Background:

Gestational diabetes mellitus (GDM) is largely asymptomatic; screening for GDM during pregnancy could identify women who could benefit from treatments to reduce adverse consequences of GDM.

Purpose:

To systematically update the 2012 evidence review used to inform United States Preventive Services Task Force (USPSTF) recommendations on benefits and harms of screening for GDM.

Data Sources:

MEDLINE, Embase, and CINAHL (2010 to May 2020), ClinicalTrials.gov, reference lists of primary studies and systematic reviews; with surveillance through June 2021. All previously reviewed studies were re-assessed for eligibility.

Study Selection:

Two investigators independently reviewed abstracts and full-text articles against a set of a priori inclusion criteria. Disagreements were resolved through discussion. We included English-language controlled trials for effectiveness of screening and treatment; observational studies on screening effectiveness and harms, diagnostic accuracy of screening tests and association between GDM and outcomes.

Data Extraction:

One investigator abstracted data and a second investigator checked data abstraction for completeness and accuracy. Two investigators independently rated quality of the included studies using design-specific criteria.

Data Synthesis (Results):

Twenty trials (different screening strategies [N=27,196]; treatment benefits and harms [N=4,235]) and 87 observational studies (screening benefits [N=4,336] and harms [N=166,082]; diagnostic accuracy [N=91,260]; outcome associations [N=105,492]) were included.

Four observational studies (N=4,336) of screening versus no screening suggested that screening may be associated with reduced risk of some pregnancy and neonatal outcomes, but findings for each outcome were based on single studies with methodological limitations. Undergoing screening or receiving a false positive result may not be associated with anxiety; GDM may be associated with unnecessary cesarean delivery.

In five trials (N=25,772), 1-step International Association of Diabetes and Pregnancy Study Group (IADPSG) versus 2-step Carpenter-Coustan (CC) screening was associated with increased likelihood of gestational diabetes (11.5% vs 4.9%) but no improved health outcomes. One trial (n=922) suggested that early versus usual timing of 2-step CC screening may not improve outcomes in obese women.

Forty-five studies (N=91,260) evaluated diagnostic accuracy. At 24 to 28 weeks’ gestation, the oral glucose challenge test using 135 or 140 mg/dL thresholds, against CC and National Diabetes Data Group (NDDG) criteria, and a fasting plasma glucose of 85 mg/dL or 90 mg/dL against CC GDM, had reasonable accuracy (sensitivities ≥81% and specificities ≥73%). Screening with the glucose challenge test against IADPSG criteria had low sensitivity.

Being diagnosed with GDM based on more (e.g., 1-step IADPSG) versus less (e.g., 2-step CC) inclusive criteria, but not treated, associated with increased risk of preeclampsia, cesarean deliveries, preterm deliveries, macrosomia, LGA, neonatal hypoglycemia, and hyperbilirubinemia. No association was found for NICU admissions.

From nine trials (N=3,982), treatment for mild GDM at or after 24 weeks’ gestation associated with decreased risk of primary cesarean deliveries (RR, 0.70 [95% CI, 0.54 to 0.91]; ARD, 5.3%), preterm deliveries (RR, 0.75 [95% CI, 0.56 to 1.01]; ARD 2.3%), preeclampsia (RR, 0.60 [95% CI, 0.35 to 1.01]; ARD, 1%; after excluding one outlier trial), shoulder dystocia (RR, 0.42 [95% CI, 0.23 to 0.77]; ARD, 1.3%), macrosomia by 8.9% (RR, 0.53 [95% CI, 0.41 to 0.68]; ARD, 8.9%), LGA (RR, 0.56 [95% CI, 0.47 to 0.66]; ARD, 8.4%), birth injuries (e.g., fracture or nerve palsies) (OR, 0.33 [95% CI, 0.11 to 0.99]; ARD, 0.2%) and NICU admissions (RR, 0.73 [95% CI, 0.53 to 0.99; ARD, 2.0%). There was no association with risk of neonatal hypoglycemia or total cesarean deliveries, or for the potential harm of small-for-gestational age. There was limited evidence on long-term health outcomes and for early versus usual timing of screening.

Limitations:

Evidence on screening versus no screening was observational; very limited evidence on early treatment; restricted to English language studies; unable to formally assess for publication bias; limited evidence for some comparisons and outcomes, and most subgroups; heterogeneity present in some analyses.

Conclusions:

While direct evidence on outcomes of screening remains very limited, screening tests can identify women with gestational diabetes at or after 24 weeks’ gestation and treatment is associated with improvement in various maternal and neonatal outcomes without serious harms. One- versus 2-step screening was not associated with improved health outcomes. Research should clarify optimal timing of screening and if risk-based tools combined with glycemic measures are better predictors of GDM and treatment-responsive outcomes.