Improving analytical methods for reducing selection bias in health economic evaluation
Lead Research Organisation:
London School of Hygiene & Tropical Medicine
Department Name: Public Health and Policy
Abstract
Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
People |
ORCID iD |
Richard Grieve (Principal Investigator) |
Publications
Grieve R
(2016)
An evaluation of the clinical and cost-effectiveness of alternative care locations for critically ill adult patients with acute traumatic brain injury.
in British journal of neurosurgery
Hartman E
(2015)
From Sample Average Treatment Effect to Population Average Treatment Effect on the Treated: Combining Experimental with Observational Studies to Estimate Population Treatment Effects
in Journal of the Royal Statistical Society Series A: Statistics in Society
Keele L
(2017)
Randomization-based Instrumental Variables methods for Binary outcomes with an Application to the 'IMPROVE' trial
in Journal of the Royal Statistical Society Series A: Statistics in Society
Kreif N
(2013)
Statistical methods for cost-effectiveness analyses that use observational data: a critical appraisal tool and review of current practice.
in Health economics
Kreif N
(2016)
Examination of the Synthetic Control Method for Evaluating Health Policies with Multiple Treated Units.
in Health economics
Kreif N
(2015)
Evaluation of the Effect of a Continuous Treatment: A Machine Learning Approach with an Application to Treatment for Traumatic Brain Injury.
in Health economics
Kreif N
(2016)
Evaluating treatment effectiveness under model misspecification: A comparison of targeted maximum likelihood estimation with bias-corrected matching.
in Statistical methods in medical research
Kreif N
(2012)
Methods for estimating subgroup effects in cost-effectiveness analyses that use observational data.
in Medical decision making : an international journal of the Society for Medical Decision Making
Description | 1. Our literature review found that studies that compare the costs and outcomes of alternative treatments rarely assess the main assumptions behind statistical analyses for addressing selection bias (Kreif et al, 2012a). The studies reviewed did not consider whether it was reasonable to assume that there were no unmeasured factors that may explain the study's findings. The studies did not assess whether the statistical model chosen to try and reduce bias was reasonable. We developed a checklist of issues that researchers need to address to reduce selection bias in cost-effectiveness studies. This checklist can improve the design of future studies and help ensure that they are analysed and interpreted more carefully. 2. We undertook several simulation studies to compare alternative statistical approaches for reducing selection bias. In these simulation studies, we generated data that suffered from selection bias and assessed how well each method performed compared with the true estimates. We considered Genetic Matching, an approach that attempts to ensure that the individuals being compared across treatments are as similar as possible. We compared Genetic Matching to more traditional approaches that use a formal statistical model, the propensity score, to match or weight the data. We found Genetic Matching gave less biased estimates of the effectiveness (Radice et al, 2012), and cost-effectiveness of alternative treatments than traditional methods (Sekhon and Grieve, 2011; Kreif et al, 2012b). 3. We considered how well the methods for addressing selection bias worked in practice. We compared Genetic Matching with traditional approaches in several real-life examples that assessed the relative costs and health outcomes following alternative medical treatments. We found that Genetic Matching ensured more similar individuals were compared across the treatment groups than traditional methods (Sekhon and Grieve, 2011; Sadique et al., 2011; Noah et al. 2011). The choice of method made a substantive difference to the studies' findings; the estimates of the average cost-effectiveness of the interventions, and the uncertainty of the estimates differed by method. We illustrated how Genetic Matching can be extended to ensure the most important patient characteristics are similar between the comparison groups (Ramsahai et al., 2011). 4. The research also considered alternative approaches for estimating the relative effectiveness and cost-effectiveness of new treatments for subgroups of patients. We found that Genetic Matching provided less biased estimates of the effectiveness and cost-effectiveness of alternative treatments for particular patient subgroups compared to traditional approaches (Radice et al., 2012; Kreif et al., 2012b). In a case study that evaluated a new treatment for patients with severe sepsis who were admitted to intensive care, we found that the choice of method had a substantial impact on the cost-effectiveness reported for each patient subgroup (Kreif et al. 2012a). 5. We developed a course for researchers working in academia and industry to explain the main issues that arise when trying to reduce selection bias. We explained how Genetic Matching can reduce selection bias compared with traditional approaches; we provided practical examples showing how to apply this method. |
Sectors | Digital/Communication/Information Technologies (including Software),Education |
URL | http://www.lshtm.ac.uk/php/hsrp/reducing-selection-bias/index.html |
Description | Analysis of patient reported outcome measures |
Amount | £817,000 (GBP) |
Organisation | Government of Wales |
Department | Department of Health |
Sector | Public |
Country | United Kingdom |
Start | 01/2009 |
End | 12/2013 |
Description | DH |
Amount | £2,000,000 (GBP) |
Organisation | Department of Health (DH) |
Sector | Public |
Country | United Kingdom |
Start | 07/2015 |
End | 06/2020 |
Description | DH |
Amount | £170,000 (GBP) |
Organisation | Department of Health (DH) |
Department | Policy Research Programme (PRP) |
Sector | Public |
Country | United Kingdom |
Start | 05/2017 |
End | 05/2021 |
Description | Policy innovation research unit |
Amount | £5,000,000 (GBP) |
Organisation | Government of Wales |
Department | Department of Health |
Sector | Public |
Country | United Kingdom |
Start | 01/2011 |
End | 12/2016 |
Title | Costs and outcomes of the use of drotrecogin alfa (DrotAA) for patients with severe sepsis and multiple organ systems failing |
Description | Health policy-makers use cost-effectiveness analyses (CEAs) to evaluate the relative costs and effectiveness of alternative ways of providing health care. Many CEAs use observational data, where the key concern is how to adjust for imbalances in baseline covariates due to the non-random assignment of the health care programs under evaluation. This research aims to improve CEA methods for addressing selection bias. We evaluate the impact of different statistical approaches on the estimated cost-effectiveness using case studies. In particular, we compare a new non-parametric method (Genetic Matching) to more traditional methods, for example propensity score (PS) matching. We present a CEA that use observational data from the Intensive Care National Audit Research Centre (ICNARC) Case Mix Program (CMP) database. The study (Kreif et al., 2012) evaluates Drotrecogin alfa (DrotAA) for patients with severe sepsis and multiple organ systems failing. Effectiveness was evaluated using hospital mortality, and age and gender-adjusted quality-adjusted life years (QALYs). Costs of procedures and hospitalisations were collected at the patient level. The dataset includes patient level records for 2726 individuals (1,076 DrotAA and 1,650 controls). The dataset includes variables recording 3 outcomes measures (hospital mortality, QALYs, total cost), and a range of case-mix variables. |
Type Of Material | Database/Collection of data |
Year Produced | 2012 |
Provided To Others? | Yes |
Description | Methods for addressing selection bias in health economic evaluation |
Organisation | Harvard University |
Department | Harvard T.H. Chan School of Public Health |
Country | United States |
Sector | Academic/University |
PI Contribution | This research has created an international collaboration with an expert methodologist, Susan Gruber based at Harvard School of Public Health (HSPH). Susan has visited the London School of Hygiene and Tropical medicine in July 2012. She gave a guest lecture as part of the short course ?Methods for addressing selection bias in health economic evaluation? (see details under outcome ?Methods for addressing selection bias in economic evaluation?, outcome type ?organised events?). Susan also gave a successful seminar talk at the Department of Statistics, and collaborated on a methodological article lead by Noemi Kreif, a PhD student funded by the grant (see ?Upgrade from MPhil to DPhil?, outcome type ?staff development?). This work involved applying cutting edge statistical methods to estimate treatment effectiveness, using patient reported outcome measure data (see outcome ?Analysis of patient reported outcome measures?, outcome type ?Further Funding?). Susan Gruber invited Noemi Kreif to give a seminar talk at the HSPH (see outcome ?An examination of doubly robust methods and bias-corrected matching for cost-effectiveness analysis?, outcome type ?Lecture?). |
Start Year | 2011 |
Description | An examination of doubly robust methods and bias-corrected matching for cost-effectiveness analysis |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Primary Audience | |
Results and Impact | Presented at Causal Inference Group Meeting, Harvard School of Public Health, 02 11 2011. |
Year(s) Of Engagement Activity | 2011 |
Description | Estimating subgroup effects in cost-effectiveness analyses (CEA) : a comparison of methods |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Primary Audience | |
Results and Impact | Seminar for Health Economics Research Unit, Aberdeen |
Year(s) Of Engagement Activity |