Effect of Consuming Vitamin A-fortified Sugar on Serum Retinol Levels in Children

A randomized, double-blind, controlled clinical trial (RCT) with a parallel-groups design was conducted on a sample of 88 children aged 3 to 6 years, of both sexes, who received a biweekly food package from a food bank and were enrolled in schools in the Agua Blanca district of Cali between July and December 2025. The sample size was calculated based on the difference in serum retinol (SR) levels observed in studies between the intervention and control groups; using the effect size and standard deviations of both study groups, the sample size formula for the difference in means in RCTs was applied. The sample size was calculated using the formula proposed in 2007 to detect differences in means in RCTs, taking into account the group assignment factor and the margin of tolerance; for this purpose, values reported by researchers in 2006 were taken into account; with a minimum detectable difference of 0.3 μmol/L, the combined variance was estimated at 0.14845, the expected power was 85% with a 95% confidence level; the proportion of imbalance considered was 75% to give greater weight to the intervention group; the margin of tolerance for the minimum detectable difference in RS was set at 0.06 μmol/L; and a 15% margin was included to account for search efforts during recruitment. The resulting sample size was 89 children (minimum 77 children), distributed between the control group (GCON): 40, and the intervention group (GINT): 49; allocation was carried out by randomly assigning the institutions to which the children belonged based on their RS levels. This decision was made in light of the possibility that vitamin A deficiency was widespread among the children (a pattern similar to that observed among the lowest socioeconomic strata in Colombia) and, in particular, due to the ethical implications of including children with vitamin A deficiency in a control group of the same size as the intervention group; furthermore, it was observed that the preparation and distribution of food in daily meals-which varied by institution-were difficult for the research team to control, which could compromise the blinding of the intervention at each institution if children from both experimental groups were included simultaneously at the same institution. Randomization and Group Definition: After deworming and based on the initial RS results for all children from the randomized centers who agreed to participate, two study groups (GCON and GINT) were formed based on the results of the first RS test from the seven centers. The 89 children from the randomized centers were assigned to the two study groups according to their RS level as follows: GCON: 40 in 3 centers; GINT: 49 in 4 centers.

Variables:

• Sociodemographic: age, sex, health insurance coverage, and self-reported ethnic origin of the child; educational level and occupation of caregivers; housing status; neighborhood of residence; family type; and family size.
• Nutritional status was assessed at baseline by estimating z-scores for the following anthropometric indicators: weight-for-height (w/h), weight-for-age (w/a), height-for-age (h/a), body mass index (BMI), and mid-upper arm circumference (MUAC). Anthropometric measurements were taken one week after the first blood sample collection, using World Health Organization (WHO) techniques; the coordinating nutritionist standardized the procedures for the team of nutritionists responsible for taking these measurements. This procedure was repeated at the end of the retinol-fortified sugar (AFR) intervention.
• The biomarkers measured were serum retinol (SR) and C-reactive protein (CRP) at enrollment, following approval of the RCT by the University's Ethics Committee, the signing of the informed consent form, and confirmation of a single-dose antiparasitic treatment, in order to control for the confounding effect of the possible presence of intestinal parasites that could affect serum iron levels and help improve nutrient absorption. On the fourth day after deworming and at the end of the RCT, blood samples were collected at each center. The necessary information was collected at the start of the RCT. Fasting blood samples were drawn in the presence of the assigned nutritionist and teacher and analyzed at a clinical laboratory outside UValle. This involved collecting 5 ml of whole blood per child via an antecubital vein puncture, performed by a bacteriology technician following the laboratory's standardized procedure. To determine subclinical vitamin A deficiency, a value of <20 µg/dL (ER) was used; ultra-sensitive PCR analysis (>0.5 mg/L) was also taken into account to adjust RS results based on the presence of clinical signs of infection at the start and end of the child's follow-up; the technique used by the laboratory for PCR was immunoturbidimetric.
• Physical health: assessed at enrollment (presence of fever, diarrhea, respiratory infection, among others), as reported by the child's teacher and confirmed by the caregiver; at the time of the initial and final blood sample collections (verified with the child's caregiver by the nutritionist); and on three occasions during the AFR intervention (three months), verified with each caregiver.
• Food intake: This was assessed after the first blood sample, both at home and at the institution, using: - The semiquantitative frequency of consumption of 22 food groups at the start of the study (with the cooks at each institution; information verified through daily cafeteria records and the children's caregivers), -Two 24-hour recall surveys (24-hour R) of consumption at the center and in the child's home (1 at enrollment and 1 at the end of the RCT), in a five-stage process using the iterative multi-step 24-hour recall method, and additionally, -semi-quantitative frequency of consumption of supplements and foods that are sources of vitamin A (ER) and carotenoids (52 foods), on three occasions during the study.

Intervention and Blinding:

The RCT began in July 2025 with the selection, training, and orientation of the field team-composed of five nutritionists (one supervising nutritionist and four field nutritionists)-and the random selection of educational institutions; Following approval by the Ethics Committee (July), the necessary data on the variables were collected (July-August), and the AFR intervention began on September 15.

The intervention consisted of a daily dose per child of 13 g of white sugar-commonly consumed in the country-refined and fortified with water-dispersible retinol palmitate (23.5 mcg ER/g) by a national sugar manufacturer and distributed by the Food Bank, to be used as a culinary ingredient in the preparation of daily meals for three months (September through December), following the children's deworming and the first blood sample collection; furthermore, at the start of the intervention, exclusion criteria that could affect vitamin A metabolism were verified. The definition of the amount of free sugar established as a reference for children's daily consumption-whether fortified or unfortified-based on moderate physical activity, age, and sex, utilized Colombia's RIEN (recommended average sugar intake by age and sex corresponding to 10% of total daily caloric energy: 32.5 g/day). Based on these data, it was decided to provide each child (control group: GCON and intervention group: GINT) with less than half the amount recommended by the RIEN for free sugars (13 g/day: 1 heaping tablespoon), with a daily retinol intake of 305.5 mcg/day for children in the GINT group of both sexes. The RCT ended on December 16, 2025, after the relevant data had been collected again. The laboratory responsible for analyzing the blood samples did not know which group each child's initial and final samples belonged to. The four field nutritionists resided in the city where the RCT was conducted and were assigned to specific institutions (ND-EE); the study was supervised at all times by another nutritionist (ND-S) with the support of the coordinating nutritionist. The pediatrician advised the ND-EE on all matters related to deworming, the children's health status, and the monitoring of potential adverse events or incidents during the AFR intervention, so that UValle could activate its event insurance. No nutritionist on the staff responsible for data collection knew which RCT group their institution belonged to, and at no time was sugar consumption promoted. All products containing sugar as an ingredient were kept at each institution and in each household; ND-EE staff verified their presence every two weeks in households and weekly at the institutions. The packaging of the two types of sugar was camouflaged; the AFR was white and similar in appearance to sugar commercially available in the country. Before the intervention began, taste acceptability tests were conducted on the AFR in liquid preparations. The clear plastic bags used to package the sugar were the same ones the Food Bank routinely used to package food distributed to the institutions.

Statistical analysis:

Descriptive analyses included absolute and relative frequencies, as well as trends, which served as an initial reference. The nutritional assessment was performed using the WHO's Anthro and AnthroPlus software programs. Classification was based on Resolution 2465 of 2016 issued by the country's Ministry of Health and Social Protection. Log-normal distributions were assumed to characterize dietary intake and SR. The comparative inference phase for the study groups (GCON and GINT) focused on determining the statistical power required to validate the effect of the experimental design (ECA) on SR levels and other indicators. Robust tests, such as the Wilcoxon signed-rank test for paired data, were applied to establish a clear distinction between the two groups (GCON vs. GINT) and the time points (before the intervention vs. after the intervention). The Durbin-Conover test was incorporated as a valid alternative to the t-test for paired samples with heterogeneous variances, allowing for comparisons between groups with unequal sample sizes. The categorical anthropometric comparison between the groups (GCON and GINT), both at baseline and at follow-up, was performed using Pearson's chi-square test and Fisher's exact test. To minimize the risk of false positives (Type I error), p-values were corrected using the Benjamini-Hochberg method for multiple comparisons. Finally, several exploratory mixed-effects models were applied to capture longitudinal dynamics and interactions among variables, which required a broader perspective than point-in-time comparisons. These models allowed the total variance in the study to be broken down into two components: fixed effects (primary variables of interest) and random effects (which captured the variance inherent to each individual participant). Unlike traditional methods, mixed-effects models can effectively handle missing data and temporal correlations among repeated measurements, allowing the model to adapt to individual RS trajectories. Analyses were performed using R-Project software v4.5.2, and the significance level for statistical interpretation was set at 5%.

Ethics:

The RCT was approved by the Uvalle Health Ethics Committee (approval no. 011-2025). It was conducted in accordance with Resolution No. 8430 of 1993 issued by the Colombian Ministry of Health and was classified by the Committee as a study involving more than minimal risk. Furthermore, it complied with the World Medical Association's Declaration of Helsinki. For the study design and conduct of the RCT, the ICH E6 (R3) guidelines and good clinical practices were followed. The confidentiality of the children's data was ensured by preserving their identity through the use of a code, and it was not necessary to modify the informed consent form.