The use of anthropometric measures in the prediction of incident gout: results from a Swedish community-based cohort study

Objectives: To study associations between different anthropometric measures and incident gout, and to find the best predictive measure. Method: We used the baseline investigation from the Malmö Diet and Cancer study, excluding cases of prevalent gout (n = 28 081). Cox regression for each anthropometric measurement was calculated per standard deviation increment for men and women, with hazard ratios (HRs) and 95% confidence intervals (CIs), using a hospital diagnosis of incident gout (M10) during follow-up as the outcome. Incremental C-statistics for each anthropometric measure were used to determine the measure with the best predictive capacity, in models adjusted for age, socio-economic data, lifestyle factors, comorbidities, and antihypertensive medications. Results: The study population included 11 049 men and 17 032 women, with 633 incident gout cases, 393 in men (3.6%) and 240 in women (1.4%). For both men and women, the five anthropometric measurements with highest C-statistics were weight, body mass index (BMI), waist circumference (WC), hip circumference, and waist-to-height ratio; in men, the measurement with the highest C-statistic was BMI (0.7361; fully adjusted HR 1.52, 95% CI 1.39-1.68), and in women WC (0.8085; fully adjusted HR 1.62, 95% CI 1.46-1.81). The increment in C-statistic with anthropometric measures was good, around 0.035. Waist-to-hip ratio, waist-to-hip-to-height ratio, body fat percentages, and especially A Body Shape Index had lower C-statistics. Conclusions: Both BMI and WC showed good predictive ability for incident gout. The clinically used cut-offs for BMI and WC appeared to be relevant in the assessment of increased risk of gout.