The consequences of uncertainty and the implications for policy. Karl Claxton 21/11/2017

Transcription

1 The consequences of uncertainty and the implications for policy Karl Claxton 21/11/2017

2 Overview Why does uncertainty matter? Clinical value of evidence Linking endpoints to outcome Dealing with costs Approval and research decisions Price and the need for evidence Irrecoverable (opportunity) costs A policy solution? Value based rebates

3 Primary endpoint captures health effects The sequence of trials of early thrombolysis using streptokinase

4 Primary endpoint captures health effects (cumulative meta-analysis) Trial name (expected benefits of additional evidence) Earlier in sequence (1969) Later in sequence (1979)

5 When was there enough evidence?

6 Primary endpoint in the meta analysis linked to other aspects of outcome (Steroids in head injury) Meta-analysis of evidence available before the CRASH trial Odds ratio for death after the CRASH trial (n= 10,008) = 1.21 ( )

7 Primary endpoint in the meta analysis linked to other aspects of outcome (Steroids in head injury) Odds ratio of dead, vegetative and severely disabled = 1.10 (0.81, 1.53) Glasgow Outcome Scale outcome Percentage of individuals (95% CI) by treatment Steroids No steroids Dead 33.5 (22.8, 45.2) 35.3 (24.8, 46.9) Vegetative 4.8 (2.8, 7.5) 3.8 (2.4, 5.9) Severe disability 13.5 (8.3, 20.1) 10.7 (7.1, 15.8) Moderate disability 11.6 (8.6, 14.8) 12.1 (9.2, 15.1) Good recovery 36.5 (28.1, 44.8) 38.0 (30.1, 45.6) Combine the effects on mortality and disability External evidence Quality of life associated with GOS outcomes Life expectancy by GOS outcomes Movement between GOS states Impact on life years adjusted for the quality in which they are likely to be lived (QALYs)

8 Primary endpoint in the meta analysis linked to other aspects of outcome (Steroids in head injury) Practice prior to CRASH trial i. No steroids, expected benefits = 1,067 QALYs ii. Steroids, expected benefits = 2,711QALYs iii. 12% on steroids, 88% no steroids, expected benefits = 1,264 QALYs

9 Assessing whether proposed research is worthwhile Was CRASH worthwhile? CRASH cost 2.2m Expected health benefits of 1,264 QALYs per year (10,266 QALYs, ) 2.2m could have been used by the NHS to generate 115 QALYs elsewhere NHS would require 19m to get similar health benefits How did it compare to other proposals that could have been funded with the same resources? Similar analysis of the health benefits of other proposals being considered CRASH offered 214 per QALY to 2015 or 1,930 per QALY in 2004 Compare value of NIHR to other NHS spend or NICE decisions What was the value of CRASH? Cannot be judged with hindsight Included the most valuable endpoints Resolved uncertainty about mortality and effects on severe disability No benefits of additional evidence after CRASH (it was a definitive trial) Was it too large or not stopped early enough?

10 What assessments are needed Distribution of expected costs and effects Estimate of health opportunity costs of health care expenditure Value of evidence Is additional evidence valuable (population NHB)? Approval may affect the prospects of acquiring it Type of research required not possible with approval Balance the benefits of early access for current patients with benefits forgone for future patients Irrecoverable costs Costs committed by approval that can not be recovered Capital costs of long lived equipment (training and learning) Initial losses (negative NHE) offset by later gains Significance depends on whether initiation of treatment can be delayed Are the benefits of early approval greater than the opportunity costs

11 Assess cost-effectiveness and population net health effects Is it cost-effective? Yes Assess irrecoverable costs Are there significant irrecoverable costs? Yes No No Assess need for evidence Go to Figure 2.2 Key Assessment Decision Guidance # Technologies expected to be cost-effective OIR if research is not possible Go to Appendix A Part II Does more research seem worthwhile? Yes What type of evidence is needed? Is the research possible with Approval? AWR Yes Assess other sources of uncertainty Will this Uncertainty be resolved over time? Yes Will research be conducted? When will it be available? How much will be resolved? Re-assess the benefits and costs of further research Are the benefits of research greater than the costs? Yes No No No No Assess other sources of uncertainty Will this Uncertainty be resolved over time? Yes Will research be conducted? When will it be available? How much will be resolved? Re-assess the benefits and costs of further research Are the benefits of research greater than the costs? Yes Assess the benefits and costs of early approval Are the benefits of approval greater than the costs? Yes Approve Approve OIR Approve Approve No No No Technologies not expected to be costeffective Reject or OIR AWR is unlikely Irrecoverable costs? OIR more likely if research is not possible OIR even if research is possible

12 A checklist of assessments Point Assessment Judgement Yes No 1 Is it expected to be cost-effective? Yes 2 Are there significant irrecoverable costs? No 3 Does more research seem worthwhile? 4 Is the research possible with approval? 5 Will other sources of uncertainty resolve over time? 6 Are the benefits of research greater than the costs? 7 Are the benefits of approval greater than the costs?

13 Assessment Guidance 1 Yes No Yes Yes Yes/No Yes - AWR 1 2 Yes No Yes Yes Yes/No No - Approve 1 Possible pathways 3 Yes No Yes No Yes/No Yes Yes Approve 2 4 Yes No Yes No Yes/No Yes No OIR 1 5 Yes No Yes No Yes/No No - Approve3 6 Yes No No Approve 4 7 No No Yes Yes Yes/No Yes - OIR 2 8 No No Yes Yes Yes/No No - Reject 1 9 No No Yes No Yes/No Yes Yes AWR 2 10 No No Yes No Yes/No Yes No Reject 2 11 No No Yes No Yes/No No - Reject 3 12 No No No Reject 4 13 Yes Yes Yes Yes Yes Yes Yes AWR 3 14 Yes Yes Yes Yes Yes Yes No OIR 3 15 Yes Yes Yes Yes Yes No Yes Approve 5 16 Yes Yes Yes Yes Yes No No Reject 5 17 Yes Yes Yes Yes No Yes Yes AWR 4 18 Yes Yes Yes Yes No Yes No OIR 4 19 Yes Yes Yes Yes No No - Approve 6 20 Yes Yes Yes No Yes Yes Yes Approve 7 21 Yes Yes Yes No Yes Yes No OIR 5 22 Yes Yes Yes No Yes No Yes Approve 8 23 Yes Yes Yes No Yes No No Reject 6 24 Yes Yes Yes No No Yes Yes Approve 9 25 Yes Yes Yes No No Yes No OIR 6 26 Yes Yes Yes No No No - Approve Yes Yes No n/a Yes n/a Yes Approve Yes Yes No n/a Yes n/a No Reject 7 29 Yes Yes No n/a No - - Approve No Yes Yes Yes Yes/No Yes - OIR 7 31 No Yes Yes Yes Yes/No No - Reject 8 32 No Yes Yes No Yes/No Yes Yes AWR 5 33 No Yes Yes No Yes/No Yes No Reject 9 34 No Yes Yes No Yes/No No - Reject No Yes No Reject 11 Part I of the algorithm Part II of the algorithm Part III of the algorithm

14 3. Does more research seem worthwhile? i. How uncertain is a decision to approve or reject ii. Do the likely consequences of uncertainty justify further research. NHE that could be gained if it could be resolved immediately Upper bound on potential benefits of more Assess need for evidence Does more research seem worthwhile? Yes No No sometimes leads directly to guidance Pathway number Assessment Guidance 6 Yes No No Approve 4 12 No No No Reject 4 35 No Yes No Reject 11

15 Assessing the consequences of uncertainty Figure 3.4b Distribution of the consequences of uncertainty for CLOP Probability : : : : : : : : : <0.01 5: 0.02 probability from PSA 5: : : < 0.01 NHS clop1 clop3 clop6 clop to to to to to to to Consequences, QALY

16 4. Is research possible with approval? i. Type of evidence needed? ii. Research required be conducted while approved? Importance of parameters (values that change the decision) Uncertainty in possible values (how likely to change) NHE that are to be gained (expected consequences) What type of evidence is needed? Is the research possible with Approval? No Yes

17 Expected consequences (importance and uncertainty) Table 3.6b Consequences of uncertainty associated with parameter values (CLOP) Natural history* Utilities* RE Costs* Expected consequences (QALYs) Decomposed by treatment choice Parameter clop12 clop6 clop3 clop1 NHS Overall 1P_die_ P_NFMI_ P_die_ P_NFMI_ P_die_ P_NFMI_ P_die_ P_NFMI_ TP_AC TP_AD TP_CD TP_BD U_Well U_Well U_NFMI U_POSTMI RR_death RR_NFMI C_Well C_MI_LT C_PostMI TC_Well_Dead C_t C_t C_t C_t C_t

18 5. Will other sources of uncertainty resolve over time? i. Other sources of uncertainty Changes in price (technology and comparators) New technologies entering Other evidence becoming available ii. Impact on benefits of research (6) iii. Impact of benefits and costs of early approval (7) Assess other sources of uncertainty Will this Uncertainty be resolved over time? No Yes

19 6. Are the benefits of research greater than the costs? i. Will the research be conducted ii. When will it be available iii. How much will be resolved? iv. Impact of other sources (5) Assessment Guidance Will research be conducted? When will it be available? How much will be resolved? Re-assess the benefits and costs of further research 1 Yes No Yes Yes Yes/No Yes - AWR 1 2 Yes No Yes Yes Yes/No No - Approve 1 5 Yes No Yes No Yes/No No - Approve3 Are the benefits of research greater than the costs? Yes No Pathway number 7 No No Yes Yes Yes/No Yes - OIR 2 8 No No Yes Yes Yes/No No - Reject 1 11 No No Yes No Yes/No No - Reject 3 19 Yes Yes Yes Yes No No - Approve 6 26 Yes Yes Yes No No No - Approve No Yes Yes Yes Yes/No Yes - OIR 7 31 No Yes Yes Yes Yes/No No - Reject 8 34 No Yes Yes No Yes/No No - Reject 10

20 When will it be available? Figure 3.7a Potential value of research and time to report (CLOP)

21 7. Are the benefits of approval greater than the costs? Benefits of approval Expected additional NHE for the target population Costs of approval are opportunity costs Potential value of research which maybe forgone NHE for future patents Irrecoverable costs committed by approval Capital costs (equipment, facilities, training and learning) Initially negative NHE (when treatment decisions can be changed) Assess the benefits and costs of early approval Are the benefits of approval greater than the costs? No Yes

22 Technologies without significant irrecoverable costs probability that resea Figure 3.11a OIR An OIR or Approve boundary (CLOP) Approve Without price change With price change time research reports (T), years

23 Technologies without significant irrecoverable costs Table 3.7a Population NHE over the technology time horizon for different policies (CLOP) Approve OIR AWR* Reject Value of AWR Uncertainty resolved at launch Value of evidence at launch Expressed in QALY T<T* (T=2) 3,680,187 3,681,480 3,682,995 3,671,660 1,515 3,684,181 2,701 T>T* (T=7) 3,680,187 3,675,487 3,680,362 3,671, ,684,181 3,994 NHE expressed in m T<T* (T=2) 73,604 73,630 73,660 73, , T>T* (T=7) 73,604 73,510 73,607 73, , Investments which might make AWR possible Value of having the evidence needed at Launch Policies for better, more relevant an timely evidence Commercial value of AWR and early evidence When should research be publically funded How should value and costs be shared Absolute and comparative advantage

24 Uncertainty, price and the need for evidence Net health effects or equivalent NHS resources Price for unrestricted access = A R Expected to be cost-effective Reject Approve 0 A R Price

25 Uncertainty, price and the need for evidence Net health effects or equivalent NHS resources Price for unrestricted access = A R Expected to be cost-effective Value of resolving uncertainty Reject Approve 0 A R Price

26 Uncertainty, price and the need for evidence Net health effects or equivalent NHS resources Research takes resources and time Research is possible with or without approval No significant irrecoverable costs Price for unrestricted access is unchanged Reject OIR AWR Approve 0 A R Price

27 Net health effects or equivalent NHS resources Uncertainty, price and the need for evidence Approve AWR Research is possible with or without approval No significant irrecoverable costs OIR Reject Reject OIR AWR Approve 0 A AWR A OIR AWR OIR A R AWR R OIR R Price

28 Net health effects or equivalent NHS resources Uncertainty, price and the need for evidence Approve Research is not possible with approval, No significant irrecoverable costs OIR Reject Reject OIR AWR Approve 0 A AWR A OIR AWR OIR A R AWR R OIR R Price

29 Technologies with significant irrecoverable costs Costs which are irrecoverable Capital costs of long lived equipment (training and learning) Initial losses (negative NHE) offset by later gains Decision might change Research reports Prices change, new technologies, other evidence Flexibility in initiation of treatment Not flexible or costs of delay for individual patient Cost of delay for capital investment Expected to be cost-effective OIR rather than AWR even if research is possible OIR rather than Approve more likely to be appropriate Reject rather than Approve is possible Not expected to be cost-effective Reject rather than OIR is more likely

30 Net health effects or equivalent NHS resources Approve Uncertainty and irrecoverable costs AWR Research is possible with or without approval No significant irrecoverable costs OIR Reject Reject OIR AWR Approve 0 A AWR A OIR AWR OIR A R AWR R OIR R Price

31 Net health effects or equivalent NHS resources Approve Uncertainty and irrecoverable costs AWR Research is possible with approval, Significant irrecoverable opportunity costs OIR Reject OIR AWR Approve 0 A AWR AWR OIR A R AWR R OIR R Price A OIR

32 Enhanced External Counterpulsation for chronic stable angina EECP is a Non-invasive procedure (adjunct to standard therapy) used to provide symptomatic relief from stable angina Compares EECP to standard therapy alone RCT evidence suggests an improvement in HRQL with EECP at 12 months but longer term effects are uncertain EECP is expected to be cost-effective but with potentially significant irrecoverable costs long lived costs associated with the purchase of equipment large initial per patient treatment costs, combined with a chronic condition where a decision not to treat a particular patient with EECP can be changed at a later date (decisions are not irreversible) when research reports or other events occur.

33 Cost-effectiveness at the patient level Figure 3.1a Cumulative incremental NHE of EECP over the patient time horizon

34 Irrecoverable costs and the profile of NHE Cumulative incremental NHE of EECP for the population 5,000 Cumulative incremental NHE at population level for EECP, QALY 0-5,000-10,000-15,000-20, Capital cost spread over 10 years Capital cost incurred in year 1 Technology time horizon -25,000 Time, years

35 Significant irrecoverable costs and research is not possible with approval 0.0 An OIR or Approve boundary (EECP) Sufficient condition for Approve Probability that research is conducted Necessary condition for OIR 4-year design 3-year design 2-year design 1-year design Time for research to report, years

36 Original results for EECP at existing prices Table 3.7b Population NHE over the technology time horizon for different policies (EECP) Approve OIR AWR Reject Value of AWR Uncertainty resolved at launch OIR even though EECP expected to be cost effective at current price AWR not valuable due to irrecoverable costs Value of having the evidence needed at Launch Policies for better, more relevant and timely evidence Value of making research results available more quickly Values depend on time research will take (T=3 and T=7) Investments which might make AWR possible Commercial value of AWR and early evidence When should research be publically funded How should value and costs be shared Value of evidence at launch Expressed in QALY T=3 1,391,001 1,397,192 1,393,578 1,389,596-3,614 1,400,288 3,096 T=7 1,391,001 1,393,608 1,392,030 1,389,596-1,578 1,400,288 6,680 Expressed in m T=3 27,820 27,944 27,872 27, , T=7 27,820 27,872 27,841 27, ,

37 Price thresholds for EECP T=3 years, no research costs AWR OIR A R

38 Summary of qualitative effects on price Research worthwhile Possible with approval Possible without approval Other sources of uncerinty Significant irrecoverable costs Price for unrestricted access Price for unconditional access 1 No - - Yes No A R AIR No change 2 No - - No Yes A R AIR 3 Yes Yes Yes Yes/No No A R = AWR OIR A AWR < A R 4 Yes No Yes Yes/No Yes/No A OIR < A R A OIR < A R Lower price 5 Yes Yes Yes Yes/No Yes AWR OIR < A R A AWR < A R 6 No - - Yes Yes A R lower A R lower Higher price Indeterminate 7 Yes Yes No Yes/No No AWR R > A R A AWR < A R 8 Yes Yes More costly Yes/No No AWR OIR > A R A AWR < A R 9 Yes Yes No Yes/No Yes AWR R > or < A R A AWR < A R 10 Yes Yes More costly Yes/No Yes AWR OIR > or < A R A AWR < A R

39 Share of the value of evidence Prices renegotiated once research reports: Manufacturers get all the value of the research Only has value if chance that VBP < generic price NHS only benefits from research when patent expires Only has value if uncertain at generic prices (effect) Prices fixed at launch: Payer gets all the value of the research Informs who should pay for or conduct the research

40 Other considerations Non marginal budget impact Health opportunity costs of larger than expected costs are greater than the gains from lower than expected Evidence suggests non marginal is quite limited Project specific risks Included in the assessment of all the consequences of uncertainty (i.e., NHB certainty equivalent) Leads to decisions that appear to be risk averse compared to expected cost-effectiveness Macroeconomic risk and prudential savings Non linear effect of uncertainty in consumption growth Limited over time horizons < 40 years Interaction of macro and project specific risks Pro or counter cyclical projects (Betas) Weak correlation and small projects assume neutral (beta=0)

41 Some implications for policy Assess whether/when accelerated access is justified Regulatory and reimbursement OIR needed as a policy option Unwilling or unable to lower price sufficiently AWR is not possible/credible Account for irrecoverable (opportunity)costs Commonly lower than A R Align Incentives for evaluative research Greater reward for relevant evidence at launch How value is shared depends on price renegotiation rules Who should conduct or pay for the research Other public investments Evidence at launch Make AWR possible Research findings more quickly available

42 What's the problem? Limits to per product price negotiation or discounts Reference pricing and parallel trade Lack of a credible demand side Threats to HCS Access without the evidence Immediate harms (health opportunity costs) Undermine evidence base (harm to future patients) Access matters Incentives for private top up insurance market Threats to manufacturers Ad hoc barriers to access Slow up take and additional costs Lack of predictability Unable to access all the value during patent Global price discrimination Internal price discrimination

43 Can we fix it?. value based rebates NICE Appraisal Benefits, costs Appropriate volume Centre (DH) Rebate required Appropriate volume Rebate beyond volume Assessment of health opportunity costs Manufacturer Agrees to include in rebate calculation Pays rebate based on sales Manufacturer Does not agree to include in rebate calculation Prescriber Pays list price to manufacturer Fully reimbursed by centre Prescriber Pays list price to manufacturer Not reimbursed Claxton K. Pharmaceutical pricing: early access, the cancer drugs fund and the role of NICE. Centre for Health Economics, University of York Mar, CHE Policy & Research Briefing. Claxton K., Briggs A., Buxton MJ., Culyer AJ, McCabe C., Sculpher MJ., and Walker S. Value-Based Pricing for NHS Drugs: an Opportunity Not to be Missed? British Medical Journal, 2008; 336:

44 Requirements and possibilities Evidence based assessment of additional effects and costs Assessment of eligible population (max volume) Evidence based assessment of health opportunity costs Recalculate rebates when patent of any comparator expires Encourage generic entry or reference price generics Rebate to increases to list price if exceed max volumes Reimburse prescribers (at price paid) Different rebates for different indications and subgroups Link rebates to need for evidence and irrecoverable costs Recalculate rebates as research reports