Björn Stollenwerk1, Afschin Gandjour, Markus Lüngen, Uwe Siebert. 1. Helmholtz Zentrum München (GmbH), Institute of Health Economics and Health Care Management, Ingolstädter Landstraße 1, 85764, Neuherberg, Germany, bjoern.stollenwerk@helmholtz-muenchen.de.
Abstract
BACKGROUND: Positive screening results are often associated not only with target-disease-specific but also with non-target-disease-specific mortality. In general, this association is due to joint risk factors. Cost-effectiveness estimates based on decision-analytic models may be biased if this association is not reflected appropriately. OBJECTIVE: To develop a procedure for quantifying the degree of bias when an increase in non-target-disease-specific mortality is not considered. METHODS: We developed a family of parametric functions that generate hazard ratios (HRs) of non-target-disease-specific mortality between subjects screened positive and negative, with the HR of target-disease-specific mortality serving as the input variable. To demonstrate the efficacy of this procedure, we fitted a function within the context of coronary artery disease (CAD) risk screening, based on HRs related to different risk factors extracted from published studies. Estimates were embedded into a decision-analytic model, and the impact of 'modelling increased non-target-disease-specific mortality' was assessed. RESULTS: In 55-year-old German men, based on a risk screening with 5% positively screened subjects, and a CAD risk ratio of 6 within the first year after screening, incremental quality-adjusted life-years were 19% higher and incremental costs were 8% lower if no adjustment was made. The effect varied depending on age, gender, the explanatory power of the screening test and other factors. CONCLUSION: Some bias can occur when an increase in non-target-disease-specific mortality is not considered when modelling the outcomes of screening tests.
BACKGROUND: Positive screening results are often associated not only with target-disease-specific but also with non-target-disease-specific mortality. In general, this association is due to joint risk factors. Cost-effectiveness estimates based on decision-analytic models may be biased if this association is not reflected appropriately. OBJECTIVE: To develop a procedure for quantifying the degree of bias when an increase in non-target-disease-specific mortality is not considered. METHODS: We developed a family of parametric functions that generate hazard ratios (HRs) of non-target-disease-specific mortality between subjects screened positive and negative, with the HR of target-disease-specific mortality serving as the input variable. To demonstrate the efficacy of this procedure, we fitted a function within the context of coronary artery disease (CAD) risk screening, based on HRs related to different risk factors extracted from published studies. Estimates were embedded into a decision-analytic model, and the impact of 'modelling increased non-target-disease-specific mortality' was assessed. RESULTS: In 55-year-old German men, based on a risk screening with 5% positively screened subjects, and a CAD risk ratio of 6 within the first year after screening, incremental quality-adjusted life-years were 19% higher and incremental costs were 8% lower if no adjustment was made. The effect varied depending on age, gender, the explanatory power of the screening test and other factors. CONCLUSION: Some bias can occur when an increase in non-target-disease-specific mortality is not considered when modelling the outcomes of screening tests.
Authors: Ken Stein; Kim Dalziel; Andrew Walker; Becky Jenkins; Alison Round; Pam Royle Journal: J Public Health (Oxf) Date: 2004-03 Impact factor: 2.341
Authors: J Stamler; R Stamler; J D Neaton; D Wentworth; M L Daviglus; D Garside; A R Dyer; K Liu; P Greenland Journal: JAMA Date: 1999-12-01 Impact factor: 56.272
Authors: A M Schott; C Ganne; D Hans; G Monnier; R Gauchoux; M A Krieg; P D Delmas; P J Meunier; C Colin Journal: Osteoporos Int Date: 2006-10-13 Impact factor: 4.507
Authors: Stephanie A K Stock; Björn Stollenwerk; Marcus Redaelli; Daniele Civello; Karl W Lauterbach Journal: J Womens Health (Larchmt) Date: 2008-04 Impact factor: 2.681