Friedreich’s ataxia (FRDA) is characterised by progressive neurological symptoms and cardiac dysfunction. It results in a reduced life span.^1,2 As the pathogenesis of FRDA has become better understood, several potential treatments have arisen. Initial beneficial treatments will probably slow progression rather than reverse morbidity. An appropriate and responsive scale for FRDA is critical if the benefits of potential treatments are to be identified. As there is no gold standard to examine disease progress in FRDA, the evaluation of measurement scales is essential to ensure that trials are efficient and their conclusions are accurate.³

The Friedreich Ataxia Rating Scale (FARS) comprises a measure of ataxia, an activities of daily living (ADL) subscale and a neurological subscale. Face and content validity and inter-rater reliability are good.⁴ Other aspects of validity have not been examined. The International Cooperative Ataxia Rating Scale (ICARS) was developed to assess pharmacotherapies in ataxia,⁵ and has good inter-rater reliability.⁶ The Functional Independence Measure (FIM)⁷ and the Modified Barthel Index (MBI)⁸ examine the assistance required to complete ADLs. They are validated instruments used widely for neurological disease.

We aimed to examine these ordinal scales proposed as outcome measures for FRDA to establish concurrent criterion validity and compare the change in score over time to determine the sensitivity of the scales and confirm which measure is most appropriate for use in clinical trials.

METHODS

The study was conducted after approval by the Southern Health Ethics in Human Research Committee. Participants were recruited by invitation through the local FRDA support association and from among those who attended a multidisciplinary FRDA clinic. All participants were homozygotic for an expanded GAA repeat in the FRDA gene. Individuals with point mutations were excluded. No active selection of participants with particular characteristics took place.

The FARS, ICARS, FIM and MBI were administered by one of three assessors (MCF, AJC and LC) according to published instructions. The FIM was assessed by team consultation following a semistructured interview and examination. Repeat measures were undertaken at an interval of approximately 12 months with the examiners being blinded to the previous results.

The FARS consists of three subscales, comprising a general score for ataxia, a score for activities of daily living (ADL) and a neurological examination. The scores can be added to make a total score ranging from 0 to 159. A higher score indicates a greater level of disability. The ICARS Scores range from 0 to 100. A higher score also indicates more disability. The FIM (possible scores 18–126) and MBI (possible scores 0–100) scores decrease with decreasing function. Clinical details including age, age at symptom onset, disease duration and drug use was recorded. Disease duration was defined as the age at initial testing minus the age of onset.

Statistical analysis

Statistical analysis was undertaken using SPSS V.12.0.1. The statistical significance of the differences in baseline characteristics between those seen once and those seen twice were assessed using Fischer’s exact test for proportions, and analysis of variance for mean scores. Spearman’s correlation coefficients were calculated to assess the concurrent criterion validity of the FARS and its subscales against the ICARS, FIM and MBI. The statistical significance of the mean change in scale score from baseline to 12 months was assessed using paired t tests. The sensitivity was assessed by calculating the effect size defined as the mean difference in scores divided by the standard deviation of the baseline score.⁹ A larger effect size indicates a more sensitive measure—that is, a greater change has been measured over the same time period. Power calculations were undertaken using Sample Power software (SPSS, Chicago, Illinois, USA) to assess the number of patients needed for a clinical trial with a treatment and placebo group with 80% power and α = 0.05. A p value <0.05 was considered significant.

RESULTS

In all, 76 participants had at least one measurement with the FARS and ICARS, and 72 of these individuals were assessed with concurrent FIM and MBI measures. In total, 43 participants were measured using all scales on two occasions 12 months apart. Table 1 shows the demographic and clinical information of the participants, which shows that the cohort had a broad range of the various parameters examined. Age of onset, disease duration and the use of antioxidants was similar between the group examined once and that assessed twice. There was a trend for less severe disease in those seen twice as judged by the four scales. Differences in FIM and MBI Scores were seen in the group who had repeat testing compared with those tested only once. The correlation between the age of disease onset and the size of GAA1 (r = −0.51; p<0.001) was comparable with previous studies.^10,11

The total FARS showed high correlation with the functional measures MBI and FIM (Spearman’s r = 0.9 and 0.94) and the ICARS (Spearman’s r>0.96), an established measure of ataxia. When the subscales of the FARS were examined, the ataxia subscale had the highest correlation with the FIM (Spearman’s r = 0.9). The neurological subscale correlated best with the ICARS (Spearman’s r = 0.97) and had a high correlation with the MBI (Spearman’s r = 0.93) and FIM (Spearman’s r = 0.9). The ADL subscale correlated with all measures (Spearman’s r = 0.9 for the ICARS and MBI; r = 0.87 for the FIM). All correlations were significant (p<0.001). Correlations between scores for the four scales and disease duration were significant for all measures. The greatest correlation was for the total FARS compared with disease duration (Spearman’s r = 0.79; p<0.001).

There was a significant decline over 12 months detected by the FARS, ICARS and FIM but not the MBI (table 2). The FARS had the greatest effect size and required fewer patients for an equivalently powered clinical trial than any of the other measures (table 2).

DISCUSSION

Of the four scales assessed, the FARS is the best for use in clinical trials of FRDA. This is based on it having the highest effect size, and power calculations that show that fewer participants are required to demonstrate the same effect of an intervention compared with other commonly used outcome measures.

The first aim of this study was to provide further validity to the FARS as an outcome measure in FRDA. As the total FARS has components to measure not only ataxia but also ADLs, we sought to perform this through correlation with both measures of ataxia and function. Concurrent criterion validity for the FARS is evident through correlations with well-established measures of disability, the MBI and FIM. The strong correlation between the scales suggests that these scales measure a common underlying construct. Of the subcomponents of the FARS, the neurological subscales have the highest correlation with the ICARS. The ataxia and ADL components correlate with all measures, indicating a clear overlap between functional assessment and ability.

As FRDA is a progressive neurological disease, we predict that scores indicating more severe disease would correlate with disease duration when examined in cross section. The correlation of all measures with disease duration was significant, but was greatest with the total FARS Score. The association of the score with disease duration provides face validity for these measures.

The second aim was to determine the most sensitive scale to measure clinical decline in FRDA. For a scale to be used in a clinical setting, especially for a clinical trial, it must be sensitive to change. This is especially true in a progressive, fatal condition such as FRDA, where there is no proved treatment to slow disease progression. Even a small benefit from a therapeutic intervention is likely to be clinically relevant. In this context, the most sensitive scale will be the most appropriate one to use.

Effect size calculations suggest that the FARS is a more sensitive measure than the other three scales. Power calculations indicated that the total FARS is the most responsive to change, in that fewer patients are required than if using the other three scales for equivalent power in a trial.

Our cohort was recruited by invitation through a patient support group and through a clinic that sees a broad range of individuals with FRDA. No selection took place except through exclusion of those with a point mutation. However, there may have been a selection bias with respect to less mobile individuals who were less likely to attend. Some evidence suggested that those seen twice were less severely affected at baseline than those seen once. However, there was still a broad range of disease severity in those with repeat measures, and the study compared the same individuals with all scales.

Despite the FARS performing better than the other three scales tested, the effect size over a period of 12 months is only moderate. This may reflect the sensitivity of the scale, or alternatively that the true amount of change seen in 12 months is modest. A longer period of assessment with greater numbers of patients will help to determine this. Four scales were used for comparison. How newly developed measures of ataxia, such as the Scale for the Assessment and Rating of Ataxia, will perform against a specific FRDA ataxia scale is yet to be determined.¹² Furthermore, being an ordinal scale, the true relationship between scale points is not defined for the FARS. Therefore, work is required to find improved measures for FRDA. Other methods may help in refining measures to behave in a more linear or continuous manner. These include techniques such as Rasch analysis of the FARS and the development of a functional composite.¹³