Figure 4.

Maternal albumin-adjusted serum calcium concentration (A), calcium:creatinine ratio (B), and serum parathyroid hormone concentration (C), by time. Group means were modeled using piecewise linear regression (panels A and B) and generalized estimative equations (GEE). (A) Knot was placed at day 60, and the change in slope at that point was statistically significant in the vitamin D group (p < 0.01), but not in the placebo group. Associations: before knot in vitamin D group = 0.044 mmol/L per month, (p < 0.001); after knot in vitamin D group = −0.028 mmol/L per month, (p = 0.141); before knot in the placebo group = 0.017 mmol/L per month, (p < 0.01); after knot in the placebo group = 0.000 mmol/L per month, (p = 0.997). (B) Calcium:creatinine ratio was square-root transformed prior to regression analysis. A knot was placed at day 14, where the change in slope was statistically significant in the vitamin D group (p < 0.01), but not in the placebo group. Associations: before knot in the vitamin D group = 0.147 mmol/mmol per month, (p < 0.05); after knot in vitamin D group = −0.042 mmol/mmol per month, (p < 0.05); before knot in placebo group = 0.052 mmol/mmol per month, (p = 0.311); after knot in placebo group = −0.048 mmol/mmol per month, (p < 0.01). Varying the knot location in sensitivity analyses did not substantially change the inferences from panels A or B. (C) Parathyroid hormone concentrations were log transformed prior to regression analyses. There was a significant increase over time in the placebo group (0.01 log-pmol/L [95% CI: 0.003, 0.008] for each day increase in follow-up time; p < 0.001), but the slope of the vitamin D group was significantly attenuated (0.007 log-pmol/L [95% CI: -0.01, -0.004] lower than in the placebo group; p < 0.001).

Roth et al. Nutrition Journal 2013 12:47   doi:10.1186/1475-2891-12-47
Download authors' original image