Reasons for and time to discontinuation of rimonabant therapy: a modified prescription-event monitoring study.

BACKGROUND: Early treatment discontinuation will have a negative effect on a drug's benefit-risk profile if discontinuation occurs earlier in time than the positive effects of treatment. This non-persistence of therapy has been associated with an increased risk of adverse health outcomes.
OBJECTIVE: The aim of this study was to explore relationships between patient characteristics and reasons for and time to discontinuation of rimonabant therapy, focusing on psychiatric events, because these were the main safety concerns for rimonabant.
METHODS: A modified prescription-event monitoring (M-PEM) study was conducted for rimonabant. Descriptive statistics were used to describe the patient population. Rate ratios with 95% confidence intervals (CIs) were calculated to explore associations between patient characteristics and selected categories of reasons for stopping (RfS). Median times to discontinuation were compared using the Mann-Whitney U test.
RESULTS: The cohort comprised 10,011 users of rimonabant, three of which were excluded from this analysis because of missing age or sex. A total of 7204 patients (72.0%) stopped using rimonabant (median observation time 323 days, interquartile range: 279-371 days). In addition, patients with a history of psychiatric illness were more likely to discontinue rimonabant therapy early for all reasons, but most pronounced due to psychiatric events (rate ratio 1.79; 95% CI 1.54, 2.09) than those without a history of psychiatric illness. In contrast, the rates of discontinuation due to lack of effectiveness, any clinical events and psychiatric events in patients with cardiovascular disease, type 2 diabetes mellitus, dyslipidaemia or hypertension tended to be lower (not all being significant) than those without. For patients who discontinued treatment due to lack of effectiveness, the median time to discontinuation was significantly shorter for patients with a history of psychiatric illness, compared with patients without a history of psychiatric illness (86 vs 97 days, p = 0.03). For patients discontinuing treatment due to psychiatric events, the difference in median time to discontinuation was also 11 days (64 vs 75 days, p = 0.38), although not statistically significant. For patients stopping due to any clinical event, median time to discontinuation was comparable for patients with and without a history of psychiatric illness (61 vs 63 days, p = 0.90).
CONCLUSIONS: In this study, reasons for and time to discontinuation were associated with patient characteristics such as medical history. Patients discontinued treatment because of psychiatric events early after starting. In general, identification and characterization of early discontinuers, and increasing the understanding of reasons for stopping, may help healthcare professionals to develop targeted interventions to further improve treatment compliance, thereby optimizing treatment benefits and drug safety.