Reliability, Validity, and Responsiveness of the Daily Sleep Interference Scale Among Diabetic Peripheral Neuropathy and Postherpetic Neuralgia Patients

Article Outline

• Abstract
• Introduction
• Methods
  • Study Sample
  • Measures
  • Data Collection Procedures
  • Statistical Analysis
• Results
  • Study Clinical and Demographic Characteristics
  • Descriptive Statistics for the DSIS and Other PRO Measures
  • Test–Retest Reliability
  • Construct Validity
  • Discriminant Validity
  • Responsiveness/Interpretation
• Discussion
• References
• Copyright

Abstract 

To evaluate the psychometric characteristics of the Daily Sleep Interference Scale (DSIS) in patients with painful diabetic peripheral neuropathy (DPN) or postherpetic neuralgia (PHN), a post hoc secondary analysis of data from eight randomized clinical trials (four DPN and four PHN) was performed. Data were pooled within patient populations when assessment weeks were the same. The DSIS was completed by 1,124 DPN and 1,034 PHN patients. Patient-reported outcomes, including a Daily Pain Diary, the Short-Form McGill Pain Questionnaire, SF-36 Health Survey, Profile of Mood States, MOS-Sleep Scale (MOS-SS), EQ-5D, and Patient Global Impression of Change, were used to validate the DSIS. Test–retest reliability was high for both samples (intraclass correlation coefficient>0.90). The DSIS showed good construct validity, with moderate to high correlations between the DSIS and weekly mean pain scores (r=0.48–0.80), MOS-SS sleep disturbance subscale (r=0.45–0.64), MPQ-SF Pain Experience (r=0.37–0.61), and VAS (r=0.42–0.72). The DSIS showed good discriminant validity in both groups; high and low MOS-SS sleep disturbance groups had significantly different DSIS scores (P<0.001). DPN patients who improved minimally on the Patient Global Impression of Change and in pain scores improved 1.5-2 DSIS points on average; for PHN, patient scores improved an average of 1–2 points. The DSIS demonstrated robust test–retest reliability, good construct and discriminant validity and responsiveness in painful DPN and PHN patients. A 1–2 point change on the DSIS might be interpreted as an important difference.

Key Words: Pain, sleep interference, patient reported outcome, psychometric validation

Back to Article Outline

Introduction 

The Neuropathy Association estimates that between 15 and 20 million Americans suffer from peripheral neuropathy.¹ Peripheral neuropathy is caused by a deterioration of the peripheral nerves, and symptoms include unpleasant sensations such as tingling, burning, itchiness, crawling sensation, dizziness, clumsiness, numbness, and chronic pain.¹ Although neuropathy commonly co-occurs with several different diseases, peripheral neuropathy associated with diabetes mellitus and the sequelae of herpes zoster are of particular interest for this study. Peripheral neuropathy is one of the most common complications associated with Type 1 and Type 2 diabetes. In a study of Type 2 diabetes patients in the United Kingdom, 63.8% reported that they experience some pain.² Chronic peripheral neuropathic pain occurs in one of six diabetic patients.³

Peripheral neuropathic pain has been shown to have significant effects upon the functional health status and other aspects of health-related quality of life (HRQL) of patients. On average, pain is quite significant, with patients reporting average pain as 6/10 on an 11-point scale.⁴ Patients report the sensations of pain as burning, electric, sharp, and dull/ache, and these symptoms are most common at night, or when patients are tired or stressed.⁴ Patients report that pain has significant impacts on enjoyment of life, and to a lesser extent impacts on recreational activities, work functioning, mobility, activity functioning, social functioning, and mood.⁴ Studies have found that one of the most significant impacts of painful peripheral neuropathy is on sleep quality.⁴, ⁵ Because there are established links between sleep and metabolic control, and because poor sleep may lead to disease progression, the impact of pain upon sleep is of large concern for this patient population.⁶, ⁷, ⁸ Clearly, sleep disturbance due to painful diabetic peripheral neuropathy (DPN) may have profound impacts on functioning and other aspects of HRQL.

Herpes zoster is another disease that is common (lifetime prevalence has been estimated to be as high as 20%) and is almost always accompanied by pain.⁹, ¹⁰ Postherpetic neuralgia (PHN) is pain that persists after the acute stage of the rash that occurs with herpes zoster. There is some evidence that PHN might be more likely to occur in herpes zoster patients over the age of 60.¹¹ PHN, like DPN, has significant impact on the HRQL for patients. On an 11-point scale, average pain among patients 65 years of age and older with PHN has been shown to be 4.6.¹² PHN has been shown to affect daily activities, mood, social functioning, and enjoyment of life. PHN also has been shown to be particularly disruptive to sleep.¹², ¹³

To reliably measure patients' experiences with pain and sleep, a Patient Reported Outcome (PRO) measure is necessary because only patients themselves can provide information about the pain quality, intensity, and sleep interference they experience. Although there are several widely used and validated pain instruments (e.g., the MPQ-SF)¹⁴ and sleep instruments (e.g., the MOS-Sleep Scale [MOS-SS])¹⁵ available, there are few instruments that measure sleep interference as a result of pain. The Daily Sleep Interference Scale (DSIS) was developed to quantify sleep interference due to pain.¹⁶, ¹⁷, ¹⁸ The DSIS is a single-item measure that is completed by patients once a day (upon awakening) to accurately capture variability in sleep interference due to pain on a daily basis, thus minimizing recall bias. The DSIS has an 11-point response scale that asks patients to “Select the number that best describes how much your pain has interfered with your sleep during the past 24hours.” Response options range from 0 (Did not interfere with sleep) to 10 (Completely interfered with sleep—unable to sleep due to pain). The DSIS is designed to be used in a patient daily diary that patients fill out upon awakening each morning.

The DSIS has recently been used to assess sleep interference due to pain with DPN and PHN patients in several randomized clinical trials of pregabalin, which has been shown to be effective in treating pain and improving sleep among DPN patients.¹⁹, ²⁰, ²¹, ²², ²³ Although the DSIS has been used as a clinical endpoint, limited information is available on the psychometric qualities of this measure to support its use for evaluating treatment effectiveness among both DPN and PHN patient populations. The purpose of this secondary analysis was to evaluate the psychometric properties of the DSIS. Previous work established the content validity of the DSIS through cognitive debriefing interviews with 16 neuropathy patients. Results from this study suggested that patients clearly understood the intent of the item and were able to easily rate their sleep interference due to pain on the 0–10 response scale. The current analyses performed on the mean Daily Sleep Interference item focused on evaluating test–retest reliability, construct validity, known-groups validity, and responsiveness separately for both the DPN and PHN patient populations. Information is also provided for exploring the minimal important difference (MID) of DSIS scores.

Back to Article Outline

Methods 

Study Sample 

This secondary analysis used data from eight different randomized clinical trial studies and includes two different patient populations: DPN patients (four studies; n=1,124) and PHN patients (four studies; n=1,034). Data from the studies were combined within patient group (four for DPN and four for PHN) when assessment weeks were the same. Table 1 summarizes each protocol, including the study title, sample size, disease population, location of study, clinical trial design, and PRO assessment schedule. Patients in both disease groups and across all studies were recruited from clinical centers. Patients were 18 years of age or older, able to understand and cooperate with study procedures, and provided written informed consent prior to participating in the study.

Table 1. Study Descriptions

Study Title	n	Population	Location	Study Duration (weeks)	Titration (weeks)	Stable Dose (weeks)	MPQ-SF	SF-36	PGIC	POMS	MOS-SS	EQ-5D
A six-week, double-blind, placebo-controlled trial of pregabalin (150, 600mg/day) for treatment of painful diabetic peripheral neuropathy (protocol 1008-14)	246	Diabetes	United States and Canada	6	2	4	−1, 0, 2, 4, 6	0, 6	6	0, 6
A five-week, double-blind, placebo-controlled trial of 3 dosages of pregabalin (75, 300, and 600mg/day) for treatment of patients with painful diabetic peripheral neuropathy (protocol 1008-029)	337	Diabetes	United States	5	1	4	−1, 0, 1, 3, 5	0, 5	5	0, 5
A five-week, double-blind, placebo-controlled, parallel group study of pregabalin (150 and 450mg/day) in patients with postherpetic neuralgia (protocol 1008-030)	255	Postherpetic	United States	5	1	4	−1, 0, 1, 3, 5	0, 5	5	0, 5
An eight-week, double-blind, placebo-controlled, parallel group study of pregabalin (150 and 300mg/day) in patients with postherpetic neuralgia (protocol 1008-045)	238	Postherpetic	International	8	1	7	−1, 0, 1, 3, 5, 8	0, 8	9
An eight-week, double-blind, placebo-controlled, parallel group study of pregabalin in patients with postherpetic neuralgia (protocol 1008-127)	173	Postherpetic	United States and Canada	8	1	7	−1, 0, 1, 3, 5, 8	0, 8	9	0, 8	0, 8
An eight-week, double-blind, placebo-controlled trial of pregabalin (300mg/day) for relief of pain in patients with painful diabetic peripheral neuropathy (protocol 1008-131)	146	Diabetes	United States	8	0	8	−1, 0, 1, 3, 5, 8	0, 8	9	0, 8
A 12-week, randomized, double-blind, multicenter, placebo-controlled study of pregabalin twice a day (BID) for relief of pain associated with diabetic peripheral neuropathy (protocol 1008-149)	395	Diabetes	International	12	1	11	−1, 0, 1, 4, 8, 12	0, 12	12		0, 12	0, 1, 4, 8, 12
A 13-week, randomized, double-blind, multicenter, placebo-controlled study of pregabalin twice a day (BID) in the treatment of postherpetic neuralgia (protocol 1008-196)	368	Postherpetic	International	13	1	12	−1, 0, 1, 4, 8, 13	0, 13	13		0, 13	0, 1, 4, 8, 13

For the painful DPN studies, inclusion criteria included diagnosis of Type 1 or 2 diabetes mellitus; diagnosis of diabetic, distal, symmetrical, sensorimotor polyneuropathy for one to five years with hemoglobin A1c levels of ≤11%; and at the baseline and randomization visits, a score of ≥40 mm on the Visual Analog Scale (VAS) of the Short-Form McGill Pain Questionnaire (SF-MPQ). Patients were excluded from the study if they had a history of comorbidities that would interfere with interpretation of results, including neurological disorders, skin conditions, and other severe pain.

For the PHN studies, inclusion criteria included pain present for more than three months after healing of the herpes zoster skin rash, but patients must not have experienced pain for more than five years, and patients at baseline must have had a score ≥40 mm on the VAS of the SF-MPQ. Patients were excluded from the study if they had a history of co-morbidities that would interfere with interpretation of results, including having undergone neurolytic or neurosurgical therapy for PHN; clinically significant hepatic, respiratory, hematological illnesses, or cardiovascular disease; patients who were immunocompromised (i.e., conditions known to be associated with an immunocompromised state); patients having had other severe pain that may confound assessment or self-evaluation of pain due to PHN; and skin conditions in the affected dermatome that could alter sensation.

Measures 

Demographic and Clinical Characteristics. Data on patient gender, age, race, and nationality were collected. For the diabetic patients, clinical characteristic data were collected on the duration of diabetes in years, creatinine clearance, diabetes type (1 or 2), and the presence of allodynia. For the PHN patients, clinical characteristics were collected on creatinine clearance and presence of allodynia.

Patient-Reported Outcome Measures. Instruments used in the psychometric analyses were completed by patients. The instruments included the DSIS Diary, the Daily Pain Diary, the SF-MPQ, the Patient Global Impression of Change (PGIC), the SF-36 Health Survey (SF-36), the Profile of Mood States (POMS), the MOS-SS, and the EuroQol Health State Index (EQ-5D). The DSIS and Daily Pain Diary were completed each morning by patients in their homes. The other PRO measures were completed during clinical visits.

Daily Sleep Interference Diary. The Daily Sleep Interference Diary consists of an 11-point Likert scale with which patients assess how pain has interfered with their sleep during the past 24 hours.¹⁶, ¹⁷, ¹⁸ On this scale a 0 indicates “pain does not interfere with sleep” and 10 indicates “pain completely interferes with sleep.” Self-assessment was performed daily at awakening. A mean DSIS was calculated at the end of each week, and these weekly mean sleep interference scores were used in analyses presented here.

Daily Pain Diary. The Daily Pain Diary consists of an 11-point Likert scale with 0 as “no pain” and 10 as “worst possible pain.”¹⁶, ¹⁷, ¹⁸ Patients describe their pain during the past 24 hours by choosing the appropriate number between 0 and 10. Self-assessment was performed daily at awakening. The mean Daily Pain score was calculated at the end of each week, and these weekly mean pain scores were used in the analyses presented here.

Short Form-McGill Pain Questionnaire (SF-MPQ). The SF-MPQ consists of 15 descriptors; 11 represent the sensory dimension of pain experience, and four represent the affective dimension.¹⁴ Each descriptor is ranked by the patient on a four-point intensity scale and totaled in each subclass. The SF-MPQ also includes a six-point Present Pain Intensity index and a VAS to provide overall intensity scores. Higher scores indicate more pain.

Patient Global Impression of Change. The PGIC is a patient-rated instrument that measures change in the patient's overall status on a seven-point scale. Scores range from 1 (very much improved) to 7 (very much worse).²⁴ Recent research has shown that single-item global improvement-type items are valid when assessing changes in pain.²⁵, ²⁶, ²⁷

SF-36 Health Survey. The SF-36 is a self-administered generic health status questionnaire that measures each of the following eight health concepts: physical functioning, role limitations due to physical problems, social functioning, bodily pain, mental health, role limitations due to emotional problems, vitality, and general health perception.²⁸ Two summary scores can be computed, the physical component and mental component summary scores. Higher scores indicate better health status.

Profile of Mood States (POMS). The POMS measures six moods: tension-anxiety, depression-dejection, anger-hostility, vigor-activity, fatigue-inertia, and confusion-bewilderment.²⁹ A score is obtained from each scale and a single overall score of “mood disturbance” is also computed.

Medical Outcomes Study-Sleep Scale (MOS-SS). The MOS-SS is a patient-rated questionnaire consisting of 12 items that assess the key constructs of sleep.¹⁵ The instrument consists of seven subscales: sleep disturbance, snoring, awaken short of breath or with headache, quantity of sleep, optimal sleep, sleep adequacy, and somnolence. Optimal Sleep was not used in this study. A sleep problems index score, the Sleep Problems Index (SPI) (nine items), was used in this study to assess common problems associated with sleep, including nonrestorative sleep and difficulty with sleep initiation and maintenance. Higher scores indicate more difficulties with the sleep attribute being measured, except sleep adequacy, which has a reverse score—higher scores indicate more sleep adequacy.

EuroQol Health State Index (EQ-5D). The EQ-5D is a patient-completed instrument designed to assess impact on HRQL in five domains: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression.³⁰ The scores from the five domains may be used to calculate a single preference-based index score. Additionally, the EQ-5D contains a VAS that asks the patient to rate their current health state from 0 to 100, where 0 represents the worst imaginable health state and 100 represents the best imaginable health state.

Data Collection Procedures 

All patients rated their Daily Sleep Interference due to pain each morning of the study about the previous night. Patients also rated their Daily Pain each morning of the study. In all studies patients also completed the MPQ-SF, SF-36, and the PGIC during different weeks of the studies (see Table 1 for study title, measures included in the study, and week of administration). In addition, the MOS-SS, POMS, and EQ-5D were included as measures in several of the clinical trials. All PRO measures other than the diary measures were completed by patients during clinical visits.

Statistical Analysis 

All analyses were conducted on blinded data using the total randomized study population. All available data were used. Psychometric analyses were conducted separately by disease population. When assessment weeks within a disease population were the same for a given measure across more than one study, data were combined.

Test–retest reliability of weekly mean DSIS over time was evaluated with two different analysis strategies by patient group. First, among the subset of patients reporting no change on the PGIC at end of treatment, intraclass correlation coefficients (ICCs), Spearman correlation coefficients, and paired t-tests were calculated between the relevant weekly mean and retest weekly mean DSIS scores. PGIC is only assessed at the end of treatment, and so stability in weekly mean DSIS scores was assessed from the week before end of treatment to the end of treatment visit. Secondly, test–retest reliability was assessed using a proxy variable from baseline (Week 0) to first treatment week (Week 1) by disease group. Specifically, among those patients who reported the same weekly mean pain score from baseline to Week 1 (within±1 point), test–retest reliability of the DSIS was assessed with ICCs, Spearman correlations, and paired t-tests between the baseline and Week 1 weekly mean daily sleep interference scores.

Construct validity refers to the extent to which the instrument measures what it is intended to measure.³¹ Construct validity of the weekly mean DSIS was evaluated through examination of the statistical relationships between the DSIS and the weekly mean pain score, the SF-MPQ (pain experience and affective dimensions and the Present Pain Intensity and VAS overall scores), the SF-36 (six subscales and two component scores), the POMS (six mood scores and overall score), the MOS-SS (six subscales and two index scores), and the EQ-5D (five domains and index score) by study population at each assessment point (baseline, Weeks 1, 2, 3, 4, 5, 6, 8, 12, and 13). Representative subsets of assessment week results are presented here. Spearman correlation coefficients were used to evaluate these relationships. It was hypothesized that moderate to large correlations would be found between the weekly mean DSIS scores and the weekly mean pain scores, the SF-36 vitality subscale, the MOS-SS sleep disturbance subscale, and the POMS fatigue subscale. Small-to-moderate relationships were expected between the weekly mean DSIS scores and the other PRO scores.

Discriminant validity is the extent to which scores from an instrument are distinguishable from groups of subjects that differ on a relevant dimension, such as clinical severity. To assess discriminant validity, subjects were stratified by the mean MOS-SS sleep problems index (nine item) and the sleep disturbance subscale scores based on the MOS study population data. The baseline visit (Week 0, as available) DSIS was compared by sleep severity level. Also to assess discriminant validity, subjects were stratified by their MPQ-SF VAS score (less pain=1–3, some pain=4–6; more pain=7–10) at Weeks 1, 2, 3, 4, 5, 6, 8, 12, and 13 (as available). A representative subset of these results is reported here. DSIS scores were compared by pain severity level. Analysis of variance (ANOVAs) were used for these analyses, with post hoc category comparisons via Scheffe's test done as part of the ANOVAs.

We examined the responsiveness of the weekly mean DSIS score and explored possible MID based on measures of patient global assessments and changes in pain through three sets of analyses. The MID is defined as the smallest difference in scores that patients perceive as important, either beneficial or harmful, and which would lead the clinician to consider a change in the patient's management. Mean change in weekly DSIS scores from baseline to end of treatment period were categorized by PGIC category at end of treatment. The PGIC was grouped into four categories: 1 and 2 (Very Much Improved and Improved); 3 (Somewhat Improved); 4 (No Change); 5, 6, and 7 (Somewhat Worse, Worse, and Very Much Worse). An analysis of covariance (ANCOVA) model was used to compare differences in mean baseline to endpoint changes between the responder groups (based on the PGIC). Responder group and baseline DSIS were entered as covariates in the ANCOVA model. Disease groups were analyzed separately. Data within disease group from protocols that had the same end of treatment period week were pooled. The estimated MID is determined by examining the differences between the stable group (PGIC score of 4) and the group defined as reporting small improvements (PGIC score of 3) based on the PGIC.

Responsiveness and MID were also examined by looking at change in DSIS scores from baseline to end of treatment period by categories of changes in weekly mean pain scores from baseline to end of treatment period. According to Farrar et al., changes of 2–3 points on 11-point pain scales can be considered clinically important effects for pain scores.²⁷ Thus, MID for sleep interference was evaluated by examining changes in DSISs that occur with clinically important changes in pain ratings. Mean change in DSISs from baseline to end of treatment period were calculated and patients were categorized by changes in weekly mean pain scores from baseline to end of treatment. Patients were grouped into three pain change categories: <2 (worsening, no change); 2–3 (minimal clinical improvement); >3 (improvement). An ANCOVA model was used to compare differences in mean baseline to endpoint changes between the responder groups (based on changes in the weekly mean daily pain scores). Responder groups and baseline DSISs were entered as covariates in the ANCOVA models. Disease groups were analyzed separately. Data within disease group from protocols that had the same end of treatment period week were pooled. The estimated MID is determined by examining change in weekly mean DSISs associated with clinically important changes in weekly mean pain scores (change score of 2–3).

Back to Article Outline

Results 

Study Clinical and Demographic Characteristics 

A total of 1,124 DPN patients were included in the DSIS validation analyses presented here. As can be seen in Table 2, 58% of the DPN sample was male and the mean age was 59 years (range, 21–85 years). Approximately 92% of the DPN sample was White and the remaining 8% was Black (4%), Hispanic (3%), Asian (<1%), and American Indian/Alaskan (<1%). Approximately 65% of the DPN sample was from North America, with the remaining patients from Australia (7%), Germany (6%), Hungary (3%), Poland (15%), South Africa (3%), and the United Kingdom (2%).

Table 2. Patient Demographic Characteristics Diabetes Population and Postherpetic Population^a

Characteristic	DPN		PHN
	n	Percent	n	Percent
Age (years) mean (SD) [min–max]	58.81 (10.85) [21–85]		71.38 (10.25) [18–100]
Sex
Male	652	58.01	485	46.91
Female	472	41.99	549	53.09
Race
White	1032	91.81	1010	97.68
Black	42	3.74	7	0.68
Hispanic	32	2.85	11	1.06
Asian/Pacific Islander	8	0.71	4	0.39
American Indian/Alaska	3	0.27
Other	7	0.62	2	0.19
Patient clinical characteristics
Creatinine clearance mean mL/min (SD) [min–max]	100.82 (34.57) [31–313]		69.71 (25.83) [23–229]
Presence of allodynia
Yes	620	55.16	394	38.10
No	502	44.66	636	61.51
Duration of diabetes in years mean (SD) [min–max]	10.54 (8.79) [0–62]
Diabetes type
Type 1 diabetes	130	11.57
Type 2 diabetes	994	88.43

In the DPN population, patients had a mean diagnosis of diabetes of 11 years and 88% of the sample had a diagnosis of Type 2 diabetes (see Table 2). Creatinine clearance averaged 100.82, which is within the normal range, and 55% of the population had the presence of allodynia (touch evoked pain).

A total of 1,034 PHN patients were included in the validation analyses presented here. As can be seen in Table 2, 53% of the sample was female and the mean age was approximately 71 years (range, 18–100 years). The PHN sample was approximately 98% White and 1% Hispanic; less that 1% of the sample was Black, Asian, or American Indian. Approximately 41% of the sample was from North America and 17% was from the United Kingdom. The remaining participants were from Australia (8%), Austria (2%), Belgium (<1%), France (3%), Germany (6%), the Netherlands (9%), Poland (6%), Portugal (<1%), Spain (6%), Sweden (<1%), and Switzerland (<1%).

As can be seen in Table 2, in the PHN population, creatinine clearance, calculated using the Cockcrof–Gault equation, averaged 69.71mL/min, which is within the normal range, and 38% of the sample had the presence of allodynia (touch evoked pain).

Descriptive Statistics for the DSIS and Other PRO Measures 

Descriptive statistics (means by assessment week of the DSIS) can be seen in Fig. 1. The average baseline DSIS score in the DPN population was 5.37 (SD=2.40), and average DSIS scores decreased over time. For the PHN population, average DSIS scores were 4.64 (SD=2.57) at baseline, and decreased over time. The weekly mean pain scores followed similar trends, with patients improving over time (data not shown). Pain and sleep interference decreased similarly over time during the studies for both DPN and PHN patients.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1 

  Weekly mean DSIS for DPN and PHN patient groups over time.

Test–Retest Reliability 

Among patients who reported no change on the PGIC, DSIS scores remained very stable from the assessment week prior to the final assessment week (Table 3). In the DPN sample, ICCs ranged between 0.95 and 0.99 and t-tests between test and retest DSIS scores were all nonsignificant. In the PHN sample, ICCs were 0.98 and t-tests between the test and retest DSIS scores were nonsignificant.

Table 3. Test–Retest Reliability (Reproducibility): Patients Reporting No Change on PGIC from Baseline to End of Treatment Week

Weekly Mean Sleep Interference Score	n	Mean EOT-1	Mean EOT	Difference	Student's t	P-Value	Spearman r2	ICC
Diabetic population
Study 1008-14 (end of treatment Week 6, test–retest Week 5–6)	66	4.69	4.74	0.05	0.56	0.58	0.96	0.96
Study 1008-029 (end of treatment Week 5, test–retest Week 4–5)	60	4.51	4.44	−0.07	−0.62	0.54	0.93	0.95
Study 1008-131 (end of treatment Week 8, test–retest Week 7–8)	40	4.51	4.38	−0.13	−1.75	0.09	0.98	0.99
Study 1008-149 (end of treatment Week 12, test–retest Week 11–12)	49	4.96	4.90	−0.06	−0.77	0.45	0.98	0.97
Postherpetic population
Study 1008-030 (end of treatment Week 5, test–retest Week 4–5)	100	4.02	4.07	0.05	0.81	0.42	0.97	0.98
Study 1008-045 and 1008-127 (end of treatment Week 8, test–retest Week 7–8)	104	3.94	3.93	−0.00	−0.04	0.97	0.98	0.98
Study 1008-196 (end of treatment Week 13, test–retest Week 12–13)	72	3.53	3.48	−0.05	−0.94	0.35	0.98	0.98

When test–retest reliability was assessed among patients with stable weekly pain scores from baseline to Week 1, good test–retest reliability was also found. ICCs were very high (0.93 for both DPN and PHN populations). Mean changes from baseline to Week 1 in DSIS scores were not large (0.19–0.35 points), but were significant at the P<0.001 level. These significant P-values may be attributable to the large sample sizes, as well as to the spread of ±0.5 allowed in the pain score change from baseline to Week 1.

Construct Validity 

The construct validity of the DSIS was evaluated though correlations with the weekly mean pain score, the SF-MPQ, the SF-36 subscales, the POMS scores, the MOS-SS subscale scores, and the EQ-5D domain and index scores across all treatment weeks by disease group. As hypothesized, moderate-to-large correlations were found between the DSIS and the weekly mean pain scores and the MOS-SS sleep disturbance subscale for both DPN and PHN patients (see Table 4). Although moderate–large relationships were expected between the DSIS and the SF-36 vitality subscale and the POMS fatigue subscale, these relationships were small to moderate for both DPN and PHN patients (POMS results not shown). A moderate correlation was observed between the DSIS and the Bodily Pain subscale of the SF-36. Moderate–large relationships were found between the DSIS and the MPQ-SF pain experience and VAS dimensions for both DPN and PHN patients. Correlations between the EQ-5D and other subscales (SF-36, POMS, and MOS-SS) were small to moderate for both patient populations, as expected. All relationships were in the expected directions; as sleep interference increased, so did sleep problems and pain, while functioning decreased. These findings support good convergent validity—those scales that measure very similar constructs as the DSIS have moderate to large correlations with the DSIS. The findings reported here also support good divergent validity—those scales that measure different but related constructs to the DSIS show small to moderate correlations with the DSIS. These correlations were consistent across the painful DPN and PHN samples and across different assessment weeks in the various studies.

Table 4. Relationships Between the Weekly Mean Sleep Interference Score and Other Scales in Diabetes and Postherpetic Patients

NRS = numeric rating scale; VAS = visual analog scale.

Discriminant Validity 

Discriminant validity was assessed in three ways: by dividing patients into higher and lower MOS-SS sleep problems index (nine item) groups (above and below normative mean); by dividing patients into higher and lower MOS-SS sleep disturbance groups (above and below the normative mean); and by dividing patients into one of three pain severity groups (using MPQ-VAS scores). DSIS scores were compared by group using ANOVA. For the DPN group, the mean DSIS scores varied significantly by sleep disturbance and pain score groups (see Fig. 2, Fig. 3), and by sleep problem groups (P<0.001; data not shown). Patients with more sleep disturbance reported greater mean sleep interference scores (P<0.001). Patients with more pain reported greater mean sleep interference scores (P<0.001). These analyses were conducted at all time points and mean DSIS scores significantly varied by sleep problem groups, sleep disturbance groups, and pain groups at all time points (all P<0.001; data not shown).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 2 

  Mean weekly Sleep Interference Score by MOS-SS Sleep Disturbance Score groups in diabetes and postherpetic patients at baseline. DPN: F=78.45; P<0.001; PHN: F=109.59; P<0.001.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 3 

  Mean weekly Sleep Interference Score by MPQ-SF VAS Score in diabetes and postherpetic patients at Week 1. DPN: F=190.15; P<0.001; PHN: F=92.07; P<0.001. Scheffe post hoc tests showed that all groups were significantly different at the P<0.001 level.

For the PHN sample, similar results were found. As can be seen in Fig. 2, Fig. 3, mean DSIS scores significantly varied by sleep disturbance groups and pain groups (both P<0.001) and by sleep problem groups (P<0.001; data not shown). Those patients who reported greater sleep problems, sleep disturbance, or pain reported higher mean sleep interference scores. These results were replicated across all study weeks (data not shown).

Responsiveness/Interpretation 

Responsiveness of the DSIS was assessed in two ways: by comparing changes in DSIS scores from baseline to end of treatment week by PGIC status, and by comparing changes in the DSIS from baseline to end of treatment week by changes in weekly mean pain scores. MID was explored by comparing differences in mean DSIS change scores between the stable (no change) and small improvements groups based on the PGIC and pain scores. As findings were very similar regardless of end of treatment week, therefore only selected weeks are reported that represent the overall trends in results for all weeks.

For the DPN group, there were statistically significant differences in baseline to 12-week mean DSIS change scores by PGIC groups (P<0.001). Those patients reporting greater improvements in clinical status also reported greater improvements in DSIS scores. Those patients who reported no change on the PGIC from baseline to end of treatment had a mean change of −0.6 points in DSIS scores compared to a mean change of −1.9 points in those patients who reported a small improvement (see Fig. 4).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 4 

  Mean change in weekly Sleep Interference Score by PGIC Scores from baseline to end of treatment (Week 12) in diabetes patients. F=29.11; P<0.001.

Baseline to Week 6 changes in DSIS scores were significantly related to baseline to Week 6 endpoint changes in weekly pain scores (F=61.27; P<0.001; data not shown). Those patients who experienced a greater than three-point improvement in pain scores reported a mean improvement of −3.43 points in DSIS scores. Patients who reported no change on their weekly mean pain score had a mean change of −0.46 points on the DSIS compared to patients who experienced a small change on the weekly mean pain score, who had a mean change in DSIS of −2.54 points. Thus, for DPN patients, differences in mean change scores between patients who reported no change and clinically important change ranged from 1.3 points to 2.08 points, depending on the method used. These results suggest that for DPN patients, an important difference for the DSIS ranges from 1.5 to 2 points.

For the PHN group, baseline to Week 8 changes in DSIS scores were significantly associated with PGIC groups (P<0.001; see Fig. 5). Mean changes in DSIS scores were greatest in those groups reporting the largest improvements in clinical status. The mean change in DSIS for the group that reported no change on the PGIC was −0.7, and the mean change in DSIS for the somewhat improved group was −1.9.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 5 

  Mean change in weekly Sleep Interference Scores by PGIC Scores from baseline to end of treatment (Week 8) in postherpetic patients. F=31.17; P<0.001.

Baseline to Week 13 changes in DSIS scores were significantly related to baseline to Week 13 endpoint changes in weekly pain scores (F=58.40; P<0.001; data not shown). Those patients who experienced a greater than three-point improvement in pain scores reported a mean improvement of −3.78 points in DSIS scores. The mean DSIS change for the group that reported no change on the weekly mean pain score was −0.81 points and the group who reported a clinically meaningful change in weekly mean pain scores reported a mean change on the DSIS of −2.95 points (see Fig. 5). Differences between mean DSIS change scores for PHN patients who did not improve and for those who had some improvement ranged from 0.82 to 2.14, suggesting that an important difference for the DSIS ranges from approximately 1–2 points.

Back to Article Outline

Discussion 

The purpose of these analyses was to assess the psychometric qualities of the DSIS to support its use as an efficacy endpoint in randomized clinical trials with DPN and PHN patients. In order for a PRO instrument to be considered robust to support drug licensing or labeling claims, it must show evidence of good reliability and validity within the patient population of interest. In this study, we assessed the test–retest reliability, construct validity, discriminant validity, and responsiveness of the DSIS in both DPN and PHN groups. Based on the current study findings, the DSIS has good evidence of reliability and validity in these two patient populations. We found evidence suggesting that an important difference for the DSIS, from the perspective of patients, may be between one and two points.

Test–retest reliability analyses demonstrated good reproducibility of the DSIS over one-week retest periods. Among those patients who reported no change on the PGIC, test–retest statistics demonstrated very little change in the DSIS scores from week before end of treatment week to end of treatment week. ICCs in the stable group were greater than 0.9 for both DPN and PHN patient populations. However, t-tests assessing differences between baseline and Week 1 DSIS scores among stable pain patients were significantly different, despite ICCs>0.90. Given that the ICCs were very high, and because the actual magnitude of change between test and retest weeks among stable patients was small, the significant t-tests may have had more to do with the relatively large sample sizes included in the analyses than real changes in the DSIS among stable patients. The test–retest reliability observed in this study indicates that the DSIS is certainly sufficient for group comparisons and may be acceptable for individual patient comparisons.

Findings reported here also showed evidence of good construct validity for the DSIS. Measures that were conceptually related to the DSIS, including the pain score, the MOS-SS sleep disturbance subscale, as well as the MPQ-SF affective dimension and VAS, were moderately to highly correlated with the DSIS across all assessment weeks and for both patient populations. Stronger relationships between measures that are conceptually similar demonstrate evidence of good convergent validity. Findings reported here suggested that both pain scores (SF-36 Bodily Pain, MPQ-SF VAS) as well as sleep interference and disturbance scores on the MOS-SS were moderately to highly related to the DSIS, providing good evidence that this measure is capturing well the construct of sleep interference due to pain. The other PRO measures that are conceptually less similar to the DSIS, including other SF-36 subscales, the POMS, the EQ-5D, and other MOS-SS subscales, showed small–moderate correlations with the DSIS, as expected. The results indicate that patients report greater sleep interference with greater pain severity and with more sleep disturbance and other sleep problems.

For both PDH and PHN patients, the DSIS showed evidence of good discriminant validity. Patients in the greater sleep disturbance and problems groups had significantly higher sleep interference scores compared to the low sleep disturbance and problems groups. In addition, the DSIS significantly discriminated between pain severity groups, with those patients reporting greater levels of pain severity also reporting more sleep interference on the DSIS. Thus, in both population groups, the DSIS showed evidence of good discriminant validity.

Finally, several analyses were conducted to evaluate the responsiveness and to provide insight for interpreting differences in DSIS scores. In both painful DPN and PHN patients, the DSIS was responsive to patients' reports of change on the PGIC scale and changes in daily pain scores. Based on examining differences between the mean changes in DSIS among stable patients and those patients reporting some improvement on the PGIC or in their daily pain scores, a change or differences of 1–2 points may be important from the patient's perspective. The results presented here will help with interpretation of the scores; patients who improve 1–2 points or more on the DSIS are more likely to have clinically meaningful improvement in their sleep interference due to neuropathic pain. Further research is needed to determine whether this value represents the MID for the DSIS.

The strengths of this study include that the DSIS was assessed for psychometric properties in two patient populations with relatively large sample sizes in longitudinal data across various treatment weeks. Although point-in-time estimates are often used to assess psychometric properties, this study was able to demonstrate the good reliability and validity of the DSIS across disease groups and consistently over multiple assessment points. The fact that reliability and validity assessments showed similar patterns across disease populations and at multiple time points lends stronger evidence as to the psychometric robustness of the DSIS.

A limitation of this study is that patients whose data were assessed as a part of this study had relatively high pain scores as a requirement for inclusion in the clinical trials. Thus, these data might not be completely generalizable to all DPN and PHN patients who might have lower pain and sleep impacts due to their neuropathy. In addition, patients with potential sleep interference comorbidities were excluded from the clinical trials and thus the validation analyses were conducted using data from a sample who were likely able to easily attribute sleep interference due to pain. For patients who might suffer other types of pain or have other reasons for sleep interference, it is possible that they might not be able to distinguish between different reasons for sleep interference, leading to more error variance and poorer psychometric performance of the DSIS. This warrants further research. However, the patient samples included in the DSIS psychometric analyses are more generalizable to patients entering clinical trials of treatments for painful DPN or PHN.

In conclusion, the results of these analyses demonstrated that the DSIS has excellent evidence supporting test–retest reliability and construct validity. The DSIS was responsive to changes in patient-reported clinical status and pain scores over time. The DSIS is brief and can be easily incorporated into daily diaries. Therefore, the DSIS can be used to assess sleep interference due to pain, and reliably assess changes in sleep interference due to pain over time in DPN and PHN patient populations. In addition, we have provided preliminary guidance as to interpretation of changes in DSIS scores. A change of 1–2 points on the DSIS might be considered to be an important improvement for patients with DPN or PHN. As sleep problems are important endpoints for clinical trials comparing treatments for PHN or painful DPN, the availability of a brief and psychometrically sound assessment of sleep interference may help physicians and their patients understand the effects of treatment on both pain and sleep outcomes.

Back to Article Outline

References 

1. Neuropathy Association . Available at http://www.neuropathy.orgAccessed April 27, 2007
   • View In Article
2. Davies M, Brophy S, Williams R, Taylor A. The prevalence, severity, and impact of painful diabetic peripheral neuropathy in type 2 diabetes. Diabetes Care. 2006;29:1518–1522
   • View In Article
   • MEDLINE
   • CrossRef
3. Daousi C, MacFarlane IA, Woodward A, et al. Chronic painful peripheral neuropathy in an urban community: a controlled comparison of people with and without diabetes. Diabet Med. 2004;21:976–982
   • View In Article
   • MEDLINE
   • CrossRef
4. Galer BS, Gianas A, Jensen MP. Painful diabetic polyneuropathy: epidemiology, pain description, and quality of life. Diabetes Res Clin Pract. 2000;47:123–128
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (72 KB)
   • CrossRef
5. Gore M, Brandenburg NA, Dukes E, et al. Pain severity in diabetic peripheral neuropathy is associated with patient functioning, symptom levels of anxiety and depression, and sleep. J Pain Symptom Manage. 2005;30:374–385
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (287 KB)
   • CrossRef
6. Zelman DC, Brandenburg NA, Gore M. Sleep impairment in patients with painful diabetic peripheral neuropathy. Clin J Pain. 2006;22:681–685
   • View In Article
   • MEDLINE
   • CrossRef
7. Spiegel K, Knutson K, Leproult R, Tasali E, Van Cauter E. Sleep loss: a novel risk factor for insulin resistance and Type 2 diabetes. J Appl Physiol. 2005;99:2008–2019
   • View In Article
   • MEDLINE
   • CrossRef
8. Kawakami N, Takatsuka N, Shimizu H. Sleep disturbance and onset of type 2 diabetes. Diabetes Care. 2004;27:282–283
   • View In Article
   • MEDLINE
   • CrossRef
9. Stankus SJ, Dlugopolski M, Packer D. Management of herpes zoster (shingles) and postherpetic neuralgia. Am Fam Physician. 2000;61:2437–24442447–2438
   • View In Article
   • MEDLINE
10. Portenoy RK, Duma C, Foley KM. Acute herpetic and postherpetic neuralgia: clinical review and current management. Ann Neurol. 1986;20:651–664
    • View In Article
    • MEDLINE
    • CrossRef
11. Helgason S, Petursson G, Gudmundsson S, Sigurdsson JA. Prevalence of postherpetic neuralgia after a first episode of herpes zoster: prospective study with long term follow up. BMJ. 2000;321:794–796
    • View In Article
    • MEDLINE
    • CrossRef
12. Oster G, Harding G, Dukes E, Edelsberg J, Cleary PD. Pain, medication use, and health-related quality of life in older persons with postherpetic neuralgia: results from a population-based survey. J Pain. 2005;6:356–363
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (172 KB)
    • CrossRef
13. van Seventer R, Sadosky A, Lucero M, Dukes E. A cross-sectional survey of health state impairment and treatment patterns in patients with postherpetic neuralgia. Age Ageing. 2006;35:132–137
    • View In Article
    • MEDLINE
    • CrossRef
14. Melzack R. The short-form McGill Pain Questionnaire. Pain. 1987;30:191–197
    • View In Article
    • Abstract
    • Full-Text PDF (468 KB)
    • CrossRef
15. Hays RD, Martin SA, Sesti AM, Spritzer KL. Psychometric properties of the Medical Outcomes Study Sleep measure. Sleep Med. 2005;6:41–44
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (78 KB)
    • CrossRef
16. Max MB, Kishore-Kumar R, Schafer SC, et al. Efficacy of desipramine in painful diabetic neuropathy: a placebo-controlled trial. Pain. 1991;45:3–9discussion 1–2
    • View In Article
    • Abstract
    • Full-Text PDF (838 KB)
    • CrossRef
17. Watson C. Neuropathic pain syndromes. New York, NY: Raven Press, Ltd.; 1991;
    • View In Article
18. Jensen MP, Karoly P. Pain-specific beliefs, perceived symptom severity, and adjustment to chronic pain. Clin J Pain. 1992;8:123–130
    • View In Article
    • MEDLINE
    • CrossRef
19. Rosenstock J, Tuchman M, LaMoreaux L, Sharma U. Pregabalin for the treatment of painful diabetic peripheral neuropathy: a double-blind, placebo-controlled trial. Pain. 2004;110:628–638
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (242 KB)
    • CrossRef
20. Lesser H, Sharma U, LaMoreaux L, Poole RM. Pregabalin relieves symptoms of painful diabetic neuropathy: a randomized controlled trial. Neurology. 2004;63:2104–2110
    • View In Article
21. Richter RW, Portenoy R, Sharma U, et al. Relief of painful diabetic peripheral neuropathy with pregabalin: a randomized, placebo-controlled trial. J Pain. 2005;6:253–260
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (201 KB)
    • CrossRef
22. Freynhagen R, Strojek K, Griesing T, Whalen E, Balkenohl M. Efficacy of pregabalin in neuropathic pain evaluated in a 12-week, randomised, double-blind, multicentre, placebo-controlled trial of flexible- and fixed-dose regimens. Pain. 2005;115:254–263
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (139 KB)
    • CrossRef
23. Frampton JE, Scott LJ. Pregabalin: in the treatment of painful diabetic peripheral neuropathy. Drugs. 2004;64:2813–2820discussion 2821
    • View In Article
    • MEDLINE
    • CrossRef
24. Guy W. ECDEU assessment manual for psychopharmacology. Washington, DC: Department of Health, Education and Welfare; 1976;
    • View In Article
25. Fischer D, Stewart AL, Bloch DA, et al. Capturing the patient's view of change as a clinical outcome measure. JAMA. 1999;282:1157–1162
    • View In Article
    • MEDLINE
    • CrossRef
26. Collins SL, Edwards J, Moore RA, Smith LA, McQuay HJ. Seeking a simple measure of analgesia for mega-trials: is a single global assessment good enough?. Pain. 2001;91:189–194
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (204 KB)
    • CrossRef
27. Farrar JT, Young JP, LaMoreaux L, Werth JL, Poole RM. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain. 2001;94:149–158
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (264 KB)
    • CrossRef
28. Ware JE, Snow KK, Kosinski MK, Gandek B. SF-36 Health Survey: Manual and interpretation guide. Boston, MA: The Health Institute, New England Medical Center; 1993;
    • View In Article
29. McNair D, Lorr M, Droppleman L. Manual for the Profile of Mood States. San Diego, CA: Educational and Industrial Testing Services; 1971;
    • View In Article
30. Rabin R, de Charro F. EQ-5D: a measure of health status from the EuroQol Group. Ann Med. 2001;33:337–343
    • View In Article
    • MEDLINE
    • CrossRef
31. Hays RD, Revicki DA. Reliability and validity (including responsiveness). In:  Fayers P,  Hays RD editor. Assessing quality of life in clinical trials: Methods and practice. 2nd ed.. Oxford: Oxford University Press; 2005;p. 25–39
    • View In Article