Introduction
Several studies have explored the effect of anthropometric risk factors on metabolic syndrome. However, no systematic effort has explored the effect of overweight and obesity on the prevalence of metabolic syndrome in India. Thus, we undertook a meta-analysis to estimate the effect of anthropometric risk factors on the prevalence of metabolic syndrome.
Methods
We searched databases PubMed Central, EMBASE, MEDLINE, and Cochrane library and search engines ScienceDirect and Google Scholar, from January 1964 through March 2021. We used the Newcastle–Ottawa scale to assess the quality of published studies, conducted a meta-analysis with a random-effects model, and reported pooled odds ratios (OR) with 95% CIs.
Results
We analyzed 26 studies with a total of 37,965 participants. Most studies had good to satisfactory quality on the Newcastle–Ottawa scale. Participants who were overweight (pooled OR, 5.47; 95% CI, 3.70–8.09) or obese (pooled OR, 5.00; 95% CI, 3.61–6.93) had higher odds of having metabolic syndrome than those of normal or low body weight. Sensitivity analysis showed no significant variation in the magnitude or direction of outcome, indicating the lack of influence of a single study on the overall pooled estimate.
Conclusion
Overweight and obesity are significantly associated with metabolic syndrome. On the basis of evidence, clinicians and policy makers should implement weight reduction strategies among patients and the general population.
Introduction
Metabolic syndrome encompasses a spectrum of disorders, including central obesity, atherogenic dyslipidemia (ie, low high-density lipoprotein cholesterol [HDL-C], elevated triglycerides, and apolipoprotein B–containing lipoproteins), elevated blood pressure, elevated blood glucose, and prothrombotic and proinflammatory states (1). Metabolic syndrome recently emerged as a significant and growing public health challenge worldwide resulting from rapid urbanization, excessive energy intake, developing obesity, and sedentary lifestyle habits (2). People with metabolic syndrome have increased risk of type 2 diabetes mellitus, cardiovascular disease, myocardial infarction, and stroke and twice the risk of death from these causes compared with people without the syndrome (3).
Metabolic syndrome is characterized by chronic low-grade inflammation, which is caused by complex interactions between genetic and environmental factors (4). Prevalence has varied from 10% to 84% worldwide, depending on both the criteria used for diagnosis (5) and differences in the geographic distribution, ethnicity, age, and sex of the population studied (6). A recent meta-analysis showed that the prevalence of metabolic syndrome in India is 30% and is more commonly seen among older adults (>60 y), women, and the urban population (7). However, research exploring factors that determine this high prevalence is limited. Factors such as genetic susceptibility, obesity, physical inactivity, smoking, and alcohol consumption are components of the syndrome’s natural history (8).
Several studies have explored the reasons why obesity and physical inactivity affect metabolic syndrome. Adipocyte hypertrophy and hyperplasia, enhanced by obesity and overweight, influence the overproduction of biologically active metabolites, known as adipocytokines, such as free fatty acids, tumor necrosis factor-α, interleukin-6, and plasminogen activator inhibitor-1 (9). These mediators initiate a localized and systemic inflammation that facilitates multiple processes, such as insulin insensitivity, oxidant stress, blood coagulation, and inflammatory responses, that in turn accelerate atherosclerosis (10). Researchers have experimented with various treatment options, such as lifestyle and diet modifications, pharmacologic therapy, weight reduction, behavioral therapy, and bariatric surgery, to reduce the syndrome’s prevalence (2).
Existing evidence on anthropometric factors related to metabolic syndrome is not country-specific; however, it is essential to know whether the influence of these factors differs from country to country. Although India has almost one-third of the world’s adult population with metabolic syndrome, no systematic effort has been made to explore the effect of overweight and obesity on the syndrome’s prevalence in India. To develop effective strategies and implement relevant policies or programs to address the prevalence of metabolic syndrome, policy makers must have information on its contributing factors. However, we found no systematic review to date that examined the association worldwide between anthropometric factors and metabolic syndrome. Hence, we undertook our meta-analysis to estimate the effect of anthropometric risk factors on the prevalence of metabolic syndrome to inform researchers in India and worldwide.
Methods
Design and registration
We conducted a systematic review and meta-analysis of observational studies. The protocol was registered in PROSPERO under the registration number CRD42019147277. The “Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement 2020” was used to report this systematic review incorporating the meta-analyses (11). The institutional review board of Jawaharlal Institute of Postgraduate Medical Education and Research declared this study exempt from review.
Eligibility criteria
We included any observational study irrespective of its design (ie, cross-sectional, case-control, cohort) that reported the relevant exposure (anthropometric risk factors) and outcome (metabolic syndrome) in India. We did not restrict studies by geographic region (rural or urban) or study setting (community, facility, workplace). Only full-text publications were included, and we excluded studies published as conference abstracts, case reports, or case series and unpublished data.
The studies included were conducted among adults in India aged 18 years or older and assessed the association of anthropometric risk factors (overweight/obesity) with metabolic syndrome. We excluded disease-specific (eg, noncommunicable diseases, mental health disorders) studies. Studies reporting prevalence of metabolic syndrome in relation to different anthropometric factors were included irrespective of the type of definition or criteria used for diagnosis (eg, National Cholesterol Education Program Adult Treatment Panel III guidelines, International Diabetes Federation guidelines, Harmonized Asia Pacific criteria).
We conducted a systematic search of literature in electronic databases (PubMed Central, EMBASE, MEDLINE, and Cochrane library) and by using search engines (ScienceDirect and Google Scholar). Both medical subject headings (MeSH) and free-text words were used for all searches (Table 1). We used a set of keywords and their synonyms for searches with appropriate truncations and wildcards and for proximity searching. We also searched for key concepts by using corresponding subject headings in each of the databases. Our final search was conducted by combining individual search results by using appropriate Boolean operators (“OR” and “AND”). The search was narrowed by using the available filters for time period (from inception of the databases [January 1964 through March 2021]), language (published in English language only), and study design (observational studies). Bibliographies of the retrieved articles were also hand-searched to identify any articles missed during the database search.
Study selection
Our study selection process involved 3 stages:
- Primary screening: Two independent investigators (Y.K. and S.R.) performed primary screening of title, abstract, and keywords by executing the literature search. Full-text articles were retrieved for the studies shortlisted on the basis of eligibility criteria.
- Secondary screening: Full text of these retrieved studies was screened by Y.K. and S.R. and assessed against our eligibility criteria. We included studies that satisfied all eligibility criteria with respect to design, participants, exposure, and outcome.
- Finalizing the study selection: Disagreements among investigators during the screening process were resolved, and final consensus on inclusion of studies was reached with the help of another investigator (S.M.). Where the required information was missing, we contacted the corresponding author of the respective study and obtained the information. If we received no response from the author, the study was excluded.
Data extraction
We manually extracted data by using a predefined, structured data extraction form that included general information about the article, such as author and year of publication, and information related to the methods section, including study design, setting, sample size, sampling strategy, study participants, inclusion and exclusion criteria, exposure and outcome assessment method, quality-related information, the number of participants in exposed and nonexposed groups, and the number of exposed and nonexposed participants with metabolic syndrome. Data were entered by S.R., and entries were double-checked by Y.K. and S.M. for accuracy. We based our criteria for a diagnosis of metabolic syndrome on those of the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) (4), the International Diabetes Federation (3), the American Heart Association and the National Heart, Lung, and Blood Institute (4), and the Harmonized Asia Pacific criteria (4) (Box).
Risk-of-bias assessment
Two independent investigators, S.R. and S.M., used the Newcastle–Ottawa Quality Assessment Scale to perform risk-of-bias assessment for observational studies (12). The scale consists of 3 domains, each of which receives a score of stars that varies by category: selection (maximum 4 stars), comparability (maximum 2 stars), and outcome (maximum 2 stars). Within each of these domains, we assessed representativeness, sample size justification, nonresponse, ascertainment of exposure, control for confounding, assessment of outcome, and statistical tests. The total score for a study ranged from 0 to 8 stars. Studies with 7 to 8 stars were considered of good quality; 5 to 6, of satisfactory quality; and 0 to 4, of unsatisfactory quality (12).
Statistical analysis
We used STATA version 14.2 (StataCorp) to perform our meta-analysis. Because all outcomes were dichotomous, the number of events and participants in each group were entered to obtain the pooled effect estimate in terms of odds ratio (OR). We used the random effects model with the inverse variance method to calculate the weight of individual studies. Evidence of between-study variance resulting from heterogeneity was assessed through a χ2 test of heterogeneity and by using I2 statistics to quantify the inconsistency. An I2 less than 25% indicates mild heterogeneity; 25%–75%, moderate; and more than 75%, substantial (13). Study-specific and pooled estimates were graphically represented through forest plots. We also performed subgroup analysis by using multiple study characteristics or covariates, study setting, geographic region, metabolic syndrome definitions, and quality of studies. We were able to perform analysis based on study design because only 1 study was a prospective study; the rest were cross-sectional.
We performed univariable meta-regression analysis with study-level characteristics. Variables with a P value less than .20 were used to perform multivariable meta-regression for identifying the source of heterogeneity between the studies. We assessed publication bias for each of the outcomes by using a funnel plot and a Doi plot (14) for visual interpretation and Egger test (13) and the Luis Furuya-Kanamori asymmetry index (LFK index) (14) for statistical interpretation. Asymmetry of the funnel plot or the Doi plot and a P value less than .10 in the Egger test indicates possible publication bias. On the basis of the LFK index value, the possibility of publication bias is classified as no asymmetry (value within ±1), minor asymmetry (value greater than ±1 but within ±2), and major asymmetry (value greater than ±2) (14).
We performed sensitivity analysis to assess the robustness of results by removing the studies one by one and checking for any significant variation in results. We also performed random-effects cumulative meta-analysis to delineate temporal changes in the magnitude and direction of the pooled association estimate because the evidence accumulates over time. First, we sorted the studies by publication year and then added them sequentially to analysis in chronological order, recalculating the pooled estimates with each added study (15).
Results
Study selection
We found 3,321 studies through our systematic literature search. We also retrieved full texts for 4 articles obtained through manual searching of the bibliographies in the retrieved studies. After removing duplicates, we screened 2,786 articles. Of these, we excluded 2,659 because the title and abstract indicated that they did not have relevant study participants or exposure or outcomes. We assessed 127 for eligibility and excluded 101 (67 because relevant risk factors weren’t assessed, 22 because required information was not available, and 12 because the studies described were conducted among metabolic syndrome patients only). A total of 26 studies with a total of 37,965 participants were included in our final review for qualitative and quantitative (meta-analysis) synthesis and met our eligibility criteria (16–41). We used a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow chart to describe the screening and selection process (Figure 1).
