The reproducibility of bone-related biochemical variables in post-menopausal women.

Seventeen measured and derived bone-related biochemical variables were determined in 60 postmenopausal women on two occasions from 1 to 35 months apart. The coefficients of determination (r2) between the first and second observations were highly significant in respect of all measured and calculated variables except for anion gap, calculated complexed calcium, fasting urinary sodium:creatinine ratio and urinary phosphate:creatinine ratio. The variables with coefficients of determination which ranked highest include plasma alkaline phosphatase, albumin, globulins, calculated ionized calcium and fasting urinary corrected calcium:creatinine and hydroxyproline: creatinine ratios. They also showed marked individuality suggesting that changes within the conventional population based reference range possibly indicate significant physiological changes in metabolism. The critical difference for two results to be significantly different (P less than 0.05) was calculated from the total variation for each variable. We conclude that most of the measured and derived bone-related biochemical variables in fasting plasma and urine samples are sufficiently reproducible in postmenopausal women to be clinically and physiologically meaningful but that urinary sodium and phosphate are inconstant and probably dependent on the previous day's dietary intake.