Meta-analysis and adverse drug reactions

CMAJ 1999;160:987

Although meta-analysis has become an accepted method for statistically pooling outcome measurements, the results of such analyses may not uniformly predict the clinical outcomes of randomized controlled trials² and must be interpreted with caution. Limitations depend upon the selection of articles combined, the outcome criteria chosen, the heterogeneity of the studies included, the statistical technique used to pool the data, duplicate publication and interpretation of results.^3,4

Furthermore, meta-analysis was developed primarily to examine treatment efficacy, not safety. Trindade and colleagues report crude occurrence rates of adverse events for 2 classes of antidepressant medications. They found that SSRIs cause significantly more serotonergic events and tricyclic antidepressants more anticholinergic events. Did this outcome warrant the use of meta-analysis? Although the presentation of overall event rates may be useful,⁵ meta-analysis comparing the adverse events of treatments with widely different side effect profiles is unnecessary, as there is no conflict to resolve.

In addition, meta-analysis results are restricted to published clinical trials which, although they are ideal for examining efficacy in controlled environments, are not appropriate for comprehensive investigations of adverse events. The clinical trials used in the analysis by Trindade and colleagues had small samples and thus had sufficient power only to detect common adverse events. Moreover, clinical trial designs examine the effect of treatments in "ideal" patients, typically men and women between the ages of 18 and 65 years, who are otherwise healthy and not taking other medications. The effects of the medication in different populations, such as elderly people, adolescents and people with comorbidity, are thus not evaluated. As a result, uncommon and potentially serious adverse events may go unnoticed. We published a large case-series analysis of hyponatremia and the syndrome of inappropriate antidiuretic hormone (SIADH) associated with SSRI use [abstract / résumé].⁶ We identified 736 spontaneous reports of SIADH to adverse-event reporting agencies and pharmaceutical manufacturers, of which 30 were published. Most of the cases occurred in elderly people. Yet despite the large number of case reports, this adverse event remains relatively unknown and unrecognized. A recent review found 35 additional published case reports since our original publication (12 of these published during the period between manuscript submission and publication). Only 4 case reports referenced the comprehensive case-series analysis. Given the serious nature of the event, we are troubled by the failure to recognize epidemiological or post-marketing surveillance study designs as valid sources for adverse event information.

Investigators who, like Trindade and colleagues, use meta-analysis in isolation may miss important, serious adverse events. Approaches that incorporate both clinical trial information and epidemiological and post-marketing surveillance research, including case reports, case series, and cohort and case­control studies, are necessary and appropriate to evaluate a complete, clinically relevant safety profile of therapeutic interventions.

Nicole Mittmann, PhD
HOPE Research Centre
Division of Clinical Pharmacology
Sunnybrook & Women's College
  Health Sciences Centre
University of Toronto
mittmann@srcl.sunnybrook.utoronto.ca

Barbara A. Liu, MD
Kunin-Lunenfeld Applied Research Unit
Baycrest Centre for Geriatric Care
Divisions of Geriatric Medicine
  and of Clinical Pharmacology
Sunnybrook & Women's College
  
Health Sciences Centre Sandra R. Knowles, BScPharm
Department of Pharmacy and
  Division of Clinical Pharmacology
Sunnybrook & Women's College Health Sciences Centre
University of Toronto
Neil H. Shear, MD
HOPE Research Centre
Division of Clinical Pharmacology
Sunnybrook & Women's College Health Sciences Centre
University of Toronto
Toronto, Ont.

Send a letter to the editor Envoyez une lettre à la rédaction

1. Trindade E, Menon D, Topfer L-A, Coloma C. Adverse effects associated with selective serotonin reuptake inhibitors and tricyclic antidepressants: a meta-analysis. CMAJ 1998;159(10):1245-52.
2. LeLorier J, Grégoire G, Benhaddad A, Lapierre J, Derderian F. Discrepancies between meta-analyses and subsequent large randomized, controlled trials. N Engl J Med 1997;337(8):536-42.
3. Cook DJ, Sackett DL, Spitzer WO. Methodologic guidelines for systematic reviews of randomized controlled trials in health care from the Potsdam consultation on meta-analysis. J Clin Epidemiol 1995;45(1):167-71.
4. Mittmann N, Knowles SR. Repetitive publication, meta-analysis, and adverse events. J Clin Psychopharmacol 1996:16:77.
5. Mittmann N, Herrmann N, Einarson TR, Busto UE, Lanctôt KL, Liu BA, et al. The efficacy, safety and tolerability of antidepressants in late life depression: a meta-analysis. J Affect Disord 1997;46:191-217.
6. Liu BA, Mittmann N, Knowles SR, Shear NH. Hyponatremia and the syndrome of inappropriate secretion of antidiuretic hormone associated with the use of selective serotonin reuptake inhibitors: a review of spontaneous reports. CMAJ 1996;155(5):519-27.