Evaluation of the prevalence of gestational diabetes mellitus in North Indians using the International Association of Diabetes and Pregnancy Study groups (IADPSG) criteria

Abstract

Objective:

Currently, there is controversy regarding the diagnosis of gestational diabetes mellitus (GDM) as per the newer International Association of Diabetes and Pregnancy Study Groups (IADPSG) criteria. We studied the prevalence and associations of GDM in North Indians, diagnosed by the IADPSG criteria.

Patients and Methods:

We conducted a cross-sectional study on 332 pregnant women, predominantly belonging to lower and middle socioeconomic strata. The women were screened for GDM between 24 weeks and 28 weeks of gestation by 75g oral glucose tolerance test (OGTT) and GDM diagnosed by the IADPSG criteria.

Results:

The prevalence of GDM was 41.9% [95% Confidence interval (CI) 36.6-47.2%]. Amongst the women diagnosed to have GDM, 91.4% had abnormal fasting plasma glucose (FPG), while 1-h and 2-h post-glucose (PG) levels were abnormal in 18.7% and 17.3% of women, respectively. No maternal factors were significantly associated with GDM. Birth weight of the neonates was similar in women with GDM as compared to those with normal glucose tolerance. In the entire group, fasting glucose levels were associated with the weight of the patient while 1-h PG levels were associated with weight, height, socioeconomic score, and parity.

Conclusions:

There is a very high prevalence rate of GDM using the IADPSG criteria in North Indian women of low and middle socioeconomic strata. Further studies are needed to assess the utility of applying these criteria in settings with limited resources.

Introduction

The prevalence of gestational diabetes mellitus (GDM) varies in different racial groups. Additionally, it varies widely according to the different criteria used to diagnose GDM. In 2010, the International Association of Diabetes and Pregnancy Study Groups (IADPSG) proposed new criteria for the diagnosis of GDM based on the findings from the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study.[1,2] The HAPO study established that fetal and maternal complications were strongly associated with maternal hyperglycemia.[2] A concern regarding the application of IADPSG criteria is the classification of less severe hyperglycemia as GDM, thereby leading to an increase in health care burden and cost. While the American Diabetes Association (ADA) now accepts either IADPSG or the older two-step approach for diagnosis of GDM, the World Health Organization (WHO) accepted the IADPSG criteria in 2013.[3] However, the National Institute of Health (NIH) panel in 2013 recommended the older two-step approach.[4] Urban South Asians have a high prevalence of type 2 diabetes mellitus (T2DM) and are at an increased risk of GDM.[5] The prevalence of GDM in Indians ranges 7-18%.[5,6] In a recent study conducted on South Asians settled outside India, the prevalence of GDM was 38% by the IADPSG criteria.[7] Studies in India, with different environmental influences, have provided conflicting results. In addition, only one of these studies has reported the effect of GDM on perinatal morbidity. In view of these lacunae in our knowledge, the ongoing controversy on the most relevant screening criteria for the diagnosis of GDM, and the paucity of studies in this field on Indian women, we analyzed the prevalence and risk factors associated with GDM using the IADPSG criteria.

Patients and Methods

We prospectively studied 332 pregnant women attending the antenatal clinics in three government-run urban hospitals (Queen Mary Hospital, Dr. Ram Manohar Lohia Institute of Medical Sciences, and Sanjay Gandhi Postgraduate Institute of Medical Sciences, all in Lucknow, Uttar Pradesh, India) from June 2012 to July 2013. Women with pregestational diabetes and those with any chronic illness were excluded from the study. The age (in years), height (in cm), weight (in kg), blood pressure, and hemoglobin of the women were measured on the day of oral glucose tolerance test (OGTT). The socioeconomic status of these women was classified as per Kuppuswamy's socioeconomic status scale (updated in 2013).[8] Gestational age of the women at delivery (n = 259) and birth weight of the neonates (n = 248) were noted. The birth weight of all the neonates was not available since many women had delivered in different hospitals and were unavailable for follow-ups. The age, socioeconomic status, gravida number, prevalence of GDM, and gestation age at delivery were comparable in women whose neonate's birth weight was available with those in whose cases it was not available.

The women were screened with a 2-h 75-g OGTT between 24 weeks and 28 weeks of gestation. They were advised to come for testing after >8 h overnight fast. Their blood samples were taken in fasting state and 1-h and 2-h after 75-g oral glucose load. The glucose samples were obtained in fluoride vacutainers. The plasma was immediately separated on site and transported to the central laboratory in ice for analysis. Glucose analysis was done within 6 h of sampling. The women were classified as GDM and nonGDM, based on the IADPSG criteria [fasting plasma glucose (FPG) ≥92 mg/dL, 1-h post-glucose (PG) value ≥180 mg/dL and 2-h PG value ≥153 mg/dL].[2,3] Plasma glucose was estimated by glucose-oxidase-peroxidase (GOD-POD) technique. Written consent was obtained from all the patients and the study was approved by the Ethics Committee of the institute.

Statistical methods

The continuous variables were expressed as mean and standard deviation (SD). The categorical variables were expressed as number and percentage. Student's t-test was used for the comparison of groups. Univariate and multivariate logistic regression analysis (by stepwise backward Logistic Regression) was used to analyze the association of variables such as maternal age, height, weight, systolic and diastolic blood pressure, gravida number, family history of (h/o) diabetes, acanthosis nigricans, and socioeconomic score with the diagnosis of GDM. For regression analysis, the continuous variables were categorized into units as follows: Age 5 years, weight 5 kg, and height 10 cm. In addition, multivariate linear regression analysis was performed to analyze the association of the continuous variables (FPG, 1-h PG, 2-h PG, and birth weight) with various different covariates. P value <0.05 (two-tailed) was considered significant. Statistical analyses were performed using SPSS16.0 (SPSS Inc., Chicago, IL, USA).

Results

Of all the pregnant women, 96% belonged to the middle and lower socioeconomic groups. The mean age of the subjects was 25.1 ± 3.9 years and height, weight, and body mass index (BMI) were 150.9 ± 5.8 cm, 54.7 ± 8.8 kg, and 24.0 ± 3.6 kg/m², respectively. The mean gestational age (in weeks) at the time of delivery was 38.5 ± 1.8 weeks and birth weight was 2749 ± 433 g. A h/o diabetes among first degree relatives was present in 41 (12.4%) women and preeclampsia was present in 11 (3.3%). Acanthosis nigricans (observed in 16 women, i.e., 4.8%) and hypertension (observed in six women, i.e., 2.5%) were uncommon. The mean FPG, 1-h PG, and 2 h PG were 89.0 ± 10.8 mg%, 136.8 ± 30.9 mg%, and 115.9 ± 25.9 mg%, respectively. None of the multiparous women reported previous h/o GDM.

GDM was diagnosed in 139 [41.9%, (36.6-47.2%, 95% CI)] women [Table 1]. Using the IADPSG criteria, 98 (70.5%) GDM women were found to have only abnormal FPG with normal PG values. Also, of all the GDM women diagnosed by the IADPSG criteria, 91.4% had abnormal FPG. In contrast, by the previous WHO criteria, the prevalence of GDM was 17.5% in these women.

Table 1.

Prevalence of GDM and abnormal glucose cutoffs by the IADPSG criteria

Criteria	IADPSG n (%)
Number of GDM	139 (41.9)
Abnormal FPG alone	98 (70.5)
Abnormal 1-h PG alone	4 (2.9)
Abnormal 2-h PG alone	6 (4.3)
Any two abnormal values	31 (9.3)
All three abnormal values	7 (5)

PG = Post glucose

Variables such as age, gravida number, socioeconomic status as well as blood pressure, hemoglobin, FPG, 1-h plasma glucose and 2-h plasma glucose (at 24-28 weeks of gestation) were not significantly different between the GDM and the nonGDM groups. The outcome parameters such as birth weight, gestational age at birth, and mode of delivery [after excluding subjects with previous lower segment cesarean section (LSCS) delivery] also did not differ between the two groups. On logistic regression analysis, no parameter (age, height, weight, systolic and diastolic blood pressure, gravida number, family h/o diabetes, acanthosis nigricans, and socioeconomic score) was associated with diagnosis of GDM.

On multivariate linear regression analysis, FPG was affected by weight of the patient (β = 0.19, P = 0.01) [Table 2a] while 1-h PG value was affected by height (β = –0.76, P = 0.02), weight of the patient (β = 0.48, P = 0.03), gravida number (β = 3.93, P = 0.02), and socioeconomic score (β = 0.79, P = 0.02) [Table 2b]. No association was noted with the 2-h PG value. The birth weight of the neonates was associated with the weight of the mother (β = 7.9, P = 0.01) and gestational age at delivery (β = 50.7, P = 0.001) [Table 3].

Table 2.

Results of multiple linear regression model to assess the factors affecting (A) FPG and (B) 1-h PG

A
Variable	β	P value
Age (years)	–0.04	0.78
Height (cm)	–0.18	0.11
Weight (kg)	0.19	0.01
Gravida number	0.15	0.81
Socioeconomic score	–0.01	0.97
Systolic BP (mmHg)	0.08	0.21
Diastolic BP (mmHg)	–0.08	0.45
B
Variable	β	P value
Age (years)	0.48	0.35
Height (cm)	–0.76	0.02
Weight (kg)	0.48	0.03
Gravida number	3.93	0.01
Socioeconomic score	0.79	0.02
Systolic BP (mmHg)	0.06	0.81
Diastolic BP (mmHg)	–0.16	0.49

*BP = Blood pressure

Table 3.

Results of multiple linear regression analysis (backward) with birth weight as the outcome variable

Variable	β	t	P value
Age (years)	–8.55	–1.18	0.24
Height (cm)	6.13	1.21	0.23
Weight (kg)	7.92	2.51	0.01
Gravid number	28.19	1.01	0.32
Socioeconomic score	6.09	0.98	0.33
Hemoglobin	–25.32	–0.79	0.43
FPG	–0.01	–0.002	0.99
1-h PG	0.45	0.45	0.66
2-h PG	1.39	1.31	0.19
Gestational age at birth (weeks)	50.72	3.29	0.001

Discussion

By applying the IADPSG criteria, a large proportion of healthy North Indian urban women from lower and middle-class families (42%) would be classified as having GDM. The frequency of GDM in our study population is higher than that of the HAPO study (17.8%). However, Indian women were not represented in the HAPO cohort. Even among the collaborating centers in the HAPO study, the prevalence rates of GDM differed widely, varying 9-25.5%.[9]

In a recent community-based study conducted in Norway, Jenum et al. studied the prevalence of GDM in the native and immigrant populations, including South Asian women. They reported 2.4 times increase in the prevalence of GDM among South Asians (2.8 times compared with the WHO criteria) applying the modified IADPSG criteria.[10] Though not exactly reported, GDM prevalence was >40%. As in our study, GDM was mainly related to an elevated FPG. Similarly, in a retrospective analysis of South Asian women in the United Arab Emirates, Agarwal et al. reported a high prevalence of GDM (38%) by the IADPSG criteria when compared with the ADA 2010 criteria (13%).[7] As noted in our results, where 91% of GDM by the IADPSG criteria had elevated FPG, the authors noted that FPG ≥92 mg/dL alone has a sensitivity of 91% for the diagnosis of GDM by the IADPSG criteria.

In contrast to these studies, studies from India have provided mixed results with GDM prevalence varying 7-45.4%.[11,12,13,14,15] Mohan et al., in a retrospective analysis of 1351 women from Chennai in South India, reported no change in the prevalence of GDM as per the WHO (1999) and the IADPSG criteria.[11] The same group studied 1,031 pregnant women attending antenatal clinics in urban and rural Tamil Nadu and reported the prevalence of GDM to be 10.3%.[12] In a study conducted at a community health center in Chennai, Seshiah et al. reported 14.6% prevalence of GDM as per the IADPSG criteria.[13] It was noted that measuring FPG only detects 9.3% of cases as having GDM. In a recent study from Pondicherry in South India conducted at a government hospital, the prevalence of GDM was as high as 27.3% as per the IADPSG criteria and FPG alone detected 63.9% of GDM cases.[14] As in our study, they noted that GDM was associated with increased perinatal risk to the neonates. The reasons for these vast differences in GDM prevalence, even among women of similar socioeconomic strata, are unclear. We found no differences in the maternal characteristics such as age, weight, family h/o diabetes, hypertension between the patients with GDM, and normal glucose tolerance. One reason for this may be that the criteria used classified a very large proportion of women as GDM, suggesting that they include a very mild form of glucose intolerance and may be inappropriate in this population. Similarly, in the study by Nayak et al.[14] among South Indian women of similar socioeconomic strata, there was no association of GDM diagnosed as per the IADPSG criteria with any maternal variable. As expected, we noted an association of FPG and 1-h FPG with the weight of the subject.

The lack of any association of GDM with birth weight may in part be due to the fact that this parameter was available in only 75% of the total cohort studied. However, we did note an expected association with the weight of the subject at the time of OGTT as well with her gestation age. An earlier study of Nayak et al.[14] also reported the lack of any difference in birth weight among women with GDM and nonGDM women.

Previous studies have documented that the intraindividual variation of glucose measured by glucose tolerance test in the general population is higher for the PG sample than the fasting sample.[16] Diagnosis of GDM based on single value of glucose above the cutoff, as suggested by IADPSG, may cause more number of false positive GDM when compared with the Carpenter and Coustan criteria that requires two or more glucose values than the cutoff for diagnosis. Based on the IADPSG cutoffs, we found that only 9.3% will be diagnosed as GDM if any two values of glucose above the cutoff were taken as positive. This will lead to about four times decrease in the number of women being diagnosed as GDM as compared with 42% using a single sample.

The strengths of this study include a meticulous recording of clinical details and performance of OGTT by a single physician, thus reducing the chance of interobserver variation. In addition, in contrast to most studies from the subcontinent, we also studied the association of plasma glucose with follow-up outcomes, such as birth weight. The limitation of this study includes the fact that the birth weight was not available in 22% of women, due to delivery being undertaken in other hospitals and further unwillingness to cooperate. As with any hospital-based study, there may be bias in the frequency of GDM noted, though we minimized this by recruiting patients belonging to lower and middle socioeconomic families from state government-run hospitals.

In summary, there is a high frequency of GDM using the IADPSG criteria in North Indian women of lower and middle socioeconomic strata. Due to the relatively small number of patients in our study, further studies with larger sample size are required to assess our observations of the high frequency of GDM using the IADPSG criteria. If confirmed, the cost-effectiveness of applying these criteria in settings with limited resources would need to be determined.