Point-of-Care Ultrasound Frailty Assessments: Comment.

Abstract

Listen

Translate article

We read with great interest the article by Canales et al., 1 who benchmarked the predictive value of four surrogates of preoperative sarcopenia and the Fried phenotype frailty tool for predefined adverse postoperative outcomes. We salute the authors on this pioneering and important investigation and would like to highlight several points.First, the authors used three novel, bedside, sonographic measurements of the quadriceps muscle, as well as a tomographic assessment of the psoas muscle. For the assessment of generalized sarcopenia, the abdominal muscles, including the psoas, are preferable to appendicular muscles due to the former relative independence of activity level.2 Moreover, dual-energy x-ray has been the standard quantitative measure for total appendicular muscle mass and appears in nearly 500 publications.3 The advantage of bedside appendicular muscle sonography should be further explored.Second, females comprised 71 and 61% of the “Not Frail” and “Frail” study groups. The authors adjusted psoas muscle area values according to body surface area and body mass index, but not by sex. The prevailing consensus, however, is to use sex-specific cutoffs for low values due to significant sex variation of human spinal and paraspinal muscles at all body habitus.4 This fact independently underscores the importance of equal representation of both sexes in the cohort.Third, for the sample power analysis, the authors provide the desired discriminative magnitude and CI; the expected frequency of frailty in the study population and expected sensitivity and specificity of sonography, however, are not given.5 Additionally, the areas under the receiver operating curves (AUCs) for each frailty surrogate and outcome measurement are presented (table 1 of Canales et al.1). The false discovery rate for these 40 statistical tests was not reported, and the authors did not analyze whether differences in AUC between the frailty surrogates were statistically significant.6Last, for comparison purposes, the study provided sonographic muscle mass measurements of healthy controls. However, the race/ethnicity makeup of the cohort and controls were very different (e.g., 79% Caucasians vs. 35%, respectively), even though skeletal muscle mass differs significantly between ethnicities at all adult ages.7 As expected, all controls were “Not Frail” per the Friend phenotype frailty assessment (table 2 of Canales et al.1), and therefore, the diagnostic quality of a frailty test cannot be assessed in this group. Simply put, a frailty test when performed on nonfrail controls would yield specificity and negative predictive values of 1, but the test’s sensitivity, positive predictive value, and AUC cannot be determined. However, sensitivity, specificity, positive and negative predictive values, AUC, and cutoff point are provided for each sonographic frailty surrogate in the controls (table 3 of Canales et al.1). Likewise, computed tomography was not performed in the controls (see, “Methods” and fig. 3A of Canales et al.1), yet the aforementioned statistical parameters are provided for psoas muscle area in controls. Perhaps table 3 should read “study cohort” instead of “heathy controls.”The authors did not comment on the surprisingly poor ability of the well established Fried phenotype frailty assessment to predict unplanned intensive care unit admission, prolonged intensive care unit and hospital length of stay, and rehospitalization (AUCs of 0.61, 0.54, 0.65, and 0.52, respectively). The sonographic and the tomographic sarcopenia assessments also seem to lack adequate discriminatory value for a prolonged hospital length of stay and rehospitalization. Unexpectedly prolonged postoperative hospitalization frequently precedes and portends unplanned nursing facility admission. The unexpected dichotomous discriminatory ability of these frailty assessments to predict the latter but not the former warrants further exploration. While we appreciate the authors’ objective to determine whether appendicular skeletal sonography could be used reliably in a preoperative setting to identify frailty, addressing these points would further strengthen their findings.The authors declare no competing interests.