Abstract
Background The Functional Gait Assessment (FGA) is commonly used to measure walking balance. The minimal clinically important difference (MCID) has yet to be determined for the FGA.
Objective The purposes of this study were to determine: (1) the MCID in the FGA for older community-dwelling adults relative to patients' and physical therapists' estimates of change and (2) the extent of agreement between patients' and physical therapists' estimates of change.
Design This study was a prospective case series.
Methods Patients and physical therapists rated the amount of change in balance while walking after an episode of physical therapy for balance retraining on a 15-point global rating of change (GROC) scale. Weighted kappa statistics were calculated to express agreement between patients' and physical therapists' GROC ratings. Functional Gait Assessment change scores were plotted on receiver operating characteristic curves. A cutoff of +3 on the GROC was the criterion used for important change. The optimal FGA change cutoff score for MCID was determined, and sensitivity (SN), specificity (SP), and likelihood ratios (LRs) were calculated.
Results One hundred thirty-five community-dwelling older adults (average age=78.8 years) and 14 physical therapists participated. There was poor agreement between the patients' and therapists' ratings of change (weighted kappa=.163). The estimated MCID value for the FGA using physical therapists' ratings of change as an anchor was 4 points (SN=0.66, SP=0.84, LR+=4.07, LR−=0.40). No accurate value for the FGA MCID could be determined based on the patients' ratings of change.
Limitations The small sample size was a limitation.
Conclusion Poor agreement between therapists' and patients' ratings indicate the need for further communication relative to patient goals. The 4-point MCID value for the FGA can be used for goal setting, tracking patient progress, and program evaluation.
Measurement of change is the cornerstone of assessing patient progress and outcome. Change scores on outcome measures presumably reflect improvement or decline in the patient's status as a result of the treatment he or she is receiving. Meaningful interpretation of change scores must go beyond calculation of raw scores to the application of constructs such as the minimal clinically important difference (MCID). Jaeschke et al defined the MCID as “the smallest difference in score in the domain of interest which patients perceive as beneficial and which would mandate, in the absence of troublesome side effects and excessive cost, a change in patient's management.”1(p408) Other authors have defined it more simply as the “smallest difference in the score that is considered worthwhile or important.”2(p419) Achieving a change in score at least equal to the MCID suggests that the patient has undergone a change in condition that reflects important change. The MCID, as a useful psychometric property of an outcome measure, can be used in setting treatment goals, describing patient progress, and program evaluation.3
The assignment of a threshold score associated with the MCID is often calculated using the anchor method.3,4 This method references an external criterion, or anchor, associated with important change as a means of identifying people for whom important change has or has not occurred. Often the anchor for important change is established by directly surveying individuals about their perception of the change that has occurred as a result of treatment. The survey instrument frequently used for this purpose is the global rating of change scale (GROC). The GROC, first developed by Jaeschke et al,1 allows individuals to subjectively rate perceived change on a Likert scale with 7 levels of improvement, 7 levels of decline, and a zero midpoint. Each level is associated with a description of the amount of change that has occurred. A rating of +3 (“somewhat better”) or greater on the GROC is often used as the threshold for important change.1,5
Although the GROC is frequently used as the anchor for MCID, alternative anchors (eg, discharge destination, functional independence, return to work) have been used as indicators of important change.6 Using different anchors may lead to different estimates of the MCID of an outcome measure. For example, using GROC scores derived from a survey of physicians as the anchor, the MCID for the Functional Independence Measure (FIM) motor subscore was determined to be 17 points in people with subacute stroke.5 This finding compares with an MCID estimate of 11-point change on the FIM when using a change in disability, as measured the Modified Rankin Scale,7 as the anchor.8 Different anchors also may lead to variable accuracy of an MCID estimate, as reflected in the sensitivity (SN) or specificity (SP) of the identified cutoff score. For example, achievement of treatment goals9 and direct survey by a GROC10 have both been used as anchors for estimating the MCID for the Roland-Morris Back Pain Questionnaire. Both anchors generated a 5-point change as an estimate for the MCID, but the estimate from the GROC anchor rendered higher combined SN and SP, indicating a more accurate estimate. These anchor-dependent differences in MCID values and accuracy suggest that the MCID for any outcome measure is best studied from a variety of perspectives and anchors.3
There is another important anchor-related topic to consider when estimating and interpreting an MCID value: “Important? Who says so?” In other words, when direct surveys such as the GROC are used, we should consider who is being asked about the importance of change. Studies in the area of physical therapy and back pain10 have shown good agreement between patient and therapist regarding estimates of change. However, when estimating important change in gait speed after stroke,11 as well as in other fields such as oncology,12 pulmonology,13 cardiology,14 mental health,15 and rheumatology,16 there was little agreement between the clinician's and the patient's estimates of change as a result of treatment. Therefore, it is important when estimating important change that the opinions of both clinician and patient be considered.
Balance is a construct often addressed and measured as part of physical therapy intervention. One commonly used outcome measure of walking balance is the Functional Gait Assessment (FGA). The FGA was developed by Wrisley et al17 as a modification of the Dynamic Gait Index (DGI) in order to improve the reliability18 and reduce the observed ceiling effect of the DGI.19 Several psychometric properties of the FGA have been established (including good reliability,17 internal consistency,17 and moderate concurrent validity20) with other balance measures in patients with vestibular disorders. In addition, Marchetti et al21 reported an 8-point MCID for the FGA in people with vestibular disorders using an 18-point change on the Dizziness Handicap Inventory (DHI)22 as the external anchor. This 18-point change on the DHI has been related to standard error of measurement of the instrument and is thought to indicate an improvement in disability level.22 The value for the MCID of the FGA, however, has not yet been estimated by way of direct survey of patients and therapists using the GROC. This approach to estimating the MCID of the FGA should be investigated.
The purposes of this study were: (1) to determine the MCID for the FGA using patients' and therapists' ratings of change on the GROC as the anchor for important change and (2) to examine the extent of agreement between patients' and therapists' estimates of change in balance while walking after receiving physical therapy for balance retraining in a sample of older adults.
Method
Participants
Community-dwelling older adults who were referred for physical therapy for balance retraining were recruited in this case series study. Patients were eligible for the study if they were 60 years of age or older, living in the community independently or with help, able to walk 3.05 m (10 ft) with or without an assistive device, able to follow a 3-step command, and receiving physical therapy for balance retraining. Individuals diagnosed with positional vertigo were excluded from this study. The physical therapists who participated were staff members at 3 outpatient clinics in the Boston area: an independent clinic, 1 associated with a tertiary care hospital, and 1 associated with a community hospital. This study was approved by the institutional review boards for human subject research of the participating institutions.
We performed an a priori power analysis based on procedures suggested by Simel et al.23 With assumed SN and SP values of 0.75, we determined that a minimum of 78 participants reporting an important change were required so that the positive likelihood ratio (LR+) would exceed 2.0 (ie, the lower end of the 95% confidence interval [95% CI] for the LR+ would be >2.0).23 We chose 2.0 as the threshold for the LR+ because it has been suggested that an LR+ of 2.0 generates at least small, but sometimes important, change in the likelihood of having the condition or disease of interest in individuals who have a positive test result.24
Procedure
Patients were recruited from the 3 outpatient sites. They were screened by their physical therapist at the time of their initial physical therapist examination and were enrolled in the study if they met the inclusion and exclusion criteria and consented to participate. The patients then underwent a course of physical therapy that included an individualized balance retraining program. Patients and their treating therapists were asked to rate the change in walking balance that had occurred since the start of the patient's physical therapy treatment on a 15-point GROC.1 Patients made subjective estimates of change in their walking balance by answering the question “How much has your balance while walking changed since beginning physical therapy?” Therapists were asked to answer the question “How much has your patient's balance during walking changed since beginning this episode of outpatient physical therapy?” The threshold for MCID was set at a GROC score of +3 (“somewhat better”) (ratings of 0, +1, and +2=unimportant change; ratings of +3 or better=important change).1,3,9
We suspected that the great majority of patients and therapists would assess the change in balance as at least minimally important after a complete course of physical therapy. Accordingly, and based on a preliminary data analysis that revealed only 2 of the first 46 patients who completed the course of physical therapy had reported unimportant change (GROC score <+3), the protocol was modified such that patients enrolled later in the study and their therapists were surveyed instead at an interim time point of 2 to 4 weeks after the start of physical therapy. The FGA was administered at the same interim time point as the GROC scales were completed. This approach resulted in each patient and his or her therapist completing a GROC scale only once during the episode of care, either at the end of the episode of care (patients enrolled early in the study) or at an interim time point (patients enrolled later in the study). We expected that at an earlier time point in their course of care, important change in balance would not yet be appreciated by at least some of the patients and therapists. This approach resulted in the collection of a broader spectrum of GROC ratings and corresponding FGA change scores.
Variables
Data extracted from the medical record were participant's age, sex, diagnosis by the 9th revision of the International Classification of Diseases (ICD-9) coding, number of treatment sessions, duration of treatment in weeks, and FGA scores. The FGA has been validated for the assessment of balance in people with vestibular dysfunction,17 community-dwelling older adults,20,25 patients with stroke,25 and patients with Parkinson disease.26 The FGA has good reliability (intrarater reliability: intraclass correlation coefficient [ICC]=.7417; interrater reliability: ICC=.8617 to .9320) and internal consistency (Cronbach alpha=.7917). Concurrent validity was moderate for the Activities-specific Balance Confidence Scale (r=.5320 to .6417), DHI (r=−.64),17 perception of dizziness symptoms (r=−.70),17 Timed “Up & Go” Test (r=−.5017 to −.8420), DGI (r=.8017 to .9420), number of falls (r=−.66),17 and Berg Balance Scale (r=.84).20 The FGA score (≤22/30) provided 100% SN, 72% SP, LR+ of 3.6, and LR− of 0 to predict prospective falls in community-dwelling older adults.20 The minimal detectable change (MDC) for the FGA in patients with stroke has been determined as 4.2 points,25 and for patients with balance and vestibular disorders, it has been determined as 6 points.27
Therapist demographic variables collected included years of experience and whether they were certified as clinical specialists by the American Board of Physical Therapy Specialties (ABPTS).
Data Analysis
Statistics were generated using PASW Statistics version 21 (IBM Corp, Armonk, New York). Descriptive statistics were generated for demographic variables and FGA scores for the whole sample of patients and for therapists' demographics. For further subgroup descriptive data and for group comparisons, as well as to calculate the FGA MCID value, patients were dichotomized in 2 different ways. First, they were dichotomized according to whether they rated themselves as having achieved MCID in walking balance as indicated by a score of +3 or greater on the GROC versus less than +3 (score of 0, +1, or +2). Second, they were dichotomized based on whether their therapists rated them as having achieved MCID with a GROC score of +3 or greater versus less than +3.
Group comparisons were performed between patients who experienced MCID and those who did not using unpaired t tests (age), Mann-Whitney U test (initial, discharge, and change FGA scores), and chi-square analysis (sex) with 2-tailed significance set at P<.05 level. Age (t test), sex (chi-square analysis), and initial FGA score (Mann-Whitney U test) were compared between patients who were assessed at the interim time point and those who were assessed after the complete episode of physical therapy care to determine if there was any difference in baseline characteristics (age, sex, and initial impairment level as reflected in initial FGA scores) that may have affected the perspective on change between the 2 groups. We also compared the total number of treatments and the number of weeks in treatment between patients who were assessed at the interim time point and those who were assessed after the complete episode of physical therapy care. Age (1-way analysis of variance), sex (chi-square analysis), and initial FGA scores (Kruskal-Wallis test) also were compared across the 3 participating clinics to determine if there were significant differences in these variables at the different sites at baseline.
The extent of agreement between patients' and therapists' estimates of change in balance while walking was calculated using a weighted kappa. First, the 8 levels of improvement on the 15-point GROC (0 to +7) were collapsed into 3 levels such that scores of 0 to +2 were considered as “no change,” +3 to +4 as “minimal change,” and +5 or greater as “moderate change.”5,10 Kappa values were then calculated with the following assigned weights: 0 for perfect agreement between the patients' and therapists' ratings, 2 for disagreements between minimal and moderate change, 4 for disagreements between minimal and no change, and 6 for disagreements between no change and moderate change.
We generated receiver operating characteristic (ROC) curves to determine the change score on the FGA that most accurately classified participants as having achieved MCID or not in their balance while walking. Participants were dichotomized based on a cutoff of +3 on the 15-point GROC scale. Change scores on the FGA (discharge FGA score minus initial FGA score) were plotted on the ROC curve. The optimal FGA cutoff was identified as the point nearest the upper left-hand corner of the ROC curve, as this point rendered the best balance between SN and SP. We also calculated LR+ (SN/1 − SP) and negative LR− (LR; 1 − SN/SP) based on the FGA cutoff score. We conducted separate analyses on subgroups of participants dichotomized as having achieved MCID or not based on patients' GROC ratings and those of the therapists.
Therapists' and patients' GROC ratings were correlated with initial and discharge FGA scores using the Spearman rho as a means to evaluate potential recall bias.28 We generated correlations for the whole sample, as well as for the patients who were surveyed at the interim time point and those patients who were surveyed at the end of their episode of care, in order to examine the effect of the time of survey on potential recall bias. Change estimates are often biased toward the current state because it is difficult to remember the initial condition.15 The criteria for assessments of change that are unbiased by recall are a negative correlation with the initial state and an equally positive correlation with the final state.28
Results
Of the 202 patients who were enrolled in the study, analysis was based on the 135 patients who completed all surveys at the interim time point or on the last physical therapy visit and had surveys completed by the treating therapist. When comparing age, sex, and initial FGA scores, there were no significant differences between those participants with complete data (N=135) and those lost to follow-up or with missing data (n=67). There were no significant differences in baseline characteristics among participants recruited from the 3 outpatient sites. We also saw no differences in the baseline characteristics between the participants surveyed at the interim time point (n=49) compared with those surveyed at the end of their episode of care (n=86). We performed these comparisons to ascertain if there were any differences in variables of sex, age, or initial impairment level that may have biased the perspective on change of one group compared with the other. Having seen no differences in these variables between groups, we combined the data from both groups for all analyses. As expected, the mean number of treatments for the interim group was 6.3 (SD=1.36), and mean number of weeks of treatment was 3.5 (SD=1.14), which were both significantly less than the full episode of care group (mean number of treatments=1.65, SD=4.4; mean number of weeks=8.32, SD=3.54).
Participants had various ICD-9 diagnoses, with the most common being gait disorders (61.7%) followed by vertigo (17.0%) and impaired balance (9.6%). In addition to the multifactorial balance disorders represented in the sample, the sample also included 2 people with stroke, 5 with diagnosed peripheral neuropathy, 2 with low back pain, 1 with pelvic fracture, and 1 with spinal stenosis. The mean age was 78.8 years, and the sample was 63.7% female. Participant descriptive statistics are presented in Table 1. The median initial FGA score was 18, below the score of 22 established as a fall risk threshold.20
Descriptive Statistics for Whole Sample and by MCID Categorizations and Group Comparisonsa
Poor agreement (weighted kappa=.163) was observed between patients' and therapists' estimates of change in balance while walking. The weighted kappa results are shown in Table 2. Based on patients' own estimates of change, the median GROC score was +5 (range=+1 to +7). The median GROC score based on therapists' estimates was +4 (range=0 to +7). Descriptive statistics and comparisons by patient group based on the patients' and therapists' MCID ratings are presented in Table 1. When participants were dichotomized based on their own ratings of whether they achieved MCID in their walking balance, there was no difference in any of the variables analyzed. By contrast, when participants were dichotomized based on the therapists' ratings of whether the patients achieved MCID, there were significant differences in the number of visits, duration of treatment, and discharge and change FGA scores (Tab. 1).
Agreement: Weighted Kappaa
The ROC analyses results are presented in Table 3. When participants were dichotomized based on their own estimates of having achieved MCID (+3 or better on the GROC), an FGA change of 5 points was identified from the ROC curve as the point that maximizes SN and SP. However, the null value of .50 was contained in the 95% CI values for the area under the curve (AUC) (95% CI=0.43, 0.77) and SP (95% CI=0.48, 0.97). The inclusion of the null value in the 95% CI of the AUC indicates that the 5-point MCID value, based on patient's ratings of change, may be no better than a 50/50 chance at distinguishing between those participants who reported important change and those who reported unimportant change. In contrast, when participants were dichotomized based on the therapists' ratings of change, a 4-point FGA change score was identified as the value associated with the MCID (AUC=0.83, SN=0.66, and SP=0.84), and none of the associated 95% CIs contained the null value (AUC: 95% CI=0.76, 0.90; SN: 95% CI=0.55, 0.76; and SP: 95% CI=0.69, 0.93). Figure 1 presents the distributions of FGA change scores according to patient subgroups of having achieved MCID. Figure 2 presents the ROC curves based on patients' and therapists' ratings of change.
MCID Cutoff Scores, Sensitivity, Specificity, and Likelihood Ratiosa
Distribution of Functional Gait Assessment (FGA) change scores by patient subgroup. (A) Distribution of FGA change scores by patient subgroups of minimal clinically important difference (MCID) or not based on patients' ratings of change. MCID ratings based on +3 or better on the global rating of change (GROC) scale; no MCID based on GROC score of 0, 1, or 2. (B) Distribution of FGA change scores by patient subgroups of MCID or not based on therapists' ratings of change.
Receiver operating characteristic (ROC) curves. Values are Functional Gait Assessment (FGA) change scores. (A) ROC curve based on patients' ratings of change. (B) ROC curve based on therapists' ratings of change. Diagonal line represents area under the curve of 0.50. Arrow indicates optimal cutoff score with the best balance between sensitivity and specificity.
The 14 therapists who participated in the study had an average of 12.1 years of experience (SD=10.8, range=1–33), and 4 were ABPTS-certified clinical specialists (3 Neurologic Certified Specialists [NCS] and 1 Geriatric Certified Specialist [GCS]).
Correlations between therapists' and patients' GROC ratings with FGA initial and discharge scores are reported in Table 4. Correlations ranged from −.159 to .298. There was some evidence of recall bias in participants in the full episode of care group, as they had GROC scores that were positively correlated with initial FGA scores.
Spearman Rho Correlations Between GROC Ratings and FGA Scoresa
Discussion
We estimated the MCID for the FGA based on the direct survey of older adult patients and their physical therapists after physical therapy for balance retraining. This is the first study to use direct survey to estimate the MCID for the FGA and to survey both patients and their therapists relative to perceived change in balance while walking. This approach appears to have revealed important discrepancies in the perception of change between patient and therapist. Poor agreement was observed between patients' and therapists' estimates of change in walking balance after balance retraining in this study. Only 11 patients rated themselves as having experienced unimportant change, whereas therapists placed 43 patients in that category. Disagreement between patient and clinician in the perception of important change is not uncommon.11–13,15,16 One reason for the disagreement on change in balance could be that therapists and patients were using different reference points for rating change. Patients may have been using their everyday life and tasks as their reference point, whereas therapists were evaluating balance while walking in the limited context of the clinical setting. The disagreement observed between patients' and therapists' ratings of change highlights the need for more explicit communication between the therapist and patient regarding goals and potential for improvement in therapy. Patient-centered care requires that we carefully and explicitly consider the patient's perspective.12,29 Careful consideration of the patient's perspective when establishing the plan of care leads to improved patient satisfaction and adherence.15
The differences in patients' and therapists' ratings also may be explained by different perspectives on potential improvement. Therapists may have relied on their clinical knowledge and based their estimates of change on what they perceived as the maximum possible improvement. To explore this possibility, we conducted a post hoc analysis to further examine the effect of the interim time point assessments on ratings of change and found that the therapists' GROC ratings were significantly lower at the interim time point than those recorded at the end of the episode of care (Mann-Whitney U test, P<.000). Patients' GROC ratings, by contrast, were not significantly different between those who completed the GROC at the interim time point assessment and those who completed the GROC at the end of the episode of care (P=.877). Furthermore, patients who did not achieve MCID based on the therapists' ratings had significantly fewer treatment sessions and a shorter duration of treatment. These findings may support the notion that therapists were rating change based on their perspective of what was possible by the end of treatment. This perspective may have contributed to the disagreement observed between patients' and therapists' ratings of change.
An accurate estimate for MCID value could not be derived from the patients' ratings of change. The inclusion of the null value in the 95% CI for the AUC, SN, and SP indicated that the 5-point value derived from the ROC curve was no better than chance in identifying patients who had undergone important change from those who did not and so renders this estimate for the MCID clinically useless. Furthermore, the AUC of 0.60 indicates an unacceptable level of diagnostic accuracy.30 One potential explanation for the lack of a precise MCID estimate based on the patients' ratings of change could be that some important and meaningful aspect of walking balance is not assessed by the items of the FGA. A qualitative study of patients' opinions on what aspects of activity are related to important change in walking balance may elucidate the underlying constructs. Another potential contributor to the inaccuracy of the estimate may be the issue of recall bias. Recall bias can affect MCID estimates when ratings of change are made retrospectively,28 as they were in the current study. In the current study, there was evidence, based on the correlations, that both the patients and therapists may have been influenced by recall bias, but the patients were affected to a greater extent than were the therapists, particularly the patients surveyed at the end of their episode of care. One way to minimize recall bias in future studies would be to conduct a prospective study of change where patients would be asked to rate change through an episode of care at frequent time points.28,31
When interpreting the current findings, it is important to consider that the MCID value for a particular measure may vary depending on the anchor, setting, patient population, and diagnosis.3 Compared with our 4-point estimate of the MCID for the FGA, Marchetti et al21 determined an 8-point change as the MCID for the FGA, with an SN of 63% and an SP of 59%, in a sample of patients with balance and vestibular disorders. They used a change in disability level as indicated by an 18-point change on the DHI as the external anchor for grouping participants as having achieved important change or not. One reason for the difference in MCID estimates between the study by Marchetti et al and the current findings may be related to the notion that change in disability level may not be closely related to the single construct of balance while walking. Our survey question focused directly on the construct of balance while walking, which may explain why our 4-point MCID estimate had higher SN and SP values than the estimate generated by Marchetti et al.21 Another consideration is that the sample in the study by Marchetti et al21 was younger (mean age=55 years) than the current sample, so the higher estimate for the FGA MCID value may reflect these younger patients' higher expectations related to their outcomes. The difference in estimates between the study by Marchetti et al and the current findings also may be a reflection of the difference in the diagnostic groups in each study. Our study included only older adults with general balance disorders and specifically excluded any patients with a diagnosis of benign paroxysmal positional vertigo. Because any single MCID value is only an estimate, further research is needed to establish MCID values for various patient groups using various anchors to best understand the range of values that can be associated with this psychometric property.
Our determination of the MCID value of 4 points is smaller than the FGA MDC value of 4.2 points that was determined for patients with stroke25 and the MDC estimate by Marchetti et al27 of 6 points in a group of patients with balance and vestibular disorders. Theoretically, the MCID value should be greater than the MDC, which is a threshold for measurement error.3 Values for MCID, as well as MDC, can vary from patient group to patient group.3,32 Differences in baseline characteristics of the groups studied may explain why the MDC estimates of the FGA exceed the MCID estimate from the current study. Another contributing factor may be the approach used to calculate the MDC, as it has been suggested that distribution-based approaches for calculating MDC may result in larger estimates of the MDC than using an anchor-based method.33 These discrepancies suggest that psychometric properties such as the MCID and the MDC need to be studied from a variety of approaches on a variety of patient groups.
In the present study, the therapists were aware of the purpose of the study and that the FGA was the outcome measure of interest. In addition, they were not given explicit instructions as to when to complete the FGA relative to completing the GROC, and, based on a post-analysis query, all but 1 of the 14 participating therapists indicated that they usually completed the GROC after scoring the FGA. Thus, therapists likely anchored their estimates of change based on both the observed balance performance of their patients and their knowledge of FGA scores and so were using more than one perspective when estimating MCID values. Future studies may standardize the sequence for scoring the FGA relative to the GROC and perhaps implement a formalized triangulation approach14 for estimating the MCID of the FGA where more than one perspective is considered in determining MCID, such as expert opinion, as well as survey estimates on a GROC.
The approach of surveying patients either earlier or at the end of the episode of care has advantages over surveying all patients only at the end of care. Because it would be expected that most patients who complete their course of therapy would indicate at least minimal change had occurred, it allowed us to capture ratings from patients for whom important change had not yet occurred. This approach consequently yielded more varied GROC ratings and corresponding FGA change scores. Another benefit of this data collection approach is that the interim group had less recall bias than the group surveyed at the end of their episode of care, attenuating this bias in the overall sample. Another approach to populating the group of patients reporting unimportant change might have been to use a GROC cutoff score of +5, rather than +3, as Fulk and colleagues11 did when determining the MCID for gait speed in patients poststroke. We were reluctant to use +5 as the cutoff because the language on the GROC scale would indicate this amount of change to be “a good deal better,” even more than “moderate” (+4), based on the language of the GROC scale, and the construct of interest in the present study was the MCID.
One of the limitations of our study was a small sample size. A larger sample size would have allowed for categorization based on initial impairment levels. This approach would be helpful because it has been demonstrated that MCID scores vary depending on baseline scores.5,34 Further research utilizing other relevant anchors, such as attainment of mobility goals, may further inform our interpretation of the MCID values for the FGA. Clinical diagnoses were based on ICD-9 codes. Further categorizing patients based on physical therapy–based diagnoses would be more descriptive and informative. The generalizability of these results is limited to older adults with characteristics similar to those of this study's sample. Caution should be used in generalizing these current findings to patients in other settings with other diagnoses or characteristics.
In summary, poor agreement between therapists' and patients' estimates of change indicate the need for further communication. This communication needs to be directed toward formation of goals and outcomes that are patient related and that the patient perceives as important. Finally, 4-point change was determined as the MCID value for the FGA in community-dwelling older adults. This value can be used for program evaluation as well as for tracking individual patient progress.
The Bottom Line
What do we already know about this topic?
Change scores on the Functional Gait Assessment (FGA) are used as indicators of improvement in walking balance. Patients and therapists, however, often disagree on how much change has occurred as a result of treatment.
What new information does this study offer?
The amount of change on the FGA that corresponds with at least minimally important improvement was 4 points. There was significant disagreement between patients and therapists on how much change occurred as a result of treatment.
If you're a patient, what might these findings mean for you?
It is important for patients to talk with their therapists about exactly what aspect of function will improve as a result of physical therapist intervention and how that improvement will be measured.
Footnotes
Dr Beninato and Ms Fernandes provided concept/idea/research design and data analysis. All authors provided writing. Dr Beninato provided data collection and project management. Dr Plummer provided consultation (including review of the manuscript before submission).
This research was presented as a poster at the American Physical Therapy Association's Combined Section Meeting; February 8–12, 2012; Chicago, Illinois.
- Received December 13, 2013.
- Accepted June 2, 2014.
- © 2014 American Physical Therapy Association
References
Issue highlights
