Abstract
BACKGROUND: Modic changes (MC) have been proposed as a cause of low back pain (LBP). However, the proposition remains controversial. There is uncertainty over the existence or degree of association between the two and whether, if there is an association, it is a causal relationship. Previous systematic reviews of the evidence have had methodological flaws. AIMS: The aim was to synthesize the current evidence to test the hypothesis that there is an association between MC and LBP and if there is, to evaluate the strength of the association. METHODS: MEDLINE, EMBASE, and CINAHL were searched for all studies up to 31 December 2018 for cohort, case-control, and cross-sectional studies. Screening, quality assessment, and data extraction were conducted by two independent reviewers. Quality was assessed using the Joanna Biggs Institute tools for observational studies. The clinical heterogeneity among these studies ruled out pooling so a narrative review was undertaken. RESULTS: Fifteen studies met the inclusion criteria, varying in patient characteristics, characteristics of MC, coexisting spinal conditions, and outcomes. The quality of evidence was poor in six and moderate in seven. There was wide clinical heterogeneity amongst the studies. The inclusion ages varied from early teens to over 65s, pain duration varied from under 6 weeks to over three months, and characteristics of the MC chosen for the studies also differed. The results were inconsistent across the studies: the odds ratios varied from showing an inverse relationship [Kovacs] with an odds ratio 0.31 (95% confidence interval, 0.1-0.95) to a very strong positive association 121.4 (11.21-1315.08) [Nakamae]. There was no consistency in associations for: type of MC, lumbar levels affected, position in relation to the vertebra, and presence of co-existing spinal conditions. The associations were possibly spurious arising from potential biases suggested by incomplete reporting: publication bias, selective reporting, and post hoc analysis. Only one study at low risk of bias found a substantial association but it was a small study of a narrow group meaning its results may not be generalizable. DISCUSSION: The inconsistency of findings and the possibility that they were spurious means that no conclusions can be drawn about an association between MC and LBP. Future research should be designed as prospective cohort studies with adherence to reporting guidelines pertaining to observational studies and to MRI. Currently, clinicians should not look for the presence or absence of MC to guide their management of patients with LBP.