Marc Aerts1, Girma Minalu2, Stefan Bösner3, Frank Buntinx4, Bernard Burnand5, Jörg Haasenritter3, Lilli Herzig6, J André Knottnerus7, Staffan Nilsson8, Walter Renier9, Carol Sox10, Harold Sox11, Norbert Donner-Banzhoff3. 1. Interuniversity Institute for Biostatistics and Statistical Bioinformatics (I-BIOSTAT), Hasselt University, I-BioStat, Agoralaan, Building D, Diepenbeek B-3590, Belgium. Electronic address: marc.aerts@uhasselt.be. 2. Interuniversity Institute for Biostatistics and Statistical Bioinformatics (I-BIOSTAT), Hasselt University, I-BioStat, Agoralaan, Building D, Diepenbeek B-3590, Belgium. 3. Department of General Practice and Family Medicine, Philipps University Marburg, Karl-von-Str. 4, Marburg 35037, Germany. 4. Department of Public Health and Primary Care, KU Leuven, Kapucijnenvoer 33, Blok J, PB 7001, Leuven 3000, Belgium; Department of General Practice, Maastricht University, Peter Debyeplein 1, P.O. Box 616, Maastricht 6200 MD, The Netherlands. 5. Institute of Social and Preventive Medicine, Lausanne University Hospital, Route de la Corniche 10, Lausanne 1010, Switzerland. 6. Institute of Family Medicine, University of Lausanne, 44 rue du Bugnon, Lausanne CH-1011, Switzerland. 7. Department of General Practice, Maastricht University, Peter Debyeplein 1, P.O. Box 616, Maastricht 6200 MD, The Netherlands. 8. Division of Community Medicine, Department of Medicine and Health Sciences, Linköping University, Linköping SE-581 83, Sweden. 9. Department of Public Health and Primary Care, KU Leuven, Kapucijnenvoer 33, Blok J, PB 7001, Leuven 3000, Belgium. 10. Department of Community and Family Medicine, Geisel School of Medicine at Dartmouth, 1 Rope Ferry Road, Hanover, NH 03755-1404, USA. 11. Department of Community and Family Medicine, Geisel School of Medicine at Dartmouth, 1 Rope Ferry Road, Hanover, NH 03755-1404, USA; Patient-Centered Outcomes Research Institute, 1828 L Street, NW, Suite 900, Washington, DC 20036, USA.
Abstract
OBJECTIVE: To construct a clinical prediction rule for coronary artery disease (CAD) presenting with chest pain in primary care. STUDY DESIGN AND SETTING: Meta-Analysis using 3,099 patients from five studies. To identify candidate predictors, we used random forest trees, multiple imputation of missing values, and logistic regression within individual studies. To generate a prediction rule on the pooled data, we applied a regression model that took account of the differing standard data sets collected by the five studies. RESULTS: The most parsimonious rule included six equally weighted predictors: age ≥55 (males) or ≥65 (females) (+1); attending physician suspected a serious diagnosis (+1); history of CAD (+1); pain brought on by exertion (+1); pain feels like "pressure" (+1); pain reproducible by palpation (-1). CAD was considered absent if the prediction score is <2. The area under the ROC curve was 0.84. We applied this rule to a study setting with a CAD prevalence of 13.2% using a prediction score cutoff of <2 (i.e., -1, 0, or +1). When the score was <2, the probability of CAD was 2.1% (95% CI: 1.1-3.9%); when the score was ≥ 2, it was 43.0% (95% CI: 35.8-50.4%). CONCLUSIONS: Clinical prediction rules are a key strategy for individualizing care. Large data sets based on electronic health records from diverse sites create opportunities for improving their internal and external validity. Our patient-level meta-analysis from five primary care sites should improve external validity. Our strategy for addressing site-to-site systematic variation in missing data should improve internal validity. Using principles derived from decision theory, we also discuss the problem of setting the cutoff prediction score for taking action.
OBJECTIVE: To construct a clinical prediction rule for coronary artery disease (CAD) presenting with chest pain in primary care. STUDY DESIGN AND SETTING: Meta-Analysis using 3,099 patients from five studies. To identify candidate predictors, we used random forest trees, multiple imputation of missing values, and logistic regression within individual studies. To generate a prediction rule on the pooled data, we applied a regression model that took account of the differing standard data sets collected by the five studies. RESULTS: The most parsimonious rule included six equally weighted predictors: age ≥55 (males) or ≥65 (females) (+1); attending physician suspected a serious diagnosis (+1); history of CAD (+1); pain brought on by exertion (+1); pain feels like "pressure" (+1); pain reproducible by palpation (-1). CAD was considered absent if the prediction score is <2. The area under the ROC curve was 0.84. We applied this rule to a study setting with a CAD prevalence of 13.2% using a prediction score cutoff of <2 (i.e., -1, 0, or +1). When the score was <2, the probability of CAD was 2.1% (95% CI: 1.1-3.9%); when the score was ≥ 2, it was 43.0% (95% CI: 35.8-50.4%). CONCLUSIONS: Clinical prediction rules are a key strategy for individualizing care. Large data sets based on electronic health records from diverse sites create opportunities for improving their internal and external validity. Our patient-level meta-analysis from five primary care sites should improve external validity. Our strategy for addressing site-to-site systematic variation in missing data should improve internal validity. Using principles derived from decision theory, we also discuss the problem of setting the cutoff prediction score for taking action.
Keywords:
Chest pain; Individual patient data meta-analysis; Medical history taking; Myocardial ischemia; Primary health care; Sensitivity and specificity; Symptom assessment
Authors: Angel M R Schols; Robert T A Willemsen; Tobias N Bonten; Martijn H Rutten; Patricia M Stassen; Bas L J H Kietselaer; Geert-Jan Dinant; Jochen W L Cals Journal: Ann Fam Med Date: 2019-07 Impact factor: 5.166
Authors: Nini H Jonkman; Marco Colpo; Jochen Klenk; Chris Todd; Trynke Hoekstra; Vieri Del Panta; Kilian Rapp; Natasja M van Schoor; Stefania Bandinelli; Martijn W Heymans; Dominique Mauger; Luca Cattelani; Michael D Denkinger; Dietrich Rothenbacher; Jorunn L Helbostad; Beatrix Vereijken; Andrea B Maier; Mirjam Pijnappels Journal: BMC Geriatr Date: 2019-06-27 Impact factor: 3.921
Authors: Ralf E Harskamp; Simone C Laeven; Jelle Cl Himmelreich; Wim A M Lucassen; Henk C P M van Weert Journal: BMJ Open Date: 2019-02-27 Impact factor: 2.692
Authors: Amy Manten; Cuny J J Cuijpers; Remco Rietveld; Emma Groot; Freek van de Graaf; Sandra Voerman; Jelle C L Himmelreich; Wim A M Lucassen; Henk C P M van Weert; Ralf E Harskamp Journal: Prim Health Care Res Dev Date: 2020-05-08 Impact factor: 1.458
Authors: Tim Norman; Joanna Young; Jo Scott Jones; Gishani Egan; John Pickering; Stephen Du Toit; Fraser Hamilton; Rory Miller; Chris Frampton; Gerard Devlin; Peter George; Martin Than Journal: BMJ Open Date: 2022-04-15 Impact factor: 3.006
Authors: Loes T C M Wouters; Dorien L M Zwart; Daphne C A Erkelens; Elisabeth J M Adriaansen; Hester M den Ruijter; Esther De Groot; Roger A M J Damoiseaux; Arno W Hoes; Maarten van Smeden; Frans H Rutten Journal: BMJ Open Date: 2022-10-05 Impact factor: 3.006