The Center of Excellence for Pediatric Quality Measurement (CEPQM) at Boston Children's Hospital developed the two pediatric readmission measures. The measures underwent a rigorous development process, including the following steps:
Literature Review and Expert Interviews
We conducted a thorough review of the literature and guidelines pertaining to both pediatric and adult readmissions and reviewed current hospital, state, and delivery system initiatives focused on readmissions. In addition, we interviewed readmission experts across the country. These interviews helped to identify key challenges in developing pediatric readmission metrics as well as strategies for responding to these challenges.
We gathered ample feedback from provider, payer, and consumer stakeholders vested in child health care quality on the development of the readmissions measures. In particular, we convened a National Stakeholder Panel (NSP) to advise on the validity, feasibility, and usefulness of pediatric readmissions measures to support public reporting and quality improvement.
Our literature review and expert interviews demonstrated that the timing, prevalence, and rates of pediatric readmissions were largely unknown. We therefore performed an epidemiological analysis of readmissions using retrospective data from 69 children’s hospitals (Berry et al. JAMA 2013;309(4):372-80). We assessed readmissions for over 300 different admission diagnoses, as well as for all hospital admission diagnoses in aggregate (referred to as all-condition readmissions). We examined the timing of readmissions; readmission variation across hospitals; and the admission diagnoses associated with the highest readmission rates, prevalence and cost.
Evaluation of Candidate Measures
Additional Medicaid and all-payer data sets from multiple data sets were used to identify and test candidate readmission measures based on a variety of criteria, including clinical relevance, prevalence, cost, presence of disparities, extent of hospital performance variation, and degree of potential preventability.
Applying the knowledge gained from the work above, we developed two readmission quality measures specifically suited for pediatric patients: 1) the Pediatric All-Condition Readmission Measure, which evaluates readmissions following index admissions for almost all pediatric medical and surgical health conditions and 2) the Pediatric Lower Respiratory Infection Readmission Measure, which evaluates readmissions following index admissions for bronchiolitis, influenza, or community-acquired pneumonia. We chose to focus a measure on lower respiratory infections based on their relatively high readmission prevalence and the variation in their readmissions. We formulated and tested the measures using multiple administrative data sets that include various types of hospitals that provide pediatric care (i.e., children’s hospitals and general community hospitals) to both publicly and privately insured patients.
Exclusion of Planned Procedure Admissions
Many admissions for planned procedures, such as an admission for a planned tonsillectomy two weeks after a hospitalization for seizures, are part of a patient’s intended course of care and are less likely to be due to suboptimal care during the initial hospitalization. We therefore exclude such admissions from readmissions. We asked clinical experts to identify within their specialties procedures that are usually planned, defined as scheduled in advance more than 80% of the time for an expected medical need. We define planned procedure admissions as those for which a procedure that is usually planned is coded as the primary procedure.
To enable appropriate performance comparisons among institutions and states, we developed readmission case-mix adjustment procedures for each candidate measure that account for effects of differences in characteristics of individuals that could influence readmission risk, including age, co-morbid conditions, and severity of illness.
To ensure that the readmission measures are useful and meaningful to end users, we partnered with the New York Office of Quality and Patient Safety to test implementation of our measures on its Medicaid and all-patient inpatient claims databases and to obtain feedback on how the measure specifications could be improved. Feedback from New York on its testing experience indicated that the measure is straightforward and easy to implement.