Net reclassification indices for evaluating risk prediction instruments: a critical review - PubMed (original) (raw)
Review
Net reclassification indices for evaluating risk prediction instruments: a critical review
Kathleen F Kerr et al. Epidemiology. 2014 Jan.
Abstract
Net reclassification indices have recently become popular statistics for measuring the prediction increment of new biomarkers. We review the various types of net reclassification indices and their correct interpretations. We evaluate the advantages and disadvantages of quantifying the prediction increment with these indices. For predefined risk categories, we relate net reclassification indices to existing measures of the prediction increment. We also consider statistical methodology for constructing confidence intervals for net reclassification indices and evaluate the merits of hypothesis testing based on such indices. We recommend that investigators using net reclassification indices should report them separately for events (cases) and nonevents (controls). When there are two risk categories, the components of net reclassification indices are the same as the changes in the true- and false-positive rates. We advocate the use of true- and false-positive rates and suggest it is more useful for investigators to retain the existing, descriptive terms. When there are three or more risk categories, we recommend against net reclassification indices because they do not adequately account for clinically important differences in shifts among risk categories. The category-free net reclassification index is a new descriptive device designed to avoid predefined risk categories. However, it experiences many of the same problems as other measures such as the area under the receiver operating characteristic curve. In addition, the category-free index can mislead investigators by overstating the incremental value of a biomarker, even in independent validation data. When investigators want to test a null hypothesis of no prediction increment, the well-established tests for coefficients in the regression model are superior to the net reclassification index. If investigators want to use net reclassification indices, confidence intervals should be calculated using bootstrap methods rather than published variance formulas. The preferred single-number summary of the prediction increment is the improvement in net benefit.
Figures
Figure 1
In each plot the solid line is the ROC curve for the “old” marker and the dotted line is the ROC curve for the “new” risk model that incorporates the new marker. The new marker has identical distribution in all four cases. _NRI>0_=0.622 in all cases, despite the fact that the prediction increment of the new marker decreases as the strength of the old model increases.
Figure 2
The same data as Figure 1 are shown here in terms of the distributions of risks for old and new risk models. Risk distributions are shown on the log odds scale. Solid lines are the risks using the established predictors X, with nonevents tending to have lower risks than events. Dotted lines are risks using the new marker Y together with X.
Comment in
- Commentary: On NRI, IDI, and "good-looking" statistics with nothing underneath.
Hilden J. Hilden J. Epidemiology. 2014 Mar;25(2):265-7. doi: 10.1097/EDE.0000000000000063. Epidemiology. 2014. PMID: 24487208 No abstract available.
Similar articles
- Nontraditional Risk Factors in Cardiovascular Disease Risk Assessment: A Systematic Evidence Report for the U.S. Preventive Services Task Force [Internet].
Lin JS, Evans CV, Johnson E, Redmond N, Burda BU, Coppola EL, Smith N. Lin JS, et al. Rockville (MD): Agency for Healthcare Research and Quality (US); 2018 Jul. Report No.: 17-05225-EF-1. Rockville (MD): Agency for Healthcare Research and Quality (US); 2018 Jul. Report No.: 17-05225-EF-1. PMID: 30234933 Free Books & Documents. Review. - Graphical assessment of incremental value of novel markers in prediction models: From statistical to decision analytical perspectives.
Steyerberg EW, Vedder MM, Leening MJ, Postmus D, D'Agostino RB Sr, Van Calster B, Pencina MJ. Steyerberg EW, et al. Biom J. 2015 Jul;57(4):556-70. doi: 10.1002/bimj.201300260. Epub 2014 Jul 18. Biom J. 2015. PMID: 25042996 - Measures for evaluation of prognostic improvement under multivariate normality for nested and nonnested models.
Enserro DM, Demler OV, Pencina MJ, D'Agostino RB Sr. Enserro DM, et al. Stat Med. 2019 Sep 10;38(20):3817-3831. doi: 10.1002/sim.8204. Epub 2019 Jun 18. Stat Med. 2019. PMID: 31211443 Free PMC article. - The Ankle Brachial Index for Peripheral Artery Disease Screening and Cardiovascular Disease Prediction in Asymptomatic Adults: A Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet].
Lin JS, Olson CM, Johnson ES, Senger CA, Soh CB, Whitlock EP. Lin JS, et al. Rockville (MD): Agency for Healthcare Research and Quality (US); 2013 Sep. Report No.: 12-05162-EF-1. Rockville (MD): Agency for Healthcare Research and Quality (US); 2013 Sep. Report No.: 12-05162-EF-1. PMID: 24156115 Free Books & Documents. Review. - Net Reclassification Improvement.
Jewell ES, Maile MD, Engoren M, Elliott M. Jewell ES, et al. Anesth Analg. 2016 Mar;122(3):818-824. doi: 10.1213/ANE.0000000000001141. Anesth Analg. 2016. PMID: 26891393
Cited by
- The SpO2/FiO2 Ratio Combined with Prognostic Scores for Pneumonia and COVID-19 Increases Their Accuracy in Predicting Mortality of COVID-19 Patients.
Zinna G, Pipitò L, Colomba C, Scichilone N, Licata A, Barbagallo M, Russo A, Coppola N, Cascio A. Zinna G, et al. J Clin Med. 2024 Oct 2;13(19):5884. doi: 10.3390/jcm13195884. J Clin Med. 2024. PMID: 39407943 Free PMC article. - Prediction of Peripheral Artery Disease Prognosis Using Clinical and Inflammatory Biomarker Data.
Li B, Shaikh F, Zamzam A, Raphael R, Syed MH, Younes HK, Abdin R, Qadura M. Li B, et al. J Inflamm Res. 2024 Jul 22;17:4865-4879. doi: 10.2147/JIR.S471150. eCollection 2024. J Inflamm Res. 2024. PMID: 39070129 Free PMC article. - Risk estimation for the primary prevention of cardiovascular disease: considerations for appropriate risk prediction model selection.
van Daalen KR, Zhang D, Kaptoge S, Paige E, Di Angelantonio E, Pennells L. van Daalen KR, et al. Lancet Glob Health. 2024 Aug;12(8):e1343-e1358. doi: 10.1016/S2214-109X(24)00210-9. Lancet Glob Health. 2024. PMID: 39030064 Free PMC article. Review. - Predicting cardiovascular disease risk using photoplethysmography and deep learning.
Weng WH, Baur S, Daswani M, Chen C, Harrell L, Kakarmath S, Jabara M, Behsaz B, McLean CY, Matias Y, Corrado GS, Shetty S, Prabhakara S, Liu Y, Danaei G, Ardila D. Weng WH, et al. PLOS Glob Public Health. 2024 Jun 4;4(6):e0003204. doi: 10.1371/journal.pgph.0003204. eCollection 2024. PLOS Glob Public Health. 2024. PMID: 38833495 Free PMC article. - Ensemble machine learning for predicting in-hospital mortality in Asian women with ST-elevation myocardial infarction (STEMI).
Kasim S, Amir Rudin PNF, Malek S, Ibrahim KS, Wan Ahmad WA, Fong AYY, Lin WY, Aziz F, Ibrahim N. Kasim S, et al. Sci Rep. 2024 May 29;14(1):12378. doi: 10.1038/s41598-024-61151-x. Sci Rep. 2024. PMID: 38811643 Free PMC article.
References
- Wilson PW, D'Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation. 1998;97(18):1837–1847. - PubMed
- Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, Mulvihill JJ. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. Journal of the National Cancer Institute. 1989;81(24):1879–86. - PubMed
- 27th Bethesda Conference: Matching the intensity of risk factor management with the hazard for coronary disease events. Journal of the American College of Cardiology. 1996;27(5):957–1047. - PubMed
- Pencina MJ, D'Agostino RB, Sr, D'Agostino RB, Jr, Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Statistics in Medicine. 2008;27:157–172. - PubMed
Publication types
MeSH terms
Grants and funding
- R01 HL085757/HL/NHLBI NIH HHS/United States
- N01 HC095167/HC/NHLBI NIH HHS/United States
- N01 HC095161/HC/NHLBI NIH HHS/United States
- N01 HC095164/HC/NHLBI NIH HHS/United States
- N01-HC-95159/HC/NHLBI NIH HHS/United States
- N01-HC-95169/HC/NHLBI NIH HHS/United States
- N01HC95159/HL/NHLBI NIH HHS/United States
- N01 HC095169/HC/NHLBI NIH HHS/United States
- N01 HC095165/HC/NHLBI NIH HHS/United States
- N01 HC095163/HC/NHLBI NIH HHS/United States
- HL085757-07/HL/NHLBI NIH HHS/United States
- N01 HC095162/HC/NHLBI NIH HHS/United States
- N01 HC095159/HC/NHLBI NIH HHS/United States
- GM054438/GM/NIGMS NIH HHS/United States
- N01HC95169/HL/NHLBI NIH HHS/United States
- N01 HC095166/HC/NHLBI NIH HHS/United States
- N01 HC095160/HC/NHLBI NIH HHS/United States
- R01 GM054438/GM/NIGMS NIH HHS/United States
- U01 DK082185/DK/NIDDK NIH HHS/United States
- R01 CA152089/CA/NCI NIH HHS/United States
- N01 HC095168/HC/NHLBI NIH HHS/United States
LinkOut - more resources
Full Text Sources
Other Literature Sources
Medical