Upright versus recumbent lumbar spine MRI: do findings differ systematically, and which correlates better with pain?...

Recumbent MRI is the most widely used image modality in people with low back pain (LBP), however, it has been proposed that upright (standing) MRI has advantages over recumbent MRI because of its ability to assess the effects of being weight-bearing. It has been suggested that this produces systematic differences in MRI parameters and differences in the correlation between MRI parameters and pain or disability in patients thus, potentially adding clinically helpful information.

This paper aims to review and summarize the available empirical evidence for or against these 2 hypotheses.

Systematic review of the literature (PROSPERO ID: CRD42017048318). Studies should be based on paired observations of MRI findings in the upright and recumbent positions.

People aged 18 or older with or without low back pain ± radiculopathy. Studies needed a minimum of 15 participants.

All continuous, ordinal, and dichotomous parameters based on MRI images. All measures of pain or disability.

Studies assessing MRI parameters both in upright and recumbent positions on the same individuals measured on continuous, ordinal, or dichotomous scales were included. For each parameter, the expected direction of the difference between recumbent and upright position was specified as an increase, no change, or decrease. Information on the observed distribution of individual differences was extracted from included studies and subjected to meta-analyses if sufficient data was available. Observed differences were then compared with the prespecified expectations. Studies were also screened for information on correlations between patients' pain and/or disability and MRI parameters or differences between patient subgroups defined by patients' pain and/or disability.

About 19 studies were identified, including 5,082 participants with LBP (16 studies) and 166 participants without low back pain (5 studies). Twenty-five MRI parameters were measured on a continuous scale, ten parameters were assessed on an ordinal scale, and 15 parameters were reported as dichotomous data. The observed differences between recumbent and upright MRI were mostly consistent with the prespecified expectations. Correlations between patients' pain or disability level and MRI parameters were reported in only 1 study, and three studies reported comparisons of MRI parameters across subgroups of patients defined by pain or disability characteristics. Higher correlations or larger effect sizes when using the upright position were observed in most results reported.

For most MRI parameters, the direction of the observed difference between assessment in recumbent and upright positions aligned with the prespecified expectation implied by the weight-bearing position. This confirms the existence of a systematic difference between the 2 positions. Performing an MRI upright instead of recumbent position may increase the correlation with pain, but final evidence for this property is still missing. The clinical significance of upright MRI is still unclear, and there is a need to directly investigate the impact of MRI findings on clinical decision-making and patient outcomes.