Table 2:

Change in emergency department (ED) visits per 1000 rostered patients

CharacteristicChange in ED visits per 1000 rostered patients (95% CI)*
Unadjusted modelAdjusted model
Proportion of virtual visits (per 1% increase)−11.0 (−11.8 to −10.1)0.2 (−0.5 to 0.9)
Physician payment model
 FHORef.Ref.
 FHG−0.2 (−0.3 to −0.2)0.3 (−0.1 to 0.7)
Patient complexity*
 Q1 (lowest)Ref.Ref.
 Q20.4 (0.4 to 0.5)3.8 (3.4 to 4.2)
 Q30.7 (0.7 to 0.8)6.2 (5.7 to 6.7)
 Q41.1 (1.0 to 1.1)8.8 (8.2 to 9.5)
 Q5 (highest)1.6 (1.5 to 1.7)13.9 (13.1 to 14.8)
Rurality
 UrbanRef.Ref.
 Semiurban0.4 (0.3 to 0.4)0.5 (−0.1 to 1.1)
 Rural2.0 (1.8 to 2.1)6.7 (5.1 to 8.4)
Physician age, yr
 25–44Ref.Ref.
 45–64−0.1 (−0.2 to −0.1)−0.4 (−0.7 to −0.04)
 65–74−0.2 (−0.3 to −0.1)−0.7 (−1.3 to −0.2)
 ≥ 75−0.02 (−0.2 to 0.1)−1.1 (−2.1 to −1.1)
Physician gender
 FemaleRef.Ref.
 Male0.2 (0.2 to 0.3)0.8 (0.2 to 1.4)
Mean patient age (per additional yr)0.03 (0.02 to 0.03)−0.2 (−0.2 to −0.1)
Proportion of male patients (per additional 1%)1.0 (0.8 to 1.3)2.2 (−0.8 to 5.2)
Total visits per roster (per additional visit)−0.003 (−0.004 to −0.001)−0.02 (−0.03 to −0.01)
2019 ED visits per roster (per additional visit)0.007 (0.006 to 0.009)0.02 (0.02 to 0.03)
R20.0400.385
No. of physician-months92 47092 470
  • Note: CI = confidence interval, FHG = family health group, FHO = family health organization.

  • * Unless indicated otherwise. Standard errors are clustered at the level of the physician. Adjusted model also controlled for month of year, census division and the interaction of month of year and census division.

  • Higher quintiles represent increasing average patient complexity, based on 2018 Canadian Institute for Health Information’s resource intensity weights.

  • The R2 statistic represents the proportion of the total variance in the outcome (ED visits) explained by the variables in the model. The R2 for models including all covariates without the main predictor (proportion of virtual visits) and with the main predictor were both 0.3849. Therefore, the incremental change in R2 when the main predictor is added to the model with all confounders was 0.