'Why reporting matters'
SSRIs are, arguably, one of the most written about classes of medicines, both for putting back quality into the lives of thousands of patients battling with serotonin-dependent depression and for their reported suicidal effects in adolescents. The latest write-up on the SSRIs is the Lancet article 'Selective serotonin reuptake inhibitors in pregnant women and neonatal withdrawal syndrome: a database analysis' by Emilio J Sanz, Carlos De-Las-Cuevas, Anne Kiuru, Andrew Bate and Ralph Edwards (Lancet 2005, 365: 482-87). Using the WHO global database for adverse drug reaction reports, the authors have investigated whether use of SSRIs in pregnant women might cause neonatal withdrawal syndrome. They conclude that 'SSRIs, especially paroxetine, should be avoided or cautiously managed in the treatment of pregnant women with a psychiatric disorder'.
One of the most important tasks of the WHO Programme for International Drug Monitoring is to identify 'signals' of drug safety problems as early as possible. Spontaneously reported cases of suspected adverse drug reactions (ADRs) are forwarded from national pharmacovigilance centres (appointed by national governments) in 75 countries (plus 13 associated countries) to the WHO Collaborating Centre for International Drug Monitoring, Uppsala Monitoring Centre (the Centre), in Sweden. The case reports, recorded in a common format, are processed and stored in the ADR database. Over 3.1 million case records are maintained by the Centre, which provides a unique source of international ADR information. Although the WHO ADR data are not homogeneous with respect to origin or the probability that the pharmaceutical product caused the adverse reaction, they have a proven use in the early detection of signals. A signal is defined by WHO as 'reported information on a possible causal relationship between an adverse event and a drug, the relationship being unknown or incompletely documented previously'.
Since 1998 the Centre has been using the Bayesian Confidence Propagation Neural Network (BCPNN) methodology to identify unexpectedly strong statistical associations between drugs and adverse reactions. The BCPNN uses a logarithmic measure of disproportionality called the information component (IC) for the calculation of 'observed to expected' reporting ratios. Using this methodology, the complete WHO database is screened quarterly generating a computerized table known as the combinations database containing more than 2000 new associations that stand out statistically from the background of all reports in the database. It is sent every quarter to the national pharmacovigilance centres (National Centres, NCs) who review its international contents for issues of relevance to their own countries. At the Centre, associations are checked against available product information literature. For drugs where the reaction is not found or not described well enough, case reports are retrieved from the WHO database. The Centre then sends the cases to the most appropriate expert in the Centre's global review panel to assess the evidence for the reaction being related to the suspected drug using their clinical experience and pharmacological knowledge.
The additional 'signal' detection work by the Centre and its review team is complementary to the work performed by national pharmacovigilance centres and not a substitute for local evaluation and decision-making.
An association between paroxetine and neonatal convulsions was identified already in December 2001 by the Centre by the above process and NCs were notified. In August 2003 the Australian Therapeutic Goods Administration (TGA) drew attention to this reported problem in its Adverse Drug Reactions Bulletin. In 2004 both the US FDA and Health Canada issued warnings about the potentials of in utero SSRI-exposure. The current 2005 Lancet paper analyses further data, accumulated up to 2003 in the WHO database, using 102 cases (reported from 11 countries) of neonatal effects associated with SSRI use and confirms these warnings. Again, as with the cyclooxygenase-2 inhibitor story (see WHO Pharmaceuticals Newsletter No.5, 2004), the WHO database helped identify, well ahead of time, the drug and the adverse effects requiring special attention. However, this is not just about reminding Member States of what the International Drug Monitoring Programme can do. Rather, it is an opportunity to reinforce that a facility exists and that it should be put to optimal use with more active collaboration from the Member States.
That the WHO database is the largest source of global information on adverse reaction reports cannot be contested. Neither can we challenge the fact that spontaneous reports form a valuable component of the 'early warning system'. The Centre encourages regular reporting, but less than ten countries send reports once per month; some countries do not send any reports at all; some countries send reports in large batches and nothing in between for long periods. This attitude needs to change in order to improve the database both qualitatively and quantitatively, to make it truly current, representative and global, and to make the early warning system work in real terms. It is all about preventing avoidable harm and regular reporting will contribute.
1. EJ Sanz et al., Lancet, 365:482-487, 2005.
2. Australian Adverse Drug Reactions Bulletin, Vol. 22, no.4, August 2003.
3. Reactions Weekly (Adis International), No.7, 26 June 2004.