Empirical evaluation of the accuracy of diagnostic criteria for Restless Legs Syndrome.

BACKGROUND

In accordance with the diagnostic criteria of the International Restless Legs Syndrome (RLS) Study Group, the diagnosis of RLS is exclusively based on subjective information. Patients must report an urge to move the legs (and arms) with or without unpleasant sensations which is engendered by rest, relieved by movement, and worse in the evening or at night than during the day (essential criteria). Objective information such as excessive periodic leg movements, positive response to dopaminergic medication, family history of RLS or findings of a neurological examination cannot substitute any of the essential criteria but are considered both supportive for the RLS diagnosis and important for decisions on differential diagnoses. In this article, we report a systematic empirical analysis of the accuracy of diagnostic decisions based on all diagnostic criteria being either "essential" or "non-essential."

METHODS

We re-analyzed data from a validation study for the RLS Diagnostic Index, in which ten items related to diagnostic criteria were compared with an expert diagnosis of RLS (n = 86) and other sleep-related diagnoses (n = 93). The value of individual diagnostic criteria and features of RLS predictive of the expert diagnoses were analyzed with logistic regression models by increasing the set of diagnostic criteria stepwise based on delta-chi(2) tests.

RESULTS

The essential diagnostic criteria share a large amount of variance in the prediction of the expert diagnosis. Three of the four essential criteria (urge to move the legs, which is engendered by rest and worse in the evening) almost completely determine the expert diagnosis. However, adding response to dopaminergic medication to the set of the essential criteria increased the effect size in the logistic regression model from 69.4% to 88.4%, indicating a relevant improvement of the accuracy in supporting or excluding the diagnosis of RLS.

CONCLUSION

On the basis of our exploratory empirical analysis we conclude that the accuracy of diagnostic decision making in patients suspicious for RLS can be improved by inclusion of objective information. Response to dopaminergics is the most important criterion which both supports the "true" diagnosis of RLS, if present, and is also relevant to exclude mimics of RLS.