Understanding Clinical Research7 April 2020Studies Using Randomized Trial Data to Compare Nonrandomized ExposuresCatharine B. Stack, PhD, Anne R. Meibohm, PhD, Joshua M. Liao, MD, MSc, and Eliseo Guallar, MD, DrPHCatharine B. Stack, PhDAmerican College of Physicians, Philadelphia, Pennsylvania (C.B.S., A.R.M.)Search for more papers by this author, Anne R. Meibohm, PhDAmerican College of Physicians, Philadelphia, Pennsylvania (C.B.S., A.R.M.)Search for more papers by this author, Joshua M. Liao, MD, MScUniversity of Washington School of Medicine, Seattle, Washington (J.M.L.)Search for more papers by this author, and Eliseo Guallar, MD, DrPHJohns Hopkins Bloomberg School of Public Health and Johns Hopkins School of Medicine, Baltimore, Maryland (E.G.)Search for more papers by this authorAuthor, Article, and Disclosure Informationhttps://doi.org/10.7326/M20-0071 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail Researchers frequently use data collected in randomized controlled trials to target questions that are beyond the scope of the original trial. For example, trial data may be used to explore the effects on outcomes of an exposure collected during the trial that was not the randomized intervention. Although such data originate from a randomized trial, the analyses do not have the protection from bias that randomization provides. Successful randomization tends to balance observed and unobserved characteristics between study groups. Comparisons of nonrandomized exposures, even when done using trial data, can be biased and require analytic approaches commonly used for observational ...References1. Hernán MA, Hernández-Díaz S, Robins JM. Randomized trials analyzed as observational studies. Ann Intern Med. 2013;159:560-2. [PMID: 24018844]. doi:10.7326/0003-4819-159-8-201310150-00709 LinkGoogle Scholar2. Rigotti NA, Tindle HA, Regan S, et al. A post-discharge smoking-cessation intervention for hospital patients: Helping HAND 2 randomized clinical trial. Am J Prev Med. 2016;51:597-608. [PMID: 27647060] doi:10.1016/j.amepre.2016.04.005 CrossrefMedlineGoogle Scholar3. Rigotti NA, Chang Y, Tindle HA, et al. Association of e-cigarette use with smoking cessation among smokers who plan to quit after a hospitalization. A prospective study. Ann Intern Med. 2018;168:613-20. [PMID: 29582077]. doi:10.7326/M17-2048 LinkGoogle Scholar4. Porta M. A Dictionary of Epidemiology. 6th ed. New York: Oxford Univ Pr; 2014. Google Scholar5. Ananth CV, Lavery JA. Biases in secondary analyses of randomized trials: recognition, correction, and implications. BJOG. 2016;123:1056-9. doi:10.1111/1471-0528.13732 CrossrefMedlineGoogle Scholar6. Rubin DB, Waterman RP. Estimating the causal effects of marketing interventions using propensity score methodology. Stat Sci. 2006;21:206-22. doi:10.1214/088342306000000259 CrossrefGoogle Scholar7. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25:1-21. [PMID: 20871802] CrossrefMedlineGoogle Scholar8. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med. 2017;167:268-74. [PMID: 28693043]. doi:10.7326/M16-2607 LinkGoogle Scholar9. Howard G, Howard VJ. Observational epidemiology within randomized clinical trials: getting a lot for (almost) nothing. Prog Cardiovasc Dis. 2012;54:367-71. [PMID: 22226006] doi:10.1016/j.pcad.2011.08.003 CrossrefMedlineGoogle Scholar Author, Article, and Disclosure InformationAffiliations: American College of Physicians, Philadelphia, Pennsylvania (C.B.S., A.R.M.)University of Washington School of Medicine, Seattle, Washington (J.M.L.)Johns Hopkins Bloomberg School of Public Health and Johns Hopkins School of Medicine, Baltimore, Maryland (E.G.)Disclosures: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M20-0071.Corresponding Author: Anne R. Meibohm, PhD, American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106; e-mail, [email protected]org.Current Author Addresses: Dr. Stack: Jazz Pharmaceuticals, 2005 Market Street, Suite 2100, Philadelphia, PA 19103.Dr. Meibohm: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106.Dr. Liao: University of Washington School of Medicine, 1959 NE Pacific Street, Seattle, WA 98115.Dr. Guallar: Welch Center for Prevention, Epidemiology and Clinical Research, Johns Hopkins Bloomberg School of Public Health, Room 2-645, Baltimore, MD 21205.Author Contributions: Conception and design: C.B. Stack, A.R. Meibohm, J.M. Liao.Drafting of the article: C.B. Stack, A.R. Meibohm, J.M. Liao.Critical revision of the article for important intellectual content: C.B. Stack, A.R. Meibohm, J.M. Liao, E. Guallar.Final approval of the article: C.B. Stack, A.R. Meibohm, J.M. Liao, E. Guallar.Statistical expertise: C.B. Stack, A.R. Meibohm, E. Guallar.This article was published at Annals.org on 17 March 2020. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetails Metrics Cited byA Narrative Review of Methods for Causal Inference and Associated Educational ResourcesEpidural Steroids for Degenerative Spondylolisthesis: Good, Bad, or Indifferent? 7 April 2020Volume 172, Issue 7Page: 492-494KeywordsCarbon monoxideClinical epidemiologyDisclosureFactor analysisNicotineObservational studiesPrevention, policy, and public healthRandomized trialsSalivaSmoking cessation ePublished: 17 March 2020 Issue Published: 7 April 2020 Copyright & PermissionsCopyright © 2020 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

Editorials15 May 2018Designed Learning: Missing Data in Clinical ResearchFREECatharine B. Stack, PhD, Trevor Butterworth, BA, MPhil, MS, and Rebecca Goldin, PhDCatharine B. Stack, PhDDeputy Editor, Statistics Annals of Internal Medicine (C.B.S.), Trevor Butterworth, BA, MPhil, MSSense About Science USA, Alexandria, Virginia (T.B.), and Rebecca Goldin, PhDGeorge Mason University, Fairfax, Virginia (R.G.)Author, Article, and Disclosure Informationhttps://doi.org/10.7326/M18-0534 SectionsAboutVisual AbstractPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail Countless articles and textbooks have been written on the topic of missing data. In 2010, the National Research Council published recommendations for the prevention and treatment of missing data in clinical trials, which were developed by a panel of experts assembled by the U.S. Food and Drug Administration (1). However, reviews on the topic provide evidence that learning is slow and the effect of these written resources on published articles is hardly noticeable (2, 3). Many factors contribute to this problem. One solution could involve creating new mechanisms to deliver the information directly to the communities of scientists at the front lines. In our continued effort to explain analytic methods to clinical researchers in new ways (4, 5), we joined forces with Sense About Science USA to communicate concepts of missing data through graphic design. Sense About Science USA enlisted the data visualization expertise of Accurat. The product, "Missing Data" (http://labs.annals.org/missingdata), is an early result of our collaboration."Missing Data" combines illustrations of key concepts (provided in the left pane of the screen) with explanatory text (in the right pane). Material updates as users move down the page, and users can also jump to specific sections by clicking on the outline in the center of the page. White vertical lines denote observed data items, and red forward slashes represent missing data items. This interactive Web site transitions from instructive scenarios where researchers commonly encounter missing data to descriptions of statistical concepts, such as missing completely at random, missing at random, and missing not at random, in order to set a foundation for understanding proper analytic methods. Two case studies based on published Annals articles allow users to see how this knowledge is applied in practice (6, 7). The module explores how researchers might go wrong and explains why a simple fix will not work.Missing data are ubiquitous in clinical research, and extracting valid conclusions from incomplete data sets requires thought and knowledge. We hope this interactive guide will help clinical researchers to better understand their missing data so that they can minimize lost data when possible, handle it more optimally within analyses, carry out meaningful sensitivity analysis, and involve statisticians when needed. We also hope this module stimulates new approaches to communicating statistical concepts to researchers.References1. National Research Council. The Prevention and Treatment of Missing Data in Clinical Trials. Washington, DC: National Academies Pr; 2010. doi:10.17226/12955 CrossrefGoogle Scholar2. Fielding S, Ogbuagu A, Sivasubramaniam S, MacLennan G, Ramsay CR. Reporting and dealing with missing quality of life data in RCTs: has the picture changed in the last decade? Qual Life Res. 2016;25:2977-83. [PMID: 27650288] CrossrefMedlineGoogle Scholar3. Bell ML, Fiero M, Horton NJ, Hsu CH. Handling missing data in RCTs; a review of the top medical journals. BMC Med Res Methodol. 2014;14:118. [PMID: 25407057] doi:10.1186/1471-2288-14-118 CrossrefMedlineGoogle Scholar4. Liao JM, Stack CB. Annals Understanding Clinical Research: implications of missing data due to dropout. Ann Intern Med. 2017;166:596-8. [PMID: 28241264]. doi:10.7326/M17-0195 LinkGoogle Scholar5. Talluri R, Ranadive R, Rao JK, Laine C, Griswold M. Bringing data to life: interactive visualizations of complex data. Ann Intern Med. 2017;167:820-1. [PMID: 29132157]. doi:10.7326/M17-2903 LinkGoogle Scholar6. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91. [PMID: 25222385]. doi:10.7326/M14-0006 LinkGoogle Scholar7. Tomasson G, Peloquin C, Mohammad A, Love TJ, Zhang Y, Choi HK, et al. Risk for cardiovascular disease early and late after a diagnosis of giant-cell arteritis: a cohort study. Ann Intern Med. 2014;160:73-80. [PMID: 24592492] LinkGoogle Scholar Comments0 CommentsSign In to Submit A Comment Author, Article, and Disclosure InformationAffiliations: Deputy Editor, Statistics Annals of Internal Medicine (C.B.S.)Sense About Science USA, Alexandria, Virginia (T.B.)George Mason University, Fairfax, Virginia (R.G.)Disclosures: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M18-0534.Corresponding Author: Catharine B. Stack, PhD, American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106-1572; e-mail, [email protected]acponline.org.Current Author Addresses: Dr. Stack: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106-1572.Mr. Butterworth: Sense About Science USA, 732 North Washington Street, Alexandria, VA 22314.Dr. Goldin: Department of Mathematical Sciences, George Mason University, 4400 University Drive, Fairfax, VA 22030.This article was published at Annals.org on 10 April 2018. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetails Metrics Cited bySLE clinical trials: impact of missing data on estimating treatment effects 15 May 2018Volume 168, Issue 10Page: 744KeywordsClinical trialsDisclosureFood and Drug AdministrationScientistsStatistical data ePublished: 10 April 2018 Issue Published: 15 May 2018 Copyright & PermissionsCopyright © 2018 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

Editorials16 October 2018Inappropriate Statistical Analysis and Reporting in Medical Research: Perverse Incentives and Institutional SolutionsA. Russell Localio, PhD, Catharine B. Stack, PhD, Anne R. Meibohm, PhD, Eric A. Ross, PhD, Eliseo Guallar, MD, DrPH, John B. Wong, MD, John E. Cornell, PhD, Michael E. Griswold, PhD, and Steven N. Goodman, MD, MHS, PhDA. Russell Localio, PhDUniversity of Pennsylvania, Philadelphia, Pennsylvania (A.R.L.)Search for more papers by this author, Catharine B. Stack, PhDAmerican College of Physicians, Philadelphia, Pennsylvania (C.B.S., A.R.M.)Search for more papers by this author, Anne R. Meibohm, PhDAmerican College of Physicians, Philadelphia, Pennsylvania (C.B.S., A.R.M.)Search for more papers by this author, Eric A. Ross, PhDFox Chase Cancer Center, Philadelphia, Pennsylvania (E.A.R.)Search for more papers by this author, Eliseo Guallar, MD, DrPHJohns Hopkins University, Baltimore, Maryland (E.G.)Search for more papers by this author, John B. Wong, MDTufts University School of Medicine, Boston, Massachusetts (J.B.W.)Search for more papers by this author, John E. Cornell, PhDUniversity of Texas Health Science Center, San Antonio, Texas (J.E.C.)Search for more papers by this author, Michael E. Griswold, PhDUniversity of Mississippi Medical Center, Jackson, Mississippi (M.E.G.)Search for more papers by this author, and Steven N. Goodman, MD, MHS, PhDStanford University School of Medicine, Stanford, California (S.N.G.)Search for more papers by this authorAuthor, Article, and Disclosure Informationhttps://doi.org/10.7326/M18-2516 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail Wang and colleagues (1) present a sobering report of a national survey of nearly 400 consulting statisticians about requests from investigators to engage in inappropriate statistical practices. Framed as an exploration of bioethical issues, the report implicitly adopts Doug Altman's mantra: “Misuse of statistics is unethical” (2). Although the survey did not ask statisticians whether they fulfilled these requests, the inappropriate methods described in this report are still used in the published literature, and thus contribute to the problem of nonreproducible research.Practices like these are extraordinarily difficult to detect in published work; identification takes either unusual transparency or a ...References1. Wang MQ, Yan AF, Katz RV. Researcher requests for inappropriate analysis and reporting: a U.S. national survey of consulting biostatisticians. Ann Intern Med. 2018;169:554-8. doi:10.7326/M18-1230 LinkGoogle Scholar2. Altman DG. Statistics and ethics in medical research. Misuse of statistics is unethical. Br Med J. 1980;281:1182-4. [PMID: 7427629] CrossrefMedlineGoogle Scholar3. Baker M. 1,500 scientists lift the lid on reproducibility. Nature. 2016;533:452-4. [PMID: 27225100] doi:10.1038/533452a CrossrefMedlineGoogle Scholar4. Rubin DB. For objective causal inference, design trumps analysis. Annals of Applied Statistics. 2008;2:808-40. CrossrefGoogle Scholar5. Rosenbaum PR. Design sensitivity and efficiency in observational studies. Journal of the American Statistical Association. 2010;105:692-702. CrossrefGoogle Scholar6. Rubin DB. The design versus the analysis of observational studies for causal effects: parallels with the design of randomized trials. Stat Med. 2007;26:20-36. [PMID: 17072897] CrossrefMedlineGoogle Scholar7. National Academies of Sciences, Engineering, and Medicine. Fostering Integrity in Research. Washington, DC: National Academies Pr; 2017. Google Scholar8. Patient-Centered Outcomes Research Institute (PCORI) Methodology Committee. PCORI Methodology Report. July 2017. Accessed at www.pcori.org/sites/default/files/PCORI-Methodology-Report.pdf on 12 September 2018. Google Scholar9. National Institutes of Health. Clinical e-Protocol Writing Tool. Accessed at https://e-protocol.od.nih.gov/#/home on 13 July 2018. Google Scholar10. Moher D, Naudet F, Cristea IA, Miedema F, Ioannidis JPA, Goodman SN. Assessing scientists for hiring, promotion, and tenure. PLoS Biol. 2018;16:e2004089. [PMID: 29596415] doi:10.1371/journal.pbio.2004089 CrossrefMedlineGoogle Scholar Author, Article, and Disclosure InformationAffiliations: University of Pennsylvania, Philadelphia, Pennsylvania (A.R.L.)American College of Physicians, Philadelphia, Pennsylvania (C.B.S., A.R.M.)Fox Chase Cancer Center, Philadelphia, Pennsylvania (E.A.R.)Johns Hopkins University, Baltimore, Maryland (E.G.)Tufts University School of Medicine, Boston, Massachusetts (J.B.W.)University of Texas Health Science Center, San Antonio, Texas (J.E.C.)University of Mississippi Medical Center, Jackson, Mississippi (M.E.G.)Stanford University School of Medicine, Stanford, California (S.N.G.)Disclosures: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M18-2516.Corresponding Author: A. Russell Localio, PhD, Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania, 617 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104; e-mail, [email protected]upenn.edu.Current Author Addresses: Dr. Localio: Department of Biostatistics, Epidemiology and Informatics, Perelman School of Medicine, University of Pennsylvania, 617 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104.Drs. Stack and Meibohm: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106.Dr. Ross: Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, PA 19111.Dr. Guallar: Welch Center for Prevention, Epidemiology, and Clinical Research, Johns Hopkins Medical Institutions, Room 2-645, Baltimore, MD 21205.Dr. Wong: Tufts Medical Center, 800 Washington Street, #302, Boston, MA 02111.Dr. Cornell: 14546 Indian Woods, San Antonio, TX 78249.Dr. Griswold: Department of Data Science, John D. Bower School of Population 08alth, University of Mississippi Medical Center, 2500 North State Street, Jackson, MS 39216.Dr. Goodman: Stanford University School of Medicine, 259 Campus Drive, T265 Redwood Building/Health Research and Policy, Stanford, CA 94305.This article was published at Annals.org on 9 October 2018. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetailsSee AlsoResearcher Requests for Inappropriate Analysis and Reporting: A U.S. Survey of Consulting Biostatisticians Min Qi Wang , Alice F. Yan , and Ralph V. Katz Metrics Cited byLes clefs d’une publication réussie dans les Eur Ann Otorhinolaryngol Head Neck Dis : analyse STROBE de la relecture des articles scientifiques soumis en 2020–2021Keys for successful publication in Eur Ann Otorhinolaryngol Head Neck Dis: A STROBE analysis of peer reviews of articles submitted in 2020–2021Collaborative biostatistics and epidemiology in academic medical centres: A survey to assess relationships with health researchers and ethical implicationsMaîtriser les statistiques descriptives utilisées en otorhinolaryngologieMastering the descriptive statistics used in otorhinolaryngologyComprehensive review of statistical methods for analysing patient-reported outcomes (PROs) used as primary outcomes in randomised controlled trials (RCTs) published by the UK’s Health Technology Assessment (HTA) journal (1997–2020)Détente: A Practical Understanding of P values and Bayesian Posterior ProbabilitiesLes statistiques des articles scientifiques publiés dans les European Annals of Otorhinolaryngology Head & Neck DiseasesStatistics in scientific articles published in the European Annals of Otorhinolaryngology Head & Neck DiseasesInaccurate Use of the Upper Extremity Fugl-Meyer Negatively Affects Upper Extremity Rehabilitation Trial Design: Findings From the ICARE Randomized Controlled TrialEnhancing Collaboration between Clinician-Researchers and Methodologists in Clinical ResearchShortcoming of Visual Interpretation of Cardiotocography: A Comparative Study with Automated Method and Established Guideline Using Statistical Analysis« Suggestif » et « significatif » sont dans un bateau…“Suggestive” and “Significant”: You can’t always get what you want…The Importance of Predefined Rules and Prespecified Statistical Analyses 16 October 2018Volume 169, Issue 8Page: 577-578KeywordsBiostatisticsClinical epidemiologyDisclosureEpidemiologyMotivationReproducibilityResearch and reporting methodsResearch designResearch reporting guidelinesStatistical data ePublished: 9 October 2018 Issue Published: 16 October 2018 Copyright & PermissionsCopyright © 2018 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

Special Articles2 October 2018Annals Understanding Clinical Research: Interpreting Results With Large P ValuesJoshua M. Liao, MD, MSc, Catharine B. Stack, PhD, and Steven Goodman, MD, MHS, PhDJoshua M. Liao, MD, MScUniversity of Washington, Seattle, Washington (J.M.L.)Search for more papers by this author, Catharine B. Stack, PhDAmerican College of Physicians, Philadelphia, Pennsylvania (C.B.S.)Search for more papers by this author, and Steven Goodman, MD, MHS, PhDStanford University School of Medicine, Stanford, California (S.G.)Search for more papers by this authorAuthor, Article, and Disclosure Informationhttps://doi.org/10.7326/M18-2003 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail Reliance on dichotomous or bright-line interpretations of study results (for example, a treatment does or does not work) based solely on P values is widely recognized as a suboptimal research practice, but it remains the dominant method of study interpretation in biomedicine and many other fields. Recognizing this, the American Statistical Association issued a policy statement on statistical significance that attempted to discourage this practice and dispel the notion that any “single index should substitute for scientific reasoning” (1).Treating all results that are not statistically significant as “negative” is a damaging misuse of P values. Doing so confuses 2 ...References1. Wasserstein RL, Lazar NA. The ASA's statement on p-values: context, process, and purpose. Am Stat. 2016;70:129-33. doi:10.1080/00031305.2016.1154108 CrossrefGoogle Scholar2. Altman DG, Bland JM. Absence of evidence is not evidence of absence. BMJ. 1995;311:485. [PMID: 7647644] CrossrefMedlineGoogle Scholar3. Alderson P. Absence of evidence is not evidence of absence [Editorial]. BMJ. 2004;328:476-7. [PMID: 14988165] CrossrefMedlineGoogle Scholar4. Goodman SN. Toward evidence-based medical statistics. 1: The P value fallacy. Ann Intern Med. 1999;130:995-1004. [PMID: 10383371]. doi:10.7326/0003-4819-130-12-199906150-00008 LinkGoogle Scholar5. Goodman SN, Berlin JA. The use of predicted confidence intervals when planning experiments and the misuse of power when interpreting results. Ann Intern Med. 1994;121:200-6. [PMID: 8017747]. doi:10.7326/0003-4819-121-3-199408010-00008 LinkGoogle Scholar6. Callahan CM, Boustani MA, Schmid AA, LaMantia MA, Austrom MG, Miller DK, et al. Targeting functional decline in Alzheimer disease: a randomized trial. Ann Intern Med. 2017;166:164-71. [PMID: 27893087]. doi:10.7326/M16-0830 LinkGoogle Scholar Author, Article, and Disclosure InformationAffiliations: University of Washington, Seattle, Washington (J.M.L.)American College of Physicians, Philadelphia, Pennsylvania (C.B.S.)Stanford University School of Medicine, Stanford, California (S.G.)Disclosures: Authors have disclosed no conflicts of interest. Forms can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M18-2003.Corresponding Author: Catharine B. Stack, PhD, American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106; e-mail, [email protected]org.Current Author Addresses: Dr. Liao: University of Washington, 1959 NE Pacific Street, BB 1240, Seattle, WA 98195.Dr. Stack: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106.Dr. Goodman: Stanford University School of Medicine, 259 Campus Drive, T265 Redwood Building/HRP, Stanford, CA 94305.Author Contributions: Conception and design: J.M. Liao, C.B. Stack.Drafting of the article: J.M. Liao, C.B. Stack.Critical revision of the article for important intellectual content: J.M. Liao, C.B. Stack, S. Goodman.Final approval of the article: J.M. Liao, C.B. Stack, S. Goodman.Statistical expertise: C.B. Stack, S. Goodman.This article was published at Annals.org on 11 September 2018. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetails Metrics Cited byP-values – a chronic conundrumOptimal dosing of zoledronate to prevent fractures: improving risk‐benefit balanceEffect of Catheter Ablation vs Antiarrhythmic Drug Therapy on Mortality, Stroke, Bleeding, and Cardiac Arrest Among Patients With Atrial FibrillationVitamin D status and bone health: a possible inverse association 2 October 2018Volume 169, Issue 7Page: 485-486KeywordsActivities of daily livingAlzheimer diseaseCaregiversConfidence intervalsConflicts of interestDementiaDisclosureForecastingNursesReasoning ePublished: 11 September 2018 Issue Published: 2 October 2018 Copyright & PermissionsCopyright © 2018 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

Editorials5 June 2018Statistical Code to Support the Scientific StoryA. Russell Localio, PhD, Steven N. Goodman, MD, PhD, Anne Meibohm, PhD, John E. Cornell, PhD, Catharine B. Stack, PhD, Eric A. Ross, PhD, ScM, and Cynthia D. Mulrow, MD, MScA. Russell Localio, PhDUniversity of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania (A.R.L.), Steven N. Goodman, MD, PhDStanford University School of Medicine, Stanford, California (S.N.G.), Anne Meibohm, PhDAmerican College of Physicians, Philadelphia, Pennsylvania (A.M., C.B.S., C.D.M.), John E. Cornell, PhDUniversity of Texas Health Science Center, San Antonio, Texas (J.E.C.), Catharine B. Stack, PhDAmerican College of Physicians, Philadelphia, Pennsylvania (A.M., C.B.S., C.D.M.), Eric A. Ross, PhD, ScMFox Chase Cancer Center and Temple University, Philadelphia, Pennsylvania (E.A.R.), and Cynthia D. Mulrow, MD, MScAmerican College of Physicians, Philadelphia, Pennsylvania (A.M., C.B.S., C.D.M.)Author, Article, and Disclosure Informationhttps://doi.org/10.7326/M17-3431 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail Clearly presented and transparently reported statistical code is a sine qua non for reproducible research. More than a decade ago, Annals began asking authors to report the availability of code that supported their statistical methods (1). Before that policy adoption, our statistical editors routinely requested and reviewed code underlying analyses of papers that we eventually published. Although we never formally graded submitted code, our experiences are similar to those reported by Assel and Vickers (2).We have found that authors increasingly apply complex statistical methods to account for such factors as correlated and repeated measures, missing data, incomplete adherence, and ...References1. Laine C, Goodman SN, Griswold ME, Sox HC. Reproducible research: moving toward research the public can really trust. Ann Intern Med. 2007;146:450-3. [PMID: 17339612] LinkGoogle Scholar2. Assel M, Vickers AJ. Statistical code for clinical research papers in a high-impact specialist medical journal. Ann Intern Med. 2018;168:832-3. doi:10.7326/M17-2863 LinkGoogle Scholar3. Peng RD. Reproducible research and biostatistics [Editorial]. Biostatistics. 2009;10:405-8. [PMID: 19535325] doi:10.1093/biostatistics/kxp014 CrossrefMedlineGoogle Scholar4. Stodden V, McNutt M, Bailey DH, Deelman E, Gil Y, Hanson B, et al. Enhancing reproducibility for computational methods. Science. 2016;354:1240-1. [PMID: 27940837] CrossrefMedlineGoogle Scholar5. International Committee of Medical Journal Editors. Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals. Updated December 2017. Accessed at www.icmje.org/icmje-recommendations.pdf on 19 January 2018. Google Scholar6. Institute of Medicine. Institute of Medicine. Sharing Clinical Trial Data: Maximizing Benefits, Minimizing Risk. Washington, DC: National Academies Pr; 2015:91-133. Google Scholar7. Akacha M, Bretz F, Ruberg S. Estimands in clinical trials—broadening the perspective. Stat Med. 2017;36:5-19. [PMID: 27435045] doi:10.1002/sim.7033 CrossrefMedlineGoogle Scholar8. Kernighan BW, Plauger PJ. The Elements of Programming Style. 2nd ed. New York: McGraw-Hill; 1978. Google Scholar9. Knuth DE. Literate programming. Comput J. 1984;27:97-111. CrossrefGoogle Scholar10. Correction: rilonacept for colchicine-resistant or -intolerant familial Mediterranean fever. Ann Intern Med. 2014;160:291-2. [PMID: 24727851]. doi:10.7326/L14-5004-6 LinkGoogle Scholar Author, Article, and Disclosure InformationAffiliations: University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania (A.R.L.)Stanford University School of Medicine, Stanford, California (S.N.G.)American College of Physicians, Philadelphia, Pennsylvania (A.M., C.B.S., C.D.M.)University of Texas Health Science Center, San Antonio, Texas (J.E.C.)Fox Chase Cancer Center and Temple University, Philadelphia, Pennsylvania (E.A.R.)Disclosures: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M17-3431.Corresponding Author: A. Russell Localio, PhD, Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania, 635 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104-6021; e-mail, [email protected]upenn.edu.Current Author Addresses: Dr. Localio: Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania, 635 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104-6021.Dr. Goodman: Stanford University School of Medicine, 259 Campus Drive, T265 Redwood Building/HRP, Stanford, CA 94305.Drs. Meibohm, Stack, and Mulrow: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106.Dr. Cornell: Department of Epidemiology and Biostatistics, University of Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, TX 78229-3900.Dr. Ross: Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, PA 19111.This article was published at Annals.org on 6 February 2018. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetailsSee AlsoStatistical Code for Clinical Research Papers in a High-Impact Specialist Medical Journal Melissa Assel and Andrew J. Vickers Metrics Cited byImplementing clinical trial data sharing requires training a new generation of biomedical researchersInsufficient transparency of statistical reporting in preclinical research: a scoping reviewStatistical programming: Small mistakes, big impactsDevelopment of a model to predict the probability of incurring a complication during spine surgeryAdherence-adjustment in placebo-controlled randomized trials: An application to the candesartan in heart failure randomized trialStatistical Analysis Must Improve to Address the Reproducibility Crisis: The ACcess to Transparent Statistics (ACTS) Call to ActionTransparency in Decision Modelling: What, Why, Who and How?Overall bias and sample sizes were unchanged in ICU trials over time: a meta-epidemiological studyShould Psychology Journals Adopt Specialized Statistical Review?Transparent and systematic reporting of meta-epidemiological studiesWhat is the treatment effect of surgery compared with nonoperative treatment in patients with lumbar spinal stenosis at 1-year follow-up?Trust, But Verify 5 June 2018Volume 168, Issue 11Page: 828-829KeywordsBiostatisticsComputersDisclosureEpidemiologyFeversLibrariesResearch reporting guidelinesSoftware toolsStatistical dataStatistical methods ePublished: 6 February 2018 Issue Published: 5 June 2018 Copyright & PermissionsCopyright © 2018 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

Special Articles15 August 2017Annals Understanding Clinical Research: Evaluating the Meaning of a Summary Estimate in a Meta-analysisJohn E. Cornell, PhD, Joshua M. Liao, MD, MSc, Catharine B. Stack, PhD, MS, and Cynthia D. Mulrow, MD, MScJohn E. Cornell, PhDFrom University of Texas Health Science Center, San Antonio, Texas, and University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania.Search for more papers by this author, Joshua M. Liao, MD, MScFrom University of Texas Health Science Center, San Antonio, Texas, and University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania.Search for more papers by this author, Catharine B. Stack, PhD, MSFrom University of Texas Health Science Center, San Antonio, Texas, and University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania.Search for more papers by this author, and Cynthia D. Mulrow, MD, MScFrom University of Texas Health Science Center, San Antonio, Texas, and University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania.Search for more papers by this authorAuthor, Article, and Disclosure Informationhttps://doi.org/10.7326/M17-1454 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail In traditional meta-analyses, researchers combine data from individual studies into a summary measure to describe the benefits or harms of an intervention. The pooled estimate is calculated by first estimating the treatment effect and 95% CI for each individual study. Each study's estimated treatment effect is then weighted, usually by its precision, to reflect the amount of information the study contains relative to the others. The pooled overall treatment effect is the weighted average of the individual treatment effect estimates.This installment of the “Understanding Clinical Research” series addresses issues (Table) that readers should consider when evaluating the meaning of ...References1. Gargiulo G, Sannino A, Capodanno D, Barbanti M, Buccheri S, Perrino C, et al. Transcatheter aortic valve implantation versus surgical aortic valve replacement: a systematic review and meta-analysis. Ann Intern Med. 2016;165:334-44. [PMID: 27272666]. doi:10.7326/M16-0060 LinkGoogle Scholar2. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557-60. [PMID: 12958120] CrossrefMedlineGoogle Scholar3. Leon MB, Smith CR, Mack MJ, Makkar RR, Svensson LG, Kodali SK, et al; PARTNER 2 Investigators.. Transcatheter or surgical aortic-valve replacement in intermediate-risk patients. N Engl J Med. 2016;374:1609-20. [PMID: 27040324] doi:10.1056/NEJMoa1514616 CrossrefMedlineGoogle Scholar4. Thourani VH, Kodali S, Makkar RR, Herrmann HC, Williams M, Babaliaros V, et al. Transcatheter aortic valve replacement versus surgical valve replacement in intermediate-risk patients: a propensity score analysis. Lancet. 2016;387:2218-25. [PMID: 27053442] doi:10.1016/S0140-6736(16)30073-3 CrossrefMedlineGoogle Scholar Author, Article, and Disclosure InformationAffiliations: From University of Texas Health Science Center, San Antonio, Texas, and University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania.Acknowledgment: The authors thank the senior clinical and statistical editors from Annals of Internal Medicine for their input and review of earlier drafts of the manuscript.Disclosures: Authors have disclosed no conflicts of interest. Forms can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M17-1454.Corresponding Author: John E. Cornell, PhD, University of Texas Health Science Center at San Antonio, 7703 Merton Minter Boulevard, San Antonio, TX 78299; e-mail, [email protected]edu.Current Author Addresses: Dr. Cornell: University of Texas Health Science Center at San Antonio, 7703 Merton Minter Boulevard, San Antonio, TX 78299.Dr. Liao: Department of Medicine, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104.Dr. Stack: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106.Dr. Mulrow: University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78229.Author Contributions: Conception and design: J.E. Cornell, J.M. Liao, C.B. Stack, C.D. Mulrow.Analysis and interpretation of the data: J.E. Cornell, J.M. Liao.Drafting of the article: J.E. Cornell, J.M. Liao, C.B. Stack, C.D. Mulrow.Critical revision of the article for important intellectual content: J.E. Cornell, J.M. Liao, C.B. Stack, C.D. Mulrow.Final approval of the article: J.E. Cornell, J.M. Liao, C.B. Stack, C.D. Mulrow.Statistical expertise: J.E. Cornell.Collection and assembly of data: J.E. Cornell.This article was published at Annals.org on 8 August 2017. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetails Metrics Cited byThe Effect of Interventions on the Prevention of Parental Vaccine Refusal and Hesitancy: A Systematic Review and Meta-analysis of Randomized Controlled TrialsThe effect of music therapy interventions on fatigue in patients with hematological cancers: a systematic review and meta-analysis of randomized controlled trialsEffects of maternal folate and vitamin B12 on gestational diabetes mellitus: a dose-response meta-analysis of observational studiesThe effect of nurse‐led care on fatigue in patients with rheumatoid arthritis: A systematic review and meta‐analysis of randomised controlled studiesComparison of the effect of educational and self‐management interventions on adherence to treatment in hemodialysis patients: A systematic review and meta‐analysis of randomized controlled trialsThe effects of music ıntervention on breast milk production in breastfeeding mothers: A systematic review and meta‐analysis of randomized controlled trialsThe effects of music listening on the management of pain in primary dysmenorrhea: A randomized controlled clinical trialFuture of evidence ecosystem series: 1. Introduction Evidence synthesis ecosystem needs dramatic changeAnnals Clinical Decision Making: Weighing Evidence to Inform Clinical DecisionsJoshua P. Metlay, MD, PhD and Katrina A. Armstrong, MD, MSSeroprevalence of toxoplasma gondii infection: An umbrella review of updated systematic reviews and meta-analysesThe effect of music intervention on patients with cancer‐related pain: A systematic review and meta‐analysis of randomized controlled trialsAdvantages and Disadvantages in Clinical TrialsStandardizing music characteristics for the management of pain: A systematic review and meta-analysis of clinical trialsThe Importance of Reporting Biases in Patient Care: Can We Trust the Evidence From Either Individual Studies or Systematic Reviews?Kay Dickersin, MA, PhD and Riaz Qureshi, MScMusic is an effective intervention for the management of pain: An umbrella review 15 August 2017Volume 167, Issue 4Page: 275-277KeywordsAortic valve replacementDatabasesDisclosureInformation storage and retrievalMortalityObservational studiesOdds ratioPopulation statisticsRandomized trialsStenosis ePublished: 8 August 2017 Issue Published: 15 August 2017 Copyright & PermissionsCopyright © 2017 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

Special Articles18 April 2017Annals Understanding Clinical Research: Implications of Missing Data Due to DropoutJoshua M. Liao, MD and Catharine B. Stack, PhDJoshua M. Liao, MDFrom University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania. and Catharine B. Stack, PhDFrom University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania.Author, Article, and Disclosure Informationhttps://doi.org/10.7326/M17-0195 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail Through a series of "Understanding Clinical Research" articles, Annals aims to help nonstatisticians assess the implications of analytic methods used in research. These articles use illustrations from published studies to identify questions that are critical to understanding particular methods and appropriate interpretation of findings. This installment addresses the issue of missing data due to dropout (for the definition of this and other terms used in the article, see the Glossary)—a common problem that can affect the validity of trial findings. Several analytic methods exist to handle this problem; each has assumptions that require explicit statement and consideration (see the Table).... References1. Bronfort G, Hondras MA, Schulz CA, Evans RL, Long CR, Grimm R. Spinal manipulation and home exercise with advice for subacute and chronic back-related leg pain: a trial with adaptive allocation. Ann Intern Med. 2014;161:381-91. [PMID: 25222385]. doi:10.7326/M14-0006 LinkGoogle Scholar2. National Research Council. Division of Behavioral and Social Sciences and Education. Committee on National Statistics. Panel on Handling Missing Data in Clinical Trials. The Prevention and Treatment of Missing Data in Clinical Trials. Washington, DC: National Academies Pr; 2010. Google Scholar3. Little RJ, D'Agostino R, Cohen ML, Dickersin K, Emerson SS, Farrar JT, et al. The prevention and treatment of missing data in clinical trials. N Engl J Med. 2012;367:1355-60. [PMID: 23034025] doi:10.1056/NEJMsr1203730 CrossrefMedlineGoogle Scholar4. Molenberghs G, Kenward M. Missing Data in Clinical Studies. London: J Wiley; 2007. Google Scholar5. Carpenter JR, Kenward MG. Missing data in randomised controlled trials - a practical guide. Publication RM03/JH17/MK. Birmingham, United Kingdom: National Institute for Health Research; 2008. Google Scholar6. Bell ML, Fairclough DL. Practical and statistical issues in missing data for longitudinal patient-reported outcomes. Stat Methods Med Res. 2014;23:440-59. [PMID: 23427225] doi:10.1177/0962280213476378 CrossrefMedlineGoogle Scholar7. Mallinckrodt C, Roger J, Chuang-Stein C, Molenberghs G, O'Kelly M, Ratitch B, et al. Recent developments in the prevention and treatment of missing data. Ther Innov Regul Sci. 2014;48:68-80. doi:10.1177/2168479013501310 CrossrefMedlineGoogle Scholar8. White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30:377-99. [PMID: 21225900] doi:10.1002/sim.4067 CrossrefMedlineGoogle Scholar Author, Article, and Disclosure InformationAffiliations: From University of Pennsylvania and American College of Physicians, Philadelphia, Pennsylvania.Acknowledgment: The authors thank the Annals senior clinical and statistical editors for their input and review of earlier drafts of the manuscript.Disclosures: Authors have disclosed no conflicts of interest. Forms can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M17-0195.Corresponding Author: Joshua M. Liao, MD, Department of Medicine, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104.Current Author Addresses: Dr. Liao: Department of Medicine, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104.Dr. Stack: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106.Author Contributions: Conception and design: J.M. Liao, C.B. Stack.Drafting of the article: J.M. Liao, C.B. Stack.Critical revision of the article for important intellectual content: J.M. Liao, C.B. Stack.Final approval of the article: J.M. Liao, C.B. Stack.Statistical expertise: C.B. Stack.This article was published at Annals.org on 28 February 2017. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetailsSee AlsoAnnals Understanding Clinical Research: Intention-to-Treat Analysis Joshua M. Liao , Catharine B. Stack , Michael E. Griswold , and A. Russell Localio Metrics Cited byThe impact of poly-traumatization on treatment outcomes in young people with substance use disordersExtra-Virgin Olive Oil Improves Depression Symptoms Without Affecting Salivary Cortisol and Brain-Derived Neurotrophic Factor in Patients With Major Depression: A Double-Blind Randomized Controlled TrialThe effect of flaxseed on physical and mental fatigue in children and adolescents with overweight/obesity: a randomised controlled trialMissing data in randomised controlled trials of rheumatoid arthritis drug therapy are substantial and handled inappropriatelySelf-reported sick leave following a brief preventive intervention on work-related stress: a randomised controlled trial in primary health careAnnals Clinical Decision Making: Weighing Evidence to Inform Clinical DecisionsJoshua P. Metlay, MD, PhD and Katrina A. Armstrong, MD, MSAn observational study on trajectories and outcomes of chronic low back pain patients referred from a spine surgery division for chiropractic treatmentSpinal manipulative therapy and exercise for older adults with chronic low back pain: a randomized clinical trialThe Importance of Reporting Biases in Patient Care: Can We Trust the Evidence From Either Individual Studies or Systematic Reviews?Kay Dickersin, MA, PhD and Riaz Qureshi, MScImprovements in clinical characteristics of patients with non-alcoholic fatty liver disease, after an intervention based on the Mediterranean lifestyle: a randomised controlled clinical trialSpinal manipulation and exercise for low back pain in adolescents: a randomized trialDesigned Learning: Missing Data in Clinical ResearchCatharine B. Stack, PhD, Trevor Butterworth, BA, MPhil, MS, and Rebecca Goldin, PhDComment on CONCEPT by Reginster et al : are the authors' interpretations supported by the data analysis?Annals Understanding Clinical Research: Intention-to-Treat AnalysisJoshua M. Liao, MD, Catharine B. Stack, PhD, Michael E. Griswold, PhD, and A. Russell Localio, PhD 18 April 2017Volume 166, Issue 8Page: 596-598KeywordsClinical trialsExerciseExercise therapyForecastingInformation technologyPostural controlRandomized trialsResearch designStatistical dataTechnicians ePublished: 28 February 2017 Issue Published: 18 April 2017 Copyright & PermissionsCopyright © 2017 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

A randomized trial aims to reduce bias in comparing 2 or more treatments. An intention-to-treat (ITT) analysis theoretically evaluates all participants within their original randomly assigned group in an attempt to fulfill this primary aim. In practice, however, a true ITT analysis is often difficult when not all patients complete the trial or fully adhere to randomized treatments. Despite the challenges of dropout and nonadherence, investigators routinely report that ITT analyses were conducted without further explanation (1). In 2010, CONSORT (Consolidated Standards of Reporting Trials) replaced the item requiring ITT analyses with a request for a more informative description of who is included in each analysis and within each treatment group (2). A more suitable term to replace ITT might be as randomized. Readers and reviewers should look deeper into how missing outcomes and nonadherence have been handled rather than simply checking to determine whether an analysis is described as ITT. In some settings, an analysis carried out as ITT may be biased or may not even answer the questions of most interest. The Example We illustrate important issues and considerations surrounding ITT analyses by using a trial that randomly assigned 169 patients older than 50 years to receive either surgical decompression or physical therapy for symptomatic lumbar spinal stenosis (3). After their operations, the patients in the decompression group participated in a postoperative graduated ambulation program and were encouraged to increase their level of walking as tolerated. Patients in the physical therapy group received a structured program that included twice-weekly meetings for 6 weeks; used a structured approach to identify strength and flexibility impairments; and emphasized general conditioning, patient education, and lumbar flexion exercises. Patients in the physical therapy group could cross over or switch to surgical decompression at any point during the study period based on shared decision making with their spine surgeons. The study protocol required patients to complete the Short Form-36 Health Survey, which assessed their physical function, at 2 years. Approximately 75% of patients in both groups completed 2-year follow-up, by which point 57% of those in the physical therapy group had switched to receive surgical decompression. The authors reported ITT analyses as well as additional analyses that tried to estimate efficacy despite the high proportion of crossovers. They concluded that surgical decompression and physical therapy led to similar improvement in physical function scores. The primary clinical question the trial was trying to answer was, does surgery or physical therapy provide better treatment? Readers may use several key related questions to assess whether an analysis based on ITT principles answers this question of interest (Table 1). Glossary Table 1. Key Questions for Assessing the Meaning of an ITT Analysis Did Patients Adhere to Their Randomized Treatment? If Not, Why Were They Nonadherent? Did Adherence to the Assigned Treatment Vary by Study Group? Not all patients received their randomly assigned treatment, and most nonadherent patients switched to receive the other treatment. Most patients who deviated from their assigned treatment were randomly assigned to physical therapy but switched to receive surgery. Within the 2-year study period, 2 patients (2%) randomly assigned to surgery declined it and received physical therapy, whereas 47 patients (57%) randomly assigned to physical therapy received surgery, more than half (31 of 47) of whom switched to surgery before week 10. Three additional patients randomly assigned to physical therapy received neither physical therapy nor surgery. Are All Randomly Assigned Patients Included in the Analyses? All available measurements from the 87 surgery and 82 physical therapy patients were included in the primary analysis within their randomly assigned groups. The primary analysis used a mixed-effects model to accommodate the repeated measurements of physical function score. Treatment was modeled as each patient's randomized assignment, and missing outcomes were assumed to be missing at random (see our previous article [4] for more on missing-data mechanisms). Based on the Answers to the Aforementioned Questions, What Clinical Question Does the ITT Analysis Answer? Within the context of this trial, the ITT or as-randomized analysis answered the question, what is the effect on patients' outcomes if they are assigned to surgery or to physical therapy? In other words, the ITT analysis addressed the effectiveness of the initial assignment to surgery versus physical therapy in those who were eligible for and would agree to surgery. Given the observed nonadherence, however, the ITT analysis did not estimate the efficacy of surgery, nor did it estimate the effect of receipt of surgery versus the receipt of physical therapy in a group of eligible patients (5). To estimate efficacy, we would want to know what would happen to a group of patients if they could actually be treated with surgery versus physical therapy. Can This Trial Help to Answer Other Clinically Relevant Questions? Are Analyses Included That Address Such Questions? The authors carried out 2 additional analyses to estimate the efficacy of surgery in different ways and under different assumptions. They modeled compliance along with outcome to estimate the complier average causal effect (CACE; also called the local average treatment effect), that is, the treatment effect in the subgroup of patients who would adhere to whatever treatment was assigned to them (68). The CACE value is not simply the contrast of outcomes among the patients who actually adhered to their assigned treatment (an as-treated analysis), as that contrast does not benefit from randomization and may be strongly biased. Instead, the CACE estimate compares patients in the interventional (physical therapy) group who adhered to their assigned treatment with those in the alternative (surgery) group who would have adhered to the interventional treatment (physical therapy) if they had been assigned to it. In this example, the CACE estimate is imprecise, with wide confidence bounds that include a potential benefit of physical therapy of as much as 31.4 points (Table 2). Table 2. Treatment Effects From ITT and Additional Analyses* Delitto and colleagues (3) also performed an analysis by using inverse probability weighting to estimate the treatment effect in patients who received their randomized assignment, while accounting for likely bias from self-selection and dropout. This analysis used data from the trial along with adherence and dropout information in an attempt to create a balanced pseudo-population of patients who received their randomized treatment and completed the study (9). Results under this analytic approach were similar to the effect of the receipt of surgery or physical therapy. If the authors had instead carried out an as-treated analysis by simply excluding the physical therapy patients who did not receive physical therapy and the surgery patients who did not receive surgery, the results would not be protected by the randomization. Such results would be subject to the bias that is present in observational data (10). Similarly, if the analysis had excluded all patients who did not follow the protocol, under a per protocol analysis, the results likely would be invalid because of selection bias. Instead, these authors performed additional analyses that are more complex but likely to provide more valid estimates for the actual efficacy questions of interest. The results from these 3 analyses support the authors' conclusion that patients and physicians should engage in shared decision-making discussions about surgical or nonsurgical treatments. Conclusion Randomization at baseline does not ensure full adherence to study protocols. As a result, analyses done in accordance with ITT principles do not automatically produce the answers clinicians want. The related questions provided here may help clinicians and peer reviewers to assess the clinical relevance of the results from ITT analyses and to decide whether results from additional analyses would be helpful.

Editorials6 January 2015TRIPOD: A New Reporting Baseline for Developing and Interpreting Prediction ModelsA. Russell Localio, PhD and Catharine B. Stack, PhD, Deputy Editor, StatisticsA. Russell Localio, PhDFrom Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, and Annals of Internal Medicine (Deputy Editor, Statistics). and Catharine B. Stack, PhD, Deputy Editor, StatisticsFrom Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, and Annals of Internal Medicine (Deputy Editor, Statistics).Author, Article, and Disclosure Informationhttps://doi.org/10.7326/M14-2423 SectionsAboutFull TextPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinkedInRedditEmail In this issue, the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) investigators propose an annotated checklist for transparent reporting of prediction or prognostic models (1). The accompanying 22 000-word "Explanation and Elaboration" (2), with more than 500 references from statistics, epidemiology, and clinical decision making as well as from the applied clinical literature, should serve as an important resource for model developers. Applications span prediction (for diagnosis from cross-sectional data) and prognosis (for outcomes after longitudinal follow-up) across many clinical settings. These TRIPOD documents represent serious efforts to synthesize best practices for authors and readers.... References1. Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): the TRIPOD statement. Ann Intern Med. 2015;162:55-63. doi:10.7326/M14-0697 LinkGoogle Scholar2. Moons KG, Altman DG, Reitsma JB, Ioannidis JP, Macaskill P, Steyerberg EW, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015;162:W1-73. doi:10.7326/M14-0698 LinkGoogle Scholar3. Laine C, Goodman SN, Griswold ME, Sox HC. Reproducible research: moving toward research the public can really trust. Ann Intern Med. 2007;146:450-3. [PMID: 17339612] doi:10.7326/0003-4819-146-6-200703200-00154 LinkGoogle Scholar4. Smits M, Dippel DW, Steyerberg EW, deHaan GG, Dekker HM, Vos PE, et al. Predicting intracranial traumatic findings on computed tomography in patients with minor head injury: the CHIP prediction rule. Ann Intern Med. 2007;146:397-405. [PMID: 17371884] doi:10.7326/0003-4819-146-6-200703200-00004 LinkGoogle Scholar5. Lautenbach E, Localio R, Nachamkin I. Clinicians required very high sensitivity of a bacteremia prediction rule. J Clin Epidemiol. 2004;57:1104-6. [PMID: 15528062] CrossrefMedlineGoogle Scholar6. Vickers AJ, Cronin AM. Everything you always wanted to know about evaluating prediction models (but were too afraid to ask). Urology. 2010;76:1298-301. [PMID: 21030068] doi:10.1016/j.urology.2010.06.019 CrossrefMedlineGoogle Scholar7. Raji OY, Duffy SW, Agbaje OF, Baker SG, Christiani DC, Cassidy A, et al. Predictive accuracy of the Liverpool Lung Project risk model for stratifying patients for computed tomography screening for lung cancer: a case–control and cohort validation study. Ann Intern Med. 2012;157:242-50. [PMID: 22910935] doi:10.7326/0003-4819-157-4-201208210-00004 LinkGoogle Scholar8. Pepe MS, Janes H, Li CI. Net risk reclassification P values: valid or misleading? J Natl Cancer Inst. 2014;106:dju041. [PMID: 24681599] doi:10.1093/jnci/dju041 CrossrefMedlineGoogle Scholar9. Daniel RM, Cousens SN, DeStavola BL, Kenward MG, Sterne JA. Methods for dealing with time-dependent confounding. Stat Med. 2013;32:1584-618. [PMID: 23208861] doi:10.1002/sim.5686 CrossrefMedlineGoogle Scholar10. van Klaveren D, Steyerberg EW, Perel P, Vergouwe Y. Assessing discriminative ability of risk models in clustered data. BMC Med Res Methodol. 2014;14:5. [PMID: 24423445] doi:10.1186/1471-2288-14-5 CrossrefMedlineGoogle Scholar Author, Article, and Disclosure InformationAuthors: A. Russell Localio, PhD; Catharine B. Stack, PhD, Deputy Editor, StatisticsAffiliations: From Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, and Annals of Internal Medicine (Deputy Editor, Statistics).Disclosures: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M14-2423.Corresponding Author: A. Russell Localio, PhD, Department of Biostatistics and Epidemiology, Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, 635 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104-6021.Current Author Addresses: Dr. Localio: Department of Biostatistics and Epidemiology, Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine, University of Pennsylvania, 635 Blockley Hall, 423 Guardian Drive, Philadelphia, PA 19104-6021.Dr. Stack: American College of Physicians, 190 N. Independence Mall West, Philadelphia, PA 19106. PreviousarticleNextarticle Advertisement FiguresReferencesRelatedDetailsSee AlsoTransparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): The TRIPOD Statement Gary S. Collins , Johannes B. Reitsma , Douglas G. Altman , and Karel G.M. Moons Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): Explanation and Elaboration Karel G.M. Moons , Douglas G. Altman , Johannes B. Reitsma , John P.A. Ioannidis , Petra Macaskill , Ewout W. Steyerberg , Andrew J. Vickers , David F. Ransohoff , and Gary S. Collins Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): The TRIPOD Statement Gary S. Collins , Johannes B. Reitsma , Douglas G. Altman , and Karel G.M. Moons Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): Explanation and Elaboration Karel G.M. Moons , Douglas G. Altman , Johannes B. Reitsma , John P.A. Ioannidis , Petra Macaskill , Ewout W. Steyerberg , Andrew J. Vickers , David F. Ransohoff , and Gary S. Collins Metrics Cited byDelirium Prediction Using Machine Learning Interpretation Method and Its Incorporation into a Clinical WorkflowRadiomics-based prediction of two-year clinical outcome in locally advanced cervical cancer patients undergoing neoadjuvant chemoradiotherapyPrognostic Research in Traumatic Brain Injury: Markers, Modeling, and Methodological PrinciplesArtificial Intelligence and Liability in Medicine: Balancing Safety and InnovationScreening the risk factors for methamphetamine use in pregnant women not receiving prenatal careDevelopment and Validation of a Hospital Indicator of Activity-Based Costs for Injury AdmissionsCurrent Clinical Applications of Artificial Intelligence in Radiology and Their Best Supporting EvidenceRadiomics for radiation oncologists: are we ready to go?Oncotype DX Predictive Nomogram for Recurrence Score Output: The Novel System ADAPTED01 Based on Quantitative Immunochemistry AnalysisFunctional Outcome Prediction in Ischemic Stroke: A Comparison of Machine Learning Algorithms and Regression ModelsExternal Validation of Two Models to Predict Delirium in Critically Ill Adults Using Either the Confusion Assessment Method-ICU or the Intensive Care Delirium Screening Checklist for Delirium AssessmentMultinational development and validation of an early prediction model for delirium in ICU patients 6 January 2015Volume 162, Issue 1Page: 73-74KeywordsBacteremiaBiostatisticsCancer epidemiologyClinical epidemiologyDecision makingDisclosureInformation storage and retrievalMedical risk factorsStatistical dataStatistical methods ePublished: 6 January 2015 Issue Published: 6 January 2015 Copyright & PermissionsCopyright © 2015 by American College of Physicians. All Rights Reserved.PDF downloadLoading ...

BACKGROUND:Obesity rates in the United States have escalated in recent decades and present a major challenge in public health prevention efforts. Currently, testing to identify genetic risk for obesity is readily available through several direct-to-consumer companies. Despite the availability of this type of testing, there is a paucity of evidence as to whether providing people with personal genetic information on obesity risk will facilitate or impede desired behavioral responses.PURPOSE:We describe the key issues in the design and implementation of a randomized controlled trial examining the clinical utility of providing genetic risk information for obesity.METHODS:Participants are being recruited from the Coriell Personalized Medicine Collaborative, an ongoing, longitudinal research cohort study designed to determine the utility of personal genome information in health management and clinical decision making. The primary focus of the ancillary Obesity Risk Communication Study is to determine whether genetic risk information added value to traditional communication efforts for obesity, which are based on lifestyle risk factors. The trial employs a 2 × 2 factorial design in order to examine the effects of providing genetic risk information for obesity, alone or in combination with lifestyle risk information, on participants' psychological responses, behavioral intentions, health behaviors, and weight.RESULTS:The factorial design generated four experimental arms based on communication of estimated risk to participants: (1) no risk feedback (control), (2) genetic risk only, (3) lifestyle risk only, and (4) both genetic and lifestyle risk (combined). Key issues in study design pertained to the selection of algorithms to estimate lifestyle risk and determination of information to be provided to participants assigned to each experimental arm to achieve a balance between clinical standards and methodological rigor. Following the launch of the trial in September 2011, implementation challenges pertaining to low enrollment and differential attrition became apparent and required immediate attention and modifications to the study protocol. Although monitoring of these efforts is ongoing, initial observations show a doubling of enrollment and reduced attrition.LIMITATIONS:The trial is evaluating the short-term impact of providing obesity risk information as participants are followed for only 3 months. This study is built upon the structure of an existing personalized medicine study wherein participants have been provided with genetic information for other diseases. This nesting in a larger study may attenuate the effects of obesity risk information and has implications for the generalizability of study findings.CONCLUSIONS:This randomized trial examines value of obesity genetic information, both when provided independently and when combined with lifestyle risk assessment, to motivate individuals to engage in healthy lifestyle behaviors. Study findings will guide future intervention efforts to effectively communicate genetic risk information.

We describe the development and implementation of a randomized controlled trial to investigate the impact of genomic counseling on a cohort of patients with heart failure (HF) or hypertension (HTN), managed at a large academic medical center, the Ohio State University Wexner Medical Center (OSUWMC). Our study is built upon the existing Coriell Personalized Medicine Collaborative (CPMC®). OSUWMC patient participants with chronic disease (CD) receive eight actionable complex disease and one pharmacogenomic test report through the CPMC® web portal. Participants are randomized to either the in-person post-test genomic counseling-active arm, versus web-based only return of results-control arm. Study-specific surveys measure: (1) change in risk perception; (2) knowledge retention; (3) perceived personal control; (4) health behavior change; and, for the active arm (5), overall satisfaction with genomic counseling. This ongoing partnership has spurred creation of both infrastructure and procedures necessary for the implementation of genomics and genomic counseling in clinical care and clinical research. This included creation of a comprehensive informed consent document and processes for prospective return of actionable results for multiple complex diseases and pharmacogenomics (PGx) through a web portal, and integration of genomic data files and clinical decision support into an EPIC-based electronic medical record. We present this partnership, the infrastructure, genomic counseling approach, and the challenges that arose in the design and conduct of this ongoing trial to inform subsequent collaborative efforts and best genomic counseling practices.

A primary goal of meta-analysis is to improve the estimation of treatment effects by pooling results of similar studies. This article explains how the most widely used method for pooling heterogeneous studies--the Der Simonian-Laird (DL) estimator--can produce biased estimates with falsely high precision. A classic example is presented to show that use of the DL estimator can lead to erroneous conclusions. Particular problems with the DL estimator are discussed, and several alternative methods for summarizing heterogeneous evidence are presented. The authors support replacing universal use of the DL estimator with analyses based on a critical synthesis that recognizes the uncertainty in the evidence,focuses on describing and explaining the probable sources of variation in the evidence, and uses random-effects estimates that provide more accurate confidence limits than the DL estimator.

This issue includes 2 systematic reviews that use the same data to address the same question: Compared with iliac crest bone grafting, does rhBMP-2 safely improve outcomes of spinal fusion surgery?...

Authors’ Views on Online-Only Publication Frauke Becher,1 Are Brean,1 Erlend Hem,1 Christine Laine,2,3 Alicia Ludwig,3 Mary Beth Schaeffer,2,3 Catharine Stack,2,3 Arlene Weissman3 Objective Many biomedical journals have both print and electronic versions, providing the opportunity for some articles to appear online only. We wanted to determine authors’ views on online-only publication in such settings. Design Using Survey Monkey, we surveyed individuals who served as corresponding author on at least one article published in 2012 in the Annals of Internal Medicine (US-based journal) and the Journal of the Norwegian Medical Association (Scandinavia-based journal). In addition to asking about authors’ reading habits, we asked them to suppose that the journal they had published in began to publish some articles online only with no print publication, specifying that online-only articles would be indexed in PubMed as were those that appeared in both electronic and print versions. We then asked whether and how the possibility of online-only publication would influence their likelihood of submitting future work to the journal. Both journals publish new issues twice a month. The US-based and Scandinavia-based journals have circulations of approximately 85,000 and 28,650, respectively, and receive about 3,000 and 1,500 manuscripts per year, respectively. Results Seventy percent (237/335) of authors at the Scandinavia-based journal and 161 of 274 (59%) authors at the US-based journal completed the survey. Forty-five percent of Scandinavia-journal authors and 26% of US-journal authors reported being less likely to submit articles if they knew their article would be published online only, with no subsequent print publication (Table 18). A smaller number of authors (5% Scandinavia, 11% US) noted being more likely to submit if their article was available online only. Reasons for being less likely

Confidence in evidence summarized in meta-analyses depends on the strength of the underlying studies. This inherent limitation of syntheses appears in the case of a meta-analysis of sodium-glucose cotransporter 2 inhibitors for the treatment of type 2 diabetes because many of the pertinent randomized trials did not handle patient dropout and "rescue" medication properly. Repudiated statistical methods, such as last observation carried forward, and unsophisticated methods for handling postrescue data produce unreliable summary estimates. Future reports of randomized studies and meta-analyses of those studies must focus on posing precise questions about the treatment effect of interest and then implement appropriate statistical methods to account for missing data, patient dropout, and use of rescue medication.

Implementation of pharmacogenomics (PGx) in clinical care can lead to improved drug efficacy and reduced adverse drug reactions. However, there has been a lag in adoption of PGx tests in clinical practice. This is due in part to a paucity of rigorous systems for translating published clinical and scientific data into standardized diagnostic tests with clear therapeutic recommendations. Here we describe the Pharmacogenomics Appraisal, Evidence Scoring and Interpretation System (PhAESIS), developed as part of the Coriell Personalized Medicine Collaborative research study, and its application to seven commonly prescribed drugs.

Purpose: Recent genome wide-association studies have identified hundreds of single nucleotide polymorphisms associated with common complex diseases. With the momentum of these discoveries comes a need to communicate this information to individuals.Methods: The Coriell Personalized Medicine Collaborative is an observational research study designed to evaluate the utility of personalized genomic information in health care. Participants provide saliva samples for genotyping and complete extensive on-line medical history, family history, and lifestyle questionnaires. Only results for diseases deemed potentially actionable by an independent advisory board are reported.Results: We present our methodology for developing personalized reports containing risks for both genetic and nongenetic factors. Risk estimates are given as relative risk, derived or reported from representative peer-reviewed publications. Estimates of disease prevalence are also provided. Presenting risk as relative risk allows for consistent reporting across multiple diseases and across genetic and nongenetic factors. Using this approach eliminates the need for assumptions regarding population lifetime risk estimates. Publications used for risk reporting are selected based on the strength of the design and study quality.Conclusion: Coriell Personalized Medicine Collaborative risk reports demonstrate an approach to communicating risk of complex disease via the web that encompasses risks due to genetic variants along with risks caused by family history and lifestyle factors.

There is a dearth of large prospective studies to determine if genetic risk factors are useful predictors of health outcomes and if reporting them to individuals or physicians changes health behavior. The Coriell Personalized Medicine Collaborative® (CPMC, NJ, USA) is a prospective observational study with three cohorts - community, cancer and chronic disease cohorts. Participants provide detailed medical history through a dynamic internet-based portal. DNA is tested and personalized risk reports are provided for potentially actionable health conditions. To date, the community cohort has enrolled 4372 participants. The internet-based portal supplies educational content, captures phenotypic data and delivers customized risk reports. The Informed Cohort Oversight Board has approved 16 health conditions to date, and risk reports with genetic and nongenetic risks for six conditions have been released. The majority (87%) of participants who completed requisite questionnaires viewed at least one report. The CPMC is a cohort study delivering customized risk reports for actionable conditions using a web interface and measuring outcomes longitudinally.

To the Editor: The recent article by Mihaescu et al.1.Mihaescu R. van Hoek M. Sijbrands E.J. Evaluation of risk prediction updates from commercial genome-wide scans.10.1097/GIM.0b013e3181b13a4fGenet Med. 2009; 11: 588-594Google Scholar makes important points regarding the impact of updates to risk factors and the limitations of disease risk estimates derived from genetic variants during a time of active discovery. However, there is a fundamental assumption made, both in this work and in the presentations of disease risk from commercial companies offering genome-wide scans, that is worthy of challenge. The threshold used is the population average. By presenting risks in relationship to the average in the overall population, clinical utility of the population average is implied, though this assumption most often is not supported. In the Rotterdam Study population used by Mihaescu et al., the average risk of type 2 diabetes, calculated using both incident and prevalent cases, is reported as 20%. Based on the single TCF7L2 variant, predicted risks were 17.6%, 20.8%, and 28.0% in the CC, CT, and TT genotype groups, respectively. So, consistent with the risk reporting used by direct-to-consumer companies offering full genome scans, risk of type 2 diabetes is deemed "below average" for CC individuals and "above average" for CT and TT individuals. However, what is the meaning of "above average" risk, particularly for the CT individuals, who make up about 40% of the population,2.van Hoek M. Dehghan A. Witteman J.C. Predicting type 2 diabetes based on polymorphisms from Genome-Wide Association Studies: a population-based study.1:CAS:528:DC%2BD1MXhsleis7c%3D10.2337/db08-0425Diabetes. 2008; 57: 3122-3128Google Scholar and in whom the predicted risk is 20.8%? In addition, what are the implications of comparing the risk in one subgroup of a population to the risk in the full population, when variants are common and subgroups make up a substantial portion of the total population (in this case 40%)? Some well-studied, clinically developed risk scores have corresponding thresholds used in clinical care. The Framingham risk score, for example, provides estimates of the 10-year risk of heart attack or dying from coronary heart disease, based on a patient's age, gender, smoking status, diabetes status, blood pressure, and cholesterol.3.Wilson P. D'Agostino R. Levy D. Belanger A.M. Silbershatz H. Kannel W.B. Prediction of coronary heart disease using risk factor categories.1:STN:280:DyaK1c3msVOjsA%3D%3D10.1161/01.CIR.97.18.1837Circulation. 1998; 97: 1837-1847Google Scholar Current guidelines for prescribing cholesterol lowering therapy from the National Cholesterol Education Program-(ACT III) incorporate ranges of Framingham 10-year risk (<10%, 10-20%, and >20%),4.Grundy S.M. Cleeman J.I. Mairey Merz C.N. National Heart, Lung, and Blood Institute. American College of Cardiology Foundation. American Heart Association. Implications of recent clinical trials for the National Cholesterol Education Program Adult Treatment Panel III Guidelines.10.1161/01.CIR.0000133317.49796.0ECirculation. 2004; 110: 227-239Google Scholar and physicians may use the threshold of 20% when making treatment decisions. Hence, in this setting, a threshold of 20% for Framingham 10-year risk has utility, and classifications and reclassifications based on this cutoff are consequential. Returning to the example of type 2 diabetes, one might attempt to put the risks of disease based on genetic variants within clinical context by referring to comparable risk estimates in the clinical literature. The relationship between body mass index (BMI) and risk of diabetes has been established. Also, Narayan et al.5.Narayan K.M. Boyle J.P. Thompson T.J. Gregg E.W. Williamson D.F. Effect of BMI on lifetime risk for diabetes in the U.S.1:STN:280:DC%2BD2szhsV2hsw%3D%3D10.2337/dc06-2544Diabetes Care. 2007; 30: 1562-1566Google Scholar estimated the remaining life-time risk of type 2 diabetes at age 18 years to be 19.8% for men of average weight (18.5 ≤ BMI <25 kg/m2), 29.7% for overweight men (25 ≤ BMI <30 kg/m2), and 57.0% for obese men (30 ≤ BMI <35 kg/m2). Similar estimates of remaining life-time risk of type 2 diabetes at age 18 years in women were 17.1%, 35.4%, and 54.6%, respectively. Based on these estimates, remaining life-time risks at age 18 years of 30-35% (overweight) and greater (55-57%, obese) provide a clinical framework, and possible thresholds, for reported risks. (To apply these cutoffs to the predicted risks presented by Mihaescu et al., however, one would have to demonstrate that the predicted risks, which might be interpreted as life-time risks of developing disease from birth to an average age of 69.5 years, are comparable with remaining life-time risk at age 18 years.) Another approach to presenting risk estimates within a clinical context is to present risk of disease due to genetic and nongenetic factors side-by-side. This option, which our research team is currently pursuing, requires knowledge of an individual's medical, lifestyle, and family history. Granted for some diseases, risks due to lifestyle behaviors or other nongenetic factors that are relevant for all demographic groups are not available. However, presenting genetic risk alone and applying a threshold of the population average most often does not place reported risks within a meaningful clinical context. Disclosure: The author declares no conflict of interest.