OP0031 The importance of assessing multiplicative and additive interaction: examining the effect of glucocorticoid therapy on mortality in patients with rheumatoid arthritisand concomitant type ii diabetes. (12th June 2018)

• Record Type: Journal Article
• Authors: Costello, R.; Marsden, A.; Movahedi, M.; Lunt, M.; Emsley, R.; Dixon, W.G.
• Is Part Of: Annals of the rheumatic diseases. Volume 77(2018)Supplement 2
• Page Start: 65
• Page End: 65
• Publication Date: 2018-06-12
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2018-eular.1755
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background: Glucocorticoids (GC) are widely used to treat rheumatoid arthritis (RA), however they are known to have risks associated with them. It has been shown that GCs increase the risk of diabetes mellitus (DM). A few studies have investigated the long-term effects of GC use on outcomes in DM, but not in RA specifically. As people with RA already have increased risk of cardiovascular (CV) disease, the additional burden of DM and GCs may be important. If the effect of GCs was dependent on DM we would say there is effect modification and this can be on the additive scale, corresponding to variation in the absolute treatment effect, e.g. the risk difference (RD), across DM status, or the multiplicative scale, corresponding to variation in the relative treatment effect e.g. the rate ratio (RR). 1 Objectives: To examine in patients with RA 1) whether all-cause and CV mortality rates differ by GC and DM status, and 2) whether DM modifies the relationship between GC and all-cause and CV mortality on multiplicative and additive scales. Methods: Patients with RA and linkage to mortality data were identified from the Clinical Practice Research Datalink (n=9085), a database of primary care electronic medical records in the UK. RR and RD for ever GC use were calculated by DM status. Cox proportional hazards (PH) regression models were fitted with an interaction term for DM and use of GC to assess multiplicative interaction. Additive interaction was measured with the Relative Excess Risk due to Interaction (RERI) 2 where a value different from zero indicates a difference in the absolute effect of treatment. Results: Those with DM and ever treated with GCs had a 3-fold increased all-cause mortality RR (95% CI: 2.27, 4.09) whilst those without DM had a slightly higher RR (3.46 (95% CI: 2.95, 4.07)). However those with DM had a higher RD: 36.46 deaths per 1000 patient years (pyrs) (95% CI: 27.5, 45.41) compared to those without DM: RD 22.83 deaths per 1000 pyrs (95% CI: 19.83, 25.82) because of higher baseline mortality rates. A similar pattern was seen for CV mortality. The adjusted Cox PH model for all-cause mortality showed no evidence of multiplicative interaction, but there was significant additive interaction (RERI 0.86 (95% CI: 0.18, 1.54)). For CV mortality there was no interaction on either scale. Conclusions: Methodologically, this study showed assessing interaction on the additive and multiplicative scales can lead to different conclusions and should be considered carefully. In this study significant interaction was seen on additive scale but not on the multiplicative scale due to higher baseline rates in patients with DM. Clinically, this study provides some evidence that long-term GC therapy may be particularly harmful in patients with RA and DM. References: [1] VanderWeele TJ, Knol MJ. A tutorial on interaction. Epidemiologic Methods2014;3(1):33–72. [2] Li R, Chambless L. Test for additive interaction in proportional hazards models. Annals of epidemiology2007Mar;17(3):227–236. Disclosure of Interest: None declared