Identifying patient characteristics that guide medication advice for COPD patients in a primary care electronic service

Aim, Several guidelines are available to assist and guide physicians in the selection of the most suitable inhalation treatment for COPD patients. Nevertheless, it is unclear which patient characteristics actually influence treatment decisions in daily clinical practice. Therefore, this study aims to determine the patient characteristics that guide medication advice for primary care COPD patients, using decision tree modelling.Method, Data from a Dutch asthma/COPD service was used. In this structured system, pulmonologists electronically support GPs in their diagnosis and treatment of patients with suspicion of obstructive lung disease. These data include for every patient the medical history, patient-reported outcomes (Clinical COPD Questionnaire (CCQ) and Asthma Control Questionnaire (ACQ)) and spirometry. The Chi-squared Automatic Interaction Detection method was used to build two decision trees. Results, The total study population consisted of 2051 patients with COPD. An ICS containing treatment was advised for 33% of the patients. The main predictor was whether the patient was treatment naïve or already on maintenance treatment; other predictors were the starting age of complaints, exacerbation frequency, FEV1 and CCQ. The most accurate tree was able to correctly predict 72% of the treatment advices; 42% of the ICS containing treatment and 86% of the other treatments.Conclusion, The most important characteristic that informs medication advice seems to be current medication use. Bronchodilators are advised for treatment of naïve patients and ICS containing treatment is advised in specific situations as a follow up treatment. Understanding pulmonologists’ prescriptions for patients of whom they only see electronic data is challenging. Nevertheless, the decision tree is partially in line with the treatment recommendations of (inter)national guidelines. Implementation Science/Service Development Research Ideas on Respiratory Conditions and Tobacco Dependency Abstract Declaration of Interest This study is funded by General Practitioners Research Institute and Boehringer Ingelheim The Netherlands bv. Boehringer Ingelheim supported this project with an unrestricted grant and did not have influence on the design, interpretation or reporting. References and Clinical Trial Registry Information