National and regional prevalence of gestational diabetes mellitus in India: a systematic review and Meta-analysis

This Systematic Review and Meta-Analysis is written in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [24] and is registered in the International Prospective Register of Systematic Reviews (PROSPERO) database (Ref.no. CRD42022335011).

We framed our research question using the PICO(S)(T) methodology (Population-pregnant women; Intervention-nil; Comparison-nil; Outcome-GDM; Study design-cross-sectional in India).

We performed searches in PubMed and Scopus using selected keywords. These results were supplemented by relevant studies from Google Scholar and ShodhGanga—Indian thesis repository (https://​shodhganga.​inflibnet.​ac.​in/​). The last day fir performing the search was 15th June 2022. No date or language restrictions were imposed. The cross-references of the included studies were explored for additional studies. Keywords were identified by iterative discussion among reviewers, and a search query was developed separately for each database. The controlled descriptors (such as MeSH terms) and Boolean operators were used to develop a robust search strategy. (See Additional file 1: Search Strategy).

The studies reporting the prevalence of GDM in pregnant women in India were included.

A reviewer independently conducted searches on all information sources from various databases and uploaded to Rayyan QCRI online software [25]. Rayyan QCRI helped in ensuring a systematic and comprehensive search and selection process. A fourth reviewer managed Rayyan QCRI software, who identified and removed the duplicate citations. Three authors independently screen titles and abstracts with turned “blind” function on. The discrepancies between the three reviewers were discussed with a fourth author for making a consensus to select the articles. Full-text copies of all selected studies were obtained to find more details. We documented the reasons for the exclusion of studies explored as full text. The study inclusion process is presented using the PRISMA flowchart. The reference management software Mendeley Desktop (https://​www.​mendeley.​com) for Windows was used to store, organize, cite, and manage all the included articles.

After selecting eligible studies, we obtained full-text articles for all included studies. Two reviewers independently performed data extraction of relevant information. Data were extracted regarding author, year of publication, study location, site (hospital- or community-based or data-based), study type, trimester, sample size, diagnostic criteria, and prevalence of GDM. We recorded investigators’ definitions of GDM and screening and diagnostic criteria used for GDM.

When a study reports the prevalence of GDM in the same population using multiple diagnostic criteria, the most recent and up-to-date criteria was selected in the following sequence-.IADPSG/ WHO 2013 > DIPSI> WHO 1999 > ADA > NICE> Carpenter and Coustan > NDDG> O’Sullivan and Mahan’s Criteria as framed after the iterative discussion.

A variety of diagnostic criteria were used in a total of 117 studies included in the review. (See Table 2: Different GDM Screening criteria).

DIPSI (29 prevalence estimates) [23] was the most common diagnostic criteria used, followed by IADPSG / WHO 2013 (38 prevalence estimates) [53], WHO 1999 (24 prevalence estimates) [54], and ADA (11 prevalence estimates) [55]. Other criteria used were Carpenter and Coustan Criteria (6 prevalence estimates) [56], NDDG (1 prevalence estimate) [57], NICE (1 prevalence estimate) [58], and O’Sullivan and Mahan’s criteria (1 prevalence estimate) [59]. There was no clear description of study criteria used in 6 studies [33, 60‐63].

We assessed the Risk of Bias using the AXIS tool [26]. Overall, 117 studies were included in the Risk of Bias assessment using the AXIS tool. A horizontal bar graph showing the Risk of bias tool result for each component is given in Fig. 2 Risk of Bias.

Majority of the study components revealed a low risk of bias namely, objective of the study, appropriateness of the study design, study population defined, appropriateness of sample frame, risk factors measured according to the objectives and with the appropriate study tool, accuracy of choice of statistical methods, measures of replicability of the study, description of the basic data, results internally consistent, all results presented and justification of discussion and conclusion.

There was no clear description of response rate bias in 48 studies. Also, there was no description of Ethical consent in 22 studies. Only 9 studies reported funding, but there was no clarity of 28 studies on their funding sources keeping them in unclear risk of bias.

The final 117 studies were used for prevalence estimates of GDM in pregnant population in India. A total of 106 studies were conducted in a hospital-based setting and 11 were community-based studies.

We found a pooled estimate (with an Inverse square heterogeneity model) of the prevalence of overall GDM in pregnant women was 13% [95% CI, 9–16%, n = 117 studies] with the heterogeneity of the studies high at 99% which restricts the generalizability of the findings (Fig. 3 Forest Plot depicting the pooled prevalence of GDM in India) The possible reasons could be studies varied widely in population type, geography, as well as the diagnostic method used. (Table 3 Sub group Analysis) The publication date of the studies ranged from 1989 to 2022.

India has a union of 28 states and 8 Union territories, divided as “North,” “South,” “East,” “Central” or “West” based on the Inter-state council secretariat classification of geographic regions of India [73]. Therefore, region-wise subgroup analysis was also conducted to get estimates of the prevalence of GDM. North region includes Haryana, Himachal Pradesh, Punjab, Delhi, Chandigarh, Uttarakhand, Jammu and Kashmir and Ladakh. States like Gujarat, Rajasthan, Maharashtra, Goa, Daman and Diu and Dadara and Nagar Haveli comprises West Region of India. South India includes Andhra Pradesh, Karnataka, Kerala, Tamil Nadu, Telangana, Andaman and Nicobar Islands, Lakshadweep and Puducherry. East and North-eastern states are Bihar, Jharkhand, Odisha, West Bengal, Arunachal Pradesh, Sikkim, Mizoram, Assam, Meghalaya, Manipur, Nagaland and Tripura. Central Zone of India includes Chhattisgarh, Uttar Pradesh and Madhya Pradesh.

The prevalence of GDM varies across the 5 zones of India. The highest prevalence of GDM was found in North region followed by South India. Areas of low prevalence include West, Central and Eastern zone of India. One of the confounding factors behind low prevalence could be lesser studies conducted in these zones to estimate the prevalence. (Fig. 4 Map of India showing the prevalence of GDM in 5 different zones of India).

The pooled prevalence of GDM in North Zone was found to be 16.1% [95% CI, 9.9–22.7, I² = 98.9%, n = 31 studies]. The maximum weightage (36.53) was by a study from Punnose J et al. conducted in 2018 [28].

Similarly, the pooled prevalence of GDM in West Zone was found to be 7% [95% CI, 3.3–11.2, I² = 98.9%, n = 17 studies]. The maximum weightage (50.24) was by a study from Naik RR et al. 2019 [74].

In Central Zone, the pooled prevalence of GDM was found to be 12.0% [95% CI, 4.3–21.1, I² = 99.29%, n = 13 studies]. A study by Jain R et al. conducted in 2016 has a maximum weightage of 66.55 [75].

The pooled prevalence of GDM in South Zone was 12.6% [95% CI, 7.8–17.8, I² = 98.38%, n = 47 studies]. The maximum weightage was held with study by Jeeyasalan L et al. conducted in 2016 [63].

In East and North-eastern Region, the pooled prevalence of 11.5% was found. [95% CI, 5.3–18.4, I² = 97.34%, n = 9 studies]. The maximum weightage (27.27) by a study done by Hussain et al. in 2020.) [76].

A total of 92 studies were conducted in urban areas, 8 studies in semi-urban areas and 17 studies in rural areas. The pooled prevalence in the rural population was 10.0% [6.0–13.0%, I²₌96%, n = 10 studies], whereas, the pooled prevalence of 12.0% [9.0–16.0%, I² = 99%, n = 88 studies] was found in the urban population. A study conducted by Seshiah V et al. in 2009 included the study participants from urban, semi-urban and rural areas of Tamil Nadu [35].

With the subgroup-analysis using diagnostic criteria, the prevalence of GDM using WHO 1999 criteria was 12.0% (9.0–16.0%), I²₌97% studies, n = 57 studies] which was slightly less than the prevalence of GDM with DIPSI criteria [23] 13.0% [3.0–24.0%, I²₌99%, n = 29 studies] The IADPSG/ WHO 2013 criteria detected a higher prevalence of GDM as 17.0% [12.0–22.0%, I² = 99%, n = 38 studies], while, ADA criteria pooled a lower prevalence of 7.0 [4.0–10.0%, I² = 86%, n = 11 studies]. There was prevalence range of 13.0% [3.0–24.0%, I₂ = 99%, n = 9 studies] was using other criteria like C&C criteria, NICE, NDDG and O′ Sullivan Criteria.

The most commonly used criteria were DIPSI followed by IADPSG/WHO 2013 and WHO 1999. With descriptive analysis, we found that the WHO 1999 criteria detected a high prevalence of GDM as compared to IADPSG and DIPSI which almost detected the pooled prevalence of 12–13%.

Das Mukhopadhyay et al. did not find any significant difference between the prevalence rates of GDM among DIPSI and IADPSG criteria [52]. But he concluded that DIPSI being simple in execution and patient friendly is close to the international consensus. In a study by Singh et al. in 2021, it was observed that DIPSI was only 37.1% sensitive as compared to IADPSG criteria [51]. Contrary to these findings, Seshiah et al. found a high concordance between DIPSI and IADPSG criteria [78]. The low sensitivity of DIPSI has been reported by studies such as Mohan et al.2014 [41]. and Herath et al. [79]. Sensitivity of DIPSI is quite low, hence to be used as screening and diagnostic tool at the same time is still questionable. This is the dire requirement of our country to have a better sensitive method for diagnosing GDM so that patients do not escape diagnosis (false-negatives cases) detected by DIPSI which later on crunch out the health system.

Indeed, in 2013, the WHO embraced the IADPSG criteria, replacing the earlier 1999 criteria. The DIPSI criteria were formulated based on the 2-hour post-glucose (PG) values of the WHO 1999 criteria, primarily focusing on the simplicity of assessing the 2-hour PG value independently. It’s important to note that the Fasting Plasma Glucose (FPG) parameter from the WHO 1999 criteria is considered outdated now, indicative of diabetes [53].

Further, IADPSG recommendation necessarily requires estimation of plasma glucose in three blood samples after administrating 75 g oral glucose load. Pregnant women resent this procedure, as they are pricked three times and feel too much of blood is drawn. Whereas, DIPSI criterion requires one blood sample drawn at 2-h for estimating the plasma glucose Future studies should compare the outcomes of the GDM cases diagnosed by different criteria as this would provide the final answer as to which criteria is more suitable for Indians.

A study by Mohan V et al. in 2014 compared the IADPSG, DIPSI and WHO 1999 criteria shows that the non-fasting OGTT has poor sensitivity compared to both the WHO 1999 criteria (27.7%) and the IADPSG criteria (22.6%) [41]. Thus, the current DIPSI guidelines of doing a single-step non-fasting OGTT using the 2-h VPG cut point of 140 mg/dl (7.8 mmol/l) to diagnose GDM would miss 72.3% of women with GDM diagnosed by the WHO 1999 criteria and 77.4% of women with GDM diagnosed by the IADPSG criteria. Similarly, a study by Tripathi R et al. 2017, a two-hour 75 g non-fasting DIPSI test was done and followed by OGTT [40]. Using OGTT as per the WHO 2013 /IADPSG criteria as gold standard, the sensitivity of 75 g non-fasting test was low. With this low sensitivity, about one quarter of women with GDM were missed. Missing such a large number is not acceptable for a diagnostic test, especially as GDM is associated with both maternal and perinatal complications. On contrary, in the study population, Seshiah V 2012, utilized both DIPSI and IADPSG criteria to ascertain the prevalence of GDM, which were 13.4 and 14.6% respectively [43].

There is a debate regarding the timing of screening for GDM, whether it should be done during the first prenatal visit or during the recommended period of 24–28 weeks of gestation. On the question of when to screen for GDM, a descriptive analysis by Li et al. 2018 showed that a substantial percentage of patients (11.4–60% of GDM cases) develop GDM in the first trimester, but that a similarly large percentage of patients (16–40% of GDM cases) are missed at the first visit [77]. Conducting the screening at later stages of pregnancy is linked to increased risks of maternal and perinatal morbidity and mortality. Many studies on GDM also suggest that early screening and dietary control of GDM can promote the curtailment of maternal and perinatal morbidities [80, 81]. Additionally, Raets et al. demonstrated that there is need for clear guidelines and criteria concerning early screening for GDM [82]. In line with the Flemish consensus of 2019 on screening for GDM, this review recommend to universally screen for diabetes in early pregnancy [83].

Therefore, the review findings indicates an early screening with an OGTT test at 24 weeks coupled with diet counselling and postpartum testing in pregnant women can improve perinatal outcomes [75]. However, this may not be a logistically feasible or cost-effective strategy for all patients, and screening may need to be risk-stratified in Low or Middle Income Country (LMIC).

In their study, Supraneni et al. conducted a comparative analysis of the diagnostic effectiveness of different fasting plasma glucose levels and the one-hour 75 g OGTT in diagnosing GDM [84]. The study found that fasting plasma glucose levels above 92 mg/dL exhibited better diagnostic effectiveness, but there was no significant difference when compared to the results obtained from the one-hour 75 g OGTT in distinguishing between pregnant women with and without GDM.

Additionally, the researchers observed that utilizing the International Association of Diabetes and Pregnancy Study Groups (IADPSG) cutoffs for fasting and one-hour 75 g OGTT demonstrated good diagnostic properties in the study population. By implementing an exit strategy based on a positive result at either the fasting or one-hour mark, it was estimated that the need for further testing could potentially be reduced in approximately one in five pregnant women. However, accessing antenatal care in a fasting state posed challenges in rural settings, as highlighted in a 2014 study by Mohan et al. [41]. On the other hand, the DIPSI (Diabetes in Pregnancy Study Group India) guidelines suggest that the GDM test can be conducted at any time during pregnancy, regardless of food intake [85]. Nevertheless, the DIPSI approach faces difficulties in effectively screening pregnant women for GDM due to low sensitivity and underdiagnosis [86].

Based on the findings of the review, it is clear that a significant need exists for well-designed and unified programs aimed at effectively managing GDM cases. Implementing such programs would be instrumental in reducing the escalating burden of diabetes in India.

There is contradictory evidence reporting varying results and conclusions regarding the accuracy and agreement between blood glucose estimation using venous plasma glucose (VPG) and capillary blood glucose (CBG) methods for diagnosing GDM.

The study by Balaji V in 2012 involving a significant number of cohorts indicated that the Accu-Chek glucometer, a CBG measurement device, provided accurate results that aligned well with laboratory measurements of VPG [72]. Similarly, another study reported that CBG values provided the closest approximation to VPG values in healthy individuals without diabetes or GDM [66]. On the other hand, Jadhav DS conducted a hospital-based clinical study in 2017 comparing VPG and CBG estimation using a glucometer based on the DIPSI criteria found a satisfactory level of agreement between the two methods with equal sensitivity. Additionally, the CBG estimation by glucometer demonstrated a small number of false positive cases due to its high specificity (99.46%) [70].

Indeed, it is interesting to note that in some studies, the capillary blood glucose (CBG) and venous plasma glucose (VPG) values were found to be similar at 1 hour (9.9 mmol/L vs. 9.6 mmol/L) and 2 hours (7.9 mmol/L vs. 7.7 mmol/L) after the glucose load [87]. These findings suggest a fair agreement between CBG and VPG measurements during the 2-hour OGTT test for (GDM.

However, it is worth mentioning that other studies have reported a slight difference between VPG and CBG values, ranging from 0.28 to 0.5 mmol/L (5–9 mg/dL) specifically at the 2-hour mark, although the difference is relatively small [88]. These discrepancies in findings may be attributed to several factors, including the specific population under study, the glucose measurement methods used, and the performance characteristics of the glucose measurement devices employed [89]. The accuracy and agreement between CBG and VPG measurements can vary across different studies and settings.

A recent study by VidyaM Sree et al. demonstrated an excellent diagnostic accuracy (99.77%) of CBG estimation using a one-step OGTT based on DIPSI criteria for GDM in an Indian population. This study highlighted the feasibility and reliability of capillary blood estimation for GDM screening, particularly in countries with limited resources [71].

This review led to the conclusion that capillary blood estimation is a feasible and reliable method for screening GDM In countries with limited resources as this approach requires less technically trained manpower and equipment. It is important for further research to explore and address these differences in order to establish standardized guidelines and protocols for the diagnosis and management of GDM, particularly in terms of blood glucose estimation methods.

The prevalence of GDM varies across different states in India, highlighting the country’s diversity. Even if a universally applicable, feasible, diagnostically accurate, and cost-effective test for GDM is discovered, the gravity of the problem remains consistent.

Supraneni et al. discovered in his study that the IADPSG criteria have good specificity, positive likelihood ratio and post-test probabilities for GDM in their study population [87]. However, the cost involved for performing IADPSG recommended procedure is high, as this procedure requires three blood tests compared to one blood test of DIPSI.

“DIPSI as one-step screening and diagnostic procedure for assessing GDM in pregnant women which is less time-consuming, economical and feasible” as stated by Mounika E et al. in her study conducted in south Indian Population [47]. But, the large extent of false negatives is a major limitation of DIPSI test which cannot be overlooked. Swaroop N et al. used one-step DIPSI criteria in his study and found it to be effective but larger studies are required to further validate its importance [90].

Thus, this review suggests that ideally, and whenever feasible, a single-step 75-g OGTT using the IADPSG criteria should be done in the fasting state as this is the accepted criteria worldwide and would help to bring about international standardization. However, in countries with less resources, DIPSI criteria may be used as a backup option in certain situations where it would be cost-effective without compromising the clinical equipoise: (a) inaccessible areas where pregnant females are not able to visit healthcare facility in fasting state in morning (b) epidemiological studies where fasting sample is unavailable (c) where OGTT is not feasible in some pregnant females due to certain specified reason.

Our review raises a valid point regarding the challenges of implementing a universal screening program for GDM in India. We have taken into account unpublished literature from the Indian database ShodhGanga to gather comprehensive information about the current scenario of GDM in different zones of India. We have made efforts to contact authors to obtain full-text articles or any necessary information for our analysis, ensuring maximum data inclusion.

The review highlights the need for policymakers to reach a consensus on a universal screening test for diagnosing GDM in pregnant women, considering various key factors. These factors include the variation in diagnostic criteria, such as fasting or non-fasting, one-step or two-step approaches, and the use of capillary or venous blood estimation. Additionally, the review considers the sensitivity and specificity of the diagnostic test, the cost-effectiveness of the screening method, and its feasibility in real-world settings.

We also conducted an analysis to assess publication bias. However, since we have included prevalence studies, the results can be generalized to the population regardless of any bias. Furthermore, we performed a sub-group analysis to provide an overview of the current pooled prevalence of GDM in different geographic zones of India.

The authors suggest that implementing a uniform approach nationwide may not be practical. Instead, they propose adopting a more focused and region-specific strategy to maximize resources and efficiently detect and address cases of GDM.

Overall, our review aims to provide evidence-based insights and encourage policymakers to develop consensus guidelines for GDM screening in India. By considering the diverse factors and conducting thorough analyses, we hope to contribute to the formulation of effective strategies for GDM diagnosis and management across the country.

Although we comprehensively searched four databases, we may have included a few more databases to include more GDM-related studies. Further, analyzing the risk factors involved in the prevalence of GDM was not in the scope of our review. Further, some studies did not provide detailed information about their population type, their GDM screening methods, trimester or the distribution between multiple different screening methods that were used. It is imperative to acknowledge the absence of a standardized screening strategy, which introduces a significant limitation to our analysis. Furthermore, we recognize the potential influence of evolving diagnostic criteria on variations in GDM prevalence. To address this concern, it would be beneficial to incorporate a comparative analysis of GDM prevalence across different regions, focusing on studies that employ consistent diagnostic criteria such as DIPSI or IADPSG (WHO 2013). Additionally, we acknowledge that differences in prevalence may be attributed to assessments conducted in distinct time periods. As a means to enhance the comprehensiveness of our review, we highlight the importance of exploring studies that specifically examine trends in GDM within a given population in India over time.