Association between Antipsychotic Drugs and Mortality in Older Persons with Alzheimer’s Disease: A Systematic Review and Meta-Analysis

Abstract

Antipsychotic drugs have been inconsistently associated with death risk of Alzheimer’s disease (AD) patients. Herein we review and quantitatively summarize the evidence from epidemiological studies. Pertinent studies were identified by searching PubMed and Cochrane Library Register of Controlled Trials through 20 December 2015. The DerSimonian and Laird random effect model was adopted as the pooling method. Twelve studies from nine articles with 11,463 participants were included. The pooled RR of observational studies was 1.36 (95% CI, 0.83–2.24; I² = 94.9%) for antipsychotic drugs users versus individuals who were not exposed to antipsychotic drugs. When the three studies that were key contributors to the high heterogeneity were excluded, the pooled RR was 2.08 (95% CI 1.39 to 3.13). The result of one double-blind randomized clinical trial indicated that antipsychotic drugs nearly doubled the risk of death in AD patients. In conclusion, there is no evidence of absence of association between antipsychotic drugs’ use with death risk of AD patients. Careful assessments of potential benefits and risks should be made before prescribing antipsychotics for treatment of psychosis symptoms and behavioral problems of AD patients.

Keywords

Alzheimer’s disease antipsychotic agents epidemiology meta-analysis

INTRODUCTION

Worldwide, 47.5 million people have dementia, and there are 7.7 million new cases every year; the total number of people with dementia is projected to increase to 75.6 million in 2030, and almost triple by 2050 to 135.5 million (http://www.who.int/mediacentre/factsheets/fs362/en/). In one cohort study based on a large population, antipsychotic medications taken by dementia patients were reported to be associated with higher mortality risk than most other medications used for neuropsychiatric symptoms [1]. Similarly, studies conducted in nursing homes (or end-stage dementia) also found an association between dead risk and conventional or atypical antipsychotics’ use [2, 3].

Alzheimer’s disease (AD) is the most common form of dementia, and may contribute to 60–70% of cases of dementia [4]. Researchers have investigated the association between prescription of antipsychotic drugs and mortality in patients with AD in several epidemiological studies, and the results have been controversial [5 –10]. One study [7] reported that antipsychotics’ use was related to increased mortality of AD patients. However, another prospective cohort study [9] showed no significant association between antipsychotic drugs and death risk in patients with AD. Additionally, other studies about the association between different types of antipsychotics (conventional and atypical) and the death risk of AD patients also had inconsistent results [6 , 11]. Musicco et al. found that the death risk of AD patients who had been prescribed a conventional antipsychotic drug was higher than that of those prescribed an atypical drug [7]. Nevertheless, another study by Liperoti et al. indicated that there was no difference between conventional and atypical antipsychotic drugs, with respect to the death risk of AD patients [6].

Therefore, we conducted a systematic review and meta-analysis by combining all available data of epidemiological studies to derive an estimation of the association between antipsychotics’ use and the death risk of AD patients.

METHODS

Literature search and selection

We followed the PRISMA guidelines [12]. A search of the literature up to 20 December 2015 was performed, using the databases of PubMed and Cochrane Library Register of Controlled Trials, with the following MeSH (Medical Subject Headings) terms: Antipsychotic Agents and Tranquilizing Agents combined with Alzheimer Disease. The search was limited to human studies reporting in English. Moreover, we reviewed the reference lists from retrieved articles to search for further relevant studies.

To be included, studies had to meet the following criteria: (1) original research papers with retrospective or prospective design (randomized controlled trials or observational studies); (2) the exposure of interest was the use of antipsychotic drugs; (3) the outcome of interest was mortality in older people with AD; (4) relative risk (RR) with 95% confidence interval (CI) (or data to calculate these) was provided; and (5) be published in English. If data from the same population had been published more than once, the most recent and complete studies were chosen. Two investigators searched articles and reviewed all retrieved studies independently. If the two investigators disagreed about the eligibility of an article, it was resolved by consensus.

Data extraction and quality assessment

The following data were extracted from each study by two investigators: (1) name of the first author; (2) publication year; (3) study population; (4) study region; (5) sample size; (6) age range or mean age of subjects; (7) study period; (8) methods to assess AD; (9) the type of antipsychotics, dosage, and the length of the treatment; (10) RR (adjusted for the largest number of confounders in the original study) with 95% CI for antipsychotic drugs and the risk of mortality; (11) adjustment for confounders.

The quality of observational studies was assessed using the Newcastle–Ottawa quality assessment scale (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp), and the quality of the randomized clinical trial was evaluated by Jadad score [13]. The maximum score on the Newcastle–Ottawa quality assessment scale is 9. “Good” was defined as a total score of 7–9; “fair,” a total score of 4–6, and “poor,” defined as a total score less than 4. For the Jadad score, a total score of 3–5 indicated high methodological quality, and a total score of 1–2 represented low quality.

Statistical analysis

We weighted the study-specific log relative risks by the inverse of their variance to calculate a summary estimate and its 95% confidence interval. The DerSimonian and Laird random effects model was selected [14]. I² of Higgins and Thompson was used to assess heterogeneity among studies [15] and I² values of 0, 25, 50, and 75% represent no, low, moderate, and high heterogeneity [14], respectively. Univariate meta-regression analyses by study region, study period, and number of adjustments were conducted to investigate the potential sources of heterogeneity. The leave-one-out analysis [16] was carried out to evaluate the key studies that have a substantial impact on the between-study heterogeneity. We conducted sensitivity analyses excluding one study at a time to investigate whether the results were due to one large study or a study with an extreme result. We also conducted subgroup analyses by type of antipsychotic drugs (conventional and atypical). Small-study effect was assessed with visual inspection of the funnel plot and Egger’s test [17].

All statistical analyses were conducted by Stata 12.0 (Stata Corp, College Station, TX). A two-tailed p < 0.05 was considered statistically significant.

RESULTS

Literature search and study characteristics

Initially, we identified 741 articles by our literature search, 689 of which were excluded after the review of titles or abstracts (Fig. 1). One additional article [7] was found in reference lists of retrieved studies. We reviewed fifty-two possibly relevant articles in full text. Five articles on mortality risk in patients with dementia were excluded, as well as seven on other drugs. In addition, four letters and two reviews without original data were excluded, along with 25 articles without RRs concerning the relation between antipsychotic drugs and death risk in patients with AD. At last, twelve studies with 11,463 subjects from nine articles [3 , 18] were included (Table 1).

Overall, nine studies [3 , 18] investigated the relationship between antipsychotic drugs and mortality risk in patients with AD, and three studies [6 , 11] discovered the association of conventional or atypical antipsychotic drugs and mortality in AD patients. With regard to study region, one study was conducted in Asia, [5] three in America [6, 9], and eight in Europe [3 , 18]. All studies reported the criteria used for AD ascertainment. The major measured confounding factors included age, gender, race, education, comorbidities, and medication use. Quality assessment showed that the Newcastle-Ottawa score of observational studies was not less than 7, and the Jadad score of the randomized clinical trial was 5, indicating that the methodological quality was generally good (Supplementary Tables 1 and 2).

Antipsychotic drugs and mortality risk

Nine included studies [3 , 18] with 11,186 participants reported death risks for AD patients prescribed antipsychotic drugs. Five of these studies showed a significant association between antipsychotic drugs use and mortality of AD patients [3 , 18], while the other four studies [8 –10] indicated no relation between antipsychotics and death of AD patients. The pooled RR of eight observational studies [5 , 18] was 1.36 (95% CI, 0.83–2.24) for users of antipsychotic drugs versus those who had not been exposed to antipsychotic drugs, with high heterogeneity (I² = 94.9%, p_{heterogeneity} = 0) (Fig. 2). The double-blind randomized clinical trial showed that there was an increased risk of mortality in patients with AD who were prescribed antipsychotic medication; antipsychotic drugs nearly doubled the risk of death [3].

Meta-regression

As seen in Fig. 2, high heterogeneity (I² = 94.9%, o_{heterogeneity} = 0) among studies was demonstrated. Univariate meta-regression analysis with the covariates of study region, study period, and number of adjustments, showed no covariate having a significant impact on between-study heterogeneity (p values were 0.054, 0.769, and 0.089, respectively).

Subgroup analyses

In subgroup analyses, the pooled RR for mortality of AD patients was (1.75, 95% CI, 0.40–7.56; I² = 97.8%) for studies [7, 9] of conventional antipsychotic drugs and (1.75, 95% CI, 0.99–3.09; I² = 87.6%) for studies [7 , 18] of atypical antipsychotic drugs.

Sensitivity analyses

In a sensitivity analysis excluding one study at a time, the pooled RR for mortality of AD patients ranged from 1.17 (95% CI, 0.73–1.89) to 1.51 (95% CI, 0.91–2.48) (Fig. 3). All of the point estimates lay within the 95% CI of the combined analysis, indicating that no individual study had excessive influence on the pooled effect between antipsychotic drugs’ use and death risk of patients with AD.

Three studies [5, 9] were found to be key contributors to this high between-study heterogeneity by the leave-one-out sensitivity analysis. After further excluding these studies, the heterogeneity was decreased (I² = 75.7%), and the pooled RR was 2.08 (95% CI 1.39 to 3.13).

Small-study effect evaluation

Visually, the funnel plot seemed asymmetrical; whereas, Egger’s test (p = 0.638) showed no evidence of small-study effects (Fig. 4).

Conventional versus atypical antipsychotic drugs

Three included studies [6 , 11] with 5,148 participants provided death risks for persons prescribed a conventional antipsychotic drug compared to those prescribed an atypical drug. The pooled result indicated that there was no difference between conventional and atypical antipsychotic drugs with respect to the death risk of AD patients (RR, 1.23; 95% CI, 0.91–1.65; I² = 30.5%, p_{heterogeneity} = 0.237). For the limited study number, we did not conduct further analysis.

DISCUSSION

To date, a few meta-analyses evaluated the relationship between antipsychotics and dementia [2 , 20]. A meta-analysis of 15 randomized placebo-controlled trials indicated an increased risk for deaths in patients with dementia receiving atypical antipsychotic drugs [2]. Another meta-analysis found a nonsignificant increase in mortality during treatment with risperidone in dementia patients [19]. A more recent meta-analysis showed higher risks for adverse events and mortality may offset the efficacy of atypical antipsychotics for treatment of dementia [20]. The study objects of the aforementioned research were patients with dementia. However, dementias may include heterogeneous clinical presentations such as AD, vascular dementia, Lewy body dementia, and frontotemporal dementia [21]. Thus, this article merely evaluated epidemiological studies investigating antipsychotic drugs’ use and the death risk of people with AD.

Overall, we included twelve studies with 11,463 participants. Meta-analysis of observational studies indicated no association between antipsychotic drugs use and mortality of AD patients. The summary RRs (95% CIs) for death risk did not substantially change in subgroup analyses and sensitivity analyses, except for the leave-one-out analysis.

Methodological considerations of meta-analysis

However, the meta-analysis result was based on observational studies [5–11 , 18], and only four studies [5 , 18] were prospective. The largest weakness of non-experimental studies is selection bias. In observational studies, an investigator makes no intervention, and patients are allocated to treatment based on clinical decisions of the physician. Confounding by indication (the patient characteristics that have made a physician prescribe a particular drug to a particular patient) makes observational studies much more vulnerable to methodological problems. In contrast, the purpose of randomization in a clinical trial is to balance variation and unknown confounders. Therefore, it is reasonable that randomized controlled trials are considered the best way of proving causality, especially when the AD groups that needed the prescription of antipsychotic drugs were different from the groups that did not need antipsychotics.

Our meta-analysis showed a very high degree of heterogeneity. This inconsistency reduces the confidence of recommendations about treatment. According to the Cochrane handbook (http://handbook.cochrane.org), meta-analysis should only be considered when a group of studies is sufficiently homogeneous in terms of participants, interventions, and outcomes to provide a meaningful summary. Our meta-regression and subgroup analyses did not find any of study region, study period, number of adjustments, and type of antipsychotic drugs as the important contributor to the between-study heterogeneity. An indeterminate number of characteristics that vary among studies could be the cause of between-study heterogeneity. Other potential contributors to the conflicting results could be: (1) different methods to diagnose AD; (2) drug-drug interactions because patients were simultaneously prescribed two or more antipsychotics (e.g., risperidone plus olanzapine) [22, 23]; (3) various side effects of antipsychotics on participants among included studies [24]; (4) the age group and age-related comorbidities of participants varied among studies; (5) the diversity of potential confounders adjusted for in studies. In the leave-one-out sensitivity analysis, when the three studies [5, 9] that were key contributors to the high heterogeneity were excluded, the pooled RR was 2.08 (95% CI 1.39 to 3.13).

Literature based on clinical trials

There is one double-blind randomized clinical trial showing that antipsychotic drugs nearly double the risk of death in AD patients [3]. The meta-analysis by Schneider et al. indirectly proved that antipsychotics increased the risk of mortality of AD patients [2]. Eighty-seven percent of the 3,353 randomized patients suffered from AD. Subgroup analysis did not reveal heterogeneity between trials of patients with psychosis of AD, compared with those trials that did not select patients on this basis. In a review, Ballard and Howard mentioned that summary data supplied by Eli Lilly suggested a similar increase in the incidence of cerebrovascular adverse events in the placebo-controlled trials of olanzapine in AD (olanzapine 1.3% versus placebo 0.4%) [25].

Limitations

A strength of the present study is the large number of subjects included, allowing a greater possibility of reaching reasonable conclusions. However, the potential limitations should also be considered. First, although the researchers had adjusted for some confounders, other potential confounding factors (e.g., genetic factors) could not be ruled out completely. Second, no included study used histological method (postmortem examination) to diagnose AD. Five studies [3 , 8–10] included patients with only probable AD, and others included probable and possible AD patients in general. Therefore, the results could be affected by the selection bias. Third, the limited information, such as the dosages of exposure to antipsychotics, provided in the included studies precluded the possibility of dose-response analysis. Fourth, for the limited number of included studies, the test power for funnel plot asymmetry might be too low to distinguish chance from real asymmetry. Hence, undetected bias in the results of meta-analysis could not be completely ruled out.

Clinical implications

The evidence provided by this meta-analysis is weak. The literature based on clinical trials is small, but suggests that antipsychotic drugs increase the risk of death in AD patients. Thus, based on the overall review of the literature, the evidence tends to favor an increased risk of death. Conservatively, there is no evidence of absence of association between antipsychotic drugs’ use with death risk of AD patients. Data from randomized controlled trials of probable AD patients are needed to better understand the association.

DISCLOSURE STATEMENT

Authors’ disclosures available online (http://j-alz.com/manuscript-disclosures/15-1207r1).

Footnotes

Appendix

References

Kales

, Valenstein

, Kim

, McCarthy

, Ganoczy

, Cunningham

, Blow

(2007) Mortality risk in patients with dementia treated with antipsychotics versus other psychiatric medications. Am J Psychiatry 164, 1568–1576;quiz 1623.

Schneider

, Dagerman

, Insel

(2005) Risk of death with atypical antipsychotic drug treatment for dementia: Meta-analysis of randomized placebo-controlled trials. JAMA 294, 1934–1943.

Ballard

, Hanney

, Theodoulou

, Douglas

, McShane

, Kossakowski

, Gill

, Juszczak

, Yu

, Jacoby

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

Burns

, Iliffe

(2009) Alzheimer’s disease. BMJ 338, b158.

Suh

, Shah

(2005) Effect of antipsychotics on mortality in elderly patients with dementia: A 1-year prospective study in a nursing home. Int Psychogeriatr 17, 429–441.

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.

Musicco

, Palmer

, Russo

, Caltagirone

, Adorni

, Pettenati

, Bisanti

(2011) Association between prescription of conventional or atypical antipsychotic drugs and mortality in older persons with Alzheimer’s disease. Dement Geriatr Cogn Disord 31, 218–224.

Gardette

, Lapeyre-Mestre

, Coley

, Cantet

, Montastruc

, Vellas

, Andrieu

(2012) Antipsychotic use and mortality risk in community-dwelling Alzheimer’s disease patients: Evidence for a role of dementia severity. Curr Alzheimer Res 9, 1106–1116.

Lopez

, Becker

, Chang

, Sweet

, Aizenstein

, Snitz

, Saxton

, McDade

, Kamboh

, DeKosky

, ReynoldsCF, 3rd, Klunk

(2013) The long-term effects of conventional and atypical antipsychotics in patients with probable Alzheimer’s disease. Am J Psychiatry 170, 1051–1058.

10.

Vilalta-Franch

, Calvo-Perxas

, Garre-Olmo

, Turro-Garriga

, Lopez-Pousa

(2013) Apathy syndrome in Alzheimer’s disease epidemiology: Prevalence, incidence, persistence, and risk and mortality factors. J Alzheimers Dis 33, 535–543.

11.

Trifiro

, Verhamme

, Ziere

, Caputi

, Ch Stricker

, Sturkenboom

(2007) All-cause mortality associated with atypical and typical antipsychotics in demented outpatients. Pharmacoepidemiol Drug Saf 16, 538–544.

12.

Moher

, Liberati

, Tetzlaff

, Altman

(2009) Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLoS Med 6, e1000097.

13.

Jadad

, Moore

, Carroll

, Jenkinson

, Reynolds

, Gavaghan

, McQuay

(1996) Assessing the quality of reports of randomized clinical trials: Is blinding necessary?. Control Clin Trials 17, 1–12.

14.

Higgins

, Thompson

, Deeks

, Altman

(2003) Measuring inconsistency in meta-analyses. BMJ 327, 557–560.

15.

Higgins

, Thompson

(2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21, 1539–1558.

16.

Patsopoulos

, Evangelou

, Ioannidis

(2008) Sensitivity of between-study heterogeneity in meta-analysis: Proposed metrics and empirical evaluation. Int J Epidemiol 37, 1148–1157.

17.

Egger

, Davey Smith

, Schneider

, Minder

(1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315, 629–634.

18.

Piersanti

, Capannolo

, Turchetti

, Serroni

, De Berardis

, Evangelista

, Costantini

, Orsini

, Rossi

, Maggio

(2014) Increase in mortality rate in patients with dementia treated with atypical antipsychotics: A cohort study in outpatients in Central Italy. Riv Psichiatr 49, 34–40.

19.

Haupt

, Cruz-Jentoft

, Jeste

(2006) Mortality in elderly dementia patients treated with risperidone. J Clin Psychopharmacol 26, 566–570.

20.

, Huang

, Cong

, Wang

, Jiang

, Gao

, Zhu

(2014) The efficacy and safety of atypical antipsychotics for the treatment of dementia: A meta-analysis of randomized placebo-controlled trials. J Alzheimers Dis 42, 915–937.

21.

Burns

, Iliffe

(2009) Dementia. BMJ 338, b75.

22.

Ereshefsky

(1996) Pharmacokinetics and drug interactions: Update for new antipsychotics. J Clin Psychiatry 57(Suppl 11), 12–25.

23.

Alexopoulos

, Streim

, Carpenter

, Docherty

(2004) Expert consensus guidelines for using antipsychotic agents in older patients. J Clin Psychiatry 65, 1–105.

24.

Masand

(1999) Side effects of antipsychotics in the elderly. J Clin Psychiatry 61, 43–49;discussion 50-41.

25.

Ballard

, Howard

(2006) Neuroleptic drugs in dementia: Benefits and harm. Nat Rev Neurosci 7, 492–500.

26.

Lamberts

, Wood

, Hofmans-Okkes

(1992) International primary care classifications: The effect of fifteen years of evolution. Fam Pract 9, 330–339.

27.

Gambassi

, Landi

, Peng

, Brostrup-Jensen

, Calore

, Hiris

, Lipsitz

, Mor

, Bernabei

(1998) Validity of diagnostic and drug data in standardized nursing home resident assessments: Potential for geriatric pharmacoepidemiology. SAGE Study Group. Systematic Assessment of Geriatric drug use via Epidemiology. Med Care 36, 167–179.