Diagnostic accuracy of the clinical examination compared to available reference standards in chronic low back pain patients

Abstract: It is often stated that most cases of persistent low back pain are diagnostically indeterminate. Diagnostic accuracy estimates are obtained by comparing the results of clinical tests with diagnostic reference standards. Although data is available from past research, most commonly used clinical tests have not been evaluated in validity studies. Significant issues regarding reference standards in back pain research exist, such as availability, cost and invasiveness.The aim of this thesis work was to explore the diagnostic accuracy of non-invasive clinical examination techniques and variables, in relation to available reference standards for diagnosis of the most common patho-anatomic sources and causes of persistent back pain. The intervertebral disc, the zygapophysial joints, sacroiliac joints are believed to be the most common sources of low back and referred pain. Spinal stenosis and herniated discs are infrequent but regularly occurring lumbar causes of causing radicular pain. A secondary aim was to evaluate the potential confounding influences of disability, psychosocial distress and inappropriate illness behaviours on diagnostic accuracy.This work comprises two projects with different patient samples attending the same specialist interventional radiology spinal diagnostics clinic in New Orleans, USA. In Project 1 48 patients received a clinical examination that included a McKenzie-styled repeated movement's examination identifying the presence or absence of the centralization phenomenon, six pain provocation sacroiliac joint tests, and controlled (double anesthetic) blocks to the sacroiliac joints as a reference standard. In Project 2, 216 patients received an extensive clinical examination that included the tests used in Project 1, and a variety of reference standard diagnostic procedures such as; provocation discography, zygapophysial joint blocks, sacroiliac joint blocks, MRI and CT scans. Physiotherapists experienced in the McKenzie method and examination of chronic low back pain patients carried out the clinical examinations. An experienced radiologist carried out the reference standard examinations. The radiologist and physiotherapists were blinded to each other's results. Pain drawings, pain intensity VAS's, and three questionnaires were used to evaluate pain characteristics, disability and psychosocial distress.The reference standard for discogenic pain was concordant pain production during hydraulic distention of a disc during discography in the presence of a negative control disc and post discography axial CT images demonstrating annular fissuring involving the outer annulus. The reference standard for ZJ pain was substantial relief following intraarticular joint or medial branch blocks. The reference standard for SIJ pain was substantial relief following intra-articular SIJ blocks. Reference standards for nerve root pain, spinal stenosis and herniated disc causing radicular pain, were based on a combination of clinical findings, CT or MRI imaging results and response to selective or caudal epidural blockade.Five papers were produced reporting diagnostic accuracy of the clinical examination for disc pain, zygapophysial joint and sacroiliac joint pain. The sixth paper reports on agreement between physiotherapy clinical diagnoses and reference standard / expert opinion diagnoses.Prediction of discography: 118 patients received discography with 70% having a positive response. The most useful clinical signs were: The centralization phenomenon (sensitivity 25%, specificity 96%); directional preference (sensitivity 49%, specificity 91%); history of persistent pain between acute episodes (sensitivity 32%, specificity 92%).Prediction of screening zygapophysial joint blocks: 120 patients received unconfounded screening ZJ blocks and 13 had 95% or more reduction in pain. Three variables had some predictive value: Age over 55 (sensitivity 39%, specificity 85%), Walking is best activity for pain (sensitivity 31%, specificity 92%), Sitting is best position for pain (sensitivity 33%, specificity 90%). A cluster of seven clinical signs and symptoms previously thought to be a useful predictive model ('Revel's criteria') were not predictive of a 75% reduction in pain.Prediction of controlled (double) sacroiliac joint blocks: 43 patients received double blocks and 11 had SIJ pain confirmed. The presence of 3 or more positive provocation SIJ tests had sensitivity and specificity of 91% and 78% respectively. In the subset of 34 patients without evidence of the centralization phenomenon, sensitivity and specificity of three or more positive SIJ tests was 91% and 87% respectively.The number of cases with spinal stenosis and herniated disc causing radicular pain was low, so diagnostic accuracy results for the clinical examination were not reported separately.Patients were typically chronic, with 51% being distressed, 60% being severely disabled, 30% exhibiting overt illness behaviours and 24% had abnormal pain drawings. Severe disability and the presence of three or more signs of inappropriate illness behaviour reduced the specificity of the centralization and directional preference for positive discography by 20%. Psychosocial distress categorization and pain intensity did not substantially confound the diagnostic accuracy of centralization or directional preference.Overall agreement between diagnoses based by the physiotherapist's examinations and the reference standard diagnoses was evaluated compared to what could be expected based on chance agreements. Agreement on patho-anatomic sources and causes of pain was 56% (kappa 0.31). Proportional Chance Criterion estimate of chance agreement was 33%.In conclusion, the physiotherapy clinical examination can identify a subset (up to 5O%) of positive discography cases; is unable to predict the outcome of screening blocks for ZJ pain; and can identify approximately 88% of painful SIJ cases. Agreement on pathoanatomic sources and causes of pain occurs in over 50% of cases.