Urinary Sodium and Potassium, and Risk of Ischaemic and Haemorrhagic Stroke (INTERSTROKE): a case-control study.

Judge, Conor; O'Donnell, Martin J; Hankey, Graeme J; Rangarajan, Sumathy; Chin, Siu Lim; Rao-Melacini, Purnima; Ferguson, John; Smyth, Andrew; Xavier, Denis; Lisheng, Liu; Zhang, Hongye; Lopez-Jaramillo, Patricio; Damasceno, Albertino; Langhorne, Peter; Rosengren, Annika; Dans, Antonio L; Elsayed, Ahmed; Avezum, Alvaro; Mondo, Charles; Ryglewicz, Danuta; Czlonkowska, Anna; Pogosova, Nana; Weimar, Christian; Diaz, Rafael; Yusoff, Khalid; Yusufali, Afzalhussein; Oguz, Aytekin; Wang, Xingyu; Lanas, Fernando; Ogah, Okechukwu S; Ogunniyi, Adesola; Iversen, Helle K; Malaga, German; Rumboldt, Zvonko; Oveisgharan, Shahram; Hussain, Fawaz Al; Yusuf, Salim

BACKGROUND Although low sodium intake (<2g/day) and high potassium intake (>3·5g/day) are proposed as public health interventions to reduce stroke risk, there is uncertainty about the benefit and feasibility of this combined recommendation on prevention of stroke and its subtypes. METHODS We obtained random urine samples from 9,275 cases of acute first stroke and 9,726 matched controls (8,761 matched pairs for conditional analysis) from 27 countries and estimated the 24-hour sodium and potassium excretion, a surrogate for intake, using the Tanaka formula. Using multivariable conditional logistic regression, we determined the associations of estimated 24-hour urinary sodium and potassium excretion with stroke and its subtypes. RESULTS The mean estimated 24-hour sodium and potassium urinary excretion was 3·29g/day and 1·57g/day, with 0·01% of participants having both low sodium (<2·0g/day) and high potassium excretion (>3·5g/day). There was a moderate positive correlation between sodium and potassium excretion (r=0·4435, P<0.001) and between sodium excretion and blood pressure (P<0.001). Compared with an estimated urinary sodium excretion of 2·8-3·5g/day (second quartile, reference), higher (>4·26g/day) (OR 1.81;95%CI,1.65-2.00) and lower (<2·8g/day) sodium excretion (OR 1.39;95%CI,1.26-1.53) were significantly associated with increased risk of stroke. The stroke risk associated with the highest quartile of sodium intake (sodium excretion >4·26g/day) was significantly greater (P<0.001) for intracerebral haemorrhage (ICH) (OR 2.38;95%CI,1.93-2.92) than for ischemic stroke (OR 1.67;95%CI,1.50-1.87), and greater for large vessel and small vessel ischemic stroke than for cardioembolic ischemic stroke. Urinary potassium was inversely and linearly associated with risk of stroke, and stronger for ischemic stroke than ICH (P=0.026). In an analysis of combined sodium and potassium excretion, the combination of high potassium intake (>1·58g/day) and moderate sodium intake (2.8-3.5 g/day) was associated with the lowest risk of stroke. CONCLUSION The association of sodium intake and stroke is J-shaped, with high sodium intake a stronger risk factor for intracerebral haemorrhage than ischemic stroke. Our data suggest that moderate sodium intake - rather than low sodium intake - combined with high potassium intake may be associated with the lowest risk of stroke and expected to be a more feasible combined dietary target.