Limitations of Formal Post-Marketing Surveillance Studies

LIMITATIONS OF FORMAL POST-MARKETING SURVEILLANCE STUDIES

A general practice based Prescription Event Monitoring (PEM) study profiled the safety of terodiline in 12 457 patients, treated between November 1986 and September 1987 (Inman et al., 1993). Of these patients, 72.5% were females. The mean age was 65.6 (range 5–98) years in males and 63.3 (range 5–102) years in females. Incontinence (47.8%), frequency (16.9%), bladder irritability (7.7%) and urgency (6.6%) accounted for the majority of the indications for use of terodiline in females. In clini-cal practice, 62.2% of the patients were receiving a maximum daily dose of 25 mg, 18.2% were receiv-ing 50 mg and a minority had used other regimes, including some up to 100 mg per day. Terodiline was reported to have been effective in 56% of the patients. Cardiovascular events reported during the first 6 months and at any time during and after treatment with terodiline, but not considered to be adverse reactions to it, included dizziness (n = 135 and 255, respec-tively), syncope (41 and 105), hypotension (15 and 30), atrial fibrillation (8 and 30), tachycardia (8 and 17), bradycardia (2 and 10), arrhythmias (2 and 8), ventricular fibrillation (0 and 3), heart block (0 and 2) and cardiac arrest (0 and 2). Even in a subsequent survey (initiated in 1990) of co-prescribing of vari-ous cardioactive medications, it could not be estab-lished whether the excess of syncope, arrhythmias, bradycardia, hypotension and other cardiovascular events was due to drug combinations or the pres-ence of co-existing cardiovascular disease. Of all the events reported in the cohort, only 51 events were suspected to be actual adverse reactions to terodi-line and these included 2 cases of dizziness. No case of cardiovascular collapse attributable to torsade de pointes could be found.

Even a retrospective study, undertaken in the after-math of the powerful signal from the spontaneous reporting system and the withdrawal of terodiline from the market, failed to better quantify the risk of cardiotoxicity of terodiline. In this study using the VAMP database (Hall et al., 1993), a prelimi-nary open study identified a total of 9176 terodiline-treated patients. A total of 77 (0.8%) of these 9176 patients had an ECG investigation during the study period. There was only one confirmed case of torsade de pointes in a 41-year-old female who had hypokalaemia at the time of the event. Apart from a 50 mg daily dose of terodiline, she was concur-rently receiving a tricyclic antidepressant. Altogether, a total of 59 patients were found to have had a cardiac arrhythmia during the follow-up period. This open study estimated the risk of terodiline-induced torsade de pointes to be 1.1 per 10 000 patients. A retrospec-tive but limited inquiry into the nature of arrhythmias in the 59 patients with cardiac arrhythmias elicited information in only 19 patients. These included 6 bradycardia, 4 heart blocks, 3 ventricular tachycar-dias, 2 ventricular conduction defects, 2 extrasystoles, 1 ‘tachy-brady syndrome’ and 1 cardiac arrest. None had previously been reported to the CSM through the yellow cards and 16 of the 19 practitioners concerned agreed to complete a yellow card.

In another retrospective cohort extension of the above VAMP study, 5705 terodiline-treated patients were compared with 9604 controls. It concluded that there was no significant difference in the risk of devel-oping an arrhythmia in the terodiline-treated patients compared with that in the controls. The relative risk compared with controls was estimated at 1.1 (95% CI: 0.64–1.90). Even the patients reporting symp-toms suggestive of cardiac arrhythmias (syncope, collapse, blackouts) were not overly represented in the terodiline-treated cohort. Only dizziness and falls were reported significantly more frequently in the terodiline-treated patients (5.13% vs. 3.35%).

Both these studies had failed spectacularly if it was intended that they would test or strengthen what is frequently, and deprecatingly, termed merely a ‘hypothesis’ when reports of serious reactions are gathered through a spontaneous reporting system.

The failure of formal post-marketing surveillance studies to detect or quantify the risk of drug-induced QT interval prolongation, with or without torsade de pointes, associated with some potent torsadogens is not unfamiliar (Pratt et al., 1994; Hanrahan et al., 1995; Staffa et al., 1995; de Abajo and Rodriguez, 1999; Layton, Key and Shakir, 2003).