Achievable Benchmarks of Care for Pediatric Readmissions

Metropolitan Teaching

The readmission ABC was calculated for all 17 diagnoses (Table 2). The ABC ranged from 0.4% in acute kidney and urinary tract infection to 7.0% in sickle cell anemia crisis. Bipolar disorder, major depressive disorders and other psychoses, and sickle cell disease (SCD) had the highest percent of outlier hospitals whose mean readmission rates confidence interval did not contain the ABC; tonsil and adenoid procedures and viral illness had the lowest.¹

Metropolitan Nonteaching

The ABC was calculated for 13 of the 17 diagnoses because ABCs were not calculated when there were fewer than three best practicing hospitals. This was the case for tonsil and adenoid procedures, diabetes, seizures, and depression except for major depressive disorder (Table 2). Seven of the 13 diagnoses had an ABC of 0.0%: viral illness, infections of the upper respiratory tract, bronchiolitis, gastroenteritis, hypovolemia and electrolyte disorders, asthma, and childhood behavioral disorders. Like the findings at the metropolitan teaching hospitals, ABCs were lowest for surgical and acute conditions while bipolar disorder, major depressive disorders and other psychoses, and SCD had the highest percent of outlier hospitals with readmission rates beyond the 95% confidence interval of their hospital type’s ABC.

Nonmetropolitan

There was a sufficient number of best practicing hospitals to calculate the ABC for six of the 17 diagnoses (Table 2). For conditions where readmission ABCs could be calculated, they were low: 0.0% for appendectomy, bronchiolitis, gastroenteritis, and seizure; 0.3% for pneumonia; and 1.3% in kidney and urinary tract disorders. None of the conditions with the highest ABCs in other hospital settings (bipolar disease, sickle cell anemia crisis, and major depressive disorders and other psychoses) could be calculated in this setting. Seizure-related readmissions exhibited the most outlier hospitals yet were less than 5%.¹

Among a nationally representative sample of different hospital types that deliver care to children, we report the mean readmission rates and ABCs for 30-day all-cause, same-hospital readmissions for the most commonly readmitted pediatric diagnoses based on hospital type. Previous studies have shown patient variables such as race, ethnicity, and insurance type influencing readmission rates.^19,20 However, hospital type has also been associated with a higher risk of readmission due to the varying complexity of patients at different hospital types.^21,22 Our analyses provide hospital-type specific national estimates of pediatric readmission ABCs for medical and surgical conditions, many less than 1%. While commonly encountered pediatric conditions like asthma and bronchiolitis had low mean readmission rates and ABCs across all hospital types, the mean rates and ABCs for SCD and mental health disorders were much higher with more hospitals performing far from the ABCs.

Diagnoses with a larger percentage of outlier hospitals may represent a national opportunity to improve care for children. Conditions such as SCD and mental illnesses have the highest percentage of hospitals whose readmission rates fall outside of the ABCs in both metropolitan teaching and metropolitan nonteaching hospitals. Hospital performance on SCD and mental health disorders may not reflect deficits in hospital quality or poor adherence to evidence-based best practices, but rather the complex interplay of factors on various levels from government policy and insurance plans, to patient and family resources, to access and availability of medical and mental health specific care. Most importantly, these diseases may represent a significant opportunity for quality improvementin hospitals across the United States.

Sickle cell disease is predominantly a disease among African-Americans, a demographic risk factor for decreased access to care and limited patient and family resources.^23-26 In previous studies evaluating the disparity in readmission rates for Black children with asthma, socioeconomic variables explained 53% of the observed disparity and readmission rates were inversely related to the childhood opportunity index of the patient’s census tract and positively related with geographic social risk.^27,28 Likewise, with SCD affecting a specific demographic and being a chronic disease, best practice policies need to account for the child’s medical needs and include the patient and family resources to ensure access to care and enhanced case management for chronic disease if we aim to improve performance among the outlier hospitals.

Similarly, barriers to care for children with mental illnesses in the United States need attention.^29,30 While there is a paucity of data on the prevalence of mental health disorders in children, one national report estimates that one in 10 American adolescents have depression.^29,31 The American Academy of Pediatrics has developed a policy statement on mental health competencies and a mental health tool-kit for primary care pediatricians; however, no such guidelines or policy statements exist for hospitalized patients with acute or chronic psychiatric conditions.^32,33 Moreover, hospitals are increasingly facing “boarding” of children with acute psychiatric illness in inpatient units and emergency departments.³⁴ The American Medical Association and the American College of Emergency Physicians have expressed concerns regarding the boarding of children with acute psychiatric illness because nonpsychiatric hospitals do not have adequate resources to evaluate, manage, and place these children who deserve appropriate facilities for further management. Coordinated case management and “bundled” discharge planning in other chronic illnesses have shown benefit in cost reduction and readmission.^35-37 Evidence-based practices around pediatric readmissions in other diagnoses should be explored as possible interventions in these conditions.³⁸

There are several limitations to this study. Our data is limited to one calendar year; therefore, admissions in January do not account for potential readmissions from December of the previous year, as patient identifiers do not link across years in the NRD. We also limited our evaluation to the conventional 30-day readmission window, but recent publications may indicate that readmission windows with different timelines could be a more accurate reflection of medically preventable readmissions versus a reflection of social determinants of health leading to readmissions.²⁴ Newborn index admissions were not an allowable index admission; therefore, we may be underreporting readmissions in the neonatal age group. We also chose to include all-cause readmissions, a conventional method to evaluate readmission within an institution, but which may not reflect the quality of care delivered in the index admission. For example, an asthmatic discharged after an acute exacerbation readmitted for dehydration secondary to gastroenteritis may not reflect a lack of quality in asthma inpatient care. Readmissions were limited to the same hospital; therefore, this study cannot account for readmissions at other institutions, which may cause us to underestimate readmission rates. However, end-users of our findings most likely have access only to their own institution’s data. The inclusion of observation status admissions in the database varies from state to state; therefore, this percent of admissions in the database is unknown.

The use of the ABC methodology has some inherent limitations. One hospital with a significant volume diagnosis and low readmission rate within a hospital type may prohibit the reporting of an ABC if less than three hospitals composed the total of the ‘best performing’ hospitals. This was a significant limitation leading to the exclusion of many ABCs in nonmetropolitan institutions. The limitation of calculating and reporting an ABC then prohibits the calculation of outlier hospitals within a hospital type for a given diagnosis. However, when the ABCs are not available, we do provide the mean readmission rate for the diagnosis within the hospital type. While the hospital groupings by population and teaching status for ABCs provide meaningful comparisons for within each hospital setting, it should be noted that there may be vast differences among hospitals within each type (eg, tertiary children’s hospitals compared to teaching hospitals with a pediatric floor in the metropolitan teaching hospital category).^39,40

As healthcare moves from a fee-for-service model to a population-health centered, value-based model, reduction in readmission rates will be more than a quality measure and will have potential financial implications.⁴¹ In the Medicare fee-for-service patients, the Hospital Readmission Reduction Program (HRRP) penalize hospitals with excess readmissions for acute myocardial infarction, heart failure, and pneumonia. The hospitals subject to penalties in the HRRP had greater reduction in readmission rates in the targeted, and even nontargeted conditions, compared with hospitals not subject to penalties.⁴² Similarly, we believe that our data on low readmission rates and ABCs for conditions such as asthma, bronchiolitis, and appendicitis could represent decades of quality improvement work for the most common pediatric conditions among hospitalized children. Sickle cell disease and mental health problems remain as outliers and merit further attention. To move to a true population-health model, hospitals will need to explore outlier conditions including evaluating patient-level readmission patterns across institutions. This moves readmission from a hospital quality measure to a patient-centric quality measure, and perhaps will provide value to the patient and the healthcare system alike.