Objective: To examine (i) the association of percentage of total energy intake from protein (protein intake %) with bone mineral density (BMD, g/cm2) and bone loss at the femoral neck, trochanter and lumbar spine (L2-L4) and (ii) Ca as an effect modifier.

Setting: The Framingham Offspring Study.

Subjects: Men (n 1280) and women (n 1639) completed an FFQ in 1992-1995 or 1995-1998 and underwent baseline BMD measurement by dual-energy X-ray absorptiometry in 1996-2000. Men (n 495) and women (n 680) had follow-up BMD measured in 2002-2005.

Design: Cohort study using multivariable regression to examine the association of protein intake % with each BMD, adjusting for covariates. Statistical interaction between protein intake % and Ca (total, dietary, supplemental) intake was examined.

Results: The mean age at baseline was 61 (sd 9) years. In the cross-sectional analyses, protein intake % was positively associated with all BMD sites (P range: 0·02-0·04) in women but not in men. Significant interactions were observed with total Ca intake (<800 mg/d v. ≥800 mg/d) in women at all bone sites (P range: 0·002-0·02). Upon stratification, protein intake % was positively associated with all BMD sites (P range: 0·04-0·10) in women with low Ca intakes but not in those with high Ca intakes. In the longitudinal analyses, in men, higher protein intake % was associated with more bone loss at the trochanter (P = 0·01) while no associations were seen in women, regardless of Ca intake.

Conclusions: This suggests that greater protein intake benefits women especially those with lower Ca intakes. However, protein effects are not significant for short-term changes in bone density. Contrastingly, in men, higher protein intakes lead to greater bone loss at the trochanter. Longer follow-up is required to examine the impact of protein on bone loss.