Epidemiology, Clinical Characteristics, and Diagnostic Testing Practices for Pneumocystis Pneumonia–Associated Hospitalizations, United States, 2019–2022 Open Access

Abstract

Among 9196 hospitalizations involving Pneumocystis pneumonia, those without HIV had higher in-hospital mortality (24.3% vs 10.5%, P < .001) when compared with those with HIV. These findings underscore the continued importance of Pneumocystis pneumonia clinical awareness and the need for comprehensive prophylaxis guidance, particularly for certain patients without HIV who are immunosuppressed.

Pneumocystis pneumonia (PCP) is a life-threatening opportunistic fungal infection associated with approximately 10 500 hospitalizations and >400 deaths in the United States annually [1, 2]. Although often considered primarily an AIDS-defining illness, PCP in the United States affects patients with immunosuppressive conditions besides advanced HIV infection as well, with patients with PCP who are non-HIV immunosuppressed experiencing greater diagnostic delays and poorer outcomes [3]. Diagnosis is challenging because the causative organism (Pneumocystis jirovecii) is nonculturable and detecting the organism with staining and direct microscopy of bronchoalveolar lavage fluid or induced sputum has low sensitivity [4]. Polymerase chain reaction (PCR) is more sensitive than staining methods [4, 5].

Previous studies of PCP-related US hospitalizations focused on patients with HIV [6], relied on data that were nearly a decade old [6–8], and lacked information on testing and treatment practices. Recent data on PCP hospitalizations are limited. Therefore, we used a large health care services database to describe features of hospitalized patients with PCP during 2019 to 2022.

METHODS

The PINC-AI Healthcare Database (PHD) contains deidentified HIPAA-compliant data (Health Insurance Portability and Accountability Act) on health care utilization, financial information, and pharmacy usage submitted by >1100 diverse nonprofit, nongovernmental, and community and teaching hospitals throughout the United States [9]. Laboratory data (including in-house and send-out testing) are available from ∼25% of hospitals in the PHD. We limited analyses of PCP-related testing to hospitals and months in which the hospital reported ≥1 laboratory test to the PHD. We used ICD-10-CM discharge diagnosis codes to identify PCP and associated conditions (Supplementary Table 1) and billing data to identify PCP-related medications. We described features of PCP hospitalizations and compared those with and without HIV using chi-square, Fisher exact, and Student t tests (α = .05). We performed logistic regression to assess differences in in-hospital mortality between PCP-related hospitalizations with and without HIV, adjusting for age, sex, race/ethnicity, and receipt of medication for PCP.

RESULTS

A total of 9196 PCP-related hospitalizations occurred at 780 hospitals. Of those, 5536 (60.2%) had HIV listed as a diagnosis (Table 1). Patients with PCP hospitalizations and HIV were more likely to be younger (mean, 44.7 vs 62.9 years; P < .001), male (73.6% vs 53.4%, P < .001), non-Hispanic Black (46.9% vs 12.6%) or Hispanic/Latino (18.1% vs 11.4%, P < .001), and hospitalized in the South (50.4% vs 35.2%, P < .001) as compared with those without HIV. Diabetes (33.3%), long-term use of steroids (32.7%), immunosuppressive disorders besides HIV (32.2%), chronic obstructive pulmonary disease (25.2%), hematologic malignancy (24.7%), and solid organ malignancy (22.0%) were common conditions among PCP hospitalizations without HIV.

Among 2624 (28.5%) PCP hospitalizations at 207 hospitals with laboratory data available, 813 patients (31.0%) were tested with a PCP-specific laboratory test; of those, 478 (58.8%) were positive (PCR, n = 296; direct fluorescent antibody, n = 157; microscopy, n = 36). Patients from 753 (28.7%) hospitalizations were also tested for (1,3)-β-d-glucan; of those, 340 (45.2%) were positive. When compared with PCP hospitalizations without HIV, those with HIV less frequently had PCP-specific testing (29.3% vs 33.2%, P = .032) and a positive PCP-specific test result (16.8% vs 20.1%, P = .029; Supplementary Table 2).

Patients with PCP hospitalizations and HIV were more likely than those without HIV to receive any medication for PCP (96.4% vs 90.2%, P < .001), specifically trimethoprim-sulfamethoxazole (83.4% vs 73.6%, P < .001), but were less likely to receive atovaquone (22.1% vs 24.6%, P = .005), clindamycin (13.4% vs 15.5%, P = .005), and primaquine (11.0% vs 12.7%, P = .011). Time from hospital admission to first receipt of PCP medication was longer for patients without HIV (mean, 5.7 vs 2.5 days; P > .001).

PCP hospitalizations without HIV were longer (mean, 15.7 vs 12.5 days; P < .001) and more frequently involved intensive care unit stay (67.2% vs 52.0%, P < .001), acute respiratory failure (69.2% vs 54.8%, P < .001), mechanical ventilation (31.3% vs 18.4%, P < .001), and in-hospital death (24.3% vs 10.5%, P < .001) as compared with those with HIV. The association between PCP hospitalizations without HIV and higher mortality persisted after adjusting for age, sex, race/ethnicity, and receipt of PCP-related medication (odds ratio, 1.90; 95% CI, 1.60–2.25).

DISCUSSION

This analysis of a large hospitalization database demonstrates that PCP remains a serious public health issue given its high morbidity and mortality among patients with HIV and those with other immunosuppressive conditions. Our findings underscore the importance of continued efforts to engage all people with HIV in care and treatment, including PCP prophylaxis when indicated [5]. Our results also support the critical need to prevent, diagnose, and treat PCP in people without HIV who are immunosuppressed given the poor hospitalization outcomes in this population.

Malignancy (43%) was the most common underlying condition among patients without HIV, similar to a previous national study of PCP hospitalizations [8]. However, that study, which spanned 2005 to 2014, did not find differences in mortality between patients with and without HIV, which could be related to differences in study design (use of primary vs all-listed discharge diagnosis codes) or could reflect true differences in the changing epidemiology of PCP. Patients with PCP and without HIV may experience diagnostic delays when compared with those with HIV, and they appear to be prescribed prophylaxis less frequently [3], which might be mirrored by our finding longer time from admission to PCP-related medication among patients without HIV. Yet, a meta-analysis of risk factors for mortality among patients with PCP but without HIV found that receipt of PCP prophylaxis and time from symptom onset to treatment were not associated with mortality; rather, predictors included older age, solid tumors, and invasive ventilation [10]. Considering in-hospital deaths (24.3%) plus discharges to hospice (6.4%), the estimated overall mortality rate among patients without HIV in our study (30.7%) was identical to this meta-analysis, despite differences in study design, setting, and time frame [10]. Overall, the severity of illness associated with PCP among patients without HIV highlights the importance of considering prophylaxis for certain groups; a Cochrane review indicated that PCP prophylaxis might be beneficial for persons without HIV who are immunosuppressed when the risk of PCP exceeds 6.2% [11]. The risk for potential drug-related adverse events must also be considered. Further studies are needed to quantify the risk for PCP among specific groups of persons without HIV who are immunosuppressed.

The low PCP testing rate (31%) that we observed may reflect empiric diagnoses as well as the challenges associated with confirming PCP diagnosis, such as the collection of deep respiratory samples in patients with critical illness [4]. It is also possible that patients received PCP-specific testing as outpatients, which would not be reflected in this analysis of hospital data and could result in underestimation of PCP testing rates. The lower PCP-specific testing rate among PCP hospitalizations in patients with HIV might reflect greater clinician comfort with empiric diagnosis of PCP in patients with HIV (vs those without HIV). Continued monitoring of PCP testing practices and implementation of newer technologies will be important for informing guidance about diagnosis and treatment.

To our knowledge, national data on PCP testing practices have not been previously described. PCR was the most common test type in this analysis and may be growing in popularity because of its rapid turnaround time, although it cannot easily distinguish between colonization and infection [4]. Over one-quarter of patients with laboratory data available were also tested for (1,3)-β-d-glucan, which is sensitive but not specific for PCP [4]. Given the challenges with laboratory detection of Pneumocystis, metagenomic next-generation sequencing may be a promising option [12]. Continued research efforts to improve PCP diagnostic testing are warranted to ensure that patients receive timely treatment and that unnecessary exposure to antimicrobials is minimized.

A main limitation is that PHD data are a convenience sample of hospitalizations nationwide and may not necessarily be representative of all hospitalizations involving PCP. In particular, the subset of PHD data with laboratory data available underrepresents the West and the Northeast, potentially skewing our findings about laboratory test types used to diagnose PCP. In addition, ∼15% of race/ethnicity data were missing, and medical coding data are inherently subject to misclassification. Patients with PCP-related hospitalizations (as defined by the presence of a PCP ICD-10-CM code) who had a negative PCP-specific laboratory test result may not have truly had PCP. Last, PHD data do not include medication indication, so we were unable to determine whether PCP-related medications were for prophylaxis or treatment.

Despite these limitations, this analysis allowed us to examine of a large number of PCP hospitalizations, demonstrating the need for continued attention to PCP prevention and clinician awareness for patients with and without HIV.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. This activity was reviewed by the Centers for Disease Control and Prevention (CDC) and was conducted consistent with applicable federal law and CDC policy (eg, 45 CFR part 46.102[l][2], 21 CFR part 56; 42 USC §241[d]; 5 USC §552a; 44 USC §3501 et seq). PHD data are fully deidentified, so this analysis was not subject to review by the CDC institutional review board. No specific funding was received for this work.

Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.

Patient consent statement. This study did not include factors necessitating patient consent.

References

Author notes