Comparing Illness Severity Classifications for Children Hospitalized With Asthma

Although guidelines define asthma severity by outpatient frequency and severity of symptoms, there is a lack of consensus on inpatient severity classification during acute illness, limiting hospital-based research on pediatric asthma.1,2 In a recent study, we considered different approaches to classify inpatient asthma illness severity before applying a novel asthma-specific classification using administrative data.3 This current analysis aimed to (1) compare how different approaches classified illness severity and (2) how the classifications aligned with All Patient Refined Diagnosis Related Groups (APR-DRG) severity of illness classification, which classifies hospital inpatients according to their reason for admission and severity of illness for hospital reimbursement.4This is a secondary analysis of a recently published multicenter, retrospective cohort study of children aged 2 to 18 years admitted with a primary diagnosis of asthma at 44 children’s hospitals between May 1, 2021, and April 30, 2022, participating in the Pediatric Health Information System (PHIS).3 A primary diagnosis of asthma was identified using published International Classification of Diseases, 10th Revision (ICD-10) codes.5 Children discharged from the emergency department were not included in the study. See Supplemental Material for supplemental methods.Three illness severity classification systems were compared to each other and to the APR-DRG severity classification using descriptive analysis. (1) Pediatric intensive care unit (PICU) admission (presence of a PICU room charge at any point during the hospitalization): (1) Yes (severe); (2) No (mild-moderate).(2) Respiratory illness–based (mirrors previously published COVID-19 and pneumonia classification systems6,7): (1) critical (PICU admission plus: invasive mechanical ventilation, vasopressors, and/or extracorporeal membrane oxygenation [ECMO]), (2) severe (PICU admission not meeting critical illness criteria), (3) mild-moderate (admission not meeting critical or severe illness criteria).(3) Asthma-specific (adapted from the National Heart, Lung and Blood Institute Guidelines and several children’s hospitals’ inpatient asthma clinical pathways2,3,8,9): (1) critical (PICU admission plus: terbutaline, heliox, epinephrine, isoflurane, sevoflurane, non-invasive positive pressure ventilation, invasive mechanical ventilation, and/or ECMO), (2) severe (PICU admission not meeting critical illness criteria or admission outside the PICU plus magnesium sulfate) or (3) mild-moderate (admission not meeting critical or severe illness criteria).Of the 19 119 hospitalizations that met inclusion, most were classified as mild/moderate illness for the PICU admission (87.2%) and respiratory illness–based (87.2%) classifications compared with the asthma-specific classification, which more equally distributed hospitalizations between severe (49.1%) and mild/moderate (47%) illness (Table 1).The majority of hospitalizations were classified as APR-DRG mild severity (64%; calculated from Table 2). Of the PICU admissions, 23.8% were APR-DRG severity of mild. Respiratory illness–based and asthma-specific classifications largely aligned with APR-DRG classification, except for asthma-specific severe hospitalizations, of which 61.8% were APR-DRG mild (Table 2).By comparing 3 illness severity classification systems for asthma hospitalizations, we found that PICU admission alone and PICU plus respiratory illness–based classifications likely underestimated the proportion of children with severe asthma illness, many of whom are managed on the general pediatric floor with magnesium. In contrast, the asthma-specific classification that applied clinical markers of asthma severity available in PHIS had a more varied distribution among illness severity groups. Overall, the majority of hospitalizations were classified as APR-DRG mild severity and did not consistently align with the illness severity classifications; in particular, the majority of severe illness cases in the asthma-specific classification were APR-DRG mild severity.Identifying which classification system to use when designing administrative database studies is essential and should be guided by the research question or quality improvement goal. The asthma-specific classification may better reflect illness severity in this population than PICU admission or respiratory illness–based classifications, yet it remains misaligned with APR-DRG classification. For example, some patients have low APR-DRG severity, yet treatments suggesting higher asthma severity given APR-DRG assign severity solely by diagnosis and procedure codes, not by asthma treatments received. Use of information on asthma treatments may help hospitals accurately categorize patients and potentially better align their cost of care and reimbursement. Our report is limited by the administration data available in PHIS. For example, we were unable to distinguish continuous albuterol vs intermittent albuterol. Moreover, we did not assess supplemental oxygen due to unreliable coding.10 Our findings demonstrate that using PICU admission and/or respiratory illness–based classification systems to distinguish illness severity may unintentionally underestimate the number of children with severe illness and skew results.We thank Dr Hugh J. Ladd for his clinical guidance on our asthma illness severity classification system.