Abstract
Dear Editor:
P
Observational studies in PLWHIV with acute infections highlight the potential for improved outcomes with timely cART treatment due to attenuation of immune dysregulation with acute viral suppression. 2,4 –6 Ongoing HIV replication drives immune activation, which is increased in PLWHIV not on antiretroviral therapy. 9 Activated T cells are known to worsen outcomes in laboratory models of sepsis. 10 cART decreases immune activation and helps normalize inflammation. 9
While long-term mortality benefits of cART are clearly known, few studies have focused on patients with acute severe infections or critical illness. Prior studies of cART in hospitalized patients have not necessarily focused on sepsis, nonopportunistic infections, or survival outcomes. 2,6,7 We aimed to examine the clinical course of acute septic shock in PLWHIV with the hypothesis that early cART treatment is associated with improved in-hospital outcomes.
With this aim, we undertook a retrospective cohort study to examine cART treatment in sepsis. Our healthcare system has a treatment pathway for patients with acute septic shock. Enrollment in the pathway requires suspicion of infection and systolic blood pressure <90 mm Hg after 20 mL/kg of intravenous fluids or serum lactate ≥4.0 mmol/L. A dedicated prospective registry of pathway patients is maintained for quality improvement purposes and was used to identify subjects for this study. All study subjects were admitted through the emergency department (ED) at one of seven hospitals in the Carolinas Healthcare Network within metropolitan Charlotte, NC. The Institutional Review Board and Privacy Board of Carolinas Healthcare System approved this study under waiver of informed consent.
Adult patients (age >18) enrolled in the sepsis pathway between January 2014 and September 2016 with an International Classification of Diseases, Ninth Revision, diagnosis of HIV were eligible. Patients admitted more than once for discrete clinical events were included as separate sepsis encounters. A standardized data collection instrument and guidance tool was developed for data collection and medical record review. Review and data abstraction were performed by two authors and data discrepancies resolved through verbal mediation. Data were specifically collected on receipt of cART both before and during admission, type of admission, and laboratories, including viral suppression and CD4 count within 6 months of admission.
Descriptive statistics were calculated for clinical measures stratified by mortality. Logistic regression was used for the primary analysis, testing mortality between patients who received cART prescription versus patient who did not receive cART prescription. Additionally, logistic regression was also used to assess factors which impacted mortality, and corresponding odds ratios were also calculated. Finally, multivariable logistic regression was used to assess a combination of factors influencing mortality. We limited the number of factors considered in the model to two due to the small sample size. SAS Enterprise Guide 6.1 software was used for the analysis assuming a two-tailed alpha of 0.05.
During the study period, 6415 patients were enrolled in the ED septic shock clinical pathway. HIV was identified as a comorbid factor in 78 acute septic shock patients; 67 patients had all data available for analysis in the study with 11 patients with missing data points, all with unknown HIV viral loads around admission. PLWHIV admitted for sepsis had a 35% inpatient mortality rate compared with an overall inpatient mortality rate of 17% for all patients in our institutional sepsis pathway database.
Most patients (56%) had a nonopportunistic infection identified as their source of sepsis with only 25% presenting with an opportunistic infection. The remaining 19% of patients did not have an infection identified. For those with opportunistic infections, disseminated mycobacterium avium complex (MAC) was most common (six patients), followed by cytomegalovirus (CMV) (three patients), pneumocystis pneumonia (three patients), tuberculosis (TB) (three patients), and one patient each with cryptococcal meningitis, histoplasmosis, Kaposi's sarcoma, mycobacterium kansasii, progressive multifocal leukoencephalopathy (PML), Salmonellosis, Toxoplasmosis, and vericella zoster (VZV) meningoencephalitis. The nonopportunistic bacterial pathogens recovered in patients were E. coli (six patients), Staph aureus and various streptococcal species (five patients each), pseudomonas and C difficile (two patients each), and Klebsiella (one patient). For these pathogens, pneumonia was the most common site of infection (19 patients) followed by the gastrointestinal (9 patients) and urinary tracts (7 patients), skin and soft tissue infection and primary bacteremia (each with 5 patients each), and meningitis (1 patient).
Only 55% of patients received inpatient cART. For 26 patients who were virologically suppressed on cART at admission, 7 patients had cART held on admission. The most common reasons for stopping ARVs on admission were acute renal failure with concern for medication-induced renal toxicity, and the inability to swallow pills due to gastrointestinal dysfunction or intubation. Of 41 patients not virologically suppressed on admission, 21 patients were prescribed inpatient cART. However, 13 of these patients were on cART before admission just not virologically suppressed. Only seven individuals were initiated on cART while inpatient, but two of these patients received just one dose of ARVs (Table 1).
cART, combination antiretroviral therapy.
Factors predictive of mortality in univariate analysis were baseline CD4 count, virologic suppression at admission, APACHE score, presence of opportunistic infection, admission to a tertiary care hospital, and inpatient prescription of cART (Table 2).
cART, combination antiretroviral therapy; CI, confidence interval; VL, viral load.
Patients not on cART while an inpatient had four times odds of in-hospital mortality controlling for impact of HIV viremia on admission in a multivariable analysis [odds ratio (OR) = 4, 95% confidence interval (CI) 1.3–12, p = 0.01]. In a separate multivariable analysis, patients not prescribed cART had 3.3 times greater odds of mortality by discharge after controlling for CD4 count (OR = 3.3, 95% CI 1–10, p = 0.04).
While many variables impact survival in PLWHIV admitted with sepsis, including baseline immunologic and virologic status, type of infection, and admission to a tertiary care hospital, prescription of cART while inpatient is a significant predictor of survival at discharge. This represents a great opportunity to improve care and potentially decrease mortality outcomes in this population as nearly half of PLWHIV admitted with sepsis do not receive cART while inpatient.
The greatest barriers to administration of ARVs in hospitalized patients both in our and other published cohorts include need to give medications crushed down enteral feeding tubes, drug interactions, and potential for renal toxicity in patients with organ failure. Recent advances in medical therapy should allow management of these concerns with modern cART being less toxic with less impact on renal function and fewer drug interactions than in the past. There is also increasing data on safety and efficacy of crushing many ARVs. 11 Our experience demonstrates that even in the difficult to treat a population of PLWHIV coinfected with mycobacteria, in which recent literature suggests poor medication tolerance, 12 cARVs can be given with 5 of our 10 coinfected patients receiving inpatient cART in approximate equivalence to our overall inpatient cART prescription rate. This should allow clinicians to prescribe cART to even the sickest patients admitted with sepsis. Inpatient prescription of cART for PLWHIV admitted with sepsis should be strongly considered to improve patient outcomes.
Footnotes
Acknowledgments
The authors would like to thank Colleen Karvetski for her invaluable assistance with data retrieval for this project, and Jing Zhao for her work in final preparation of statistics and tables on the article.
Author Disclosure Statement
One author, Christopher Polk, has served on Viiv and Gilead Advisory Boards and participates in clinical trials funded by Gilead and Viiv. There was no funding supporting research described in this article.