By the mid-1970s, several researchers in the United States and Canada had already been collecting longitudinal data on their rheumatology patients. An initial group of six institutions (including Joseph Levinson, MD, and Dr. Hess at the University of Cincinnati; Donald Mitchell, MD, at the University of Saskatchewan, Saskatoon; Thomas Medsger, MD, at the University of Pittsburgh, and others) agreed to pool data, with Dr. Fries’ Stanford group serving as coordinators of the entire project. NIH-sponsored conferences quickly followed, to agree upon terminology and uniform data storage units.
You Might Also Like
Explore This IssueJune 2007
Also By This Author
A Rich Vein of Information
First funded in 1976, ARAMIS was intended to be a framework for enrolling patients with rheumatic diseases and following them for life. Five years into the project, however, investigators discovered that there was an 80% dropout rate. The unit of collection was through the investigator, so patients were being lost as they changed physicians or investigators came and went. In addition, tests were not being performed uniformly on all patients at standard intervals.
To improve follow-up and correct the biased data collection, the ARAMIS team changed from the doctor to the patient as their unit of follow-up, and imposed a protocol on the observational study. Thus, the team developed the Health Assessment Questionnaire and instituted intense follow-up to ensure patient compliance.3 As Dr. Fries puts it, “We had been accidentally compelled to invent patient-reported outcomes.”
Other historical phases followed, such as the creation of infrastructure methodology when investigators developed the complete rheumatic disease criteria sets. In addition, the wealth of data in ARAMIS allowed the investigators to make a series of other breakthroughs, such as identifying a successful treatment for scleroderma renal crisis, discovering NSAID gastropathy, and reversing the standard disease treatment pyramid. Looking at the results from a variety of treatment strategies and patterns over time it became clear, says Dr. Fries, that physicians were missing therapeutic opportunities. NSAIDs, regarded as more benign than disease-modifying antirheumatic drugs (DMARDs), were in fact more toxic. And the latter, if prescribed earlier in the disease continuum, improved health outcomes.
These revelations had direct effects on improving treatments for arthritis patients. Pharmaceutical companies were spurred to develop NSAIDs using selective COX-1 sparing mechanisms as a way of protecting the stomach and physicians began to view these drugs in a more realistic light. The ARAMIS bibliography is now at 931 articles. (The bibliography is classified by author and year and available on the ARAMIS Web site.)
Dr. Fries and his colleagues presented several studies at last year’s ACR meeting in Washington, D.C.: two longitudinal studies of aging and musculoskeletal disability showing that health habits can delay onset of disability; a study demonstrating that hydroxychloroquine (Plaquenil), a DMARD used to treat RA and lupus, prevents the development of diabetes mellitus 2; and a paper on item improvement. The last paper, presented by Dr. Fries, addresses one aspect of a newly mounted NIH initiative to develop improved outcome assessment instruments. Dr. Fries is also a lead investigator in this new initiative, called PROMIS (Patient Reported Outcomes Medical Information System). Drawing on educational testing theory, this effort uses computerized adaptive testing to refine questions to elicit increasingly precise measures on individuals. The hoped-for result will be to reduce the standard error of measurement, and thus reduce required sample sizes for clinical trials by 25% to 40%. The savings for clinical trials sponsors, such as the NIH, would be considerable.