Can adverse drug reactions be detected earlier? A comparison of reports by patients and professionals

BMJ 1996; 313 doi: (Published 31 August 1996) Cite this as: BMJ 1996;313:530
  1. Toine C G Egberts, research fellowa,
  2. Maartje Smulders, studentb,
  3. Fred H P de Koning, pharmacistb,
  4. Ronald H B Meyboom, physicianb,
  5. Hubert G M Leufkens, associate professor of pharmacoepidemiologya
  1. a Department of Pharmacoepidemiology and Pharmacotherapy, Utrecht Institute for Pharmaceutical Sciences, Utrecht University, PO Box 80.082, 3508 TB Utrecht, Netherlands,
  2. b Netherlands Pharmacovigilance Foundation LAREB, 5026 RH Tilburg, Netherlands
  1. Correspondence to: Dr Egberts.
  • Accepted 8 March 1996

The occurrence of previously unknown adverse reactions after the marketing of a new drug is inevitable given the limitations of preregistration clinical trials. Nevertheless, their impact on public health should be minimised by ensuring that reactions are detected as early as possible. The reporting of suspected adverse drug reactions by health care professionals to monitoring agencies1 and to medical journals2 has been important in alerting doctors to drug safety problems, but Mitchell et al suggested that the time lag to the first reports of adverse reactions might be shortened if patients themselves reported adverse events.3 For a newly introduced antidepressant we compared the time to reporting of adverse drug reactions by patients and by health care professionals.

Methods and results

A telephone medicines information service was started in 1990 by the Dutch Ministry of Health and the Royal Dutch Association for the Advancement of Pharmacy to promote the correct use of drugs and identify problems related to drug use. Anonymously and free of charge, patients can consult a pharmacist. The pharmacist summarises each call on a standard form. All reports produced by the service in 1992-4 were searched for those indicating an adverse reaction associated with the antidepressant paroxetine. This drug was introduced just before the study (September 1991) and has been prescribed in considerable volume. The time lag between marketing of paroxetine and the date of the phone calls was calculated and compared with the time lag between marketing of paroxetine and the date of suspected reactions reported by health professionals to the Netherlands Pharmacovigilance Foundation. This is a typical spontaneous adverse drug reaction reporting system, which in 1994 covered a population of about two million people in seven regions of the Netherlands. We analysed separately the time lag in both systems for those drug reactions that were not mentioned in the patient information leaflet at the time of the study.

Out of 23 625 calls to the telephone service 120 suggested an adverse reaction to paroxetine. Of the 7665 suspected adverse reactions reported to the Netherlands Pharmacovigilance Foundation 89 were related to paroxetine. The mean time lag for all suspected reactions to paroxetine was 229 days less for the telephone service than for the Pharmacovigilance Foundation (95% confidence interval 160 to 298; Student's t test; fig 1). The proportion of new suspected reactions (those not mentioned in the patient information leaflet) was similar for both systems (21% v 18%). The nine new reactions identified by both systems were: rigours (first reported by a patient at 390 days, first reported by a health professional at 578 days), bleeding disorders (396, 1025), hypertension (432, 1192), mydriasis (597, 660), taste disturbance (635, 827), apathy (943, 1124), flushing (967, 1100), menstrual disorders (971, 1162), and rash (1002, 1125). The mean difference in time lag for these reactions was 273 (89 to 458) days.

Fig 1
Fig 1

Cumulative number of reports of suspected adverse drug reactions associated with use of paroxetine received from patients and from health care professionals since marketing (t = 0).


Our findings show that reporting by patients may contribute to earlier detection of known and unknown adverse drug reactions. The information available from patients in the telephone service was, however, often crude and incomplete in comparison with professional adverse drug reaction reports. Therefore the telephone service cannot be relied on as an independent reporting system; it might generate too many false alarms. A combination of information from patients and health care professionals might, however, enable earlier detection. Additional information is relevant because alerts are often based on only a few cases,4 but too high demands on quality of the reports may impair their capacity for early warning.5 We suggest that patient reporting may play an additional role in pharmacovigilance.

We thank Mrs L van der Schoot, Mrs A van Berkel, and Mr H ter Steege for their help and the Royal Dutch Association for the Advancement of Pharmacy for the permission to analyse the forms of the telephone medicines information service.


  • Funding Netherlands Pharmacovigilance Foundation LAREB and Department of Pharmacoepidemiology and Pharmacotherapy, Utrecht University.

  • Conflict of interest None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
View Abstract