Aging Clinical and Experimental Research https://doi.org/10.1007/s40520-019-01127-4 ORIGINAL ARTICLE Individual and cumulative association of commonly used biomarkers on frailty: a cross-sectional analysis of the Mexican Health and Aging Study Mario Ulises Pérez‑Zepeda1,2  · Carmen García‑Peña1 · María Fernanda Carrillo‑Vega3 Received: 16 July 2018 / Accepted: 11 January 2019 © Springer Nature Switzerland AG 2019 Abstract Frailty has been recognized as a common condition in older adults, however, there is scarce information on the association between frailty and commonly used biomarkers. The aim of this study was to assess the individual and cumulative associa- tion of biomarkers with frailty status. This is a cross-sectional analysis of the 2012 wave of the Mexican Health and Aging Study. A sub-sample of 60-year or older adults with anthropometric measurements was analyzed. Frailty was defined with a 31-item frailty index and those considered frail had a score ≥ 0.21. Biomarkers were further categorized as normal/abnormal and tested both one by one and grouped (according to their usual cutoff values). Adjusted logistic models were performed. A total of 1128 older adults were analyzed and their mean age was 69.45 years and 51.24% of them were women. 26.7% (n = 301) were categorized as frail. Individual biomarkers associated with frailty after adjusting for confounding were: hemoglobin [odds ratio (OR) 1.67, 95% confidence interval (CI) 1.13–2.46, p = 0.009], glycated hemoglobin (OR 2.04, 95% CI 1.54–2.7, p < 0.001) and vitamin D (OR 1.53, 95% CI 1.13–2.07, p = 0.005). Those with ≥ 4 abnormal biomarkers had an independent association with frailty when compared to those without any abnormal biomarker (OR 2.64, 95% CI 1.3–5.25, p = 0.005). Aside from the individual associations of specific biomarkers, our findings show that an incremental associa- tion of abnormal biomarkers increases the probability of frailty, accounting for the multidimensional nature of frailty and the possible interplay between components of the system that potentiate to give rise to a negative condition such as frailty. Keywords Frail older adult · Geriatric epidemiology · Biomarkers · Aging Introduction Adverse outcomes usually arise when the older adult is exposed to stressors [2], that in normal conditions would Frailty has been recognized as a condition with multiple have no effect—or minimal—on the overall health status of causality, characterized by an increase in the individual’s the individual. The fact that frailty is a common condition vulnerability for developing adverse outcomes (e.g., dis- in older adults and that this group of the population has an ability, dependency, institutionalization and/or death) [1]. accelerated growth—in comparison to other age groups— increases the need for reliable information that helps to char- acterize it in different settings [3, 4], and in turn improves Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s40520 -019-01127 -4) contains older adult care. In addition to narrowing the gap on the supplementary material, which is available to authorized users. knowledge of frailty, it also helps in providing data on how commonly used biomarkers are part of some chronic dis- * María Fernanda Carrillo-Vega eases or even multimorbidity, closely related to the genesis marifercave@yahoo.com.mx of frailty. Having this in mind, older adults with abnormal 1 Instituto Nacional de Geriatría, Mexico, México biomarkers could be screened with readily available tools 2 that include clinical features of the condition or may help Instituto de Envejecimiento, Facultad de Medicina, Pontificia Universidad Javeriana, Bogotá, Colombia to identify older adults that could benefit from a thorough 3 geriatric assessment [5]. This is particularly true in those Instituto Nacional de Geriatría, Periférico Sur 2767, colonia San Jerónimo Lídice, delegación Magdalena Contreras, settings with scarce resources specialized for care of the 10200 Mexico, Mexico older adults [6]. Vol.:(012 3456789) Aging Clinical and Experimental Research In particular, biomarkers are substances that can be meas- community-dwelling Mexican older adults. A set of ques- ured objectively in the body, reflecting underlying processes tionnaires (e.g., socio-demographic, health-related, cogni- either of normal or abnormal physiology, and some of them tive performance, functional status, etc.) were applied head- are used commonly in the health care of older adults, as to-head by standardized interviewers. Each wave included ‘routine labs’ (e.g., vitamin D, cholesterol, thyroid hormones a sub-sample in which anthropometry and biomarkers were and C-reactive protein) [7, 8]. There is evidence that these included. Regarding biomarkers, technical specificities for biomarkers have an association with frailty as a whole or each of the biomarkers are available upon request. with some of its components (i.e., physical performance, For the purposes of this report, only 2012 data was ana- exhaustion, low physical activity, specific deficits, etc); in lyzed. A total of 18,465 participants who were 50 years or addition, some other conditions such as geriatric syndromes older were assessed in this wave in which a sub-sample of have been also associated to abnormal values of biomarkers 1128 individuals was included (with biomarkers and anthro- (e.g., falls, late-life depressive symptoms, cognitive impair- pometric measurements). ment, etc.). In particular, vitamin D has been associated with frailty [9], in different contexts and populations [10–14], and impacting the physical component of frailty through its Measurements effect in muscle strength. Moreover, different studies have shown that C-reactive protein (CRP) is associated with Dependent variable decreased gait speed and lower handgrip strength, along with overall lower physical function [15]. Finally, some evi- A frailty index (FI) constructed with standard procedures dence shows that high levels of thyroid-stimulating hormone [22] was used to categorize older adults as frail or non- (TSH) are associated with disability, cognitive dysfunction, frail. The FI included 31 deficits from different domains: osteoporosis and cardiovascular disease [16]. self-rated health, comorbidities, mental health and somatic There is a complex interrelation of the physiologic symptoms (see Sect. 4 of the supplementary material). Each systems that gives rise to frailty [17]. For example, some deficit was transformed into a score of 0 (deficit absent) to evidence points to the fact that a set of hormones have a 1 (deficit present) with possible intermediate scores; after- stronger association with frailty when combined [18]. This wards all the scores were summed and divided by the total goes in the same line on how the multi-causality of frailty number of deficits (i.e., 31) for each participant. The final could be better explained by synergic etiology rather than score ranged from zero (no deficits, lowest frailty score pos- by the one-cause one-disease paradigm [19]. This has been sible) to one (all deficits present, highest frailty score pos- shown by Rockwood et al., when composing a frailty index sible). A cutoff value of 0.21 or higher was used to define with biomarkers (i.e., addition of abnormal laboratory val- frailty, a value validated for Mexican older adults previously ues), that had similar association as the conventional frailty [23]. index with adverse outcomes [8]. To our knowledge, there is no current information on Independent variables the association of commonly used biomarkers and frailty in community-dwelling older adults. Therefore, the aim of the Biomarkers were obtained from a blood sample of periph- present study was to assess the individual and cumulative eral venipuncture by trained personnel. Collected samples association of biomarkers with frailty status. We hypoth- were centrifuged 30 min after the venipuncture at 2,500RPM esized that the association between frailty and serum mark- for 15–20  min; serum was separated by this technique ers would be stronger when added in comparison to any of and preserved in 2 ml tubes under refrigeration (2–8 °C). the biomarkers alone. These measurements were performed between October and November of 2012. Cutoff values to define abnormal- ity were defined as follows: CRP ≥ 3 mg/dL, total choles- Methods terol ≤ 200  mg/dL, high-density lipoprotein cholesterol (HDL-c) ≥ 40 mg/dL, thyroid-stimulating hormone (TSH) Design and settings 0.45–4.12 mIU/mL, hemoglobin ≥ 13.5 g/dL for men and ≥ 12.0 g/dL for women, vitamin D ≥ 20 ng/ml and glycated This is a cross-sectional analysis of the third (2012) wave hemoglobin ≤ 6.5%. In addition to individual biomarkers, from the Mexican Health and Aging Study (MHAS), a a composite variable was constructed by adding abnormal prospective cohort conducted in Mexico since 2001. The biomarkers (ranging from 0 = no abnormal biomarkers to aim and design of the MHAS are published elsewhere 4 or more abnormal biomarkers). These scores were then [20, 21]. In brief, there are four waves of this study (2001, contrasted between frail and non-frail older adults (having as 2003, 2012, and 2015) with a representative sample of the reference group those without any abnormal biomarker). 1 3 Aging Clinical and Experimental Research Confounding Ethical issues To further describe the study population, socio-demo- The Institutional Review Boards or Ethics Committees of graphic characteristics included, age, sex, marital status the University of Texas Medical Branch in the United States, (married or not married), and years of education were the Instituto Nacional de Estadística y Geografía, the Insti- included as well as body mass index (BMI). In addition, tuto Nacional de Salud Pública and the Instituto Nacional de these variables were also used in the adjusted models to Geriatría in Mexico approved the study. All study subjects consider confounding. signed an informed consent form. Statistical analysis Results Descriptive statistics were performed, quantitative vari- From a total of 1128, 26.7% were categorized as frail ables are presented as means (± SD) and categorical (n = 301). Their mean age was 69.5 years (± SDS 7.8), and variables as relative frequencies (percentage). Univariate the frail older adults were significantly older (p < 0.001). analysis was performed to compare frail and non-frail par- Regarding years in school, frail older adults had significantly ticipants in baseline demographics and biomarkers, using fewer completed years in school compared to those without independent samples t test, and Chi square test for cat- frailty (p < 0.001). There is no difference between frail and egorical variables. A multiple logistic regression model non-frail people for BMI (p < 0.05). There was a significant was fitted with frailty as the dependent variable, odds ratio higher proportion of abnormal and glycated hemoglobin, (OR) with 95% confidence intervals (CI) were reported in vitamin D levels and CRP in frail older adults (see Table 1). unadjusted and adjusted (for age, sex, marital status, years As shown by the multivariate logistic regression adjusted in school and BMI) fashion. All analyses were performed models (Table 2), frail older adults who had lower levels of with statistical package software STATA 14® (StataCorp hemoglobin had 1.67 times the risk of being frail (p < 0.05) 4905, Lakeway Drive, College Station, TX 77845 USA). compared with people with higher levels and lower levels Table 1 General description of Total (n = 1128) Frail (n = 301 [26.68%]) Non-frail p value the sample by frailty status (n = 827 [73.32%]) Age, mean (SD) 69.45 (7.77) 71.93 (9.26) 68.54 (6.94) < 0.001 Women, n (%) 578 (51.24) 204 (35.29) 374 (64.71) < 0.001 Married, n (%) 654 (57.98) 143 (21.87) 511 (78.13) < 0.001 Years in school, mean (SD) 4.5 (4.26) 3.03 (3.32) 5.11 (4.43) < 0.001 Body mass index, mean (SD) 28.41 (5.3) 28.89 (6.12) 28.24 (4.96) 0.068 Hemoglobin, n (%) 173 (14.6) 99 (57.2) 74 (42.7) 0.015 Glycated hemoglobin, n (%) 467 (39.5) 267 (57.1) 200 (42.8) < 0.001 Total cholesterol, n (%) 167 (14.7) 77 (46.1) 90 (53.8) 0.366 HDL cholesterol, n (%) 538 (47.6) 270 (50.1) 268 (49.8) 0.586 Thyroid-stimulating hormone, n (%) 204 (17.2) 109 (53.4) 95 (46.5) 0.133 Vitamin D, n (%) 416 (36.8) 236 (56.7) 180 (43.2) < 0.001 C-reactive protein, n (%) 478 (40.4) 271 (56.7) 207 (43.3) < 0.001 Number of abnormal biomarkers  0 86 (7.62) 14 (16.28) 72 (83.72) < 0.001  1 229 (20.3) 192 (83.84) 37 (16.16)  2 368 (32.62) 85 (28.24) 283 (76.9)  3 275 (24.38) 92 (33.45) 183 (66.55)  ≥ 4 170 (15.07) 73 (42.94) 97(57.06) Frailty was considered present in those older adults with a frailty index score ≥ 0.21 Cutoff values to define abnormality were defined as follows: C-reactive protein (CRP) 3 mg/dL, total cho- lesterol < 200  mg/dL, high-density lipoprotein cholesterol [HDL-c] ≥ 40  mg/dL, thyroid-stimulating hor- mone (TSH) 0.45–4.12 mIU/mL, hemoglobin ≥ 13.5  g/dL for men and ≥ 12.0  g/dL for women, vitamin D ≥ 20 ng/ml and glycated hemoglobin ≤ 6.5% 1 3 Aging Clinical and Experimental Research Table 2 Multivariate logistic Unadjusted OR (95% CI) p value Adjusted OR (95% CI)a p value regression for frailty, individual abnormal biomarkers and Hemoglobin 1.42 (1.01–2.03) 0.049 1.67 (1.13–2.46) 0.009 number of abnormal biomarkers Glycated hemoglobin 1.88 (1.44–2.44) < 0.001 2.04 (1.54–2.7) < 0.001 Total cholesterol 1.01 (0.7–1.47) 0.934 1.04 (0.7–1.55) 0.822 HDL cholesterol 0.88 (0.68–1.15) 0.377 0.87 (0.65–1.16) 0.347 Thyroid-stimulating hormone 1.33 (0.92–1.93) 0.129 1.19 (0.8–1.7) 0.375 Vitamin D 1.7 (1.3–2.22) < 0.001 1.53 (1.13–2.07) 0.005 C-reactive protein 1.7 (1.36–2.17) < 0.001 1.4 (1.08–1.81) 0.01 Number of abnormal biomarkers  None Reference  1 0.9 (0.5–1.94) 0.979 0.8 (0.39–1.61) 0.53  2 1.54 (0.82–2.87) 0.17 1.24 (0.65–2.39) 0.503  3 2.58 (1.38–4.82) 0.003 2.02 (1.04–3.8) 0.035  4 or more 3.87 (2.02–7.39) < 0.001 2.64 (1.3–5.25) 0.005 Frailty was considered present in those older adults with a frailty index score ≥ 0.21 Cutoff values to define abnormality were defined as follows: C-reactive protein (CRP) 3 mg/dL, total cho- lesterol < 200  mg/dL, high-density lipoprotein cholesterol (HDL-c) ≥ 40  mg/dL, thyroid-stimulating hor- mone (TSH) 0.45–4.12 mIU/mL, hemoglobin ≥ 13.5  g/dL for men and ≥ 12.0  g/dL for women, vitamin D ≥ 20 ng/ml and glycated hemoglobin ≤ 6.5% a Adjusted models for: age, sex, marital status, years in school and body mass index of vitamin D indicates 1.53 times the risk of frailty in com- also associated with frailty. Moreover, CRP as a marker of parison with people who had higher concentrations. Moreo- inflammatory status was also associated with frailty in our ver, older adults with higher levels of glycated hemoglobin study, as already shown in a previous work [28]. have 2.04 times the risk of being frail. Regarding CRP, the When adding the number of abnormal biomarkers, the adjusted model showed an OR of 1.4 (95% CI 1.08–1.81, strength of association with frailty was higher; this finding p = 0.01). Although other parameters did not demonstrate is more in line with the proposed deficit accumulation path statistical significance, trends are shown in the expected to frailty, that has also shown to be associated with adverse direction. Regarding the incremental association of the outcomes [29]. In this respect, it is known that the sum of addition of abnormal biomarkers, as the number of altered altered biomarkers can be useful in identifying the individual parameters increases, the risk of frailty also increased, with risk of frailty (because the biomarkers make part of chronic the highest significance for ≥ 4 abnormal biomarkers (OR diseases or multimorbidity that lead to frailty) [2, 8], func- 2.64; 95% CI 1.3–5.25; p < 0.05), when compared to no tional decline [30] and disability [11]. Notwithstanding, Van abnormal biomarker. Hemelrijck et. al. constructed a mortality score based on the number of abnormal biomarkers, and noted that those older adults who had more than three altered biomarkers were Discussion at significantly higher risk for 3- and 7-year mortality than those with one or two biomarkers (p < 0.01) [28], similar According to our results, there is an association between to our results, that showed an association with frailty when frailty and commonly used biomarkers individually—espe- three or more abnormal biomarkers were present, but not cially for those that have shown to be related to the physi- with one or two. It is important to stress the fact that inter- opathology of frailty—and also an incremental association vening in those conditions related to abnormal biomarkers when adding abnormal biomarkers. Our results showed or as a group, could finally impact on the development of that hemoglobin and vitamin D are associated individually frailty or its progression. For example, the supplementation with frailty, results that are similar with those reported by of vitamin D has shown to improve muscle strength, this in Schoufour et al. where hemoglobin was inversely correlated turn would turn in a better physical status that could prevent with frailty in both the unadjusted and adjusted models frailty or even halt its progression. (p < 0.001) and with those from Sanchis et al. where levels Our study has a number of relevant limitations that of vitamin D were lower in frail people in comparison with should be considered to interpret the results appropriately. non-frail (p < 0.05) [24, 25]. In addition, an association has The first one is that from our data, no causal relationship also been established between diabetes and frailty [26, 27], could be inferred due to its cross-sectional nature. Future as in our study a higher level of glycated hemoglobin was research should focus on the pathophysiological mechanisms 1 3 Aging Clinical and Experimental Research that underlie abnormal biomarkers and the impact of these undifferentiated older emergency department patients: a sys- alterations in trajectories of frailty and its adverse outcomes tematic review and meta-analysis. Acad Emerg Med 22:1–21 such as disability. Although a vast number of biomarkers 6. Gutierrez-Robledo LM (2002) Looking at the future of geriatric care in developing countries. J Gerontol A Biol Sci Med Sci had been reported to be linked with frailty and other condi- 57:M162–M167 tions in older adults, in the Mexican context, those reported 7. Blodgett JM, Theou O, Howlett SE et al (2017) A frailty index in the present work are the most used in the clinical asset, from common clinical and laboratory tests predicts increased so that comparisons with other reports can be difficult to risk of death across the life course. GeroScience 39:447–455 8. Rockwood K, Mcmillan M, Mitnitski A et al (2015) A frailty done. However, other biomarkers could have been studied index based on common laboratory tests in comparison with a that have been associated with frailty in other populations; in clinical frailty index for older adults in long-term care facilities. this work we were limited to those available in MHAS. One J Am Med Dir Assoc 16:842–847 common problem when it comes to older adults is the use 9. Gutierrez-Robledo LM, Avila-Funes JA, Amieva H et al (2016) Association of low serum 25-hydroxyvitamin D levels with of cutoff values, the vast majority of these reference values the frailty syndrome in Mexican community-dwelling elderly. come from younger population or not the same population of Aging Male 19:58–63 the manuscript; another approach could be based on the dis- 1 0. Buta B, Choudhury PP, Xue QL et al (2017) The association of tribution of the biomarkers data, however, we decided using vitamin D deficiency and incident frailty in older women: the role of cardiometabolic diseases. J Am Geriatr Soc 65:619–624 commonly used reference values to make the interpretation 11. Ensrud KE, Blackwell TL, Cauley JA et al (2011) Circulating of the associations easier. 25-hydroxyvitamin D levels and frailty in older men: the osteo- Specific biomarkers were associated with frailty, particu- porotic fractures in men study. J Am Geriatr Soc 59:101–106 larly those involved in the pathophysiology of this condition. 12. Ensrud KE, Ewing SK, Fredman L et al (2010) Circulating 25-hydroxyvitamin D levels and frailty status in older women. In addition, a higher number of abnormal biomarkers was J Clin Endocrinol Metab 95:5266–5273 also associated with frailty irrespective of which was, point- 13. Orces CH (2017) Prevalence of clinically relevant muscle weak- ing also to the deficit accumulation pathway to frailty. ness and its association with vitamin D status among older adults in Ecuador. Aging Clin Exp Res 29:943–949 14. Puts MT, Visser M, Twisk JW et  al (2005) Endocrine and inflammatory markers as predictors of frailty. Clin Endocrinol Funding The Mexican Health and Aging Study was supported in its (Oxf) 63:403–411 2012 version by National Institutes of Health/National Institute on 15. Cesari M, Penninx BW, Pahor M et al (2004) Inflammatory Aging (R01AG018016, R Wong, PI). markers and physical performance in older persons: the InCHI- ANTI study. J Gerontol A Biol Sci Med Sci 59:242–248 Compliance with ethical standards 1 6. Virgini VS, Wijsman LW, Rodondi N et al (2014) Subclini- cal thyroid dysfunction and functional capacity among elderly. Thyroid 24:208–214 Conflict of interest The authors declare that they have no conflict of 1 7. Fried LP, Xue QL, Cappola AR et al (2009) Nonlinear multisys- interest. tem physiological dysregulation associated with frailty in older women: implications for etiology and treatment. J Gerontol A Ethical approval All procedures performed in the present study were Biol Sci Med Sci 64:1049–1057 in accordance with the ethical standards of the National Geriatrics 18. Cappola AR, Xue QL, Fried LP (2009) Multiple hormonal defi- Institute research committee and with the 1964 Helsinki declaration ciencies in anabolic hormones are found in frail older women: and its later amendments or comparable ethical standards. the women’s health and aging studies. J Gerontol A Biol Sci Med Sci 64:243–248 Informed consent Informed consent was obtained from all individual 19. Ferrucci L, Cavazzini C, Corsi A et al (2002) Biomarkers of participants included in the study. frailty in older persons. J Endocrinol Investig 25:10–15 20. Wong R, Michaels-Obregon A, Palloni A (2017) Cohort profile: the Mexican Health and Aging Study (MHAS). Int J Epidemiol 46:e2 References 21. Wong R, Michaels-Obregon A, Palloni A et al (2015) Progres- sion of aging in Mexico: the Mexican Health and Aging Study 1. Morley JE, Vellas B, Van Kan GA et al (2013) Frailty consensus: (MHAS) 2012. Salud Publica Mex 57:S79–S89 a call to action. J Am Med Dir Assoc 14:392–397 22. Searle SD, Mitnitski A, Gahbauer EA et al (2008) A standard 2. Mitnitski A, Song X, Rockwood K (2013) Assessing biologi- procedure for creating a frailty index. BMC Geriatr 8:24 cal aging: the origin of deficit accumulation. Biogerontology 23. Garcia-Gonzalez JJ, Garcia-Pena C, Franco-Marina F et  al 14:709–717 (2009) A frailty index to predict the mortality risk in a popula- 3. Rockwood K, Theou O, Mitnitski A (2015) What are frailty instru- tion of senior Mexican adults. BMC Geriatr 9:47 ments for? Age Ageing 44:545–547 24. Schalk BW, Visser M, Deeg DJ et al (2004) Lower levels of 4. Walston J, Hadley EC, Ferrucci L et al (2006) Research agenda for serum albumin and total cholesterol and future decline in func- frailty in older adults: toward a better understanding of physiology tional performance in older persons: the longitudinal aging and etiology: summary from the American Geriatrics Society/ study Amsterdam. Age Ageing 33:266–272 National Institute on Aging Research Conference on Frailty in 2 5. Schoufour JD, Echteld MA, Boonstra A et al (2016) Biochemi- Older Adults. J Am Geriatr Soc 54:991–1001 cal measures and frailty in people with intellectual disabilities. 5. Carpenter CR, Shelton E, Fowler S et al (2015) Risk factors Age Ageing 45:142–148 and screening instruments to predict adverse outcomes for 1 3 Aging Clinical and Experimental Research 2 6. Castrejon-Perez RC, Aguilar-Salinas CA, Gutierrez-Robledo LM 29. Shardell M, Hicks GE, Miller RR et al (2009) Association of low et al (2017) Frailty, diabetes, and the convergence of chronic dis- vitamin D levels with the frailty syndrome in men and women. J ease in an age-related condition: a population-based nationwide Gerontol A Biol Sci Med Sci 64:69–75 cross-sectional analysis of the Mexican nutrition and health sur- 3 0. Van Hemelrijck M, Harari D, Garmo H et al (2012) Biomarker- vey. Aging Clin Exp Res 30:935–941 based score to predict mortality in persons aged 50 years and 27. Castrejon-Perez RC, Gutierrez-Robledo LM, Cesari M et al (2017) older: a new approach in the Swedish AMORIS study. Int J Mol Diabetes mellitus, hypertension and frailty: A population-based, Epidemiol Genet 3:66–76 cross-sectional study of Mexican older adults. Geriatr Gerontol Int 17:925–930 Publisher’s Note Springer Nature remains neutral with regard to 2 8. Zuliani G, Volpato S, Romagnoni F et al (2004) Combined meas- jurisdictional claims in published maps and institutional affiliations. urement of serum albumin and high-density lipoprotein choles- terol strongly predicts mortality in frail older nursing-home resi- dents. Aging Clin Exp Res 16:472–475 1 3