Risk of Mortality Associated with Antipsychotic Monotherapy and Polypharmacy Among Community-Dwelling Persons with Alzheimer’s Disease

Abstract

We aimed to analyze the risk of non-cancer mortality according to duration of antipsychotic use and to compare the risk associated with polypharmacy and monotherapy among community-dwellers with Alzheimer’s disease (AD). The risk of mortality between most frequently used antipsychotic drugs was compared. Data from a nationwide register-based MEDALZ study that included all 70,718 community-dwellers newly diagnosed with AD during 2005–2011 in Finland was utilized. Death, excluding cancer as direct cause of death, was extracted from Causes of Death Register. Incident antipsychotic use was compared with time without antipsychotics with Cox proportional hazard models. Antipsychotic use was associated with an increased risk of mortality (adjusted hazard ratio [aHR] 1.61; 95% Confidence Interval [CI] 1.53–1.70). The absolute difference in mortality rate was 4.58 (95% CI 4.53–4.63) deaths per 100 person-years. The risk of mortality was increased from the first days of use and attenuated gradually but remained increased even after two years of use (aHR 1.30; 95% CI 1.16–1.46). Compared with nonuse, antipsychotic polypharmacy (aHR 2.88; 95% CI 2.38–3.49) was associated with an increased risk of mortality than monotherapy (aHR 1.57; 95% CI 1.49–1.66). Haloperidol was associated with higher risk of mortality (aHR 1.52; 95% CI 1.14–2.02) and quetiapine with lower risk (aHR 0.84; 95% CI 0.75–0.94) compared with risperidone. In conclusion, the findings support current treatment guidelines on having a high threshold for antipsychotic initiation among persons with AD. Antipsychotic polypharmacy and long-term use should be avoided and drug choice should be weighed against risk/benefit evidence.

Keywords

Alzheimer’s disease antipsychotics mortality pharmacoepidemiology

INTRODUCTION

The first warnings on increased risk of mortality among atypical antipsychotic users with dementia were issued in 2005 based on data from 17 placebo-controlled trials [1]. In 2008, the warnings were expanded to cover typical antipsychotics [2, 3] after observational studies also indicated that typical antipsychotics were associated with an increased risk of mortality [4, 5]. Since then, several other cohort studies with new user design have reported higher mortality among typical than atypical antipsychotic users [6 –11]. A recent systematic review focused on studies conducted among persons with Alzheimer’s disease (AD) reported an increased risk of mortality associated with antipsychotic use [12]. In addition, a recent nationwide population-based cohort study found that cumulative dosages of antipsychotic drugs were associated with increased mortality rates in Danish patients with AD [13]. Although many studies have focused on group differences, there is evidence that the mortality risk might be different between individual antipsychotics [5 , 14–19]. Comparisons between individual drugs are important for guiding the treatment practices and drug choices.

Most of the previous cohort studies have focused on the first 180 days after antipsychotic initiation [4–9 , 21]. However, antipsychotics are frequently used for longer time among community-dwellers with AD [22], despite recommendations of time-limited use [23 –26]. Only one large cohort study reported the risk of long-term mortality over 360 days after first dispensing of antipsychotic drug among persons with dementia but it was not able to account for discontinuation of use [17]. Thus, there is a need for studies assessing the safety of long-term antipsychotic use. Although antipsychotics had small benefits on aggression and psychosis over 6–12 weeks in clinical trials [27 –29], there is less evidence of the long-term efficacy [30, 31].

Despite safety concerns and modest efficacy, antipsychotic use has remained frequent among persons with dementia. Although prevalence of antipsychotic use has declined in some countries [32 –35], this trend has not been observed everywhere [36, 37]. Even antipsychotic polypharmacy (use of ≥2 antipsychotics concomitantly) has been observed among community-dwellers with AD [38], although there is no evidence of efficacy and safety of antipsychotic polypharmacy in the treatment of behavioral and psychological symptoms of dementia (BPSD). There is a place for antipsychotic use in the treatment of most severe BPSD including severe aggression, psychosis, and agitation when nonpharmacological treatment options have failed or the symptoms cause serious risk of harm to the patient or others [23 –26]. However, antipsychotic polypharmacy is notrecommended.

The objective of our study was to analyze the risk of non-cancer mortality according to duration of antipsychotic use and to compare the risk associated with polypharmacy and monotherapy among community-dwellers with AD. We also studied whether the risk of mortality differed between the most frequently used antipsychotics.

MATERIALS AND METHODS

Study cohort

The Finnish nationwide register-based MEDALZ (Medication and Alzheimer’s disease) cohort study includes all community-dwellers with newly diagnosed AD during 2005–2011 (n = 70,718). Persons with AD were identified from the Special Reimbursement Register maintained by the Social Insurance Institution of Finland (SII). This register contains persons entitled to higher reimbursement of certain drugs due to chronic diseases such as AD [39]. The SII grants reimbursement for anti-dementia drugs if predefined diagnostic criteria, based on the NINCDS-ADRDA [40] and DSM-IV criteria for AD, are met. These include exclusion of alternative diagnoses, computed tomography or magnetic resonance imaging scan, symptoms consistent with AD, and confirmation of diagnosis by geriatrician or neurologist. Since 2007, persons with dementia related to Parkinson’s disease have been entitled to receive special reimbursement but these persons were excluded from this study.

Antipsychotic drug exposure

Data on reimbursed prescription drug purchases during 1995–2012 were extracted from the Finnish Prescription Register. The purchased amount is summarized in number of Defined Daily Doses (DDDs), which is the assumed average maintenance dose per day for a drug used for its main indication in adults [41]. Drugs are classified according to Anatomical Therapeutic Chemical (ATC) classification system [42]. In this study, antipsychotics were defined as ATC class N05A excluding lithium (N05AN01) and prochlorperazine (N05AB04).

Drug exposure periods were modeled from the drug purchases with a previously used PRE2DUP method described in detail by Tanskanen et al. [43]. Briefly, this mathematical modeling method calculates sliding averages of daily dose (in DDDs) for each person and each drug from individual purchase histories. The method joins purchases to the same continuous drug use period if the purchased amount is enough to last to the next purchase with personal temporal daily dose taking into account stockpiling of drugs, personal purchase regularity, and stays at hospital or nursing home care when drug use is not recorded in the Prescription Register. For the comparison of mortality risk between antipsychotic use and nonuse, drug use periods of individual antipsychotic drugs were combined to derive the use of “any antipsychotics”. Thus, persons were allowed to change from one antipsychotic drug to another during “any antipsychotic” use. For the monotherapy/polypharmacy comparisons, concomitant use of ≥2 antipsychotics (polypharmacy) was identified by combining overlapping drug use periods of individual antipsychotics.

Outcome

The date and cause of death for each person was received from the Statistics Finland. The main outcome was non-cancer mortality. Cancer deaths (direct cause ICD-10 codes C00-C97) were not considered as they are unlikely to be caused by antipsychotic use and their inclusion could overestimate the mortality related to antipsychotic use if antipsychotics were used for cancer-related nausea treatment. Persons were censored for cancer death.

Covariates

Data on comorbidities (1972–2012) were extracted from the Special Reimbursement Register and the Hospital Discharge Register. Diagnoses recorded any time since start of registers until the date of cohort entry were considered. In the comparisons of antipsychotic use and nonuse, the cohort entry was the date of AD diagnosis and in the drug-drug comparisons cohort entry was the date of first initiation of antipsychotic use. Persons with history of stroke (ICD-10 codes I60-I64), hip fracture (S72.0–S72.2) and ischemic cardiac events (I20–I25) or revascularization procedures by bypass or angioplasty (NOMESCO Classification of Surgical Procedures codes FNA*, FNC*, FNE*, FNG00, FNG10, FN1AT, FN1BT, FN1YT) before the follow-up were identified from the Hospital Discharge Register. Modified baseline Charlson Comorbidity Index score (CCI) [44] was computed on the basis of comorbidities extracted from the Special Reimbursement Register. Chronic asthma or obstructive pulmonary disease, coronary artery disease, heart failure, diabetes, rheumatoid arthritis, and disseminated connective tissue disease each received a score of 1. Uremia requiring dialysis, severe anemia in connection with chronic renal failure, leukemia and other malignant diseases of blood and bone marrow, malignant neoplasms, and gynecological, breast, and prostate cancers received a score of 2. Persons with history of schizophrenia, schizotypal, or delusional disorders (ICD-10 codes F20-29), or bipolar disorder (F30-31) were excluded as focus of this study was the risks of antipsychotic use in the treatment of BPSD. For psychiatric disorders, data from the beginning of Hospital Discharge register (1972) until 5 years before the diagnosis of AD were considered. The five-year gap before the AD diagnoses was inserted as, according to our unpublished data, sometimes persons with prodromal symptoms of AD are mistakenly diagnosed as suffering from psychiatric disorders. According to the Finnish legislation, no ethics committee approval was required as de-identified register-based data was used.

Statistical analyses

In the primary analyses, mortality during antipsychotic use was compared with time without antipsychotics with antipsychotic use as a time-dependent variable. To restrict analyses to incident users and avoid prevalent user bias, one-year washout period was applied preceding the cohort entry, i.e., the diagnosis of AD and persons using antipsychotics during this period were excluded from all analyses (n = 7,632). The follow-up was censored after the first discontinuation of use. As the Prescription Register does not include drugs used in hospitals or nursing homes, we excluded persons who were hospitalized or institutionalized for ≥6 months during the washout period or had an ongoing hospital stay of ≥3 months at the end of washout period, or were hospitalized the entire follow-up. Furthermore, persons with hospital-based diagnoses of cancer or antineoplastic drug use during one year before the start of follow-up were excluded from analyses. The formation of the final study sample (n = 57,755) is presented in Fig. 1. The follow-up began on the date of AD diagnosis. Person-time was accounted for nonusers until or if person initiated antipsychotic use, after which person-time was accounted for antipsychotic use, until death or censoring event whichever occurred first. The follow-up was censored at the first occurrence of the following: death, long-term institutionalization/hospitalization, discontinuation of use, and the end of study period (December 31, 2012). In addition, in the analyses of antipsychotic monotherapy and polypharmacy compared with time without antipsychotics, person time was classified time-dependently as antipsychotic monotherapy when only one antipsychotic drug was used and as polypharmacy during time periods when more than two antipsychotics were used concomitantly. Hazard ratios (HRs) with 95% confidence intervals (CI) were calculated with Cox proportional hazards model. Analyses were adjusted for baseline variables including age, sex, CCI score, and history of serious events including stroke, hip fracture, and ischemic cardiac events. In addition, analyses were adjusted for time-dependent covariates during the entire follow-up including use of benzodiazepines and related drugs (N05BA, N05CF, N05CD), antidepressants (N06A) and antidepressants in combination with psycholeptics (N06CA), and opioids (N02A). Similar results were obtained in sensitivity analyses that accounted for baseline use of these drugs instead of time-dependence. To analyze whether the risk of mortality varied by duration of exposure, the duration of antipsychotic use was categorized time-dependently as ≤30 days, 31–90 days, 91–180 days, 181–365 days, 366–730 days, and >730 days. To better control for the impact of AD severity and progression of the disease in the analyses, we conducted additional analyses by matching two nonusers to each antipsychotic user for the start date by incidence density sampling (exposure matched cohort). The matching criteria for nonuser-controls were time since AD diagnoses (≤90 days), age (±2 years), and gender. In this design, antipsychotic users were compared with nonusers. The duration of antipsychotic use was categorized time-dependently as in the main analysis and was compared with nonuse. Similarly, use of antipsychotics in monotherapy or polypharmacy was included as time-dependent variable and was compared with nonusers.

In drug-drug comparisons (user only-design), mortality risk was compared between most frequently used antipsychotics. The follow-up started from the first initiation of antipsychotic use after the diagnosis of AD. The washout period was one year preceding the antipsychotic initiation and persons using antipsychotic during the washout period were excluded from analyses. In addition to the exclusion criteria of the primary analyses, persons who initiated antipsychotic use with polypharmacy were excluded to compare the mortality risk between drugs. The formation of the study sample (n = 17,731) for drug-drug comparisons is summarized in Fig. 2. Analyses were restricted to the three most frequently used antipsychotics, risperidone, quetiapine, and haloperidol, due to the low number of users of other antipsychotic drugs (Supplementary Table 1). In drug-drug comparisons, the follow-up was censored if the user switched to a different antipsychotic drug, started antipsychotic polypharmacy, or discontinued use. In addition, the follow-up was censored at the date of death, start of long-term institutionalization/hospitalization, after 1,000 days of antipsychotic use, or the end of study period; whichever occurred first. The follow-up time was restricted to the first 1,000 days of use to account for data sparsity. Risperidone was used as the reference group because it was the most frequently used antipsychotic drug and it has been commonly used as a reference group in previous studies [5 , 14–19]. Analyses were adjusted for age, sex, time since AD diagnosis, CCI score, history of stroke, hip fracture, and ischemic cardiac event, and use of benzodiazepines, antidepressants, and opioids at the time of antipsychotic initiation. To take account for differences between users of different antipsychotics at baseline, we conducted a sensitivity analysis by weighting the analysis comparing quetiapine and haloperidol use to risperidone use with stabilized inverse probability of treatment weights (IPTW) based on propensity scores. First, we derived a propensity score by logistic regression to predict the probability of antipsychotic choice at baseline. The propensity score included the following factors: age, sex, occupational social class (according to classification by Statistics Finland in population census), time since AD diagnoses until start of antipsychotic use, diabetes, asthma/COPD, cardiovascular disease, ischemic heart disease, rheumatoid arthritis and corresponding conditions, history of serious events including stroke and hip fracture, history of substance abuse, any mental or behavioral disorder, history of delirium, metastatic cancer, hemiplegia, history of alcohol abuse, any tumor, any chronic pulmonary disease, anemia, fluid and electrolyte disorders, liver disease, peripheral vascular disorder, pulmonary circulation disorder, use of the following drugs ever before the baseline: antidepressants, BZDRs, opioids, antiepileptics; and the sum of the following drugs at the start of antipsychotic use (classified as 0, 1, 2, 3, 4, ≥5): antidepressants, BZDRs, opioids, antiepileptics. Second, we stabilized the inverse probability with the prevalence of antipsychotic use at the baseline. Covariate balance after IPTW weighting was checked using standardized difference. In other sensitivity analyses, drug-drug comparisons were further restricted to the first 180 days after the start of antipsychotic use and two analyses were conducted: as-treated and intention to treat (ITT). In ITT analyses, the analyses were not censored to any changes in antipsychotic use or hospitalizations.

In addition, the dose-dependent effects were analyzed. The dose per day represents average dose from the entire antipsychotic use period. Risperidone doses per day were categorized into ≤0.5 mg (0.1 DDD) (reference category) and >0.5 mg, quetiapine doses per day into ≤50 mg (0.125 DDD) and >50 mg, and haloperidol doses per day into ≤1 mg (0.125 DDD) and >1 mg. All analyses were performed with SAS (Version 9.3; SAS Institute Inc., Cary, NC, USA).

RESULTS

Median follow-up time was 2.4 years (interquartile range; IQR 1.3–3.9). During the follow-up, 27.4% (15,806/57,755) persons with AD initiated antipsychotic use and 14.1% (n = 2,230) of antipsychotic initiators used ≥2 antipsychotics concomitantly at some point. Compared with non-users, antipsychotic initiators were younger, and more likely to use benzodiazepines and antidepressants at the date of AD diagnosis (Table 1). They had fewer comorbidities, were less likely to have history of stroke, hip fracture, and ischemic cardiac events, and were less likely to use opioids. Those with antipsychotic polypharmacy were more likely to be younger and male than monotherapy users. They also had fewer comorbidities, were less likely to have a history of hip fracture, and used opioids less frequently.

Antipsychotic use was associated with an increased risk of mortality (adjusted HR 1.61; 95% CI 1.53–1.70) (Table 2). The absolute difference in mortality rate was 4.58 (95% CI 4.53–4.63) deaths per 100 person-years. The risk of mortality was increased from the first days of use and attenuated gradually but remained increased even after two years of use (adjusted HR 1.30; 95% CI 1.16–1.46).

Compared with nonuse, antipsychotic polypharmacy (adjusted HR 2.88; 95% CI 2.38–3.49) was associated with an increased risk of mortality compared with monotherapy (adjusted HR 1.57; 95% CI 1.49–1.66) (Table 2). The absolute difference in mortality rates between antipsychotic polypharmacy and monotherapy was 7.49 (95% CI 7.16–7.83) deaths per 100 person-years. Median time on antipsychotic monotherapy was 205 days (IQR 80–495) and 78 days (IQR 39–124) on polypharmacy. The most frequently used combination was quetiapine and risperidone (54.4%) followed by haloperidol and risperidone (8.6%), haloperidol and quetiapine (7.7%), olanzapine and risperidone (6.3%) and olanzapine and quetiapine (5.2%). The results comparing antipsychotic users to matched nonusers showed similar results although HRs somewhat attenuated (Supplementary Table 2).

Drug-drug comparisons included 11,144 new risperidone users, 5,186 quetiapine users, and 612 haloperidol users. Median follow-up time was 179 (IQR 69–461) days. Median use time was 172 days (IQR 61–428) for risperidone, 234 days (IQR 99–563) for quetiapine and 116 days (IQR 68–204) for haloperidol. Haloperidol and quetiapine users were more likely to be male and use benzodiazepines and opioids at the time of antipsychotic initiation than risperidone users (Table 1). Haloperidol users were more likely to have a history of stroke and quetiapine users were more likely to be younger, users of antidepressants and have a history of ischemic cardiac events compared with risperidone users. Haloperidol was associated with higher risk of mortality (adjusted HR 1.52; 95% CI 1.14–2.02) and quetiapine with lower risk (adjusted HR 0.84; 95% CI 0.75–0.94) compared with risperidone (Table 3). IPTW weighted analyses resulted in similar results. In as treated and ITT analyses restricted to the first 180 days no major changes were noticed. Median doses per day were 0.7 (IQR 0.5–1.0) mg for risperidone, 33.0 (IQR 22.9–54.2) mg for quetiapine and 1.0 (IQR 0.8–1.6) mg for haloperidol. Compared with low-dose (≤0.5 mg per day) risperidone use, higher risperidone doses per day (>0.5 mg) (adjusted HR 1.57; 95% CI 1.36–1.81) and both haloperidol doses per day, ≤1 mg and >1 mg (adjusted HR 1.66; 95% CI 1.10–2.50 and 2.55; 95% CI 1.70–3.85, respectively), were associated with an increased risk of mortality. Quetiapine dose ≤50 mg per day was associated with slightly higher mortality (adjusted HR 1.18; 95% CI 1.00–1.38) while the risk among higher dose quetiapine users (>50 mg per day) was similar (adjusted HR 0.97; 95% CI 0.78–1.21) with low-dose risperidone users.

DISCUSSION

Mortality risk was increased shortly after the antipsychotic initiation and remained increased even after two years of use among community-dwellers with AD. Few previous studies have reported the risk of mortality for long-term antipsychotic use [17, 45]. In a placebo-controlled withdrawal trial, those who were randomized to continue antipsychotic use for 12 months had increased risk of mortality at 12 months compared with those who received placebo, and the difference was more pronounced after the first year [45]. In a cohort study, persons with dementia using antipsychotics had twice as high risk of mortality from the first dispensing of an antipsychotic drug compared with those using solely other psychotropics and the risk remained consistently higher for over 6 years [17]. Our results provide additional evidence that the mortality risk remains elevated in long-term use and confirm the need to limit duration of antipsychotic use and attempt withdrawal regularly according to recommendations of current treatment guidelines [23 –26].

Compared with nonuse, antipsychotic polypharmacy was associated with 2.9 times higher risk of mortality and monotherapy with 1.6 times higher risk in community-dwellers with AD. Similarly, in a recent population-based cohort study, antipsychotic polypharmacy was associated with higher risk of major adverse cardiovascular events and noncardiovascular mortality compared with risperidone monotherapy among new antipsychotic users aged ≥70 years [18]. There is no evidence of effectiveness of antipsychotic polypharmacy in the treatment of BPSD. Current treatment guidelines recommend antipsychotics for the most severe BPSD including severe psychosis, aggression, and agitation if symptoms cause serious risk of harm to the patient or others [23 –26]. Thus overall, there should be a high threshold for initiating antipsychotic use and there is no rational ground for antipsychotic polypharmacy in the treatment of BPSD. Our results demonstrate that antipsychotic polypharmacy is not safe among persons with AD. The register-based data did not include indications for use. Thus, it is not known whether two or more antipsychotics were used concomitantly to treat the same or different symptoms. The most frequent combination in our study was quetiapine and risperidone. Risperidone is the only antipsychotic with an approved indication for the treatment of BPSD in Finland. On the other hand, quetiapine is frequently used to treat insomnia [46]. If severe aggression, agitation, or psychotic symptoms develop and a trial of risperidone is deemed necessary, possible use of quetiapine for the treatment of insomnia should be reviewed and discontinued to avoid excess mortality risk.

In our study, haloperidol was associated with a 52% increase in mortality risk and quetiapine with a 16% decrease compared with risperidone use among community-dwellers with AD. Our findings are consistent with studies reporting higher mortality for haloperidol [5 , 14–19] and lower mortality for quetiapine [14–16 , 19] compared with risperidone users with and without dementia living in community or nursing homes. Previous studies have indicated that these differences between haloperidol, quetiapine, and risperidone persist after dose adjustment [14 –16]. Further, a dose-response in mortality risk has been shown for risperidone and haloperidol [14 , 47]. Accordingly compared with low-dose risperidone use in the present study, haloperidol use with dose ≤1 mg per day was associated with 1.7 times higher risk of mortality and use with dose >1.0 mg per day was associated with 2.6 times higher risk. In addition, use of higher risperidone doses per day (>0.5 mg) were associated with 1.6 times higher risk than low-dose risperidone use. Although we found that in general quetiapine was associated with lower mortality compared with risperidone, when comparing quetiapine use with low-dose risperidone use, risk of mortality was slightly higher with quetiapine doses ≤50 mg per day but similar with quetiapine doses >50 mg per day. This might be due to misclassification: dose estimates are more reliable for those who survive longer as average dose calculus is based on more than two purchases. Doses of single purchases were based on the most frequently used dose per day for the purchased package in the study cohort and thus, were more likely to be categorized in the low-dose category due to frequent use of low doses in the cohort. This limitation applies to the dose analyses in general and thus, the results should be interpreted with caution. It is more likely that quetiapine is associated with similar risk of mortality as low-dose risperidone. None of the previous studies have observed a dose-response in mortality for quetiapine [14 , 47]. Gerhard et al. [16] discussed that this might be due to less variation in clinical dosing of quetiapine. In our study, approximately three quarters used quetiapine with dose ≤50 mg per day similarly as in the previous studies [14, 16].

The difference in the mortality risk between antipsychotics could be explained by confounding by indication. Haloperidol is used as a standard treatment of delirium, and delirium itself is associated with an increased risk of mortality [48]. On the other hand, quetiapine is frequently used with low doses in the treatment of insomnia [46]. Thus, it is possible that haloperidol and risperidone were more frequently used to treat delirium and more severe BPSD, which could contribute to the observed higher mortality risks. The possibility of confounding by indication could not be excluded as the register-based data did not include indications for use or information on severity of BPSD or AD. However, some previous studies were able to adjust for delirium to some extent and still found an increased risk of mortality with haloperidol use compared with risperidone [14 –16] or use of atypical antipsychotics [49]. Based on accumulated evidence, it seems reasonable to try to avoid haloperidol use among older persons. Despite the lower mortality risk, quetiapine use cannot be considered safe. Maust et al. recently reported number needed to harm (NNH) of 50 (95% CI 30–150) for quetiapine, 26 (95% CI 15–99) for haloperidol, and 27 (95% CI 19–46) for risperidone when compared with matched nonusers during 180 days of follow-up [21]. In addition, there is a lack of evidence of efficacy of quetiapine use in the treatment of BPSD [29] limiting the potential benefit of somewhat lower mortality risk. On the other hand, there is evidence of modest efficacy of risperidone, olanzapine, and aripiprazole in the treatment of aggression and psychosis [28, 29]. Due to a low number of aripiprazole (n = 11) and olanzapine (n = 257) users, we were unable to study their mortality risk compared to risperidone. Four previous studies have not found difference in mortality risk between olanzapine and risperidone [5 , 18]. However, Schmedt et al. found lower risk for olanzapine as well as Gerhard et al. after adjusting for dose [16, 19]. More comparative studies on efficacy and adverse events of individual antipsychotics are needed.

Strengths and limitations

A major strength of our study is the large nationwide data of community-dwellers with clinically verified AD. This allowed us to assess the mortality risk during antipsychotic polypharmacy and monotherapy and to compare most frequently used antipsychotics. The long follow-up time enabled us to study the mortality risk according to duration of antipsychotic use. We restricted the analyses to new users to avoid prevalent user bias as prevalent users have survived from the early period of antipsychotic use. Our results are generalizable to community-dwellers with AD. However, previous studies have reported similar results regarding drug-drug comparisons among older persons with and without dementia living in community and nursing homes[5 , 14–19].

The Finnish Prescription Register has previously been shown to be a valid data source on antipsychotic exposure among older persons [50]. A general limitation of the register is lack of information on prescribed dose. However, we were able to model the duration and dose of antipsychotic use [43]. The dose per day represents the average dose during the entire antipsychotic use period. It does not necessarily reflect dose at a certain time point such as dose at the time of death, due to the possible temporal changes in dosage. The dose of a single purchase reflects the most typical dose per day for that particular drug package which might have caused misclassification of dose categories. The results regarding dose should be interpreted with these limitations in mind.

We excluded persons with history of schizophrenia and bipolar disorder to ensure that antipsychotic use was most likely to be due to BPSD. However, as the registers did not include data on indication and severity of BPSD, we were unable to control for disease severity. Drug-drug comparisons were adjusted for time from AD diagnosis to initiation of antipsychotic use to account for the possible differences in severity of AD. In addition, adjusting for benzodiazepine and antidepressant use may have partly controlled for the BPSD and AD severity. Further, we conducted additional analyses by matching antipsychotic users to nonusers on the basis of age, sex, and disease duration, indicated by time since AD diagnosis. In these matched analyses, the results did not change. However, confounding by indication cannot be ruled out as users have an indication for antipsychotic use whereas nonusers may not have.

A limitation in register-based data is that it does not indicate whether two antipsychotics are used concomitantly or switched from one drug to another. Thus, there might be some misclassification in person-time between antipsychotic monotherapy and polypharmacy as some of the time classified as antipsychotic polypharmacy might constitute of drug switches. However, if polypharmacy would have been defined as overlapping use of at least, e.g., 60 days, this would have caused immortal time bias, i.e., to be classified as a concomitant user of two or more antipsychotics person must survive the first 60 days of concomitant use biasing the results. However, we cannot rule out the possibility of residual confounding, i.e., that persons on polypharmacy have more severe symptoms that contribute to higher mortality risk.

CONCLUSIONS

Our findings support current treatment guidelines on having a high threshold for antipsychotic initiation among persons with AD. If antipsychotic use is deemed necessary, the lowest effective dose should be used and duration should be limited with regular attempts of withdrawal, as the mortality risk remains elevated in long-term use. Antipsychotic polypharmacy should be avoided as the risk of mortality was higher than in monotherapy. The antipsychotic drug choice should be weighed against risk/benefit evidence. More research is needed on the comparative efficacy and the risk of adverse events of individual antipsychotics.

DISCLOSURE STATEMENT

Authors’ disclosures available online (http://j-alz.com/manuscript-disclosures/16-0671r1).

Footnotes

Appendix

References

U.S. Food and Drug Administration (FDA), Public health advisory: Deaths with antipsychotics in elderly patients with behavioral disturbances. FDA ALERT [4/11/2005], http://www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/ucm053171.htm, Last updated August 16, 2013, Accessed on July 7, 2015.

U.S. Food and Drug Administration (FDA), Information for healthcare professionals. Conventional antipsychotics. FDA ALERT [6/16/2008], http://www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/ucm124830.htm, Last updated August 15, 2013, Accessed on July 7, 2015.

European Medicines Agency (EMEA). CHMP Assessment report on conventional antipsychotics. Procedure under Article 5(3) of Regulation (EC) No 726/2004. EMEA/CHMP/590557/2008, http://www.ema.europa.eu/docs/en_GB/document_library/Report/2010/01/WC500054057.pdf, Last updated November 20, 2008, Accessed on July 7, 2015.

Gill

, Bronskill

, Normand

, Anderson

, Sykora

, Lam

, Bell

, Lee

, Fischer

, Herrmann

, Gurwitz

, Rochon

(2007) Antipsychotic drug use and mortality in older adults with dementia. Ann Intern Med 146, 775–786.

Schneeweiss

, Setoguchi

, Brookhart

, Dormuth

, Wang

(2007) Risk of death associated with the use of conventional versus atypical antipsychotic drugs among elderly patients. CMAJ 176, 627–632.

Setoguchi

, Wang

, Brookhart

, Canning

, Kaci

, Schneeweiss

(2008) Potential causes of higher mortality in elderly users of conventional and atypical antipsychotic medications. J Am Geriatr Soc 56, 1644–1650.

Liperoti

, Onder

, Landi

, Lapane

, Mor

, Bernabei

, Gambassi

(2009) All-cause mortality associated with atypical and conventional antipsychotics among nursing home residents with dementia: A retrospective cohort study. J Clin Psychiatry 70, 1340–1347.

Huybrechts

, Rothman

, Silliman

, Brookhart

, Schneeweiss

(2011) Risk of death and hospital admission for major medical events after initiation of psychotropic medications in older adults admitted to nursing homes. CMAJ 183, e411–e419.

Aparasu

, Chatterjee

, Mehta

, Chen

(2012) Risk of death in dual-eligible nursing home residents using typical or atypical antipsychotic agents. Med Care 50, 961–969.

10.

Vasilyeva

, Biscontri

, Enns

, Metge

, Alessi-Severini

(2013) Adverse events in elderly users of antipsychotic pharmacotherapy in the province of Manitoba: A retrospective cohort study. J Clin Psychopharmacol 33, 24–30.

11.

Sikirica

, Marino

, Gagne

, De Palma

, Maio

(2014) Risk of death associated with the use of conventional vs. atypical antipsychotic medications: Evaluating the use of the Emilia-Romagna Region database for pharmacoepidemiological studies. J Clin Pharm Ther 39, 38–44.

12.

Zhai

, Yin

, Zhang

(2016) Association between antipsychotic drugs and mortality in older persons with Alzheimer’s disease: A systematic review and meta-analysis. J Alzheimer Dis 52, 631–639.

13.

Nielsen

, Lolk

, Valentin

, Andersen

(2016) Cumulative dosages of antipsychotic drugs are associated with increased mortality rate in patients with Alzheimer’s dementia. Acta Psychiatr Scand 134, 314–320.

14.

Huybrechts

, Gerhard

, Cystal

, Olfson

, Avorn

, Levin

, Lucas

, Schneeweiss

(2012) Differential risk of death in older residents in nursing homes prescribed specific antipsychotic drugs: Population based cohort study. BMJ 344, e977.

15.

Kales

, Kim

, Zivin

, Valenstein

, Seyfried

, Chiang

, Cunningham

, Schneider

, Blow

(2012) Risk of mortality among individual antipsychotics in patients with dementia. Am J Psychiatry 169, 71–79.

16.

Gerhard

, Huybrechts

, Olfson

, Schneeweiss

, Bobo

, Doraiswamy

, Devanand

, Lucas

, Huang

, Malka

, Levin

, Crystal

(2014) Comparative mortality risks of antipsychotic medications in community-dwelling older adults. Br J Psychiatry 205, 44–51.

17.

Langballe

, Engdahl

, Nordeng

, Ballard

, Aarsland

, Selbæk

(2014) Short- and long-term mortality risk associated with the use of antipsychotics among 26,940 dementia outpatients: A population-based study. Am J Geriatr Psychiatry 22, 321–331.

18.

Sahlberg

, Holm

, Gislason

, Køber

, Torp-Pedersen

, Andersson

(2015) Association of selected antipsychotic agents with major adverse cardiovascular events and noncardiovascular mortality in elderly. J Am Heart Assoc 4, e001666.

19.

Schmedt

, Kollhorst

, Enders

, Jobski

, Krappweis

, Garbe

, Schink

(2016) Comparative risk of death in older adults treated with antipsychotics: A population-based cohort study. Eur Neuropsychopharmacol 26, 1390–1400.

20.

Wang

, Schneeweiss

, Avorn

, Fischer

, Mogun

, Solomon

, Brookhart

(2005) Risk of death in elderly users of conventional vs. atypical antipsychotic medications. N Engl J Med 353, 2335–2341.

21.

Maust

, Kim

, Seyfried

, Chiang

, Kavanagh

, Schneider

, Kales

(2015) Antipsychotics, other psychotropics and the risk of death in patients with dementia. Number needed to harm. JAMA Psychiatry 72, 438–445.

22.

Koponen

, Taipale

, Tanskanen

, Tolppanen

, Tiihonen

, Ahonen

, Hartikainen

(2015) Long-term use of antipsychotics among community-dwelling persons with Alzheimer’s disease: A nationwide register-based study. Eur Neuropsychopharmacol 25, 1706–1713.

23.

Finnish Medical Society Duodecim, Helsinki, 2010. Current care: Memory disorders. [in Finnish with English summary], www.kaypahoito.fi, Last updated August 13, 2010, Accessed on June 28, 2015.

24.

Azermai

, Petrovic

, Elseviers

, Bourgeois

, Van Bortel

, Vander Stichele

(2012) Systematic appraisal of dementia guidelines for the management of behavioural and psychological symptoms. Ageing Res Rev 11, 78–86.

25.

Zuidema

, Johansson

, Selbæk

, Murray

, Burns

, Ballard

, Koopmans

(2015) A consensus guideline for antipsychotic drug use for dementia in care homes. Bridging the gap between scientific evidence and clinical practice. Int Psychogeriatr 27, 1849–1859.

26.

APA Practice Guideline Writing group, Reus

, Fochtmann

, Eyler

, Hilty

, Horvitz-Lennon

, Jibson

, Lopez

, Mahoney

, Pasic

, Tan

, Wills

(2016) The American Psychiatric Association practice guideline on the use of antipsychotics to treat agitation or psychosis in patients with dementia. American Psychiatric Association, https://dx-doi-org.web.bisu.edu.cn/10.1176/appi.books.9780890426807

27.

Lonergan

, Luxenberg

, Colford

(2002) Haloperidol for agitation in dementia. Cochrane Database Syst Rev CD002852.

28.

Ballard

, Waite

(2006) The effectiveness of atypical antipsychotics for the treatment of aggression and psychosis in Alzheimer’s disease. Cochrane Database Syst Rev CD003476.

29.

Maher

, Maglione

, Bagley

, Suttorp

, Hu

, Ewing

, Wang

, Timmer

, Sultzer

, Shekelle

(2011) Efficacy and comparative effectiveness of atypical antipsychotic medications for off-label uses in adults: A systematic review and meta-analysis. JAMA 306, 1359–1369.

30.

Ballard

, Corbett

(2013) Agitation and aggression in people with Alzheimer’s disease. Curr Opin Psychiatry 26, 252–259.

31.

Declercq

, Petrovic

, Azermai

, Vander Stichele

, De Sutter

, van Driel

, Christiaens

(2013) Withdrawal versus continuation of chronic antipsychotic drugs for behavioural and psychological symptoms in older people with dementia. Cochrane Database Syst Rev 3, CD007726.

32.

Guthrie

, Clark

, Reynish

, McCowan

, Morales

(2013) Differential impact of two risk communications on antipsychotic prescribing to people with dementia in Scotland: Segmented regression time series analysis 2001-2011. Plos One 8, e68976.

33.

Martinez

, Jones

, Rietbrock

(2013) Trends in the prevalence of antipsychotic drug use among patients with Alzheimer’s disease and other dementias including those treated with antidementia drugs in the community in the UK: A cohort study. BMJ Open 3, e002080.

34.

Gallini

, Andrieu

, Donohue

, Oumouhou

, Lapeyre-Mestre

, Gardette

(2014) Trends in use of antipsychotics in elderly patients with dementia: Impact of national safety warnings. Eur Neuropsychopharmacol 24, 95–104.

35.

Nørgaard

, Jensen-Dahm

, Gasse

, Hansen

, Waldemar

(2016) Time trends in antipsychotic drug use in patients with dementia: A nationwide study. J Alzheimers Dis 49, 211–220.

36.

Schulze

, van den Bussche

, Glaeske

, Kaduszkiewicz

, Wiese

, Hoffmann

(2013) Impact of safety warnings on antipsychotic prescriptions in dementia: Nothing has changed but the years and the substances. Eur Neuropsychopharmacol 23, 1034–1042.

37.

Taipale

, Koponen

, Tanskanen

, Tolppanen

, Tiihonen

, Hartikainen

(2014) High prevalence of psychotropic drug use among persons with and without Alzheimer’s disease in Finnish nationwide cohort. Eur Neuropsychopharmacol 24, 1729–1737.

38.

Taipale

, Koponen

, Tanskanen

, Tolppanen

, Tiihonen

, Hartikainen

(2014) Antipsychotic polypharmacy among a nationwide sample of community-dwelling persons with Alzheimer’s disease. J Alzheimers Dis 41, 1223–1228.

39.

Tolppanen

, Taipale

, Koponen

, Lavikainen

, Tanskanen

, Tiihonen

, Hartikainen

(2013) Use of existing data sources in clinical epidemiology: Finnish health care registers in Alzheimer’s disease research - the Medication use among persons with Alzheimer’s disease (MEDALZ-2005) study. Clin Epidemiol 5, 277–285.

40.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34, 939–944.

41.

WHO Collaborating Centre for Drug Statistics Methodology. Defined daily dose. http://www.whocc.no/ddd/definition_and_general_considera/, Last updated September 29, 2016, Accessed on October 11, 2016.

42.

WHO Collaborating Centre for Drug Statistics Methodology. The Anatomical Therapeutic Chemical Classification System. http://www.whocc.no/atc_ddd_index/, Last updated December 16, 2015, Accessed on October 11, 2016.

43.

Tanskanen

, Taipale

, Koponen

, Tolppanen

, Hartikainen

, Ahonen

, Tiihonen

(2015) From prescription drug purchases to drug use periods - a second generation method (PRE2DUP). BMC Med Inform Decis Mak 15, 21.

44.

Charlson

, Pompei

, Ales

, MacKenzie

(1987) A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 40, 373–383.

45.

Ballard

, Hanney

, Theodoulou

, Douglas

, McShane

, Kossakowski

, Gill

, Juszczak

, Yu

, Jacoby

, for the DART-AD investigators (2009) The dementia antipsychotic withdrawal trial (DART-AD): Long-term follow-up of a randomised placebo-controlled trial. Lancet Neurol 8, 151–157.

46.

Hermes

, Sernyak

, Rosenbeck

(2013) Use of second-generation antipsychotic agents for sleep and sedation: A provider survey. Sleep 36, 597–600.

47.

Rossom

, Rector

, Lederle

, Dysken

(2010) Are all commonly prescribed antipsychotics associated with greater mortality in elderly male veterans with dementia? J Am Geriatr Soc 58, 1027–1034.

48.

Inouye

, Westendorp

RGJ

, Saczynski

(2014) Delirium in elderly people. Lancet 383, 911–922.

49.

Park

, Franklin

, Schneeweiss

, Levin

, Crystal

, Gerhard

, Huybrechts

(2015) Antipsychotics and mortality: Adjusting for mortality risk scores to address confounding by terminal illness. J Am Geriatr Soc 63, 516–523.

50.

Rikala

, Hartikainen

, Sulkava

, Korhonen

(2010) Validity of the Finnish Prescription Register for measuring psychotropic drug exposures among elderly Finns: A population-based intervention study. Drugs Aging 27, 337–349.