Dr. Fries, who is now an emeritus professor of immunology and rheumatology at Stanford, concurs: “The data banks have demonstrated the importance of long-term observational studies in providing us with identifiable and verifiable information on issues like mortality, costs of the illness, and drug side effects. You can’t do randomized, clinical trials on those topics.”
Through databases, researchers have developed validated techniques to allow them to compare the effectiveness of different drugs and drug combinations used for treating rheumatologic disorders, says Dr. Fries. “Optimal treatment depends on using the best, not necessarily the newest, drugs, and for obvious reasons, drug companies won’t sponsor those studies,” he says. Private databases pick up the slack.
Long View on Chronic Disease
“Chronic diseases typically are studied for too short a time,” explains Theodore Pincus, MD, professor of medicine and microbiology at Vanderbilt University Medical Center in Nashville. “Two or three months might be sufficient to study the outcomes of an illness like acute pneumonia or myocardial infarction, but chronic conditions like rheumatoid arthritis must be studied for five to 20 years. Longitudinal databases are essential to that task.” Many of Dr. Pincus’ patients are enrolled in the NDB, and he has analyzed his own patients at Vanderbilt.
Dr. Pincus and Tuulikki Sokka, MD, PhD, a rheumatologist at Jyväskylä Central Hospital in Jyväskylä, Finland, have described several limitations of short-term clinical assessments that can be overcome by studying large numbers of patients for many years (Clin Exp Rheumatol; 23(Suppl 39):S1-S9).
Overall, the best way to assess outcomes and the functional impact of the disease is through long-term patient self-reports on issues like pain, fatigue, and global status, Drs. Pincus and Sokka write. Indeed, says Dr. Wolfe, “Loss of functional ability is the best predictor of adverse outcomes, and the only way to really assess that is through patient self-reports” such as those gathered by the NDB.
Databases are also an invaluable resource for studying rare events. Such an event might be a drug side effect, such as malignancy, which might not be picked up in a clinical trial either because too few patients were studied or because they weren’t studied long enough—or because a disorder itself might be rare.
“If you have reports from 3,000 rheumatologists, each of whom has two patients with polymyositis or scleroderma, that’s 6,000 patients and an informative database,” says Dr. Pincus.
Finally, databases provide important lessons on the fine points of data collection. Back in the early 1970s, Dr. Fries and his colleagues envisioned the project that eventually became ARAMIS as a way of maintaining objective, longitudinal data that would use then-current computer techniques to make clinical research more efficient. They quickly encountered problems, including a high dropout rate and data that were sketchy or missing altogether.