Discussion

We did not find transplantation (SOT and HSCT), malignant tumor, or primary immunodeficiencies to be independently associated with MRSA or P. aeruginosa isolation from respiratory cultures. Instead, our findings reinforce prior evidence that influenza, substance use disorder, and lung abscess/empyema are associated with MRSA, whereas bronchiectasis and COPD are associated with P. aeruginosa.^{Reference Arancibia, Bauer and Ewig11–Reference Wooten and Winston16} These findings are reassuring for ASPs at centers with large transplant and oncology populations where optimizing use of BSAs is often challenging.

Our study adds to a limited but growing body of evidence that immunocompromising conditions are not independently associated with MRSA or P. aeruginosa in CAP.^{Reference Di Pasquale, Sotgiu and Gramegna17,Reference Saravolatz, Gandhi and Vaughn18} A secondary analysis of the Global Initiative for MRSA Pneumonia (GLIMP) database found that, while P. aeruginosa CAP was more prevalent among immunocompromised patients, immunocompromise was not a risk factor after multivariate adjustment for confounders.^{Reference Di Pasquale, Sotgiu and Gramegna17} In a separate observational study of moderately immunocompromised patients who did not meet IDSA/ATS criteria for BSAs, no mortality difference was found between those treated with standard versus BSA.^{Reference Saravolatz, Gandhi and Vaughn18} Both of these studies evaluated immunosuppressive therapies, such as corticosteroids, in addition to immunocompromised conditions, such as malignant tumor and transplantation.

We observed a higher prevalence of P. aeruginosa (5.5%) compared with prior studies (1.2–2.1%), likely reflecting selection for a sicker cohort due to the requirement for respiratory cultures.^{Reference Saravolatz, Gandhi and Vaughn18–Reference Jones, Jones and Huttner20} This is supported by a high ICU admission rate (41.6%). In studies with high rates of respiratory cultures (>85%), P. aeruginosa has been reported to be around 4%.^{Reference Di Pasquale, Sotgiu and Gramegna17,Reference Cilloniz, Gabarrus and Ferrer21} Notably, the isolation of MRSA or P. aeruginosa in respiratory culture was not associated with ICU admission in a secondary analysis. Collectively, these findings suggest that immunocompromising conditions alone may be insufficient to guide empiric BSAs in CAP.

Two unexpected findings were that PUD was associated with both MRSA and P. aeruginosa, and liver disease was inversely associated with P. aeruginosa. Observational data have demonstrated associations between acid-suppressive therapies and pneumonia, as well as increased intestinal colonization with multidrug-resistant organisms (MDROs).^{Reference Laheij, Sturkenboom, Hassing, Dieleman, Stricker and Jansen22,Reference Willems, van Dijk, Ket and Vandenbroucke-Grauls23} However, an association of acid-suppression with MDROs, including P. aeruginosa and MRSA, within the context of CAP has not been evaluated. Given that we did not capture whether patients were on acid-suppressive therapies, these findings would need to be formally evaluated as the association likely reflects residual confounding. For liver disease, chronic fluoroquinolone exposure could theoretically influence P. aeruginosa CAP risk. However, we did not capture whether these patients were on spontaneous bacterial prophylaxis.

Our study has some important limitations. First, we relied on ICD-10 codes and the CCI to identify immunocompromising conditions. This limited our ability to evaluate all the conditions qualifying for immunocompromise in CAP as outlined by CHEST, specifically those related to pharmacotherapies.^{Reference Ramirez, Musher and Evans24} These pharmacotherapy-related conditions were evaluated in the studies by Di Pasquale and Saravolatz and were not found to be independently associated with MRSA or P. aeruginosa, as previously mentioned.^{Reference Di Pasquale, Sotgiu and Gramegna17,Reference Saravolatz, Gandhi and Vaughn18} However, with increased use of immunosuppressing agents since those reports, this merits further study. Since we did not capture antimicrobial exposures, we are unable to comment on the utilization or appropriateness of BSAs within our cohort. We also did not capture the timing of CAP hospitalization relative to date of transplantation as this may influence infection risk profiles. Our findings may not be generalizable to other institutions, as our cohort had a high ICU admission rate—likely reflecting higher acquisition of respiratory cultures, which are more commonly obtained in the ICU setting. Additionally, we did not stratify our analysis to evaluate temporal fluctuations in respiratory viruses, such as influenza or COVID-19. However, neither respiratory virus was found to have a significant association with either MRSA or P. aeruginosa pneumonia besides influenza with MRSA pneumonia [aOR 7.64]. Lastly, since we included only patients who had respiratory cultures, we may have missed patients with MRSA or Pseudomonas CAP who did not have respiratory cultures obtained.

In conclusion, our results support prior studies that did not find immunocompromising conditions to be associated with MRSA or P. aeruginosa in patients hospitalized with CAP. Our results affirm previously demonstrated risk factors for MRSA and P. aeruginosa in CAP and add additional confidence that immunocompromising conditions evaluated in this cohort were not associated with MRSA or P. aeruginosa CAP. Overall, these findings support an individualized approach to empiric CAP therapy, prioritizing established clinical risk factors over immunocompromised status alone.