Dataset Information

Implications of Nine Risk Prediction Models for Selecting Ever-Smokers for Computed Tomography Lung Cancer Screening.

ABSTRACT: Background:Lung cancer screening guidelines recommend using individualized risk models to refer ever-smokers for screening. However, different models select different screening populations. The performance of each model in selecting ever-smokers for screening is unknown. Objective:To compare the U.S. screening populations selected by 9 lung cancer risk models (the Bach model; the Spitz model; the Liverpool Lung Project [LLP] model; the LLP Incidence Risk Model [LLPi]; the Hoggart model; the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial Model 2012 [PLCOM2012]; the Pittsburgh Predictor; the Lung Cancer Risk Assessment Tool [LCRAT]; and the Lung Cancer Death Risk Assessment Tool [LCDRAT]) and to examine their predictive performance in 2 cohorts. Design:Population-based prospective studies. Setting:United States. Participants:Models selected U.S. screening populations by using data from the National Health Interview Survey from 2010 to 2012. Model performance was evaluated using data from 337 388 ever-smokers in the National Institutes of Health-AARP Diet and Health Study and 72 338 ever-smokers in the CPS-II (Cancer Prevention Study II) Nutrition Survey cohort. Measurements:Model calibration (ratio of model-predicted to observed cases [expected-observed ratio]) and discrimination (area under the curve [AUC]). Results:At a 5-year risk threshold of 2.0%, the models chose U.S. screening populations ranging from 7.6 million to 26 million ever-smokers. These disagreements occurred because, in both validation cohorts, 4 models (the Bach model, PLCOM2012, LCRAT, and LCDRAT) were well-calibrated (expected-observed ratio range, 0.92 to 1.12) and had higher AUCs (range, 0.75 to 0.79) than 5 models that generally overestimated risk (expected-observed ratio range, 0.83 to 3.69) and had lower AUCs (range, 0.62 to 0.75). The 4 best-performing models also had the highest sensitivity at a fixed specificity (and vice versa) and similar discrimination at a fixed risk threshold. These models showed better agreement on size of the screening population (7.6 million to 10.9 million) and achieved consensus on 73% of persons chosen. Limitation:No consensus on risk thresholds for screening. Conclusion:The 9 lung cancer risk models chose widely differing U.S. screening populations. However, 4 models (the Bach model, PLCOM2012, LCRAT, and LCDRAT) most accurately predicted risk and performed best in selecting ever-smokers for screening. Primary Funding Source:Intramural Research Program of the National Institutes of Health/National Cancer Institute.

SUBMITTER: Katki HA

PROVIDER: S-EPMC6557386 | biostudies-literature | 2018 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Implications of Nine Risk Prediction Models for Selecting Ever-Smokers for Computed Tomography Lung Cancer Screening.

Katki Hormuzd A HA Kovalchik Stephanie A SA Petito Lucia C LC Cheung Li C LC Jacobs Eric E Jemal Ahmedin A Berg Christine D CD Chaturvedi Anil K AK

Annals of internal medicine 20180515 1

<h4>Background</h4>Lung cancer screening guidelines recommend using individualized risk models to refer ever-smokers for screening. However, different models select different screening populations. The performance of each model in selecting ever-smokers for screening is unknown.<h4>Objective</h4>To compare the U.S. screening populations selected by 9 lung cancer risk models (the Bach model; the Spitz model; the Liverpool Lung Project [LLP] model; the LLP Incidence Risk Model [LLPi]; the Hoggart ...[more]

PMID: 29800127

Similar Datasets

Project description:BackgroundLung cancer screening with chest computed tomography (CT) reduces lung cancer death. Centers for Medicare & Medicaid Services (CMS) eligibility criteria for lung cancer screening with CT require detailed smoking information and miss many incident lung cancers. An automated deep-learning approach based on chest radiograph images may identify more smokers at high risk for lung cancer who could benefit from screening with CT.ObjectiveTo develop and validate a convolutional neural network (CXR-LC) that predicts long-term incident lung cancer using data commonly available in the electronic medical record (EMR) (chest radiograph, age, sex, and whether currently smoking).DesignRisk prediction study.SettingU.S. lung cancer screening trials.ParticipantsThe CXR-LC model was developed in the PLCO (Prostate, Lung, Colorectal, and Ovarian) Cancer Screening Trial (n = 41 856). The final CXR-LC model was validated in additional PLCO smokers (n = 5615, 12-year follow-up) and NLST (National Lung Screening Trial) heavy smokers (n = 5493, 6-year follow-up). Results are reported for validation data sets only.MeasurementsUp to 12-year lung cancer incidence predicted by CXR-LC.ResultsThe CXR-LC model had better discrimination (area under the receiver-operating characteristic curve [AUC]) for incident lung cancer than CMS eligibility (PLCO AUC, 0.755 vs. 0.634; P < 0.001). The CXR-LC model's performance was similar to that of PLCOM2012, a state-of-the-art risk score with 11 inputs, in both the PLCO data set (CXR-LC AUC of 0.755 vs. PLCOM2012 AUC of 0.751) and the NLST data set (0.659 vs. 0.650). When compared in equal-sized screening populations, CXR-LC was more sensitive than CMS eligibility in the PLCO data set (74.9% vs. 63.8%; P = 0.012) and missed 30.7% fewer incident lung cancers. On decision curve analysis, CXR-LC had higher net benefit than CMS eligibility and similar benefit to PLCOM2012.LimitationValidation in lung cancer screening trials and not a clinical setting.ConclusionThe CXR-LC model identified smokers at high risk for incident lung cancer, beyond CMS eligibility and using information commonly available in the EMR.Primary funding sourceNone.

Project description:ImportanceThe American College of Radiology (ACR) has recognized the importance of minimizing radiation doses used for lung cancer screening (LCS) computed tomography (CT). However, without standard protocols, doses could still be unnecessarily high, reducing screening margin of benefit.ObjectiveTo characterize LCS CT radiation doses and identify factors explaining variation.Design, setting, and participantsWe prospectively collected LCS examination dose metrics, from 2016 to 2017, at US institutions in the University of California, San Francisco International Dose Registry. Institution-level factors were collected through baseline survey. Mixed-effects linear and logistic regression models were estimated using forward variable selection. Results are presented as percentage excess dose and odds ratios (ORs) with 95% confidence intervals (CIs). The analysis was conducted between 2018 and 2019.Main outcomes and measuresLog-transformed measures of (1) mean volume CT dose index (CTDIvol, mGy), reflecting the average radiation dose per slice; (2) mean effective dose (ED, mSv), reflecting the total dose received and estimated future cancer risk; (3) proportion of CT scans using radiation doses above ACR benchmarks (CTDIvol >3 mGy, ED >1 mSv); and (4) proportion of CT scans using radiation doses above 75th percentile of registry doses (CTDIvol >2.7 mGy, ED >1.4 mSv).ResultsData were collected for 12 529 patients undergoing LCS CT scans performed at 72 institutions. Overall, 7232 participants (58%) were men, and the median age was 65 years (interquartile range [IQR], 60-70). Of 72 institutions, 15 (21%) had median CTDIvol and 47 (65%) had median ED above ACR guidelines. Institutions allowing any radiologists to establish protocols had 44% higher mean CTDIvol (mean dose difference [MDD], 44%; 95% CI, 19%-69%) and 27% higher mean ED (MDD, 27%; 95% CI, 5%-50%) vs those limiting who established protocols. Institutions allowing any radiologist to establish protocols had higher odds of examinations exceeding ACR CTDIvol guidelines (OR, 12.0; 95% CI, 2.0-71.4), and 75th percentile of registry CTDIvol (OR, 19.0; 95% CI, 1.9-186.7) or ED (OR, 8.5; 95% CI, 1.7-42.9). Having lead radiologists establish protocols resulted in lower odds of doses exceeding ACR ED guidelines (OR, 0.01; 95% CI, 0.001-0.1). Employing external vs internal medical physicists was associated with increased odds of exceeding ACR CTDIvol guidelines (OR, 6.1; 95% CI, 1.8-20.8). Having medical physicists establish protocols was associated with decreased odds of exceeding 75th percentile of registry CTDIvol (OR, 0.09; 95% CI, 0.01-0.59). Institutions reporting protocol updates as needed had 27% higher mean CTDIvol (MDD, 27%; 95% CI, 8%-45%).Conclusions and relevanceFacilities varied in LCS CT radiation dose distributions. Institutions limiting protocol creation to lead radiologists and having internal medical physicists had lower doses.

Dataset Information

Implications of Nine Risk Prediction Models for Selecting Ever-Smokers for Computed Tomography Lung Cancer Screening.

Publications

Implications of Nine Risk Prediction Models for Selecting Ever-Smokers for Computed Tomography Lung Cancer Screening.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets