Abstract
Dear Professor Johnson,
We would like to comment on the paper of J.H. Campbell, et al., “OASIS I: Retrospective analysis of four different microprocessor knee types”, published in the Journal of Rehabilitation and Assistive Technology Engineering 2020, Volume 7, pages 1–10. We commend the authors for their impressive study that is a valuable contribution to the body of evidence as it confirms the generalizability of smaller studies with microprocessor-controlled knees (MPK) to real-life conditions in a large sample and assessed injurious falls that may serve as a solid foundation for future cost-effectiveness analyses. We congratulate the authors on the big achievement to have completed a study with 602 participants successfully. Nevertheless, we believe that some points in the paper require clarification.
First, the authors refer to their study as “retrospective” as the most important aspect of the study design. However, we think it would be more important to note that the study design was cross-sectional, which has considerable consequences for the interpretation of data and foundation of conclusions. By nature, cross-sectional study designs assess outcomes at only one point in time and, thus, do not allow for any conclusions on causal relationships to explain the results. Therefore, all the interpretations and conclusions offered by the authors should have been clearly marked as hypotheses that require further study to confirm or reject rather than facts supported by the results. Of course, the authors have all the right to state their opinion which of the hypotheses they deem the most likely ones to explain the results. Nevertheless, it would have been appropriate to also discuss other possible explanations for the results (see below), especially in lieu of the claimed “absence of bias”.
Second, as this study did not assess a baseline for using non-microprocessor prosthetic knees (NMPK), the absence of statistical differences between the outcomes does not necessarily mean that there are no differences in the effects of the four MPK. As there is no information on the baseline status of the patients included, the study was unable to control for a potential clinician bias in the selection of the different MPK. If clinicians preferred a certain MPK for patients with increased risk of falling and lower baseline mobility, the “parity” of outcomes would conceal a clear difference in effects of the four MPK on injurious falls and mobility. This limitation is important as there is comparative evidence that has already demonstrated that there are meaningful technical and functional differences between the MPK (e.g. Bellmann et al., 2010; Thiele et al., 2014; Thiele et al., 2019; Bellmann et al., 2019). Thus, another hypothesis to explain the results would have been that the Hanger prosthetists apparently choose MPK wisely when fitting their patients, making sure that the final outcomes of patients with different levels of baseline mobility are not significantly different.
Third, we find the use of the term “parity” confusing and scientifically inadequate. “Parity” is more of a legal rather than a scientific term. The established statistical approaches to compare the differential effects and outcomes of similar interventions are equivalence and non-inferiority. However, these would have required the use of completely different statistical methods as the simple absence of statistical differences does neither establish equivalence nor non-inferiority.
Fourth, another scientifically inappropriate wording affects the interpretation of outcomes of patients in the various age bins. The use of the term “decline” to characterize the differences in outcomes suggests a time series that reflects a decrease in outcomes over time when patients age with their MPK. However, this is not what this study measured. Given the facts that the study was cross-sectional and, thus, assessed outcomes only once, did not provide any information on how long the patients had been amputated and using MPK, and that the single-most frequent amputation etiology was vascular disease/diabetes, it is reasonable to assume that the majority of patients were relatively recent rather than long-time amputees. Thus, the study did not find a decline over time but the unsurprising fact that patients who undergo an amputation later in life start, on average/median, with a lower baseline mobility than younger amputees. Inasmuch, the finding that “C-leg … users saw declines with advanced aging.” may even be interpreted as support for the above-mentioned hypothesis of a clinician bias with preference for the C-Leg for patients with low baseline mobility.
Fifth, the authors stated “The ability for the C-Leg and Orion to significantly reduce injurious falls compared to nMPKs would suggest that when stability and falls reduction is considered a primary aim for MPK prescription, these models may be more indicated.” Though the differences in injurious falls between the four MPK were not statistically significant, they may be considered clinically meaningful. Based on 100 patients fitted with each MPK, the C-Leg may spare injurious falls in 10 patients compared to NMPK, in 8 patients compared to the Rheo Knee, in 7 patients compared to the Plié, and in 3 patients compared to the Orion. These differences should not be neglected. At any rate, it is unfortunate that the authors stopped short of drawing the conclusion to clearly recommend to routinely assess the risk of falling in transfemoral amputees to guide the selection of an MPK.
We would be grateful if the authors could weigh in on our comments.
Sincerely yours, Andreas Kannenberg and Andreas Hahn
