Advertisement

Low magnesium intake is associated with increased knee pain in subjects with radiographic knee osteoarthritis: data from the Osteoarthritis Initiative

Open ArchivePublished:February 15, 2018DOI:https://doi.org/10.1016/j.joca.2018.02.002

      Summary

      Objective

      As magnesium mediates bone and muscle metabolism, inflammation, and pain signaling, we aimed to evaluate whether magnesium intake is associated with knee pain and function in radiographic knee osteoarthritis (OA).

      Methods

      We investigated the associations between knee pain/function metrics and magnesium intake from food and supplements in 2548 Osteoarthritis Initiative cohort participants with prevalent radiographic knee OA (Kellgren–Lawrence score ≥2). Magnesium intake was assessed by Food Frequency Questionnaire (FFQ) at baseline. WOMAC and Knee Injury and Osteoarthritis Outcome Score (KOOS) scores were reported annually with total follow up of 48 months. Analyses used linear mixed models.

      Results

      Among participants with baseline radiographic knee OA the mean total magnesium intake was 309.9 mg/day (SD 132.6) for men, and 287.9 mg/day (SD 118.1) for women, with 68% of men and 44% of women below the estimated average requirement. Subjects with lower magnesium intake had worse knee OA pain and function scores, throughout the 48 months (P < 0.001). After adjustment for age, sex, race, body mass index (BMI), calorie intake, fiber intake, pain medication use, physical activity, renal insufficiency, smoking, and alcohol use, lower magnesium intake remained associated with worse pain and function outcomes (1.4 points higher WOMAC and 1.5 points lower KOOS scores for every 50 mg of daily magnesium intake, P < 0.05). Fiber intake was an effect modifier (P for interaction <0.05). The association between magnesium intake and knee pain and function scores was strongest among subjects with low fiber intake.

      Conclusion

      Lower magnesium intake was associated with worse pain and function in knee OA, especially among individuals with low fiber intake.

      Keywords

      Introduction

      At least 10% of the United States population aged 60 years or older have symptomatic knee osteoarthritis (OA)
      • Zhang Y.
      • Jordan J.M.
      Epidemiology of osteoarthritis.
      . In a study from the United Kingdom, 15% of subjects in the general population aged over 55 years have had restricted activity because of knee pain occurring on most days in 1 month during the preceding year
      • McAlindon T.E.
      • Cooper C.
      • Kirwan J.R.
      • Dieppe P.A.
      Knee pain and disability in the community.
      .
      Magnesium is the second most abundant intracellular cation, and is a critical cofactor for any reaction powered by ATP. It also acts as a calcium channel antagonist and thus plays an important role in activities regulated by intracellular calcium concentration fluxes such as muscle contraction and insulin release
      • Sanders G.T.
      • Huijgen H.J.
      • Sanders R.
      Magnesium in disease: a review with special emphasis on the serum ionized magnesium.
      . Additionally, magnesium likely plays a role as a pain mediator, and has been shown to alter the levels of inflammatory cytokines and neurotransmitters in human and animal models
      • Simental-Mendía L.E.
      • Rodríguez-Morán M.
      • Guerrero-Romero F.
      Oral magnesium supplementation decreases C-reactive protein levels in subjects with prediabetes and hypomagnesemia: a clinical randomized double-blind placebo-controlled trial.
      • Weglicki W.B.
      • Phillips T.M.
      Pathobiology of magnesium deficiency: a cytokine/neurogenic inflammation hypothesis.
      . Magnesium intake comes primarily from ingested food and water sources. Approximately half of the US population has been shown to consume less than the daily requirement of magnesium from foods
      • McGuire S.
      Scientific report of the 2015 Dietary Guidelines Advisory Committee. Washington, DC: US Departments of Agriculture and Health and Human Services, 2015.
      . This is likely related to the increased consumption of processed foods, intensification of farming practices, and increased intake of filtered water vs ground water, all of which have, or result in, decreased magnesium content and intake. The prevalence of magnesium deficiency in arthritis patients is not known, however we recently found low serum magnesium concentrations in 76 of 200 patients (38%) in a Rheumatology outpatient clinic
      • Ghattaura N.
      • Murphy C.
      • Valen P.
      High prevalence of hypomagnesemia and its relation to BMI, type 2 diabetes, and clinical disease measures in a Va outpatient rheumatology clinic population: abstract number: 2260.
      .
      The etiology of OA is complex and likely multi-factorial. There is some evidence to support a link between low magnesium intake and the likelihood of knee OA, but previous studies have been limited by a cross-sectional or case-control design. In a cross-sectional analysis in a population-based cohort, a modest inverse threshold association was found between dietary magnesium intake and knee OA in Caucasians, but not African Americans
      • Qin B.
      • Shi X.
      • Samai P.S.
      • Renner J.B.
      • Jordan J.M.
      • He K.
      Association of dietary magnesium intake with radiographic knee osteoarthritis: results from a population-based study.
      . In a case-control twin study to assess the relationship between radiographic knee OA and both bone turnover and calcium regulation, the discordant twin pair analysis revealed a significant reduction in the odds of having radiographic OA with a higher magnesium level
      • Hunter D.J.
      • Hart D.
      • Snieder H.
      • Bettica P.
      • Swaminathan R.
      • Spector T.D.
      Evidence of altered bone turnover, vitamin D and calcium regulation with knee osteoarthritis in female twins.
      . In a cross-sectional study of 2855 subjects, there was a significant association between serum magnesium concentration and radiographic knee OA after adjustment for age, sex, and body mass index (BMI)
      • Zeng C.
      • Wei J.
      • Li H.
      • Yang T.
      • Zhang F.
      • Pan D.
      • et al.
      Relationship between serum magnesium concentration and radiographic knee osteoarthritis.
      .
      Based on the data suggesting a possible association between magnesium and knee osteoarthritis, as well as a potential role for magnesium in OA pain, we investigated the association between magnesium intake and knee pain scores in a large prospective cohort of patients with prevalent radiographic knee OA. Our primary objective was to evaluate the associations of magnesium intake from diet and supplements with measures of knee pain and function over 4 years of follow up. Our secondary objective was to estimate the prevalence of insufficient magnesium intake in this knee OA cohort.

      Methods

      Study design

      A prospective cohort study.

      Participants

      We used existing data from the Osteoarthritis Initiative, an ongoing multi-center cohort study of knee OA in the United States. 4796 subjects with or at risk for knee OA were recruited between 2006 and 2010 and followed annually with detailed clinical assessments and questionnaires. For this analysis, subjects with radiographic knee OA in at least one knee at baseline (Kellgren and Lawrence score ≥2) were selected (N = 2548). Subjects with knee replacements at baseline were excluded (N = 58).

      Variables

      Dietary information was obtained by Block Brief 2000 food frequency questionnaire (FFQ) at baseline. This questionnaire collects usual intake information on 72 common food items, as well as questions about regularly used vitamin supplements (multivitamins, individual vitamin supplements, and antacid combinations (Tums)). Magnesium intake from self-reported dietary components and supplements (exposure) was calculated separately for OAI by NutritionQuest, the makers of the Block Brief 2000 FFQ. Cumulative magnesium intake from diet and supplements was analyzed in calorie-adjusted sex specific quintiles (Fig. 1), and as a continuous calorie-adjusted variable (milligrams per 1000 calories). Magnesium intake from supplements was also analyzed separately, without calorie adjustment. Total fiber intake from foods was also estimated in grams per 1000 calories. Outcomes included self-reported annual Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and its pain and function subscales, as well as Knee Injury and Osteoarthritis Outcome Score (KOOS), measured at baseline and every year, up to 4 years of follow up. WOMAC total score characterizes average knee pain, stiffness, and functional limitations in the past 7 days, and was assessed on a 0–96 point scale, with higher scores indicative of worse pain and lower function. WOMAC pain and function/disability subscale scores ranged from 0 to 20 points and 0 to 68 points respectively. The KOOS pain questionnaire is a brief assessment tool that evaluates knee pain with twisting, bending, and straightening the knee in the past 7 days, scores range from 0 to 100 with lower scores indicative of worse pain. While there is a high agreement between WOMAC and KOOS, KOOS may be more sensitive in younger and more active subjects
      • McConnell S.
      • Kolopack P.
      • Davis A.M.
      The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC): a review of its utility and measurement properties.
      • Roos E.M.
      • Roos H.P.
      • Lohmander L.S.
      • Ekdahl C.
      • Beynnon B.D.
      Knee Injury and Osteoarthritis Outcome Score (KOOS)–development of a self-administered outcome measure.
      , hence we reported both measures. If the subject had one knee with radiographic OA, pain and function outcomes in the affected knee were followed for the entire study period. If both knees had radiographic OA, the knee with worse pain and function scores was picked at each time point.
      Figure thumbnail gr1
      Fig. 1Study diagram with magnesium intake quintiles. OA – osteoarthritis. KL – Kellgren and Lawrence score. Q1–Q5 – sex specific, calorie-adjusted quintiles of total magnesium intake from diet and supplements, based on the FFQ at baseline. WOMAC – Western Ontario and McMaster Universities Osteoarthritis Index. KOOS – Knee Injury and Osteoarthritis Outcome Score. PASE – Physical Activity Scale for the Elderly.

      Addressing bias

      Potential confounders were selected based on previously reported risk factors for increased OA pain (age, sex, race, BMI, OA radiographic severity, pain medication use on the day of assessment, physical activity metrics), and factors associated with altered magnesium excretion (history of renal insufficiency, alcohol use, and smoking status). We additionally adjusted for total calorie and fiber intake, as dietary markers of overall healthy diet
      • Park Y.
      • Subar A.F.
      • Hollenbeck A.
      • Schatzkin A.
      Dietary fiber intake and mortality in the NIH-AARP diet and health study.
      • Gil Á.
      • Martinez de Victoria E.
      • Olza J.
      Indicators for the evaluation of diet quality.
      , that could potentially confound magnesium effects. Age, sex, and race were self-reported by subjects at baseline. BMI was measured at each annual clinical visit. OA radiographic severity was assessed by Kellgren and Lawrence scores from flexed knee radiographs at baseline. Pain medication use was ascertained at each annual study visit. Subjects were asked whether they took any pain medication on the day of visit, including both prescription and over-the-counter pain medication for any type of pain. Self-reported alcohol use, smoking status, history of renal insufficiency, and physical activity (by Physical Activity Scale for the Elderly)
      • Washburn R.A.
      • Smith K.W.
      • Jette A.M.
      • Janney C.A.
      The Physical Activity Scale for the Elderly (PASE): development and evaluation.
      were available from standardized annual questionnaires.

      Statistical methods

      After pre-processing, data for unadjusted analyses were available on 2548 subjects. Full dietary assessment, outcomes and covariates data during the entire follow up period were available on 2253 subjects. The rest contributed shorter follow up. ANOVA, repeated measures ANOVA, and Chi-square tests were used for descriptive statistics. Linear mixed effects models with limited and multivariable adjustment were used for longitudinal analyses. Time-varying covariates at each annual visit included BMI, pain medication use, and physical activity. Appropriateness of linear mixed effects models was confirmed by analysis of residuals. A 95% confidence level was set for all tests of significance. All statistical analyses were performed in SAS 9.4 (SAS Institute, Inc., Cary, NC).

      Results

      The majority of the 2548 Osteoarthritis Initiative subjects with baseline radiographic knee OA (analytical cohort) were between 45 and 69 years old (72.7%), female (57.8%), Caucasian (77.2%) and overweight (mean BMI 29.6). The top sources of dietary magnesium in the OAI were vegetables (carrots, green beans, broccoli), beans, milk, cold and cooked cereals, and white potatoes. 61% of subjects reported magnesium intake from dietary supplements. 46% 100 mg/day, 15% between 30 and 70 mg/day. Mean total magnesium intake from diet and supplements in the cohort was 297.2 ([standard deviation] 124.9) mg/day and the mean total reported daily calorie intake was 1422 (631) calories. Both magnesium and total calorie intake differed by sex. Mean magnesium intake for women was 287.9 (118.1) mg/day, and for men 309.9 (132.6) mg/day (P < 0.001). Mean calorie intake for women was 1292 (555.6) calories/day and for men 1598 (683.9) calories/day (P < 0.001). Women reported higher magnesium intake per 1000 calories than men, and quintile cutoffs were also higher for women than for men (Fig. 1). A large proportion of subjects with radiographic knee OA reported below recommended magnesium intake. 44.5% of women with OA and 68.1% of men with OA consumed less than the estimated average requirement of magnesium; 64.2% of women with OA and 82.7% of men with OA consumed less than the recommended dietary allowance of magnesium for their age and sex.
      There were differences in demographic and behavioral characteristics across caloric-adjusted sex specific quintiles of magnesium intake (Table I). Subjects in the lowest magnesium intake quintile were more likely to be younger (P < 0.001) and African American (P < 0.001), had higher BMI (P < 0.001), and lower total fiber intake (P < 0.001). They were also more likely to drink eight or more servings of alcohol per week than those in higher magnesium intake quintiles (P = 0.01). There were no significant differences among magnesium intake quintiles in baseline radiographic OA severity (P = 0.9), pain medication use on the day of pain assessment (P = 0.3), smoking status (P = 0.4), history of renal insufficiency (P = 0.6), and physical activity scores (P = 0.9).
      Table IBaseline characteristics of Osteoarthritis Initiative participants with baseline radiographic knee osteoarthritis (KL ≥ 2) in at least one knee (N = 2548)
      Total sample

      N = 2548
      Sex specific magnesium intake quintilesP value
      Q1

      N = 508
      Q2

      N = 510
      Q3

      N = 510
      Q4

      N = 510
      Q5

      N = 510
      Age, %:
      45–591002 (39.3)53.246.538.231.227.5<0.001
      60–69851 (33.4)27.730.235.736.736.7
      70 and older695 (27.3)19.123.326.132.235.9
      Race, %:
      White1972 (77.4)68.473.980.884.379.8<0.001
      Black514 (20.2)30.523.317.112.617.5
      Asian/Other61 (2.4)1.22.82.23.12.8
      BMI, Mean (SD)29.6 (4.8)31.2 (5.2)30.2 (4.8)29.5 (4.7)28.8 (4.3)28.3 (4.4)<0.001
      OA radiographic severity, %:
      KL = 21352 (53.5)54.550.653.354.352.60.9
      KL = 3884 (35.0)33.537.334.533.534.7
      KL = 4292 (11.6)11.811.811.011.011.8
      Alcohol use, %:
      <1 drink/week1485 (58.4)60.054.956.855.664.60.01
      1–7 drinks/week710 (27.9)25.029.428.530.526.1
      8 or more drinks/week349 (13.7)15.015.714.714.09.2
      Smoking status, %:
      Never2014 (79.4)80.880.579.477.878.50.4
      Former446 (17.6)15.616.518.918.018.8
      Current77 (3.0)3.63.01.84.12.8
      Total fiber intake, g/1000 cal, Mean (SD)10.9 (4.2)7.6 (2.1)9.8 (2.9)10.7 (3.3)11.8 (3.7)14.6 (4.7)<0.001
      History of renal insufficiency, %37 (1.5)2.21.01.41.41.40.6
      Pain medication use, %334 (13.1)12.411.012.814.714.90.3
      PASE score, Mean (SD)156.2 (81.2)153.9 (79.9)157.7 (82.5)156.4 (81.2)157.3 (82.6)154.7 (78.7)0.9
      WOMAC pain score:
      Mean (SD)3.9 (3.9)4.8 (4.5)4.0 (3.8)3.7 (3.9)3.6 (3.7)3.3 (3.6)<0.001
      WOMAC function score:
      Mean (SD)12.4 (12.7)15.6 (14.5)12.8 (12.2)11.9 (12.9)11.8 (12.1)10.2 (11.1)<0.001
      WOMAC total score:
      Mean (SD)18.4 (17.6)22.8 (19.9)19.2 (16.9)17.3 (16.9)17.3 (16.9)15.3 (15.5)<0.001
      KOOS pain score:
      Mean (SD)76.0 (20.2)71.5 (22.1)74.8 (19.8)76.9 (20.1)77.4 (19.6)79.3 (18.5)<0.001
      Q1–Q5 – sex specific, calorie-adjusted quintiles of total magnesium intake from diet and supplements, based on the FFQ at baseline. Quintile ranges for women (mg of magnesium per day per 1000 calories): Q1 < 164.7, Q2 164.7–208.6, Q3 208.7–248.1, Q4 248.2–293.7, Q5 > 293.7. Quintile ranges for men Q1 < 145.9, Q2 145.9–177.2, Q3 177.3–210.5, Q4 210.6–253.5, Q5 > 253.5. SD – standard deviation. OA – osteoarthritis. KL – Kellgren and Lawrence score. PASE – Physical Activity Scale for the Elderly. WOMAC – Western Ontario and McMaster Universities Osteoarthritis Index. KOOS – Knee Injury and Osteoarthritis Outcome Score. P-values presented for Chi-square tests for categorical variables and ANOVA for continuous variables.
      The mean baseline WOMAC scores for the entire sample were 3.9 (median 3.0) for WOMAC pain subscale, 12.4 (median 8.5) for WOMAC function subscale, and 18.4 (median 13.4) for total WOMAC (Table I). The mean KOOS pain score was 76 (median 80.0). These scores correspond to mild knee pain and functional impairment on average. WOMAC and KOOS scores were strongly correlated (Pearson coefficient – 0.92, P < 0.001). Subjects in the lowest magnesium intake quintile had worse WOMAC and KOOS scores than those in the higher quintiles (P < 0.001 for all metrics). This trend persisted throughout the 4-year observation period in a dose-dependent fashion (Fig. 2). While overall mean WOMAC and KOOS scores stayed stable to slightly improved over time, they remained consistently worse in lower magnesium intake quintiles over 4 years (repeated measures ANOVA P-value < 0.001 for all outcomes).
      Figure thumbnail gr2
      Fig. 2Mean knee pain and function scores over 4 years of follow up in Osteoarthritis Initiative participants with baseline radiographic knee osteoarthritis (KL ≥2) across quintiles of magnesium intake. Q1–Q5 – sex specific, calorie-adjusted quintiles of total magnesium intake from diet and supplements, based on the FFQ at baseline. Circles indicate mean pain and function scores reported at annual clinical assessments; color of circle corresponds to quintile of baseline magnesium intake. Vertical bars correspond to standard deviation.
      The association between baseline magnesium intake and knee pain/function outcomes over 4 years of follow up was also observed in linear mixed effects models using calorie-adjusted magnesium intake as a continuous variable (Table II) . Over 48 months 173 subjects (6.8%) were lost to follow up, outcomes and covariates data during the entire follow up period were available on 2253 subjects. After adjustment for age, sex, race, BMI, total calorie intake, total fiber intake, pain medication use, physical activity, renal insufficiency, smoking, and alcohol use, magnesium intake remained significantly associated with pain and function outcomes (Table II). The estimated difference attributed to magnesium intake in fully adjusted models was −0.3 points per 50 mg of magnesium on the WOMAC pain subscale, −1.0 points per 50 mg of magnesium on the WOMAC function subscale, and −1.4 points on the total WOMAC scale. Hence a minimal clinically important difference in WOMAC (20% from the sample mean) corresponded to about 130 mg of magnesium intake per 1000 calories. The difference for KOOS pain score was 1.5 points per 50 mg of magnesium, however the minimal clinically important difference for KOOS is not defined. Full adjusted model estimates are available in Appendix 1.
      Table IIMagnesium intake-attributed differences in pain and function scores in OA knees over 4 years of follow up among Osteoarthritis Initiative participants with baseline radiographic knee osteoarthritis (KL ≥ 2)
      Unadjusted estimateMinimally adjusted estimate
      Adjusted for age, sex, race, total calorie intake.
      Fully adjusted estimate
      Adjusted for age, sex, race, BMI, total calorie intake, pain medication use, physical activity, renal insufficiency, smoking, alcohol use, and fiber intake (including an interaction with fiber intake). Full estimates for all covariates are supplemented in Appendix 1.
      WOMAC pain score (SE)−0.3 (0.0)***−0.2 (0.1)***−0.3 (0.1)**
      WOMAC function score (SE)−1.0 (0.1)***−0.9 (0.2)***−1.0 (0.3)**
      WOMAC total score (SE)−1.5 (0.2)***−1.2 (0.2)***−1.4 (0.5)**
      KOOS pain score (SE)1.7 (0.2)***1.2 (0.2)***1.5 (0.5)**
      Estimates represent a mean difference in knee pain and function scores for every additional 50 mg/1000 cal of magnesium intake. SE – standard error.
      *P < 0.05 **P < 0.01 ***P < 0.001.
      Adjusted for age, sex, race, total calorie intake.
      Adjusted for age, sex, race, BMI, total calorie intake, pain medication use, physical activity, renal insufficiency, smoking, alcohol use, and fiber intake (including an interaction with fiber intake). Full estimates for all covariates are supplemented in Appendix 1.
      In fully adjusted models, fiber intake was an effect modifier for the relationship between magnesium intake and knee pain and function outcomes, as illustrated in Fig. 3 (P for interaction = 0.01 for WOMAC Pain, 0.03 for WOMAC Function, 0.02 for Total WOMAC, and 0.03 for KOOS Pain score). The association between magnesium intake and knee pain and function scores was strongest among subjects who had a low fiber intake, and became less robust in individuals with a higher fiber intake.
      Figure thumbnail gr3
      Fig. 3Fiber intake modifies the effect of magnesium intake on knee pain in Osteoarthritis Initiative participants with baseline radiographic knee osteoarthritis (KL ≥2). The association between magnesium intake and the total WOMAC/KOOS Pain scores is modeled at three levels of fiber intake within the range of the Osteoarthritis Initiative dietary intake data: 5 g of fiber/1000 calories, 10 g of fiber/1000 calories, and 15 g of fiber 1000/calories. Each color represents a regression line for the corresponding fiber intake group with its 95% confidence limits. The association between magnesium intake and knee pain and function scores was strongest among subjects who had a low fiber intake, and became less robust in individuals with a higher fiber intake.
      Dietary magnesium intake and dietary fiber intake were highly correlated (Pearson correlation coefficient 0.82, P < 0.001), while magnesium intake from supplements and dietary fiber intake were not (Pearson correlation coefficient 0.11, P < 0.001). We performed additional analyses to evaluate the associations between magnesium intake from supplements only and pain and function outcomes, without adjustment for fiber intake. We found similar inverse associations between magnesium intake from supplements and WOMAC and KOOS outcomes (Appendix 2).
      In summary, insufficient magnesium intake was highly prevalent among subjects with radiographic knee OA, and was more common in men than in women. Subjects with low magnesium intake had worse knee OA pain and function scores, throughout the 48 month follow up period, even when adjusted for age, sex, race, BMI, total calorie intake, total fiber intake, pain medication use, physical activity, renal insufficiency, smoking, and alcohol use. Fiber intake was an effect modifier. Magnesium effect on knee OA pain and function scores was strongest in subjects with low fiber intake.

      Discussion

      In a cohort of adults with prevalent knee osteoarthritis, we found that low magnesium intake at baseline was associated with worse pain and function in the affected knee over 48 months of follow-up. These results add to previous findings of an association between low magnesium intake and serum levels with prevalent radiographic knee osteoarthritis
      • Qin B.
      • Shi X.
      • Samai P.S.
      • Renner J.B.
      • Jordan J.M.
      • He K.
      Association of dietary magnesium intake with radiographic knee osteoarthritis: results from a population-based study.
      • Hunter D.J.
      • Hart D.
      • Snieder H.
      • Bettica P.
      • Swaminathan R.
      • Spector T.D.
      Evidence of altered bone turnover, vitamin D and calcium regulation with knee osteoarthritis in female twins.
      • Zeng C.
      • Wei J.
      • Li H.
      • Yang T.
      • Zhang F.
      • Pan D.
      • et al.
      Relationship between serum magnesium concentration and radiographic knee osteoarthritis.
      .
      There are several mechanisms by which low magnesium intake may contribute to OA pain, including proinflammatory effects of magnesium deficiency, and its role in pain sensitization. Although OA has traditionally been thought of as a degenerative process primarily affecting hyaline articular cartilage through aberrant mechanical loading, there is evidence that inflammation plays an important role as well
      • Konttinen Y.T.
      • Sillat T.
      • Barreto G.
      • Ainola M.
      • Nordström D.C.E.
      Osteoarthritis as an autoinflammatory disease caused by chondrocyte-mediated inflammatory responses.
      • Goldring M.B.
      • Otero M.
      Inflammation in osteoarthritis.
      . Low magnesium intake has been associated with elevated serum levels of C-reactive protein (CRP). Adults who consumed less than the RDA of magnesium were 1.48–1.75 times more likely to have an elevated CRP than those with normal magnesium intake, after controlling for demographic and cardiovascular risk factors
      • King D.E.
      • Mainous A.G.
      • Geesey M.E.
      • Woolson R.F.
      Dietary magnesium and C-reactive protein levels.
      . Conversely, oral magnesium supplementation decreased hsCRP levels in healthy subjects with prediabetes and hypomagnesemia
      • Simental-Mendía L.E.
      • Rodríguez-Morán M.
      • Guerrero-Romero F.
      Oral magnesium supplementation decreases C-reactive protein levels in subjects with prediabetes and hypomagnesemia: a clinical randomized double-blind placebo-controlled trial.
      . In subjects with metabolic syndrome, severe hypomagnesemia was associated with elevated levels of both CRP and TNF-α
      • Guerrero-Romero F.
      • Bermudez-Peña C.
      • Rodríguez-Morán M.
      Severe hypomagnesemia and low-grade inflammation in metabolic syndrome.
      .
      Hyperalgesia seen in early OA may be related to inflammation as well, likely caused by sensitization of joint afferents by inflammatory mediators. In a rat OA model of hyperalgesia, the inflammatory cytokines TNF and IL-6 were elevated at early stages
      • Orita S.
      • Ishikawa T.
      • Miyagi M.
      • Ochiai N.
      • Inoue G.
      • Eguchi Y.
      • et al.
      Pain-related sensory innervation in monoiodoacetate-induced osteoarthritis in rat knees that gradually develops neuronal injury in addition to inflammatory pain.
      . In more advanced stages of OA, patients show evidence of more widespread pain, suggesting the possibility of central sensitization. The mechanism of central sensitization in OA is not fully understood, but in OA animal models, the spinal content of substance P and calcitonin gene-related peptide (CGRP) is enhanced
      • Ferland C.E.
      • Laverty S.
      • Beaudry F.
      • Vachon P.
      Gait analysis and pain response of two rodent models of osteoarthritis.
      . These neuropeptides facilitate the generation of spinal hyperexcitability
      • Schaible H.-G.
      • Richter F.
      • Ebersberger A.
      • Boettger M.
      • Vanegas H.
      • Natura G.
      • et al.
      Joint pain.
      . There is increasing evidence that CGRP plays an important role not only in central nociceptive transmission and neurogenic inflammation, but also in peripheral mechanisms of OA pain. Knee structures are innervated by CGRP-expressing sensory neurons
      • Fernihough J.
      • Gentry C.
      • Bevan S.
      • Winter J.
      Regulation of calcitonin gene-related peptide and TRPV1 in a rat model of osteoarthritis.
      • Schwab W.
      • Bilgiçyildirim A.
      • Funk R.H.
      Microtopography of the autonomic nerves in the rat knee: a fluorescence microscopic study.
      . Synovial CGRP-immunoreactive nerve fiber density is greater in OA patients with pain compared to symptomatic controls
      • Saxler G.
      • Löer F.
      • Skumavc M.
      • Pförtner J.
      • Hanesch U.
      Localization of SP- and CGRP-immunopositive nerve fibers in the hip joint of patients with painful osteoarthritis and of patients with painless failed total hip arthroplasties.
      . A recent study demonstrated that CGRP plays an important role in maintaining peripheral sensitization in experimentally induced OA
      • Bullock C.M.
      • Wookey P.
      • Bennett A.
      • Mobasheri A.
      • Dickerson I.
      • Kelly S.
      Peripheral calcitonin gene-related peptide receptor activation and mechanical sensitization of the joint in rat models of osteoarthritis pain.
      .
      The mechanism linking hypomagnesemia and OA pain likely involves altered levels of neurotransmitters and N-methyl-d-aspartate (NMDA) receptor function. Weglicki has shown that serum levels of the neuropeptides substance P and CGRP were significantly elevated within 3 days after initiating a low magnesium diet in a rodent model
      • Weglicki W.B.
      • Phillips T.M.
      Pathobiology of magnesium deficiency: a cytokine/neurogenic inflammation hypothesis.
      . The source of the acute elevation of the neurotransmitters may be C-fibers in the dorsal root ganglia
      • Leeman S.E.
      Substance P and neurotensin: discovery, isolation, chemical characterization and physiological studies.
      . These neurotransmitters may then depolarize postsynaptic neurons to activate quiescent NMDA receptors, leading to hyperalgesia
      • Basbaum A.I.
      • Bautista D.M.
      • Scherrer G.
      • Julius D.
      Cellular and molecular mechanisms of pain.
      . Magnesium is an NMDA receptor antagonist that acts by blocking the receptor channel complex in a voltage-dependent manner
      • Mayer M.L.
      • Westbrook G.L.
      • Guthrie P.B.
      Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones.
      . Hypomagnesemia may accentuate pain by unblocking the NMDA receptor and increasing NMDA receptor expression
      • Rondón L.J.
      • Privat A.M.
      • Daulhac L.
      • Davin N.
      • Mazur A.
      • Fialip J.
      • et al.
      Magnesium attenuates chronic hypersensitivity and spinal cord NMDA receptor phosphorylation in a rat model of diabetic neuropathic pain.
      .
      We also found that the average magnesium intake from diet and supplements in the Osteoarthritis Initiative cohort was below recommended amounts. The rates of insufficient magnesium intake appear similar to those reported by the USDA in the National population from the same time period, although direct comparison is not feasible due to differences in diet assessment methods (the Block Brief 2000 FFQ method used here tends to underestimate total nutrient intakes). About half the US population consumes less than the daily requirement (RDA) of magnesium from foods

      Moshfegh A, Goldman JD, Ahuja J, Rhodes D, Lacomb R. What We Eat in America, NHANES 2005-2006; Usual Nutrient Intakes from Food and Water Compared to 1997; Dietary Reference Intakes for Vitamin D, Calcium, Phosphorus, and Magnesium. Available at: http://www.ars.usda.gov/ba/bhnrc/fsrg. [Accessed 17 January 2018].

      , and the dietary intake of magnesium in the Western world appears to be declining
      • Ford E.S.
      • Mokdad A.H.
      Dietary magnesium intake in a national sample of US adults.
      . Multiple factors may contribute to a decreasing intake of magnesium from dietary sources. Drinking water accounts for about 10% of daily magnesium intake
      • Marx A.
      • Neutra R.R.
      Magnesium in drinking water and ischemic heart disease.
      . Increased consumption of filtered water, which has a lower concentration of magnesium than ground/well water, may be a contributing factor. Studies have shown that serum magnesium was significantly lower in the population consuming filtered water compared to those drinking non-filtered water
      • Kanadhia K.C.
      • Ramavataram D.V.S.S.
      • Nilakhe S.P.D.
      • Patel S.
      A study of water hardness and the prevalence of hypomagnesaemia and hypocalcaemia in healthy subjects of Surat district (Gujarat).
      . Food processing procedures, such as refining of grains, remove nutrient rich germ and bran, and result in lower magnesium content
      • Lakshmanan F.L.
      • Rao R.B.
      • Kim W.W.
      • Kelsay J.L.
      Magnesium intakes, balances, and blood levels of adults consuming self-selected diets.
      • Elin R.J.
      Magnesium metabolism in health and disease.
      .
      Additionally, we gained insight into the relationship between magnesium intake and fiber intake. The association between magnesium intake and knee pain and function scores was strongest among subjects who had a low fiber intake, and became less robust in individuals with a higher fiber intake in our study. Fiber intake typically correlates with unmeasured indicators of a healthy diet
      • Park Y.
      • Subar A.F.
      • Hollenbeck A.
      • Schatzkin A.
      Dietary fiber intake and mortality in the NIH-AARP diet and health study.
      • Gil Á.
      • Martinez de Victoria E.
      • Olza J.
      Indicators for the evaluation of diet quality.
      , and hence adjustment for fiber intake is important to isolate the effect of a specific nutrient. Low magnesium intake and low fiber intake have both been associated with radiographic osteoarthritis
      • Qin B.
      • Shi X.
      • Samai P.S.
      • Renner J.B.
      • Jordan J.M.
      • He K.
      Association of dietary magnesium intake with radiographic knee osteoarthritis: results from a population-based study.
      • Hunter D.J.
      • Hart D.
      • Snieder H.
      • Bettica P.
      • Swaminathan R.
      • Spector T.D.
      Evidence of altered bone turnover, vitamin D and calcium regulation with knee osteoarthritis in female twins.
      • Zeng C.
      • Wei J.
      • Li H.
      • Yang T.
      • Zhang F.
      • Pan D.
      • et al.
      Relationship between serum magnesium concentration and radiographic knee osteoarthritis.
      • Dai Z.
      • Niu J.
      • Zhang Y.
      • Jacques P.
      • Felson D.T.
      Dietary intake of fibre and risk of knee osteoarthritis in two US prospective cohorts.
      . While many foods high in magnesium are also high in fiber, there are many foods with high magnesium and low fiber content as well. Some of the top sources of dietary magnesium reported in Americans, for example, were milk, meat, coffee, and beer
      • Ford E.S.
      • Mokdad A.H.
      Dietary magnesium intake in a national sample of US adults.
      . Hence an interaction between magnesium and fiber intake is biologically plausible. Other nutrition studies that did not test for interaction and simply treated fiber intake as a confounder in predictive models were unable to separate the effect of magnesium from the effect of fiber intake
      • Kirii K.
      • Iso H.
      • Date C.
      • Fukui M.
      • Tamakoshi A.
      JACC Study Group
      Magnesium intake and risk of self-reported type 2 diabetes among Japanese.
      • Bo S.
      • Durazzo M.
      • Guidi S.
      • Carello M.
      • Sacerdote C.
      • Silli B.
      • et al.
      Dietary magnesium and fiber intakes and inflammatory and metabolic indicators in middle-aged subjects from a population-based cohort.
      . The presence of an interaction with fiber rather than confounding in our study indicates a role for dietary magnesium in osteoarthritis pain, that is outside the general healthy diet phenotype.
      Other nutrients, foods, and diets have been studied in relationship to OA pain, including vitamin C, vitamin D, milk, and the Mediterranean diet. Vitamin C was not significantly associated with knee symptoms in OA, and the results for vitamin D were inconclusive
      • Hung M.
      • Bounsanga J.
      • Voss M.W.
      • Gu Y.
      • Crum A.B.
      • Tang P.
      Dietary and supplemental vitamin C and D on symptom severity and physical function in knee osteoarthritis.
      . Mediterranean diet was associated with lower WOMAC scores and better quality of life
      • Veronese N.
      • Stubbs B.
      • Noale M.
      • Solmi M.
      • Luchini C.
      • Smith T.O.
      • et al.
      Adherence to a Mediterranean diet is associated with lower prevalence of osteoarthritis: data from the osteoarthritis initiative.
      , and higher milk consumption was linked to reduced radiographic OA progression in women
      • Lu B.
      • Driban J.B.
      • Duryea J.
      • McAlindon T.
      • Lapane K.L.
      • Eaton C.B.
      Milk consumption and progression of medial tibiofemoral knee osteoarthritis: data from the Osteoarthritis Initiative.
      . Interestingly, milk-based dietary supplements have also been shown to reduce OA pain symptoms in two randomized clinical trials
      • Colker C.M.
      • Swain M.
      • Lynch L.
      • Gingerich D.A.
      Effects of a milk-based bioactive micronutrient beverage on pain symptoms and activity of adults with osteoarthritis: a double-blind, placebo-controlled clinical evaluation.
      • Zenk J.L.
      • Helmer T.R.
      • Kuskowski M.A.
      The effects of milk protein concentrate on the symptoms of osteoarthritis in adults: an exploratory, randomized, double-blind, placebo-controlled trial.
      . As milk and components of the Mediterranean diet are high in magnesium, it is plausible that magnesium intake is the common mechanism of their effect on pain and function. This argument is further strengthened by our finding that magnesium intake from supplements is associated with better OA pain and function outcomes. At the same time, as dietary components are closely correlated, it is possible that this effect is due to an unmeasured nutrient, and not magnesium. Hence, our findings should be interpreted with caution, and further evaluated in clinical trials.
      Our study's strengths include data from a large, well-established osteoarthritis cohort with standardized assessments of radiographic knee OA and knee pain/function scores. Evaluating pain outcomes in relationship to magnesium is likely more relevant to patients and clinicians than radiographic outcomes, and can aid in determining the expected effect size for clinical trials of magnesium supplementation in treatment of knee OA. The limitations include inaccuracies in dietary assessment by FFQ, reporter bias for multiple self-reported parameters, and collider bias due to studying prevalent rather than incident knee OA. Based on calorie counts, the overall dietary intake was likely underestimated
      • Subar A.F.
      • Thompson F.E.
      • Kipnis V.
      • Midthune D.
      • Hurwitz P.
      • McNutt S.
      • et al.
      Comparative validation of the block, Willett, and National Cancer Institute food frequency questionnaires: the Eating at America's table study.
      . Hence calorie adjustment was used for analyses. Our study did not use an objective measure of magnesium deficiency, such as serum magnesium levels or overall magnesium balance, however, in long-term magnesium balance studies dietary intake correlated closely with serum magnesium and blood cell magnesium
      • Mayer M.L.
      • Westbrook G.L.
      • Guthrie P.B.
      Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones.
      . It should also be noted that there was little change in mean pain and function scores over 4 years of follow up among subjects in different quintiles of magnesium intake, hence results were largely determined by the baseline value (Fig. 2). Further studies in different populations are required to evaluate the effects of magnesium intake on change in pain and function over time. Additionally, as the Osteoarthritis Initiative primarily included white educated US adults, caution should be used when generalizing results to other populations.
      In conclusion, the reported magnesium intake among adults with prevalent radiographic knee OA was low, though likely similar to the general US population. Low magnesium intake was associated with worse pain and function scores in radiographically affected knees at baseline and over 4 years of follow up, after adjustment for multiple confounders. A similar association was observed between magnesium intake from supplements and pain/disability outcomes. Further prospective experimental studies are needed to address the effects of magnesium on changes in OA pain over time, and to reliably isolate magnesium effects from other nutritional factors.

      Conflict of interest

      Neither of the authors have any significant conflict of interest relevant to this manuscript.

      Funding sources

      Internal University of Minnesota funds. University of Minnesota paid researchers' salaries and provided protected research time for this project.

      Other contributors

      None.

      Acknowledgements

      Author contributions: All authors have read and approved the manuscript. Drs Anna Shmagel, Naoko Onizuka, Tien Vo, Lisa Langsetmo and Robert Foley were directly involved in study design, data analysis, generating tables and figures, and manuscript writing and editing. Drs Kristine Ensrud and Peter Valen were directly involved in study design, and manuscript writing and editing.

      Appendix A. Supplementary data

      References

        • Zhang Y.
        • Jordan J.M.
        Epidemiology of osteoarthritis.
        Clin Geriatr Med. 2010; 26: 355-369https://doi.org/10.1016/j.cger.2010.03.001
        • McAlindon T.E.
        • Cooper C.
        • Kirwan J.R.
        • Dieppe P.A.
        Knee pain and disability in the community.
        Br J Rheumatol. 1992; 31: 189-192
        • Sanders G.T.
        • Huijgen H.J.
        • Sanders R.
        Magnesium in disease: a review with special emphasis on the serum ionized magnesium.
        Clin Chem Lab Med. 1999; 37: 1011-1033https://doi.org/10.1515/CCLM.1999.151
        • Simental-Mendía L.E.
        • Rodríguez-Morán M.
        • Guerrero-Romero F.
        Oral magnesium supplementation decreases C-reactive protein levels in subjects with prediabetes and hypomagnesemia: a clinical randomized double-blind placebo-controlled trial.
        Arch Med Res. 2014; 45: 325-330https://doi.org/10.1016/j.arcmed.2014.04.006
        • Weglicki W.B.
        • Phillips T.M.
        Pathobiology of magnesium deficiency: a cytokine/neurogenic inflammation hypothesis.
        Am J Physiol. 1992; 263: R734-R737
        • McGuire S.
        Scientific report of the 2015 Dietary Guidelines Advisory Committee. Washington, DC: US Departments of Agriculture and Health and Human Services, 2015.
        Adv Nutr. 2016; 7: 202-204https://doi.org/10.3945/an.115.011684
        • Ghattaura N.
        • Murphy C.
        • Valen P.
        High prevalence of hypomagnesemia and its relation to BMI, type 2 diabetes, and clinical disease measures in a Va outpatient rheumatology clinic population: abstract number: 2260.
        Arthritis Rheumatol. 2015; 67: 2725-2726
        • Qin B.
        • Shi X.
        • Samai P.S.
        • Renner J.B.
        • Jordan J.M.
        • He K.
        Association of dietary magnesium intake with radiographic knee osteoarthritis: results from a population-based study.
        Arthritis Care Res (Hoboken). 2012; 64: 1306-1311https://doi.org/10.1002/acr.21708
        • Hunter D.J.
        • Hart D.
        • Snieder H.
        • Bettica P.
        • Swaminathan R.
        • Spector T.D.
        Evidence of altered bone turnover, vitamin D and calcium regulation with knee osteoarthritis in female twins.
        Rheumatology (Oxford). 2003; 42: 1311-1316https://doi.org/10.1093/rheumatology/keg373
        • Zeng C.
        • Wei J.
        • Li H.
        • Yang T.
        • Zhang F.
        • Pan D.
        • et al.
        Relationship between serum magnesium concentration and radiographic knee osteoarthritis.
        J Rheumatol. 2015; 42: 1231-1236https://doi.org/10.3899/jrheum.141414
        • McConnell S.
        • Kolopack P.
        • Davis A.M.
        The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC): a review of its utility and measurement properties.
        Arthritis Rheumatol. 2001; 45: 453-461
        • Roos E.M.
        • Roos H.P.
        • Lohmander L.S.
        • Ekdahl C.
        • Beynnon B.D.
        Knee Injury and Osteoarthritis Outcome Score (KOOS)–development of a self-administered outcome measure.
        J Orthop Sports Phys Ther. 1998; 28: 88-96https://doi.org/10.2519/jospt.1998.28.2.88
        • Park Y.
        • Subar A.F.
        • Hollenbeck A.
        • Schatzkin A.
        Dietary fiber intake and mortality in the NIH-AARP diet and health study.
        Arch Intern Med. 2011; 171: 1061-1068https://doi.org/10.1001/archinternmed.2011.18
        • Gil Á.
        • Martinez de Victoria E.
        • Olza J.
        Indicators for the evaluation of diet quality.
        Nutr Hosp. 2015; 31: 128-144https://doi.org/10.3305/nh.2015.31.sup3.8761
        • Washburn R.A.
        • Smith K.W.
        • Jette A.M.
        • Janney C.A.
        The Physical Activity Scale for the Elderly (PASE): development and evaluation.
        J Clin Epidemiol. 1993; 46: 153-162
        • Konttinen Y.T.
        • Sillat T.
        • Barreto G.
        • Ainola M.
        • Nordström D.C.E.
        Osteoarthritis as an autoinflammatory disease caused by chondrocyte-mediated inflammatory responses.
        Arthritis Rheumatol. 2012; 64: 613-616https://doi.org/10.1002/art.33451
        • Goldring M.B.
        • Otero M.
        Inflammation in osteoarthritis.
        Curr Opin Rheumatol. 2011; 23: 471-478https://doi.org/10.1097/BOR.0b013e328349c2b1
        • King D.E.
        • Mainous A.G.
        • Geesey M.E.
        • Woolson R.F.
        Dietary magnesium and C-reactive protein levels.
        J Am Coll Nutr. 2005; 24: 166-171
        • Guerrero-Romero F.
        • Bermudez-Peña C.
        • Rodríguez-Morán M.
        Severe hypomagnesemia and low-grade inflammation in metabolic syndrome.
        Magnes Res. 2011; 24: 45-53https://doi.org/10.1684/mrh.2011.0281
        • Orita S.
        • Ishikawa T.
        • Miyagi M.
        • Ochiai N.
        • Inoue G.
        • Eguchi Y.
        • et al.
        Pain-related sensory innervation in monoiodoacetate-induced osteoarthritis in rat knees that gradually develops neuronal injury in addition to inflammatory pain.
        BMC Musculoskelet Disord. 2011; 12: 134https://doi.org/10.1186/1471-2474-12-134
        • Ferland C.E.
        • Laverty S.
        • Beaudry F.
        • Vachon P.
        Gait analysis and pain response of two rodent models of osteoarthritis.
        Pharmacol Biochem Behav. 2011; 97: 603-610https://doi.org/10.1016/j.pbb.2010.11.003
        • Schaible H.-G.
        • Richter F.
        • Ebersberger A.
        • Boettger M.
        • Vanegas H.
        • Natura G.
        • et al.
        Joint pain.
        Exp Brain Res. 2009; 196: 153-162https://doi.org/10.1007/s00221-009-1782-9
        • Fernihough J.
        • Gentry C.
        • Bevan S.
        • Winter J.
        Regulation of calcitonin gene-related peptide and TRPV1 in a rat model of osteoarthritis.
        Neurosci Lett. 2005; 388: 75-80https://doi.org/10.1016/j.neulet.2005.06.044
        • Schwab W.
        • Bilgiçyildirim A.
        • Funk R.H.
        Microtopography of the autonomic nerves in the rat knee: a fluorescence microscopic study.
        Anat Rec. 1997; 247: 109-118
        • Saxler G.
        • Löer F.
        • Skumavc M.
        • Pförtner J.
        • Hanesch U.
        Localization of SP- and CGRP-immunopositive nerve fibers in the hip joint of patients with painful osteoarthritis and of patients with painless failed total hip arthroplasties.
        Eur J Pain. 2007; 11: 67-74https://doi.org/10.1016/j.ejpain.2005.12.011
        • Bullock C.M.
        • Wookey P.
        • Bennett A.
        • Mobasheri A.
        • Dickerson I.
        • Kelly S.
        Peripheral calcitonin gene-related peptide receptor activation and mechanical sensitization of the joint in rat models of osteoarthritis pain.
        Arthritis Rheumatol. 2014; 66: 2188-2200https://doi.org/10.1002/art.38656
        • Leeman S.E.
        Substance P and neurotensin: discovery, isolation, chemical characterization and physiological studies.
        J Exp Biol. 1980; 89: 193-200
        • Basbaum A.I.
        • Bautista D.M.
        • Scherrer G.
        • Julius D.
        Cellular and molecular mechanisms of pain.
        Cell. 2009; 139: 267-284https://doi.org/10.1016/j.cell.2009.09.028
        • Mayer M.L.
        • Westbrook G.L.
        • Guthrie P.B.
        Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones.
        Nature. 1984; 309: 261-263
        • Rondón L.J.
        • Privat A.M.
        • Daulhac L.
        • Davin N.
        • Mazur A.
        • Fialip J.
        • et al.
        Magnesium attenuates chronic hypersensitivity and spinal cord NMDA receptor phosphorylation in a rat model of diabetic neuropathic pain.
        J Physiol (Lond). 2010; 588: 4205-4215https://doi.org/10.1113/jphysiol.2010.197004
      1. Moshfegh A, Goldman JD, Ahuja J, Rhodes D, Lacomb R. What We Eat in America, NHANES 2005-2006; Usual Nutrient Intakes from Food and Water Compared to 1997; Dietary Reference Intakes for Vitamin D, Calcium, Phosphorus, and Magnesium. Available at: http://www.ars.usda.gov/ba/bhnrc/fsrg. [Accessed 17 January 2018].

        • Ford E.S.
        • Mokdad A.H.
        Dietary magnesium intake in a national sample of US adults.
        J Nutr. 2003; 133: 2879-2882
        • Marx A.
        • Neutra R.R.
        Magnesium in drinking water and ischemic heart disease.
        Epidemiol Rev. 1997; 19: 258-272
        • Kanadhia K.C.
        • Ramavataram D.V.S.S.
        • Nilakhe S.P.D.
        • Patel S.
        A study of water hardness and the prevalence of hypomagnesaemia and hypocalcaemia in healthy subjects of Surat district (Gujarat).
        Magnes Res. 2014; 27: 165-174https://doi.org/10.1684/mrh.2014.0373
        • Lakshmanan F.L.
        • Rao R.B.
        • Kim W.W.
        • Kelsay J.L.
        Magnesium intakes, balances, and blood levels of adults consuming self-selected diets.
        Am J Clin Nutr. 1984; 40: 1380-1389
        • Elin R.J.
        Magnesium metabolism in health and disease.
        Dis Mon. 1988; 34: 161-218
        • Dai Z.
        • Niu J.
        • Zhang Y.
        • Jacques P.
        • Felson D.T.
        Dietary intake of fibre and risk of knee osteoarthritis in two US prospective cohorts.
        Ann Rheum Dis. 2017; 76: 1411-1419https://doi.org/10.1136/annrheumdis-2016-210810
        • Kirii K.
        • Iso H.
        • Date C.
        • Fukui M.
        • Tamakoshi A.
        • JACC Study Group
        Magnesium intake and risk of self-reported type 2 diabetes among Japanese.
        J Am Coll Nutr. 2010; 29: 99-106
        • Bo S.
        • Durazzo M.
        • Guidi S.
        • Carello M.
        • Sacerdote C.
        • Silli B.
        • et al.
        Dietary magnesium and fiber intakes and inflammatory and metabolic indicators in middle-aged subjects from a population-based cohort.
        Am J Clin Nutr. 2006; 84: 1062-1069
        • Hung M.
        • Bounsanga J.
        • Voss M.W.
        • Gu Y.
        • Crum A.B.
        • Tang P.
        Dietary and supplemental vitamin C and D on symptom severity and physical function in knee osteoarthritis.
        J Nutr Gerontol Geriatr. 2017; 36: 121-133https://doi.org/10.1080/21551197.2017.1317315
        • Veronese N.
        • Stubbs B.
        • Noale M.
        • Solmi M.
        • Luchini C.
        • Smith T.O.
        • et al.
        Adherence to a Mediterranean diet is associated with lower prevalence of osteoarthritis: data from the osteoarthritis initiative.
        Clin Nutr. 2017; 36: 1609-1614https://doi.org/10.1016/j.clnu.2016.09.035
        • Lu B.
        • Driban J.B.
        • Duryea J.
        • McAlindon T.
        • Lapane K.L.
        • Eaton C.B.
        Milk consumption and progression of medial tibiofemoral knee osteoarthritis: data from the Osteoarthritis Initiative.
        Arthritis Care Res (Hoboken). 2014; 66: 802-809https://doi.org/10.1002/acr.22297
        • Colker C.M.
        • Swain M.
        • Lynch L.
        • Gingerich D.A.
        Effects of a milk-based bioactive micronutrient beverage on pain symptoms and activity of adults with osteoarthritis: a double-blind, placebo-controlled clinical evaluation.
        Nutrition. 2002; 18: 388-392
        • Zenk J.L.
        • Helmer T.R.
        • Kuskowski M.A.
        The effects of milk protein concentrate on the symptoms of osteoarthritis in adults: an exploratory, randomized, double-blind, placebo-controlled trial.
        Curr Ther Res. 2002; 63: 430-442https://doi.org/10.1016/S0011-393X(02)80049-2
        • Subar A.F.
        • Thompson F.E.
        • Kipnis V.
        • Midthune D.
        • Hurwitz P.
        • McNutt S.
        • et al.
        Comparative validation of the block, Willett, and National Cancer Institute food frequency questionnaires: the Eating at America's table study.
        Am J Epidemiol. 2001; 154: 1089-1099