Whilst the consumption of red meat is often recommended for its iron content [1], there is emerging evidence that red meat and processed meat are associated with carcinogenesis at several anatomic sites and increased mortality [2–5]. The mechanisms for these adverse effects are various but have been related to the constituent nutrients of red meat and methods of processing and cooking. Meat is a source of several multisite carcinogens including N-nitroso compounds, heterocyclic amines, and polycyclic aromatic hydrocarbons, some of which are formed during high-temperature cooking of meat [2]. Iron in red meat may increase oxidative damage and increase the formation of N-nitroso compounds [6]. Meat is also a major source of saturated fat which has been positively associated with cancer [7]. Current dietary recommendations are to consume small to moderate amounts of red meat and processed meat as a way of reducing the risk of a number of chronic diseases including cardiovascular disease and cancer [8].

Meat consumption may have beneficial effects on the musculoskeletal system since there is evidence that high animal protein is associated with high bone mineral density and low risk of hip fracture [9–11]. Osteoarthritis (OA) is a common chronic joint disease involving the whole joint including the articular cartilage, subchondral bone and soft tissues. There is evidence that dietary nutrients may protect against OA [12]. However, it is unknown whether meat consumption may affect the risk of OA, and the existing evidence is conflicting. A recent cross-sectional study reported a higher prevalence of degenerative arthritis associated with greater meat consumption [13], whereas another cross-sectional study found no relationship between consumption of meat/poultry and fish and the prevalence of knee OA [14].

Total joint replacement is the only established treatment for those with symptomatic end-stage OA where conservative management fails, thus it may be considered a proxy outcome measure for severe end-stage OA. We analyzed data from the Melbourne Collaborative Cohort Study (MCCS) on consumption of red meat, processed meat, chicken, and fish to determine the relation to the risk of primary knee and hip joint replacement due to OA.

The study profile is shown in Figure 1. Over an average of 4.8 (SD 0.7) years of follow-up per person (2001-5), 888 participants were identified with incident primary joint replacement for OA including 411 hip replacements and 477 knee replacements. The baseline (1990-4, an average of 12.6 years prior to the joint replacement) demographic characteristics and meat consumptions of the study population are presented in Tables 1, 2 and 3. Meat was eaten more frequently by men than by women: men's weekly median frequencies of consumption were 5.0 times for fresh red meat and 2.5 times for processed meat compared to 4.0 and 2.0 times for women. Italian and Greek immigrants consumed fresh red meat (median 5.0 versus 4.5 times/week) and chicken (median 2.0 versus 1.5 times/week) more frequently than those born in Australia and United Kingdom. Participants with lower education level consumed fresh red meat more frequently than those with a higher level of education (median 4.5 versus 4.0 times/week). The meat consumptions at baseline and first follow up were moderately correlated, with the correlation coefficient ranging from 0.37 to 0.40 for fresh red meat, chicken and fish.

The ratio of fresh red meat consumption to chicken and fish consumption was also examined (Table 4). There was a negative dose-response association observed for the risk of hip replacement (HR of 0.87 for increase of 1 unit of the ratio, 95% CI 0.77 - 0.97, P for trend 0.01), but no association observed for the risk of knee replacement (HR 0.97, 95% CI 0.88 - 1.08, P for trend 0.62). The statistical evidence for heterogeneity in these dose-response associations between the hip and knee was weak (test for homogeneity of trends, P = 0.12).

There was no statistical evidence that country of birth, gender, or educational level modified the associations between meat consumption and the risk of hip or knee replacement (P value for effect modification ranged from 0.11 to 0.92).

In this prospective cohort study, we found that a high level consumption of fresh red meat was associated with a decreased risk of hip, but not knee joint replacement for OA. Consumption of processed meat, chicken and fish had little association with the risk of either hip or knee joint replacement.

No previous study has examined the relationship between meat consumption and the risk of joint replacement, a proxy outcome measurement for severe OA. The data regarding the association between meat consumption and the risk of OA are sparse. In a community-based cross-sectional study, Kacar and colleagues found that the intake of meat/poultry and fish was not associated with the prevalence of symptomatic knee OA where the diagnosis of knee OA was made clinically or clinically and radiologically according to the American College of Rheumatology criteria [14]. In contrast, another cross-sectional study showed an independent and moderately increased prevalence of degenerative arthritis associated with greater meat, poultry, and fish consumption in both men and women [13]. However, this latter study used self-reported information on the prevalence of degenerative arthritis and rheumatism as outcome measure, including not only OA but also other forms of arthritis and rheumatism. Moreover, cross-sectional studies are subject to recall bias when trying to capture the dietary patterns which may contribute to the discrepancies. We used prospectively collected dietary data at two time points and followed up a large population for approximately 13 years to address the effect of meat consumption on the risk of joint replacement. In contrast to the findings from the two previous cross-sectional studies [13, 14], we found that higher consumption of fresh red meat approximately 13 years prior to joint replacement was related to a decreased risk of hip, but not knee, joint replacement due to OA, whereas consumption of processed meat, chicken and fish was not associated with the risk. The association between fresh red meat consumption and hip replacement risk was further supported by the findings when self-reported meat consumption, recorded approximately 8 years prior to joint replacement, was examined. Similar effect (same direction and magnitude) of fresh red meat consumption was observed, although less significant due to reduced sample size.

The relationship between different food items and disease is sometimes difficult to interpret due to the inter-correlation between these items. Reported red meat consumption may be negatively correlated with reported consumption of white meat [23]. It has been recommended that red meat should be replaced by white meat, and the ratio of red meat to white meat consumption has been used as an indicator of compliance with this recommendation. Several studies on breast and colorectal cancer, including the Nurses Health Study [23] and the Health Professionals Follow-up Study [24], have placed great emphasis on this ratio. It has implications in the recommendation of consumptions of different food items and public health to examine the association between the ratio of red meat to white meat consumption and diseases. In our study we found consistent results for red meat consumption and the ratio of intake of fresh red meat to chicken and fish (Table 4).

Some methodological issues must be considered when interpreting the findings of this study. Residual confounders, dietary or non-dietary, may provide an alternative explanation for the observed associations. In this study, we have taken into account physical activity, smoking, and dietary intake of fruit, vegetable, vitamin C, vitamin E, beta-carotene, and polyunsaturated fatty acids as potential confounders. We have controlled for educational level and country of birth which may have captured some differences in terms of socioeconomic status in the analysis, since there is evidence that socioeconomic factors affect the utilization of joint replacement procedure [25, 26]. However, it is possible that some residual confounding due to socioeconomic factors persists. Those of higher socioeconomic status are more likely to adhere to dietary recommendations regarding reduction in red meat intake. This would tend to bias the results towards an inverse relationship between meat consumption and the risk of joint replacement if residual confounding was the main explanation. Moreover, we would expect the same effect to be seen for knee and hip replacement which was not the case. In addition, the beneficial effect we observed was for fresh red meat but not for processed meat, suggesting that the effect is not simply due to residual confounding since consumption of these often follows similar socioeconomic patterns [27]. Multiple testing is another issue that needs to be considered, which may lead to the possibility of a spurious finding. When examining the association between meat consumption and the risk of joint replacement, we had 8 main analyses, 4 for the knee and 4 for the hip, for the association of fresh red meat, processed meat, chicken and fish. The ratio of fresh red meat to chicken and fish was a composite of the others and the finding was consistent. Thus using Bonferroni correction, a p value of 0.006 (0.05/8) would be considered significant. Our results were similar (Table 4). However, this correction does not take into account the dose-response relationship we observed, so needs to be considered with caution. Although there is the potential issue of multiple testing, this study is still a valid hypothesis generating study.

The strength of the study was the prospectively collected meat consumption data prior to joint replacement surgery. We also examined meat consumption at baseline and at first follow-up. At baseline, a FFQ was used which provided a validated accurate method for assessing meat consumption and determining energy intake. At first follow-up a simplified, limited questionnaire was administered to determine meat intake. Despite using the different methods, a similar effect was observed for fresh red meat consumption and the reduced risk of hip, but not knee replacement, providing further support for the beneficial effect of fresh red meat consumption on hip replacement risk and also suggesting that the observed effect was not dependent on the instrument used to determine meat intake. This is perhaps not surprising since we found a moderate correlation between baseline and follow up meat consumption in this healthy population, suggesting relatively stable meat consumption over the 4 years period, and supporting the notion that nutrient intake is relatively stable and tends to be more stable with increasing age [28, 29].

Random error in measuring meat consumption at baseline and first follow-up may have occurred in this study, but is most likely to have attenuated the associations we observed. Because the FFQ did not measure portion sizes, the associations would be further attenuated if between-person differences in portion size contribute to between-person variability in amount consumed. Another potential limitation is the moderate reproducibility of the FFQ for the assessment of meat consumption, with kappa values ranged from 0.42 to 0.60 for the reliability study which assessed the FFQ twice at 12 months apart. However, the reliability is similar to others [30, 31], and use of self-reported meat intake at MCCS baseline and first follow up provided similar results. Furthermore, this within-person error in reporting meat consumption is likely to be non-differential misclassification and thus may have diluted rather than identified a relationship. We did not have complete and reliable joint replacement data for the study participants prior to 2001. Although we excluded those MCCS participants who reported a joint replacement prior to 2001 at the second follow-up, this information may be unreliable and is only known for 68% of the original cohort. As a result, some misclassification of joint replacement status may have occurred. It is likely to have been non-differential in relation to the meat consumption measures and thus, may have attenuated any of the observed associations.

The mechanism for the relationship between fresh red meat consumption and the decreased risk of hip joint replacement at this stage is speculative. Bones are in a constant state of change through the process of remodeling which continues throughout life. Red meat is a rich source of protein and minerals, especially iron, zinc, selenium and magnesium, and a great source of vitamin B12 and long-chain omega-3 fatty acids. Iron, zinc, selenium and magnesium play an important role in bone metabolism and dietary intake of these minerals is positively associated with bone mass and strength [32]. Red meat is also a useful source of vitamin D which is essential for bone health. Higher animal protein consumption has been shown to be associated with increased bone mineral density and decreased bone loss [9, 10]. There is increasing evidence that hip joint shape is modifiable and that subtle changes in hip joint shape are a risk factor for hip OA [33, 34]. This is in contrast to knee OA where joint structure does not seem to be as important. It has been shown that certain deformities, such as the flattening of the femoral head, especially in conjunction with a sharper transition from the femoral head to the lower part of the neck are associated with increased risk of hip OA [34]. It may be that fresh red meat consumption reduces the risk of hip OA and subsequent hip replacement via increased bone strength and therefore less change in hip joint shape and less development of the types of hip deformities associated with hip OA. Whether this is the case will require further work.

A high level consumption of fresh red meat was associated with a decreased risk of hip, but not knee, joint replacement for OA. One possible mechanism to explain these differential associations may be via an effect of meat intake on bone strength and hip shape. This should be taken as a hypothesis generating study, and further confirmatory studies are warranted.