For epidemiological studies, screening instruments used to diagnose restless legs syndrome in the absence of clinical interviews aren’t cutting it, experts say.
By Risa Kerslake, BSN-RN
An international team led by Stephany Fulda, PhD, a sleep researcher and clinician at the sleep medicine unit of the Neurocenter of Southern Switzerland, has carried out a large systematic review and meta-analysis to assess the accuracy of screening instruments in their ability to diagnose restless legs syndrome (RLS) in 52 studies from 2002 to 2018. The findings are published in Sleep Medicine Reviews.1
Only 14 studies of those reviewed used the reference standard, in this case, a clinical interview, with either all study participants or all screen positives and a small number of screen negatives. The analysis found a median sensitivity of 0.88 and a specificity of 0.90. Although these values sound good, assuming an RLS general population prevalence of 5%, the positive predictive value (PPV) was only 31%.
Concerns With RLS Prevalence Studies
Patients with RLS experience uncontrollable urges to move, particularly in the evening or at night, and report itching and crawling sensations, particularly in the legs. RLS has been reported to increase the risk of cardiovascular disease, hypertension, psychiatric illness, and death. However, Fulda cautions that studies reporting the increased risks all used RLS screening questionnaires to identify RLS cases.
John Winkelman, MD, PhD, who helped author the paper, has done numerous prevalence studies in RLS in several cohorts. “As much as I would like to trust the RLS diagnoses in those cohorts, I am concerned about the findings of this review, which suggest that some of those we thought had RLS really did not. These are chronic issues in RLS research, which is diagnosed by history, rather than by several ‘objective tests,’” says Winkelman, who is a professor of psychiatry at Harvard Medical School and chief of the sleep disorders clinical research program at Massachusetts General Hospital.
Since there is no biological test for RLS, patients typically receive a diagnosis based on a clinician evaluating their subjective descriptions of their symptoms. Noted in the review, most RLS screening instruments didn’t include questions that would differentiate true RLS from its mimics. Most epidemiological studies only include screening instruments, and the authors assume that the estimated prevalences in the studies that didn’t utilize clinical interviews had overestimated RLS prevalence.
Winkelman acknowledges the complexity of RLS research and the importance of specificity in epidemiological prevalence studies. He says, “Sensitivity is: were you able to identify everybody in the population who had RLS? Specificity means: of the people you identify as having RLS, how many actually did have it?”
Positive predictive value means the people whom you’ve identified with RLS actually have the condition, explains Winkelman. If the specificity and sensitivity are both 90% and the prevalence in the population is 5%, then the positive predictive value is only about 0.35—meaning only a 35% chance of being right.
Positive predictive value “is important because we’re taking five people who we said have restless legs and comparing them to 95 people who didn’t have RLS, and then we’re following them for years to see what happens to them. So those five people are really key,” says Winkelman. “If something bad happens, such as a heart attack, to one of those five, that’s a big deal for the conclusions of our study.”
Mimics Can Mask True RLS Prevalence
J. Andrew Berkowski, MD, a neurologist and sleep medicine specialist with ReLACS Health, has been concerned about the accuracy of RLS screening for years. “The treatments for RLS can have very severe side effects and consequences. If people think that they have the condition, they could end up doing themselves a great degree of harm by using treatments for RLS when they don’t have it.”
This is because the historic first-line treatments for RLS are dopamine agonists—medications that stimulate dopamine receptors and include ropinirole, pramipexole, and rotigotine. These medications have been shown to make the condition worse with time.
“It’s extremely important not to be misdiagnosed,” says Berkowski, stressing that there’s no downside in choosing not to treat RLS. “If someone were to prescribe these drugs, it could actually cause RLS to develop in somebody who didn’t already have it.”
Berkowski, who was not involved with the paper, didn’t realize his concerns with poor screening methods were attached to the low prevalence of the condition itself. The seemingly optimistic sensitivity and specificity are only good numbers when they’re applied to common conditions. “When you apply those numbers to a condition where it’s only 5% of the population, then it really makes the [positive predictive value] low. Now you have all these people—who maybe half of them don’t even have the condition, and you think they might—based on the screening tool.”
Mimics are a particular challenge for RLS screening questionnaires, says Fulda. RLS mimics are conditions that formerly fulfill the four core diagnostic criteria established by the International Restless Legs Syndrome Study Group for RLS: an urge to move the legs, symptoms are worse while resting, symptoms improve with movement, and symptoms are worse at night. A fifth criterion was later added to include ruling out the possibility of these mimics. Leg cramps, peripheral neuropathy, arthritis, positional discomfort, and pronounced, unconscious leg movements are the most common mimics.2
Fulda gives the example of nocturnal leg cramps. “If this patient is given an RLS screening questionnaire asking for the four core diagnostic criteria, this patient would faithfully and correctly answer all questions with ‘yes.” Even though they don’t have RLS but have an RLS-mimicking condition.
Clinician Interview As a Gold Standard?
The Sleep Medicine Review paper indicates a clinical interview is rarely feasible in epidemiologic studies. Nonetheless, the team concluded that for future epidemiologic studies, screening should follow a full or partial verification by clinician interview with someone who is familiar with the diagnostic criteria as well as RLS mimics.
Clinically, more time needs to be focused on RLS screening questions, particularly on the mimics, says Berkowski. While it’s understood that symptoms can be caused by another condition, clinicians aren’t asking questions about those other conditions. Berkowski adds, “They focus only on restless legs when there might be a need to focus more on what else it could be to make sure they’ve excluded those other possibilities.”
Winkelman addresses this diagnosis question in a commentary published in the Journal of Clinical Sleep Medicine titled, “Opioids and RLS: a double-edged sword.”3 As an example, a Mayo Clinic study in the same issue of the journal uses screening questionnaires to assess opioid withdrawal-related RLS, without including participant interviews.4
“It complicates all of our research, unfortunately,” Winkelman says, speaking generally of studies done using only screening instruments. “If you just do questionnaires you lose a lot of precision, whereas back-and-forth clinical interaction, even if it’s brief, can clarify things much better than a questionnaire can.”
But even a clinician interview, the so-called gold standard, isn’t fully accurate. “The gold standard is changing from each of these studies,” notes Berkowski, referencing the review. “You’re not doing a blood test and comparing it to something such as an MRI scan. The clinicians in one study might be much different than the clinicians in another study.”
Fulda is hopeful that one day there will be a more accurate screening tool for RLS. “Technically this will be very feasible with all the personal data that is theoretically available,” she says. “The success, however, will depend on a considerable research effort that generates a large gold-standard data set with RLS diagnoses made by RLS experts.” Any testing, such as polysomnography or wearables, needs to empirically show it can identify RLS cases with high accuracy but also reliably reject non-RLS cases, according to Fulda. Only a few studies have attempted this, and none of them have been independently confirmed for methods other than questionnaires and clinical interviews for RLS diagnosis, she says.
Winkelman isn’t so sure about a diagnostic tool identifying “an urge to move the legs” based on subjective experience. “How do you have an objective test for an urge?” he asks. He acknowledges RLS can have biomarkers, things highly associated with the condition, but he’s not sure medicine will get to the point where there are highly sensitive and specific biomarkers.
There may not be true objective tests for RLS, says Berkowski, and supporting tests such as iron studies or genetic testing can help support an RLS diagnosis. “But nothing is really going to confirm the condition other than clinical judgment with these [screening] questions,” he says. Nothing, he adds, is going to say for sure a patient has RLS even if they meet all the criteria.
1. Fulda S, Allen RP, Earley CJ, et al. We need to do better: A systematic review and meta-analysis of diagnostic test accuracy of restless legs syndrome screening instruments. Sleep Med Rev. 2021 Aug;58:101461.
2. Hening WA, Allen RP, Washburn M, et al The four diagnostic criteria for Restless Legs Syndrome are unable to exclude confounding conditions (“mimics”). Sleep Med. 2009 Oct;10(9):976-81.
3. Winkelman JW. Opioids and RLS: a double-edged sword. J Clin Sleep Med. 2023 Feb 17. Epub ahead of print.
4. McCarter SJ, Labott JR, Mazumder MK, et al. Emergence of restless legs syndrome during opioid discontinuation. J Clin Sleep Med. 2023 Jan 24. Epub ahead of print.