Accuracy of clinical neurological examination in diagnosing lumbo-sacral radiculopathy: a systematic literature review

Lumbo-sacral radiculopathy, a substantial cause of disability and morbidity, represents one distinct presentation of low back-related leg pain, which constitutes between 23% - 57% of LBP cases [1]. Lumbo-sacral radiculopathy refers to a pathologic process involving the lumbo-sacral nerve roots causing radicular symptoms into a lower extremity [2], which may or may not be accompanied by other radicular irritation symptoms and/or symptoms of decreased function [3]. Lumbar IVD protrusion is the most common cause underlying nerve root irritation and subsequent radiculopathy [1‐3]. However, other mechanical factors including, lumbar vertebrae osteophytes, lumbar facet joint hypertrophy or ligamentum flavum hypertrophy may also cause lumbar nerve root compression [3]. Radicular symptoms may also be primarily caused by inflammatory reactions of the neural or surrounding musculo-articular structures [4], hence suggesting that lumbar radiculopathy is not always mechanically mediated, and that mechanical nerve root compression on its own does not necessarily determine radicular symptoms as seen on positive MRI findings on asymptomatic subjects [5].

In clinical practice, the diagnosis of lumbo-sacral radiculopathy involves the use of various tools and procedures including neuropathic pain screening, clinical neurological examination, electro-diagnosis, nerve root blockage and radiological imaging [3‐5]. Clinical neurological tests include sensory, motor, reflex, neuro-dynamic and nerve trunk palpation procedures designed to assess the physiological and bio-mechanical status of specific lumbar nerve roots thought to be responsible for the patient's signs and symptoms [5]. Determination of the presence or absence of radiculopathy is dependent upon the examiner's awareness of clinical signs and symptoms, physical examination, knowledge of possible pathology, mechanisms of injury and ability to perform the tests correctly [6‐8]. The clinical usefulness of neurological examination tests is largely determined by the accuracy with which they determine the presence or abscence of the suspected patho-neuro-physiology.

MRI is frequently utilized in detecting nerve root compression, one of the many causes of radiculopathy [4, 9]. While the accuracy of MRI in detecting alterations in both the anatomy and tissue properties is well established, the relationship between the detected anatomical abnormalities and clinical history and patients outcomes remain controversial [6].

Although MRI is being used as a diagnostic tool of choice by clinicians in practice and a gold standard by researchers in primary diagnostic accuracy studies [10, 11], there are several limitations proposed in the literature. One, MRI embraces the patho-anatomical model yet radiculopathy is not always mechanically mediated by IVD nerve root compression as earlier reported [12]. Two, there is not an acceptable gold standard diagnostic tool to which MRI can be compared [13‐15] This is because, even though conventional electro-diagnostic procedures are sometimes used as gold standard for detecting nerve involvement, experts argue that they leave the function of small caliber afferent fibers unexplored, and therefore, there is no basis for positive findings [7‐9]. Current perception threshold testing [7], electro-myelography [8], and nerve root blocks [9] on the other hand are used mainly to confirm symptomatic structures.

Early and accurate diagnosis of lumbar radiculopathy is crucial to ensure target-specific treatment and avoid chronicity, disability and work loss [14, 15] and clinical neurological examination forms a vital component of the initial diagnostic work-up for patients with clinical suspicion of lumbar radiculopathy. Clinical neurological examination tests could be used to discriminate patients with radiculopathy distinct from other low back pain sub-types like non-specific low back pain of somatic origin, lumbar facet or intervetebral joint derangement disorder. These tests are easy to perform, cost-effective and run a relatively very low health risk to patients. It is therefore imperative to identify those which have a reported acceptable diagnostic sensitivity and/or specificity through a structured systematic review. The available systematic literature reviews which have been published recently have an evident variation in case definition of lumbar radiculopathy and have also focused on detection of disc herniation or protrusion as the only cause of nerve root compression and subsequent radiculopathy [10, 11]. Different from this trend and for the purposes of this review, our operational clinical definition for lumbo-sacral radiculopathy was: “Objective loss of sensory and motor function with or without accompanied spinal and/or referred leg pain following a mechanical or bio-chemical dysfunction of lumbar and sacral spinal nerve roots and their associated dorsal root ganglions (DRGs)”. This review therefore aimed at determining the accuracy of clinical neurological tests in diagnosing lumbo-sacral radiculopathy.

This review was conducted using the diagnostic tests accuracy (DTA) protocol [10].

The search on relevant electronic data bases retrieved a total of 1568 articles (Fig. 1) by the first hit of the key terms and the mesh terms. After screening the title, key words and abstract of all articles and removal of duplicates, 39 articles were selected as potentially suitable for inclusion and were retrieved as full articles for further analysis. Out of the 39, 24 were selected from those that were generated by the entry of the key terms while 15 were selected from the output of the mesh terms. Full screening of the 39 articles was independently done by two reviewers (NT & ID) using a PICO analysis and disagreements were resolved through adjudication by a third reviewer (AR). Twenty-eight studies were further excluded for not meeting the inclusion criteria. An additional reference hand-searching of all included studies and subject specific journals was done by one reviewer (NT) but did not yield any more relevant studies. A total of 12 studies whose characteristics are summarized in Table 2 (Albeck 1996 [13], Haldeman et al. 1998 [18], Vroomen et al. 2002 [10], Rabin 2007 [19], Majlesi et al. 2008 [20], Bertilson et al. 2010 [14], Lee-Robinson et al. 2010 [2], Coster et al. 2010 [7], Suri et al. 2010 [15], Suri et al. 2011 [21], Trainor and Pinnington 2011 [16], Iversen et al. 2013) met the inclusion criteria (Published as a full article in a peer-reviewed journal, in English; Evaluated the sensitivity and/or specificity of clinical neurological test (s) in diagnosing lumbar/sacral radiculopathy; Incorporated a comparator test (s); Study subjects presented with clinical signs and symptoms consistent with lumbo-sacral radiculopathy as diagnosed by the referring clinicians). Of the 12 studies included in this review, 11 were cohorts while 1 was a case control study.

The clinical neurological examination tests assessed by the included studies were the standard sensory (soft touch and pin prick), motor (functional tests and resisted isometric contractions), deep tendon reflex (patella) and neuro-dynamic (Straight Leg Raise and Femoral nerve) tests. MR imaging was used as a reference standard in 8 of the included studies while 2 studies used EMG, one electro-diagnostics and CT, and the other one intra-operative findings. Eleven studies were carried out in secondary and tertiary care settings while one was a primary care diagnostic study. The search history is displayed in Fig. 1.

The QUADAS scores for each of the 12 included studies across all QUADAS items ranged from 50% – 92% (vertically) and the scores for all included studies per QUADAS item ranged from 0 – 100% (horizontally). The Bertilson et al. (2010) [14] study had the highest score of 92% across all QUADAS items followed by Suri et al. 2011 [21], Rabin et al. 2007 [19], Vroomen et al. 2002 [10], Haldeman et al. 1998 [18] and Iversen et al. 2013 while the Albeck 1996 [13] study had the lowest score of 58%. All studies fulfilled items 1, 5 and 6 which concern a representative spectrum of study subjects, verification bias and clear explanation of index test execution respectively; while none of the 12 studies met criteria 11 on reporting uninterpretable or intermediate index test results. The scores are displayed in Table 1.

The current review evaluated 12 primary diagnostic accuracy studies that specifically assessed the performance of various individual clinical neurological tests in detecting nerve root impingement. Different from previous reviews [11‐13], we did not consider disc herniation as the cause of nerve root impingement and subsequent radiculopathy. A meta-analysis of pooled data for individual tests was not performed due to heterogenity of the included studies.

The current review analysed the accuracy of index tests for diagnosing lumbo-sacral radiculopathy (sensory, motor, reflex and neuro-dynamic) by comparing them to MR imaging, electro-diagnostics or intra-operative findings either in generally detecting nerve root impingement at mid-lumbar (L2-L4) or lower-lumbar (L4-S1) or at specific segmental nerve root levels (L2, L3, L4, L5, S1).

All the studies after the year 2000, that evaluated the diagnostic performance of sensory testing, used MR imaging as a reference standard. However, the oldest study by Albeck (1996) [13] which compared clinical assessment with surgical findings, demonstrated the best sensitivity (confidence interval 95%) 0.61 (0.47-0.73) with a moderate specificity of 0.63 (0.38-0.84). Higher specificity in this study may be attributed to the fact that patients who are scheduled for surgery are carefully selected compared to those whom surgey is not contemplated. Hence the probability of a positive index test result becomes relatively higher in the surgical than imaging group. The results of the reviewed studies indicate that sensory testing of superficial soft touch and superficial pain are very specific and could therefore be used to rule in the diagnosis of lumbo-sacral radiculopathy among patients presenting with low back and radiating leg symptoms.

Motor tests evaluated in the reviewed studies were mostly functional tests of heel walk, heel raise, sit-to-stand, and resisted isometric contractions for hip flexion, knee extension, great toe extension, ankle dorsi- and planter flexion. The test in all studies was determination of paresis or muscle weakness. Sensitivity was poor to moderate, ranging from 0.13 (0.04-0.31), in the study of Bertilson et al. (2010) [14] to 0.61 (0.36-0.83), in the study of Suri et al. (2011) [21]. The clinical implication of these findings is that motor tests are not ideal for ruling out the diagnosis of lumbo-sacral radiculopathy. The highest specificity was reported in the Suri et al. (2011) [21] for detecting S1 nerve root impingement. A clear description of the actual execution of motor tests, which permits duplication, was provided in the Bertilson (2010) [14] study.

Deep tendon reflex testing focused on evaluation of the patella and Achilles’ tendon reflexes. Generally, in the studies where reflex testing was included, diagnostic performance of reflex tests across the studies was notably good with specificity ranging from 0.60 (0.51-0.69) in the recent Iversen et al. study to 0.93 (0.87-0.97) in the Vroomen (2002) [10] study. However, the sensitivity was moderate with the highest being 0.67 (0.21-0.94) in the Iversen (2013) study. Therefore the results of this review present evidence for use of deep tendon reflex tests as confirmatory tests in the diagnosis of lumbo-sacral radiculopathy. However, index test procedure, together with the cut-off values for positivity, were not provided in some of the studies, and where provided, there were outright procedural variations.

There seem to be a disconnect among researchers regarding the diagnostic utility of lower limb neuro-dynamic tests which include the SLR test for the sacral plexus and the femoral nerve stretch test for the lumbar plexus. In some studies, these tests were intended to detect the presence of disc herniation and subsequent nerve root compression [7], and in some [10, 16] studies they were proposed to test mechanical sensitivity of the femoral and sacral plexii. Also, the procedural difference between the SLR test and Lassegue’s sign is not clear to some authors of primary diagnostic test accuracy studies. There is thus a high probability that such variations would negatively impact on the reported diagnostic performance of the neuro-dynamic tests. A good sensitivity and specificity 1.00 (0.40-1.00) and 0.83 (0.52-0.98) respectively was reported in the Trainor & Pinnington (2011) [16] study with the rest of the studies recording a poor and moderate diagnostic performance. Therefore in light of these findings, lower limb neuro-dynamic tests (FNST and SLRT) are more sensitive than specific hence ideal for ruling out the diagnosis of lumbo-sacral radiculopathy.

In this review, the diagnostic accuracy of most clinical neurological tests range from low to moderate. This finding may stem from several factors ranging from variations in operational case definition of the target condition, outcome of clinical testing, that is, detection of radiculopathy due to disc-related nerve root compression among others.

The outcome of previous systematic reviews on diagnostic accuracy of clinical neurological testing could be questioned due to inconsistencies in specific objectives of diagnostic tests for the primary study selection, and therefore the criteria used to select studies.

Verification bias may also contribute towards the minimal utility of clinical neurological tests reported since the commonly utilized reference standard is MR imaging whose value and accuracy is known only in detecting visible structural nerve root impingement which does not necessarily mediate radicular symptoms yet the evaluated index tests are intended to detect radicular symptoms.

Another contributing factor to the rather poor performance of sensory tests is the variability of dermatomal maps for sensory testing. These tests are guided by published dermatome maps indicating the cutaneous fields of the suspected spinal nerve roots, however, there are reported variations among these maps [19, 21]. Dermatomes are also known to overlap and vary across individuals due to possible extra-dural anomalies where two pairs of nerve roots may arise from a single dural sleeve or extra-dural anastomosis [22].

While clinical neurological tests remain a vital component of the initial diagnostic work-out of patients suspected of radiculopathy, and for researcher and clinicians to establish their actual clinical utility, a common ground must be reached in terms of operational definition of the target condition, the index test outcome and the homogeneity of reviewed studies. This would improve the reported accuracy and ultimately the diagnostic credibility of clinical tests.

Sensory testing has moderate sensitivity in the detection of lumbo-sacral radiculopathy and prior knowledge of MRI results is a source of bias in sensory testing. This review highlights the inconsistencies in execution of motor tests and grading of test results, such methodological di-similarities could be attributed to the variations in motor tests sensitivities as reported in the primary diagnostic studies analysed in this review. Similarly, SLR test and Lassegue’s sign have been used interchangeably with variation on the expected diagnostic outcome on whether they detect IVD prolapse and subsequent nerve root impingement or hypersensitivity of the lumbar and sacral plexii to mechanical loading. There is however an acceptable level of consistency and similarities in execution and reporting of deep tendon reflex tests which in this review showed good sensitivity in detecting lumbo-sacral radiculopathy. However, in clinical practice, the diagnosis of lumbo-sacral radiculopathy should always be arrived at through consolidation of sensory, motor and deep tendon reflex test results and not isolated single test results.