Handoffs are among the most fragile moments in hospital care. Information, responsibility and clinical intuition all have to move from one clinician to another without losing anything that matters. In resident heavy settings, poorly managed handoffs have been linked to preventable harm, adverse events and higher costs. Much less is known about handoffs between attending physicians, particularly hospitalists who carry a large share of general medical inpatient care.
Farid and colleagues asked a very specific question: when hospitalised Medicare patients are cared for by hospitalists who are close to the end of their work block and are therefore more likely to hand off care, does this translate into a measurable difference in 30 day mortality (Farid, Tsugawa, & Jena, 2021)?
Study Question And Context
In typical hospitalist models, physicians work blocks of consecutive days, then hand over their entire inpatient panel to a colleague who is starting a new block. Intuitively, these transitions might pose risks, through fragmented information, loss of subtle clinical cues or delays in key decisions. Previous research on handoffs has focused mainly on trainees. Farid et al. extended the discussion to hospitalists and examined whether block end timing, used as a proxy for handoff probability, is associated with mortality among Medicare inpatients (Farid et al., 2021).
Design And Methods In Plain Language
The study used a quasi experimental design that exploits the “natural experiment” created by hospitalist work schedules. Instead of directly observing handoff events, the authors inferred the probability of handoff from the admission date relative to the last scheduled workday in the hospitalist’s block.
The data comprised a 20 percent random sample of Medicare fee for service beneficiaries admitted between 1 January 2011 and 31 December 2016 with general medical conditions. From more than one million admissions, the analysis focused on 597,288 hospitalisations with a length of stay of 14 days or less, in order to reduce the chance of multiple handoffs within a single admission. The average patient was 75.9 years old, 57.4 percent were women and 82.1 percent were White (Farid et al., 2021).
The key exposure was defined as follows. A high handoff probability group included patients admitted 1 to 2 days before the hospitalist’s last scheduled workday in a block. These patients were much more likely to experience a handoff to a new hospitalist. This group contained 366,287 hospitalisations. A low handoff probability group included patients admitted 6 to 7 days before the last scheduled workday, who were more likely to remain under the same hospitalist through most of their stay. This group contained 231,001 hospitalisations. Because both groups were defined purely by small differences in timing, they were clinically very similar on observed characteristics, which strengthens causal interpretation.
Thirty day mortality was compared between the two groups using models adjusted for age, sex, race or ethnicity, eleven chronic conditions, day of week, calendar year and hospital fixed effects. Hospital fixed effects control for all time invariant hospital characteristics, such as baseline staffing models or culture, that might otherwise confound the association (Farid et al., 2021).
Main Findings
In the full cohort of Medicare medical inpatients, high versus low handoff probability was not associated with differences in 30 day mortality.
In adjusted analyses, the 30 day mortality rate was 10.6 percent in the high handoff probability group, with a 95 percent confidence interval of 10.5 to 10.7 percent, and 10.6 percent in the low handoff probability group, with the same confidence interval. The adjusted absolute difference was 0.0 percentage points, with a 95 percent confidence interval from minus 0.2 to 0.1 percentage points (Farid et al., 2021). At the population level, block end timing as a proxy for handoff probability did not signal a broad mortality problem.
The picture changed when attention shifted to the sickest patients. In an exploratory analysis restricted to patients in the highest quartile of predicted mortality risk, higher handoff probability was associated with a modest but statistically significant increase in 30 day mortality. In this high severity subgroup, adjusted 30 day mortality was 27.8 percent for the high handoff probability group, compared with 26.8 percent for the low handoff probability group. The adjusted absolute difference was 1.0 percentage point, with a 95 percent confidence interval of 0.5 to 1.4 percentage points (Farid et al., 2021).
The authors also examined teaching and non teaching hospitals separately. The mortality increase among high risk patients with higher handoff probability appeared in both settings. In non teaching hospitals, the absolute difference was roughly 1.0 percentage point, while in teaching hospitals it was approximately 0.8 percentage points, with confidence intervals that excluded zero in both cases (Farid et al., 2021). This pattern suggests that the risk is not confined to trainee driven handoffs, but may be intrinsic to the handoff process among hospitalists themselves.
How Large Is A One Percentage Point Difference?
A one percentage point difference can look small in isolation. However, in a population where baseline 30 day mortality among the highest risk patients is close to 27 percent, an additional percentage point represents a non trivial number of excess deaths. The exposure is not a single drug or procedure, but a routine organisational process. This means that improvements in handoff quality have the potential to shift outcomes for large numbers of patients over time.
Clinical And Organisational Implications
The findings support a nuanced conclusion. For the average Medicare inpatient, current hospitalist handoff patterns do not appear to increase 30 day mortality in a detectable way. For the sickest patients, however, handoffs may carry a measurable incremental risk (Farid et al., 2021).
This has three practical implications.
First, hospitals should systematically strengthen handoff processes, especially for high risk patients. Structured tools such as I PASS and other standardised communication protocols can reduce information loss, clarify responsibilities and support anticipatory planning. If any group stands to gain from higher reliability during transitions, it is patients with unstable conditions and narrow safety margins.
Second, triage and scheduling policies can be aligned with risk. The study design relies on admission timing relative to block end, which is usually treated as an operational detail. The same information can be used strategically. High severity patients can be preferentially assigned to hospitalists at the beginning of a block rather than near its end. Flexible staffing plans can aim to avoid clustering of high acuity admissions at times when handoffs are more likely.
Third, fatigue and continuity need to be considered together with handoffs. Although Farid et al. performed sensitivity analyses to probe whether cumulative days worked, as a proxy for fatigue, explained the results, and found that fatigue alone did not account for the observed patterns, real world practice inevitably combines both effects. A hospitalist nearing the end of a long run is both more likely to hand off and more likely to be tired. Policy responses should therefore integrate block length, rest periods, team based care models and escalation rules for critically ill patients around transition points (Farid et al., 2021).
Limitations
Despite its strengths, the study remains observational. Handoffs were not directly observed. The exposure was inferred from scheduling and billing data, which may misclassify some patients. Residual confounding by unmeasured clinical factors cannot be completely ruled out. The analysis of the highest risk quartile was exploratory and requires replication in other datasets or prospective designs. Finally, the study focused on 30 day mortality among Medicare beneficiaries, so findings may not generalise to younger populations or non Medicare payers (Farid et al., 2021).
Why This Study Matters
Farid and colleagues provide a large scale, policy relevant signal. At the system level, hospitalist handoffs appear reasonably safe for most patients. At the same time, the study highlights a vulnerable minority for whom seemingly small process features, such as the exact timing of a hospitalist’s work block, may influence survival. The message is not to eliminate handoffs, which are unavoidable, but to concentrate safety efforts where the combination of acuity and transition risk is highest.
In practice, this means viewing handoffs as a predictable stress test for the care system. When the patient is stable and risk is low, the system can tolerate imperfections in the transition. When the patient is severely ill, that same level of imperfection may be the difference between recovery and deterioration. Designing hospitalist schedules, triage rules and communication practices with this asymmetry in mind is a logical next step.
Reference
Farid, M., Tsugawa, Y., & Jena, A. B. (2021). Assessment of care handoffs among hospitalist physicians and 30-day mortality in hospitalized Medicare beneficiaries. JAMA Network Open, 4(3), e213040. https://doi.org/10.1001/jamanetworkopen.2021.3040