Cluster Analysis of Acute Care Use Yields Insights for Tailored Pediatric Asthma Interventions
Published in: Annals of Emergency Medicine, Volume 70, Issue 3 (September 2017), Pages 288-299.e2. doi: 10.1016/j.annemergmed.2017.06.024
Posted on RAND.org on January 31, 2018
- Can patterns of acute care use among pediatric asthma patients be used to inform interventions that would decrease acute care needs—and improve outcomes?
We undertake this study to understand patterns of pediatric asthma-related acute care use to inform interventions aimed at reducing potentially avoidable hospitalizations.
Hospital claims data from 3 Camden city facilities for 2010 to 2014 were used to perform cluster analysis classifying patients aged 0 to 17 years according to their asthma-related hospital use. Clusters were based on 2 variables: asthma-related ED visits and hospitalizations. Demographics and a number of sociobehavioral and use characteristics were compared across clusters.
Children who met the criteria (3,170) were included in the analysis. An examination of a scree plot showing the decline in within-cluster heterogeneity as the number of clusters increased confirmed that clusters of pediatric asthma patients according to hospital use exist in the data. Five clusters of patients with distinct asthma-related acute care use patterns were observed. Cluster 1 (62% of patients) showed the lowest rates of acute care use. These patients were least likely to have a mental health-related diagnosis, were less likely to have visited multiple facilities, and had no hospitalizations for asthma. Cluster 2 (19% of patients) had a low number of asthma ED visits and onetime hospitalization. Cluster 3 (11% of patients) had a high number of ED visits and low hospitalization rates, and the highest rates of multiple facility use. Cluster 4 (7% of patients) had moderate ED use for both asthma and other illnesses, and high rates of asthma hospitalizations; nearly one quarter received care at all facilities, and 1 in 10 had a mental health diagnosis. Cluster 5 (1% of patients) had extreme rates of acute care use.
Differences observed between groups across multiple sociobehavioral factors suggest these clusters may represent children who differ along multiple dimensions, in addition to patterns of service use, with implications for tailored interventions.
- Our cluster analysis found five distinct groups of asthma-related acute care use.
- Little difference in age distribution was found among the five clusters, which suggests that asthma-related use of the emergency department (ED) and inpatient care is similar across age groups.
- Presence of high rates of non–asthma-related ED use among the cluster with highest acute care usage suggests that interventions must focus on more than just asthma control.
- Interventions that home in on differences between groups and across factors, such as presence of mental health comorbidity or the number of facilities visited, may be more effective in improving the health of asthmatic children than interventions that are not tailored.