Rational diagnosis and treatment in unclassified arthritis: how clinical data may guide requests for Lyme serology and antibiotic treatment.

To improve the appropriateness and efficiency of diagnostic serological tests and subsequent antibiotic treatment, clinical data from 102 patients with unclassified arthritis were analysed to investigate whether the presence of positive IgG antibodies to Borrelia burgdorferi could be predicted. The clinical data were blindly ranked from 1 to 4 (1, Lyme arthritis unlikely; 4, Lyme arthritis very likely). Antibodies to B burgdorferi were positive in nine of 102 patients (9%). Six of 15 (40%) patients with rank numbers 3 and 4 were positive for antibodies to B burgdorferi, in contrast with only three of 87 (3%) patients with rank numbers 1 and 2. The likelihood ratio of positive Lyme serology for patients ranked 3 and 4 was 12.0, for patients ranked 2 to 4, 4.5, and for patients with arthritis of the knee, 3.0. These likelihood ratios were associated with a post-test probability of 55, 30, and 20% respectively. The clinical history in patients with unclassified arthritis can largely predict the presence of antibodies to B burgdorferi. The absolute value of a likelihood ratio can be a contributing factor in deciding to request tests for antibodies to B burgdorferi in patients with unclassified arthritis.