This was a multicenter cross-sectional study. The data utilized in this study, including information on breastfeeding practices, birth characteristics, physical and nutritional indicators, etc., were collected through the survey of “Infant Body Composition Study” [17]. These research findings are consistent with the broader objectives of the “Infant Body Composition Study”.

All participants were recruited from well-child health visits at four medical institutions between March 2021 and November 2022: Jinan, Liaocheng, Tengzhou, and Dongying in Shandong Province, China. Healthy singleton term infants who were born between 37+ 0 and 41+ 6 weeks of GA, and whose age at the time of the visit ranged from 0 to 12 months, were recruited. Exclusion criteria included congenital malformations, syndromes, and short- or long-term diseases. In our local context, infant well-child visits are generally scheduled at 42 days, 3 months, 6 months, 8 or 9 months, and 12 months post-birth, leading to a concentration of participants within these age intervals.

Ethical approval was granted from the Medical Ethics Committee of The First Affiliated Hospital of Shandong First Medical University (YXLL-KY-2022 (017)) and the Medical Ethics Committee of Shandong Provincial Maternal and Child Health Care Hospital Affiliated with Qingdao University (No. 2023-025). Written informed consent was obtained from all participants of their parents.

Mothers of the participants received nutritional and feeding guidance based on current World Health Organization (WHO) guidelines from pediatricians at routine well-child visits [18]. The recommendations included: (1) exclusively breastfeeding for the first 6 months of life and (2) the introduction of complementary foods at 6 months, along with continued breastfeeding up to 2 years of age or beyond. This study obtained information on the implementation of feeding patterns before 6 months of age and the timing of complementary foods introduction through a questionnaire survey during the visit.

Information about infants at birth included ethnicity (“Han”, others), gender (“female”, “male”), birth weight (kg), and GA (week). This information was obtained from the medical records of child health care centers or birth certificates of infants. In this study, we reclassified GA into two categories for analysis: 0. “full-term”, defined as GA 39+ 0–41+ 6 weeks, and 1. “early-term”, defined as GA 37+ 0–38+ 6 weeks [11].

Data on nutritional practices was obtained from a parental questionaire at visit, which included breastfeeding practices and complementary foods introduction. There were five categories for feeding patterns in the questionnaire: “1. exclusive breastfeeding”, “2. mainly breastfeeding”, “3. breastfeeding accounts for half”, “4. mainly formula-feeding”, and “5. exclusive formula-feeding”, which were reclassified as “1. exclusive breastfeeding”, “2. mixed feeding” (including “mainly breastfeeding”, “half breastfeeding”, and “mainly formula-feeding” in the questionnaire), and “3. exclusive formula-feeding”) for analysis. For infants under 6 months old, parents were requested to document the current breastfeeding routines (point-in-time data), i.e., the breastfeeding practices in previous 24 h before the visit; while for infants 6 months and older, retrospective data about the breastfeeding practices when they got 6 months age was documented. Complementary foods introduction was documented as “no” or “yes” for infants under 6 months old and as “introduced at 4–5 months” or “introduced at ≥ 6 months” for infants 6 months and older.

Maternal information during gestation included gestational diabetes mellitus (GDM, “no”, “yes”) and gestational hypertension (“no”, “yes”), which were also documented through the parental questionnaire at the visit.

All participants have complete birth information documented in their medical records or birth certificates. Parental questionnaires were administered on-site by trained staff during recruitment and thoroughly checked for relevant variables to guarantee the completeness of data for all participants.

The measurements of length and weight for all infants were conducted according to standard procedures, as detailed in our previous study [17]. The Z-scores of length, weight and weight for length were calculated in accordance with WHO growth standards and recorded as LAZ, WAZ and WLZ, respectively. Three types of malnutrition were documented: moderate and severe stunting (MSS) [19], defined as LAZ <-2; moderate and severe acute malnutrition (MSAM) [20], defined as WLZ <-2; and overweight and obesity, defined as WLZ > 2.

A total of 7299 singleton term infants with complete data were enrolled in the study. To ensure data quality and accuracy, rigorous data cleaning was conducted: any LAZ, WAZ, or WLZ values outside the range of -4 to 4 were considered potentially suspicious outliers. Participants displaying one or more of these outliers were excluded from the final analysis, resulting in the removal of 29 participants. Among the remaining 7270 infants, 4066 were aged 0–5 months and 3204 were aged 6–12 months (Fig. 1).

Flowchart of data sampling

Categorical variables are presented as numbers (n) and percentages (%), while continuous variables are presented as the means (standard deviations) or means (95% CIs).

In both age groups, covariates adjusted for included gender, ethnicity, birth weight, GDM, gestational hypertension, parental heights, and age at the visit. Considering that most infants under 6 months did not reach the recommended time for introducing complementary foods, the covariate complementary foods introduction was only included in the analysis of infants aged ≥ 6 months. Linear regression models were utilized to investigate the correlation of GA at birth (full-term and early-term) and feeding practices (exclusive breastfeeding, mixed feeding and exlusive formula-feeding) with LAZ and WLZ, with β values indicating these relationships. Logistic regression models were employed to explore the associations of GA at birth and feeding practices with the risks of MSS, MSAM, and overweight/obesity, with OR values representing the likelihood of these outcomes. A significance level of P < 0.05 was considered to indicate statistical significance.

Given the potential influence of macrosomic infants (birth weight > 4.0 kg) and low birth weight infants (birth weight < 2.5 kg) on observed outcomes, we performed a sensitivity analysis in children of normal birth weight (2.5–4.0 kg) to assesse the robustness of study findings. The covariates adjusted for and the statistical methods utilized in the sensitivity analysis were entirely consistent with those previously described, with the exception of incorporating children with normal birth weight.

All the statistical analyses were conducted using IBM SPSS Statistics 21 software (Chicago, IL, USA) and R software (version 4.0.3; creator: John Chambers and colleagues; location: Jersey City, NJ, USA).