Associations Between Yogurt Consumption and Obesity Risk: A Systematic Review and Meta-analysis of Observational Studies

Background: Plenty of studies have shown that yogurt have specific benefits for human health, but the association between yogurt consumption and obesity risk is still indistinct from previous research. Objectives: The aim of this study was to make a systematic review of recent epidemiological studies about yogurt consumption and obesity risk, and conduct a meta-analysis. Methods: Related studies were searched in electronic databases up to 1 November 2020. The meta-analyses synthesize included obesity, overweight and abdominal obesity comparing extreme categories of yogurt consumption. Results: A total of 35 studies from 27 articles were carried out in the review. Yogurt intake could decrease the risk of obesity (OR:0.83, 95% CI: 0.79-0.87), and abdominal obesity (OR: 0.80, 95% CI: 0.69-0.92), but showed no associations with overweight happen (OR: 0.88, 95% CI: 0.58-1.36). A dose-response analysis shows that the risk of obesity de-creased by approximately 44% with increasing intake of yogurt up to ~165 g/day. Moreover, most prospective cohort studies reveal that the intake of yogurt in a long term can reduce weight and waist circumference, but not affect the change of BMI. Conclusions: The meta-analysis indicates that yogurt consumption would reduce the overall obesity and abdominal obesity. In addition, long term consumption of yogurt may contribute to some obesity-related anthropometry change.


Introduction
Obesity has always been a significant public health issue, and the prevalence of overweight and obesity continues to increasing globally [1]. Many epidemiological studies have shown that obesity is a risk factor for many chronic diseases, including cardiovascular disease, diabetes [2], chronic kidney disease [3], and various cancers [4,5]. As of 2015, about 107.7 million children and 603.7 million adults worldwide were obese, the overall prevalence of obesity was 5.0% and 12.0% respectively [6]. The latest research has predicted that about 1 in 2 adults will be obese in the United States by 2030 [7]. Some factors, such as individual behavior, environment, and gene may lead to imbalance in energy intake and excretion, causing obesity [8]. In order to maintain a good balance of energy metabolism, a sensible diet and physical exercise are necessary for the prevention and treatment of obesity [8]. Furthermore, substantial evidence supports the beneficial role of dairy products in dietary energy distribution [9], may decrease the risk of obesity [10,11].
Dairy products are a major part of the daily diet and contribute many important nutrients to our daily diet. Yogurt, one of the dairy products, is loved by many people because of its unique flavor and ingredients. Compared with milk, in addition to valuable nutrients such as protein, minerals, vitamins, and fatty acids [12], yogurt also includes some probiotics [13], which can alter the composition of the intestinal flora as well as affect food intake, appetite, weight, and metabolic functions [14,15]. A meta-analysis suggested that yogurt intake could reduce the metabolic syndrome risk [16]. With the increase in the consumption rate of yogurt, its impact on human health especially obesity has attracted much attention.
Several studies have explored the relationship between yogurt intake and obesity risk. A 2016 meta-analysis suggests a negative correlation between yogurt and the risk of central adiposity [17]. Subsequently, Sayón-Orea et al. [18] have reviewed the prospective study of yogurt on obesity, overweight, and metabolic syndrome, but the correlation has not been confirmed. So we updated related research reviews and performed a meta-analysis with published reports to clarify the relationship between yogurt and obesity incidence.

Materials and Methods
This systematic review with meta-analysis was designed, implemented, analyzed, and reported following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) [19] protocol. This systematic review and meta-analysis was registered at PROSPERO as CRD42020191376.

Search Strategy and Selection Criteria
We searched from PubMed, Embase, and Web of Science for relevant articles that described the associations between yogurt and obesity published up to 1 November 2020, only English articles. We use the following search terms: (yogurt OR yoghurt OR sour milk OR fermented milk OR fermented dairy OR cultured milk) AND (obesity OR overweight OR "Body composition" OR "Body constitution" OR "Weight status" OR body mass index OR BMI OR waist circumference OR WC OR Metabolic Syndrome OR anthropometry) and (prospective OR follow-up OR cohort OR longitudinal OR cross-sectional). Besides, all relevant references that appeared in retrieved articles were searched and reviewed manually to obtain other qualified publications. The search for articles only included full-text journal English articles from the original study and has been done independently by three investigators (J.W. and L.L.), subject to approval by a third reviewer (Y.X.). Any disputes were resolved through group discussions until a consensus was reached.
Studies were included in this systematic review if the following conditions were met: 1) studies were prospective observational or cross-sectional design; 2) the yogurt was the exposed factor; 3) the outcomes were weight or waist circumference change, overweight, obesity, and abdominal obesity, and outcome criteria were appropriately well defined. Furthermore, for meta-analysis also includes; 4) the relationship of yogurt consumption and obesity, overweight, abdominal adiposity are reported with adjusted odds ratio (OR), hazard ratio (HR), or relative risk (RR) and their 95% confidence interval (95% CI) (or provide information that can be used in risk calculations; 5) for the dose-response analysis, yogurt intake in each category was provided. If the same data were used in multiple articles, our analysis use the latest data.
We excluded the study if: 1) the data was only reported other dairy products such as milk, butter or the ingredients of yogurt; 2) the participants of studies were pregnant or lactating females; 3) the study type is meta-analyses, reviews, and case-reports; 4) other non-specified outcomes, such as malnutrition.

Data Extraction and Quality Assessment
The author (J.W. and L.L.) extracted data and assessed the eligibility of studies from included literature, and were checked by a third investigator (Y.X). The following data were extracted from 32 studies: publication year, the first author's name, geographic location, study design (prospective study or cross-sectional study), participants (number, and Percentage of men), age range, event and cases number, intake comparison, dietary assessment, type of yogurt, outcome type, change of anthropometry (BMI, body weight, waist circumference) and the corresponding OR or HR with 95% CI. The ORs and HRs were extracted of which most adjustment or calculated by the numbers of cases and controls. If appropriate, the extracted information will be used for subgroup analysis.
The methodological quality were assessed using the Newcastle-Ottawa Scale (NOS) [20] for cohort studies with evaluation score ranged from 0 to 9 and using the Agency for Healthcare Research and Quality (AHRQ) methodology checklist [21] for cross-sectional studies with evaluation score ranged from 0 to 11 [22]. The scores of 7-9 and 8-11 are considered as good quality of cohort and cross-sectional studies, respectively. Any inconsistencies were resolved through group discussion with an additional investigators (Q.L., H.W. and F.W.).

Statistical Methods
We assessed the association between yogurt intake and obesity, overweight, abdominal obesity risk with the pooled OR and 95% CI in this meta-analysis. Studies with a clear number of cases and controls and effect size with 95% CI was included in the synthesis. If quantitative synthesis is not appropriate, we extracted effect size and summarized in the table.
We use the Q test and I 2 statistic to evaluated the heterogeneity, and P <0.10 suggested significant heterogeneity [23]. If I 2 ≤50%, Mantel-Haenszel fixed-effects model was used, otherwise, Der Simonian and Laird random-effects model was used [24]. A meta-regression was performed to test which heterogeneous sources could have a significant impact on the differences between any studies [25]. To assess the effect of potential key covariates, stratified analyses were performed for geographic location (the U.S., Europe, and South Korea), dietary assessment (validated, non-validated, or not available [N/A]), and adjustment by physical activity/energy intake/smoking/alcohol (yes or no).
Dose-response analyses were conducted with the method described by Greenland and Longnecker [26] for the risk of obesity and the increased intake of yogurt. We extracted information from the selected articles with yogurt consumptions grading. The midpoint of the range was used if the median or mean was not reported. The width of the adjacent category to the top-level category was used if the top category is open-ended, and the lowest boundary was set to zero if the lowest category was open-ended [27]. Then, we evaluated possible non-linear associations between the consumption of yogurt and the risk of obesity by using restricted cubic splines with three knots at the 10%, 50%, and 90% of the distribution [27,28]. We calculated the non-linear p-value by testing the null hypothesis that the second spline coefficient was equal to zero.
We performed a "leave-one-out" sensitivity analysis by removing one study at a time to evaluate the possible influence of each included study [29]. Moreover, we used quantitative Egger [30], and Begg [31] tests to assess the publication bias. All data were analyzed with STATA software (version 11.0; Stat Corp, College Station, TX, USA) and reported P values were 2-sided, with P < 0.05 was considered statistically significant.

Search Results for Systematic Review and Meta-analysis
We initially retrieved 1138 articles in databases and added 3 manually retrieved articles, then, excluded 465 articles due to duplicates, and 612 articles that did not meet our inclusion criteria. In the rest 64 potentially eligible articles, 11 reviews are excluded, 12 articles were not yogurt consumption, 14 articles were not required outcomes. After all, there were 27 articles included in the review. After we excluded insufficient and confusing data (n= 17 records), 10 articles were included in the final meta-analysis, where 7 articles include cross-sectional studies, and 3 include cohort studies. All the included articles were evaluated by PRISMA. (Figure 1)

Study Characteristics
This review included 27 articles reporting 35 studies listed in Table 1. 8 studies for meta-analysis contain 90,496 controls and 24,648 cases of obesity [32][33][34][35][36][37]. Four studies with 25,664 cases [32,33,38] and five studies with 28,465 cases [18,33,39,40] carry out meta-analysis for overweight and abdominal adiposity, respectively. Where two studies were performed in the United States [34,37], seven in Europe [18,32,33,35,38,40], and five in South Korea [36,39]. Participants were aged from 13 to 90 years. All information on yogurt consumption was collected through questionnaires except one for 24h-recall [36], and total yogurt exposure was used in this meta-analysis. The studies quality was evaluated with the NOS or AHRQ methodology checklist in Table 2 and Table 3 respectively. Two cross-sectional studies [33,40] and three prospective studies [18,37,39] have good quality, and the rest were medium quality in this meta-analysis.
This review also includes studies on anthropometric changes related to obesity, including BMI, body weight, and waist circumference. Except for the every four years changes of body weight in three studies [41], the remaining studies were annual change. in anthropometry with increased consumption of yogurt per day [42][43][44][45][46]. In terms of the type of yogurt, there were 8 studies on the whole yogurt [41,[43][44][45][46] and 3 studies on whole-fat or low-fat yogurt [44,46,47].
Three of them reported that yogurt has a significant protective effect on obesity [33,35,53], nine studies' result were not significant [32, 34-37, 50, 52], but two studies concluded that sour milk consumption will increase the risk of obesity [51]. Only one study shown that eating whole-fat yogurt could reduce the incidence of obesity, while eating low-fat yogurt won't [49] (Table 4). Eight studies with no specific number of cases were excluded while extracting data for statistical analysis [48][49][50][51][52].
Then we found an inverse association between yogurt intake and obesity in the meta-analysis, the pooled OR of obesity risk for the highest level versus the lowest level of yogurt intake was 0.83 (95% CI: 0.79, 0.87; I 2 = 0.0%) ( Figure  2). Geographical subgroup ( Whether the race was adjusted has no effect on the pooled OR. A non-linear dose-response association was observed (P-trend = 0.03) from 3 studies [32,33,36] which shows the risk of obesity decreased by approximately 44% with increasing intake of yogurt up to ~165 g/day ( Figure 3).  For each study, the evaluation score was ranged from 0 to 9, and the scores of 0-3, 4-6 and 7-9 were considered as bad, medium and good quality respectively.  An item would be scored "1" if it was answered "YES", and if the answer was "NO" or "UNCLEAR" it would be scored "0". For each study, the evaluation score was ranged from 0 to 11, and the scores of 0-3, 4-7 and 8-11 were considered as low, moderate and high quality respectively.

Yogurt Intake and Risk of Overweight
We also found Yogurt intake does not affect the risk of the overweight occurs with a pooled OR of 0.88 (95% CI: 0.58, 1.36; I 2 = 82.8%) (Figure 4) calculated using a random effects model. Only one study with 64,939 participants adjusted for confounding factors (OR: 0.94; 95% CI: 0.91, 0.98) [33]. The OR of the other three studies was calculated by cases and the number of controls [32,38] (Table 6). We did not conduct a dose-response analysis due to insufficient number of studies. There still have 4 articles about yogurt consumption and the risk of being overweight or obese that do not meet the meta-analysis conditions [55][56][57][58] (Table 5). The data from a Prospective cohort study after a median follow-up of 6.6 years was analyzed by Martinez-Gonzalez et al. [57] and 1860 individuals were identified as overweight/obesity. The group of high (>7 servings/week) intake of total and whole-fat yogurt had lower incidence of overweight/obesity than lower intake (0-2 servings/week) (OR:0.80; 95% CI:0.68, 0.94; P-trend= 0.001; and OR=0.62; CI:0.47, 0.82; P-trend <0.001). In participants with higher fruit consumption this trend was more obvious. But this inverse association was not found in the low-fat yogurt category (OR:0.84; 95% CI:0.61, 1.15; P-trend =0.601). A longitudinal analysis was conducted by Johansson et al. [56] after 8-12 years follow-up of 27,682 participants in Northern Sweden. Higher intake (Quartile 5) of fermented milk had lower incidence of overweight/obesity than lower intake (Quartile 1) in men (OR:0.92; 95% CI:0.85, 0.99; P-trend =0.035), not in the women participants (OR:0.98; 95% CI:0.91, 1.06). In the study of Cormier et al. [55] obesity incidence was no difference between yogurt consumers and non-consumers. Calculated from a prospective cohort study about middle-aged and older women, Rautiainen et al. [58] found yogurt consumption will lead a higher risk of overweight/obesity (HR:1.16; 95% CI:1.02, 1.31; P-trend =0.002). As for the yogurt subgroup analysis, only one study shown that eating whole-fat yogurt could reduce the incidence of obesity, while eating low-fat yogurt won't [57].

Yogurt Intake and Risk of Abdominal Obesity
The pooled OR of abdominal obesity risk of yogurt intake was 0.80 (95% CI: 0.69, 0.92; I 2 = 88.8%) (Figure 5), indicating that increased consumption of yogurt would degrade the occurrence of abdominal obesity. We also did not conduct a dose-response analysis due to lack of sufficient data.
Two studies have used OR [18,33], and two studies have used HR [39,54] and one has used RR [40]. Only one study has suggested that yogurt consumption would not affect abdominal obesity [40], while the others showed that the yogurt consumption would reduce the possibility of being abdominal obesity [18,33,39,54]. There five other studies not included in the analysis because of no cases number (Table  8). In a follow-up study, the HR of abdominal obesity was 0.74 (95% CI:0.61, 0.91) Comparing tertile 3 and tertile 1 of yogurt consumption in elderly individual [54]. Song et al. [52] has found a higher intake (≥100 g/d) of yogurt had a lower incidence of abdominal obesity in men compared to no intake (OR:0.41; 95% CI:0.24, 0.70; P-trend<0.01), rather than women (OR:0.65; 95% CI:0.37, 1.13; P-trend<0.12). And in a survey from Korea yogurt did not affect the incidence of abdominal obesity if it's consumption above 1 once per day [50]. Nevertheless, after five years follow-up, the risk of abdominal obesity has increased with annual rates of change in yogurt consumption (OR:1.21, 95% CI: 1.01, 1.44; P-trend<0.05) in HANDLS cohort [48].   Total yogurt include all kind of yogurt and undefined type. 2 Hazard Ratio and 95% CI. 3 Relative Risk and 95% CI. 4 Odds Ratio and 95% CI of reversion of abdominal obesity and yogurt intake. As for the yogurt subgroup analysis, three studies have protective effects in whole-fat yogurt [49,54,60], while the other has no effect [40]; low-fat yogurt has one protective effect [54] and one promoting effect [49] respectively, and the other two have no effect [18,40]. In addition, the result of the other one study is that long-term consumption of whole-yogurt would increase the reversion of abdominal obesity, but total/low-fat yogurt have not effect [44].

Yogurt Intake and Anthropometry Change
Eight studies have evaluated the relationship between yogurt intake and anthropometry (BMI, body weight, waist circumference) changes. The results shown in Figure 6 express the anthropometry changes of one year or four years by the consumption of one serving (125g) yogurt or in the highest quintile per day. Two studies have mentioned that total yogurt consumption would not affect the change of BMI, respectively Snijder et al. [45] ( :0.1; 95% CI: -0.04, 0.24) and Trichia et al, [46] ( : -0.09; 95% CI: -0.18, 0).
As shown in Figure 6, four studies have shown that total yogurt consumption can reduce body weight [41,46]. In the study of Wang et al. [61], the annualized body weight change of the high-dose (3.74; 95% CI:3.00, 24.50 servings/week) group was also lower than that of the low-dose (0; 95% CI:0, 0.74 servings/week) group (0.07±0.04 kg vs. 0.16±0.02 kg; P-trend =0.03). Vergnaud et al. [59] have assessed body weight and waist circumference changes in middle-aged French adults who consumed yogurt. The participants were divided into normal-weight/overweight at baseline groups. The weight of people who eat more yogurt (Quartile 4) decreased than the weight of people who eat less (Quartile 1) in normal-weight women at baseline (P-trend =0.04) and overweight men at baseline (P-trend =0.01). We can also see the change in waist circumference in the Figure 6. Three studies have showed that total yoghurt intake reduced waist circumference [43,46], while two also decreased but without statistical difference [44,45]. Wang et al. [61] have found that the waist circumference annualized increase of the high-dose group was lower than that of the low-dose group (0.57±0.04 cm vs. 0.71±0.02 cm; P-trend =0.008). Another study conducted by Vergnaud et al. [59] has founded high yogurt intake changed less than low-intake only in overweight men at baseline (0.33 ± 0.43 cm vs. 1.62 ± 0.42 cm; P-trend =0.03), and others did not change. When the yogurt was classified, the influence trend of whole-fat ( Figure 7) and low-fat yogurt (Figure 8) on the anthropometry change was not too uniform.

Sensitivity Analysis and Publication Bias
Sensitivity analysis shows that no separate study affects the combined effect. In the funnel plots, no potential publication bias for obesity risk was observed ( Figure 9). Then, we looked for publication bias with Egger [30] and Begg [31] tests and found that no significant publication bias for obesity risk (P value for Begg: 0.54; Egger: 0.87). No publication bias testing has been performed, as there were few studies on overweight and abdominal obesity.

Discussion
This systematic review includes 35 studies, updates the review about the relationship between yogurt intake and obesity risk. As far as we know, only one systematic review [18] adopted ten cohort studies to sum up the association between yogurt intake and the changes in body weight, waist circumference, as well as the risk of overweight and obesity. In that review, Sayón-Orea et al. have concluded that the inverse relationship between yogurt intake and obesity risk was still not clear. Therefore, we collected more research and extracted some of the available data for meta-analysis to further clarify the relationship between yogurt intake and the risk of obesity, overweight and abdominal obesity.
The results showed a clear inverse relationship between yogurt consumption and obesity. The underlying mechanism has been revealed in some studies. Yogurt contains calcium that can reduce fat content in fat cells [62,63], inhibits fat cell hypertrophy [64], trigger mature fat cell apoptosis [65,66], enhance thermogenesis [67], and increase excretion of fecal fat [68,69]. Furthermore, yogurt consumption may increase the proportion of beneficial gut flora by regulating energy absorption and extraction, and these microorganisms are thought to be related to weight maintenance [41,70]. One review mentioned that eating yogurt can improve intestinal health and reduce chronic inflammation by enhancing immune responses, intestinal barrier function, lipid distribution, as well as regulating appetite [71]. In the analysis of Barengolts et al. [72], it has not been shown that the consumption of probiotics compared to traditional yogurt could improve glycemic control in diabetic or obese patients..
In the subgroup analysis of geographical, no relevance for United States populations, which is possibly because that the baseline level of BMI is already overweight [37] or that confounding factors have not been adjusted [34]. Even many Americans are not following the recommendations of the Dietary Guidelines for Americans (DGA) still consuming more than the recommended amount of sugar and saturated fat and lower vegetables, fruit, dairy [73]. Only one study population was younger than 18 years and showed no effect of yogurt on obesity [32]. The OR value was calculated without adjusting for confounding factors. In the subgroup analysis, physical activity, energy intake and alcohol consumption were identified as confounder separately. Only one article was adjusted for race [35], and the results were no different from others.
Several studies have different results on yogurt consumption and obesity or abdominal obesity. Song et al. [52] have explained that yogurt has a specific regulatory effect on the function of the gut, and the degree of effect on obesity and abdominal obesity was different. In the study conducted by Beydoun et al. [48], the reason maybe yogurt intake at a low average consumption may not be reliable like other dairy foods. Lahti-Koski et al. [51] have found that people may start drinking milk and change into a healthy lifestyle after becoming obese.
The other finding showed there were no significant associations observed for the occurrence of overweight. Overweight, like obesity, is defined by BMI. So, the underlying influence mechanism was similar, and might be explained by the heterogeneity associated with yogurt.
The results also showed that yogurt intake is inversely related to the risk of abdominal obesity, which is consistent with the result of Schwingshackl et al. [17]. The underlying mechanism behind this is similar to obesity. Moreover, the meta-analysis results show that yogurt intake does not affect overweight, which is possibly because the number of related studies was not enough and the heterogeneity. After we eliminated an article that reduced heterogeneity, the results showed a negative correlation between yogurt consumption and overweight. When it comes to the relationship between yogurt consumption and obesity or overweight, Martínez-González et al. [57]  Most of the studies included in this review have shown that total yogurt or low-fat yogurt can make some anthropometric indicators healthier [41][42][43][44]46]. Trichia et al. [46] have explained the possible mechanisms that the intestinal flora may reduce the caloric effects of various macronutrients or affect appetite-related gastrointestinal hormones to reducing obesity [74]. Santiago et al. [44] have found that people who eat yogurt regularly are more likely to have a healthy lifestyle, as well as had more calcium intake and probiotics [55]. The difference in the effect of whole-fat yogurt and low-fat yogurt on body weight may be that obese or normal-weight people have different weight control strategies.
Yogurt is a nutrient-rich, easy-to-digest food that contains bioactive proteins, lipids, live microorganisms, and specific amino acids [75,76]. Furthermore, yogurt can be used as a carrier of some probiotics and/or prebiotic compounds to exert health benefits [76,77]. Obesity is a risk factor for many diseases, and a lot of research evidence suggests that yogurt seems to play a protective role in these diseases. Certain nutrients in yogurt, such as proteins, specific lipids, vitamin D, calcium, magnesium, and bioactive nutrients, may have favorable effects on cardiac metabolic diseases risk factors [78,79]. Frequent consumption of yogurt is associated with better diet quality and a healthier insulin situation in American children [80], reduces the intake of high-calorie foods [81], lower the intakes of grains, alcohol [34], and increasing satiety [82]. A cross-sectional study [83] conducted that yogurt intake is associated with better diet quality and metabolic profile among US people. In a randomized controlled trial [84], compared to non-dairy control foods, the biomarkers of chronic inflammation and endotoxin in premenopausal women were reduced after eating low-fat yogurt for 9 weeks. A randomized controlled study has mentioned that the overweight/obese premenopausal women had significantly lower serum levels of lipoproteins after 8 weeks of the intervention of kefir than those in the control group [70]. Yogurt may reduce blood lipids, inflammation, oxidative stress, and LPS to improve insulin resistance in obese women with both nonalcoholic fatty liver disease and metabolic syndrome [85]. Much evidence strongly suggests that eating yogurt can prevent the occurrence of type 2 diabetes [86][87][88].
Our systematic review contains more comprehensive research on prospective observational or cross-sectional studies about yogurt consumption and the risk of obesity. Furthermore, our meta-analysis has three strengths. First, to our knowledge, this is the first study to analyze the consumption of yogurt and the risk of obesity with epidemiological data. Second, we evaluated the evaluation of dose-response relationship for obesity risk. Third, our results cover a wide range of ages.
The limitations of our study include the following aspects: 1) Most of the data used in our meta-analysis came from cross-sectional studies. 2) Insufficient data are available for subgroup analysis of yogurt types.

Conclusions
Overall, the result of our meta-analysis shows that total yogurt intake could reduce obesity risk in adults and daily consumption 165g yogurt can reduce the risk of overall obesity approximately 44% among all people. Yogurt intake might reduce the occurrence of abdominal obesity and the anthropometric variables such as body weight and waist circumference. However, yogurt consumption would not affect overweight incidence. In general, yogurt has a protective effect on the excessive increase of BMI, which is beneficial to human health. In order to clarify the role of different types of yogurt on the risk of obesity there is still a need to identify the more sizeable prospective cohort studies and randomized controlled trials.