A new critique of the JASMINE Trial is drawing attention to one of the most difficult problems in clinical research: determining whether an observed health effect is caused by an intervention itself or by the nutritional circumstances surrounding the people who receive it. In a paper published in the Journal of Perinatology, B.C. King and A. Malhotra examine what they describe as nutritional confounding and reporting limitations in the trial. Their analysis highlights how apparently convincing results can become difficult to interpret when participants’ diets, nutritional status, or access to food are not measured and reported with sufficient detail.
Nutritional confounding occurs when nutrition is linked both to the treatment being studied and to the outcome researchers are trying to explain. If participants who receive an intervention also differ in their intake of calories, protein, vitamins, minerals, or culturally specific foods, those differences may influence the results independently of the intervention. The problem is especially important in perinatal research, where maternal nutrition can affect fetal growth, birth weight, gestational development, metabolic health, and a range of outcomes observed during pregnancy or shortly after birth. Without carefully accounting for nutrition, an apparent treatment effect may partly reflect the biological consequences of diet.
The critique of the JASMINE Trial arrives at a moment when medical researchers are increasingly reassessing how clinical evidence is generated and communicated. Randomized trials are widely regarded as the strongest method for testing whether an intervention works because random allocation is intended to balance known and unknown characteristics between groups. Yet randomization does not automatically eliminate every source of uncertainty. Imbalances can arise by chance, and some factors may change during the study. If nutritional exposure is incompletely measured, poorly categorized, or omitted from the published report, readers may be unable to judge whether it was adequately controlled.
Nutrition is unusually difficult to capture in a clinical trial. A person’s dietary intake can vary from day to day, and food-frequency questionnaires depend on memory and estimates of portion size. Standard food records may alter behavior because participants know they are being monitored. Blood or urine biomarkers can provide more objective evidence for certain nutrients, but they are expensive, may reflect only recent intake, and do not measure the full complexity of a diet. Socioeconomic conditions add another layer: food availability, household income, education, cultural practices, supplement use, and access to healthcare can all shape nutritional status and may also influence participation, adherence, and outcomes.
The authors’ focus on reporting limitations is therefore central to the scientific debate. A study may have collected relevant information but failed to present it clearly, leaving outside readers unable to reconstruct the analysis. Important details can include how dietary exposure was defined, when it was measured, whether supplements were allowed, how missing data were handled, and whether nutritional variables were included in the statistical model. Researchers must also distinguish between factors measured before treatment, factors changed by treatment, and variables that lie on the causal pathway between an intervention and an outcome. Treating all of them as ordinary “confounders” can introduce a different form of bias.
Technically, confounding can be represented with a causal diagram in which nutrition influences both the intervention received and the outcome. If that common cause is not controlled, the estimated treatment effect may combine several pathways. Statistical adjustment can help, but only when the relevant variables have been measured with reasonable accuracy and the model reflects the underlying biology. Adjusting for a noisy or incomplete nutritional measure may leave residual confounding, while adjusting for a variable affected by the intervention can create collider or mediation bias. These distinctions matter because a small change in analytical strategy can substantially alter the interpretation of a clinical result.
The issue is not merely academic. Perinatal trials often inform recommendations that affect pregnant people, newborns, clinicians, and public-health systems. If a reported benefit or harm is partly attributable to differences in nutrition, applying the finding to another population may produce misleading expectations. A population with different rates of food insecurity, anemia, obesity, micronutrient deficiency, or supplement use may not experience the same outcome. External validity—the degree to which results travel beyond the original study group—depends on understanding these background conditions, not simply on knowing the intervention and the headline result.
King and Malhotra’s critique also reinforces the importance of transparent trial reporting. Readers need enough information to evaluate participant characteristics, protocol deviations, adherence, missing observations, and prespecified versus post hoc analyses. Nutrition-related reporting should ideally include the tools used to assess diet, the timing of measurements, relevant biomarkers, supplement exposure, and the methods used to classify nutritional risk. When these details are absent, uncertainty is not proof that the trial is wrong, but it does limit the strength of the conclusions that can responsibly be drawn from it.
The debate surrounding the JASMINE Trial illustrates a broader shift in science: credibility increasingly depends not only on obtaining a statistically significant result, but also on showing how that result survives alternative explanations. A finding that remains stable after careful nutritional adjustment is more persuasive than one that has not been tested against plausible dietary influences. Future perinatal studies may benefit from integrating nutritional assessment into trial design from the beginning, publishing complete baseline data, using validated biomarkers where possible, and making analytic code and protocols available for independent review.
The paper does not make the larger question of clinical evidence simpler, but it makes the stakes clearer. Trials involving pregnancy and newborn health operate within complex biological and social environments, and nutrition is one of the most powerful forces shaping those environments. By calling attention to possible confounding and incomplete reporting, the authors invite researchers to examine not only what a trial found, but also what it may have been unable to distinguish. For clinicians and readers, the message is equally important: a promising result deserves attention, while a transparent result—one that openly accounts for uncertainty—deserves trust.
Subject of Research: Nutritional confounding and reporting limitations in the JASMINE Trial
Article Title: Nutritional confounding and reporting limitations in the JASMINE Trial
Article References: King, B.C., Malhotra, A. “Nutritional confounding and reporting limitations in the JASMINE Trial.” J Perinatol (2026). https://doi.org/10.1038/s41372-026-02863-y
Image Credits: AI Generated
DOI: https://doi.org/10.1038/s41372-026-02863-y
Keywords: JASMINE Trial, nutritional confounding, perinatal research, clinical trials, maternal nutrition, trial reporting, epidemiology, causal inference, evidence quality
