Involved in the continued development of the field of decision science, members of the Center for Health Decision Science focus on theoretical and methodological issues. This work encompasses research in decision theory as well as the development of methodologies for applied research. Additionally, novel and creative uses of existing methodologies from other disciplines applied to decision analytic problems, characterize and support a variety of these endeavors.
Theoretical work in decision theory and outcomes research
-
Exploring theoretical relationship between monetary and health-utility based methods of valuing health risk and how these measures depend on characteristics of the individual (e.g., age, health, income) and the health risk (e.g., disease type, controllability). Members have contributed to the development of quality-adjusted life years (QALYs), disability-adjusted life years (DALYs), and value of a statistical life years (VSLYs).
-
Methods to elicit and measure health. These include development and application of revealed- and stated-preference methods for characterizing preferences over health risks, with applications to high- and low-income countries. Measurement issues in subpopulations such as children are also considered.
-
Ethical considerations such as distributional equity for research allocation.
Methodologies for the construction of decision models
-
Methods for simulation and optimization such as discrete-event simulation and linear programming applied to population health. Researchers utilize a wide-range of modeling techniques including Markov models, state-transition models, discrete-event simulation, first-order Monte Carlo simulation, and dynamic transmission models. As each technique is suitable for answering specific policy questions, it is important to understand the strengths and weaknesses of each modeling method.
-
Methods to reflect parameter uncertainty in simulation models that consider a large number of alternative strategies.
-
Methods for variance reduction and computational efficiency in large-scale disease simulation models. We employ methods from engineering and operations research to enhance the efficiency of our simulation models.
-
Integrating different methods from operations research, including linear programming, to explicitly consider non-monetary constraints (such as a shortage of human resources or equipment) in addition to monetary constraints, and to quantify their impacts on program effectiveness, feasibility, and cost-effectiveness.
Parameterization, calibration and validation practices for decision models
-
Methods to combine subjective probability distributions, such as those elicited from multiple experts, for use in subsequent analyses. Additionally, meta-analytic and Bayesian techniques are used to synthesis data for model parameterization.
-
Methods for empirical calibration of model input parameters to multiple sources of epidemiological data, including likelihood-based and Bayesian approaches, for various types of decision-analytic models, and validation of these models using independent empirical data from heterogeneous sources.
- Detailed modeling of the underlying natural history of disease. Through calibration and validation methods, these models can be used to examine hypotheses about the often unknown biologic process of disease onset and progression.
Use and combination of epidemiologic and economic data
-
Methods for incorporating different levels of interaction among individuals in a population in infectious disease simulation models.
-
Use of multi-country epidemiologic data to gain additional insights into the natural history of disease and effects of risk factors across countries and through time.
-
Development of life tables specific for segments of populations with different risk factor status. Examples include splitting the overall US life table by body mass index and smoking status, as a way to document the joint impact of obesity and smoking on life expectancy.
- Methods for the estimation of full economic cost of health services. Examples include estimation of direct non-medical costs such as patient transport and time costs, laboratory transport costs, loss to follow-up and maintaining programs at high levels.
Development of guidelines for best practice
-
Members serve on a wide-range of US and international panels and committees aimed at building consensus about and fostering standardization of methodologies. Notably, members have participated in the Panel on Cost-Effectiveness in Health and Medicine, Institute of Medicine panels on health outcome measures, and the International Society for Pharmacoeconomics and Outcomes Research Task Force on Good Research Practices for Modeling Studies.
-
Methods for reporting results from cost-effectiveness and benefit-cost analyses. Examples include methods for probabilistic cost-effectiveness analyses and other techniques to reflect uncertainty in model outcomes.