Intended for healthcare professionals


Lessons from the withdrawal of rofecoxib: France has policy for overall assessment of public health impact of new drugs

BMJ 2004; 329 doi: (Published 02 December 2004) Cite this as: BMJ 2004;329:1342
  1. Lucien Abenhaim (lucien.abebhaim{at}, professor of public health
  1. Faculty of Medicine, 24 rue du Faubourg St Jacques, Paris 75014, France

    EDITOR—In their editorial commenting on the withdrawal of rofecoxib, Dieppe et al advocate a series of measures before a drug can be licensed.1 Several experiences at the directorate general of the Health of France relating to this issue have led to the formulation of a new policy.

    Whenever a drug is likely to be used on a large scale, pharmaceutical companies must present a pre-reimbursement assessment and organise a postmarketing study of the public health impact of the drug. This impact assessment goes far beyond single end points, as is the case in trials and classic epidemiological (aetiological) research. It currently works within the framework of a formal agreement signed in May 2003 between the health product economic committee (CEPS) and the association of drug enterprises (LEEM).

    Large scale epidemiological evaluations are needed to measure the potential shifts in disease related morbidity and mortality in populations, and risk assessment and the use of other concerned drugs need to be evaluated. Plans should ensure the optimal public health impact of the drug (restricted target populations, selective prescribing, close follow up).

    The first example of this new policy was to ask in 2001 for an independent, large scale cohort study of 40 000 patients treated with rofecoxib, celecoxib, or traditional nonsteroidal anti-inflammatory drugs. More than 50 such studies have now been agreed for various drugs between pharmaceutical firms and the French authorities. Some of these agreements have included limiting the size of the population for which the drug should be reimbursed and to achieve a stepwise introduction of the drug—a goal similar to that advocated by Dieppe et al. However, the policy has been in operation for only a year and its efficiency will have to be evaluated.

    Requirements for public health impact assessment of drugs (2003)

    • Description of the treated population and comparison with the target population for the drug. Identification, within the population of treated patients, of the fraction in whom disease is poorly controlled with previous treatments and who should be treated with the new drug alone

    • Evaluation of the effect of co-prescriptions, comorbidities, and other risk factors

    • Epidemiological assessment of the impact of the new drug on the incidence and prevalence of complications of the treated disease in populations, including morbidity and mortality associated with treated disease and side effects and risks of the drugs

    • Measure of the impact of other new comparative drugs in the same population

    • Evaluation of the impact of the new drug on the health system (medical and hospital services, use of diagnostic procedures and other treatments, including other drugs)

    The box lists the elements required for public health impact assessment by the general health inspectorate.


    • LA is a former general director of Health of France (1999-2003) and a full professor of epidemiology and biostatistics, McGill University.

    • Competing interests LA has been active in or acted as a consultant for studies funded by Roche, United Therapeutics, Negma and Astra-Zeneca, but none for non-steroidal anti-inflammatory drugs or COX-2 inhibitors.


    1. 1.
    View Abstract