Editor's Note 

What if Readmission Prevents Death?—The Challenge of Composite Outcomes

Deborah Grady, Raegan W. Durant, Timothy S. Anderson

JAMA Intern Med Published Online: August 11, 2025

doi: 10.1001/jamainternmed.2025.3709

Composite outcomes, which combine several different events into a single outcome, are frequently used in clinical trials. Clinical trials testing transition of care interventions often measure distinct outcomes such as readmission rates, emergency department visits, and mortality.¹ Combining these outcomes into a single composite outcome generally increases the frequency of outcome events, which increases the power and reduces the sample size and cost of the trial. The validity of the composite outcomes is strengthened if each element is clinically important and affected similarly by the intervention. However, composite outcomes can also be misleading if the intervention has divergent impacts on different elements of the composite outcome. For example, an intervention might be beneficial for some of the elements of the composite, but harmful for others, and the overall effect could be null.

In JAMA Internal Medicine, Taylor et al² conducted a stepped-wedge, cluster randomized trial to evaluate the effect of a sepsis transition and recovery program compared to usual care on a composite primary outcome of readmission or mortality at 90 days. The trial randomized 7 hospitals in 1 system and included 3548 patients who had been hospitalized for sepsis. The composite primary outcome of readmission or death did not differ between the groups. However, there was a lower rate of death among intervention patients compared to usual care (17.3% vs 20.5%; P = .04) but a higher rate of readmission (35.9% vs 33.5%; P = .24). Lower mortality is always a desirable outcome, but the effectiveness of transition of care interventions is also usually judged based on whether a reduction in readmissions is achieved. The authors concluded that the intervention did not reduce the rate of the composite outcome but speculated that close observation may have led to more frequent readmission resulting in lower mortality.

Readmission shortly after an initial hospital stay has long been considered to indicate suboptimal clinical care or problems transitioning from hospital to home. Readmission rate has become a common and financially important quality metric, although some argue that the focus on readmission has had minimal benefit while distracting from other important quality concerns.³ Many studies of transition of care interventions have resulted in lower rehospitalization rates.⁴ However, it makes sense that close observation of patients who were hospitalized with sepsis could lead to appropriate rehospitalization to address treatable clinical issues, thus resulting in lower mortality. Other trials of hospital transition interventions have similarly shown no decrease⁴ or an increase in the rate of rehospitalization.⁵

The primary preplanned outcome of the trial by Taylor et al² was null, so we cannot conclude that the intervention is effective in reducing mortality. However, speculation that the increased rate of rehospitalization may have reduced mortality is intriguing. Perhaps the main lesson is that clinical trials of transitions of care should examine rehospitalizations independently if there is a chance of divergent results among other would-be composite outcome elements.⁶