Hostname: page-component-7c8c6479df-xxrs7 Total loading time: 0 Render date: 2024-03-28T16:59:09.572Z Has data issue: false hasContentIssue false

Associations of dietary and lifestyle inflammation scores with mortality due to CVD, cancer, and all causes among Black and White American men and women

Published online by Cambridge University Press:  10 May 2022

Alyssa N. Troeschel
Affiliation:
Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, GA 30322, USA
Doratha A. Byrd
Affiliation:
Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, GA 30322, USA
Suzanne Judd
Affiliation:
Department of Biostatistics, School of Public Health, University of Alabama, Birmingham, AL, USA
W. Dana Flanders
Affiliation:
Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, GA 30322, USA Winship Cancer Institute, Emory University, Atlanta, GA, USA
Roberd M. Bostick*
Affiliation:
Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, GA 30322, USA Winship Cancer Institute, Emory University, Atlanta, GA, USA
*
*Corresponding author: Dr R. M. Bostick, fax +404 727 8737, email rmbosti@emory.edu
Rights & Permissions [Opens in a new window]

Abstract

One potential mechanism by which diet and lifestyle may affect chronic disease risk and subsequent mortality is through chronic systemic inflammation. In this study, we investigated whether the inflammatory potentials of diet and lifestyle, separately and combined, were associated with all-cause, all-CVD and all-cancer mortality risk. We analysed data on 18 484 (of whom 4103 died during follow-up) Black and White men and women aged ≥45 years from the prospective REasons for Geographic and Racial Differences in Stroke study. Using baseline (2003–2007) Block 98 FFQ and lifestyle questionnaire data, we constructed the previously validated inflammation biomarker panel-weighted, 19-component dietary inflammation score (DIS) and 4-component lifestyle inflammation score (LIS) to reflect the overall inflammatory potential of diet and lifestyle. From multivariable Cox proportional hazards models, the hazards ratios (HR) and their 95 % CI for the DIS–all-cause mortality and LIS–all-cause mortality risk associations were 1·32 (95 % CI (1·18, 1·47); P for trend < 0·01) and 1·25 (95 % CI (1·12, 1·38); P for trend < 0·01), respectively, among those in the highest relative to the lowest quintiles. The findings were similar by sex and race and for all-cancer mortality, but weaker for all-CVD mortality. The joint HR for all-cause mortality among those in the highest relative to the lowest quintiles of both the DIS and LIS was 1·91 (95 % CI 1·57, 2·33) (P for interaction < 0·01). Diet and lifestyle, via their contributions to systemic inflammation, separately, but perhaps especially jointly, may be associated with higher mortality risk among men and women.

Type
Research Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2022. Published by Cambridge University Press on behalf of The Nutrition Society

CVD and cancer are among the leading causes of death globally(1). Emerging evidence suggests that chronic low-grade systemic inflammation may be a unifying mechanism underlying the development and progression of CVD and cancer(Reference Koene, Prizment and Blaes2). Circulating concentrations of inflammation biomarkers have been associated with higher risk for cancer(Reference Brenner, Scherer and Muir3), CVD(Reference Blake and Ridker4,Reference Kaptoge and Di Angelantonio5) and mortality(Reference Li, Zhong and Cheng6Reference Singh-Manoux, Shipley and Bell8). Diet and several lifestyle factors, including obesity, physical inactivity and alcohol and tobacco use are thought to contribute to inflammation(Reference Geovanini and Libby9). While diet and lifestyle are accepted risk factors common to CVD, cancer and mortality risk(Reference Willett, Koplan, Nugent, Jamison, Breman and Measham10,Reference Loef and Walach11) , whether inflammation is the primary mechanism through which diet and lifestyle affect mortality risk remains unclear.

Most previous studies that reported associations of the inflammatory potential of diet with all-cause and cause-specific mortality risk assessed diet using the dietary inflammatory index (DII). The largely nutrient-based DII was developed a priori based on its individual components’ reported effects on various inflammation biomarkers (mostly C-reactive protein) in in vitro and animal model studies and human trials, and associations with such biomarkers in human observational studies(Reference Shivappa, Steck and Hurley12). In a 2017 meta-analysis of prospective cohort studies, the DII was positively associated with all-cause (four studies), all-CVD (three studies) and all-cancer (four studies) mortality risk(Reference Zhong, Guo and Zhang13). Studies published since also support these findings for all-cause(Reference Garcia-Arellano, Martínez-González and Ramallal14Reference Agudo, Masegú and Bonet21) and all-CVD(Reference Park, Kang and Wilkens15Reference Agudo, Masegú and Bonet21) mortality, though findings for all-cancer mortality were more mixed(Reference Park, Kang and Wilkens15,Reference Okada, Shirakawa and Shivappa16,Reference Shivappa, Schneider and Hébert18Reference Agudo, Masegú and Bonet21) . However, the largely nutrient rather than food-based nature of the DII may not fully account for all the dietary constituents that may act and interact amongst themselves to affect inflammation. Moreover, the inflammatory potential of other lifestyle factors, such as physical inactivity, obesity and tobacco use, together with diet, may act collectively to affect mortality risk.

We recently developed and validated two novel scores, an a priori, largely whole foods and beverages-based dietary inflammation score (DIS) and a lifestyle inflammation score (LIS), based on FFQ and lifestyle questionnaire responses, and weighted to reflect dietary and lifestyle contributions to inflammation(Reference Byrd, Judd and Flanders22). The DIS was more strongly associated with high circulating concentrations of inflammation biomarkers than was the DII in three populations(Reference Byrd, Judd and Flanders22). The DIS and LIS were also positively associated with all-cause, all-cancer and all-CVD mortality risk separately, and especially jointly, among older White women in Iowa(Reference Li, Gao and Byrd23). However, DIS–mortality and LIS–mortality risk associations have not been investigated in a population comprising Black and White men and women. Accordingly, in the present study, we aimed to investigate the separate and combined associations of the DIS and LIS with all-cause, all-CVD and all-cancer mortality risk in a large, diverse cohort of US men and women. We hypothesised that the separate and, especially, the combined scores would be directly associated with all three mortality outcomes. We also compared associations of the weighted DIS and LIS (representing their components’ inflammatory potential) with mortality to those for an equal-weight DIS and LIS (representing their components’ overall mechanisms, not just inflammation-related ones) to explore the extent to which associations with risk may be inflammation-related.

Methods

Study population and data collection

We analysed data from 30 183 participants in a previously described prospective cohort study, REasons for Geographic and Racial Differences in Stroke (REGARDS)(Reference Howard, Cushman and Pulley24). Briefly, adults ≥45 years old were enrolled in REGARDS January 2003–October 2007 using a random sampling design within race-sex-geographic strata to recruit White and Black American men and women in both ‘stroke belt’ and non-stroke belt regions of the contiguous forty-eight states of the USA. REGARDS was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the University of Alabama-Birmingham Institutional Review Board (approval # IRB-020925004). Written informed consent was obtained from all study participants at enrolment. After enrolment, participants completed a telephone interview to collect information on demographics, medical conditions, lifestyle and other factors, followed by an in-home physical exam to obtain anthropometrics, including BMI. Dietary and alcohol data were derived from a previously validated(Reference Boucher, Cotterchio and Kreiger25), self-administered 110-item Block 98 FFQ, which was given to participants during the in-home visit to complete and return by mail. Physical activity was assessed via an open-ended question regarding the frequency in which the participant engaged in ‘intense physical activity, enough to work up a sweat’. Smoking status was derived from two questions regarding: (1) whether the participant smoked ≥100 cigarettes in their lifetime and (2) whether they currently smoked. Individuals who smoked <100 cigarettes were classified as never smokers. Individuals who smoked ≥100 cigarettes were classified as former smokers if they did not currently smoke, and current smokers if they did. We excluded participants according to criteria as shown in Fig. 1, yielding an analytic sample of 18 484.

Fig. 1. Exclusion flow chart among participants in the REGARDS study. REGARDS, REasons for Geographic and Racial Differences in Stroke. aThose who did not return the FFQ, returned a blank FFQ or those who skipped >15 % of the FFQ.

Exposure assessment

We created the exposures of interest, the DIS and LIS, to be mechanistic exposure scores (as opposed to diet or lifestyle ‘quality’ scores) to reflect the overall inflammatory potential of diet and lifestyle, respectively, and calculated them exactly as previously described(Reference Byrd, Judd and Flanders22). Accordingly, both scores comprised sums of components weighted according to their strengths of associations with a panel of systemic biomarkers of inflammation in a diverse population(Reference Byrd, Judd and Flanders22).

Briefly, the 19-component DIS comprises eighteen food (whole foods and beverages) groups (leafy greens and cruciferous vegetables; tomatoes; apples and berries; deep-yellow or orange fruits and vegetables; other fruits and real fruit juices; other vegetables; legumes; fish; poultry; red and organ meats; processed meats; added sugars; high-fat dairy products; low-fat dairy products; coffee and tea; nuts; other fats; refined grains and starchy vegetables) and one vitamin/mineral supplement score (Table 1). For the supplement score, first, we categorised individuals according to sex-specific tertiles of the distribution for each of the seventeen supplemental micronutrients considered (listed below). We assigned individuals in the lowest, middle and highest intake tertiles values of 0, 1 and 2, respectively. Then, we multiplied the tertile values for the hypothesised anti-inflammatory micronutrients (vitamins A, B12, B6, C, D and E; and β-carotene, folate, niacin, riboflavin, Ca, Mg, Se, thiamin and Zn) by +1, and the values for hypothesised pro-inflammatory micronutrients (Cu and Fe) by −1. We then summed the values to yield the supplement score. We standardised each of the eighteen food groups (g/d) and the supplement score to a mean of 0 and a sd of 1 based on the distribution in the analytic population. We then multiplied the resultant values for the nineteen DIS components by their respective weights (Table 1) and summed them to yield the DIS.

Table 1. Components and construction of the DIS and the LIS in the REGARDS study

DIS, dietary inflammation score; LIS, lifestyle inflammation score; REGARDS, REasons for Geographic and Racial Differences in Stroke.

* Dietary components were standardised to the sample at baseline, by sex, to a mean of 0 and a sd of 1.

Disaggregated from the following FFQ line items: refried beans or bean burritos; chili with beans (with or without meat); vegetable stew; vegetable soup, vegetable beef, chicken vegetable or tomato soup; any other soup, like chicken noodle, chowder, mushroom and instant soups; spaghetti, lasagna or other pasta with tomato sauce; cheese dishes without tomato sauce like macaroni and cheese; pizza, including carry-out; tacos, burritos, enchiladas, tamales, etc., with meat or chicken.

Vitamin and mineral supplemental intakes (self-reported by the participant from multivitamin/mineral and individual supplements) were ranked into tertiles of intake and assigned a value from 0 (low or no intake) to 2 (highest intake) for hypothesised anti-inflammatory micronutrients (all listed micronutrients except for Fe and Cu) or from 0 (low or no intake) to −2 (highest intake) for hypothesised pro-inflammatory micronutrients (Fe and Cu).

For the LIS, we categorised each component as follows: alcohol (non-drinkers (0 drinks/d), moderate drinkers (>0–≤1 drink/d for women and >0–≤2 drinks/d for men) and heavy drinkers (>1 drink/d for women and >2 drinks/d for men)); physical activity (inactive (0 times/week), moderately active (1–3 times/week) and heavily active (≥4 times/week)); BMI (normal (<25 kg/m2), overweight (25–<30 kg/m2) and obese (≥30 kg/m2)); and smoking (current and not current). We created dummy variables for each of the components and multiplied them by their assigned weights (Table 1) and summed them to yield the LIS.

Outcome assessment

REGARDS participants or their designated proxies were contacted by study staff every 6 months to ascertain deaths. If a death was reported, all associated records were collected, including medical records and the death certificate, and cause of death (through December 2016) was adjudicated by a committee of trained adjudicators(Reference Halanych, Shuaib and Parmar26). Our primary outcome of interest was all-cause mortality, defined as deaths due to any cause. Secondary outcomes of interest included all-CVD mortality, defined as deaths due to myocardial infarction, stroke, sudden death, heart failure, pulmonary embolism, other cardiac causes of death (e.g. myocarditis) and non-cardiac but other CVD deaths (e.g. ruptured aortic aneurysm), and all-cancer mortality, defined as deaths due to any type of cancer.

Statistical analyses

We summarised the participants’ baseline characteristics overall and within quintiles of the DIS and LIS distributions among all participants in the analytic cohort. We produced cumulative incidence functions for all-cause, all-CVD and all-cancer mortality within strata of the DIS and LIS quintiles. To estimate associations of DIS and LIS quintiles with mortality outcomes, we used multivariable Cox proportional hazards regression models to calculate cause-specific hazards ratios (HR) and their 95 % CI. To assess potential interaction between the DIS and LIS, we conducted joint/combined analyses to estimate the separate and combined associations of the DIS and LIS with all-cause, all-CVD and all-cancer mortality risk. For all analyses, follow-up began on the date of baseline questionnaire completion and ended at death or 31 December 2016, whichever was earliest. We assessed proportional hazards assumptions using likelihood ratio tests to compare models with and without a survival time exposure of interest interaction term; we observed no violations.

We identified and selected covariates as potential confounders based on biological plausibility and previous literature(Reference Park, Kang and Wilkens15Reference Shivappa, Schneider and Hébert18,Reference Veronese, Li and Manson20,Reference Agudo, Masegú and Bonet21,Reference Shivappa, Blair and Prizment27Reference Shivappa, Steck and Hussey29) . In all multivariable models, we adjusted for age (years), sex/current hormone therapy use (male, female-hormone therapy and female-no hormone therapy), race (White and Black American), annual household income (<$20 k, 20–34 k, 35–74 k, ≥75 k and missing), education (<high school, high school graduate, some college and college graduate or more), marital status (married, single and other), health insurance (yes and no), region of residence (stroke belt and non-stroke belt), regular (≥twice/week) non-aspirin non-steroidal anti-inflammatory drug (NSAID) or aspirin use (yes and no), regular statin use (yes and no), total energy intake (kcal/d), and co-morbid conditions (diabetes, heart disease (surgery or procedure on arteries, angioplasty or stenting of coronary arteries, repair of an aortic aneurism, self-reported myocardial infarction or evidence of a myocardial infarction via electrocardiogram) or kidney disease (based on glomerular filtration rate >60 ml/min/1·73 m2 or a urinary albumin:creatinine ratio >30 mg/g)) at baseline; scored 0–3). Multivariable models for the DIS additionally included all LIS components individually (for this purpose, we operationalised smoking as smoking pack-years). LIS models additionally included an equal-weight DIS and former smoking.

We also conducted several supplemental (secondary and sensitivity) analyses. First, to assess potential effect modification, we investigated potential interactions of the DIS and LIS with selected participant characteristics (sex, age, race, region, non-aspirin NSAID use, aspirin use, statin use, co-morbidities at baseline and tobacco use). Second, we calculated and investigated associations of equal-weight DIS and LIS (all components multiplied by 1 or −1 and summed) with mortality risk. We did this because the inflammation biomarker-weighted DIS and LIS were intended to be mechanistic (as opposed to ‘diet quality’) scores to represent the inflammatory potential of diet and lifestyle (i.e. each component’s contribution to the score is constrained by its strength of association with a panel of systemic biomarkers of inflammation and thus likely does not capture other potential mechanistic effects the components may have on disease or mortality risk). So, the equal-weight DIS and LIS were intended to represent the scores’ overall potential (inflammation-related plus other disease risk mechanisms). We hypothesised that, since inflammation is unlikely to be the only mechanism through which diet and lifestyle affect mortality risk, the equal-weight scores would be more strongly associated with mortality risk than would the weighted scores. Third, to compare DIS–mortality and DII–mortality associations, we calculated the DII(Reference Shivappa, Steck and Hurley12), the most commonly reported index for assessing the inflammatory potential of diet, exactly as previously described(Reference Byrd, Judd and Flanders22), based on thirty-four of the forty-five components available in REGARDS. For all DII components, we calculated Z-scores using the published global means and standard deviations, which we then converted to normalised percentiles, centred and multiplied by their respective weights. Based on previous findings that the DIS was more strongly associated with inflammation biomarkers than was the DII in three populations(Reference Byrd, Judd and Flanders22), we hypothesised that the DIS–mortality associations would be stronger than the DII–mortality associations. Fourth, we assessed the effects on our estimated associations of excluding ‘co-morbidities’ as a model covariate, because it may be a mediating factor (e.g. if individuals consume similar diets over time, a pro-inflammatory diet may increase the risk of co-morbidities present at baseline). Finally, because pharmalogical doses of β-carotene supplements may actually be pro-oxidant/pro-inflammatory and increase lung cancer and mortality risk among individuals at high risk for lung cancer(Reference Fortmann, Burda and Senger30), we assessed the effects on our estimated associations of excluding from analysis individuals with β-carotene supplement intakes above the study population’s 95th percentile (≥4·2 mg).

We conducted our statistical analyses using SAS (version 9.4) and produced our graphs using R (R Foundation for Statistical Computing), version 3.5.2. All statistical tests were two-sided, and we considered P ≤ 0·05 or 95 % CI that excluded 1·0 to be statistically significant.

Results

Over a median of 10·3 years of follow-up (range: 0·1–13·9), 4103 partipants died (1287 from CVD and 1072 from cancer). Participants in the highest relative to the lowest DIS quintile were more likely to be Black American, live in the US ‘stroke belt’ region, have lower household incomes and have less formal education (Table 2). Similar trends were observed across LIS quintiles.

Table 2. Baseline characteristics according to quintiles of DIS and LIS among participants in REGARDS (n 18 484), USA, 2003–2007

(Mean values and standard deviations; number and percentages)

DIS, dietary inflammation score; LIS, lifestyle inflammation score; HRT, hormone replacement therapy; NSAID, non-steroidal anti-inflammatory drug; REGARDS, REasons for Geographic and Racial Differences in Stroke.

* DIS quintile ranges were as follows: quintile 1, −10·4 to −2·1; quintile 3, −0·7 to 0·6; quintile 5, 2·1 to 10·0.

LIS quintile ranges were as follows: quintile 1, −1·1 to −0·2; quintile 3, 0·5 to 0·8; quintile 5, 1·4 to 2·4.

The following variables had missing values: income (11·5 %), insurance (<0·1 %), education (<0·1 %), regular NSAID use (0·3 %) and regular aspirin use (<0·1 %).

§ North Carolina, South Carolina, Georgia, Arkansas, Tennessee, Alabama, Mississippi and Louisiana.

Included diabetes, heart disease (surgery or procedure on arteries, angioplasty or stenting of coronary arteries, repair of an aortic aneurism, self-reported myocardial infarction, or evidence of a myocardial infarction via electrocardiogram) or kidney disease (based on glomerular filtration rate > 60 ml/min/1·73 square metres or a urinary albumin:creatinine ratio > 30 mg/g) at baseline (score 0–3).

At least twice/week.

** >1 drink (>14 g ethanol)/d for women; >2 drinks (>14 g ethanol)/d for men.

The 12-year cumulative incidences of all-cause, all-CVD and all-cancer mortality were higher among participants in the highest DIS quintile (33·8 %, 10·1 % and 8·8 %, respectively) than in the lowest (21·8 %, 7·3 % and 5·4 %, respectively) (Fig. 2, online Supplemental Table 1). The 12-year cumulative incidences of all-cause, all-CVD and all-cancer mortality were higher among participants in the highest LIS quintile (31·4 %, 10·3 % and 7·9 %, respectively) than in the lowest (23·9 %, 6·9 % and 6·2 %, respectively) (Fig. 3, online Supplemental Table 2).

Fig. 2. Cumulative incidence of all-cause, all-CVD and all-cancer mortality according to quintiles of the baseline distribution of the DIS in REGARDS (n 18 484), USA, 2003–2016. DIS, diet inflammation score; REGARDS, REasons for Geographic and Racial Differences in Stroke. Results are unadjusted.

Fig. 3. Cumulative incidence of all-cause, all-CVD and all-cancer mortality according to quintiles of the baseline distribution of the LIS in REGARDS (n 18 484), USA, 2003–2016. LIS, lifestyle inflammation score; REGARDS, REasons for Geographic and Racial Differences in Stroke. Results are unadjusted.

In multivariable models, the DIS was positively associated with all-cause and all-cancer mortality risk (P for trend < 0·01) (Table 3). For example, individuals in the highest relative to the lowest DIS quintile had statistically significant 32 % higher hazards of all-cause mortality (95 % CI (18, 47)) and statistically significant 39 % higher hazards of all-cancer mortality (95 % CI (11, 73)). The DIS association with all-CVD mortality was less clear. The LIS was postively associated with all three mortality outcomes (all P for trend ≤ 0·03). For example, individuals in the highest relative to the lowest LIS quintile had statistically significant 25 % (95 % CI (12, 38)), 26 % (95 % CI (5, 52)) and 33 % (95 % CI (9, 63)) higher hazards of all-cause, all-CVD and all-cancer mortality, respectively.

Table 3. Associations of DIS* and LIS with all-cause, all-CVD and all-cancer mortality risk among participants in REGARDS (n 17 757), USA, 2003–2016

(Hazard ration and 95 % confidence intervals)

DIS, dietary inflammation score; LIS, lifestyle inflammation score; REGARDS, REasons for Geographic and Racial Differences in Stroke; HR, hazards ratio; Ref, referent; HRT, hormone replacement therapy; NSAID, non-steroidal anti-inflammatory drug.

* The 19-component DIS was calculated as described in the text and Table 1; a higher DIS reflects a more pro-inflammatory diet.

The 4-component LIS was calculated as described in the text and Table 1; a higher LIS reflects a more pro-inflammatory lifestyle.

Death counts are from multivariable models that exclude participants with missing data for covariates. Death counts from models adjusting for age only may be higher than those shown.

§ From age-adjusted Cox proportional hazards models.

From multivariable Cox proportional hazards models. Models for DIS adjusted for age, sex/HRT use, race, income, education, insurance, marital status, region, co-morbidities (score 0–3), aspirin/NSAID use, statin use, total energy intake, physical activity (none, 1–3 times/week and ≥ 4 times/week), BMI (healthy weight, overweight and obese), alcohol intake (none, moderate and heavy) and tobacco use (pack-years). Models for LIS adjusted for age, sex/HRT use, race, income, education, insurance, marital status, region, co-morbidities (score 0–3), aspirin/NSAID use, statin use, total energy intake, former smoking status (yes and no) and the DIS (equal weights); excludes n 727 participants with missing data for covariates.

DIS quintile ranges were as follows: quintile 1, −10·4 to −2·1; quintile 2, −2·2 to −0·6; quintile 3, −0·7 to 0·6; quintile 4, 0·6 to 2·2; and quintile 5, 2·1 to 10·0.

** P for trend calculated by assigning the median of each DIS or LIS quintile to each quintile and treating this quintile exposure as a continuous variable.

†† LIS quintile ranges were as follows: quintile 1, −1·1 to −0·2; quintile 2, −0·2 to 0·5; quintile 3, 0·5 to 0·8; quintile 4, 0·9 to 1·3; and quintile 5, 1·4 to 2·4.

In our joint/combined analyses, the highest hazards for all-cause mortality risk was among those in the highest joint DIS and LIS quintile relative to those in the joint lowest (Table 4); risk was statistically significantly 91 % higher (95 % CI (57, 133)) (P for multiplicative interaction < 0·01). This compares to statistically significant 46 % higher hazards among those in the lowest LIS quintile who were in the highest relative to the lowest DIS quintile, and statistically significant 48 % higher hazards among those in the lowest DIS quintile who were in the highest relative to the lowest LIS quintile. The sample sizes for joint/combined analyes for CVD and cancer mortality risk were more limited, and no P for interaction from these analyses was statistically significant. However, the findings for all-cancer mortality risk were very similar to those for all-cause mortality, for example, the highest HR was among those in the highest joint DIS and LIS quintile relative to those in the joint lowest (HR 2·17; 95 % CI (1·57, 3·01)) (online Supplemental Table 3). On the other hand, there was no clear pattern for joint/combined DIS and LIS analysis findings in relation to CVD mortality (online Supplemental Table 4).

Table 4. Joint/combined (cross-classification) associations* of the DIS and LIS with all-cause mortality risk in REGARDS (n 17 757), USA, 2003–2016

(Hazard ration and 95 % confidence intervals)

DIS, diet inflammation score; LIS, lifestyle inflammation score; REGARDS, REasons for Geographic and Racial Differences in Stroke; HR, hazards ratio; Ref, referent; HRT, hormone replacement therapy; NSAID, non-steroidal anti-inflammatory drug.

* From multivariable Cox proportional hazards model adjusting for age, sex/HRT use, race, income, education, insurance, marital status, region, co-morbidities (score 0–3), aspirin/NSAID use, statin use and total energy intake. The interaction between the DIS and the LIS was modelled by including dummy variables representing the 2nd through 5th quintiles of the DIS and the LIS (four dummy variables each, with quintile 1 as the referent group) as well as their product terms (sixteen total product terms). Excludes 727 participants with missing data for covariates.

LIS quintile ranges were as follows: quintile 1, −1·1 to −0·2; quintile 2, −0·2 to 0·5; quintile 3, 0·5 to 0·8; quintile 4, 0·9 to 1·3; and quintile 5, 1·4 to 2·4.

DIS quintile ranges were as follows: quintile 1, −10·4 to −2·1; quintile 2, −2·2 to −0·6; quintile 3, −0·7 to 0·6; quintile 4, 0·6 to 2·2; and quintile 5, 2·1 to 10·0.

§ From DIS x LIS interaction term in the Cox proportional hazards model; relative excess risk due to interaction = −0·02, 95 % CI: (−0·48, 0·44); likelihood ratio test for multiplicative interaction: χ 2 = 40·8, P < 0·01.

There were no clear differences in the estimated DIS–all-cause mortality association across the strata of various other risk factors. However, there were suggestions that the estimated associations were stronger among those who were younger (<65 years) or formerly or currently smoked (online Supplemental Table 5). The LIS–all-cause mortality association was stronger among those who were younger (<65 years) (online Supplemental Table 6). The samples sizes for stratified analyses for CVD and all-cancer mortality were limited and the findings were too unstable for meaningful interpretation (online Supplemental Tables 710). However, there were suggestions that the DIS–all-CVD mortality association was stronger among women, younger participants and those without baseline co-morbidities.

The equal-weight DIS–mortality associations were minimally stronger and the equal-weight LIS–mortality associations were substantially stronger than those for their respective inflammation biomarker-weighted scores (online Supplemental Table 11). As examples, the HR for all-cause mortality risk among those in the highest relative to the lowest weighted and equal-weight DIS quintles were 1·32 and 1·38, respectively, and the corresponding values for the weighted and equal-weight LIS were 1·25 and 1·66, respectively.

Multivariable-adjusted DII–mortality associations were weaker than the corresponding DIS–mortality associations (online Supplemental Table 12). After excluding co-morbidities as a covariate from models, DIS–mortality associations were nearly identical, while LIS–mortality associations appeared stronger (online Supplemental Table 13). Results excluding participants with extreme β-carotene supplement intakes (online Supplemental Table 14) were similar, albeit slightly stronger, to those when those participants were included.

Discussion

Our results suggest that diets and lifestyles with greater inflammatory potentials, separately, but perhaps especially jointly, may be associated with higher mortality risk due to all causes, cancer and CVD. Our results also suggest that, although dietary and lifestyle contributions to mortality risk via inflammation may be substantial, for diet, inflammation may be the primary contribution, whereas lifestyle may also contribute substantially via other mechanisms.

Our findings that the DIS and LIS were positively associated with all-cause, all-cancer and all-CVD mortality (although the estimated positive DIS–all-CVD association among men and women combined was weaker and not statistically significant) are mostly supported by the only previous reported study of DIS–mortality and LIS–mortality associations, conducted by Li et al. (Reference Li, Gao and Byrd23), with some exceptions. For example, Li et al., in a prospective study of White Iowa women, found a positive DIS–all-CVD mortality association that was similar to ours, but slightly stronger and statistically significant(Reference Li, Gao and Byrd23). In our subgroup analyses, we found a positive DIS–all-CVD mortality association among women that was statistically significant and stronger than that among men. Moreover, in a recent meta-analysis, the DII, an alternative measure of the inflammatory potential of diet, was positively associated with CVD incidence or mortality risk among women but not men(Reference Shivappa, Godos and Hébert31). Other studies(Reference Park, Kang and Wilkens15,Reference Okada, Shirakawa and Shivappa16) , though not all(Reference Agudo, Masegú and Bonet21), also found stronger DII–all-CVD mortality associations among women than among men.

Other previous literature mostly supports our observation that a pro-inflammatory diet is associated with higher all-cause and all-cancer mortality risk, though most previous studies assessed inflammation from diet using the DII. According to recent meta-analyses, those in the highest relative to the lowest DII quantile had 21–23 % higher all-cause mortality risk(Reference Garcia-Arellano, Martínez-González and Ramallal14,Reference Namazi, Larijani and Azadbakht32) and 28 % higher all-cancer mortality risk(Reference Namazi, Larijani and Azadbakht32) – results that are comparable to our estimates using the DII (22 % and 29 % for all-cause and all-cancer mortality, respectively). Studies published since the meta-analyses also support our positive associations with all-cause mortality(Reference Park, Kang and Wilkens15,Reference Okada, Shirakawa and Shivappa16,Reference Veronese, Cisternino and Shivappa33) but were mixed with regard to all-cancer mortality, with some(Reference Park, Kang and Wilkens15,Reference Shivappa, Hebert and Kivimaki19,Reference Agudo, Masegú and Bonet21) , but not all(Reference Okada, Shirakawa and Shivappa16,Reference Shivappa, Schneider and Hébert18,Reference Veronese, Cisternino and Shivappa33) , studies supporting our findings. In our study, the DIS was more strongly directly associated with all-cause and all-cancer mortality than was the DII. These findings are consistent with those from our previous report, in which the DIS was more strongly associated with circulating inflammation biomarkers than was the DII(Reference Byrd, Judd and Flanders22). This may be due to the food-based nature of the DIS (as opposed to the largely nutrient-based DII), which may more comprehensively account for known and unknown dietary constituents that may affect inflammation, and the complex interactions among them(Reference Willett34).

In contrast to recent meta-analyses(Reference Zhong, Guo and Zhang13,Reference Shivappa, Godos and Hébert31,Reference Namazi, Larijani and Azadbakht32,Reference Ji, Hong and Chen35) , in our study, a pro-inflammatory diet assessed using the DIS or the DII was not statistically significantly associated with all-CVD mortality, although the estimated HR were >1·0. However, we did find evidence to suggest that the DIS may be more strongly associated with all-CVD mortality among women (as discussed above), younger participants and those without underlying baseline co-morbidities, though the CI for corresponding estimates across strata were wide and overlapped. Our observation of a possible stronger association among younger participants is in contrast to most(Reference Okada, Shirakawa and Shivappa16,Reference Agudo, Masegú and Bonet21,Reference Veronese, Cisternino and Shivappa33) , but not all(Reference Hodge, Bassett and Dugué17), studies, and could be due to chance. Our observation of a possible stronger DIS–CVD mortality association among those without co-morbid conditions at baseline could suggest that participants with a baseline co-morbidity may have changed their diets from their long-term unhealthy diets to healthier diets, but that the disease process was sufficiently advanced such that diet could no longer have a substantial effect.

One other study, a prospective cohort study in Sweden(Reference Kaluza, Håkansson and Harris36), reported associations of a diet inflammation score (the anti-inflammatory dietary index) with mortality risk(Reference Kaluza, Harris and Melhus37). The food-based anti-inflammatory dietary index, derived using a data-driven approach and scored in the opposite direction of the DIS and DII, was inversely associated with all-cause, all-CVD and all-cancer mortality risk(Reference Kaluza, Håkansson and Harris36). Discrepancies in findings for all-CVD mortality could be due to the aforementioned reasons.

Our study findings, except for our weak estimated DIS–CVD mortality risk association, generally align with those from previous studies that investigated various healthy diet pattern scores, albeit our findings were marginally weaker. In REGARDS, both Paleolithic and Mediterranean diet scores (with higher scores indicating ‘healthier’ diets) were inversely associated with all-cause, all-CVD and all-cancer mortality(Reference Whalen, Judd and McCullough38). We expected that other non-mechanism-oriented, ‘diet quality’ dietary pattern scores and our equal-weight DIS would be more strongly associated with mortality than was our weighted DIS, because the DIS was designed to assess the effects of diet through inflammation, not through the collective effects of all mechanisms. However, the similarity of the findings using our equal-weight and weighted diet scores suggest that inflammation may be the primary mechanism through which diet affects mortality risk.

Our study is among the first to report that a LIS comprising components weighted according to their contributions to inflammation is associated with higher all-cause, all-CVD and all-cancer mortality risk. Our findings using the equal-weight LIS were subtantially stronger than those using the weighted LIS, suggesting that lifestyle may affect mortality risk through inflammation as well as other mechanisms. For example, mortality risk may be increased by tobacco smoke mutagens(39), obesity through alterations in hormones and adipocytes(Reference Golemis, Scheet and Beck40,Reference Kyrou, Randeva, Tsigos, Feingold, Anawalt and Boyce41) , and physical inactivity through its effects on neuroendocrine and physiological responses to stressors(Reference Silverman and Deuster42). Heavy alcohol intake may affect mortality risk through a variety of mechanisms, including damage to DNA and organs (e.g. liver), weakening the immune system, and increasing injury risk(Reference Seitz and Stickel4345). Findings from previous studies that reported non-mechanistic lifestyle scores (i.e. the components were not weighted according to their strengths of associations with inflammation biomarkers) generally align with ours(Reference Loef and Walach11,Reference Veronese, Li and Manson20,Reference Ding, Rogers and van der Ploeg46Reference Ford, Bergmann and Boeing56) , though there is substantial heterogeneity in how lifestyle scores were constructed and modelled.

Our findings should be considered in context with our study’s limitations. First, information on diet and lifestyle were self-reported and may be subject to measurement error. However, such misclassification is likely non-differential due to the prospective nature of the study and is not expected to account for our positive findings. Second, we lacked detailed information on diet, lifestyle and other factors related to chronic disease risk and subsequent mortality over the life course. For example, if someone previously had a diet and lifestyle with a high inflammatory potential and was diagnosed with CVD prior to study enrolment, it is possible that they made lifestyle changes in hopes of reducing CVD mortality risk. As a result, this participant would appear to have a diet and lifestyle with a low inflammatory potential but would still be at high risk for CVD-related mortality, which would likely attenuate results. We tried to mitigate this potential by controlling for and considering possible heterogeneity by co-morbidities and certain CVD-related medications (e.g. aspirin and statins) at baseline, but bias due to residual or unmeasured confounding cannot be ruled out. Also, our FFQ did not allow separation of types of oils (e.g. olive oil) consumed. Last, our study results may not be generalisable to adults <45 years of age and non-Black and non-White Americans. Strengths of our study include the prospective study design, large and racially diverse study population, and use of validated DIS and LIS(Reference Byrd, Judd and Flanders22).

In summary, our findings, taken together with previous literature, suggest that diets and lifestyles (summarised through physical activity, obesity, and alcohol and tobacco use) with higher inflammatory potentials, both alone and especially in combination, may be associated with higher mortality risk. In addition, inflammation may be the primary mechanism through which diet affects mortality risk, and although inflammation may be a major mechanism through which lifestyle affects mortality risk, other mechanisms also likely substantially contribute; further investigations in these regards are needed. Our findings also support the use of our DIS and LIS. Finally, if our findings were to be consistently replicated, studies to test the effects of more anti-inflammatory diets and lifestyle on inflammation biomarkers and chronic disease incidence would be indicated.

Acknowledgements

The authors thank the other REGARDS study investigators who are not listed as co-authors on the present manuscript, as well as the staff and participants of the REGARDS study for their valuable contributions. A full list of participating REGARDS investigators and institutions can be found at: https://www.uab.edu/soph/regardsstudy/.

This work was supported by a cooperative agreement (SJ, grant number U01 NS041588) co-funded by the National Institute of Neurological Disorders and Stroke (NINDS) and the National Institute on Aging (NIA), National Institutes of Health, Department of Health and Human Service, and by R01 HL80477 from the National Heart Lung and Blood Institute (NHLBI); additional funding was provided by The Anne and Wilson P. Franklin Foundation (RMB). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS, NIA, NHLBI, or The Anne and Wilson P. Franklin Foundation. Representatives of the NINDS were involved in the review of the manuscript but were not directly involved in the collection, management, analysis or interpretation of the data.

A. T. assisted in designing the study, conducting and interpreting data analyses, and drafting the article. R. B. and D. B. assisted in the conception and design of the study, interpretating the data, and reviewing and revising the article. S. J. assisted in the study conception, acquisition of data, data interpretation, as well as reviewing and revising the article. W. D. F. assisted in interpretation of data and reviewing and revising the article. All authors read and approved the final version of this manuscript.

Dr Flanders owns ‘Epidemiologic Research & Methods, LLC’ which does some consulting work for a variety of clients. He knows of no conflicts of interest. All other authors have no conflicts of interest to disclose.

Supplementary material

For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114522001349

References

GBD 2017 Causes of Death Collaborators (2018) Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 392, 17361788.CrossRefGoogle Scholar
Koene, RJ, Prizment, AE, Blaes, A, et al. (2016) Shared risk factors in cardiovascular disease and cancer. Circulation 133, 11041114.CrossRefGoogle ScholarPubMed
Brenner, DR, Scherer, D, Muir, K, et al. (2014) A review of the application of inflammatory biomarkers in epidemiologic cancer research. Cancer Epidemiol Biomark Prev 23, 17291751.CrossRefGoogle ScholarPubMed
Blake, GJ & Ridker, PM (2002) Inflammatory bio-markers and cardiovascular risk prediction. J Intern Med 252, 283294.CrossRefGoogle ScholarPubMed
Emerging Risk Factors Collaboration, Kaptoge, S, Di Angelantonio, E, et al. (2010) C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet 375, 132140.Google ScholarPubMed
Li, Y, Zhong, X, Cheng, G, et al. (2017) Hs-CRP and all-cause, cardiovascular, and cancer mortality risk: a meta-analysis. Atherosclerosis 259, 7582.CrossRefGoogle ScholarPubMed
Kabagambe, EK, Judd, SE, Howard, VJ, et al. (2011) Inflammation biomarkers and risk of all-cause mortality in the reasons for geographic and racial differences in stroke cohort. Am J Epidemiol 174, 284292.CrossRefGoogle ScholarPubMed
Singh-Manoux, A, Shipley, MJ, Bell, JA, et al. (2017) Association between inflammatory biomarkers and all-cause, cardiovascular and cancer-related mortality. CMAJ 189, E384E390.CrossRefGoogle ScholarPubMed
Geovanini, GR & Libby, P (2018) Atherosclerosis and inflammation: overview and updates. Clin Sci 132, 12431252.CrossRefGoogle ScholarPubMed
Willett, WC, Koplan, JP, Nugent, R, et al. (2006) Prevention of chronic disease by means of diet and lifestyle changes. Chapter 44, In Disease Control Priorities in Developing Countries [Jamison, DT, Breman, JG, Measham, AR et al., editors]. Washington, DC: World Bank. Co-published by Oxford University Press, New York. Available from: https://www.ncbi.nlm.nih.gov/books/NBK11795/.Google Scholar
Loef, M & Walach, H (2012) The combined effects of healthy lifestyle behaviors on all cause mortality: a systematic review and meta-analysis. Prev Med 55, 163170.CrossRefGoogle ScholarPubMed
Shivappa, N, Steck, SE, Hurley, TG, et al. (2014) Designing and developing a literature-derived, population-based dietary inflammatory index. Public Health Nutr 17, 16891696.CrossRefGoogle ScholarPubMed
Zhong, X, Guo, L, Zhang, L, et al. (2017) Inflammatory potential of diet and risk of cardiovascular disease or mortality: a meta-analysis. Sci Rep 7, 6367.CrossRefGoogle ScholarPubMed
Garcia-Arellano, A, Martínez-González, MA, Ramallal, R, et al. (2019) Dietary inflammatory index and all-cause mortality in large cohorts: the SUN and PREDIMED studies. Clin Nutr 38, 12211231.CrossRefGoogle ScholarPubMed
Park, S-Y, Kang, M, Wilkens, LR, et al. (2018) The dietary inflammatory index and all-cause, cardiovascular disease, and cancer mortality in the multiethnic cohort study. Nutrients 10, 1844.CrossRefGoogle ScholarPubMed
Okada, E, Shirakawa, T, Shivappa, N, et al. (2019) Dietary inflammatory index is associated with risk of all-cause and cardiovascular disease mortality but not with cancer mortality in middle-aged and older Japanese adults. J Nutr 149, 14511459.CrossRefGoogle Scholar
Hodge, AM, Bassett, JK, Dugué, P-A, et al. (2018) Dietary inflammatory index or Mediterranean diet score as risk factors for total and cardiovascular mortality. Nutr Metab Cardiovasc Dis 28, 461469.CrossRefGoogle ScholarPubMed
Shivappa, N, Schneider, A, Hébert, JR, et al. (2018) Association between dietary inflammatory index, and cause-specific mortality in the MONICA/KORA Augsburg cohort study. Eur J Public Health 28, 167172.CrossRefGoogle ScholarPubMed
Shivappa, N, Hebert, JR, Kivimaki, M, et al. (2017) Alternative healthy eating index 2010, dietary inflammatory index and risk of mortality: results from the Whitehall II cohort study and meta-analysis of previous dietary inflammatory index and mortality studies. Br J Nutr 118, 210221.CrossRefGoogle ScholarPubMed
Veronese, N, Li, Y, Manson, JE, et al. (2016) Combined associations of body weight and lifestyle factors with all cause and cause specific mortality in men and women: prospective cohort study. BMJ 355, i5855.CrossRefGoogle ScholarPubMed
Agudo, A, Masegú, R, Bonet, C, et al. (2017) Inflammatory potential of the diet and mortality in the Spanish cohort of the European prospective investigation into cancer and nutrition (EPIC-Spain). Mol Nutr Food Res 61. doi: 10.1002/mnfr.201600649.CrossRefGoogle ScholarPubMed
Byrd, DA, Judd, SE, Flanders, WD, et al. (2019) Development and validation of novel dietary and lifestyle inflammation scores. J Nutr 149, 22062218.CrossRefGoogle ScholarPubMed
Li, Z, Gao, Y, Byrd, DA, et al. (2021) Novel dietary and lifestyle inflammation scores directly associated with all-cause, all-cancer, and all-cardiovascular disease mortality risks among women. J Nutr 151, 930939.CrossRefGoogle ScholarPubMed
Howard, VJ, Cushman, M, Pulley, L, et al. (2005) The reasons for geographic and racial differences in stroke study: objectives and design. Neuroepidemiology 25, 135143.CrossRefGoogle ScholarPubMed
Boucher, B, Cotterchio, M, Kreiger, N, et al. (2006) Validity and reliability of the Block98 food-frequency questionnaire in a sample of Canadian women. Public Health Nutr 9, 8493.CrossRefGoogle Scholar
Halanych, JH, Shuaib, F, Parmar, G, et al. (2011) Agreement on cause of death between proxies, death certificates, and clinician adjudicators in the reasons for geographic and racial differences in stroke (REGARDS) study. Am J Epidemiol 173, 13191326.CrossRefGoogle ScholarPubMed
Shivappa, N, Blair, CK, Prizment, AE, et al. (2016) Association between inflammatory potential of diet and mortality in the Iowa Women’s Health study. Eur J Nutr 55, 14911502.CrossRefGoogle ScholarPubMed
Shivappa, N, Harris, H, Wolk, A, et al. (2016) Association between inflammatory potential of diet and mortality among women in the Swedish Mammography Cohort. Eur J Nutr 55, 18911900.CrossRefGoogle ScholarPubMed
Shivappa, N, Steck, SE, Hussey, JR, et al. (2017) Inflammatory potential of diet and all-cause, cardiovascular, and cancer mortality in National Health and Nutrition Examination Survey III Study. Eur J Nutr 56, 683692.CrossRefGoogle ScholarPubMed
Fortmann, SP, Burda, BU, Senger, CA, et al. (2013) Vitamin, Mineral, and Multivitamin Supplements for the Primary Prevention of Cardiovascular Disease and Cancer: A Systematic Evidence Review for the U.S. Preventive Services Task Force. Rockville, MD: Agency for Healthcare Research and Quality.Google Scholar
Shivappa, N, Godos, J, Hébert, JR, et al. (2018) Dietary inflammatory index and cardiovascular risk and mortality-a meta-analysis. Nutrients 10, 200.CrossRefGoogle ScholarPubMed
Namazi, N, Larijani, B & Azadbakht, L (2018) Dietary inflammatory index and its association with the risk of cardiovascular diseases, metabolic syndrome, and mortality: a systematic review and meta-analysis. Horm Metab Res 50, 345358.Google ScholarPubMed
Veronese, N, Cisternino, AM, Shivappa, N, et al. (2020) Dietary inflammatory index and mortality: a cohort longitudinal study in a Mediterranean area. J Hum Nutr Diet 33, 138146.CrossRefGoogle Scholar
Willett, W (2013) Nutritional Epidemiology, 3rd ed. New York: Oxford University Press.Google Scholar
Ji, M, Hong, X, Chen, M, et al. (2020) Dietary inflammatory index and cardiovascular risk and mortality: a meta-analysis of cohort studies. Medicine 99, e20303.CrossRefGoogle ScholarPubMed
Kaluza, J, Håkansson, N, Harris, HR, et al. (2019) Influence of anti-inflammatory diet and smoking on mortality and survival in men and women: two prospective cohort studies. J Intern Med 285, 7591.CrossRefGoogle ScholarPubMed
Kaluza, J, Harris, H, Melhus, H, et al. (2018) Questionnaire-based anti-inflammatory diet index as a predictor of low-grade systemic inflammation. Antioxid Redox Signal 28, 7884.CrossRefGoogle ScholarPubMed
Whalen, KA, Judd, S, McCullough, ML, et al. (2017) Paleolithic and Mediterranean diet pattern scores are inversely associated with all-cause and cause-specific mortality in adults. J Nutr 147, 612620.CrossRefGoogle ScholarPubMed
Centers for Disease Control and Prevention (US), National Center for Chronic Disease Prevention and Health Promotion (US) & Office on Smoking and Health (US) (2010) How Tobacco Smoke Causes Disease: The Biology and Behavioral Basis for Smoking-Attributable Disease: A Report of the Surgeon General. 1, Introduction, Evaluation of Evidence on Mechanisms of Disease Production, and Summary. Atlanta, GA: Centers for Disease Control and Prevention.Google Scholar
Golemis, EA, Scheet, P, Beck, TN, et al. (2018) Molecular mechanisms of the preventable causes of cancer in the United States. Genes Dev 32, 868902.CrossRefGoogle ScholarPubMed
Kyrou, I, Randeva, HS, Tsigos, C, et al. (2000) Clinical problems caused by obesity. In Endotext [Internet] [Feingold, KR, Anawalt, B, Boyce, A et al., editors]. South Dartmouth, MA: MDText.com, Inc. Available from: https://www.ncbi.nlm.nih.gov/books/NBK278973/.Google Scholar
Silverman, MN & Deuster, PA (2014) Biological mechanisms underlying the role of physical fitness in health and resilience. Interface Focus 4, 20140040.CrossRefGoogle ScholarPubMed
Seitz, HK & Stickel, F (2007) Molecular mechanisms of alcohol-mediated carcinogenesis. Nat Rev Cancer 7, 599612.CrossRefGoogle ScholarPubMed
Rehm, J, Baliunas, D, Borges, GLG, et al. (2010) The relation between different dimensions of alcohol consumption and burden of disease: an overview. Addiction 105, 817843.CrossRefGoogle ScholarPubMed
World Health Organization (2018) Global Status Report on Alcohol and Health 2018. Geneva: WHO.Google Scholar
Ding, D, Rogers, K, van der Ploeg, H, et al. (2015) Traditional and emerging lifestyle risk behaviors and all-cause mortality in middle-aged and older adults: evidence from a large population-based Australian cohort. PLoS Med 12, e1001917.CrossRefGoogle ScholarPubMed
Petersen, KEN, Johnsen, NF, Olsen, A, et al. (2015) The combined impact of adherence to five lifestyle factors on all-cause, cancer and cardiovascular mortality: a prospective cohort study among Danish men and women. Br J Nutr 113, 849858.CrossRefGoogle ScholarPubMed
Prinelli, F, Yannakoulia, M, Anastasiou, CA, et al. (2015) Mediterranean diet and other lifestyle factors in relation to 20-year all-cause mortality: a cohort study in an Italian population. Br J Nutr 113, 10031011.CrossRefGoogle Scholar
Yun, JE, Won, S, Kimm, H, et al. (2012) Effects of a combined lifestyle score on 10-year mortality in Korean men and women: a prospective cohort study. BMC Public Health 12, 673.CrossRefGoogle ScholarPubMed
Lee, I, Kim, S & Kang, H (2019) Lifestyle risk factors and all-cause and cardiovascular disease mortality: data from the Korean longitudinal study of aging. Int J Environ Res Public Health 16, 3040.CrossRefGoogle ScholarPubMed
Behrens, G, Fischer, B, Kohler, S, et al. (2013) Healthy lifestyle behaviors and decreased risk of mortality in a large prospective study of U.S. women and men. Eur J Epidemiol 28, 361372.CrossRefGoogle Scholar
Bonaccio, M, Di Castelnuovo, A, Costanzo, S, et al. (2019) Impact of combined healthy lifestyle factors on survival in an adult general population and in high-risk groups: prospective results from the Moli-sani study. J Intern Med 286, 207220.Google Scholar
Li, Y, Pan, A, Wang, DD, et al. (2018) Impact of healthy lifestyle factors on life expectancies in the US population. Circulation 138, 345355.CrossRefGoogle ScholarPubMed
Larsson, SC, Kaluza, J & Wolk, A (2017) Combined impact of healthy lifestyle factors on lifespan: two prospective cohorts. J Intern Med 282, 209219.CrossRefGoogle ScholarPubMed
Carlsson, AC, Wändell, PE, Gigante, B, et al. (2013) Seven modifiable lifestyle factors predict reduced risk for ischemic cardiovascular disease and all-cause mortality regardless of body mass index: a cohort study. Int J Cardiol 168, 946952.CrossRefGoogle ScholarPubMed
Ford, ES, Bergmann, MM, Boeing, H, et al. (2012) Healthy lifestyle behaviors and all-cause mortality among adults in the United States. Prev Med 55, 2327.CrossRefGoogle ScholarPubMed
Figure 0

Fig. 1. Exclusion flow chart among participants in the REGARDS study. REGARDS, REasons for Geographic and Racial Differences in Stroke. aThose who did not return the FFQ, returned a blank FFQ or those who skipped >15 % of the FFQ.

Figure 1

Table 1. Components and construction of the DIS and the LIS in the REGARDS study

Figure 2

Table 2. Baseline characteristics according to quintiles of DIS and LIS among participants in REGARDS (n 18 484), USA, 2003–2007(Mean values and standard deviations; number and percentages)

Figure 3

Fig. 2. Cumulative incidence of all-cause, all-CVD and all-cancer mortality according to quintiles of the baseline distribution of the DIS in REGARDS (n 18 484), USA, 2003–2016. DIS, diet inflammation score; REGARDS, REasons for Geographic and Racial Differences in Stroke. Results are unadjusted.

Figure 4

Fig. 3. Cumulative incidence of all-cause, all-CVD and all-cancer mortality according to quintiles of the baseline distribution of the LIS in REGARDS (n 18 484), USA, 2003–2016. LIS, lifestyle inflammation score; REGARDS, REasons for Geographic and Racial Differences in Stroke. Results are unadjusted.

Figure 5

Table 3. Associations of DIS* and LIS† with all-cause, all-CVD and all-cancer mortality risk among participants in REGARDS (n 17 757), USA, 2003–2016(Hazard ration and 95 % confidence intervals)

Figure 6

Table 4. Joint/combined (cross-classification) associations* of the DIS and LIS with all-cause mortality risk in REGARDS (n 17 757), USA, 2003–2016(Hazard ration and 95 % confidence intervals)

Supplementary material: File

Troeschel et al. supplementary material

Troeschel et al. supplementary material

Download Troeschel et al. supplementary material(File)
File 133.9 KB