Professor Underwood has received honoraria (less than $10,000) from the National Institute of Health and Clinical Excellence (NICE) for service as Chair of the group that developed NICE guidelines on early management of persistent nonspecific low back pain.
Diagnosing acute nonspecific low back pain: Time to lower the red flags?
Version of Record online: 29 SEP 2009
Copyright © 2009 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 60, Issue 10, pages 2855–2857, October 2009
How to Cite
Underwood, M. (2009), Diagnosing acute nonspecific low back pain: Time to lower the red flags?. Arthritis & Rheumatism, 60: 2855–2857. doi: 10.1002/art.24858
- Issue online: 29 SEP 2009
- Version of Record online: 29 SEP 2009
- Manuscript Accepted: 29 JUN 2009
- Manuscript Received: 6 MAY 2009
Episodes of acute low back pain are a universal human experience (1). Usually, this is a benign, self-limiting disorder that does not require professional advice or specific treatment (2). Once someone with acute low back pain presents for care, a plethora of management guidelines are available to treating clinicians. These guidelines typically recommend that serious disorders (cauda equina syndrome, fracture, infection, inflammatory disorders, malignancy) should be excluded, using a number of “red flag” questions, before making a diagnosis of nonspecific low back pain (3). In contrast to the high-quality data available for at least some treatments of nonspecific low back pain, systematic reviews of the use of red flag questions to identify malignancy and fractures show that the strength of the evidence underpinning these screening questions is weak (4, 5). Furthermore, few of the data that are available were collected in a primary care setting, where most consultations for back pain take place.
In this issue of Arthritis & Rheumatism, Henscke and colleagues report on the performance of commonly recommended red flags in a prospective study of 1,172 consecutive patients presenting to a primary care setting because of acute low back pain (duration of >24 hours but <6 weeks) (6). After careful followup over 1 year, a serious cause for back pain was identified in only 11 patients (0.9%; 95% confidence interval [95% CI] 0.5–1.7%). The incidence of spinal fractures (8 [0.7%] of 1,172 patients [95% CI 0.4–1.3%]) and malignancy (none of 1,172 patients [95% CI 0–0.3%]) was much lower than the commonly quoted values for osteoporotic fractures and malignancy (∼4% and ∼0.7%, respectively) (7).
An important strength of this study is that data were collected from members of different health professions who provide first-contact care for acute back pain in a primary care setting. This approach also results in an important weakness of the study: only first presentations for a new episode of low back pain were considered. Serious disorders causing low back pain are likely to be more common in some other patient groups. Persons in such groups include those presenting for a second, third, or subsequent primary care consultation because of pain that is not resolving, those presenting to an emergency room, and those who had been referred for specialist care.
Thus, even though the incidence of serious disease is very low at the time of first consultations for new episodes of acute low back pain, clinicians do need to keep their diagnosis of nonspecific low back pain under review during the subacute and early chronic phases of low back pain (8). Indeed, these serious disorders can develop in patients with established disabling chronic low back pain and thus cannot be disregarded regardless of how long the patient has been experiencing low back pain.
With this extremely low incidence of serious disease identified following a first consultation for a new episode of acute low back pain, the large number of patients with 1 or more red flag symptoms (80%) is of considerable concern. If nearly everyone with acute low back pain has a red flag, then the presence of a red flag will not help the clinician in deciding whether any further investigation or treatment is needed. Nearly all of the individual red flag questions were uninformative. The only diagnostic decision rule that Henschke and colleagues could generate was for fracture (positive response for 3 of 4 factors, female sex, age >70 years, prolonged use of corticosteroids, and significant trauma), which had a high positive likelihood ratio of 218. This was not, however, substantially different from the positive likelihood ratio of 194 for a clinical diagnosis made by clinicians without using a formalized diagnostic decision rule. This resonates with the observation that screening tools for psychosocial “yellow flags,” designed to identify people with nonspecific low back pain with a poor prognosis, did not perform substantially better than clinical judgment (9).
Any recommendations for the use of red flags need to consider how likely it is that patients who are seen in primary care will have a serious disorder causing their low back pain and the consequences if the diagnosis is overlooked.
Cauda equina syndrome due to disc prolapse can have catastrophic consequences, and early diagnosis and surgical treatment are probably helpful (10). However, the incidence of cauda equina syndrome is so low that most general (family) practitioners in the UK will not see a true case in their practicing lifetime (10). The clinical diagnosis of cauda equina syndrome is difficult, with a false-positive rate of 43% even when the diagnosis is made by experienced clinicians (10). There is a need to be vigilant for the new onset of perianal sensory change or bladder symptoms in patients with low back pain of any duration, with a low threshold for referral for expert assessment (10). However, this vigilance needs to be tempered by the reality that many of us who work in primary care will never see a case of cauda equina syndrome.
Osteoporotic fractures of the spine are a relatively common presentation in primary care but represent <1% of the presentations in the study by Henschke and colleagues (6). Clinician judgment, however, has a positive predictive value similar to that of the formulaic red flag–based diagnostic rule (6). In cases of acute back pain, appropriate management for most patients will be pain relief and resumption of normal activities as soon as possible; such an approach is similar to the management of nonspecific low back pain. In the absence of convincing evidence that vertebroplasty or kyphoplast are superior to medical management, there is little need for urgent investigation and referral to specialists (11). Diagnosing and treating osteoporosis to prevent further fractures are important, and possibly more important than diagnosing the fracture itself, in persons at risk of a fragility fracture. In my opinion, there is little point in doing further investigations for most patients with known osteoporosis who present with a new episode of low back pain.
Infections may account for 0.01% of cases of low back pain (7). It is hardly surprising that Henschke et al did not identify any cases of infection in their study of only 1,172 patients, even though reasonable numbers had positive responses for the red flags suggested to screen for infection (6). In practical terms, this means that spinal infection as a cause of low back pain will, like cauda equina syndrome, be a once-in-a-lifetime diagnosis for most practitioners working in primary care. Nearly 3% of patients with acute low back pain have the commonest red flag for infection (constant, progressive nonmechanical pain) at the time of their first consultation (6). There is a need to be vigilant for the patient with deteriorating back pain who is systemically unwell, but this vigilance needs to be tempered by knowledge of the rarity of spinal infections.
Ankylosing spondylitis and other inflammatory disorders are an uncommon cause of low back pain. These are chronic disorders, the diagnosis of which is commonly delayed for several years. What is less clear is whether the majority of patients with mild disease are harmed by this delay, because many will be treated with nonsteroidal antiinflammatory drugs and advised to exercise irrespective of the diagnosis. In primary care patients with chronic low back pain, established ankylosing spondylitis with unequivocal radiographic change is rare, although using wider diagnostic criteria for an axial spondylarthropathy, the prevalence may be as high as 5% (12). The risk here from the use of red flags is swamping of secondary care services, because of the high number of false-positive results for the screening questions. Three of the screening questions used by Henschke and colleagues had positive replies from more than one-fourth of the patients; these patients represent a population with acute pain who would not at first presentation satisfy any diagnostic criteria for an axial spondylarthropathy. The possibility of ankylosing spondylitis as a diagnosis needs to be considered, but only in patients who are not experiencing improvement after more than 3 months.
Malignancy is a diagnosis that practitioners would not wish to miss. It is reassuring that none of the 46 patients with a past history of cancer in the Henschke study had malignancy as a cause of their back pain (6). The formulaic use of a red flag of a past history of cancer is too blunt an instrument to be used in routine practice without considering the type of cancer and how long ago it was diagnosed. Except in the context that a cancer is being, or has recently been, treated, clinician judgment is needed to decide which cases may need further investigation. In cases of increasing pain or failure to improve, malignancy needs to be considered, because it may be that it is at the second or third consultation (not studied by Henschke) when malignancy needs to be primarily considered.
Too great a focus on addressing red flag questions may distract the clinician from delivering key information to the patient: reassurance as to the benign nature of the disorder for the vast majority of patients and the benefits of avoiding bed rest and maintaining normal activity, including work. It is worrying that some investigators are advocating comprehensive recording of answers to all red flag questions as a desirable aspiration in back pain management (13). The indiscriminate use of red flag symptoms as a trigger to order further investigations will lead to unnecessary investigations that are themselves harmful, through a combination of overmedicalizing a benign usually self-limiting disorder, the harmful effects of radiation from obtaining unnecessary radiographs and computed tomography scans, and the consequences of these investigations themselves producing false-positive results.
That factors known to be associated with a specific diagnosis were not found to be helpful in this study might appear surprising. It is important to recognize that a statistically significant association between a screening tool and the condition of interest does not mean that its positive and negative predictive values in any particular population is sufficient to justify its use in clinical practice (14). It is tempting to call for further research to determine which patients to investigate for specific causes of low back pain in different populations. However, any study with sufficient statistical power to produce robust estimates of the sensitivity and specificity of single and multiple variables, which can then be used to produce positive and negative predictive values in different populations, is likely to be many times larger than this current study. It may be difficult to persuade research funders that this is good use of resources.
Few people will come to significant harm if the diagnosis of a serious cause for their back pain is delayed for a moderate period of time. Taking this and the poor performance of red flags used to identify people with specific causes for their back pain into consideration, we should refocus our attention away from recording an exhaustive list of red flags to considering a small number of disorders in which early diagnosis and treatment might make a big difference, and use time as a diagnostic tool for the remainder. Specifically, we should consider cauda equina syndrome, major intraabdominal pathology, focal infections, and fractures. If, in the clinician's judgment, the patient may have one of these, then appropriate investigation and treatment are needed. This judgment may need to be informed by combining multiple observations, perhaps made over 1 or more consultation. Because we do not have enough data to create formal decision rules, this needs to rely on the skills and experience of the treating clinician. This approach, which is grounded in considering the clinical features of disorders of interest, is likely to be more discriminatory than formulaic application of red flags and decision rules.
I am grateful to Shilpa Patel, Dawn Carnes, and David Evans for their comments on earlier versions of this article.