Assessing sleep problems and daytime functioning: a translation, adaption, and validation of the Athens Insomnia Scale for non-clinical application (AIS-NCA)

Abstract

Objective & Design: Sleep problems are common and have been linked to health problems, diminished well-being, and impaired performance. Many scales to diagnose clinically relevant sleep problems are time-consuming, complex, and difficult to administer in non-clinical and multi-thematic studies. Through a multi-stage translation (from English to German) and scale testing process, we developed a parsimonious measure of sleep problems and daytime functioning for non-clinical applications based on the Athens Insomnia Scale. Results: Exploratory (N_{Study 1} = 25,140) and confirmatory (N_{Study 2} = 14,797) factor analyses suggest a two-dimensional structure with the subscales “sleep problems” and “daytime functioning”. Internal scale consistency was acceptable. Measurement invariance was found across time, gender, age, and diagnosed sleep disorders. The scale discriminates between people with and without sleep disorders and predicts emerging sleep disorders. Short-term retest reliability was acceptable (N_{Study 3} = 78). Convergent validity with other sleep measures and discriminant validity with indicators of well-being were observed (N_{Study 4} = 341). After a multi-stage translation to English, we confirmed the factor structure and found measurement invariance across languages (N_{Study 5} = 623). Conclusion: Our short 7-item scale has good psychometric properties and is suitable for self-administration, making it useful in measuring sleep problems and daytime functioning efficiently and reliably, especially for large population studies.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by the German Research Foundation [SA 2992/2-1; to Sebastian Sattler]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The views expressed do not necessarily reflect the policies of the funder. The authors declare that there is no conflict of interest.

Notes

1 The abbreviation ADM stands for "Arbeitskreis Deutscher Markt- und Sozialforschungsinstitute e.V." (Association of German Market and Social Research Institutes). Within this association, there is a group of market research agencies in Germany that is responsible for the sampling frame to member agencies. This is a sampling frame used for representatively selecting telephone samples of the population living in private households (Gäbler & Häder 1998; ADM, 2012).

2 The sample is based on an initial sample (37,003 invitees, 24,085 consenting individuals, and 22,024 completers) with an extended sample (10,403 invitees, 3,064 consenting individuals, and 2,785 completers) to counteract demographic imbalances (due to selective participation of more difficult-to-reach participants), and thus, to increase representation.

3 In order to ease interpretation and to enable a comparison to the standard normal distribution, we adjusted the kurtosis by subtracting the value 3, meaning that any value other than zero implies excess kurtosis.

4 We ran a logit regression model to test if sleep problems resulted in selective dropout and found no such effect (OR = 1.027, p = 0.213).

5 Respondents who indicated a diagnosed sleep disorder at t₁ were excluded from the analysis, since only new diagnoses were of interest.

6 In addition, we tested this model including the omitted item 4 (in this case the analytic sample decreased to N = 14,792). Modification indices pointed toward a cross-loading between item 4 and the second factor (impaired daytime functioning). When we included this cross-loading, standardized factor loadings of item 4 were 0.382 on the first factor and 0.311 on the second factor, which confirms our decision to omit item 4 from the scale.

7 Please note that more than half of the participants in Study 3 were females, which may limit the generalizability of the results. We suggest that future studies use samples that are more representative.

8 We ran a logit regression model to test if sleep problems resulted in selective dropout and found a tendency towards a lower probability of re-participation with increasing sleep problems (OR = 0.93). This effect failed to reach conventional levels of statistical significance (p = 0.890).

9 The provider takes care of data quality by an elaborate scoring and control process of the panel members.

Additional information

Funding

This work was supported by the German Research Foundation [SA 2992/2-1; to Sebastian Sattler]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The views expressed do not necessarily reflect the policies of the funder. The authors declare that there is no conflict of interest.