Intended for healthcare professionals


Red flags for back pain

BMJ 2013; 347 doi: (Published 12 December 2013) Cite this as: BMJ 2013;347:f7432
  1. Martin Underwood, director, Warwick Clinical Trials Unit1,
  2. Rachelle Buchbinder, professor, Department of Epidemiology and Preventive Medicine, School of Public Health and Preventive Medicine, Monash University 2
  1. 1Warwick Medical School, University of Warwick, Coventry CV4 7AL, UK
  2. 2Monash Department of Clinical Epidemiology, Cabrini Hospital, Cabrini Medical Centre, Malvern, VIC 3144, Australia
  1. m.underwood{at}

A popular idea that didn’t work and should be removed from guidelines

Nearly all guidelines on low back pain have stressed the importance of red flag signs—positive answers to key questions about medical history that indicate the possibility of a serious underlying condition—since the publication of influential reports in the United Kingdom and United States in 1994.1 2 In a linked systematic review (doi:10.1136/bmj.f7095), Downie and colleagues challenge this orthodoxy by investigating the accuracy of red flags to screen for fracture or cancer in people presenting with low back pain.3

Although intended to prompt clinicians to think about the possibility of serious conditions such as cancer, infection, or fracture, it is worrying that doctors are advised to investigate or refer patients on the basis of the presence or absence of red flags alone. For example, a BMJ 10 Minute Consultation on chronic back pain advised that “If red flag signs are present refer [patient] to a specialist for further evaluation.”4 NHS clinical knowledge summaries advise “Further investigation and referral with appropriate urgency may be indicated, and should always be considered, if one or more red flags are present.”5 Australian guidance states that “if [red flag] features are present, further investigation or referral is warranted.”6 Other guidelines suggest that demonstrating that red flags have been considered should be an audit standard when managing back pain.7 Most people with back pain and red flags will not have serious conditions, so taking this advice literally can cause harm. These, harms include unnecessary exposure to radiation and labelling effects from unwarranted imaging, as well as unnecessary treatments, including surgery.6

Starting from a previous review of eight back pain guidelines in which no two guidelines endorsed the same set of red flags for cancer (24 different red flags) or fracture (29 different red flags), Downie and colleagues have synthesised the evidence for different red flags.8 Few studies provided precise definitions of the red flags and only a minority of the red flags had been evaluated in more than one study. Most studies focused on individual red flags. Only six considered the accuracy of combinations of factors that are likely to be of greater clinical utility.

For fracture, only age, trauma, glucocorticoid use, and presence of contusion were confirmed as risk factors. However the increased risk of fracture was too low to suggest that these were clinically useful—a 15% increase in risk of fracture for older age and trauma. Three studies reported on the accuracy of combinations of factors, but none have been validated in independent samples. These combinations can have impressive predictive values. For example, in one study a positive response to three or more of age over 70 years, clinically significant trauma, being female, and prolonged use of glucocorticoids produced a post-test probability of 90% (95% confidence interval 34% to 99%).3 This combination does not, in fact, perform substantially better that clinical judgment.9

The data on red flags for the identification of spinal cancer are even weaker; the only informative red flag is a history of cancer. Even here, the data were from one primary care study from 1988 and one emergency department study from 1998.10 11 In 2009, Henschke and colleagues found no cases of cancer in a primary care inception cohort of 1172 people presenting to primary care with acute low back pain.12 A generic red flag of “history of cancer” is a blunt instrument. It does not consider the type of cancer or the time since diagnosis. For example, the risk of spinal metastases in a man with castration resistant prostate cancer and a woman who had curative treatment for carcinoma of the body of the uterus 20 years earlier would be very different. Another recent review found only a weak evidence base for the early diagnosis of metastasis or spinal cord compression as a consequence of metastatic disease.13

Over recent decades we have made substantial progress in developing treatment guidelines for non-specific low back pain that are grounded in evidence. However, there is inadequate evidence to support the formulaic use of red flags as a screening tool to inform clinicians when further investigation for a specific cause of low back pain is indicated. Such mnemonics may be useful prompts to guide clinical reasoning and the hypothetico-deductive approach to diagnosis. Nevertheless, guidance producers, and those disseminating or implementing guidance, need to remove such formulaic guidance forthwith. It should be reinstated only when (and if) a robust evidence base emerges that supports its use.

A positive diagnosis that leads to appropriate treatment is important for people with a specific cause of low back pain. Clinicians need to be familiar with the clinical picture seen in people with back pain as a result of cancer, fracture, infection, or ankylosing spondylitis and actively consider these diagnoses. In many cases the diagnosis will not be made on the first consultation. In most cases the patient will come to little long term harm if the diagnosis is not made at this time. Cauda equina syndrome is an exception. Although this is not a cause of back pain, a high degree of vigilance is needed to ensure the symptoms of cauda equina compression are not missed in those presenting with back pain, with or without radicular pain. The need for further investigation may become clear only as the clinical picture evolves over time—hence the need to keep the diagnosis under review.14


Cite this as: BMJ 2013;347:f7432


  • Research, doi:10.1136/bmj.f7095
  • Competing interests: We have read and understood the BMJ Group policy on declaration of interests and declare the following interests: None.

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


View Abstract