Our study investigated the rates of all spinal x-ray, CT, and MRI examinations amongst adults over the previous two decades in Ontario (2002 to 2019) and Manitoba (2001 to 2011), Canada. We found a 27% increase in the rate of spine imaging utilization in Ontario, mainly driven by a greater than four-fold increase in the rate of MR spinal imaging. Over 18.2 million spinal x-ray, CT and MR imaging examinations were performed in Ontario between 2002 and 2019, costing the province over $1 billion dollars (CAD). Costs attributable to spinal imaging rose by 54% during this time, driven primarily by increased use of MRI. Similar trends were seen in Manitoba, where the rate of spine MRI utilization increased 4.5-fold and constituted 40% of total spine imaging costs provincewide as of 2010/11.
In contrast with MRI, we found that overall rates of spine x-ray and CT imaging declined or remained stable in Ontario and Manitoba; however, Ontarians over the age of 75 demonstrated increasing rates of x-ray and CT utilization over the 2002 to 2019 study period. Reduced bone mineral density with increasing age, as well as higher risk of accidental falls, raises the fracture risk in elderly patients and may explain the higher rate of first-line x-ray imaging amongst older patients [23, 24]. We found a pronounced reduction in the utilization of spinal x-rays and CTs in Ontario in 2012, and this may be related to an amendment to the Schedule of Benefits for Physician Services [25] that limited the eligibility of lumbar spine imaging payment to cases of low back pain with suspected or known pathology [26].
Our observation of increased MRI utilization compared with relatively steady or decreasing rates of spine x-rays and CT imaging over time in Ontario and Manitoba likely reflects MRI’s replacement of CT as the “gold standard” imaging modality for spine-related complaints, such as spinal cord and nerve root pathologies [27]. Results of a 2014 survey of Canadian spine surgeons [7] found that 78% of respondents required imaging studies to accompany all physician spine-related referrals, most commonly MRI (48%), possibly further contributing to this trend. Of note, a prior study has found that most MRIs requested for surgical referrals are unnecessary [28]. An awareness among physicians and patients of the risks associated with ionizing radiation exposure of x-rays and CT imaging may have also played a role in the trends we observed [27].
Rising use of imaging for spine-related complaints is incongruent with prior [29] and current imaging guidelines and recommendations [8–12, 15, 16], many of which were already in place at some point during the 2002/19 (Ontario) and 2001/11 (Manitoba) time periods investigated. The prevalence of inappropriate spine imaging has been well-documented, with previous reports indicating guideline-discordant imaging in up to 54% of cases involving the cervical spine and 80% for the lumbar spine [1]. A prospective analysis of outpatient MRI requisitions at two Canadian hospitals [14] found that in 56% of patients referred for lumbar spine MRI, the imaging study was deemed inappropriate or of uncertain value. Similarly, a 2018 systematic review and meta-analysis of 33 studies concluded that lumbar spine x-ray, CT or MR imaging was inappropriately performed 44% of the time in patients presenting for care, as judged by duration of episode (28%; 95% CI, 21–35%), absence of red flags (9%; 95% CI, 7–11%), or lack of clinical suspicion of pathology (7%; 95% CI, 2–23%) [10]. Previous research has also demonstrated poor correlation between lumbar spine diagnostic imaging findings, clinical symptoms [30] and healthcare utilization [31]. Patients’ expectations and physicians’ fear of litigation have been identified as common reasons for inappropriate spine imaging referrals [14].
Excessive or inappropriate utilization of diagnostic imaging for spine-related complaints in hospital-based and primary care settings may be related to poor implementation of accepted criteria for appropriate spine imaging use [2, 32]. For example, acquiescing to a patient’s request for imaging is more efficient than explaining why a given test is either of uncertain value or inappropriate. Our results and those of other studies [13, 14, 28] suggest that ongoing overutilization of spine-related imaging in Canada and elsewhere [10] poses a significant burden on healthcare systems, and limits timely access to diagnostic imaging for patients most in need [14, 33]. Accordingly, there is a need for implementation and enforcement of clinical decision supports [34] and spine care pathways such as the Rapid Access Clinics for Low Back Pain program in Ontario, Canada [35], where a 31% reduction in MRI referrals was demonstrated within the first three years of its inception [17]. Unnecessary imaging also poses risks to patients, including false positive tests leading to psychological stress and unnecessary invasive procedures, as well as the risk with x-rays and CT scans of exposure to ionizing radiation [28, 32]. Understanding current utilization trends may help guide future health system interventions aimed at improving appropriateness of spine imaging [1, 17].
Limitations
This study had several limitations. Due to the lack of clinical findings and outcomes, we were unable to characterize rates of appropriate imaging use; however, several prior studies [13, 14, 28] have found that most spinal MRIs are unnecessary. Second, our cost analyses of imaging are underestimated as technical fees/operating budgets could not be included. Further, most data available for our study was from ICES, which only provides healthcare utilization for Ontario and may limit the applicability of results in other jurisdictions. We included data from Manitoba, available to the year 2010/11, to increase the generalizability of our findings. We relied on historical data and were unable to examine imaging rates or appropriateness in different clinical settings (e.g., outpatient vs. Emergency Department) or between different referring physician groups (e.g., family doctors, radiologists, nurse practitioners, chiropractors, or physiotherapists) because this information was not available in the databases utilized. The inability of the accessed databases to divide CT and MR spinal imaging into cervical, thoracic and lumbar levels is also a limitation. We also did not control for additional confounding factors (e.g., operating room access), or include a comparison group to determine if the rate of spine MRI increased at a higher rate than other musculoskeletal or organ systems (e.g., knee, liver). These data would provide insight into whether the increased rates of spinal MRI utilization were related to increased MRI capacity (i.e., reduced wait times for appropriate imaging) versus inappropriate spinal imaging, as implied.