Validation of the minimum question set for diagnosis of the restless legs syndrome in epidemiology study in the population of Czech pregnant women

Abstract

Background: The aim of this study was to estimate diagnostic accuracy of the minimum question set used to estimate the Restless Legs Syndrome (RLS) prevalence using self administered questionnaire. The original English version was published in 2003 by Richard P. Allen et al. in Sleep Medicine. The data set was constructed along with the diagnostic criteria and was indented to be used in epidemiology studies. The minimum set consists of three questions to diagnose RLS and the forth question estimating frequency of the symptoms. The questions were translated to Czech language and blindly reversely translated to assure no shift in the original meaning. Pregnant women were asked to fill the questionnaire between 34–38 week of pregnancy at the Department of Obstetrics of the First Medical Faculty, Charles University in Prague. Subsequently and independently, the women were contacted over telephone by an sleep physician trained in diagnosing RLS. The results of both methods were compared to estimate the diagnostic profile of the minimum question set. Any women, answering first question as ”no” were set to answer all three questions as negative. In total 187 questionnaires were validated, in 57 the personal interview confirmed the diagnosis of RLS. When the patient answered all three questions positively, in twelve cases RLS was not confirmed by the telephone interview. Thus positivity of all three questions yields specificity of 90.8%, but only 79.0% both sensitivity and positive predictive value. When a patient answered “no” to the first question, then RLS was found to be present in only three women. Thus positivity of only the first question has sensitivity 94.7% and negative predictive value 97.4%. Diagnosing RLS using self-administered questionnaire is a fast and cheap method, representing a major advantage when screening larger population. Our study validated Czech version of the questions and found high total diagnostic accuracy, even when using only the first question, with highest sensitivity, thus being very usefull for screening. Adding the other two questions led to increase in specificity, but at the costs of losing some patients. Such approach is encouraged e.g. in genetic studies, where including a false positive patient may lead to distortion of the result. Supported by grant IGA-NT 12141–3 and MSM 0021620849.