Background/Purpose:

Low socioeconomic status (SES) negatively impacts health outcomes in the general population, as well as in systemic lupus erythematosus (SLE), but the impact on healthcare costs is unknown. In addition, there are little data on the long term costs of SLE cases beginning from diagnosis. To address these knowledge gaps, we examined the relationship between SES at diagnosis, and direct medical costs for 10 years following, in a general population-based context.

Methods:

Data Source: Our administrative data captured all provincially-funded outpatient encounters and hospitalizations (1990-2010), and all dispensed medications (1996-2010) regardless of funding source, in British Columbia, Canada. Sample: We assembled a population-based cohort of all incident cases of SLE who received care from 1996-2010, based on the following validated algorithm: a) two ICD-9-CM codes for SLE at least 2 months apart but within a 2 year period by a non-rheumatologist physician; or b) one ICD code by a rheumatologist or hospitalization. Statistics Canada neighborhood income quintile data for the year of SLE diagnosis was used to define SES. Cost Calculation: Costs for outpatient services and prescriptions were summed directly from billing data. Case-mix methodology was used for hospitalizations. Statistical Analysis: While most cases (87%) had less than 10 years follow-up, estimating long-term costs exclusively from cases with complete follow-up can introduce survival bias, and underestimate costs. Thus, costs were predicted using data from all cases. To account for censoring, follow-up was divided into 90-day periods, and costs per-period were weighted by the person-specific inverse probability of being alive in each period. A generalized linear model was used to predict the cumulative 10-year costs (adjusted for censoring) for cases in each SES group. Parametric bootstrapping was used to obtain 95% confidence intervals (CI). Costs are reported in 2010 Canadian dollars.

Results:

We identified 4,209 incident SLE cases (86% female, mean age 49 years) contributing 18,028 person-years. The 10-year costs from all cases totaled $132,762,777 with 30% from outpatient, 41% from hospitalizations and 30% from medications.

After adjusting for sex, age and baseline Charlson’s comorbidity index, predicted costs were significantly greater (21%, p=0.01) for the lowest SES cases compared to the highest, and averaged $12,489 more per-patient over 10 years.

Hospitalization costs over 10 years were 64% greater for the lowest-SES cases versus the highest ($13,097 vs. $8,001, p=0.05), with most hospitalizations (85%) occurring within 12 months of SLE diagnosis. Medication costs were 18% greater for the lowest SES than the highest, but unlike hospitalization costs, these costs continued over time.

Conclusion:

Lower SES at SLE diagnosis is associated with higher healthcare costs, with medication costs driving this disparity over the long term.