## Abstract

Changes in health policy have underlined the importance of evidence-based clinical practice and rigorous evaluation of patient-based outcomes. As patient-based outcome measurement is particularly important in treatment trials of multiple sclerosis, a number of disease-specific instruments have been developed recently. One limitation of these instruments is that none was developed using the standard psychometric approach of reducing a large item pool generated from people with multiple sclerosis. Consequently, an outcome measure for clinical trials of multiple sclerosis that is disease specific and combines patient perspective with rigorous psychometric methods will complement existing instruments. The aim of this study was to develop such a measure. Standard psychometric methods were used. A pool of 129 questionnaire items was generated from interviews with 30 people with multiple sclerosis, expert opinion and literature review. The questionnaire was administered by postal survey to 1530 people selected randomly from the Multiple Sclerosis Society membership database. Redundant items and those with limited measurement properties were removed. The remaining items (n = 41) were grouped into scales using factor analysis, and then refined to form the Multiple Sclerosis Impact Scale (MSIS-29), an instrument measuring the physical (20 items) and psychological (nine items) impact of multiple sclerosis. Five psychometric properties of the MSIS-29 (data quality, scaling assumptions, acceptability, reliability and validity) were examined in a separate postal survey of 1250 Multiple Sclerosis Society members. A preliminary responsiveness study of the MSIS-29 was undertaken in 55 people admitted for rehabilitation and intravenous steroid treatment of relapses. The MSIS-29 satisfied all psychometric criteria. Data quality was excellent, missing data were low (maximum 3.9%), item test–re-test reliability was high (r = 0.65–0.90) and scale scores could be generated for >98% of respondents. Item descriptive statistics, item convergent and discriminant validity, and factor analysis indicated that it was legitimate to generate scores for MSIS-29 scales by summing items. MSIS-29 scales showed good variability, small floor and ceiling effects, high internal consistency (Cronbach's alpha ≤0.91) and high test–re-test reliability (intraclass correlation ≤0.87). Correlations with other measures and the analysis of group differences provided evidence that the MSIS-29 measures the physical and psychological impact of multiple sclerosis. Effect sizes (physical scale = 0.82, psychological scale = 0.66) demonstrated preliminary evidence of good responsiveness. These results indicate the MSIS-29 is a clinically useful and scientifically sound patient-based outcome measure of the impact of multiple sclerosis suitable for clinical trials and epidemiological studies.

## Introduction

Changes in health policy have underlined the importance of evidence-based clinical practice and the need to evaluate outcomes that are important to patients. These changes are particularly relevant to multiple sclerosis, a chronic, disabling, condition of young people for which a number of costly interventions are available that purport to improve quality of life. As decisions about the effectiveness of these treatments influence patient welfare and the expenditure of public funds, it is essential that evaluations are based on scientifically rigorous outcome measures. If treatments are to be evaluated using outcomes that are important to patients, and are intended to incorporate their perspective, the instruments used should be developed from and completed by patients.

Over the last two decades, outcome measurement in multiple sclerosis has relied heavily on the Expanded Disability Status Scale (EDSS) (Kurtzke, 1983). Although the EDSS evaluates disability, it was developed before psychometric methods became widely available to clinicians, was not based on recognized techniques of scale construction (Nunnally and Bernstein, 1994) and did not directly involve people with multiple sclerosis. Moreover, the EDSS is rated by neurologists rather than by patients themselves and has limited measurement properties (Sharrack et al., 1999; Hobart et al., 2000b).

The lack of validated multiple sclerosis-specific measures has led to the use of generic measures, such as the Medical Outcomes Study 36-item Short-Form Health Survey (SF-36) (Ware et al., 1993), which have the advantage of enabling comparisons across diseases. However, generic measures may fail to address important areas of impact that are disease specific (Peto et al., 1995) or may have limited responsiveness (Patrick and Deyo, 1989). Psychometric limitations of the SF-36 in multiple sclerosis include significant floor and ceiling effects (Freeman et al., 2000), limited responsiveness (Freeman et al., 2000), underestimation of mental health problems (Nortvedt et al., 2000) and a failure to satisfy assumptions about scaling summary scores (Hobart et al., 2000a).

A number of multiple sclerosis-specific measures have been developed in the last 5 years. These include the Functional Assessment of Multiple Sclerosis (FAMS) (Cella et al., 1996), the MSQOL-54 (Vickrey et al., 1995), the Multiple Sclerosis Functional Composite (MSFC) (Rudick et al., 1997), the Guy's (now UK) Neurological Disability Scale (GNDS/UKNDS) (Sharrack and Hughes, 1999), the Multiple Sclerosis Quality of Life Inventory (MSQLI) (LaRocca et al., 1996) and the Health-Related Quality of Life Questionnaire for Multiple Sclerosis (HRQOL-MS) (Pfennings et al., 1999). One of the limitations of these disease-specific measures is that none was developed using the standard psychometric approach of reducing an item pool generated de novo from people with multiple sclerosis. The FAMS and MSQOL-54 were developed by adding multiple sclerosis-specific items to existing measures, an approach that has some limitations (Freeman et al., 1999). The HRQOL-MS was developed from the statistical analysis of items from two generic and one multiple sclerosis-specific measure, whilst the MSQLI combines a large number of existing disease-specific and generic instruments. Items for the GNDS were developed through expert clinical opinion rather than on the basis of interviews with people with multiple sclerosis. Consequently, an outcome measure for clinical trials that is multiple sclerosis specific and combines patient perspective with rigorous psychometric methods will complement existing instruments. The aim of this study was to develop such a measure.

## Method

### Overview

The MSIS-29 was developed in three stages. First, a 129-item questionnaire was generated from 30 patient interviews, expert opinion and literature review. The questionnaire was then administered by postal survey to 1530 randomly selected members of the Multiple Sclerosis Society of Great Britain and Northern Ireland; standard item reduction techniques were used to develop a 29-item scale measuring the physical (20 items) and psychological (nine items) impact of multiple sclerosis (see Appendix I). Finally, the psychometric properties of the MSIS-29 (i.e. data quality, scaling assumptions, acceptability, reliability and validity) were evaluated in an independent sample of 1250 members of the Multiple Sclerosis Society. A preliminary study of the responsiveness of the MSIS-29 has been conducted in 55 in-patients at the National Hospital for Neurology and Neurosurgery (NHNN). The ethics committee of the NHNN approved the study.

### Item generation

An initial pool of 129 items concerning the health impact of multiple sclerosis was generated from three sources: semi-structured interviews of people with multiple sclerosis, multidisciplinary expert opinion and a comprehensive literature review. All 30 people with multiple sclerosis who were invited for interview, selected to represent the diversity of the illness, agreed to participate. Interviews lasted an average of 1 h, were tape recorded, transcribed and then content analysed. Statements concerning the health impact of multiple sclerosis were extracted, grouped into themes and examined for redundancy by the study team. A total of 3750 health impact statements were extracted from the interviews (mean 125; range 64–212). These statements generated 91 questionnaire items. Although no new themes appeared after the first 20 interviews, all 30 were analysed. A further 38 items were generated from interviews with health professionals at the NHNN (i.e. neurologists, neuropsychologists, nurses, occupational therapists, physiotherapists, social workers, and speech and language therapists) who were involved in the care of people with multiple sclerosis and from a comprehensive literature review.

Examination of the content of the 129 items indicated that two distinct question stems and response scales were required. The majority of items (n = 97) were best represented by the stem How much have you been bothered by …' with a five-point response option (1 = not at all; 5 = extremely). The remaining items (n = 32) that referred specifically to activity limitations were best represented by the stem How much has your multiple sclerosis limited your ability to …' with a six-point response option (1 = not at all limited; 6 =unable to do this activity). The time frame for all questions was the preceding 2 weeks.

The preliminary 129-item questionnaire was reviewed for content, wording and clinical appropriateness by patients and clinicians who were involved in its development. It was then pre-tested formally in an independent and heterogeneous sample of 20 people with multiple sclerosis who were attending the NHNN. They identified items and instructions that were unclear, ambiguous, irrelevant, misleading or offensive, and made suggestions for alterations to the questionnaire.

### Item reduction and development of scales (first field test)

The 129-item questionnaire was administered by postal survey to 1530 people, randomly selected and geographically stratified, from the membership database of the Multiple Sclerosis Society of Great Britain and Northern Ireland. This sampling frame has the advantage of being truly representative. The disadvantage is that not all members have multiple sclerosis. Therefore, based on results of a pilot study (Hobart et al., 2000c), we chose a target sample size of 1530 to ensure 500 completed questionnaires with no missing data. A subsample of 400 people was randomly selected from the larger sample to study item test–retest reproducibility to ensure 125 completed questionnaires on two occasions with no missing data. Patients in the test–retest sample received two questionnaires in the same envelope: one to complete immediately (time 1) and a second in a sealed envelope with instructions to open and complete 10 days later (time 2). A postcard reminder to complete the time 2 questionnaire was sent at day 7. Non-responders received reminders (letter and questionnaire) at 2 and 4 weeks (Dillman, 1978). In the test–retest subsample, non-responders to the time 2 questionnaire did not receive a reminder.

#### Item reduction

The following psychometric properties (descriptive statistics and reliability estimates) were examined for each item: percentage missing data; frequency distributions for each response option; maximum endorsement frequency (response option most frequently endorsed); mean score, standard deviation, skewness, floor and ceiling effects; and test–retest reproducibility (product–moment correlations). Items with >10% missing data were eliminated (WHOQOL Group, 1998). Correlations among the items were then examined to identify redundant items (item–item correlations ≥0.70; Juniper et al., 1997). For each item–item correlation ≥0.70, the item with the least favourable psychometric properties was eliminated. When items had similar psychometric properties, a consensus clinical decision determined which item to retain. Finally, the psychometric properties of the remaining items were examined. Items were eliminated if: floor effects, ceiling effects or maximum endorsement frequencies exceeded 40%; the sum of the endorsement frequencies for any two adjacent item response categories was <10% (WHOQOL Group, 1998); or if item test–retest reproducibility was <0.50 (Duruoz et al., 1996).

#### Development of scales

Scales were developed using an iterative process. First, all items were entered into a principal components analysis without rotation to determine whether there were any rogue items that should be eliminated (Ferguson and Cox, 1993). Next, principal axis factoring with varimax rotation was undertaken (Fayers and Machin, 1998). Multiple criteria were used to determine how many factors to rotate: Eigenvalues exceeding unity (Guttman, 1954); the scree test (Cattell, 1966); the 5% rule (Guertin and Bailey, 1970); and trial rotations (Ware et al., 1980). All potential factor solutions were examined for cross loading [items loading on two or more factors by >0.40, and items loading on two or more factors within 0.1 of each other (Ferguson and Cox, 1993)], clinical interpretability of item content and replicability of results in random split half samples. Item groups modelled through factor analysis were then examined to determine if they satisfied recommended criteria for summed rating scales and were acceptable, reliable and valid (methods described below).

### Psychometric evaluation of the MSIS-29 (second field test)

Item reduction analyses produced a 29-item measure that includes two scales: physical impact (20 items) and psychological impact (nine items). All items could be referenced back to statements made by patients during the interviews. The two summary scores are generated by summing individual items and then transformed to a 0–100 scale. High scores indicate worse health. For respondents with missing data, but where at least 50% of the items in a scale had been completed, a respondent-specific mean score computed from the completed items was imputed (Ware et al., 1993).

The psychometric properties of the MSIS-29 were evaluated comprehensively in two independent samples. A second and separate postal survey of randomly selected and geographically stratified members of the Multiple Sclerosis Society (n = 1250) was undertaken to evaluate data quality, scaling assumptions, acceptability, reliability and validity. Responsiveness was evaluated in 55 people with multiple sclerosis admitted to the NHNN for in-patient rehabilitation or intravenous steroids for multiple sclerosis relapses.

#### Postal survey

The postal survey sample was divided randomly into three subsamples (n = 500, 500 and 250). Respondents in the two larger subsamples completed the MSIS-29, demographic questions and three other health measures. Respondents in sample 1 completed the SF-36, EuroQol (EQ-5D) (EuroQol Group, 1990) and postal Barthel Index (BI) (Gompertz et al., 1994), whilst respondents in sample 2 completed the FAMS, EQ-5D and 12-item version of the General Health Questionnaire (GHQ-12) (Goldberg and Hillier, 1979). Respondents in the smaller test–retest subsample completed the MSIS-29 on two occasions separated by a 10-day interval. The survey methods were the same as those used in the first field test.

Five psychometric properties of the MSIS-29 were evaluated using standard methods (Nunnally and Bernstein, 1994; Streiner and Norman, 1995; Lohr et al., 1996). Data quality (McHorney et al., 1994) was determined by calculating the percentage missing data for items, percentage computable scale scores and item test–retest reproducibility (intraclass correlation coefficient, ICC; Bartko, 1966). Scaling assumptions examine whether it is legitimate to generate scores by summing items without weighting or standardization, and whether items are grouped correctly into scales. Items can be summed to generate scores when items have similar response option frequency distributions, equivalent mean scores and variances, and substantial (r > 0.30) and equivalent item–total correlations (Likert, 1932). Items are grouped correctly into scales when item–own scale correlations exceed item–other scale correlations by at least two standard errors (1/√n; Ware et al., 1997), and when the results of factor analysis support hypothesized item groups.

Acceptability was determined by examining score distributions. Acceptability is supported when observed scores are well distributed (Stewart and Ware, 1992), mean scores are near the scale mid-point (Eisen et al., 1979), floor and ceiling effects are <20% (McHorney and Tarlov, 1995) and skewness statistics range from –1 to +1 (Holmes et al., 1996). Two types of reliability, internal consistency (Cronbach's alpha coefficients; Cronbach, 1951) and scale test–retest reproducibility (ICC), were examined. Estimates should exceed 0.80 (Nunnally and Bernstein, 1994).

The aim of the validity studies was to examine evidence that the MSIS-29 was a measure of the physical and psychological impact of multiple sclerosis. Three types of validity were examined. Internal validity (Bohrnstedt, 1983) was determined by examining the intercorrelation between MSIS-29 scales. A moderate correlation (r = 0.30–0.70) was predicted. Convergent and discriminant validity (Cronbach and Meehl, 1955) was determined by examining the extent to which correlations between MSIS-29 scales and other measures (SF-36, BI, EQ-5D, FAMS and GHQ) and variables (age, sex and duration of multiple sclerosis) were consistent with predictions. For example, we predicted that the MSIS-29 physical impact scale would correlate highly (r > 0.70) with other measures of physical health (e.g. SF-36 physical functioning dimension, BI, FAMS mobility scale and EQ-5D mobility dimension). Group differences validity was determined by examining MSIS-29 scores for groups of patients. We predicted that: people who were retired due to their multiple sclerosis would have higher scores than people who were still employed; people with increasing difficulties in mobility and self-care as defined by the EQ-5D would have greater differences in their physical scores than their psychological scores; people with increasing anxiety or depression as defined by the EQ-5D would have greater differences in their psychological scores than their physical scores; men and women would have similar scores; and people with or without a degree would have similar scores.

#### Responsiveness study

A preliminary responsiveness study has been undertaken in consecutive admissions to the NHNN between February 1 and August 1, 2000 for rehabilitation and intravenous steroid treatment. People were excluded if they appeared to have severe cognitive impairment substantiated by neuropsychological testing. People admitted for rehabilitation completed the MSIS-29 on admission and discharge, whilst those admitted for intravenous steroid treatment completed the MSIS-29 on admission and 6 weeks later. Responsiveness was determined by calculating effect sizes (ES; Kazis et al., 1989), mean change score (admission minus discharge) divided by the standard deviation of admission scores. These are interpreted (Cohen, 1969) as either small (ES < 0.20), medium (ES = 0.50) or large (ES > 0.80). The statistical significance of the change scores was determined using paired sample t tests (Deyo et al., 1991).

## Results

### Item generation, item reduction and development of scales

The remaining 41 items were entered into a principal components analysis. All items loaded onto the first component by >0.40, indicating a common underlying dimension. Neither principal components analysis nor principal axis factoring indicated a clear solution. Therefore, all solutions with two to seven factors were evaluated. The two-factor solution was judged to be the most appropriate. However, three items that loaded on both factors by >0.40, indicating a limited ability to discriminate between the two factors, were removed. Whilst the clinical interpretation of the two factors led these to be labelled the physical (25 items) and psychological (13 items) impact of multiple sclerosis, the consensus opinion of the investigators was that five items were not entirely consistent with this interpretation. These items were thus removed, resulting in a 33-item instrument with two scales: physical impact (22 items) and psychological impact (11 items).

Preliminary psychometric evaluation of the MSIS-33 indicated that three items had similar correlations with the two scales and, therefore, were considered probable scaling failures. These items were removed to minimize measurement overlap between the two scales (Ware et al., 1997). When the psychometric properties of the 30-item measure were re-tested, all criteria were satisfied except one item that was classified as a probable scaling failure. This item was therefore removed to produce the final 29-item MSIS (20-item physical scale; nine-item psychological scale). The MSIS-29 includes 26 items with five-point response options and three items with six-point response options. The latter three items were re-scaled (category 5 combined with 6) so that all items have the same number of response options. Preliminary psychometric analyses, based on data collected in the first field test, indicated that the MSIS-29 satisfied standard criteria for acceptability, reliability and validity (results not reported).

### Psychometric evaluation of the MSIS-29

#### Postal survey

A total of 1023 (81.8%) questionnaires were returned, of which 310 were returned blank (change of address or deceased n = 63, did not have multiple sclerosis n = 155, did not wish to participate n = 64, no reason given n = 28). The second postal survey generated data for 713 people giving a response rate of 69.1% (1023 – 310/1250 – 63 – 155) that was similar to the first field test. In the test–retest subsample, 90.6% (n = 136) of people who returned the time 1 questionnaire returned the time 2 questionnaire. The characteristics of samples for the first and second field tests were similar (Table 1). There were no significant differences in demographic characteristics between patients in the three subsamples.

##### Data quality (Table 2).

Missing data for items were low (range 1.1–3.6%). Eighty-four per cent of respondents endorsed all 29 items (100% complete data), 8.4% of respondents missed out one item and 3.2% of respondents missed out two items. Ninety-seven per cent of respondents had ≥90% complete data. Therefore, MSIS-29 scale scores could be computed for 703 respondents (98.6%). Item test–retest reproducibility was high. These results indicate that data quality was high.

##### Scaling assumptions (Table 2).

Frequency distributions for item response scales were quite symmetrical and not unduly skewed (range –0.86 to +0.41), and items within each scale had similar mean scores and standard deviations. All item–own scale correlations were high (range 0.49–0.84) and exceeded item–other scale correlations by at least two standard errors (range 0.12–0.39). Principal axis factoring of the 29 items, cross-validated in random split half samples, generated two factors whose item contents were consistent with the hypothesized physical and psychological scales. These results indicate that the MSIS-29 satisfied tests of scaling assumptions.

##### Acceptability and reliability (Table 2).

Scale scores spanned the entire scale range and were not notably skewed, mean scores were near the scale mid-point, and floor and ceiling effects were negligible (maximum 3.9%). Internal consistency and test–retest reproducibility exceeded the recommended criterion for group comparisons of 0.80. There were no statistically significant differences in MSIS-29 scores between the three subsamples and between time 1 and time 2 scores for the test–retest reproducibility subsample. These results indicate that the MSIS-29 satisfied criteria for acceptability and reliability.

## Discussion

The aim of this study was to develop a multiple sclerosis-specific outcome measure that combines the patient perspective with a rigorous scientific approach. We tried to achieve these aims by generating items from in-depth patient interviews, using the self-report method of administration, selecting items on the basis of psychometric performance in a large field test and rigorously applying psychometric methods. In the samples we have studied, the MSIS-29 satisfies criteria as a summed rating scale and is acceptable, reliable and valid. Furthermore, there is preliminary evidence that the MSIS-29 detects change. Finally, all items could be referenced back to statements made by patients during the interviews.

The MSIS-29 is a measure of the physical and psychological impact of multiple sclerosis from the patients' perspective. This description has been chosen as it best defines the health constructs that we intended to measure, and because the terms health-related quality of life and disablement, both of which could be used to categorize the MSIS-29, have several different definitions (Fitzpatrick et al., 1998). We feel it is important to be as specific as possible when defining the purpose of a measure, to guide prospective users of any scale.

Stringent criteria for item selection were adopted in an attempt to develop an instrument with strong psychometric properties. In order to create a responsive scale, items were selected that discriminated well between individuals, while items with maximum endorsement frequencies >40% were eliminated. Similarly, in order to reduce overlap between the two MSIS-29 scales, we eliminated items with limited item convergent and discriminant validity. Such a rigorous approach to health measurement is important because the results of studies are dependent on the quality of the measures used for data collection, and the limitations of measures cannot be overcome easily by improvements in study design and powerful statistical methods (Fleiss, 1986). Results concerning the responsiveness of the MSIS-29 must be considered preliminary due to the small sample size, and further evaluations of responsiveness are needed in different samples and settings.

There are potential limitations in using the Multiple Sclerosis Society membership database to define our sampling frame. It is known that not all members have multiple sclerosis (calculations based on our postal surveys estimate this to be a minimum of 56%), and that many members of the Multiple Sclerosis Society are partners, friends or relatives of people with multiple sclerosis. Therefore, we specifically asked people who did not have multiple sclerosis to tick a box on the front of the questionnaire and return it blank. However, the percentage of people in the database with a neurologist-confirmed diagnosis of clinically definite multiple sclerosis, the disease type of those with multiple sclerosis and the representativeness of people who join charitable groups is unknown. Our estimates indicate, however, that we have randomly sampled from ~35% (28 000) of the total UK multiple sclerosis population.

As the psychometric properties of health measurement instruments are sample dependent and cannot be established in a single study (Stewart et al., 1988), further evaluations of the MSIS-29 are needed. Critical evaluations in different settings will define the strengths and weaknesses of the MSIS-29, further define its role in clinical practice and research, and help to determine whether the development process may have been biased by people without multiple sclerosis completing the questionnaire. As traditional psychometric methods were used to develop and evaluate the MSIS-29, it is also important that newer psychometric methods such as Rasch (Rasch, 1960) and Item Response Theory (Lord and Novick, 1968) models are used to evaluate the MSIS-29. Finally, comparisons between the MSIS-29 and widely used measures for multiple sclerosis, such as the Multiple Sclerosis Functional Composite (Cutter et al., 1999) and GNDS (Sharrack and Hughes, 1999), should be undertaken. These studies will determine the advantages and disadvantages of different instruments, how they complement each other, and provide an evidence-based framework to guide the selection of outcome measures for research and audit. Over time, the accumulation of such data will also establish normative values and content-based interpretation of scores and score changes (McDowell and Jenkinson, 1996).

This study produced some unexpected results. Only two distinct dimensions of health, physical and psychological impact, appear to underlie the diverse 129-item pool. Although many other dimensions of health such as symptoms were included in the initial version of the questionnaire, psychometric analyses did not support multiple dimensions. These results support previous findings (Ware et al., 1994; Pfennings et al., 1999) that have observed that a two-dimensional model, consisting of physical and psychological health, appears to underpin the construct of subjective health status. Another unexpected result is that the MSIS-29 physical and psychological scales have the same correlation with the FAMS emotional well-being scale. However, it is encouraging to note that correlations between the MSIS-29 physical scale and other measures of psychological distress (GHQ, SF-36 mental health dimension and EQ-5D anxiety/depression dimension) are low to moderate.

This study has important implications for clinical trials and epidemiological studies. The MSIS-29 can be used in cross-sectional studies to describe the impact of multiple sclerosis, in longitudinal studies to monitor the natural history of the disorder and, most importantly, in clinical trials to evaluate therapeutic effectiveness from the patients' perspective. Furthermore, the availability of reliable, valid and responsive patient-based outcome measures is central to an improved understanding of the impact of multiple sclerosis and its relationships with other indicators of disease activity, such as neuroimaging and neurophysiology.

In addition to physical and psychological impact scores, an overall impact score could be reported as the total scale satisfies criteria as a summed rating scale. Although a single summary score would simplify data analysis, we do not recommend use of an overall summary score for clinical trials or epidemiological studies. This is because evidence indicates that the two scales are measuring related but distinct constructs (intercorrelation between scales = 0.62; factor analysis supports two dimensions). Combining these distinct aspects of outcome into an overall score could mask important (and possibly opposite) differential effects of treatment on physical and psychological health.

## Appendix I

### Multiple Sclerosis Impact Scale (MSIS-29)

• The following questions ask for your views about the impact of MS on your day-to-day life during the past two weeks

• For each statement, please circle the one number that best describes your situation

• Please answer all questions

In the past two weeks, how much has your MS limited your ability to… Not at all A little Moderately Quite a bit Extremely
1. Do physically demanding tasks?
2. Grip things tightly (e.g. turning on taps)?
3. Carry things?
In the past two weeks, how much has your MS limited your ability to… Not at all A little Moderately Quite a bit Extremely
1. Do physically demanding tasks?
2. Grip things tightly (e.g. turning on taps)?
3. Carry things?
In the past two weeks, how much have you been bothered by… Not at all A little Moderately Quite a bit Extremely
Please check that you have answered all the questions before going on to the next page
®2000 Neurological Outcome Measures Unit
4. Problems with your balance?
5. Difficulties moving about indoors?
6. Being clumsy?
7. Stiffness?
8. Heavy arms and/or legs?
9. Tremor of your arms or legs?
10. Spasms in your limbs?
11. Your body not doing what you want it to do?
12. Having to depend on others to do things for you?
In the past two weeks, how much have you been bothered by… Not at all A little Moderately Quite a bit Extremely
Please check that you have answered all the questions before going on to the next page
®2000 Neurological Outcome Measures Unit
4. Problems with your balance?
5. Difficulties moving about indoors?
6. Being clumsy?
7. Stiffness?
8. Heavy arms and/or legs?
9. Tremor of your arms or legs?
10. Spasms in your limbs?
11. Your body not doing what you want it to do?
12. Having to depend on others to do things for you?
In the past two weeks, how much have you been bothered by… Not at all A little Moderately Quite a bit Extremely
®2000 Neurological Outcome Measures Unit
Please check that you have circled ONE number for EACH question
Copies of the scale can be obtained from the corresponding author.
13. Limitations in your social and leisure activities at home?
14. Being stuck at home more than you would like to be?
15. Difficulties using your hands in everyday tasks?
16. Having to cut down the amount of time you spent on work or other daily activities?
17. Problems using transport (e.g. car, bus, train, taxi, etc.)?
18. Taking longer to do things?
19. Difficulty doing things spontaneously (e.g. going out on the spur of the moment)?
20. Needing to go to the toilet urgently?
21. Feeling unwell?
22. Problems sleeping?
23. Feeling mentally fatigued?
24. Worries related to your MS?
25. Feeling anxious or tense?
26. Feeling irritable, impatient, or short tempered?
27. Problems concentrating?
28 Lack of confidence?
29. Feeling depressed?
In the past two weeks, how much have you been bothered by… Not at all A little Moderately Quite a bit Extremely
®2000 Neurological Outcome Measures Unit
Please check that you have circled ONE number for EACH question
Copies of the scale can be obtained from the corresponding author.
13. Limitations in your social and leisure activities at home?
14. Being stuck at home more than you would like to be?
15. Difficulties using your hands in everyday tasks?
16. Having to cut down the amount of time you spent on work or other daily activities?
17. Problems using transport (e.g. car, bus, train, taxi, etc.)?
18. Taking longer to do things?
19. Difficulty doing things spontaneously (e.g. going out on the spur of the moment)?
20. Needing to go to the toilet urgently?
21. Feeling unwell?
22. Problems sleeping?
23. Feeling mentally fatigued?
24. Worries related to your MS?
25. Feeling anxious or tense?
26. Feeling irritable, impatient, or short tempered?
27. Problems concentrating?
28 Lack of confidence?
29. Feeling depressed?
Table 1

Characteristics of samples

Variable* Sample Semi-structured interviews First field test Second field test Responsiveness
*All values are percentages unless specified otherwise; for whom both physical and psychological scale scores could be computed; n = 27 admitted for in-patient rehabilitation and n = 28 admitted for intravenous steroids; §question not asked.
n 30 766 713 55
Gender
Female 56 74 71 66
Age
Mean (SD) 41 (12) 51 (12) 52 (12) 45 (13)
Range 23–70 23–87 18–82 23–83
Ethnicity
White 100 98 98 95
Years since MS onset
Mean (SD) 12 (11) 19 (12) 19 (11) 16 (12)
Range 1–36 1–56 1–59 1–60
Mobility indoors
Walks unaided 40 § 32 24
Walks with an aid 23 – 40 49
Uses a wheelchair 37 – 28 27
Mobility
Can walk N/A 79 –
Cannot walk N/A 21 –
Marital status
Married 77 66 70 64
Living with others  83 81 82
Employment status
Retired due to MS 63 54 56 31
Employed  18 19 44
Type of MS (%)
Primary progressive 13.3 Unknown Unknown 5.5
Secondary progressive 43.4 Unknown Unknown 47.3
Relapsing–remitting 43.3 Unknown Unknown 47.3
Variable* Sample Semi-structured interviews First field test Second field test Responsiveness
*All values are percentages unless specified otherwise; for whom both physical and psychological scale scores could be computed; n = 27 admitted for in-patient rehabilitation and n = 28 admitted for intravenous steroids; §question not asked.
n 30 766 713 55
Gender
Female 56 74 71 66
Age
Mean (SD) 41 (12) 51 (12) 52 (12) 45 (13)
Range 23–70 23–87 18–82 23–83
Ethnicity
White 100 98 98 95
Years since MS onset
Mean (SD) 12 (11) 19 (12) 19 (11) 16 (12)
Range 1–36 1–56 1–59 1–60
Mobility indoors
Walks unaided 40 § 32 24
Walks with an aid 23 – 40 49
Uses a wheelchair 37 – 28 27
Mobility
Can walk N/A 79 –
Cannot walk N/A 21 –
Marital status
Married 77 66 70 64
Living with others  83 81 82
Employment status
Retired due to MS 63 54 56 31
Employed  18 19 44
Type of MS (%)
Primary progressive 13.3 Unknown Unknown 5.5
Secondary progressive 43.4 Unknown Unknown 47.3
Relapsing–remitting 43.3 Unknown Unknown 47.3
Table 2

Data quality, scaling assumptions, acceptability, reliability and responsiveness of the MSIS-29

Psychometric property MSIS-29 scale Physical impact Psychological impact
*Intraclass correlation coefficient; percentage of times where item–own scale correlation exceeds item–other scale correlation by at least 2 SE (2 × 1/÷n); time 1 minus time 2; §mean change score divided by standard deviation of time 1 score.
Data quality (n = 713)
Item missing data % 1.7–3.6 1.1–1.8
Item test–retest reproducibility*: range 0.65–0.90 0.72–0.82
(mean) (0.81) (0.78)
Computable scale scores % 98.0 98.7
Scaling assumptions (n = 703)
Item mean scores: range 2.54–3.83 2.57–3.28
Item SD: range 1.20–1.56 1.27–1.37
Item skewness: range –0.86 to +0.41 –0.29 to +0.40
Definite scaling successes 100% 100%
Item–own factor loading: range 0.58–0.85 0.47–0.79
Item–other factor loading: range 0.19–0.38 0.19–0.36
Acceptability (n = 703)
Possible score range 0–100 0–100
Observed score range 0–100 0–100
Mean score (SD) 56.0 (26.6) 45.5 (25.2)
Floor/ceiling effect % 0.9/3.9 1.7/1.9
Skewness –0.285 +0.172
Reliability
Cronbach's alpha (n = 703) 0.96 0.91
Scale test–re-test reproducibility (n = 128)* 0.94 0.87
Responsiveness (n = 55)
Time 1 score: mean (SD) 64.4 (23.0) 48.4 (26.7)
Time 2 score: mean (SD) 45.6 (23.4) 30.7 (22.3)
Change score: mean (SD); P 18.8 (19.6); <0.001 17.7 (24.6); <0.001
Effect size§ 0.82 0.66
Psychometric property MSIS-29 scale Physical impact Psychological impact
*Intraclass correlation coefficient; percentage of times where item–own scale correlation exceeds item–other scale correlation by at least 2 SE (2 × 1/÷n); time 1 minus time 2; §mean change score divided by standard deviation of time 1 score.
Data quality (n = 713)
Item missing data % 1.7–3.6 1.1–1.8
Item test–retest reproducibility*: range 0.65–0.90 0.72–0.82
(mean) (0.81) (0.78)
Computable scale scores % 98.0 98.7
Scaling assumptions (n = 703)
Item mean scores: range 2.54–3.83 2.57–3.28
Item SD: range 1.20–1.56 1.27–1.37
Item skewness: range –0.86 to +0.41 –0.29 to +0.40
Definite scaling successes 100% 100%
Item–own factor loading: range 0.58–0.85 0.47–0.79
Item–other factor loading: range 0.19–0.38 0.19–0.36
Acceptability (n = 703)
Possible score range 0–100 0–100
Observed score range 0–100 0–100
Mean score (SD) 56.0 (26.6) 45.5 (25.2)
Floor/ceiling effect % 0.9/3.9 1.7/1.9
Skewness –0.285 +0.172
Reliability
Cronbach's alpha (n = 703) 0.96 0.91
Scale test–re-test reproducibility (n = 128)* 0.94 0.87
Responsiveness (n = 55)
Time 1 score: mean (SD) 64.4 (23.0) 48.4 (26.7)
Time 2 score: mean (SD) 45.6 (23.4) 30.7 (22.3)
Change score: mean (SD); P 18.8 (19.6); <0.001 17.7 (24.6); <0.001
Effect size§ 0.82 0.66
Table 3

Convergent and discriminant construct validity of the MSIS-29

Instrument Scale/dimension/variable MSIS-29 scale* r (n
Physical Psychological
*Multiple Sclerosis Impact Scale: high scores = worst health; Pearson product–moment correlation coefficients; Medical Outcomes Study 36-item Short Form Health Survey (n = 263–280): high scores = best health; §Functional Assessment of Multiple Sclerosis (n = 233–259): high scores = best health; EuroQol: high scores = worst health (n = 520–550); **General Health Questionnaire (n = 248 and 249, respectively): high scores = worst health; ††high scores best health (n = 260 and 243, respectively).
SF-36 Physical functioning –0.79 –0.41
Role limitations physical –0.43 –0.40
Bodily pain –0.45 –0.50
General health perception –0.48 –0.53
Vitality –0.49 –0.55
Social functioning –0.64 –0.56
Role limitations emotional –0.29 –0.52
Mental health –0.41 –0.76
FAMS§ Mobility –0.88 –0.50
Symptoms –0.55 –0.64
Emotional well-being –0.68 –0.68
General contentment –0.64 –0.58
Thinking and fatigue –0.56 –0.73
Family/social well-being –0.37 –0.50
EQ-5D Mobility 0.61 0.23
Self-care 0.69 0.37
Usual activities 0.69 0.42
Pain/discomfort 0.44 0.43
Anxiety/depression 0.36 0.68
GHQ-12** Total score 0.46 0.68
Postal Barthel Index†† Total score –0.71 –0.35
Demographic variables Age (n = 678) 0.22 0.03
Sex (n = 686) 0.05 –0.05
Years since diagnosis (n = 629) 0.19 0.03
Instrument Scale/dimension/variable MSIS-29 scale* r (n
Physical Psychological
*Multiple Sclerosis Impact Scale: high scores = worst health; Pearson product–moment correlation coefficients; Medical Outcomes Study 36-item Short Form Health Survey (n = 263–280): high scores = best health; §Functional Assessment of Multiple Sclerosis (n = 233–259): high scores = best health; EuroQol: high scores = worst health (n = 520–550); **General Health Questionnaire (n = 248 and 249, respectively): high scores = worst health; ††high scores best health (n = 260 and 243, respectively).
SF-36 Physical functioning –0.79 –0.41
Role limitations physical –0.43 –0.40
Bodily pain –0.45 –0.50
General health perception –0.48 –0.53
Vitality –0.49 –0.55
Social functioning –0.64 –0.56
Role limitations emotional –0.29 –0.52
Mental health –0.41 –0.76
FAMS§ Mobility –0.88 –0.50
Symptoms –0.55 –0.64
Emotional well-being –0.68 –0.68
General contentment –0.64 –0.58
Thinking and fatigue –0.56 –0.73
Family/social well-being –0.37 –0.50
EQ-5D Mobility 0.61 0.23
Self-care 0.69 0.37
Usual activities 0.69 0.42
Pain/discomfort 0.44 0.43
Anxiety/depression 0.36 0.68
GHQ-12** Total score 0.46 0.68
Postal Barthel Index†† Total score –0.71 –0.35
Demographic variables Age (n = 678) 0.22 0.03
Sex (n = 686) 0.05 –0.05
Years since diagnosis (n = 629) 0.19 0.03
Table 4

MSIS-29 group differences and relative validity

Variable MSIS-29 score: mean (SD)
Physical Psychological
*One-way ANOVA with Duncan's post hoc comparisons; calculated as the ratio of paired F values using the largest as the denominator; independent samples t tests, equality of variances not assumed.
Employment status
Employed (n = 107) 30.6 (23.1) 31.1 (22.5)
Retired due to MS (n = 390) 64.3 (23.0) 49.9 (24.9)
Mean difference (P–33.7 (<0.001) –18.8 (<0.001)
EQ-5D mobility dimension
No problems in walking about (n = 61) 17.5 (17.2) 27.7 (23.1)
Some problems in walking about (n = 389) 56.4 (21.7) 46.6 (23.5)
Confined to bed (n = 70) 82.6 (16.8) 51.4 (28.1)
F (P)* 164.3 (<0.001) 19.3 (<0.001)
Relative validity  1.0  0.12
EQ-5D self-care dimension
No problems with self-care (n = 227) 35.5 (22.1) 34.5 (23.0)
Some problems with self-care (n = 235) 66.6 (16.9) 51.5 (22.0)
Unable to wash or dress myself (n = 76) 85.2 (15.3) 58.9 (27.3)
F (P256.7 (<0.001) 46.2 (<0.001)
Relative validity  1.0  0.18
EQ-5D anxiety/depression dimension
Not anxious or depressed (n = 229) 45.8 (27.4) 27.1 (17.8)
Moderately anxious or depressed (n = 277) 62.2 (22.7) 55.6 (19.3)
Extremely anxious or depressed (n = 38) 75.3 (22.7) 81.6 (16.3)
F (P39.5 (<0.001) 231.3 (<0.001)
Relative validity  0.17  1.0
Gender
Female (n = 489) 55.0 (26.9) 46.1 (25.8)
Male (n = 197) 58.1 (26.1) 43.5 (23.6)
Mean difference (P)  –3.1 (0.165)  2.6 (0.197)
Degree or professional qualification
Yes (n = 183) 53.2 (26.7) 41.6 (25.8)
No (n = 491) 56.7 (26.5) 46.8 (24.8)
Mean difference (P –3.5 (0.131)  –5.3 (0.133)
Variable MSIS-29 score: mean (SD)
Physical Psychological
*One-way ANOVA with Duncan's post hoc comparisons; calculated as the ratio of paired F values using the largest as the denominator; independent samples t tests, equality of variances not assumed.
Employment status
Employed (n = 107) 30.6 (23.1) 31.1 (22.5)
Retired due to MS (n = 390) 64.3 (23.0) 49.9 (24.9)
Mean difference (P–33.7 (<0.001) –18.8 (<0.001)
EQ-5D mobility dimension
No problems in walking about (n = 61) 17.5 (17.2) 27.7 (23.1)
Some problems in walking about (n = 389) 56.4 (21.7) 46.6 (23.5)
Confined to bed (n = 70) 82.6 (16.8) 51.4 (28.1)
F (P)* 164.3 (<0.001) 19.3 (<0.001)
Relative validity  1.0  0.12
EQ-5D self-care dimension
No problems with self-care (n = 227) 35.5 (22.1) 34.5 (23.0)
Some problems with self-care (n = 235) 66.6 (16.9) 51.5 (22.0)
Unable to wash or dress myself (n = 76) 85.2 (15.3) 58.9 (27.3)
F (P256.7 (<0.001) 46.2 (<0.001)
Relative validity  1.0  0.18
EQ-5D anxiety/depression dimension
Not anxious or depressed (n = 229) 45.8 (27.4) 27.1 (17.8)
Moderately anxious or depressed (n = 277) 62.2 (22.7) 55.6 (19.3)
Extremely anxious or depressed (n = 38) 75.3 (22.7) 81.6 (16.3)
F (P39.5 (<0.001) 231.3 (<0.001)
Relative validity  0.17  1.0
Gender
Female (n = 489) 55.0 (26.9) 46.1 (25.8)
Male (n = 197) 58.1 (26.1) 43.5 (23.6)
Mean difference (P)  –3.1 (0.165)  2.6 (0.197)
Degree or professional qualification
Yes (n = 183) 53.2 (26.7) 41.6 (25.8)
No (n = 491) 56.7 (26.5) 46.8 (24.8)
Mean difference (P –3.5 (0.131)  –5.3 (0.133)

We wish to thank the people with multiple sclerosis who freely participated in this study, the Multiple Sclerosis Society of Great Britain and Northern Ireland for their collaboration, Mr Peter Cardy and Dr Iain Smith (Advisory Committee), Ms Irene Richardson for undertaking the patient interviews and research assistance during the first field test, Dr Sarah Smith for assistance with the psychometric analyses during the item reduction process, Ms Sara Schroter and Mr Stefan Cano for their contribution to the item reduction process, and Ms Laura Camfield for research assistance during the second field test. The study was funded by a grant from the NHS Health Technology Assessment Programme, but the views and opinions expressed do not necessarily reflect those of the NHS Executive.

## References

Bartko JJ. The intraclass correlation coefficient as a measure of reliability.
Psychol Rep

1966
;
19
:
3
–11.
Bohrnstedt GW. Measurement. In: Rossi PH, Wright JD, Anderson AB, editors. Handbook of survey research. New York: Academic Press; 1983. p. 69–121.
Cattell RB. The scree test for the number of factors.
Multivar Behav Res

1966
;
1
:
245
–76.
Cella DF, Dineen K, Arnason B, Reder A, Webster KA, Karabatsos G, et al. Validation of the functional assessment of multiple sclerosis quality of life instrument.
Neurology

1996
;
47
:
129
–39.
Cohen J. Statistical power analysis for the behavioral sciences. New York: Academic Press; 1969.
Cronbach LJ. Coefficient alpha and the internal structure of tests.
Psychometrika

1951
;
16
:
297
–334.
Cronbach LJ, Meehl PE. Construct validity in psychological tests.
Psychol Bull

1955
;
52
:
281
–302.
Cutter GR, Baier ML, Rudick RA, Cookfair DL, Fischer JS, Petkau J, et al. Development of a multiple sclerosis functional composite as a clinical trial outcome measure.
Brain

1999
;
122
:
871
–82.
Deyo RA, Diehr P, Patrick DL. Reproducibility and responsiveness of health status measures: statistics and strategies for evaluation.
Control Clin Trials

1991
;
12
: (4 Suppl)
142s
–58s.
Dillman DA. Mail and telephone surveys: the total design method. New York: Wiley; 1978.
Duruoz MT, Poiraudeau S, Fermanian J, Menkes C-J, Amor B, Dougados M, et al. Development and validation of a rheumatoid hand functional disability scale that assesses functional handicap.
J Rheumatol

1996
;
23
:
1167
–72.
Eisen M, Ware JE Jr, Donald CA, Brook RH. Measuring components of children's health status.
Med Care

1979
;
17
:
902
–21.
EuroQoL Group. EuroQoL: a new facility for the measurement of health-related quality of life.
Health Policy

1990
;
16
:
199
–208.
Fayers PM, Machin D. Factor analysis. In: Staquet MJ, Hays RD, Fayers PM, editors. Quality of life assessment in clinical trials: methods and practice. Oxford: Oxford University Press; 1998. p. 191–223.
Ferguson E, Cox T. Exploratory factor analysis: a user's guide.
Int J Select Assess

1993
;
1
:
84
–94.
Fitzpatrick R, Davey C, Buxton MJ, Jones DR. Evaluating patient-based outcome measures for use in clinical trials. [Review].
Health Technol Assess

1998
;
2
:
i
–iv, 1–74.
Fleiss JL. The design and analysis of clinical experiments. New York: Wiley; 1986.
Freeman JA, Hobart JC, Langdon DW, Thompson AJ. Improving measurement scales: is adding items the answer? [abstract].
Ann Neurol

1999
;
46
:
507
.
Freeman JA, Hobart JC, Langdon DW, Thompson AJ. Clinical appropriateness: a key factor in outcome measure selection. The 36-item Short Form Health Survey in multiple sclerosis.
J Neurol Neurosurg Psychiatry

2000
;
68
:
150
–6.
Goldberg DP, Hillier VF. A scaled version of the General Health Questionnaire.
Psychol Med

1979
;
9
:
139
–45.
Gompertz P, Pound P, Ebrahim S. A postal version of the Barthel Index.
Clin Rehabil

1994
;
8
:
233
–9.
Guertin WH, Bailey JP Jr. Introduction to modern factor analysis. Ann Arbor (MI): Edwards Brothers; 1970.
Guttman LA. Some necessary conditions for common-factor analysis.
Psychometrika

1954
;
19
:
149
–61.
Hobart J, Freeman J, Lamping D, Fitzpatrick R, Thompson AJ. The medical outcomes study 36-item Short-Form Health Survey in multiple sclerosis: why assumptions must be tested [abstract].
Ann Neurol

2000
;
48
:
495
.
Hobart J, Freeman J, Thompson A. Kurtzke scales revisited: the application of psychometric methods to clinical intuition.
Brain

2000
;
123
:
1027
–40.
Hobart JC, Lamping DL, Fitzpatrick R, Thompson AJ. Patient based outcome measures for multiple sclerosis are needed and can be developed [abstract].
J Neurol Neurosurg Psychiatry

2000
;
69
:
420
.
Holmes W, Bix B, Shea J. SF-20 score and item distributions in a human immunodeficiency virus-seropositive sample.
Med Care

1996
;
34
:
562
–9.
Juniper EF, Guyatt GH, Streiner DL, King DR. Clinical impact versus factor analysis for quality of life questionnaire construction.
J Clin Epidemiol

1997
;
50
:
233
–8.
Kazis LE, Anderson JJ, Meenan RF. Effect sizes for interpreting changes in health status.
Med Care

1989
;
27
(3 Suppl):
S178
–89.
Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS).
Neurology

1983
;
33
:
1444
–52.
LaRocca NG, Ritvo PG, Miller DM, Fischer JS, Andrews H, Paty DW. `Quality of life' assessment in multiple sclerosis clinical trials: current status and strategies for improving multiple sclerosis clinical trial design. In: Goodkin DE, Rudick RA, editors. Multiple sclerosis: advances in clinical trial design, treatment and future perspectives. London: Springer-Verlag; 1996. p. 145–60.
Likert RA. A technique for the development of attitudes.
Arch Psychol

1932
;
140
:
5
–55.
Lohr KN, Aaronson NK, Alonso J, Burnam MA, Patrick DL, Perrin EB, et al. Evaluating quality-of-life and health status instruments: development of scientific review criteria.
Clin Ther

1996
;
18
:
979
–92.
Lord FM, Novick MR. Statistical theories of mental test scores. Reading (MA): Addison-Wesley; 1968.
McDowell I, Jenkinson C. Development standards for health measures.
J Health Serv Res Policy

1996
;
1
:
238
–46.
McHorney CA, Tarlov AR. Individual-patient monitoring in clinical practice: are available health status surveys adequate?
Qual Life Res

1995
;
4
:
293
–307.
McHorney CA, Ware JE Jr, Lu JF, Sherbourne CD. The MOS 36-Item Short-Form Health Survey (SF-36): III. Tests of data quality, scaling assumptions, and reliability across diverse patient groups.
Med Care

1994
;
32
:
40
–66.
Nortvedt MW, Riise T, Myer K-M, Nyland HI. Performance of the SF-36, SF-12, and RAND-36 summary scales in a multiple sclerosis population.
Med Care

2000
;
38
:
1022
–8.
Nunnally JC, Bernstein IH. Psychometric theory. 3rd edn. New York: McGraw-Hill; 1994.
Patrick DL, Deyo RA. Generic and disease-specific measures in assessing health status and quality of life. [Review].
Med Care

1989
;
27
(3 Suppl):
S217
–32.
Peto V, Jenkinson C, Fitzpatrick R, Greenhall R. The development and validation of a short measure of functioning and well being for individuals with Parkinson's disease.
Qual Life Res

1995
;
4
:
241
–8.
Pfennings LE, Van der Ploeg HM, Cohen L, Bramsen I, Polman CH, Lankhorst GJ, et al. A health-related quality of life questionnaire for multiple sclerosis patients.
Acta Neurol Scand

1999
;
100
:
148
–55.
Rasch G. Probabilistic models for some intelligence and attainment tests. Chicago: University of Chicago Press; 1960.
Rudick R, Antel J, Confavreux C, Cutter G, Ellison G, Fischer J, et al. Recommendations from the National Multiple Sclerosis Society Clinical Outcomes Assessment Task Force.
Ann Neurol

1997
;
42
:
379
–82.
Sharrack B, Hughes RA. The Guy's Neurological Disability Scale (GNDS): a new disability measure for multiple sclerosis.
Mult Scler

1999
;
5
:
223
–33.
Sharrack B, Hughes RA, Soudain S, Dunn G. The psychometric properties of clinical rating scales used in multiple sclerosis.
Brain

1999
;
122
:
141
–59.
Stewart AL, Ware JE Jr, editors. Measuring functioning and well-being: the medical outcomes study approach. Durham (NC): Duke University Press; 1992.
Stewart AL, Hays RD, Ware JE Jr. The MOS Short-Form General Health Survey: reliability and validity in a patient population.
Med Care

1988
;
26
:
724
–35.
Streiner DL, Norman GR. Health measurement scales: a practical guide to their development and use. 2nd edn. Oxford: Oxford University Press; 1995.
Vickrey BG, Hays RD, Harooni R, Myers LW, Ellison GW. A health-related quality of life measure for multiple sclerosis.
Qual Life Res

1995
;
4
:
187
–206.
Ware JE Jr, Brook RH, Davies-Avery A, Williams KN, Stewart AL, Rogers WH, et al. Conceptualization and measurement of health for adults in the health insurance study: Vol. I: model of health and methodology. Santa Monica (CA): Rand Corporation; 1980.
Ware JE Jr, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey manual and interpretation guide. Boston (MA): Nimrod Press; 1993.
Ware JE Jr, Kosinski MA, Keller SD. SF-36 physical and mental health summary scales: a user's manual. Boston (MA): Health Institute, New England Medical Center; 1994.
Ware JE Jr, Harris WJ, Gandek B, Rogers BW, Reese PR. MAP-R for windows: multitrait/multi-item analysis program—revised user's guide. Boston (MA): Health Assessment Laboratory; 1997.
WHOQOL Group. World Health Organization Quality of Life Assessment (WHOQOL): development and general psychometric properties.
Soc Sci Med

1998
;
46
:
1569
–85.