Intended for healthcare professionals


Off-label prescribing of antidepressants

BMJ 2017; 356 doi: (Published 21 February 2017) Cite this as: BMJ 2017;356:j849
  1. Daniel R Morales, general practitioner and discovery fellow,
  2. Bruce Guthrie, professor of primary care medicine
  1. Population Health Sciences Division, University of Dundee Medical School, Dundee DD2 4BF, UK
  1. Correspondence to: B Guthrie b.guthrie{at}

Strength of evidence matters more than presence or absence of a specific licence

In most countries, medicines have a product licence that describes how they should be used. Licensing is intended to ensure that medicines meet acceptable standards of efficacy, safety, and quality for a particular indication in a particular group of patients.1 “Off-label” use occurs when a drug is prescribed for an unlicensed indication, to an unlicensed patient group (such as children), and/or as an unlicensed dosage or formulation. In their paper (doi:10.1136/bmj.j603), Wong and colleagues found that almost a third of antidepressants were prescribed for off-label indications, most commonly pain, insomnia, and migraine.2

Clinicians can legally prescribe off-label, and professional licensing agencies recognise that off-label use is necessary if licensed medicines are ineffective, if they are associated with adverse effects, or if the licensed dose or formulation does not meet the patient’s needs. Professional responsibility in these circumstances is fundamentally the same as for on-label prescribing. As the UK medical regulator says, “We expect you to carefully consider any treatment that you prescribe, and we expect you to be able to justify your decisions and actions when prescribing, administering and managing medicines regardless of whether they are licensed or unlicensed.”3 Although off-label prescribing may need more explicit justification, the evidence supporting prescribing is actually more important than the presence or absence of a licence in itself.

Only 16% of the off-label prescribing identified by Wong and colleagues was directly supported by strong evidence, with a further 40% having indirect support from strong evidence for other drugs in the same class.2 Although it may seem odd that off-label prescribing can have strong evidence, this often occurs when new indications for old drugs are evaluated in trials but pharmaceutical companies have not judged alteration of existing marketing authorisations worthwhile because the drug is off-patent and the required regulatory process is complex and expensive.

Amitriptyline for chronic pain is an example of evidence based and guideline recommended off-label prescribing,45 and it accounted for 14% of off-label antidepressant prescribing in Wong and colleagues’ study.2 Extending the range of uses of long established medicines in this way is attractive for healthcare professionals and patients, because these drugs are perceived to have familiar safety profiles and are cheaper than drugs still under patent protection. Similarly, clinical guidelines may recommend antidepressant drug classes for extended indications such as anxiety rather than recommending only specific licensed medicines,6 even though this requires a fairly strong assumption that all drugs in the same class are equally effective or equally safe.

These mismatches between licensing, evidence, and guidelines within countries complicate clinical and shared decision making, and variations in licensing and guidelines between countries add further confusion. Variation between countries even occurs in which adverse events are considered significant for licensed drugs.7 For all prescribing, patients (or their parents or carers) should be given enough information to allow them to make an informed decision whether to take a medicine. This should include whether the intended use is off-label, but more importantly prescribers should discuss the strength of the evidence base underlying their recommendation.3 How often this occurs in practice remains uncertain. Electronic prescribing systems offer opportunities to provide more point-of-care information to prescribers and patients, including whether an intended use is off-label or lacks evidence relevant to the patient in front of them. However, such systems need careful design and evaluation to avoid unintended consequences of implementation in time constrained clinical workflows.

Off-label prescribing matters because it is usually (but not always) associated with substantial uncertainty about the balance of benefit and harm.8 Prescribers should therefore be cautious when they prescribe an off-label medicine on the basis of an extrapolation of evidence for a different indication, in a different patient group, or for a substantially different dose or formulation.

Equally, however, on-label prescribing also often involves extrapolation, most commonly because the patient needing treatment is very different from the patients included in trials. For example, the evidence underpinning on-label use of antidepressants to treat depression comes from trials in people with more severe depression and less psychiatric and physical comorbidity than is typical in everyday practice.910 As a result, most people with well characterised major depressive disorder in everyday practice would be ineligible for the trials on which licensing and treatment recommendations are based.11 Evidence also shows that patients who would be ineligible for trials are less likely to respond to antidepressants and more likely to experience adverse events.12

As Wong and colleagues show, off-label prescribing is common and is often poorly supported by evidence or relies heavily on extrapolating evidence from one situation to another. These pitfalls are not confined to off-label drugs, however. Patients and prescribers should be cautious about all extrapolations of evidence whether the proposed treatment is “on-label” or “off-label.”


  • Research, doi: 10.1136/bmj.j603
  • Competing interests: We have read and understood the BMJ policy on declaration of interests and declare the following interests: none.

  • Provenance and peer review: Commissioned, not peer reviewed.


View Abstract

Log in

Log in through your institution


* For online subscription