Did intense adverse media publicity impact on prescribing of paroxetine and the notification of suspected adverse drug reactions? Analysis of routine databases, 2001–2004

Abstract

Aim

To document the impact on clinical practice in England of media attention around possible adverse effects of paroxetine.

Design

Analysis of national selective serotonin reuptake inhibitor (SSRI) prescribing trends and yellow-card adverse drug reaction reports, 2001–2004.

Results

From a steady state in 2001, paroxetine prescribing declined sharply from April 2002, coinciding with a USA regulatory action; the subsequent decline in paroxetine prescribing was 1.87% per month (95% confidence interval −2.06, −1.68). Other SSRI prescribing increased by 1% per month until a major UK review of SSRIs in children in December 2003, after which prescribing plateaued. Media publicity was associated with short-term peaks in yellow-card reports related to paroxetine.

Conclusion

Falls in paroxetine and other SSRI prescribing in the UK coincided, respectively, with regulatory communications from the USA and the UK, but associations may have noncausal or other explanations. Reports of adverse reactions to paroxetine appeared to increase after adverse media publicity about the drug.

Introduction

In the last 3 years there has been intense media and regulatory scrutiny of possible adverse effects of selective serotonin reuptake inhibitors (SSRIs). Publicity has particularly centred on paroxetine [1–3], until recently the second most widely prescribed SSRI in England. The US Food and Drug Administration (FDA) warned of severe withdrawal symptoms associated with paroxetine in January 2002 [1] and, from October 2002, suggested links between paroxetine and suicidal behaviour [2, 3] were the focus of three Panorama programmes (October 2002, May 2003 and October 2004) and five regulatory communications (June 2003, September 2003, December 2003, March 2004 and October 2004) with UK health professionals. We examined monthly trends in SSRI prescribing and voluntary notification of suspected adverse drug reactions between 2001 and 2004, focusing a priori on paroxetine, to assess the extent to which a period of sustained negative publicity influenced clinical practice.

Methods

Data sources

Information on monthly prescribing of SSRIs (available as numbers of single items written on prescription forms) by general practitioners in England between January 2001 and November 2004 was obtained from the Prescription Statistics section (SD1E) of the Department of Health. Only annual data were available before 2001, the data could not be identified by age of the patient and prescriptions written in hospitals/secondary care clinics were not included, even if dispensed in the community. Monthly numbers of reports of suspected adverse drug reactions (yellow-card reports) for SSRIs were obtained from the Medicines and Healthcare products Regulatory Agency (MHRA) for the same time period.

Statistical analysis

Analysis of prescribing trends was conducted by join point regression, in which trend data are described by a number of contiguous linear segments and ‘join points’ (points at which trends change). Join point regression was used to estimate monthly percentage change in number of prescription items and the number and location of join points [4]. Models were based on linear regression with the log number of prescription items as the dependent variable and month as the independent variable. To identify the best-fitting combination of line segments and join points, a series of permutation tests was performed, first testing the null hypothesis (H_o = no join points) vs. the alternative hypothesis (H_a = 3 join points). Hypothesis testing proceeded sequentially, increasing the number of join points under H_o by 1 if the null hypothesis was rejected and decreasing the number of join points under the alternative hypothesis if H_o was accepted. The maximum number of join points tested was three in each analysis. For each model, the locations of the best-fitting join points were identified using a grid search algorithm [5]. A Bonferonni correction was applied by conducting each test at the α/3 level, ensuring that the probability of a type I error (i.e. concluding that there are one or more join points when there are in fact none) was at most 0.05. Analyses were conducted using Joinpoint software (version 3, April 2005) made available by the National Cancer Institute (srab.cancer.gov/joinpoint).

For the analysis of yellow-card data we calculated the rate of reporting of suspected adverse drug reactions to paroxetine as the number of yellow-card reports per 100 000 paroxetine prescriptions for each month of the study.

Results

Prescribing of paroxetine was relatively steady between January 2001 and April 2002, but there was strong evidence that the number of paroxetine items prescribed per month declined after April 2002 [95% confidence interval (CI) December 2001 to June 2002] (P < 0.0001 for test of null hypothesis that difference between first and second slope = 0), so that the estimated monthly percentage change after this time was −1.87% (95% CI −2.06, −1.68; P < 0.0001 for test of null hypothesis that monthly percentage change = 0) (Figure 1A). There was no evidence that the rate of decline in prescribing of paroxetine changed again up to November 2004.

Figure 1.

Monthly number of prescription items (thousands), 2001–2004 (dotted line). The solid line is fitted using join point regression [allowing estimation of the monthly percentage change in number of prescription items and the number and location of join points (points at which trends change)]. Arrows show dates of Panorama programmes (solid arrows) or regulatory communications (dashed arrows). Y axes do not include zero. (A) Paroxetine. (B) All other SSRIs

In contrast, prescribing of all other SSRIs combined (mainly fluoxetine and citalopram) increased by 1.15% (95% CI 1.01, 1.29; P ≤ 0.0001) per month between January 2001 and December 2003. After December 2003 the rate of increase declined (0.33%; 95% CI −0.39; 1.05; P = 0.4) (P = 0.03, for difference between slope 1 and 2) (Figure 1B), although the 95% CI for a join point at December 2003 was wide (April 2002 to June 2004). Analysis of yellow-card data suggests that adverse publicity, particularly the three Panorama programmes, was associated with marked, short-term peaks in reporting (Figure 2). The mean rate of yellow-card reporting (per 100 000 prescriptions) in the month before each regulatory announcement was 7.6 vs. 8.0 in the month after. In contrast, the mean rate of yellow-card reporting in the month before each media publicity episode was 8.3 vs. 13.4 in the month after. An identical pattern of peaks and troughs in yellow-card reporting of suspected adverse drug reactions to paroxetine was observed when the temporal trend in number of reports was plotted (data not shown).

Figure 2.

Suspected adverse drug reactions to paroxetine (yellow-card reports) per 100 000 paroxetine prescriptions by year and month between 2001 and 2004. Arrows indicate dates of Panorama programmes (solid arrows) or regulatory communications (dashed arrows)

Discussion

The commencement of the secular decline in prescribing of paroxetine coincided with a USA FDA high-profile warning in January 2002 of severe withdrawal symptoms associated with the drug [1]. Regulatory communications from the USA are not directed at UK prescribers, so the mechanism linking the FDA action to prescribing in the UK is not immediately obvious. The USA action was communicated to UK prescribers in the medical media in February 2002 [1] at the same time as a declaration by the International Federation of Pharmaceutical Manufacturers Associations that GlaxoSmithKline (who manufacture paroxetine) was guilty of misleading the public about paroxetine in the USA [1]. The trigger for the fall in prescribing of paroxetine may have been the cumulative effect of these events combined with other influences occurring over a longer period of time, including clinical experience [6], published case reports [7], earlier UK regulatory communications [8] and other publications [9, 10]. Alternatively, the observed temporal associations between regulatory event and prescribing decline may have arisen by chance, or could be explained by other factors not studied in this report.

Adverse media and regulatory publicity occurring between October 2002 and October 2004 in England did not appear to increase the rate of reduction in paroxetine prescribing, but such exposure may have maintained the decline. In contrast, prescribing of the other SSRIs continued to increase at over 1% per month until December 2003, but with no evidence of any acceleration in prescribing after April 2002 to compensate for falls in paroxetine use. The decline in other SSRI prescribing after December 2003 coincided with the public communication of a major review of SSRIs in children undertaken by an Expert Working Group of the MHRA. The data are largely limited to prescribing in general practice and temporal patterns of hospital prescribing may differ. The data only capture dispensed prescriptions, but it seems unlikely that changes in primary noncompliance rates altered over time in such as way as to explain the results.

Media publicity was associated with sharp but only short-term peaks in voluntary reporting of suspected adverse drug reactions to paroxetine, highlighting a limitation of the yellow-card scheme as a robust method of adverse event monitoring [11].

Conflict of interests

D.G. was a member of the MHRA's Expert Working group on the safety of SSRIs. He acted as an independent advisor, receiving travel expenses and a small fee for meeting attendance and reading materials in preparation for the meeting.