jfa journal

AND option

OR option



A.J. Mayhew1,2,3, S.M. Phillips4, N. Sohel1,2,3, L. Thabane1,5, P.D. McNicholas6, R.J. de Souza1,7, G. Parise4, P. Raina1,2,3


1. Department of Health Research Methods, Evidence, and Impact, McMaster University, Hamilton, Ontario, Canada; 2. Labarge Centre for Mobility in Aging, Hamilton, Ontario, Canada; 3. McMaster Institute for Research on Aging, Hamilton, Ontario, Canada; 4. Department of Kinesiology, McMaster University, Hamilton, Ontario, Canada; 5. Biostatistics Unit, Research Institute at St Joes, St. Joseph’s Healthcare Hamilton, Hamilton, Ontario, Canada; 6. Department of Mathematics & Statistics, McMaster University, Hamilton, Ontario, Canada; 7. Population Genomics Program, Chanchlani Research Centre, McMaster University, Hamilton, Ontario, Canada.
Corresponding author: Parminder Raina, PhD, Department of Health Research Methods, Evidence, and Impact, McMaster University, MIP 309A, 175 Longwood Road South, Hamilton, Ontario, L8P 0A1, Canada, Tel: 905 525 9140 x 22197, e-mail: praina@mcmaster.ca

J Frailty Aging 2020;in press
Published online September 29, 2020, http://dx.doi.org/10.14283/jfa.2020.48



Background: Using residual values calculated from models regressing appendicular lean mass on fat mass and height is one of several suggested strategies for adjusting appendicular lean mass for body size when measuring sarcopenia. However, special consideration is required when using this technique in different subgroups in order to capture the correct individuals as sarcopenic. Objectives: To provide guidance about how to conduct stratified analyses for the regression adjustment technique using age groups as an example. Design: Cross-sectional study. Setting: Data collected at baseline (2012-2015) for the Canadian Longitudinal Study on Aging. Participants: Community dwelling participants of European descent aged 45 to 85 years (n=25,399). Measurements: Appendicular lean mass, height, and weight were measured. Sex-specific residuals were calculated in participants before and after stratifying participants by age group (45-54, 55-64, 65-74, 75-85 years). Cut offs corresponding to the sex-specific 20th percentile residual values in participants ≥65 years were determined first in the residuals calculated in all participants and residuals calculated in only those aged ≥65 years. For each set of cut offs, the percentage of age and sex-stratified participants with low appendicular lean mass were compared for the residuals calculated in all participants and the residuals calculated after stratifying by age. Results: In 12,622 males and 12,737 females, regardless of the cut off used, the percentage of participants with low appendicular lean mass decreased with age when residuals were calculated after age stratification. When the residuals were calculated in all participants, the percentage of participants with sarcopenia increased from the youngest to the oldest age groups. Conclusions: Sex-specific residuals in all participants should be calculated prior to stratifying the sample by age group, or other stratification variables, for the purposes of developing appendicular lean mass cut offs or subgroup analyses.

Key words: Appendicular lean mass, CLSA, muscle, residuals, sarcopenia, skeletal muscles.



Sarcopenia refers to the decline in muscle mass, muscle strength, and muscle function that occurs with age (1). It is associated with an increased risk of falls and fractures, activities of daily living limitations, and mortality (2–5). Given the profound individual and societal costs of sarcopenia, there has been substantial interest in finding ways to prevent and treat sarcopenia. However, the field of sarcopenia research has been hindered by the lack of a clear definition and standardized diagnostic criteria (6).
Four expert-group definitions for sarcopenia define sarcopenia as the combination of low muscle mass, typically measured as appendicular lean mass (ALM), with either low muscle strength or impaired physical performance (6–10). There is a consensus among the definitions that ALM should be adjusted for body size due to the strong correlation between ALM with height and weight, however there is little agreement about which measure of body size should be utilized (6, 11). Four techniques are recommended; dividing by height squared, body mass, body mass index (BMI), and regressing ALM on height and fat mass (6–10). Of these methods, regressing ALM on height and fat mass may most accurately identify individuals with low ALM as it adjusts for two measures of body size whereas the other techniques only adjust for one measure of body size (12). This technique involves creating a regression model (ALM = intercept + height (m2) + fat mass (kg)) in a sample of individuals. For each individual, a predicted value of ALM is calculated based on the regression equation. Subtracting the estimated value of ALM from the actual value of ALM for each person provides a residual value. Positive residual values indicate that the individual has more ALM than would be expected given their height and weight and negative residual values indicate the individual has less ALM than would be expected given their height and weight.
Unlike adjusting ALM by height, weight, or BMI which are done at the individual level and are not influenced by other participants, calculating residuals is dependent on the sample. For height, weight, and BMI adjustment, the adjusted values refer to the same amount of ALM relative to the anthropometric measure adjusted for regardless of the person or sample. In contrast, the residual value for each person is dependent on the regression equation which in turn is dependent on the distribution of the variables in the sample. Consequently, even if low ALM offs are developed in a random, population-based sample, they cannot be appropriately applied to another population unless the two samples have identical joint distributions of ALM, fat mass, and height. Due to the unavailability of cut offs, studies that have investigated sarcopenia using the residual adjustment technique have considered the lowest quintile of sex-specific residual values as sarcopenic (13–19). However, a consequence of using the lowest quintile is that sarcopenia prevalence is the same for all studies, regardless of age, which is problematic for a condition for which the prevalence increases with age. This poses additional challenges for studies with a wide range of ages which want to conduct age stratified analyses.
To our knowledge, there has not been any discussion in the literature about the implications of stratifying a sample by age when applying the residual technique. We aimed to provide the necessary guidance for how to handle age stratification when calculating residual values for ALM adjusted for height and fat mass.



Setting and study population

We used data from the Canadian Longitudinal Study on Aging (CLSA), a national longitudinal research platform. There were 51,338 participants aged 45 to 85 years recruited from the ten Canadian provinces at baseline. Participants had to be physically and cognitively able to participate on their own as well as not living in institutions such as long term care to be eligible for the study. The participants were recruited in to one of two cohorts, the Tracking cohort and the Comprehensive cohort. Participants from all ten provinces were randomly selected for the Tracking cohort (n=21,241) and were interviewed by telephone. The Comprehensive cohort participants (n=30,097) lived within 25-50kg of one of 11 Data Collection Sites located in seven provinces. The Comprehensive cohort participants were interviewed in-person and also completed in-depth physical assessments and provided blood and urine samples. Details on the study design have been described elsewhere (20). Only participants from the Comprehensive cohort (n=30,097) were included in these analyses as the physical assessment data was required. The sample was further limited to those identifying as European as ALM, muscle strength, and physical function have shown to vary by ethnicity (21–23). This project uses data collected at baseline (September 2011 to May 2015). Ethics approval was received by the Hamilton Research Ethics Board (#2686).

Clinical measurements

Trained research assistants collected data on height, weight, and muscle mass. Height was measured twice using a stadiometer and the mean value of the two measurements was used in the analyses. The Hologic Discovery ATM DXA machine was calibrated daily using a spine phantom, weekly using a whole body step phantom, and yearly using a gold standard phantom. DXA provides a valid measures of ALM and fat mass when compared to the gold standards of computerized tomography (CT) and magnetic resonance imaging (MRI) scans (24, 25).
All analyses were stratified by sex. We used multiple linear regression models with ALM as the dependent variable and height (m2) and fat mass (kg) as the independent variables to estimate the predicted value of ALM for each participant. The residual values were calculated as the predicted value of ALM subtracted from the actual value of ALM. To test the impact of age stratification on the residual values, we first calculated residuals based on the regression model including participants aged 45 to 85 years. We then calculated residuals based on regression models run separately for each age strata (45 to 54, 55 to 64, 65 to 74, and 75 to 85 years). We followed the EWGSOP recommendation of using the lowest sex-specific 20th percentile of residual values as the cut off for low ALM (7). We chose to limit the sample for calculating cut offs to participants ≥65 years based on guidance from the literature (7). To explore the impact of age stratification on the values of the residual cut offs, we determined the cut offs for the residuals in the model that included all participants aged 45 to 85 years, as well as for residual values based on a model that only included participants 65 years and older.
The cut-offs detertmined using the non-age stratified residuals and the residuals calculated in just participants aged ≥65 years were applied to the residuals calculated in the whole sample and the age-stratified residuals. Therefore, there were four different strategies used to identify participants: Strategy 1: all residuals calculated in all participant; Strategy 2: individual residuals calculated in all participants, cut offs developed in participants ≥65 years; Strategy 3: individual residuals calculated in specific age groups, cut offs developed in all participants; Strategy 4: individual residuals calculated in specific age groups, cut offs developed in participants ≥65 years.

Statistical anaylses

Of the 30,097 participants at baseline, 1324 were excluded as they were non-European, 3356 were excluded for missing ALM, grip strength, gait speed, or BMI data resulting in a final sample size of 25,399 participants. All statistical analyses were completed using SAS (version 12.3).
The percentage of age and sex-stratified participants categorized as having low ALM by each of the four strategies for handling age-stratification for the development of cut offs and individual residual values were determined. Bootstrap percentile confidence intervals were calculated for each estimate. This technique involves resampling with replacement and calculating the proportion of participants with sarcopenia for each resample (26). We resampled 10,000 times and identified the values corresponding to the 2.5th and 97.5th percentiles of the 10,000 resamples in order to estimate the 95% confidence interval. This technique has the advantage of only including valid values of parameter estimates in the confidence interval (26).



Participant characteristics

The mean (SD) age of the participants was 62.8 (10.2) years and 49.9% of the sample were males (Table 1). Younger males and females had greater mean (SD) ALM: 27.2kg (4.2) and 17.9kg (3.4), grip strength: 47.3kg (9.1) and 28.6kg (5.6), and gait speed: 1.03m/s (0.18) and 1.02m/s (0.19) compared to older males and females (ALM: 24.4kg (3.7) and 16.3kg (2.9), grip strength: 39.4kg (8.5) and 23.6kg (5.2), and gait speed: 0.94m/s (0.19) and 0.90m/s (0.19).

Table 1
Participant characteristics

1. Heart disease includes angina, myocardial infarction, and heart disease; 2. Cardiovascular disease includes stroke and transient ischemic attack; 3. Neurological conditions include multiple sclerosis, epilepsy, migraine headaches, and Parkinson’s Disease

Distribution of residuals

The overall distribution of the residual values was calculated in all participants versus calculating the residuals in age-stratified groups. In males, the mean (SD) for all participants was 0 (2.90), while the mean of the residuals for all age-stratified residuals pooled together was 0 (3.13). The corresponding values were 0 (2.08) and 0 (2.16) in females. However, the distribution of the data within each age group was markedly different. In both males and females, when the residuals were calculated after stratifying the sample by age, the residuals of each age group had a mean of 0. In contrast, when the residuals were calculated in the whole sample, there was a gradient of mean values when stratified by age group. The mean residual value for males 45 to 54 years was 1.36 and for females was 0.84 which decreased to -1.95 in males and -0.67 in females aged 75 to 85 years (Supplementary Appendix 1).

Muscle mass cut off estimates

The lowest 20th percentile cut offs corresponded to -3.51 for males and -2.15 for females when the residual values were calculated all participants, then restricted to participants aged ≥65 years. When the residuals were calculated in only participants ≥65 years, the 20th percentile cut offs were -2.23 for males and -1.58 for females.

Low muscle mass prevalence

The lower cut offs determined using the non-age stratified residual values of -3.51 for males and -2.23 for females identified fewer participants as having low muscle mass compared to the age-stratified residual values of -2.15 for males and -1.58 for females (Figure 1). For these cut offs, the prevalence of low muscle mass was 12.3% for males and 14.6% for females when the individual residuals were not age stratified (Strategy 1) and 10.3% for males and 13.8% for females when the individual residuals were age stratified (Strategy 3). The cut offs developed using residual values calculated in only participants ≥65 years, identified 23.8% of males and 22.8% of females as having low ALM when the non-age stratified residual values (Strategy 2) and 21.7% of males and 21.9% of females as having low ALM when the age-stratified values were used (Strategy 4).
When looking at the percentage of people with low muscle mass within each age group, the percentage of males and females with low muscle mass increased with age when the individual residuals were not age-stratified, regardless of the cut offs used (Strategy 1 and Strategy 2). In contrast, the percentage of males and females with low muscle mass decreased with age when the age-stratified residuals were used (Strategy 3 and Strategy 4).

Figure 1
Percentage of participants with low ALM adjusted for height and fat mass stratified by age group and sex



To our knowledge, this is the first study to investigate the implications of age stratification when using the residual values for ALM after regressing on height and fat mass. We determined that residual values should be calculated in all participants before stratifying by age for the purposes of subgroup analyses or developing muscle mass cut offs (Strategy 1).
Stratifying the sample by age prior to calculating residuals for the purpose of subgroup analyses based on age or for developing cut offs proved problematic. When the sample was stratified by age before calculating the residuals (Strategy 3 and Strategy 4), the percentage of participants with low ALM decreased from the youngest to the oldest age groups (Figure 1) because of how the residuals are calculated. The maximum likelihood estimation technique used in linear regression to calculate the residuals requires that the sum of the residuals for the sample to equal zero. When the sample was stratified by age before calculating the residuals, the mean value of the residuals for each age group was zero. However, the standard deviation decreased with age (Supplementary Appendix 1). The greater the standard deviation for the age group, the more participants were below the low ALM cut off and therefore the higher the percentage of people with low ALM.
The problems we encountered stratifying our sample by age before calculating the residuals extend to any situation in which residuals calculated in one sample are combined or applied to another sample. Residual values are sample dependent and therefore unless two groups of participants have identical joint distributions of ALM, height, and fat mass, the residuals from one study will not identify people with the same amount of ALM relative to height and fat mass. This means that cut offs for the residual technique, even if developed in a population-based random sample with cut offs validated against relevant health outcomes, cannot be meaningfully applied to another sample. For this reason, in our analyses Strategy 1 which calculates the residuals in all participants before limiting to those ≥65 years to determine the lowest quintile is the appropriate strategy.
To resolve the issue of residual values and corresponding cut offs not being comparable between studies, prediction equations, similar to those that have been used for lung function can be developed (27). A sample of representative older adults could be used to create sex-specific prediction equations for ALM based on height and fat mass. Variables such as age, ethnicity, and other body composition variables could be explored for inclusion in the equation, as well as possible interactions between variables. These equations would allow for results to be meaningfully compared between studies and would also allow clinicians to use this technique to diagnose low ALM in individuals. Low ALM cut offs, ideally determined by assessing which cut offs best predict health outcomes relevant to sarcopenia, could be established and used differents studies.
To our knowledge, only one study has assessed the relationship between low ALM operationalized using the residual adjustment technique with health (12, 28). Cawthon et al. observed that low ALM adjusted for height and fat mass was significantly associated with risk of functional limitations and mortality, but not recurrent falls or hip fractures (12). Studies operationalizing sarcopenia as low ALM only often do not find significant associations with health, therefore the associations found with functional limitations and mortality are particularly notable (12, 29, 30). Given this evidence as well as the strong face validity for adjusting ALM simultaneously for height and fat mass, future studies are required to determine if adjusting ALM for height and fat mass, alone and in combination with muscle strength or function, better identifies people at poor risk for health compared to the other adjustment techniques.
In conclusion, adjusting ALM for height and fat mass using the regression technique is a promising method of operationalizing low ALM that warrants greater inclusion in future sarcopenia studies. In this study, we show that to appropriately apply the residual technique to a stratified sample, the regression equation must be calculated in all participants before stratifying the sample in order to identify the correct individuals as sarcopenic.

Acknowledgements: This research was made possible using the data/biospecimens collected by the Canadian Longitudinal Study on Aging (CLSA). Funding for the Canadian Longitudinal Study on Aging (CLSA) is provided by the Government of Canada through the Canadian Institutes of Health Research (CIHR) under grant reference: LSA 94473 and the Canada Foundation for Innovation. This research has been conducted using the CLSA dataset, Baseline Comprehensive Dataset version 4.0, under Application Number 160608. The CLSA is led by Drs. Parminder Raina, Christina Wolfson and Susan Kirkland. The opinions expressed in this manuscript are the author’s own and do not reflect the views of the Canadian Longitudinal Study on Aging.
Funding: No funding to report.
Conflict of interest: None declared
Author contributions: AJM, SMP, and PR conceptualized this project with feedback from NS, LT, PDM, RJd and GP. AJM and NS completed the analysis of the data. AJM, NS, and PR interpreted the results. AJM completed the draft of the manuscript with revisions from the remaining authors. All authors provided approval for the final version to be published and agree to be accountable for all aspects of the work.
Ethical standards: Ethics approval for this project was received by the Hamilton Research Ethics Board (#2686).


1. Rosenberg IH. Sarcopenia: origins and clinical relevance. J Nutr 1997;127:990S–911S
2. Kim JH, Lim S, Choi SH, et al. Sarcopenia: an independent predictor of mortality in community-dwelling older Korean men. J Gerontol A Biol Sci Med Sci 2014;69:1244–1252
3. Landi F, Cruz-Jentoft AJ, Liperoti R, et al. Sarcopenia and mortality risk in frail older persons aged 80 years and older: results from ilSIRENTE study. Age Ageing 2013;42:203–209
4. Janssen I, Heymsfield SB, Ross R. Low relative skeletal muscle mass (sarcopenia) in older persons is associated with functional impairement and physical disability. J Am Geriatr Soc 2002;50:889–896
5. Yeung SSY, Reijnierse EM, Pham VK, et al. Sarcopenia and its association with falls and fractures in older adults: A systematic review and meta-analysis. J Cachexia Sarcopenia Muscle 2019;10:485–500
6. Cruz-Jentoft AJ, Bahat G, Bauer J, et al. Sarcopenia: Revised European consensus on definition and diagnosis. Age Ageing 2019;48:16–31
7. Cruz-Jentoft A, Baeyens J, Bauer J, et al. Sarcopenia: European consensus on definition and diagnosis: Report of the European Working Group on Sarcopenia in Older People. Age Ageing 2010;39:412–423
8. Studenski SA, Peters KW, Alley DE, et al. The FNIH sarcopenia project: Rationale, study description, conference recommendations, and final estimates. Journals Gerontol Med Sci 2014;69:547–558
9. Fielding RA, Vellas B, Evans WJ, et al. Sarcopenia: An undiagnosed condition in older adults. Current consensus definition: Prevalence, etiology, and conseuqences. International Work Group on Sarcopenia. Am Med Dir Assoc 2011;12:249–256
10. Chen L-K, Liu L-K, Woo J, et al. Sarcopenia in Asia: Consensus report of the Asian Working Group for Sarcopenia. J Am Med Dir Assoc 2014;15:95–101
11. Gallagher D, Visser M, De Meersman RE, et al. Appendicular skeletal muscle mass: effects of age, gender, and ethnicity. J Appl Physiol 2017;83:229–239
12. Cawthon PM, Blackwell TL, Francisco S, et al. An evaluation of the usefulness of consensus definitions of sarcopenia in older men: results from the observational Osteoporotic Fractures in Men (MrOS) cohort study. J Am Diet Assoc 2016;63:2247–2259
13. Menant JC, Weber F, Lo J, et al. Strength measures are better than muscle mass measures in predicting health-related outcomes in older people: time to abandon the term sarcopenia? Osteoporos Int 2016;1–12
14. Delmonico MJ, Harris TB, Lee J-S, et al. Alternative definitions of sarcopenia, lower extremity performance, and functional impairment with aging in older men and women. J Am Geriatr Soc 2007;55:769–774
15. Newman AB, Kupelian V, Visser M, et al. Sarcopenia: Alternative Definitions and Associations with Lower Extremity Function. J Am Geriatr Soc 2003;51:1602–1609
16. Figueiredo CP, Domiciano DS, Lopes JB, et al. Prevalence of sarcopenia and associated risk factors by two diagnostic criteria in community-dwelling older men: The São Paulo Ageing & Health Study (SPAH). Osteoporos Int 2014;25:589–596
17. Domiciano DS, Figueiredo CP, Lopes JB, et al. Discriminating sarcopenia in community-dwelling older women with high frequency of overweight/obesity: The São Paulo Ageing & Health Study (SPAH). Osteoporos Int 2013;24:595–603
18. Chalhoub D, Cawthon PM, Ensrud KE, et al. Risk of nonspine fractures in older adults with sarcopenia, low bone mass, or both. J Am Geriatr Soc 2015;63:1733–1740
19. Scott D, Chandrasekara SD, Laslett LL, Cicuttini F, Ebeling PR, Jones G. Associations of sarcopenic obesity and dynapenic obesity with bone mineral density and incident fractures over 5 – 10 years in community-dwelling older adults. Calcif Tissue Int 2016;99:30–42
20. Raina PS, Wolfson C, Kirkland SA, et al. Cohort profile: The Canadian Longitudinal Study on Aging (CLSA). Can J Aging 2009;28:221–229
21. Capistrant BD, Glymour MM, Berkman LF. Assessing mobility difficulties for cross-national comparisons: Results from the World Health Organiation Study on Global AGEing and Adult Health. J Am Geriatr Soc 2015;62:329–335
22. Leong DP, Teo KK, Rangarajan S, et al. Prognostic value of grip strength: findings from the Prospective Urban Rural Epidemiology (PURE) study. Lancet 2015;386:266–273
23. Silva AM, Shen W, Heo M, et al. Ethnicity-related skeletal muscle differences across the lifespan. Am J Hum Biol 2010;22:76–82
24. Buckinx F, Landi F, Cesari M, et al. Pitfalls in the measurement of muscle mass: a need for a reference standard. J Cachexia Sarcopenia Muscle 2018;9:269–278
25. Kullberg J, Brandberg J, Angelhed JE, et al. Whole-body adipose tissue analysis: Comparison of MRI, CT and dual energy X-ray absorptiometry. Br J Radiol 2009;82:123–130
26. Carpenter J, Bithell J. Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians. Stat Med 2000;19:1141–1164
27. Falaschetti E, Laiho J, Primatesta P, Purdon S. Prediction equations for normal and low lung function from the health survey for England. Eur Respir J 2004;23:456–463
28. Beaudart C, Zaaria M, Reginster J. Health outcomes of sarcopenia: A systematic review and meta-analysis. PLoS One 2017;12:e0169548
29. Schaap LA, Schoor NM Van, Lips P, Visser M. Associations of sarcopenia definitions, and their components, with the incidence of recurrent falling and fractures: The Longitudinal Aging Study Amsterdam. Journals Gerontol Med Sci 2018;73:1199–1204
30. Bischoff-Ferrari HA, Orav JE, Kanis JA, et al. Comparative performance of current definitions of sarcopenia against the prospective incidence of falls among community-dwelling seniors age 65 and older. Osteoporos Sarcopenia 2015;26:2793–2802



J. Muscedere1, P.M. Kim2, J. Afilalo3, C. Balion4, V.E. Baracos5, D. Bowdish6, M. Cesari7, J. D. Erusalimsky8, T. Fülöp9, G. Heckman10, S.e. Howlett11, R.G. Khadaroo12, J.L. Kirkland13, L. Rodriguez Mañas14, E. Marzetti15, G. Paré4, P. Raina16, K. Rockwood17, A. Sinclair18, C. Skappak19, C. Verschoor16, S. Walter20 for the Canadian Frailty Network


1. Department of Critical Care Medicine, Queen’s University; 2. Canadian Frailty Network, Kingston, ON, Canada; 3. Division of Cardiology and Centre for Clinical Epidemiology, Jewish General Hospital, McGill University; 4. Department of Pathology and Molecular Medicine, McMaster University; 5. Department of Oncology, University of Alberta; 6. Department of Pathology and Molecular Medicine, Master University; 7. Department of Clinical Sciences and Community Health, Università di Milano; and Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan, Italy; 8. Department of Biomedical Sciences, Cardiff Metropolitan University, Cardiff, UK; 9 Research Center on Aging, University of Sherbrooke, Québec, Canada; 10. School of Public Health and Health Systems, Schlegel University of Waterloo Research Institute for Aging, University of Waterloo; 11. Departments of Pharmacology and Medicine (Geriatric Medicine), Faculty of Medicine, Dalhousie University; 12. Department of Surgery and Critical Care Medicine, University of Alberta; 13. Robert and Arlene Kogod Center on Aging, Mayo Clinic; 14. Department of Geriatrics, Hospital Universitario de Getafe, Madrid; 15. Fondazione Policlinico Universitario «Agostino Gemelli» IRCCS, Università Cattolica del Sacro Cuore, Rome, Italy; 16. Department of Health Research Methods, Evaluation and Impact, McMaster University; 17. Division of Geriatric Medicine, Dalhousie University; 18. Foundation for Diabetes Research in Older People at Diabetes Frail Ltd; 19. Schwartz/Reisman Emergency Medicine Institute, Mount Sinai Hospital, University of Toronto, and the Division of Emergency Medicine, McMaster University; 20. Fundación de Investigación Biomédica Hospital Universitario de Getafe, Getafe, Spain. Dept. of Epidemiology and Biostatistics, University of California San Francisco, California, United States of America.
Corresponding author: Dr. John Muscedere, Kingston Health Sciences Centre, 76 Stuart Street, Kingston, Ontario, Canada, Email: john.muscedere@kingstonhsc.ca
J Frailty Aging 2019;in press
Published online April 30, 2019, http://dx.doi.org/10.14283/jfa.2019.12



The Canadian Frailty Network (CFN), a pan-Canadian not-for-profit organization funded by the Government of Canada through the Networks of Centres of Excellence Program, is dedicated to improving the care of older Canadians living with frailty. The CFN has partnered with the Canadian Longitudinal Study on Aging (CLSA) to measure potential frailty biomarkers in biological samples (whole blood, plasma, urine) collected in over 30,000 CLSA participants.  CFN hosted a workshop in Toronto on January 15 2018, bringing together experts in the field of biomarkers, aging and frailty. The overall objectives of the workshop were to start building a consensus on potential frailty biomarker domains and identify specific frailty biomarkers to be measured in the CLSA biological samples. The workshop was structured with presentations in the morning to frame the discussions for the afternoon session, which was organized as a free-flowing discussion to benefit from the expertise of the participants. Participants and speakers were from Canada, Italy, Spain, United Kingdom and the United States.   Herein we provide pertinent background information, a summary of all the presentations with key figures and tables, and the distillation of the discussions.  In addition, moving forward, the principles CFN will use to approach frailty biomarker research and development are outlined.  Findings from the workshop are helping CFN and CLSA plan and conduct the analysis of biomarkers in the CLSA samples and which will inform a follow-up data access competition.

Key words: CFN, CLSA, frailty, frailty index, seniors, older adults, biomarkers, aging.



Older adults living with frailty represent an increasing concern for health systems due to increased vulnerability to acute stressors, increased risk of functional impairment, increased healthcare utilization including emergency department visits, hospitalizations, and increased mortality (1-4). Therefore, it is critically important to identify, as early as possible, those at imminent risk of frailty and those living with frailty. In addition, there is a need to be able to assess the severity of frailty as objectively as possible for both prognostic purposes and monitoring of response to therapeutic interventions.
There are two main views of frailty; as a syndrome or as a state arising from an accumulation of deficits. Considerable progress and increased understanding of both approaches has occurred since their introduction (5). Frailty as a phenotype can be measured by Fried or Bergman phenotype criteria (1, 6). This model derives from an attempt to characterize the clinical manifestations of vulnerability outside of multimorbidity and disability. Frailty as an accumulation of deficits can be measured by the Frailty Index (FI), which is the number of physiological deficits affecting an individual divided by the total number of deficits measured (7, 8). The theoretical underpinning for the deficit model arises from the observation that deficits increase variably with age in humans and in animals and that the risk of morbidity, functional decline and death increases with an increasing number of deficits both individually and on a population basis (9, 10).
Although the concept of frailty is widely accepted, how to best detect and measure the severity of frailty remains controversial and as a consequence, there are many clinical frailty assessment instruments utilized in practice and research (11-13). Shortcomings of current instruments include the requirement for relatively large amounts of data, the use of specialized procedures, subjective assessments and the lack of responsiveness to therapeutic interventions (14-16). Frailty biomarkers have the potential to complement the clinical evaluation of frailty including aiding in its diagnosis, assessment of severity, and evaluation of prognosis (17, 18). Although clinical frailty assessment instruments may be more effective from a population-based approach, whereas frailty biomarkers, may be able to individualize diagnosis/prognosis and personalize care by determining an individual’s biological frailty profile.
The utility of biomarkers has been demonstrated in the diagnosis of some cancers and other disorders (19-21), and to assess treatment responses and disease progression in various diseases (22, 23). In addition, biomarkers offer the promise of precision medicine where a person’s care and treatment is based on one’s genetics and biology and there is a need to extend and explore this promise to the challenges posed by frailty.
A promising venue to study frailty biomarkers are longitudinal research initiatives that collect clinical data on aging in large numbers of participants coupled with biological samples. The Canadian Longitudinal Study on Aging (CLSA) (www.clsa-elcv.ca/) which studies the aging process in over 50,000 Canadians is one of the largest and most comprehensive initiatives in the world. Selected participants undergo comprehensive clinical evaluations including frailty assessments supplemented with hand grip-strength, timed up-and-go, chair rise, 4-metre walk and standing balance. Using a FI of 90 possible health deficits with 0.25 as the threshold for frailty, approximately 7% of CLSA participants are frail, increasing to approximately 11% in those over the age of 75 (24). Biological specimens are collected in a selected cohort every three years including whole blood, serum, plasma and buffy coat containing peripheral blood mononuclear cells (Table 1).  For a biomarker to be considered for inclusion in the CLSA, it needs to have at least preliminary evidence to link the biomarker to a pathophysiological frailty pathway or mechanism.
Here we report the proceedings of a symposium convened by the Canadian Frailty Network (CFN) (www.cfn-nce.ca) in collaboration with the CLSA. The overall objective of this meeting was to inform future efforts of CFN to improve the availability of frailty biomarkers and to guide further CFN funded analyses of biological samples held by the CLSA. Specific objectives of the symposium were to:
1.    Explore the current state of evidence of biomarkers for frailty.
2.    Obtain an understanding of other frailty biomarker initiatives around the world.
3.    Obtain guidance on biomarkers that could be measured in the CLSA biological samples, in addition to those already being assayed, to identify the biological, biochemical and genetic factors/markers associated with the onset and progression of frailty in order to develop predictive, prognostic and diagnostic tests to aid in the care and treatment of people living with frailty.

Table 1 Biomarkers currently being analysed in the CLSA samples

Table 1
Biomarkers currently being analysed in the CLSA samples

Workshop details

The workshop was held in Toronto, Canada on January 15th 2018. Participants for the workshop were international stakeholders, key opinion leaders and frailty and/or biomarker experts.  They were identified using one or more of the following criteria: they were leading large scale initiatives involving frailty and the measurement of biomarkers, they were investigators on CFN-funded research grants studying frailty and measuring biomarkers, they had published peer-reviewed studies on frailty and related biomarkers, were clinicians caring for older adults living with frailty and/or were in relevant decision-making roles. The twenty-two delegate attendees included basic researchers (e.g., biochemist, pharmacologist, immunologist), clinician researchers (e.g., intensivist, geriatrician, cardiologist) and health-care administrators/policy experts.

The current state of frailty biomarkers and frailty assessment

Multiple clinical frailty assessment instruments exist, each with advantages and disadvantages and it is not clear which instrument is optimal for a particular care setting (15). In addition, due to variability in their measurement characteristics, the prevalence of frailty depends on the instrument utilized (25, 26). This variability together with the subjective nature of some frailty assessments have generated increasing interest in integrating clinical assessments with objective laboratory-based biomarker tests.   The working definition for this discussion was that a biomarker is, “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.” (NIH working group) (27). Some potential uses of biomarkers conceived by the 2001 NIH working group are listed in Table 2.

Table 2 Potential utility of biomarkers

Table 2
Potential utility of biomarkers


Biomarkers may lead to the identification of mechanisms and pathophysiological pathways leading to or causing frailty, and the early identification of frailty which, importantly, may be reversible in early stages (28, 29). In addition, identification and modification of treatment plans in those identified as frail may ameliorate poor outcomes from healthcare interventions (30-32). Lastly biomarkers may help with selecting interventions, monitor response to treatments, and identify those for whom some interventions are unlikely to be of benefit.
Although no single biomarker has yet proven to be of sufficient diagnostic or prognostic utility to be clinically useful for frailty, there has been increasing work on integrating laboratory evaluation in the assessment of frailty (33). As an example, a FI composed of laboratory results has been developed and shown to be effective in identifying frailty in individuals living in the community and those in long-term care facilities (17). This index of biomarkers (the FI-Lab) predicts increased risk of death in community-dwelling and institutionalized individuals and it can be combined with a clinical FI to improve prediction of outcomes in older adults (17, 18, 34-36).
A recent systematic review of frailty biomarker studies in community-dwelling individuals of trials using a validated definition of frailty and comparing two or more biomarkers was recently reported (33). Biomarkers related to immune function, inflammation, endocrine function and metabolic syndrome were the most frequently reported. There have been conflicting results regarding the utility of single biomarkers for the detection and assessment of frailty. Associations with frailty have been shown for inflammatory markers (e.g., interleukin-6 (IL-6), Tumor necrosis factor-alpha (TNF-alpha), C-reactive protein (CRP)), reduced total lymphocyte count and other markers of immunocompetence, although this has not been observed with some markers that have been associated with aging such as telomere length and oxidative stress, when studied individually (37). Other systematic reviews have come to the same conclusion. The variety of biomarkers reported in recent systematic reviews are summarized in Table 3 (33, 38-40).

Table 3 Summary of biomarkers examined in prior studies

Table 3
Summary of biomarkers examined in prior studies


Specific Frailty Biomarker Considerations

Frailty Index (FI) approach to biomarkers

Consistent with the deficit accumulation approach, Mitnitski et al. (18) re-analysed the data in Collerton et al. (37) to create an FI composed of 40 biomarkers of cellular ageing, inflammation and haematology. Like the FI-Lab, this suggests that currently available single biomarkers have relatively low information value by themselves when compared to a combination of multiple markers and an FI approach with a panel of biomarkers may be a more promising avenue of investigation, and more closely correspond to how deficits propagate to give rise to frailty (41, 42).
The deficit accumulation approach has been validated in animal models of frailty, where both laboratory-based and clinical FI tools have been developed and behave similarly as in humans (9, 43-45). Animal studies also indicate that levels of pro-inflammatory cytokines increase in proportion to FI scores (45). Further studies suggest that mechanisms that give rise to frailty are present at the cellular/subcellular levels and scale up to impact function at the level of the organ and ultimately the organism (46, 47). The availability of animal models of frailty with biomarker and clinical criteria has led to the study of possible interventions for frailty in pre-clinical models using resveratrol, caloric restriction and an angiotensin converting enzyme inhibitor (48, 49). All this is consistent with the multifactorial nature of frailty, thus requiring the evaluation of a biological network (and not a standalone biomarker) for adequately capturing the complexity of the state.
In conclusion, measuring a larger set of biomarkers (e.g., 40) is likely to be better than measuring a smaller set. Given current evidence, inflammatory markers should be included in any frailty biomarker panel as should measures of metabolism (large panel metabolomics) and other measures related to aging processes (e.g., telomere length, oxidative stress, DNA damage and repair).  Following a standard procedure for creating a FI from clinical data, for a biomarker to be included as a deficit in a FI, the biomarker should have the following characteristics, as outlined by Searle et al. (8):
•    Abnormal biomarker levels should be associated with a negative health-related outcome.
•    The risk associated with abnormal levels of the biomarker should be higher as age increases.
•    The biomarker should be neither too rare (i.e., < 1%) nor too common (i.e., > 80% by age 80).
•    Biomarker data should be available for at least 80% of individuals.

Frailty and skeletal muscle

Muscle mass, has for the most part, been neglected in frailty assessment instruments, perhaps because there are no good practical tools to measure it.  Most current methods are crude (e.g., calipers, body mass index), complex or difficult to use or access (e.g., magnetic resonance imaging). Computerized tomography (CT) scans are of increasing promise due to their availability and growing evidence indicates that muscle mass seen on CT scans, as evidenced by psoas muscle area, correlates with outcomes including mortality (50-52). In particular, muscle mass analysis can be an important measure for frailty assessment in patients who are already undergoing CT scans for other medical reasons and can be of added prognostic value to other biochemical biomarkers/tests. Although CTs are of potential utility, further research is required before they can be recommended for routine clinical use. Bioelectrical impedance analysis (BIA) and dual-energy X-ray absorptiometry (DEXA) are also commonly used for the assessment of muscle mass but their availability are limited and their ability to guide clinical practice requires further evaluation (53, 54). Overall, skeletal muscle assessment should be tailored to the objectives desired and the availability of the diagnostic test. Also different tests may be required for different care settings (55).

Anemia and hypoalbuminemia

Anemia has been shown to increase rapidly with age and is associated with adverse geriatric outcomes such as disability, reduced physical function, cognitive dysfunction and dementia (56, 57). The presence of anemia is also associated with an increased likelihood of frailty (58). It is unknown whether this is due to reduced oxygen delivery, resulting in inactivity and fatigue, or if anemia is the result of other conditions linked to frailty, including chronic inflammation and/or malnutrition. In addition, it is unknown if the successful treatment of anemia can reverse or attenuate frailty?
Reduced serum albumin has been associated with disability and mortality (59). While hypoalbuminemia was previously thought to be a marker of malnutrition, it is now known that it is more likely due to systemic inflammation (60). The incorporation of albumin and anemia into clinical frailty criteria has been shown to be predictive of death and worsening disability in selected surgical populations (e.g., transcatheter aortic valve replacement) (www.frailtytool.com) (61, 62). Moreover, these two parameters are commonly measured clinically thereby increasing their availability for the assessment of frailty.

Other biomarkers

Other biomarkers such as vitamin D and androgen levels have also been linked to frailty.  A systematic review suggests that low levels of vitamin D are associated with an increase in the risk of frailty (63), although potential benefits of vitamin D supplementation on frailty requires additional evidence (64). There is no consensus in the literature with respect to the biological role of androgens in frailty. An earlier review concluded that more work needs to be done (65), although recent longitudinal studies show an association between testosterone and FI scores (66, 67). Evidence to date suggests that testosterone treatment may be beneficial in older people (68-70). However, further evidence is required before recommending that these biomarkers be incorporated into future frailty biomarker panels.


Current frailty studies and biomarker databases

Participants of this CFN sponsored workshop are leading a number of international initiatives investigating the potential utility of biomarkers for frailty assessment. These initiatives are discussed below.

Sarcopenia & Physical fRailty IN older people: multi-componenT Treatment strategies (SPRINTT); www.mysprintt.eu/en/wp

The SPRINTT project (5 years) was started in 2014 as a private/public partnership between the European Commission and the European Federation of Pharmaceutical Industries and Associations.  The overall goal of SPRINTT is to develop innovative therapeutic interventions against physical frailty and sarcopenia as a prototype geriatric indication. SPRINTT includes a definitive large phase III randomized clinical trial (RCT) (71) comparing the efficacy of a multi-component intervention in preventing mobility disability, based on long-term structured physical activity, personalised nutritional counselling/dietary intervention and an information and communication technology intervention, versus a healthy ageing lifestyle education program.  The RCT completed recruitment in November 2018 and follow-up is planned to be completed in 2019.  One of the accomplishments of SPRINTT was the operationalization of a new condition incorporating physical frailty and sarcopenia, whose methodology was endorsed by the European Medicines Agency (EMA) (72).
For each participant biological samples are collected (whole blood, serum, plasma and urine) for future analysis (Table 4).  The collection of these biospecimens has two objectives: 1) development of discriminant biomarkers that will allow for comparison of biomarker levels between physical frailty and sarcopenia (at baseline) versus physical frailty and no sarcopenia. 2) development of prognostic and predictive biomarkers for the identification of study participants at high risk for the development of mobility disability and prediction of response to the multi-component intervention.

Table 4 Candidate biomarkers being considered for SPRINTT

Table 4
Candidate biomarkers being considered for SPRINTT

FRAILOMIC; www.frailomic.org

The main goal of the FRAILOMIC initiative is to develop clinical instruments composed of clinical data, laboratory biomarkers and omics-based laboratory biomarkers. Specifically, the three objectives of FRAILOMIC are to develop clinical instruments that can: 1) improve diagnostic accuracy of frailty in day-to-day practice, 2) predict the risk of frailty and 3) assess the prognosis of frailty in terms of death, disability and other adverse outcomes.  FRAILOMIC has four main phases, with the first two phases now complete, which have identified the biomarkers listed in Table 5.  Preliminary results show that biomarkers potentially useful for the diagnosis of frailty are different from those useful for assessing frailty risk and prognosis.

Table 5 Biomarkers identified in exploratory phase. Biomarkers identified according to their relationship with function, muscle and longevity. Selection of biomarkers attempted to be as inclusive as practical

Table 5
Biomarkers identified in exploratory phase. Biomarkers identified according to their relationship with function, muscle and longevity. Selection of biomarkers attempted to be as inclusive as practical



Biomarkers, definition of frailty and fit within clinical practice

Key to the selection of biomarkers for frailty assessment will be an agreed upon definition of frailty. However, coming to a consensus definition of frailty and choosing the appropriate clinical instruments have proven to be difficult.  Current frailty instruments/tools in use today, whether based on a deficit accumulation model or a phenotypic model have varied sensitivity, specificity, and positive and negative predictive values (73-76). Moreover, frailty assessment tools may perform differently in different populations and settings (11-13). Overall, it is very likely that there will be no single best reference clinical frailty assessment instrument, so to proceed with biomarker work, a single measurement paradigm will need to be selected and used consistently.
In regards to the selection of the reference paradigm, it should be connected with the underlying biology of frailty. It is generally agreed that frailty is a result of the age-associated accumulation of deficits across multiple systems and therefore the FI paradigm seems to be a reasonable starting point. The FI is consistent in different care settings, countries and species because of its basis in systems biology. In addition, the FI is multi-dimensional and allows for grading of frailty severity, which is an important determinant of outcome. One drawback to the FI is that it may be impractical as a clinical assessment tool since it requires large amounts of data, stemming from the results of a comprehensive evaluation of the individual but this is less of a disadvantage when used for research purposes as a reference standard.
Any adopted biomarker needs to pair with the clinical instrument used to better diagnose, predict risk and determine therapeutic responsiveness (77). The utility of the FI is that it can utilize readily measured clinical deficits, many of which are currently recorded and even exist in electronic health records (10). The FI can utilize readily available and routine laboratory tests, that can also be augmented with novel biomarkers. This would apply to laboratory tests such hemoglobin, albumin, estimated glomerular filtration rate, low density lipoprotein and glycosylated hemoglobin. If a combination of these perform well in a frailty assessment tool such as the FI, the utility of more complex, more difficult to measure and more expensive biomarkers would need to be demonstrated before they are adopted. Overall, to be most applicable and advance the possibility of personalized medicine for people living with frailty, it would be better to start broadly with the FI and then tailor biomarkers to the specific needs of the person.

Clinical frailty assessment instruments versus biomarker-based frailty assessment tools

Biomarkers may direct clinicians, but do not replace geriatric expertise. Any biomarker(s) chosen will need to assist clinical activities beyond what already exists, be easier to use and be more cost-effective. Biomarkers that are difficult and expensive to measure or require specialized equipment should be reserved for environments dealing with complex cases or for research purposes.  Some assessments, such as gait speed combined with grip strength, may already provide data that biomarkers promise to do (74, 75). Biomarkers will need to increase the sensitivity, specificity, negative predictive value and positive predictive value of clinical evaluations or instruments, especially for risk prediction and this will need to be demonstrated in clinical use. Also, for their adoption, it will need to be demonstrated that their utilization is associated with better outcomes.
When a clinical frailty assessment tool is used in combination with biomarkers, it is important that the biomarker(s) increase the utility of the tool as follows: 1) It allows the tool to better identify a population as truly frail than the clinical tool alone, and 2) It allows for better selection of a care plan for an individual than the clinical tool alone. To determine if a biomarker has these key characteristics, further research will be required once candidate biomarkers are identified. As an example, if a biomarker(s) identifies a nutrient deficiency, we need to know that acting on that information would improve outcomes. This will likely only be answered by doing randomized clinical trials with the ability to detect changes in patient-centered outcomes such as quality of life. Overall, the identification of frailty in a more rigorous manner, with attendant care plans, would be a major step forward.

Generally available versus novel biomarkers

Biomarkers that are only associated with a particular disease may not be of high interest when considering frailty since they are likely specific for the disease. Biomarkers more applicable to frailty are likely those that are closer to the physiology and basic biology of aging. Studies of the basic biology of aging has led to a number of new interventions targeting fundamental aging processes that have been found to impact development of age-related disorders, including many chronic diseases, geriatric syndromes and loss of resilience/loss of ability to respond to stressors in animal models (78, 79). These new interventions will need to be tested in human clinical trials. Biomarkers may be better able to characterize the populations enrolled in these trials to avoid the heterogeneity that is characteristic of frailty and aging. If successful, these interventions may be able to fundamentally change geriatric medicine, with the ultimate goal of increasing health-span and not necessarily life-span.

Need for a variety of biomarkers

An ideal biomarker would present early in the course of frailty and correlate with its severity. There are indications that different biomarkers are needed for risk stratification, diagnosis and prognosis. Overall, it is unlikely that a single biomarker can address all these objectives (80). Moreover, the utility of a diagnostic biomarker is dependent on the prevalence of the condition which in frailty are care setting dependent increasing from a small percentage of the general population to as much as 60% in hospital, and majority in long-term care (81-84). In very high prevalence environments, diagnostic frailty biomarkers may have limited utility. For prognosis, we need to be as precise as possible as to what they predict to make them useful to clinicians; risk of poor functional outcomes may be as important as mortality.

Broader versus narrower sets of biomarkers

For discovery, there may be value in adopting a biomarker approach based on metabolomic profiling, understanding that this approach might not have clinical value as yet but will help to generate a wide range of markers, which from a research perspective may have great utility. The added benefit of utilizing metabolomics on whole blood, plasma or urine is that once mass spectrometry (MS) or nuclear magnetic resonance (NMR) data are captured, these data can be reanalyzed in the future for new markers. Metabolomics has already produced results that are leading to frailty interventions, such as ketogenic compounds, for which clinical trials are about to start (85, 86). There are at least 20 proteins or peptides in mice related to frailty where preliminary human trials are being considered (87, 88). Also with newer techniques, proteins can be measured in very small quantities of sample (e.g., 10 microlitres). The caveat is that although MS and NMR are both powerful technologies with great discovery (research) potential, at this time translating their application to clinical settings is more difficult than other approaches, such as determining inflammation by measuring interleukins or acute phase CRP.
There are large studies utilizing metabolomics but no such large study exists for frailty investigation. However, this work has already commenced in the CLSA with 1,000 people having metabolomics analysis. Further metabolomics analysis within the CLSA database will be a unique opportunity to significantly add to the evidentiary base.



This workshop brought together experts in the field of biomarkers, aging and/or frailty to discuss frailty biomarkers. This discussion will inform CFN and CLSA in planning for the analysis of a select group of biomarkers in the CLSA samples and a follow-up data access competition to utilize these new biomarker data in conjunction with the wealth of clinical data collected by the CLSA.
There was considerable discussion regarding the appropriate way to think about frailty and how the frame of reference will influence the choice of biomarkers to analyze.  Considerations discussed included probabilistic medicine and population-based care versus precision medicine or individualized medicine and how these overall approaches should influence the use of biomarkers and their incorporation into practice and ultimately influence policy.  There was debate about care settings and biomarker utility across care settings. In addition, the overall objective of using biomarkers as diagnostic tools for frailty versus their use to quantify frailty risk versus their use as frailty prognostic tools were topics of high interest and discussion. By the end of the workshop, there was a clear sense that research and clinical trials need to be conducted regarding frailty biomarkers and the need for the development of biomarker tools to be used either alone or with currently existing clinical frailty assessment instruments such as the FI.
Arising from the meeting, the list of principles to guide future CFN biomarker initiatives including its partnership with the CLSA, are as follows:
1)    Biomarkers should reflect a pathophysiological pathway or mechanism that is fundamental to frailty onset, development/progression and severity. Conceptually there maybe two categories of biomarkers:
i)    Biomarkers that are linked with frailty but are not causal to the pathophysiology of frailty. These would not be actionable.
ii)    Biomarkers that are a component of the pathophysiology of frailty and have a causal role. These would be actionable such that the modulation of the biomarker could directly affect the onset or severity of frailty and/or progression of frailty.
2)    The utility of biomarkers can be classified into two different types:
i)    Biomarkers to increase the utility of (or support) existing clinical frailty measures (e.g., FI).
ii)    Biomarkers to be used independently of clinical frailty measures.
3)    Biomarkers should be able to be embedded in clinical assessments and tools, but more research on how to best achieve this is needed. Concomitant use of both a clinical frailty assessment instrument and biomarkers is likely to be the optimal method to bring about personalized frailty assessment and individualized care plans.
4)    Biomarkers chosen for a clinical tool should be evaluated on their ability to accomplish the ultimate clinical purpose. For instance, biomarkers used for diagnosis may be different from those used for risk assessment, which may differ from those used for prognosis.
5)    Different care settings are likely to require different biomarkers due to variation in prevalence of both frailty and biomarkers in different populations.
6)    Any candidate biomarker should be validated in different populations, care settings and environments.
7)    An ideal frailty biomarker would be able to measure the effectiveness of an intervention.
8)    Practical considerations related to ease of measurement (i.e., special instruments and/or expertise required) and ease of securing biological samples (e.g., tissue biopsy vs blood sample collection) should be considered when selecting frailty biomarkers.


Acknowledgements: The Canadian Frailty Network (CFN) is a pan-Canadian network focused on the care of older citizens living with frailty. CFN is comprised of nearly 3,500 corporate and non-profit partners, researchers, scientists, health-care professionals, citizens, students, trainees, educators, and decision-makers. CFN supports and catalyzes original research and innovations to improve the care and quality of life of Canadians living with frailty across all settings of care. The Network also trains the next generation of health-care professionals and scientists. CFN is funded by the Government of Canada through the Networks of Centres of Excellence (NCE) Program. In early 2017, in recognition of the work done in its first five years of operation, the Government of Canada announced funding for a second five-term (2017-2022).

Conflict of interest disclosures: Matteo Cesari, Emanuele Marzetti and Leocadio Rodriguez Mañas are partners of the SPRINTT consortium, which is partly funded by the European Federation of Pharmaceutical Industries and Associations (EFPIA) . Kenneth Rockwood is President and Chief Science Officer of DGI Clinical, which in the last five years has contracts with pharma and device manufacturers (Baxter, Baxalta, Shire, Hollister, Nutricia, Roche, Otsuka) on individualized outcome measurement. In 2017 he attended an advisory board meeting with Lundbeck. He is Associate Director of the Canadian Consortium on Neurodegeneration in Aging, which is funded by the Canadian Institutes of Health Research, and with additional funding from the Alzheimer Society of Canada and several other charities, as well as, in first phase (2013-2018), from Pfizer Canada and Sanofi Canada.   He receives career support from the Dalhousie Medical Research Foundation as the Kathryn Allen Weldon Professor of Alzheimer Research, and research support from the Canadian Institutes of Health Research, the Nova Scotia Health Research Foundation, the Capital Health Research Fund and the Fountain Family Innovation Fund of the Nova Scotia Health Authority Foundation.



1.    Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. 2001;56(3):M146-56.
2.    Shamliyan T, Talley KM, Ramakrishnan R, Kane RL. Association of frailty with survival: a systematic literature review. Ageing Res Rev. 2013;12(2):719-36.
3.    Fried LP, Ferrucci L, Darer J, Williamson JD, Anderson G. Untangling the concepts of disability, frailty, and comorbidity: implications for improved targeting and care. J Gerontol A Biol Sci Med Sci. 2004;59(3):255-63.
4.    Bagshaw SM, Stelfox HT, McDermid RC, Rolfson DB, Tsuyuki RT, Baig N, et al. Association between frailty and short- and long-term outcomes among critically ill patients: a multicentre prospective cohort study. CMAJ. 2014;186(2):E95-102.
5.    Rockwood K, Howlett SE. Fifteen years of progress in understanding frailty and health in aging. BMC Med. 2018;16(1):220.
6.    Bergman H, Ferrucci L, Guralnik J, Hogan DB, Hummel S, Karunananthan S, et al. Frailty: an emerging research and clinical paradigm–issues and controversies. J Gerontol A Biol Sci Med Sci. 2007;62(7):731-7.
7.    Mitnitski AB, Mogilner AJ, Rockwood K. Accumulation of deficits as a proxy measure of aging. ScientificWorldJournal. 2001;1:323-36.
8.    Searle SD, Mitnitski A, Gahbauer EA, Gill TM, Rockwood K. A standard procedure for creating a frailty index. BMC Geriatr. 2008;8:24.
9.    Rockwood K, Blodgett JM, Theou O, Sun MH, Feridooni HA, Mitnitski A, et al. A Frailty Index Based On Deficit Accumulation Quantifies Mortality Risk in Humans and in Mice. Sci Rep. 2017;7:43068.
10.    Clegg A, Bates C, Young J, Ryan R, Nichols L, Ann Teale E, et al. Development and validation of an electronic frailty index using routine primary care electronic health record data. Age Ageing. 2016;45(3):353-60.
11.    Sternberg SA, Wershof Schwartz A, Karunananthan S, Bergman H, Mark Clarfield A. The identification of frailty: a systematic literature review. J Am Geriatr Soc. 2011;59(11):2129-38.
12.    Cesari M, Gambassi G, van Kan GA, Vellas B. The frailty phenotype and the frailty index: different instruments for different purposes. Age Ageing. 2014;43(1):10-2.
13.    Abbasi M, Rolfson D, Khera AS, Dabravolskaj J, Dent E, Xia L. Identification and management of frailty in the primary care setting. CMAJ. 2018;190(38):E1134-E40.
14.    Gilardi F, Capanna A, Ferraro M, Scarcella P, Marazzi MC, Palombi L, et al. Frailty screening and assessment tools: a review of characteristics and use in Public Health. Ann Ig. 2018;30(2):128-39.
15.    Dent E, Kowal P, Hoogendijk EO. Frailty measurement in research and clinical practice: A review. Eur J Intern Med. 2016;31:3-10.
16.    de Vries NM, Staal JB, van Ravensberg CD, Hobbelen JS, Olde Rikkert MG, Nijhuis-van der Sanden MW. Outcome instruments to measure frailty: a systematic review. Ageing Res Rev. 2011;10(1):104-14.
17.    Howlett SE, Rockwood MR, Mitnitski A, Rockwood K. Standard laboratory tests to identify older adults at increased risk of death. BMC Med. 2014;12:171.
18.    Mitnitski A, Collerton J, Martin-Ruiz C, Jagger C, von Zglinicki T, Rockwood K, et al. Age-related frailty and its association with biological markers of ageing. BMC Med. 2015;13:161.
19.    Sylman JL, Mitrugno A, Atallah M, Tormoen GW, Shatzel JJ, Tassi Yunga S, et al. The Predictive Value of Inflammation-Related Peripheral Blood Measurements in Cancer Staging and Prognosis. Front Oncol. 2018;8:78.
20.    Yu IS, Cheung WY. A Contemporary Review of the Treatment Landscape and the Role of Predictive and Prognostic Biomarkers in Pancreatic Adenocarcinoma. Can J Gastroenterol Hepatol. 2018;2018:1863535.
21.    Shang J, Yamashita T, Fukui Y, Song D, Li X, Zhai Y, et al. Different Associations of Plasma Biomarkers in Alzheimer’s Disease, Mild Cognitive Impairment, Vascular Dementia, and Ischemic Stroke. J Clin Neurol. 2018;14(1):29-34.
22.    Chen RY, Via LE, Dodd LE, Walzl G, Malherbe ST, Loxton AG, et al. Using biomarkers to predict TB treatment duration (Predict TB): a prospective, randomized, noninferiority, treatment shortening clinical trial. Gates Open Res. 2017;1:9.
23.    Kostikas K, Brindicci C, Patalano F. Blood Eosinophils as Biomarkers to Drive Treatment Choices in Asthma and COPD. Curr Drug Targets. 2018;19(16):1882-96.
24.    Kanters DM, Griffith LE, Hogan DB, Richardson J, Patterson C, Raina P. Assessing the measurement properties of a Frailty Index across the age spectrum in the Canadian Longitudinal Study on Aging. J Epidemiol Community Health. 2017;71(8):794-9.
25.    van Iersel MB, Rikkert MG. Frailty criteria give heterogeneous results when applied in clinical practice. J Am Geriatr Soc. 2006;54(4):728-9.
26.    Kehler DS, Ferguson T, Stammers AN, Bohm C, Arora RC, Duhamel TA, et al. Prevalence of frailty in Canadians 18-79 years old in the Canadian Health Measures Survey. Bmc Geriatr. 2017;17(1):28.
27.    Atkinson A, Jr; Magnuson, Warren G.; Colburn, Wayne A.; DeGruttola, Victor G; DeMets, David L.; Downing, Gregory J; Hoth, Daniel F; Oates, John A; Peck, Carl C; Schooley, Robert T; Spilker, Bert A; Woodcock, Janet; Zeger, Scott L. . Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001;69(3):89-95.
28.    Jha SR, Hannu MK, Newton PJ, Wilhelm K, Hayward CS, Jabbour A, et al. Reversibility of Frailty After Bridge-to-Transplant Ventricular Assist Device Implantation or Heart Transplantation. Transplant Direct. 2017;3(7):e167.
29.    Gustavson AM, Falvey JR, Jankowski CM, Stevens-Lapsley JE. Public Health Impact of Frailty: Role of Physical Therapists. J Frailty Aging. 2017;6(1):2-5.
30.    Mosquera C, Spaniolas K, Fitzgerald TL. Impact of frailty on surgical outcomes: The right patient for the right procedure. Surgery. 2016;160(2):272-80.
31.    Sepehri A, Beggs T, Hassan A, Rigatto C, Shaw-Daigle C, Tangri N, et al. The impact of frailty on outcomes after cardiac surgery: a systematic review. J Thorac Cardiovasc Surg. 2014;148(6):3110-7.
32.    Fagard K, Leonard S, Deschodt M, Devriendt E, Wolthuis A, Prenen H, et al. The impact of frailty on postoperative outcomes in individuals aged 65 and over undergoing elective surgery for colorectal cancer: A systematic review. J Geriatr Oncol. 2016;7(6):479-91.
33.    Lee L, Patel T, Hillier LM, Maulkhan N, Slonim K, Costa A. Identifying frailty in primary care: A systematic review. Geriatr Gerontol Int. 2017;17(10):1358-77.
34.    Rockwood K, McMillan M, Mitnitski A, Howlett SE. A Frailty Index Based on Common Laboratory Tests in Comparison With a Clinical Frailty Index for Older Adults in Long-Term Care Facilities. J Am Med Dir Assoc. 2015;16(10):842-7.
35.    Blodgett JM, Theou O, Howlett SE, Rockwood K. A frailty index from common clinical and laboratory tests predicts increased risk of death across the life course. Geroscience. 2017.
36.    Ritt M, Jager J, Ritt JI, Sieber CC, Gassmann KG. Operationalizing a frailty index using routine blood and urine tests. Clin Interv Aging. 2017;12:1029-40.
37.    Collerton J, Martin-Ruiz C, Davies K, Hilkens CM, Isaacs J, Kolenda C, et al. Frailty and the role of inflammation, immunosenescence and cellular ageing in the very old: cross-sectional findings from the Newcastle 85+ Study. Mech Ageing Dev. 2012;133(6):456-66.
38.    Fernandez-Garrido J, Ruiz-Ros V, Buigues C, Navarro-Martinez R, Cauli O. Clinical features of prefrail older individuals and emerging peripheral biomarkers: a systematic review. Arch Gerontol Geriatr. 2014;59(1):7-17.
39.    Sanchez-Flores M, Marcos-Perez D, Costa S, Teixeira JP, Bonassi S, Pasaro E, et al. Oxidative stress, genomic features and DNA repair in frail elderly: A systematic review. Ageing Res Rev. 2017;37:1-15.
40.    Soysal P, Isik AT, Carvalho AF, Fernandes BS, Solmi M, Schofield P, et al. Oxidative stress and frailty: A systematic review and synthesis of the best evidence. Maturitas. 2017;99:66-72.
41.    Farrell SG, Mitnitski AB, Theou O, Rockwood K, Rutenberg AD. Probing the network structure of health deficits in human aging. Phys Rev E. 2018;98(3).
42.    Rutenberg AD, Mitnitski AB, Farrell SG, Rockwood K. Unifying aging and frailty through complex dynamical networks. Exp Gerontol. 2018;107:126-9.
43.    Parks RJ, Fares E, Macdonald JK, Ernst MC, Sinal CJ, Rockwood K, et al. A procedure for creating a frailty index based on deficit accumulation in aging mice. J Gerontol A Biol Sci Med Sci. 2012;67(3):217-27.
44.    Whitehead JC, Hildebrand BA, Sun M, Rockwood MR, Rose RA, Rockwood K, et al. A clinical frailty index in aging mice: comparisons with frailty index data in humans. J Gerontol A Biol Sci Med Sci. 2014;69(6):621-32.
45.    Kane AE, Keller KM, Heinze-Milne S, Grandy SA, Howlett SE. A Murine Frailty index based on Clinical and Laboratory Measurements: Links between Frailty and Pro-inflammatory Cytokines differ in a Sex-specific Manner. J Gerontol A Biol Sci Med Sci. 2018.
46.    Jansen HJ, Moghtadaei M, Mackasey M, Rafferty SA, Bogachev O, Sapp JL, et al. Atrial structure, function and arrhythmogenesis in aged and frail mice. Sci Rep. 2017;7:44336.
47.    Feridooni HA, Sun MH, Rockwood K, Howlett SE. Reliability of a Frailty Index Based on the Clinical Assessment of Health Deficits in Male C57BL/6J Mice. J Gerontol A Biol Sci Med Sci. 2015;70(6):686-93.
48.    Kane AE, Hilmer SN, Boyer D, Gavin K, Nines D, Howlett SE, et al. Impact of Longevity Interventions on a Validated Mouse Clinical Frailty Index. J Gerontol A Biol Sci Med Sci. 2016;71(3):333-9.
49.    Keller K, Kane A, Heinze-Milne S, Grandy SA, Howlett SE. Chronic treatment with the ACE inhibitor enalapril attenuates the development of frailty and differentially modifies pro-and anti-inflammatory cytokines in aging male and female C57BL/6 mice. J Gerontol A Biol Sci Med Sci. 2018.
50.    Mamane S, Mullie L, Piazza N, Martucci G, Morais J, Vigano A, et al. Psoas Muscle Area and All-Cause Mortality After Transcatheter Aortic Valve Replacement: The Montreal-Munich Study. Can J Cardiol. 2016;32(2):177-82.
51.    Zuckerman J, Ades M, Mullie L, Trnkus A, Morin JF, Langlois Y, et al. Psoas Muscle Area and Length of Stay in Older Adults Undergoing Cardiac Operations. Ann Thorac Surg. 2017;103(5):1498-504.
52.    Bibas L, Saleh E, Al-Kharji S, Chetrit J, Mullie L, Cantarovich M, et al. Muscle Mass and Mortality After Cardiac Transplantation. Transplantation. 2018;102(12):2101-7.
53.    Cruz-Jentoft AJ, Bahat G, Bauer J, Boirie Y, Bruyere O, Cederholm T, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48(1):16-31.
54.    Studenski SA, Peters KW, Alley DE, Cawthon PM, McLean RR, Harris TB, et al. The FNIH sarcopenia project: rationale, study description, conference recommendations, and final estimates. J Gerontol A Biol Sci Med Sci. 2014;69(5):547-58.
55.    Beaudart C, McCloskey E, Bruyere O, Cesari M, Rolland Y, Rizzoli R, et al. Sarcopenia in daily practice: assessment and management. Bmc Geriatr. 2016;16(1):170.
56.    Penninx BW, Pahor M, Cesari M, Corsi AM, Woodman RC, Bandinelli S, et al. Anemia is associated with disability and decreased physical performance and muscle strength in the elderly. J Am Geriatr Soc. 2004;52(5):719-24.
57.    Hong CH, Falvey C, Harris TB, Simonsick EM, Satterfield S, Ferrucci L, et al. Anemia and risk of dementia in older adults: findings from the Health ABC study. Neurology. 2013;81(6):528-33.
58.    Chaves PH, Semba RD, Leng SX, Woodman RC, Ferrucci L, Guralnik JM, et al. Impact of anemia and cardiovascular disease on frailty status of community-dwelling older women: the Women’s Health and Aging Studies I and II. J Gerontol A Biol Sci Med Sci. 2005;60(6):729-35.
59.    Corti MC, Guralnik JM, Salive ME, Sorkin JD. Serum albumin level and physical disability as predictors of mortality in older persons. JAMA. 1994;272(13):1036-42.
60.    Don BR, Kaysen G. Serum albumin: relationship to inflammation and nutrition. Semin Dial. 2004;17(6):432-7.
61.    Robinson TN, Eiseman B, Wallace JI, Church SD, McFann KK, Pfister SM, et al. Redefining geriatric preoperative assessment using frailty, disability and co-morbidity. Ann Surg. 2009;250(3):449-55.
62.    Afilalo J, Lauck S, Kim DH, Lefevre T, Piazza N, Lachapelle K, et al. Frailty in Older Adults Undergoing Aortic Valve Replacement: The FRAILTY-AVR Study. J Am Coll Cardiol. 2017;70(6):689-700.
63.    Zhou J, Huang P, Liu P, Hao Q, Chen S, Dong B, et al. Association of vitamin D deficiency and frailty: A systematic review and meta-analysis. Maturitas. 2016;94:70-6.
64.    Bruyere O, Cavalier E, Buckinx F, Reginster JY. Relevance of vitamin D in the pathogenesis and therapy of frailty. Curr Opin Clin Nutr Metab Care. 2017;20(1):26-9.
65.    Afilalo J. Androgen deficiency as a biological determinant of frailty: hope or hype? J Am Geriatr Soc. 2014;62(6):1174-8.
66.    Swiecicka A, Eendebak R, Lunt M, O’Neill TW, Bartfai G, Casanueva FF, et al. Reproductive Hormone Levels Predict Changes in Frailty Status in Community-Dwelling Older Men: European Male Ageing Study Prospective Data. J Clin Endocrinol Metab. 2018;103(2):701-9.
67.    Hsu B, Cumming RG, Naganathan V, Blyth FM, Le Couteur DG, Seibel MJ, et al. Associations between circulating reproductive hormones and SHBG and prevalent and incident metabolic syndrome in community-dwelling older men: the Concord Health and Ageing in Men Project. J Clin Endocrinol Metab. 2014;99(12):E2686-91.
68.    Srinivas-Shankar U, Roberts SA, Connolly MJ, O’Connell MD, Adams JE, Oldham JA, et al. Effects of testosterone on muscle strength, physical function, body composition, and quality of life in intermediate-frail and frail elderly men: a randomized, double-blind, placebo-controlled study. J Clin Endocrinol Metab. 2010;95(2):639-50.
69.    Toma M, McAlister FA, Coglianese EE, Vidi V, Vasaiwala S, Bakal JA, et al. Testosterone supplementation in heart failure: a meta-analysis. Circ Heart Fail. 2012;5(3):315-21.
70.    Neto WK, Gama EF, Rocha LY, Ramos CC, Taets W, Scapini KB, et al. Effects of testosterone on lean mass gain in elderly men: systematic review with meta-analysis of controlled and randomized studies. Age (Dordr). 2015;37(1):9742.
71.    Landi F, Cesari M, Calvani R, Cherubini A, Di Bari M, Bejuit R, et al. The “Sarcopenia and Physical fRailty IN older people: multi-componenT Treatment strategies” (SPRINTT) randomized controlled trial: design and methods. Aging Clin Exp Res. 2017;29(1):89-100.
72.    Cesari M, Marzetti E, Calvani R, Vellas B, Bernabei R, Bordes P, et al. The need of operational paradigms for frailty in older persons: the SPRINTT project. Aging Clin Exp Res. 2017;29(1):3-10.
73.    Warnier RM, van Rossum E, van Velthuijsen E, Mulder WJ, Schols JM, Kempen GI. Validity, Reliability and Feasibility of Tools to Identify Frail Older Patients in Inpatient Hospital Care: A Systematic Review. J Nutr Health Aging. 2016;20(2):218-30.
74.    Lee L, Patel T, Costa A, Bryce E, Hillier LM, Slonim K, et al. Screening for frailty in primary care: Accuracy of gait speed and hand-grip strength. Can Fam Physician. 2017;63(1):e51-e7.
75.    Lee LP, Tejal; Hillier, Loretta, M; Locklin, Jason; Milligan, James; Pefanis, John; Costa. Andrew; Lee, Joseph; Slonim, Karen; Giangregorio, Lora; Hunter, Susan; Keller, Heather; Boscart, Veronique Frailty Screening and Case-Finding for Complex Chronic Conditions in Older Adults in Primary Care. Geriatrics. 2018;3(3).
76.    van Loon IN, Goto NA, Boereboom FTJ, Bots ML, Verhaar MC, Hamaker ME. Frailty Screening Tools for Elderly Patients Incident to Dialysis. Clin J Am Soc Nephrol. 2017;12(9):1480-8.
77.    Rodriguez-Manas L, Fried LP. Frailty in the clinical scenario. Lancet. 2015;385(9968):e7-e9.
78.    Tchkonia T, Kirkland JL. Aging, Cell Senescence, and Chronic Disease: Emerging Therapeutic Strategies. JAMA. 2018;320(13):1319-20.
79.    Xu M, Pirtskhalava T, Farr JN, Weigand BM, Palmer AK, Weivoda MM, et al. Senolytics improve physical function and increase lifespan in old age. Nat Med. 2018;24(8):1246-56.
80.    Cardoso AL, Fernandes A, Aguilar-Pimentel JA, de Angelis MH, Guedes JR, Brito MA, et al. Towards frailty biomarkers: Candidates from genes and pathways regulated in aging and age-related diseases. Ageing Res Rev. 2018;47:214-77.
81.    Collard RM, Boter H, Schoevers RA, Oude Voshaar RC. Prevalence of frailty in community-dwelling older persons: a systematic review. J Am Geriatr Soc. 2012;60(8):1487-92.
82.    Le Maguet P, Roquilly A, Lasocki S, Asehnoune K, Carise E, Saint Martin M, et al. Prevalence and impact of frailty on mortality in elderly ICU patients: a prospective, multicenter, observational study. Intensive Care Med. 2014;40(5):674-82.
83.    Kojima G. Prevalence of Frailty in Nursing Homes: A Systematic Review and Meta-Analysis. J Am Med Dir Assoc. 2015;16(11):940-5.
84.    Chong E, Ho E, Baldevarona-Llego J, Chan M, Wu L, Tay L. Frailty and Risk of Adverse Outcomes in Hospitalized Older Adults: A Comparison of Different Frailty Measures. J Am Med Dir Assoc. 2017;18(7):638 e7- e11.
85.    Longo VD, Antebi A, Bartke A, Barzilai N, Brown-Borg HM, Caruso C, et al. Interventions to Slow Aging in Humans: Are We Ready? Aging Cell. 2015;14(4):497-510.
86.    Newman JC, Covarrubias AJ, Zhao M, Yu X, Gut P, Ng CP, et al. Ketogenic Diet Reduces Midlife Mortality and Improves Memory in Aging Mice. Cell Metab. 2017;26(3):547-57 e8.
87.    Lange KWL, Katharina M.; Makulska-Gertruda, Ewelina; Nakamura, Yukiko; Reissmann, Andreas; Kanaya, Shigehiko; Hausera, Joachim. Ketogenic diets and Alzheimer’s disease. Foo Science and Human Wellness. 2017;6(1):1-9.
88.    Han YM, Bedarida T, Ding Y, Somba BK, Lu Q, Wang Q, et al. beta-Hydroxybutyrate Prevents Vascular Senescence through hnRNP A1-Mediated Upregulation of Oct4. Mol Cell. 2018;71(6):1064-78 e5.