Diet and Rheumatoid Arthritis Development
Diet and Rheumatoid Arthritis Development
In general, the accumulated evidence regarding the associations between dietary factors and risk of RA is limited and results are not consistent. This may, in part, be related to the different methodologies that researchers have used in analyzing such a complex exposure as diet.
First of all, the results could be influenced by the choice of study design. Case–control studies may be affected by recall bias, a systematic error due to a different recall of the exposure status between cases and controls. Receiving a diagnosis of RA may cause a more accurate recall and better reporting of dietary habits compared with healthy controls who are not as focused on their health. Therefore it is difficult to draw conclusions based on findings from case–control studies that could either be an overestimate or underestimate of the true risk owing to this type of bias. The prospective cohort design is preferred as it is not affected by recall bias, since the collection of dietary information occurs when all members of the cohort are not affected by the disease. Moreover, cases included in a case–control design may have already changed their diet at the time of the interview owing to the developing of RA, leading to biased estimates of the risk. However, prospective, as well as retrospective, studies can be affected by nondifferential misclassification of the exposure. In fact, people tend to report their food consumption according to what they think is socially acceptable. For example, women tend to report a lower alcohol consumption since high alcohol consumption is considered an unhealthy behavior, while they tend to report higher intake of fruits and vegetables that are considered healthy. Such type of misclassification usually leads to biased estimates.
All studies included in this review used a food frequency questionnaire, with the exception of the studies conducted by Pattison et al. who used a 7-day food diary. The food diary is a more precise way to collect information regarding daily diet, but the time period of only 1 week does not allow an assessment of long-term diet and seasonal changes. Moreover, to take into account changes in diet over time and better assess the influence of food consumption on the risk of a disease it is important to collect dietary information at different points in time. Of the studies presented in this review, only the SMC and the NHS and NHS II collected information at two or more occasions using a food frequency questionnaire.
Incident cases of RA were identified in multiple ways by the studies. Case–control studies identified cases from the rheumatology or internal medicine departments of hospitals, while prospective cohort studies identified cases in two different ways: some studies linked the cohort to national registers, while other studies validated self-reported RA cases by collecting medical records. The use of self-reported and subsequently validated RA avoids the inclusion of misclassified RA cases; however, true RA cases that have not self-reported their status are missed and included in the study as noncases.
The studies conducted in the EPIC-Norfolk cohort analyzed inflammatory polyarthritis cases, of which only 40% satisfied the ACR criteria for RA definition. The authors argued that they decided to use inflammatory polyarthritis and not RA definition because the RA criteria did not perform well in the setting of early disease. They conducted analysis stratified by RA status, which did not reveal any difference from the results reported for inflammatory polyarthritis.
Finally, some studies could have been affected by problems related to a low statistical power. One nested case–control study included as few as 14 cases, while the EPIC-Norfolk study identified only 73, and later 88 cases of inflammatory polyarthritis. Among prospective cohort studies, the DCH cohort identified only 69 RA cases in a cohort of 57,053 men and women during an average follow-up of 5.3 years. The cohort with the largest number of cases was the NHS with 546 cases identified among 82,063 women (Table 1 & Table 2).
Studies on subtypes of RA, such as RF positive or negative and ACPA positive or negative, are also very limited. Only six studies stratified accordingly to subtypes of RA and found different results between positive and negative RA cases. The reasons for this could be either the small number of RA cases identified may prevent further stratifications or problems in retrieving information regarding the subtype's classification for each case.
Methodological Considerations
In general, the accumulated evidence regarding the associations between dietary factors and risk of RA is limited and results are not consistent. This may, in part, be related to the different methodologies that researchers have used in analyzing such a complex exposure as diet.
First of all, the results could be influenced by the choice of study design. Case–control studies may be affected by recall bias, a systematic error due to a different recall of the exposure status between cases and controls. Receiving a diagnosis of RA may cause a more accurate recall and better reporting of dietary habits compared with healthy controls who are not as focused on their health. Therefore it is difficult to draw conclusions based on findings from case–control studies that could either be an overestimate or underestimate of the true risk owing to this type of bias. The prospective cohort design is preferred as it is not affected by recall bias, since the collection of dietary information occurs when all members of the cohort are not affected by the disease. Moreover, cases included in a case–control design may have already changed their diet at the time of the interview owing to the developing of RA, leading to biased estimates of the risk. However, prospective, as well as retrospective, studies can be affected by nondifferential misclassification of the exposure. In fact, people tend to report their food consumption according to what they think is socially acceptable. For example, women tend to report a lower alcohol consumption since high alcohol consumption is considered an unhealthy behavior, while they tend to report higher intake of fruits and vegetables that are considered healthy. Such type of misclassification usually leads to biased estimates.
All studies included in this review used a food frequency questionnaire, with the exception of the studies conducted by Pattison et al. who used a 7-day food diary. The food diary is a more precise way to collect information regarding daily diet, but the time period of only 1 week does not allow an assessment of long-term diet and seasonal changes. Moreover, to take into account changes in diet over time and better assess the influence of food consumption on the risk of a disease it is important to collect dietary information at different points in time. Of the studies presented in this review, only the SMC and the NHS and NHS II collected information at two or more occasions using a food frequency questionnaire.
Incident cases of RA were identified in multiple ways by the studies. Case–control studies identified cases from the rheumatology or internal medicine departments of hospitals, while prospective cohort studies identified cases in two different ways: some studies linked the cohort to national registers, while other studies validated self-reported RA cases by collecting medical records. The use of self-reported and subsequently validated RA avoids the inclusion of misclassified RA cases; however, true RA cases that have not self-reported their status are missed and included in the study as noncases.
The studies conducted in the EPIC-Norfolk cohort analyzed inflammatory polyarthritis cases, of which only 40% satisfied the ACR criteria for RA definition. The authors argued that they decided to use inflammatory polyarthritis and not RA definition because the RA criteria did not perform well in the setting of early disease. They conducted analysis stratified by RA status, which did not reveal any difference from the results reported for inflammatory polyarthritis.
Finally, some studies could have been affected by problems related to a low statistical power. One nested case–control study included as few as 14 cases, while the EPIC-Norfolk study identified only 73, and later 88 cases of inflammatory polyarthritis. Among prospective cohort studies, the DCH cohort identified only 69 RA cases in a cohort of 57,053 men and women during an average follow-up of 5.3 years. The cohort with the largest number of cases was the NHS with 546 cases identified among 82,063 women (Table 1 & Table 2).
Studies on subtypes of RA, such as RF positive or negative and ACPA positive or negative, are also very limited. Only six studies stratified accordingly to subtypes of RA and found different results between positive and negative RA cases. The reasons for this could be either the small number of RA cases identified may prevent further stratifications or problems in retrieving information regarding the subtype's classification for each case.
Source...