The association between sex and body size and stroke risk and survival in patients with atrial fibrillation - an analysis of 71,589 patients from the COMBINE-AF study

Abstract Background In patients with atrial fibrillation (AF), the risk of stroke is higher and mortality is lower in females than males. Whether a smaller body size in female patients is associated with outcomes is uncertain. Purpose We assessed the risk of stroke/systemic embolism (SE), major bleeding, cardiovascular (CV) death and all-cause death in female vs male patients with AF treated with warfarin or a DOAC, and sought to determine whether the risks are modified by demographics and clinical characteristics including weight and height. Methods We used pooled patient-level data from 4 large RCTs investigating DOACs vs warfarin (ARISTOTLE, ENGAGE AF-TIMI 48, RE-LY, ROCKET) from the COMBINE-AF cohort. Stepwise Cox-regression analyses were used to adjust for factors modifying the associations with outcomes. Four models were constructed as follows: Model 0: no adjustment; Model 1: age + trial + randomized treatment; Model 2A: model 1 + prior stroke/TIA + congestive heart failure + diabetes + hypertension + vascular disease; Model 2B: model 1 + creatinine + weight + height; and Model 3: model 2 (A+B) + region + type of AF + VKA experience. Results Data were available for 71,589 patients (26,659 females). Median age was 73 [IQR 67-78] and 71 [63-77] years, median height 160.0 [155.0 – 165.0] and 174.0 [168.0 – 180.0] cm, median weight 73.0 [63.0 – 85.0] and 85.3 [75.0 – 98.2] kg in females and males, respectively. During 2.3 [1.50, 2.67] years median follow-up, there were 2,494 stroke/SE, 1,960 ischaemic stroke/SE, 648 intracranial bleeding, 1,584 gastrointestinal bleeding, 3,660 CV death and 5,818 all-cause death. Yearly incidence rates of stroke/SE 1.99 [1.88, 2.12] vs 1.59 [1.51, 1.68] and ischaemic stroke/SE 1.60 [1.49, 1.71] vs 1.23 [1.16, 1.31] were higher in females than males. In Cox models, female sex was not independently associated with stroke after adjustment for height and weight (Figure 1). Rates of gastrointestinal bleeding 1.16 [1.06, 1.26] vs 1.32 [1.24, 1.40], CV death 2.23 [2.10, 2.35] vs 2.69 [2.59, 2.80] and all-cause death 3.56 [3.40, 3.72] vs 4.27 [4.13, 4.40] were lower in females vs males. Female sex was independently associated with less bleeding and lower mortality, and with a greater magnitude after adjustment for height and weight (Figure 1). Figure 2 displays the inverse associations between height and weight across the range for stroke and up to 80 kg body weight and 170 cm length for CV mortality. Conclusion In patients with AF treated with an OAC, there is a higher rate of stroke in females than males but no independent association between female sex and the risk of stroke/SE. There is a continuous independent inverse association between a smaller body size and a higher risk of stroke/SE, and also an independent inverse association between a smaller body size and higher CV death up to a body weight of 80 kg and height of 170 cm.Figure 1.Cox regression analyses Figure 2.Predicted event rates