Key Points

QuestionÌý What antipsychotics are associated with pneumonia, which is the fourth leading cause of death in patients with schizophrenia?

FindingsÌý In this cohort study of 61 889 people with schizophrenia, antipsychotics with high anticholinergic burden were associated with an increased incidence of pneumonia. Significantly higher risks were detected for clozapine (≥180 mg/d), quetiapine (≥440 mg/d), and olanzapine (≥11 mg/d).

MeaningÌý Findings of this study suggest that in patients with schizophrenia, pneumonia is associated with the use of specific antipsychotics, setting the stage for personalized prevention strategies to decrease the incidence of this life-threatening condition in patients with schizophrenia.

ImportanceÌý Antipsychotic drugs (particularly clozapine) have been associated with pneumonia in observational studies. Despite studies of the associations between antipsychotic use and incident pneumonia, it remains unclear to what degree antipsychotic use is associated with increased risk of pneumonia, whether dose-response associations exist, and what agents are specifically associated with incident pneumonia.

ObjectiveÌý To estimate pneumonia risk associated with specific antipsychotics and examine whether polytherapy, dosing, and receptor binding properties are associated with pneumonia in patients with schizophrenia.

Design, Setting, and ParticipantsÌý This cohort study identified patients with schizophrenia or schizoaffective disorder (hereafter, schizophrenia) aged 16 years or older from nationwide Finnish registers from 1972 to 2014. Data on diagnoses, inpatient care, and specialized outpatient care were obtained from the Hospital Discharge Register. Information on outpatient medication dispensing was obtained from the Prescription Register. Study follow-up was from 1996 to 2017. Data were analyzed from November 4, 2022, to December 5, 2023.

ExposuresÌý Use of specific antipsychotic monotherapies; antipsychotics modeled by dosage as low (<0.6 of the World Health Organization defined daily dose [DDD] per day), medium (0.6 to <1.1 DDDs per day), or high dose (≥1.1 DDDs per day); antipsychotic polypharmacy; and antipsychotics categorized according to their anticholinergic burden as low, medium, and high.

Main Outcomes and MeasuresÌý The primary outcome was hospitalization for incident pneumonia. Pneumonia risk was analyzed using adjusted, within-individual Cox proportional hazards regression models, with no antipsychotic use as the reference.

ResultsÌý The study included 61 889 persons with schizophrenia (mean [SD] age, 46.2 [16.0] years; 31 104 men [50.3%]). During 22 years of follow-up, 8917 patients (14.4%) had 1 or more hospitalizations for pneumonia and 1137 (12.8%) died within 30 days of admission. Compared with no antipsychotic use, any antipsychotic use overall was not associated with pneumonia (adjusted hazard ratio [AHR], 1.12; 95% CI, 0.99-1.26). Monotherapy use was associated with increased pneumonia risk compared with no antipsychotic use (AHR, 1.15 [95% CI, 1.02-1.30]; P = .03) in a dose-dependent manner, but polytherapy use was not. When categorized by anticholinergic burden, only the use of antipsychotics with a high anticholinergic burden was associated with pneumonia (AHR, 1.26 [95% CI, 1.10-1.45]; P < .001). Of specific drugs, high-dose quetiapine (AHR, 1.78 [95% CI, 1.22-2.60]; P = .003), high- and medium-dose clozapine (AHR, 1.44 [95% CI, 1.22-1.71]; P < .001 and AHR, 1.43 [95% CI, 1.18-1.74]; P < .001, respectively), and high-dose olanzapine (AHR, 1.29 [95% CI, 1.05-1.58]; P = .02) were associated with increased pneumonia risk.

Conclusions and RelevanceÌý Results of this cohort study suggest that in patients with schizophrenia, antipsychotic agents associated with pneumonia include not only clozapine (at dosages ≥180 mg/d) but also quetiapine (≥440 mg/d) and olanzapine (≥11 mg/d). Moreover, monotherapy antipsychotics and antipsychotics with high anticholinergic burden are associated with increased pneumonia risk in a dose-dependent manner. These findings call for prevention strategies aimed at patients with schizophrenia requiring high-risk antipsychotics.

Observational studies have found associations between antipsychotic use and pneumonia in people with^1,2 and without known pre-existing pulmonary conditions.^3-8 Nonetheless, most schizophrenia guidelines provide little to no guidance on how to manage pneumonia and mitigate pneumonia risk in people using antipsychotics.^9,10 This situation may result from the limited number of available studies examining associations between pneumonia and antipsychotic use,^3-8 their risk of bias, the limited statistical power in some studies, and the low overall quality and low certainty of the evidence.¹¹ In addition, although antipsychotic dosage has been suggested to be a determinant of antipsychotic-related pneumonia, firm evidence for such a dose-response association is lacking.¹² Finally, given the conflicting evidence on associations between specific antipsychotic agents and the risk of pneumonia,^6,7,13 well-powered studies examining pneumonia risk related to individual antipsychotics are urgently needed.

Concerning the possibility that properties of antipsychotics moderate pneumonia risk in people taking them, several theories have been proposed. One relates to the dopamine D2 receptor (D2R)–blocking capacity of antipsychotics, resulting in bradykinesia of pulmonary muscles, in turn jeopardizing pulmonary clearance capacity. Alternatively, anticholinergic binding may result in esophageal dilatation and hypomotility,¹⁴ increasing aspiration pneumonia risk. Moreover, sedation may decrease pharyngeal mobility, predisposing individuals taking antipsychotics to aspiration pneumonia. Furthermore, sialorrhea may arise, resulting in higher chances of aspiration pneumonia.¹⁵ In addition, compared with other antipsychotics, secondary antibody deficiency may emerge during treatment with clozapine, with sinopulmonary infections as a possible consequence.¹⁶ This and other properties of clozapine may explain higher incidences of pneumonia observed in clozapine users than in those using other antipsychotics.^6,17 Moreover, many people with schizophrenia (those taking clozapine and those taking other antipsychotics) use tobacco, and smoking increases the risk for pneumonia.¹⁸ Finally, protopathic bias from delirium due to pneumonia and hence higher antipsychotic prescription rates in patients with schizophrenia have also been suggested to explain a (reverse) association between antipsychotic use and pneumonia,⁷ but a lack of age-dependent outcomes does not support this notion.¹⁹

Identifying which antipsychotic agents in what doses are associated with an increased risk of pneumonia in patients with schizophrenia is clinically and scientifically relevant for several reasons. First, pneumonia constitutes the fourth leading cause of death in schizophrenia.²⁰ Identification of antipsychotic drugs that are associated with pneumonia risk may better inform prevention programs (eg, vaccinations). Second, the availability of pneumonia risk estimates for individual antipsychotics and for groups of antipsychotics may foster personalized prescribing guidelines. Third, should an antipsychotic dose-response association with pneumonia be established for specific antipsychotics, patients and clinicians may become more mindful of the possibility of pneumonia when doses are increased. Finally, identification of antipsychotics with specific binding properties that are associated with incident pneumonia may provide pathophysiological clues for understanding the increased incidence of pneumonia in patients taking antipsychotics.

One approach to shed light on comparative adverse drug reaction profiles of treatment options is a large and unselected registry study. Because schizophrenia is a lifelong illness, observational studies with relatively long-term follow-up may enable a more nuanced understanding of specific risks of antipsychotics than clinical trials. The long-term follow-up in registries also allows for within-individual analyses, whereby periods of specific medication use within the same person are contrasted with intervals wherein different compounds are used, thus minimizing risks of confounding due to time-varying variables. Furthermore, patients enrolled in randomized clinical trials do not fully represent the general population with schizophrenia seen in clinical practice.²¹

Given the lack of unequivocal evidence about the associations between specific antipsychotic agents and incident pneumonia¹⁹ as well as the knowledge gap about dose-response associations, we examined pneumonia risks using nationwide registry data from patients with schizophrenia treated in Finland. We hypothesized that due to bradykinetic effects on the pharyngeal phase of swallowing, the use of the D2R-blocking agent haloperidol would be associated with pneumonia risk. We also posited that higher antipsychotic doses and antipsychotic polytherapy would increase risk of developing pneumonia due to the generally higher combined antidopaminergic activity in antipsychotic polytherapy compared with monotherapy.^22,23 To test these hypotheses, we applied within-individual analyses to a large cohort of patients with schizophrenia and schizoaffective disorder, computing pneumonia risk by age and sex for antipsychotics overall and for specific antipsychotics, for antipsychotic monotherapy and polytherapy, and finally for antipsychotics with different levels of anticholinergic burden.

This cohort study was approved by the Ethics Committee of the Finnish National Institute for Health and Welfare, The Social Insurance Institution of Finland, and Statistics Finland. We abided by the Declaration of Helsinki.²⁴ Per applicable local legislation, this registry-based study did not require informed patient consent as no contact was made with the participants.

The study population included all persons who received inpatient care due to schizophrenia (defined as International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] codes F20 for schizophrenia and F25 for schizoaffective disorder) in Finland (N = 61 889) between 1972 and 2014 (eMethods in Supplement 1). Data from all patients aged 16 or older in the registry were included. Hereafter we refer to this group as patients with schizophrenia since no systematic differences in treatment effects between schizophrenia and schizoaffective disorder exist in this registry.²⁵ Data on diagnoses and hospitalizations were obtained from the Hospital Discharge Register, including periods of inpatient care (since 1972) and visits for specialized outpatient care (since 1998). Information on drug use was gathered from the Prescription Register (since 1995), including Anatomical Therapeutic Chemical (ATC) codes, dispensing dates, purchased amount, strength, package size, and drug formulation. Dates and causes of death were obtained from the Causes of Death register. Follow-up for this study started January 1, 1996 (1 year after the start of the Prescription Register). Persons entered the cohort and follow-up started on January 1, 1996, or after the first diagnosis of schizophrenia for those diagnosed between 1996 and 2014. Follow-up ended at the end of data linkage on December 31, 2017, or at death (whichever occurred first).

Data on start and end dates of drug use periods were extracted using the PRE2DUP (From Prescriptions to Drug Use Periods) method²⁶ (eMethods in Supplement 1). For dose-specific analyses, an additional tool for modeling drug use in a time-dependent manner was used to further split use periods according to dose (sum of all antipsychotics used concomitantly).²⁷ As explained in a previous study,²⁸ dose categories were <0.6 of the World Health Organization defined daily dose²⁹ (DDD) per day (hereafter, low dose), 0.6 to less than 1.1 DDDs/d (hereafter, medium dose), and 1.1 or more DDDs/d (hereafter, high dose). Drug use periods were further categorized into monotherapy (1 antipsychotic used) vs polytherapy (≥2 antipsychotics used concomitantly) to examine whether antipsychotic monotherapy and polytherapy were associated with pneumonia risk. The anticholinergic burden of antipsychotics was categorized as low, medium, and high based on previous evidence³⁰ and additional literature searches for antipsychotics not included in this list³⁰ (eMethods and eTable 1 in Supplement 1). The primary outcome was hospitalization due to pneumonia (ICD-10 codes J12-J18) as the main diagnosis for hospital admission (eMethods in Supplement 1).

First, we measured the incidence of pneumonia by 5-year age categories to describe to what extent age, which is an independent risk factor for pneumonia,³¹ is associated with the risk of pneumonia in schizophrenia. Pneumonia incidence was calculated as the number of events per outpatient time (reported as per 100 000 person-years) within each age stratum (from 20-24 years to 85-89 years) by censoring to death and end of data linkage after cohort entry. We calculated pneumonia incidence by age group and sex. To compare incidence rates of pneumonia per age group, we computed incidence rate ratios (IRRs) with 95% CIs, with the age group of 20 to 24 years as a reference.

To examine associations of specific antipsychotics and anticholinergic loading with incident pneumonia, a within-individual study design was used to minimize selection bias, which in the case of pneumonia would most likely be related to inherent differences in the risk of this outcome between people who are prescribed different antipsychotics. Exposure periods of each participant were thus compared with nonexposure periods within the same individual, a strategy with proven benefit to resolve important clinical dilemmas.^32,33 To minimize the risk that variables known to be associated with pneumonia risk could impact the results, these within-individual analyses were adjusted time-dependently for age, concomitant use of other psychotropic drugs, and a modified pneumonia severity index that we created by adapting the widely accepted Pneumonia Severity Index³⁴ according to data availability. To create our modified pneumonia severity index, we extracted all variables used to compute the Pneumonia Severity Index³⁴ that were available in our dataset (eTable 2 in Supplement 1) and computed our modified pneumonia severity index according to the original tool’s instructions.³⁴ Additionally, our modified pneumonia severity index was updated in the model each time any change in medication (specific antipsychotics used or their dosage category [low, medium, or high]) occurred. Use of other psychotropic drugs was defined as follows: antidepressants (ATC code N06A), mood stabilizers (carbamazepine [ATC code N03AF01], lamotrigine [ATC code N03AX09], lithium [ATC code N05AN01], and valproic acid [ATC code N03AG01]), benzodiazepines (ATC codes N05BA and N05CD), and Z-drugs (zopiclone, zaleplon, and zolpidem; ATC code N05CF). Their use was included as time-varying covariates and were updated constantly in the model.

Although smoking habits in patients with schizophrenia are generally stable with very low cessation likelihoods,³⁵ one could hypothesize that in some patients smoking habits may change over time. As the registry lacks information about smoking habits, we performed sensitivity analyses, adjusting all of our models for chronic obstructive pulmonary disease (COPD) as a proxy for smoking because the population attributable fraction of smoking for COPD in Scandinavian countries is 74.6% to 76.2%^36,37 and previous research has found that time-varying smoking exposures are associated with COPD.³⁸ Consequently, sensitivity analyses were adjusted for COPD diagnoses (ICD-10 code J44) during follow-up.

To examine whether anticholinergic loading from agents other than antipsychotics impacted the results, in an additional sensitivity analysis, we adjusted the models for anticholinergic loading as described previously.³⁹ We used the Anticholinergic Cognitive Burden scale, which is a validated scale describing the anticholinergic properties of medications and rating them as 1 for low or minimal anticholinergic activity, 2 for moderate anticholinergic activity, and 3 for strong or definitive anticholinergic activity. We also included biperiden as a strong or definitive anticholinergic, although it is not included in the original categorization. We adjusted for time-varying use of medications other than antipsychotics given an Anticholinergic Cognitive Burden score of 3, also including biperiden as a potent anticholinergic (score 3).³⁹ Finally, we also examined whether the anticholinergic loading scale chosen altered results by applying this scale (which was slightly different from the aforementioned scale used to generate eTable 1 in Supplement 1) to the antipsychotics and repeating the analyses examining associations between categories of anticholinergic burden and pneumonia risk.

Within-individual data were analyzed with stratified Cox proportional hazards regression to calculate adjusted hazard ratios (AHRs) with 95% CIs, an approach that was similar to previous studies, with nonuse of antipsychotics as a reference.^27,40 Data were analyzed from November 4, 2022, to December 5, 2023. Data management and analyses were conducted with SAS, version 9.4 (SAS Institute Inc). As in a previous study,²¹ we refrained from assigning statistical significance to a specific P value cutoff but instead focused on precision estimates using 95% CIs. Thus, P values are provided only as additional measures.

The study population included 61 889 persons with schizophrenia (mean [SD] age, 46.2 [16.0] years at cohort entry), of whom 31 104 were men (50.3%) and 30 785 were women (49.7%). The percentage of women was relatively high in this long-term cohort as men die younger than women.⁴¹

During up to 22 years of follow-up (mean [SD], 14.2 [7.2] years), 8917 patients (14.4%) had at least 1 hospital admission due to pneumonia (mean [SD] age, 63.1 [14.4] years at first admission due to pneumonia). In total, we recorded 15 265 hospitalizations due to pneumonia. Of the 8917 patients admitted for pneumonia, 67.4% (6008 patients) had 1 hospital admission for pneumonia, 17.1% (1527 patients) had 2, and 15.5% (1382 patients) had 3 or more.

In all, 1137 patients with schizophrenia died within 30 days after hospitalization for pneumonia, corresponding to 12.8% of persons with hospital-treated pneumonia and 7.4% of hospitalizations for pneumonia. Mean (SD) age at fatal pneumonia was 71.2 (11.4) years; 337 deaths (30%) occurred in individuals younger than 65 years.

The incidences of pneumonia in patients with schizophrenia by 5-year age categories are shown in Figure 1 and are listed in eTable 3 in Supplement 1. Pneumonia incidence was relatively stable until age 45 to 49 years: relative to the group aged 20 to 24 years, the risk more than doubled for people aged 50 to 54 years (IRR, 2.20; 95% CI, 1.77-2.77). This approximate 50% increase in risk per 5-year increment in age group was relatively stable up to and including the last age group (85-89 years; Figure 1A and eTable 3 in Supplement 1).

When stratified by sex, pneumonia risk was higher in men than in women starting at age 40 to 44 years (IRR, 1.26 [95% CI, 1.06-1.50] relative to women of the same age). Differences in pneumonia risk by sex increased as age increased: in the older age groups, the risk of pneumonia was 2- to 3-fold higher for men compared with women of the same age (Figure 1B and eTable 4 in Supplement 1). For example, at ages 45-49, the IRR for pneumonia in men vs women with schizophrenia was 1.32 (95% CI, 1.16-1.51), while for the group aged 85 to 89 years, the IRR was 2.76 (95% CI, 2.33-3.26) (eTable 4 in Supplement 1).

Any antipsychotic use was not associated with increased pneumonia risk (AHR, 1.12; 95% CI, 0.99-1.26). Compared with no antipsychotic use, low-dose antipsychotic use (AHR, 1.00; 95% CI, 0.88-1.15) and medium-dose antipsychotic use (AHR, 1.10; 95% CI, 0.96-1.26) were not associated with pneumonia risk, while high-dose antipsychotic use was (AHR, 1.20 [95% CI, 1.06-1.37]; P = .005).

Antipsychotic monotherapy was used by 50 797 persons (mean [SD] age, 49.1 [14.7] years; 25 506 women [50.2%]) and polytherapy was used by 41 895 persons (mean [SD] age, 49.6 [13.1] years; 20 729 women [49.5%]), with most people having used both monotherapy and polytherapy during the follow-up period. Compared with no antipsychotic use, antipsychotic monotherapy was associated with increased pneumonia risk (AHR, 1.15 [95% CI, 1.02-1.30]; P = .03), whereas antipsychotic polytherapy was not (AHR, 1.05 [95% CI, 0.92-1.19]; Figure 2). We detected a dose-response association in pneumonia risk associated with monotherapy: only high-dose antipsychotic use was associated with pneumonia (AHR, 1.36 [95% CI, 1.18-1.56]; P < .001), whereas none of the 3 antipsychotic polytherapy dose groups was associated with pneumonia (Figure 2).

No first-generation antipsychotic was associated with pneumonia risk, whereas several high-dose, second-generation antipsychotics were (eTable 5 in Supplement 1). Of specific antipsychotics used in monotherapy according to dose categories, pneumonia risk was highest with the use of high-dose (≥440 mg/d) quetiapine (AHR, 1.78 [95% CI, 1.22-2.60]; P = .003), followed by high- (≥330 mg/d) and medium-dose (180 to <330 mg/d) clozapine (AHR, 1.44 [95% CI, 1.22-1.71]; P < .001 and AHR, 1.43 [95% CI, 1.18-1.74]; P < .001, respectively), and high-dose (≥11 mg/d) olanzapine (AHR, 1.29 [95% CI, 1.05-1.58]; P = .02) (Figure 3A and eTable 5 in Supplement 1).

During follow-up, 24 284 persons used antipsychotics with low anticholinergic loading, 33 233 used antipsychotics with medium anticholinergic loading, and 27 358 used antipsychotics with high anticholinergic loading, implying that most participants had used antipsychotics of 2 or more categories over time. Only the use of antipsychotics with high anticholinergic potency was associated with pneumonia risk (AHR, 1.26 [95% CI, 1.10-1.45]; P < .001). Notably, when evaluating antipsychotics by dose, increased pneumonia risks were observed for high doses of medium-potency anticholinergic antipsychotics (AHR, 1.28 [95% CI, 1.09-1.52]; P = .004), as well as for medium doses (AHR, 1.28 [95% CI, 1.07-1.52]; P = .006) and high doses (AHR, 1.38 [95% CI, 1.18-1.61]; P < .001) of high-potency anticholinergics (Figure 4).

Results from the sensitivity analyses were in line with our main results (Figure 3 and eResults in Supplement 1). Although the pattern and strength of association between high olanzapine dosages and pneumonia were similar in both sensitivity analyses, the 95% CI changed to 1.00-1.51 in the COPD-adjusted sensitivity analysis, while in the second sensitivity analysis the 95% CIs were highly similar to those reported for the main analysis. Similarly, the association of monotherapy antipsychotics as a group with pneumonia remained positive but was not significant in sensitivity analyses. The 95% CIs and significance levels of the other sensitivity results closely matched the main results.

In this cohort study, by analyzing data from 61 889 individuals with schizophrenia, we computed age- and sex-stratified incidence rates and explored associations between antipsychotic drugs and incident pneumonia. We found that pneumonia incidence rose with increasing age, particularly from middle age, and showed how men with schizophrenia are at increased risk of pneumonia after age 40 years relative to women. With regard to groups of antipsychotics, we found that monotherapy was associated with incident pneumonia compared with no antipsychotic use, while polytherapy was not. Moreover, contrary to our hypothesis, we did not find that strong D2R-blocking agents, such as haloperidol, were associated with elevated pneumonia risk. Instead, higher anticholinergic burden was associated with incident pneumonia. With regard to specific antipsychotic agents, the association of high-dose quetiapine with pneumonia was the most pronounced, followed by high- and medium-dose clozapine and high-dose olanzapine.

Possible mechanisms underlying relatively high risks of pneumonia while taking medium- or high-potency anticholinergic antipsychotics include the peripheral and central anticholinergic effects of antipsychotics. Peripheral effects include reduced mucous secretions, leading to a loss in saliva production and consequential weakening of its beneficial antimicrobial properties, thus impairing the body’s natural defense mechanism against airborne pathogens. Other peripheral mechanisms of an increase in pneumonia risk during anticholinergic antipsychotic use include depressed mucociliary transport and reduced lower esophageal sphincter pressure.⁴² With bidirectional associations between cognitive function and pneumonia widely recognized,⁴³ a central mechanism potentially explaining high pneumonia risk when anticholinergic agents are being used relates to their known cognitive adverse effects.^44,45

Our finding of a lower risk of pneumonia associated with antipsychotic polytherapy compared with monotherapy aligns with those of a previous study,⁴⁶ extending these findings to another relevant adverse outcome in schizophrenia: pneumonia. A plausible explanation behind our finding is that high doses of monotherapy incur risks associated with saturated receptor binding properties.⁴⁶

Clinical implications of this work include the possibility of raising awareness about pneumonia risk for specific groups of patients treated with high doses of quetiapine, olanzapine, or clozapine. Relative to periods of nonuse, pneumonia risks moderately increased during use of antipsychotics with medium- and high-potency anticholinergic burden. These findings call for monitoring eating habits and examining swallowing capacity in such patients. Particularly in patients using agents at increasing dosages associated with a higher risk of pneumonia, healthy eating and swallowing may be promoted to reduce aspiration risks, especially in elderly patients with difficulties chewing or swallowing. For example, although rarely performed in most clinical psychiatry settings, information about eating habits may be elicited from caregivers, and swallowing capacity may be examined during physical examinations. Other prevention strategies targeting at-risk groups of patients with schizophrenia include pneumococcal pneumonia, influenza, or SARS-CoV-2 vaccination programs. Future work may encompass studies examining pathophysiological mechanisms, potentially focusing on anticholinergic effects associated with pneumonia in patients with schizophrenia taking clozapine, quetiapine, and olanzapine by leveraging comparative mucociliary transport and lower esophageal sphincter measurements.

Strengths of this study include the large group of unselected participants, the nationwide recording of outcomes, and adjustment for a wide range of covariates in the within-individual Cox proportional hazards regression models. Nonetheless, the findings should be interpreted in the context of several limitations. First, limitations applicable to analysis of registry data have been discussed extensively elsewhere and encompass lack of generalizability to countries with different health care systems, absence of medication adherence assessments, and the possibility of residual confounding affecting some of the results.⁴⁶ For example, we did not have the data to correct for all possible risk factors that may increase pneumonia risk in individuals with schizophrenia, such as smoking and other lifestyle habits. However, we used within-individual analysis in which each individual served as their own control, which allowed us to compare the risk of pneumonia between use of a specific antipsychotic and nonuse within the same individual. Although it is unlikely that patients start or stop smoking at the same moment when their status of medication changes, we conducted a sensitivity analysis by adjusting for COPD, observing that the results did not change. We also adapted the Pneumonia Severity Index³⁴ according to data availability.^47-49

Second, a limitation concerning the interpretation of the anticholinergic burden finding is that agents with high anticholinergic burden also have high antihistaminergic burden. In clinical practice, this fact is less relevant as it is the risk per compound and not mechanisms that is relevant for pneumonia risk and preferential drug prescribing.³³ Third, a strength pertinent to our chosen outcome is that in inpatient settings, a diagnosis of pneumonia is more reliable given possibilities of intense observation and examination. A weakness is that cases of pneumonia not requiring admission could not be included; therefore, our findings possibly generalize only to cases of relatively severe pneumonia. Finally, within-individual analyses cannot be conducted for mortality; consequentially, findings related to fatal pneumonia should be interpreted with caution due to potential confounding by older age and higher morbidity levels associated with a higher risk of death during periods of antipsychotic nonuse.

In contrast to previous research reports^6,17 and statements in certain guidelines,⁵⁰ we found that antipsychotic agents associated with incident pneumonia include not only clozapine (in dosages ≥180 mg/d) but also quetiapine (≥440 mg/d) and olanzapine (≥11 mg/d). Moreover, monotherapy antipsychotics and antipsychotics with high anticholinergic burden are associated with increased pneumonia risk in a dose-dependent manner. These findings call for prevention strategies in patients with schizophrenia who take these antipsychotics to curtail the incidence of pneumonia.

Accepted for Publication: April 15, 2024.

Published Online: June 26, 2024. doi:10.1001/jamapsychiatry.2024.1441

Corresponding Author: Jurjen J. Luykx, MD, PhD, Department of Psychiatry, Amsterdam University Medical Center, location VUmc, De Boelelaan 1117, Amsterdam 1081 HV, the Netherlands (j.j.luykx@amsterdamumc.nl).

Author Contributions: Dr Taipale had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Luykx, Correll, Manu, Tiihonen, Taipale.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Luykx, Manu, Hasan.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Tanskanen, Taipale.

Administrative, technical, or material support: Luykx, Tanskanen.

Supervision: Luykx, Correll, Hasan.

Conflict of Interest Disclosures: Prof Correll reported receiving grants from Janssen and Takeda Pharmaceuticals and also reported serving as a consultant or advisor to or receiving honoraria from AbbVie, Acadia, Adock Ingram, Alkermes, Allergan, Angelini, Aristo, Biogen, Boehringer Ingelheim, Bristol Myers Squibb, Cardio Diagnostics, Cerevel, CNX Therapeutics, Compass Pathways, Darnitsa, Delpor, Denovo, Gedeon Richter, Hikma, Holmusk, Intra-Cellular Therapies, Jamjoom Pharma, Janssen/Johnson & Johnson, Karuna, LB Pharma, Lundbeck, MedAvante-ProPhase, Medincell, Merck, MindPax, Mitsubishi Tanabe Pharma, Mylan, Neurocrine, Neurelis, Newron, Noven, Novo Nordisk, Otsuka, Pharmabrain, PPD Biotech, Recordati, Relmada, Reviva, Rovi, Sage, Seqirus, SK Life Science, Sumitomo Pharma America, Sunovion, Sun Pharma, Supernus, Tabuk, Takeda, Teva, Tolmar, Vertex, and Viatris outside the submitted work; providing expert testimony for Janssen and Otsuka; serving on data safety monitoring boards for Compass Pathways, Denovo, Lundbeck, Relmada, Reviva, Rovi, Supernus, and Teva; receiving grant support from Janssen and Takeda; receiving royalties from UpToDate; and holding stock options in Cardio Diagnostics, Kuleon Biosciences, LB Pharma, MindPax, and Quantic. Dr Tanskanen reported receiving institutional support from Eli Lilly and Janssen outside the submitted work. Prof Hasan reported receiving personal fees from AbbVie, Advanz Pharma, Janssen, Lundbeck, Recordati, Rovi, and Otsuka outside the submitted work; serving as editor of the German schizophrenia treatment guidelines and as first author of the World Federation of Societies of Biological Psychiatry schizophrenia treatment guidelines; and serving on advisory boards of Janssen, Lundbeck, Recordati, Rovi, and Otsuka. Prof Tiihonen reported receiving grants from Janssen-Cilag paid to his employing institution and personal fees from HLS Therapeutics, Inc, Janssen, Orion Pharma Ltd, Teva, WebMed Global, Lundbeck, and Otsuka outside the submitted work. Dr Taipale reported receiving grants from the Academy of Finland, Janssen, and Eli Lilly and personal fees from Gedeon Richter, the Sigrid Juselius Foundation, Janssen, Lundbeck, and Otsuka outside the submitted work. No other disclosures were reported.

Funding/Support: This study was funded by the Finnish Ministry of Social Affairs and Health through the developmental fund for Niuvanniemi Hospital.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.