Intended for healthcare professionals


Including elderly people in clinical trials

BMJ 1997; 315 doi: (Published 25 October 1997) Cite this as: BMJ 1997;315:1033

Better information could improve the effectiveness and safety of drug use

  1. Jerry Avorn, Associate professor of medicinea
  1. a Harvard Medical School, Boston, MA 02115, USA

    Practitioners face a difficult paradox in prescribing for the elderly. Those aged over 65 comprise only about 14% of the population in most industrialised countries, yet they consume nearly a third of all drugs. Ample evidence indicates that, even in healthy elderly people, aging impairs the way the body handles drugs. In ill elderly people these changes can be exaggerated considerably.1

    In an ideal world data from premarketing and postmarketing surveillance studies would describe how a given drug is likely to affect older patients differently from younger ones.2 Unfortunately, rather than being oversampled in clinical trials, to reflect their distribution in the drug consuming population, elderly people are inadequately represented.3 4 Moreover, when “old” patients are enrolled in trials, they are likely to be in their mid-60s and in quite good health. Nevertheless, we do not know to what extent “the elderly” in such studies are represented primarily by robust 66 year olds, since the oldest stratum is often simply described as “≥65,” a classification which is useless to any geriatrician, primary care practitioner, or person who has watched a grandparent age over time.

    Why is there such a mismatch between the routine conduct of premarketing trials and the information needs of prescribers and patients? The answers lie in economics, statistics, and politics. The “old old” are a messy lot physiologically. They are far likelier than the young to have coexisting medical problems, for which they are likely to be taking other potentially interacting drugs. They also have the distressing property of being more likely in the middle of a trial to suffer an infarct of heart or brain or simply to drop dead. They are bad news for the drug development process.

    All adverse events and deaths occurring in a premarketing trial must be reported and scrutinised. One of the costliest components of drug development is the cost of the capital tied up before a product is approved. And if the voices of economics needed any reinforcement, our statistical colleagues would be quick to say that the increased variance introduced by a heterogenous population of older subjects will reduce the precision of study estimates, requiring larger samples or increased duration to achieve the same study power. Not surprisingly, therefore, trial designers are often reluctant to enrol many truly elderly patients. In the absence of a compelling countervailing force to include more older patients, the regulatory requirements remain disquietingly loose on the topic of age.5

    After approval of a drug elderly people are penalised again. Postmarketing surveillance studies could provide a second opportunity for systematically studying the effects of a drug in populations that include many older patients.6 However, here too most nations fail to encourage a coherent, robust response. Insights about important, even life threatening, side effects are too often left to arise from ad hoc observations by alert practitioners, rather than through any proactive method of public sector surveillance.7

    Fortunately, these problems are soluble through a few straightforward measures. Firstly, any drug likely to be used by elderly people should be required to undergo premarketing testing in patients with an age distribution comparable to that expected when the drug is in routine use. Age stratification terms such as “≥65” should be replaced by depiction of age by decade, at least to 85. Secondly, premarketing evaluation should include assessing whether important age related differences exist in efficacy and toxicity, with such differences reported for all newly marketed drugs. Thirdly, because unexpected differences may emerge in effectiveness or side effects when a drug is used routinely by large numbers of elderly patients, particularly those too frail to be included in trials, plans for postmarketing surveillance should be required at the time that a drug is approved. The increasing availability of computerised datasets of drug use and clinical events in large populations should make this requirement practical and affordable.

    Each year, the lives of elderly patients are improved by the development of new drugs, as well as the intelligent use of existing agents. However, each year a significant (albeit smaller) number of them experience serious adverse events which could have been anticipated and prevented if better information were available to the practitioner.8 Prudent science policy and regulatory approaches should make it possible for the former number to continue to grow while the latter number becomes ever smaller.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    View Abstract