The predictive value of attributes of pain to classify rheumatoid arthritis

Out of 2300 patients with rheumatic diseases 1627 were analysed to develop a classification of rheumatoid arthritis based on clinical attributes of pa in. Of these, 641 patients had the disease and 986 were controls with other rheumatic conditions. For traditional format classification, six of eight variables were selected: pain at a fixed joint; symmetrical pain; continuou s pain; pain mainly present at night or in the morning; pain following join t pressure; and pain decreased by load/movement. The occurrence of four or more of these features was associated with a 72.1% sensitivity and a 79.1% specificity. A classification tree constructed on four features that showed the greatest diagnostic power (symmetrical pain, pain mainly present at ni ght or in the morning, pain at joint pressure, continuous pain), was associ ated with a 75.8% sensitivity and a 77.0% specificity.