The best magazine
Results
Of the 11 857 patients who had a first visit for RA in 1995 (i.e. no visit for RA in the previous 5 years), there were 3890 patients with a confirmed diagnosis of RA by a rheumatologist. Of those, 1051 patients were first diagnosed with RA by a non-rheumatologist and included in our study. The mean age at diagnosis was 55.7 (s.d. 15.4) years, and 68.2% were female. Of those persons first diagnosed by a non-rheumatologist, 533 (50.7%) saw a rheumatologist within 3 months (Table 1); their median delay to see a rheumatologist was 22 days (interquartile range 7–44 days). Among the 518 who saw a rheumatologist late (i.e. past 3 months), the median delay from first diagnosis by a non-rheumatologist until rheumatology consultation was 719 days (interquartile range 209–1702 days). The patients contributed a maximum of 12 years of follow-up. During this period 210 (20.0%) persons in the cohort died, and 215 (20.5%) had orthopaedic surgery during the observation interval.
Table 2 summarizes the results of simple unadjusted analyses of the categorical factors potentially associated with an increased need for orthopaedic surgery. Persons who consulted a rheumatologist later (i.e. at least 3 months after being first diagnosed) were more likely to undergo orthopaedic surgery than those who consulted earlier (P < 0.01). Additional unadjusted analyses of continuous variables showed that persons who underwent orthopaedic surgery had an average of 4.0 (95% CI 3.6, 4.3) medical visits to a rheumatologist in the first 12 months after being diagnosed with RA compared with a mean of 3.5 (95% CI 3.3, 3.7) for those who did not undergo orthopaedic surgery (P = 0.03). Similarly the number of visits to other physicians in the first year after diagnosis was higher for those who later underwent orthopaedic surgery [15.0 (95% CI 13.5, 16.6) vs 13.1 (95% CI 12.2, 13.9); P = 0.04]. Those who underwent orthopaedic surgery tended to be older [57.1 (95% CI 55.2, 58.9)] as compared with those who did not undergo orthopaedic surgery [55.2 (95% CI 54.2, 56.4)] (P = 0.11).
Kaplan–Meier curves in Figure 1 suggest that early consulters had a lower rate of surgery across the entire 12-year follow-up period (P < 0.01 for the unadjusted log rank test). Although there is an increasing rate of orthopaedic surgery during the 12-year period in general, early consulters had a consistently lower rate. Furthermore, as time increases, the rate of increase of orthopaedic surgery is higher among late consulters. Table 3 summarizes the results of multivariate time-to-event analyses, based on two Cox PH models with time-dependent covariates. Both models fitted the data equally well (difference in AIC <1), and in both models the PH assumption was verified (data not shown). In model 1 (left part of Table 3), patients who consulted with a rheumatologist within the 3 months after being first diagnosed with RA (early consulters) had shorter time to first orthopaedic surgery than the late consulters [adjusted hazard ratio (HR) = 0.60; 95% CI 0.44, 0.82]. Consistent with this finding, in model 2 (right part of Table 3) the rate of orthopaedic surgery increased significantly with increasing time to consultation. For example, a 3-month delay and a 1-year delay in time to consultation were associated with risk increases of 3.4% (HR = 1.034; 95% CI 1.01, 1.05) and 14.4% (HR = 1.14; 95% CI 1.06, 1.24), respectively. Exploratory analyses, based on the flexible extension of the Cox model, indicated that the effect of time to consultation was non-linear (P < 0.01 for test of linearity). Specifically, the reduction in risk is especially marked while comparing patients who consulted very early (e.g. in the first 3 or 6 months) with others (seen later). On the other hand, there were no systematic differences between those who consulted with a rheumatologist only 4–5 years after the initial diagnosis and those who consulted even later (data not shown).
(Enlarge Image)
Figure 1.
Time from first RA diagnosis until first orthopaedic surgery.
The figure depicts the failure distribution function for the time until orthopaedic surgery for early consulters (i.e. those who consulted a rheumatologist within 3 months of initial diagnosis) vs late consulters (i.e. those who consulted a rheumatologist after 3 months of being initially diagnosed). The red curve depicts the late consulters; the blue curve portrays early consulters. The y axis indicates the percentage of patients having orthopaedic surgery, and the x axis indicates time (in years). The figure shows that although there is an increasing rate of orthopaedic surgery during the 12-year period, early consulters had a consistently lower rate. Furthermore, as time increases, the rate of increase of orthopaedic surgery is higher among late consulters.
In both models 1 and 2, older age at diagnosis and having had more visits with other physicians in the first year after diagnosis were associated with an increased rate of orthopaedic surgeries (Table 3). No other patient characteristics had statistically significant associations with time to orthopaedic surgery (Table 3).
Source: ...