The Ritchie Index Paper: some historical background on an early effort to establish objective measures in rheumatic diseases

Introduction

A very influential paper in rheumatology is the paper by Ritchie et al. ¹ which introduced what became known as the ‘Ritchie Articular Index’ (RAI), a measure of joint tenderness in patients with rheumatoid arthritis. The paper is notable for its impact in providing an objective measure of joint pain. It was written at a time when there was considerable interest in the development of outcome measures in rheumatoid arthritis. Outcome measures were desired both for observational studies and for clinical trials. The RAI was used as an outcome measure in early trials of Disease Modifying Drugs in rheumatoid arthritis ² but has had its greatest impact through its inclusion in the Disease Activity Score, ³ a composite outcome measure developed in the early 1990s. In addition to its notable scientific input, the paper deserves note as well as an example of a multi-author paper which benefitted particularly from being a collaborative effort. That is the primary focus of this short note in which the historical background to the paper and the individuals involved in its preparation are discussed.

Historical background

The use of joint counts, with or without a scoring system, is now an established tool in rheumatological research. The paper by Ritchie et al., published in 1968, was an early contribution to the development of such measures. A decade earlier, efforts ⁴ had been made to investigate single measures of disease activity by considering composite indices based on various measures of disease activity such as duration of morning stiffness, grip strength and erythrocyte sedimentation rate along with a count of the number of painful joints. Co-operative work undertaken by the American Rheumatism Association (ARA),^5,6 the forerunner of the American College of Rheumatology, gave the search for such measures renewed momentum in the late 1960s although opinions on their usefulness varied. As part of these efforts, the ARA introduced a particular joint count which was the number of clinically active joints determined by any one of the following: tenderness on pressure, pain on passive movement and swelling other than bony proliferation. Some joints were excluded such as the cervical spine with the number of joints included being 68 although a 66 joint count with hips excluded was also used. This was the forerunner of the ACR 66 and 68 joint counts still in widespread use today.

During this period of time, in 1965, through the efforts of the Member of Parliament Margaret Herbison, the Centre for Rheumatic Disease (CRD) was opened in Glasgow in Scotland, which previously had no designated rheumatology service. Watson Buchanan, a senior lecturer in the University of Glasgow, was appointed to lead this new Centre, which was part of the University’s Department of Medicine, Royal Infirmary. ⁷ The Ritchie paper represents an early research project from this Centre, with funding from the Arthritis Research Council in Great Britain, the principal investigator likely being Buchanan.

A brief summary of the paper

The Ritchie paper is self-described as confined to developing a measure of joint tenderness. The index used a grading of tenderness with scores from 0 to 3 measuring severity rather than a simple yes/no determination of tenderness. Some groups of joints were regarded as a single unit and all joints (or sets of joints) scored equally. There were 26 tenderness grades that contributed to the measure. Four investigations were reported in the paper. The first compared the proposed index with the ARA joint count and showed good correlation although more high scores were observed with the ARA measure. Inter-observer error was examined in 72 patients with 4 experienced observers. One observer examined all patients and 3 others examined 24 patients each. The observed differences were such that the use of the index by different observers could not be recommended for comparison purposes. In 18 patients, one observer examined each patient twice in one morning with a high degree of reproducibility. Within-patient variation was examined in 21 stable patients who were assessed on 2 successive days at the same time of day. The average difference was small. The final investigation was based on carrying out a randomised clinical trial in which patients were given 1 week of treatment with each of a placebo, aspirin and prednisolone, with the order of treatments randomised. Demonstrable improvements were seen between the active treatments and placebo and between prednisolone and aspirin. A final section of the results illustrated that the articular index scores were not correlated with grip strength and minimally correlated with time to walk 50 ft, two measures of function.

The authors

The Ritchie paper had seven authors. Watson Buchanan, who headed up the new CRD, was the senior author on the paper. Buchanan’s leadership led to the CRD soon becoming known for its clinical excellence, postgraduate training and research. The paper was one of the key research outputs from the CRD and must have been one of the earlier efforts by Buchanan to establish a research climate. Buchanan^8,9 was born in Glasgow and studied medicine there. He then took a post as a research fellow at the Western Infirmary and specialised in endocrinology, receiving an MD from the University of Glasgow in 1962 for a thesis on autoimmune thyroiditis. He subsequently spent 2 years at the National Institutes of Health (NIH) in Washington DC during which he produced a significant paper on Sjogren’s syndrome. After returning to Glasgow to head up the CRD as a senior lecturer, he received a professorial appointment in 1972. He departed Glasgow in 1979 to take up a professorship in rheumatology at McMaster University in Canada. The rest of his career was spent at McMaster although an anecdotal recollection (personal communication) suggests he was not enamoured of the cold Canadian winters. This man was described in an obituary as having a ‘larger than life, charismatic personality’ who could ‘transmit his own unquenchable intellectual curiosity and enthusiasm to those who had the pleasure and privilege of working with him’. ⁸ In addition, however, he was ‘also seriously and outspokenly critical of some of the developments in problem-based undergraduate medical education, and what he saw as the growing fashion and tyranny of evidence based medicine’. ⁸ He died in January 2006, a day after collapsing while reciting a poem at a Burn’s club in Ontario, Canada.

Buchanan, in 1971, published the highly regarded textbook, Clinical Rheumatology, with his co-author J. Anthony (Tony) Boyle. ¹⁰ Boyle ¹¹ also was born and medically trained in Glasgow, being granted, in 1965, a MD for work on non-toxic goitre. He joined the CRD at the same time as Buchanan and left in the mid-1970s to enter the pharmaceutical industry, eventually moving to the United States where he died in 2008. Boyle was the second author on the paper.

Two other medically qualified authors on this paper were Mukundrai K. Jasani ¹² and Theodore G. Dalakos (https://obits.syracuse.com/us/obituaries/syracuse/name/theodore-dalakos-obituary?id=18256631) who had posts at the CRD during the development of the paper. Jasani carried heavy family responsibilities during his youth in Kenya but nevertheless earned a bursary to study medicine in the United Kingdom, graduating first in his class at the University of Glasgow. Choosing to specialise in rheumatology, at least partly motivated by his sister’s debilitating juvenile inflammatory arthritis, he was awarded a research fellowship to study rheumatoid arthritis at the CRD. Dalakos was born in Greece and received medical training there after a difficult childhood under Nazi occupation during World War II and during the subsequent civil war. He was appointed to a residency in endocrinology at the Glasgow Royal Infirmary, likely being part of the CRD during this time. It seems most plausible that Jasani and Dalakos, along with Buchanan and Boyle, were the four ‘observers’ in the studies in the Ritchie paper and provided the necessary joint assessments. By the time of publication of the paper, Jasani had moved to the CIBA Laboratories in Horsham Sussex as a clinical scientist. He subsequently worked at Guy’s Hospital in London and practised privately until 2017, dying in 2024. Similarly, Dalakos had moved by the time of the paper’s publication, in his case, to take up a fellowship at Upstate Medical University in Syracuse, New York, United States. He spent 7 years in research before moving to private practice where he worked until 2005, dying in 2020.

It can be conjectured that, in trying to establish a research culture at the CRD, Buchanan and Boyle would need collaborative help for some projects. For the Ritchie paper, it appears that there was a significant link between Glagow and the CIBA laboratories in Sussex. Another author of the paper was Phillida Grieveson whose present address on the paper is listed as the CIBA labs, as is Jasani. She is described in the paper as being in receipt of a grant from the National Institutes of General Medical Sciences in the United States, and this grant was awarded to the CIBA laboratories for research related to the metabolism, pharmacokinetics and absorption of various drugs. Other work supported by this grant on which Grieveson is an author is a study of the metabolism of chlorpropamide in diabetic patients. ¹³ Grieveson’s role in the work reported in the paper is not specified although she may have provided the link between Glasgow and Sussex as the CIBA laboratories provided the aspirin, prednisolone and placebo treatments for the reported clinical trial, although another individual at the laboratories was thanked for advice and the actual preparation of the treatments, Dr. Denis Burley.

The final ‘medical’ author of the Ritchie paper is the first author, Dorothy M. Ritchie. Ritchie, an occupational therapist, is described, on a web page devoted to the history of rheumatology in Glasgow (https://www.gla.ac.uk/schools/medicine/mus/ourfacilities/history/20thcentury/1948-2018/rheumatology), as recruited by Buchanan in the early days of the CRD. It is entirely conjecture as no information is available, but Ritchie may well have played a major role in recruiting and supporting the patients who were examined for the purposes of this paper.

One additional author of the Ritchie paper is John McInnes. Although his middle initial is given on the paper as M, his middle name is in fact Wallace. McInnes provided the statistical analyses reported in the paper. His address on the paper is listed as G. and J. Weir Holdings Ltd, Glasgow. This company was formed in 1959 and had its origins in the G. and J. Weir and Co., a partnership of consulting engineers formed in 1871/1872 in Liverpool which moved to Glasgow in 1873. Since 1969 it has been known as the Weir Group (https://www.gracesguide.co.uk/G._and_J._Weir). McInnes was employed in a metallurgy group in this company and recalled, in 2025, that he was seconded to the Glasgow Royal Infirmary for a period of 6 months to work with the group developing an articular index. He remembers Watson Buchanan as charismatic and persuasive, and it was Buchanan who had a link to people in engineering and iron manufacture, and a direct connection with the subsidiary company Weir Pumps Engineering in which McInnes worked. It is possible that Buchanan may have known people through his father who is described in an obituary of Buchanan as a ‘sales representative in iron and steel manufacturing’. ⁸ In any event, Buchanan’s character would certainly be consistent with him searching out the statistical help that he needed for this research from any available source. Interestingly, also in 2025, McInnes had the somewhat vague recollection that the choice of the first author was made randomly in the first instance, and he declined to be chosen given his non-medical background. An index which might be thought to be surprisingly linked to the name of an occupational therapist might perhaps even more surprisingly have been linked to the name of an industrial statistician.

The statistics

John McInnes recalled in 2025 that the majority of the statistical methods used in the Ritchie paper were ‘regression methods’ which were ‘brought across’ from his industrial work. Computation was carried out on a Burroughs 2500 mainframe computer, which typically had Cobol and Fortran compilers. A simple correlation coefficient, r = 0.89, was used to compare the proposed index with the ARA index. No uncertainty was associated with the coefficient for this purpose. To study inter-observer agreement, paired t-tests were used for comparing one observer (A) who evaluated all 72 patients with the three others (B, C and D) who examined 24 patients each. No statistically different mean differences were observed between observer A and the other three, although it was noted that the variability associated with the observed patient differences varied across observers B, C and D. This was formally examined for the largest observed difference by an F-test. In addition, a statistical comparison between the highest and lowest correlation coefficients resulting from these comparisons was deemed significant (significance level = 0.015). It can be inferred from this result that Fisher’s z transformations of the correlation coefficients were used in making this comparison, necessary because the correlations, 0.91 and 0.98, were high. Simple histograms were also produced of how often observers differed in classifying a joint as painful (any degree) or not, with a notable number of patients with differences in 10 or more joints being observed. This would appear to relate to the set of single joints rather than the 26 sets of joints examined in the index, but it is not explicitly reported. Paired t-tests were also used to examine intra-observer differences and within patient differences across 2 days, along with histograms showing much less disagreement on activity in individual joints than that between observers. The analysis of the clinical trial of placebo, aspirin and prednisolone was also based on paired t-tests using both the proposed index and the ARA index. Largely comparable conclusions would be drawn although it was highlighted that the proposed index produced larger differences and larger standard errors, as might be expected since one assigns a grade to joint pain and the other is based only on a yes/no evaluation. All three possible comparisons were reported. Additionally, two figures are given displaying the relationship between other activity measures and the joint index. Comparing the index applied to the hands with a measurement of grip strength, no relationship was evident. However, there was some correlation, r = 0.55, between the index in the legs and the time to walk 50 ft. Some general observations are made on the figures, and it is pointed out that a formal regression analysis relating time to walk 50 ft to the proposed index would conclude that approximately 30% of the variation would be ‘explained’ by the joint index. The impact of the distribution of the ‘explanatory variable’, the index, in the two comparisons is notably highlighted with the larger number of joint sets in the legs compared with the hand meaning that there is more variation in the index values observed and the regression analysis is likely to be more informative. A final remark on the statistical aspects of the Ritchie paper is that it illustrates not only the value of statistical input but also the early days of its reporting. The details given of the statistical methodology used and of the data examined, including simple things such as sample sizes involved in all the various analyses, are often minimal or missing.

‘Helpful criticism’

Two individuals are acknowledged in the Ritchie paper for helpful criticism. William O’Brien from Virginia, United States, who was a key figure in the ARA Co-operating Clinics trial publications using activity measures, was one. He is cited in the paper as having carried out a large-scale review of the impact of objective measures on assessments of drug efficacy for one of the ARA studies. ⁵ The other, also a key figure in these same publications in terms of statistical advice, was Donald Mainland.^14,15 (https://www.jameslindlibrary.org/articles/donald-mainland-anatomist-educator-thinker-medical-statistician-trialist-rheumatologist/) (https://www.jameslindlibrary.org/articles/mainlands-elementary-medical-statistics-1952-a-pivotal-text-in-statistical-pedagogy/) The latter also published, based on his work with the ARA, a single-author paper on the estimation of activity in rheumatoid arthritis using composite indices ¹⁶ which was cited by Ritchie et al. Mainland was very influential in the development of medical statistics in North America. He was medically qualified and worked in anatomy before shifting his focus to statistics based on his recognition of the value of statistical methods through contact with the noted statistician RA Fisher in the United Kingdom. After taking a position as Professor of Medical Statistics at the New York University Medical Center, his main medical collaborations appear to have been in the area of rheumatology, particularly in clinical trials, an area of work which he moved to New York to enable. One can infer that both these individuals helped in understanding both current efforts in the development of indices and the required research rigour necessary to contribute to this field. It is not known whether Watson Buchanan would have met these individuals during his 2 years in the United States, but it is not beyond the realm of possibility given the research emphasis at the NIH.

Conclusion

The early development and assessment of an objective measure of joint pain in rheumatoid arthritis reported by Ritchie et al. had a major impact on subsequent research efforts, both through its application to clinical investigations and the motivation it gave to subsequent efforts to improve the assessment of outcomes as randomised trials became increasingly common in rheumatology. A major boost to trials, particularly in Europe, was the development of the DAS index ³ which used a weighted average of the RAI, a swollen joint count, the erythrocyte sedimentation rate and a general health assessment as a composite disease activity measure. The DAS, and subsequent simplifications of it which did not use the RAI specifically but relied on an unweighted tender joint count, have been used to develop, for use in trials, the European Alliance of Associations for Rheumatology (EULAR) definitions of response, remission and low, medium and high disease activity.

Central to the development of the RAI, were the efforts of Watson Buchanan in establishing a research culture in Glasgow, through the recruitment of young rheumatologists and an allied health professional to the CRD to support this and his contacts with researchers in other locations, including the United States. And Buchanan should also be commended for seeking the necessary statistical support to assess and report the results of his clinical research. He also modelled the importance of this in arranging the secondment of a statistician to the CRD for 6 months to significantly engage with this work. The value of such a collaborative research structure is now well known, and those who pioneered this model should not be forgotten.