Predicting family physician physical activity electronic medical record inputs

To compare characteristics of patients with and without physical activity noted in primary care electronic medical records.

We used pan-Canadian family physician electronic medical record data from the Canadian Primary Care Sentinel Surveillance Network (CPSSSN) to compare patient and provider characteristics on one visit per patient selected at random. Since patients were nested by providers, univariate statistics were explored then a multilevel model was constructed.

The dataset included 769,185 patients, of whom 14,828 (1.9%) had physical activity information documented. Male patients, aged 25-34.9, no comorbidities prior to the random visit date, moderate or elevated blood pressure risk categories prior to the random visit date, the least materially deprived quintile, and with median body mass index in the normal category prior to the random visit date had the most physical activity mentions. Of the 879 family physicians in the sample, just over half (56.1%) documented physical activity at least once across their patients. More female physicians and physicians who practised in academic sites documented physical activity. In a two-level logistic model to predict physical activity documented in the randomly selected visit: older than mean patient age, having fewer comorbidities, younger than mean family physician age, academic teaching sites, and electronic medical record systems were statistically significant covariates.

This work adds to existing literature by describing the frequency and the patient and family physician characteristics of physical activity documentation in the Canadian primary care context. Overall, patient physical activity was rarely documented in electronic medical records.