Background/Purpose: Logistical issues and
poor cultural environments in tertiary care create barriers to specialized care
for Aboriginal patients with rheumatoid arthritis (RA). Aboriginal patients are
thus more likely than non-Aboriginal patients to see a primary care provider
and less likely to see a rheumatologist for their RA care, and may result in
differences in disease management and clinical outcomes. We used health
services data to estimate the effect of these differences reflected in health
system costs.

Methods: The Alberta Biologics
Pharmacosurveillance Program (ABioPharm) is a longitudinal RA cohort study,
linked to population-based administrative databases. These databases capture
hospitalization, emergency room, and outpatient clinic visits which have an
associated clinical modifier group, and physician visits which have an
associated claim cost, from which health service utilization costs are
estimated. Given the skewed nature of the data and that it may contain many
zero values, we used a two-part modeling strategy for mixed discrete-continuous
outcomes. In the first part, a binary choice model was estimated for the probability
that the patient has had (or not had) health costs (dichotomous). In the second
part, a generalized linear model with gamma family (with propensity score
analysis) was used to estimate the difference in health costs between
Aboriginal and non-Aboriginal groups, conditional on a cost having been
incurred. Costs were adjusted for inflation to 2011/2012 using the Canadian
Consumer Price Index from Statistics Canada.

Results: The cohort included 1,545 patients
(n=83 Aboriginal) with 8,145 person-years of follow-up. Mean and median total costs
and specific RA-related costs for hospitalizations, emergency room, outpatient
clinic and physician costs are presented in Table 1, demonstrating the skewed
nature of the data. Emergency room costs were higher in Aboriginal patients
using standard analysis (median $1,003 CAD (IQR 2,833) vs $262 CAD (IQR 1,071)
per patient per year, Mann-Whitney p<0.001). In the two-part model (Table 1),
cost estimates for Aboriginal patients showed a numerical trend to lower
hospital, outpatient clinic and physician visit total and RA-related costs, but
with higher emergency room costs compared to non-Aboriginal patients, although not
reaching statistical significance.

Conclusion: Health service utilization costs
did not vary between Aboriginal and non-Aboriginal patients, although with
limitations of sample size. Differences in health service use may explain disparate
clinical outcomes observed in our cohort, which could be remedied by increased
collaboration with primary care providers, and creating health care environments
that deliver culturally competent care.