Objective
We sought to examine the recurrence risk of gestational diabetes mellitus (GDM) in a subsequent pregnancy and determine whether recurrence risk is modified by race/ethnicity.
Study Design
We used the Kaiser Permanente Southern California longitudinally linked records (1991-2008) to study women with first 2 (n = 65,132) and first 3 (n = 13,096) singleton pregnancies. Adjusted odds ratios (ORs) were used to estimate the magnitude of recurrence.
Results
Risks of GDM in the second pregnancy among women with and without previous GDM were 41.3% and 4.2%, respectively (OR, 13.2; 95% confidence interval, 12.0–14.6). The recurrence risk of GDM in the third pregnancy was stronger when women had GDM in both prior pregnancies (OR, 25.9; 95% confidence interval, 17.4–38.4). Hispanics and Asian/Pacific Islanders have higher risks of recurrence.
Conclusion
A pregnancy complicated by GDM is at increased risk for subsequent GDM. The magnitude of risk increases with the number of prior episodes of GDM. These recurrence risks also showed heterogeneity by race-ethnicity.
Gestational diabetes mellitus (GDM) is defined as carbohydrate intolerance of variable severity, with onset or first recognition during pregnancy. GDM complicates 3-14% of pregnancies in the United States, and it is responsible for an important proportion of fetal and maternal morbidity including pregnancy-induced hypertension, preeclampsia, urinary tract infections, cesarean delivery, fetal macrosomia, birth trauma, and a variety of relatively minor, transient neonatal maladaptations to extrauterine life (eg, hypoglycemia), and developing future diabetes. In the United States, there is an increasing trend in GDM. A recent study reported that the prevalence of GDM increased gradually between 1989-1990 and 2003-2004 by 122%, from 1.9-4.2%, respectively.
Early detection and initiation of treatment is important because unrecognized or untreated GDM is likely to lead to adverse maternal and fetal outcomes. Although maternal deaths are low, fetal and neonatal mortality remains much higher than in the general population. Potential risk factors that have been associated with GDM include advanced maternal age, being from minority race/ethnicity groups, being overweight or obese during pregnancy, being multigravida, and having a previous diagnosis of GDM.
Limited numbers of studies examined the recurrence of GDM in subsequent pregnancies. In their study, Gaudier et al noted a 52% increased risk of recurrence of GDM in subsequent pregnancies. Similarly, Moses found a 35% recurrence rate of GDM after following up 100 subsequent pregnancies. Despite the increased tendency of GDM to recur, little is known with regard to the extent to which the number of prior GDM can influence risk of GDM in future pregnancies. Furthermore, whether the risk is modified by maternal race/ethnicity is unknown.
The purposes of this study were to examine risks of recurrence: (1) of GDM in the first 2 and first 3 subsequent pregnancies; and (2) in relation to maternal race/ethnicity.
Materials and Methods
Data source
The study utilized population-based data from Kaiser Permanente Southern California (KPSC) from 1991 through 2008 (n = 540,956). We matched the perinatal service system, hospital inpatient, and physician encounter records through a unique identifier for each individual. The matched records were further linked longitudinally to subsequent pregnancies. Information extracted from the perinatal service system records includes maternal sociodemographic and behavioral characteristics, fetal and neonatal outcomes, and labor and delivery complications. Information extracted from the hospital inpatient and physician encounter records include behavioral factors, and medical and obstetric complications and procedures.
For this retrospective cohort study design, we formed 2 cohorts: a cohort of linked first 2 singleton births and a second cohort of linked first 3 consecutive singleton deliveries in all KPSC facilities.
Definition of variables
Data on maternal characteristics were based on the study cohort that consisted of the first 2 births. Self-reported maternal race was grouped as non-Hispanic white (white), non-Hispanic black (African American), Hispanic, and non-Hispanic Asian/Pacific Islander. Other independent variables considered as either potential determinants or confounders for recurrence of GDM included: maternal age (<20, 20-29, 30-34, ≥35 years), maternal education (<12, 12, and ≥13 years of completed schooling), initiation of prenatal care (early or first trimester and none or late initiation), maternal smoking and alcohol consumption during pregnancy (yes/no), and interval between a birth and a subsequent pregnancy (<6, 6-11, 12-17, 18-23, 24-29, 30-35, 36-41, 42-47, and ≥48 months). The International Classification of Diseases, Ninth Revision, Clinical Modification ( ICD-9-CM ) code 648.8 was used to identify physician diagnosis of GDM (yes/no). We validated the accuracy of hospital-based ICD-9-CM diagnosis codes for GDM by using a random sample of 1000 patients. Laboratory records were reviewed and the results were compared with the diagnosis codes collected electronically. The estimated positive predictive value and negative predictive value for GDM were 90% and 85%, respectively. Based on this preliminary finding, the hospital-based ICD-9-CM diagnosis codes seemed reasonable quality in the ascertainment of GDM cases in this study.
Data exclusions
From a total of 540,956 singleton births in the KPSC hospitals from 1991 through 2008, we excluded the following categories: births at <20 weeks of gestation and at <500 g (n = 9357), women with only 1 pregnancy during the study period, and those who had ≥1 pregnancy <1991 (n = 443,178). We also excluded women of “other” race/ethnicity group due to small number (approximately 3% of the cohort), as well as observations with missing data on race/ethnicity (n = 1756). The justification for limiting the analysis to singleton pregnancies is that multiple pregnancies are inherently at different risks of adverse pregnancy outcomes. We excluded births with gestational age <20 weeks and birth weights of <500 g to decrease errors in gestational age estimation and to minimize medical center differences in reporting live births that were at borderline of viability. Furthermore, women who have a diagnosis of type 2 diabetes prior to their first pregnancy (n = 863), and those women who developed type 2 diabetes between their first 2 (n = 2850) and between their second and third pregnancies (n = 617) were excluded from the study. To accomplish this, we used data extracted from postpartum medical encounter records.
Statistical analysis
We performed a retrospective cohort analysis to assess recurrence risks of GDM among women with singleton pregnancies. Statistical analysis was performed in 4 steps. First, we examined the distribution of maternal demographic and behavioral characteristics by GDM status using first and second births. Second, a conditional logistic regression model was fitted to examine the recurrence of GDM in subsequent pregnancies (first 2 and first 3 pregnancies) after controlling for potential confounding variables (maternal age, maternal race, maternal education, initiation of prenatal care, interpregnancy interval [IPI], smoking during pregnancy, and year of delivery). We assessed the interaction between GDM in a pregnancy and the intervals between the first 2 pregnancies. Third, we repeated the logistic regression analysis after stratifying the data by race/ethnicity in an attempt to clarify risk of recurrences in the various categories. Odds ratios (ORs) and their 95% confidence intervals (CIs) were used to estimate risks of recurrence. Furthermore, first 2 as well as second and third birth IPI-specific risks of recurrence were estimated.
All statistical analyses were performed using SAS (version 9.1; SAS Institute, Cary, NC). The study was approved by the KPSC Institutional Review Board.
Results
There were 65,132 pregnancies with first 2 and 13,096 with first 3 consecutive singleton births in the KPSC hospitals from 1991 through 2008, resulting in an incidence rate of 56 and 69 GDM pregnancies per 1000 births, respectively.
As shown in Table 1 , GDM is more likely to occur in women giving birth at age ≥30 years, with ≥13 years schooling, Hispanic and Asian/Pacific Islander race/ethnicity group, and long IPIs.
