Improving Our Understanding of Physiological Age Through the Use of Clinical Prediction Models

Editor's Note: Please see associated Patient Case Quiz


The adequate assessment of physiological age is important for clinicians caring for older adults with cardiovascular disease. Geriatric cardiologists and geriatricians often utilize concepts such as multimorbidity,1,2 frailty,3 and disability to improve on their assessment of physiological age and prediction of downstream cardiovascular and noncardiovascular events.4 However, and unfortunately, chronological age is often used as a sole surrogate for physiological age. This brief review focuses on maximizing our understanding of the data when chronological age is used as a surrogate for physiological age. We also discuss the use of clinical prediction models (CPM) that appropriately model the variable chronological age, and incorporate geriatric specific variables. The eventual goal is to standardize and improve the front-line clinicians' assessment of physiological age when confronted with older adult patients with cardiovascular disease.

The Eyeball Test in Predicting Cardiovascular Events

Clinicians often believe that their gut instinct, or the "eyeball test," for future risk prediction in their patients is satisfactory. This "eyeball test" is often based on chronological age and other subjective measures. Suffice it to say that CPM often predict risk better than subjective clinician prediction for long-term events.5 It is important to note that clinicians on average are good at risk prediction in patients who are either extremely healthy or extremely sick. Objective risk prediction, in the heterogeneous intermediate risk group, is felt to be improved by utilizing CPM.6-8

Appreciating Physiological Age Through Appropriate Statistical Modeling of Chronological Age

Chronological age is often used as a surrogate for physiological age. We demonstrated in the associated Patient Case Quiz the pitfalls of discretizing a continuous variable such as chronological age.9-12 In the following example, we show two standard and accepted modeling techniques (linear and cubic spline) of the variable chronological age that enhances our understanding of the relationship between chronological age and an outcome.

This example is from an observational study of the effects of chronological age on repeat coronary revascularization in patients with end-stage renal disease.13 In this particular example (Figure 1), an entirely opposite interpretation can again result based on which technique is used (repeat coronary revascularization decreases with age or increases after approximately 70 years). The primary objective of this example is to ensure that there is a search for nonlinear relationships between chronological age and outcomes, with the caveat that linear modeling in most instances is probably adequate in representing this relationship.

Figure 1: Two Standard Modeling Strategies for the Continuous Variable Age as a Function of Repeat Revascularization in the Fit of an Unadjusted Parametric Survival Regression

Figure 1
Green solid line = age modeled as a cubic spline; Green dashed line = age as a linear variable The graph presents the hazard ratio graphically as a function of age exponentially in the hazard scale. Adapted with permission from Krishnaswami et al.6

The Use of Clinical Prediction Models in Risk Assessment of Older Adults

CPM are extremely useful tools to objectively predict downstream events. They incorporate the wealth of data needed for prognostication. Currently, there is reported to be 796 published CPM in the area of cardiovascular disease. Unfortunately, limited use of these CPM due to end-user difficulty is common.14 An integration of CPM into the modern electronic health records in novel ways should decrease the burden of its use for frontline clinicians. Common examples of CPM are risk scores such as CHADS2, CHADSVaSc, and SYNTAX.

Although not proven, the utility of routine use of CPM promises to deliver precision medicine to older adults with cardiovascular disease. However, it is important to note that current CPM are often developed using data from randomized clinical trials where it is known that older adults with multimorbidity are mostly excluded.15 Increasing the recruitment of older adults in clinical trials and the subsequent incorporation of geriatric specific variables (such as frailty, disability, etc.) within CPM are critically needed to objectively deliver on the promise of precision medicine.16

It is important to mention that the development of CPM focused on older adults they should take into account some fundamental concepts: parsimony of variables selected, functional form of continuous variables, the effects of one variable on another (interaction or effect modification).17 Lastly, although originally described over two decades ago, the concept of age-scaled comorbidity should be further studied and perhaps, if validated, be incorporated into clinical prediction models of risk prediction in older adults.18

Geriatric Cardiology

Current clinical practice guidelines based on randomized clinical trials are not often useful for clinical care and risk prediction in older adults with multiple chronic conditions/multimorbidity. The burgeoning field of geriatric cardiology is entrenched in the complex care of older adults.19 The eyeball test can no longer be the sole method to determine physiological age and risk. Routine assessment of frailty, understanding the ramification of multimorbidity, along with the routine use of geriatric specific CPM should become a standard of care in older adults with cardiovascular disease.

Take Home Message(s):

  1. Clinicians should be cautious when reviewing data where age is discretized. Chronological age is a spectrum and should be viewed as such.
  2. Modeling techniques that display the relationship between chronological age and an outcome as a spectrum is needed to guide future health policies.
  3. A search for nonlinear relationships should be routinely undertaken with the caveat that a linear relationship may fully represent the relationship.
  4. The development of future de novo CPM and recalibration of prior CPM should incorporate geriatric specific variables such as frailty, multimorbidity, disability, etc.
  5. Ensuring future ease of CPM use will hopefully standardize the assessment of physiological age in older adults with cardiovascular disease.



  1. Rich MW. Multimorbidity in Older Adults With Cardiovascular Disease. Sep 19, 2016. Accessed May 23, 2017.
  2. Boyd CM, Darer J, Boult C, Fried LP, Boult L, Wu AW. Clinical practice guidelines and quality of care for older patients with multiple comorbid diseases: implications for pay for performance. JAMA 2005;294:716-24.
  3. Afilalo J, Alexander KP, Mack MJ, et al. Frailty assessment in the cardiovascular care of older adults. J Am Coll Cardiol 2014;63:747-62.
  4. Afilalo J, Karunananthan S, Eisenberg MJ, Alexander KP, Bergman H. Role of frailty in patients with cardiovascular disease. Am J Cardiol 2009;103:1616-21.
  5. Jain R, Duval S, Adabag S. How accurate is the eyeball test?: a comparison of physician's subjective assessment versus statistical methods in estimating mortality risk after cardiac surgery. Circ Cardiovasc Qual Outcomes 2014;7:151-6.
  6. Alba AC, Agoritsas T, Jankowski M, et al. Risk prediction models for mortality in ambulatory patients with heart failure: a systematic review. Circ Heart Fail 2013;6:881-9.
  7. Califf RM, Pencina MJ. Predictive models in heart failure: who cares? Circ Heart Fail 2013;6:877-8.
  8. Pons JM, Borras JM, Espinas JA, Moreno V, Cardona M, Granados A. Subjective versus statistical model assessment of mortality risk in open heart surgical procedures. Ann Thorac Surg 1999;67:635-40.
  9. Royston P, Altman DG, Sauerbrei W. Dichotomizing continuous predictors in multiple regression: a bad idea. Stat Med 2006;25:127-41.
  10. Altman DG, Lausen B, Sauerbrei W, Schumacher M. Dangers of using "optimal" cutpoints in the evaluation of prognostic factors. J Natl Cancer Inst 1994;86:829-35.
  11. Bennette C, Vickers A. Against quantiles: categorization of continuous variables in epidemiologic research, and its discontents. BMC Med Res Methodol 2012;12:21.
  12. Sauerbrei W, Abrahamowicz M, Altman DG, le Cessie S, Carpenter J, STRATOS initiative. STRengthening analytical thinking for observational studies: the STRATOS initiative. Stat Med 2014;33:5413-32.
  13. Krishnaswami A, Alloggiamento T, Forman DE, Leong TK, Go AS, McCulloch CE. Association of age to mortality and repeat revascularization in end-stage renal disease patients: implications for clinicians and future health policies. Perm J 2016;20:4-9.
  14. Wessler BS, Lai Yh L, Kramer W, et al. Clinical prediction models for cardiovascular disease: Tufts Predictive Analytics and Comparative Effectiveness Clinical Prediction Model Database. Circ Cardiovasc Qual Outcomes 2015;8:368-75.
  15. Skolnick AH, Alexander KP. Older adults in clinical research and drug development: closing the geriatric gap. Circ Cardiovasc Qual Outcomes 2015;8:631-3.
  16. Krishnaswami A, Forman, E, Maurer MS, Lei SJ. A decision-making framework for objective risk assessment in older adults with severe symptomatic aortic stenosis. Curr Geriatr Rep 2015;4:338-46.
  17. Krishnaswami A, Christle JW, Froelicher V. Can we improve mortality estimation in women after treadmill testing by using sex-specific scores? JAMA Cardiol 2017;2:22-4.
  18. Normand SL, Morris CN, Fung KS, McNeil BJ, Epstein AM. Development and validation of a claims based index for adjusting for risk of mortality: the case of acute myocardial infarction. J Clin Epidemiol 1995;48:229-43.
  19. Bell SP, Orr NM, Dodson JA, et al. What to expect from the evolving field of geriatric cardiology. J Am Coll Cardiol 2015;66:1286-99.

Keywords: Cardiovascular Diseases, Comorbidity, Geriatrics, Kidney Failure, Chronic, Risk Assessment, Ticlopidine

< Back to Listings