Mediterranean Diet and Changes in Frequency, Severity, and Localization of Pain in Older Adults: The Seniors-ENRICA Cohorts

Abstract

Background

Although some components of the Mediterranean diet have shown benefits in pain risk through its anti-inflammatory/antioxidant properties, no population-based studies have investigated the effect of adherence to this diet on changes in pain over time.

Methods

We used data from 864 and 862 older adults recruited in the Seniors-ENRICA-1 and Seniors-ENRICA-2 cohorts and followed-up for 2.8 and 2.4 years, respectively. Adherence to the Mediterranean diet was assessed with the MEDAS score at baseline. Frequency, severity, and locations of pain obtained at baseline and follow-up were used to compute a pain scale. Analyses were performed using multinomial logistic regression models, and adjusted for the main confounders.

Results

Participants had a mean (SD) age of 71.5 (5.1) years, 36.8% were men, and 78.3% had chronic conditions. In the pooled cohorts, compared with participants in the lowest quartile of the MEDAS score (lowest adherence to the Mediterranean diet), those in the highest quartile showed a higher frequency of pain improvement versus worsening (relative risk ratio [95% confidence interval]: 1.43 [1.03, 1.99]). This association was also evidenced in 2 components of the pain scale: improvement in pain severity (1.43 [1.01, 2.04]) and reduction in pain locations (1.54 [1.08, 2.20]), but a tendency to pain frequency improvement (1.34 [0.92, 1.93]) was also observed. The main contributors to these associations were high consumption of fruit and vegetables, and low consumption of sugar-sweetened beverages.

Conclusions

A higher adherence to the Mediterranean diet was related to a subsequent improvement in pain characteristics in older adults, suggesting that improving diet quality may help reduce the high health impact of pain.

Chronic pain, Cohort study, Dietary patterns

Pain is a common problem with a considerable impact on health and daily functioning. In 2016, 2 of the 10 more prevalent causes of disease burden worldwide were pain conditions: tension headache and migraine, with 1.9 and 1.0 billion people, respectively. In addition, the leading causes of years lived with disability globally were low back pain and migraine, and neck pain was among the top 10 (1). The magnitude of these problems is greater in older adults, where the prevalence of chronic pain conditions may increase up to 60% (2,3), constituting a growing public health concern.

There are many risk factors for chronic pain, including sociodemographic, psychological, clinical, and biological factors (4), but among them, lifestyles are of great relevance because they may influence the development and impact of chronic pain, and, most importantly, because they are modifiable (5,6). Identifying and modifying these health-related behaviors, including diet, may prevent chronic pain, or reduce its duration or severity.

Previous research on the role of diet in pain has focused on nutritional interventions to alleviate chronic pain. Thus, increased fiber intake, reduced fat consumption, and supplementation with vitamins, omega-3 fatty acids, polyphenols, soy protein, and certain amino acids have been found helpful in the management of chronic pain (7,8). Some dietary patterns, such as vegetarian or vegan diets, have also shown some benefits, which have been attributed to the anti-inflammatory or antioxidant properties of its components (9,10). In this context, the Mediterranean dietary pattern may have beneficial effects on pain by improving pain itself through its anti-inflammatory effects (11–13), but also by lowering the risk of chronic conditions associated with chronic pain, such as cardiovascular disease, type 2 diabetes, obesity, depression, and frailty (14). This is important because by improving pain, the overuse of analgesic drugs, which is a cause of concern in older adults (15), would also be reduced. However, to our knowledge, no population-based studies have investigated the effect of the Mediterranean diet on changes in pain and its characteristics over time.

Thus, this paper aimed to test the hypothesis that a greater adherence to the Mediterranean diet is associated with a clinically meaningful improvement in the frequency, severity, and number of pain sites among 2 cohorts of older adults in Spain, both independently and pooled. Our main objective was to evaluate how pain characteristics evolved over a median of 2.6 years among a subset of participants who experienced pain either at baseline or follow-up.

Method

Study Design and Population

We used data from the Seniors-ENRICA-1 (16) and Seniors-ENRICA-2 (17) cohorts. Participants in the Seniors-ENRICA-1 cohort were selected between 2008 and 2010 by stratified cluster random sampling of the community-dwelling population aged 60 and older in Spain, whereas participants in the Seniors-ENRICA-2 cohort were recruited between 2015 and 2017 by stratified random sampling of all community-dwelling individuals aged 65 and older holding a national health care card and living in the city of Madrid (Spain) or 4 surrounding large towns.

The instruments and procedures used for data collection were the same for both studies and have been reported in detail elsewhere (18). Briefly, at Wave 0, a computer-assisted telephone interview was used to obtain information on sociodemographic data, lifestyles, and morbidity, and 2 home visits were conducted to collect biological samples, perform a physical examination, and obtain a diet history. Information was updated in 2012 (Wave 1) and 2015 (Wave 2) in the Seniors-ENRICA-1, and in 2019 (Wave 1) in the Seniors-ENRICA-2. Since in the Seniors-ENRICA-1 cohort information on pain was not collected at Wave 0, for the purpose of this study Wave 1 was considered the baseline assessment. The median (full range) follow-up was 2.8 years (2.1–3.3) in the Seniors-ENRICA-1, and 2.4 years (1.8–3.3) in the Seniors-ENRICA-2. All participants provided written informed consent, and the Clinical Research Ethics Committee of “La Paz” University Hospital in Madrid approved both studies.

Study Variables

Adherence to the Mediterranean diet

Habitual dietary intake in the previous year was estimated with a validated diet history developed from the one used in the EPIC cohort study in Spain, with good correlation with seven 24-hour recalls obtained over 1 year (19). This diet history collected information on 861 foods, including 184 recipes for dishes commonly eaten in Spain, and used photographs to better quantify portion sizes. Energy and nutrient intake was estimated using standard composition tables (19).

Adherence to the Mediterranean diet was measured with the Mediterranean Diet Adherence Screener (MEDAS) (20). This instrument was developed to assess compliance with the dietary intervention of the PREDIMED trial, and consists of 2 questions on food intake habits considered characteristic of the Spanish Mediterranean diet and 12 questions on food consumption frequency. However, we did not ask these questions to the participants; instead, we obtained the information directly from the diet history. One point was given for each target achieved: using olive oil as the main fat for cooking; preferring white meat over red meat; ≥4 tablespoons/day of olive oil; ≥2 servings/day of vegetables; ≥3 pieces/day of fruit; <1 serving/day of red or processed meats; <1 serving/day of butter, margarine, or cream; <1 cup/day of sugar-sweetened beverages; ≥7 servings/week of wine; ≥3 servings/week of legumes; ≥3 servings/week of fish or seafood; <2 servings/week of commercial baked goods; ≥3 servings/week of nuts; and ≥2 servings/week of a dish with a traditional sauce of tomatoes, garlic, onion, or leeks sautéed in olive oil (sofrito). Thus, the MEDAS score ranges from 0 (lowest) to 14 (highest adherence to the Mediterranean diet) (20).

Pain

Based on the questionnaire used in the Survey on Chronic Pain in Europe (21), we obtained information on 3 characteristics of pain at baseline and follow-up:

• 1) pain frequency in the last 6 months: 0 points for participants reporting no pain in the last 6 months, 1 point for those reporting sporadic pain (less often than 1 time per month, 1–3 times per month, or 1 time per week), and 2 points for those with persistent pain (at least 2 times per week, daily, or at all times);

• 2) pain severity, based on pain impact on habitual activities of daily living: 0 points for participants with no pain, 1 point for those with light pain (pain not troubling at all, or troubling a little), and 2 points for those with moderate-to-severe pain (pain troubling moderately, a lot, or completely);

• 3) number of pain locations among 6 sites (head and neck, back, bones and joints, legs, arms, and other sites): 0 points for participants reporting zero pain sites, 1 point for those with 1 or 2 pain sites, and 2 points for those with 3 or more pain sites.

We then computed a pain scale that ranged from 0 (no pain) to 6 (worst pain) at baseline and follow-up, by summing the scores of its 3 components, and calculated changes in the overall pain scale and its 3 components from baseline to follow-up.

Participants were also categorized into pain improvement (reduction of at least 1 point in the pain scale), no change, or pain worsening (increase of at least 1 point). A 1-point change is close to one-half the SDs of the pain scale in our cohorts, which ranged from 2.2 to 2.5 (Table 1), a criterion that is usually employed to consider a certain change in a scale as a clinically important difference (22,23). It also renders large standardized response means (calculated by dividing the mean score change by the SD of the change) of −2.14 and 2.15 in the pain improvement and worsening categories, respectively, well above the Cohen’s widely used lower bound of 0.80 to consider a size effect as large (23).

Table 1.

Baseline Characteristics of Older Adults Participating in the Seniors-ENRICA-1 and Seniors-ENRICA-2 Cohorts by Categories of Adherence to the Mediterranean Diet

		MEDAS Score (quartiles)†
	Total	Q1	Q2	Q3	Q4 (highest)
Seniors-ENRICA-1 Cohort	n = 864	n = 252	n = 197	n = 186	n = 229
Age (years), mean (SD)	71.3 (5.8)	71.9 (6.2)	71.4 (5.9)	71.0 (5.5)	70.8 (5.6)
Sex—men, n (%)	333 (38.5)	88 (34.9)	79 (35.5)	74 (39.8)	101 (44.1)
Educational level—primary or less, n (%)	475 (55.0)	137 (54.4)	111 (56.4)	102 (54.8)	125 (54.6)
Tobacco smoking—current smoker, n (%)	69 (8.0)	22 (8.7)	20 (10.2)	13 (7.0)	14 (6.1)
Alcohol consumption—moderate drinker, n (%)	488 (56.5)	135 (53.6)	107 (54.3)	103 (55.4)	143 (62.5)
Leisure-time physical activity (MET-h/week), mean (SD)	21.0 (14.6)	18.4 (14.2)	20.8 (15.4)	22.5 (14.6)	22.8 (13.8)*
Time watching TV (h/day), mean (SD)	2.89 (1.56)	3.12 (1.61)	3.03 (1.65)	2.73 (1.22)	2.66 (1.63)*
BMI (kg/m2), mean (SD)	28.7 (4.5)	29.3 (4.4)	28.8 (4.5)	28.3 (4.3)	28.3 (4.5)*
Energy intake (kcal/day), mean (SD)	2008 (465)	1978 (441)	1973 (472)	1995 (463)	2084 (481)*
Number of chronic diseases,‡ mean (SD)	1.40 (0.98)	1.48 (1.02)	1.38 (0.92)	1.42 (0.99)	1.32 (0.97)
Use of analgesics, n (%)	458 (53.0)	135 (53.6)	116 (58.9)	96 (51.6)	111 (48.5)
Pain scale score, mean (SD)	3.70 (2.29)	3.63 (2.39)	3.95 (2.18)	3.64 (2.34)	3.62 (2.21)
Change in the pain scale from baseline to follow-up, mean (SD)	−0.38 (3.76)	−0.12 (3.77)	−0.49 (3.55)	−0.32 (3.94)	−0.61 (3.76)
Seniors-ENRICA-2 Cohort	n = 862	n = 284	n = 212	n = 191	n = 175
Age (years), mean (SD)	71.8 (4.3)	71.7 (4.3)	71.9 (4.2)	71.4 (4.2)	72.1 (4.8)
Sex—men, n (%)	302 (35.0)	94 (33.1)	66 (31.1)	66 (34.6)	76 (43.4)
Educational level—primary or less, n (%)	552 (64.0)	142 (62.0)	142 (67.0)	119 (62.3)	115 (65.7)
Tobacco smoking—current smoker, n (%)	63 (7.3)	32 (11.3)	13 (6.1)	14 (7.3)	4 (2.3)*
Alcohol consumption—moderate drinker, n (%)	572 (66.4)	186 (65.5)	138 (65.1)	130 (68.1)	118 (67.4)
Leisure-time physical activity (MET-h/week), mean (SD)	26.4 (19.0)	24.9 (17.8)	26.8 (19.2)	26.2 (20.3)	28.4 (19.0)
Time watching TV (h/day), mean (SD)	3.25 (1.55)	3.29 (1.57)	3.28 (1.54)	3.26 (1.65)	3.12 (1.40)
BMI (kg/m2), mean (SD)	28.4 (4.6)	28.2 (4.9)	28.2 (4.9)	28.3 (4.7)	28.9 (4.9)
Energy intake (kcal/day), mean (SD)	1928 (362)	1891 (339)	1900 (356)	1938 (356)	2012 (397)*
Number of chronic diseases,‡ mean (SD)	1.10 (0.90)	1.15 (0.88)	1.05 (0.94)	1.08 (0.91)	1.11 (0.88)
Use of analgesics, n (%)	407 (47.2)	127 (44.7)	97 (45.8)	104 (54.5)	79 (45.1)
Pain scale score, mean (SD)	3.26 (2.48)	3.15 (2.53)	3.23 (2.44)	3.48 (2.45)	3.25 (2.48)
Change in the pain scale from baseline to follow-up, mean (SD)	0.19 (3.79)	0.30 (3.91)	0.29 (3.75)	0.03 (3.72)	0.05 (3.75)

Notes: BMI = body mass index; MEDAS = Mediterranean Diet Adherence Screener; SD = standard deviation.

^†MEDAS score quartiles: Q1: ≤6, Q2: 7, Q3: 8, Q4: ≥9.

^‡Including diabetes, cardiovascular disease, chronic lung disease, musculoskeletal disease, cancer, and depression.

*p < .05 for differences among groups, analyzed with chi-squared tests for categorical variables or 1-way analysis of variance for continuous variables.

Potential confounders

At baseline, we collected information on sociodemographic and lifestyle characteristics including sex, age, educational level (primary or less, secondary, or university), tobacco smoking (never, former, or current), alcohol consumption (never, former, moderate [<24 g/day in women and <40 g/day in men], or heavy), leisure-time physical activity (in metabolic equivalents of task-hour/week), sedentary behavior (usual time spent watching TV), and energy intake (in kcal/day). Morbidity was ascertained by asking the study participants if they had been previously diagnosed with diabetes, cardiovascular disease (myocardial infarction, stroke, or heart failure), chronic lung disease (asthma or chronic obstructive pulmonary disease), musculoskeletal disease (osteoarthritis or arthritis), cancer, or depression requiring treatment. Also, we measured weight and height in standardized conditions (24) to calculate the body mass index as the weight (in kg) divided by the squared height (in m). Finally, we assessed drug treatments, including analgesics, based on participants’ self-reports and checks of the reported treatment against the drug packages kept at home.

Statistical Analysis

In the Seniors-ENRICA-1 cohort, from the initial sample of 2519 participants at baseline, we excluded 95 who had died at follow-up, 600 who were lost to follow-up, 180 who had no information on pain at baseline or follow-up, 166 with no diet history or implausible energy intake (<800 kcal or >5000 kcal for men, and <500 kcal or >4000 kcal for women) at baseline, and 2 with missing information on potential confounders at baseline. From the remaining 1476 participants, we also excluded 612 who reported no pain at baseline and follow-up, because we were interested in how adherence to the Mediterranean diet influenced pain evolution. It is of note that individuals without pain were younger and had better lifestyle habits and health than study participants (Supplementary Table 1), so it is unlikely that excluding them may lead to overestimation of the study associations. Thus, the analytical sample in the Seniors-ENRICA-1 comprised 864 individuals.

In the Seniors-ENRICA-2 cohort, from the initial sample of 3273 participants at baseline, we excluded 46 who had died at follow-up, 1333 who were lost to follow-up, 56 who had no information on pain at baseline or follow-up, 16 with no diet history or implausible energy intake at baseline, and 9 with missing information on potential confounders at baseline. From the remaining 1813 participants, we also excluded 951 who reported no pain at both baseline and follow-up. Thus, the analytical sample in the Seniors-ENRICA-2 comprised 862 individuals.

Differences in the participants’ baseline characteristics among quartiles of adherence to the Mediterranean diet were analyzed with chi-squared tests for categorical variables or 1-way analysis of variance for continuous variables.

The association of the MEDAS score at baseline with categories of change in the pain scale from baseline to follow-up in each cohort was summarized with relative risk ratios (RRRs) and their 95% confidence intervals (CIs), obtained from multinomial logistic regression. The MEDAS score was modeled in quartiles, and the lowest one was used as reference. Two models were tested: Model 1 adjusted for sex, age, and educational level; Model 2 further adjusted for tobacco smoking, alcohol consumption, leisure-time physical activity, time watching TV, body mass index, energy intake, number of chronic diseases, and use of analgesics at baseline. The association of the MEDAS score at baseline with change in the pain scale score from baseline to follow-up (as a continuous variable) was also summarized in each cohort with β coefficients and their 95% CIs, obtained from linear regression. Again, 2 models were built: Model 3 adjusted as Model 1, and Model 4 adjusted as Model 2. The associations of each item of the MEDAS score with categorized or continuous changes in the pain scale were assessed similarly, as well as the associations of the MEDAS score with changes in each component of the pain scale (frequency, severity, and locations of pain).

To determine whether the associations were modified by the cohort, we tested interaction terms defined as the product of categories of the MEDAS score and the cohort. Because no interactions were present, we replicated the analyses in the pooled cohorts adjusting for the cohort. Besides, as a sensitivity analysis, we conducted a random effects meta-analyses of the results obtained in each cohort separately. We also assessed whether study results varied by sex by testing interaction terms defined as the product of categories of the MEDAS score and sex. Since no interactions were found, results for men and women are presented combined. Lastly, we replicated the analyses including those individuals who reported no pain at baseline and follow-up. Statistical significance was set at 2-sided p value < .05. Analyses were performed with Stata, version 15.

Results

Participants in the Seniors-ENRICA-1 cohort had a mean age of 71.3 years and 38.5% were men. The corresponding figures in the Seniors-ENRICA-2 were 71.8 years and 35%. In the Seniors-ENRICA-1 cohort, participants with higher MEDAS scores were more physically active and less sedentary, and had a lower body mass index and a higher energy intake at baseline than those with lower scores; in the Seniors-ENRICA-2 cohort, participants with higher MEDAS scores also had a higher energy intake and were less often smokers (Table 1).

Compared with participants in the lowest quartile of the MEDAS score at baseline, those in the highest quartile had a higher frequency of pain improvement versus pain worsening over the follow-up. This association was not statistically significant when the 2 cohorts were analyzed separately (fully adjusted RRR [95% CI]: 1.48 [0.93, 2.35] in the Seniors-ENRICA-1, and 1.39 [0.85, 2.28] in the Seniors-ENRICA-2), but it reached statistical significance when they were pooled (1.43 [1.03, 1.99]). A decrease in the pain scale over the follow-up was also observed in participants in the highest versus the lowest quartile of the MEDAS score when the cohorts were pooled, although it was not statistically significant (β [95% CI]: −0.35 [−0.77, 0.07]). No association was evident for adherence to the Mediterranean diet with the frequency of no change in the pain scale versus pain worsening (Table 2). Similar results were obtained using random effects meta-analysis (Supplementary Table 2) and when including in the analyses those individuals who reported no pain at baseline and follow-up (RRR [95% CI]: 1.38 [1.01, 1.89] of pain improvement versus pain worsening for the highest versus the lowest quartile of the MEDAS score in the pooled cohorts). Also, adjusting or not for prevalent or incident comorbidities did not materially change the results (Supplementary Table 3). Among incident chronic conditions, only musculoskeletal disease was independently associated with a lower frequency of pain improvement versus worsening (RRR [95% CI]: 0.57 [0.39, 0.82]), and with an increase in the pain scale (β [95% CI]: 0.59 [0.13, 1.05]) over time.

Table 2.

Association of Adherence to the Mediterranean Diet With Changes in the Pain Scale Over 2.6 Years in Older Adults Participating in the Seniors-ENRICA-1 and Seniors-ENRICA-2 Cohorts

	MEDAS Score (quartiles)†
	Q1	Q2	Q3	Q4 (highest)	p Trend
Seniors-ENRICA-1 cohort (n = 864)
No change in the pain scale vs pain worsening
cases/n	54/252	39/197	34/186	33/229
Model 1: RRR (95% CI)‡	Ref.	0.98 (0.57, 1.65)	0.88 (0.51, 1.51)	0.73 (0.42, 1.25)	.24
Model 2: RRR (95% CI)§	Ref.	0.90 (0.49, 1.63)	0.96 (0.52, 1.78)	0.89 (0.49, 1.63)	.77
Pain improvement vs pain worsening
cases/n	101/252	89/197	82/186	116/229
Model 1: RRR (95% CI)‡	Ref.	1.23 (0.81, 1.88)	1.11 (0.73, 1.70)	1.36 (0.91, 2.04)	.19
Model 2: RRR (95% CI)§	Ref.	1.07 (0.66, 1.73)	1.20 (0.73, 1.96)	1.48 (0.93, 2.35)	.09
Change in the pain scale
n	252	197	186	229
Model 3: β (95% CI)||	Ref.	−0.37 (−1.07, 0.34)	−0.19 (−0.91, 0.52)	−0.49 (−1.16, 0.19)	.22
Model 4: β (95% CI)¶	Ref.	−0.12 (−0.73, 0.50)	−0.24 (−0.87, 0.39)	−0.55 (−1.15, 0.05)	.07
Seniors-ENRICA-2 cohort (n = 862)
No change in the pain scale vs pain worsening
cases/n	38/284	37/212	32/191	27/175
Model 1: RRR (95% CI)‡	Ref.	1.35 (0.80, 2.30)	1.30 (0.75, 2.26)	1.50 (0.84, 2.69)	.18
Model 2: RRR (95% CI)§	Ref.	1.40 (0.77, 2.55)	0.96 (0.51, 1.81)	1.70 (0.88, 3.28)	.27
Pain improvement vs pain worsening
cases/n	115/284	81/212	74/191	79/175
Model 1: RRR (95% CI)‡	Ref.	1.00 (0.68, 1.48)	0.98 (0.66, 1.47)	1.36 (0.90, 2.05)	.22
Model 2: RRR (95% CI)§	Ref.	0.98 (0.62, 1.57)	0.72 (0.44, 1.17)	1.39 (0.85, 2.28)	.51
Change in the pain scale
n	284	212	191	175
Model 3: β (95% CI)||	Ref.	−0.05 (−0.73, 0.62)	−0.25 (−0.94, 0.45)	−0.32 (−1.04, 0.40)	.32
Model 4: β (95% CI)¶	Ref.	0.02 (−0.54, 0.57)	0.22 (−0.36, 0.79)	−0.18 (−0.77, 0.42)	.81
Pooled cohorts (n = 1726)
No change in the pain scale vs pain worsening
cases/n	92/536	76/409	66/377	60/404
Model 1: RRR (95% CI)‡	Ref.	1.12 (0.78, 1.63)	1.06 (0.72, 1.55)	1.00 (0.68, 1.49)	1.00
Model 2: RRR (95% CI)c	Ref.	1.10 (0.73, 1.67)	0.95 (0.62, 1.47)	1.16 (0.75, 1.80)	.65
Pain improvement vs pain worsening
cases/n	216/536	170/409	156/377	195/404
Model 1: RRR (95% CI)‡	Ref.	1.10 (0.83, 1.46)	1.05 (0.78, 1.40)	1.34 (1.01, 1.78)*	.08
Model 2: RRR (95% CI)§	Ref.	1.01 (0.73, 1.41)	0.92 (0.66, 1.30)	1.43 (1.03, 1.99)*	.08
Change in the pain scale
n	536	409	377	404
Model 3: β (95% CI)||	Ref.	−0.19 (−0.68, 0.30)	−0.22 (−0.72, 0.28)	−0.37 (−0.86, 0.12)	.14
Model 4: β (95% CI)¶	Ref.	−0.01 (−0.43, 0.40)	−0.01 (−0.43, 0.42)	−0.35 (−0.77, 0.07)	.14

Notes: CI = confidence interval; MEDAS = Mediterranean Diet Adherence Screener; RRR = relative risk ratio.

^†MEDAS score quartiles: Q1: ≤6, Q2: 7, Q3: 8, Q4: ≥9.

^‡Model 1: Multinomial logistic regression model adjusted for sex, age, educational level (primary or less, secondary, or university), and cohort in the pooled analyses.

^§Model 2: As Model 1 and further adjusted for smoking status (never, former, or current), alcohol consumption (never, former, moderate, heavy), leisure-time physical activity (MET-hours/week [tertiles]), sedentary behavior (TV hours/day [tertiles]), body mass index (<25, ≥25 to <30, ≥30 kg/m²), energy intake (kcal/day [tertiles]), number of chronic diseases (including diabetes, cardiovascular disease, chronic lung disease, musculoskeletal disease, cancer, and depression), and use of analgesics at baseline.

^||Model 3: Linear regression model adjusted for sex, age, educational level (primary or less, secondary, or university), and cohort in the pooled analyses. Positive values indicate pain worsening, whereas negative values indicate pain improvement.

^¶Model 4: As Model 3 and further adjusted for smoking status (never, former, or current), alcohol consumption (never, former, moderate, heavy), leisure-time physical activity (MET-hours/week [tertiles]), sedentary behavior (TV hours/day [tertiles]), body mass index (<25, ≥25 to <30, ≥30 kg/m²), energy intake (kcal/day [tertiles]), number of chronic diseases (including diabetes, cardiovascular disease, chronic lung disease, musculoskeletal disease, cancer, and depression), and use of analgesics at baseline.

*p < .05.

In the pooled cohorts, a higher adherence to the Mediterranean diet was also associated with a higher frequency of improvement versus worsening of pain severity (RRR [95% CI] for the highest versus the lowest quartile of the MEDAS score: 1.43 [1.01, 2.04]), and pain locations (1.54 [1.08, 2.20]), and with a tendency to pain frequency improvement (1.34 [0.92, 1.93]) (Table 3). No associations were found for the frequency of no change versus worsening of the pain scale components.

Table 3.

Association of Adherence to the Mediterranean Diet With Changes in the Components of the Pain Scale Over 2.6 Years in Older Adults Participating in the Seniors-ENRICA-1 and Seniors-ENRICA-2 Cohorts (Pooled Cohorts, n = 1726)

	MEDAS Score (quartiles)†
	Q1	Q2	Q3	Q4 (highest)	p Trend
Pain frequency
No change vs worsening
cases/n	161/536	136/409	115/377	109/404
RRR (95% CI)‡	Ref.	1.36 (0.91, 2.02)	1.00 (0.67, 1.51)	1.22 (0.81, 1.84)	.54
Improvement vs worsening
cases/n	176/536	140/409	131/377	157/404
RRR (95% CI)‡	Ref.	1.18 (0.81, 1.72)	0.96 (0.66, 1.42)	1.34 (0.92, 1.93)	.24
Change in pain frequency
n	536	409	377	404
β (95% CI)§	Ref.	−0.02 (−0.18, 0.13)	−0.01 (−0.18 0.15)	−0.10 (−0.26, 0.06)	.26
Pain severity
No change vs worsening
cases/n	161/536	121/409	108/377	89/404
RRR (95% CI)‡	Ref.	0.89 (0.60, 1.32)	0.87 (0.58, 1.31)	0.94 (0.62, 1.42)	.75
Improvement vs worsening
cases/n	185/536	141/409	140/377	185/404
RRR (95% CI)‡	Ref.	0.84 (0.58, 1.21)	0.92 (0.64, 1.34)	1.43 (1.01, 2.04)*	.06
Change in pain severity
n	536	409	377	404
β (95% CI)§	Ref.	0.05 (−0.10, 0.20)	0.01 (−0.14, 0.16)	−0.15 (−0.30, 0.00)	.07
Pain locations
No change vs worsening
cases/n	128/536	112/409	98/377	99/404
RRR (95% CI)‡	Ref.	1.34 (0.91, 1.98)	1.17 (0.78, 1.74)	1.50 (1.01, 2.24)	.08
Improvement vs worsening
cases/n	196/536	157/409	147/377	171/404
RRR (95% CI)‡	Ref.	1.19 (0.83, 1.70)	1.10 (0.77, 1.59)	1.54 (1.08, 2.20)*	.03
Change in pain locations
n	536	409	377	404
β (95% CI)§	Ref.	−0.04 (−0.17, 0.10)	−0.00 (−0.14, 0.13)	−0.10 (−0.24, 0.03)	.22

Notes: CI = confidence interval; MEDAS = Mediterranean Diet Adherence Screener; RRR = relative risk ratio.

^†MEDAS score quartiles: Q1: ≤6, Q2: 7, Q3: 8, Q4: ≥9.

^‡Multinomial logistic regression model adjusted for sex, age, educational level (primary or less, secondary, or university), cohort, smoking status (never, former, or current), alcohol consumption (never, former, moderate, heavy), leisure-time physical activity (MET-hours/week [tertiles]), sedentary behavior (TV hours/day [tertiles]), body mass index (<25, ≥25 to <30, ≥ 30 kg/m²), energy intake (kcal/day [tertiles]), number of chronic diseases (including diabetes, cardiovascular disease, chronic lung disease, musculoskeletal disease, cancer, and depression), and use of analgesics at baseline.

^§Linear regression model adjusted for the same variables.

*p < .05.

Lastly, when we explored the relationship between individual components of the MEDAS score at baseline and changes in the pain scale over the follow-up, the items that were associated with higher frequency of pain improvement versus worsening in the pooled cohorts were high intake of vegetables (RRR [95% CI]: 1.41 [1.04, 1.91]) and fruit (1.38 [1.06, 1.79]) and low consumption of sugar-sweetened beverages (1.64 [1.06, 2.53]) (Table 4).

Table 4.

Association of Each Item of the MEDAS Score With Changes in the Pain Scale Over 2.6 Years in Older Adults Participating in the Seniors-ENRICA-1 and Seniors-ENRICA-2 Cohorts (Pooled Cohorts, n = 1726)

	No Change in the Pain Scale vs Pain Worsening	Pain Improvement vs Pain Worsening	Change in the Pain Scale
	RRR (95% CI)†	RRR (95% CI)†	β (95% CI)‡
Olive oil as the main fat for cooking (Yes vs No)	0.68 (0.42, 1.09)	0.84 (0.58, 1.21)	0.15 (−0.33, 0.61)
Olive oil consumption (≥4 vs <4 tablespoons/day)	1.06 (0.63, 1.79)	1.02 (0.69, 1.50)	0.13 (−0.36, 0.63)
Vegetables (≥2 vs <2 servings/day)	1.45 (0.97, 2.15)	1.41 (1.04, 1.91)*	−0.32 (−0.71, 0.07)
Fruit (≥3 vs <3 servings/day)	1.03 (0.73, 1.45)	1.38 (1.06, 1.79)*	−0.30 (−0.63, 0.03)
Red and processed meats (<1 vs ≥1 servings/day)	0.89 (0.52, 1.51)	0.86 (0.58, 1.29)	0.02 (−0.49, 0.53)
Butter, margarine or cream (<1 vs ≥1 servings/day)	0.51 (0.27, 0.97)*	0.67 (0.39, 1.16)	0.42 (−0.26, 1.09)
Sugar-sweetened beverages (<1 vs ≥1 cups/day)	1.39 (0.81, 2.38)	1.64 (1.06, 2.53)*	−0.50 (−1.04, 0.04)
Wine (≥7 vs <7 servings/week)	0.81 (0.50, 1.30)	0.97 (0.70, 1.34)	0.15 (−0.27, 0.57)
Legumes (≥3 vs <3 servings/week)	0.99 (0.59, 1.68)	1.02 (0.68, 1.53)	−0.21 (−0.72, 0.30)
Fish or seafood (≥3 vs <3 servings/week)	1.07 (0.75, 1.52)	1.03 (0.78, 1.35)	0.07 (−0.27, 0.41)
Commercial pastries (<2 vs ≥2 servings/week)	1.24 (0.91, 1.70)	1.17 (0.92, 1.50)	−0.20 (−0.51, 0.11)
Nuts (≥3 vs <3 servings/week)	0.89 (0.64, 1.24)	0.99 (0.77, 1.27)	−0.05 (−0.36, 0.27)
Preferring white meat over red meat (Yes vs No)	1.07 (0.75, 1.51)	1.28 (0.98, 1.69)	−0.18 (−0.52, 0.16)
Sofrito (≥2 vs <2 servings/week)	0.79 (0.51, 1.23)	0.89 (0.64, 1.25)	0.06 (−0.37, 0.49)

Notes: CI = confidence interval; MEDAS = Mediterranean Diet Adherence Screener; RRR = relative risk ratio.

^†Multinomial logistic regression model adjusted for sex, age, educational level (primary or less, secondary, or university), cohort, smoking status (never, former, or current), alcohol consumption (never, former, moderate, heavy), leisure-time physical activity (MET-hours/week [tertiles]), sedentary behavior (TV hours/day [tertiles]), body mass index (<25, ≥25 to <30, ≥30 kg/m²), energy intake (kcal/day [tertiles]), number of chronic diseases (including diabetes, cardiovascular disease, chronic lung disease, musculoskeletal disease, cancer, and depression), and use of analgesics at baseline.

^‡Linear regression model adjusted for the same variables.

*p < .05.

Discussion

In these 2 cohorts of older adults in Spain, a higher adherence to the Mediterranean diet at baseline (adjusting for sociodemographic, lifestyle, and clinical characteristics) was associated with a higher frequency of pain improvement over 2.6 years. All 3 characteristics of pain (severity, frequency, and number of sites) improved with increasing adherence, although the association with pain frequency improvement was not statistically significant. The main contributors to the observed associations were high consumption of fruit and vegetables, and low consumption of sugar-sweetened beverages.

There is evidence of the effect of the Mediterranean diet on the reduction of inflammatory markers, mainly C-reactive protein, interleukin-6, and several adhesion molecules (11–13). Also, a higher adherence to this dietary pattern has been previously associated with lower risk of inflammatory conditions, such as symptomatic knee osteoarthritis and associated pain worsening (25), whereas a higher inflammatory potential of diet has been linked with poor health outcomes, including higher mortality risk (26). Therefore, the anti-inflammatory properties of the Mediterranean diet may have played an important role in the evolution of pain in the study participants.

Another potential pathway for the Mediterranean diet to improve pain might be the reduction of the risk of chronic disease. The beneficial effects of the Mediterranean diet on cardiovascular disease (27,28), type 2 diabetes (29), obesity (30), depression, (31) and frailty (32–34) are widely recognized. These conditions are, in turn, associated with higher risk of chronic pain (35–43). Also, there is evidence that multi-morbidity is independently associated with chronic pain (44). Potential mechanisms for these benefits of the Mediterranean diet may include decreased levels of inflammation, oxidation, endothelial dysfunction, and insulin resistance (14,45), which would prevent chronic disease and, consequently, lower the risk of pain. However, in our study, the association of a higher adherence to the Mediterranean diet with pain improvement did remain after adjustment for prevalent or incident chronic conditions, suggesting that the effect of this dietary pattern is independent of those chronic diseases, including musculoskeletal disease, at least during the short follow-up of our cohorts. In fact, although incident musculoskeletal disease was independently associated with a lower frequency of pain improvement versus worsening and with an increase in the pain scale over time, and these associations were rather strong, this did not prevent us to find a beneficial effect of the Mediterranean diet on pain change when adjusting for this variable.

As regards individual foods and nutrients, a higher consumption of fruit and vegetables, which are rich in phenolic antioxidants, has been previously associated with lower levels of inflammatory biomarkers and improved immune cells (12,46). Also, monounsaturated fatty acids from virgin olive oil and omega-3 fatty acids from fish have been related to reduced inflammation, whereas saturated fats have shown pro-inflammatory effects (47,48). Thus, higher intakes of red and processed meats may increase oxidative stress and inflammation (49,50). Lastly, low glycemic index and glycemic load diets had been found to reduce levels of inflammation (51,52). The main drivers of the favorable association of the Mediterranean diet with pain changes found in our study, that is, high intake of fruit and vegetables, and low intake of sugar-sweetened beverages, are in line with this evidence. However, associations with pain changes for the remainder components were not apparent. This can be explained by the fact that the effects of individual foods and nutrients may be too small to be detected, and they may interact with each other (53). Since foods and nutrients are consumed in combination, using dietary patterns may be a more adequate approach to study associations with health. Thus, the effects of the Mediterranean diet may be the composite effect of high concentrations of beneficial substances with antioxidant and anti-inflammatory properties found in the different components, so that the effect of the dietary pattern is greater than that of each single food or nutrient.

We are not aware of previous population-based studies examining the effects of the Mediterranean diet on pain changes over time. Therefore, our study adds an important contribution to the body of knowledge about modifiable pain risk factors. However, our results should be confirmed by future studies examining the effects of the Mediterranean diet, as well as of other healthy dietary patterns, on the development and evolution of pain in older adults.

The main strengths of this study are its prospective design, using a validated diet history (19), and adjusting for many potential confounders. Among the limitations is the self-reported nature of dietary information, which may have led to some degree of misclassification, and of information on morbidities. Besides, the pain questionnaire has not been validated, although the items were similar to other widely used questionnaires, and information on the etiology or the type of pain was not available. In addition, the study did not include biomarkers of inflammation. Also, as in any observational study, we cannot entirely rule out residual confounding. Lastly, the high rate of losses to follow-up may have biased the study results; however, because study participants were younger and had a higher educational level and similar lifestyle habits and health at baseline than those excluded from the analyses because of losses to follow-up or missing data (Supplementary Table 1), it is unlikely that our results overestimate the actual study associations.

In conclusion, a higher adherence to the Mediterranean diet was related to a subsequent improvement of pain and its main characteristics in older adults, suggesting that improving diet quality may help reduce the high health impact of pain.

Funding

This work was supported by Instituto de Salud Carlos III, State Secretary of R+D+I and FEDER/FSE (FIS grants 16/609, 16/1512, 18/287, 19/319); CIBERESP (16/01); JPI-A Healthy Diet for a Healthy Life, State Secretary of R+D+I (PCIN-2016-145); and Cátedra de Epidemiología y Control del Riesgo Cardiovascular at UAM (#820024). The funding agencies had no role in study design, data collection and analysis, interpretation of results, manuscript preparation, or the decision to submit this manuscript for publication.

Conflict of Interest

None declared.

Author Contributions

R.O. and F.R.-A. designed the research. R.O. performed the statistical analyses. All authors contributed to results interpretation. R.O. and F.R.-A. drafted the manuscript. All authors reviewed the manuscript for important intellectual content, read, and approved the final manuscript.