Skip to main content
. 2012 Jan 12;71(7):1128–1133. doi: 10.1136/annrheumdis-2011-200702

Table 2.

Relationship between serious infection risk and oral glucocorticoid exposure for the 10 conventional models and the best fitting cumulative weighted dose model

Model Among cases (% or mean)* Among controls (% or mean)* OR (95% CI) OR for 5 mg increase (95% CI) AIC AIC–AIC of the WCD model
Binary conventional models
(1) Current use 42.1% 24.7% 1.84 (1.64 to 2.06) - 6175.0 115.2
(2) Any use past 30 days 52.9% 31.1% 2.08 (1.86 to 2.33) - 6122.1 62.3
(3) Any use past 90 days 58.6% 34.7% 2.26 (2.02 to 2.54) - 6083.6 23.8
(4) Ever use 74.4% 58.5% 1.72 (1.53 to 1.94) - 6201.5 141.7
Dose-specific conventional models
(5) Current dosea 11.0 mg PEQ 7.8 mg PEQ 1.04 (1.04 to 1.05) 1.24 (1.19 to 1.30) 6171.3 111.4
(6) Average dose in past 30 daysb 8.3 mg PEQ 6.2 mg PEQ 1.07 (1.06 to 1.08) 1.41 (1.33 to 1.49) 6124.0 64.2
(7) Average dose in past 90 daysb 7.1 mg PEQ 5.4 mg PEQ 1.09 (1.08 to 1.11) 1.56 (1.46 to 1.67) 6094.8 34.9
(8) Average dose since study entryb 5.4 mg PEQ 3.7 mg PEQ 1.08 (1.06 to 1.09) 1.45 (1.35 to 1.56) 6166.6 106.8
(9) Peak dose in past 30 daysc 13.6 mg PEQ 9.3 mg PEQ 1.03 (1.02 to 1.03) 1.14 (1.10 to 1.17) 6187.3 127.5
(10) Peak dose in past 90 daysc 15.2 mg PEQ 10.9 mg PEQ 1.02 (1.01 to 1.02) 1.10 (1.07 to 1.12) 6200.3 140.5
Flexible model incorporating weighting by recency of treatment
(11) Final WCD (3-year with 3 degrees of freedom) § § § § 6059.8 0 (minimum AIC)

OR, adjusted for all a priori confounders.

*

For models representing exposure by binary indicator variables (1–4), the value provided is the proportion of cases/controls with the defined glucocorticoid exposure. For models with continuous exposure (5–10), the mean value of the exposure variable among glucocorticoid therapy users for cases/controls is given (seea–c below).

For dose-specific models (5–10), OR represents risk per 1 mg PEQ increase.

A lower AIC value indicates a better fit to data.

§

Because the (non-parametric) WCD model estimates exposure effect using flexible spline functions, the estimated effect cannot be summarised by a single parameter. See figure 1 for the estimated weight function and table 3 for the WCD model estimates of adjusted OR associated with selected exposure patterns.

a

Mean current dose=mean dose in current users.

b

Average dose in past 30/90 days/since study entry=average dose over the past 30/90 days/since study entry in subjects who used glucocorticoid therapy for at least 1 day in past 30/90 days since study entry.

c

Mean peak dose in past 30/90 days=mean peak dose in past 30/90 days in subjects who used glucocorticoid therapy for at least 1 day in past 30/90 days.

AIC, Akaike information criterion; PEQ, prednisolone equivalent; WCD, weighted cumulative dose.