Optimizing community-level surveillance data for pediatric asthma management.

Area of Science:

• Pediatric Asthma Management
• Health Services Research
• Data Analytics in Public Health

Background:

• Community-level pediatric asthma management often uses claims data and school surveys.
• Integrating these data sources can provide a more comprehensive understanding of asthma risk patterns.

Purpose of the Study:

• To combine school-based asthma survey data with Medicaid claims data.
• To identify demographic and healthcare utilization patterns associated with pediatric asthma severity in adolescents.
• To evaluate the utility of recursive partitioning for asthma severity prediction.

Main Methods:

• Matched symptom data from school surveys with Medicaid claims for 561 adolescent students.
• Utilized recursive partitioning to develop classification trees for asthma severity.
• Analyzed patterns of demographics and healthcare utilization linked to high and low asthma severity.

Main Results:

• 86.1% of students were classified as high-severity asthma.
• Classification trees identified eight risk subsets, predicting severity based on factors like rescue medication use and nebulizer treatments.
• The model achieved 66.7% predictive accuracy, correctly identifying 91.1% of high-severity and 42.3% of low-severity cases.

Conclusions:

• Combining complementary datasets enhances community-level insights into pediatric asthma.
• Recursive partitioning is a valuable method for exploring complex asthma severity predictors.
• Findings can inform targeted interventions for adolescent asthma management.