The MOS-SS was included in a stand-alone, longitudinal study conducted between May and September 2007 designed to evaluate the psychometric properties of several outcomes assessment instruments in patients with FM. Patients were recruited through newspapers, support groups, and the Internet, and were compensated for participation. Individuals were included if they were ≥ 18 years old and provided a confirmed physician diagnosis of FM for at least 3 months prior to enrollment; the participants' clinicians forwarded written confirmation of the diagnosis of FM to the study investigators. Individuals were required to have a current pain rating of > 2 on an 11-point numerical rating scale (NRS) to enable enrollment of individuals with a broad range of pain severity, since another outcome of this study was validation of cutpoints representing moderate and severe pain as previously described 26. It is also likely that use of this rating resulted in a more representative population than limited by the score ≥ 4 generally required for inclusion in clinical trials of FM 27282930313233. The ability to read/understand English and cooperate with investigators and study procedures were also required. Exclusion criteria included a previous diagnosis of rheumatoid arthritis or systemic lupus erythematosus and/or any other chronic painful condition that could confound the ability to distinguish other chronic pain from pain related to FM. Eligible participants completed informed consent and study participation forms; protocol and study documents were approved by the appropriate Institutional Review Board.

The MOS-SS is a 12-item questionnaire that aims to evaluate key constructs of sleep, with derived subscales for the domains of sleep disturbance (4 items), quantity of sleep (1 item), snoring (1 item), awakening due to short of breath or with headache (1 item), sleep adequacy (2 items), and somnolence (3 items) 19. Additionally, a 9-item Sleep Problems Index can be generated which assesses overall sleep problems. It includes the 4 sleep disturbance and the 2 sleep adequacy items, 2 of the somnolence items, and awakening short of breath/headache; higher scores indicate greater sleep impairment, and this index is often used in clinical trials as an indication of sleep quality.

The MOS-SS was administered as a mailed questionnaire. All patients completed both the 4-week and 1-week recall period versions of the MOS-SS, with the 4-week recall period questionnaires administered first. The test and retest of the MOS-SS using the 4-week recall period were separated by an interval that ranged between 1 and 3 days, and for the 1-week recall period, the test and retest were separated by a 7-day interval. At the time of the retest, patients also evaluated their impression of change in sleep (Patient Impression of Change; PIC) for the period between the test and retest. The PIC was adapted from the Patient Global Impression of Change for the purpose of this study by specifying a change in sleep due to FM, but was based upon methodology widely used to assess degrees of change 34. As with the PGIC, it is answered on a 7-point scale of 1 = very much improved; 2 = much improved; 3 = minimally improved; 4 = no change; 5 = minimally worse; 6 = much worse; 7 = very much worse.

Paired t-tests were used to determine significance of the difference between test and retest scores. To evaluate test-retest reliability of the subscales, the intraclass correlation coefficient (ICC) using Shrout-Fleiss reliability (single-score statistic) was calculated from paired values for each recall period 35. A value greater than the conventionally accepted threshold of 0.70 was considered an indication of reliability 36. Pearson correlation coefficients were also calculated as confirmation of the reliability estimates. This assessment for reliability was performed on data from stable subjects with respect to the PIC, i.e. patients who reported "no change."

A total of 129 patients with FM were enrolled; 91.3% were female, and the mean age was 49.4 ± 11.0 years. Self-rated FM severity was at least moderate in 88.1% of patients, and 88.3% reported a duration of FM of at least 2 years since diagnosis. Approximately two-thirds of the patients (68.3%) reported taking medications for their FM.

The mean test and retest scores for the MOS-SS domains and the 9-item Sleep Problem Index, along with general US population norms that were derived in a validation study using the 4-week recall period 21, are shown in Figure 1. Scores were similar using the 4-week and 1-week recall periods, and generally showed little variation between the test and retest. The only significant differences between test and retest values were observed for the domains of Daytime Somnolence (p = 0.0062) and the 9-item Sleep Problems Index (p = 0.01) for the 4-week recall period.

All domain scores for the 4-week recall period showed substantial differences from population norms, indicating the presence of sleep impairment. Sleep Disturbance and Daytime Somnolence were the MOS-SS domains which had the greatest negative impact on sleep quality for both the 4-week and 1-week recall period. The Sleep Problems Index scores, which are shown in Table 1, ranged from 57.2 ± 14.5 to 61.9 ± 15.8 across all assessments, and were more than twice as high as reported for the general population norm (25.8) 21, demonstrating substantial sleep problems in these patients.

Although 60.4% and 56.9% of patients reported no change in sleep status on the PIC between the test and retest for the 4-week and 1-week recall periods, respectively, improvement and worsening were both reported by patients. For the 4-week recall period, 8.2% of patients reported improvement and 23.8% reported worse sleep. Similarly, for the 1-week recall period, 16.5% reported improved sleep and 26.7% reported worse sleep.

The test-retest reliability of the MOS-SS domains and the 9-item Sleep Problems Index was assessed for the 78 patients (60.4%) and 62 patients (48.1%) who reported no change on the PIC for the 4-week and 1-week recall periods, respectively. The ICC and Pearson correlations for these patients are shown in Table 2. Except for the Sleep Adequacy domain, the ICC required threshold of 0.70 was exceeded for both recall periods with slightly higher values for the 4-week period. For the Sleep Adequacy domain, although the ICC did not achieve the threshold value, the 1-week recall period resulted in a higher value (0.69) than the 4-week recall period (0.63). Values of the Pearson correlation coefficient were comparable to the ICC.

The MOS-SS is a validated instrument for evaluating the impact of disease on sleep 1921. Its utility has been further demonstrated in chronic pain conditions by characterization of clinically important differences in patients with neuropathic pain 22 and FM 24. In an effort to conform to recent recommendations for the use of patient-reported instruments with short recall periods 25, we evaluated the reliability of a 1-week recall period. The data reported here indicate that the MOS-SS produces comparable results regardless of the use of a 1-week or 4-week recall period. All domains and the 9-item Sleep Problems Index demonstrated adequate reliability that was similar for both recall periods when no change occurred in the underlying concept (e.g., sleep interference). Similar values suggesting high correlation were obtained for ICC and Pearson coefficients; ICC is considered a more conservative estimate of association than Pearson.

A limitation of this study is that these reliability estimates were based on patients who showed no change in sleep status between the test and retest. While the MOS-SS in general is sensitive to treatment effects, further evaluation of the 1-week recall period may be required under conditions characterized by a change in sleep disturbance (clinical trials and clinical practice). However, it should be noted that the psychometric evaluation of the MOS-SS in patients with FM utilized data from two clinical trials, one with a 4-week recall, and the other with a 1-week recall; both recall periods showed similar psychometric characteristics and sensitivity to change 24.

The fact that patients were compensated for participation is another limitation which may have introduced bias, since it is not known what effect the compensation may have had on the selection of patients for this study.

The previously demonstrated psychometric soundness of the MOS Sleep Scale subscales and overall Sleep Problems Index combined with the current demonstration of the reliability of a 1-week recall period suggests the appropriateness of this instrument in the evaluation of sleep disturbance in patients with FM. However, further corroboration of the reliability of a 1-week recall period in clinical trials in patients with FM may be warranted.

FM: fibromyalgia; MOS-SS: Medical Outcomes Study Sleep scale; ICC: intraclass correlation coefficient; PIC: Patient Impression of Change

Alesia Sadosky and Ellen Dukes are employees of Pfizer, Inc.; Chris Evans is an employee of Mapi Values, an outcomes research consulting company, which received funding from Pfizer to perform the analysis.

All authors jointly contributed to the design of the study, data analysis and interpretation of results, and development of the manuscript. All authors have read and approved the content of the final manuscript.