Relationship between diet and relative risk of pain in a cross-sectional analysis of the REGARDS longitudinal study

Abstract

Aim:

Determine if dietary patterns affect risk of pain.

Methods:

Data from 16,061 participants (55.4% females, 32.3% Black, age 65 ± 9 years) in the REGARDS study were categorized based on the adherence to previous dietary patterns reflecting the prevalent foods within each (convenience, alcohol/salads, plant-based, sweets/fats and ‘Southern’). A modified Poisson regression model was used to determine whether dietary patterns were associated with relative risk (RR) of pain.

Results:

High adherence to ‘Southern’ dietary pattern was associated with a 41% (95% CI: 23, 61%) increase in RR of pain. High adherence to a plant-based dietary pattern showed a 22% (95% CI: 11, 31%) decrease in the RR of pain.

Conclusion:

Poor quality dietary patterns increase the RR of pain, while plant-based patterns lowered the RR. Diet patterns should be incorporated into medical history.

Lay abstract

Chronic pain is a concern for many people and diet may influence the development and maintenance of pain. It is possible that the types of foods that people consume changes their likelihood of having pain. We know that diet interventions can reduce chronic pain, but it is not known how diet patterns are related to the risk of reporting pain. Here, we examined data from over 16,000 people and looked to see whether the types of foods that they ate was related to the likelihood of reporting pain in the last 4 weeks. People who ate more processed meats, sweetened beverages and fried foods (the ‘Southern’ diet pattern) were more likely to report pain than those that ate less of these items. However, people whose diet had more plant-based foods (vegetables, fruit, beans) had a decreased risk of reporting pain. Generally, those people who had a higher quality diet were less likely to report pain in the last 4 weeks. Not only can diets be used as a treatment for pain, but eating healthier foods may be protective of developing pain and reduce the likelihood of pain.

Chronic pain is a significant health problem affecting approximately one in five individuals in USA [1]. Unfortunately, chronic pain and access to pain treatment is not equally shared among the population. Black individuals and females report higher rates of chronic pain disorders compared with their non-Hispanic White (NHW) or male counterparts [2–4]. It has also been reported that females and Black individuals are more sensitive to noxious stimuli delivered in quantitative sensory testing (QST) [5,6] and that certain groups are at higher risk for developing pain [7–9]. Prevailing explanations regarding pain and sensitivity differences in humans have primarily focused on psychosocial variables [10–12], as well as genetic predispositions [13]. However, recent work has highlighted the role that diet plays in the development and predisposition to chronic pain.

Assessing pain-related nutrition has involved the examination of specific nutrients and types of food [14]. Research has shown that diet has the potential to exacerbate and alleviate pain [15,16], suggesting that dietary manipulation can be used as a safe means to help manage pain. Our previous experiments demonstrated that a standard American diet (SAD) high in refined carbohydrates, poly-unsaturated fatty acids (PUFAs) and saturated fat – can lead to increased immune cell activation, glucose intolerance, obesity, systemic inflammation and prolongation of recovery from injury [17]. In contrast, a low-glycemic diet that is high in anti-inflammatory food compounds can promote recovery and reverse many of the physiological and behavioral effects of the SAD in mice [18]. Additionally, we have shown that a low-carbohydrate diet can reduce daily pain, evoked pain and oxidative stress in adults with knee osteoarthritis [19]. Clinical studies have suggested that diets and specific foods can reduce pain across some conditions [16]. However, it is currently unknown whether differences in diet patterns may underlie differences in the prevalence of chronic pain from a population-based perspective. Additionally, employing a diet-pattern analysis may be beneficial compared with studying certain specific nutrients or food groups alone, as individuals typically eat more than one type of nutrient or food group in a meal [20]. It should be noted that diet patterns have been investigated in specific chronic pain conditions [21–25] and we, and others, have shown that changes to diet can affect chronic pain [19,26–28]. However, there is a need to continue to understand if and how diet can generally impact the risk of pain in the first place in a general sample, and if it may contribute to differences seen in populations in order to create better pain management regimes.

The REasons for Geographic and Racial Differences in Stroke (REGARDS) is a large longitudinal study that was designed to identify factors associated with racial and regional differences in stroke in USA. Since its implementation, REGARDS has grown to be multifaceted and expanded its reach beyond stroke to areas from renal care to cognition. Rigorous dietary analysis has resulted in specific dietary patterns in the REGARDS sample population are generalizable and have shown that both race and region are associated with nutritional differences [29,30]. The objective in this study was to use the previously derived dietary patterns to ascertain their associations on the prevalence of pain, as well as examine relationships with race within a geographically diverse population.

Methods

The description of the REGARDS cohort and methods have been described elsewhere [31] and sections have been reproduced for consistency between publications and for convenience. The data have been utilized to explore relations between diet and a number of measures including stroke [32], kidney disease [33], socioeconomic status [34], falls [35], cardiovascular disease [36], cognitive function [31] and diabetes [37].

Study sample

Details on the design and methods of REGARDS have been published in previous work [38]. To summarize, the REGARDS study is a national cohort of 30,239 community-dwelling Black and NHW individuals aged 45 years and older. Participants were recruited through the years of 2003 to 2007 via mail and telephone by using commercially available lists of US residents from Genesys, Inc. The study oversampled both Black Americans and residents of the southeastern USA. Upon entry into the study, the cohort of participants had a mean age of 64.8 years (ranging from 45 to 98 years), was approximately 42% Black, 55% females and 56% living in the southeastern USA. Criteria for exclusion included race other than Black and NHW, active treatment for cancer, chronic medical conditions precluding long-term participation, cognitive impairment, current or impending residence in a nursing home and the inability to understand verbal or written English.

The data were collected primarily by using computer-assisted telephone interviewing and an in-home examination. The initial phone call collected information regarding participant demographics, socioeconomic status, medical history and medication use. Following the telephone interview, an in-home examination by a trained medical professional was conducted to assess anthropometrics, blood and urine, blood pressure and an electrocardiogram (ECG) activity. Additionally, several self-administered questionnaires were given to the participant to complete and return. Additional details have been described elsewhere [38]. Verbal and written informed consent was obtained from all participants and the study was approved by the institutional review boards at all participating institutions (Columbia University, the University of Indiana, the University of Alabama at Birmingham, the University of Arkansas, Wake Forest University, the University of Cincinnati, Drexel University and the University of Vermont). Presence of absence of pain was determined at the first at-home visit in the study and, as such, our analysis uses a cross-sectional design with a single time point for each participant.

Dietary assessment

Dietary intake of the participants was assessed using the Block 98 Food Frequency Questionnaire (FFQ), which aims to assess usual dietary intake over the past year by asking about both frequency and portions of various food items [39]. The FFQ assesses food frequency by asking participants how often they consume each food item, with the following possible answers: never, a few times per year, once per month, 2–3-times per month, once per week, 3–4-times per week, 5–6–times per week or every day. The FFQ also assesses the usual quantity of food of the food item they consume, on average, each time it is consumed. For food consumed in individuals units such as eggs, bacon and doughnuts, participants were asked to pick the number that best represents the usual quantity of food consumed (i.e., 1 egg, 2 eggs, 3 eggs or 4 eggs). To aid in estimating usual quantity consumed for other items such as cereal or ice cream, participants were shown a photograph that illustrated several common portions of foods (i.e., ¼ cup, ½ cup, 1 cup or 2 cups of foods on plates and in bowls). The FFQ has been previously validated using food records for comparison. Correlations between the FFQ and diet records were in the 0.5–0.9 range, and were similar to those achievable by a food record for multiple days [40,41], suggesting that recall with the FFQ was comparable to 24- or 48-h food recall. The FFQ was completed by participants after the in-home visit.

Dietary patterns

The dietary patterns used in the present analysis were derived from the FFQ assessment [29]. These dietary patterns have been associated with other diseases such as incident stroke [42], coronary artery disease [30], sepsis [43] and progression to end-stage renal disease in patients with chronic kidney disease [33]. In brief, the 107 food items from the FFQ were combined into 56 food groups for use in principal component analysis (PCA), based on nutritional value and culinary use. Using a random split-sample technique to ensure validity and replication of patterns, PCA with varimax rotation was conducted in the first half of the sample. Factor solutions were examined for interpretability and separate PCA were conducted to test for congruence by region, sex and race. Congruence coefficients were used to determine if the dietary patterns represented the entire sample or should be derived into subgroups. In the second half of the sample, a confirmatory factor analysis including the food groups with absolute value loadings ≥0.20 was used independently to validate results from the PCA and test for model fit. Screen testing using eigen values >1.5 and examination of congruence coefficients to achieve optimal coherence across region, sex and race subgroups. The analysis retained five factors, and a final PCA with varimax rotation was performed on the full sample. Factor scores were calculated for each participant for each dietary pattern by multiplying the factor loading of each food group by each participant's average consumption of the foods.

The five dietary patterns were named to reflect the types of foods that were highly prevalent within them. Factor one was labeled ‘Convenience’ and consisted of mixed dishes with pizza, Chinese food and Mexican dishes; factor two was labeled ‘Plant-based’ given that it consisted mainly of vegetables, fresh fruit, beans and fish; factor three included miscellaneous sugars, desserts, candy, sweetened breakfast foods and added fats and was therefore deemed the ‘Sweets/Fats’ factor; factor four was named the ‘Southern’ pattern due to its high factor loadings of added fats, fried foods, eggs and egg dishes, organ meats, processed meats, refined sugar and sweetened beverages; and factor five was named ‘alcohol/salads’ due to the high presence of leafy-green vegetables, tomatoes, salad dressing, wine and liquor. Final factor loadings and additional details regarding dietary pattern calculation have been described at great length elsewhere [29]. It is worth noting that the comparative fit index (CFI) for the entire 56 items was 0.7 in the original derivation and the five identified factors had a CFI value of 0.675, indicating that these five factors explained the majority of the variance in the model.

Pain outcome ascertainment

Due to the fact that the REGARDS study was initially created to understand stroke risk, a bivariate (yes/no) pain variable was created using relevant questions available from the self-reported data. Questions regarding the usage of medication for pain (i.e., aspirin, opioids, acetaminophen and non-steroidal anti-inflammatories), as well as “during the past 4 weeks, how much did pain interfere with your normal work, including both work outside the home and housework? Did it interfere not at all, a little bit, moderately, quite a bit or extremely?)” were used to create the variable, with ‘yes’/‘A little bit-extremely’ indicating the presence of pain. Both questions regarding medication use and pain outcomes have been used across other validated pain assessment questionnaires aimed at assessing self-reported pain [44,45].

Statistical analysis

Before completing any analyses, the data were cleaned in order to remove those who were missing any of the diet and pain variables, as well as covariates. Questions regarding pain state were added later to the study, reducing the number in the total dataset that could provide this information. Chi-squared tests and t-tests were used to observe pain based on different socio-demographics (socioeconomic status (income + education), age, race, sex and caloric intake), lifestyle factors (smoking, mobility and quality of life (QOL)), and comorbid conditions (previous or current history of stroke, CVD and diabetes). It should be noted that the mobility variable included data concerning quantity and type of physical activity, as well as daily tasks. The QOL variable included data regarding physical, mental and emotional health and well-being. Quintiles were created to categorize individuals based on level of adherence to each dietary pattern. Bivariate analyses were used to assess the distribution of race and sex across the quintiles of each dietary pattern. Modified Poisson regression models were used to estimate the relative risk (RR) of pain based on the five different dietary patterns – convenience, plant-based, sweets/fats, Southern and alcohol/salads. In each model, quintile 1 was used as the reference when comparing all other quintiles. Three covariate models were applied to the unadjusted model to assess the relationship between diet pattern and RR of pain. Model A included age, sex, race, SES, and energy intake; Model B included Model A plus smoking, mobility and QOL; and Model C included Model B plus comorbid conditions (diabetes, stroke, and CVD). All analyses were performed using SAS 9.4 (SAS Institute, NC, USA).

Results

Of the 30,239 initially enrolled participants, 56 had data anomalies and 14,406 did not have complete data available for this analysis. Pain and some diet information was collected after the study had commenced, rendering it unavailable for some participants in the study. Caloric intake exclusions for males were <800 kcal or >5000 kcal, and for females <500 kcal and >4500 kcal, based on previous criteria of implausible daily energy intake [46]. The remainder of the participants were excluded due to missing variables needed to conduct the analysis. The details of participant selection are shown in Figure 1. Supplementary Table 1 identifies the number of individuals of each sex/race group that met exclusion criteria for the analysis. Black participants were somewhat more likely to have improper FFQ data, but NHW participants were more likely to have missing variables. The final analytic sample included 16,061 participants, which included 55.4% females, 32.3% who were Black, 17.1% whose total household income was <$20,000 per year, and an average age of 65 years (± 9 years). Demographic, socioeconomic, medical history and lifestyle characteristics by pain indication can be seen in Table 1.

Figure 1. . Details of participant selection from the REGARDS study.

Table 1. . Baseline characteristics of participants with and without pain.

Variable	Pain
	Yes, n = 11,527	No, n = 4534	p-value
Age (years)			<0.001
Age ≥65	5515 (47.8%)	2711 (59.8%)
Age <65	6012 (52.2%)	1823 (40.2%)
Race			<0.001
Black	3946 (34.2%)	1252 (27.6%)
White	7581 (65.8%)	3282 (72.4%)
Sex			<0.001
Female	7059 (61.2%)	1910 (42.1%)
Male	4468 (38.8%)	2624 (57.9%)
Income			<0.001
Income ≤$20 000	2265 (19.6%)	499 (11.0%)
Income >$20 000	9262 (80.4%)	4035 (89.0%)
Education			<0.001
Less than high school diploma	1353 (11.7%)	334 (7.4%)
More than high school diploma	10,174 (88.3%)	4200 (92.6%)
Current smoker	1736 (15.1%)	531 (11.7%)	<0.001
Adequate† QOL	2493 (21.6%)	2213 (48.8%)	<0.001
Adequate‡ mobility	2734 (23.7%)	2232 (49.2%)	<0.001
Hypertension	6948 (60.3%)	2725 (60.1%)	0.839
Stroke	717 (6.2%)	264 (5.8%)	0.342
Diabetes	2579 (22.4%)	971 (21.4%)	0.187
CVD	2073 (18.0%)	1143 (25.2%)	<0.001
Caloric intake (kcal/day), mean (SD)	1730 (732.6)	1676.0 (658.5)	<0.001

^†The term ‘adequate’ refers to the level of QOL in which the participant is able to enjoy and complete tasks without being hindered by physical, mental or emotional ailments.

^‡The term ‘adequate’ refers to an appropriate amount of bodily mobility necessary to complete daily living tasks and physical exercise without hindrance. Chi-squared tests and t-tests were used to observe pain based on different socio-demographics (socioeconomic status (income + education), age, race, sex and caloric intake), lifestyle factors (smoking, mobility and quality of life [QOL]), and comorbid conditions (previous or current history of stroke, CVD and diabetes).

Note: All results are presented as N (%) unless otherwise noted.

CVD: Cardiovascular disease; QOL: Quality of life.

In comparison to those without pain, participants indicating the presence of pain were more likely to be <65 year old, Black, females, with an income ≤$20K and less than a high school diploma (Table 1). In addition, those with pain were less likely to be a smoker with history of CVD and less likely to report a high QOL and high mobility, that is, adequate QOL and mobility that does not interfere with daily activities. There were no associations seen between those who had pain and the history of diabetes, stroke or hypertension. Bivariate analysis showed that the most pronounced difference in any of the dietary patterns was in a strong adherence (quintile 5) to the Southern pattern. A larger percentage of Black individuals (38%) compared with NHW individuals (11.4%) strongly adhered to this diet pattern. The bivariate analysis distribution of race (Black/NHW) and sex (male/female) across the dietary quintiles can be seen in Supplementary Table 2.

There were no significant interactions to report between the variables, including between sex and race. The following results are representative of those who indicated that they had pain present at baseline, and the full data analysis can be seen in Table 2. It should be noted that the results presented are those after applying Model C, which included all covariates used in the analysis.

Table 2. . Relative risk of pain outcomes among REGARDS diet patterns.

Diet pattern	Unadjusted		Model A†		Model B†		Model C†
	RR	CI	RR	CI	RR	CI	RR	CI
Factor 1 convenience
Q1 (N = 3212)	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Q2 (N = 3212)	0.93	(0.83–1.03)	0.97	(0.87–1.09)	0.95	(0.85–1.07)	0.94	(0.84–1.06)
Q3 (N = 3213)	0.94	(0.85–1.05)	1.01	(0.90–1.13)	1.00	(0.89–1.13)	0.99	(0.88–1.12)
Q4 (N = 3212)	0.99	(0.88–1.10)	1.05	(0.94–1.18)	1.03	(0.91–1.17)	1.02	(0.90–1.15)
Q5 (N = 3212)	1.09	(0.97–1.21)	1.06	(0.94–1.21)	1.05	(0.92–1.21)	1.04	(0.91–1.19)
Factor 2 plant based
Q1 (N = 3212)	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Q2 (N = 3212)	0.98	(0.88–1.09)	0.97	(0.87–1.09)	1.00	(0.88–1.12)	1.01	(0.89–1.14)
Q3 (N = 3213)	0.91	(0.81–1.01)	0.86	(0.77–0.96)	0.94	(0.83–1.06)	0.96	(0.85–1.08)
Q4 (N = 3212)	0.90	(0.80–1.00)	0.80	(0.72–0.90)	0.91	(0.80–1.02)	0.92	(0.82–1.04)
Q5 (N = 3212)	0.82	(0.73–0.91)	0.65	(0.58–0.73)	0.76	(0.67–0.87)	0.78	(0.69–0.89)
Factor 3 sweets/fats
Q1 (N = 3212)	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Q2 (N = 3212)	0.97	(0.87–1.08)	1.02	(0.91–1.13)	1.02	(0.90–1.14)	1.02	(0.91–1.14)
Q3 (N = 3213)	0.96	(0.86–1.06)	1.01	(0.91–1.14)	0.97	(0.86–1.09)	0.97	(0.86–1.09)
Q4 (N = 3212)	1.15	(1.03–1.28)	1.20	(1.06–1.35)	1.11	(0.98–1.25)	1.10	(0.97–1.25)
Q5 (N = 3212)	1.28	(1.14–1.43)	1.21	(1.06–1.39)	1.09	(0.97–1.24)	1.09	(0.94–1.26)
Factor 4 Southern
Q1 (N = 3212)	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Q2 (N = 3212)	1.07	(0.96–1.18)	1.11	(0.99–1.23)	1.03	(0.92–1.15)	1.07	(0.95–1.20)
Q3 (N = 3213)	1.27	(1.14–1.42)	1.36	(1.22–1.52)	1.20	(1.07–1.35)	1.25	(1.11–1.41)
Q4 (N = 3212)	1.29	(1.15–1.43)	1.37	(1.22–1.54)	1.16	(1.03–1.31)	1.19	(1.05–1.35)
Q5 (N = 3212)	1.68	(1.50–1.88)	1.64	(1.44–1.87)	1.48	(1.29–1.68)	1.41	(1.23–1.61)
Factor 5 alcohol/salads
Q1 (N = 3212)	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Q2 (N = 3212)	0.90	(0.81–1.01)	0.98	(0.88–1.10)	1.00	(0.89–1.13)	0.99	(0.88–1.12)
Q3 (N = 3213)	0.87	(0.78–0.97)	0.98	(0.87–1.10)	1.00	(0.89–1.13)	1.00	(0.88–1.13)
Q4 (N = 3212)	0.81	(0.73–0.91)	0.94	(0.84–1.05)	0.99	(0.88–1.12)	0.98	(0.86–1.10)
Q5 (N = 3212)	0.71	(0.64–0.79)	0.80	(0.71–0.90)	0.90	(0.80–1.02)	0.89	(0.78–1.01)

^†Model A – age, sex, race, SES and energy intake; Model B – Model A + smoking, mobility and QOL; Model C – Model B + comorbid conditions: CVD, stroke and diabetes.

Modified Poisson regression models were used to estimate the relative risk of pain based on the five different dietary patterns – convenience, plant-based, sweets/fats, Southern and alcohol/salads. In each model, quintile 1 was used as the reference when comparing all other quintiles. Three covariate models were applied to the unadjusted model to assess the relationship between diet pattern and relative risk of pain.

Factor 1: convenience diet pattern

In the unadjusted model and adjusted models (A, B and C), there was no significant effect of diet on the RR of pain.

Factor 2: plant-based diet pattern

There was a decrease RR in quintile 5 after adjusting for model C (RR: 0.78; 95% CI: 0.69–0.89).

Factor 3: sweets/fats diet pattern

Upon adjusting for model A, increases in RR remained significant in quintile 4 (RR: 1.20; 95% CI: 1.06–1.35) and quintile 5 (RR: 1.21; 95% CI: 1.06–1.39). There were no significant changes in RR after adjusting for model B or C.

Factor 4: Southern diet pattern

After adjusting for model C, quintile 3 had an increased RR (RR: 1.25; 95% CI: 1.11–1.41), as did quintile 4 (RR: 1.19; 95% CI: 1.05–1.35) and 5 (RR: 1.41; 95% CI: 1.23–1.61).

Factor 5: alcohol/salads diet pattern

After adjusting for model A, only quintile 5 showed significant reduction in RR (RR: 0.80; 95% CI: 0.71–0.90). Models B and C revealed no significant RR.

Discussion

Chronic pain is a major issue that has a tremendous impact on quality of life and health [47,48]. However, there is a paucity of data examining the association of dietary patterns on pain in a nationwide sample. To our knowledge, this is the first epidemiological analysis on the effects of dietary pattern on RR for pain. The demographic characteristics of those who reported pain in our sample align with previous literature documenting disparities [49,50]. Males and females tended to have similar percentages of the groups adhering to the five quintiles in all of the dietary patterns. Differences in the percentages of Black versus NHW individuals adhering to the five quintiles in the dietary patterns were larger, with the most prominent difference being found in a high adherence (quintile 5) to a Southern dietary pattern (Black 38%; NHW 11%). There was a significant increase in the RR of pain among those who strongly adhered to the Southern dietary pattern (41%), even after controlling for other factors [51–55]. The increases seen in this dietary patterns were not seen in diet patterns of ‘higher quality’ (higher in essential nutrients and lower in calories, saturated fats and refined carbohydrates), such as the plant-based pattern of eating where we identified a decreased RR of pain among those who strongly adhered to a plant-based dietary pattern (22%). Although non-significant, there was a slight increase in the RR for pain in the sweets/fats diet pattern and a minor reduction for the alcohol/salads diet pattern. It may be that specific aspects of each of the two previous diet patterns counteract the beneficial and detrimental aspects of the patterns and reduce the positive/negative impact.

There are a number of reasons to support the notion that diet can influence the likelihood for pain. Characteristics of the poorer quality diet patterns include high amounts of refined and added carbohydrates, saturated fats and promotion of excess consumption of calories. These patterns are often accompanied by a general lack of fruits, vegetables and subsequent vitamins and minerals that are essential for typical human functioning [29]. Diets such as these have been implicated in the progression of many disease states [56–60]. Excess carbohydrates, saturated fats and calories promote the formation of unusually high amounts of reactive oxygen species (ROS) and subsequent oxidative stress [61] – a natural phenomenon necessary for normal functioning, but can become harmful with increased ROS and lack of antioxidants [62]. Excess carbohydrates can form advanced glycation end products, which in turn can activate the immune system, stimulate nociceptors as well as cause cellular damage and inflammation [63]. Saturated fats and omega-6 fatty acids promote lipid peroxidation which can be overwhelming to the body in excess [64,65]. Excessively high caloric intake will allow more substrates to enter the mitochondrial electron transport chain, and essentially create a backlog of electrons that eventually are given to molecular oxygen forming the ROS superoxide [66]. It is thought that oxidative stress is key in the development of many conditions [67], including painful conditions [19,68,69]. Normally, the body has biochemical mechanisms to neutralize ROS before they can cause any damage [70,71], but these antioxidants are in low supply in the poorer quality diets. Plant-based dietary patterns provide individuals with adequate amounts of antioxidants and are often much lower in refined carbohydrates and saturated fats [72]. To that end, we have previously shown that reduction of oxidative stress via diet intervention may be beneficial to those with pain due to knee osteoarthritis [19].

Previous literature has shown that diets similar to the Southern diet pattern can create an environment of chronic immune system activation both centrally and peripherally in rodents [17,18]. In addition, such diet patterns have been associated with obesity and immune activation via the adipokine leptin [73,74]. Immune system activation has been established as one of the major driving forces behind chronic pain due to its ability to promote an inflammatory environment [75]. Taken together, diet and obesity induced immune system activation may predispose individuals to chronic pain.

From our dataset, a larger percentage of Black individuals (38.0%) were strongly adherent (quintile 5) to the Southern diet pattern compared with NHWs (11.4%). Whereas we show that higher adherence to poor-quality diet patterns increases RR regardless of race, it is important to acknowledge racially specific factors surrounding diet. Systemic racism, necessity and cultural factors may be related to dietary patterns seen in Black populations [76]. There are traditions and beliefs about food in Black communities that influence dietary patterns, and these must be respected when addressing or modifying dietary habits [77,78]. Predominantly Black communities in USA also have higher rates of food insecurity [79] and this is often related to poor quality diets, as access to healthier foods is limited or unavailable [80,81]. Food insecurity is defined as a household-level economic and social condition of limited or uncertain access to adequate food [82]. The REGARDS studies have shown that individuals in food deserts have a difficult time adhering to a healthier dietary pattern compared with those who do not live in a food desert [83]. Foods included in the Southern diet pattern often are more readily available and less expensive than those in the plant-based or alcohol/salads in areas experiencing food insecurity [84]. The relationship between race and food insecurity is also related to other factors such as gaps in income, poverty, unemployment, incarceration and disability [79,85]. Consequently, in areas of the USA where food insecurity is more prevalent and income is lower, adherence to a Southern diet pattern may be more likely. These regions often have greater populations of Black individuals, contributing to the higher prevalence of the Southern diet pattern in Black individuals noted above. In order to achieve better diet and pain outcomes, it is critical that social factors be addressed in a culturally sensitive manner.

There are a number of limitations in this study. First, the pain outcome variable was created using parts of other validated methods of reporting pain as opposed to one validated self-report method, such as the Brief Pain Inventory [86]. Future studies should look to include such methods of assessing pain, as well as including quantitative sensory testing measures. The present study also only assessed Black and NHW participants, and future studies should be more racially diverse. Given the impact of gender identity on pain [8], future studies should also include data on not only sex, but gender identity in order to better serve the entire population. In the current study, the analyses were conducted on the population as a whole with race and sex as variables within the analysis. Future work will carry out separate analyses in each racial group and sex. While we believe that dietary pattern is contributing to the RR of having pain, we understand that the nature of the study does not allow a determination of causality. It is possible that the presence of pain has a negative effect on dietary patterns as well. Future longitudinal studies are encouraged to determine this relationship. It would also be efficacious for future studies to look at the effect of diet pattern on specific types of painful disorders in order to gain a better understanding of the underlying mechanisms behind dietary influence on pain severity.

Conclusion

The potential impact of diet on various health outcomes has been overlooked for some time. While foodstuffs have been studied in other disease models, they have not been the focus of substantial study in the context of pain. In the present study, we showed that adherence to a diet pattern high in refined sugars and unhealthy fats, such as the Southern dietary pattern, was associated with an increased RR of having pain. This increase in RR was not seen in healthier diet patterns. While other factors in the pain experience are indeed important, it is clear that diet may be worthy of intervention as a treatment option in clinical practice. Additionally, guidance may need to be tailored based on differing demographic characteristics, such as race. It is important to keep in mind that there are a variety of factors determining one's dietary habits that must be considered in future interventions. In addition, as evidenced by intervention studies [19,26–28], diet may be suitable and effective as a preventative measure to reduce risk of chronic pain development.

Summary points.

• Chronic pain is a growing concern and diet may contribute to its development.

• Diet can be used as treatments for chronic pain.

• This study aimed to assess whether specific dietary patterns were associated with an increased risk of reporting pain.

• Data from a national sample of 16,061 individuals as part of the REGARDS study were analyzed based on levels of adherence to previously defined dietary patterns.

• Higher adherence to the Southern diet pattern was associated with a 41% increase in the relative risk (RR) for pain.

• Black individuals in our sample showed higher adherence to the Southern diet pattern than non-Hispanic White individuals.

• Higher adherence to a plant-based dietary pattern was associated with a decreased RR for pain (22%).

• Males and females showed no differences in diet patterns.

• In addition to their use as treatments for pain, dietary patterns may be useful as preventative measures for those at risk for chronic pain.