Introduction

Cardio-metabolic syndrome (CMS) is a complex clustering of metabolic and cardiovascular risk factors including dyslipidemia, systemic hypertension, central obesity, and dysglycemia [1]. Nowadays, with obesity and diabetes growing in prevalence, CMS is on the rise around the world [2, 3]. Increased prevalence of CMS and its subsequent outcomes in the cardiovascular, cerebrovascular, renal, and endocrine systems impose harmful effects on the world population [4].

An even more alarming fact is that cardiometabolic risk factors (CMRFs) are not limited to adults. In most countries, the prevalence of CMRFs has been increasing in the pediatric population owing to the rise in childhood overweight and obesity [2]. Dietary pattern and nutritional factors are among the major modifiable risk factors of obesity and metabolic disorders [5, 6]. Epidemiological studies have shown the priority of overall dietary pattern assessment in preventing chronic diseases compared with evaluation of single nutrients or food effects [7, 8]. Assessment of dietary diversity is a dietary approach to ensure high quality diets [9, 10]. Furthermore, dietary diversity is associated with adequate intake of all groups of foods and nutrients such as macronutrients and micronutrients that are protective elements against chronic diseases [11, 12].

Dietary diversity score (DDS) has been known as a useful, convenient, and inexpensive indicator of overall diet quality in large-scale surveys [13, 14]. DDS appraises all food groups diversity within a diet based on a dietary guidance [15]. Accordingly, a higher DDS is indicative of a healthier diet, and is associated with meeting all the essential nutrient requirements [16]. Previous studies have reported the association between DDS and health problems such as diabetes, metabolic syndrome, cardiovascular disease and cancer [17, 18]. In recent years, many investigators have evaluated the association between DDS and some CMRFs such as hyperlipidemia [8, 19], hypertension [8], high blood glucose [20], and abdominal obesity [21, 22]. These studies have shown controversial results in terms of beneficial or negative associations. Therefore, there is no conclusive result about the effect of DDS on cardio-metabolic risk factors. In 2016, the authors of the first systematic review and meta-analysis on the relationship between DDS and obesity failed to show any significant relationship between BMI status and high or low DDS [23].

To our knowledge, up to now, there is no comprehensive systematic review and meta-analysis about the association between DDS and CMRFs. To clarify this association, we performed a systematic review and meta-analysis of the literature.

Methods

To assess the association of DDS with cardio-metabolic risk factors, we conducted a systematic review through which all related documents were searched in international databases including PubMed and NLM Gateway, Scopus, and Institute of Scientific Information (ISI) up to October 2019. The relevant references in the selected articles were also reviewed. The protocol of present study is registered on the PROPERO with code CRD42020157127.

Search strategy

The main routes for the development of search strategies were extracted from "dietary diversity score”, "DDS", "diet quality score”, "food variety score" and all related terms to cardio metabolic risk factors (Table 1).

Table 1 Search strategy
Full size table

Inclusion and exclusion criteria

Studies were eligible for inclusion in this review as follows:

(a) Quantitative observational research design (e.g., cross-sectional); (b) Assess associations between DDS and CMRFs; (c) No limitation for time of researches or publications; (d) No limitation for language of papers; (e) No limitation for age and gender. Experimental studies in animals and those with duplicate citations were excluded. The studies that were limited to subgroup populations also were excluded. For multiple unique source publications, the most complete source of data included.

Quality assessment and data extraction strategy

Results of searches saved in Endnote software. Through three phase of refinement; titles, abstracts, and full texts were reviewed, respectively. The quality assessment of the included documents was conducted by two researchers, independently. We assessed the quality of the included studies using the Newcastle–Ottawa Quality Assessment scale [24]. Regarding quality assessment, the kappa statistic for agreement was 0.94. Probable discrepancy resolved by third expert opinion. The data were extracted from moderate and high quality studies.

The data were extracted using a checklist including general information of the study and its citation; population characteristics; methodological information of study (study region, scope of study (local study or survey), sample size, age and sex and interested measured outcomes which include overweight and/or obesity, abdominal obesity, metabolic syndrome, diabetes, BMI, waist circumference, level of lipid profile and blood pressure.

Statistical analysis

Odds ratio (OR) with 95% confidence interval (CI) was considered as the effect size for association of DDS with CMRFs as binary outcomes (obesity, overweight, abdominal obesity, diabetes, MetS) in meta-analysis. Moreover, in some studies, means (SD) of CMRFs were compared between the highest and lowest DDS, for which we used standardized mean difference (SMD) as the effect size in meta-analysis. The Chi-square based Q test and I square statistics used to assess the heterogeneity between studies. The result of Q test was regarded to be statistically significant at P < 0.1. In case of severe heterogeneity between studies regarding the reported effect sizes, pooled effect size was estimated using random-effect meta-analysis model (with the Der-Simonian and Laird method). Moreover, a forest plot was used to present the result of meta-analysis schematically. To identify possible sources of heterogeneity, subgroup analysis was done. Publication bias was evaluated using Egger’s test and it was presented schematically using funnel plots. The analyses were conducted using STATA 11 software. P values less than 0.05 were considered statistically significant.

Ethical considerations

The protocol of this study was approved by the ethical committee of Alborz University of Medical Sciences. All the included studies would be cited in all reports and all future publications.

Results

Search results and characteristics of the included studies

Figure 1 shows the detailed process of data refinement and studies selection. Using the above-mentioned search engines, 843 original research studies were identified (PubMed: 97, ISI: 215, Scopus: 531). After reviewing titles, abstracts, and full texts, 825 articles were excluded, because their design and population were not interested or there was not assessing DDS for individuals. In the final step, full texts were reviewed for cited references and five more articles were included. Finally, 23 observational studies (22 cross-sectional and one cohort) [8, 12, 13, 16, 19,20,21,22, 25,26,27,28,29,30,31,32,33,34,35,36,37,38,39] published between 2005 and 2018 were included according to the inclusion/exclusion criteria. Characteristics of the included studies were presented in Table 2. Fourteen studies were conducted in Asia [8, 12, 13, 19,20,21,22, 25, 26, 29, 35,36,37,38], five in Africa [28, 30, 32,33,34], two in North America [16, 39], one in South America [31] and one in Europe [27]. A relevancy assessment and quality control identified a total of 50,287 study subjects based on the inclusion/exclusion criteria (The sample size of included studies ranged from 113 to 17,825) were studied. Among 23 enrolled studies, 18 articles reported required data (OR and/or mean and SD across DDS categories) for the meta-analysis. Therefore, meta-analysis was performed on them. Five other studies were systematically reviewed [12, 16, 30, 37, 39]. Nineteen articles studied on healthy adults (ranged from 18 to 67 years) and 4 articles studied on children [12, 16, 32, 34]. 24-h recalls (n = 18) and FFQ (n = 5) [8, 13, 20, 22, 27] were the most common dietary assessment methods. To score dietary diversity, nine studies used five main groups according to the Food Guide Pyramid [8, 13, 16, 19, 22, 26, 27, 37, 39], 1 study used six [36] and another one used seven food groups [28], 4 studies used nine [12, 20, 34, 38], four articles used 12 [21, 25, 32, 35], one used 14 [30] and three studies used 24 food groups [29, 32, 33].

Fig. 1
figure 1

Flow chart of the number of studies selected into the meta-analysis

Full size image
Table 2 Characteristics of included studies in the systematic review
Full size table

Qualitative synthesis

Data from five other studies, including children (two studies) [12, 16] and healthy adults (three studies) [30, 37, 39], could not be pooled. All of the five articles evaluated the relation of DDS with BMI. Only in Kant et al. study, in addition to body mass index, the relationship between DDS and lipid profile and blood pressure was also examined [39]. The results of Fernadez et al. study [16] and Zhao et al. study [12] which investigate the association between DDS and BMI in children were inconsistent. Fernadez et al. reported that girls and older children had greater DDS (P = 0.04, P = 0.03) and greater healthy variety score, overall variety, and DDS were associated with greater annual increases in z-score of BMI prospectively (β[SE] = 0.009 [0.004], P = 0.04; β[SE] = 0.007 [0.003], P = 0.02; β[SE] = 0.003 [0.001], P = 0.02, respectively). Whereas Zhao et al. reported that in children > 6 years old, dietary diversity indicators were negatively associated with BMI and BAZ, just opposite to that of children ≤ 6 years old. Furthermore, in comparison of 3 food diversity indicators between overweight or obese children and normal-weight children, the results failed to find a significant association between food diversity indicators and obesity status, too. Kant et al. and Tiew et al. [37] reported a negative association between DDS and BMI. Their results showed that those being overweight or obese, and those with lower WHR were found to have a significantly poorer dietary diversity, while Keding et al. [30] revealed that BMI directly correlated with DDS, suggesting that the greater the diversity of foods and food groups eaten, the higher the BMI. Moreover, in Kant et al. study, a negative relationship was observed between DDS and lipid profile and blood pressure. A meta-analysis was not performed due to insufficient data. These results of included studies in systematic review (qualitative synthesis) are shown in Table 3.

Table 3 Results of included studies in systematic review
Full size table

Quantitative synthesis

Overall, of the studies included in the meta-analysis, eight had examined the relation of DDS and obesity [22, 25, 26, 31, 32, 34, 35, 38], five articles reported the relation between DDS and overweight [12, 22, 26, 31, 33] and four articles assessed relationship between DDS and metabolic syndrome [19, 20, 29, 36], four reported association of DDS with abdominal obesity [19, 21, 25, 38], BMI and WC [8, 13, 21, 27]. In addition to these indices, 2 studies also evaluated the relation of DDS and diabetes [21, 28], lipid profiles such as TG, TC, LDL and HDL, and blood pressure (systolic and diastolic blood pressure) [8, 13]. These results of included studies in meta-analysis are shown in Table 4.

Table 4 Meta-analysis of association between diet diversity score and cardio-metabolic risk factors
Full size table

The association of DDS with obesity-related outcomes

DDS with obesity, abdominal obesity, and overweight

Our meta-analysis based on eight papers [22, 25, 26, 31, 32, 34, 35, 38] with a total of 21,796 subjects failed to reach a significant association between DDS and obesity (OR = 0.98; 95% CI 0.92,1.05) with substantial heterogeneity (Q = 39.57; P = ˂ 0.001; I2% = 82.3).

Subgroup analysis based on age group revealed that there was no significant relationship between DDS and obesity in both adults (OR = 0.98; 95% CI 0.91, 1.04) and children (OR = 3.3; 95% CI 0.07,6.54). In the subgroup analysis, a significant inverse relationship between higher DDS and reduced risk of obesity was observed among studies whose participants were women (OR = 0.41; 95% CI 0.18, 0.7), whereas the analysis on studies with both sexes did not show a significant association (OR = 1.19; 95% CI 0.86, 1.51) and one study which assessed the relationship between DDS and obesity in men showed a significant direct association (OR = 3.4; 95% CI 1.2, 5.6).

The results of meta-analysis on articles which studied the relation of DDS and abdominal obesity (4 studies, number of participants = 2887) and overweight (5 studies, number of participants = 19,578) failed to show significant relationship (OR = 1.17; 95% CI 0.6, 1.73) with absolute heterogeneity (Q = 22.45; P = ˂ 0.001; I2% = 86.6) and (OR = 1.03; 95% CI 0.73, 1.31) with substantial heterogeneity (Q = 14.73; P = 0.005; I2% = 72.9).

DDS with BMI and waist circumferences

The results from the meta-analyses failed to show a significant relationship between DDS and body mass index (4 studies, number of participants = 3898, SMD: 0.32, 95% CI − 0.01, 0.65) and waist circumference (4 studies, number of participants = 3898, SMD: 0.05, 95% CI − 0.31, 0.41) but there was considerable heterogeneity among estimates (Q = 27.59; P = ˂ 0.001; I2% = 89.1) and (Q = 32.89; P = ˂ 0.001; I2% = 90.9), respectively.

The association of DDS with MetS

Our meta-analysis on five studies with 5701 participants that assessed the relation of DDS with MetS showed there was no significant association between quartiles of DDS and risk of metabolic syndrome (OR = 0.80; 95% CI 0.71, 1.81) with substantial heterogeneity (Q = 25.65; P = 0.005; I2% = 84.4).

The association of DDS with diabetes

There was no significant association between DDS and risk of diabetes (2 studies, number of participants = 4410, OR = 1.06; 95% CI 0.94, 1.18) with non-considerable heterogeneity (Q = 1.73; P = 0.19; I2% = 42.3) and blood pressure (2 studies, number of participants = 348).

Systolic blood pressure (OR = − 2.43; 95% CI − 5.83, 0.97), substantial heterogeneity (Q = 117.28; P = ˂ 0.001; I2% = 99.1) and diastolic blood pressure (OR = − 0.94; 95% CI − 2.64, 0.76) with considerable heterogeneity (Q = 41.18; P = ˂ 0.001; I2% = 97.6).

The association of DDS with lipid profile

The result of meta-analysis on 2 studies with 348 participants which evaluate relation of DDS with lipid profile revealed an inverse association between higher DDS and lower level of TG (SMD − 0.23, 95% CI − 0.45, − 0.01) without considerable heterogeneity (Q = 0.48; P = 0.49; I2% = 0.0), whereas there was no significant relation between DDS and level of TC (SMD − 0.37, 95% CI − 1.13, 0.39) with substantial heterogeneity (Q = 8.88; P =  < 0.001; I2% = 88.7), LDL (SMD − 2.0, 95% CI − 5.92, 1.92, 0.19) with obvious heterogeneity (Q = 164.38; P =  < 0.001; I2% = 99.4) and HDL (SMD 0.19, 95% CI − 0.029, 0.41) without obvious heterogeneity (Q = 0.01; P = 0.93; I2% = 0.0).

Publication bias

The funnel plot assessment by visual inspection was done and revealed that it is quite symmetrical and no publication bias for any variable was demonstrated. Similarly, the Egger's test also showed that there was no publication bias.

Risk of Bias

Table 5 shows the individual and summary risk of bias assessments of the included articles based on Newcastle–Ottawa Quality Assessment scale. The majority of studies were assessed as having low risk of bias across domains.

Table 5 Risk of bias assessment of included studies
Full size table

Discussion

To the best of our knowledge, the present systematic review and meta-analysis is the first comprehensive study assessing the relationship between DDS and cardio-metabolic risk factors including overweight, obesity, abdominal obesity, metabolic syndrome, diabetes, blood pressure and lipid profile.

Our results showed that in the case of BMI, five studies reported that increased DDS was positively associated with BMI [8, 13, 16, 21, 30], whereas three studies showed an inverse relationship between DDS and BMI [12, 27, 39]. Overall, ten studies had studied obesity and overweight in relation to DDS. Two studies failed to find an association between obesity/overweight and DDS [32, 33]. Six studies reported a direct association between obesity and DDS [23, 26, 31, 34, 35, 38], while two studies reported an inverse association [25, 37]. Out of four studies which assessed the association between abdominal obesity and DDS, two articles reported positive association [21, 38] and two study reported negative [19, 25]. Consistent with a previous systematic review and meta-analysis [23], our results failed to show a significant association between DDS, BMI, and overweight/obesity/abdominal obesity. However, in a systematic review by Asghari and colleagues [40], scoring on the basis of dietary diversity has direct association with weight gain. One of the reasons for the discrepancy between the results of their study (Asghari et al.) and the present study may be the smaller number of included articles and, therefore, the smaller sample size.

To find out the source of heterogeneity, subgroup analysis was performed based on sex and age. Subgroup analyses according to sex in studies reporting ORs of overweight/obesity suggested a complete absence of heterogeneity in studies whose participants were only women or men. Furthermore, another subgroup analysis based on age showed the absence of heterogeneity in studies that assessed the association of DDS with obesity in children, while those studies that assessed this association in adults had a substantial heterogeneity. Perhaps the reason for non- significant results was the inconsistency of the results of primary studies and there was a great deal of contradiction between the primary studies. Moreover, our results showed a protective effect of DDS on obesity in women. The different effects and contradictory findings in men and women may be due to the effects of hormonal changes and more weight gain over time in women, as well as receiving higher amounts of energy in men compared to women and as a result being in challenge with controlling fat and sodium intake in men.

Three studies reported blood pressure and lipid profile as an outcome. Two of these articles showed negative association between systolic blood pressure and lipid profile with DDS, whereas one study showed no significant relationship. The results of our study did not show a significant relationship between DDS and blood pressure and lipid profile except the level of TG. According to our analysis, higher quartile of DDS had protective and decreasing relationship with level of TG. These findings were inconsistent with a previous study which showed a relationship between FGDS (food group dietary score) and serum lipid levels [41]. The medium FGDS category had the highest mean triglyceride level, consistent with the higher obesity rate, whereas the low FGDS category had an association with the lowest total serum cholesterol, LDL-cholesterol and highest HDL-cholesterol levels. An inverse association was found between overall dietary score and blood pressure, high LDL-cholesterol levels [8, 42], hypercholesterolaemia, diabetes [8], and triglycerides [42], all known CVD risk factors in adults.

Two studies reported that increased DDS was significantly associated with decreased risk of metabolic syndrome, while three studies failed to identify a significant association between DDS and metabolic syndrome. Furthermore, in the two studies evaluating the relationship between the diet diversity score and OR of diabetes, no significant relationship was reported.

Unfortunately, regarding the relationship between diet diversity score and metabolic syndrome and diabetes, so far, not only no secondary study was done, but also the number of initial studies is too small to see definitive results.

This study had several limitations. First, some studies reported the DDS–BMI/blood pressure and lipid profile association only at baseline [8, 13, 21, 27]. Unfortunately, because researchers used different food groups to calculate DDS, we were unable to show to what extent the DDS contributed to individual foods or food groups. In fact, studies used a range of defined food groups to score diversity and most studies used dietary recall rather than a food frequency questionnaire to examine the diversity of the diet. Since the tool used to calculate and assessing dietary diversity scores plays an important role in determining the relationship with risk factors such as obesity, it seems that the use of a food frequency questionnaire is the most appropriate tool for assessing long-term diet diversity. Considering the fact that overweight individuals adopt a healthier diet to manage their weight, conflicting results could also be explained partly by this confounding factor. Moreover, most of the included studies did not adjust for two important confounders: drug therapy, supplements intake, energy intake and socioeconomic status. Adjustment for energy intake and socioeconomic status in future studies is highly recommended. It should be borne in mind that due to the lack of data, we were unable to include five studies in the meta-analysis. Moreover, studies that clearly stated that they used DDS were included in our meta-analysis and other dietary scores such as FDS and FGDS did not enter in our study. In the present meta-analysis, we compared cardio-metabolic risk factors between participants with the highest and lowest DDS to evaluate the linear association between DDS and variables; however, the association may be curvilinear (U-shaped). Despite these limitations, the study has notable strengths such as using comprehensive assessment of a variety of cardio-metabolic risk factors in adults and children, with a relatively large participant number and the use of a random effects model for overall analysis.

In conclusion, our systematic review and meta-analysis showed that there was no significant association between DDS and most of the CMRFs. On the other hand, our results showed an inverse association between DDS and TG. The situation may be due to different methods for assessing DDS as well as various methods used to evaluate dietary intake. Reassessment of the overall DDS tool as a criterion of diet quality and production of new and valid DDS standard tools is highly desirable. In addition, further longitudinal studies and field trials are recommended to confirm these findings.

What is already known on this subject?

In primary studies, there is no conclusive result about the beneficial or negative effect of DDS on cardio-metabolic risk factors and up to now, there is no comprehensive systematic review and meta-analysis about the association between DDS and CMRFs. To clarify this association, we performed a systematic review and meta-analysis of the literature.

What does this study add?

Our results showed an inverse association between DDS and TG but we don’t observed any significant association between DDS and other CMRFs (obesity and overweight, BMI, diabetes, metabolic syndrome and lipid profile). These results may be due to the lack of a unified method in estimating DDS, so it seems that it needs to reassessment of the overall DDS tool and production of new and valid DDS standard tools to achieve conclusive results.