Assessing Quality of Sleep in Patients with Rheumatoid Arthritis

PREPARING FOR THE PATIENT PERSPECTIVE WORKSHOP ON SLEEP

In preparing for OMERACT 9, the working group met periodically by teleconference and E-mail in addition to in-person meetings at the American College of Rheumatology and European League Against Rheumatism conferences in 2007. A systematic literature review of instruments designed to assess various aspects of sleep was conducted in January 2007, and during 2007 these instruments were evaluated on their response characteristics, psychometric properties, and domains of sleep assessed. The deliverables for OMERACT 9 were to present the results of the systematic literature review on sleep instruments and their truth and feasibility of use in RA. The objective for OMERACT 9 was to select candidate instruments based on truth, discrimination, and feasibility that measure sleep domains of interest.

Systematic review of sleep instruments

In conducting the systematic review the methodology of the Cochrane Collaboration was adhered to and the following steps were undertaken: a comprehensive literature search was conducted (keywords and MeSH terms: sleep, insomnia, sleep disorders, questionnaires, interviews, health surveys, psychometrics, health status, quality of life); citations and articles were selected using predefined criteria by 2 independent reviewers; information on the instruments was extracted from the articles using 2 independent reviewers; characteristics of the instruments were summarized including format properties, number of items, response format, timeline, and psychometric properties (reliability, validity, responsiveness). The literature search included: Medline (1966 to January 2007), PsychINFO (1806 to January 2007), Web-based databases (MAPI Research Institute and Educational Testing Service Test Collection), sleep assessment textbook chapters, bibliographies of sleep research, and review articles. Self-report instruments designed to assess sleep and sleep disorders in adults were selected. Instruments developed to measure sleep disruption secondary to other medical conditions (e.g., Parkinson’s disease, sleep apnea) were excluded, with the exception of chronic pain.

The search resulted in 3751 citations from Medline (1966 to January 2007) and 174 citations from PsychINFO (1806 to January 2007). After applying the selection criteria, 45 instruments were identified that assessed a variety of domains related to sleep (Table 1)^15–64. In particular, the domains related to sleep that were assessed in the sleep instruments were identified and summarized, their applicability to chronic diseases, and in particular RA, were evaluated, and the psychometric properties and feasibility aspects of the instruments were reviewed.

The various domains related to sleep that were assessed in the sleep instruments identified in the systematic review were itemized and summarized (Table 2). Fourteen domains were identified and presented at the EULAR 2007 conference. At a meeting of the working group at EULAR 2007, the applicability of these domains to chronic diseases, and in particular RA, was evaluated and confirmed. Also, the response characteristics and psychometric properties of the instruments were identified and summarized and, in preparation for OMERACT 9, a Delphi process reduced the number of instruments for consideration at OMERACT to 15 instruments. Selection of the instruments followed a similar process that would be used at the OMERACT meeting. The response characteristics of the instruments are summarized in Table 3. The number of items typically ranged from 1 to 30 items, the response format was usually a Likert scale (4 or 5 point) or visual analog scale, and the timeline ranged from “recent” to 3 months. Most of the instruments were multidomain, and a summary of their psychometric properties based on the primary report of the instrument is provided in Appendix A.

BREAKOUT GROUP SESSIONS FOR THE PATIENT PERSPECTIVE WORKSHOP ON SLEEP

At OMERACT 9, the session on sleep was part of the Patient Perspective Workshop that was designed to consider: a Patient Core Set, Sleep, Effective Consumer, and Psychological and Educational Interventions. At the plenary session for the Patient Perspective Workshop, the preparatory work was described and the tasks of 3 breakout groups for sleep were outlined. Each of the breakout groups considered different aspects of sleep: sleep domains, feasibility, and psychometric properties. The rapporteur for each breakout group reported back to the Patient Perspective Workshop on the deliberations of their group. They described the process and any key points raised during the breakout session and provided a summary of the rankings and scorings.

Sleep domains

For this breakout group, a deck of 14 cards was given to each participant. On each card was the identification of a domain related to sleep and a brief description (Table 2), and the participant was to reorder the cards from the most important to the least important domain based on their opinion. Although the domain descriptions were self-explanatory and were in lay language, if needed, the Chair of the breakout group could briefly review the domains. Once the task was completed, each participant returned the card deck ordered from the most important to the least important domain. In reporting back to the Patient Perspective Workshop the following were the 4 highest ranked domains: 1. Sleep Adequacy; 2. Sleep Maintenance; 3. Sleep Initiation; 4. Daytime Functioning.

Feasibility

A package of 15 sheets was given to each participant. On each sheet the identification of the instrument and a summary of the format of the instrument were provided. In addition there was a description of the instrument taken from the primary publication, which could vary from the instrument itself to a listing of the items in the instrument to a simple text description. If needed the Chair of the breakout group could review the “feasibility” component of OMERACT filter of “Truth, Discrimination and Feasibility.” The participant was to reorder the sheets from the most feasible to the least feasible to use based on their opinion. In reporting back to the Patient Perspective Workshop the following were the 4 highest ranked sleep instruments based on feasibility: Athens Insomnia Scale, 2.3; MOS Sleep Measure, 4.0; Insomnia Severity Index, 4.9; Women’s Health Insomnia Rating Scale, 5.5.

Psychometric properties

A package of 15 sheets was given to each participant. On each sheet the identification of the instrument and a summary of the reliability and validity results were provided. The statistics and the details varied by instrument but provided psychometric results given in the primary publication of the instrument. Given the difficulty of the task in evaluating some of the statistical methodology and descriptions of the psychometric properties, the breakout group was divided into 3 subgroups, and each subgroup reviewed 5 instruments. After their review, each subgroup provided opinions on the instruments they reviewed, and the breakout group chairs coordinated a discussion among all the breakout group participants and reached an accord on the scoring of the instruments (with a high score indicating good psychometric properties). The 4 highest scored instruments were: Athens Insomnia Scale, 13.6; Sleep Assessment Questionnaire, 13.0; Pittsburgh Sleep Diary, 12.0; MOS Sleep Measure, 11.0.

CONSENSUS BASED ON SLEEP DOMAINS, FEASIBILITY AND PSYCHOMETRIC PROPERTIES

Based on results from the 3 breakout groups (sleep domains considered most important, highest ranked sleep instruments on feasibility, and the highest scored sleep instruments on psychometric properties), the 4 sleep instruments identified for further consideration were the Athens Insomnia Scale, the MOS Sleep Measure, the Pittsburgh Sleep Diary and the Women’s Health Insomnia Rating Scale (Table 4). Only the Athens Insomnia Scale assessed each of the top 4 sleep domains, and the MOS Sleep Measure and the Pittsburgh Sleep Diary assessed 3 of the domains and partially assessed the fourth domain of daytime functioning. Although the Pittsburgh Sleep Diary scored high on truth, it is difficult to complete and ranked low on feasibility. On the other hand, the Women’s Health Insomnia Rating Scale was easy to complete and so ranked high on feasibility but did not score high on truth. Both the Athens Insomnia Scale and the MOS Sleep Measure scored high on truth and ranked high on feasibility.

In summary, sleep instruments have been evaluated on the domains assessed, feasibility, and psychometric properties. In terms of the OMERACT filter: truth relates to the domains assessed (content validity) and psychometric properties of validity and reliability; feasibility is directly related to administrative burden and applicability; and discrimination relates to the psychometric property of sensitivity or responsiveness.

A number of domains related to sleep have been reviewed, and several sleep instruments have been identified that may be applicable to RA patients, namely: Athens Insomnia Scale, the MOS Sleep Measure, the Pittsburgh Sleep Diary, and the Women’s Health Insomnia Rating Scale. To further evaluate the sleep instruments identified, they should be considered in planned clinical trials of RA patients to assess their applicability. To further establish acceptability and applicability of the domains and the specific instruments, a Delphi exercise involving RA patients to further understand sleep quality from their perspective should be performed.