary analy- (Matching on Comorbidity Scores or Propensity Scores) sis (Table 4). In the analysis involving propensity-based matching, HR (95% CI) 12 891 drug users (65.1%) were each matched to 4 con- Matching on Matching on trol subjects, 6811 drug users (34.4%) were matched to Comorbidity Propensity a b 2 controls, and the remaining 101 drug users (0.5%) were Outcome Scores Scores each matched to 1 control. Propensity-based matching Syncope Unadjusted 1.71 (1.49-1.96) 1.73 (1.50-1.99) produced results that were consistent with those from c the primary analysis and comorbidity-based matching Adjusted 1.81 (1.57-2.10) 1.77 (1.53-2.04) Bradycardia (Table 4). Unadjusted 1.72 (1.27-2.32) 1.87 (1.38-2.54) We confirmed the absence of any relationship be- Adjustedc 2.08 (1.47-2.96) 1.77 (1.28-2.43) tween cholinesterase inhibitor use and the 2 outcomes we Permanent pacemaker insertiond anticipated would be unrelated to drug use. Specifically, Unadjusted 1.57 (1.10-2.23) 1.51 (1.07-2.14) cholinesterase inhibitor use was not significantly associ- Adjustedc 1.72 (1.17-2.54) 1.49 (1.04-2.14) ated with either subsequent hospitalization for pulmo- Hip fracture Unadjusted 1.20 (1.03-1.39) 1.19 (1.02-1.39) nary embolism (HR, 0.97; 95% CI, 0.55-1.72) or subse- Adjustedc 1.21 (1.03-1.43) 1.26 (1.07-1.48) quent cataract extraction (HR, 1.00; 95% CI, 0.91-1.10). Abbreviations: CI, confidence interval; HR, hazard ratio. a Up to 3 control cohort subjects matched to each drug cohort subject, COMMENT with matching based on comorbidity score (0, 1, or Ն2 points). b Up to 3 control cohort subjects matched to each drug cohort subject, with matching based on propensity scores. This population-based study identifies increased risks of c Covariates for risk adjustment for each outcome are listed in the text. syncope, bradycardia, permanent pacemaker insertion, d Patients with previously inserted pacemakers were excluded from this analysis, leaving 19 803 − 314 = 19 489 drug users and 61 499 − 1091 = 60 408 and hip fracture in community-dwelling older adults who controls who were eligible to be matched for analysis of this outcome. received cholinesterase inhibitors. To our knowledge, this is the first large study to focus on the risk of syncope with these commonly prescribed medications, and the first to Our findings are generally consistent with the lim- document a link to pacemaker insertion and hip frac- ited published research on this topic. Schneider41 sug- ture. Although our study is observational, the relation- gested that drug-related syncope might lead to more in- ships we found are biologically plausible, the cohorts were juries in people prescribed cholinesterase inhibitors; our comparable at baseline, the results were consistent in the results now bolster these suspicions. The risk of syn- primary analysis and 2 sets of matched analyses, and we cope found in our study is similar to results of a review confirmed the absence of potentially spurious associa- by Birks2 of RCTs (odds ratio [OR], 1.90; 95% CI, 1.09- tions with unrelated outcomes. 3.33). However, the review’s wider CI reflects the fact that The impact of these adverse events on patient out- data on syncope were pooled from 5 RCTs with a total comes merits serious concern for several reasons. First, of only 2206 subjects. Similar limitations affect the risks emergency department assessments and hospitalizations are presented for accidental injury (OR, 1.35; 95% CI, 0.86- costly and expose cognitively impaired older patients to fur- 2.10) and fracture (OR, 0.96; 95% CI, 0.53-1.74) in the ther risks.36 Second, the prognosis following hip fracture review by Birks.2 Although these risks seem insignifi- in older patients with dementia is grim.16 Finally, al- cant, imprecise estimates were unavoidable in this re- though the retrospective nature of our study did not allow view because data could only be pooled from the few trials us to judge the appropriateness of individual procedures, reporting on these outcomes. Thus, the overall number RCTs have questioned the role of pacemaker insertion to of events was small.2 Vandenbroucke and Psaty42 high- treat neurocardiogenic syncope.11 This invasive proce- light the many challenges of assessing harm in meta- dure can involve serious complications, including venous analyses of RCTs. Newer RCTs not yet included in Birks’ thrombosis, pericardial tamponade, and infection.11 We review2 also involve too few adverse events to permit de- speculate that bradycardia and further episodes of syn- finitive conclusions, but some of these trials have re- cope might have been prevented in some subjects if their ported a slight excess of accidental injuries and frac- cholinesterase inhibitors had been discontinued. As a re- tures in subjects receiving cholinesterase inhibitor sult of these multiple consequences of syncope, the im- treatment.37,43 Few postmarketing studies have been pub- pact of cholinesterase inhibitor treatment on overall health lished on this topic, and most represent small case se- care costs may be currently underestimated, and the cost- ries or case reports.18,19,44 As a result, many clinicians do effectiveness of these drugs remains debatable.4-6 Prescrip- not recognize the potential for syncope-related adverse tion of cholinesterase inhibitors is steadily rising in many events with these drugs, and prescribing patterns vary countries,3 in part owing to expanding indications, includ- widely in patients with cardiovascular conditions.45 ing severe Alzheimer disease37 and non-Alzheimer demen- Cholinesterase inhibitors are thought to work by re- tias.38,39 Despite the fact that RCTs have failed to establish ducing inactivation of the neurotransmitter acetylcho- benefits from the use of these drugs to treat mild cognitive line, thereby increasing its activity in the synaptic cleft. This impairment, many physicians report prescribing them off- action is not limited to the central nervous system; well- label for patients with this condition.40 documented peripheral cholinergic effects include nau-

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 9), MAY 11, 2009 WWW.ARCHINTERNMED.COM 871

Downloaded From: https://jamanetwork.com/ on 09/23/2021 sea, diarrhea, muscle cramps, and urinary inconti- to events following cohort entry. Further research is nence.2,46,47 In the cardiovascular system, increased needed to establish the safety of these drugs in patients cholinergic activity leads to a variety of changes, includ- with prior syncopal episodes. ing bradyarrhythmias.10 It is reasonable to implicate cho- Additional research is needed to confirm our find- linesterase inhibitors as potential precipitants of syncope ings. The best approach may require combining evi- even when bradycardia is not present. For example, our dence from RCTs and observational studies42 because results demonstrate that clinical recognition of bradycar- many RCTs provide inadequate data on adverse events.52 dia is less frequent than is a presentation of syncope. We Older adults with dementia are vulnerable to adverse drug speculate this is because bradyarrhythmias may be tran- effects, and future RCTs evaluating treatments targeted sient (ie, present to provoke the episode of syncope but re- to this population should therefore provide comprehen- solved before a clinical assessment is performed). It is also sive documentation of common and serious outcomes possible that mechanisms other than bradycardia are re- such as falls (syncopal or otherwise) and injuries.53 sponsible for syncope associated with cholinesterase in- In conclusion, use of cholinesterase inhibitors is as- hibitor use.12 Older adults with dementia are particularly sociated with increased rates of syncope, bradycardia, susceptible to syncope and fall-related injuries; Kenny et pacemaker insertion, and hip fracture in older adults with al48 found that neurocardiovascular instability is highly dementia. The risk of these previously underrecognized prevalent in this population, and treatment with cholin- serious adverse events must be carefully weighed against esterase inhibitors may worsen these deficits. the drugs’ generally modest benefits. Physicians caring for people with dementia should be alert to the risks we document herein. Treatment deci- Accepted for Publication: December 18, 2008. sions need to be individualized and should involve a dis- Correspondence: Sudeep S. Gill, MD, MSc, Room 1-152, cussion with patients and their caregivers about the ex- Chapel Wing, St Mary’s of the Lake Hospital, 340 Union pected benefits and potential risks of treatment.49 In St, Kingston, ON K7L 5A2, Canada. clinical practice, syncope represents a common present- Author Contributions: Dr Gill had full access to all the ing symptom that can be provoked by many conditions. data in the study and takes responsibility for the integ- Clinicians evaluating a patient with syncope should be rity of the data and the accuracy of the data analysis. Study open to the possibility that several potential contribu- concept and design: Gill, Anderson, Fischer, Bell, Li, Nor- tors could exist, and they should not immediately as- mand, and Rochon. Analysis and interpretation of data: sume syncope is exclusively the result of exposure to a Gill, Anderson, Fischer, Bell, Li, Normand, and Ro- cholinesterase inhibitor.8,19 Many patients in our study chon. Drafting of the manuscript: Gill. Critical revision of had additional risk factors for syncope. Clinicians must the manuscript for important intellectual content: Gill, also appreciate that drug-induced syncope does not al- Anderson, Fischer, Bell, Li, Normand, and Rochon. Sta- ways occur immediately after treatment with the drug is tistical analysis: Gill, Li, and Normand. Obtained fund- started. The timing of adverse events relative to initia- ing: Gill, Anderson, and Rochon. Administrative, techni- tion of cholinesterase inhibitor treatment may be influ- cal, or material support: Fischer and Li. Study supervision: enced by several factors, including the slow upward dose Anderson, Normand, and Rochon. titration commonly recommended for these drugs, po- Financial Disclosure: Dr Fischer was employed by Bayer tential drug-drug interactions, and periodic develop- Canada from September 2003 to October 2004. Dr Fischer ment of other factors (such as hypovolemia) that might has no current financial interest, relationship, or affili- contribute to neurocardiogenic syncope. ation relevant to the subject matter of the article. Bayer Our study has potential limitations. First, all obser- does not manufacture any cholinesterase inhibitor men- vational studies are vulnerable to residual confounding tioned in this study or any competing dementia drug. and hidden bias. Although we used a variety of tech- Funding/Support: This study was funded by the Clini- niques to assess for these possibilities, we cannot rule out cal Teachers Association of Queen’s (CTAQ) Endow- residual confounding by unknown or unmeasured fac- ment Fund and a Chronic Disease New Emerging Team tors. Our ability to control for differences in the cohorts program grant (No. 54010) from the Canadian Insti- was limited to the variables recorded in the administra- tutes of Health Research (CIHR). The New Emerging tive databases. Nonetheless, observational studies that fo- Team program receives joint sponsorship from the Ca- cus on unrecognized adverse drug effects often provide nadian Diabetes Association, the Kidney Foundation of credible results.23,42 Second, we did not compare event Canada, the Heart and Stroke Foundation of Canada, and rates for individual cholinesterase inhibitors, but the 3 the CIHR Institutes of Nutrition, Metabolism and Dia- drugs we studied are thought to possess similar benefits betes, and Circulatory and Respiratory Health. Dr Gill and risks.2,50 Third, we did not examine for dose- is supported by an Ontario Ministry of Health and Long- response relationships, in part because these drugs are term Care Career Scientist Award. Dr Anderson is sup- started at low doses and only gradually increased to tar- ported by a Chair in Health Management Strategies from get doses. Fourth, the only fall-related injury we exam- the University of Toronto. Dr Bell is supported by a CIHR ined was hip fracture. Further research is required to de- New Investigator Award. termine associations between drug exposure and other Role of the Sponsors: The study sponsors had no role fall-related injuries including fractures at other sites, closed in the study design; data collection, analysis, or inter- head injuries, and motor vehicle crashes.51 Finally, we pretation; or writing of the report. excluded patients with a recent history of syncope to make Disclaimer: The opinions, results, and conclusions are it more likely that the drugs were the main contributor those of the authors. No endorsement by the Ontario Min-

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 9), MAY 11, 2009 WWW.ARCHINTERNMED.COM 872

Downloaded From: https://jamanetwork.com/ on 09/23/2021 istry of Health and Long-term Care, the Institute for Clini- work. The underrecognized burden of influenza in young children. N Engl J Med. 2006;355(1):31-40. cal Evaluative Sciences, or the funding sources is in- 26. Getchell WS, Larsen GC, Morris CD, McAnulty JH. Epidemiology of syncope in tended or should be inferred. hospitalized patients. J Gen Intern Med. 1999;14(11):677-687. Additional Contributions: Kathy Sykora, MSc, pro- 27. Grossman SA, Van Epp S, Arnold R, et al. The value of cardiac enzymes in el- vided thoughtful assistance with data analysis. derly patients presenting to the emergency department with syncope. J Geron- tol A Biol Sci Med Sci. 2003;58(11):1055-1058. 28. Essebag V, Hadjis T, Platt RW, Pilote L. Amiodarone and the risk of bradyar- REFERENCES rhythmia requiring permanent pacemaker in elderly patients with atrial fibrillation and prior myocardial infarction. J Am Coll Cardiol. 2003;41(2):249-254. 29. Juurlink D, Preyra C, Croxford R, et al. Canadian Institute for Health Information 1. Ferri CP, Prince M, Brayne C, et al; Alzheimer’s Disease International. Global preva- Discharge Abstract Database: A Validation Study. Toronto, Ontario, Canada: In- lence of dementia: a Delphi consensus study. Lancet. 2005;366(9503):2112- stitute for Clinical Evaluative Sciences; 2006. 2117. 30. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying 2. Birks J. Cholinesterase inhibitors for Alzheimer’s disease. Cochrane Database prognostic comorbidity in longitudinal studies: development and validation. Syst Rev. 2006;(1):CD005593. J Chronic Dis. 1987;40(5):373-383. 3. Morden NE, Zerzan JT, Larson EB. Alzheimer’s disease medication: use and cost 31. Stukel TA, Fisher ES, Wennberg DE, Alter DA, Gottlieb DJ, Vermeulen MJ. Analy- projections for Medicare Part D. J Am Geriatr Soc. 2007;55(4):622-624. sis of observational studies in the presence of treatment selection bias: effects 4. Courtney C, Farrell D, Gray R, et al; AD2000 Collaborative Group. Long-term do- of invasive cardiac management on AMI survival using propensity score and in- nepezil treatment in 565 patients with Alzheimer’s disease (AD2000): ran- strumental variable methods. JAMA. 2007;297(3):278-285. domised double-blind trial. Lancet. 2004;363(9427):2105-2115. 32. D’Agostino RB Jr. Propensity score methods for bias reduction in the compari- 5. Ames D, Kaduszkiewicz H, van den Bussche H, et al. For debate: is the evidence son of a treatment to a non-randomized control group. Stat Med. 1998;17(19): for the efficacy of cholinesterase inhibitors in the symptomatic treatment of Alz- 2265-2281. heimer’s disease convincing or not? Int Psychogeriatr. 2008;20(2):259-292. 33. Austin PC, Grootendorst P, Anderson GM. A comparison of the ability of differ- 6. Qaseem A, Snow V, Cross JT Jr, et al; American College of Physicians/American ent propensity score models to balance measured variables between treated and Academy of Family Physicians Panel on Dementia. Current pharmacologic treat- untreated subjects: a Monte Carlo study. Stat Med. 2007;26(4):734-753. ment of dementia: a clinical practice guideline from the American College of Phy- 34. Brookhart MA, Schneeweiss S, Rothman KJ, Glynn RJ, Avorn J, Sturmer T. sicians and the American Academy of Family Physicians. Ann Intern Med. 2008; Variable selection for propensity score models. Am J Epidemiol. 2006;163 148(5):370-378. (12):1149-1156. 7. Kapoor WN. Syncope. N Engl J Med. 2000;343(25):1856-1862. 35. Rosenbaum PR, Rubin DB. Constructing a control group using multivariate matched 8. Chen LY, Gersh BJ, Hodge DO, Wieling W, Hammill SC, Shen WK. Prevalence sampling methods that incorporate the propensity score. Am Stat. 1985;39 and clinical outcomes of patients with multiple potential causes of syncope. Mayo (1):33-38. Clin Proc. 2003;78(4):414-420. 36. Creditor MC. Hazards of hospitalization of the elderly. Ann Intern Med. 1993;118 9. Krahn AD, Klein GJ, Yee R, Manda V. The high cost of syncope: cost implica- (3):219-223. tions of a new insertable loop recorder in the investigation of recurrent syncope. 37. Winblad B, Kilander L, Eriksson S, et al; Severe Alzheimer’s Disease Study Group. Am Heart J. 1999;137(5):870-877. Donepezil in patients with severe Alzheimer’s disease: double-blind, parallel- 10. Masuda Y. Cardiac effect of cholinesterase inhibitors used in Alzheimer’s dis- group, placebo-controlled study. Lancet. 2006;367(9516):1057-1065. ease: from basic research to bedside. Curr Alzheimer Res. 2004;1(4):315-321. 38. Erkinjuntti T, Kurz A, Gauthier S, Bullock R, Lilienfeld S, Damaraju CV. Efficacy 11. Grubb BP. Neurocardiogenic syncope and related disorders of orthostatic of galantamine in probable vascular dementia and Alzheimer’s disease com- intolerance. Circulation. 2005;111(22):2997-3006. bined with cerebrovascular disease: a randomised trial. Lancet. 2002;359(9314): 12. Mosqueda-Garcia R, Furlan R, Tank J, Fernandez-Violante R. The elusive patho- 1283-1290. physiology of neurally mediated syncope. Circulation. 2000;102(23):2898-2906. 39. Emre M, Aarsland D, Albanese A, et al. Rivastigmine for dementia associated with 13. Brignole M, Alboni P, Benditt DG, et al; Task Force on Syncope, European So- Parkinson’s disease. N Engl J Med. 2004;351(24):2509-2518. ciety of Cardiology. Guidelines on management (diagnosis and treatment) of syn- 40. Raschetti R, Albanese E, Vanacore N, Maggini M. Cholinesterase inhibitors in cope: update 2004. Europace. 2004;6(6):467-537. mild cognitive impairment: a systematic review of randomised trials. PLoS Med. 14. Strickberger SA, Benson DW, Biaggioni I, et al; American Heart Association Coun- 2007;4(11):e338. cils on Clinical Cardiology, Cardiovascular Nursing, Cardiovascular Disease in 41. Schneider LS. Treatment of Alzheimer’s disease with cholinesterase inhibitors. the Young, and Stroke; Quality of Care and Outcomes Research Interdiscipli- Clin Geriatr Med. 2001;17(2):337-358. nary Working Group; American College of Cardiology Foundation; Heart Rhythm 42. Vandenbroucke JP, Psaty BM. Benefits and risks of drug treatments: how to com- Society; American Autonomic Society. AHA/ACCF Scientific Statement on the evalu- bine the best evidence on benefits with the best data about adverse effects. JAMA. ation of syncope: from the American Heart Association Councils on Clinical Car- 2008;300(20):2417-2419. diology, Cardiovascular Nursing, Cardiovascular Disease in the Young, and Stroke, 43. Howard RJ, Juszczak E, Ballard CG, et al; CALM-AD Trial Group. Donepezil for and the Quality of Care and Outcomes Research Interdisciplinary Working Group; the treatment of agitation in Alzheimer’s disease. N Engl J Med. 2007;357(14): and the American College of Cardiology Foundation: in collaboration with the Heart 1382-1392. Rhythm Society: endorsed by the American Autonomic Society. Circulation. 2006; 44. Suleyman T, Tevfik P, Abdulkadir G, Ozlem S. Complete atrioventricular block 113(2):316-327. and ventricular tachyarrhythmia associated with donepezil. Emerg Med J. 2006; 15. Rubenstein LZ, Josephson KR. The epidemiology of falls and syncope. Clin Geri- 23(8):641-642. atr Med. 2002;18(2):141-158. 45. Malone DM, Lindesay J. Cholinesterase inhibitors and cardiovascular disease: a 16. Nightingale S, Holmes J, Mason J, House A. Psychiatric illness and mortality af- survey of old age psychiatrists’ practice. Age Ageing. 2007;36(3):331-333. ter hip fracture. Lancet. 2001;357(9264):1264-1265. 46. Hashimoto M, Imamura T, Tanimukai S, Kazui H, Mori E. Urinary incontinence: 17. Gill SS, Bronskill SE, Mamdani M, et al. Representation of patients with demen- an unrecognised adverse effect with donepezil. Lancet. 2000;356(9229):568. tia in clinical trials of donepezil. Can J Clin Pharmacol. 2004;11(2):e274-e285. 47. Gill SS, Mamdani M, Naglie G, et al. A prescribing cascade involving cholines- 18. Bordier P, Garrigue S, Lanusse S, et al. Cardiovascular effects and risk of syn- terase inhibitors and anticholinergic drugs. Arch Intern Med. 2005;165(7): cope related to donepezil in patients with Alzheimer’s disease. CNS Drugs. 2006; 808-813. 20(5):411-417. 48. Kenny RA, Kalaria R, Ballard C. Neurocardiovascular instability in cognitive im- 19. Bordier P, Lanusse S, Garrigue S, et al. Causes of syncope in patients with Alz- pairment and dementia. Ann N Y Acad Sci. 2002;977:183-195. heimer’s disease treated with donepezil. Drugs Aging. 2005;22(8):687-694. 49. Lindstrom HA, Smyth KA, Sami SA, et al. Medication use to treat memory loss 20. Levy AR, O’Brien BJ, Sellors C, Grootendorst P, Willison D. Coding accuracy of in dementia: perspectives of persons with dementia and their caregivers. Dementia. administrative drug claims in the Ontario Drug Benefit database. Can J Clin 2006;5(1):27-50. Pharmacol. 2003;10(2):67-71. 50. Hogan DB, Goldlist B, Naglie G, Patterson C. Comparison studies of cholines- 21. Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. terase inhibitors for Alzheimer’s disease. Lancet Neurol. 2004;3(10):622-626. Am J Epidemiol. 2003;158(9):915-920. 51. Kannus P, Sievanen H, Palvanen M, Jarvinen T, Parkkari J. Prevention of falls 22. Psaty BM, Koepsell TD, Lin D, et al. Assessment and control for confounding by and consequent injuries in elderly people. Lancet. 2005;366(9500):1885-1893. indication in observational studies. J Am Geriatr Soc. 1999;47(6):749-754. 52. Lee PE, Fischer HD, Rochon PA, et al. Published randomized controlled trials of 23. Vandenbroucke JP. When are observational studies as credible as randomised drug therapy for dementia often lack complete data on harm [published online trials? Lancet. 2004;363(9422):1728-1731. July 9, 2008]. J Clin Epidemiol. doi:10.1016/j.jclinepi.2007.09.012. 24. Taylor DH Jr, Fillenbaum GG, Ezell ME. The accuracy of Medicare claims data in 53. Ioannidis JP, Evans SJ, Gotzsche PC, et al; CONSORT Group. Better reporting of identifying Alzheimer’s disease. J Clin Epidemiol. 2002;55(9):929-937. harms in randomized trials: an extension of the CONSORT statement. Ann In- 25. Poehling KA, Edwards KM, Weinberg GA, et al; New Vaccine Surveillance Net- tern Med. 2004;141(10):781-788.

(REPRINTED) ARCH INTERN MED/ VOL 169 (NO. 9), MAY 11, 2009 WWW.ARCHINTERNMED.COM 873

Downloaded From: https://jamanetwork.com/ on 09/23/2021