Serious Pathologies in the Lumbar Spine: Prevalence and Diagnostic Accuracy of Red Flag Questions

Street, Kathryn J.
White, Steve
Vandal, Alain
Item type
Degree name
Master of Health Science
Journal Title
Journal ISSN
Volume Title
Auckland University of Technology

Low back pain is a common problem that can be seriously or chronically disabling. It is one of the most common reasons for people to visit their general practitioner and is associated with high healthcare costs. Low back pain is frequently benign, but in rare cases may be due to underlying serious pathology. However, the actual likelihood of a patient presenting with a serious pathology to primary, secondary, or tertiary care is largely unknown. Without knowledge of prevalence it is not possible for clinicians to estimate the probability of a patient having a serious pathology. Additionally, knowledge of incidence is required to understand disease aetiology, including which age or ethnic groups may be more at risk. Clinical identification of serious pathologies can be difficult, with evidence that the diagnosis of these cases is often delayed and not uncommonly missed. Therefore, screening questions, known as ‘red flag’ questions, have been widely recommended to assist with early recognition of serious pathologies in the lumbar spine and are included in a number of international guidelines. However, the diagnostic accuracy and utility of these questions is poorly understood, and several authors have expressed concerns regarding their use, given the lack of supportive evidence. The aims of this thesis were to determine the prevalence of serious pathologies that commonly affect the lumbar spine (vertebral fracture, malignancy, cauda equina syndrome, and infection) in secondary and tertiary care settings, and to determine the incidence of serious pathologies in the geographic region of Counties Manukau, Auckland, New Zealand. This thesis also aimed to determine the diagnostic accuracy of red flag questions commonly used to screen for serious pathologies in patients presenting with low back pain. The primary component of this research was a retrospective audit of 2,383 lumbar MRI scans. Adult participants who were referred for a lumbar MRI scan for the investigation of low back pain were consecutively recruited over a 10-month period. Target condition prevalence was calculated as a percentage of the study population and the prevalence specific to secondary care and tertiary care was also calculated. The incidence of serious pathologies was determined by comparing the 2013 census results from the Counties Manukau region to the data collected from Middlemore Hospital (tertiary care). Data was subcategorised into age groups, gender, and ethnicity, to allow investigation of which groups may be most at risk. The prevalence of serious pathologies varied from 0.12% for spinal infection in secondary care to 6.7% for vertebral fractures in tertiary care. The prevalence was significantly higher in the tertiary care setting than in the secondary care setting for all serious pathologies. The total incidence varied from 2.5 per 100,000 person-years for cauda equina syndrome to 12.9 per 100,000 person-years for vertebral fracture. Overall there was no significant difference between genders. However, the risk of developing a serious pathology increased significantly with age and peaked at 249 per 100,000 person-years in the 85 years and over group. To determine the diagnostic accuracy of the commonly recommended red flag questions, 552 consecutive participants referred for an MRI scan for LBP were consecutively and prospectively recruited. All participants were required to complete a questionnaire that contained 37 questions related to specific spinal pathologies and a body chart to provide further detail of their symptoms. This cohort was a subgroup of the full cohort included in the study designed to determine prevalence and incidence. Data collection for these two studies occurred concurrently. This study demonstrated that a number of red flag questions or index tests had negative likelihood ratios less than 0.1, indicating that the absence of these findings results in a conclusive shift in probability that the target pathology will be absent. These index tests were: age greater than 35 years for vertebral fracture, age greater than 42 years or ‘worsening pain’ for malignancy, and age greater than 55 years, insidious onset of pain or ‘night pain that wakes you from sleep’ for spinal infection. Hence, these index tests have sufficient diagnostic accuracy to suggest that they can be used as screening tests. For cauda equina syndrome, no index test had a negative likelihood ratio less than 0.1. This study also demonstrated that only two red flag questions had positive likelihood ratios >10, indicating that the presence of these findings results in a conclusive shift in the probability of that pathology being present. With respect to vertebral fracture, the only index test that met this criteria (LR+ >10) was a history of concomitant HIV or AIDS. For spinal infection, a history of immunosuppressant use was the only red flag question with a positive likelihood ratio greater than 10.. Hence, these tests have potential diagnostic utility as risk factors. No red flag questions for cauda equina syndrome or malignancy displayed a conclusive increase in probability. This thesis has provided important new information related to the prevalence and incidence of serious pathologies within a population of low back pain patients presenting to secondary or tertiary care. In addition, this thesis has determined the diagnostic accuracy of all commonly recommended red flag questions to screen for serious pathologies. Hence, this study has provided information that can be used to determine pre and post-test probability to assist with clinical decision-making, and facilitate early diagnosis and treatment to improve patient outcomes.

Serious pathologies , Low back pain , Red flags , Prevalence , Diagnostic accuracy
Publisher's version
Rights statement