Microsoft Word - NEF119BF

Dr. Troels Ring, MD, Medical Department C, Aalborg Hospital, Aalborg (Denmark) Dear Sir, Two recent papers comparing calcium carbonate with calcium acetate both conclude that calcium acetate is the better phosphorus binder. In the paper by Caravaca et al. [1], the serum phosphate concentration was 1.80 ± 0.50 (SD) mmol/l while on calcium acetate, and 1.93 ± 0.48 on calcium carbonate in a parallel study on 31 plus 35 patients. Using the pooled estimate of variance, Varriiff = SD? · (n, – 1) + SO\ · (n2 – 1) n, + n2 – 2 and SDerrordiff = ]/vardiff(– +–j, the standard error on the difference is readily obtained as 0.121 mmol/l and hence a 95% confidence interval of -0.37 mmol/l to 0.11 mmol/l, i.e. a nonsignificant difference. The authors report a significant increase from 1.77 ± 0.35 (while on aluminum hydroxide) to 1.93 ± 0.48 mmol/l on calcium carbonate, with p = 0.03 using a paired t test. In contrast, a nonsignificant increase from 1.73 ± 0.25 to 1.80 ± 0.50 mmol/l was found comparing aluminum hydroxide to calcium acetate. From that, the authors conclude that calcium acetate is better. In the first t test, however, we can see that t must have been 2.2650, and the standard error of the mean difference therefore (1.93-1.77)/t, that is 0.07064. Likewise, the standard error in the second test is 0.06097. Since the standard deviations in the two differences can be computed as the square root of the group size multiplied by the standard errors of the means above and since a pooled estimate of the variance of the difference between the two differences and the relevant standard error of the mean difference can be found as above, an unpaired t test of the difference between the difference yields = 0.95. (1.93 – 1.77) – (1.80 – 1.73) 0.0945 With 64 degrees of freedom this gives a p value of 0.35. Hence, the authors’ informal comparison of a significant with an insignificant change appears to be mistaken. In the paper by Morinière et al. [2] 8 patients were compared while being treated with half the dose of elemental calcium as calcium acetate compared to calcium carbonate. The authors were unable to find a difference between 1.74 ± 0.32 mmol/l (on calcium carbonate) and 1.75 ± 0.38 mmol/l (second course of calcium acetate). The power of the paired t test used was not specified. Hence it is difficult to accept this as a proof of the superiority of calcium acetate, especially since more magnesium hydroxide was given with calcium acetate. An estimate of the confidence interval is obtained using the relation σ2diff = σ, + σ2 – 2 · σ, · σ2 · p (1,2),