The Guy's Neurological Disability Scale (GNDS) was originally developed in the United Kingdom to measure disability in multiple sclerosis (MS). The GNDS is an outcome measure that is a simple, user‐friendly, multidimensional self‐report instrument. It was developed when the need for a comprehensive outcome measure was identified from a survey of 49 leading international neurologists from Europe, North America, Australia, and New Zealand (Sharrack & Hughes, 1999).
Because the GNDS was originally developed and tested in the United Kingdom, it reflects the vocabulary and phrases used there. Reliability and validity testing is needed before the GNDS is used to measure neurological disability in individuals from the United States. According to Rossier and Wade (2002), an examination of the sensitivity of the GNDS to change and its ability to detect differences has yet to be established. The purpose of this study was to test the reliability, validity, and sensitivity of the Americanized version of the GNDS.
The initial development of the GNDS by Sharrack and Hughes (1999) included the compilation, from a review of the literature, of a detailed list of various disabilities relevant to MS. This list was supplemented by open interviews with five individuals living with MS. Twelve separate functional domains were identified and used to construct the GNDS: cognitive, mood, visual, speech and communication, swallow, upper limb, lower limb, bladder, bowel, sexual, fatigue, and other disabilities (e.g., pain, spasms, vertigo). Each domain is graded from 0 to 5 according to severity and impact on the individual, as determined by the help required to perform these functions (0 = normal status; 5 = total loss of function —maximal help required). The possible range of scores on the GNDS is 0 (no disability) to 60 (maximum possible disability).
Face and content validity were determined by the same 49 leading international neurologists, who identified the need for a more comprehensive measure. The neurologists confirmed and approved the proposed GNDS. Interrater reliability was confirmed from a sample of 50 patients attending an MS clinic. Each patient was interviewed in the same session by two different neurologists using the GNDS. The intraclass correlation coefficient was 0.98. The GNDS was also found to be moderately responsive to clinical change in the participants' neurological disability over a 9‐month period in a small sample (Sharrack & Hughes, 1999).
Convergent validity was tested by Sharrack and Hughes (1999), who examined for a relationship of the GNDS with similar measures. Significant relationships were found with all measures as follows: Scripps Neurological Rating Scale (r = ‐0.78), The Expanded Disability Status Scale (EDSS; r = 0.75), The London Handicap Scale (r = 0.52), the Functional Independence Measure (r = ‐0.81), and the physical functioning domain of the Short Form 36® (SF‐36) Health Survey (r = ‐0.81). The strongest correlations were with other disability measures rather than those measuring impairment handicap or healthrelated quality‐of‐life measures. Rossier and Wade (2002) also tested for convergent validity. A strong inverse relationship was found between the GNDS and the Barthel Activities of Daily Living (ADL) Index score (r = ‐0.76), and a strong positive relationship was found with the EDSS (r = 0.64).
A factor analysis by Sharrack and Hughes (1999) suggested a four‐factor solution that accounted for 58.7% of the total variance. The investigators stated that each item of the GNDS loaded on one factor only. The four factors are the Spinal Factor (Factor 1; lower limb, bladder, bowel, and sexual disabilities); Mental Factor (Factor 2; cognitive, mood, and fatigue disabilities); Bulbar Factor (Factor 3; speech and swallowing disabilities); and Upper‐Limb, Vision, and Other Factor (Factor 4; upper‐limb, vision, and other disabilities). The investigators did not provide the detail of steps or parameters when running the factor analysis. Factor 4. does not have similar items loading together to form the subscale and seems to need renaming, if indeed there are four factors. Because a small sample size was used for the factor analysis (n = 50), the results should be considered tentative.
Sharrack and Hughes (1999) reported an internal consistency for the GNDS as 0.79 using a Cronbach's alpha in a sample of 50 individuals with MS attending an MS clinic in the United Kingdom. Rossier and Wade (2002) reported a 2‐3 week test‐retest reliability of 0.97 when administered face‐to‐face and 0.90 when administered via mail to a sample of 43 individuals with MS. The investigators concluded that the GNDS was a valid and reliable instrument for the assessment of individuals with MS, whatever the degree of disability.
The psychometric properties of the Americanized version of the GNDS are unknown. Therefore, the reliability, validity, and sensitivity need to be investigated.
1. Is there convergent validity of the Americanized version of the GNDS as evidenced by a significant relationship with the SF‐36 and ADL Self‐Care for MS Scale?
2. Does the Americanized version of the GNDS have the same four‐factor structure as reported in the psychometric testing of the original GNDS?
3. Is the Americanized version of the GNDS a reliable measure when examined with a 2‐week retest?
4. Does the Americanized version of the GNDS have internal consistency?
5. Is the Americanized version of the GNDS sensitive to changes in neurological disability in MS over time?
Potential participants included approximately 1,000 individuals with MS who the principal investigator originally recruited for another study using the NARCOMS patient database (Yale University) and postings on MS Watch and MS Society Web sites and newsletters. The pool of participants requested to be notified via e‐mail of any additional MS studies by the investigator for which they might qualify to participate. Main eligibility criteria for participants included being diagnosed with relapsing‐remitting MS, having a history of one or more relapses in the previous 2 years, being 18 years of age or older, being willing and able to complete all evaluations related to the study, and providing informed consent. Exclusion criteria for participants included significant medical illnesses other than MS, inability to complete the study instruments or to provide informed consent, and previous participation in the study.
As previously discussed, the Guy's Neurological Disability Scale (Sharrack & Hughes, 1999) comprises 12 functional domains of disability: cognitive, mood, visual, speech and communication, swallow, upper limb, lower limb, bladder, bowel, sexual, fatigue, and other disabilities (i.e., pain, spasms, vertigo). Each domain is graded on a scale of 0‐5 according to severity and impact on the individual, as judged by the help needed to perform these functions (0 = normal status; 5 = total loss of function—maximal help required). The possible range of scores on the GNDS is 0 (no disability) to 60 (maximum possible disability).
The Americanized version of the GNDS includes some modifications to the wording of questions on the GNDS to reflect vocabulary and phrases used in the United States, as well as some refinement and clarification of the instrument. The Americanized version was modified as follows
* cognitive disability: the question “Are you having someone help you with this survey?” was added
* mood disability: “If you had a mood problem…” was changed to “If you have been feeling anxious, irritable, or depressed…”
* visual disability: “…with ordinary glasses, if worn” was changed to “…with regular glasses, if worn”
* speech and communication disability: the question “Do you have to repeat yourself when speaking to strangers?” was added; the question “Is the patient able to communicate effectively using these methods” was changed to “Are you able to communicate effectively by using sign language or the help of your caregivers?”
* swallowing disability: the phrase “Do you have to take care when swallowing…” was changed in two places to “Do you have to be careful when swallowing…”
* upper‐limb disability: no modification was made
* lower‐limb disability: the word “stick” was changed to “cane” in three places; the word “sticks” was changed to “canes” in two places
* bladder disability: the question “On your current medication, are you having any problem with your bladder?” was added; the line “If you are having any problem with your bladder” was added before the question “Do you have to rush to the toilet…”; the words “been incontinent” was changed in three places to “had a problem controlling your urine”; the question “Do you use a sheath to collect your urine?” was changed to “Do you use a condom catheter to collect your urine?”
* bowel disability: the question “On your current medication, are you having any problems?” was added
* sexual disability: the phrase “If the patient agrees to respond” was changed to “If you agree to respond”
* fatigue disability: no modification was made
* other disabilities: the question “With your current medication, is/are these problem(s) bothering you?” was added.
The SF‐36 Health Survey (Stewart, Hays, & Ware, 1988) is a 36‐item instrument developed to measure health outcomes. Higher scores indicate better health. The instrument includes eight subscales, with the range of raw scores as follows: physical functioning (10‐30), role limitations due to physical health problems (4‐8), bodily pain (2‐12), general health (5‐25), vitality (energy/fatigue; 4‐24), social functioning (2‐10), role limitations due to emotional problems (3‐6), and mental health (psychological distress and psychological well‐being; 5‐30). Raw scores for each subscale are then transformed using a formula and range from 0 to 100. Reliability is reported as ranging from 0.73 to 0.96. The SF‐36 is a well‐developed and extensively tested outcome measure (Ware, Kosinski, & Gandek, 2003).
The ADL Self‐Care for MS Scale (Gulick, 1987) is a 15‐item instrument that measures essential activities encountered by all persons in everyday life, and these items were especially important to persons with MS. It is used to determine what functions the person with MS presently performs. The Likert scale ranges from 0 (never) to 5 (always) for performance of the 15 listed ADLs. There are four subscales: motor (eating, dressing, bathing, walking, travel), intimacy, sensory and communication, and recreation and socializing. All items can be summed to yield a total score between 0 and 75; higher scores suggest a higher level of functioning. Reliability for internal consistency was reported as 0.96 in 629 individuals with MS. Test‐retest over a period of 2‐4 weeks among a subset of 89 participants was 0.86. Administration time is approximately 5 minutes. The measure is well developed, reliable, and valid.
The initial e‐mail invitation to participate included the purpose of the study, what was expected by participants, questions regarding inclusion/exclusion, information about compensation for participation, and contact information for the principal investigator. The initial sociodemographic questionnaire (Time 1) included questions about date of diagnosis with MS, use of immunomodulators for MS, age, gender, ethnicity, marital status, level of education, household income, and contact information for follow‐up at 2 weeks (Time 2) and 6 months (Time 3).
The 6‐month follow‐up questionnaire (Time 3) requested confirmation of contact information so that payment could be sent to the participant; it also asked about the current use of immunomodulators for MS. Two additional questions were included to evaluate the sensitivity of the Americanized version of the GNDS. The first question was “Have you had a flare‐up of your MS symptoms in the past 6 months?” Participants could respond “yes” or “no.” The second question was “How would you rate your MS symptoms at this time as compared with 6 months ago?” Participants could respond “much worse,” “somewhat worse,” “about the same,” “somewhat better,” and “much better.”
Institutional review board approval was obtained from Monmouth University in West Long Branch, NJ. Participants were invited to participate in this prospective study via e‐mail. They received a description of the study, an eligibility screen, and an informed consent agreement. They were invited to ask any questions before signing and returning the informed consent agreement to the principal investigator. After that document was returned, participants received the study survey package in an e‐mail. If preferred, the questionnaire was mailed to the participant with a postage‐paid return envelope. The survey package included the Americanized version of the GNDS, the SF‐36, the ADL Self‐Care for MS Scale, and a sociodemographic questionnaire. Participants were asked to return the survey package to the principal investigator when it was completed. Two weeks later participants received another copy of the Americanized version of the GNDS to complete and return for test‐retest reliability. At 6 months from baseline participants received the final copy of the study survey package to test for sensitivity to change over time. It took participants approximately 20 minutes to complete the survey packet at the first and third administration and approximately 5‐10 minutes to complete the Americanized version of the GNDS at the 2‐week retest.
Each participant was assigned a number that was used on the instruments and sociodemographic questionnaire. Any identifying information on the completed questionnaire, such as the participant's e‐mail address, was masked with a black gel pen. The principal investigator's computer remained password protected, and a secure firewall was maintained. Electronic data was stored on the researcher's computer during the study, then it was removed and stored on a CD‐ROM in a locked drawer. The data will be stored for a period of 5 years, then destroyed.
To encourage respondents to complete all three survey packages, they were told in the informed consent agreement that they could receive up to $50 for their participation if they completed all surveys —$20 for the first full study survey package, $10 for the single scale survey package, and $20 for completing the second full study survey package. Participants who did not complete the entire study were compensated for the study survey packages they did complete. Participants were paid at the end of the study.
The data were entered into the Statistical Package for the Social Sciences, version 11.5 (SPSS 11.5) for analysis. Convergent Validity of the Americanized version of the GNDS was tested using the Pearson Correlations for the GNDS and the SF‐36 and ADL Self‐Care for MS Scale. Construct validity was examined using factor analysis. The 2‐week test‐retest reliability of the Americanized version of the GNDS was examined using the Pearson Correlation. Internal consistency was determined by the Cronbach's Alpha. The comparison of instruments completed at Time 1 and Time 3 (month 6) was evaluated with a paired t test. Descriptive statistics were run for all study data.
The sample included 253 participants—87% women (n = 219) and 13% men (n = 32). Two participants did not specify their gender. Participants ranged in age from 22 to 77 years (M = 46, SD = 9). Almost all of the participants were Caucasian (95%); however, there also were participants who were African American (2.4%), Hispanic Latino (1.2%), Asian or Pacific Islander (0.4%), Asian Indian (0.4%), and other (1.2%). The participants were well educated; 68% had a college education and 32% had a high school diploma. Household income was reported as being below $60,000 for 41% of the sample and $60,000 or more for 45%. Fourteen percent chose not to answer the question. More than half of the participants were married (59%); 13% were never married, and 17% were divorced or separated. Eighty‐four percent were taking an injectable immunomodulator for their MS (Avonex 39%, Betaseron 9%, Copaxone 23%, and Rebif 13%). The remaining 16% were untreated. Three participants were lost to follow‐up at Time 3 (month 6). The final sample at Time 3 was 250 participants.
Convergent validity of the Americanized version of the GNDS was supported by significant inverse relationships with the eight subscales of the SF‐36 and the ADL Self‐Care for MS Scale. The correlations ranged from ‐0.33 to ‐0.66 (Table 1).
Factor analysis is used to identify which items of an instrument should be grouped as a factor or subscale and which should be dropped. A factor is a group of items that appear to belong together. A person who scores high on one item of a particular factor is likely to score high on other items of the same factor. There is a high correlation between items of the same factor. There is a low correlation between items from different factors. The results of a factor analysis will confirm the number of factors or subscales, which are then named according to the type of items that have loaded together. The minimal sample size for factor analysis is 10 participants for each item on the instrument (Munro, 2001).
The first step is to run a Pearson's correlation. An item‐to‐item correlation is run to examine for significant relationships between all the items on the instrument. Total scores are not included. Correlations are reviewed, and only items that correlate between 0.30 and 0.70 are included (Munro, 2001). The Pearson correlation was run using the items on the Americanized version of the GNDS. All items were significantly correlated and ranged between 0.30 and 0.70.
The number of factors identified by the factor analysis indicates the number of subscales on the instrument. Eigen values are calculated from the unrotated factor matrix and represent the total amount of variance explained by a factor. An Eigen value of at least 1 is desired, representing 5% of the variance explained. A Scree Plot, or graph, confirmed the number of factors or subscales (Munro, 2001).
After the factor analysis, items are reviewed to see how they loaded on each factor. There must be a 0.2 difference between factors to retain the item. If items are not retained, a factor analysis excluding that item is repeated (Munro, 2001).
For this study, a four‐factor solution for the Americanized version of the GNDS was forced based on previous research suggesting four subscales. The sample size was 253 the first and second times the factor analysis was run and 250 for the third analysis. The factor analysis was run for the data collected at Time 1, Time 2, and Time 3. It explained 59% of the variance at each time, but items did not fall into the factors described by Sharrack and Hughes (1999), or the items loaded differently each time. Items did not fall into a logical order based on neurophysiology. Most of the items that loaded on Factor 2 also loaded similarly on Factor 1 or Factor 3 using Time 1 data. The Time 2 data revealed a different configuration of items loading on the factors.
When the factor analysis was run using the traditional criteria of Eigen values of 1 or more, 52% of the variance was explained each of the three times it was run. The Time 1 data for the Americanized version of the GNDS loaded items in a logical pattern, although four items still loaded similarly on two factors. Time 2 data loaded more cleanly on three factors and explained 52% of the variance. The exception was upper limb, which loaded almost equally on all three factors. The upper‐limb item was excluded on the follow‐up factor analysis. The exclusion of the upper‐limb item resulted in 53% of the variance explained and items loaded cleanly with at least 0.24‐0.61 difference between the factors. One exception was cognitive disability, which had a 0.15 difference between Factor 1 and Factor 2. The Time 2 data revealed three factors or subscales: Factor 1—cognitive, mood, visual, speech and communication, and swallow disabilities; Factor 2—sexual, fatigue, and other (i.e., pain, spasms, dizziness) disabilities; Factor 3—lowerlimb, bladder, and bowel disabilities.
A factor analysis run with Time 3 data revealed a third variation of item loading: Factor 1—cognitive, mood, fatigue, and other disabilities; Factor 2—visual, speech and communication, swallow, upper‐limb, and lower‐limb disabilities; Factor 3—bladder, bowel, and sexual disabilities.
The findings of three‐factor analyses in this study do not support the four‐factor solution suggested by Sharrack and Hughes (1999). It is evident that each item on the Americanized version of the GNDS measures a specific disability that a person with MS might experience at any given time and the different combinations of disabilities at various times in the process of MS. The investigators conclude that items of the Americanized version of the GNDS should not be conceptualized as falling together to form consistently identifiable factors or subscales.
A problem was noted with participants responding to item 12, which assesses “other disabilities” and asks, “Do you have any problems due to MS such as pain, spasms, or dizziness that have not been mentioned so far?” This question groups several common problems in MS into one item. The item also assesses the extent to which disability affects participants' lives. In addition, although participants did complete this entire item, 21% (n = 53) failed to write in specifically whether they experienced pain, spasms, or dizziness at Time 3. Of those who did write in a specific problem, 16% (n = 41) reported having pain, 15% (n = 37) spasms, 10% (n = 25) dizziness, and 4% (n = 10) numbness at Time 3. Thirty‐four percent (n = 84) denied having any other disability and scored a 0 on this item. Because pain, spasms, dizziness, and numbness are common problems in MS, each problem needs a more complete assessment.
The Americanized version of the GNDS was found to be a reliable measure when examined with a 2‐week test‐retest. The Pearson correlation revealed a very strong relationship (r = 0.91, p = .000), indicating an excellent reliability. The Cronbach's alpha was 0.79 at Time 1, 0.78 at Time 2, and 0.80 at Time 3, indicating good internal consistency.
At Time 3 participants were asked two additional questions to evaluate the sensitivity of the Americanized version of the GNDS. The first question was “Have you had a flare‐up of your MS symptoms in the past 6 months?” Participants could respond “yes” or “no.” Of the 250 participants remaining at Time 3, 100 indicated they'd had a flare‐up in the previous 6 months. This self report of a flare‐up was not verified by a neurologist, nor was any further detail of the flare‐up obtained. When examining the mean scores on the GNDS from Time 2 to Time 3 using a paired t test, a significant increase in neurological disability was found for those who reported having a flare‐up (t = ‐3.76, p = .000). During this 51/2‐month period, the mean score on the Americanized version of the GNDS increased from 16.7 (SD = 8.2) to 18.5 (SD = 8.8). For those who reported that they had not had a flare‐up (n = 149), there was no significant difference in level of disability during the same period (t = ‐.25, p = .805). The mean scores during the 51/2 ‐month period was 12.1 (SD = 7.6) at the 2‐week retest time and then 12.2 (SD = 7.9) at the end of the study. Four participants did not respond to the question about flare‐ups in the past 6 months.
The second question asked at Time 3 was “How would you rate your MS symptoms at this time as compared to 6 months ago?” Participants could respond “much worse,” “somewhat worse,” “about the same,” “somewhat better,” and “much better.” The higher the score on the Americanized version of the GNDS, the higher the level of self‐reported neurological disability. When examining the scores for changes in levels of disability from the beginning of the study to the end of the study, participants' perceptions of their level of disability was reflected in the Americanized version of the GNDS scores. Those who thought they were much worse (n = 5) than 6 months prior had a mean increase in their level of disability of 5.6. Participants reporting being somewhat worse (n = 77) had a mean increase in disability of 0.32. Those who thought they were about the same (n = 125) had a mean decrease in their level of disability of 0.17. Participants who thought they were somewhat better (n = 25) had a mean decrease in level of disability of 1.68. Those who thought they were much better (n = 18) had a mean decrease of 3.39 in their level of disability (Table 2).
The participants in this study were mostly welleducated, Caucasian women, and the majority of participants were married. Males and other ethnic groups were minimally represented in the sample of participants with MS.
Because participants were invited to participate via e‐mail, the study was limited to those who had a current e‐mail address and could use a computer. The requirement of signed informed consent agreements sometimes caused a delay in individuals participating until the signed consent was received by the principal investigator through the mail. Participants that preferred to receive and return study questionnaires through regular mail may have been inconvenienced by the need to read, complete, and mail back their questionnaire packet. An addressed envelope and return postage was provided to help facilitate this process.
This was a longitudinal study over a 6‐month period with some loss of participants. Of the 253 that began the study, 250 completed the final set of questionnaires at Time 3. The loss of participants over time was minimized by good communication via e‐mail, regular mail, and phone calls between the principal investigator and the participants.
The purpose of this study was to test the reliability, validity, and sensitivity of the Americanized version of the GNDS. Convergent validity was supported by a significant inverse relationship with the SF‐36 and ADL Self‐Care for MS Scale. The factor analysis did not support the four‐factor structure as reported in the psychometric testing of the original GNDS (Sharrack & Hughes, 1999). When the factor analysis was run using the traditional criteria of Eigen values of 1 or more, a three‐factor solution was found. When each of the three Americanized versions of the GNDS were administered and factor analyses were run, the items loaded differently on each of the three factors or subscales.
According to Miller (2001), lesions in MS may occur in the brain, spinal cord, and optic nerves, and the course of MS is highly variable. This suggests that individuals completing a self‐report of level of neurological disability may have a variety of different symptoms at any given time, depending on the number, size, and location of demyelinated areas. Therefore, the investigators of this study conclude that items on the Americanized version of the GNDS should not be conceptualized as loading together to form consistently identifiable factors or subscales. Instead, the items should be viewed as measuring different aspects of disability in MS. The scale was found to be a reliable measure when examined with a 2‐week retest and 6‐month followup. The scale has good internal consistency and appears to be sensitive to changes in neurological disability in MS over time, as perceived by those who are living with MS.
The Americanized GNDS appears to be a good selfreport measure of level of disability in MS, although some limitations were revealed during the research process. Based on the findings of this study, the investigators strongly suggest revising and expanding the Americanized version of the GNDS. The changes include (a) revising and shortening item 6 (upper‐limb disability), (b) deleting item 12 (other disabilities), and (c) adding items 12 (pain), 13 (spasm), 14 (dizziness), and 15 (sensory). In addition, because the principal investigator hand scored each completed Americanized version of the GNDS using a coded instrument, subscript numbers in a small font were inserted into the proposed revised and expanded scale to facilitate the scoring process. Psychometric testing is needed to evaluate the new 15‐item Americanized version of the GNDS in a sample of at least 150 participants. This expanded version with 15 items shall be known as the Guy's Neurological Disability Scale US‐15 Version (GNDS US‐15).
The use of a self‐report measure of neurological disability has several benefits, including the ability to easily monitor neurological status over time, creation of a starting point for discussion between individuals and healthcare providers during followup care, and, most importantly, a means to assess neurological status when individuals living with MS are not in the physical presence of healthcare professionals.
This study found the Americanized version of the GNDS to be a reliable, valid, and sensitive multidimensional measure of neurological disability for individuals with MS. The expansion of the measure from 12 to 15 items provides a more comprehensive assessment of pain, spasm, and dizziness, in addition to a sensory disability item.
This study was funded by Teva Neuroscience.