The Six-Minute Walk Test Predicts Ambulation After TKA
The Six-Minute Walk Test Predicts Ambulation After TKA
This study deepens our understanding of walking ability of individuals one year after TKA surgery. The prevailing norm in this area is to use short duration walk tests to evaluate functional mobility after surgery. It is unclear whether such tests truly reflect ambulatory capacity more typical of daily requirements in this population. This study contributes to the existing literature on both ambulation capacity after TKA and tests used to measure capacity. For the first time, we have data which provide comparisons between age-matched norms and TKA recipients in terms of distance walked over a 30 minute period. Likewise, we now have confirmation that the 6MW test is an excellent predictor of longer duration ambulation after TKA. Consequently, our study provides long overdue construct validity for the use of 6MW test as a measure of functional ambulation in TKA recipients. Because of the close relationship between the 6MW and 30MW tests, contrary to our expectations, the 30MW test did not demonstrate greater capacity to differentiate between individuals nor have stronger correlations with patient-reported outcomes compared to the 6MW test. As such, an extended walk test is no more useful as a research tool than the shorter walk tests in this population and yet is more arduous.
We observed that the 30-minute walking capacity of TKA recipients at one year after surgery remains significantly inferior to their healthy counterparts. Given the similar effort observed during the 30MW test between TKA recipients and controls and the subjective reporting by participants, the inferior performance is likely explained by the presence of a prosthetic joint and pain in the index joint, as well as other factors shown to influence recovery such as the presence of significant co-morbidities, severe other joint disease, and obesity. Lower physical activity levels reported by the TKA cohort, which are a likely manifestation of the co-morbidities, joint disease and obesity associated with patients requiring TKA, may also be a contributing factor to their poorer performance. Despite their limitations, all TKA recipients completed the test without adverse events, indicating that though the test is arduous, it appears to be safe. The 30MW test was observed to be highly repeatable. The healthy cohort walked significantly further by 50 m in the second test. This difference may be due to a practice effect, which has been described by some authors in the 6MW test. It is likely that a significant increase was not seen in the TKA cohort because there was greater variability in their performance, as indicated by a larger standard deviation of the difference between the first and second test (TKA SD 215 m vs Controls SD 93 m). The reluctance by patients to repeat the test reflects that it may not be embraced in this population, but this does not alter the new insights gained about ambulatory capacity, the various associations between the walk tests, or between the walk tests and patient-reported outcomes.
The observed residual impairment in mobility amongst our TKA cohort is further supported by comparisons between their 6MW test performance with published standard values of healthy elderly populations. Using the reference equation published by Troosters et al., our healthy cohort performed very similarly to their predicted distance (582 m vs predicted 576 m), while the performance of our TKA cohort was much lower than their predicted distance (424 m vs predicted 538 m). As this reference equation takes into account age, gender, height and weight (the constituents of BMI), it demonstrates that the higher BMI of the TKA cohort only makes a minor contribution to the large difference in walking performance compared to the healthy cohort. Using reference equations published by Enright et al., which have been considered by some to underestimate normal walking distance, the performance of our TKA cohort was still inferior, with the predicted distance for TKA group of 464 m.
The near-perfect correlation between the 6MW and 30MW tests and the high predictability of 30MW distance by 6MW distance were unexpected. This is because participants were instructed to walk as far as possible but safely, and as such we expected that some participants would not be able to maintain the same speed in both tests, presuming a comparatively poor fitness level in the cohort. We observed that the healthy cohort walked faster in the longer test. This may be due to less frequent deceleration and turning associated with the longer lap in the 30MW test. In contrast, the walking speeds of TKA recipients were identical in both tests. The gain in walk speed associated with a longer track was not seen in the TKA cohort, suggesting that the 30MW test was indeed more fatiguing for this cohort. For some participants, the overall walk speed was reduced as a result of the rests taken during the 30MW test and these people demonstrated the slowest speeds in the 6MW test. For those who did not rest, there was a trend for a slightly faster speed but this did not quite reach statistical significance. Consequently, whilst fatigue likely did become a factor in the longer test, its effects were small, and those most affected were those who performed worst in the 6MW test anyway. Thus the prediction of the 30MWD and 6MWD was high. The knowledge that a strong correlation exists between the 6MW and 30MW tests is useful both in clinical practice and research, as it means that the 6MW test is a suitable tool to evaluate interventions aimed at improving longer duration ambulation; specifically, the 6MW test will predict how far a TKA recipient can walk over 30 minutes with acceptable accuracy.
All three walk tests moderately predicted self-reported function. The 6MW and 30MW tests produced similar models for reasons described above, with TUG producing the strongest model. A possible explanation for the latter is that, unlike the 30MW and 6MW tests, TUG performance is a combination of walking ability, lower limb strength and balance owing to the sit-to-stand component, and is based on activities similar to items included in the WOMAC questionnaire. Others have found similar, modest relationships between measured ambulation and self-reported function. Rossi et al. found moderate correlation between TUG and perceived function amongst participants 17 months after TKA. Another study examining short-term recovery also found only moderate correlation between self-reported function and performance based outcomes. Our results, based on a cohort 1 to 1.5 years post-surgery, add to the growing body of literature justifying the collection of both performance-based and self-reported measures in assessing longer-term recovery after TKA.
Physical activity levels reported by the TKA cohort were significantly lower than the control group. The average time spent performing physical activities by TKA recipients (202 minutes) was also lower than reported in a similarly aged population-based sample (255 minutes) using the same survey, whilst the activity levels of our healthy cohort (mean 504 min) were much greater. Whilst these comparisons with population-based data appear to corroborate the representativeness of both our patient and healthy cohorts in terms of self-reported physical activity, we remain unclear as to the relationship between self-reported physical activity and measured ambulation. TUG was a significant, but weak predictor of self-reported physical activity in our TKA cohort. A possible explanation for the weak relationship is that physical activity is determined by other factors such as self–efficacy, personality, knowledge, beliefs and social support and not just physical mobility. Alternatively, patient-reported physical activity may not reliably measure actual physical activity. It is not known how accurately the survey questions we used correlate with actual performance. We used questions used by a large Australian population-based study - the National Physical Activity Survey - for the purposes of having a population-based comparison. The trade-off for this approach is the lack of certainty with which these questions do reliably reflect actual physical activity.
A limitation to the study is that the results are limited to the TKA population one year post-surgery. A minor limitation of the study is the sample size of the TKA cohort as it necessarily restricted the number of variables included in the regression modeling. A larger sample would have allowed the inclusion of other potential predictors, such as age and co-morbidity which are known predictors of the shorter walk tests. This notwithstanding, as 6MW distance was such a strong predictor of the 30MW distance, it is unlikely that inclusion of other variables would change the definitive model substantially. Furthermore, factors such as age, gender and BMI are accounted for to some degree by their association with 6MW distance. A potential limitation of the study was that our TKA cohort included TKA recipients with and without the common mobility-limiting co-morbidities. A sample which excluded patients with significant co-morbidity would have enabled a more straight-forward comparison between age-matched peers and TKA recipients, and eliminate the variability in walk tests due to factors other than the TKA surgery. However, such a sample would not be representative of typical TKA cohorts. We note that subgroup analyses of TKA recipients with minimum co-morbidities still demonstrated significantly inferior performance compared to healthy controls. Not randomizing the order of testing to eliminate the effect of fatigue during testing could be viewed as another potential limitation in this study. However, the amount of time required after the 30MW test to ensure sufficient rest made this impractical. Furthermore, it was unlikely that fatigue affected testing because amongst the TKA cohort, in whom fatigue would have a larger potential impact, those who returned for a repeat 30MW test (which did not follow a 6MW test) did not perform significantly better in the second test.
Discussion
This study deepens our understanding of walking ability of individuals one year after TKA surgery. The prevailing norm in this area is to use short duration walk tests to evaluate functional mobility after surgery. It is unclear whether such tests truly reflect ambulatory capacity more typical of daily requirements in this population. This study contributes to the existing literature on both ambulation capacity after TKA and tests used to measure capacity. For the first time, we have data which provide comparisons between age-matched norms and TKA recipients in terms of distance walked over a 30 minute period. Likewise, we now have confirmation that the 6MW test is an excellent predictor of longer duration ambulation after TKA. Consequently, our study provides long overdue construct validity for the use of 6MW test as a measure of functional ambulation in TKA recipients. Because of the close relationship between the 6MW and 30MW tests, contrary to our expectations, the 30MW test did not demonstrate greater capacity to differentiate between individuals nor have stronger correlations with patient-reported outcomes compared to the 6MW test. As such, an extended walk test is no more useful as a research tool than the shorter walk tests in this population and yet is more arduous.
We observed that the 30-minute walking capacity of TKA recipients at one year after surgery remains significantly inferior to their healthy counterparts. Given the similar effort observed during the 30MW test between TKA recipients and controls and the subjective reporting by participants, the inferior performance is likely explained by the presence of a prosthetic joint and pain in the index joint, as well as other factors shown to influence recovery such as the presence of significant co-morbidities, severe other joint disease, and obesity. Lower physical activity levels reported by the TKA cohort, which are a likely manifestation of the co-morbidities, joint disease and obesity associated with patients requiring TKA, may also be a contributing factor to their poorer performance. Despite their limitations, all TKA recipients completed the test without adverse events, indicating that though the test is arduous, it appears to be safe. The 30MW test was observed to be highly repeatable. The healthy cohort walked significantly further by 50 m in the second test. This difference may be due to a practice effect, which has been described by some authors in the 6MW test. It is likely that a significant increase was not seen in the TKA cohort because there was greater variability in their performance, as indicated by a larger standard deviation of the difference between the first and second test (TKA SD 215 m vs Controls SD 93 m). The reluctance by patients to repeat the test reflects that it may not be embraced in this population, but this does not alter the new insights gained about ambulatory capacity, the various associations between the walk tests, or between the walk tests and patient-reported outcomes.
The observed residual impairment in mobility amongst our TKA cohort is further supported by comparisons between their 6MW test performance with published standard values of healthy elderly populations. Using the reference equation published by Troosters et al., our healthy cohort performed very similarly to their predicted distance (582 m vs predicted 576 m), while the performance of our TKA cohort was much lower than their predicted distance (424 m vs predicted 538 m). As this reference equation takes into account age, gender, height and weight (the constituents of BMI), it demonstrates that the higher BMI of the TKA cohort only makes a minor contribution to the large difference in walking performance compared to the healthy cohort. Using reference equations published by Enright et al., which have been considered by some to underestimate normal walking distance, the performance of our TKA cohort was still inferior, with the predicted distance for TKA group of 464 m.
The near-perfect correlation between the 6MW and 30MW tests and the high predictability of 30MW distance by 6MW distance were unexpected. This is because participants were instructed to walk as far as possible but safely, and as such we expected that some participants would not be able to maintain the same speed in both tests, presuming a comparatively poor fitness level in the cohort. We observed that the healthy cohort walked faster in the longer test. This may be due to less frequent deceleration and turning associated with the longer lap in the 30MW test. In contrast, the walking speeds of TKA recipients were identical in both tests. The gain in walk speed associated with a longer track was not seen in the TKA cohort, suggesting that the 30MW test was indeed more fatiguing for this cohort. For some participants, the overall walk speed was reduced as a result of the rests taken during the 30MW test and these people demonstrated the slowest speeds in the 6MW test. For those who did not rest, there was a trend for a slightly faster speed but this did not quite reach statistical significance. Consequently, whilst fatigue likely did become a factor in the longer test, its effects were small, and those most affected were those who performed worst in the 6MW test anyway. Thus the prediction of the 30MWD and 6MWD was high. The knowledge that a strong correlation exists between the 6MW and 30MW tests is useful both in clinical practice and research, as it means that the 6MW test is a suitable tool to evaluate interventions aimed at improving longer duration ambulation; specifically, the 6MW test will predict how far a TKA recipient can walk over 30 minutes with acceptable accuracy.
All three walk tests moderately predicted self-reported function. The 6MW and 30MW tests produced similar models for reasons described above, with TUG producing the strongest model. A possible explanation for the latter is that, unlike the 30MW and 6MW tests, TUG performance is a combination of walking ability, lower limb strength and balance owing to the sit-to-stand component, and is based on activities similar to items included in the WOMAC questionnaire. Others have found similar, modest relationships between measured ambulation and self-reported function. Rossi et al. found moderate correlation between TUG and perceived function amongst participants 17 months after TKA. Another study examining short-term recovery also found only moderate correlation between self-reported function and performance based outcomes. Our results, based on a cohort 1 to 1.5 years post-surgery, add to the growing body of literature justifying the collection of both performance-based and self-reported measures in assessing longer-term recovery after TKA.
Physical activity levels reported by the TKA cohort were significantly lower than the control group. The average time spent performing physical activities by TKA recipients (202 minutes) was also lower than reported in a similarly aged population-based sample (255 minutes) using the same survey, whilst the activity levels of our healthy cohort (mean 504 min) were much greater. Whilst these comparisons with population-based data appear to corroborate the representativeness of both our patient and healthy cohorts in terms of self-reported physical activity, we remain unclear as to the relationship between self-reported physical activity and measured ambulation. TUG was a significant, but weak predictor of self-reported physical activity in our TKA cohort. A possible explanation for the weak relationship is that physical activity is determined by other factors such as self–efficacy, personality, knowledge, beliefs and social support and not just physical mobility. Alternatively, patient-reported physical activity may not reliably measure actual physical activity. It is not known how accurately the survey questions we used correlate with actual performance. We used questions used by a large Australian population-based study - the National Physical Activity Survey - for the purposes of having a population-based comparison. The trade-off for this approach is the lack of certainty with which these questions do reliably reflect actual physical activity.
A limitation to the study is that the results are limited to the TKA population one year post-surgery. A minor limitation of the study is the sample size of the TKA cohort as it necessarily restricted the number of variables included in the regression modeling. A larger sample would have allowed the inclusion of other potential predictors, such as age and co-morbidity which are known predictors of the shorter walk tests. This notwithstanding, as 6MW distance was such a strong predictor of the 30MW distance, it is unlikely that inclusion of other variables would change the definitive model substantially. Furthermore, factors such as age, gender and BMI are accounted for to some degree by their association with 6MW distance. A potential limitation of the study was that our TKA cohort included TKA recipients with and without the common mobility-limiting co-morbidities. A sample which excluded patients with significant co-morbidity would have enabled a more straight-forward comparison between age-matched peers and TKA recipients, and eliminate the variability in walk tests due to factors other than the TKA surgery. However, such a sample would not be representative of typical TKA cohorts. We note that subgroup analyses of TKA recipients with minimum co-morbidities still demonstrated significantly inferior performance compared to healthy controls. Not randomizing the order of testing to eliminate the effect of fatigue during testing could be viewed as another potential limitation in this study. However, the amount of time required after the 30MW test to ensure sufficient rest made this impractical. Furthermore, it was unlikely that fatigue affected testing because amongst the TKA cohort, in whom fatigue would have a larger potential impact, those who returned for a repeat 30MW test (which did not follow a 6MW test) did not perform significantly better in the second test.