Diagnostic accuracy of behavioral, activity, ferritin, and clinical indicators of restless legs syndrome

Lack of a valid diagnostic measure of restless legs syndrome (RLS) for persons with dementia, who do not have the cognitive ability to report complex symptoms, impedes RLS treatment and research in this population. The aim of this study was to determine the sensitivity and specificity of a combination of indicators for identifying RLS that could eventually be used to diagnose RLS in persons with dementia. 3-day, prospective instrument validation. Sleep laboratory. Cognitively intact, 107 with RLS, 105 without RLS. N/A. Serial 20-min observations with a new measure, the Behavioral Indicators Test-Restless Legs (BIT-RL); leg movements with 3 nights of the Periodic Activity Monitor-Restless Legs (PAM-RL); ferritin; sleep history; clinical data; polysomnography; Hopkins Telephone Diagnostic Interview of RLS Symptoms. The best-fitting diagnostic model for identifying RLS included previous history of iron deficiency (odds ratio [OR] 7.30), leg discomfort (OR 6.47), daytime fatigue (OR 6.15), difficulty falling asleep (OR 3.25), RLS family history (OR 2.60), BIT-RL (OR 1.49), and absence of diabetes (OR 0.27), with sensitivity 78%, specificity 79%, and 77% correctly classified. This model retained its predictive accuracy even with co-morbid sleep apnea. When compared to those without RLS, persons with RLS have observable behaviors, such as rubbing the legs, that differentiate them, but the behaviors have no circadian and activity-related variability. The final model of clinical and sleep historical data and observation for RLS behaviors using the BIT-RL had good diagnostic accuracy.