The “exposure window”

We used the hospitalization database to identify all BC residents aged 18 to 55 years old who had been discharged from an episode of hospitalization with a main diagnosis of asthma (ICD-9: 493.xx, ICD-10: J45, J46). A previous chart review study showed that the main diagnosis of asthma in a discharge record had a sensitivity of 87% and a positive predictive value of 90%. We excluded the pediatric patient population because in Canada Pharmacy pediatricians can act as both generalists and specialists.

Patients were categorized as receiving primary care if they had had at least one outpatient service record, with asthma as the reason for the service, generated by a GP within 2 months of discharge from the index hospitalization, and had no code generated by any specialist during this time window. Patients were considered to be under secondary care if they had at least one outpatient service record for asthma generated by an internal medicine, a respiratory medicine, or an allergy/clinical immunology specialist. Of note, in Canada, where a publicly funded health-care system is in place, all internists practice as consultants, providing secondary care and requiring a referral from a general practitioner.

The “exposure window” of 2 months was chosen to cover the wait time for visiting a specialist and to account for situations in which the patient sought care sometime after finishing the medications provided on the discharge day (usually supplied for 30 days). Individuals not satisfying these exposure definitions were excluded from this analysis. An index date was assigned as the 60th day after discharge from the index hospitalization. 

Similar to the concept of “intention-to-treat” analysis, in the main analysis we retained the individuals’ exposure status for the entire follow-up period regardless of the subsequent changes in the type of care. Each individual could potentially contribute more than one index hospitalization and its corresponding follow-up period, provided that such follow-up times did not overlap.

Patients were followed for up to 12 months after the index date. Patients who exited the provincial coverage and those who died prior to 12 months after the index date were excluded from this analysis.

A critical issue in comparing the outcomes of primary vs secondary care is to adjust for the case mix, because patients under secondary care are more likely to have more severe asthma. To rigorously adjust for the case mix, we created a propensity-score-matched cohort.