Consequences of the Use of Images Not Recommended in Non-Specific Low Back Pain

Low back pain is the second most common symptomatic reason for visits to the doctor’s office and the associated healthcare costs are increasing rapidly. One reason for this increase is the inappropriate use of advanced imaging , particularly magnetic resonance imaging (MRI), for uncomplicated nonspecific low back pain . Routine use of MRIs for low back pain is common, with 16% to 21% of low back pain patients on commercial health care plans and 12% of Medicare patients receiving an MRI. Clinical guidelines recommend that new episodes of nonspecific low back pain that are not complicated by red flag conditions should be treated with conservative therapy and that MRI is not indicated in the first 6 weeks of onset. However, studies have found that 26 to 44 percent of spinal MRIs do not agree with guidelines.

There is increasing evidence that the consequences of inadequate imaging for uncomplicated non-specific low back pain extend beyond the direct costs of an MRI. Possible downstream consequences include subsequent referrals and interventions performed as a result of imaging. Rates of lumbar spine procedures, including surgery, epidural steroid injections, and facet joint injections, are also increasing and are part of the higher costs associated with inadequate imaging.

There is no evidence that these additional procedures lead to better outcomes, and they may even result in more harm than good.

The downstream effects of early lumbar spine MRI on costs, procedures, and opioid use remain underexplored in the literature, while the effects on pain outcomes remain unstudied. This study fills these gaps, exploring the association between early MRI and surgery, opioid use, cost, and patient pain course.

Background

Contrary to guidelines, magnetic resonance imaging (MRI) is often ordered in the first 6 weeks of new episodes of uncomplicated nonspecific low back pain.

Aim

Determine the subsequent consequences of early images.

Design

Retrospective matched cohort study using electronic health record data from US Department of Veterans Affairs primary care clinics.

Participants

Patients seeking primary care for nonspecific low back pain without a red flag condition or low back pain encounter in the previous 6 months (N=405,965).

Exposure

MRI of the lumbar spine within 6 weeks of the initial primary care visit.

Main Measures

Covariates included patient demographics, health history in the past year, and baseline pain. Outcomes were lumbar surgery, prescription opioid use, acute healthcare costs, and last pain score recorded within 1 year prior to the index visit.

Key results

Early MRI was associated with more back surgery (1.48% vs. 0.12% in episodes without early MRI), greater use of prescription opioids (35.1% vs. 28.6%), a final score of higher pain (3.99 vs. 3.87) and higher acute care costs ($8082 vs. $5560), p < 0.001 for all comparisons.

Limitations

Reliance on data collected in normal clinical care and the potential for residual confounding despite the use of coarse exact match weights to adjust for baseline differences.

Conclusions

The association between early imaging and increased utilization was evident even in a setting largely unaffected by the incentives of fee-for-service care. The reduced cost of imaging is only part of the motivation to improve compliance with guidelines for the use of MRI. Early screening is associated with excess surgery, higher costs for other care, and worse outcomes, including potential harms from prescription opioids.

Discussion

This study found that an MRI of the lumbar spine provided early in episodes of nonspecific low back pain was associated with more surgeries, higher costs of care, greater use of prescription opioids, and worse pain at follow-up. Lumbar surgery was 13 times more likely in the early MRI group compared to those without an early scan (1.48% vs. 0.09%). This is consistent with other observational studies, where surgery was 5 to 20 times more likely among those who received an early MRI. However, the absolute rates of surgery in this study were much lower than in these other studies. In those studies, lumbar surgery was provided in 14% to 22% of those with early exploration and in 1% to 3% without early exploration. These studies had approximately 10 times the rate of lumbar surgery that we observed in the VA.

Despite additional surgeries and higher rates and doses of opioid prescriptions, recipients of early scans had worse pain at follow-up relative to the comparison group. This result was consistent with studies that found no health benefit from early MRI. This study found that acute care costs incurred in the follow-up period were 1.4 times higher in the early screening group ($8082 vs. $5560). This was less than in other studies, where the MRI group had 3 to 8 times the cost of the comparison group.

The finding that early MRI was associated with increased opioid use during the follow-up period appears to be a unique contribution of this study, which is especially important given concerns about the risks resulting from overprescribing opioids for the pain. We found only one clinical trial that considered this outcome and found no significant association between early MRI and prescription opioid use. 29 Other studies did not consider this result.

Early MRI was provided to 2.46% of this cohort. This is a lower rate than in other studies in which the denominator was MRIs rather than primary care visits for low back pain. It may also be lower than expected because VA providers are salaried and are not affected by the financial incentives present in fee-for-service settings. Self-referral drives some of the high use of LS-MRI in Medicare-sponsored patients, although federal statutes on self-referral may have reduced this practice. Additionally, VA providers may be less likely to practice “defensive medicine ,” since liability for malpractice in the VA system is the responsibility of the federal government.

This study used electronic health records to assemble the largest cohort used to estimate the downstream consequences of early MRI. The size of this cohort (N = 405,965) was more than 10 times the size of other observational studies and more than 400 times the size of clinical trials addressing this topic. While large cohort studies have the advantage of collecting evidence from widespread real-world experience , they must address the issue of selection bias : patients were not randomized to receive early screening. Older methods, including simple covariate adjustment and the inclusion of a propensity score as a covariate, have now been replaced by more advanced matching methods.