Abstract
IMPORTANCE: Evidence on the association between breastfeeding duration and cognitive development remains inconsistent, partly due to confounding by socioeconomic status (SES). Whether this association exists across populations with distinct socioeconomic patterning of breastfeeding is unclear. OBJECTIVES: To investigate the association between breastfeeding duration and cognitive outcomes among youths in China. DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional analysis used data from the China Family Panel Studies for youths aged 10 to 15 years who underwent cognitive assessments in 2010, 2014, or 2018 and were born between 1995 and 2009. Data were analyzed from February 2025 to January 2026. EXPOSURES: Breastfeeding duration (≤6 months vs >6 months and continuous). MAIN OUTCOMES AND MEASURES: Cognitive performance was assessed using mathematics and word recognition tests and was standardized as age-adjusted z scores. Poor cognitive performance was defined as scoring at or below the 15th percentile. Survey-weighted generalized linear models were fitted with progressive adjustment for demographic, perinatal, and socioeconomic factors. Restricted cubic splines were used to explore potential nonlinear associations. RESULTS: Among 5436 youths included in the study (mean [SD] age, 12.03 [1.58] years; 52.5% male), 750 (13.8%) were breastfed for 6 months or less and 4686 (86.2%) for longer than 6 months. Longer breastfeeding duration was more common among youths from lower-income households (1229 [26.2%] vs 133 [17.7%]) and those with parents of lower educational attainment (maternal, 1464 [31.2%] vs 181 [24.1%]; paternal, 810 [17.3%] vs 99 [13.2%]). In unadjusted analyses, breastfeeding duration was not associated with mathematics (β = 0.03; 95% CI, -0.06 to 0.13) or word recognition (β = -0.02; 95% CI, -0.13 to 0.09) z scores. In models adjusted for SES, breastfeeding for more than 6 months was associated with higher mathematics (β = 0.14; 95% CI, 0.05-0.22) and word-recognition (β = 0.12; 95% CI, 0.02-0.21) z scores and lower odds of poor mathematics (odds ratio [OR], 0.65; 95% CI, 0.48-0.88) and word recognition (OR, 0.64; 95% CI, 0.47-0.87) performance. Restricted cubic spline analyses indicated that the associations may not be strictly linear. CONCLUSIONS AND RELEVANCE: In this cross-sectional study of Chinese youths, longer breastfeeding duration was not associated with cognitive performance in the unadjusted model but was associated with better cognitive performance after adjustment for SES. Longer breastfeeding was more common among families with low SES. Continued policy support, such as extended maternity leave and workplace lactation accommodations, may be warranted alongside further longitudinal research.