Preventing Pediatric Readmissions: How Does the Hospital Fit In?

Pediatric readmission rates are increasingly used as a quality marker for pediatric hospitals. Yet, there is angst about the appropriateness of using readmission rates for payment given the lack of evidence regarding the potential preventability of these events.1–3 Previous studies have assessed the preventability of pediatric readmissions through medical record review4,5 and through application of proprietary software to administrative data.6 These studies yielded estimates of 20% (medical record review) and almost 50% (proprietary software) preventability. In addition, significant discrepancy exists in preventability assessments between medical record review and the use of preventability software in children with ventricular shunt readmissions.7In this issue of Pediatrics, the analysis by Toomey et al8 identified 30% of 30-day readmissions as being potentially preventable. This study is the most comprehensive assessment of pediatric readmission preventability to date. The authors conducted >1000 interviews with parents, adolescent patients, inpatient nurses, attending physicians, social workers, case managers, subspecialist physicians, and primary care providers. The investigators ensured high fidelity in data collection, with response rates for the various groups often >90%. Interview materials were combined with structured medical record review, and a consensus review process was used to determine readmission preventability.Although post-readmission interview data have acknowledged limitations (response bias, differing stakeholder perceptions9), Toomey et al found that interview data led to a change in reviewer assessments of preventability in 12% of readmissions. The change in opinion from nonpreventable to potentially preventable was almost sixfold more frequent than the change in opinion from potentially preventable to nonpreventable. As such, relying only on medical record review without adjunctive interview data likely advantages the hospital by underestimating preventability.The in-depth, 360-degree evaluation allowed the investigators to further examine the etiology of potentially preventable readmissions, distinguishing between hospital, patient, and primary care provider factors. Not surprisingly, the hospital won out, with almost 80% of preventable readmissions having a hospital factor contributing to the readmission. Comparatively fewer preventable readmissions had a primary care provider (39%) or patient (14%) factor identified. Toomey et al do not present the percentage of preventable readmissions attributed solely to hospital factors (without a patient, primary care, or other factor), which would be important in further understanding the hospital’s responsibility in preventing readmissions.With regard to contributing factors, an assortment of usual suspects, including issues with patient/caregiver education, handover communication, and postoperative complications, were attributed to hospital care. Interestingly, the most common factor attributed to the hospital involved the category of patient assessment. According to the STate Action on Avoidable Rehospitalizations (STAAR) tool, patient assessment factors include items such as “unrealistic optimism of patient and family to manage at home,” “failure to recognize worsening clinical status in the hospital,” or “not addressing the whole patient.”10 The identification of patient assessment issues argues for continued inquiry into less traditional transition research concepts such as discharge readiness,11,12 caregiver confidence,13 communication/shared understanding,14,15 and inpatient staff to patient ratios.16There were some other important results from this study. Echoing the findings by Wallace et al,4 the authors found that the time to readmission was shorter for potentially preventable than for non–potentially preventable readmissions (a median of 4 days versus 9 days). Potentially preventable readmissions were more likely to be causally related to the index hospitalization than nonpreventable readmissions. However, operationalization of a causally related readmission metric would be effectively impossible without case-by-case review. Surprisingly, patient complexity was not associated with preventability. However, the study institution serves only a small percentage of patients without any chronic conditions (18%), which limits the generalizability of this finding.The numbers also provide a sobering reminder of the difficulty of moving the bar in pediatric readmissions. Toomey et al report a 30-day unplanned readmission rate of 6.5%. If 30% of unplanned readmissions are truly preventable, then, on average, only 2 in 100 discharges will result in preventable 30-day readmission. Unfortunately, none of the study findings will help hospitals predict preventability in real time. Without accurate prediction models to prospectively identify children at high risk of preventable readmission who may benefit from enhanced discharge support, quality improvement leaders must weigh costs and benefits of transition interventions to prevent readmission across the board. Perhaps this is why, in a recent commentary, Berry and Gay17 suggested that we pivot away from the obsession with pediatric readmissions and focus on improving the hospital experience and discharge process for children and their families.With regard to readmission as a hospital quality metric, we now know that we may be able to prevent the stress and disruption of a 30-day readmission in 30% of cases. Evidence also suggests that a shorter readmission time interval may be a more meaningful measure of hospital quality than the 30-day interval. Whether or not readmission is the “right” quality metric, Toomey et al present compelling evidence that children’s hospitals have room to improve transitional care.