Text version
Hi. Jim Phelps here. Now, let’s look at the risk of pneumonia associated with the use of antipsychotic medications. Well, antipsychotics have plenty of risks associated with their use, including for nearly all of them weight gain and increased cardiovascular risk. And in older patients, one can add the risk of sedation and associated falls and even death, as noted in the FDA black box warning for death associated with antipsychotics in that older population. So, presumably, already without looking at yet another article, we’re already only prescribing antipsychotics when we have no alternatives and we feel we absolutely must, right? In that case, then, adding another risk shouldn’t really change our clinical practice very much. So, the not very surprising or not clinically practice-changing result of a 2018 review on pneumonia and antipsychotics is that, yes, there is an association there, especially in the elderly, although not exclusively. But the risk isn’t really big enough to change our clinical practice, I think. And that’s reflected in the statistic that they cite, which is number needed to harm. So, actually, what I think is useful in this article is an opportunity to think through number needed to harm (the concept) and look at this as an example and compare it some other risks.
So, for example, number needed to harm for pneumonia associated with antipsychotics is 86 and actually may be even as high as 1100 depending on which background rate one uses. I’ll just remind you that number needed to harm is based on the incidence of the problem in those who are exposed to the treatment minus the risk in those unexposed. And if that latter figure varies from study to study, like the incidence of pneumonia in the general population, the number needed to harm figure will vary. And that’s why there’s a difference of as much as 86 all the way up to 1100 in this review. Simply put, we can say that number needed to harm for antipsychotics causing pneumonia is a big number—86 at least and maybe much higher than that.
By comparison, let’s play with NNHs a little bit. Olanzapine: Weight gain associated with olanzapine; number needed to harm is 6. For sedation and somnolence associated with quetiapine, that’s 5. So, I’m sure you, like me, have used a ton of those medications. I would’ve thought the number needed to harm would be more like 2, meaning nearly everybody who gets the medication gets the side effect. But remember, number needed to harm is calculated using the incidence in the treated population versus a comparison group. In this case, in a randomized trial, that’s the placebo group. So, for things like weight gain or sedation, which might be relatively common in the placebo group, that will tend to inflate number needed to harm. The sedation, for example, might be coming from the placebo. So, before we can attribute it to quetiapine, we would have to see more cases. So, the number needed to harm, the number needed to reliably attribute the side effect to quetiapine, would be larger. The point here is that number needed to harm for really common things like those side effects is in the single digit.
By comparison, how about akathisia caused by lurasidone? Number needed to harm there is 15, so not as common as sedation or weight gain but not up there in the 25 range, which is where we might start to see lamotrigine and significant rash, for example. Now, by comparison, remember the number needed to harm for haloperidol and sudden death is 8. So, number needed to harm allows you to kind of ballpark the risk of a medication for a particular risk. And in the case of pneumonia here, we can say that even though there is a causal association that’s demonstrated in this review, it’s not among the big risks of antipsychotics.
Thank you for following me as I try to drag you back through number needed to harm.
Abstract
Antipsychotic Drug Use and Pneumonia: Systematic Review and Meta-Analysis
O Dzahini, N Singh, D Taylor, and PM Haddad
Objectives: The purpose of this study was to investigate the association of antipsychotic exposure to the incidence and mortality of pneumonia.
Methods: The design of this study involved meta-analysis of observational studies identified from electronic databases.
Results: In total, 19 studies were included in the systematic review and 14 in the meta-analysis. Risk of pneumonia was increased by first-generation antipsychotics (risk ratio 1.69, 95% confidence interval 1.34–2.15; five studies), second-generation antipsychotics (risk ratio 1.93, 95% confidence interval 1.55–2.41; six studies) and all antipsychotics (risk ratio 1.83, 95% confidence interval 1.60–2.10; seven studies) compared with no antipsychotic use. Pneumonia risk did not differ in seven studies comparing first-generation antipsychotics with second-generation antipsychotics (risk ratio 1.07, 95% confidence interval 0.85–1.35). Case fatality rate was not different in pneumonia cases associated with antipsychotic exposure versus cases without exposure (risk ratio 1.50; 95% confidence interval 0.76–2.96; two studies). All antipsychotics with data from ⩾2 studies allowing meta-analysis, were associated with a significantly increased pneumonia risk (i.e. haloperidol, olanzapine, clozapine, risperidone, quetiapine, zotepine).
Conclusion: Exposure to both first-generation antipsychotics and second-generation antipsychotics is associated with an increased pneumonia risk. Clinicians need to be vigilant for the occurrence of pneumonia in patients commencing antipsychotics, especially those with other risk factors for pneumonia including older age, chronic respiratory disease, cerebrovascular disease, dysphagia and smoking.
Download PDF and other files
Reference
Dzahini, O., Singh, N., Taylor, D., & Haddad, P. (2018). Antipsychotic drug use and pneumonia: Systematic review and meta-analysis. Journal of Psychopharmacology, 32(11), 1167-1181
