Health Economic Evaluation Nusaraporn Kessomboon, MSc(Health Economics), PhD
Content Definitions, principles Methods Some examples
Unique nature of health as a good Non-transferable goods Outcome of an intervention is always uncertain for an individual Supply induce demand Externality
Health Economics Applying economic principles and theories to health and to the health care sector
Health Economic Evaluations Are Just One Part of Health Economics HealthEconomics Health Economic Evaluation other topics in health economics: optimal size of hospitals, optimal payment for physicians, optimal level of co-payment by patients,….
Definition of Health Economic Evaluation The comparative analysis of alternative courses of action in terms of BOTH their costs and health consequences Pharmaco-economic evaluation = if at least one drug is involved
The Different Steps of Evidence Can it work? = Efficacy Does it work in reality? = Effectiveness Efficiency Is it worth doing it, compared to other things we could do with the same money? = Cost-effectiveness = Efficiency
Difficult questions and difficult answers... which services to provide? how much to provide? at what stage in the disease process to provide it? to whom it should be provided?
Economic Evaluation costs (inputs) and consequences (outputs) comparison of two or more alternatives
Decision making Method consequ ence Decision rule CMA Identical Lowest cost CEA ‘Natural units’ Lowest cost per improved outcome CUA Quality adjusted life years (QALYs) Lowest cost per QALY gained CBA BahtHighest net benefits
The use of CE or CU ratios as a decision rule CE ratio = the difference in costs divided by the difference in outcome ∆C = C A - C B ` ∆E QALY A - QALY B
Unit cost determination System analysis NRPCC RPCC PS LC+MC+CC LC+MC+CC LC+MC+CC TDC TDC TDC IDC from NRPCC IDC from RPCC (cost allocation) Full cost of PS = (IDC+DC) Volume of care provided Unit cost
Monetary valuation : benefits Single outcome : effectiveness Multiple outcome : utility Consequence s analysis
Monetary valuation : benefits Human capital approach Revealed preference Stated preference
Revealed preference extra earnings of construction workers in risky occupations over safe occupations not appropriate in the healthcare field due to consumer ignorance and zero or subsidized price at the point of use (Arrow,1963; Culyer,1971; Mooney,1986)
Stated preferenc e Contingent valuation Hypothetical scenarios Conjoint analysis
Single outcome : effectiveness Immediate outcome : symptom free Intermediate outcome : no of ulcer prevented Final outcome : life years saved
Multiple outcome : utility Non-preference- based measures of health status : QOL not utility Preference-based measures of health status : QALYs
Non-preference-based measures of health status standardized questionnaires to assess patient health across broad areas : symptoms, physical functioning, work and social activities, and mental well- being
can be disease- specific or generic can generate a profile of scores, or a single index usually, scoring procedures (e.g.SF-36 assumes equal weighting for most items.) Non-preference-based measures of health status (cont.)
to assess the relative efficiency of interventions in very limited circumstances 3 components to the scoring: (1) equal weighting (e.g.the SF-36) (2) weightings to combine items (3) combined into an overall total score using a set of weigh.(not usually done) Non-preference-based measures of health status (cont.)
For clinical purposes : present separate scores by dimension. Non-preference-based measures of health status (cont.)
Preference-based measures of health status standardized questionnaires assess patient health across broad areas including symptoms, physical functioning, work and social activities, and mental well-being can be disease-specific or generic a single index based on people preferences (e.g.EQ- 5D, HIU)
value of is equivalent to full health 0 is dead known as health state utilities used to calculate quality- adjusted life-years, QALYs Preference-based measures of health status (cont.)
5 preference-based measures of health Quality of Well-Being Scale (QWB) : lengthier interview Rosser’s disability/ distress sale : self- administration Health Utility Index (HUI; mark I to III): self-administration Preference-based measures of health status (cont.)
EQ-5D (EuroQoL) : self- administration EQ-15D : self- administration no consensus amongst health economists as to which is better. Preference-based measures of health status (cont.)
consumer theory predicting the choices of individuals between different bundles of commodities (Deaton and Muelbauer,1980) Theoretical basis of preference-based
assumes individuals choose the bundle of commodities which maximizes utility subject to budget constraint utility is an indicator of the consumer’s strength of preference Theoretical basis of preference-based (cont.)
a person deciding whether or not to purchase health services will consider –the likely effects they are expected to have on their health –whether the benefits of these effects are worth the costs of the health care Theoretical basis of preference-based (cont.)
Trading e.g. have an operation associated with the risk of mortality VS life extending chemotherapy with side effects Theoretical basis of preference-based (cont.)
The main economic theory of decision- making under uncertainty is expected utility theory (EUT) Individuals choose between prospects as to maximize their expected utility (Von Neumann and Morgenstern,1947) Theoretical basis of preference-based (cont.)
Practice of measuring preferences for health Paired Comparison (PC) Visual analogue scale (VAS) Magnitude estimation (ME) Standard gamble (SG) Time trade-off (TTO :Torrance, 1986) Person trade-off (PTO : Nord, 1992)
Visual analogue scale (VAS) Category rating (CR) Rating scale (RS) Visual aids e.g., “feeling thermometer” are used widely used to value health states : QWB, HUI-II and HUI-III transform VAS values into SG
Standard gamble two alternatives 1 : treatment with two possible outcomes: return to normal health and lives for an additional t years(P), or dies immediately (1-P) 2 : has the certain outcome of chronic state i for life (t years) (Torrance, 1986)
Probability P is varied until the respondent is indifferent between the two alternatives, at which point the required preference value for state i is simply P, that is Ui = P (Torrance, 1986) Standard gamble
Time trade-off two alternatives 1:state i for time t (life expectancy of an individual with the chronic condition) followed by death 2: healthy for time x; x < t followed by death (Torrance, 1986)
time x is varied until the respondent is indifferent between the two alternatives, at which point the required preference value for state i is given by Ui = x/t (Torrance, 1986) Time trade-off
Time trade-off (TTO :Torrance, 1986) x*1.0 = t*Ui x / t = Ui
introduced by Klarman et al for chronic renal failure combine into a single interval-scale measure based on relative desirability of the different outcomes (patient preferences) from perfect health to death QALYs = Utility weight x Years of life Quality Adjusted Life Years (QALYs)
Perfect health = With programme 1. Without programme Death = 0.0 Health- related quality of life (Utility weights) Duratio n (Years of life) Death 1 Death 2 Y1Y1 Y2Y2 QALYs gained
QALYs Calculation Duration Health state Weight * 3 monthsHospital dialysis monthsHome confinement for TB yearsHome dialysis yearsMastectomy for breast cancer0.48 * based o TTO on a random sample of the general public (Sackett and Torrance, 1978)
Exercise : QALYs Calculation Sketch QALY diagram and determine QALYs gained assume no discounting
Exercise :QALY Calculation Sketch QALY diagram and determine QALYs gained 1. Eight-year life extension on home dialysis 2. Three-month life extension on hospital dialysis 3. Preventing a TB case treated at home for three months
4. Assume a breast-cancer patient will become symptomatic, have a mastectomy, and live a additional six years. By screening, the cancer will be detected one year earlier, the surgery will be done one year earlier, and the life will be extended two years (compared with no screening). Exercise :QALY Calculation
Issues in CUA Who is the right person to ask to get the utility score? Patients, Physicians, or other. QALYs count quality and quantity of life in a one to one ratio. 2 years in a.5 condition is assumed to be the same as 1 year in perfect health.
Issues in CUA Adding 1 QALY to 75 people is counted the same as adding 75 QALYs to one person. QALYs are counted the same for all people-- young or old, rich or poor. Evaluation with QALYs takes the average QALY answer and applies it to everyone. Note that it is possible that no person gave the average answer.
Using Costs and QALYs This brings us to the task of maximizing QALYs for a given budget. This is done the same way as in CEA. In CEA we covered how to maximize the desired effect for a given budget for one patient population. An example with two patient populations (i.e. two groups with different disease states).
Steps Step 1—Put the Costs and QALYs in a table ordered by Costs. Step 2—Eliminate all dominated and second order dominated treatments within each group. Step 3—Given the budget you allocate money to treatments in order of incremental cost utility ratios. Remember that the budget must cover both groups. Typically you end up splitting treatments in one of the groups— this is done solving two equations and two unknowns exactly as with one group.
Example L16100D K1268C 1260J8 B 1040I1050A QALYsCo sts Gro up2 QAL Ys Cos ts Gro up1
Dominated Treatments There are no dominated treatments in group 2. In group 1, B is dominated, and C is Second Order Dominated. –The incremental cost utility ratio between A and C is 9 and the incremental ratio between C and D is 8. We end up with a final table with incremental cost utility ratios (ICURs) for all of the non-dominated treatments.
Final Table L K J8 1/ D I A ICU R QCG2ICU R QCG1
Example—Allocating Budgets Again you allocate a budget based on the incremental cost utility ratios. You allocate in order of the incremental ratios—and you always spend all the money. Small budget examples--Note that the ICUR of I is 4 and for A it is 5. I costs 40 and A costs 50. Hence for any budget under 40 you buy only as large a percentage of I as you can. For any budget between 40 and 90 you buy I for everyone in Group2 and as much a percentage of A as you can with the remaining money (e.g. with 65 you buy I for everyone in Group2 and A for half the patients in group 1).
One Example of Splitting Take a total budget of 120. At this level you want to give treatment I to all of group 2, costing 40, and then use the remaining 80 to treat group 1. With 80 for group 1 you want to split between A and D. Solve Pa + Pd = 1, Pa(50)+Pd(100) = 80
Answer Pa = 2/5, Pd = 3/5. Overall, group 2 gets I, 2/5 of group 1 gets A and 3/5 of group 1 gets D. –Total cost = /5(50) + 3/5(100) = 120 –Total QALYs = 10 +2/5(10)+3/5(16) = 23 3/5
Allocating Budgets Continued The order of treatments is I, A, D, J, K, L. –First, total budget from 0-40, I goes to part of group 2. –Second, total budget 40-90, I for all of group 2 and A for part of group1. –Third, total budget , I to all of group2 and a mix of A and D to group 1. Note that the budget for group 1 in this range is between 50 and 100. –Fourth, total budget , D to all of group 1 and a mix of I and J to group 2. Note that the budget for group 2 is between 40 and 60.
STEPS TO AN ECONOMIC ANALYSIS Understand and describe the problem Choose appropriate analysis method Collect cost and outcome data Perform analysis Assess sensitivity of results