How long should treatments be continued?

BMJ 2010; 341 doi: https://doi.org/10.1136/bmj.c4102 (Published 19 August 2010) Cite this as: BMJ 2010;341:c4102
  1. Jonathan C Craig, senior staff specialist, paediatric nephrology1,
  2. Angela C Webster, staff specialist, nephrology1,
  3. Clement Loy, staff specialist, neurology1
  1. 1Sydney School of Public Health, University of Sydney, Sydney NSW, Australia
  1. [email protected]

    Data from randomised trials and cohort studies are needed to answer this question

    To make evidence based decisions, clinicians and patients need to know the answers to three questions about an intervention. Does it do more good than harm? How intense should treatment be? For how long should it be given? In the linked study (doi:10.1136/bmj.c4024), Chen and colleagues tackle the third question by assessing relapse rates after a first episode of psychosis in patients who either continued or discontinued antipsychotic drugs after at least one year of treatment.1

    Because of the explosion of randomised controlled trials and systematic reviews of trials in the past 50 years,2 we have high quality evidence on the effects of many interventions. These days, almost all issues of major general and specialist journals will include at least one randomised controlled trial that will provide reasonable certainty about the effects of a treatment. Implicit in the design will be the dose or intensity of the intervention evaluated, which is informed by previous phase I and II studies and may be the subject of specific phase III trials.

    In comparison, we know little about how long interventions should be given, both for common acute conditions like respiratory or urinary tract infections,3 4 and for chronic conditions like mental illness, cardiovascular illness, and chronic kidney disease, where patients are exposed to potentially harmful interventions for months, years, or even decades. Trials are rarely designed to evaluate how long people should be treated. Does this matter? Yes. Interventions given for too short a time can result in preventable adverse outcomes, such as relapse or recurrence. If interventions are given for too long then patients are at unnecessary risk of adverse effects and money is wasted.

    Chen and colleagues’ randomised trial was unusual in that the research question was treatment duration.1 They randomised patients who had recovered from their first psychosis to maintenance quetiapine or placebo, and they found that the risk of recurrence was 41% (95% confidence interval 29% to 53%) in the quetiapine group compared with 79% (68% to 90%) in the placebo group. The risk of discontinuation because of adverse events was about 10% higher in the quetiapine group (18% v 8%; relative risk 2.29, 0.99 to 5.28) over the next year, however. The trial provides reliable evidence on the benefits and harms of maintenance quetiapine and shows that psychotic relapse can be prevented by prolonged treatment with this drug.

    How do we make informed decisions about the duration of interventions in the absence of such trials? The table details a suggested framework for answering questions about the duration of interventions.

    A framework for making informed choices about treatment duration

    View this table:

    Ideally, the effects of a shorter or longer duration can be compared directly, as in Chen and colleagues’ trial.1 Alternatively, the effects of different durations can be inferred from trials that evaluate a single duration of an intervention but display the control adjusted treatment course over time, usually as a survival plot. In Chen and colleagues’ trial the outcomes are similar for the first 60 days of treatment and then diverge over the remaining period of the trial, which suggests many months of treatment are needed. In contrast, in a randomised trial of antibiotics to prevent urinary tract infection in children, most benefit occurs within the first six months of the 12 month randomised exposure period.5 Although trials have not been specifically designed to determine how long women with early breast cancer should be treated with tamoxifen, data provided by the Early Breast Cancer Trialists’ Collaborative Group clearly show improved survival with longer duration of exposure.6 The third, but potentially biased information comes from indirect comparisons across different trials that have evaluated the same intervention but with different treatment durations. For example, a systematic review found that the most cost effective duration for trastuzumab as adjuvant treatment for early breast cancer was uncertain.7

    The course of the chronic disease can also help to guide the length of treatment. But experiencing a symptom for the first time does not necessarily imply that the disease will have a chronic course. After a first seizure, the two year seizure recurrence rate is 40-50%, so giving everyone an antiepileptic drug after a first seizure is unnecessary. This is illustrated by the Medical Research Council Multicentre Trial for Early Epilepsy and Single Seizures,8 where patients were randomised to immediate or deferred antiepileptic treatment. Even though only 40% of patients in the deferred treatment group received an antiepileptic, seizure-free rate and quality of life at five years were similar in both groups.

    Once it is known how chronic the disease is, the next question is whether it has a progressive or relapsing-remitting course. For patients with non-progressive infrequent relapses and full recovery in remission, evidence from trials of acute treatment can be directly applied. An example of such a situation is the use of triptans in acute migraine attacks.9 However, most chronic diseases have a progressive course, with progressive deterioration. In a disease with progressive deterioration, treatment efficacy may decline as the disease worsens. In contrast, in some circumstances, such as childhood asthma, the underlying disease process may progressively improve over time, in which case withdrawal of inhaled corticosteroids may be safe.10

    In some cases, treatment decisions can be individualised if the course of a condition can be predicted using periodic measurement of biomarkers of disease or monitoring.11 Monitoring in chronic disease aims to guide treatment and control the disease or to predict relapse and re-establish control afterwards by measuring change in the disease marker. Disease markers are generally surrogate measures correlated with outcomes that are relevant to patients, usually through associations measured in cohort studies, but ideally within randomised trials, where a predictable and quantitative change in the biomarker correlates with patient outcome. However, most disease markers are imperfect, and true change can be hard to separate from expected variability and measurement error.12

    More trials of the duration of treatment are needed. Until then clinicians, consumers, and policy makers may use existing trials and observational studies within the framework described to guide treatment decisions.


    Cite this as: BMJ 2010;341:c4102


    • Research, doi:10.1136/bmj.c4024
    • Competing interests: All authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

    • Provenance and peer review: Commissioned; not externally peer reviewed.


    View Abstract

    Sign in

    Log in through your institution

    Free trial

    Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
    Sign up for a free trial