Inpatient hospital performance is associated with post-discharge sepsis mortality

Sepsis is a life-threatening condition that affects 1.7 million Americans annually, with an in-hospital mortality of 16% [1]. Costing over $60 billion each year, sepsis is the most expensive acute medical condition for which Medicare reimburses hospitals [2]. Early treatment has been shown to improve survival to hospital discharge [1, 3, 4], but sepsis case-fatality continues to vary significantly by hospital (41% for highest-mortality decile of hospitals vs. 29% for lowest-mortality decile, p < 0.001) [5].

The burden of sepsis continues beyond hospital discharge. Many sepsis survivors have persistent neurocognitive deficits, neuromuscular weakness, symptoms of depression and post-traumatic stress disorder, and poor quality of life [6‐10]. Readmissions after discharge are common (21% within 30 days) [11], and sepsis survivors have increased post-discharge mortality [12‐14]. Post-discharge deaths have been associated with premorbid health status and hospital factors [15, 16], but it is unclear whether these factors are modifiable, and whether hospital care can alter the post-discharge course.

The objective of this study was to determine the relationship between quality of hospital care (measured by hospital-specific inpatient adjusted sepsis survival) and post-discharge outcomes, including mortality, readmissions, and skilled nursing facility discharge (Fig. 1). We postulate that hospitals that provide higher quality care have improved sepsis risk-adjusted inpatient mortality and that this quality leads to improved post-discharge mortality in those who survive hospitalization. This analysis differs from prior work, in that we are evaluating a primary exposure of hospital performance (a more direct measure of hospital quality) rather than administrative or epidemiologic factors than may be related with quality indirectly, and our outcome includes only those who survive hospitalization [15‐18]. Such a relationship would suggest that the impact of inpatient sepsis therapies may extend beyond hospital discharge, and that early treatment elements may change long-term immunity, organ function, and debility.

A total of 830,720 age-qualifying Medicare beneficiaries were hospitalized for sepsis between 2013 and 2014 (Fig. 2). We excluded 123,807 admissions in 1922 low-volume hospitals with fewer than 200 cases during the sample period. Most patients (68%) were hospitalized with either a respiratory or urinary infection source, 14% had a co-existing malignancy, and 9% were admitted from a skilled nursing facility (Table 1). Of all cases, 1% were transferred between hospitals. In-hospital mortality was 20% (n = 167,854), with 30-day, 90-day, and 180-day cumulative mortality of 35% (n = 291,708), 43% (n = 353,147), and 48% (n = 398,914), respectfully. Of sepsis discharges, 37% (n = 247,045) were readmitted to a hospital within 180 days, with the median time to readmission approximately 32 days (IQR 12–73 days).

The relationship between sepsis and post-discharge mortality has been well established [12‐14, 27], but whether post-discharge mortality has modifiable treatment-related risks remained unclear. This question has been challenging to answer, since the specific biological mechanisms by which post-sepsis mortality is elevated have not been well elucidated. Mechanistic data suggest that sepsis survivors remain immune suppressed [28, 29], have increased rates of atherosclerosis [13, 30], and that epigenetic regulation may play a role in modulating these effects [31]. The question that mechanistic studies have not answered, however, is whether improved inpatient care can attenuate the impact on late post-discharge mortality. By analyzing variation between U.S. hospitals, our data suggests that it may.

Inpatient quality can be difficult to measure [32, 33], and the factors associated with sepsis quality are myriad [34, 35]. The Early Management Bundle for Severe Sepsis/Septic Shock (SEP-1) [19] reported by the Centers for Medicare and Medicaid Services is one parameter for ranking hospitals on adherence with a single bundle of very early care, but this measure is not a comprehensive measure of sepsis quality of care and the specific causal relationship between SEP-1 performance and outcomes has been questioned [36‐40]. By using an inclusive measure of risk-adjusted inpatient sepsis mortality aggregated at the facility level, we have attempted to capture quality-of-care by its outcome. This study design cannot identify which factors are associated with preventing late mortality or readmissions, but it provides evidence that care factors that affect inpatient outcomes may continue to influence patients after hospital discharge.

What factors might be responsible for the magnitude of the association? Factors associated with post-discharge death could fall into one of four categories: early resuscitation interventions, sub-acute organ support and infection management, de-escalation of therapy, and post-discharge planning and care coordination. Early care interventions, such as timely antibiotics and hemodynamic resuscitation could decrease organ failure and limit the degree of subclinical organ dysfunction persisting at hospital discharge. Immunomodulatory effects of resuscitation may prevent both early organ failure and persistent immune dysfunction [43].

The overall elements of care that are performed after initial resuscitation, including antibiotic use and discontinuation, diuresis, early mobility, delirium management, nutrition, and other factors may also contribute to sepsis outcomes [4, 44‐46], but the specific manner in which these practices function as predictors and effect modifiers is not well described. Early mobility, timely extubation, and delirium prevention strategies are well-described methods of decreasing mortality and disability from critical illness, and thus are attractive targets to investigate in future studies [47‐49]. Finally, discharge planning and care coordination may play a significant role in post-discharge follow-up, access to care, medication management, and the transition to community living [50]. Discharge planning, care coordination, and degree of pre-discharge medical recovery also may explain the relationship between inpatient O:E mortality, new skilled nursing discharges, and readmissions, but none of these elements were tested in our analysis. Gadre, et al. recently reported readmissions after nearly 18% of sepsis discharges, and they were associated with more comorbidities, longer index hospital stays, and residence in skilled nursing facilities [51]. Although conclusive causal relationships have not been established, one might predict that (1) providing high-quality sepsis care that limits deconditioning and chronic organ dysfunction, (2) scheduling discharge timing and managing chronic disease well, and (3) developing strong relationships with primary care providers, community organizations, and post-acute care providers to provide a seamless post-acute care transition may effectively limit the need for both skilled nursing and readmissions.

This study has several limitations. First, our use of observational data limits our statistical models to variables commonly recorded in administrative data, allowing for the possibility of unobserved confounders. These available data also led to our use of hospital O:E mortality as a surrogate for hospital quality—likely an incomplete surrogate with some potential for residual confounding and sensitive to differences in coding practices between hospitals. Second, we included only Medicare beneficiaries in these data. While two-thirds of sepsis cases in the U.S. are Medicare beneficiaries, using these cases alone isolates a population over age 65 with reliable payment for healthcare services.

In conclusion, hospitals with the lowest risk-adjusted sepsis inpatient mortality also have reduced post-discharge mortality. These observations suggest that some factors influencing in-hospital mortality also influence post-discharge outcomes, and some of these factors may be modifiable. Future work should focus on identifying factors associated with post-discharge outcomes and optimizing inpatient and transitional care to improve the likelihood of functional long-term recovery.