Korea's National Institute of Health announcement has a headline that can be frightening: diabetes risk in children of mothers with gestational diabetes may be more than four times higher. The number belongs to a particular offspring type 2 diabetes comparison, not a mother's personal odds and not the probability that a child will develop diabetes. For a family, the more useful step is to prepare two different conversations.

AI-generated image: Blood-glucose meter without a visible result and testing supplies

Conversation one: the mother after delivery

The Korean agency's background sheet, page 6, advises blood glucose testing 6–12 weeks after childbirth for a person who had gestational diabetes, followed by regular testing. Postpartum glucose often returns to normal, but that does not erase the mother's later type 2 diabetes concern. If the window has passed, retrieve any test result and ask your obstetric or internal-medicine clinician what test and follow-up schedule now fit your situation. The fourfold statistic from the birth cohort does not estimate a mother's future diabetes risk.

An American College of Obstetricians and Gynecologists (ACOG) patient FAQ separately says 4–12 weeks postpartum and, if the result is normal, diabetes testing every 1–3 years. That is US patient guidance, not a substitute for a Korean clinician's individual plan. Do not splice the two organizations' intervals together and call the result a single national rule.

AI-generated image: Vegetables, grains and a blank notebook on a kitchen counter

Conversation two: the child's medical history

The original birth-cohort paper studied 3,491,680 births involving 2,520,815 mothers in Korea between 2009 and 2018, with records followed through the end of 2022 (median follow-up 9.5 years, maximum 14 years). Of the children, 424,185 were born to mothers with gestational diabetes; 30,003 were in the group with a maternal insulin-treatment record during pregnancy. Those cohort shares are not the prevalence of diagnosed gestational diabetes among all pregnant people in Korea. Table 2 compares three groups of children: 3,067,495 whose mothers did not have gestational diabetes (reference), 394,182 whose mothers had it without a recorded insulin treatment, and 30,003 whose mothers had it with an insulin-treatment record. These are the exposure-group denominators for both offspring outcomes below, not the numbers of observed cases. In its adjusted model:

  • For offspring type 2 diabetes, the hazard ratio (HR) was 1.281 (95% CI 1.146–1.433) if the mother had gestational diabetes without a recorded insulin treatment during pregnancy; it was 4.329 (3.555–5.270) in the insulin-treated gestational-diabetes group. Observed case counts were 3,125 in the much larger no-GDM reference group, 352 without a maternal insulin record and 104 with one.
  • For offspring type 1 diabetes, the corresponding adjusted HRs were 0.857 (0.696–1.054) without insulin and 1.936 (1.228–3.052) with insulin. The first interval includes 1, so this study did not establish a clear type 1 association for the non-insulin group. Observed case counts were 1,166 in the reference group, 99 without a maternal insulin record and 19 with one. The groups differ in size and follow-up; the counts alone are not comparable personal risks.

An HR is a relative comparison of event rates over observed time after adjustment; 4.329 does not mean one child in four will develop diabetes. Type 1 and type 2 results must not be pooled. The paper counted claims-defined disease: an E10 diagnosis plus insulin prescription for type 1, or E11–E14 diagnosis plus antidiabetic medication for type 2. It did not measure every borderline blood test or predict lifetime outcomes.

Why an insulin record is not a verdict on insulin

This is a retrospective observational analysis, not a trial assigning insulin. The treatment group may differ in the severity of gestational diabetes and other metabolic or inherited characteristics. The authors adjusted for several recorded maternal and child factors but lacked detailed maternal obesity, family history and health-behavior data. They explicitly caution that causal inference is not possible and that claims coding can misclassify cases. Their result does not show that an insulin prescription causes a child's diabetes; do not stop or change prescribed treatment based on the comparison. ACOG's FAQ describes insulin as a recommended option for controlling blood glucose during pregnancy when medication is needed.

Tell the child's pediatrician about the gestational-diabetes history and whether insulin was used during pregnancy. Ask whether growth, weight and the child's own risk warrant blood glucose evaluation. Korea's agency particularly stresses individualized ongoing attention for children of mothers whose gestational diabetes required insulin. Neither the cohort nor the checked guidance establishes a universal test-at-age-X, every-year rule for all such children. A child's testing need and timing belong in a clinical conversation, separate from the mother's postpartum test.

About the Author

Life Mokcha's editors compared the Korean agency's original seven-page release and appendices with the original paper's full XML and Table 2, and checked the ACOG patient FAQ for separately attributed US guidance. This is an evidence-based reading aid, not personal medical advice or a firsthand care account.

References