Inflammatory potential of diet and risk of cardiovascular disease or mortality: A meta-analysis

Inconsistent findings have reported on the inflammatory potential of diet and cardiovascular disease (CVD) and mortality risk. The aim of this meta-analysis was to investigate the association between the inflammatory potential of diet as estimated by the dietary inflammatory index (DII) score and CVD or mortality risk in the general population. A comprehensive literature search was conducted in PubMed and Embase databases through February 2017. All prospective observational studies assessing the association of inflammatory potential of diet as estimated by the DII score with CVD and all-cause, cancer-related, cardiovascular mortality risk were included. Nine prospective studies enrolling 134,067 subjects were identified. Meta-analyses showed that individuals with the highest category of DII (maximal pro-inflammatory) was associated with increased risk of all-cause mortality (hazard risk [HR] 1.22; 95% confidence interval [CI] 1.06–1.41), cardiovascular mortality (RR 1.24; 95% CI 1.01–1.51), cancer-related mortality (RR 1.28; 95% CI 1.04–1.58), and CVD (RR 1.32; 95% CI 1.09–1.60) than the lowest DII score. More pro-inflammatory diets, as estimated by the higher DII score are independently associated with an increased risk of all-cause, cardiovascular, cancer-related mortality, and CVD in the general population, highlighting low inflammatory potential diet may reduce mortality and CVD risk.

the United States 13,15 , Spain 17,18 , Sweden 12 , France 14,20 , and Australia 19 . The follow-up duration ranged from 1.24 to 20.7 years. All nine studies achieved moderate to high methodological quality with a score from 6-8 stars and the mean NOS stars of the included studies was 7.0.
Publication bias and sensitivity analyses. The small number of studies included in the individual outcome prevented us from conducting the publication bias using the Begg's test and Egger's test. Sensitivity analysis showed that individual study did not significantly impact on estimated overall effect size, indicating the reliability of our pooling results (data not shown).

Discussion
The main finding of this meta-analysis indicates that more pro-inflammatory diets, as estimated by the higher DII score are independently associated with the increased risk of CVD and all-cause, cancer-related, cardiovascular mortality in the general population. Participants with the highest category of DII (maximal pro-inflammatory) led to 32% higher risk of CVD, 22% higher risk of all-cause mortality, 28% higher risk of total cancer-related mortality, and 24% higher risk of cardiovascular mortality. These findings strengthen the idea that there are incremental disadvantages to increase the pro-inflammatory dietary ingredients.
There are several approaches to assess the dietary quality. The Healthy Eating Index 23 , Alternate Healthy Eating Index 24 , and Dietary Approaches to Stop Hypertension Score 25 are the most frequently used diet indexes. All of these healthy dietary patterns were associated with a decreased risk of all-cause mortality 24 . In contrast to above dietary score, DII represents a new dietary quality index that specifically focuses on the dietary inflammatory potential. DII is a literature-derived population-based dietary score to assess a total of 45 food parameters including various macronutrients, vitamins, minerals, flavonoid, and specific food items 11 . A higher DII score is representative of more pro-inflammatory diets. In practice, DII score was computed from dietary intake assessed using a validated food frequency questionnaire or 24-h recall dietary records. Findings from the Energy Balance Study indicated that the DII score was associated with above mentioned dietary indices 26 . Studies that not satisfied the inclusion criteria for this meta-analysis also investigated the association of dietary inflammatory potential with CVD or mortality risk. In the European Prospective Investigation into Cancer and Nutrition study 27 , dietary inflammatory potential was assessed by means of an inflammatory score of the diet (ISD). Subjects with the highest ISD (more pro-inflammatory diet) had an HR of 1.42 for all-cause mortality, 1.89 for cardiovascular mortality, and 1.44 for cancer-related mortality. National Health and Nutrition Examination Survey (NHANES) III study 28 showed that pro-inflammatory diet, as indicated by higher DII score,  Table 1. Characteristics of studies included in meta-analysis. Abbreviations: OR, odds ratio; HR, hazard ratio; CVD, cardiovascular disease; CV, cardiovascular; DM, diabetes mellitus; SBP, systolic blood pressure; DBP, diastolic blood pressure; DII, dietary inflammatory index; FFQ, food frequency questionnaire; HRT, hormone replacement therapy; PA, physical activity. was associated with an increased risk of all-cause, CVD, all-cancer mortality among prediabetic patients. While in the NHANES study 29 , previously diagnosed CVD patients with the highest DII score significantly increased by 30% prevalence odds ratio of combined circulatory disorders, suggesting limiting pro-inflammatory diets may contribute to reduce the recurrence in these patients.
Chronic inflammation has a negative impact on the human body and health. The first step in the prevention and treatment of many chronic diseases is to follow a healthy dietary habit. Our meta-analysis reveals that limiting pro-inflammatory diets may be a healthy eating strategy to reduce CVD and mortality risk. An anti-inflammatory diet has beneficial effects on aging and health by slowing down telomere shortening 21 . In order to adopt a healthier diet, several anti-inflammatory foods including fruits, vegetables, fish or fish oil, walnuts, brown rice, and bulgur wheat should be part of the diet 30 . Moreover, avoid refined or processed foods and cut back on red meat or full-fat dairy foods are recommended.
This meta-analysis had some limitations. First, DII score was collected by self-report from 24-hour dietary recall or food frequency questionnaire, which carried an inherent degree of bias. Second, DII score was estimated at baseline and these data might change during the follow-up duration. However, adult dietary habits seem to relatively stable over time 31,32 . Third, substantial heterogeneity was observed across studies pooling the mortality risk. Potential sources of heterogeneity included differences in the food items considered in the DII score, demographic characteristics, and follow-up duration. Fourth, the wide range of follow-up period (from 1.24 to 20.7  Scientific RepoRts | 7: 6367 | DOI:10.1038/s41598-017-06455-x years) among the included studies was another limitation. However, sensitivity analyses showed no significant impact on the overall risk estimates when we excluded one study 14 with the shortest follow-up duration. Finally, the majority of study participants were of European descent. Therefore, generalization of these findings to other ethnic populations should be taken with caution.
In conclusion, this meta-analysis suggests that more pro-inflammatory diets, as estimated by the DII score are independently associated with the increased risk of CVD and all-cause, cancer-related, cardiovascular mortality in the general population. However, whether promoting dietary patterns with low inflammatory potential could improve health status should be verified by more well-designed randomized controlled trials.

Methods
Search strategy. This meta-analysis followed the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement. A comprehensive literature search was carried out in PubMed and Embase through February 2017 with the following search terms in various combinations: (inflammatory potential of diet OR dietary inflammatory index OR pro-inflammatory diet OR anti-inflammatory diet) AND (cardiovascular disease OR mortality OR death OR survival) AND (prospective OR longitudinal OR follow-up). In order to find additional studies, we also manually searched the reference lists of all relevant articles. Study selection. Studies were included if: (1) study participants were the general population; (2) study was prospective observational design; (3) the exposure of interest was the inflammatory potential of diets as estimated by DII score; and (4) reporting multivariable-adjusted risk estimates for the highest DII score (maximal pro-inflammatory diets) versus the lowest DII score (lowest pro-inflammatory diets) with respect to the all-cause, cardiovascular, cancer-related mortality, or CVD. Exclusion criteria were: (1) participants had a specific disease; (2) DII score as a continuous variable; (3) other inflammatory score of the diets as exposure; and (4) cross-sectional study, review or conference abstract.
Data extraction and quality assessment. For each included study, information on first author's surname, publication year, study location, study design, sample sizes, percentage of women, mean age or age range, dietary assessment method, DII score comparison, number of events, most fully adjusted risk estimate, follow-up duration, and adjustment for variables. The methodological quality of included studies was assessed by a nine-star Newcastle-Ottawa Scale (NOS) for cohort studies 33 . The NOS is categorized into selection, comparability, and ascertaining of outcome. Studies with ≥7 stars were defined as high quality. Data extraction and quality assessment were performed independently by two authors. Any discrepancies between two authors were resolved by consensus.
Statistical analyses. All the analyses were conducted by the STATA software version 12.0 (Stata Corporation, College Station, Texas, USA). The multivariable-adjusted hazard risk ratio (HR) or odds ratio (OR) with 95% confidence interval (CI) was pooled for the highest versus the lowest DII score. The heterogeneity across studies was explored using the Cochrane Q and I 2 statistic. If the Cochrane Q test ≤0.10 or I 2 > 50%, the heterogeneity was considered as statistically significant 34 . Thus, a random effect model was selected to compute the summary effect; otherwise, a fixed-effect model was applied. Publication bias assessment with the Begg's test 35 and Egger's test 36 was planned when more than 10 studies were retrieved 37 . Sensitivity analysis was conducted by sequentially removed one study at a time.