Pneumonia may be more frequent and have more fatal outcomes with clozapine than with other second‐generation antipsychotics

In 2005, the US Food and Drug Administration (FDA) requested a warning in the package insert of second-generation antipsychotics (SGAs), indicating that the use of those medications was associated with increased mortality in elderly people with behavioral disturbances. One of the causes of death mentioned in the FDA report was pneumonia. Since then, evidence has progressively suggested that patients on clozapine are particularly prone to developing pneumonia. In Taiwan, Hung et al1 found that clozapine was the only antipsychotic associated with a clear dose-dependent increase in the risk for recurrent pneumonia (adjusted risk ratio = 1.40), with patients re-exposed to the drug having a higher risk than those receiving it only prior to the baseline pneumonia. In a 25-month retrospective US study, pneumonia occurred in 34% of clozapine patients (odds ratio = 4.07 compared with the general population), whereas there was no significantly increased risk of pneumonia associated with the use of risperidone2. In a 12-year study in another US hospital, pneumonia was the top cause of medical admissions in clozapine patients (19% of cases)3. The increased risk of pneumonia from clozapine may be explained by swallowing disturbances common to all SGAs; clozapine's increased risk of sedation and hypersalivation; and clozapine's not-well-understood effects on the immune system4. Pneumonia during clozapine treatment may be particularly lethal. In a British drug discontinuation study, five deaths due to pneumonia were found in 529 clozapine patients, and none among 250 patients on long-acting risperidone5. In the Danish registry, Rodhe et al6 explored myocarditis within 2 months of 3,262 clozapine initiations in outpatients. None of the 26 deaths was caused by myocarditis, while pneumonia caused 7 deaths (2.1 per 1,000 patients), making it by far the primary cause of death. Since 2002, many cases of clozapine intoxication during pneumonia or other severe infections have been published. A systematic literature review through 2016 identified 40 cases7. Inflammation releases cytokines which inhibit the main metabolic enzyme of clozapine, CYP1A2, which increases clozapine levels. Clozapine has a narrow therapeutic range, and severe inflammation during pneumonia can cause clozapine intoxication, possibly leading to death by further increasing the risk of hypersalivation, sedation, aspiration, or even arrhythmia4. Here we provide new data obtained from VigiBase, the World Health Organization (WHO)'s global database, which receives spontaneously reported cases of suspected adverse drug reactions from 134 countries around the world, and currently includes nearly 20 million reports. On April 9, 2019, we searched the WHO database for reports of pneumonia related to clozapine and three other SGAs, the most frequently used worldwide – risperidone, quetiapine and olanzapine. For clozapine we found 4,865 reports of pneumonia, which corresponds to 3.5% of all clozapine reports in the database and exceeds the 1,195 expected if clozapine had followed the general reporting in the database (0.9% of the reports relating to pneumonia). A standard statistical analysis for these data, combining statistical shrinkage and Bayesian confidence intervals for the observed-to-expected ratio, indicates a robust statistical association (p<0.001). Moreover, such association was observed in separate analyses for all adult age groups and across the Americas, Asia, Europe and Oceania, reducing the likelihood of being due to report artifacts or case duplication. In contrast, there were fewer reports than expected in VigiBase for the other SGAs: 393 vs. 845 for risperidone, 622 vs. 650 for quetiapine, and 493 vs. 529 for olanzapine. More importantly, the number of pneumonia reports with fatal outcomes was dramatically higher for clozapine (1,577) than for the other SGAs (141, 105 and 147). As the 1,577 fatal outcomes from pneumonia appeared very high, we compared this finding with other reported fatal outcomes associated with clozapine. The second reported fatal outcome was cardiac arrest, with 943 cases. Only 212 fatal outcomes were reported in 4,775 agranulocytosis cases. We found 144 fatal outcomes in 2,694 myocarditis reports. Our analyses of the WHO database highlight pneumonia as a possible major cause of death in patients on clozapine, but alternative explanations for the observed reporting patterns must be considered. Risperidone, quetiapine and olanzapine are each prescribed much more frequently than clozapine worldwide, but the total number of reports in VigiBase is highest for clozapine. This may reflect the closer monitoring of patients on clozapine, although it cannot explain the higher proportion of reports on pneumonia, including associated fatal outcomes. On the other hand, clozapine is typically used in more treatment-refractory patients, who may be more severely ill, so it cannot be ruled out that these patients are at greater risk of developing or dying from pneumonia4. The literature clearly suggests long-term clozapine treatment to be less frequently associated with all-cause mortality than other antipsychotic use8, but preventing deaths from pneumonia may further reduce mortality, since pneumonia may be killing many more clozapine patients than agranulocytosis or myocarditis, which are emphasized in the package insert. If other studies, particularly from the reliable Scandinavian registries, confirm and extend the Danish finding that pneumonia starts killing clozapine patients within the first 2 months of treatment, there may be a need to warn about pneumonia in the clozapine package insert. Until then, two simple measures may help clinicians decrease the mortality risk in clozapine patients during pneumonia. First, patients and families should be told to call the psychiatrist if the patient develops a fever or any obvious sign of a severe infection. Second, if the patient has fever or an elevation in C-reactive protein, the clinician should pay attention to any sign of clozapine intoxication (including sedation, hypersalivation or myoclonus) and urgently measure trough (early morning before medication) clozapine levels to rule out clozapine intoxication. In the absence of rapid access to clozapine levels, the psychiatrist may consider halving the clozapine dose until the infection/inflammation has resolved.