Estimating asthma control questionnaire (ACQ) scores from claims data.

Area of Science:

• Health Services Research
• Clinical Informatics
• Respiratory Medicine

Background:

• Asthma control is a primary treatment goal, typically assessed via prospective surveys like the Asthma Control Questionnaire (ACQ).
• Prospective surveys present challenges for large-scale population health studies.
• Efficient methods are needed to assess asthma control in real-world, large-scale settings.

Purpose of the Study:

• To develop and validate a predictive algorithm for estimating Asthma Control Questionnaire, 5-item version (ACQ-5) scores.
• To utilize readily available healthcare claims data for asthma control assessment.
• To enable population-level analysis of asthma control.

Main Methods:

• A prediction algorithm was developed and validated using prospective survey data (including ACQ-5) and retrospective medical/pharmacy claims from Kaiser Permanente of Colorado.
• Statistical methods included stepwise regression, ordinary least squares, Tobit, and Median regression.
• Validation involved 100 split-sample iterations, assessing mean squared error (MSE) and mean absolute error (MAE).

Main Results:

• The study included 2,657 individuals with complete baseline data.
• Statistically significant predictors of ACQ-5 scores included gender, use of oral corticosteroids, short-acting beta agonists, asthma drug classes, and healthcare utilization (emergency/outpatient visits).
• The prediction algorithm demonstrated similar performance (MSE, MAE) in both estimation and validation samples.

Conclusions:

• This study demonstrates the feasibility of predicting ACQ-5 scores using accessible medical and pharmacy claims data.
• The developed algorithm offers a potential tool for public health and population-level asthma control analyses.
• Further validation in diverse populations is recommended to confirm the algorithm's generalizability.