Bias	Author's judgment	Support for judgment
Risk of bias arising from the randomization process	Some concerns	Centers were randomized but no explanation of the randomization process, no reporting of allocation conceallment. Baseline characteristics were balanced except for maternal weight, suggesting appropriate randomization.
Risk of bias due to deviations from the intended interventions (effect of assignment to intervention)	High risk	Because of the nature of the intervention, participants and personnel were made aware of their assignment during the trial, and no reporting of analysis to determine the effect of intervention assignment.
Risk of bias due to deviations from the intended interventions (effect of adhering to intervention)	High risk	There were difficulties experienced by the nurses and the program's coordinator in carrying out home visits during the winter months, when many of the county's mountainous villages are cut off by snow for days or weeks at a time, and no reporting of special analysis.
Missing outcome data	High risk	Incomplete sets of assessments were imbalanced across study arms, analysis therefore only focused on paired intergroup comparisons in dietary Intake at different points in time.
Risk of bias in measurement of the outcome	High risk	Nurses who delivered the intervention measured the outcomes. However, to minimize the possible bias associated with the nurses simultaneously delivering nutrition counseling and recording the effects in the intervention group (in terms of dietary histories and weight gain), the program's nurse coordinator periodically accompanied each nurse on their home visits to observe the data gathering and to sort out any associated problems.
Risk of bias in selection of the reported result	Low risk	data that produced this result was analyzed in accordance with a pre‐specified analysis plan that was finalized before unblinded outcome data were available for analysis.