Random serial sampling to evaluate efficacy of iron fortification: a randomized controlled trial of margarine...

Random serial sampling is widely used in population pharmacokinetic studies and may have advantages compared with conventional fixed time-point evaluation of iron fortification.

Our objective was to validate random serial sampling to judge the efficacy of iron fortification of a low-fat margarine.

We conducted a 32-wk placebo-controlled, double-blind, iron-intervention trial in 18-40-y-old Swiss women (n = 142) with serum ferritin (SF) concentrations <25 μg/L. Women were randomly assigned to 3 groups to receive 20 g margarine, with 14 mg added iron as either micronized ground ferric pyrophosphate (MGFePP) or sodium iron edetate (NaFeEDTA), or placebo daily. We measured hemoglobin and iron status of subjects at 2 fixed time points (at baseline and the endpoint) plus 3 randomly assigned time points between 4 and 28 wk. With the use of bootstrapping, the number of observations per individual was reduced to 3 and then compared with the 5-time-point data. Mixed-effects models were used to estimate iron repletion over time for random sampling, and analysis of covariance was used for fixed time-point sampling.

Body iron stores increased in women who received MGFePP or NaFeEDTA compared with women who received placebo (P < 0.05). The increase in body iron stores with NaFeEDTA fortification was 2-3 times the increase with MGFePP fortification (P < 0.05); the difference was more marked in women with baseline SF concentrations <15 μg/L (P < 0.05). Random serial sampling reduced the required sample size per group to one-tenth of that for 2 fixed time points. Compared with the 5-time-point analysis, the 3-time-point sparse sampling generated comparable estimates of efficacy.

When used to evaluate the efficacy of iron fortificants, random serial sampling can reduce the sample size, invasiveness, and costs while increasing sensitivity. Random serial sampling more clearly describes the pattern of iron repletion and may prove useful in evaluating other micronutrient interventions.