Your recent systematic review sought to estimate the extent to which MRI findings predict future low back pain,
associated disability and global recovery. Was your approach to reviewing the literature similar to that of your previous review on this topic?

This systematic review was an update of a previous systematic review led by Steffens et al 2013. 

We followed an identical approach and only included longitudinal studies, which also incorporated secondary analyses of randomised controlled trials. 

We excluded randomised trials that included treatments designed to alter lumbar spine MRI findings such as lumbar surgery or antibiotic treatment for Modic changes.

We also only included studies if they had participants with an MRI at baseline, reported on at least one lumbar spine finding (eg, disc degeneration), provided an outcome measure at follow-up (eg, pain or disability outcomes) and provided enough data to calculate an association between the MRI finding and low back pain or disability outcomes.

We specifically wanted to see if MRI findings could predict future low back pain in three distinct groups and separated our analysis into patients with current low back pain, patients without current low back pain and a mixed sample of patients with and without current low back pain.

Given that MRI changes occur in both people with low back pain and people without pain, what is the rationale behind thinking that MRI findings might be predictive of future low back pain?

It is true that some previous systematic reviews have shown that some MRI changes are present in people both with and without low back pain; however, the prevalence of MRI findings is often higher in those with low back pain compared to those without low back pain.

Another point to consider is that most of these previous reviews are based on cross-sectional studies. 

Cross-sectional studies only provide information about individuals at a single point in time and do not provide information about any causal relationships or about whether MRI findings predict future outcomes.

Given the limitations of cross-sectional studies, we decided to update the Steffens et al 2013 review, which included longitudinal studies. 

Longitudinal studies are better at providing prognostic information and information about causal relationships because they measure relationships over a period of time. 

Given the limited research in this area, we wanted to explore whether particular MRI findings might predict worse pain or disability outcomes in the future.

Were any predictors in people without pain at baseline identified? 

Christopher Han's systematic review on the use of MRI findings in predicting low back pain is in the April 2023 issue of the Journal of Physiotherapy.

In people without low back pain at baseline, our meta-analysis found that those with disc degeneration may have worse pain and disability outcomes in the future. 

It is important to note that we were not able to perform meta-analysis for any other MRI findings due to a lack of data.

What about those with pain at baseline?

In people with low back pain at baseline, our meta-analysis found that Modic type 1 and 2 changes, Modic type 1 changes alone and disc degeneration were associated with worse pain and disability outcomes in the future.

What about the studies in mixed populations?

In samples with a mixed population, we were not able to perform any meta-analysis due to a lack of data. 

Results based on single studies (which should therefore be interpreted with a high level of caution) did demonstrate that Modic changes and disc herniation at baseline were associated with future worse pain outcomes.

So are these associations strong enough to warrant getting imaging routinely just to make a weak prognosis?

The short answer is no. We need more robust research in this area. 

For most MRI findings, whether there is, or is not, a relationship with future low back pain remains uncertain.

However, many patients will present to physiotherapy with scans already. 

What we do recommend is that clinicians incorporate this information (when available) into their overall impression of a patient’s likely prognosis but we do not currently recommend imaging purely to gain this information.

What further research is needed in this area?

Future research should focus on standardising the reporting of MRI findings and clinical outcomes so that researchers can utilise the vast amount of data already available in the literature. 

It should also investigate the impact that age and psychosocial factors may have on associations between MRI findings and clinical outcomes. 

Lastly, we hope that this study presents ideas for future research to potentially define subgroups within the ‘non-specific low back pain’ label to better understand the role of diagnostics in low back pain and to develop new specific treatments.

>> Christopher Han APAM is a musculoskeletal physiotherapist currently working in a private outpatient service and a PhD student at the University of Sydney and the Institute for Musculoskeletal Health.