By Betty Tran, MD, MSc, Editor

SYNOPSIS: Using a latent profile analysis in observational cohort studies of patients hospitalized for sepsis, investigators identified subtypes of patients based on inpatient healthcare facility use in the year prior to sepsis hospitalization and correlated to 90-day mortality.

SOURCE: Prescott HC, et al. Paths into sepsis: Trajectories of presepsis healthcare use. Ann Am Thorac Soc 2019;16:116-123.

Sepsis is a heterogeneous condition, making it difficult to identify with certainty and treat successfully in many cases. As such, it remains a significant public health problem. Neither the incidence nor combined outcome of death or discharge to hospice changed much between 2009 and 2014, despite advances in medical knowledge, better clinical awareness, and changes in definitions and timely management.1

To elucidate clinically relevant contributors to sepsis heterogeneity, Prescott et al hypothesized that differences in sepsis outcomes may result from differences in a patient’s clinical course leading up to a sepsis hospitalization. They studied three cohorts of patients hospitalized with sepsis: 1,512 participants in the U.S. Health and Retirement Study (HRS) in fee-for-service Medicare hospitalized 1998-2005 served as the derivation cohort, while 1,992 HRS cohort patients from 2006-2012 served as one validation cohort. Further, 32,525 Department of Veterans Affairs beneficiaries in 2009 served as a second validation cohort. The authors identified subgroups of patients with sepsis defined by their trajectory of presepsis inpatient healthcare facility use in the year before hospitalization. To identify these subgroups, the authors used a latent profile analysis, whereby the number of subgroups is determined by minimizing within-group differences and maximizing between-group differences. The authors determined differences in patient characteristics between subgroups and measured their association with 90-day mortality was measured with adjustment for multiple variables, including sex, race, age, acute illness severity, chronic disease burden, and presepsis functional limitations.

Researchers found a three-class model that best characterized presepsis trajectories of healthcare use. Half of patients defined as low use did not spend any days in an inpatient facility. Patients defined as rising use increased their healthcare facility use in the months immediately preceding sepsis hospitalization (median = 55 days spent in a healthcare facility). Patients defined as high use spent a significant amount of time in inpatient healthcare facilities over the year (median = 119 days). None of these trajectories resembled the overall mean trajectory. The three-class model from the derivation cohort remained robust when applied to the validation cohorts in terms of patient characteristics and distribution.

Overall, the low use group was healthier. They demonstrated fewer functional limitations and comorbidities before developing sepsis. High use patients were chronically ill, exhibiting more presepsis disabilities and higher comorbidity burdens. Those in the rising use group were older than other patients but showed fewer functional disabilities and comorbidities than the high use class. The authors attributed most inpatient healthcare use to time spent in long-term acute care hospitals or skilled nursing facilities. Researchers observed no differences in mechanical ventilation or ICU use between classes.

Across all cohorts, 90-day mortality was highest in the rising use class. The low use class demonstrated the best survival rate. Adjusted odds of 90-day mortality were 1.3- to 2.2-fold higher in the rising use class vs. the low use class. Specifically, adjusted mortality for the rising use class was 58% vs. 39% for the low use class in the derivation cohort, 44% vs. 31% in the first validation cohort, and 27% vs. 23% in the second validation cohort. Across all cohorts, more rising use patients died in the first 30 days and used inpatient healthcare facilities more often during the 90 days after hospital admission for sepsis.

COMMENTARY

Often, the authors of sepsis studies focus on improving delivery of care during hospitalization, with increasing efforts devoted to highlighting posthospitalization rates of readmission and sequelae. This study is novel in that it focuses on presepsis risk factors that can affect sepsis-related outcomes. The results from this observational cohort study suggest that sepsis outcomes correlate to patients’ overall health status before they are even hospitalized for sepsis; how this is linked to underlying sepsis pathobiology and the epidemiology of nosocomial exposures has yet to be elucidated. Notably, this study implies that the acuity of the change in host immunity in the rising use patients accounts for worse outcomes in this group, rather than the magnitude of healthcare facility use prior to sepsis hospitalization. Merely quantifying the amount of inpatient days in the year prior to sepsis hospitalization does not explain a patient’s risk for sepsis-related mortality. This is supported by the finding that rising use patients experienced poorer outcomes than high use patients, and that none of the three trajectories resembled the pattern of mean use in all cohorts.

How can one explain higher mortality rates among rising use patients? The acute decline in host defense mechanisms may outpace any protective stress response. Recent high inpatient use can lead to microbiome disruption, making patients more vulnerable to sepsis. Also, recent exposure to antibiotics and nosocomial pathogens can increase the risk of sepsis.

The findings from this study could be used to identify high-risk patients through electronic health records to inform staff regarding care delivery and outcomes, researchers conducting clinical trials regarding types of patients enrolled, and translational investigators regarding whether the host response to sepsis is different based on presepsis trajectory.

REFERENCE

  1. Rhee C, et al. Incidence and trends of sepsis in US hospitals using clinical vs claims data, 2009-2014. JAMA 2017;318:1241-1249.