Background/Purpose:
Increased mortality in
RA is believed to be a consequence of inflammation. With improved treatment, mortality
would be expected to decrease over time. The objective of our study was to compare
mortality risk in RA vs. general population controls across RA incident cohorts
of earlier vs. later RA onset.

Methods:  We conducted a retrospective cohort
study of a population based cohort of all incident RA cases in BC with matched general
population controls, using administrative health data. RA cases were selected if they had
≥2 MD visits > 2 mos apart for RA in 01/1996-12/2006, with no prior RA
visit since 1990. Cases were excluded if they had ≥2 subsequent MD visits
for another inflammatory arthritis; if they saw a rheumatologist and the RA
diagnosis was never confirmed; or if there were no subsequent RA visits over a F/U
>5 yrs. General
population controls were randomly selected, matched 1:1 on birth yr, gender and
calendar yr of inclusion. Cohorts were divided into earlier (RA onset 1996-2000) and later (2001-2006)
cohorts. Data were
obtained on all physician visits and vital statistics data including cause of
death from death certificates, until Dec 2010.

Person-years
(PY) of F/U were calculated from index date to end of F/U, last health care use,
or death. Cases and controls were right censored at 5 yrs of F/U to ensure
equal F/U time when comparing mortality rates between earlier and later
incident cohorts. Sensitivity analyses including all years of F/U yielded
similar results. All cause and cause-specific mortality rates and 95% CI were
calculated for RA cohorts and controls, along with mortality rate ratios. In
addition, we obtained hazard ratios (HRs) for mortality in RA vs. controls via
exponential regression models adjusting for age, testing differences between incident
cohorts via an interaction term.

Results: The sample included 24,914 RA cases
and controls (66.5% female; mean [SD] age 57.3 [17.4] years) contributing 112,431
and 113,100 PY of follow-up, resp., with 2747 and 2123 deaths observed in RA
and controls, yielding all-cause mortality rates of 24.43 and 18.77 per 1000 PY
in RA and controls, resp., with a mortality rate ratio of 1.30 (95%CI: 1.23;1.38).
Mortality risk in RA vs. controls differed across incident cohorts. Mortality was
significantly increased in RA vs. controls in earlier, but not later, incident
cohorts, for all-cause as well as cause-specific mortality (Table 1). In the age-adjusted
exponential mortality models, a significant interaction between RA (vs.
controls) and incident cohort (early vs. late) was found (p<0.001) for all-cause,
CVD and cancer, but not infection (p=0.31).

Conclusion:
In our
population-based incident RA cohort, the risk of mortality compared to the
general population has improved over time. The mortality gap between RA and the
general population present in people with RA onset on or before 2000 was no
longer present in people with RA onset after 2000.