Many sleep scales have been “validated,” a term that people automatically equals to ready for clinical use without questioning anything else. However, the fact that a sleep scale has been validated simply says that it asks about things related to sleep problems (face validity), with more or less appropriate coverage of the sleep-related problems (content validity), more or less appropriate relationship with other aspects of the disease (construct validity), and with the items composing the scale having some internal consistency [28]. These conditions, however, are apparently not very difficult to achieve and other possible scales could be made with different questions and very likely be successfully validated. Validation does not mean that all the questions of the scale have common sense or that they are all necessary. Two examples of this – the PDSS and the SCOPA-sleep scales – are illustrative.

The PSDSS-1 and PDSS-2 have both been validated and yet have important differences in their scoring system (a 10-cm visual analog scale from 0 = severe and always present to 10 = not present, vs a numeric score) as well as in the items asked (items 5, 9 and 15 of PDSS-1 are no longer present in PDSS-2 and items 4, 10 and 11 have different wording). In addition, there are items whose scores were not significantly different in patients and controls (items 2, 1, 14) and a few items are overrepresented, like items 4, 5, and 10, in which a patient with RLS will very likely answer yes three times for the same problem. A lower number of items would perhaps give similar relevant information and have similar clinimetric properties than the 15-item scale (see Table 5.2). In fact, it is unclear why the authors decided to make a 15-item scale.

The SCOPA-sleep scale is another validated scale, with repetitive questions. For instance in the NS part, a positive response to item 2 (have woken too often), item 3 (lying awake too long at night), or item 4 (have woken up too early) implies very likely a positive response to item 5 (had too little sleep at night) and the same happens in the Daytime sleepiness (DS) part where a positive answer to items 1, 2, 3, 4 implies very likely a positive answer to item 5. These repetitions will artificially increase the difference between patients and controls. Finally, the construct validity of the SCOPA-sleep was proved by showing a significant correlation with the ESS and the PSQI. However, items 2, 3, and 4 in the DS-sleep part are very similar to items 1, 2, 6 and probably 3 in the ESS, whereas items 1 and 3 of the NS-sleep part are very similar to items 5a and 5b in the PSQI. It is not rare then that SCOPA-sleep correlated well with ESS and PSQI.

Two other areas for improvement can be found. One is that in most sleep scales the questions asked have not been extracted directly from the patients’ own way of describing their complaints – that is from how the patients describe their own problems – rather from how their doctors (ESS, PDSS, SCOPA) interpret it. An extreme case is the SCOPA, where the questions were chosen from “the literature,” and the authors in fact did not see directly any of the patients or controls that participated in the study, but only the written responses to their questionnaire.

The other is that all sleep scales have assessed their construct validity by calculating the correlation between the scores on the scale with those in other scales that have addressed similar constructs, but not with objective sleep recordings. Unfortunately, the older scales – such as the ESS – had limited or controversial correlations with objective tests [1].

Using scales has advantages and limitations. Advantages are the homogeneity of the questions asked, which is ideal for multicenter studies or to always record the same type of information. It may also help some patients to be aware of a sleep problem that they did not consider relevant enough to tell their doctor. Limitations are inherent to the methodology: homogeneity results in lack of flexibility in evaluating symptoms that do not perfectly fit with the questions. They may also induce an error, which is relying uniquely in a scale to diagnose or measure a sleep problem.