A cluster of at least 3 of 4 positive tests was found to be useful for CLI diagnosis
In absence of a true clinical gold standard this conclusion must be taken with caution
Combined with a thorough history, this cluster may help to evaluate the presence of lumbar instability
Lumbar instability is believed to be a significant contributor to LBP-subgroups (prevalence ranging from 13-33%) and is often missed in clinical practice. Instability is defined as a decreased capacity of the spinal stabilizers to maintain the neutral zone within the physiological limits of movement. Up until now clinicians diagnosing clinical lumbar instability mostly rely on medical imaging, which is quite remarkable given that this imaging assumes structural instability. However, findings on medical imaging do not always correspond to pathology and therefore, this gold standard can be questioned. We highlight this paper as it made a significant effort to investigate not so much structural but rather clinical lumbar instability (CLI) from several findings and tests proposed in literature earlier on.
A cross-sectional study was conducted which included 200 participants between 40-60y with LBP over a period of 3 months. Patients were classified in 2 groups: a CLI group and a group of other spinal pathologies.
A cluster of 4 tests for diagnosing clinical lumbar instability was carried out by a physical therapist:
This cluster was then tested against a self-constructed reference standard containing 13 recommended history-taking signs and 6 physical examination findings as proposed in earlier studies. An orthopedic surgeon assessed whether these findings were present. The reference standard was considered positive when 7 and 3 signs from respectively the history-taking and physical examination were present.
A cluster of 3/4 positive tests was the most accurate cluster of tests overall, with the highest LR+ (5.8) and the second highest specificity (91.7%) but the second lowest sensitivity (47.8%) and LR- (0.6). The cluster of two of four clinical tests showed the second highest sensitivity (89.1%), LR+ (2.4), and LR- (0.2).
While the authors have made a substantial effort to study the diagnosis of CLI, this study shows several flaws. First and foremost, participants were recruited from an orthopedic department in a hospital, which may cause that patients with more severe LBP have entered the study, thus limiting generalizability. A technique of “convenience sampling” was used, which recruits patients from a group that is easily accessible or highly interested in participating. Therefore, the sample may not fully reflect all LBP patients. Furthermore, patients presenting with an inability to actively move the lumbar spine due to severe pain/muscles spasms were excluded, which we think may be a sign of CLI. Also it was not indicated whether the required sample size was determined a priori or a posteriori and the flow chart did not indicate how many patients in total were assessed for eligibility.
Further problems arise with the so-called “gold standard” reference test. As it is composed of several imperfect tests, we cannot ascertain the reference test being a true gold standard, possibly introducing imperfect reference standard bias and thus caution with interpreting the diagnostic accuracy of the proposed cluster is warranted. Yet it was a good option to consider this composite reference standard as it avoided the diagnosis of CLI through structural findings as evaluated on medical imaging. The orthopedic surgeon evaluated the presence of a positive or negative result on the reference standard. This may be of issue since orthopedic surgeons may look differently to LBP as physiotherapists do. Furthermore, it is unsure if the choice of 7 history findings and 3 physical examination findings having to be present for a positive reference standard was determined arbitrarily or based on proposed evidence from the literature. Last but not least the risk of incorporation bias is present as some of the index tests make part of the reference test.
Link to the paper: https://pubmed.ncbi.nlm.nih.gov/33099174/