Rationale for Proportional Methods

RATIONALE FOR PROPORTIONAL METHODS

A basic question to be asked is whether the use of the proportion of reports for a particular reaction (compared with all reports) is sensible. As a first step, it is reasonable to examine the trends in proportions of reports over time within a major database. Figure 19.1 below gives the cumulative total number of reports and the cumulative number of reactions reported as suspected ADRs in the UK MCA database (Waller, Coulsen and Wood, 1996). This database, called Adverse Drug Reactions Online Information and Tracking (ADROIT), has suspected ADRs reported on Yellow Cards since 1964. Over this period, the number of reports in the database has risen dramatically but the pattern in proportions in different SOCs has remained relatively stable.

Figure 19.1 shows the cumulative number of reports and reactions by year. Figure 19.2 shows the cumulative proportion (on a log scale) of reactions reported for a cross section of the SOCs. These show that cardiovascular reports are stable at a high level, while ‘ear’ reports are fairly stable at a low level. Eye reactions show a notable rise in 1974 – a result of practolol. Skin reactions are stable at a high level with a peak in 1982 (benoxaprofen).