Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 Mar 22.
Published in final edited form as: Arch Intern Med. 2010 Mar 22;170(6):566–569. doi: 10.1001/archinternmed.2010.8

Individualized Medical Decision Making: Necessary, Achievable, but Not Yet Attainable

Liana Fraenkel 1,2, Terri R Fried 1,2
PMCID: PMC2845319  NIHMSID: NIHMS173937  PMID: 20308644

Abstract

There is an urgent need to provide older persons with individualized information regarding the benefits and harms of different diagnostic and treatment strategies. This need results from the growing recognition of the heterogeneity in outcomes among older persons with differing comorbidity profiles. The importance of heterogeneity in outcomes has been most thoroughly described in cancer screening. The heterogeneity of benefits and harms resulting from treatment is not yet as well appreciated. Warfarin versus aspirin for the reduction of stroke risk in nonvalvular atrial fibrillation (NVAF) provides an example of a treatment for which the benefit to harm ratio may actually reverse according to an older person’s comorbidities, thus highlighting the importance of basing this treatment decision on individualized outcome data Despite the wealth of studies in NVAF, many assumptions are necessary to calculate patient-specific outcomes, and these assumptions may lead to substantial over- or under-estimation of benefits and harms. Improving care for patients with co-morbidities will require substantive increases in the efforts and resources allocated towards the collection and dissemination of outcome data for patients with varying comorbidities.

Keywords: Comorbidity, outcomes, decision making

Comorbidities and the need for individualized medical decision-making

Evidence is growing for the need to utilize patient-specific data regarding the expected benefits and harms of different diagnostic and treatment strategies to inform medical decision making. It is well recognized that the “average” benefits and risks as measured in randomized controlled trials (RCTs) may not apply to the individual patient.1, 2 This is particularly true for older patients, in whom comorbid conditions and functional disability can diminish the benefits of standard diagnostic and therapeutic strategies.3, 4 For example, an 81 year old woman with no comorbidities has a life expectancy of 13.8 years following a diagnosis of stage 1 colon cancer, whereas an 81 year old woman with three or more comorbidities has a life expectancy of only 4.9 years.5

Treatment decisions in older adults with varying comorbidities are frequently even more complex than those involving testing because they require an individualized assessment of outcomes associated with multiple options. To illustrate the importance of basing treatment decisions on individualized patient data in clinical practice, we present the example of anticoagulation in nonvalvular atrial fibrillation (NVAF). This is a clinical scenario in which benefits (i.e. stroke risk reduction) and harms (i.e. increased bleeding risk) of treatment vary considerably, such that the benefit to harm ratio can reverse according to the patients’ specific comorbid conditions. Significant advances have been made to improve decision making at the individual patient level, including the use of prediction rules, risk calculators, and decision aids.6 However, despite these advances, our ability to provide individualized outcome assessments based on currently available data remains limited. Because the gaps in knowledge regarding outcomes can result in under- and over-estimates of both benefits and harms, it is expected that treatment decisions would differ if based on more accurate outcome data.

Variability in expected outcomes

The decision regarding therapy for reducing stroke risk in NVAF involves trade-offs among bleeding, stroke, and inconveniences associated with three options: warfarin, aspirin and no therapy. A recent meta-analysis of 29 RCTs (mean age of 71 years, 35% women) found a greater absolute reduction in stroke risk and a small incremental risk of major hemorrhage associated with warfarin compared to aspirin.7 As a result, guidelines recommend the use of warfarin for patients at moderate or high risk for stroke who do not have an absolute contraindication to warfarin.8 In contrast, a number of observational studies have demonstrated that the risk of bleeding associated with warfarin is not small.9–12 In a population-based study utilizing Medicare data, the rate of bleeding resulting in hospitalization ranged from 1.9 to 12.3 per 100 patient-years.9

Several observational studies have demonstrated that both the risk of stroke and the risk of bleeding vary according to patients’ comorbidities.9–11, 13 Because of this variability in risk, there is a large range in the incremental risks and benefits associated with warfarin, aspirin and, no treatment. The table provides examples of the 5-year risks of stroke and bleed, converted from annualized outcome rates,14 associated with each option for two 70-year old men with different comorbidities. The baseline stroke risk and risk of bleeding with warfarin were based on validated risk calculators derived from observational, population-based data.9, 15 The risks of stroke with warfarin and aspirin were derived by applying the 67% and 21% reduction in stroke risk associated with these two therapies published in meta-analysis of RCT data.7 Risk calculators are not available for risk of bleeding with no treatment and with aspirin; therefore, these risks were taken from a systematic review.16

Table.

Five-year risk of stroke and bleeding associated with no treatment, aspirin, or warfarin in the treatment of NVAF according to comorbid conditions:

70 year old man with well-controlled hypertension, heart failure, diabetes mellitus, and non- ulcer-related abdominal pain:
No medication Aspirin Warfarin
Stroke 26% 21% 9%
Bleed 4% 7% 9%
70 year old man with poorly controlled hypertension, renal disease, and history of a fall:
No medication Aspirin Warfarin
Stroke 13% 10% 5%
Bleed 2% 4% 34%
Open in a new tab

As can be seen in the table, expected outcomes calculated using the best available data vary markedly according to the individual’s comorbidities. Moreover, specific comorbidities differentially affect the risk of stroke and bleed. For example, in the case of a 70 year old man with well-controlled hypertension, heart failure, diabetes mellitus, and non-ulcer related abdominal pain, the first three comorbid conditions increase his baseline risk of stroke but not bleeding, and the last comorbid condition modestly increases his risk of bleeding with aspirin but not warfarin. In contrast, for a 70 year old man with poorly controlled hypertension, renal disease, and a history of a fall, only the first comorbid condition increases his baseline stroke risk while all the comorbidities increase his bleeding risk.

Despite the availability of a large number of studies and formal meta-analyses, the best available calculations of individualized benefit and harm in NVAF still depend upon a number of assumptions because of the absence of data needed to provide individualized estimates. The following paragraphs outline how the absence of these data affects decision making at the individual patient level.

Gaps in currently available data

Duration of follow-up

The decision on whether to initiate aspirin or coumadin is based on the trade-off between the expected reduction in stroke risk and increase in bleeding risk over the long term, yet RCTs have an average follow-up of less than 1.5 years per patient.7 The absolute number of outcomes over this short time frame is small and fails to reflect the larger absolute difference in outcomes associated with therapy over the long term that may be more meaningful to patients. In order to generate long-term outcome data, outcome rates must be extrapolated from person-year rates. This conversion assumes that rates remain constant over time. This assumption, while widely applied, may not be accurate. On one hand, the risks of bleeding with coumadin have been shown to be highest in the months following initiation of therapy and to subsequently decrease over time.17 Therefore, long-term bleeding rates calculated from short-term studies may overestimate risk of bleeding. On the other hand, a patient in whom comorbid conditions accumulate over time might be expected to be at even higher risk of bleed.

Baseline risk of adverse events

The recognition of the need to quantify the risk of adverse events resulting from treatment18 has generally not been accompanied by recognition of the need to quantify these same outcomes without treatment. The lack of data regarding baseline risks of adverse events supports the assumption that these rates are negligible, as evidenced, for example, by the presentation of bleeding risk as zero in an atrial fibrillation decision aid.19 Failure to present baseline rates of adverse outcomes, however, leads to overestimates of the harm associated with therapy. For example, if baseline bleeding risks were not included in the table, the incremental risks associated with aspirin and warfarin would appear larger. Unfortunately, the data available for calculating baseline bleeding risk are limited to a paper which adjusted for only the most basic risk factors,16 using relative risks pooled from heterogeneous studies.20

Choice of outcomes

Risks and harms of different treatment options are generally presented in terms of disease-specific outcomes. Yet, a number of studies have demonstrated that the outcomes of greatest importance to individuals are the sequelae of these diseases. In NVAF, what may matter most to patients is not the risk of stroke or bleed, but rather the risks of functional and cognitive disability.21–23 Functional outcomes for stroke can be extrapolated from other stroke cohorts, but there are no studies examining these outcomes among patients with NVAF, and there are very limited data available describing what happens to patients surviving a major bleed.17 One study demonstrated a 30-day mortality rate after major hemorrhage that exceeded the rate of intracranial bleeding10 suggesting that a proportion of extracranial bleeds were fatal, but no population-based study has examined survival and functional outcomes associated with different subtypes of bleeding.

Effects of treatment of comorbid conditions

Coronary artery disease is a prevalent comorbid condition in patients with NVAF. Both aspirin and coumadin are frequently recommended for patients with these two conditions.8 This combination does not improve stroke prevention24 and may not provide added protection against myocardial infarction.24 Combination therapy, does however, increase overall bleeding risk.24 In addition, despite the lack of data delineating the incremental benefit or risks of triple therapy, there is a rising use of prolonged dual antiplatelet therapy plus coumadin among patients with NVAF and coronary artery disease who undergo percutaneous coronary interventions.25 Even fewer data are available regarding the incremental harms and benefits for this treatment regimen.26

Categorization of risk

Even when data are available to calculate individualized outcomes, risk categories (e.g. low versus high stroke risk),19 rather than absolute risks, are frequently used to simplify the calculation and presentation of the outcomes. These categories are, however, defined by arbitrary cut-off points. For example, one set of guidelines recommends coumadin for any patient with a single stroke risk factor included in the CHADS2 risk index and aspirin for patients without these factors. A patient with any one of these factors has a risk of 2.8 per 100 patient-years of having a stroke. A patient without any of these factors has a risk of stroke of 1.9 per 100 patient-years. The absolute difference between these rates is not large. Because patients vary in the amount of risk they are willing to accept to prevent a stroke, relying on the same “cut-off” for all patients does not respect individual patient values. Moreover, several studies have shown that patients’ values often differ from those of physicians,27–30 and it is probable that many patients would disagree with the population-based cut-offs chosen by investigators.

Addressing the gaps in currently available data

The illustration of treatment decision making in NVAF demonstrates that, despite a wealth of clinical trials and epidemiologic studies of atrial fibrillation, substantial gaps remain in our ability to determine patient-specific outcomes. These gaps are not specific to NVAF but rather are indicative of limitations in current approaches to the collection of outcome data. It has previously been argued that obtaining individualized assessments of the risks and benefits related to available options requires that RCT data be supplemented by comprehensive observational data.18 This effort, however, needs to go beyond the call for the use of observational data to identify adverse events.18 Observational data are also required to generate expected rates of adverse outcomes without treatment and estimates of treatment-related outcomes for patients with varying comorbidities over meaningful time periods. Databases should include a catalogue of a broad set of health outcomes, including sequelae of disease-specific physical, cognitive, and psychosocial outcomes among representative patient populations possessing a wide range of comorbid conditions.

Obtaining these data will require considerable expansion of current cohort studies. Comprehensive systematic assessments across large and diverse patient populations are now possible given the use of unified electronic medical record (EMR) systems.31 The Veterans Aging Cohort Study demonstrates the feasibility of combining clinical, laboratory, and pharmacy data to facilitate the development of computerized individualized decision support systems.32 The single study in NVAF examining the functional sequelae of bleeds was conducted within a cohort of persons receiving their care within Kaiser Permanente of Northern California.17 Quality measures, mandating the use of functional assessment questionnaires, are an example of the potential means by which functional status and mental health can be tracked over time.33

Informed decision making depends upon the EMR not only for its “inputs,” but also for its “outputs.” The EMR allows for the possibility of capturing patients’ relevant risk factors and calculating updated individualized outcome estimates which 1) eliminates the presentation of categorical, rather than continuous, risk estimates, 2) allows for a re-examination of outcomes as the patients’ risk profile changes and 3) allows relevant patient information to be available in clinical offices in real time, so that it can be more fully utilized in decision making.

Conclusion

Different comorbidity profiles can have clinically significant effects on the expected harms and benefits related to available treatment options. This variability in outcomes highlights the potentially harmful consequences of utilizing average data to inform medical decisions and provide a strong argument that decision making must be based on the expected risks and benefits for each individual patient. However, enabling clinicians to make medical decisions based on individualized expected outcomes will require substantive increases in the efforts and resources allocated towards the collection and dissemination of data for patients with varying comorbidities.

Acknowledgments

Dr. Fraenkel is supported by K23 AR048826. Dr. Fried is supported by K24 AG28443. The project described was supported by the Donaghue Foundation Practical Benefit Initiative DF #06-205.

References

  • 1.Kattan M, Vickers A. Incorporating predictions of individual patient risk in clinical trials. Urol Oncol. 2004;22:348–52. doi: 10.1016/j.urolonc.2004.04.012. [DOI] [PubMed] [Google Scholar]
  • 2.Kent DM, Hayward RA. Limitations of applying summary results of clinical trials to individual patients. JAMA. 2007;298:1209–12. doi: 10.1001/jama.298.10.1209. [DOI] [PubMed] [Google Scholar]
  • 3.Tinetti ME, Bogardus ST, Jr, Agostini JV. Potential pitfalls of disease-specific guidelines for patients with multiple conditions. N Engl J Med. 2004;351:2870–4. doi: 10.1056/NEJMsb042458. [DOI] [PubMed] [Google Scholar]
  • 4.Huang ES, Zhang Q, Gandra N, Chin MH, Meltzer DO. The effect of comorbid illness and functional status on the expected benefits of intensive glucose control in older patients with type 2 diabetes: a decision analysis. Ann Intern Med. 2008;149:11–9. doi: 10.7326/0003-4819-149-1-200807010-00005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Gross CP, McAvay GJ, Krumholz HM, Paltiel AD, Bhasin D, Tinetti ME. The effect of age and chronic illness on life expectancy after a diagnosis of colorectal cancer: implications for screening. Ann Intern Med. 2006;145:646–53. doi: 10.7326/0003-4819-145-9-200611070-00006. [DOI] [PubMed] [Google Scholar]
  • 6.Wyatt JC, Altman DG. Commentary: Prognostic models: clinically useful or quickly forgotten? [Accessed January 2010.];BMJ. 1995 311:1539–1541. http://www.bmj.com/cgi/content/full/311/7019/1539.
  • 7.Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med. 2007;146:857–67. doi: 10.7326/0003-4819-146-12-200706190-00007. [DOI] [PubMed] [Google Scholar]
  • 8.Fuster V, Ryden LE, Asinger RW, et al. ACC/AHA/ESC guidelines for the management of patients with atrial fibrillation: executive summary. A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the European Society of Cardiology Committee for Practice Guidelines and Policy Conferences. J Am Coll Cardiol. 2001;38:1231–66. doi: 10.1016/s0735-1097(01)01587-x. [DOI] [PubMed] [Google Scholar]
  • 9.Gage BF, Yan Y, Milligan PE, et al. Clinical classification schemes for predicting hemorrhage: results from the National Registry of Atrial Fibrillation (NRAF) Am Heart J. 2006;151:713–9. doi: 10.1016/j.ahj.2005.04.017. [DOI] [PubMed] [Google Scholar]
  • 10.Shireman TI, Mahnken JD, Howard PA, Kresowik TF, Hou Q, Ellerbeck EF. Development of a contemporary bleeding risk model for elderly warfarin recipients. Chest. 2006;130:1390–6. doi: 10.1378/chest.130.5.1390. [DOI] [PubMed] [Google Scholar]
  • 11.Beyth RJ, Quinn LM, Landefeld CS. Prospective evaluation of an index for predicting the risk of major bleeding in outpatients treated with warfarin. Am J Med. 1998;105:91–9. doi: 10.1016/s0002-9343(98)00198-3. [DOI] [PubMed] [Google Scholar]
  • 12.Wysowski DK, Nourjah P, Swartz L. Bleeding complications with warfarin use: a prevalent adverse effect resulting in regulatory action. Arch Intern Med. 2007;167:1414–9. doi: 10.1001/archinte.167.13.1414. [DOI] [PubMed] [Google Scholar]
  • 13.Wang TJ, Massaro JM, Levy D, et al. A risk score for predicting stroke or death in individuals with new-onset atrial fibrillation in the community: the Framingham Heart Study. JAMA. 2003;290:1049–56. doi: 10.1001/jama.290.8.1049. [DOI] [PubMed] [Google Scholar]
  • 14.Beck JR, Pauker SG, Gottlieb JE, Klein K, Kassirer JP. A convenient approximation of life expectancy (the “DEALE”):II. Use in medical decision-making. Am J Med. 1982;73:889–97. doi: 10.1016/0002-9343(82)90787-2. [DOI] [PubMed] [Google Scholar]
  • 15.Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285:2864–70. doi: 10.1001/jama.285.22.2864. [DOI] [PubMed] [Google Scholar]
  • 16.Hernandez-Diaz S, Rodriguez LAG. Cardioprotective aspirin users and their excess risk of upper gastrointestinal complications. [Accessed January 2010.];BMC Med. 2006 4:22. doi: 10.1186/1741-7015-4-22. http://www.biomedcentral.com/1741-7015/4/22. [DOI] [PMC free article] [PubMed]
  • 17.Fang MC, Go AS, Chang Y, et al. Death and disability from warfarin-associated intracranial and extracranial hemorrhages. Am J Med. 2007;120:700–5. doi: 10.1016/j.amjmed.2006.07.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Vandenbroucke JP, Psaty BM. Benefits and risks of drug treatments: how to combine the best evidence on benefits with the best data about adverse effects. JAMA. 2008;300:2417–9. doi: 10.1001/jama.2008.723. [DOI] [PubMed] [Google Scholar]
  • 19.Man-Son-Hing M, Laupacis A, O’Connor AM, et al. Development of a decision aid for atrial fibrillation who are considering antithrombotic therapy. J Gen Intern Med. 2000;15:723–30. doi: 10.1046/j.1525-1497.2000.90909.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Hernandez-Diaz S, Rodriguez LAG. Association between nonsteroidal anti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation: an overview of epidemiologic studies published in the 1990s. Arch Intern Med. 2000;160:2093–9. doi: 10.1001/archinte.160.14.2093. [DOI] [PubMed] [Google Scholar]
  • 21.Fried TR, McGraw S, Agostini JV, Tinetti ME. Views of older persons with multiple morbidities on competing outcomes and clinical decision-making. J Am Geriatr Soc. 2008;56:1839–44. doi: 10.1111/j.1532-5415.2008.01923.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Fried TR, Bradley EH, Towle VR, Allore H. Understanding the treatment preferences of seriously ill patients. N Engl J Med. 2002;346:1061–6. doi: 10.1056/NEJMsa012528. [DOI] [PubMed] [Google Scholar]
  • 23.Rosenfeld KE, Wenger NS, Kagawa-Singer M. End-of-life decision making: a qualitative study of elderly individuals. J Gen Intern Med. 2000;15:620–5. doi: 10.1046/j.1525-1497.2000.06289.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Gorelick PB. Combining aspirin with oral anticoagulant therapy: is this a safe and effective practice in patients with atrial fibrillation? Stroke. 2007;38:1652–4. doi: 10.1161/STROKEAHA.107.485250. [DOI] [PubMed] [Google Scholar]
  • 25.Lip GYH, Karpha M. Anticoagulant and antiplatelet therapy use in patients with atrial fibrillation undergoing percutaneous coronary intervention: the need for consensus and a management guideline. Chest. 2006;130:1823–7. doi: 10.1378/chest.130.6.1823. [DOI] [PubMed] [Google Scholar]
  • 26.Watson T, Lip GY. Combining antiplatelet drugs and oral anticoagulant therapy in atrial fibrillation: acute coronary syndromes and/or percutaneous coronary intervention/stenting revisited. Stroke. 2007;38:107–8. doi: 10.1161/STROKEAHA.107.492488. [DOI] [PubMed] [Google Scholar]
  • 27.Elstein AS, Chapman GB, Knight SJ. Patients’ values and clinical substituted judgments: the case of localized prostate cancer. Health Psychol. 2005;24(suppl):S85–S92. doi: 10.1037/0278-6133.24.4.S85. [DOI] [PubMed] [Google Scholar]
  • 28.Sawyer SM, Fardy HJ. Bridging the gap between doctors’ and patients’ expectations of asthma management. J Asthma. 2003;40:131–8. doi: 10.1081/jas-120017983. [DOI] [PubMed] [Google Scholar]
  • 29.Spoorenberg A, van Tubergen A, Landewe R, et al. Measuring disease activity in ankylosing spondylitis: patient and physician have different perspectives. Rheumatology. 2005;44:789–95. doi: 10.1093/rheumatology/keh595. [DOI] [PubMed] [Google Scholar]
  • 30.Longacre AV, Imaeda A, Garcia-Tsao G, Fraenkel L. A pilot project examining the predicted preferences of patients and physicians in the primary prophylaxis of variceal hemorrhage. Hepatology. 2008;47:169–76. doi: 10.1002/hep.21945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Altman DG, Vergouwe Y, Royston P, Moons KGM. Prognosis and prognostic research: validating a prognostic model. BMJ. 2009 May 28;338:b605. doi: 10.1136/bmj.b605. [DOI] [PubMed] [Google Scholar]
  • 32.Justice AC, Dombrowski E, Conigliaro J, et al. Veterans Aging Cohort Study (VACS): Overview and description. Med Care. 2006;44:S13–S24. doi: 10.1097/01.mlr.0000223741.02074.66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Jones N, Jones S, Miller N. The Medicare Health Outcomes Survey program: Overview, context, and near-term prospects. Health Qual Life Outcomes. 2004;2:33. doi: 10.1186/1477-7525-2-33. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES