The Journal of Nutrition Letters to the Editor
Response to Dr. Sempos and Dr. Picciano1
TABLE 1
Subgroup analysis comparing baseline and 1-y change in 25(OH)D among those who changed oral contraceptive use during the study period in the placebo group and in those receiving 20 mg/d cholecalciferol1
Contraceptives discontinued Placebo Treatment Contraceptives started Placebo Treatment Remaining participants Placebo Treatment All participants Placebo Treatment
n
Baseline 25(OH)D
1 5
74.75 71.9 6 39.0
added back into the revised data analysis, the overall results reported in the manuscript did not change substantially. Although means and frequencies were slightly different, there were no differences in significance between any of the analyses for baseline data or for the response to supplementation. However, there was a significant inverse correlation between baseline 25(OH)D and the seasonal increase in serum 25(OH)D (r ¼ 20.237; P ¼ 0.017), which was not seen in the original analysis. The requested analysis of covariance data for the 1-y change in serum 25(OH)D concentration is presented in Table 2. After the revised data analysis for those participants supplemented with 20 mg/d cholecalciferol, 81% of the participants had optimal serum 25(OH)D concentrations at the end of winter, compared with 37% of participants in the placebo group. The net treatment effect resulting from the revised data analysis was 24.7 nmol/L, which corresponds to an increase of 1.1 nmol/L for every mg of daily supplemental cholecalciferol intake. Urinary calcium to creatinine ratio was used in addition to serum calcium levels to monitor safety of the 20 mg/d supplements of cholecalciferol, according to the discussion by Vieth et al. (3). A slight elevation in the calcium to creatinine ratio in the context of normal serum calcium was not considered a meaningful adverse reaction.
TABLE 2
1-y change in 25(OH)D
Analysis of covariance for variables affecting the 1-y change in 25(OH)D concentration in those participants receiving a placebo (n ¼ 34) or 20 mg/d cholecalciferol (n ¼ 61)
nmol/L 217.0 15.6 6 39.8a
3 3
39.5 6 8.8 35.9 6 2.1
26.6 6 8.2 85.1 6 21.9b
31 55
61.9 6 22.6 62.1 6 24
10.9 6 16.9 35.3 6 23.2a
35 63
60.3 6 22.1 61.9 6 25.1
11.4 6 17.3 36.1 6 27.2
1
Data are means 6 SD. ANOVA with Tukey’s post-hoc analysis used to compare groups. There was no significant differences in baseline 25(OH)D in the placebo or treatment groups or in the 1-y change in the placebo group. Means in a column without a common letter differ, P , 0.05.
1 Author disclosures: B. W. Hollis is a paid consultant for Diasorin Corp; M. L. Nelson, J. M. Blum, C. Rosen, and S. S. Sullivan, no conflicts of interest.
Source Corrected model Intercept Baseline 25(OH)D Seasonal change Estrogen dose Skin type Mean percent body fat Change in dietary vitamin D intake 2005 to 2006 Change in sun exposure Tanning bed use 2005–2006 Started or quit contraceptives Randomization Error Total Corrected total 1
Type III Sum of Squares 34619.336 2081.110 4781.337 6122.586 3612.357 5.682 1654.072 55.022
1
2.988 879.865 1203.328 13626.789 32790.835 141820.188 67410.171
df
F
P
10 1 1 1 1 1 1 1
8.868 5.331 12.248 15.684 9.254 0.015 4.237 0.141
0.000 0.023 0.001 0.000 0.003 0.904 0.043 0.708
1 1 1 1 84 95 94
0.008 2.254 3.083 34.908
0.930 0.137 0.083 0.000
R Squared ¼ 0.514 (Adjusted R Squared ¼ 0.456).
0022-3166/08 $8.00 ª 2009 American Society for Nutrition. J. Nutr. 139: 1205–1206, 2009. Manuscript received April 13, 2009. Initial review completed April 17, 2009. Revision accepted April 22, 2009. First published online May 6, 2009; doi:10.3945/jn.109.108738.
1205
Downloaded from jn.nutrition.org by guest on February 19, 2013
We thank Dr. Sempos and Dr. Picciano (1) for their comments on our article (2) and for the opportunity to clarify the data analysis. We have reanalyzed the data as suggested, adding back all participants who were excluded during the original data analysis. Mean baseline serum 25-hydroxyvitamin D [25(OH)D] levels are presented in Table 1. The mean seasonal (February to September) change in serum 25(OH)D was 40.7 6 26.6 nmol/L (n ¼ 100), an increase of 68%, after the revised data analysis. There were no significant differences in baseline characteristics between the treatment (n ¼ 67) and placebo (n ¼ 36) groups. Upon reanalysis of the data, 27% of participants had optimal serum 25(OH)D concentrations ($75 nmol/L) at baseline and 43% had concentrations ,50 nmol/L. The 4 participants who were excluded from the study because their 25(OH)D concentrations were outside the range established for inclusion were excluded in March 2005 based on safety concerns. In the original data analysis, we excluded data from 12 women who either initiated or discontinued using hormonal contraceptives. The change in exogenous estrogen supply between data collection points biased the serum 25(OH)D response to supplementation (Table 1). However, when these 12 participants were
Monica L. Nelson Susan S. Sullivan* Department of Food Science and Nutrition, University of Maine, Orono, ME 04469
Literature Cited 1.
2.
James M. Blum Dietary Supplement Research Group, Saco ME 04072 Bruce W. Hollis Department of Pediatrics, Medical University of South Carolina, Charleston, SC 29425 Clifford Rosen Maine Medical Center Research Institute, Scarborough, ME 04041
3.
Sempos CT, Picciano MF. The intention to treat principal, and the potential impact of excluding data from the analysis of clinical trial data. J Nutr. 2009;139:1204. Nelson ML, Blum JM, Hollis BW, Rosen C, Sullivan SS. Supplements of 20mg/day cholecalciferol optimized serum 25-hydroxyvitamin D concentrations in 80% of premenopausal women in winter. J Nutr. 2009;139:540–6. Vieth R, Chan PCR, MacFarlane GD. Efficacy and safety of vitamin D3 intake exceeding the lowest observed adverse effect level. Am J Clin Nutr. 2001;73:288–94.
* To whom correspondence should be addressed. E-mail:
[email protected].
Downloaded from jn.nutrition.org by guest on February 19, 2013
1206
Letters to the Editor
