Table 1. Characteristics of the cohort studies included in the systematic review and meta-analysis (n = 12).
| Reference and type of study | Maternal age | Maternal BMI | Maternal diabetes diagnosis | Diabetes control | Sample size † (n) | Child age(years) | SES | Parental education | Other confounders | Cognitive test and primary cognitive outcome* | Findings | Risk of bias | Quality score |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Fraser et al. 2012 Cohort prospective, ALSPAC (UK) | ND | GDM women were overweight | Questionnaire | ND | PGDM n = 20, GDM n = 23, ctrl n = 5079 | 8 | Yes | Yes | Sex and maternal age at birth, pre-pregnancy BMI, maternal smoking, parity, mode of delivery, gestational age, birth weight standardized for gestational age and duration of breastfeeding | WISC-III Total IQ: p = 0.09, Ctrl: 105.3 (16.3), PGDM: 103.2 (17.7), GDM: 98.7 (19.9). Adjusted total IQ: Mean differences (95% CI): Ctrl: 0, PGDM: -0.54 (-9.61, 8.52), GDM: -5.93 (-14.24, 2.38) | Support: GDM and PGDM were consistently associated with lower offspring cognition | L | 8 |
| Nomura et al. 2012 Cohort retrospective (Queens College, Flushing, New York, USA) | ND | ND | Mother’s retrospective report | ND | GDM n = 12, ctrl n = 97 | 4 | No | No | Age of mother, mother’s alcohol use and smoking during pregnancy, age, sex, race/ethnicity, and birth weight of the child, and maternal ADHD symptoms, paternal ADHD symptoms, and risk-group status. | WPPSI-III (full-scale) Ctrl: 113.6 (3.5), GDM: 109.2 (1.4), P˂0.001 | Support: Children of GDM mothers especially those raised in lower SES showed signs of suboptimal neurocognitive and behavioral development | L | 7 |
| Hod et al. 1999 Cohort prospective (Israel) | ND | ND | Management and control of maternal diabetes at the Rabin Medical Center | Diet+ insulin | T1DM n = 21, T2DM n = 10, ctrl n = 41 | 1 | No | No | No significant differences in maternal age, gestational age at delivery, incidence of premature delivery, birth weight, or neonatal complications | BSID-II MDI: p˂0.05, Ctrl: 98.15 (12.05), PGDM: 91.04 (9.01). PDI: p˂0.05, Ctrl: 95.54 (18.14), PGDM: 85.15 (14.53). By subgroup: MDI: N.S., T1DM: 90.4 (8.4), T2DM: 92.1 (10.3). PDI: p˂0.01, T1DM: 89.3 (12.8), T2DM: 78.1 (15.2) | Support: Infants of PGDM women scored lower on mental and psychomotor measures | M | 5 |
| Yamashita et al. 1996 Cohort prospective (Kurume, Japan) | DM: 30.6 years; Ctrl: 29.6 years | ND | OGTT | Diet n = 7 Diet+insulin n = 26 | T2DM n = 24, T1DM n = 6, GDM n = 3, ctrl n = 34 | 3–4 | No | No | No significant differences in birth weight, in duration of pregnancy, maternal age and age at time of IQ testing, but there were in duration of pregnancy and maternal toxoanemia | Tanaka-Binet Intelligence scale Ctrl: 113.4 (15.3), DM: 98.4 (17.4), p˂0.0001 | Support: The offspring of DM mothers showed a poorer intellectual development than ctrls | H | 6 |
| DeBoer et al. 2005 Cohort prospective (Minnesota, USA) | ND | ND | OGTT | Diet with or without insulin | DM n = 13, ctrl n = 16 | Up to 1 | No | No | Gestational age | BSID-II MDI: p˂0.05, Ctrl: 103 (10), DM: 95 (8); F (1,27) = 5.50. PDI: p = 0.06, Ctrl: 102 (13), DM: 89 (21); F (1,26) = 3.93 | Mixed: Significant differences were found on the MDI scale, but not on the PDI | H | 7 |
| Nelson et al. 2003 Cohort prospective (Minnesota, USA). | ND | ND | OGTT | Diet with or without insulin | DM n = 52, ctrl n = 75. | 1 | No | Yes | No significant differences in gestational age, and maternal age, but there were in birth weight | BSID-II MDI: p˂0.03, Ctrl: 104 (8), DM: 100 (9). PDI: N.S., Ctrl: 101 (13), DM: 98 (17) | Mixed: Children of DM mothers showed significantly lower MDI scores, but no differences on the PDI | L | 7 |
| Ornoy et al. 1998§ Cohort retrospective (Israel) | ND | ND | Laboratory examinations | Low sugar diet and insulin | PGDM n = 57, ctrl n = 57 | 5–12 (mean 8) | No | Yes | Children were matched by age and school placement, by gestational age and birth order | WISC-R Total IQ: p = 0.6, Mean (SE):Ctrl: 118.5 (1.3), PGDM: 117.7 (1.7) | Mixed: Pregnancy diabetes adversely affected some fine neurological functions (motor), but not their IQ- cognitive scores. No correlation with the degree of glycemic control | L | 7 |
| Sells et al. 1994 Cohort 3-year follow-up (University of Washington) in the collaborative DIEP (USA) | T1DM: 26.6 years Ctrl: 30.5 years | ND | Described in detail elsewhere35 | Better in “early entry” mothers. “Late entry” mothers had significantly higher mean glycosylated hemoglobin during the 1st and 2nd trimesters than “early entry” | For MDI: T1DM n = 93, ctrl n = 83. For PDI: T1DM n = 93, ctrl n = 83. For IQ: T1DM n = 62, ctrl n = 65 | 1, 2 and 3 | No | Yes | Not adjusted | BSID-I MDI: 1 year: N.S., Ctrl: 117 (12.5); T1DM “early entry”: 113 (15.3), T1DM “late entry”: 112 (13.5). 2 years: N.S., Ctrl: 118 (18.4), T1DM “early entry”: 118 (19.4), T1DM “late entry”: 112 (16.8). PDI: 1 year: N.S., Ctrl: 103 (15.8), T1DM “early entry”: 104 (15.5), T1DM “late entry”: 102 (17.2). 2 years: N.S., Ctrl: 110 (18.2), T1DM “early entry”: 108 (17.5), T1DM “late entry”: 112 (21.5). Stanford-Binet Intelligence Scale (4th Edition) IQ:3 years: N.S., Ctrl:110 (9.6), T1DM “early entry”: 109 (7.9), T1DM “late entry”: 103 (11) | Mixed: T1DM mothers who maintained good glycemic control during pregnancy can expect to have infants with a normal neurodevelopment as compared to those whose diabetes is less well controlled Non-significant differences were found | L | 9 |
| Rizzo et al. 1991 Cohort prospective (Prentice Women's Hospital, Chicago, IL, USA) | ND | ND | OGTT | The “high risk” group received insulin | For MDI: PGDM n = 75, GDM n = 82, ctrl n = 29. For IQ: PGDM n = 80, GDM n = 79, ctrl n = 27 | 3–5 | Yes | No | Race or ethnicity origin | BSID-I MDI: 2 years: N.S., Ctrl: 89 (13), PGDM: 89 (18), GDM: 90 (14). Stanford-Binet Intelligence Scale IQ: Average score at 3,4 and 5 years: N.S, Ctrl: 92 (10), PGDM: 89 (14), GDM: 93 (11) | Mixed: Inverse correlations between MDI scores and plasma β-hydroxybutyrate. Inverse correlations of Stanford-Binet scores with β-hydroxybutyrate and free FA levels of the mother. No differences in cognitive function. | L | 8 |
| Townsend et al. 2005 Cohort retrospective (Minnesota, USA) | ND | ND | ND | ND | DM n = 15, ctrl n = 15 | 4 | No | No | Gestational age | WPPSI-R IQ: N.S, Ctrl:121 (21), DM:118 (15). | No support: “Low-risk” children born to DM mothers (good glycemic control) demonstrated typical long-term neurocognitive development. | M | 5 |
| Nelson et al.2000 Cohort prospective (University of Minnesota, USA) | ND | ND | OGTT | Diet with or without insulin | DM n = 32, ctrl n = 25 | 1 | No | No | Not adjusted | BSID-II MDI: N.S., Ctrl: 105 (8.7), DM: 103 (7.5). PDI: N.S., Ctrl:102 (11.9), DM: 102 (9.5) | No support: The BSID exam failed to distinguish the effect between both children groups. | H | 6 |
| De Regnier et al. 2000 Cohort prospective (New York, USA) | ND | ND | Medical history | Only n = 12 treated with insulin | DM n = 22, ctrl n = 27 | 1 | No | Yes | No significant differences in gestational age, birth weight and birth head circumference | BSID-II MDI: p = 0.31, Ctrl: 104.7 (1.6), DM: 102.6 (1.4). PDI: p = 0.31, Ctrl: 103.4 (1.9), DM: 100.6 (3.4) | No support: No difference in MDI or PDI scores nor in the language examination | H | 4 |
Studies list is ordered by date and by type of outcome (Most recent studies and “support” are showed first).
ALSPAC, Avon Longitudinal Study of Parents and Children; ND, Not defined; GDM, Gestational Diabetes Mellitus; PGDM, Pre-gestational Diabetes Mellitus; Ctrl, Control; BMI, Body Mass Index; WISC-III/R, Wechsler Intelligence Scale for Children-3rd Edition, R, revised; IQ, Intelligence Quotient; ADHD, Attention Deficit Hyperactivity Disorder; WPPSI-III/R, Wechsler Preschool and Primary Scale of Intelligence–3rd Edition, R, Revised; SES, Socio Economic Status; T1DM, Type 1 Diabetes Mellitus; T2DM, Type 2 Diabetes Mellitus; BSID-I/II, Bayley Scales of Infant Development- 1st/2nd Edition; MDI, Mental Development Index; PDI, Psychomotor Development Index; OGTT, Oral Glucose Tolerance Test; DIEP, Diabetes in Early Pregnancy; FA, Fatty Acids.
Risk of bias classification (GRADE): L, Low; M, Medium; H, High.
Quality score (Newcastle-Ottawa): from 0 (lowest) to 9 (highest).
† Number of children included in the studies.
*All values refer to mean (Standard Deviation), otherwise it is stated.
§ Three references are related to the same subjects, so only the oldest study was included.