For

participants in England, the last date of follow-up was March 31, 2008; and for participants in Scotland the last date of follow-up was December 31, 2008. Cox regression models with attained age as the underlying time variable were used to estimate relative risks (RR) and 95% confidence intervals AC220 for incident ankle, wrist, and hip fractures by BMI and physical activity. Analyses were stratified by recruitment region (ten regions) and adjusted for: socio-economic status (quintiles using the Townsend index [22]), smoking status (current, past, never), alcohol consumption (0, 1–2, 3–6, 7–14, ≥ 15 drinks per week), menopausal hormone therapy use (never, past, current), diabetes (yes, no), history of heart disease/thrombosis (yes, no), history of osteo/rheumatoid arthritis (yes, no),

thyroid disease (yes, no), and height (< 155, 155.0 to 159.9, 160 to 164.9, 165.0 to 169.9, or ≥ 170 cm). Depending on the model, additional adjustments included: BMI (< 20, 20.0–22.4, 22.5–24.9, 25.0–27.4, 27.5–29.9, ≥ 30.0 kg/m2), and strenuous physical activity (rarely/never (inactive), BIBF 1120 molecular weight at most once per week, or more than once per week). Missing data for the adjustment variables (generally < 2% for each variable) were assigned to an additional category. The RRs were treated as floated absolute risks [23] when more than two categories were used for risk comparisons, and given with corresponding floated confidence intervals (FCIs), so that valid comparisons can be made between any two groups. When only two categories are compared or when log-linear

trends in risk are quoted, conventional confidence intervals are used. To ensure that the impact of measurement error was minimised, category specific relative risks based on self-reported data were plotted against mean measured BMI values within each category. Age-specific incidence rates per 100 women over 5 years were calculated for each fracture site for 5-year age groups from 50–54, to 80–84 years. Cumulative risks from ages 50 to 85 were calculated for each fracture site, taking the average hazard rate over this time period to be the uniformly age-standardised incidence rate per person-year. Cumulative absolute incidence rates for women aged Linifanib (ABT-869) from 50 to 79 were also calculated for each fracture site according to BMI and strenuous physical activity categories. To allow for potential non-proportional hazards, such as might be associated with the dramatic increase in incidence of hip fractures with age, we analysed the data in 10 year age bands. For each fracture type and exposure, category-specific relative risks were converted to incidence rates by multiplying them by the appropriate age-specific incidence rate, divided by a weighted average of all relative risks [24]. These incidence rates were age-standardised across the full age range from 50 to 79 and used to compute cumulative risks as above.