Type 2 diabetes and pre-diabetes mellitus: a systematic review and meta-analysis of prevalence studies in women of childbearing age in the Middle East and North Africa, 2000–2018

Background Investing in women’s health is an inevitable investment in our future. We systematically reviewed the available evidence and summarized the weighted prevalence of type 2 diabetes (T2DM) and pre-diabetes mellitus (pre-DM) in women of childbearing age (15–49 years) in the Middle East and North African (MENA) region. Methods We comprehensively searched six electronic databases to retrieve published literature and prevalence studies on T2DM and pre-DM in women of childbearing age in the MENA. Retrieved citations were screened and data were extracted by at least two independent reviewers. Weighted T2DM and pre-DM prevalence was estimated using the random-effects model. Results Of the 10,010 screened citations, 48 research reports were eligible. Respectively, 46 and 24 research reports on T2DM and pre-DM prevalence estimates, from 14 and 10 countries, were included. Overall, the weighted T2DM and pre-DM prevalence in 14 and 10 MENA countries, respectively, were 7.5% (95% confidence interval [CI], 6.1–9.0) and 7.6% (95% CI, 5.2–10.4). In women sampled from general populations, T2DM prevalence ranged from 0.0 to 35.2% (pooled, 7.7%; 95% CI, 6.1–9.4%) and pre-DM prevalence ranged from 0.0 to 40.0% (pooled, 7.9%; 95% CI, 5.3–11.0%). T2DM was more common in the Fertile Crescent countries (10.7%, 95% CI, 5.2–17.7%), followed by the Arab Peninsula countries (7.6%, 95% CI, 5.9–9.5%) and North African countries and Iran (6.5%, 95% CI, 4.3–9.1%). Pre-DM prevalence was highest in the Fertile Crescent countries (22.7%, 95% CI, 14.2–32.4%), followed by the Arab Peninsula countries (8.6%, 95% CI, 5.5–12.1%) and North Africa and Iran (3.3%, 95% CI, 1.0–6.7%). Conclusions T2DM and pre-DM are common in women of childbearing age in MENA countries. The high DM burden in this vital population group could lead to adverse pregnancy outcomes and acceleration of the intergenerational risk of DM. Our review presented data and highlighted gaps in the evidence of the DM burden in women of childbearing age, to inform policy-makers and researchers. Systematic review registration PROSPERO CRD42017069231


Background
The global burden of type 2 diabetes mellitus (T2DM) is rapidly increasing, affecting individuals of all ages. The global T2DM prevalence nearly doubled in the adult population over the past decade from 4.7% in 1980 to 8.5% in 2014 [1]. The global burden of T2DM in people 20-79 years is further projected to increase to 629 million in 2045 compared to 425 million in 2017 [1]. Lowand middle-income countries will be the most affected with the rise in the burden of T2DM. For the period between 2017 and 2045, the projected increase in the prevalence of T2DM in the Middle East and North Africa (MENA) region is 110% compared to 16% in Europe, 35% in North Africa and the Caribbean, and 62% in South and Central America [1]. Pre-diabetes (pre-DM) or intermediate hyperglycaemia is defined as blood glucose levels above the normal range, but lower than DM thresholds [1]. The burden of pre-DM is increasing worldwide. By 2045, the number of people aged between 20 and 79 years old with pre-DM is projected to increase to 587 million (8.3% of the adult population) compared to 352.1 million people worldwide in 2017 (i.e., 7.3% of the adult population of adults aged 20 to 79 years) [1]. About three quarters (72.3%) of people with pre-DM live in low-and middle-income countries [1].
Pre-DM or T2DM are associated with various unfavorable health outcomes. People with pre-DM are at high risk of developing T2DM [1]. Annually, it is estimated that 5-10% of people with pre-DM will develop T2DM [2,3]. Pre-DM and T2DM are also associated with early onset of nephropathy and chronic kidney disease [4][5][6][7], diabetic retinopathy [6,8,9], and increased risk of macrovascular disease [10,11]. T2DM is also reported to increase the risk of developing active [12] and latent tuberculosis [13]. The rising levels of different modifiable key risk factors, mainly body overweight and obesity, driven by key changes in lifestyle, are the attributes behind the continued burgeoning epidemics of pre-DM and T2DM [14][15][16]. Women of childbearing age (15-49 years) [17] are also affected by the global rise in pre-DM and T2DM epidemics. Rising blood glucose levels in women of childbearing age has pre-gestational, gestational, and postpartum consequences, including increased intergenerational risk of DM [18].
The total population in 20 countries (Algeria, Bahrain, Djibouti, Egypt, Iran, Iraq, Jordan, Kuwait, Lebanon, Libya, Malta, Morocco, Oman, Palestine, Qatar, Saudi Arabia, Syria, Tunisia, the United Arab Emirates, and Yemen) in the Middle East and North Africa region comprises almost 6.7% (~421 million people) of the world's population, with about 200 million females as of July 1, 2015 [19]. In adults ≥ 18 years, T2DM prevalence rose sharply by 2.3 times in each of the Eastern Mediterranean regions and the African region, between 1980 and 2014 [20]. This sharp increase in these two regions is higher than that reported in the region of the Americas (1.7 times), the European region (1.4 times), and the Western Pacific Region (1.9 times) [20].
Key pre-DM and T2DM risk factors, body overweight and obesity, are highly prevalent in people in the MENA countries. In 2013, the age-standardized prevalence of overweight and obesity among women ≥ 20 years was 65.5% (obese 33.9%) [21]. The high burden of overweight and obesity in several MENA countries attributed to the interrelated economic, dietary, lifestyle behavioral factors. The nutrition transitions and changes in the food consumption habits were supported by the witnessed economic development in most of the MENA countries. For instance, in the past five decades, the economic development in the Arab Gulf countries linked to the discovery of oil and gas reserves led to changes in eating habits towards the consumption of foods rich in fat and calories as well as increasing behavioral habits towards a sedentary lifestyle [22,23]. This is particularly true with the significant shift from the consumption of traditional low-fat food to fat-rich foods, as well as with a major change from an agricultural lifestyle to an urbanized lifestyle that is often accompanied by decreased levels of physical activity. The urbanized lifestyle increases exposure to fast foods through the high penetration of fast food restaurants serving fat-rich foods, the reliance on automobiles for transport, and the increasing penetration of cell phones, all of which facilitate low levels of physical activity. Globally, physical inactivity is estimated to cause around 27% of diabetes cases [24]. In eight Arab countries, based on national samples, low levels of physical activity in adults ranged from 32.1% of the population in Egypt in 2011-2012 to as high as 67% of the population in Saudi Arabia in 2005 [25]. Furthermore, fruit and vegetable consumption is inversely associated with weight gain [26]. Studies indicated a low intake of fruit and vegetables in some of the MENA countries [27,28]. The growing burden of the possible risk factors of body overweight and obesity in women may further affect and exacerbate the burden of DM and its associated complications in the MENA countries.
To develop effective prevention and control interventions, there is a need for understanding the actual burden of pre-DM and T2DM epidemics in vital population groups, such as women of childbearing age (15-49 years), in the MENA region. Thus, individual studies need to be compiled and summarized. According to our previously published protocol (with a slight deviation) [29], here, we present the results of the systematically reviewed published quantitative literature (systematic review "1"), to assess the burden (prevalence) of T2DM and pre-DM in women of childbearing age in the MENA region, from 2000 to 2018.
Investing in women's health paves the way for healthier families and stronger economies. Societies that prioritize women's health are likely to have better population health overall and to remain more productive for generations to come [30]. Against this background, our review was aimed at characterizing the epidemiology of T2DM and pre-DM in population groups of women of childbearing age in the MENA through (1) systematically reviewing and synthesizing all available published records of T2DM and pre-DM and (2) estimating the mean T2DM and pre-DM prevalence at national, sub-regional, and regional levels, from January 2000 to July 2018. The findings of the review fill an evidence gap to inform policy-makers on the epidemiologic burden of T2DM and pre-DM in women of childbearing age.

Methods
Following our published protocol [29] that is registered with the International Prospective Registry of Systematic Reviews (PROSPERO registration number "CRD42017069231" dated 12/06/2017), we reported here systematic review "1". This review adheres to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) 2009 guidelines [31][32][33]. The PRISMA checklist is provided in the Additional file 1.

Data source and search strategy
To identify eligible studies on T2DM and pre-DM prevalence measures in MENA countries, we implemented a comprehensive computerized search of six electronic databases (MEDLINE, EMBASE, Web of Science, SCOPUS, Cochrane library, and Academic Search Complete) from January 1, 2000, to July 12, 2018, using variant Medical Subject Headings (MeSH) and free-text (Text) terms. The detailed search strategy is presented in an additional box file (see Additional file 2). We also hand-searched the reference lists of eligible studies for further studies that might have been missed.
We defined the participants, exposure, comparator, outcome(s), and type of study "PECO(T)". The PECO(T) statement provides the framework for the identification and selection of studies for inclusion [34]. As we were looking for prevalence studies, we only considered participants and the outcomes.

Inclusion and exclusion criteria
Participants: Women of childbearing age were defined according to the World Health Organization (WHO) as women aged between 15 and 49 years (thereafter, women of childbearing age) [35]. Pregnant women were also considered in this review as long as they were tested for T2DM and/or pre-DM according to what was reported in the individual studies.
Outcomes: T2DM and pre-DM. The included studies should have reported quantitative or calculable pre-DM or T2DM prevalence estimate(s) in women of childbearing age regardless of the sample size, pregnancy status, or pre-DM/T2DM ascertainment methodology, in any of the 20 MENA region countries [36]. We excluded studies of selfreported pre-DM/T2DM not supported with either anti-DM medications or a documented diagnosis. We also excluded studies on metabolic syndrome as long as there was no clear information on the proportion of women of childbearing age with pre-DM or T2DM. Studies were also excluded if they pooled women of childbearing age with pre-DM/T2DM with other non-communicable diseases in the same category, or together with males, or for each gender separately but without age stratification. We excluded studies with incalculable pre-DM/T2DM prevalence after attempting to contact the authors at least twice with no response.
Types of studies: We included observational studies if they were cross-sectional, comparative cross-sectional, casecontrol (not comparing T2DM/pre-DM vs. no T2DM/pre-DM), or cohort study designs. We excluded observational studies of other study designs.
Detailed eligibility criteria are available in the published protocol [29]. The PRISMA flow chart for the selection of studies is shown in Fig. 1.

Identifying eligible studies
Titles and abstracts of the remaining citations were screened independently by four reviewers (AI, KA, MM, and MQ) for any potential study on pre-DM/T2DM in childbearing age women. Full-texts of the identified potentially eligible studies were thoroughly screened and independently assessed by the four reviewers. The qualities of the extracted studies were independently assessed by two other reviewers (RHA and FA). Discrepancies in data extraction were discussed and resolved.

Data extraction
Data from fully eligible studies were extracted into a pre-defined data extraction excel file using a pre-defined list of variables [29]. Our outcome of interest was the national/regional weighted pooled prevalence of T2DM and pre-DM in women of childbearing age in the MENA. We extracted the following data on the baseline characteristics of the eligible research reports (author names, year of publication, country, city, and study setting), study methodology (design, time period, sampling strategy, and T2DM/pre-DM ascertainment methodology), and study population (age, pregnancy status, co-morbidity, and number of women with the outcomes of interest).
In research reports which provided stratified T2DM/ pre-DM prevalence estimates, the prevalence of the total sample was replaced with the stratified estimates keeping the rule of having at least 10 tested subjects per strata, otherwise we extracted information on the whole tested sample. We followed a pre-defined sequential order when extracting stratified prevalence estimates. Outcome measures stratified according to body mass index (BMI) were prioritized, followed by age and year. This prioritization scheme was used to identify the strata with more information on the tested women. When the strata were not prioritized, the overall outcome prevalence measured was extracted. For a research report that stratified the prevalence of the outcome of interest at these different levels (i.e., age and BMI), one stratum per research report was considered and included to avoid double counting. If the outcome measure was ascertained by more than one ascertainment guideline, we extracted relevant information based on the most sensitive and reliable ascertainment assay (i.e., prioritizing fasting blood glucose "FBG"

Meta-bias
We generated a funnel plot to explore the small-study effect on the pooled prevalence estimates. The funnel plot was created by plotting each prevalence measure against its standard error. The asymmetry of the funnel plot was tested using the Egger's test [37] (see Additional files 3 and 4).

Quality appraisal and risk of bias
We assessed the methodological quality and risk of bias (ROB) of the studies on T2DM or pre-DM prevalence measures using six-quality items adapted from the National Heart, Lung, and Blood Institute (NIH) tool [38]. Of the 14 items proposed for observational studies on the NIH tool, eight items were not used as they are relevant only for cohort studies assessing the relationship between an exposure and an outcome [38]. We also assessed the robustness of the implemented sampling methodology and the ascertainment methodology of the measured outcome(s) using three additional quality criteria (sampling methodology, ascertainment methodology, and precision of the estimate). Studies were considered as having "high" precision if at least 100 women tested for T2DM/pre-DM; a reasonable precision, given a pooled prevalence of 7.2% for T2DM or 7.6% for pre-DM estimated in this study, was obtained. We computed the overall proportion of research reports with potentially low risk of bias across each of the nine quality criteria. We also computed the proportion (out of nine) of quality items with potentially a low risk of bias for each of the included research reports.

Quantitative synthesis: meta-analysis
Meta-analyses of the extracted data to estimate the weighted pooled prevalence of T2DM and pre-DM and the corresponding 95% confidence interval (CI) were executed. The variances of prevalence measures were stabilized by the Freeman-Tukey double arcsine transformation method [39,40]. The estimated pooled prevalence measures were weighted using the inverse variance method [40], and an overall pooled prevalence estimate was generated using a Dersimonian-Laird random-effects model [41]. Heterogeneity measures were also calculated using the Cochran's Q statistic and the inconsistency index; Isquared (I 2 ) [42]. In addition to the pooled estimates, the prevalence measures were summarized using ranges and medians. The prediction interval, which estimated the 95% interval in which the true effect size in a new prevalence study will lie, was also reported [42,43]. Country-level pooled estimates were generated according to the population group of tested women (general population, pregnant, non-pregnant with history of gestational DM (GDM), and patients with co-morbidity), and the overall country-level pooled prevalence, regardless of the tested population and study period. To assess if the prevalence of T2DM and pre-DM is changing over time, we stratified studies into two time periods: 2000-2009 and 2010-2018. In order not to miss any important data when estimating country-level, sub-regional, and regional prevalence, the period for studies that overlapped these two periods was defined as "overlapping". In studies with an unclear data collection period, we used the median (~2 years) that was obtained from subtracting the year of publication from the year of data collection to estimate the year of data collection in those studies. The "patients with co-morbidity" included women of childbearing age with organ transplant, kidney dialysis, cancer, HIV, chronic obstructive pulmonary disease, polycystic ovarian syndrome (PCOS), or schizophrenia. Categorization of the study period was arbitrary with an aim to estimate the change in T2DM and pre-DM at the country-level and overall, over time.
We also estimated the weighted pooled prevalence, regardless of country, according to the tested women's population group, study period, T2DM/pre-DM ascertainment guidelines (WHO guidelines, American DM Association (ADA) guidelines, International DM Association (IDF) guidelines, or medical records/anti-DM medications/selfreported), and sample size (< 100 or ≥ 100). The overall weighted pooled prevalence of T2DM and pre-DM regardless of the country, tested population, study period, ascertainment guidelines, and sample size was also generated. Providing pooled estimates regardless of the ascertainment guidelines was justified by the fact that the subject women were defined and treated as T2DM or pre-DM patients following each specific ascertainment guidelines.
To provide prevalence estimates at a more sub-regional level, countries in the MENA region were re-grouped into three sub-regions, namely, "Arab Peninsula, Fertile crescent, and North Africa and Iran." The pooled prevalence in these three sub-regions was estimated according to the tested population group, study period, ascertainment guidelines, and sample size, as well as overall for each sub-region.
We also estimated the weighted pooled prevalence of T2DM and pre-DM according to age group. We categorized women of childbearing age into three age groups (15-29 years, 30-49 years) and not specified/overlapping. The "not specified/overlapping" category covers women who did fell in the other two age groups. For example, women with an age range of 25-34 years or 18-40 years. The age group weighted pooled prevalence produced regardless of the country, sub-region, and tested population as well as study period.

Sources of heterogeneity: meta-regression
Random-effects univariate and multivariable metaregression models were implemented to identify sources of between-study heterogeneity and to quantify their contribution to variability in the T2DM and pre-DM prevalence. In univariate meta-regression models, analysis was performed by country, tested population, study period, ascertainment guidelines, and sample size. All variables with a p < 0.1, in the univariate models, were included in the multivariable model. In the final multivariable model, a p value ≤ 0.05 was considered statistically significant, contributing to heterogeneity in prevalence estimates.
All meta-regression analyses were performed using the metareg package in Stata/SE v15 [44].

Meta-bias in T2DM prevalence
The asymmetry in the funnel plot examining the smallstudy effects on the pooled T2DM prevalence among women of childbearing age indicates evidence for the presence of a small-study effect (Egger's test p < 0.0001). The funnel plot is presented in an additional figure file (see Additional file 3).

Predictors of heterogeneity in T2DM prevalence
In the univariate meta-regression models, all variables except study period, T2DM ascertainment criteria, and sample size were associated with T2DM prevalence at p value < 0.1. In the adjusted meta-regression model, none of the included variables was significantly associated with T2DM prevalence at p value < 0.05. In two studies       Q: Cochran's Q statistic is a measure assessing the existence of heterogeneity in estimates of T2DM prevalence 2 I 2 : a measure assessing the percentage of between-study variation that is due to differences in T2DM prevalence estimates across studies rather than chance 3 Prediction interval: estimates the 95% confidence interval in which the true T2DM prevalence estimate in a new study is expected to fall 4 General populations could include healthy population, health care workers, migrant workers, or employees 5 Year range does not cover every single year within that range. In studies with unclear information on when the study was conducted, we subtracted 2 years from the publication year as this was the median of the data collection period and the publication year for the other studies with full information 6 Study period was before and after 2009 7 Pooled estimate, regardless of the tested population, sample size, and data collection period, used the most updated criteria when T2DM was ascertained, based on different criteria in the same population 8 Patients could be those on kidney dialysis, or with arthritis, organ transplant, cancer, HIV, COPD, PCOS, or schizophrenia 9 Regardless of the year of the guidelines for the most updated criteria when T2DM was ascertained, based on different criteria in the same population 10 Overall pooled estimate in the 15 countries regardless of the tested population, sample size, and data collection period, using the most updated criteria when T2DM ascertained using different criteria in the same population NE not estimable, CI confidence interval calculated using the exact binomial method,

Scope of reviewed pre-DM reports
The 24 research reports on pre-DM prevalence yielded 52 pre-DM prevalence studies and were from 10  [62] and 2018 in Saudi Arabia [81]. Thirteen (25.0%), 11 (21.2%), and 11 (21.2%) of the pre-DM prevalence studies were from Iran, Saudi Arabia, and UAE, respectively. Approximately 87.0% of the pre-DM prevalence studies tested women of childbearing age sampled from general populations. The pre-DM prevalence estimates ranged from 0.0% in various age groups in multiple countries [51,60,70] to 40.0% in Iraq in women aged 20-39 years, recruited from the general population [55] (Table 1).

Pooled pre-DM prevalence
In the 10 countries, the weighted pre-DM prevalence in women of childbearing age was estimated at 7.6% (95% CI, 5.2-10.4%, I 2 , 99.0%) (  (Table 3). In 13 studies in Iran (7 from the general population), the prevalence of pre-DM ranged from 0.0 to 21.4% with an overall weighted prevalence of 3.8% (95% CI, 1.2-7.6%, I 2 , 98.3%). The 11 pre-DM studies in Saudi Arabia were in women of childbearing age sampled from the general population, with an overall weighted pre-DM prevalence of 6.6% (95% CI, 3.  2) by publication year.

Meta-bias in pre-DM prevalence measures
The asymmetry in the funnel plot examining the smallstudy effects on the pooled pre-DM prevalence among women of childbearing age indicates evidence for the presence of a small-study effect (Egger's test p < 0.0001).
The funnel plot is presented in an additional figure file (Additional file 4).

Predictors of heterogeneity in pre-DM prevalence
Country, study period, and pre-DM ascertainment criteria were associated with a difference in the pre-DM prevalence in the univariate meta-regression models at p value < 0. Quality assessment of the T2DM/pre-DM research reports Findings of our summarized and research report-specific quality assessments for relevant DM prevalence studies can be found in Additional file 12. Briefly, all the 48 research reports clearly stated their research questions or objectives, clearly specified and defined their study populations, and selected or recruited the study subjects from the same or similar populations. There was a clear gap in the reporting or justifying of the sample size calculation in 79.2% of the research reports. The majority (87.5%) of the research reports tested ≥ 100 women of childbearing age, and they were classified as having high precision. Overall, the 48 research reports were of reasonable quality with potentially low ROB in an average of 7.2 items (range, 6-9). Four (8.3%) of the 48 reports had potentially low ROB in all the measured nine quality items [66,82,83,86] (see Additional file 12).

Discussion
We provided, to our knowledge, the first regional study that comprehensively reviewed and estimated the regional, sub-regional, and country-level burden of T2DM and pre-DM in various populations of women of childbearing age in the MENA. Based on the available data from 14 and 10 studies in MENA countries, the present findings document      I 2 : a measure assessing the percentage of between-study variation that is due to differences in pre-DM prevalence estimates across studies rather than chance 3 Prediction interval: estimates the 95% confidence interval in which the true pre-DM prevalence estimate in a new study is expected to fall 4 General populations could include healthy population, health care workers, migrant workers, or employees 5 Year range does not cover every single year within that range. In studies with unclear information on when the study was conducted, we subtracted 2 years from the publication year as this was the median of the data collection period and the publication year for the other studies with full information 6 Study period was before and after 2009 7 Pooled estimate, regardless of the tested population, sample size, and data collection period, used the most updated criteria when pre-DM was ascertained, based on different criteria in the same population NE not estimable, CI confidence interval calculated using the exact binomial method, the comparable burden of T2DM (7.5%, 95% CI 6.9-9.0%) and pre-DM (7.6%, 95% CI 5.2-10.4%) in women of childbearing age. The estimated prevalence of T2DM and pre-DM in 14 countries in the MENA is similar to the estimated worldwide crude diabetes prevalence of 8.2% (95% credible interval (CI) 6.6-9.9%) in adult women in 2014 (age-standardized 7.9%, 95% CI 6.4-9.7%) [91]. The T2DM and pre-DM prevalence in women of childbearing age varied across the three sub-regions in the MENA, by population group, time period, DM ascertainment criteria, and sample size. The obvious common prevalence of T2DM and pre-DM in women of childbearing age in the MENA countries reflects the highest prevalence of adult diabetes estimated for the MENA [91]. In this region, the crude diabetes prevalence in adult women increased from 5.0% in 1980 to 9.0% in 2014 [91]. This increase in diabetes prevalence among adult populations in the MENA over time is higher than many other regions including Europe and Central and West Africa [91]. The highest national adult diabetes prevalence estimates documented in the MENA is 5-10 times greater than the lowest national prevalence estimates documented in Western European countries [91]. T2DM is a significant public health problem in both developed and developing countries that can lead to various health complications including increased overall risk of dying prematurely [20]. The common burden of T2DM and pre-DM in women of childbearing age, which is reflected in the high burden of adult diabetes in this region [91], might be mainly driven by the sociodemographic changes in this region. In recent decades, there was an increase in median age, sedentary lifestyle, and physical inactivity in the MENA [92]. These lifestyle changes are linked to an increase in the burden of body overweight and obesity that are shared predisposing factors for pre-DM and T2DM [20]. At the population level, physical inactivity was very common in many MENA countries (Saudi Arabia 67.  [25]. The burden of body overweight and obesity is higher in many lowincome and middle-income countries in the MENA than in Europe and Asia Pacific countries [93]. Obesity in women in several Middle Eastern countries was 40-50% [93]. The age-standardized prevalence of obesity was 32.0% in Egypt, 35.5% in Jordan, 30.4% in Iraq, 32.5% in Libya, and 35.4% in Saudi Arabia [94]. In Tunisia, 43.7% and 24.1% of 35-70-year-old females in urban and rural areas, respectively, were obese [95]. In 2016, in almost all of the countries in MENA, the mean BMI for people aged ≥ 18 years was ≥ 25.0 [96]. To curb the burden of DM and its associated complications in women of childbearing age in the MENA countries, our results suggest three main implications for care. First, based on the estimated 5-10% progression rate from pre-DM to T2DM [3,10], out of the 47, 958 tested women of childbearing age for pre-DM (Table 3), we estimate that 2398 to 4796 women are expected to progress to T2DM. This risk of progression to T2DM could be reduced through lifestyle and drugbased interventions as it was reported elsewhere [97][98][99]. In England, 55-80% of participants with hyperglycemia at baseline had normal glycaemia at 10 year follow-up [3]. The high burden of DM along with pre-DM in women of childbearing age could accelerate maternal complications including GDM leading to increased intergenerational risk of DM. Programs to halt the growing epidemic of DM among different population groups could start by addressing the key risk factors including sedentary lifestyle and increased body weight. Addressing this problem would require social and public policies and efforts to reduce the national and regional burden of increased body weight and obesity through enhancing healthy eating behaviors and physical activity. Second, there is a critical need for strengthened surveillance systems that match the scale and nature of the DM epidemic in women of childbearing age in the MENA. Enhancing early detection and management of high-risk individuals requires accessible and affordable health care systems, outreach campaigns to raise public awareness, and social and medical support to induce and maintain a healthy lifestyle. Adult people at increased risk of T2DM and pre-DM can be predicted based on good screening tools from the Centers for Disease Control and Prevention (CDC) [100] and the American Diabetes Association (T2DM Risk Test) [101]. Early screening and detection will require government-funded prevention programs. Third, controlling the burden of T2DM and pre-DM in MENA countries requires strong and successful partnerships between public health and clinical departments. Physicians have a fundamental role in the care of individual patients to screen, diagnose, and treat both pre-DM and T2DM in clinical settings. In addition, physicians have a fundamental role in working to raise awareness and participating in developing prevention programs and engaging communities. Concerted efforts and partnership between physicians, health departments, and community agencies are needed to strengthen health care services, encouraging and facilitating early screening and detection, and promoting healthy diets and physical activity.
Providing summary estimates and up-to-date mapping gaps-in-evidence of T2DM and pre-DM prevalence in women of childbearing age in different MENA countries provides the opportunities for future public health interventions and research to better characterize the T2DM and pre-DM epidemiology nationally and regionally. Nevertheless, present review findings suggest that the DM burden in women of childbearing age in MENA countries is capturing only the tip of the iceberg. Identifying gaps-inevidence through systematically reviewing and summarizing the literature has public health research implications. Our review shows that in many countries, the estimation of the burden of T2DM or pre-DM in women of childbearing age in general populations occurred more than a decade ago (Table 1). Additionally, the review shows that there was no data on the burden of T2DM and pre-DM in women of childbearing age in several countries in the MENA region. This lack of evidence on a key public heath outcome requires a strongly resourced research capacity and research funding schemes. There is evidence that federally funded research can impact important health issues that affect a large segment of the population [102].

Strengths
This robust approach to the literature search and review as well as in retrieving and extracting relevant data from the published literature allowed us to provide summary estimates on the burden of T2DM and pre-DM in women of childbearing age from the 14 and 10 countries in the MENA, respectively. Once the diagnosis was established, regardless of the ascertainment criteria, patients were treated as having diabetes or pre-diabetes. Thus, generating pooled estimates, regardless of the DM ascertainment criteria, stratified according to various population groups, provided more insights into the actual burden of T2DM and pre-DM in various populations of women of childbearing age. The meta-regression analysis identified sources of variations in T2DM and pre-DM prevalence and sources of between-study heterogeneity in prevalence estimates. (Additional files 9 and 11 show these in more detail). The country-stratified and populationstratified T2DM and pre-DM prevalence reports revealed gaps in evidence that can help strengthen research and DM control programs in the most affected countries and populations. The use of probability sampling was very common in the studies included, which may provide broader insights on the representation of our findings to the general or specific group of women of childbearing age at the national, but not at the regional, level.

Limitations
There are important but unavoidable limitations when interpreting the results of our review. Despite the estimated DM prevalence, the actual DM burden could have been underestimated, at country, sub-regional, or regional level, due to several reasons. The inaccessibility of data on pre-DM or T2DM in women of childbearing age from several countries in the MENA may not necessarily mean an actual lack of data. To meet the aim of our review of estimating the burden of pre-DM and T2DM in women of childbearing age, in several published studies reviewed, women of childbearing age were found to have been combined with those of other age groups or with men. The presented overall pooled estimates, regardless of the tested population group, should not be interpreted as the total burden of the outcome at the population level. Utilizing data on T2DM and pre-DM from only 14 and 10 countries may limit the findings from being generalizable to the entire MENA region. Although we followed a thorough and well-defined search strategy, there is a potential of publication bias as shown in funnel plots (Additional files 3 and 4). The estimated T2DM and pre-DM prevalence suggest that only the tip of the iceberg was captured. The presented estimates may not be representative of the true prevalence for each population. This underestimation may be particularly true in low-resource settings where necessary resources and capacity in investigating pre-DM at the community level are lacking. The wide array of blood glucose cut-off points and criteria used for T2DM and pre-DM ascertainment also suggests that overestimation and underestimation bias cannot be excluded. Unless estimated from individual population-based studies only, the presented weighted pooled estimates at the country, sub-regional, or regional level should not be interpreted as the burden of the measured outcomes at the population level. Also, the presented pooled estimates according to the two time periods, from 2000 to 2009 and from 2010 to 2018, should not be interpreted as an over-time change in the burden of the measured outcomes. While our meta-analyses revealed substantial heterogeneity across studies, the meta-regression analyses identified the potential sources of between-study heterogeneity within the framework of the present study and the level of detail that can be used in describing these sources (Tables 1 and 2). Thus, much of the variability in T2DM and pre-DM prevalence across studies might remain unexplained.
Despite these potential limitations, our study provided a characterization of the scale of T2DM and pre-DM among women of childbearing age in several MENA countries based on the best available evidence. Data presented in this review can be used to (a) understand the burden of T2DM and pre-DM among a vital population group and to identify at high-risk populations within this specific population group; (b) guide the planning, implementation, and evaluation of programs to prevent and control DM; (c) implement immediate public health actions to prioritize the allocation of public health resources; and (d) formulate research hypotheses and provide a basis for epidemiologic studies. Future research opportunities should prioritize large country-level and multicenter comparable studies, to determine the prevalence of T2DM and pre-DM in various population groups of women of childbearing age. A definitive characterization of the burden of DM in women of childbearing age at the regional and sub-regional level would require comparable and empirical studies using standardized methodology and comparable DM ascertainment assays.

Conclusions
In conclusion, women of childbearing age in the MENA region bear an appreciable burden of T2DM and pre-DM. The estimated burden of T2DM and pre-DM was higher in the Arabian Peninsula and Fertile Crescent countries compared to the rest of the MENA countries identified with prevalence estimates in this review.