ILAR was first published in 1995 on a proposed system of classification and nomenclature of juvenile arthritis for all nations [1]. Prior to this there had been little uniformity with several different systems of classification and naming in use. The third and final revision (to date) of the ILAR classification was published in 2004 [2] and provided international standardisation for classification and the new name “juvenile idiopathic arthritis” (JIA) which replaced “juvenile rheumatoid arthritis” (JRA) and “juvenile chronic arthritis” (JCA). JIA described seven homogenous mutually exclusive subgroups and one category (“undefined” or “other arthritis”) for conditions that did not fit precisely into only one single category. This classification system has made epidemiological studies more standardised, but many problems remain. Meantime, it is becoming apparent that with better understanding of the various subgroups of JIA, a new classification system may be due, particularly in relation to systemic arthritis (see below). A classification system should be viewed as a living organism, ready to change as the sub-speciality grows and matures, and knowledge of the area becomes more precise.

The diagnosis of JIA remains a clinical one. There are no tests available to confirm or exclude the diagnosis. Hence, an experienced clinician is required to make a diagnosis by defining swollen, tender or joints with limited range of movements. For the inexperienced it is sometimes difficult to do this. For the experienced clinician as well, it can pose a challenge. It is known that many children with JIA remain undiagnosed for long stretches of time, partly because joint pain itself seems to be less in children, and swelling may not be noticed by a parent [28]. This impacts on case ascertainment in epidemiological studies and largely explains the wide variance between true community-based studies and case-based studies, further discussed below. It is mainly for this reason that case-based epidemiology studies will always show lower prevalence than studies based on the clinical examination of a cohort of children presumed to be healthy. Children with the milder forms of JIA such as persistent oligoarthritis are the most likely to remain undiagnosed [11, 29].

Since JIA is a diagnosis of exclusion, one first needs to know what to exclude. The differential diagnosis may be wide. Septic arthritis is a condition that requires prompt treatment. It can sometimes resemble JIA (oligoarthritis form) at presentation, but there are disastrous consequences if the diagnosis is missed. However, it is not uncommon for an unsuspecting orthopaedic surgeon to needlessly open a swollen joint due to JIA, thinking it to be septic. Many a child with JIA bears the scars of having had a joint opened for no good purpose. It is preferable just the same, to overdiagnose septic arthritis initially than to under-diagnose it, as untreated septic arthritis is quickly destructive.

Since there is a requirement for the presence of joint inflammation for 6 weeks for a diagnosis of JIA, this can delay a diagnosis until time has run its course. The most common cause of a swollen joint in a child is a viral illness. A viral swollen joint does not generally last for 6 weeks, and more commonly it has abated in less than 2 weeks.

It is now apparent that one subgroup of JIA, systemic arthritis, is fundamentally different to all other subgroups. It does not comfortably sit in the category of “autoimmune disease” implying adaptive immune system pathology [30]. Instead it has features of increased activity of the innate immune system and increased secretion of certain cytokines, interleukin-1 (IL-1) and interleukin-6 (IL-6). Its behaviour is similar to that of a polygenic auto-inflammatory disease [30]. This reminds us that classification systems should be constantly renewed as we understand more about a condition.

Commonly, case-based studies are utilised in epidemiology studies of JIA, i.e. previously diagnosed cases of JIA. Where so-called “healthy” children are checked within the community for the presence of JIA, such studies can be categorised as true community-based studies (CBSs). It is known that true CBSs will include significant numbers of children who have remained undiagnosed with JIA, usually disease at the milder end of the spectrum such as persistent oligoarthritis [11]. It is logical to presume that true CBSs will show higher results for prevalence and incidence of JIA. True CBSs are expensive and time-consuming and rely heavily on experienced clinicians. In addition, the numbers of children needing to be examined range in the thousands, to identify relatively small numbers of cases for an uncommon disease. By their nature, such studies allow for significant chance variation, i.e. the resulting prevalence figures will have wide 95 % confidence intervals and a need for careful interpretation of the data. In true CBSs such as those of Mielants [4] and Manners [11], there is an increased number of children with the more mild forms whose condition had remained unnoticed by the family and medical authorities. In true CBSs the prevalence is usually significantly higher because of the previously undiagnosed cases. However, in the study of Aggarwal [26] in the Indian subcontinent, a true CBS did not show numbers of undiagnosed children with the more mild forms of JIA. This important study suggests that in India there is a scarcity of the more mild forms of JIA seen often in Caucasian population studies. This provides strong evidence that there is true genetic variation between races (see below).