Antipsychotic use in elderly patients and the risk of pneumonia

Abstract

Antipsychotics are frequently and increasingly prescribed off-label for the treatment of behavioral and psychological symptoms associated with dementia, despite their modest efficacy. Instead, the safety profile of antipsychotics has been questioned repeatedly in recent years with various concerns, including death. Meta-analyses of randomized controlled trials found that one of the major causes of death associated with atypical antipsychotics use was pneumonia. Only few observational studies, however, have investigated the risk of pneumonia in elderly patients, especially among those receiving conventional antipsychotics. The aim of this editorial is to synthesize the current evidence from observational studies regarding the risk of pneumonia in elderly patients receiving either conventional or atypical antipsychotics. The studies conducted so far document that the risk of pneumonia is two- to threefold increased in a dose-dependent fashion with both classes compared to nonuse, with a possibly higher risk attributable to atypical antipsychotics. The risk seems to peak at the beginning of treatment (e.g., 7 – 30 days), and dissipates over time for both conventional and atypical antipsychotics. The risk–benefit ratio suggests that there will be 1 excess hospitalization for pneumonia for every 2 – 5 patients receiving any clinical improvement in symptoms. Considering the modest improvement in terms of efficacy, the risks associated with antipsychotics in elderly patients may outweigh their benefit.

Keywords:

• antipsychotics
• behavioral disorders
• dementia
• elderly
• observational studies
• pneumonia

Antipsychotics are commonly prescribed in both community and nursing home setting. These drugs are traditionally classified as conventional or atypical agents. The two classes differ in terms of pharmacological profile: conventional antipsychotics (such as haloperidol) are D2-receptor antagonists while atypical antipsychotics (such as quetiapine, olanzapine and risperidone) are 5HT-receptor antagonists although they also bind to other receptor types.

In the United Kingdom, psychiatric drugs made up nearly 9% of all prescriptions in 2010, with olanzapine, quetiapine and risperidone accounting for 24, 23 and 17% of all prescriptions, respectively [1]. Antipsychotics are primarily indicated in the treatment of schizophrenia and in the manic phases of bipolar disorder. However, these drugs are also frequently used off-label. In particular, in the last years there is an increasing use of antipsychotics worldwide for the treatment of behavioral and psychological symptoms of dementia (BPSD). Only risperidone is approved for the treatment of aggression, one of the several symptoms of BPSD.

The safety of antipsychotics when used in elderly people with dementia has been seriously questioned. The FDA issued a warning in April 2005 about an almost twofold increased risk of all-cause mortality when antipsychotics were used to control BPSD, with pneumonia being one of the leading causes of death [2]. The FDA later extended such warning to conventional antipsychotics, inferring that the risk of pneumonia may be similarly increased by conventional agents [3].

After the initial warning, several observational studies have investigated the association between antipsychotics use in elderly patients and pneumonia (Table 1). In the Netherlands, Knol et al. assessed the risk of pneumonia leading to hospitalization using a record-linkage administrative database [4], while Trifirò et al. investigated fatal and nonfatal pneumonia using a nationwide general practice database [5]. Interestingly, both studies found an increased risk of pneumonia with antipsychotic use, although twice as high with atypical as compared to conventional antipsychotic agents. The higher risk of pneumonia associated with atypical antipsychotics has been confirmed by Huybrechts et al. in North America [6,7]. Similarly, a differential risk of pneumonia associated with antipsychotic class was also observed by Barnett et al. using a retrospective cohort study design [8]. Others studies found no differential risk or provided insufficient information [9,10].

Table 1. Observational studies investigating pneumonia as an outcome following antipsychotic exposure in elderly persons.

Author (year)	Study design	Setting – population	Outcome	Exposure	Main risk estimates
Knol et al. (2008) [4]	Nested case–control	Dutch PHARMO database – patients (≥ 65 years) newly treated with antipsychotics (n = 22,944; n = 543 cases)	Hospital admission due to pneumonia	AAP or CAP (nonuse as comparator)	Adj. OR -current use of AAP: 3.1 (1.9 – 5.1) -current use of CAPs: 1.5 (1.2 – 1.9)
Trifirò et al. (2010) [5]	Nested case–control	Dutch general practice database – older people (≥ 65 years) newly treated with AP (n = 258)	Fatal and nonfatal community-acquired pneumonia	AAP or CAP (past use of any AP as comparator)	Adj. OR Fatal/nonfatal pneumonia: -AAPs: 2.61 (1.5 – 4.6) -CAPs: 1.8 (1.2 – 2.5) Fatal pneumonia: - AAPs: 6.0 (1.5 – 24.0) -CAPs: 1.7 (0.8 – 3.9)
Huybrechts et al. (2011) [6]	Retrospective cohort study	Administrative data from the British Columbia Ministry of Health – elderly persons ≥ 65 years newly admitted to a nursing home (n = 10,900)	All noncancer deaths, including pneumonia	AAPs, CAPs (AAPs as comparator when comparing AP classes), benzodiazepines and antidepressants	Propensity-adj. RR: -CAPs: 0.94 (0.56 – 1.58)
Huybrechts et al. (2012) [7]	Retrospective cohort study	Medicaid–Medicare, Minimum Data Set and Online Survey Certification and Reporting data for patients from nursing homes in 45 US states – residents ≥ 65 years who initiated antipsychotic treatment in nursing home (n = 83,959)	Hospitalization for myocardial infarction, cerebrovascular events, serious bacterial infections (including pneumonia), and hip fracture within 180 days of antipsychotic initiation	AAPs (AAPs as comparator when comparing classes; risperidone as comparator when comparing individual APs)	Propensity-adj. HR -CAPs: 0.81 (0.65 – 1.01) -aripiprazole: 0.99 (0.72 – 1.35) -olanzapine: 0.94 (0.84 – 1.05) -quetiapine: 0.91 (0.80 – 1.03)
Barnett et al. (2006) [8]	Retrospective cohort study	US Veterans Administration database – patients hospitalized due to pneumonia (n = 16,931)	In-hospital mortality	AAP or CAP (no use of neuropsychiatric drugs as comparator)	Adj. OR -CAP: 1.5 (1.0 – 2.2) -AAPs: 1.2 (1.0 – 1.5)
Gau et al. (2010) [9]	Case–control study	Rural community hospital in Ohio (USA) – patients aged 65 years or older (n = 194)	Hospital admission due to community-acquired pneumonia	AAP (nonuse as comparator)	Adj. OR -AAP: 2.26 (1.23 – 4.15)
Aparasu et al. (2013) [10]	Retrospective cohort study	Dual eligible (Medicare Medicaid) nursing home patients new antipsychotic users in four US states (n = 49,904)	Hospital claim for pneumonia within 6 months of treatment	CAPs (AAPs as comparator)	Adj. HR 1.24 (0.94 – 1.64) Adj. HR < 50 days: 1.17 (0.83 – 1.66) Adj. HR 50 – 180 days: 1.36 (0.87 – 2.14)
Wang et al. (2007) [12]	Retrospective cohort study	The Pharmaceutical Assistance Contract for the Elderly Information from the Pennsylvania Pharmaceutical Assistance Contract for the Elderly and Pennsylvania Medicare – patients ≥ 65 years with at least one prescription for an antipsychotic (n = 22,890)	Acute myocardial infarction; ventricular arrhythmia; cerebrovascular events; congestive heart failure; pneumonia; other serious bacterial infections	CAPs (AAPs as comparator)	Adj. HR in 30 days: 1.11 (0.76 – 1.63) -Adj. HR in 60 days: 1.03 (0.76 – 1.38) -Adj. HR in 120 days: 0.84 (0.66 – 1.05)
Pratt et al. (2011) [13]	Self-controlled case series	Australian Government Department of Veterans’ Affairs Health Care Claims Database – elderly patients ≥ 65 years (n = 13,324 patients hospitalized for pneumonia)	Hospitalization rates for hip fracture and pneumonia	AAPs or CAPs (nonuse as comparator)	Adj. IRR -CAPs at 1 week: 1.51 (1.07 – 2.14) -CAPS at 2 – 8 weeks: 1.62 (1.37 – 1.92) -CAPS at 9 – 12 weeks:1.69 (1.32 – 2.16) -CAPS at over 12 weeks: 1.63 (1.36 – 1.96) AAPs at 1 week: 1.73 (1.31 – 2.29) AAPs at 2 – 8 weeks: 1.70 (1.48 – 1.95) AAPs at 9 – 12 weeks: 1.67 (1.37 – 2.04) AAPs at over 12 weeks: 1.70 (1.51 – 1.93)
Star et al. (2010) [17]	Self-controlled cohort	UK IMS Health Disease Analyzer database – adults aged 65 years or older (number of patients not provided)	Acute chest infections, bronchopneumonia, pneumonia	Distribution over time of AAPs and CAPs with respect to date of diagnosis of the study outcomes	In elderly patients (≥ 65): -higher rate of acute chest infections following AAPs and much less CAPs -higher rate of bronchopneumonia following either AAP or CAP prescriptions.
Hatta et al. (2013) [18]	Prospective observational study	Thirty-three general hospitals, where at least one psychiatrist worked full time – patients who developed delirium during hospital admission and received antipsychotics (N = 2453 of which 30% had dementia)	Serious adverse events, including aspiration pneumonia as secondary outcome	All APs prescribed	Aspiration pneumonia, n (%): -with all APs 17 (0.7) -with risperidone 7 (0.8) -with quetiapine 4 (0.5) -with risperidone 0 (0) -with olanzapine 2 (2.3) -with aripiprazole 0 (0) -with haloperidol 3 (0.6) -with ‘other’ antipsychotics 1 (0.8)

Risk estimates are reported with 95% CI.

Adj: Adjusted; AAPs: Atypical antipsychotics; APs: Antipsychotics; CAPs: Conventional antipsychotics; HR: Hazard ratio; IRR: Incidence rate ratio; OR: Odds ratio; RR: Risk ratio.

While these and other findings indicate an increased risk of pneumonia with antipsychotics, observational studies are liable to confounding by indication because dementia patients have a higher baseline risk of aspiration pneumonia [11]. In addition, frail elderly persons may experience delirium as a prodromal symptom of pneumonia and, as a consequence, receive antipsychotic treatment. This increases the risk of protopathic bias in observational studies, that is, wrong attribution of the onset of pneumonia to antipsychotic administration [4]. Nevertheless, Trifirò et al. [5] and Huybrechts et al. [7] observed that the risk of pneumonia appears to be dose-dependent, strengthening the hypothesis that antipsychotics are involved in the causal pathway.

As regards the temporal pattern of pneumonia associated with antipsychotics, current findings suggest that the risk peaks initially and decreases over time. Wang et al. [12] reported that the risk is higher within 30 days of antipsychotic initiation, (hazard ratio [HR] 1.11 [95% CI: 0.76 – 1.63]), decreases after 60 days (HR 1.03 [95% CI: 0.76 – 1.38]) and it is not further evident after 120 days of continuous treatment (HR 0.84 [95% CI: 0.66 – 1.05]). However, these findings were not statistically significant. Trifirò et al. [5] documented that the risk of pneumonia in elderly patients treated with antipsychotics appears to be higher during the very first week of treatment (odds ratio [OR] 4.62 [95% CI: 2.05 – 10.38]) and decreases thereafter. Similarly, reduced risk with continuous treatment was found in a self-case controlled series by Pratt et al. [13], but only for atypical antipsychotics.

The evaluation of antipsychotic-associated risk of pneumonia by class is limited by the heterogeneity of individual antipsychotics. Based on the different receptor-binding profile of antipsychotic drugs, it has been hypothesized that the risk of pneumonia might differ by individual agent [13]. However, there are only few studies investigating the risk of pneumonia with individual antipsychotic agents. A nested case–control study in 2560 elderly patients observed that the risk of pneumonia was highest for risperidone (OR 3.51 [95% CI: 1.94 – 6.36]), followed by haloperidol (OR 1.95 [95% CI: 1.20 – 3.17]) [5]. A similar study using risperidone as a comparator, however, found no statistically significant differences in the risk of pneumonia for olanzapine, quetiapine and aripiprazole [7].

More evidence is needed to confirm that the risk of antipsychotic-associated pneumonia varies with individual antipsychotics.

The biological pathways underlying antipsychotic-induced pneumonia are not currently known although plausible hypotheses exist. Antipsychotics may lead to aspiration pneumonia in elderly patients through extrapyramidal adverse events, dysphagia or sedation, as a result of modulation of dopaminergic, muscarinic, and H1-histaminergic receptor systems, respectively [5]. It should be noted, however, that in most of the observational studies there were insufficient information about the severity of the underlying condition and inconsistent data regarding comorbidities [14]. Nonetheless, the use of antipsychotics has been linked to an increased risk of pneumonia also in large nationwide cohorts of much younger patients affected by schizophrenia, and by bipolar disorder with a substantially lower burden of comorbid conditions [15,16]. In these patients, lithium and other mood stabilizers are not associated with the risk of pneumonia. Receptor affinities for histaminergic and muscarinic are considered the most plausible explanation for the association seen with antipsychotics in such cohorts. To further support this biological mechanism, it is a consistent finding that the concomitant use of more than one antipsychotic drug is far more dangerous [15,16].

Expert opinion

The studies conducted so far suggest an association between antipsychotic drug use in elderly persons and pneumonia. Three studies demonstrated that the risk of pneumonia varies by antipsychotic class, with a higher risk being attributed to atypical antipsychotics [4,6-9]. A more recent finding is the demonstration of a differential risk associated with individual antipsychotic drugs [5,7]. Further studies are needed to confirm whether there are differences in risk of pneumonia associated with individual antipsychotics and to identify possible risk factors of antipsychotic-induced pneumonia, particularly in frail elderly with dementia. The role of severity of dementia as a risk factor was generally not considered in any of the studies.

As pneumonia associated with antipsychotic in elderly patients is more likely to occur at higher dosage, it is important to start the therapy with the lowest dosage possible, followed by careful dose titration.

In addition, the evidence about the temporal relation between the initiation of an antipsychotic and the onset of pneumonia suggests that a more intense patient monitoring is needed immediately after initiation and during the early phases of antipsychotic treatment, particularly among nursing home residents.

It is currently not known which clinical risk factors predispose elderly patients prescribed antipsychotics to develop pneumonia. This limits clinicians in their ability to identify elderly persons for whom the risks of antipsychotic treatment exceed the benefit.

Future research should provide more details on individual antipsychotics, the lowest-risk dose and the role of severity of dementia and other possible risk factors in the antipsychotic–pneumonia association. To conclude, the risk of pneumonia associated with antipsychotics has been quantified by several studies and these have provided valuable information. Any antipsychotic use is associated with an increased risk of pneumonia compared to nonuse but atypical antipsychotics were often found to be associated with a higher risk of pneumonia compared to conventional agents. Pratt and co-authors have estimated the risk–benefit ratio suggesting that there will be 1 excess hospitalization for pneumonia for every 2 – 5 patients receiving any clinical improvement in symptoms [13]. Considering the modest improvement in terms of efficacy, the risks associated with antipsychotics in elderly patients may outweigh their benefit.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.