A B C Cost A Cost B Outcome A Outcome B Cost-effectiveness ratio - - PowerPoint PPT Presentation
R ESULTS PRESENTATION O UTLINES Model transparency and validation (lecture 7) Deterministic results Incremental costeffectiveness ratio (ICER) Costeffectiveness plane and Efficiency frontier Deterministic sensitivity analysis
139
140
CostA OutcomeA
CostB OutcomeB
std i std i
Average cost-effectiveness ratios (ACERs)
Dispute about definitions
Treeage, dividing a therapy’s total costs by its total outcomes Evaluate cost and outcomes of each intervention against its
baseline option (e.g., “do nothing” or current practice).
Incremental cost-effectiveness ratios (ICERs)
Comparison of costs and outcomes among the
alternative options
When there are only 2 options being evaluated,
the average and incremental cost-effectiveness ratios are the same
141
STOOL GUAIAC TEST
142
# Guaiac test Cost Cases detected Avg cost/case detected *
1 7.75 0.00659469
10.77 0.00714424 5,495 3 13.02 0.00719004 8,852 4 14.81 0.00719385 11,783 5 16.31 0.00719417 14,279 6 17.63 0.00719420 16,480
1 1
E E C C
i i
Source: Neuhauser and Lewicki, NEJM, 1975;293:226‐8.
143
# Gua. test Cos t Cases detected ACE R* Inc r. cos t Incr. cases ICER
1 7.75 0.00659469 2 10.77 0.00714424 5,495 3.02 0.00054955 5,495 3 13.02 0.00719004 8,852 2.25 0.00004580 49,127 4 14.81 0.00719385 11,783 1.79 0.00000381 469,816 5 16.31 0.00719417 14,279 1.50 0.00000032 4,687,500 6 17.63 0.00719420 16,480 1.32 0.00000003 44,000,000
1 1
E E C C
i i
Source: Neuhauser and Lewicki, NEJM, 1975;293:226‐8.
144
Method Cases Detected Cost Incr. cases Incr. cost ICER 1 7 160,733 3 9 166,477 2 5,744 2,872 4 9 194,540 28,063 Dominated (by M3) 2 10 191,959 1 25,482 25,482 FPG 10 231,790 39,831 Dominated (by M2)
Example: Average ratios and screening methods of Type2 DM
Rearrangement
145
146
ICER = 2,900
ICER = 25,500
147
1. Rank options in ascending order of either outcomes or cost 2. Eliminate options that are strongly dominated (i.e. have increased cost and reduced outcomes compared with at least one other alternative) 3. Calculate ICERs for each adjacent pair of outcomes (e.g. between option 1 and 2, 2 and 3, 3 and 4, etc.) 4. Eliminate options that are weakly dominated (i.e. have less effective but higher cost‐effectiveness ratio than the next highest ranked option) 5. Recalculate the ICERs (e.g. between option 2 and 4) 6. Repeat step 4 and 5 if necessary
Six treatment options
A: Intermittent proton-pump inhibitor (PPI) B: Maintenance PPI C: Maintenance H2 receptor antagonists (H2RA) D: Step-down maintenance prokinetic agent (PA) E: Step-down maintenance H2RA F: Step-down maintenance PPI
148
Source: Goeree et al., PharmacoEconomics, 1999; 16:679‐97.
ORDER OF EITHER OUTCOMES OR COST
149
Strategy 1-year cost per patient ($) Weeks with GORD per patient in 1 year C: Maintenance H2RA 657 10.41 A: Intermittent PPI 678 7.78 E: Step-down maintenance H2RA 748 6.17 D: Step-down maintenance PA 805 12.60 F: Step-down maintenance PPI 955 5.54 B: Maintenance PPI 1093 4.82
STRONGLY DOMINATED
150
Strategy 1-year cost per patient ($) Weeks with GORD per patient in 1 year C: Maintenance H2RA 657 10.41 A: Intermittent PPI 678 7.78 E: Step-down maintenance H2RA 748 6.17 D: Step-down maintenance PA 805 12.60 Dominated by C, A, E F: Step-down maintenance PPI 955 5.54 B: Maintenance PPI 1093 4.82
151
Strategy 1-year cost per patient ($) Weeks with GORD per patient in 1 year Incr. cost ($) Weeks averted ICER ($/GORD week averted) C 657 10.41 A 678 7.78 21 2.63 8 E 748 6.17 70 1.61 44 F 955 5.54 207 0.63 329 B 1093 4.82 138 0.72 192
152
Strategy 1-year cost per patient ($) Weeks with GORD per patient in 1 year Incr. cost ($) Weeks averted ICER ($/GORD week averted) C 657 10.41 A 678 7.78 21 2.63 8 E 748 6.17 70 1.61 44 F 955 5.54 207 0.63 329 F has less effective but higher ICER than B B 1093 4.82 138 0.72 192
153
Strategy 1-year cost per patient ($) Weeks with GORD per patient in 1 year Incr. cost ($) Weeks averted ICER ($/GORD week averted) C 657 10.41 A 678 7.78 21 2.63 8 E 748 6.17 70 1.61 44 D 805 12.60 dominated F 955 5.54 dominated B 1093 4.82 345 1.35 256
154
155
― Showing how results depend on uni‐, multi‐ parameters ― Defensible range should be taken (e.g. SE, 95% CI)
― Identifying parameter’s value needed in order to change / still do not change the decision
― Several discrete values (e.g. risk group, acceptance rate, tool’s accuracy) have impact to the outcome, instead of a continuous range
156 Source: Leelahavarong et al. BMC Health Services Research; 2010; 10:209.
“Tornado diagram” showing the impact of uncertainty on the outcome of a decision model
Percentage change in ICER ranged from -65 to 30 when utility of BT-ICT was varied from 0.3 to 0.9
Point estimate = 80,000
189 81 71 57 189 38 32 24 20 40 60 80 100 120 140 160 180 200 ไม่ต้องฉีดซ้ํา ฉีดซ้ําทุก 20 ป ฉีดซ้ําทุก 15 ป ฉีดซ้ําทุก 10 ป ราคาวัคซีนต่อเข็ม (บาท) กระตุ้น 1 เข็ม กระตุ้น 3 เข็ม
* กรณีให้วัคซีนที่อายุ 15 ปี + คัดกรองทุก 5 ปีแก่สตรีอายุ 30-60 ปี (ความครอบคลุม 70:70) เปรียบเทียบกับ มาตรการพื้นฐานคือคัดกรอง VIA + PAP ทุก 5 ปีแก่สตรีอายุ 30-60 ปี (ความครอบคลุม 80%)
ราคาสูงสุดที่ยอมรับได้ กรณีไม่ต้องการลงทุนเพิ่ม (BREAK EVEN PRICE)
158
500,000
5
ICER 120,000
3
C Rc=100,000 ฿/LYs
THRESHOLD AT BREAK-EVEN POINT (COST-SAVING)
159
500,000
5
ICER 120,000
3
C Rc = 0 ฿/LYs
160
Two‐way threshold analysis Three‐way threshold analysis
Source: Briggs AH, Weinstein MC, Fenwick EAL, et al. Model Parameter Estimation and Uncertainty: A Report of the ISPOR‐SMDM Modeling Good Research Practices Task Force‐6. Value Health 2012;15:835‐842.
161
New treatment More effective New treatment less effective New treatment More costly New treatment less costly
C
New treatment more effective But more costly Existing treatment dominated New treatment dominated New treatment less costly But less effective
162
163
500,000
5 4 5
ICER 25,000 ICER 400,000 ICER -200,000 ICER 120,000
3 2 1
ICER -200,000
C
ICER 60,000
6
Rc=100,000 ฿/LYs
164
500,000
5 4 5
ICER 25,000 ICER 400,000 ICER -200,000 ICER 120,000
3 2 1
ICER -200,000
C
ICER 60,000
6
Rc=100,000 ฿/LYs
Accept the technology if ICER < ceiling ratio
Willingness to pay for one unit of outcome Commonly unknown its true value Depends on social value of health May have more than one values at the same system! Useful for producing cost-effectiveness acceptability
curves
165
166
167
£0 £10,000 £20,000 £30,000 £40,000 £50,000
1 2 3 4
Incremental life-years Incremental costs
Point estimate
£0 £10,000 £20,000 £30,000 £40,000 £50,000
1 2 3 4
Incremental life-years Incremental costs
Source: Briggs A (2004)
RC=£0/LY RC=£/LY RC=£15,000/LY RC=£5,000/LY RC=£30,000/LY RC=£50,000/LY RC=£100,000/LY
Source: Briggs A (2004)
RC=£0/LY 0.05
Source: Briggs A (2004)
RC=£15,000/LY 0.5
Source: Briggs A (2004)
RC=£50,000/LY 0.87
Source: Briggs A (2004)
RC=£/LY 0.92
Source: Briggs A (2004)
80% interval exists RC=£2,000/LY RC=£72,000/LY
Cost‐effectiveness acceptability curves: Confidence intervals
Source: Briggs A (2004)
RC=£/LY 0.92 RC=£50,000/LY 0.87 RC=£15,000/LY 0.5 RC=£5,000/LY 0.15 RC=£0/LY 0.05
Cost‐effectiveness acceptability curves: Confidence surfaces
100,000 105,000 110,000 115,000 120,000 125,000 130,000 135,000 140,000 1720.6 1720.7 1720.8 1720.9 1721 1721.1 1721.2 1721.3 1721.4
Cost (Baht) QALYs (years)
3+1 schedule with indirect vaccine effects
No vaccine PCV10 PCV13
100,000 105,000 110,000 115,000 120,000 125,000 130,000 135,000 140,000 1720.6 1720.7 1720.8 1720.9 1721 1721.1 1721.2 1721.3 1721.4
Cost (Baht) QALYs (years)
100,000 105,000 110,000 115,000 120,000 125,000 130,000 135,000 140,000 1720.6 1720.7 1720.8 1720.9 1721 1721.1 1721.2 1721.3 1721.4
Cost (Baht) QALYs (years)
100,000 105,000 110,000 115,000 120,000 125,000 130,000 135,000 140,000 1720.6 1720.7 1720.8 1720.9 1721 1721.1 1721.2 1721.3 1721.4
Cost (Baht) QALYs (years)
175
No vaccine PCV 10 PCV 13
SIMULATION RESULTS OF PROVIDING PCV VACCINE, 3+1 SCHEDULE
WITHOUT INDIRECT EFFECTS, SOCIETAL PERSPECTIVE
176
2,000 4,000 6,000 8,000 10,000 12,000 29.19 29.2 29.21 29.22 29.23 29.24 29.25 29.26
Cost (Baht) QALYs (years)
3+1 schedule without indirect vaccine effects
No vaccine PCV10 PCV13
177
0.0 0.2 0.4 0.6 0.8 1.0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000
Probability of favouring each option Value of ceiling ratio (THB per QALY)
3+1 schedule without indirect vaccine effects
No vaccine PCV10 PCV13
The CE decision rule: Accept the technology if ICER <
ceiling ratio
Rearranging: Comparing:
C
C
Edited from: Tambour et al, 1998 IJTAHC
178
B B C A A C
Calculate net-benefit for each simulation option Calculate average net-benefit Optimal option has greatest average net-benefit No need to worry about positive/negative cost and health
Easy to implement
179
FRONTIER
Cost-effectiveness acceptability curve look at
In fact, decision should be drawn across the
180
Source: Fenwick et al., Health Economics; 2001.
181
0.0 0.2 0.4 0.6 0.8 1.0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000
Probability of favouring each option Value of ceiling ratio (THB per QALY)
3+1 schedule without indirect vaccine effects
No vaccine PCV10 PCV13
0.0 0.2 0.4 0.6 0.8 1.0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000
Probability of favouring each option Value of ceiling ratio (THB per QALY)
3+1 schedule without indirect vaccine effects
No vaccine PCV10 PCV13
0.0 0.2 0.4 0.6 0.8 1.0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000
Probability of favouring each option Value of ceiling ratio (THB per QALY)
3+1 schedule with indirect vaccine effects
No vaccine PCV10 PCV13
0.0 0.2 0.4 0.6 0.8 1.0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000
Probability of favouring each option Value of ceiling ratio (THB per QALY)
3+1 schedule with indirect vaccine effects
No vaccine PCV10 PCV13