A Self-Fulfilling Hypothesis*
The population in the study by Ehlenbach et al (4) consisted of 135,370 patients 65 years old or older who were randomly selected from the Medicare & Medicaid Services Chronic Condition Data Warehouse (5). These patients had an episode of severe sepsis while in an acute care hospital (not necessarily in the ICU) during 2005–2009 and survived their hospital encounter. Their information was cross-linked with CMS’ Long Term Care Minimum Data Set (MDS) (6). The MDS is a government mandated set of data containing copious information on cognition and functionality for patients residing in a SNF/nursing home. Additional information on patients was obtained from health insurance claims.
The study had a surfeit of objectives, which unfortunately obfuscated the article’s message. One objective was to determine if patients who had previously been residing in a SNF had more severe cognitive and functional impairments than patients newly discharged to a SNF. Another aim of the study was to see if discharge location (SNF vs other) was associated with a patient receiving critical care while hospitalized. Finally, mortality within a year after hospital discharge was assessed based on discharge location. Again, all of these patients had had a diagnosis of severe sepsis while hospitalized.
The findings from the study by Ehlenbach et al (4) showed that patients who had a diagnosis of sepsis and were subsequently discharged to a SNF had increased cognitive impairment and reduced functionality compared with similarly afflicted patients discharged elsewhere. Although 1-year mortality was high among all patients in the study, the highest mortality rate was found in previous SNF patients discharged back to a SNF, followed by patients discharged de novo to a SNF. The lowest mortality occurred in patients not discharged to a SNF.
After disentangling the various clinical endpoints, this study’s (4) findings are unremarkable. Why else would a patient be discharged to a SNF if not for an increased level of impairment? In the authors’ (4) defense, prior studies on identifying risk factors for postdischarge outcomes among elderly critically ill patients showed analogous results. In one study (7), a multivariable model was developed for predicting discharge to a postacute care facility for critically ill patients. A low Glasgow Coma Score and neurologic diagnoses, implying reduced cognitive ability, were some of the major risk factors. A study by Baldwin et al (8) looked at 6-month mortality for critically ill patients discharged alive from the hospital. Among the most important risk factors were admission from or discharge to a skilled care facility.
What is particularly troubling about the study by Ehlenbach et al (4) is their inclusion of patients who were residing in a SNF before admittance to the hospital. It is difficult to construct a scenario whereby a patient admitted to the hospital from a SNF is subsequently discharged home. In essence, one of the study’s main hypotheses is self-fulfilling.