A New Functional Scale and Ambulatory Functional Classification of Duchenne Muscular Dystrophy: Scale Development and Preliminary Analyses of Reliability and Validity

Stage 1: Item generation

The MEDLINE database was searched from January 1965 to June 2009 using the following combinations of terms: (Duchenne muscular dystrophy OR muscular dystrophy) AND (functional evaluation OR functional scale OR activities of daily living). The search identified 305 articles, describing the following key functional scales: the Revised Amyotrophic Lateral Sclerosis Functional Rating Scale [12]; the Barthel Index [13]; the Modified Barthel Index [14]; the Muscular Dystrophy-Specific Functional Rating Scale [15]; the Modified Hammersmith Functional Motor Scale [16]; the Motor Performance Test [17]; the Pediatric Evaluation of Disability Inventory [18]; the Vignos Scale; the Brooke Scale [19]; the Wee Functional Independence Measure [20]; the Gross Motor Function Measure; the Physical Self-Maintenance Scale [21] and the measure of activities of daily living (ADL) and instrumental ADL [21]. A preliminary pool of items was generated following a series of expert panel meetings. Refinement of the preliminary item selection was conducted via a two-step process. In the first step, a closed questionnaire was sent to the members of the expert panel to obtain individual information pertaining to the importance of each item and sub-item (e.g., sitting item and related sub-items for each posture and activity, as described in Supplementary Table S1). The perceived importance of items and sub-items was measured on a Likert scale, ranging from 1 to 5, with the following descriptions provided: very unimportant, unimportant, not important or unimportant, important, and very important. In the second step, the results of the closed-questionnaire were tabulated and summarized as a total score and a mean (SD) score for each item and sub-item, and provided to the expert panel.

Each panel member was asked to review the results and to re-score items and sub-items as they deemed applicable.

Stage 2: Primary item reduction and scale development

Following the second round of scoring, ranked items were shared with the expert panel and classified into functional domains. A modified Delphi approach, in conjunction with subsequent meetings of the expert panel, was used to select relevant items and sub-items from the pool of items identified at the end of stage 1.

Stage 3: Preliminary evaluation of the inter-rater reliability and validity of the CFSD and AFCSD

The inter-rater reliability of the CFSD and the AFCSD was evaluated in 48 boys diagnosed with DMD, with a mean age of 5.29±3.51 years. All the items and sub-items were tested by two clinical experts on the same day, a pediatric physiatrist and a pediatric physical therapist. Items derived from established objective measurements, such as the Cobb angle of scoliosis or intelligence quotients, were not included in the evaluation of reliability.

Inter-rater reliability of the CFSD and AFCSD, as well as of the Vignos and the Brooke Scales, was evaluated using the intraclass correlation coefficient (ICC), with ICC values >0.7 suggesting acceptable level of reliability. Bland-Altman plots and limits of agreement (mean±1.96 SD) were also calculated.

The concurrent validity of the CFSD and AFCSD was evaluated against established functional scales, namely the Vignos and the Brooke Scales. Spearman correlation coefficients were calculated between the total scores of the AFCSD and both the Vignos and the Brooke scales, to provide a global measure of validity. Specific validity of the CFSD was evaluated by comparing total scores on the CFSD to scores for each grade of function on the Vignos and the Brooke scales, and for the 5 levels of the AFCSD. The level of concordance between the grades of the Vignos and Brooke scales and the levels of the AFCSD was assessed using the Kruskal-Wallis test, with the Wilcoxon rank sum test used for post-hoc analyses of differences detected between the scales, with corrections for multiple comparisons according to the false detection rate (FDR) method.

All the statistical analyses were performed using the R3.1.0 GUI, with the level of significance set at p<0.05.

The new functional scale for Duchenne muscular dystrophy

A preliminary pool of 174 items, described in Supplementary Table S1, was identified from the structured review of the literature and the series of the expert panel meetings. These items were classified into the following 9 functional domains: mobility, ADL, muscle power, joint contracture, pulmonary function, cardiac function, scoliosis, pain, and cognitive and social function. Highly scored items in each of the 9 domains of function were extracted. In circumstances involving two or more items within a domain sharing similar component activities (e.g., transferring from the bed to a wheelchair and from the wheelchair to a toilet seat), only one item was selected. In addition, when items with a broader meaning yielded scores similar to items measuring specific attributes (e.g., the activity of eating in comparison to individual items of feeding, swallowing, and drinking, which contribute to this activity), the item with the broader meaning was included. The importance of peak cough flow, pulmonary function tests, electrocardiograms, and echocardiograms was acknowledged, and these items were excluded as they are known to be independently associated with the functional prognosis of patients with DMD.

Following the methods for item reduction, 21 items and 78 sub-items were retained across the following 7 domains: mobility (8 items), ADL (6 items), contractures (2 items), pulmonary function (2 items), cardiac function (1 item), scoliosis (1 item), and cognition (1 item) (Table 1). A variable grading system was developed to produce a range of qualitatively different levels of functional performance. For example, the item ‘using drugs for heart disease’ was graded on a 2-point scale, 0 and 3 representing negative and affirmative responses, respectively. The total score on the CFSD ranged from a minimum score of 0 to a maximum score of 63, with higher scores representing higher function, and hence, less severe diseaserelated impairment. The full items and scoring systems are described in Table 1.

Table 1.

A comprehensive functional scale for Duchenne muscular dystrophy (CFSD)

Reliability and validity

Reliability

The ICCs between two examiners are reported in Table 2. Two out of the 48 subjects were excluded from the analysis due to missing discrete item scores. Interrater ICC values were below the cutoff of 0.7, set a priori, for the following four CFSD items: indoor activities (ICC=0.680), personal computer utilization (ICC=0.638), ventilator assist (ICC=0.662), and presence of dyspnea (ICC=0.453). Evaluation of the limits of agreement of the total AFCSD score between the two examiners indicated disagreement for a discrete outlier within one level (Fig. 2). With this exception, the overall limits of agreements ranged between -6.21 and 3.11 (Fig. 3A). These limits of agreement narrowed to a range of -5.54 to 3.19 when the items scoring ICC values <0.7 were removed from the calculation (Fig. 3B).

Table 2.

ICC for each item between two independent examiners

Fig. 2.

Bland-Altman plot for agreements between examiners 1 and 2 in the ambulatory functional classification system for Duchenne muscular dystrophy (AFCSD). Blue horizontal lines show a 97.5% limit, median, and 2.5% limit of the agreements.

Fig. 3.

Bland-Altman plot for agreements between examiners 1 and 2 in the total score of the new functional scale. (A) Total score ranging from item 1 to 19; (B) items with acceptable (>0.7) intraclass correlation coefficients. Blue horizontal lines show a 97.5% limit, median, and 2.5% limit of the agreements.

Validity

The Spearman correlation coefficient between the total score on the CFSD and on the Vignos scale was 0.833 (p<0.001), and 0.714 (p<0.001) between the total score of the CFSD and the Brooke Scale. The following significant correlations were calculated between the total CFSD score and the Vignos and Brooke scales or AFCSD: -0.818, -0.645, and -0.686 for the Vignos Scale, the Brooke Scale, and the AFCSD, respectively (p<0.001 for all pairs).

The mean (SD) scores for the CFSD score and each grade of the Vignos and Brooke scales, and the levels of the AFCSD are presented in Table 3. As shown in Fig. 4A, the total scores varied significantly between the following functional grades of the Vignos scale: 1 and 2; 1 and 4; 1 and 9; 2 and 9; 3 and 9; and 4 and 9. For the Brooke scale, the total scores were distinct for functional grades 2 and 3 (Fig. 4B). For the AFCSD, the total scores varied significantly between the functional levels 1 and 5 and 2 and 5 (Fig. 5).

Table 3.

Summary of the total score in each level and grade of the AFCSD, the Vignos and the Brooke Scales

Fig. 4.

Total scores of the comprehensive functional scale under each grade: (A) Vignos scale, and (B) Brooke scale. In the Vignos scale, scores varied significantly between grades 1–2, 1–4, 1–9, 2–9, 3–9, and 4–9. In the Brooke scale, scores between grade 2 and 3 differ from each other. Lower and upper margins of the box present the lower quartile (Q1) and the upper quartile (Q3) of the total score. The band inside the box indicates median, and the whiskers range from Q1+1.5 interquartile range (IQR) to Q3+IQR. Hollow circles represent outliers. Horizontal lines above the box graph depict significant differences in the Wilcoxon rank sum test (adjusted p<0.05).

Fig. 5.

Total scores in each level of the ambulatory functional classification system for Duchenne muscular dystrophy (AFCSD). Significant differences in the total score were found between 1–5, and 2–5. Lower and upper margins of the box present the lower quartile (Q1) and the upper quartile (Q3) of the total score. The band inside the box indicates median, and the whiskers range from Q1+1.5 interquartile range (IQR) to Q3+IQR. Hollow circles show outliers. Horizontal lines above the box graph depict significant differences in the Wilcoxon rank sum test (adjusted p<0.05).

A new functional scale for Duchenne muscular dystrophy

Novel items of the CFSD evaluating indoor activities, personal computer utilization, ventilator assistance, and presence of dyspnea scored ICC values <0.7. A review of these items indicates that the low reliability of the scores may reflect the vagueness of terms used, such as ‘active’ and ‘simple’ social interactions, which may be interpreted differently by different examiners. Language subtleties may influence the reliability scores. As an example, while mechanical ventilators and mechanical insufflators and exsufflators have adistinct meaning in English, a common wording is used in Korean (‘hoheup-bojogi’ vs. ‘kichimbojogi’). The low reliability scores for items on dyspnea may specifically reflect the existing vagueness surrounding the description of respiratory symptoms in children with DMD. For example, dyspnea associated with ventilation insufficiency rather than oxygenation insufficiency tends to be poorly differentiated until symptoms become severe and require intervention.

The total CFSD scores correlated strongly with the total scores on the Vignos and the Brooke scales. These high correlation coefficient scores might be useful to differentiate mild functional impairments, as the scores varied significantly from those for grades 1 and 2 on the Vignos scale (Fig. 4A). This sensitivity of the CFSD and AFCSD in differentiating grades of function in children in the early (mild) phases of DMD progression might be important in measuring changes after medical interventions and rehabilitation in the early stages of DMD. However, we must also consider the benefit of less granular classification grades; as an example, grade 2 on the Brooke scale includes a broad range of total scores (Fig. 4B), which might reflect diverse compensatory strategies that children with DMD adopt to accomplish functional goals during the early stages of the disease.

Ambulatory functional classification system for Duchenne muscular dystrophy

The high Spearman correlation coefficients between the AFCSD and the Vignos and the Brooke scales indicate concurrent validity of the AFCSD. However, as 34 of the 46 participants in this study (73.9%) were classified under level 1 or 2 on the AFCSD, with only a few children classified under levels 3 and 4, the total CFSD scores varied significantly only between levels 1 and 5 and levels 2 and 5. As the scores were not normally distributed, the mean (SD) values of total scores within each classification level could not be calculated.

Limitations

The study group manifested a skewed distribution of disease severity, with the majority of children in the mild phase of the disease: 28/46 (60.9%) and 33/46 (71.7%) classified under grades 1 and 2 of the Vignos and Brooke scales, respectively, whereas 34/46 (73.9%) were classified under levels 1 and 2 of the AFCSD. Validation of the CFSD and AFCSD for children at more progressive stages of disease severity requires large-scale clinical trials, with a normal distribution of participants across all stages of DMD disease severity.

Each stage of the AFCSD represents the functional levels of children with DMD. However, it does not contain age-specific or developmental stage-specific information. Further validation and modification of each stage requires testing in a larger population of children with DMD.

The intra-rater reliability of each item of the CFSD was not tested in this study as the participants and their legal guardians refused to undergo prolonged evaluation. Intra-rater reliability of each item should be further investigated in a subsequent study. The validity of the CFSD also needs to be tested further using comprehensive evaluation tools such as the Gross Motor Functional Measure, Modified Barthel Index, Pediatric Evaluation of Disability Inventory, and Functional Independence Measure for Children (WeeFIM).

Relatively new functional scales, such as North Star Ambulatory Assessment [23] or motor function measurement scale were not included in the item generation because they were not widely used at the time this study was conducted.

Conclusions

We have developed a new scale, and associated classification system, to assess function in patients diagnosed with DMD. Preliminary evaluation of the psychometric properties of the functional scale and classification system indicate sufficient reliability and concurrent validity to include the CFSD and AFCSD in large-scale clinical trials.