Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The Problem and Promise of Prognosis Research

The Problem and Promise of Prognosis Research A recent article referred to diagnosis as “the other half of medicine.”1 But there is more than therapy and diagnosis to medicine. Patients and their families want to know answers to questions like “Will she live?” “Will this happen again?” and “How long until we can return home?” Prognosis—the act of foretelling the course of health or illness—has historically played a major role in medicine2 and remains relevant today. Consider, for example, the current statin guidelines, which no longer recommend therapy on the sole diagnosis of hypercholesterolemia but on a patient’s risk for future disease.3 In broad terms, the importance of prognosis in medical decision making informs recent debates about the appropriateness of interventions for precancerous breast tissues, prostate-specific antigen screening for prostate cancer, and, in our area of research, counseling and intervention surrounding extremely preterm birth.4 In this Viewpoint, we explore difficulties in conducting and interpreting prognosis research to guide clinical practice and offer suggestions to improve the approach to this important aspect of medicine. Difficulties in Conducting Prognosis Research Scientific evidence for medical therapy has improved dramatically during recent decades with elaboration of clinical trial methods and widespread education of clinicians in how to interpret and apply trial data in practice. However, the methods and interpretation of prognosis research have not improved in tandem.5,6 A primary difficulty of conducting prognosis research is that it is generally observational. That is, it involves considering many factors—such as a patient’s age, sex, social factors, genetic make-up, and comorbidities—that cannot be assigned by investigators. Importantly, prognosis research is not merely about understanding what is expected in the absence of treatment. It also provides understanding regarding the potential benefit or harm of treatment. Two patients in the same clinical trial having different baseline risks for an outcome (eg, mortality) differ in the absolute potential benefit that each may gain from the study therapy, assuming therapy provides the same relative benefit to each of them—an assumption investigators and consumers of clinical trials often make. For a therapy that decreases the risk of 5-year mortality by half (relative risk, 0.5), the potential gain is much greater for a patient with a 40% 5-year risk of death (absolute risk reduction, 20%) than one with a 5% risk (absolute risk reduction, 2.5%). Although methods to incorporate information on prognosis in the analysis of clinical trials have been proposed,7 they rely on traditional tools of observational research such as multivariable regression modeling and are subject to many of the same limitations. Difficulties in Interpreting Prognosis Research One area where both the importance and limitations of prognosis research come into focus is in the use of prognosis research to inform end-of-life decisions. In neonatology, for instance, there has long been debate about the earliest gestational ages at which extremely preterm infants should be actively resuscitated. With improvements in clinical practice and technology, the gestational age at which a good outcome following resuscitation becomes possible has decreased. The best available evidence to support whether to resuscitate an infant born at 23 or 24 weeks’ gestation consists of recent studies regarding how other infants born at these early gestations have fared. Because infants cannot be randomized to resuscitation, prognostic data to inform the decision to resuscitate are, by necessity, observational. At institutions in which newborns with certain characteristics (eg, gestational age) are not resuscitated, infants with these characteristics rarely survive. Prognosis research from such institutions will therefore show a very high rate of death for such infants. A problem arises when this research informs future decisions not to resuscitate and results in subsequent studies showing poor outcomes, creating the possibility of a self-fulfilling prognosis (Figure).8 Figure. Diagnosis, Prognosis, and Therapy View LargeDownload In clinical practice, prognosis links diagnosis and therapy by indicating what is expected. The relationship of prognosis informed by diagnosis and informing therapy is shown with solid arrows. Therapy—through treatment or prevention—may also influence both prognosis and diagnosis, as shown by the dashed arrows. Other issues affecting the interpretation of prognosis research include variation among studies in their definitions of outcomes and prognostic factors and differences in the treatment milieu to which patients are exposed. Interpreting prognosis research of uncommon conditions may also be affected by statistical uncertainty resulting from small sample sizes. Opportunities to Improve Prognosis Research Prognosis has an important role in medicine—in particular in avoiding overtreatment and undertreatment and facilitating shared decision making with patients. Yet research on prognosis may not be held to the same high standards as research about diagnosis and therapy and can be difficult to interpret and apply in practice.5,6 To improve prognosis research, issues such as self-fulfilling prognoses in end-of-life care deserve further attention. In particular, attention should be given to specifying the denominator in prognosis research relevant to the treatment decision. In the case of extremely preterm birth, for example, it is important to specify whether the infants in the denominator of the survival statistic were resuscitated.9 Prognosis research may also be improved by requiring the presentation of statistical uncertainty (eg, 95% confidence intervals) for rates of study outcomes, registration of prognosis study protocols to prespecify study methods, and the development of more robust methods for prognostic reviews and meta-analyses. We are encouraged by the recent promotion of prognosis research guidelines from organizations including the Cochrane Collaboration.10 Such guidelines, if widely adopted, may help clinicians, researchers, and journal editors in their respective roles to improve the use of prognosis research in clinical practice. Clinicians may sometimes be reluctant to make statements about a patient’s prognosis due to concerns that such pronouncements may influence the patient’s course. Concerns about giving depressing news or, conversely, giving false hope might influence whether a prognosis is painted as overly rosy or bleak, or whether one is offered at all.2 However, in situations as diverse as whether to use a statin to decrease the likelihood of cardiovascular disease to whether to resuscitate an extremely preterm infant, prognosis research matters. With recent developments in “personalized” medicine, “-omics” research (eg, genomic, epigenomic, proteomic, metabolomic), and the predictive analytics made possible by electronic medical records, both the complexity and potential of prognosis research continue to grow. It is time to improve the conduct and interpretation of research on prognosis to avoid waste and increase the value of this important aspect of clinical practice. Back to top Article Information Corresponding Author: Matthew A. Rysavy, MD, PhD, Department of Pediatrics, University of Wisconsin Hospital and Clinics, 600 Highland Ave, Madison, WI 53792 (mrysavy2@uwhealth.org). Published Online: March 14, 2016. doi:10.1001/jamapediatrics.2015.4871. Conflict of Interest Disclosures: None reported. Additional Contributions: We thank Edward Bell, MD (University of Iowa), Paul Christine, MPH (University of Michigan), and Martha Carvour, MD, PhD (University of New Mexico), for their thoughtful feedback on this work. They received no compensation. References 1. Singh H, Graber ML. Improving diagnosis in health care—the next imperative for patient safety. N Engl J Med. 2015;373(26):2493-2495.PubMedGoogle ScholarCrossref 2. Christakis NA. Death Foretold: Prophecy and Prognosis in Medical Care. Chicago, IL: University of Chicago Press; 1999. 3. Stone NJ, Robinson JG, Lichtenstein AH, et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129(25)(suppl 2):S1-S45.PubMedGoogle ScholarCrossref 4. Croft P, Altman DG, Deeks JJ, et al. The science of clinical practice: disease diagnosis or patient prognosis? evidence about “what is likely to happen” should shape clinical practice. BMC Med. 2015;13:20.PubMedGoogle ScholarCrossref 5. Hemingway H, Riley RD, Altman DG. Ten steps towards improving prognosis research. BMJ. 2009;339:b4184.PubMedGoogle ScholarCrossref 6. Hemingway H. Improving prognosis research: standards primary, secondary, and tertiary. J Clin Epidemiol. 2007;60:865-866.Google ScholarCrossref 7. Kent DM, Rothwell PM, Ioannidis JPA, Altman DG, Hayward RA. Assessing and reporting heterogeneity in treatment effects in clinical trials: a proposal. Trials. 2010;11:85.PubMedGoogle ScholarCrossref 8. Wilkinson D. The self-fulfilling prophecy in intensive care. Theor Med Bioeth. 2009;30(6):401-410.PubMedGoogle ScholarCrossref 9. Rysavy MA, Li L, Bell EF, et al; Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. Between-hospital variation in treatment and outcomes in extremely preterm infants. N Engl J Med. 2015;372(19):1801-1811.PubMedGoogle ScholarCrossref 10. Scope of our work. Cochrane Methods Prognosis. http://prognosismethods.cochrane.org/scope-our-work. Accessed February 1, 2016. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JAMA Pediatrics American Medical Association

The Problem and Promise of Prognosis Research

JAMA Pediatrics , Volume 170 (5) – May 1, 2016

Loading next page...
 
/lp/american-medical-association/the-problem-and-promise-of-prognosis-research-StGdIujrLp

References (12)

Publisher
American Medical Association
Copyright
Copyright © 2016 American Medical Association. All Rights Reserved.
ISSN
2168-6203
eISSN
2168-6211
DOI
10.1001/jamapediatrics.2015.4871
Publisher site
See Article on Publisher Site

Abstract

A recent article referred to diagnosis as “the other half of medicine.”1 But there is more than therapy and diagnosis to medicine. Patients and their families want to know answers to questions like “Will she live?” “Will this happen again?” and “How long until we can return home?” Prognosis—the act of foretelling the course of health or illness—has historically played a major role in medicine2 and remains relevant today. Consider, for example, the current statin guidelines, which no longer recommend therapy on the sole diagnosis of hypercholesterolemia but on a patient’s risk for future disease.3 In broad terms, the importance of prognosis in medical decision making informs recent debates about the appropriateness of interventions for precancerous breast tissues, prostate-specific antigen screening for prostate cancer, and, in our area of research, counseling and intervention surrounding extremely preterm birth.4 In this Viewpoint, we explore difficulties in conducting and interpreting prognosis research to guide clinical practice and offer suggestions to improve the approach to this important aspect of medicine. Difficulties in Conducting Prognosis Research Scientific evidence for medical therapy has improved dramatically during recent decades with elaboration of clinical trial methods and widespread education of clinicians in how to interpret and apply trial data in practice. However, the methods and interpretation of prognosis research have not improved in tandem.5,6 A primary difficulty of conducting prognosis research is that it is generally observational. That is, it involves considering many factors—such as a patient’s age, sex, social factors, genetic make-up, and comorbidities—that cannot be assigned by investigators. Importantly, prognosis research is not merely about understanding what is expected in the absence of treatment. It also provides understanding regarding the potential benefit or harm of treatment. Two patients in the same clinical trial having different baseline risks for an outcome (eg, mortality) differ in the absolute potential benefit that each may gain from the study therapy, assuming therapy provides the same relative benefit to each of them—an assumption investigators and consumers of clinical trials often make. For a therapy that decreases the risk of 5-year mortality by half (relative risk, 0.5), the potential gain is much greater for a patient with a 40% 5-year risk of death (absolute risk reduction, 20%) than one with a 5% risk (absolute risk reduction, 2.5%). Although methods to incorporate information on prognosis in the analysis of clinical trials have been proposed,7 they rely on traditional tools of observational research such as multivariable regression modeling and are subject to many of the same limitations. Difficulties in Interpreting Prognosis Research One area where both the importance and limitations of prognosis research come into focus is in the use of prognosis research to inform end-of-life decisions. In neonatology, for instance, there has long been debate about the earliest gestational ages at which extremely preterm infants should be actively resuscitated. With improvements in clinical practice and technology, the gestational age at which a good outcome following resuscitation becomes possible has decreased. The best available evidence to support whether to resuscitate an infant born at 23 or 24 weeks’ gestation consists of recent studies regarding how other infants born at these early gestations have fared. Because infants cannot be randomized to resuscitation, prognostic data to inform the decision to resuscitate are, by necessity, observational. At institutions in which newborns with certain characteristics (eg, gestational age) are not resuscitated, infants with these characteristics rarely survive. Prognosis research from such institutions will therefore show a very high rate of death for such infants. A problem arises when this research informs future decisions not to resuscitate and results in subsequent studies showing poor outcomes, creating the possibility of a self-fulfilling prognosis (Figure).8 Figure. Diagnosis, Prognosis, and Therapy View LargeDownload In clinical practice, prognosis links diagnosis and therapy by indicating what is expected. The relationship of prognosis informed by diagnosis and informing therapy is shown with solid arrows. Therapy—through treatment or prevention—may also influence both prognosis and diagnosis, as shown by the dashed arrows. Other issues affecting the interpretation of prognosis research include variation among studies in their definitions of outcomes and prognostic factors and differences in the treatment milieu to which patients are exposed. Interpreting prognosis research of uncommon conditions may also be affected by statistical uncertainty resulting from small sample sizes. Opportunities to Improve Prognosis Research Prognosis has an important role in medicine—in particular in avoiding overtreatment and undertreatment and facilitating shared decision making with patients. Yet research on prognosis may not be held to the same high standards as research about diagnosis and therapy and can be difficult to interpret and apply in practice.5,6 To improve prognosis research, issues such as self-fulfilling prognoses in end-of-life care deserve further attention. In particular, attention should be given to specifying the denominator in prognosis research relevant to the treatment decision. In the case of extremely preterm birth, for example, it is important to specify whether the infants in the denominator of the survival statistic were resuscitated.9 Prognosis research may also be improved by requiring the presentation of statistical uncertainty (eg, 95% confidence intervals) for rates of study outcomes, registration of prognosis study protocols to prespecify study methods, and the development of more robust methods for prognostic reviews and meta-analyses. We are encouraged by the recent promotion of prognosis research guidelines from organizations including the Cochrane Collaboration.10 Such guidelines, if widely adopted, may help clinicians, researchers, and journal editors in their respective roles to improve the use of prognosis research in clinical practice. Clinicians may sometimes be reluctant to make statements about a patient’s prognosis due to concerns that such pronouncements may influence the patient’s course. Concerns about giving depressing news or, conversely, giving false hope might influence whether a prognosis is painted as overly rosy or bleak, or whether one is offered at all.2 However, in situations as diverse as whether to use a statin to decrease the likelihood of cardiovascular disease to whether to resuscitate an extremely preterm infant, prognosis research matters. With recent developments in “personalized” medicine, “-omics” research (eg, genomic, epigenomic, proteomic, metabolomic), and the predictive analytics made possible by electronic medical records, both the complexity and potential of prognosis research continue to grow. It is time to improve the conduct and interpretation of research on prognosis to avoid waste and increase the value of this important aspect of clinical practice. Back to top Article Information Corresponding Author: Matthew A. Rysavy, MD, PhD, Department of Pediatrics, University of Wisconsin Hospital and Clinics, 600 Highland Ave, Madison, WI 53792 (mrysavy2@uwhealth.org). Published Online: March 14, 2016. doi:10.1001/jamapediatrics.2015.4871. Conflict of Interest Disclosures: None reported. Additional Contributions: We thank Edward Bell, MD (University of Iowa), Paul Christine, MPH (University of Michigan), and Martha Carvour, MD, PhD (University of New Mexico), for their thoughtful feedback on this work. They received no compensation. References 1. Singh H, Graber ML. Improving diagnosis in health care—the next imperative for patient safety. N Engl J Med. 2015;373(26):2493-2495.PubMedGoogle ScholarCrossref 2. Christakis NA. Death Foretold: Prophecy and Prognosis in Medical Care. Chicago, IL: University of Chicago Press; 1999. 3. Stone NJ, Robinson JG, Lichtenstein AH, et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129(25)(suppl 2):S1-S45.PubMedGoogle ScholarCrossref 4. Croft P, Altman DG, Deeks JJ, et al. The science of clinical practice: disease diagnosis or patient prognosis? evidence about “what is likely to happen” should shape clinical practice. BMC Med. 2015;13:20.PubMedGoogle ScholarCrossref 5. Hemingway H, Riley RD, Altman DG. Ten steps towards improving prognosis research. BMJ. 2009;339:b4184.PubMedGoogle ScholarCrossref 6. Hemingway H. Improving prognosis research: standards primary, secondary, and tertiary. J Clin Epidemiol. 2007;60:865-866.Google ScholarCrossref 7. Kent DM, Rothwell PM, Ioannidis JPA, Altman DG, Hayward RA. Assessing and reporting heterogeneity in treatment effects in clinical trials: a proposal. Trials. 2010;11:85.PubMedGoogle ScholarCrossref 8. Wilkinson D. The self-fulfilling prophecy in intensive care. Theor Med Bioeth. 2009;30(6):401-410.PubMedGoogle ScholarCrossref 9. Rysavy MA, Li L, Bell EF, et al; Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network. Between-hospital variation in treatment and outcomes in extremely preterm infants. N Engl J Med. 2015;372(19):1801-1811.PubMedGoogle ScholarCrossref 10. Scope of our work. Cochrane Methods Prognosis. http://prognosismethods.cochrane.org/scope-our-work. Accessed February 1, 2016.

Journal

JAMA PediatricsAmerican Medical Association

Published: May 1, 2016

Keywords: confidence interval,outcomes research,gestational age,infant, premature,resuscitation,observation in research,observational studies,biomedical research,health care decision making,premature birth,statins,interpretation of findings,prognostic study,guidelines,absolute risk reduction,end-of-life care,prognostic factors

There are no references for this article.