Introduction: Computed tomography (CT) can accurately measure muscle mass, which is necessary for diagnosing sarcopenia, even in dialysis patients. However, CT-based screening for such patients is challenging, especially considering the availability of equipment within dialysis facilities. We therefore aimed to develop a bedside prediction model for low muscle mass, defined by the psoas muscle mass index (PMI) from CT measurement. Methods: Hemodialysis patients (n = 619) who had undergone abdominal CT screening were divided into the development (n = 441) and validation (n = 178) groups. PMI was manually measured using abdominal CT images to diagnose low muscle mass by two independent investigators. The development group’s data were used to create a logistic regression model using 42 items extracted from clinical information as predictive variables; variables were selected using the stepwise method. External validity was examined using the validation group’s data, and the area under the curve (AUC), sensitivity, and specificity were calculated. Results: Of all subjects, 226 (37%) were diagnosed with low muscle mass using PMI. A predictive model for low muscle mass was calculated using ten variables: each grip strength, sex, height, dry weight, primary cause of end-stage renal disease, diastolic blood pressure at start of session, pre-dialysis potassium and albumin level, and dialysis water removal in a session. The development group’s adjusted AUC, sensitivity, and specificity were 0.81, 60%, and 87%, respectively. The validation group’s adjusted AUC, sensitivity, and specificity were 0.73, 64%, and 82%, respectively. Discussion/Conclusion: Our results facilitate skeletal muscle screening in hemodialysis patients, assisting in sarcopenia prophylaxis and intervention decisions.

Sarcopenia, characterized by loss of skeletal muscle mass and function, is a serious problem, not only in the older population but also in younger patients with comorbidities, because sarcopenia is strictly correlated with physical disability, poor quality of life, and death [1]. In hemodialysis patients, who represent a weak population, those with sarcopenia have significantly higher mortality than those without (11%/year vs. 0.88%/year) [1‒3]. Sarcopenia is highly prevalent in patients with cardiovascular disease (CVD) and in those with chronic kidney disease (CKD), especially those undergoing dialysis [4, 5]. Sarcopenia prevalence is approximately 29% among patients with CKD. Because sarcopenia is more prevalent in old and weak patients, its prevalence in hemodialysis patients (40%) is higher than that in patients with CKD [3, 6‒8]. Furthermore, low skeletal muscle alone has a high impact on patient prognosis. Regarding the relationship between low muscle mass and CVD incidence, defined using the skeletal muscle index of CKD patients, the CVD incidence rate in the low muscle mass group was higher than that in the high muscle mass group [9‒11].

Computed tomography (CT) and magnetic resonance imaging (MRI) are the most accurate gold standard methods for measuring skeletal muscle mass, which is an indicator of sarcopenia [12‒15]. However, measuring skeletal muscle mass using CT or MRI in routine practice is not easy because both are costly and difficult to mobilize. Moreover, patient exposure to radiation must be considered in CT measurements, and long exposure of patients to loud noise must be considered in MRI measurements [16]. Therefore, to measure skeletal muscle mass for sarcopenia diagnosis in the usual practice, dual-energy X-ray absorptiometry (DEXA) and bioelectrical impedance analysis (BIA) are used as substitutes, which can be easily measured and are recommended by the European Working Group on Sarcopenia (EWGS) and Asian Working Group on Sarcopenia (AWGS) [17, 18]. However, hemodialysis patients have higher water content, rendering both DEXA and BIA measurements inaccurate [16]. In particular, the DEXA method may be overestimated compared to skeletal muscle mass measurement by CT or MRI in dialysis patients [17], and its use as the gold standard is inappropriate. Moreover, BIA only measures the electrical impedance depending on the water in separate parts of the body and does not directly measure skeletal muscle mass, whereas DEXA measures body water components, such as fat infiltrated into the lower limb tendons and skeletal muscles, as well as subdermal blood, lymph, and bone marrow fluid. Additionally, measured values may differ significantly depending on the estimation formula built into each device. Although BIA has high comparability within individuals, it has low comparability between individuals and minimal value in the prognosis-prediction model [18].

Muscle strength is an important component of sarcopenia, and grip strength, in particular, is used for sarcopenia screening in the general older population, according to the EWGS and AWGS [14, 15]. Grip strength measurement can be performed in a wide range of occupations and is well understood by patients; therefore, it can be easily used at the bedside. However, there are no reports on the accurate measurement of skeletal muscle mass using CT and MRI in hemodialysis patients, or on the inclusion of the older population. In this study, we created a prediction model for low muscle mass that can be applied at the bedside, using the psoas muscle mass index (PMI) measured on CT as the gold standard.

Research Design and Setting

We conducted a multicenter prospective cohort study in 18 Japanese dialysis centers. All centers had at least one CT system, 10 of the 18 centers had hospitalization facilities in addition to outpatient facilities, and 12 of the 18 centers were in the Kansai region. This study was performed in accordance with the tenets set forth in the Declaration of Helsinki. All patients provided written informed consent before enrollment in the study. Data collection was performed anonymously. This study was approved by the Ethics Committee of Kyoto University (approval No.: R2008-08).

Inclusion and Exclusion Criteria

We consecutively included outpatients who had been on dialysis for >6 months and were screened using noncontrast-enhanced abdominal CT in their respective facilities between June 29, 2019, and December 31, 2020. Of these, we excluded patients who had been hospitalized a month before or after CT screening, patients who did not have visible psoas muscles at the umbilical level, or patients who could not grip the dynamometer using each upper limb.

Main Outcome Measures

The primary outcome was low muscle mass defined by PMI and calculated using the patient’s height and area of psoas muscles from abdominal CT. The cross-sectional area of the bilateral psoas muscles, as an indicator of skeletal muscle mass, was measured using manual tracing at the umbilical level of the ilium using the RadiAnt DICOM Viewer (Medixant, Poznan, Poland) by two independent investigators (Fig. 1). If the umbilical level was not clear, the L4/5 level was used. The inter- or intra-rater reliability is shown in online supplementary material (for all online suppl. material, see www.karger.com/doi/10.1159/000526866). PMI was calculated as follows: PMI = cross-sectional area of the bilateral psoas muscle/height2 (cm2/m2). We set the PMI cut-off values as 7.17 and 5.13 cm2/m2 for men and women, respectively [10].

Fig. 1.

Cross-sectional CT image and ROI of the umbilical level. The white line shows the bilateral psoas muscle measured by manual tracing. PMI = psoas muscle area/height2 (cm2/m2). Low muscle mass was defined as PMI ≤7.17 cm2/m2 for men and ≤5.13 cm2/m2 for women. Representative CT images of low PMI (3.86 cm2/m2) (a) and high PMI (11.74 cm2/m2) (b). CT, computed tomography; ROI, region of interest; PMI, psoas muscle mass index.

Fig. 1.

Cross-sectional CT image and ROI of the umbilical level. The white line shows the bilateral psoas muscle measured by manual tracing. PMI = psoas muscle area/height2 (cm2/m2). Low muscle mass was defined as PMI ≤7.17 cm2/m2 for men and ≤5.13 cm2/m2 for women. Representative CT images of low PMI (3.86 cm2/m2) (a) and high PMI (11.74 cm2/m2) (b). CT, computed tomography; ROI, region of interest; PMI, psoas muscle mass index.

Close modal

Sample-Size Calculation

To develop the multivariable logistic regression model, at least ten events per variable were required [19]. Ten variables were to be included in the model. Therefore, at least 100 events were required for model validation [20]. Thus, a total of 200 events were required in this study. The proportion of low skeletal muscle mass cases was estimated to be 40%. Based on the above, we calculated that we required at least 500 patients (250 patients in the development cohort and 250 in the validation cohort) in this study.

Imputation

Multivariate imputation using chained equations was used to impute the missing variables. Twenty copies of data were used, each with suitably imputed missing values. Using the stepwise method for variable selection, we estimated mean odds ratios and adjusted 95% confidence intervals using Rubin’s rules [21].

Development of a Prediction Model

Based on geographical factors (shown in online suppl. material), we divided our study population into two cohorts: the development cohort (in the Kansai area) and the validation cohort (outside the Kansai area). The model was developed using multivariable logistic regression in the development cohort. The variables in online supplementary Table 1 were included in the stepwise logistic regression. Imputation data were dealt with as follows: (1) we selected variables using a multivariable logistic regression backward elimination technique for each stratum of the imputed cohort. (2) We voted for the selected variables in each stratum and summated the votes. (3) We selected the variables in order of the sum of their votes. The number of variables was limited to ≤10, according to the sample size. (4) We adapted the selected variables using multivariable logistic regression. We estimated the β coefficients for each stratum. (5) We estimated the mean β coefficients and used them as a developed model.

Internal and External Validation of the Model

We performed bootstrap validation repeated 100 times as an internal validation. We evaluated the performance of the internally validated model based on the calculated performance optimism. For external validation, we used a developed prediction model for our validation cohort (geographic validation) and evaluated both calibration and discrimination [22]. Calibration exhibited the accuracy of absolute risk estimates, whereas discrimination revealed the effectiveness of the developed model in differentiating those at higher risk from those at lower risk [23]. Model performance was evaluated by means of goodness-of-fit and calculated using the le Cessie-van Houwelingen normal test statistic for the unweighted sum of squared errors, the description of the calibration slope for evaluating the calibration ability, and the descriptions of the receiver operating characteristic (ROC) curve and area under the curve (AUC) for evaluating discriminative ability.

Statistical Analysis

Data were analyzed using R (version 4.11; R Foundation for Statistical Computing, Vienna, Austria). The statistical significance of the le Cessie-van Houwelingen normal test was set at p > 0.05. The data are displayed as number (%) for gender and low muscle mass, and as mean (standard deviation) for age and dialysis history.

Study Population

Figure 2 shows the flowchart of the research group. Between June 29, 2019, and December 31, 2020, 630 maintenance hemodialysis patients from 19 joint research facilities were selected as the study population. Of the 630 patients on maintenance hemodialysis, those whose grip strength could not be measured using a grip strength meter (n = 5) and those with abdominal CT data deficiency (n = 6) were excluded. Of the 619 included patients, 441 and 178 were placed in the development and validation groups, respectively, based on geographical location and use in analysis. Table 1 shows the baseline characteristics of the overall study subjects, and the characteristics of the patients in the development and validation groups. Among the included patients, 366 (63%) were male, the average age was 69 years (12), and the dialysis history was 13 years (11). Of the 619 patients, 226 (37%) were diagnosed with low muscle mass.

Table 1.

Patient characteristics

 Patient characteristics
 Patient characteristics
Fig. 2.

Flowchart illustrating the selection of study participants.

Fig. 2.

Flowchart illustrating the selection of study participants.

Close modal

Predictive Model for Low Muscle Mass

In the development group, 150 patients were diagnosed with low muscle mass. We created a logistic regression model using variables from clinical information as predictive variable candidates. Variable selection was performed using a stepwise approach. The final low muscle mass prediction model included ten variables from clinical information, including each grip strength, sex, height, dry weight, primary cause of end-stage renal disease, diastolic blood pressure at start of session, pre-dialysis potassium and albumin level, and dialysis water removal in a session.

Model Performance

The ROC curves for the development and validation groups are shown in Figure 3. The development group’s adjusted AUC, sensitivity, and specificity were 0.81, 60%, and 87%, respectively. The validation group’s adjusted AUC, sensitivity, and specificity were 0.73, 64%, and 82%, respectively. Model calibration was evaluated by comparing the observed and predicted risks in both the development and validation groups using calibration plots (shown in Fig. 3).

Fig. 3.

The ROC curves for the development and validation groups. The ROC curves for the development (n= 441) (a) and validation (n= 178) (b) groups are shown. Calibration plots comparing predicted and observed diagnostic rates for the development (c) and validation (d) group models. AUC, area under the curve; ROC, receiver operating characteristic.

Fig. 3.

The ROC curves for the development and validation groups. The ROC curves for the development (n= 441) (a) and validation (n= 178) (b) groups are shown. Calibration plots comparing predicted and observed diagnostic rates for the development (c) and validation (d) group models. AUC, area under the curve; ROC, receiver operating characteristic.

Close modal

We developed a predictive model for low muscle mass in patients undergoing hemodialysis. Our model exhibited reasonable performance and calibration in both the development and validation groups, with AUCs of 0.81 and 0.73, respectively. Ten variables were selected as predictors for the final model: each grip strength, sex, height, dry weight, primary cause of end-stage renal disease, diastolic blood pressure at start of session, pre-dialysis potassium and albumin level, and dialysis water removal in a session.

One of the strengths of this study is that the reported predictive model is easy to use in routine bedside work. The model predictor variables employ items that can be extracted from medical records. Moreover, grip strength measurement can be easily used in a wide range of occupations. It is also easy to use at the bedside for hemodialysis patients of all ages without the need for more expensive equipment and the consideration of any contraindications. Second, this model includes not only the concept of muscle mass but also the concept of muscle function. This prediction model is consistent with the concept of sarcopenia and can evaluate muscle strength or predict muscle mass, for which no easy measurement method existed before, thus leading to early sarcopenia diagnosis or prevention. Third, muscle mass was measured using PMI evaluated from CT images. CT imaging is the gold standard for assessing muscle mass, as they are not affected by body water content and can accurately evaluate muscle mass in hemodialysis patients.

There are several predictive models for low muscle mass that can be used at the bedside of hemodialysis patients, as the model reported in this study [24, 25]. For hemodialysis patients, a model using lower leg maximum circumference as a predictor and the DEXA method as the responder was associated with a good AUC for predicting appendicular fat-free mass. However, it has significant methodological errors. The difference between the normal and low muscle mass groups in the population was clearly dissociated, and the data continuum was considered not permissible. Therefore, their data led to the model’s overvaluation. Moreover, it is necessary to adapt their data to the real-world population, because the target population was younger than the real-world population. The population of hemodialysis patients is aging worldwide. According to the United States Renal Data System Report, the population on dialysis in both the USA and Europe is aging [26]. Moreover, compared to other continents (especially the USA and Europe), the number of hemodialysis patients in Asia is also increasing significantly.

In our study’s screening process, we collected information on the grip strength, which can be easily measured, and the remaining necessary model prediction variables were obtained from the patients’ medical records. Although BIA and DEXA are often used to measure muscle volumes, they are both associated with limitations. BIA, for instance, is contraindicated in patients with pacemakers and has measurement errors in patients with metal implants [27]. In addition, the BIA test uses the most reliable equation for predicting the body fat percentage; the effectiveness of retesting is ±3.5% points, and measurement errors have been reported [28]. Moreover, BIA is easily affected by overhydration and dehydration. Furthermore, DEXA is too expensive to be used regularly. Conversely, our model only requires a low additional cost and does not involve additional exposure to radiation; therefore, its regular use can be enabled.

Sarcopenia diagnosis requires not only low muscle mass, but also low physical function or low intensity. In addition to the loss of skeletal muscle mass, which is the main cause of sarcopenia, muscle and physical function weakness are intricately linked, thus rendering proper diagnosis difficult. However, despite weakness being one of the main aspects of sarcopenia, most studies reported so far have ignored it. To address this parameter, in our model we included each grip strength. Nonetheless, although in our study we predicted low muscle mass, we could not predict low intensity. Therefore, models included in future studies might be necessary to predict low intensity evaluated by muscle radiodensity loss [29].

However, this study has some limitations that should be acknowledged. First, it did not directly compare the proposed model with BIA with DEXA and with other methods used in traditional sarcopenia screening, such as lean tissue mass index and creatinine index. Moreover, this model was not proven to be superior to BIA or DEXA, and it is unclear whether it can replace these methods. However, previous studies have demonstrated that manually measured PMI in hemodialysis patients is highly correlated with skeletal muscle index measured using the BIA method, suggesting that the model presented here may be an alternative method [30, 31]. By comparing it with BIA and DEXA in the future, this predictive model may be an alternative to sarcopenia muscle mass screening. Second, we used the cut-off based on Harada’s criteria. We selected Harada’s criteria because they were related to cardiovascular events in CKD patients [10], and because the cut-off points for low muscle mass are not well defined for these measurements in the EWGSOP guidelines [32]. However, some investigators have also proposed other low muscle mass criteria for the Asian population [9, 33]. In this study, although these were not confirmed in relationship with hard endpoints, predicting them may have produced better results. Third, the model cannot be used for people whose grip strength is difficult to measure, such as those with hemiplegia and dementia. Fourth, because the timing of taking the abdominal CT imaging varied at each facility, PMI might be affected by the body water content. In addition, PMI measurements were manually measured without the use of automated software. As a result, it is possible that the reliability of the evaluators varied, which may have reduced the reproducibility of the analysis. Fifth, it is unclear whether this model is applicable to hemodialysis patients abroad, as it has not been evaluated in practice. In the future, the utility of this model in global hemodialysis patients will need to be evaluated.

Mizuho Ikenoue R.N. measured psoas area as an independent investigator. We appreciate her accurate work. Editage provided language editing, writing, technical editing, language proofreading, and grammar proofreading support. We would like to take this opportunity to thank you from the bottom of our hearts.

This study was performed in accordance with the tenets set forth in the Declaration of Helsinki. All patients provided written informed consent before enrollment in the study. Data collection was performed anonymously. This study was approved by the Ethics Committee of Kyoto University (approval number: R2008-08). A person in charge of each research facility was assigned, and informed consent was obtained from the participants in order to participate in this study.

The authors have no conflicts of interest to declare.

This study was supported by the 2019 Japanese Society for Dialysis Therapy and Allied Health Professional Research Grant. The funding body played no role in the preparation of data or the manuscript.

Daiki Senzaki involved in conceptualization, data curation, funding acquisition, investigation, project administration, visualization, writing – original draft, review, and editing, and is a lead author. Tatsuyoshi Ikenoue involved in conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, supervision, validation, visualization, writing – the original draft, review, and editing, and funding acquisition. Nobuo Yoshioka, Osamu Nagakawa, Emi Inayama, Takafumi Nakagawa, Hidehito Takayama, Toko Endo, Fumitaka Nakajima, Masayoshi Fukui, Yasuaki Kijima, Yasuo Oyama, Risshi Kudo, Tadashi Toyama, Yosuke Yamada, Kiyoshi Tsurusaki, Naoki Aoyama, Takayasu Matsumura, Hideki Yamahara, Kenro Miyasato, and Tetsuya Kitamura were involved in conceptualization, data collection and recruiting, and review and editing.

The data used in this study are available upon reasonable request. However, the abdominal CT cannot be shared in order to protect the patients’ privacy.

1.
Ren
H
,
Gong
D
,
Jia
F
,
Xu
B
,
Liu
Z
.
Sarcopenia in patients undergoing maintenance hemodialysis: incidence rate, risk factors and its effect on survival risk
.
Ren Fail
.
2016
;
38
(
3
):
364
71
.
2.
Isoyama
N
,
Qureshi
AR
,
Avesani
CM
,
Lindholm
B
,
Bàràny
P
,
Heimbürger
O
,
.
Comparative associations of muscle mass and muscle strength with mortality in dialysis patients
.
Clin J Am Soc Nephrol
.
2014 Oct 7
;
9
(
10
):
1720
8
.
3.
Mori
K
,
Nishide
K
,
Okuno
S
,
Shoji
T
,
Emoto
M
,
Tsuda
A
,
.
Impact of diabetes on sarcopenia and mortality in patients undergoing hemodialysis
.
BMC Nephrol
.
2019 Mar 28
;
20
(
1
):
105
.
4.
Foley
RN
,
Wang
C
,
Ishani
A
,
Collins
AJ
,
Murray
AM
.
Kidney function and sarcopenia in the United States general population: NHANES III
.
Am J Nephrol
.
2007
;
27
(
3
):
279
86
.
5.
Domanski
M
,
Ciechanowski
K
.
Sarcopenia: a major challenge in elderly patients with end-stage renal disease
.
J Aging Res
.
2012
:
754739
.
6.
Keith
DS
,
Nichols
GA
,
Gullion
CM
,
Brown
JB
,
Smith
DH
.
Longitudinal follow-up and outcomes among a population with chronic kidney disease in a large managed care organization
.
Arch Intern Med
.
2004 Mar 22
;
164
(
6
):
659
63
.
7.
Souza
VA
,
Oliveira
D
,
Barbosa
SR
,
Corrêa
JODA
,
Colugnati
FAB
,
Mansur
HN
,
.
Sarcopenia in patients with chronic kidney disease not yet on dialysis: analysis of the prevalence and associated factors
.
PLoS One
.
2017
;
12
(
4
):
e0176230
.
8.
Pacifico
J
,
Geerlings
MAJ
,
Reijnierse
EM
,
Phassouliotis
C
,
Lim
WK
,
Maier
AB
.
Prevalence of sarcopenia as a comorbid disease: a systematic review and meta-analysis
.
Exp Gerontol
.
2020 Mar
;
131
:
110801
.
9.
Hamaguchi
Y
,
Kaido
T
,
Okumura
S
,
Kobayashi
A
,
Hammad
A
,
Tamai
Y
,
.
Proposal for new diagnostic criteria for low skeletal muscle mass based on computed tomography imaging in Asian adults
.
Nutrition
.
2016 Nov–Dec
;
32
(
11–12
):
1200
5
.
10.
Harada
K
,
Suzuki
S
,
Ishii
H
,
Aoki
T
,
Hirayama
K
,
Shibata
Y
,
.
Impact of skeletal muscle mass on long-term adverse cardiovascular outcomes in patients with chronic kidney disease
.
Am J Cardiol
.
2017 Apr 15
;
119
(
8
):
1275
80
.
11.
Funamizu
T
,
Nagatomo
Y
,
Saji
M
,
Iguchi
N
,
Daida
H
,
Yoshikawa
T
.
Low muscle mass assessed by psoas muscle area is associated with clinical adverse events in elderly patients with heart failure
.
PLoS One
.
2021
;
16
(
2
):
e0247140
.
12.
Shen
W
,
Punyanitya
M
,
Wang
Z
,
Gallagher
D
,
St-Onge
MP
,
Albu
J
,
.
Total body skeletal muscle and adipose tissue volumes: estimation from a single abdominal cross-sectional image
.
J Appl Physiol
.
2004 Dec
;
97
(
6
):
2333
8
.
13.
Pahor
M
,
Manini
T
,
Cesari
M
.
Sarcopenia: clinical evaluation, biological markers and other evaluation tools
.
J Nutr Health Aging
.
2009 Oct
;
13
(
8
):
724
8
.
14.
Cruz-Jentoft
AJ
,
Baeyens
JP
,
Bauer
JM
,
Boirie
Y
,
Cederholm
T
,
Landi
F
,
.
Sarcopenia: European consensus on definition and diagnosis – report of the European Working Group on Sarcopenia in Older People
.
Age Ageing
.
2010 Jul
;
39
(
4
):
412
23
.
15.
Chen
LK
,
Liu
LK
,
Woo
J
,
Assantachai
P
,
Auyeung
TW
,
Bahyah
KS
,
.
Sarcopenia in Asia: consensus report of the Asian Working Group for Sarcopenia
.
J Am Med Dir Assoc
.
2014 Feb
;
15
(
2
):
95
101
.
16.
Chien
MY
,
Huang
TY
,
Wu
YT
.
Prevalence of sarcopenia estimated using a bioelectrical impedance analysis prediction equation in community-dwelling elderly people in Taiwan
.
J Am Geriatr Soc
.
2008 Sep
;
56
(
9
):
1710
5
.
17.
Shih
R
,
Wang
Z
,
Heo
M
,
Wang
W
,
Heymsfield
SB
.
Lower limb skeletal muscle mass: development of dual-energy X-ray absorptiometry prediction model
.
J Appl Physiol
.
2000 Oct
;
89
(
4
):
1380
6
.
18.
Ishiguro
N
,
Kanehisa
H
,
Miyatani
M
,
Masuo
Y
,
Fukunaga
T
.
A comparison of three bioelectrical impedance analyses for predicting lean body mass in a population with a large difference in muscularity
.
Eur J Appl Physiol
.
2005 May
;
94
(
1–2
):
25
35
.
19.
Wang
QQ
,
Yu
SC
,
Qi
X
,
Hu
YH
,
Zheng
WJ
,
Shi
JX
,
.
Overview of logistic regression model analysis and application
.
Zhonghua Yu Fang Yi Xue Za Zhi
.
2019 Sep 6
;
53
(
9
):
955
60
.
20.
Vergouwe
Y
,
Steyerberg
EW
,
Eijkemans
MJC
,
Habbema
JDF
.
Substantial effective sample sizes were required for external validation studies of predictive logistic regression models
.
J Clin Epidemiol
.
2005 May
;
58
(
5
):
475
83
.
21.
Azur
MJ
,
Stuart
EA
,
Frangakis
C
,
Leaf
PJ
.
Multiple imputation by chained equations: what is it and how does it work
.
Int J Methods Psychiatr Res
.
2011
;
20
(
1
):
40
9
.
22.
Collins
GS
,
Reitsma
JB
,
Altman
DG
,
Moons
KGM
.
Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement
.
Eur Urol
.
2015 Jun
;
67
(
6
):
1142
51
.
23.
Alba
AC
,
Agoritsas
T
,
Walsh
M
,
Hanna
S
,
Iorio
A
,
Devereaux
PJ
,
.
Discrimination and calibration of clinical prediction models: users’ guides to the medical literature
.
JAMA
.
2017 Oct 10
;
318
(
14
):
1377
84
.
24.
Kaysen
GA
,
Zhu
F
,
Sarkar
S
,
Heymsfield
SB
,
Wong
J
,
Kaitwatcharachai
C
,
.
Estimation of total-body and limb muscle mass in hemodialysis patients by using multifrequency bioimpedance spectroscopy
.
Am J Clin Nutr
.
2005 Nov
;
82
(
5
):
988
95
.
25.
Bellafronte
NT
,
Sizoto
GR
,
Vega-Piris
L
,
Chiarello
PG
,
Cuadrado
GB
.
Bed-side measures for diagnosis of low muscle mass, sarcopenia, obesity, and sarcopenic obesity in patients with chronic kidney disease under non-dialysis-dependent, dialysis dependent and kidney transplant therapy
.
PLoS One
.
2020
;
15
(
11
):
e0242671
.
26.
Shoaf
C
,
Genaidy
A
,
Karwowski
W
,
Huang
SH
.
Improving performance and quality of working life: a model for organizational health assessment in emerging enterprises
.
Hum Factors Ergon Manufacturing
.
2004
;
14
(
1
):
81
95
.
27.
Alexandrou
ME
,
Balafa
O
,
Sarafidis
P
.
Assessment of hydration status in peritoneal dialysis patients: validity, prognostic value, strengths, and limitations of available techniques
.
Am J Nephrol
.
2020
;
51
(
8
):
589
612
.
28.
Aandstad
A
,
Holtberget
K
,
Hageberg
R
,
Holme
I
,
Anderssen
SA
.
Validity and reliability of bioelectrical impedance analysis and skinfold thickness in predicting body fat in military personnel
.
Mil Med
.
2014 Feb
;
179
(
2
):
208
17
.
29.
Lee
J
,
Lin
JB
,
Wu
MH
,
Jan
YT
,
Chang
CL
,
Huang
CY
,
.
Muscle radiodensity loss during cancer therapy is predictive for poor survival in advanced endometrial cancer
.
J Cachexia Sarcopenia Muscle
.
2019 Aug
;
10
(
4
):
814
26
.
30.
Ito
K
,
Ookawara
S
,
Imai
S
,
Kakuda
H
,
Bandai
Y
,
Fueki
M
,
.
Muscle mass evaluation using psoas muscle mass index by computed tomography imaging in hemodialysis patients
.
Clin Nutr ESPEN
.
2021 Aug
;
44
:
410
4
.
31.
Takata
T
,
Motoe
A
,
Tanida
K
,
Taniguchi
S
,
Ida
A
,
Yamada
K
,
.
Feasibility of computed tomography-based assessment of skeletal muscle mass in hemodialysis patients
.
J Nephrol
.
2021 Apr
;
34
(
2
):
465
71
.
32.
Cruz-Jentoft
AJ
,
Bahat
G
,
Bauer
J
,
Boirie
Y
,
Bruyere
O
,
Cederholm
T
,
.
Sarcopenia: revised European consensus on definition and diagnosis
.
Age Ageing
.
2019 Jan 1
;
48
(
4
):
601
31
.
33.
Wu
CH
,
Chao
CT
,
Liang
PC
,
Shih
TTF
,
Huang
JW
.
Computed tomography-based sarcopenia in patients receiving peritoneal dialysis: correlation with lean soft tissue and survival
.
J Formos Med Assoc
.
2022 Feb
;
121
(
2
):
500
9
.