Introduction

As with all tissues in our bodies, the sacroiliac joint may cause symptoms of lower back pain. Many times, research has tried to evaluate provocation tests for sacroiliac joint pain, but results are often very variable. Depending on the reference standard used, the prevalence of sacroiliac joint pain ranges from 10-64%. Some studies clearly advise against the use of these provocation tests, while others recommend its use in clinical practice. To unravel the uncertainty around sacroiliac joint pain, this study tried to systematically review the diagnostic accuracy of sacroiliac joint provocation tests, using meta-analyses to pool the evidence.

Methods

A systematic review with meta-analysis was conducted including diagnostic accuracy studies. In these diagnostic accuracy studies, an index test is compared to a reference test (so called gold standard) to evaluate the diagnostic ability of the index test.

The index test chosen for this study were pain provocation tests for the sacroiliac joint in the form of a cluster. This was compared to local intra-articular anesthetic blocks for relieving sacroiliac joint pain, which is considered to be the reference or the so-called gold standard test. The target condition was the presence of sacroiliac joint pain with the cluster of tests applied.

Results

From 1957 potential records, 5 studies were included in the review. The provocation clusters examined combinations of following tests: distraction test, thigh thrust test, Gaenslen’s test, compression test, sacral thrust test, flexion abduction external rotation (FABER) test, sacral distraction test, lateral compression test, Patrick’s test, Yeoman’s test, Newton’s test, and the ipsilateral Gaenslan’s test.

Pooled values were: sensitivity 0.83 (95%CI: 0.62, 0.93), specificity 0.86 (95%CI: 0.36, 0.79) and the false positive rate 0.41 (95%CI: 0.21, 0.64). The corresponding positive likelihood ratio was 2.13 (95%CI: 1.2, 3.9) and the negative likelihood ratio was 0.33 (95% CI: 0.11, 0.72). The diagnostic odds ratio was 9.01 (95% CI: 1.72, 28.4). There was evidence for a substantial amount of heterogeneity in the pooled studies.

After a positive test and considering a pre-test probability (prevalence) of 20% with the pooled positive likelihood ratio of 2.13, the post-test probability of sacroiliac joint pain rises slightly to 35%. A positive test is thus not very helpful in the clinical diagnosis of sacroiliac joint pain. In case of a higher pre-test probability (prevalence) of 30%, the post-test odds of sacroiliac joint pain being the cause of symptoms slightly increase to 48%. Still not enough to confidently blame the sacroiliac joint for the symptoms of your patient.

With a negative test and a pre-test probability of 20%, the negative likelihood ratio of 0.33 leads to a post-test probability of 8%. When the higher prevalence of 30% is used, the post-test odds of sacroiliac joint causing the symptoms are 12%. It seems that this cluster can be more confidently used to rule out the sacroiliac joint as a cause of symptoms.

Questions and thoughts

Rather than depicting sensitivity and specificity alone, the authors tried to make the findings easily applicable for the clinician by reporting likelihood ratios. From these likelihood ratios and from your pre-test suspicion of the possibility of sacroiliac joint pain causing the symptoms, the post-test chances can be calculated. The authors used 2 different prevalence rates of sacroiliac joint pain, based on population prevalences found in previous systematic reviews: 20 and 30%. This leads to a respective 35% and 48% post-test probability in case of a positive test. As you can see, you might as well flip a coin to evaluate if the sacroiliac joint causes your patients’ symptoms.

“The multifactorial nature of pain is unlikely to be captured alone by a pain provocation test or basing diagnosis on the introduction of anesthetic into the SIJ.”

Moreover, these prevalence rates of 20 and 30% may have been overestimated. Some studies report a prevalence of only 2%, and in this case, the post-test probability would be 4% in the case of a positive test and 1% in case of a negative test. This means that ruling in sacroiliac joint pain after a positive test is not possible. In case of a negative test you can be more confident to rule out the sacroiliac joint as a cause of symptoms, but why would you perform this cluster if your suspicion is already so low (2%)? Another problem arises when looking at the reference standard. The use of anesthetic blocks cannot be regarded as a solid reference standard as this procedure also gives false positive results. In this light, it is necessary to cautiously interpret these findings.

Talk nerdy to me

The authors did really well on the methodological aspects as they followed all standards for proper reporting of systematic reviews and meta-analyses. The review was prospectively registered on PROSPERO and was reported according to the PRISMA guidelines for diagnostic test accuracy reviews. Studies were evaluated with the QUADAS and the available evidence was rated according to the GRADE recommendations.

A thorough search was conducted in several databases from inception up to September 2020. No filters were applied and to complete the search, reference lists were screened for potentially eligible articles that the search may have missed. Handsearching of the included trials was also performed. We can therefore assume all relevant articles were included in this review. A limitation of this study lies in the language filter that restricted the studies to be included in English and German language only.

Substantial heterogeneity exists in the results and because of the small number of included studies, it was not able to perform subgroup analyses. As all studies were at high risk of bias, the certainty of the evidence was therefore downgraded to very low. An important aspect to note is that most of the studies included preselected study participants. This implies that patients with pain in the region of the sacroiliac joint who were referred for invasive procedures in specialist units were included. This diminishes the generalizability of the findings to the population of patients with lower back pain presenting in general physiotherapy practice.

Take home messages

There is very low certainty evidence that clusters of pain provocation tests provide low accuracy for detecting or ruling out the SIJ as the cause of pain. You might as well flip a coin. Sacroiliac joint pain can be ruled out with more confidence in case of a negative result on the cluster. The results may not be generalizable to general physiotherapy practice as patients in the included studies were preselected and referred to specialized clinics.

Reference

Saueressig, T., Owen, P. J., Diemer, F., Zebisch, J., & Belavy, D. L. (2021). Diagnostic accuracy of clusters of pain provocation tests for detecting sacroiliac joint pain: systematic review with meta-analysis. journal of orthopaedic & sports physical therapy, 51(9), 422-431.