Selection of eligible participants for screening for lung cancer using primary care data

Lung cancer screening is effective if offered to people at increased risk of the disease. Currently, direct contact with potential participants is required for evaluating risk. A way to reduce the number of ineligible people contacted might be to apply risk-prediction models directly to digital primary care data, but model performance in this setting is unknown.MethodThe Clinical Practice Research Datalink, a computerised, longitudinal primary care database, was used to evaluate the Liverpool Lung Project V.2 (LLPv2) and Prostate Lung Colorectal and Ovarian (modified 2012) (PLCOm2012) models. Lung cancer occurrence over 5–6 years was measured in ever-smokers aged 50–80 years and compared with 5-year (LLPv2) and 6-year (PLCOm2012) predicted risk.ResultsOver 5 and 6 years, 7123 and 7876 lung cancers occurred, respectively, from a cohort of 842 109 ever-smokers. After recalibration, LLPV2 produced a c-statistic of 0.700 (0.694–0.710), but mean predicted risk was over-estimated (predicted: 4.61%, actual: 0.9%). PLCOm2012 showed similar performance (c-statistic: 0.679 (0.673–0.685), predicted risk: 3.76%. Applying risk-thresholds of 1% (LLPv2) and 0.15% (PLCOm2012), would avoid contacting 42.7% and 27.4% of ever-smokers who did not develop lung cancer for screening eligibility assessment, at the cost of missing 15.6% and 11.4% of lung cancers.ConclusionRisk-prediction models showed only moderate discrimination when applied to routinely collected primary care data, which may be explained by quality and completeness of data. However, they may substantially reduce the number of people for initial evaluation of screening eligibility, at the cost of missing some lung cancers. Further work is needed to establish whether newer models have improved performance in primary care data.

Download Full-text

Can We and Should We Produce a model for Emergency Admissions derived from Primary Care Data?

International Journal for Population Data Science ◽

10.23889/ijpds.v1i1.295 ◽

2017 ◽

Vol 1 (1) ◽

Author(s):

Kerry Bailey

Keyword(s):

Primary Care ◽

Statistical Model ◽

Risk Prediction ◽

Prediction Models ◽

Structured Interview ◽

Added Value ◽

Emergency Admissions ◽

Risk Prediction Models ◽

Primary Care Data ◽

Literature Reviews

ABSTRACT BackgroundSeveral Risk Prediction models exist for predicting emergency admissions. One exists derived solely from primary care data but this is only available in England. All the models currently available have limitations. The aim was to utilise the SAIL databank to develop a predictive model/tool using only primary care data that has comparable accuracy to the currently used risk stratification tools but overcomes the limitations of the current models available. ApproachA multidimensional approach was taken including literature reviews and qualitative interviews to the model to situate the statistical model. The need was identified to develop the contextual background in order to situate the statistical model development and understand the need for an improved model. Stakeholder groups were identified – patients, managers, General Practitioners and policy makers – and purposive sampling was used to obtain breadth and depth of representation into the subject area. Interviews were carried out in two stages initially to identify key themes and the breadth of the interest in stakeholder and influenced hypothesis generation. The second stage was to provide depth and clarity on specific themes with a more structured interview schedule. Results of the qualitave work triangulated with literature reviews influenced the statistical model build. DiscussionThe statistical challenge of building Risk Prediction models well documented. Understanding the shortcoming of current models as perceived by stakeholders is necessary to produce a model of added value which performs statistically similar however has a different functionality. The qualitative work that was carried out to situate this model added value to the models development but also raised some interesting ethical issues. The validity and performance of the models although important is only one aspect that influences whether GPs would use a risk prediction model and how acceptable it is to stakeholders. None of the patients interviewed in this thought that their data was, or would be without their knowledge, used in risk prediction models yet all were patients at practices where this was happening or had morbidities which would have necessitated the use of several in primary care.

Download Full-text

Faculty Opinions recommendation of Risk prediction models for selection of lung cancer screening candidates: A retrospective validation study.

Faculty Opinions – Post-Publication Peer Review of the Biomedical Literature ◽

10.3410/f.727476053.793541151 ◽

2018 ◽

Author(s):

Robert Keith ◽

Melissa New

Keyword(s):

Lung Cancer ◽

Cancer Screening ◽

Risk Prediction ◽

Validation Study ◽

Prediction Models ◽

Lung Cancer Screening ◽

Risk Prediction Models ◽

Selection Of

Download Full-text

A Comparative Modeling Analysis of Risk-Based Lung Cancer Screening Strategies

JNCI Journal of the National Cancer Institute ◽

10.1093/jnci/djz164 ◽

2019 ◽

Vol 112 (5) ◽

pp. 466-479 ◽

Cited By ~ 13

Author(s):

Kevin ten Haaf ◽

Mehrad Bastani ◽

Pianpian Cao ◽

Jihyoun Jeon ◽

Iakovos Toumazis ◽

...

Keyword(s):

Lung Cancer ◽

Cancer Screening ◽

Risk Prediction ◽

Prediction Models ◽

Lung Cancer Screening ◽

Comparative Modeling ◽

Risk Prediction Models ◽

Life Years ◽

Screening Strategies

Abstract Background Risk-prediction models have been proposed to select individuals for lung cancer screening. However, their long-term effects are uncertain. This study evaluates long-term benefits and harms of risk-based screening compared with current United States Preventive Services Task Force (USPSTF) recommendations. Methods Four independent natural history models were used to perform a comparative modeling study evaluating long-term benefits and harms of selecting individuals for lung cancer screening through risk-prediction models. In total, 363 risk-based screening strategies varying by screening starting and stopping age, risk-prediction model used for eligibility (Bach, PLCOm2012, or Lung Cancer Death Risk Assessment Tool [LCDRAT]), and risk threshold were evaluated for a 1950 US birth cohort. Among the evaluated outcomes were percentage of individuals ever screened, screens required, lung cancer deaths averted, life-years gained, and overdiagnosis. Results Risk-based screening strategies requiring similar screens among individuals ages 55–80 years as the USPSTF criteria (corresponding risk thresholds: Bach = 2.8%; PLCOm2012 = 1.7%; LCDRAT = 1.7%) averted considerably more lung cancer deaths (Bach = 693; PLCOm2012 = 698; LCDRAT = 696; USPSTF = 613). However, life-years gained were only modestly higher (Bach = 8660; PLCOm2012 = 8862; LCDRAT = 8631; USPSTF = 8590), and risk-based strategies had more overdiagnosed cases (Bach = 149; PLCOm2012 = 147; LCDRAT = 150; USPSTF = 115). Sensitivity analyses suggest excluding individuals with limited life expectancies (<5 years) from screening retains the life-years gained by risk-based screening, while reducing overdiagnosis by more than 65.3%. Conclusions Risk-based lung cancer screening strategies prevent considerably more lung cancer deaths than current recommendations do. However, they yield modest additional life-years and increased overdiagnosis because of predominantly selecting older individuals. Efficient implementation of risk-based lung cancer screening requires careful consideration of life expectancy for determining optimal individual stopping ages.

Download Full-text

Assessing the Predictive Validity of Simple Dementia Risk Models in Harmonized Stroke Cohorts

Stroke ◽

10.1161/strokeaha.120.027473 ◽

2020 ◽

Vol 51 (7) ◽

pp. 2095-2102

Author(s):

Eugene Y.H. Tang ◽

Christopher I. Price ◽

Louise Robinson ◽

Catherine Exley ◽

David W. Desmond ◽

...

Keyword(s):

Risk Factors ◽

General Population ◽

Risk Prediction ◽

Prediction Models ◽

Disease Risk ◽

Risk Index ◽

Area Under The Curve ◽

Risk Prediction Models ◽

C Statistic ◽

Dementia Risk

Background and Purpose: Stroke is associated with an increased risk of dementia. To assist in the early identification of individuals at high risk of future dementia, numerous prediction models have been developed for use in the general population. However, it is not known whether such models also provide accurate predictions among stroke patients. Therefore, the aim of this study was to determine whether existing dementia risk prediction models that were developed for use in the general population can also be applied to individuals with a history of stroke to predict poststroke dementia with equivalent predictive validity. Methods: Data were harmonized from 4 stroke studies (follow-up range, ≈12–18 months poststroke) from Hong Kong, the United States, the Netherlands, and France. Regression analysis was used to test 3 risk prediction models: the Cardiovascular Risk Factors, Aging and Dementia score, the Australian National University Alzheimer Disease Risk Index, and the Brief Dementia Screening Indicator. Model performance or discrimination accuracy was assessed using the C statistic or area under the curve. Calibration was tested using the Grønnesby and Borgan and the goodness-of-fit tests. Results: The predictive accuracy of the models varied but was generally low compared with the original development cohorts, with the Australian National University Alzheimer Disease Risk Index (C-statistic, 0.66) and the Brief Dementia Screening Indicator (C-statistic, 0.61) both performing better than the Cardiovascular Risk Factors, Aging and Dementia score (area under the curve, 0.53). Conclusions: Dementia risk prediction models developed for the general population do not perform well in individuals with stroke. Their poor performance could have been due to the need for additional or different predictors related to stroke and vascular risk factors or methodological differences across studies (eg, length of follow-up, age distribution). Future work is needed to develop simple and cost-effective risk prediction models specific to poststroke dementia.

Download Full-text