Chronic inflammatory diseases are the leading cause of death worldwide(Reference Pahwa, Goyal and Jialal1). Further, more than 50% of deaths can be attributed to inflammation-related diseases such as type 2 diabetes mellitus (T2DM), stroke, ischaemic heart disease and non-alcoholic fatty liver disease (NAFLD)(Reference Furman, Campisi and Verdin2). Inflammation is a necessary, protective immune response often triggered by cell damage, pathogens and toxins(Reference Chen, Deng and Cui3). However, repeated exposure to cellular insults and tissue damage, without adequate resolution, can cause sustained release of pro-inflammatory cytokines, which have harmful effects including alterations in tissues and organs, and impaired cellular physiology and immune function(Reference Namazi, Larijani and Azadbakht4). Additionally, oxidative stress, which is characterised by an imbalance in oxidant and antioxidants levels, can damage critical cellular components and lead to chronic inflammation(Reference Hussain, Tan and Yin5,Reference Moylan and Reid6) . Chronic inflammation and oxidative stress are involved in the pathogenesis of several cardiometabolic diseases, both of which can be moderated through diet(Reference Biswas7).
The role of individual nutrients and foods in modulating levels of inflammation and oxidative stress is well-established(Reference Ricker and Haas8). Research indicates that foods rich in refined, high glycaemic-load carbohydrates and trans-fatty acids have pro-inflammatory effects; while unprocessed foods high in polyphenols contain protective anti-inflammatory properties(Reference Ricker and Haas8). However, increasingly studies have recognised that individual foods and nutrients are rarely consumed in isolation and a synergy exists where the health effects of overall dietary patterns may be more important than the individual foods and nutrients consumed(Reference Roberts, Cade and Dawson9). As such, there is a trend in research towards exploring overall dietary patterns(Reference Barbaresko, Koch and Schulze10).
Consequently, several dietary indices have been developed to characterise overall dietary patterns as pro-inflammatory or pro-oxidant, or anti-inflammatory or antioxidant, and to predict circulating biomarker levels of inflammation and oxidative stress(Reference Barbaresko, Koch and Schulze10). These indices can be useful in predicting the impact a diet with a higher inflammatory potential has on the risk of mortality(Reference Li, Chen and Zhang11). In addition, identifying the inflammatory or oxidation potential of an overall diet is valuable as it provides insight into the collective synergistic or antagonistic effect of biochemical interactions, which can be translated into clinical practice(Reference Barbaresko, Koch and Schulze10).
Previous reviews have predominantly focused on the Dietary Inflammatory Index (DII) rather than capturing the full range of dietary indices and scores. The current version of the DII was developed in 2014 and was validated based on its ability to predict elevated C-reactive protein (CRP) levels in a healthy adult population(Reference Shivappa, Steck and Hurley12). In 2019, the DII was reported to have been used in > 450 papers, making it the most widely used tool of its kind(Reference Marx, Veronese and Kelly13). A recent systematic review and meta-analysis examined the association between DII and elevated CRP, including 14 studies; a single unit increase in DII score was associated with 10% increased odds of higher CRP(Reference Mohammadi, Hosseinikia and Ghaffarian-Bahraman14). This review focussed only on CRP, and the included studies were predominantly (11 out of 14) in healthy people, rather than populations that might present clinically for dietary advice in health services and who may be prescribed medications to manage various disease risk factors.
With regard to cardiometabolic disease populations, in which dietary therapy has the potential to impact associated chronic low-grade inflammation and oxidative stress, numerous observational studies have explored the relationship between dietary inflammatory indices and incident disease risk. For example, recent meta-analyses have been published on the DII and risk of each of CVD, T2DM, chronic kidney disease (CKD) and NAFLD, as well as cardiometabolic diseases and mortality as collective outcomes(Reference Ji, Hong and Chen15–Reference Aslani, Sadeghi and Heidari-Beni19). These studies have reported mixed results, and in some meta-analyses, there is significant heterogeneity among the studies(Reference Namazi, Larijani and Azadbakht4,Reference Ji, Hong and Chen15,Reference Aslani, Sadeghi and Heidari-Beni19–Reference Kenđel Jovanović, Pavičić Žeželj and Klobučar Majanović24) . These evidence syntheses do not report on whether the DII was able to predict levels of inflammatory biomarkers in people who had developed these cardiometabolic conditions. The validity of the DII and other dietary inflammation and oxidative stress indices to accurately predict the inflammatory or oxidation potential of dietary intake in these populations therefore remains unclear(Reference Mohammadi, Hosseinikia and Ghaffarian-Bahraman14).
A synthesis and appraisal of studies that have investigated the association between dietary inflammatory or oxidative stress indices and various inflammatory or oxidative stress biomarkers in populations with established cardiometabolic disease is needed. Therefore, this systematic review aims to examine the association between all existing dietary inflammatory or oxidative stress indices and measured biomarkers of inflammation or oxidative stress in adults with cardiometabolic conditions. The findings of this review could inform future studies with regard to the selection of indices which are validated to predict dietary inflammation or oxidative stress potential based on particular biomarkers of interest, and could assist clinicians in the assessment of dietary intake or targeting dietary changes to impact inflammation or oxidative stress in specific cardiometabolic disease populations.
Methods
A systematic literature review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines (online Supplementary files)(Reference Page, McKenzie and Bossuyt25). The review protocol was registered with the International Prospective Register for Systematic Reviews on March 1, 2021 (PROSPERO number: CRD42021223186).
Data sources
Four electronic databases (Embase, CINAHL, MEDLINE and CENTRAL) were searched for published literature in any language from database inception to October 28, 2020, by two researchers (CP, CD). The advice of a specialist search librarian was sought, and the search was updated with a revised search strategy from October 28, 2020, to June 28, 2023, by one researcher (AI). The search strategies were developed in PubMed and translated for use in other databases (online Supplementary Table 1) using the Systematic Review Accelerator Polyglot Search tool(Reference Clark, Sanders and Carter26). A forward citation search of all the included studies (n 16) was performed using SpiderCite, a forward and backward citation searching tool(27), from June 2023 (the date of the most recent search) until September 2024.
Study eligibility
Empirical studies of adults diagnosed with at least one cardiometabolic condition reporting on the association between any dietary inflammatory or oxidative stress index or score, and biomarker/s of inflammation or oxidative stress were included. Table 1 summarises the eligibility criteria and was designed using the PICOS format to identify studies. A dietary inflammatory or oxidative stress index was defined as any tool that used dietary intake data to calculate an overall score reflecting the inflammatory or oxidative stress status of an individual’s diet. Eligible studies reported an analysis of the association between the dietary index used and at least one biomarker of inflammation or oxidative stress. Examples of relevant biomarkers of inflammation or oxidative stress included, but were not limited to, CRP, tumour necrosis factor-α (TNF-α), interleukin-6 (IL-6), IL-8, IL-10, IL-18, fibrinogen, adiponectin, e-selectin, platelet-activating factor, lipoprotein-associated phospholipase A2, reactive oxygen species, pentraxin3 and reactive nitrogen species.
Eligibility criteria according to the PICOS format
CKD, chronic kidney disease; DASH, dietary approaches to stop hypertension; MetS, metabolic syndrome; NAFLD, non-alcoholic fatty liver disease; T2DM, type 2 diabetes mellitus.
Study selection
All identified records were imported into reference management software Endnote X9 and deduplicated(28). The remaining records were screened using Covidence systematic review software (version 5.3)(29). Subsequent title, abstract and full-text screening were conducted in duplicate by two pairs of researchers (CP and CD; KOJ and AI)(29). All citing articles sourced through SpiderCite underwent title and abstract screening by two authors (CE and KOJ), and full texts were reviewed in duplicate. Disagreements were resolved through discussion with a third author (HM, CE or DR).
Data extraction
Data were independently extracted by two authors (CP or AI, and CE) and checked for accuracy by at least one other author (CD, DR or KOJ). Study characteristics extracted were: first author, year and country of publication, study design, setting, population size, sex, inclusion criteria including primary disease state, age reported as a mean and range, diet assessed, method of dietary intake assessment, the dietary inflammatory or oxidative stress index used, the number of associated dietary parameters assessed, and the inflammatory or oxidative stress biomarkers assessed. Outcome data reporting the association between the dietary indices and biomarkers of interest were extracted as reported including odds ratios (OR), adjusted means, beta estimates with 95% confidence intervals (CI), mean or median and standard deviation or interquartile range, and Pearson’s or Spearman’s coefficient. The authors of included studies were contacted for missing data at least twice by email.
Study quality
Included studies were critically appraised independently in pairs between four researchers (CP and CD; AI and KOJ) using the Academy of Nutrition and Dietetics Quality Criteria Checklist for primary research(30). Discrepancies were discussed with a third researcher (HM, CE or DR) and managed by consensus. The Quality Criteria Checklist is a validated critical appraisal tool comprising four relevance questions that explore applicability to practice and ten validity questions that evaluate scientific soundness(30). The Quality Criteria Checklist categorises the overall study quality as positive, neutral or negative and reflects bias across recruitment, generalisability, intervention design, data collection and analysis(30).
Data synthesis
A narrative synthesis was considered likely at the time of protocol registration, due to expected limited homogeneity across included studies with regard to population, dietary indices and outcomes. Following data extraction, as pre-specified in the protocol, a meta-analysis was explored. Heterogeneity of indices and outcomes reported, as well as the presentation of the results data (not reported as outcome data that could be extracted; not reported as consistently continuous or categorical data) across the included studies that included populations with similar diagnoses was confirmed. It was deemed inappropriate to quantitatively synthesise such heterogeneous findings even when the same score or index appeared to be reported with the same inflammatory biomarker in populations with similar diagnoses. Results are therefore presented as a narrative synthesis.
Results
Search results and study quality
Of the 18,966 records identified, 6680 duplicates were removed (Figure 1). Following screening, 172 full-text articles were assessed for eligibility, and 17 studies were included in this review.
PRISMA flow diagram of the searching and selection process.
Study characteristics
Characteristics of the 17 included studies are summarised in Table 2. All studies were published between 2014 to 2023. Four studies were conducted in America(Reference Banerjee, McCulloch and Crews31–Reference Marks, Hartman and Judd34), six in Iran(Reference Arab, Karimi and Nazari35–Reference Farhangi and Najafi40), three in Turkey(Reference Kizil, Tengilimoglu-Metin and Gumus41–Reference Demirer, Yardımcı and Erem Basmaz43), two in China(Reference Ren, Zhao and Wang44,Reference Zeng, Lin and He45) and two in Australia(Reference Mayr, Itsiopoulos and Tierney46,Reference Nikbakht, Singh and Vider47) . One prospective parallel design randomised controlled trial (RCT)(Reference Mayr, Itsiopoulos and Tierney46), one open-label clinical trial(Reference Nikbakht, Singh and Vider47), two case-control studies(Reference Moradi, Heidari and Teimori37,Reference Toprak, Görpelioğlu and Özsoy42) , two cohort studies(Reference Banerjee, McCulloch and Crews31,Reference Huang, Zhang and Zeng33) and 11 cross-sectional studies were included(Reference Huang, Zeng and Zhang32,Reference Marks, Hartman and Judd34–Reference Farhangi and Najafi36,Reference Abaj, Rafiee and Koohdani38–Reference Kizil, Tengilimoglu-Metin and Gumus41,Reference Demirer, Yardımcı and Erem Basmaz43–Reference Zeng, Lin and He45) . The studies included participants diagnosed with metabolic syndrome (MetS)(Reference Ren, Zhao and Wang44,Reference Nikbakht, Singh and Vider47) , T2DM(Reference Abaj, Rafiee and Koohdani38,Reference Toprak, Görpelioğlu and Özsoy42,Reference Demirer, Yardımcı and Erem Basmaz43,Reference Nikbakht, Singh and Vider47) , CVD(Reference Farhangi and Najafi36,Reference Mohammadi, Abdollahzad and Rezaeian39,Reference Farhangi and Najafi40,Reference Demirer, Yardımcı and Erem Basmaz43,Reference Mayr, Itsiopoulos and Tierney46) , CKD(Reference Banerjee, McCulloch and Crews31–Reference Arab, Karimi and Nazari35,Reference Kizil, Tengilimoglu-Metin and Gumus41,Reference Zeng, Lin and He45) and NAFLD(Reference Moradi, Heidari and Teimori37). The number of study participants ranged from 65 to 7207 and the mean age ranged from 36·5 to 75·6 years. Dietary intake data was collected using a 7-day food diary(Reference Mayr, Itsiopoulos and Tierney46), a 3-day dietary recall(Reference Kizil, Tengilimoglu-Metin and Gumus41), a 3-day food diary(Reference Demirer, Yardımcı and Erem Basmaz43,Reference Nikbakht, Singh and Vider47) , a 24-hour dietary recall(Reference Banerjee, McCulloch and Crews31–Reference Huang, Zhang and Zeng33,Reference Arab, Karimi and Nazari35,Reference Ren, Zhao and Wang44) or food frequency questionnaires (FFQ)(Reference Marks, Hartman and Judd34,Reference Farhangi and Najafi36–Reference Farhangi and Najafi40,Reference Toprak, Görpelioğlu and Özsoy42,Reference Zeng, Lin and He45) . Eight authors were contacted for missing data(Reference Marks, Hartman and Judd34,Reference Farhangi and Najafi36,Reference Farhangi and Najafi40,Reference Nikbakht, Singh and Vider47–Reference Tian, Zhang and Xie51) ; two responded with both studies included in this review(Reference Marks, Hartman and Judd34,Reference Nikbakht, Singh and Vider47) . Seven (41%) of the 17 included studies applied the dietary inflammatory or oxidative stress tools retrospectively to previously collected dietary intake data(Reference Huang, Zeng and Zhang32–Reference Marks, Hartman and Judd34,Reference Farhangi and Najafi36,Reference Farhangi and Najafi40,Reference Kizil, Tengilimoglu-Metin and Gumus41,Reference Ren, Zhao and Wang44) .
Study characteristics of the included studies (n 17)
ADII, Adapted Dietary Inflammatory Index; CABG, coronary artery bypass graft surgery; CKD, chronic kidney disease; E-DII, energy-adjusted DII; HD, haemodialysis; MetS, metabolic syndrome; NAFLD, non-alcoholic fatty liver disease; NHANES, National Health and Nutrition Examination Survey; RCT, randomised controlled trial.
Dietary inflammatory index
The DII is an a priori literature-based index comprised of 45 individual nutrient, food or flavonoid parameters that were identified to be associated with at least one of six inflammatory biomarkers; IL beta-1, IL-4, IL-6, IL-10, TNF-α and CRP(Reference Shivappa, Steck and Hurley52). A lower DII score reflects a more anti-inflammatory diet and is more favourable, while a higher score reflects a more pro-inflammatory diet(Reference Shivappa, Steck and Hurley52). The development(Reference Shivappa, Steck and Hurley52) and validation(Reference Kotemori, Sawada and Iwasaki53) of the DII have been previously reported. Of the 17 included studies, 12 assessed the DII: one utilised an adapted DII(Reference Banerjee, McCulloch and Crews31), two utilised an energy-adjusted dietary inflammatory index (E-DII)(Reference Huang, Zhang and Zeng33,Reference Arab, Karimi and Nazari35) , and the remaining nine studies used the DII.(Reference Huang, Zeng and Zhang32,Reference Farhangi and Najafi36,Reference Moradi, Heidari and Teimori37,Reference Kizil, Tengilimoglu-Metin and Gumus41,Reference Toprak, Görpelioğlu and Özsoy42,Reference Ren, Zhao and Wang44–Reference Nikbakht, Singh and Vider47) Six studies included adults diagnosed with CKD(Reference Banerjee, McCulloch and Crews31–Reference Huang, Zhang and Zeng33,Reference Arab, Karimi and Nazari35,Reference Kizil, Tengilimoglu-Metin and Gumus41,Reference Zeng, Lin and He45) , two included adults with CVD(Reference Farhangi and Najafi36,Reference Mayr, Itsiopoulos and Tierney46) , one included people with T2DM(Reference Toprak, Görpelioğlu and Özsoy42), one included people diagnosed with T2DM and people with MetS(Reference Nikbakht, Singh and Vider47), one included adults with MetS(Reference Ren, Zhao and Wang44) and one included adults with NAFLD(Reference Moradi, Heidari and Teimori37). For calculations of the DII, one study assessed all 45 parameters(Reference Mayr, Itsiopoulos and Tierney46), one study assessed 44 parameters(Reference Toprak, Görpelioğlu and Özsoy42), 10 studies assessed between 21 and 28 parameters(Reference Banerjee, McCulloch and Crews31–Reference Huang, Zhang and Zeng33,Reference Arab, Karimi and Nazari35–Reference Moradi, Heidari and Teimori37,Reference Kizil, Tengilimoglu-Metin and Gumus41,Reference Ren, Zhao and Wang44,Reference Zeng, Lin and He45,Reference Nikbakht, Singh and Vider47) .
Other dietary indices
The oxidative balance score (OBS) is a tool based on 13 a priori defined pro- and antioxidant dietary factors including alcohol and nutrients such as vitamin E and C; a lower score reflects a pro-oxidant diet while a higher score is more favourable as it indicates a more antioxidant diet(Reference Ilori, Sun Ro and Kong54). One study used the OBS in people with CKD; 11 of the 14 score parameters were assessed(Reference Marks, Hartman and Judd34). Another study used the OBS in people with T2DM but did not report the number of parameters assessed(Reference Demirer, Yardımcı and Erem Basmaz43).
The dietary total antioxidant capacity (DTAC) indicates the effect of the cumulative total antioxidants present in the diet which remove free radicals, prevent cell damage, and are associated with levels of CRP or hs-CRP(Reference Mozaffari, Daneshzad and Surkan55). The total antioxidant capacity is calculated using the ferric reducing ability of plasma (FRAP), oxygen radical absorbance capacity, total radical-trapping antioxidant parameter (TRAP), Trolox equivalent antioxidant capacity and vitamin C equivalent antioxidant capacity (VCEAC)(Reference Mozaffari, Daneshzad and Surkan55). A lower score reflects a pro-oxidant diet while a higher score is preferable as it indicates a more antioxidant diet(Reference Mozaffari, Daneshzad and Surkan55). One study used the DTAC in people with CVD; it does not report the parameters that were assessed(Reference Mohammadi, Abdollahzad and Rezaeian39). Another study used DTAC in people with T2DM; with all four parameters assessed(Reference Abaj, Rafiee and Koohdani38).
The empirically developed dietary inflammatory pattern (EDIP) uses an empirical hypothesis-oriented approach and is based on 18 food group-based parameters that are predictive of three inflammatory biomarkers: CRP, IL-6 and TNF-α receptor type 2(Reference Tabung, Smith-Warner and Chavarro56). One study explored the EDIP in CVD populations; 15 of the 18 food group-based parameters were measured(Reference Farhangi and Najafi40).
Outcomes
The results of the included studies relevant to the biomarker outcomes are summarised in Table 3. Seventeen inflammatory biomarkers were reported across the studies. No biomarkers of oxidative stress were reported.
Outcomes of the association between dietary indices and inflammatory biomarkers reported in the included studies (n 17)
DII, Dietary Inflammatory Index; CKD, chronic kidney disease; HD, haemodialysis; E-DII, energy-adjusted DII; CRP, C-reactive protein; CABG, coronary artery bypass graft surgery; Cys-C, cystatin C; dAGE, dietary advanced glycation end products; DTAC, dietary total antioxidant capacity; EDIP, empirically developed dietary inflammatory potential; FRAP, ferric reducing ability of plasma; Hs−, high sensitivity; Int, interval; MDA, malondialdehyde; NSAIDS, non-steroidal anti-inflammatory drugs; Q, quartile/quintile; SES, socio-economic status; sICAM, soluble intercellular adhesion molecule-1; sVCAM-1, soluble vascular cell adhesion molecule-1; T, tertile; WBC, white blood cell. Statistically significant p-values have been bolded.
C-reactive protein
CRP was assessed in 15 studies. Three studies analysed the association between a dietary index and high-sensitivity (hs)-CRP in CVD populations(Reference Farhangi and Najafi36,Reference Farhangi and Najafi40,Reference Mayr, Itsiopoulos and Tierney46) . The first study reported on individuals with CHD in a randomised dietary trial and replicated the validation method used for the original DII. At baseline, they reported that DII score had a nonsignificant association with an increased odds of elevated hs-CRP (i.e. >3 mg/L). Using pooled data, they also distributed participants that completed the healthy diet interventions into tertiles of change in DII score from baseline to 6 months(Reference Mayr, Itsiopoulos and Tierney46). The adjusted mean hs-CRP values at 6 months were the highest in tertile 2 and lowest in tertile 3. This represented a significant difference across the tertiles (P = 0·004) but no clear association between DII change and hs-CRP levels was observed. The second study reported that male coronary artery bypass grafting surgery candidates in the third DII quartile had significantly higher hs-CRP values compared with those in the second quartile (P < 0·05)(Reference Farhangi and Najafi36). However, no statistically significant trends were reported across the DII quartiles for males or females(Reference Farhangi and Najafi36). No statistically significant findings were reported in the third study, which examined the association between hs-CRP and the EDIP in coronary artery bypass grafting surgery candidates(Reference Farhangi and Najafi40).
One study examined the association between CRP and the DII in a MetS population and found that in adjusted models, CRP beta-estimates in tertiles 2 and 3 were significantly higher compared with tertile 1 (P < 0·05)(Reference Ren, Zhao and Wang44). These results highlighted that as DII scores increased, CRP increased. Similarly, a significant positive association was identified between CRP and the DII when the continuous form of DII was used (P < 0·05)(Reference Ren, Zhao and Wang44).
Seven studies explored the association between CRP and dietary indices in CKD. One study reported a medium positive correlation between the DII and CRP as continuous variables (r 0·35, P < 0·001). The same paper also reported that CRP values significantly increased across tertiles of DII score after adjustments were made for gender, marital status and education level (P = 0·001)(Reference Kizil, Tengilimoglu-Metin and Gumus41). Three studies reported a statistically significant association between DII score and CRP in CKD (P < 0·001)(Reference Huang, Zeng and Zhang32,Reference Huang, Zhang and Zeng33,Reference Zeng, Lin and He45) . These studies reported that CRP values increased across tertiles of DII score(Reference Huang, Zeng and Zhang32,Reference Huang, Zhang and Zeng33,Reference Zeng, Lin and He45) . One study analysed the association between the OBS and CRP in people with CKD and reported no statistically significant associations(Reference Marks, Hartman and Judd34). A single study that analysed the association between the adapted DII score and CRP in people undergoing kidney failure replacement therapy, such as dialysis or a kidney transplant, reported no statistically significant associations(Reference Banerjee, McCulloch and Crews31). Finally, a study assessing the association between the E-DII and CRP in haemodialysis patients reported no statistically significant difference in CRP between E-DII tertiles(Reference Arab, Karimi and Nazari35).
Four studies examined the association between CRP and T2DM. One study reported a large negative correlation between OBS and CRP in individuals with T2DM(Reference Demirer, Yardımcı and Erem Basmaz43). One study reported a medium positive correlation between DII and CRP in adults with type 2 diabetes(Reference Toprak, Görpelioğlu and Özsoy42). Two other studies did not report any significant results between DTAC or DII and CRP(Reference Abaj, Rafiee and Koohdani38,Reference Nikbakht, Singh and Vider47) .
IL-6
IL-6 was assessed in four studies. One dietary RCT study in individuals with CHD reported that the adjusted mean hs-IL-6 values at 6 months post-diet intervention were higher in tertiles 2 and 3 compared with tertile 1 of DII change(Reference Mayr, Itsiopoulos and Tierney46). This represented a significant trend across the tertiles (P = 0·006) and indicated that a greater prospective reduction in DII scores was associated with lower hs-IL-6 levels(Reference Mayr, Itsiopoulos and Tierney46). The same study identified a medium positive correlation between a reduction in DII scores and lower IL-6 values (r 0·34 (95% CI: 0·05, 0·56)). One case–control study in individuals with insulin-dependent T2DM reported a positive correlation between DII and IL-6 (r 0·375 (P = 0·017))(Reference Toprak, Görpelioğlu and Özsoy42). One study examined the association between IL-6 and dietary indices in CKD and MetS populations but reported no statistically significant findings(Reference Marks, Hartman and Judd34). Further, one study examined the association between DII and IL-6 in people with T2DM or MetS, reporting no significant findings(Reference Nikbakht, Singh and Vider47).
Other biomarkers
Fourteen other inflammatory biomarkers were assessed once across the included studies. One cross-sectional study reported no statistically significant association between the OBS and fibrinogen, white blood cell count, cystatin C or IL-10 in participants with CKD(Reference Marks, Hartman and Judd34). The same study identified a positive association between OBS and IL-8, with IL-8 values found to be 14·7% higher for every five-unit increase in OBS (P = 0·04)(Reference Marks, Hartman and Judd34). This positive trend was consistent across intervals of OBS exposure (P = 0·05)(Reference Marks, Hartman and Judd34). Another cross-sectional study of adults with T2DM reported no statistically significant association between DTAC groups (FRAP, Trolox equivalent antioxidant capacity, oxygen radical absorbance capacity) and IL-18; however, there was an association with the DTAC group TRAP and IL-18, with participants with a higher intake of TRAP reporting higher IL-18 concentrations (P = 0·03)(Reference Abaj, Rafiee and Koohdani38). The same study found no associations between any of the DTAC groups assessed with pentrexin3(Reference Abaj, Rafiee and Koohdani38).
A study examining the association between DTAC tertiles and intercellular and vascular cell adhesion molecules in adults with CVD reported a significant difference, with tertile three reporting lower values of intercellular and vascular cell adhesion molecules compared to tertiles one and two(Reference Mohammadi, Abdollahzad and Rezaeian39). The same study reported no statistically significant association between DTAC tertiles and IL-17(Reference Mohammadi, Abdollahzad and Rezaeian39). A separate study reported no association between DII tertiles and malondialdehyde in NAFLD participants(Reference Moradi, Heidari and Teimori37). A further study reported no association between DII and IL-8, IL-18 or IL-1Rα or TNFα in either adults with MetS or those diagnosed with T2DM(Reference Nikbakht, Singh and Vider47).
The RCT in people with CHD also reported on adiponectin, an anti-inflammatory biomarker, and found no statistically significant associations with the DII(Reference Mayr, Itsiopoulos and Tierney46). One study assessed the association between the E-DII and soluble vascular cell adhesion molecule-1, soluble intercellular adhesion molecule-1, soluble endothelial leukocyte adhesion molecule-1 (sE-selectin) and malondialdehyde in haemodialysis patients(Reference Arab, Karimi and Nazari35). The study reported that participants of the highest tertile of E-DII compared to the lowest tertile had significantly higher soluble vascular cell adhesion molecule-1(Reference Arab, Karimi and Nazari35). In regression analysis, there was a statistically significant trend across the E-DII tertiles in all models for soluble vascular cell adhesion molecule-1 and models 2 and 3 for sE-selectin(Reference Arab, Karimi and Nazari35). There were no statistically significant associations between tertiles of the E-DII and soluble intercellular adhesion molecule-1 or malondialdehyde(Reference Arab, Karimi and Nazari35).
Study quality
Three papers were rated ‘positive’ for study quality(Reference Arab, Karimi and Nazari35,Reference Mayr, Itsiopoulos and Tierney46,Reference Nikbakht, Singh and Vider47) , while the remainder (14 studies) were rated ‘neutral’ (Table 4)(Reference Banerjee, McCulloch and Crews31–Reference Marks, Hartman and Judd34,Reference Farhangi and Najafi36–Reference Zeng, Lin and He45) . Papers were commonly downgraded to neutral due to insufficient detail regarding potential bias, funding sources, withdrawals, or study limitations.
Risk-of-Bias assessment using Academy of Nutrition and Dietetics Quality Criteria Checklist (n 17)(Reference Clark, Sanders and Carter26)
Footnote: Green (+) = Yes; Yellow (?) = Unclear; Grey (no symbol) = N/A; Red (-) = No.
a Relevance questions (n 4):
1. Would implementing the studied intervention or procedure (if found successful) result in improved outcomes for the patients/clients/population group?
2. Did the authors study an outcome (dependent variable) or topic that the patients/clients/population group would care about?
3. Is the focus of the intervention or procedure (independent variable) or topic of study a common issue of concern to dietetics practice?
4. Is the intervention or procedure feasible?
b Validity questions (n 10):
1. Was the research question clearly stated?
2. Was the selection of study subjects/patients free from bias?
3. Were study groups comparable?
4. Was method of handling withdrawals described?
5. Was blinding used to prevent introduction of bias?
6. Were intervention/therapeutic regimens/exposure factor or procedure and any comparison(s) described in detail? Were intervening factors described?
7. Were outcomes clearly defined and the measurements valid and reliable?
8. Was the statistical analysis appropriate for the study design and type of outcome indicators?
9. Are conclusions supported by results with biases and limitations taken into consideration?
10. Is bias due to study’s funding or sponsorship unlikely?
Discussion
This review examined the association between dietary inflammatory or oxidative stress indices and inflammatory or oxidative stress biomarkers in adults with cardiometabolic conditions to inform the likely validity of these indices in these population groups. Seventeen studies were included, which is surprising given the high volume of literature noted particularly for the DII over the last decade(Reference Marx, Veronese and Kelly13). Twelve included studies reported on the DII, two studies used DTAC, two studies used the OBS and one study used EDIP to estimate the inflammatory or oxidative stress potential of dietary intake. Studies reported only on associations between the dietary indices and inflammatory biomarkers and not specific biomarkers of oxidative stress. There were no consistent findings across studies investigating associations for a range of cardiometabolic conditions. There were only three studies: a prospective RCT design(Reference Mayr, Itsiopoulos and Tierney46), an open-label intervention(Reference Nikbakht, Singh and Vider47) and a prospective, cross-sectional study(Reference Arab, Karimi and Nazari35) that were categorised as positive in study quality, with the remaining studies rated as neutral, indicating a paucity of rigorous research. Furthermore, 15 of the included studies were observational studies, seven of which were retrospective. Observational studies, particularly those with retrospective designs, are subject to recall and selection biases and may present overestimated results.
Of the 12 studies exploring the DII, no clear association was identified between inflammatory biomarkers and the index. Nine studies reported a statistically significant association between CRP and the DII(Reference Huang, Zeng and Zhang32,Reference Huang, Zhang and Zeng33,Reference Farhangi and Najafi36,Reference Kizil, Tengilimoglu-Metin and Gumus41,Reference Toprak, Görpelioğlu and Özsoy42,Reference Ren, Zhao and Wang44–Reference Mayr, Itsiopoulos and Tierney46) and of these, two included incorrect DII score calculations(Reference Farhangi and Najafi36,Reference Kizil, Tengilimoglu-Metin and Gumus41) which made it difficult to draw meaningful conclusions(Reference Hébert, Shivappa and Wirth57). One study reported an analysis between adiponectin and the DII and found no significant association between the index and adiponectin(Reference Mayr, Itsiopoulos and Tierney46). Adiponectin is an anti-inflammatory adipocytokine predominantly secreted by adipose tissue and notably was not one of the six biomarkers included in the development of the DII(Reference Shivappa, Steck and Hurley52) making conclusions difficult. Another study found that the E-DII was a predictor of higher soluble vascular cell adhesion molecule-1 (sVCAM-1) and sE-selectin(Reference Arab, Karimi and Nazari35).
In addition to the DII, three other dietary indices were identified in the review; however, their use and validity were limited. The EDIP tool, based exclusively on 18 food-group parameters,(Reference Tabung, Smith-Warner and Chavarro56) was used in one study(Reference Farhangi and Najafi40). Notably, three parameters, wine, beer and low-energy beverages, were excluded from analysis as consumption of those items was deemed uncommon in Iranian populations(Reference Farhangi and Najafi40). The OBS was used in one study and was modified from the precursor study to exclude two non-diet index parameters, smoking and nonsteroidal anti-inflammatory drugs(Reference Marks, Hartman and Judd34).
The use of the tools within the included studies varied. Dietary intake data required to calculate DII scores(Reference Shivappa, Steck and Hurley52) encompassed five different dietary assessment methods across the 11 included DII studies; a 24-hour dietary recall(Reference Banerjee, McCulloch and Crews31–Reference Huang, Zhang and Zeng33,Reference Arab, Karimi and Nazari35,Reference Ren, Zhao and Wang44) , 3-day dietary recall(Reference Kizil, Tengilimoglu-Metin and Gumus41), 3-day diary(Reference Nikbakht, Singh and Vider47), 7-day food diary(Reference Mayr, Itsiopoulos and Tierney46) and FFQ(Reference Farhangi and Najafi36,Reference Moradi, Heidari and Teimori37,Reference Toprak, Görpelioğlu and Özsoy42,Reference Zeng, Lin and He45) . Although the DII was developed to be used with any dietary assessment method, 24-hour dietary recall or food records were recommended(Reference Shivappa, Steck and Hurley52) to allow for the inclusion of all 45 parameters for calculation of the index. No statistical analyses have been conducted to investigate sensitivity of the DII where less than 45 parameters are used(Reference Shivappa, Steck and Hurley52). Within the studies that used the DII included in the current review, the number of parameters used to calculate the index ranged from 21 to 45, with only one study(Reference Mayr, Itsiopoulos and Tierney46) using all 45 parameters. It is difficult to draw conclusions about associations in studies using less than the full 45 parameters, particularly as many omitted specific food items (e.g. tea, ginger, garlic, herbs and spices) or other components (e.g. flavonoids, caffeine and alcohol) which have anti-inflammatory effect scores. Indeed, a previous review suggested dietary intake assessment tools should include a forgotten foods checklist to prevent overlooking relevant DII food components(Reference Vicente, dos Santos Lucio Quaresma and de Maria Melo58).
The method used to calculate DII scores also requires careful consideration. The DII score calculations are complex, and therefore, the original DII authors often provide technical assistance. One of the 11 DII studies in the current review listed two of the original DII development authors as co-authors(Reference Mayr, Itsiopoulos and Tierney46). The remaining 10 studies independently calculated the DII scores; however, two studies(Reference Farhangi and Najafi36,Reference Kizil, Tengilimoglu-Metin and Gumus41) were later criticised for clear miscalculations and the use of incorrect algorithms(Reference Hébert, Shivappa and Wirth59). One of the 10 studies used an adapted DII which excluded some food items to avoid overestimation of their inflammatory effect(Reference Banerjee, McCulloch and Crews31). However, this index has also been subject to criticism by the developers of the original DII score(Reference Hébert, Shivappa and Wirth57). The energy-adjusted DII developed by the original DII authors was also used by two of the 10 studies, one adjusted for both alcohol and energy(Reference Huang, Zhang and Zeng33) and one adjusted for energy only(Reference Arab, Karimi and Nazari35). Whilst the tool itself was rigorously developed, issues related to quality are inevitable if independent researchers are unable to accurately calculate the score. Additionally, as was evident from the results of the current review, tools such as the DII need to be further investigated prospectively in interventional studies to inform whether changes in DII will modulate cardiometabolic disease risk.
Inflammation is influenced by diet; however, other influences are known to impact inflammatory biomarkers. The current review specifically sought to investigate associations in adults with cardiometabolic conditions, who are more likely to be taking medications (for example statins and aspirin) which have downstream anti-inflammatory effects(Reference Mayr, Itsiopoulos and Tierney46). For this group, diet is not likely to have the same association with inflammatory markers compared to healthy adults without these confounding influences. Other influences on inflammatory biomarkers such as physical activity, smoking and mental stress also need to be considered. A study which utilised data from the UK Biobank for 188,433 adults aged 39–72 years found that a low-inflammatory diet is associated with a lower risk of death, noting that a favourable lifestyle beyond diet, including not smoking, regular physical activity and maintaining a body weight consistent with a recommended BMI, may reinforce the protective role of a low-inflammatory diet against death(Reference Da, Yang and Liang60). Of the four indices identified in this review, only the OBS attempts to account for some of these factors including medications and smoking(Reference Ilori, Sun Ro and Kong54), although the single study that used the OBS excluded these factors in their calculation(Reference Marks, Hartman and Judd34).
Strengths and potential limitations
This review was strengthened by the robust and broad search strategies using four databases with no language or date restrictions, purposefully designed to capture any dietary inflammatory or oxidative stress index from inception. Previous reviews have focussed on the DII only and have often overlooked the full range of indices that exist. A limitation of this review was the heterogeneity observed across the studies, which meant meta-analysis was not feasible. Consequently, funnel plots were unable to be generated, and publication bias could not be evaluated.
Implications and future directions
The findings of the current review are inconclusive and may be attributed to the heterogeneity of indices, disease states and biomarkers reported. The included studies reported on five different disease states, including MetS, T2DM, CVD, CKD and NAFLD. Only one study reported on participants with NAFLD(Reference Moradi, Heidari and Teimori37), indicating a research gap in this population group.
Dietary-related inflammation and oxidative stress are rapidly expanding research areas that continue to capture the interest of researchers and the general population alike. When used appropriately, dietary indices are valuable tools that enable researchers to predict specific biomarker levels in a cost effective and timely manner. Additionally, tools for assessing the inflammatory potential of diet, such as the DII, could be useful in clinical practice; however, RCT’s studying the effect of changing DII scores are needed before the DII can be recommended for use in CVD and T2DM(Reference Hariharan, Odjidja and Scott61). Future studies should aim to demonstrate high levels of scientific rigour and be prospectively designed. Further, researchers should aim to use dietary assessment data that allows the full range of parameters to be used in calculating indices. Strengthening the quality of research in this area will allow for meta-analyses to determine whether an association exists between dietary indices and biomarkers of inflammation or oxidative stress. Currently, there is insufficient evidence in cardiometabolic conditions to conclude this, limiting the interpretation of these dietary indices as robust measures of inflammation or oxidation potential in these populations.
Conclusions
The volume of research using dietary indices which intend to measure dietary potential to impact inflammation or oxidative stress is rapidly increasing. However, this systematic review demonstrates that results regarding the validity of these tools to accurately reflect an association with biomarkers of inflammation and oxidative stress in cardiometabolic disease populations are varied. Additional rigorous, prospective research is required to strengthen the evidence base and ensure that the results obtained from dietary indices can be confidently relied upon.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114525000686
Acknowledgements
The authors acknowledge Sarah Bateup, Faculty Librarian, Health Sciences and Medicine at Bond University for her assistance with the search strategy.
Financial support: Nil
Authorship: Study conceptualisation: C. J. E., D. P. R., H. L. M. and K. O-J.; search design, execution of search, screening of search results against eligibility criteria, quality assessment and data extraction: C. G. P., C. D., A. I., K. O-J., D. P. R., C. E. J.; data analysis and interpretation: all authors; drafting the manuscript: C. G. P., A. I., K. O- J., C. J. E; critical review of the manuscript: all authors.
Conflict of interest disclosure: H. L. M. authored one of the studies included in this review, however, was not involved in study selection or data extraction for this review. All other authors declare no actual or potential conflicts of interests.
Open access
