Article Text

Extended report
Dietary intake of fibre and risk of knee osteoarthritis in two US prospective cohorts
  1. Zhaoli Dai1,
  2. Jingbo Niu1,
  3. Yuqing Zhang1,
  4. Paul Jacques2,
  5. David T Felson1,3
  1. 1 Department of Medicine, Boston University School of Medicine, Clinical Epidemiology Research and Training Unit, Boston, Massachusetts, USA
  2. 2 Jean Mayer USDA Human Nutrition Research Center on Aging and Friedman School of Nutrition Science and Policy, Tufts University, Boston, Massachusetts, USA
  3. 3 University of Manchester and Central Manchester Foundation Trust, Manchester, UK
  1. Correspondence to Dr Zhaoli Dai, Clinical Epidemiology Research and Training Unit, Boston University School of Medicine, 650 Albany St, Boston, MA 02118, USA; joyzdai{at}


Objectives Dietary fibre reduces body weight and inflammation both of which are linked with knee osteoarthritis (OA). We examined the association between fibre intake and risk of knee OA.

Methods We used data from the Osteoarthritis Initiative (OAI) of 4796 participants and Framingham Offspring Osteoarthritis Study (Framingham) of 1268 persons. Dietary intake of fibre was estimated at baseline, and incident radiographic OA (ROA) and symptomatic OA (SxOA) were followed annually until 48 months in OAI and assessed 9 years later in Framingham. Knee pain worsening was also examined in OAI. Generalised estimating equations were applied in multivariable regression models.

Results In OAI, we identified 869 knees with SxOA, 152 knees with ROA and 1964 knees with pain worsening among 4051 subjects with valid dietary intake (baseline mean age: 61.2 years; mean body mass index (BMI): 28.6). In Framingham, 143 knees with SxOA and 176 knees with ROA among 971 such subjects (baseline mean age: 53.9 years; mean BMI: 27.0) were identified. In both cohorts, dietary total fibre was inversely associated with risk of SxOA (p trend <0.03) with significantly lower risk at the highest versus lowest quartile (OR (95% CI): 0.70 (0.52, 0.94) for OAI and 0.39 (0.17, 0.88) for Framingham). Furthermore, dietary total and cereal fibre were significantly inversely associated with knee pain worsening in OAI (p trend <0.02). No apparent association was found with ROA.

Conclusions Findings from two longitudinal studies consistently showed that higher total fibre intake was related to a lower risk of SxOA, while the relation to ROA was unclear.

  • Knee Osteoarthritis
  • Epidemiology
  • Treatment

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Dietary fibres are carbohydrates, primarily from plant-based foods such as cereal grains, fruits and vegetables, and nuts and legumes. Fibre is not digestible or absorbable in the small intestine but is partially or fully fermented in the colon.1 Epidemiological studies suggest that fibre is associated with lower risks of cardiovascular diseases,2 type 2 diabetes3 4 and mortality.5 Furthermore, clinical trials have shown the beneficial effect of dietary fibre in reducing body weight,6–8 blood pressure9 and circulating C reactive protein (CRP),10 and in improving glycaemic control.11 12 Such protective effects of dietary fibre are likely attributed to its intrinsic property of entrapping sugars and fats to promote satiety and reduce calorie intake,13 reduce serum cholesterol2 and improve glycaemic control12 as well as to its prebiotic function through fermentation in the gut to stimulate desirable microbiome to lower infection and inflammation.14

As the most common joint disorder, osteoarthritis (OA) is highly prevalent among adults aged ≥60 years.15 16 Largely due to pain and limited physical function, OA is a leading cause of disability and impairment of quality of life;17 however, no effective structure-modifying treatment is available to date. Like metabolic diseases, there is a strong link between obesity, inflammation and knee OA, where obesity both increases loading in weight-bearing joints18–21 and proinflammatory cytokines and adipokines;22 and increased inflammation and synovitis23 24 have been shown to be associated with joint pain.

In light of the relation of dietary fibre to body weight and inflammation, we initially examined the association between fibre intake and knee OA in the Osteoarthritis Initiative (OAI) and, as shown below, found a protective association of fibre intake with symptomatic OA (SxOA) and knee pain worsening. We sought confirmation using a second cohort in hope of replicating these results and better addressing dietary confounders in the Framingham Offspring Study, which was designed to more comprehensively assess diet. The objective of this study was to examine the association between dietary fibre and knee OA phenotypes including symptomatic knee OA (SxOA), radiographic knee OA (ROA) and knee pain worsening in two US cohorts.


Study population

Osteoarthritis Initiative (OAI)

OAI is a multi-centre, longitudinal prospective cohort of 4796 US men (41.5%) and women aged 45–79 years with or at risk of knee OA recruited from 2004 to 2006 to investigate OA risk factors. Details of the study protocol can be found elsewhere.25 After enrolment, participants were followed for OA assessment annually until 48 months.

Framingham Offspring cohort (Framingham)

Participants from the Framingham Offspring cohort, which was assembled in 1971 consisting of adult children of the Original Framingham Study and spouses of the offspring participants.26 The goal of the Offspring OA study was to investigate inheritance of OA among study participants who represent a community-based population unselected for OA or knee pain.27 During a callback visit after the Offspring exam 5 (1993–1994), 1268 eligible subjects (men: 43.9%, mean age 53.2 years) as baseline of the OA study, weight-bearing radiographs of both knees and surveys about knee pain were obtained. The same assessments were performed among the same subjects after an average of 9.5 years at exam 7 (2002–2005).28

In both studies, we excluded those who had rheumatoid arthritis or other forms of inflammatory arthritis. Institutional Review Board approval was obtained from all OAI study sites in OAI and from Boston University for the Framingham Study.

Baseline assessment

In both studies, information was collected including demographics, history of knee injury and surgery (including total knee replacement), medication use, tobacco and alcohol use, physical activity assessed by the Physical Activity in the Elderly Scale (PASE) and habitual diet recorded using the validated food frequency questionnaires (FFQs). Based on the major food sources of fibre, total dietary fibre was derived as sum of fibre from cereal grain, fruits and vegetables, and nuts and legumes. We only included participants with valid fibre (no missing values for total dietary fibre) and calorie intake (≥500 kcal and <4200 kcal for men and <4000 kcal for women). Anthropometric parameters including body weight, height and abdominal (OAI)/waist (Framingham) circumference were measured.25 26 Knee symptom assessment using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) was available at all exams in OAI but was only assessed in exam 7 in Framingham.

Baseline dietary fibre estimation

Osteoarthritis Initiative

Dietary fibre intake was estimated using the Block Brief 2000 FFQ of 60 food items.29 30 For each food item, frequency consumption was surveyed according to nine predetermined categories ranging from never to everyday with illustrations of standard portion sizes. Estimation of nutrient intake was calculated based on the nutrient composition values developed in the Second National Health and Nutrition Examination Survey.29

Framingham Study

Habitual dietary intake was recorded at baseline (exam 5) and 4 years later at exam 6 (1998–2001) via the Harvard validated semi-quantitative FFQ.31 The FFQ listed 126 items of foods in standard serving sizes and nine frequency categories ranging from ‘never or less than once per month’ to ‘6+ per day.’ Participants were allowed to add up to three additional foods, types of breakfast cereal and cooking oil that are not listed in the FFQ. Nutrient values were estimated and primarily relied on the US Department of Agriculture food composition database and supplemented by other published data.31 Additionally, glycaemic load for each food was calculated to estimate glucose effect. The Dietary Guidelines Adherence Index (DGAI-2010) score was applied to the baseline FFQ to determine diet quality according to the Dietary Guidelines for Americans 2010.32 Detail of the DGAI-2010 has been described33 and scores range from 0 (lowest adherence) to 100 (highest adherence). Because dietary fibre is one of the 24 components in the DGAI-2010, the modified DGAI-2010 used in this study excluded fibre.

Assessment of knee OA outcomes

Osteoarthritis Initiative

At each exam, subjects were asked about knee pain using the following question: ‘During the past 30 days, have you had pain, aching, or stiffness in your right/left knee on most days?’ and a positive response was used to define a painful knee. Knee pain was also assessed using the WOMAC pain subscale ranging from 0 (no pain) to 20 (most pain) points. All subjects obtained a fixed flexion posterior–anterior radiograph, which was read centrally for Kellgren and Lawrence (KL) grade. OA outcomes of interest included incident (1) ROA if a knee developed a KL grade ≥2 at follow-up, (2) SxOA, new onset of a combination of a painful knee and ROA and (3) knee pain worsening, if WOMAC pain score difference of a knee between baseline and each annual exam ≥14% of the base score defined in the published estimates for the minimal clinical important difference in WOMAC.34 Additional detail for knee pain worsening can be found in online supplementary file text 1. For incident SxOA or ROA, we excluded those with prevalence at baseline for the corresponding outcome.

Supplementary Material

Supplementary file

Framingham Study

After subjects had completed the baseline and follow-up exams, radiographs at both exams were read independently by two study readers among a sample of 1268 eligible subjects.28 The same definitions as OAI were applied to Framingham to define incident ROA and SxOA after excluding prevalent knees with these conditions.

Statistical analysis

For both studies, we calculated the residual of dietary fibre by regressing fibre intake on total calories within each cohort.35 Quartile cut-offs for dietary fibre were created based on the calculated residuals35 within men and women separately to account for sex differences in dietary intake. The rationale to use the residual method is described in online supplementary file text 2. Dietary intake was estimated by a semi-quantitative FFQ, which is suitable to rank individuals with regard to their intake.36 We examined the relation of dietary fibre to the risk of knee OA by comparing a higher quartile (Q2, Q3 and Q4) to the lowest quartile (Q1: reference category), using generalised estimating equations to account for correlation between two knees for each participant37 in both studies and for the analysis of pain worsening to account for repeated WOMAC pain measures annually in OAI.

In the base model (model 1), we adjusted for age (years), sex (men vs women), race (white vs non-white in OAI) and total energy intake (kcal); model 2 was further adjusted for education attainment (below vs college or above), annual household income (<US$50 000 vs ≥US$50 000), smoking status (never, former and current smokers), physical activity (PASE, continuous) and other dietary factors including dietary vitamin C (mg/day), K (µg/day), polyunsaturated fat (g/day), saturated fats, which have been shown to relate to OA in previous studies.38–42 Non-steroidal anti-inflammatory drugs use (yes vs no) was further adjusted for pain worsening in OAI. In Framingham, we also controlled for glycaemic load, which has been found to be inversely related to the Mediterranean diet in model 2.43 44 In a separate model (model 3), we adjusted for the same covariates in model 2 except replacing the dietary variables with DGAI-2010 to account for diet quality. In addition, we further adjusted for baseline body mass index (BMI) (kg/m2) in addition to the full model in both studies. Linear trends were tested using the median value of each quartile of dietary fibre as a continuous variable in the model. Because there was no significant difference between genders in the fibre–OA associations, we combined men and women for the descriptive characteristics and regression analysis.

Consistency of our results was tested by stratifying socioeconomic status and replacing BMI with abdominal/waist circumference (cm) as a covariate in the full model. All statistical analysis was conducted using SAS V.9.3. A p value <0.05 (two sided) was considered statistically significant.


At the end of the study course, 540 (11%) among 4796 participants in OAI and 131 persons (11.9%) among 1268 persons in Framingham were lost to follow-up. Figure 1 describes the eligible participants and knees included in the final analyses.

Figure 1

Eligible participants in (A) OAI and (B) Framingham Offspring Study. OAI, Osteoarthritis Initiative; ROA, radiographic OA; SOA, symptomatic OA; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index.

We summarised baseline characteristics across quartiles of total fibre intake by each cohort study (table 1). In both studies, those who consumed more fibre were older, had lower BMI and were more educated. They were less likely to have knee pain symptoms and consumed higher amount of vitamins C and K and less dietary saturated fats.

Table 1

Participants’ baseline characteristics according to quartile intake of total fibre in OAI and Framingham OA Study

The associations between dietary fibre and phenotypes of knee OA in OAI are presented in table 2. For total dietary fibre, compared with the lowest quartile (Q1), there was a significant dose-dependent inverse relationship for the risks of SxOA and knee worsening (p trend ≤0.005): at the highest quartile (Q4), risks of SxOA and knee pain worsening were lowered significantly by 30% (OR (95% CI): 0.70 (0.52 to 0.94)) and 19% (OR (95% CI): 0.81 (0.71 to 0.94)), respectively. Although non-statistically significant, similar results for fibre from cereal grains and from fruits and vegetables were observed. Significant inverse relationship was also found between cereal fibre and pain worsening (p trend <0.02) with a 14% lower risk in those at Q4 relative to Q1. No association was found for incident ROA or between fibre from nuts and legumes and knee OA.

Table 2

OR (95% CI) to estimate total effect of dietary fibre on knee incident SxOA (n=2876 persons), incident ROA (n=1675 persons) and knee pain worsening (n=3976 persons) for all eligible participants in OAI

The association of dietary fibre with knee OA in Framingham is presented in table 3. There was a significant dose-dependent inverse relationship between dietary total fibre and risk of SxOA in all models (p trend <0.03). Participants who consumed the highest quartile (Q4) of total fibre compared with the lowest quartile (Q1) had significantly lower risk by 61% [OR (95% CI): 0.39 (0.17 to 0.88)] in the full model adjusted for diet quality (DQAI-2010). Similarly inverse trends for fibre from other sub-categories were noted although these were not statistical significant. The associations with incident ROA were not apparent.

Table 3

OR (95% CI) to estimate total effect of dietary fibre on knee incident SxOA and incident ROA where radiographs are centrally viewed for all eligible participants in the Framingham Offspring OA study (n=971 persons)

In the full model with further adjustment for BMI, we found an attenuated but significant inverse relation of dietary total fibre to risk of SxOA (p trend=0.03 in both studies) and to knee pain worsening (p trend=0.03 in OAI). Results for the association between dietary fibre and incident SxOA or knee pain worsening did not differ by stratum of obesity (BMI <30 vs BMI ≥30) as evidenced by a p value for interaction greater than 0.05 between obesity status and quartile dietary total fibre. Secondary analysis showed that dietary fibre was inversely associated with risk of SxOA or knee pain worsening independent of waist circumference in both studies (see online supplementary file tables 1-2) and regardless of socioeconomic status (see online supplementary file tables 3-4).


To our knowledge, this was the first study in the literature investigating the association between dietary fibre and OA outcomes. Our results consistently showed that in two prospective US cohorts with different study designs and study populations, those who consumed higher fibre intake were less likely to develop SxOA or to experience worsening knee pain during the study course regardless of socioeconomic or obesity status. These data demonstrate a consistent protective association between total fibre intake and symptom-related knee OA in two study populations with careful adjustment for potential confounders.

There is increasing evidence suggesting that OA shares similar metabolic characteristics including obesity, dyslipidaemia and inflammation with cardiovascular diseases and diabetes.45 Our findings are consistent with those showing that higher dietary total fibre intake reduced risks of cardiovascular diseases46 and reduced fasting glucose in type 2 diabetes.12 The results of non-statistical associations for fibre from other diet sources with OA are similar with findings in cardiovascular diseases2 47 and mortality5 after controlling for various risk factors and covariates. Furthermore, these results are consistent with our report on high fibre intake with lower risks of moderate and severe knee pain trajectories over 8 years in OAI.48 In OAI and Framingham, an inverse relationship was found between dietary fibre and BMI and between BMI and incident SxOA (p<0.01). Additional adjustment for baseline BMI attenuated the results compared with those without such adjustment. We have not tested for whether fibre’s association with OA is mediated by BMI which could, in part, explain the dilution of the risk estimates after controlling for BMI.

The study design and participants’ characteristics are different between the cohorts. Additionally, there were strengths and weaknesses in the assessment of exposure (fibre intake) and OA outcomes of interest between the studies (see online supplementary file text 3). The inverse association between total fibre intake and knee pain worsening further confirmed the results of SxOA. Using data from two cohorts to assess the relation of detary fibre to OA reduced biases due to potential confounding and misclassification of exposures and outcomes.

The estimation for total fibre intake on average is 18.8 (SD 7.8 g/day) in Framingham and is 15.0 (7.3) in OAI. Such estimated values for daily fibre intake are comparable with the average intake (15 g/day) among Americans.13 Furthermore, because those who consume more fibre tend to better adhere to the dietary guidelines (table 1), we adjusted for diet quality (DGAI-2010) and found a similar significant inverse association in Framingham. Additionally, a statistically significant inverse relationship was found between the average intake of dietary fibre from baseline to 4 years later and incident SxOA (p trend=0.047) in Framingham. The similar effect estimates in all models further suggest that the association of dietary total fibre and risk of symptom related knee OA may not be affected by the risk factors, lifestyle and dietary confounders we accounted for.

In this study, although we did not find a statistically significant association between dietary fibre and radiographic OA, we cannot rule out the possibility that fibre may reduce the risk of ROA by lowering BMI20 which may not be detectable because of a bias due to depletion of susceptibility to knee ROA49 (see online supplementary file text 4). Also, serum CRP level, a marker of systemic inflammation, has been shown to play a greater role in symptoms than radiographic changes in OA.24 Additionally, synovitis has been related to knee pain in persons with ROA.50 If one of the biological mechanisms for dietary fibre and knee OA is to lower inflammation, its effect on SxOA may be more apparent compared with structural deterioration presented by ROA.

Strengths of this study included using data from two prospective cohorts where one is a multicentre and multiethnic cohort targeting to older Americans at risk or with OA and the other is a community-based population study. Therefore, using two different studies to assess fibre intake and OA outcomes in studying the association between dietary fibre and risk of knee OA improves internal validity. However, as a limitation, participants in OAI had higher BMI, and those in Framingham were selected regardless of their risk for OA. Such selection criteria in OAI might introduce collider bias by conditioning on BMI, where an unknown confounder (say, a genetic factor for obesity) affects BMI and knee OA and may negatively relate to dietary fibre. Our reported risk estimates for the association between fibre intake and knee OA could be diluted towards the null due to such collider bias. However, this would not have been true for analyses of Framingham data. Another limitation is that self-reported dietary data are prone to biases and potentially results in non-differential misclassification of fibre consumption, even though this is likely to lead to an underestimate of the observed associations. Inevitably, results from observational studies do not prove causality and may still raise concerns about residual confounding. Ultimately randomised trials are needed to prove causation, but observational and other studies are needed first to provide empirical evidence. As for dietary fibre, evidence from randomised clinical trials have demonstrated its effects on body weight,6–8 blood pressure,9 CRP10 and glycaemia control.11 12 Many of these are of relevance to OA. This suggests that the association of fibre with OA uncovered in our data may be genuine and unlikely due to the threats to internal validity.51 Nonetheless, findings in these two studies need further corroboration.

In conclusion, data from two US prospective cohorts demonstrated that higher dietary total fibre intake was associated with lower risks of SxOA and pain worsening of the knee, while the association with radiographic OA was not apparent. Such results support the current recommended daily fibre for older Americans.52


The authors would like to thank Gail Vanca Rogers, senior statistician at Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University for her generous assistance in expediting the distribution of the fibre variables and DGAI-2010 in the Framingham Study.


View Abstract

Supplementary materials

  • Lay summary

    Disclaimer: This is a summary of a scientific article written by a medical professional (“the Original Article”). The Summary is written to assist non medically trained readers to understand general points of the Original Article. It is supplied “as is” without any warranty. You should note that the Original Article (and Summary) may not be fully relevant nor accurate as medical science is constantly changing and errors can occur. It is therefore very important that readers not rely on the content in the Summary and consult their medical professionals for all aspects of their health care and only rely on the Summary if directed to do so by their medical professional. Please view our full Website Terms and Conditions.
    Copyright © 2017 BMJ Publishing Group Ltd & European League Against Rheumatism. Medical professionals may print copies for their and their patients and students non commercial use. Other individuals may print a single copy for their personal, non commercial use. For other uses please contact our Rights and Licensing Team.


  • Contributors Study design: ZJD, YZ and DF. Data analysis: ZJD and JN. Data interpretation: ZJD, YZ, PJ and DF. Drafting manuscript: ZJD. All authors contributed to the final draft of the manuscript.

  • Competing interests None declared.

  • Ethics approval Institutional Review Board from all Osteoarthritis Initiative study sites and from Boston University.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Correction notice This article has been corrected since it published Online First. The first author affiliation has been corrected and the abstract updated.

Linked Articles