The costeffectiveness of alternative chemotherapy regimens in advanced colorectal cancer: economic a

1
The cost-effectiveness of alternative
chemotherapy regimens in advanced colorectal
cancer economic analysis using data from the
FOCUS trial

• Yolanda Bravo Vergel, Christian Asseburg,
• Andrea Manca and Mark Sculpher
• Centre for Health Economics,
• University of York

MRC Clinical Trials Unit, 5th December 2005
2
Acknowledgments

• Matt Seymour, Gareth Griffiths and Denise Blake
  
• National Cancer Research Institute Colorectal
  Clinical Studies Group
• Research funded by the MRC

3
Outline

• Background
• Trial design
• Aims
• Data sources
• Modelling
• Markov model structure
• Input parameters
• Assumptions
• Cost-effectiveness results
• Conclusions

4
Background

• 2nd commonest cause of cancer in the UK
• Risk developing colorectal cancer is 120
• Annual incidence 31,000
• Unresectable or metastatic - poor prognosis
  (median survival under 12 months)
• Aim treatment is to improve duration and QoL of
  patients remaining life.
• Mid 90s - Standard treatment infusional 5FU FA.
• Administration - modified version of the Gramont
  schedule (MdG), giving FA and bolus 5FU on day 1
  followed by 5FU infusion over 46 h.

5
Trial design(1)

• New therapies, different mechanisms
• Irinotecan,
• Oxaliplatin
• More expensive
• Trials to date efficacy single treatment
  episodes.
• FOCUS trial - what is the optimum treatment
  strategy?
• Standard approach of sequential single-agent
  therapy (1st MdG 2nd Iri)
• Combination schedule with MdG (IrMdG or OxMdG)
• Adult patients with measurable, advanced
  colorectal cancer unsuitable for surgery (n
  2,135)

6
Trial design(2)

• Multicentre 5-year RCT (in the ratio 21111).
• FOCUS - trial design
• Plan A 1st line MdG 2nd Irinotecan
• Plan B 1st line MdG 2nd IrMdG
• Plan C 1st line IrMdG
• Plan D 1st line MdG 2nd OxMdG
• Plan E 1st line OxMdG
• Supportive care alone / salvage chemotherapy

7
Aims

• To estimate the cost-effectiveness of five
  chemotherapy regimens for the treatment of
  advanced colorectal cancer, using individual
  patient level data (FOCUS trial).

8
When is an intervention cost-effective?
Benefits of new treatment
More
Less
?
Costs of new treatment
More
dominated
Less
?
dominates
What is the opportunity cost of finding the
additional resources to provide the new
treatment? Can the extra resources be obtained
from doing less of something else and TOTAL
benefits increase?
9
What do we need to know to make decisions about
resources?

• Outcomes in terms of mortality and side effects
• Measured in generic units comparable with other
  services
• Time horizon of a patients lifetime

• Costs on chemotherapy drugs
• Chemotherapy administration costs
• Hospitalisation and other costs over a patients
  lifetime

10
Data sources (1)

• Measurement in trial
• Survival
• Quality of life (EQ-5D)
• Number of drug vials
• Number of chemotherapy cycles
• Resource use chemo administration
• Hospital GP visits

11
Data sources (2)

• Costing of resource use
• National statistics
• NHS Reference Costs 2004,
• Review Body for Nursing Staff 2003,
• CIPFA 2003/04.
• BNF v.49 2005
• Literature review

12
Modelling

• Probabilistic economic model for assessing the
  cost-effectiveness of five management strategies
• Primary outcome measure cost per QALY.
• Costs from the perspective of the health care
  provider (UK NHS), expressed in 2004/05 price.
• Costs and benefits discounted at 3.5 following
  the latest NICE guidelines (2004).
• Markov model - Cycle length 12 weeks
• Reasons for treatment interruption (toxicity,
  holidays)
• Radiological assessments at 12-week intervals
• Collection cost-related information

13
Markov model structure

• Markov model with health states equivalent to
  coincide with each chemotherapy plan (A to E) and
  treatment line (1st line prior to 2nd line, 2nd
  line prior to salvage), plus death.

Prior to 2nd-line
Prior to Salvage
salvage
dead
For plans A, B and D, the Markov model has the
above structure. Plans C and E have only one line
of treatment previous to salvage.
14
Model parameters(1)

• 3 main categories of parameters
• Transition probabilities, relating to the move
  from 1st line to 2nd line, salvage treatment or
  death
• Quality of life (generic EQ-5D instrument)
• Key resource use components
• Administration chemotherapy plan
• Primary care related (GP attendance, GP visit,
  district nurse visit)
• Hospital related (inpatient chemotherapy,
  general/acute inpatient, day case, outpatient
  medical oncology visits, ICU HDU)
• Salvage chemotherapy palliative care

15
Model parameters(2)

• Cost per chemotherapy cycle include
• initial clinical radiological assessment,
• drug acquisition (vials),
• adjunctive drugs and devices,
• chemotherapy administration (day case /
  outpatient ward, nurse time)
• regular monitoring tests.
• Number of drug vials based on actual dosage (mg)
  and number of chemotherapy cycles were estimated
  based on direct analysis of FOCUS dataset.

16
Assumptions

• Allowance for dosage wastage
• Chemotherapy cycles administered on a day case /
  outpatient basis.
• Primary care visits and hospital attendance
  recorded in the trial forms reflect side effects
  and SAE treatment costs.
• EQ-5D reflect SAE impact on health.
• All patients incur a cost of 2 weeks as hospital
  at home when they die.

17
Results
Totals
BREAKDOWN OF TOTAL COSTS
Note Means. Costs discounted at 3.5.
18
Survival curve
Totals
Mean trial follow-up
19
Results
Totals
POPULATION IN MARKOV STATES DETAIL
Note Means. Discounted LYs at 3.5. This slide
shows how much of the estimated life-time falls
into each Markov state.
20
Results
Totals
5
5
1
2
2
2
5
1
1
3
3
3
4
4
4
Note Mean and 90-CrIs. Discounted at 3.5
21
Cost-effectiveness results
Totals
Note Means. Discounted at 3.5
22
Totals
Cost-Effectiveness Acceptability Curves
A MdG -gt Ir
B MdG -gt IrMdG
C IrMdG
D MdG -gt OxMdG
E OxMdG
0
20,000
40,000
60,000
80,000
100,000
23
Conclusions

• Results indicate that the combination therapy
  MdG-gtIrMdG is cost-effective for the treatment of
  advanced colorectal cancer based on a lifetime
  horizon.
• At conventional levels of cost-effectiveness
  considered by UK decision makers to provide value
  for money to the NHS (aprox 30,000 per QALY),
  the probability that the combination regimen MdG
  -gtIrMdg is cost-effective is 60.
• In spite of superior survival QALY gain, from a
  UK perspective, IrMdG does not appear to be
  cost-effective. The rest of combination therapies
  were dominated.
• All combination chemotherapy regimens resulted in
  a superior mean survival compared to standard
  therapy

24
Appendix 1. Results
Markov states
AVERAGE DRUG DOSES (mg.)
Note Medians and 90-CIs. Doses given in mg per
patient per chemotherapy cycle and include the
wastage.
25
Appendix 2. Results
Markov states
ANNUAL NUMBER OF CHEMOTHERAPY CYCLES
Note Medians and 90-CIs.The numbers are per
patient per year.
26
Appendix 3. Results
Markov states
CHEMOTHERAPY ANNUAL COSTS
Note Medians and 90-CIs. The costs are shown
for a hypothetical patient who stays in the
relevant Markov state for a whole year.