Health Economics Modeling Workshop - 3. Incremental cost-effectiveness anlaysis

Use of CEA in two situations

Shopping Spree: Decision problem has non-competing programs/interventions.

Each program is compared to a null alternative; therefore, you’re calculating an “average” cost-effectiveness ratio.

Use of CEA in two situations

Competing Choice: Decision problem has competing programs/interventions for the same purpose; these choices are mutually exclusive.

Two or more active alternatives in addition to the null option.
You need to calculate an “incremental cost- effectiveness ratio”, which gives us the added cost per unit of added benefit of an option, relative to the next less expensive choice

Non-Competing (Shopping Spree) Decision Problem

How can we measure the relative priority of various health programs that compete for limited resources?

Cardiovascular disease program
Safe motherhood program
HIV prevention initiative
Child vaccination
Depression screening

Assumptions

Program alternatives are assumed to be independent
Budget constraint is only limitation
Neither the net cost nor the net effectiveness depend on what other programs are selected
Programs are assumed to be divisible [programs can be partially implemented]

Objectives: Shopping Spree Problem

Maximize the total net effectiveness (health benefit) of the programs selected.

Stay within budget.

Shopping Spree Problem

Rule out programs that cost $ but have negative health effects
Dominated by alternative of “no program”

Shopping Spree Problem

Select programs that are cost-saving & offer benefit – net savings can also be added to budget
Cost-saving compared to alternative of no program

Shopping Spree Problem

Rank other programs in ascending order by their cost-effectiveness ratio (lowest to highest).
Programs are then selected from the LEAST to the MOST expensive until the budget is expended.
Final array of programs selected will depend on the budget constraint.

Shopping Spree Problem

Program	Cost	QALYs	Status
A	27	30
B	30	20
C	56	70
D	20	40
E	30	50
F	50	75
G	40	-30	Ruled Out
H	-20	20	Adopted

Shopping Spree Problem

Initial budget: $80
Budget savings: $20.
Total budget: $80 + $20 = $100

Program	Cost	QALYs	Status
A	27	30
B	30	20
C	56	70
D	20	40
E	30	50
F	50	75
G	40	-30	Ruled Out
H	-20	20	Adopted

Shopping Spree Problem

Calculate average cost-effectiveness ratio.

Program	Cost	QALYs	C/E
A	27	30	0.90
B	30	20	1.50
C	56	70	0.80
D	20	40	0.50
E	30	50	0.60
F	50	75	0.67

Shopping Spree Problem

Calculate average cost-effectiveness ratio.
Sort (by C/E) in ascending order .

Program	Cost	QALYs	C/E
D	20	40	0.50
E	30	50	0.60
F	50	75	0.67
C	56	70	0.80
A	27	30	0.90
B	30	20	1.50

Shopping Spree Problem

Calculate cumulative costs
Determine what is adoptable based on global budget constraint ($100)
Calculate cumulative effects (QALYs)

Shopping Spree Problem

Budget: $100

Program	Cost	QALYs	C/E	Cumulative Cost	Cumulative QALYs
D	20	40	0.50	20	40
E	30	50	0.60	50	90
F	50	75	0.67	100	165
C	56	70	0.80	156	235
A	27	30	0.90	183	265
B	30	20	1.50	213	285

Shopping Spree Problem

Budget: $100

Program	Cost	QALYs	C/E	Cumulative Cost	Cumulative QALYs
D	20	40	0.50	20	40
E	30	50	0.60	50	90
F	50	75	0.67	100	165
C	56	70	0.80	156	235
A	27	30	0.90	183	265
B	30	20	1.50	213	285

Budget	Adopted	Effect	Threshold
100	D, E, F, H	165	0.67

Shopping Spree Problem

Budget: $150

Program	Cost	QALYs	C/E	Cumulative Cost	Cumulative QALYs
D	20	40	0.50	20	40
E	30	50	0.60	50	90
F	50	75	0.67	100	165
C	56	70	0.80	156	235
A	27	30	0.90	183	265
B	30	20	1.50	213	285

Budget	Adopted	Cost	Effect	Threshold	Remaining
150	D, E, F, H	100	165	0.67	50

Shopping Spree Problem

Budget: $150

Program	Cost	QALYs	C/E	Cumulative Cost	Cumulative QALYs
D	20	40	0.50	20	40
E	30	50	0.60	50	90
F	50	75	0.67	100	165
C	56	70	0.80	156	235
A	27	30	0.90	183	265
B	30	20	1.50	213	285

Budget	Adopted	Cost	Effect	Threshold	Remaining
150	D, E, F, H	100	165	0.67	50

Shopping Spree Problem

Budget: $150

Program	Cost	QALYs	C/E	Cumulative Cost	Cumulative QALYs
D	20	40	0.50	20	40
E	30	50	0.60	50	90
F	50	75	0.67	100	165
C (89.3%)	56	70	0.80	156	235
A	27	30	0.90	183	265
B	30	20	1.50	213	285

Budget	Adopted	Cost	Effect	Threshold	Remaining
150	D, E, F, C (89.3%), H	150	226.6	0.8	0

Summary: Shopping Spree Problem

Maximize the total net effectiveness (health benefit) .

Stay within budget.

Can do the same with other objectives (e.g., Minimize costs, subject decision to ‘minimum benefit’ constraint, etc.)

Use of CEA in two situations

Shopping Spree: Decision problem has non-competing programs/interventions.

Use of CEA in two situations

Competing Choice: Decision problem has competing programs/interventions for the same purpose; these choices are mutually exclusive.

Objectives: Competing Choice Problem

Cannot implement more than one strategy at a time.

Incremental cost-effectiveness ratio is below a pre-specified adoption threshold.

What’s different?

Shopping Spree

Can select multiple programs.
Different costs & effects associated with each.
Requires calculation of an Average Cost-Effectiveness Ratio

Competing Choice

Programs are mutually exclusive.
Different costs & effects associated with each.
Requires calculation of an Incremental Cost-Effectiveness Ratio (ICER).

Incremental CEA

Calculate costs and effects for each strategy.
Sort table by costs in ascending order.¹
Calculate ICER based on difference in costs and effects.
Determine dominated strategies (ICER<0).
Re-calculate ICERs after eliminating dominated strategies.
Determine strategies ruled out by extended dominance.
Re-calculate ICERs after ruling out all dominated strategies.
Repeat 5-7 as needed.

Incremental CEA

1. Calculate costs and effects for each strategy.

Strategy	Cost	QALYs
A	16,454	17.33
D	24,504	17.49
C	33,443	17.58
B	21,457	17.41
E	43,332	17.49

Incremental CEA

Calculate costs and effects for each strategy.

2. Sort table by costs in ascending order.¹

Strategy	Cost	QALYs
A	16,454	17.332
B	21,457	17.409
D	24,504	17.491
C	33,443	17.580
E	43,332	17.491

Incremental CEA

Calculate costs and effects for each strategy.
Sort table by costs in ascending order.¹

3. Calculate ICER based on difference in costs and effects.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332	9,888.4	17.491	-0.08825	-112,048

Incremental CEA

Calculate costs and effects for each strategy.
Sort table by costs in ascending order.¹
Calculate ICER based on difference in costs and effects.

4. Determine dominated strategies (ICER<0)

Determining Dominated Strategies

Let’s take a look at our table.
Notice that strategy E has a negative ICER. Why is this?
Strategy E raises costs but lowers QALYs.
Therefore, we’d be better off by selecting strategy C (we would get more health gain for less money…)

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332	9,888.4	17.491	-0.08825	-112,048

Determining Dominated Strategies

Strong dominance refers to situations where one strategy is preferred over another on both costs and health effects (e.g., QALYs).
When we identify a strongly dominated option, we remove it from the table and re-calculate ICERS based on the remaining strategies.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332	9,888.4	17.491	-0.08825	-112,048	Dominated

Incremental CEA

Calculate costs and effects for each strategy.
Sort table by costs in ascending order.¹
Calculate ICER based on difference in costs and effects.

4. Determine dominated strategies (ICER<0)

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332	9,888.4	17.491	-0.08825	-112,048	Dominated

Incremental CEA

Calculate costs and effects for each strategy.
Sort table by costs in ascending order.¹
Calculate ICER based on difference in costs and effects.

Determine dominated strategies (ICER<0).

5. Re-calculate ICERs after eliminating dominated strategies.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332		17.491		-112,048	Dominated

Incremental CEA

Calculate costs and effects for each strategy.
Sort table by costs in ascending order.¹
Calculate ICER based on difference in costs and effects.

Determine dominated strategies (ICER<0).
Re-calculate ICERs after eliminating dominated strategies.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332		17.491

Determining Dominated Strategies

We’re not quite done yet.
Notice something odd about strategy B?
It’s ICER is higher than the next most costly alternative (strategy D).
Let’s look briefly at each strategy’s Cost and QALYs plotted in a graph.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332		17.491

Determining Dominated Strategies

Suppose strategy A is the status quo, and we may be willing to implement a strategy that increases health at some cost.
Is our next best option strategy B?
Let’s zoom in for a minute on strategies A, B, and D.

Determining Dominated Strategies

Perhaps instead of implementing B, we could implement some fraction of strategy D and obtain better health outcomes at lower cost?

Determining Dominated Strategies

If we partially implement strategy D (e.g., A for 99% of the population, D for 1%) we would have a strategy that falls on a line connecting A to D.
What happens at a strategy of roughly 50% strategy A and 50% Strategy D?

Determining Dominated Strategies

A strategy of B for 54% of the population and A for 46% of the population yields greater health benefit at lower cost than B!
Therefore, strategy B is dominated through a concept known as extended dominance.

Determining Dominated Strategies

Strategies in the interior of the cost-effectiveness frontier can be ruled out due to dominance (strong or extended)

Determining Dominated Strategies

A telltale sign of extended dominance in a (sorted) CEA table is a strategy with a higher ICER than the next most expensive option.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895	Dominated (Extended)
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332		17.491			Dominated

Incremental CEA

Calculate costs and effects for each strategy.
Sort table by costs in ascending order.¹
Calculate ICER based on difference in costs and effects.

Determine dominated strategies (ICER<0).
Re-calculate ICERs after eliminating dominated strategies.

6. Determine strategies ruled out by extended dominance.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
B	21,457	5,002.6	17.409	0.07709	64,895	Dominated (Extended)
D	24,504	3,047.5	17.491	0.08239	36,989
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332		17.491			Dominated

Incremental CEA

7. Re-calculate ICERs after ruling out all dominated strategies.

Strategy	Cost	dCost	QALYs	dQALYs	ICER
A	16,454		17.332
D	24,504	8,050.1	17.491	0.15948	50,478
C	33,443	8,939.2	17.580	0.08825	101,292
E	43,332		17.491			Dominated
B	21,457		17.409			Dominated (Extended)

CEA Thresholds

So now we have our ICERs, but how do we make a decision?
We must define a threshold – a value that determines whether or not we implement a given strategy.
What are common threshold and how are they determined?

CEA Thresholds

Decision should be informed by the value of what will be given up as a consequence of those cost.
- Known as the “opportunity cost.”
If resources are committed to the funding of one intervention, then they are not available to fund and deliver others.
The opportunity cost of a commitment of resources is the health forgone because these “other” interventions that are available to the health system cannot be delivered.
Source: Woods et al. (2016)

CEA Thresholds in LMICs

WHO: intervention is “very cost-effective” if cost per DALY averted is less than GDP per capita (GDPpc).
- “Cost effective” if DALY averted is less than 3x GDPpc.

CEA Thresholds in LMICs

Some have argued the WHO’s guidelines may be too high and result in adoption of interventions that displace existing services that provide greater health benefit.
- Suggest 0.5 GDPpc is a more appropriate benchmark for low-income countries and 0.71 GDPpc for middle-income countries.