Intended for healthcare professionals

CCBYNC Open access

Red flags to screen for malignancy and fracture in patients with low back pain: systematic review

BMJ 2013; 347 doi: (Published 11 December 2013) Cite this as: BMJ 2013;347:f7095

Re: Red flags to screen for malignancy and fracture in patients with low back pain: systematic review

Dear Ed

I note the article by Downie et al which is in part based on the Cochrane review headed by one of the co-authors Nicholas Henschke(Ref 1).

I would like to point out a potential source of error in the selection of the main reference in the analysis of history of cancer (in the emergency department setting) used as a red flag in relation to presence of spinal malignancy. The authors has concluded the "history of cancer" was found to be "informative" as it provides a post-test probability of 33%.

I believe that their conclusion is based solely on the analysis performed in the Cochrane review (Ref 1) in which the study by Reinus et al (Ref 2) was the sole research performed in emergency department setting.

From the analysis detailed in the Cochrane review, the study based on patient assessment questionnaire matched with radiological findings (n=482) appears to have concluded that the proportion of true-positives/ false-positives/ true-negatives/ false-negative is 7/13/0/442. Even assuming the FN=442 was a typo in the review pdf (expected true FN=462) the basis of analysis in this way based solely on published data represents a leap of faith.

In summary of the Reinus article on their data relating to malignancy, of the 482 patients referred for lumbosacral spine (LSS) X-ray crossed referenced with the indications for imaging completed by referring clinician, 20 patients was investigated for concerns with previous history of malignancy but only 3 has malignancy detected on LSS xray. Another 4 with no malignancy history identified in the questionnaire, was found to have vertebral metastasis on X-ray, but chart review of these patients found to have history of bone metastasis.

Furthermore chart review was not routinely performed on all patients; specifically the Reinus et al stated that they did not review the charts of patients with LSS Xray reports showing either normal findings or spondylosis (although it appears they have reviewed charts of at least distinct group, those with trauma history n =173 and those with positive malignancy LSS findings). This makes it near impossible to determine the proportion of patient with chart history of malignancy but not detected by questionnaire and has normal/spondylosis LSS findings. This makes the handling of data by Henschke et al (Cochrane review) inconsistent at best, since they chose to admit the 3 patients with negative malignancy on questionnaire (but positive on chart) and LSS reported as malignancy but cannot account for the accuracy of the other patients whose history is based on questionnaire but chart is not reviewed (n = 482 - 173 - 20 - 3 = 286)

As a result this represent a serious flaw by Dowie et al in accepting the misguided analysis as presented by the Cochrane group, and subsequently I believe no reasonable conclusion can be made from the published data by Reinus et al on the diagnostic accuracy of history of maligancy as red flag for spinal malignancy.

1. Henschke N, Maher CG, Ostelo RW, de Vet HCW, Macaskill P, Irwig L. Red flags to screen for malignancy in patients with low-back pain. Cochrane Database Syst Rev2013
2. Reinus WR, Strome G, Zwemer FL, Jr. Use of lumbosacral spine radiographs in a level II emergency department. AJR Am J Roentgenol1998;170:443-7

Competing interests: No competing interests

12 December 2013
Shyan Goh
Orthopaedic Surgeon
Sydney, Australia