Materials and methods: A survey of trace metals in 429 waterworks, supplying 64% of the population in Norway,
was linked geographically to the home addresses of patients with incident hip fractures (1994–2000). Drinking
water mineral concentrations were divided into “low” (belowand equalwaterworks average) and “high” (above
waterworks average). Poisson regression models were fitted, and all incidence rate ratios (IRRs) were adjusted
for age, geographic region, urbanization degree, type of water source, and pH of the water. Effect modifications
were examined by stratification, and interactions between calcium and magnesium, copper, zinc, iron and
manganese were tested both on the multiplicative and the additive scale. Analyses were stratified on gender.