Validity of the Functional Gait Assessment in Patients With Parkinson Disease: Construct, Concurrent, and Predictive Validity

Construct Validity

Construct validity refers to how well a measurement tool is capable of measuring the theoretical concept being investigated and to the number of dimensions that the score can explain.³² Assessment of construct validity involves major factor analysis and aims to identify scale potential structure and to reduce the number of items. Wrisley et al⁴ tested the construct validity of the FGA in patients with vestibular disorders and found 3 extracted factors that may represent separate domains of the scale. In this study of patients with PD, we found that all 10 items of the FGA measured one common factor, which was functional gait status. Generally, a minimum of 60.0% variance is necessary for satisfactory accomplishment.³³ The common factor in this study accounted for 64.0% of the variance, providing sufficient evidence that the FGA has good construct validity in patients with PD.

Concurrent Validity

We also analyzed concurrent validity of the FGA, which is a type of criterion-related validity. Concurrent validity examines the degree of closeness between the scale in question and another standard measurement or the “gold standard.” We selected several scales that are commonly used in the clinic and have been tested for their reliability and validity as standard measures. Our results showed that the correlation coefficients for the FGA compared with other measures ranged from .57 to .85, indicating their strength was moderate to strong.⁹ Of these scales, the FGA showed the strongest correlations with the BBS and FAC (.85 and .78, respectively), as the core contents in these scales are similar to those assessed with the FGA. This finding is consistent with the findings of the study by Leddy et al,⁹ in which the correlations for the FGA compared with the BBS, ABC, MDS-UPDRS-3, and modified Hoehn and Yahr scale were .78, .71, .70, and .67, respectively, in patients with PD. These values are similar to those we report for the present study (.85, 0.72, .66, and .70, respectively). These results provide sufficient evidence to support the use of the FGA for assessment of balance and gait in patients with PD.

Conversely, the FGA had the lowest degree of correlation (.57) with the TUG. One possible explanation for this low degree of correlation is that TUG measures a single functional task that records only the time it takes for a patient to stand up from a chair, walk, and then return to sitting in the chair. It does not reflect other characteristics of balance and gait disorders such as stepping over an object, standing with one leg, stepping up and stepping down, standing on a narrow support area, or walking. These results suggest that the FGA can provide a more comprehensive view of gait and balance than the TUG.

Predictive Validity

Clinical perspective needs to be taken into account when considering what is the appropriate cutoff score for identifying those at risk for falling. Inappropriately high specificity will cause an increase in false-negative rate, resulting in missing individuals who are at risk for falling. Conversely, inappropriately high sensitivity can lead to a high false-positive rate, which may result in unnecessary intervention. Based on the findings of our study of patients with PD, we propose an FGA cutoff score of ≤18/30 for identifying those at risk for falling, with an AUC of 0.84 and a positive likelihood ratio of 4.03. The AUC suggests that the FGA had an 84% chance of making a correct prediction of falls using this cutoff, and patients with an FGA score of ≤18 would be 4 times as likely to have a fall compared with those with an FGA score of >18.

Wrisley and Kumar⁷ proposed an FGA score of ≤22/30 as an effective cutoff for predicting fall risk in community-dwelling older adults. Although an FGA score of ≤20/30 (AUC=0.92, sensitivity=100%, specificity=83%, and positive likelihood ratio=5.8) (Tab. 4) provided the optimum metrics in identifying older adults who had unexplained falls, they argued that clinicians should use a more conservative criterion score of ≤22/30 to lead to earlier intervention.⁷ Leddy et al,⁹ in their study of community-dwelling individuals with PD, suggested an FGA cutoff value of 15/30 (AUC=0.80, sensitivity=72%, specificity=78%, and positive likelihood ratio=3.24) (Tab. 4) for identifying those at risk for falling.⁹

The differences in cutoff values in the current study and the studies by Wrisley and Kumar⁷ and Leddy et al⁹ may be explained, at least partially, by the differences in patient populations, sample sizes, and methods of assessment. To our knowledge, our study is the largest study to date to validate the FGA. Wrisley and Kumar's research was conducted in community-dwelling older adults. Additionally, their definition of a fall was more stringent than ours in that they defined a “faller” as a patient who reported 2 or more unexplained falls. Therefore, a certain number of patients who sustained only a single fall were not classified as fallers as they were in our study. In Leddy and colleagues' study, the participants were community-dwelling patients with PD. In their study, a faller again was defined as someone who reported 2 or more falls in the previous 6 months, without further distinguishing explained or unexplained falls. It is possible that patients with explained falls were counted as fallers, which may have contributed to the different results from our study. On the other hand, the adjustment of anti-PD medications after the FGA evaluation in our study may have improved patients' balance and gait function. These improvements may have led to a different cutoff score for identifying fallers, with those scoring higher than 18 points during our testing potentially being at a higher risk for falling. The current study was undertaken during the clinically defined “on” medication phase (1 hour after taking anti-PD medication). Foreman et al⁸ showed different assessment results when performed during clinical “on” versus “off” medication states. Physicians and therapists should be aware of the testing conditions.

Falls often occur in patients with PD, resulting in serious consequences such as bone fractures and traumatic brain injury, which further increase morbidity, mortality, and health care costs as well as the burden on patients and their families. It is of great importance to be able to predict those patients with PD who are at increased risk for falling. There are many factors that may affect fall risk such as lower extremity weakness, balance disorders, cognitive impairment, visual defects, medication factors, environmental factors, and care factors. The current study enrolled hospitalized patients with PD. Based on defined inclusion and exclusion criteria, we were able to exclude confounding conditions such as disease or trauma that may affect balance and gait (eg, stroke, lower extremity weakness, impaired vision, cognitive impairment). Evaluation time also was carefully planned to be 1 hour after a patient took the anti-PD medication, thus avoiding the potential effects of the medication. Additionally, the criteria for determination of fall excluded explained falls, thus excluding the environmental factors and nursing factors. Overall, the study design allowed us to exclude various confounding factors and to evaluate falls that are due to balance impairment. The risk of falling could be increased in patients with other risk factors mentioned.

We propose to use an FGA score of 18/30 as a threshold with high sensitivity and high specificity to predict those at risk for falling in the next 6 months. Physicians and therapists should be alerted to the risk of falling and provide appropriate intervention to the patient and caregivers to prevent falls, such as optimizing PD medication, offering physical therapy, and providing appropriate equipment and training. It is worthwhile emphasizing to physicians and therapists that the FGA threshold score may be different in other populations. When the FGA score is between 18 and 20, there is a high risk for falling, and appropriate precautions should be taken to avoid potentially serious adverse outcomes.

Limitations

One caveat related to the accuracy of fall incident recall is that patients were not provided with a logbook for recording fall events just after occurrence but rather relied on memory recall of fall events as recollected during follow-up by telephone interview 6 months after hospital discharge. A patient or caretaker may not able to recall fall history accurately due to various reasons such as cognitive and memory impairment and may be biased to recall only more recent or serious falls. As the follow-up period in this study was 6 months, yet PD is a progressively degenerative disease of the nervous system, the conclusions of this study should not be extended to a patient's long-term fall risk. In our study, the adjustment of anti-PD medications after the FGA evaluation may have improved patients' balance and gait function, which may have led to a different cutoff score for identifying fallers. Future research may focus on lengthening the follow-up time and exploring long-term fall risk as predicted by the FGA.