Toward a definition and method of assessment of treatment failure and treatment effectiveness: The case of leflunomide versus methotrexate

Objective. Time to treatment discontinuation (TTD) is an accepted method of assessing treatment effectiveness in the community, but is susceptible to channeling bias and secular and cohort effects. In addition, TTD does not consider the addition of new disease modifying antirheumatic drugs (DMARD) to insufficiently effective therapies. We expand the definition of treatment failure to include discontinuation or addition of a second DMARD (1) to examine leflunomide (LEF) versus methotrexate (MTX) effectiveness in clinical practice; (2) to obtain an estimate of overall clinical effectiveness; and (3) to identify factors associated with treatment successes and failure. In addition, (4) we test the feasibility of performing a clinical trial using a longitudinal data bank. Methods. Using the National Data Bank for Rheumatic Diseases longitudinal data bank, 1431 patients with rheumatoid arthritis (RA) who began taking LEF or MTX as part of their routine medical care were followed from 1998 through 2001. None of the 1431 patients had received either treatment previously. Patients were assessed at 6 month intervals for periods up to 36 months by mailed questionnaires concerning DMARD therapy and demographic and RA severity factors. Kaplan-Meier survivor functions and Cox regression analyses were used to assess treatment failure, defined as time to discontinuation or to the addition of a second DMARD. Results. For 756 patients taking LEF, the failure rate was 55.5 per 100 patient-years, and the median time to failure was 15 (95% CI 13, 17) months. For 675 patients taking MTX the failure rate was 57.3 per 100 patient-years, and the median failure time was 14 (95% CI 12, 18) months. These differences were not statistically significant. The overall rate of discontinuation was 68.7% of the failure rate. Discontinuation was predicted by adverse effects [hazard ratio 1.76 (95% CI 1.51, 2.04)] and by clinical status prior to starting DMARD, and these results were not affected by specific DMARD treatment. Discontinuation was more common with LEF, and addition of a second DMARD was more common with MTX. More than 77% of treatment failures, defined by use of additional therapy, resulted in starting anti-tumor necrosis factor treatment rather than a conventional DMARD. Conclusion. In an observational clinical trial using a contemporary longitudinal data bank, with time to treatment failure as the outcome, LEF and MTX had equal effectiveness as measured by time to treatment failure. Treatment failure rates were substantially greater than noted historically. Given the availability of many efficacious additional treatment options, this increase in failure rate appears to reflect a greater propensity to discontinue and/or add therapy.