Re: Death and readmissions after hospital discharge during the December holiday period: cohort study
I read with interest this large and robust cohort study on a challenging topic relevant to both clinical practice and health policy. This paper could be cited in medical and mainstream media and may add fuel to healthcare management and political fires in some regions regarding staffing of hospitals during the December holiday season. The primary outcome was death or readmission, defined as a visit to an emergency department or urgent rehospitalisation, within 30 days. The paper highlights that patients discharged over the December holiday period are less likely to have scheduled follow up and more likely to attend the emergency department, both of which are undesirable and unsatisfactory outcomes for patients and hospitals and the results as presented supporting this are meaningful. However, the authors' selection, presentation of the composite end-point and their emphasis on an increased rate of death stated in the abstract,conclusion and frequently repeated throughout the text could be challenged.
Composite end-points, if not appropriately selected, are both confusing and a source of misleading information and this has been highlighted previously in the BMJ in the context of clinical trials.  It can equally be applied to cohort studies. Furthermore, when the outcomes involve vastly different degrees of importance to patients, clinicians or even administrators the choice of outcomes must be questioned. If an outcome is an outcome is an outcome, then a trip to the emergency department is not a readmission is not a death.
I have read the full paper and the supplementary material. The 30 day deaths rate was 1.5% in each group which is an unadjusted risk difference of 0%. In subgroups often associated with unplanned readmissions such as patients with congestive cardiac failure and COPD there also was no difference in death rate, in fact a few less deaths were observed in COPD patients discharged over the holiday season (2.2 v 2.8%, appendix table 7). The difference in 30 day rehospitalisation rate for the entire population was 0.4% (11.8 v 11.4%).
Emphasising death in the composite end-points observed could detract from the authenticity and usefulness of otherwise important results demonstrated by the authors. To paraphrase from the conclusion on unwanted gifts, unwanted results, perhaps inconvenient to the narrative, should not be buried in the appendices.
1. Montori VM et al. Validity of composite end points in clinical trials
BMJ 2005; 330 doi: https://doi.org/10.1136/bmj.330.7491.594
Competing interests: No competing interests