Messages from paper are incorrect

EDITOR,—Paul Brennan and colleagues found that rheumatoid factor, disease lasting longer than three months, the involvement of at least two large joints, and male sex were significant predictors of radiological erosions in patients with arthritis in primary care.1 The first two findings are consistent with, but the latter two conflict with, previous data. Is this important, and why has it arisen?

Entry criteria—The authors used the American College of Rheumatology's criteria for rheumatoid arthritis. These criteria, however, were not devised or intended to be used for early presentation in hospital, let alone in the community.

Study design—Baseline observations were correlated with radiological findings at roughly one year; thus an existing erosion (which in a community setting may be asymptomatic and unrelated to disease2) was equated with a new erosion. Radiological examination was also performed a variable time after the start of drug treatment (which influences the development of erosions). Finally, clinical features were not validated: in early disease repeated observation may be required to confirm a diagnosis.

Male sex—In table 5 of the authors' paper the entry “No” in the column headed “Male sex” means that there was a positive association with female sex. The one association with male sex shown in this table is incorrect as the paper showed increased radiological damage in women. The paper is said to include all recent studies of over 100 patients. However, the largest paper is cross sectional and cannot relate to prediction, several papers are not confined to early disease, and the well recognised genetic association with erosions is not discussed—nor is a longitudinal study that, consistent with all subsequent longitudinal studies, showed associations between rheumatoid factor, genetic factors, and erosions4 (and validated the previous original observation5).

The new element in the algorithm proposed by the authors is the involvement of two large joints. Misleadingly, the authors do not discuss the pathogenic relation in rheumatoid arthritis whereby small joint disease leads sequentially to large joint involvement. “Isolated” large joint disease has a good prognosis; the paper shows that only two of 10 patients with this developed erosive disease.

The new messages from this paper are not only incorrect but possibly harmful to patients. Doctors may think that women have a good prognosis and delay referral until large joints are involved. The more relevant question of whether disability is truly preventable will be answered by interventional studies currently in progress.