Potential impact of populationbased colorectal cancer screening in Canada: An application of the POp

1
Potential impact of population-based colorectal
cancer screening in CanadaAn application of the
POpulation HEalth Model (POHEM)
2
Context

• Randomized controlled trials had shown efficacy
  of screening for colorectal cancer with faecal
  occult blood test (FOBT)
• National Committee on Colorectal Cancer Screening
  established in 1998 by Health Canada to study
  impact and feasibility of colorectal cancer
  screening in Canada
• POHEM used to evaluate effectiveness of
  population-based screening for Canada

3
Randomized Controlled Trials

• Funen, Denmark Kronborg et. al., Lancet 1996
• Nottingham, UK Hardcastle et. al., Lancet 1996
• Minnesota, USA Mandel et. al., New England
  Journal of Medicine, 1993

4
Colorectal Cancer in the year 2000
Second cause of cancer deaths in Canada

• Estimated
• Incidence Deaths
• MEN 9,200 3,500
• WOMEN 7,900 3,000
• TOTAL 17,100 6,500

Source Canadian Cancer Statistics 2000
5
Incidence Rate for Colon and Rectal Cancer, 1995
6
Stage Distribution at Diagnosisof Colorectal
Cancer
Source Ottawa Chart Review 1991-92 (N700)
7
Disease-Specific SurvivalRectal Cancer
Source Ottawa Regional Cancer Center chart
review of 700 cases diagnosed in 1991-1992,
Dukes staging converted one to one from TNM
8
METHODS

• Population Health Model (POHEM)
• FOBT screening protocol established in
  consultation with experts from National Committee
• Validated with randomized controlled trials
• Performed sensitivity analysis on parameters
• Evaluated multiple screening scenarios
• Performed cost-effectiveness analysis
• differential cost per life year gained between
  screening and not screening
• included the estimated total cost of screening
  and cost of lifetime treatment

9
Population Health Model

• Monte Carlo microsimulation model
• Generates health biographies for synthetic
  individuals from empirical observations
• Continuous time
• Competing risks
• Colorectal cancer incidence/progression module
• Screening protocol module added

10
Screening
11
Screening Protocol

• Screening Test Faecal Occult Blood Test (FOBT)
• - Hemoccult II not rehydrated
• Eligibility no history of CRC, target age range
• Recruitment year of introduction length of
  follow-up
• Participation 1st and subsequent invitations
• Frequency biennial vs annual
• Sensitivity how good the test is at detecting a
• cancer when one is present (Funen RCT)
• Specificity how good the test is at determining
• that no cancer is present (Funen RCT)
• Sojourn time period when cancer is present and
• detectable by a screening test but
• not clinically detectable
• Follow-up Colonoscopy follow-up for positive
  FOBT
• Follow-up when polyps detected.

12
FOBT Screening Paths
Pre-clinically detected cancer

True positive
Colonoscopy
-
Cancer missed, will show before next screen
-

False positive
Colonoscopy
No Cancer Remove polyps if any Re-invited to
later screening
FOBT
True negative
-
No Cancer Go on to next screening
False negative
Cancer missed, will show before next screen
13
Sensitivity Specificity
True positive (TP)
Sensitivity TP/(TPFN) how good the test is at
detecting a cancer when one is present

False positive (FP)
FOBT
Specificity TN/(TNFP) how good the test is at
determining that no cancer is present
True negative (TN)
-
False negative (FN)
14
Simulating outcomes of FOBT
True positive
Apply sensitivity of FOBT
Yes
False negative
Pre-clinical cancer?
True negative
Apply specificity of FOBT
No
False positive
15
Estimated unit costs of national screening program
Treatment costs documented elsewhere (Maroun et
al, CDIC 2003)
16
Stage Distributions
17
Validation of Model

• Objective To reproduce the colorectal cancer
  mortality reduction observed in a randomized
  controlled trial (RCT) using model simulation
• Method Input characteristics of Funen RCT into
  the simulation and estimated 10-year mortality
  reduction
• Funen RCT was population-based and had a clear
  recruitment strategy that could be reproduced in
  the model to generate similar follow-up periods
  (10-year)
• Results
• CRC mortality reduction in Funen trial was 18
  (1-32)
• based on approximately 60 000 participants
• CRC mortality reduction from model was 17.9
  (16.9-18.9)
• based on approximately 7 million cases

18
Parameter Sensitivity AnalysisBiennial Screening

• 10 Year CRC
• Type of Run Mortality Reduction
• Funen parameters 17.2
• From Funen to Canadian stage data 17.5
• From 67 to 50 FOBT compliance 10.5
• From 93 to 80 rescreen rate 9.9
• From 89 to 80 colonoscopy compliance 9.3
• From 45-75 to 50-74 age group 8.9
• runs of approximately 2 million cases parameter
  changes were additive

19
Cohort Types
20
Canadian Screening Scenarios

• Reference scenario
• 0. No screening
• incidence based on observed rates from Canadian
  Cancer Registry
• Base scenario
• Biennial screening 67 participation
• 51 sensitivity, 98 specificity (Funen RCT)
• Stage shift as observed in Funen applied to
  Canadian stage distribution
• Alternative scenarios
• Annual screening 67 participation
• 80.8 sensitivity, 97.7 specificity (Minnesota
  RCT)
• Stage shift as observed in Minnesota applied to
  Canadian stage distribution
• Biennial screening 50 participation
• Biennial screening gradual ramp-up to 67
  participation over 5 years
• Individual Risk Assessment Full participation
  to biennial screening
• For all screening scenarios
• Complications per 10,000 colonoscopy 2 deaths,
  17 perforations, 3 haemorrages

21
Results fromMortality Reduction Analysis
22
Percent mortality reduction over timeCanadian
Base Scenario Biennial screening, 67
participation
23
Results
24
Results fromCost-effectiveness Analysis
25
Undiscounted
Discounted at 5
26
At what age is it cost-effective to begin
screening ?

• starting cohort aged 40-44 in 2000
• screening ending at age 75

X
?
/LYG incremental cost per life-year gained
27
Comparison of life expectancy gains of screening
by age screening started
28
At what age is it cost-effective to end screening
?

• cohort aged 50-54 in 2000
• screening started at age 50

?
?
X
29
Comparison of life expectancy gains of screening
by age screening ended
30
Results from Resource Impact Analysis
31
Volume of FOBTs
32
Volume of Colonoscopies
33
Summary

• 15 increase in the number of colonoscopies over
  current practice as estimated for year 2000
• 838 FOBTs and 17 colonoscopies were needed to
  pre-clinically detect one CRC
• 1278 FOBTs and 27 colonoscopies were needed to
  avoid one CRC death
• Average cost of screening (discounted at 5)
• 112 million per year (33 of total cost)
• 5 reduction in treatment cost
• Incremental cost of 355 per person

34
Results from Individual Risk Assessment Analysis
35
What are the potential gains a 50 year old might
expect from full participation in a screening
program until age 75 ?

• Mortality Reduction from CRC
• 34.5 after 10 years
• 31.4 after 25 years
• Potential gain in life expectancy
• 0.10 years (37 days) for cohort
• 1.75 years for persons diagnosed with CRC
• Lifetime probability of having a colonoscopy
  0.25
• Lifetime rate of complications per 10,000
• Death 0.5, Perforations 4.3 and Haemorrage 0.8

36
Discussion

• POHEM results were used to inform recommendations
  by the National Committee on Colorectal Cancer
  Screening in 2000
• http//www.phac-aspc.gc.ca/publicat/ncccs-cndcc/in
  dex.html
• POHEM was used to generate province-specific
  results for Alberta and Ontario
• Ontario began population-based screening in
  spring 2007