Economic evaluation of health programmes Department of Epidemiology, Biostatistics and Occupational Health Class no. 25: Dec 1, 2008
Plan of class Discussion of assignment no 4 Using PBMA; Multi-criteria decision analysis Course wrap-up
Question 1: Additional assumptions If treatment at ER is unsuccessful, and only if it is unsuccessful, person is hospitalized – prob = Following relief (however obtained), 48 hours elapse until a new attack may occur If recurrence occurs, it does so according to probabilities previously specified
Comments on question 1 24 pathways (see diagrams) Taking time into account – see spreadsheet Prorate cost of hospitalization? Atypical decision tree analysis: Time built-in; utility “payoffs” depend on relative time duration in different states See example next slide
Example from which assignment drawn: Briggs, Claxton, Sculpher, Decision Modelling for Health Economic Evaluation, Oxford, 2006
Comments on question 2 Straightforward calculations Small difference in transition probabilities + most people staying well leads to small difference in utility/QALYs; cost difference driven by difference in cost of meds Best modelling approach depends greatly on extent to which past events influence likelihood of future ones
Using Programme Budgeting and Marginal Analysis and Multi-criteria decision analysis
The problem NICE’s problem described last time Cost-effectiveness insufficient, scale, regional differences and other factors affect true opportunity cost Problem magnified at local level Even if you have national guidelines, not clear just how to implement them at the local level Many possible criteria – how to weigh them?
Source: Peacock and Ruta, 2006
Four types of analysis Evidence-based medicine What works? Burden of disease analysis Cost-effectiveness analysis Equity analysis Distributional impact – to what extent do the poor or other disadvantaged groups benefit compared to better-off groups? Types of analysis developed separately from each other
Ad hoc priority setting Source: Baltussen and Niessen, 2006 Intuition is inadequate to process all this information rationally
Rational priority setting Methods exist to analyze this information systematically Source: Baltussen and Niessen, 2006
Analysing a performance matrix Qualitative E.g., look for dominance Quantitative Construct scales to represent preferences for consequences (so programs can be compared dimension by dimension) Weight the scales for relative importance Calculate weighted averages across scales Often done using linear additive model if reasonable to think criteria preferentially independent of each other
Source: Baltussen and Niessen, 2006
Questions for discussion What are the key ideas that struck you about the course? Anything that really interested you? What topics more effectively taught, what less well? What made the difference? Comments on assignments? Nature and quantity of work outside class?