Imaging incidence and type in primary care patients with low back pain: a cross-sectional study on new referrals to an Australian specialist spinal surgical centre

Keywords: back pain, clinical guidelines, general practitioners, health financing, health funding, health policy, musculoskeletal disease, neurosurgeons, New Zealand health strategy, primary health care.

Low back pain (LBP) is a significant cause of morbidity in adults;^1,2 however, approximately 90% of cases are termed ‘non-specific’ in etiology³ and often resolve within a few weeks.⁴ Spinal imaging for LBP in primary care is therefore not recommended by guidelines⁵ unless, among other things, a serious cause of pathology is suspected,⁴ radicular pain has been present for 6–8 weeks or severe neurological deficits are present.⁶ Despite these recommendations, many patients with LBP undergo inappropriate imaging in primary care,^4,7,8 and many patients that should have imaging do not receive it.⁷ Inappropriate imaging causes clinical harm^4,9,10 and is a financial burden to the government and often also the patient.¹¹

A 2016 report on diagnostic imaging for LBP by the Medicare Benefits Schedule (MBS) Review Taskforce has attempted to improve patient primary care management by limiting use of CT and increasing access to subsidised MRI, where specific clinical indications are met.¹² In Australia, certain imaging modalities for LBP are subsidised only under specialist care,¹¹ and we have continued to note a high incidence of inappropriate spinal CT scanning in new patients with LBP referred to our Centre for specialist surgical opinion.¹³ Though data are available on imaging incidence^14,15 and conformity to guidelines⁷ in primary care, no such data are available from within the population group of patients referred for surgical opinion. Thus, an audit was undertaken that focused on radiological studies in the 6 months prior to the new patient’s referral. The primary objective of this study was to explore both the incidence and type of imaging used for these patients. The secondary objective was to determine if spinal imaging was warranted by comparing the patient’s presenting complaint and clinical features with guidelines for imaging in primary care.

This was a single-centre, cross-sectional observational study.

This study was set in an interdisciplinary, independent spinal centre in an Australian capital city.¹³

Included in this study were 100 sequential new patient referrals for LBP management from primary care to surgeons at the spinal centre between January and February 2022. Eligible referrals for this audit met the criteria outlined in Table 1.

Table 1.  Inclusion and exclusion criteria for patient referrals.

Variables included epidemiological data, duration of LBP, radiation of pain, presence of ‘red flags’ and all lumbar imaging performed in the prior 6 months to referral. All data were collected from patient referral documents. Diagnostic imaging was divided into three categories: lumbar plain radiograph (XR) and diagnostic spinal CT and MRI. Interventional lumbar imaging was also documented. These variables enabled patients to be separated into 3 subgroups that corresponded to local guidelines (The South Australia Health Spinal Imaging Recommendations; SA Health guidelines).⁶ Table 2 describes these subgroups and lists all ‘red flag’ symptoms according to these guidelines.

Table 2.  Three patient subgroups and subgroup criteria based on imaging recommendations, as per SA Health guidelines.⁶

This study used simple descriptive data analysis with calculations of 95% confidence intervals (95% CI).

This study is classified as negligible risk research according to the guidelines of the NHMRC (National Health and Medical Research Council)¹⁶ and the University of Adelaide Human Research Ethics Committee (see Supplementary File S1). All patients gave written permission for their medical records to be used in clinical research and an audit.

To obtain 100 eligible referrals, 227 consecutive new patient referrals were needed (109 were for complaints other than LBP; seven patients had complaints of both LBP and thoracic or cervical pain; 11 had no comment on the duration of symptoms). The median age of patients was 55 years, 60% were male and 40% female.

Eighty-six percent of patients (95% CI 78–92%) received diagnostic lumbar spinal imaging, and four others had imaging as part of an interventional procedure for an overall imaging incidence of 90% (95% CI 83–95%). Many patients (36% : 95% CI 27–46%) underwent two or more different imaging modalities. The frequency of different types of radiological imaging, and their permutations, are provided as Supplementary Table S1. Fifty-two percent of the cohort (95% CI 42–62%) underwent diagnostic CT, 42% (95% CI 33–52%) underwent MRI and 28% (95% CI 20–37%) also underwent CT-guided interventional spinal procedures (nine ‘epidural injections’; seven ‘facet joint injections’; nine ‘nerve root injections’ and three ‘rhizolysis procedures’). The different types of imaging and their frequencies performed in the three subgroups, according to guideline indications for imaging, are summarised in Table 3.

Table 3.  Frequency and types of radiological imaging in the three different patient subgroups, together with 95% CI (N = 100).

Sixty-two percent of patients were appropriately imaged or appropriately not imaged. Thirty-three percent of patients were inappropriately imaged (33 of 42 in subgroup 3) and 5% inappropriately not imaged (four in subgroup 1 and one in subgroup 2). Table 4 summarises the proportions of patients who were investigated appropriately according to the imaging guidelines.⁶

Table 4.  Frequency of appropriateness and inappropriateness of radiological investigations according to cohort, as defined by the imaging guidelines.

The incidence of diagnostic radiology for new patients referred from primary care for a specialist spinal surgical opinion was 86% (95% CI 78–92%). The high, but appropriate, incidences of diagnostic radiology in subgroups 1 (88%) and 2 (96%) revealed that, in these clinical categories, guidelines are being followed for imaging, but often not for the recommended modality because CT imaging was more common than MRI scanning. Patients without guideline indications for spinal imaging (subgroup 3) are commonly inappropriately investigated (79%), again with the overuse of CT. Though it was noted that rates of inappropriate over-imaging were high and the rates of inappropriate ‘non-imaging’ were low, direct comparison of these data with datasets from primary care – where both the incidence of imaging and use CT and MRI are respectively around 3 and 8-fold lower^7,8 – is not appropriate because of the specific nature of our cohort. Referred patients, which accounted for only 5.2% of 846 patient with LBP in a prospective French study,¹⁷ incur higher imaging rates for reasons such as patient demand,¹⁸ GP diagnostic uncertainty and desire to streamline and expedite the referral process.¹⁹ Additionally, whether imaging patterns vary according to patient referral to specific medical or surgical sub-specialty is unknown.

The relative high incidence of CT use compared to MRI was not an unexpected finding given it was one of the main reasons for the audit. However, it is disconcerting in view of the better tissue imaging with MRI, and absence of any radiation exposure,^3,20 the guideline recommendations⁶ and recent statements released by the Australian Commission on Safety and Quality in Healthcare (ACSQHC),²¹ recommending MRI as the modality of choice when imaging is indicated for LBP.²¹ The continued use of spinal CT in preference to MRI is perhaps an unwarranted legacy of the MBS rebate system in Australia, which subsidises general practitioner (GP) lumbar CT requests, but not lumbar MRI requests,¹¹ and has remained unchanged in over a decade.²² The findings of this pilot study support the MBS Review Taskforce’s recommendations regarding MRI and CT subsidy changes:¹² improving primary care access to MRI subsidy when guidelines recommend its use, while restricting access to CT where clinical criteria are not met.¹² Considering the ACSQHC’s recent statements on best practice imaging for LBP²¹ and this audit’s findings, the necessity for change to these subsidies is becoming more cogent.

In New Zealand, the government’s Accident Compensation Corporation (ACC) has introduced operational guidelines for the subsidised use of GP referred MRI, in the context of presentations for knee, lumbar or cervical spine injury.^23,24 The ACC’s strategy of increasing GP access to subsidised MRI within the framework of increased GP training may act as a useful template for a similar process to be adopted in Australia.²³ Given recent statements by the ACC, which extol the effectiveness of this program in reducing MRI wait times and improving patient care,²⁵ it would be prudent to consider the development and pilot of a similar training and imaging access program in Australia.

Some limitations of this preliminary study include the assumption that new referrals constitute a new case of LBP and not a recurrence of such pain. To comprehensively define the cohort and their past medical histories would require interrogation of additional data sources; however, previous attempts at such an investigational approach have revealed vital data remains missing.²⁶ The size of the cohort selected for this preliminary study resulted in small subgroup sizes (post-hoc analysis suggests that increasing the sample size to 228 patients per subgroup would enable detection of statistically significant differences); however, the pilot study has revealed some useful data and trends in radiological investigations and practice in primary care. One unexpected finding was that 28 patients had undergone interventional spinal therapeutic procedures using image guidance. Previous studies on LBP imaging practices in primary care lack data on the use of interventional radiology.^7,8 The indications for, outcomes of and relative costs of such practice is worthy of further study.

Although this was a study at a single centre over a limited time period, the centre receives 2200 new patient referrals annually from many primary care providers across the state, and therefore is representative of current imaging practices in predominantly privately insured patients. The latter fact may limit the generalisability of these findings because whether non-insured patients undergo the same imaging pathway in primary care, before referral to public, healthcare spinal specialists, is unknown. Although there are many different national and international clinical imaging guidelines for LBP, they are now substantially similar,²⁷ so this variable is unlikely to be a source of major variance in future studies in this area.

This pilot study of patterns of lumbar spinal imaging in patients with LBP referred to an Australian spinal surgical specialist centre found most referring GPs conformed to guideline recommendations when imaging was indicated, but they over investigated patients in whom guidelines recommended no imaging. The preference for CT over MRI imaging and frequency of interventional image-guided spinal procedures prior to referral both warrant further study to determine the causes of these practices.

The data that support this study cannot be publicly shared due to ethical or privacy reasons, but may be shared, if appropriate, upon reasonable request to the corresponding author.

The authors declare no conflicts of interest.

Any publication fees associated with this work are supported by an Educational Grant from Radiology SA, of up to $1500 AUD.

Supplementary material is available online.