TABLE 3

Adipokines at 3 years postpartum, by exclusive breastfeeding duration: predicted values* from linear regression models

nMonths of exclusive lactationP
None>0 to <11 to <33 to <6≥6
Mean (95% CI)Mean (95% CI)Mean (95% CI)Mean (95% CI)Mean (95% CI)
Leptin (ng/mL) (n)551218115130
 Unadjusted43812.8 (10.2–15.3)10.2 (8.9–11.6)8.8 (7.2–10.4)9.0 (7.4–10.6)9.3 (6.5–12.1)0.03
 Multivariable adjusted43811.4 (9.5–13.3)8.8 (7.5–10.1)7.8 (6.4–9.2)8.5 (7.1– 9.8)9.0 (6.9–11.1)0.006
Adiponectin (μg/mL) (n)561208115330
 Unadjusted44019.8 (16.2–23.4)21.5 (19.6–23.5)21.4 (19.1–23.7)20.7 (18.4–22.9)24.5 (20.6–28.5)0.26
 Multivariable adjusted44020.4 (16.9–24.0)23.6 (21.1–26.1)23.3 (20.7–25.9)21.9 (19.4–24.4)25.9 (21.9–29.9)0.05
Ghrelin (pg/mL) (n)1139225014
 Unadjusted136665.1 (422.9–907.4)721.5 (602.2–840.9)728.7 (585.1–872.2)794.6 (646.6–942.6)1,034 (811.9–1,257)0.05
 Multivariable adjusted136706.3 (484.8–927.8)754.4 (617.9–890.9)865.0 (723.1–1,007)810.0 (667.9–952.1)920.9 (710.5–1,131)0.2
PYY (pg/mL) (n)1139215014
 Unadjusted13561.0 (47.1–78.9)55.5 (48.8–63.0)56.9 (48.6–66.6)62.0 (52.9–72.6)53.5 (42.2–67.9)0.46
 Multivariable adjusted13557.0 (42.6–76.2)56.7 (47.4–67.9)56.4 (46.5–68.4)60.2 (50.0–72.6)51.2 (38.8–67.4)0.63
  • Data from 570 participants in Project Viva who presented for follow-up at 3 years postpartum without an intervening birth.

  • *For the multivariate-adjusted model, data presented are mean predicted values for a participant with a prepregnancy BMI of 25.1 kg/m2, the mean for the study population, modeled using a three-knot quadratic spline model for BMI. This participant continued to breastfeed for up to 3 months after introducing formula or solid foods and is a white woman aged 35–40 years, has two children, has no parental history of diabetes, had normal glucose tolerance, is a nonsmoker, and gained 15 kg during the index pregnancy.

  • †Partial F test P values for differences among categories.

  • ‡Results presented for PYY are geometric means. Because this outcome was not normally distributed, it was modeled on the log scale. Predicted values are exponentiated for presentation to improve interpretability.