Context

The study by Delitto and colleagues gives options to patients who otherwise may feel back surgery is inevitable. Spinal stenosis is common,1 yet there is no professional consensus on diagnostic criteria or indications for surgery. The label is frequently assigned in MRI reports and may quickly lead to surgery. However, arthritic changes in the lumbar spine are common even among asymptomatic individuals,2 so diagnosis based on the pattern and severity of a patient's symptoms.3 Surgery is an option, but treatment choice depends on a patient's knowledge, preferences and values.

Findings

The surgery and PT groups achieved similar relief of pain and improvement in physical function at 2 years following enrolment. More than half (57%) of the patients randomised to PT chose instead to have surgery within 10 weeks. These crossover patients actually received more intense PT than the patients randomised to the non-surgical arm (average of 7.8 vs 3.4 visits).

Commentary

This was a well-done and well-reported study. The investigators bypassed diagnostic uncertainty by only enrolling patients who had already consented for decompression surgery. High crossover rates make randomisation difficult to interpret, even after complex statistical adjustment.4 Because as-treated analysis can result in bias, the authors present ‘complier average causal effect analysis and inverse probability weighting’, but these analyses were indeterminate because of wide CI. The benefit observed in the PT group may have been largely due to surgery. Nearly half of the patients who consented for surgery did not need it. They achieved the same degree of improvement in pain and function as those who had surgery.

RCTs have limited generalisability. The six surgeons who participated in the study may not be representative of the entire spine surgeon community. Patients who enrolled may not be representative of the broader community of patients who have symptomatic spinal stenosis. Recruiting 169 participants took 5 years. Of the 5119 patients screened, only 481 (9%) were deemed eligible, of which 312 (65%) declined randomisation.

Implications for practice

Spinal stenosis rarely causes neurological loss, but this fear often drives patients to surgery. This study provides reassurance that many patients who currently may be scheduled for surgery actually still have effective alternatives. Natural history of spinal stenosis is fluctuating symptoms with gradual progression. Definitive treatment is laminectomy.5 Lumbar decompression relieves leg symptoms but does not improve back pain.6 Epidural steroid injections are not effective.7 Surgery to increase the space between vertebrae is less invasive, but it only works short term.8 Lumbar fusion is more invasive and has a higher complication rate than decompression,9 without proven advantages.

Patients suffering from severe back problems are vulnerable to promises of quick relief and fear of paralysis. Rather than imaging-based diagnosis and treatment, what is needed is a data-driven discussion with patients about the severity of their symptoms, comorbidity and predicted benefits and risks. This is now achievable with Patient-reported Outcome Measures (PROMs) that quantify how a patient feels and functions. Benchmarked PRO scores can help engage patients by showing them how they compare to other patients with similar characteristics, such as age, sex, weight, height and comorbidity. An example of personalized information for knee replacement is shown at www.peerstudy.org. Similar outcomes prediction models based on benchmarked PRO scores would be helpful for patients with spinal stenosis. Discretionary spine surgery is an irreversible, life-changing event; patients need more time to understand personalised predictions than possible in a typical 10–20 min office visit with a spine surgeon.