• rheumatoid arthritis
• spondyloarthritis
• psoriatic arthritis

With interest I read the Viewpoint of Landewé in which he discussed six domains in rheumatology in which progress has been made in the last decennia, and where the occurrence of overdiagnosis and subsequent overtreatment could be a disadvantageous side effect.1 I appreciate and agree with his concerns. Indeed, the burden of unnecessary treatment should be avoided for individual patients and for society as a whole, especially if expensive drugs (eg, biologics) are applied.

The Viewpoint is written from the general concept, referring to advances in rheumatoid arthritis (RA), spondyloarthritis and psoriatic arthritis. With respect to the paradigm ‘early diagnosis’, Landewé included a remark concerning the recent efforts to identify patients at risk for RA in the symptomatic phase that precedes the development of clinically evident arthritis.1 When discussing the risks of overtreatment as a drawback of early identification of RA, it is important to put the discussion in the correct context.

Arthralgia suspicious for progression to RA (or clinically suspect arthralgia, CSA) is a novel and relatively young area of research. At present, there is no evidence that treatment in this phase is more effective than treatment initiated in the phase of clinical arthritis. Several proof-of-concept trials are investigating this assumption in autoantibody-positive arthralgia or CSA2; results are awaited in the next 5 years. As long as there is no evidence, patients with CSA are not treated or should not be treated with disease-modifying antirheumatic drugs (DMARDs) in clinical practice. Thus, overtreatment is not yet an issue in CSA. Importantly, when positive results of the ongoing proof-of-concept trials would appear and will be reason to treat patients with arthralgia, overtreatment may become an issue since the European League Against Rheumatism definition of CSA has insufficient specificity, when used alone, for RA development and the optimal combination of risk factors to be used on top of the clinical definition is not yet known.3 4 However, let us keep in mind that the field of CSA is a novel area of research of which the results are not yet ready for implementation in rheumatological practice.

With respect to the early identification of RA and the risk of overtreatment, the discussion should focus on the phase of early undifferentiated arthritis (UA). DMARDs are often started in patients presenting with UA, while all placebo-controlled randomised clinical trials evaluating the efficacy of starting treatment in this phase are negative, as no statistically significant beneficial effect was shown.5–9 This lack of evidence is possibly caused by insufficient power, as a recent meta-analysis revealed a reduced risk of onset of RA when the results of these different trials were combined (HR 0.68–0.74 depending on medication used).10 Insufficient risk stratification of patients with UA may also have contributed to negative findings, as suggested by a recent post hoc analysis on data from the PROMPT trial. On group level, the risk to progress from UA to RA was 30%; after risk stratification, a high-risk group (risk of RA development in the next year >80%) was identified. Analysis of data from this high-risk group showed that up to 50% of patients treated with a 1-year course of methotrexate did not progress to RA and >35% obtained a sustained DMARD-free status, whereas all placebo-controlled patients progressed to RA and none achieved a sustained DMARD-free status.11 This example illustrates that adequate risk stratification resulted in an increased power to detect a risk-reducing effect and might considerably reduce overtreatment.

Currently, there is insufficient data on the degree of overtreatment in UA. However, overtreatment is inevitable in medicine and the most important question is to what extent overtreatment is considered acceptable. Persons who have a cervical intraepithelial neoplasia 3 (this includes cervical carcinoma in situ) have a 31% risk to progress to cervical cancer in the next 10 years, based on the natural disease course.12 Overtreatment will occur here as well, but intervention is accepted by the large majority of women in this situation. Another example is the treatment of hypertension to prevent stroke. The number of patients needed to treat (NNT) to prevent 1 stroke over 5 years of treatment is up to 118.13 This is considered acceptable and treatment of hypertension is widely recommended. The NNT in the post hoc analysis of the PROMPT trial was 2.11 I agree with Landewé that more research is needed, especially more trials in the field of UA that use risk stratification while including patients14 and relevant outcome measures (eg, sustained DMARD-free status or workability). In addition, discussions on the acceptability of early treatment of UA should be continued and the severity of the outcome (RA might be considered different from cancer and stroke) and the burden of treatment should be included. Nonetheless, the presently available data suggest that these discussions could result in the risk of overtreatment of patients with UA being considered acceptable, especially if conventional DMARDs are used.