James Bishop, Parisa Glass, Elizabeth Tracey, Margaret Hardy, Kylie Warner, Koji Makino, Adam Gordois, Jodie Wilson, Carmel Guarnieri, Jun Feng and Lynn Sartori In collaboration with IMS Health Health Economics Review of Bowel Cancer Screening in Australia August 2008 Cancer Institute NSW Monograph
107
Embed
Health Economics Review of Bowel Cancer Screening in Australia€¦ · Health Economics Review of Bowel Cancer Screening in Australia 5.1 Evidence from the literature 58 5.2 Cost-effectiveness
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
James Bishop, Parisa Glass, Elizabeth Tracey, Margaret Hardy, Kylie Warner, Koji Makino,
Adam Gordois, Jodie Wilson, Carmel Guarnieri, Jun Feng and Lynn Sartori
In collaboration with IMS Health
Health Economics Reviewof Bowel Cancer Screening
in Australia
August 2008
Cancer Institute NSW Monograph
Cancer Institute NSW catalogue number:
SM-2008-1
National Library of Australia Cataloguing–in–Publication data:
Predictive value of positive for colorectal cancers
and adenomas 80
vi
Health Economics Review of Bowel Cancer Screening in Australia
Figures
Figure 1
Incidence of bowel cancer (diagnosed) and extent
of disease at diagnosis – simulation results xvi
Figure 2
Literature inclusion and exclusion criteria 4
Figure 3
Simplifi ed natural history of bowel cancer used in the
economic model 21
Figure 4
Screening pathway used in the economic model 23
Figure 5
Age distribution in the simulation cohort
at baseline 24
Figure 6
Incidence of bowel cancer (diagnosed) and
extent of disease at diagnosis – simulation results 56
Figure 7
Bowel cancer – odds ratio – fi xed model 81
Figure 8
Bowel cancer – relative risk – fi xed model 81
Figure 9
Bowel cancer – risk difference – fi xed model 81
Figure 10
Adenoma – odds risk – fi xed model 82
Figure 11
Adenoma – relative risk – fi xed model 82
Figure 12
Adenoma – risk difference – fi xed model 82
Figure 13
Bowel cancer deaths – odds risk – fi xed model 83
Figure 14
Bowel cancer deaths – relative risk – fi xed model 83
Figure 15
Bowel cancer deaths – risk difference
– fi xed model 83
Figure 16
Dukes’ A – odds risk – fi xed model 84
Figure 17
Dukes’ A – relative risk – random model 84
Figure 18
Dukes’ A – risk difference – fi xed model 84
Figure 19
Dukes’ B – odds ratio – fi xed model 85
Figure 20
Dukes’ B – relative risk – fi xed model 85
Figure 21
Dukes’ B – risk difference – fi xed model 85
Figure 22
Dukes’ C – odds ratio – random model 86
Figure 23
Dukes’ C – relative risk – random model 86
Figure 24
Dukes’ C – risk difference – random model 86
Figure 25
Dukes’ D – odds ratio – fi xed model 87
Figure 26
Dukes’ D – relative risk – random model 87
Figure 27
Dukes’ D – risk difference – fi xed model 87
Figure 28
All-cause mortality – odds ratio – fi xed model 88
Figure 29
All-cause mortality – relative risk – fi xed model 88
Figure 30
All-cause mortality – risk difference
– fi xed model 88
vii
Abbreviations
ABS Australian Bureau of Statistics
AHTAC Australian Health Technology Advisory Committee
AIHW Australian Institute of Health and Welfare
CI confi dence interval
FOBT faecal occult blood test
iFOBT immunochemical faecal occult blood test
NHMRC National Health and Medical Research Council
OR odds ratio
RCT randomised controlled trial
RD risk difference
RR relative risk
TNM tumour, node, metastasis
viii
Health Economics Review of Bowel Cancer Screening in Australia
Foreword from the Minister
The NSW Government has made substantial
commitments to improving the outcomes for cancer
in the State. The NSW Cancer Plan 2004–2006 was an
Australian fi rst and reported important areas of progress
by 2006. The NSW Cancer Plan 2007–2010 renews our
commitment to ongoing improvement in cancer results.
Australians have a particular problem with bowel cancer,
recording one of the highest incidence amongst comparable
developed countries. Bowel screening offers real hope to
substantially improve outcomes in this disease. In the NSW
Cancer Plan 2007–2010, the Cancer Institute NSW and
NSW Health are committed to support the roll out of this
important national program.
This report provides valuable insight into the cost-
effectiveness of bowel screening. The intervention compares
favourably with many services and programs we take for
granted. However, it is not without initial costs, with benefi ts
occurring as the program is established. This report assists
us in defi ning the costs and benefi ts of this important
national program.
I commend this report to you.
Hon. Verity Firth MPMinister for Climate Change and the Environment
Minister for Women
Minister for Science and Medical Research
Minister Assisting the Minister for Health (Cancer)
ix
Chief Cancer Offi cer’s report
Bowel cancer is the second most common cancer in men
behind prostate cancer, and in women it is second behind
breast cancer. It is the second largest cause of cancer death
in both men and women in NSW.i Australia has one of the
highest incidence rates of bowel cancer, surpassing the UK
and USA.
Major risk factors for bowel cancer are family history,
consumption of red or processed meat, alcohol consumption
and body and abdominal obesity.ii Conversely dietary fi bre,
physical activity, calcium and milk appear protective.
The fi ve-year survival of bowel cancer is currently 65 per
cent in NSW compared to 88 per cent for breast cancer.i
The substantial reductions in cancer mortality from breast
cancer of 18 per cent over the past 10 years has been
equally attributed to breast cancer screening and treatment
improvements.i,iii It is estimated that complete deployment
of bowel cancer screening in Australia would deliver
mortality reductions of 13 to 17 per cent for patients with
bowel cancer.
Currently only around 34 per cent of bowel cancer cases are
localised on presentation, compared to 60 per cent of breast
cancer cases.iv The fi ve-year survival for localised bowel
cancer is much better at 87 per cent. Screening should result
in a larger number of cases that are localised, increasing those
patients chances of long-term survival.
This report looks at the best health economic approach
to introduce bowel cancer screening into the population.
It concludes that universal screening for all persons aged
50 to 74 years is cost-effective and more so than other
age scenarios. It provides evidence that a high level of
colonoscopy following a positive faecal occult blood test
(FOBT) is most cost-effective, with participation rates also
somewhat cost-effective.
This report provides a good rationale for the early roll out
of more widespread bowel cancer screening using the FOBT
followed by colonoscopy as the screening tool. The report
can be used by health planners to further monitor and
improve bowel screening in Australia. Based on evidence
available, it is hoped that bowel screening will substantially
improve the outcomes for bowel cancer in our community.
Professor Jim F Bishop AO MD MMED MBBS FRACP FRCPA
Chief Cancer Offi cer and CEO, Cancer Institute NSW
Professor of Cancer Medicine, University of Sydney
Tracy E, Baker D, Chen W, Stravrou E, Bishop J. Cancer in New South Wales: Incidence, Mortality and Prevalence 2005, Sydney, Cancer Institute NSW, i. November 2007.
World Cancer Research Fund. American Institute for Cancer Research. Food, nutrition, physical activity and the prevention of cancer: A Global Perspective. ii. Washington DC: AICR, 2007.
Berry DA, Cronin KA, Plevritis S, et al. Effects of screening and adjuvant therapy on mortality from breast cancer. NEJM 2005; 353:17841792.iii.
Tracey, Chen S, Baker D, Bishop J, Jelfs P. Cancer in NSW South Wales: Incidence and mortality 2004. Sydney, Cancer Institute NSW, November 2006.iv.
x
Health Economics Review of Bowel Cancer Screening in Australia
Executive Summary
In NSW, screening for breast and cervical cancers has had
a major impact on mortality associated with these diseases.
In the past decade, mortality rates for breast and cervical
cancers have declined by 22% and 52%, respectively.1 The
principal cause of these mortality rate improvements comes
from effective population-based screening programs.
Bowel cancer is the second most common cause of cancer
related death in NSW. Approximately one in 17 men and
one in 26 women will develop bowel cancer before the age
of 75.1 Population-based screening improves the likelihood
of early detection of pre-cancerous lesions and early stage
malignancies. Detection of pre-cancerous or early stage
bowel cancers reduces morbidity and mortality associated
with the disease.2
The National Bowel Cancer Screening Program is a nationally
coordinated, population-based initiative that commenced
in August 2006. The Program currently targets Australians
who turn 55 or 65 years of age each year, and those who
participated in the Bowel Cancer Screening Pilot Program.
This report investigates current evidence relating to the
effi cacy of bowel cancer screening by conducting a systematic
review of the literature. A systematic literature review was
performed to demonstrate the clinical evidence of screening
instruments used in the National Bowel Cancer Screening
Program – immunochemical faecal occult blood testing
(iFOBT) and colonoscopy.
This report also examines whether a national bowel
cancer screening program represents value for money
for the Australian health systems. The extent of fi nancial
implications associated with implementing the program was
also estimated.
Systematic review of available studies
Effi cacy of bowel cancer screening
A search of relevant literature was conducted using
EMBASE.com and Cochrane Library databases. The trials
included in the systematic review involved general screening
populations of asymptomatic participants aged from 45 to
80 years of age. Participants undertook faecal occult blood
tests (FOBT). A positive FOBT result on more than one
occasion required further investigation by colonoscopy or
double-contrast barium enema, where colonoscopy was
contraindicated or incomplete. Participants with cancer
diagnoses exited studies for treatment. Participants whose
FOBT results were negative were re-invited to undergo
biennial testing and were followed up.
A meta-analysis was conducted using results from three
large international randomised controlled trials – Minnesota,
USA (1993), Funen, Denmark (1996), and Nottingham, UK
(1996) – to assess the effi cacy of bowel cancer screening
to detect early cancers, and any subsequent reduction in
bowel cancer mortality.3-19 Each trial involved participants
undertaking FOBTs and subsequent diagnostic colonoscopy
for participants who had one or more positive FOBT results.
Outcomes from both the screening and control groups were
included in the analysis. Proportions of detected bowel
cancers and adenomas, Dukes’ stages at diagnosis, deaths
from bowel cancer, and all-cause mortality were included. It
was found that:
adenoma detection rates in the screening group tested ■biennially were higher, with a relative risk on average
of 2.60, compared with the non-screening group
(p<00001). Early detection of adenomas, which have
potential for malignancy if untreated, in the screened
population resulted in their prompt removal and
reduction of subsequent risk of progression to
bowel cancer
overall, bowel cancer detection rates were similar ■between the screening and control groups (p<0.27)
there were more diagnoses of Dukes’ A stage bowel ■
Biennial screening is associated with 13–17 per cent reduced rates of bowel cancer mortality.
xi
cancer in the screening group tested biennially, with a
relative risk on average of 1.56, compared with the non-
screening group (p<0.001),
diagnoses of Dukes’ B, Dukes’ C and Dukes’ D stages ■in the screening group were less frequent, with relative
risks on average of 0.93, 0.95 and 0.91, respectively,
compared with the non-screening group, however, were
not signifi cant. Increased detection of Dukes’ A resulted
in participants leaving the study to receive treatment
bowel cancer related death in the screening group tested ■biennially was lower, with a relative risk on average of
0.85, compared with the non-screening group
biennial screening is associated with 13–17% reduced ■rates of bowel cancer mortality during follow up periods
between 11.7 and 18 years.
FOBT accuracy
Faecal occult blood testing (FOBT) is an easy-to-use, non-
invasive technique for asymptomatic people to detect the
presence of occult (hidden) blood in stools which may be
attributable to bowel cancer. The National Bowel Cancer
Screening Program selected immunochemical FOBT
(iFOBT), over guaiac FOBT, because it is more sensitive,
does not require any dietary or medication changes before
use, and is well accepted by users. A search to identify
relevant literature that considered iFOBT accuracy among an
asymptomatic population was conducted using
EMBASE.com. Inclusion of a symptomatic population would
have overestimated the sensitivity and specifi city of the
iFOBT, so studies that enrolled symptomatic people were
excluded from the systematic review.
Nakazato et al (2006) conducted a cross-sectional analysis of
3,090 asymptomatic people, average age 53.4 (± 8.2), who
underwent iFOBT followed by colonoscopy.20
In this instance:
reported sensitivity and specifi city of iFOBT for cancer ■was 52.6% and 87.2%, respectively
reported sensitivity and specifi city of iFOBT for large ■adenomas (diameter 10 mm) was 24.5% and
87.1%, respectively.
Colonoscopy accuracy and safety
FOBT results do not necessarily equate with diagnoses.
Follow up colonoscopy is recommended for people whose
FOBT fi ndings are positive. Colonoscopy is an invasive
procedure performed under sedation that is safe and
relatively pain free. A systematic review of studies that
reported sensitivity and specifi city of colonoscopy used
as a diagnostic test and any associated complications was
performed to assess the procedure’s accuracy and safety.
NHMRC Guidelines for the prevention, early detection
and management of colorectal cancer (2005) reported
that colonoscopy sensitivity for detection of cancer and
adenomas is 95% and 85% respectively. Specifi city is 100%.
Colonoscopy is considered to be the gold standard to detect
adenomas and cancers.21
A review of retrospective studies indicated that colonoscopy
has been associated with few complications. For every
10,000 colonoscopies performed, the perforation rate
reported in studies ranged between 0 and 19; the
occurrence of bleeding ranged between 20 and 25 times, and
the mortality rate ranged between 0 and 5.
Cost-effectiveness of the National Bowel Cancer
Screening Program
The cost-effectiveness of bowel cancer screening has been
assessed by numerous studies.22-24 In the current evaluation,
the modelled cost-effectiveness analysis was conducted to
examine whether the National Bowel Cancer Screening
Program represents value for money for the Australian
health system. To this end, the likely economic and health
outcomes consequences of the current Program were
compared with a scenario without a nationally coordinated
screening program.
A Markov decision-analytic model was developed to simulate
possible scenarios for assessment. This approach allowed
a comprehensive analysis of short-term outcome effects,
such as screening costs and incidence of colonoscopy
complications, as well as long-term outcomes – in this case,
survival – in the absence of empirical data reported from
the Program.
xii
Health Economics Review of Bowel Cancer Screening in Australia
The model simulated bowel cancer disease history, including
prevalence and incidence, disease progression, symptomatic
presentation and diagnosis, and patient survival. It also
replicated National Bowel Cancer Screening Program
screening practices. The Program initially targets people
turning 55 or 65 years of age each year, who are re-invited
for screening biennially thereafter until they reach 75 years
of age. The National Bowel Cancer Screening Program uses
immunochemical faecal occult blood testing (iFOBT) as a
fi rst-line test. Participants whose test results are positive are
referred for follow up colonoscopy.
A hypothetical cohort of people aged between 50 and 74
years was applied to simulate the 55 or 65 year old scenario.
This age group was selected because the risk of bowel
cancer has been reported to increase after the age of 50.2
The timing of the fi rst invitation to attend screening was
consistent with the eligibility age group being considered (i.e.
as people turn 55 or 65 years of age).
The relative cost-effectiveness of the Program was expressed
in terms of the cost per additional life-year saved over the
cohort’s lifetime. This enabled the cost-effectiveness of the
National Bowel Cancer Screening Program to be compared
with other healthcare programs whose aim is to improve
survival. Results from the cost-effectiveness analysis are
presented in Table 1.
Under the 55 and 65 years scenario, the Program was estimated to produce a cost per life-year saved of $48,921, when compared with a scenario where screening was not provided. A value of $50,000 – 60,000 per life-year saved was generally regarded as an upper threshold of acceptable cost-effectiveness in the Australian healthcare system.
If the current Program is to continue over a suffi ciently long period of time, the population screened would eventually be made up of people who received their fi rst invitations as they turn 55 years old. To this end, a scenario where all people are aged 55 years old at the baseline was also explored. This scenario captured the long-term cost-effectiveness of the current Program in which screening would eventually cover all Australians aged between 55 and 74 years should it receive continued funding. The program produced costs per additional life year saved of $41,321 (Table 2). Under
this scenario, the model simulated the total number of bowel cancers detected by the program to be 52 cancers per 10,000 people over the cohort’s life time, giving a cost per cancer detected of approximately $85,000.
The long-term cost-effectiveness of screening scenarios covering all people aged between 45 and 74 years and 50 and 74 years was also investigated (Table 2). Screening was shown to reduce mortality and generate additional life years among the screened population in all eligibility age scenarios. Screening was also likely to represent a cost-effective strategy in the long run among all eligibility age groups.
The cost-effectiveness of bowel cancer screening using FOBT has been assessed by numerous studies.2,22-24 The current fi ndings support previously reported results.
It is important to note that the practicality and feasibility of expanding the age of eligibility should be assessed in relation to additional healthcare resource requirements and associated fi nancial costs with each age range.
A number of data inputs and assumptions were applied in the model to perform the simulation. The ability to generalise these results is dependent on their validity and accuracy. The current model was, wherever possible, informed by Bowel Cancer Screening Program Pilot data. Additional inputs were derived from the literature. A series of sensitivity analyses were performed to examine uncertainty associated with the simulation results.
Under the 55 years old scenario (see Table 2), the base case results were derived using the level of screening participation observed in the Pilot program.2 The iFOBT participation rate was reported to be only moderate – 45.4% of all invitees completed the tests. A moderate level of compliance with colonoscopy follow up was also reported in the Pilot program – the rate was 55% among people with positive iFOBT results. When a colonoscopy follow up rate of 80% was incorporated in the model, the incremental effectiveness offered by screening improved to 155 life-years per 10,000 invited people, and the incremental cost effectiveness ratio improved to $38,698. On the other hand, should the compliance rate be 20%, the incremental cost-effectiveness ratio declined to $63,744 due to a signifi cant deterioration in the Program’s effectiveness.
xiii
Sensitivity analysis demonstrated that the simulation results were sensitive to cancer survival estimates incorporated in the model. Recent estimates from the American Cancer Society (2007)v may be interpreted to indicate more favourable fi ve-year survival estimates than the available Australian estimates.25 The Program was no longer considered to be cost-effective when the American Cancer Society estimates were applied. Survival determines the health benefi t resulting from early cancer detection achieved
by screening. Hence, improvement in cancer survival erodes the cost-effectiveness of the screening program despite it detecting similar numbers of cancers. The current analysis may need to be revised as more recent cancer survival data become available in Australia.
American Cancer Society. Detailed Guide: Colon and Rectum Cancer [Online]. 2007; v. URL: http://www.cancer.org/docroot/CRI/content/CRI_2_4_3X_How_is_colon_and_rectum_cancer_staged.asp?sitearea.
Table 1 Cost-effectiveness of the National Bowel Cancer Screening Program
Note: All cost and outcome estimates were discounted using a 5% discount rate.
Table 2 Cost-effectiveness of national biennial bowel cancer screening: various screening ages
Note: All cost and outcome estimates were discounted using a 5% discount rate.
Lifetime cost per 10,000 invited participants
($ million)Life-years saved per 10 000 invited participants
Incremental cost per life-year saved
($)Screening Diagnostic
follow up
Cancer
management
Total
Current management – – 6.3 6.3 – –
Screening program 0.4 0.5 6.4 7.3 18.8 48 921
Lifetime cost per 10 000 invited people($ million) Life-years saved
per 10 000 invited people
Incremental cost per life-year saved($)
Screening Diagnostic follow
up
Cancer
management
Total
Program initiating screening for people turning 45 years of age
No national screening – – 5.5 5.5 – –
Screening program 1.5 3.7 5.9 11.1 123.5 44 955
Program initiating screening for people turning 50 years of age
No national screening – – 5.8 5.8 – –
Screening program 1.3 3.6 6.1 11.1 145.5 36 080
Program initiating screening for people turning 55 years of age
No national screening – – 6.2 6.2 – –
Screening program 1.2 3.2 6.4 10.8 112.8 41 321
xiv
Health Economics Review of Bowel Cancer Screening in Australia
Expected extent of fi nancial implications of the
National Bowel Cancer Screening Program
Feasibility and practicality of implementing a national
screening program relies in part on associated
budgetary impacts.
The estimated fi nancial implications of implementing the
National Bowel Cancer Screening Program were estimated.
These estimates exclude costs potentially borne by individual
participants. The costs presented are for the 10 years of the
Program, in which people become eligible as they turn 55 or
65 years of age. Eligible participants are re-invited
biennially thereafter.
Estimation of the fi nancial implications were also performed
for alternative eligibility age groups considered in the cost-
effectiveness analysis, covering all people aged between 45
and 74 years, 50 and 74 years, and 55 and 74 years.
The estimated sizes of eligible populations for the
considered scenarios are presented in Table 3. Under the
current program, the number of people covered expands
gradually over time as additional people enter the eligible
population each year (approximately 45,000–57,000
people annually). The gradual rollout of the program is
expected to reach optimum coverage of 5.1 million people
by the tenth year following introduction. When compared
with the current program, other eligibility age scenarios
cover larger populations, especially during the early years
of implementation, which would be translated to larger
healthcare resource requirements and associated costs under
these scenarios.
The estimated fi nancial implications of implementing the
National Bowel Cancer Screening Program were determined
and are presented in Table 4. Pilot data regarding
participation rates, iFOBT positivity rate and compliance
with the recommended diagnostic follow up were applied to
inform the estimate.
The total costs of the Program (nationally) are estimated to
be $21.9 million in Year 1, increasing to $126.3 million by Year
10 as screening coverage expands over time.
The likely 10-year extent of fi nancial implications of a national
screening program targeting all people aged between 45 and
74 years, 50 and 74 years, and 55 and 74 years, respectively,
was similarly determined.
The national costs of a screening program targeting all people
aged between 45 and 74 years was estimated to be $168.6
million in Year 1, increasing to $209.5 million by Year 10.
The national costs for a screening program targeting all
people aged between 50 and 74 years were estimated to be
$130.8 million in Year 1, increasing to $169.7 million by
Year 10.
The national costs of a screening program targeting all
people aged between 55 and 74 years was estimated to be
$96.6 million in Year 1, increasing to $131.8 million by Year 10.
In contrast to the current program, where screening is
gradually phased into effect, these eligibility age scenarios
would create an increase in people requiring diagnostic
follow up. Further work is now required on the training,
workforce and cost of a fully implemented national screening
program from a state and territory perspective.
xv
Table 3 Population coverage by the National Bowel Cancer Screening Program
Source: Australian Bureau of Statistics (2003).
Based on the 2008 population estimate.a.
People turning 55 and 65 years of age each year enter the screened population. b.
Table 4 Estimated costs of the screening program for years 1–10 (current age eligibility
– initial invitation at 55 and 65 years of age)
Note: These cost estimates were not discounted.
Cost of iFOBT ($10) includes supply of test kits, postages and reminder letter, and other coordination costs. An additional $20 for pathology and information a. management is incurred for each test completed and returned by the participant.
Unit cost estimates for colonoscopy and polypectomy were based on the National Hospital Cost Data Collection Cost Report Round 7 Public Sector b. ($1,082; additional $524 with polypectomy;. Costs of GP consultations were also included ($32.1; Level B GP consultation).26 Incidence of adverse events (perforation) was estimated using a risk of 0.001.27 Costs of perforation were based on information presented by O’Leary et al (2004),22 adjusted to 2004 prices ($17 662).28
9% of FOBT screening costs. This was based on national population cervical screening data.c. vi
Cost ($A)
Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9 Year 10
Current age eligibility – initial invitation at 55 and 65 years of age
Kronborg O, Jorgensen OD, Fenger C, Rasmussen M. Randomized study of biennial screening with a faecal occult blood test:
Results after nine screening rounds. Scand J Gastroenterol 2004; 39:846–851
6
Health Economics Review of Bowel Cancer Screening in Australia
2.1.3 Characteristics of the comparative randomised trials
Randomisation
Each of the three RCTs sent invitations to participate in
a bowel cancer screening program. Participants were
randomised to either a screening or non-screening group.
Participants in the screening group were invited for
screening biennially.
Trial design
The RCTs provided biennial screening to participants. If
adenoma or cancer was detected, participants left the
study for treatment. Table 7 lists details of the RCTs and
participants recruited.
Screening regime
Each trial involved participants undertaking FOBT, and
follow up diagnostic colonoscopy in the event of one or
more positive FOBT results. Table 8 lists each study and its
screening regime.
Reported outcomes
Outcomes addressed by randomised controlled trials and
used in the meta-analysis are presented in Table 9. The
screened and unscreened groups were compared for each
outcome. Other outcomes that were identifi ed through
trials, but which were not considered for meta-analysis, are
presented in Appendix B.
Modifi ed Dukes’ staging system
Bowel cancers are classifi ed into stages and levels of severity
depending on the extent and spread of malignancy. Earlier
stage cancers generally have better outcomes than cancers
diagnosed at later stages. In Australia, the Dukes’ staging
system is generally applied to describe the extent and spread
of bowel cancer (Table 10).
2.1.4 Analysis of the comparative randomised trials
Meta-analysis
The identifi ed RCTs were considered for meta-analysis.3-19 Outcomes from these trials are presented in Table 6. All analyses were performed using Review Manager Version 4.2.7 and presented outcomes applied the mean and 95% confi dence intervals (CI) for relative risk (RR). Relative risk provides a value expressing the likelihood that an event would occur compared with the non-screening group. Odds ratio (OR) and risk difference (RD) values for each comparison were also presented. A chi-squared test for heterogeneity was performed on each analysis, and was considered signifi cant if p < 0.05. Where there was signifi cant heterogeneity between the studies, a random-effects model was applied to control for between study variance.
Table 7 Meta-analysed RCTs
Abbreviation: RCT, randomised controlled trial.
Study and location Design Participants Duration
Mandel et al (1993)
Minnesota, USA
RCT 50–80 years
Exclusions: People with histories of bowel cancers, familial polyposis or
chronic ulcerative colitis
18 years
Hardcastle et al (1996)
Nottingham, UK
RCT 45–74 years
Exclusions: People with serious illness, including bowel cancer, diagnosed
within the past fi ve years
11.7 years
Kronborg et al (1996)
Funen, Denmark
RCT 45–75 years
Exclusions: People with known bowel cancers, adenomas, or any metastatic
malignancies
17 years
7
Table 8 Screening regime
Abbreviation: FOBT, faecal occult blood test.
Table 9 Reported outcomes
Table 10 Modifi ed Duke staging system33
Study FOBT regime Diagnostic follow up
Minnesota Hemoccult; six guaiac impregnated paper slides;
two samples from each of three consecutive
stools; dietary restrictions in place up to 24
hours before; rehydration
One or more slides of the six testing positive were offered a hospital
evaluation that included colonoscopy, or double-contrast barium enema
where necessary
Nottingham Haemoccult guaiac FOBT kit; two samples from
each of three consecutive stools; not rehydrated;
restricted diet for two days before collecting
samples, plus retest
Five or more positive squares at fi rst test, and those with one or more
positive squares at re-test, were offered colonoscopy; double-contrast barium
enema performed when full colonoscopy could not be done
Funen Hemoccult-II; guaiac FOBT; restricted diet;
completed slides not rehydrated; two faecal
samples from each of three consecutive stools
People with positive FOBT results were invited for interview, physical
examination and colonoscopy, or double-contrast barium enema where
necessary
Outcome Defi nition
Bowel cancers Proportion of bowel cancers detected
Adenomas Proportion of adenomas detected
Dukes’ A Proportion of Dukes’ A diagnosed
Dukes’ B Proportion of Dukes’ B diagnosed
Dukes’ C Proportion of Dukes’ C diagnosed
Dukes’ D Proportion of Dukes’ D diagnosed
Bowel cancer deaths Proportion of deaths due to bowel cancer
All cause mortality Proportion of deaths due to all causes
Stage Criteria
Dukes’ A The tumour penetrates into the mucosa of the bowel wall but no further
Dukes’ B
B1
The tumour penetrates into, but not through the muscularis propria (the muscular layer) of the bowel wall
B2 The tumour penetrates into and through the muscularis propria of the bowel wall
Dukes’ C
C1
The tumour penetrates into, but not through the muscularis propria of the bowel wall; there is pathological evidence of colon
cancer in the lymph nodes
C2 The tumour penetrates into and through the muscularis propria of the bowel wall; there is pathological evidence of colon
cancer in the lymph nodes
Dukes’ D The tumour, which has spread beyond the confi nes of the lymph nodes (to organs such as the liver, lung or bone)
8
Health Economics Review of Bowel Cancer Screening in Australia
2.1.5 Results of the comparative randomised trials
The Forest plots of all comparisons are presented in Appendix C.
Adenomas
The proportions of adenomas detected in the included trials are presented in Table 11. It was determined that 1.32% and 0.51% of patients were diagnosed with adenomas in the screening and control groups, respectively. These results suggest that those screened were 2.60 times more likely to have adenomas detected at screening compared with the control group (average relative risk of 2.60 [2.35, 2.87]).
Bowel cancers
All stages
The proportions of bowel cancers detected in the included trials are presented in Table 12.
It was determined that 2.11% and 2.17% of patients were diagnosed with bowel cancer in the screening and control groups, respectively. An average relative risk of 0.97 [0.92, 1.02] showed that there were no signifi cant differences in overall cancers detected within each group. This was not adjusted for different stages of bowel cancer.
Dukes’ A
The proportion of patients diagnosed with Dukes’ A bowel cancer in the included trials is presented in Table 13. It was determined that 21.4% and 13.7% of patients were diagnosed with Dukes’ A bowel cancer in the screening and control groups, respectively. Therefore, those screened were 1.56 times more likely to have Dukes’ A stage bowel cancer detected compared with their control counterparts (average relative risk of 1.56 [1.19, 2.05]).
Dukes’ B
The proportions of patients diagnosed with Dukes’ B stage bowel cancer in the included trials are presented in Table 14. It was determined that 31.4% and 34.0% of patients were diagnosed with Dukes’ B bowel cancer among the screening and control groups, respectively. Detection of Dukes’ B was
more likely among people in the control group compared with their screened counterparts (average relative risk of 0.93 [0.85, 1.02]).
Dukes’ C
The proportions of patients diagnosed with Dukes’ C stage bowel cancer in the included trials are presented in Table 15. It was determined that 23.5% and 26.0% of patients were diagnosed with Dukes’ C bowel cancer among the screening and control groups, respectively. Detection of Dukes’ C was more likely to occur among people in the control group compared with their screened counterparts (average relative risk of 0.95 [0.72, 1.25]), however, this was not signifi cant.
Dukes’ D
The proportions of patients diagnosed with Dukes’ D stage bowel cancer in the included trials are presented in Table 16. It was determined that 18.7% and 20.5% of patients were diagnosed with Dukes’ D bowel cancer in the screening and control groups, respectively. Detection of Dukes’ D was more likely to occur among people in the control group compared with their screened counterparts (average relative risk of 0.91 [0.80, 1.03]).
Bowel cancer deaths
The proportions of bowel cancer deaths in the included trials are presented in Table 17. It was determined that 0.90% and 1.05 % of deaths among the screening and control groups, respectively, were due to bowel cancer. Bowel cancer death is less likely to occur among people who are screened compared with their control counterparts (average relative risk of 0.85 [0.79, 0.97]).
All-cause mortality
The proportions of all-cause mortalities detected in the included trials are presented in Table 18. It was determined that 30.8% and 30.7% of patients died from all-cause deaths in the screening and control groups, respectively. Randomisation of participants to screening and control groups was adequate, and there were no signifi cant differences between groups (average relative risk of 1.00 [0.99, 1.01]).
9
Table 11 Rates of adenoma detection in the included RCTs
Table 12 Rates of overall bowel cancer detection in the included RCTs
Trial ID Screening Control Odds ratio (95%) Relative risk(95% CI)
and specifi city, feces analysis, diagnostic accuracy,
haemoccult, immunochemical tests, elisa, inform,
clinical trial
1,081 (1,082)
Manual search 0
Total 1,081
Cancer Condition determined by colonoscopy
True False
Test outcome Positive test True positive = 10 False positive = 394 Positive predictive value = 2.5%
Negative test False negative = 9 True negative = 2,677 Negative predictive value = 99.7%
Sensitivity = 52.6% Specifi city = 87.2%
Large adenoma Condition determined by colonoscopy
True False
Test outcome True True positive = 13 False positive = 391 Positive predictive value = 3.2%
Negative test False negative = 40 True negative = 2,646 Negative predictive value = 98.5%
Sensitivity = 24.5% Specifi city = 87.1%
2.2.3 Results
Nakazato et al (2006) reported iFOBT sensitivity of 52.6%
for bowel cancer and specifi city was 87.2%. FOBT sensitivity
and specifi city for large adenomas was 24.5% and 87.2%
respectively.20 The proportions of patients with positive
results who were correctly diagnosed with cancer or larger
adenomas were 2.5% and 3.2%, respectively (Table 20).
Studies of follow up colonoscopy in asymptomatic
populations were very limited because people whose
FOBT tests were negative are rarely tracked to or beyond
colonoscopy. The design of the National Bowel Cancer
Screening Program includes re-inviting people who have had
one or more negative FOBT test results to be re-screened
every two years.
16
Health Economics Review of Bowel Cancer Screening in Australia
2.3 Colonoscopy accuracy
Because FOBT does not defi nitively represent diagnoses of adenoma and bowel cancer, follow up colonoscopy is required for positive FOBT test results. Colonoscopy is considered the gold standard to detect adenomas and cancers.21 As was the case for FOBT, there is diffi culty in obtaining colonoscopy sensitivity and specifi city data, especially false negative rates in asymptomatic populations. This is because follow up colonoscopy is not mandatory for people who have negative FOBT test results. There is also debate about whether people who have negative FOBT test results should be subjected to invasive procedures such as colonoscopy.
2.3.1 Descriptions of the search strategies for relevant data
A systematic literature review of colonoscopy accuracy was performed using studies that reported the sensitivity and specifi city of the diagnostic test. The search was conducted using EMBASE.com. The search strategy is presented in Appendix A. After removing duplicates, 925 articles were identifi ed. A manual search of bibliographies of the retrieved articles was conducted which yielded no additional studies. All included references were retrieved and reviewed before further exclusions were made (Table 21).
There were no reports of colonoscopy sensitivity and specifi city in a general screening, average-risk population found in the literature.
2.3.2 Sensitivity and specifi city results from NHMRC Guidelines
Given that no reports of colonoscopy sensitivity and specifi city values were found in the literature, these values were adopted from the NHMRC Guidelines for the prevention, early detection and management of colorectal cancer (2005). It was estimated that the sensitivity of colonoscopy for detecting bowel cancers was 95% and specifi city was 100%. The reported sensitivity was 85%, and specifi city 100% for detecting small adenomas.21
2.4 Colonoscopy safety
Colonoscopy, although an invasive procedure provided under sedation, is safe and relatively pain free. The Minnesota USA trial reported colonoscopy perforation and bleeding rates of 0.03% and 0.09%, respectively.3 The Nottingham trial reported a perforation rate of 0.5%.7 No deaths due to complications arising from colonoscopy were reported by either trial. Retrospective reviews of medical evidence from 1999 were used to address the safety issues of colonoscopy in this systematic review.
2.4.1 Description of the search strategies for relevant data
A search was conducted using EMBASE.com to identify retrospective reviews of medical records that documented major complications associated with colonoscopy (see Appendix A). After removing duplicates, the search provided 1277 articles. All potential references were retrieved and reviewed before further exclusions were made. A manual search of bibliographies yielded no additional studies (Table 22).
The included studies in the systematic review of colonoscopy safety were retrospective reviews of medical records that addressed complications such as perforation, bleeding and mortality rates. A total of 15 studies post-1999 were included.
2.4.2 Results from retrospective reviews of medical records
Table 23 presents key complication rates reported by each study. For every 10,000 colonoscopies performed, the perforation rate reported in studies ranged from 0 to 19, the occurrence of bleeding ranged from 20 to 25, and the mortality rate ranged from 0 to 5. Colonoscopies were performed by, or under the supervision of, trained endoscopists, gastroenterologists, colorectal or general surgeons.
17
Databases Search terms Number of articles
Colonoscopy – Sensitivity and specifi city
EMBASE.com
(includes EMBASE and Medline)
colonoscopy, colorectal cancer, carcinoma, tumor,
adenoma, cancer screening, sensitivity and specifi city,
2.4.3 Retrospective reviews of medical records containing complications of colonoscopy
There were 15 retrospective reviews of medical records identifi ed from the search (Table 24). Medical records of patients who had undergone colonoscopies were examined for procedural complications. Data concerning perforation, bleeding and mortality rates associated with colonoscopy were extracted.
Table 21 Literature search results: colonoscopy accuracy
Table 22 Literature search results: colonoscopy complications
Health Economics Review of Bowel Cancer Screening in Australia
Table 23 Complication rates27,35-48
Abbreviation: NR, not reported.
2.5 Summary
The key fi ndings of the literature review include:
Early detection of adenomas among the screened ■population contributed to prompt follow up and treatment, which in turn reduced the subsequent risk of progression to bowel cancer.
Bowel cancers and adenomas detected at earlier stages ■reduced associated mortality. Biennial bowel cancer screening is associated with mortality reduction of 13–17%.
There were more diagnoses of Dukes’ A disease among ■the biennially tested screening group compared with people in the unscreened population. Early detection and treatment of bowel cancers reduces morbidity and mortality.
Dukes’ A disease detected by screening resulted in ■affected participants’ exiting the study for follow up and treatment.
Trial Perforation Bleeding/haemorrhage Mortality
Achiam 2001 12/4,000 (0.3%) NR 2/4,000 (0.05%)
Anderson 2000 20/10,486 (0.19%) NR 2/10,486 (0.02%)
The reported sensitivity of FOBT for cancer was 52.6% ■and specifi city was 87.2%.
The reported FOBT sensitivity and specifi city for ■detection of large adenomas was 24.5% and 87.2%, respectively.
The NHMRC ■ Guidelines for the prevention, early detection and management of colorectal cancer (2005) reported sensitivity of colonoscopy to detect cancers and small adenomas at 95% and 85%, respectively. Specifi city was 100%.21
Retrospective reviews of medical records showed ■that there were few complications associated with colonoscopy. For every 10,000 colonoscopies performed, the perforation rate reported in studies ranged from 0 to 19, the occurrence of bleeding ranged from 20 to 25, and the mortality rate ranged from 0 to 5.
19
Table 24 Literature search results: retrospective reviews of colonoscopy complications
Trial Reference
Achiam 200135 Achiam M, Rosenberg J. Quality of colonoscopy and surgical treatment of perforations. Ugeskr Laeg 2001; 163(6):775–778
Anderson 200036 Anderson ML, Pasha TM, Leighion JA. Endoscopic perforation of the colon: Lessons from a 10-year study. Am J Gastroenterol
Table 41 Cost of colonoscopy with and without polyp removal
Abbreviation: DRG, diagnosis related group.
Source: National Hospital Cost Data Collection Cost Report Round 7.xiii
Note: Public sector estimates are used as they capture the associated resource use more comprehensively than the private sector counterparts, hence more accurately representing economic value of the resource requirements from a societal perspective.
Variable Value Reference/note
Invitation/FOBT test kit $10 Estimate
FOBT pathology/information provision $20 Estimate
GP visit and referral $32.10 MBS Book November 2006
Colonoscopy -
Casemix Round 7 (2004)Without polypectomy $1,082
With polypectomy $1,606
Treatment of complication (perforation) $17,662 O’Leary et al (2004)22, AIHW (2006)
Lifetime bowel cancer treatment -
O’Leary et al (2004)22, AIHW (2006)Dukes’ stage A bowel cancer $17,148
Dukes’ stage B bowel cancer $33,364
Dukes’ stage C bowel cancer $25,771
Dukes’ stage D bowel cancer $6,264
DRG Code Description Cost estimates
Colonoscopy without polypectomy
G44C Other colonoscopy, same day $1,082
Colonoscopy with polypectomy
G43Z Complex therapeutic colonoscopy $1,606
Department of Health and Ageing 2004xiii.
38
Health Economics Review of Bowel Cancer Screening in Australia
3.2 Results
3.2.1 Base case analysis
Results from the cost-effectiveness analysis that compared
the program targeting those who turn 55 or 65 years of
age each year with no screening program are summarised
in Table 42. The assessment of cost-effectiveness of the
national screening program was made in terms of the cost
per additional life-year saved by avoiding bowel cancer-
related death during the cohort’s life time.
Screening was estimated to generate a cost per additional
life-year saved of approximately $48,921. These results
should represent the cost-effectiveness of the National
Bowel Cancer Screening Program, as assessed amongst
Australian people currently aged 55 and 65 years.
A value of $50,000–$60,000 per life-year saved was
generally regarded as an upper threshold of acceptable cost-
effectiveness for pharmaceutical treatments in the Australian
healthcare system.57 This point is discussed further in
Section 5.
Life-time costs of screening were estimated to be $726
($1,671 when undiscounted) per person. Life-time costs
were estimated to be $634 ($1,240 when undiscounted)
per person in the no screening arm, which was attributable
to cancer treatment costs. In the screening arm, the
invitation to screening (incorporating iFOBT provision costs)
and pathology analysis of completed iFOBTs accounted
for approximately 5% of the total costs in the screening
arm. Diagnostic colonoscopy follow up accounted for
approximately 7% of the costs (Table 42).
A large proportion of the total costs were attributed to
managing detected cancers. Most of these costs were
incurred regardless of screening being implemented. The
total cancer management costs were estimated to be
slightly higher in the screening arm than in the no screening
arm over the cohort’s life time. These costs accounted
for treatment of detected cancers as well as surveillance
of people with history of large adenomas (ie, fi ve-yearly
colonoscopies; see Section 3.1.1). When the surveillance
costs were excluded from the calculation, the total costs of
cancer management were lower in the screening arm than
in the no screening arm, at approximately $6.3 million per
10,000 people or $628 per person over the cohort’s life
time. This refl ects that the model predicted screening to
detect more malignancies at early stages (see Section 4).
The number of life-years saved offered by screening was
estimated to be 19 years per 10,000. When undiscounted,
this was estimated to be 167 years per 10,000. This estimate
equates to 0.002 life years per person (0.017 life years when
undiscounted). Considering the small absolute risk of bowel
cancer, it was expected that the screening program would
generate a small per-person benefi t over no screening in
terms of life years.
Table 43 presents an exploration of the value for money
that a bowel cancer screening program would offer with
alternative eligibility age ranges of 45–74 years, 50–74 years,
and 55–74 years. Under these scenarios, the screening was
introduced targeting all people within these age groups.
The 45–74 years age range scenario was assessed using a
cohort of people aged between 45 and 74, while other two
scenarios were assessed using a cohort aged between 50 and
74 as per the 55 and 65 years scenario.
The analyses, although valid, impart a distorted view of
the screening program’s long-term cost-effectiveness. The
relevance of these results would decline over time, because
they only related to a phase-in period of the program. This
occurred because of the baseline age distribution applied to
a cohort (see Figure 5). For the 55 and 65 years scenario,
this meant that people who were above the age of 65 years
at the baseline did not become eligible for screening at all,
thereby affecting the overall effectiveness of screening in
the analysis. For all the four analyses so far, it also meant
that an increasing number of people in the cohort became
ineligible for screening as the simulation progressed. This
was particularly relevant to the 55–74 age range scenario
because proportionally more people in this age range were
offered fewer opportunities for screening compared with
other eligibility age scenarios considered by this evaluation.
To address this limitation, analyses of a theoretical program
that initiated screening among people turning 45, 50 and 55
years old each year (ie, lower eligibility age cut-off for each
39
of the four alternative eligibility ages) were conducted (Table 44). Each analysis included a hypothetical cohort consisting
of people at the respective lower eligibility age cut-off.
These analyses depicted the long-term cost-effectiveness of
a screening program. This is because, if the program is to
continue to be implemented over a suffi ciently long period
of time, the population screened would eventually be made
up of people who received their fi rst invitation as they
turn these ages. To this end, the 55 years scenario is most
relevant to the current program as this scenario provided
the long-term perspective should the program become a
long-term public healthcare commitment offered by the
government through continued funding.
Table 44 also presents results from an analysis performed
in a cohort of people aged 55 or 65 years. Age distribution
within the cohort was performed on the basis of the current
Australian population data (ie, 60% aged 55 and 40% aged
65 years). This analysis still suffered the aforementioned
distortion due to age progression during simulation, but to a
smaller extent than the previous analyses.
The incremental cost-effectiveness ratios under these
scenarios differed only slightly from each other. Slightly more
incremental life years were observed under the 55 years
scenario, making this scenario relatively more cost effective
than others. Nonetheless, screening was shown to be cost-
effective under all the age scenarios considered here.
Under the 55 years old scenario in which the long-term
effectiveness of the current program was depicted, the
model simulated the total number of bowel cancers detected
by the program to be 52 cancers per 10,000 people over the
cohort’s life time. This fi gure corresponded with the average
cost per cancer detection of approximately $85,000. The
model also simulated a shift in cancer stages at diagnosis –
more cancers were diagnosed at earlier stages. Given that
bowel cancer can be associated with poor survival, especially
among people with late stage disease; screening was shown
to be reasonably cost-effective, as represented by the
incremental cost-effectiveness ratio expressed using the
number of life-years saved (Table 42).
These analyses clearly indicate that screening reduces
mortality and, thus, generate additional life years amongst
screening population. It is also shown that screening is likely
to represent a cost-effective strategy in the long run in all
eligibility age groups considered in the current evaluation. It
should be acknowledged that generalisability of these results
was limited by the data inputs and assumptions applied in
the model. The practicality and feasibility of expanding
the eligibility age should be assessed against the additional
healthcare resource requirements and associated fi nancial
costs. This is explored in Section 4.
Table 42 Cost effectiveness of a national bowel cancer screening program (people turning 55 or 65 years)
Note: All cost and outcome estimates are discounted using a 5% discount rate.
Lifetime cost per 10,000 invited people ($ million) Life-years saved per 10 000 invited people
Incremental cost per life-year saved($)
Screening Diagnostic follow up
Cancer management
Total
No national screening – – 6.3 6.3 – –
Screening program 0.4 0.5 6.4 7.3 18.8 48 921
40
Health Economics Review of Bowel Cancer Screening in Australia
Table 43 Cost-effectiveness of a national biennial bowel cancer screening: various eligibility age groups
Note: All cost and outcome estimates are discounted using a 5% discount rate.
Table 44 Cost-effectiveness of a national biennial bowel cancer screening: various initial screening ages
Note: Biennial screening is discontinued from 75 years of age. All cost and outcome estimates are discounted using a 5% discount rate.
Lifetime cost per 10,000 invited people ($ million) Life-years saved per 10,000 invited people
Incremental cost per life-year saved ($)Screening Diagnostic
follow upCancer management
Total
Program covering 45–74 years of age
No national screening – – 6.2 6.2 – –
Screening program 1.1 2.9 6.4 10.4 82.7 50,749
Program covering 50–74 years of age
No national screening – – 6.3 6.3 – –
Screening program 0.9 2.6 6.6 10.2 71.4 53,648
Program covering 55–74 years of age
No national screening – – 13.0 13.0 – –
Screening program 1.6 4.7 13.5 19.8 39.7 170,744
Lifetime cost per 10,000 invited people ($ million) Life-years saved per 10,000 invited people
Incremental cost per life-year saved($)
Screening Diagnostic follow up
Cancer management
Total
Program initiating screening for people turning 45 years of age
No national screening – – 5.5 5.5 – –
Screening program 1.5 3.7 5.9 11.1 123.5 44,955
Program initiating screening for people turning 50 years of age
No national screening – – 5.8 5.8 – –
Screening program 1.3 3.6 6.1 11.1 145.5 36,080
Program initiating screening for people turning 55 years of age
No national screening – – 6.2 6.2 – –
Screening program 1.2 3.2 6.4 10.8 112.8 41,321
Program initiating screening for people turning 55 or 65 years of age (a cohort aged 55 or 65 only)
No national screening – – 6.4 6.4 – –
Screening program 1.0 2.8 6.6 10.4 99.1 40,943
41
3.2.2 Sensitivity analysis
Validity and generalisability of the model-based economic
valuation are dependent on the accuracy of data inputs and
assumptions assigned to the model. A series of sensitivity
analyses were performed to examine the robustness of the
presented cost-effectiveness results. Sensitivity analyses also
aimed to identify and examine the program’s key elements
that may have important cost-effectiveness implications.
The following analyses were conducted in the context of
the 55 years scenario (Table 44). This scenario allowed
assessment of long-term cost-effectiveness of the current
program. Conducting sensitivity analyses using this scenario
would better inform the decision makers the impacts of
changing data input in the model, assisting them consider
whether continued funding of the program represents value
for money under various circumstances.
Participation is an integral part of any screening program
and an important contributor to overall effectiveness.
Screening participation for iFOBT was varied from the
base case level of 45.4% to 30% and 70%. The base case
fi gure was obtained from Pilot program data.2 The cost
and effectiveness estimates of a program running at these
participation rates are presented in Table 45.
It is signifi cant that the cost-effectiveness of screening
remained relatively stable when ranges of participation
rates were applied. This effect was created because while
increasing the participation rate improved the overall
effectiveness of the program, there was an associated
increase in costs. At 70% participation rate, the total number
of bowel cancers detected was estimated to increase to 79
cancers per 10,000 from the base case estimate of 52 per
10,000 over the cohort’s life time.
Compliance with diagnostic colonoscopy follow up was also
expected to be an important determinant for screening
program cost-effectiveness. The base case analysis
incorporated a compliance rate of 55%, as reported by the
Pilot program.2 The Pilot evaluation report noted that the
colonoscopy follow up rate was affected by missing data.
When a colonoscopy follow up rate of 80% was
incorporated in the model, the incremental cost-effectiveness
ratio improved slightly to $38,698. As expected, an increase
in screening effectiveness was observed. On the other
hand, a large decline in the effectiveness was observed
under an assumption of 20% colonoscopy compliance rate,
deteriorating the cost-effectiveness ratio to $63,744 per life
year saved.
Cancer survival represents another key determinant of the
relative cost-effectiveness of screening. Survival determines
the health benefi t, that is, additional life-years, resulting from
early cancer detection achieved by screening. The base case
analysis incorporated fi ve-year survival rates reported by a
Victorian Cancer registry study of bowel cancer
surgical patients.25
Recent bowel cancer fi ve-year survival estimates quoted by
American Cancer Society (2007) are summarised in Table 47. Compared with McLeish et al (2002), survival estimates
differ greatly, especially for a Dukes’ C disease, depending on
conversion from the Dukes’ and TNM staging systems.25
Results from an analysis incorporating fi ve-year survival rates
of 93%, 85%, 64% and 8% for Dukes’ A to Dukes’ D bowel
cancers, respectively, are presented in Table 48.
The incremental health benefi ts provided by screening
deteriorated with improved cancer survival at a cost per
life year saved of $121,034. This was an expected outcome:
health benefi ts, in terms of life-years provided by detecting
bowel cancer declines under this scenario. This is despite
the program detecting similar numbers of cancers, generating
costs per cancer detection similar to the base case analysis.
Results of other sensitivity analyses are presented in Table 49. All sensitivity analyses results, other than those relating
to discounting, produced cost-effective outcomes within the
range generally accepted as representing value for money of
less than $50,000 per life-year saved.
42
Health Economics Review of Bowel Cancer Screening in Australia
Table 45 Cost-effectiveness of national biennial bowel cancer screening program – differing participation rates
Note: All cost and outcome estimates are discounted using a 5% discount rate.
Versus the base case no screening arm.a.
Table 46 Cost-effectiveness of national biennial bowel cancer screening program – higher colonoscopy follow up rate
Note: All cost and outcome estimates are discounted using a 5% discount rate
Versus the base case no screening arma.
Table 47 Bowel cancer fi ve-year survival estimates, American Cancer Society (2007)XIV
Note: All cost and outcome estimates are discounted using a 5% discount rate.
Abbreviations: AJCC; American Joint Committee on Cancer, TNM, tumour, node, metastasis.
Lifetime cost per 10,000 invited people ($ million) Life-years saved per 10,000 invited people
American Cancer Society. Detailed Guide: Colon and Rectum Cancer [Online]. 2007; xiv. URL: http://www.cancer.org/docroot/CRI/content/CRI_2_4_3X_How_is_colon_and_rectum_cancer_staged.asp?sitearea
43
Table 48 Cost-effectiveness of a national biennial bowel cancer screening program – improved cancer survival
(Dukes’ C≈TNM IIIB)
Note: All cost and outcome estimates are discounted using a 5% discount rate.
Abbreviation: ACS, American Cancer Society.
Five-year survival of Dukes’ A (TNM stage I) = 93%, Dukes’ B (TNM stage IIA) = 85%, Dukes’ C (TNM stage IIIB) = 64%, Dukes’ D (TNM stage IV) = 8%.a.
Versus the base case no screening arm.b.
Table 49 Sensitivity analyses around key assumptions in the economic model
Lifetime treatment costs of detected cancer were $17,608; $17,608; $28,027 and $24,024 for Dukes’ stages A–D, respectively, based on data from Bolin et a. al (1999)23, updated to 2004 prices.
Lifetime cost per 10,000 invited people ($ million) Life-years saved per 10,000 invited people
Incremental cost per life-year saved($)
Screening Diagnostic follow up
Cancer management
Total
No national screening – – 6.2 6.2 – –
Screening—base case survival 1.2 3.2 6.4 10.8 112.8 41,321 a
Screening—ACS estimates a 1.2 3.2 6.5 10.8 38.2 121,034
Total lifetime cost per 10,000 invited people ($ million) Life-year saved per 10,000 people
Incremental cost per life-year saved ($)Screening Diagnostic
follow upCancer management
Total
Cost variables
High iFOBT costs–50% increase
No national screening – – 6.2 6.2 – –
Screening program 1.8 3.2 6.4 11.4 112.8 46,515
Alternative cancer treatment costs, Bolin et al (1999)a
No national screening – – 6.1 6.1 – –
Screening program 1.2 3.6 6.3 11.1 112.8 44,518
FOBT sensitivity and specifi city estimates
Alternative diagnostic accuracy–Nakazato et al (2006)
No national screening – – 6.2 6.2 – –
Screening program 1.2 3.9 6.5 11.6 122.9 44,005
Lower sensitivity for non-malignant polyps–10.6% (half of the base case estimate, see Table 38)
No national screening – – 6.2 6.2 – –
Screening program 1.2 2.4 6.4 10.0 108 35,549
Discount rate
No discounting
No national screening – – 13.9 13.9 – –
Screening program 1.7 4.9 14.5 21.1 381.2 19,112
Discounting at 10% per annum
No national screening – – 3.5 3.5 – –
Screening program 0.9 2.3 3.6 6.8 39 85,860
44
Health Economics Review of Bowel Cancer Screening in Australia
4. Financial implications
Average annual cost of bowel cancer treatment in screening ages 50 years and over was $191.3 million and $189.6 million without screening.
The estimated fi nancial implications of implementing the
National Bowel Cancer Screening Program were calculated.
The presented costs are for 10 years of the program.
Enrolment is based on recruiting eligible people as they
turn 55 or 65 years of age. Eligible screening participants
are re-invited biennially after initial screening. People who
have abnormalities detected by iFOBT are recommended to
receive diagnostic follow up using colonoscopy. An analysis
of the value for money represented by these costs to the
Australian healthcare system is presented in Section 3.
A cost-effectiveness analysis is reported as cost per
life-year saved.
An estimate was also derived for alternative eligibility age
groups considered in the cost-effectiveness analysis. These
alternative age groups included all people aged between 45
and 74 years, those between 50 and 74 years, and people
aged from 55 to 74 years.
The size of eligible population was fi rst determined for each
of the screening age scenarios (Section 4.1). The expected
resource requirements for each of the 10 years in focus
were estimated on the basis of expected participation rate,
iFOBT positivity rate, and colonoscopy follow up compliance
(Section 4.2). These resource requirements were then
applied, with unit cost estimates, to determine associated
costs (Section 4.2). The economic model was used to
derive the expected number of cancer detections each year
with and without the implementation of the program, to
determine annual costs of cancer treatment over the 10 year
period (Section 4.3).
4.1 National Bowel Cancer Screening Program
eligible population size
The size of the program’s eligible population – people who
turn 55 or 65 years of age each year, and who would receive
invitations to participate in the program – is shown in
Table 50. The number of people eligible to participate in the
program expands gradually over time as additional people
become eligible each year (approximately 45,000–57,000
people would enter annually).
The gradual rollout of the Program is expected to provide
overall coverage for 5.1 million people by the tenth year
following implementation.
The number of people invited by the program each year
is also presented in Table 50. Because participants are
invited for screening biennially, the number of invitations
sent out would be less than the population coverage each
year after the fi rst two years. These estimates represent a
slight overestimation because people with personal or family
histories of bowel cancer, and people with bowel cancer
symptoms, may not be invited to participate in the program,
but encouraged to undertake active surveillance. The
effects on the fi nal cost estimates resulting from this slight
overestimation are, however, likely to be negligible.
The size of eligible population under each alternative
eligibility age scenario – covering all people aged between
45 and 74 years, 50 and 74 years, and 55 and 74 years – can
be similarly determined. Under these age scenarios, people
become ineligible for screening as they reach 75 years of
age over the 10 year period, and others become newly
eligible annually. The number of people invited each year is
presented in Table 51. It was assumed that screening would
be initially rolled out to the entire eligible population over
a two year period, at the outset inviting roughly half of the
eligible population in the fi rst year and the remainder in the
second year. Screening was repeated biennially thereafter.
45
Tab
le 5
0
Siz
e of popula
tion e
ligib
le for
the n
atio
nal
bow
el s
creenin
g pro
gram
and n
um
ber
of in
vite
d p
eople
eac
h y
ear
So
urc
e: A
ust
ralia
n B
ure
au o
f St
atis
tics
(2003).
Bas
ed o
n t
he
2008 p
opula
tio
n e
stim
ate.
a.
Adju
sted fo
r th
e e
stim
ated n
um
ber
of deat
hs
eac
h y
ear
. b.
Year
1a
Year
2Ye
ar 3
Year
4Ye
ar 5
Year
6Ye
ar 7
Year
8Ye
ar 9
Year
10
Num
ber
turn
ing
55 o
r 65
yea
rs o
ld
Tota
l (new
invi
tations)
448,1
48
467,2
63
483,3
03
498,6
53
534,7
96
541,6
31
549,4
96
558,1
01
570,8
31
569,7
54
Num
ber
who
rem
ain
elig
ible
b
Invi
ted in
Yea
r 1
448,1
48
446,6
89
444,9
47
442,9
41
440,7
32
438,3
46
435,5
91
432,4
80
429,0
77
425,3
20
Invi
ted in
Yea
r 2
467,2
63
465,7
13
463,8
72
461,7
61
459,4
36
456,9
28
454,0
40
450,7
81
447,2
18
Invi
ted in
Yea
r 3
483,3
03
481,7
24
479,8
55
477,7
13
475,3
57
472,8
06
469,8
73
466,5
63
Invi
ted in
Yea
r 4
498,6
53
497,0
54
495,1
68
493,0
02
490,6
19
488,0
40
485,0
73
Invi
ted in
Yea
r 5
534,7
96
532,9
76
530,8
51
528,4
29
525,7
69
522,8
86
Invi
ted in
Yea
r 6
541,6
31
539,8
93
537,8
52
535,5
22
532,9
58
Invi
ted in
Yea
r 7
549,4
96
547,8
21
545,8
46
543,5
82
Invi
ted in
Yea
r 8
558,1
01
556,4
58
554,5
14
Invi
ted in
Yea
r 9
570,8
31
569,2
07
Invi
ted in
Yea
r 10
569,7
54
Cum
ula
tive
tota
l (sc
reen
ing
cove
rage
)
448,1
48
913,9
52
1,3
93,9
63
1,8
87,1
90
2,4
14,1
98
2,9
45,2
70
3,4
81,1
18
4,0
22,1
48
4,5
72,1
97
5,1
17,0
75
Num
ber
of p
eopl
e in
vite
d (a
nnua
l)
Nat
ional
448,1
48
467,2
63
928,2
50
962,5
25
1,4
55,3
83
1,4
96,2
35
1,9
91,2
95
2,0
40,6
12
2,5
41,3
96
2,5
89,5
17
46
Health Economics Review of Bowel Cancer Screening in Australia
Tab
le 5
1
Siz
e of popula
tion e
ligib
le for
the
nat
ional
bow
el s
creenin
g pro
gram
and n
um
ber
of in
vite
d p
eople
eac
h y
ear
– a
ltern
ativ
e e
ligib
ility
age
sce
nar
ios
So
urc
e: A
ust
ralia
n B
ure
au o
f St
atis
tics
(2003).
Bas
ed o
n t
he
2008 p
opula
tio
n e
stim
ate.
a.
Adju
sted fo
r th
e e
stim
ated n
um
ber
of deat
hs
eac
h y
ear
.b.
Initia
lly r
olle
d o
ut
to a
chie
ve c
om
ple
te p
opula
tio
n c
ove
rage
ove
r tw
o y
ear
s.c.
Year
1a
Year
2Ye
ar 3
Year
4Ye
ar 5
Year
6Ye
ar 7
Year
8Ye
ar 9
Year
10
Elig
ibili
ty a
ge g
roup
45–
74
Scre
enin
g co
vera
ge
(tota
l) b
6,9
14,3
36
7,0
82,9
62
7,2
41,4
63
7,3
86,5
95
7,5
26,3
34
7,6
67,5
91
7,8
14,6
95
7,9
63,0
35
8,1
27,9
33
8,2
79,9
99
Num
ber
of peo
ple
invi
ted (
annual
) c
– N
atio
nal
3,4
57,1
68
3,6
97,1
29
3,6
16,6
27
3,8
47,6
07
3,7
58,2
04
3,9
90,7
42
3,9
08,1
32
4,1
40,7
81
4,0
75,6
29
4,2
96,5
96
Elig
ibili
ty a
ge g
roup
50–
74
Scre
enin
g co
vera
ge
(tota
l) b
5,3
65,7
06
5,5
23,6
33
5,6
82,5
93
5,8
39,6
33
5,9
92,5
03
6,1
43,9
33
6,2
86,3
00
6,4
17,8
32
6,5
41,0
18
6,6
55,3
02
Num
ber
of peo
ple
invi
ted (
annual
) c
– N
atio
nal
2,6
82,8
53
2,9
12,7
90
2,8
43,0
98
3,0
74,9
89
2,9
97,9
75
3,2
28,2
28
3,1
43,3
47
3,3
61,3
02
3,2
69,3
73
3,4
79,0
89
Elig
ibili
ty a
ge g
roup
55–
74
Scre
enin
g co
vera
ge
(tota
l) b
3,9
62,4
07
4,0
87,5
43
4,2
17,0
61
4,3
43,4
27
4,4
64,0
03
4,5
92,5
43
4,7
23,9
16
4,8
55,5
26
4,9
90,1
47
5,1
17,0
75
Num
ber
of peo
ple
invi
ted (
annual
) c
– N
atio
nal
1,9
81,2
04
2,1
78,1
90
2,1
12,1
70
2,3
09,5
39
2,2
34,9
69
2,4
39,7
15
2,3
69,5
17
2,5
72,7
41
2,5
07,2
17
2,7
03,0
30
47
4.2 Estimated extent of resource requirements
and associated fi nancial implications
The expected extent of resource requirements associated
with implementation of the screening program was
estimated. This was performed using the participation rate,
iFOBT positivity rate and compliance with recommended
diagnostic follow up observed during the Pilot study,
as shown in Table 52. The incidence of colonoscopy
complications was estimated using information presented by
Viiala et al (2003) as per the cost-effectiveness model.27
The likely extent of resource requirements associated
with the National Bowel Cancer Screening Program each
year over the fi rst 10 years of screening program can be
determined by combining these estimates with the expected
number of iFOBT invitations (Table 53).
These resource requirements do not account for treatment
of bowel cancers detected through the screening program.
The costs of treatment arising from screening were
estimated in Section 4 using the economic model described
in Section 3.
The estimated extents of resource requirements for other
alternative age scenarios were similarly determined, as shown
in Table 54–Table 56. Due to the wider population coverage
under these scenarios, especially during the early years of
implementation, the resource requirements were signifi cantly
more extensive than the current program (Table 53). To
exemplify, the current program was estimated to invite about
1.5 million people by the fi fth year, increasing to 2.6 million
people by the tenth year; this estimates were found to be
3.8 million in the fi fth year and 4.3 million in the tenth year if
the program was to target all people aged 45 to 74 years old
(Table 54).
The estimated fi nancial implications of implementing the
National Bowel Cancer Screening Program were determined,
as presented in Table 57. Costs are presented for 10 years
of the program, and accommodate eligible people as they
turn 55 or 65 years of age who would be invited to repeat
screening biennially until they reach age 75years, pending
their continuing eligibility.
The total costs of the program are estimated to be $21.9
million in Year 1, increasing to $126.3 million by Year 10 as
screening coverage extends over time.
At the national level, iFOBT and pathology accounted
for approximately 40% of annual total costs; and the
remaining 60% of total costs is attributable to activities
relating to diagnostic follow up, including GP consultations,
colonoscopies, polypectomies and management of
colonoscopy complications. These estimates were based
on the positivity rate observed during the Pilot program
(9%). The screening program will also involve ongoing
administrative, coordination, and management costs. These
costs were estimated to be $0.8 million in Year 1, increasing
to $4.5 million by Year 10, as screening coverage extends
over time. Estimates were based on experiences from the
current cervical cancer screening program (Table 57). In
practice, administrative and information management costs
are unlikely to increase proportionally to screening coverage.
These values may therefore require further review as
additional information becomes available.
Some people at average risk are currently screened using
colonoscopy.58 It is uncertain if such screening colonoscopies
would be disallowed by the program, and if so, to what
extent. An improvement in community awareness about
bowel cancer may increase interest in screening colonoscopy.
The likely extent of fi nancial implications from a national
screening program targeting all people aged between 45 and
74 years, between 50 and 74 years, and between 55 and 74
years over ten years, respectively, was similarly determined,
as presented in Table 58. Screening was assumed to be
rolled out initially to the entire eligible population over two
years, covering about half the eligible population in the fi rst
year and the remainder in the second year.
As expected, when compared with the current program
where screening is gradually rolled out, the costs of screening
implementation were estimated to be considerably more
extensive, especially during the early years.
The national costs of a screening program targeting all people
aged between 45 and 74 years was estimated to be $168.6
million in Year 1, increasing to $209.5 million by Year 10.
48
Health Economics Review of Bowel Cancer Screening in Australia
The national costs for a screening program targeting
all people aged between 50 and 74 years of age, were
estimated to be $130.8 million in Year 1, increasing to $169.7
million by Year 10.
The national costs of a screening program targeting all
people aged between 55 and 74 years was estimated to be
$96.6 million in Year 1, increasing to $131.8 million by Year 10.
Table 52 Assumptions in the estimation of screening resource requirements
Health Economics Review of Bowel Cancer Screening in Australia
Tab
le 5
4
Est
imat
ed r
eso
urc
e re
quirem
ents
of th
e s
creenin
g pro
gram
for
year
s 1–10
(ag
e e
ligib
ility
betw
een 4
5 a
nd 7
4 y
ear
s)
Abbre
viat
ion: i
FOBT, i
mm
uno
chem
ical
fae
cal o
ccult b
loo
d t
est
.
Po
lypect
om
y is p
erf
orm
ed w
here
nece
ssar
y–20%
of al
l co
lonosc
opie
s—bas
ed o
n t
he
est
imat
ed p
reva
lence
of po
lyps.
a.
Uni
ts
Year
1Ye
ar 2
Year
3Ye
ar 4
Year
5Ye
ar 6
Year
7Ye
ar 8
Year
9Ye
ar 1
0
Nat
iona
l est
imat
es
iFO
BT
invi
tation (
see
Table
50)
3,4
57,1
68
3,6
97,1
29
3,6
16,6
27
3,8
47,6
07
3,7
58,2
04
3,9
90,7
42
3,9
08,1
32
4,1
40,7
81
4,0
75,6
29
4,2
96,5
96
Com
ple
ted
and r
eturn
ed
iFO
BT
s
1,5
73,0
11
1,6
82,1
94
1,6
45,5
65
1,7
50,6
61
1,7
09,9
83
1,8
15,7
87
1,7
78,2
00
1,8
84,0
55
1,8
54,4
11
1,9
54,9
51
Num
ber
of
positive
iFO
BT
s
141,5
71
151,3
97
148,1
01
157,5
59
153,8
98
163,4
21
160,0
38
169,5
65
166,8
97
175,9
46
GP
consu
ltat
ions
due
to p
ositive
iFO
BT
87,9
16
94,0
18
91,9
71
97,8
44
95,5
71
101,4
84
99,3
84
105,3
00
103,6
43
109,2
62
Dia
gnost
ic
colo
nosc
opy
a
77,8
64
83,2
69
81,4
55
86,6
58
84,6
44
89,8
81
88,0
21
93,2
61
91,7
93
96,7
70
Colo
nosc
opy
com
plic
atio
n
78
83
81
87
85
90
88
93
92
97
51
Tab
le 5
5
Est
imat
ed r
eso
urc
e re
quirem
ents
of th
e sc
reenin
g pro
gram
for
year
s 1–10
(ag
e elig
ibili
ty b
etw
een 5
0 a
nd 7
4 y
ear
s)
Abbre
viat
ion: i
FOBT, i
mm
uno
chem
ical
fae
cal o
ccult b
loo
d t
est
.
Po
lypect
om
y is p
erf
orm
ed w
here
nece
ssar
y – 2
0%
of al
l co
lonosc
opie
s – b
ased o
n t
he
est
imat
ed p
reva
lence
of po
lyps.
a.
Uni
ts
Year
1Ye
ar 2
Year
3Ye
ar 4
Year
5Ye
ar 6
Year
7Ye
ar 8
Year
9Ye
ar 1
0
Nat
iona
l est
imat
es
iFO
BT
invi
tation (
see
Table
50)
2,6
82,8
53
2,9
12,7
90
2,8
43,0
98
3,0
74,9
89
2,9
97,9
75
3,2
28,2
28
3,1
43,3
47
3,3
61,3
02
3,2
69,3
73
3,4
79,0
89
Com
ple
ted
and r
eturn
ed
iFO
BT
s
1,2
20,6
98
1,3
25,3
19
1,2
93,6
10
1,3
99,1
20
1,3
64,0
78
1,4
68,8
44
1,4
30,2
23
1,5
29,3
92
1,4
87,5
64
1,5
82,9
85
Num
ber
of
positive
iFO
BT
s
109,8
63
119,2
79
116,4
25
125,9
21
122,7
67
132,1
96
128,7
20
137,6
45
133,8
81
142,4
69
GP
consu
ltat
ions
due
to p
ositive
iFO
BT
68,2
25
74,0
72
72,3
00
78,1
97
76,2
38
82,0
94
79,9
35
85,4
78
83,1
40
88,4
73
Dia
gnost
ic
colo
nosc
opy
a
60,4
25
65,6
03
64,0
34
69,2
56
67,5
22
72,7
08
70,7
96
75,7
05
73,6
34
78,3
58
Colo
nosc
opy
com
plic
atio
n
60
66
64
69
68
73
71
76
74
78
52
Health Economics Review of Bowel Cancer Screening in Australia
Tab
le 5
6
Est
imat
ed r
eso
urc
e re
quirem
ents
of th
e sc
reenin
g pro
gram
for
year
s 1–10
(ag
e elig
ibili
ty b
etw
een 5
5 a
nd 7
4 y
ear
s)
Abbre
viat
ion: i
FOBT, i
mm
uno
chem
ical
fae
cal o
ccult b
loo
d t
est
.
Po
lypect
om
y is p
erf
orm
ed w
here
nece
ssar
y – 2
0%
of al
l co
lonosc
opie
s – b
ased o
n t
he
est
imat
ed p
reva
lence
of po
lyps.
a.
Uni
ts
Year
1Ye
ar 2
Year
3Ye
ar 4
Year
5Ye
ar 6
Year
7Ye
ar 8
Year
9Ye
ar 1
0
Nat
iona
l est
imat
es
iFO
BT
invi
tation (
see
Table
50)
1,9
81,2
04
2,1
78,1
90
2,1
12,1
70
2,3
09,5
39
2,2
34,9
69
2,4
39,7
15
2,3
69,5
17
2,5
72,7
41
2,5
07,2
17
2,7
03,0
30
Com
ple
ted
and r
eturn
ed
iFO
BT
s
901,4
48
991,0
76
961,0
37
1,0
50,8
40
1,0
16,9
11
1,1
10,0
70
1,0
78,1
30
1,1
70,5
97
1,1
40,7
84
1,2
29,8
78
Num
ber
of
positive
iFO
BT
s
81,1
30
89,1
97
86,4
93
94,5
76
91,5
22
99,9
06
97,0
32
105,3
54
102,6
71
110,6
89
GP
consu
ltat
ions
due
to p
ositive
iFO
BT
50,3
82
55,3
91
53,7
12
58,7
31
56,8
35
62,0
42
60,2
57
65,4
25
63,7
58
68,7
38
Dia
gnost
ic
colo
nosc
opy
a
44,6
22
49,0
58
47,5
71
52,0
17
50,3
37
54,9
48
53,3
67
57,9
45
56,4
69
60,8
79
Colo
nosc
opy
com
plic
atio
n
45
49
48
52
50
55
53
58
56
61
53
Tab
le 5
7
Est
imat
ed c
ost
s of th
e sc
reenin
g pro
gram
for
year
s 1–10
(cu
rrent
age
elig
ibili
ty –
initia
l invi
tation a
t 55 a
nd 6
5 y
ear
s of ag
e)
Note
: These
cost
est
imat
es
were
not
dis
counte
d.
Cost
of iF
OBT
($10
) in
cludes
supply
of te
st k
its,
post
ages
and r
em
inder
lett
er,
and o
ther
coo
rdin
atio
n c
ost
s. A
n a
dditio
nal
$20 fo
r pat
ho
logy
and info
rmat
ion m
anag
em
ent
is
a.
incu
rred fo
r eac
h t
est
co
mple
ted a
nd r
etu
rned b
y th
e par
tici
pan
t.
Unit c
ost
est
imat
es
for
colo
nosc
opy
and p
oly
pect
om
y w
ere
bas
ed o
n t
he
Nat
ional
Hosp
ital
Cost
Dat
a C
olle
ctio
n C
ost
Repo
rt R
ound 7
Public
Sect
or
($1,
082; a
dditio
nal
$524
b.
with p
oly
pect
om
y)26;.
Cost
s of G
P c
onsu
ltat
ions
were
als
o incl
uded (
$32.1
; Leve
l B G
P c
onsu
ltat
ion)
. Inci
dence
of ad
vers
e e
vents
(perf
ora
tio
n) w
as e
stim
ated u
sing
a ri
sk o
f 0.0
01.2
7 C
ost
s of perf
ora
tio
n w
ere
bas
ed o
n info
rmat
ion p
rese
nte
d b
y O
’Lear
y et
al (
2004)2
2, a
dju
sted t
o 2
004 p
rice
s ($
17,6
62; A
IHW
2006).
9%
of FO
BT
scr
eenin
g co
sts.
This w
as b
ased o
n n
atio
nal
po
pula
tio
n c
erv
ical
scr
eenin
g dat
a.c.
2
Cos
t ($
mill
ion)
Year
1Ye
ar 2
Year
3Ye
ar 4
Year
5Ye
ar 6
Year
7Ye
ar 8
Year
9Ye
ar 1
0
Cur
rent
age
elig
ibili
ty—
initi
al in
vita
tion
at 5
5 an
d 65
yea
rs o
f age
Nat
ional
est
imat
es
Scre
enin
g a
8.6
8.9
17.7
18.4
27.8
28.6
38.0
39.0
48.5
49.5
Dia
gnost
ic
follo
w u
p b
12.5
13.1
25.9
26.9
40.7
41.8
55.6
57.0
71.0
72.4
Dev
elopm
ent/
coord
inat
ion
cost
s c
0.8
0.8
1.6
1.7
2.5
2.6
3.4
3.5
4.4
4.5
Tota
l – n
atio
nal
21.9
22.8
45.3
46.9
71.0
73.0
97.1
99.5
123.9
126.3
54
Health Economics Review of Bowel Cancer Screening in Australia
Tab
le 5
8
Est
imat
ed c
ost
s of th
e s
creenin
g pro
gram
for
year
s 1–10
(va
rious
elig
ibili
ty a
ges)
Note
: These
cost
est
imat
es
were
not
dis
counte
d.
Cos
t ($
mill
ion)
Year
1Ye
ar 2
Year
3Ye
ar 4
Year
5Ye
ar 6
Year
7Ye
ar 8
Year
9Ye
ar 1
0
Age
elig
ibili
ty b
etw
een
45 a
nd 7
4 ye
ars
Nat
ional
est
imat
es
Scre
enin
g 66.0
70.6
69.1
73.5
71.8
76.2
74.6
79.1
77.8
82.1
Dia
gnost
ic
follo
w u
p
96.6
103.3
101.1
107.5
105.0
111.5
109.2
115.7
113.9
120.1
Dev
elopm
ent/
coord
inat
ion
cost
s
5.9
6.4
6.2
6.6
6.5
6.9
6.7
7.1
7.0
7.4
Tota
l–nat
ional
168.6
180.3
176.4
187.6
183.3
194.6
190.6
201.9
198.7
209.5
Age
elig
ibili
ty b
etw
een 5
0 a
nd 7
4 y
ears
Nat
ional
est
imat
es
Scre
enin
g 51.2
55.6
54.3
58.7
57.3
61.7
60.0
64.2
62.4
66.5
Dia
gnost
ic
follo
w u
p
75.0
81.4
79.4
85.9
83.8
90.2
87.8
93.9
91.4
97.2
Dev
elopm
ent/
coord
inat
ion
cost
s
4.6
5.0
4.9
5.3
5.2
5.5
5.4
5.8
5.6
6.0
Tota
l–nat
ional
130.8
142.0
138.6
149.9
146.2
157.4
153.3
163.9
159.4
169.7
Age
elig
ibili
ty b
etw
een 5
5 a
nd 7
4 y
ears
Nat
ional
est
imat
es
Scre
enin
g 37.8
41.6
40.3
44.1
42.7
46.6
45.3
49.1
47.9
51.6
Dia
gnost
ic
follo
w u
p
55.4
60.9
59.0
64.5
62.5
68.2
66.2
71.9
70.1
75.5
Dev
elopm
ent/
coord
inat
ion
cost
s
3.4
3.7
3.6
4.0
3.8
4.2
4.1
4.4
4.3
4.6
Tota
l–nat
ional
96.6
106.2
103.0
112.6
109.0
119.0
115.5
125.5
122.3
131.8
55
4.3 Estimated number of cancer detection and
cancer treatment costs
Bowel cancer screening aims to detect cancers and
abnormalities with potential for malignancy. Screening
also aims to improve disease outcomes by detecting early
stage disease. These factors underpin the foundations of
health benefi ts provided by screening: reducing mortality
and morbidity caused by bowel cancer. It is important to
acknowledge that screening infl uences healthcare resource
requirements associated with bowel cancer treatment.
As well as fi nancial costs directly related to implementing
a screening program (Section 4.2), the costs of cancer
treatment should also be investigated.
The economic model was used to estimate the expected
number of cancers detected and associated costs of cancer
treatment both in the presence and absence of the screening
program. Details of the economic model are described
in Section 3. The participation rate observed in the Pilot
(45.4%) was applied. Model outputs were then infl ated
to the national levels, based on the size of the eligible
populations as presented in Table 50.
A slight increase in the costs of cancer treatment was
simulated to occur with the implementation of the program
under the 55 and 65 years scenario, as shown in
Table 59. The national costs of bowel cancer treatment were
estimated to be $31.4 million in Year 1, increasing to $191.5
million by Year 10. In the absence of a nationally coordinated
screening program, the costs of cancer treatment among the
population who would have participated were estimated to
be $16.6 million in Year 1, increasing to $189.4 million in
Year 10.
The model predicted that the national costs of bowel cancer
treatment would be approximately $190 million in the tenth
year, by which time full coverage of the population aged
55–74 years would be achieved. The Australian Institute
of Health and Welfare reported that the national costs of
bowel cancer were $162.5 million in 1993–1994 and $235.1
million in 2000–2001.xv If it is considered that the modelled
age groups account for the ranges where most bowel
cancers would be expected to occur, the current estimates
can be accepted as reasonable.
It was observed that the simulated annual incidence of
cancer diagnoses in the modelled cohort that resulted from
symptomatic presentation, and that therefore required
diagnostic investigation, fl uctuated slightly in both the
screening and non-screening arms. These fl uctuations
represent an inevitable feature of simulation-based analysis,
and therefore, estimated annual fi gures should be regarded
as indicative (see Table 59).
Over 10 years, the average annual national costs of bowel
cancer treatment among people aged between 45 and 74
years; 50 and 74 years and 55 and 74 years were estimated
to be $247.5 million, $191.3 million and $138.9 million,
respectively, with implementation of the national screening
program. Without the screening program, these costs were
estimated to be $244.7 million, $189.6 million and $138.0
million for people aged between 45 and 74 years, between
50 and 74 years, and between 55 and 74 years, respectively.
Estimates were derived using model outputs under the
50–75 years scenario. The biggest component of cost
differences between programs targeting the three age
groups is the eligible population sizes (see Table 51). Hence,
this approach, although a proxy, appropriately captured
the relative extent of fi nancial implications associated with
alternative screening populations.
Under the 55 and 65 years scenario, the model predicted
the annual incidence of bowel cancer diagnoses to average
15.1 cases per 10,000 people over a 10 year period with
no screening program. Australian Institute of Health and
Welfare data indicated similar incidence for the age groups
under consideration.xvi The model predicted that the
program would create a shift in cancer stages at diagnosis
(Figure 6). In the absence of the program, 32% of cancer
diagnoses were expected to occur at earlier stages (Dukes’ A
and B). In contrast, implementation of the program escalated
the anticipated proportion of early diagnoses to 42%. This
estimate supports the overall benefi t of the program: early
cancer diagnoses are generally associated with superior
survival and less intensive treatment.
AIHW 2005.xv.
AIHW 2004.xvi.
56
Health Economics Review of Bowel Cancer Screening in Australia
Figure 6 Incidence of bowel cancer (diagnosed) and extent of disease at diagnosis – simulation results
Table 59 Estimated costs of cancer treatment for years 1–10
(current age eligibility – initial invitation at 55 and 65 years)a
Costs of cancer treatment based on O’Leary et al (2004)a. 22, adjusted to 2004 prices (AIHW 2006). The surveillance costs in patients with a history of neoplasm diagnosis are not included.
For simplicity, the number of cancers detected in the simulation cohort each year was assumed to remain constant at the mean value of annual estimates b. over 10 years.
0
20
40
60
80
100
120
140
160
Total cancer Dukes A Dukes B Dukes C Dukes D
Cancer stage at diagnosis
Can
cer
diag
nosi
s (p
er 1
0,00
0; 1
0-ye
ar t
otal
)
Screening
No screening
Costs ($ million)
Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9 Year 10
National Bowel Cancer Screening Program (Pilot participation rate: 45%)
Total costs of bowel cancer treatment
National total 31.4 49.1 69.5 81.8 99.5 113.4 133.3 150.5 174.5 191.5
No national screening programb
Total costs of bowel cancer treatment
National total 16.6 33.8 51.6 69.8 89.3 109.0 128.8 148.9 169.2 189.4
Cost differences
National total 14.8 15.3 17.9 12.0 10.2 4.4 4.5 1.7 5.3 2.2
57
Findings from the current evaluation are discussed in this
section. The main fi ndings were:
The systematic review conducted on the effi cacy of ■bowel cancer screening provided high quality evidence
on the effectiveness of the bowel cancer screening
program. Results from three large international
randomised controlled trials – Minnesota, USA (1993),
Funen, Denmark (1996), and Nottingham, UK (1996)
– found that compared with no screening, biennial
screening was associated with bowel cancer mortality
reductions of 13–17% over follow up periods between
11.7 and 18 years (p<0001).3-19 These results highlight
the importance of early detection in leading improved
health outcomes.
A systematic literature review was conduced to ■address test accuracies of the instruments used in the
program – iFOBT and colonoscopy. Nakazato et al
(2006) conducted a cross-sectional analysis of 3,090
asymptomatic people involving iFOBT with follow up
colonoscopy.22 The reported iFOBT sensitivity was
52.6% and specifi city 87.2% for bowel cancer, and 24.5%
and 87.2% for adenomas, respectively. It has been
recommended that people with a positive iFOBT result
on one or more occasion be referred for colonoscopy
follow up.2
Colonoscopy is considered to be the gold standard to ■detect adenomas and bowel cancers. There is little
evidence in the literature concerning colonoscopy
accuracy. NHMRC guidelines have estimated
colonoscopy sensitivity and specifi city at 95% and 100%
respectively.21 A number of retrospective reviews of
medical records concluded that colonoscopy,
performed under sedation, is considered safe and has
few complications.
The cost-effectiveness of bowel cancer screening using ■FOBT has been assessed by numerous studies.2,22-24 The
National Bowel Cancer Screening Program, which is
currently phased into the targeted population through
selective invitations to people turning 55 or 65 years of
age, would eventually cover all people aged between 55
and 74 years should continuing funding be available.
In the long term, bowel screening could be considered to provide value for money for the Australian health system.
5. Discussion and recommendations
The current modelled economic analysis indicated that
in the long term the program could be considered to
provide value for money for the Australian health care
system. Other eligibility age scenarios – screening
coverage for people aged between 45 and 74 years
and 50 and 74 years – were also investigated. Under
all considered scenarios, the provision of bowel cancer
screening was demonstrated to be cost-effective by the
current model.
Gradual expansion of screening coverage as was ■implemented for the current program also created a
gradual increase in resource requirements and associated
fi nancial costs over time. Achieving immediate coverage
of the targeted population, regardless of eligibility age
criteria – age groups between 45 and 74 years, 50 and
74 years or 55 and 74 years – would create a sudden
infl ux of healthcare resource demands and substantial
fi nancial outlays, which is likely to undermine the
practicality and feasibility of the screening program.
Issues such as screening’s impact on quality of life or ■indirect economic effects, which the current analysis
has not explored so far, are also discussed. These
issues should be considered when evaluating the overall
implications of screening implementation may have on
the welfare of Australian people.
58
Health Economics Review of Bowel Cancer Screening in Australia
5.1 Evidence from the literature
Bowel cancer is the second most frequently occurring cancer
in Australia (excluding non-melanocytic skin cancers), and the
second leading cause of cancer death. The National Bowel
Cancer Screening Program was introduced in 2006, following
the success of the Bowel Cancer Screening Pilot Program.
The national program’s objective is to improve detection
rates of pre-malignant bowel abnormalities, or early stage
cancers, and to initiate appropriate follow up and treatment.
Results from three large international randomised
controlled trials (RCTs) – Minnesota, USA, Funen, Denmark,
Nottingham, UK – identifi ed from the systematic review
indicated that early detection of adenomas and bowel
cancers contribute to lower mortality rates.3-19 These
fi ndings are dependent on variables such as screening
instrument accuracies and participation.
The National Bowel Cancer Screening Program provides
immunochemical faecal occult blood testing (iFOBT) to
asymptomatic people in the targeted age groups in the
Australian community. Participants with a positive FOBT
are notifi ed and encouraged to seek follow up colonoscopy.
iFOBT is used in the fi rst instance to screen for disease
symptoms. This allows for endoscopic resources to be
allocated to those who are more likely to benefi t. It is also
more likely to facilitate program participation because it
is more user friendly compared with invasive procedures
such as fl exible sigmoidoscopy and colonoscopy. This is
consistent with recommendation from the Pilot evaluation
report.2 iFOBT was selected over the guaiac test for use
in the screening program because it offered some distinct
advantages: iFOBT does not require users to modify dietary
intake before use, was considered to be more reliable than
guaiac testing and was better accepted by users. Guaiac
testing requires dietary restrictions, including avoiding red
meat and some fruits and vegetables; no supplementary
vitamin C, aspirin, anti-infl ammatory or anticoagulant drugs
for at least three days before and throughout testing period.
Evidence of iFOBT accuracy among a general asymptomatic
population was limited because trial designs do not often
incorporate follow up of participants with negative iFOBT
results. Findings from a study by Nakazato et al (2006),
who did follow up asymptomatic patients with colonoscopy,
indicated that iFOBT correctly detected 52.6% of bowel
cancers and 24.5% of large adenomas (>10 mm diameter).22
This study also reported that iFOBT was 87% specifi c
for detecting adenomas and cancers. iFOBT detects
about 50% of bowel cancers and about 75% of potentially
progressive abnormalities among people screened using this
technology. The likelihood of returning positive results in the
absence of these abnormalities was 13%.22 As a result, it is
recommended that people with positive iFOBT results on
one or more occasions are referred for colonoscopy follow
up, as is practiced by the National Bowel Cancer
Screening Program.
Colonoscopy accuracy evidence was sought, but the
systematic literature review did not identify any RCTs that
investigated colonoscopy accuracy in an asymptomatic
population. An asymptomatic population would have
been ideal to develop more accurate assumptions, because
symptomatic populations would lead to overestimation of
the procedure’s accuracy. Asymptomatic participants are not
recalled for follow up colonoscopy. The NHMRC reports
95% and 85% colonoscopy sensitivity for bowel cancer and
85% for adenoma detection, respectively; specifi city was
100%.21 As with any diagnostic procedure, overall safety
needs to be addressed and complication risks identifi ed. A
review of retrospective studies of medical records found
that colonoscopies conducted by trained endoscopists,
gastroenterologists, colorectal or general surgeons, to be
safe, well tolerated and with few complications, such as
bowel wall perforation and bleeding. Mortality of between
zero and fi ve per 10,000 procedures was observed. Costs
involved in the implementation of a bowel cancer screening
program should consider need for trained specialists to
conduct safe colonoscopy procedures and to counteract the
increase in demand for colonoscopy services as observed in
the Pilot. The Pilot report identifi ed that colonoscopy results
were not available from the register at the time of the
Pilot evaluation.
59
Neither iFOBT nor colonoscopy can detect all adenomas
and cancers in people who are screened. This creates
implications for prognoses and healthcare resources.
There is a heightened need to defi ne optimum screening
intervals considering the absence of screening technologies
that are 100% accurate while also safe, cost-effective and
acceptable to the targeted population. The National Bowel
Cancer Screening Program has therefore recommended
that participants be screened for bowel cancer biennially.
Biennial bowel cancer screening was associated with
13–17% mortality reduction as a result of early detection
and intervention. It is also recommended that this policy
be reviewed as new evidence on screening outcomes
and bowel cancer related mortality comparing annual and
biennial screening intervals comes to light. Re-screening is
an important element of the program because undetected
adenomas and bowel cancers at an initial screen can develop
further. New adenomas or bowel cancers can also develop
between screenings. A dramatic increase in bowel cancer
incidence is reported among people in the targeted screening
eligibility age range, further supporting the implementation of
frequent re-screening.
Screening program participation rates have a signifi cant
infl uence on reducing bowel cancer mortality. The Pilot
program achieved a participation rate of 45.4%. The three
key RCTs on FOBT bowel cancer screening identifi ed
in the systematic review reported fi rst screening round
participation rates of 66.8% and 53.4% in the Funen,
Denmark and Nottingham, UK trials, respectively; and in the
Minnesota trial, 78% of people participated in at least one
of six screening rounds.3-19 Participants who attended the
fi rst round were more likely to attend the next round of
screening, compared with participants who did not initially
attend. Participants who attended the fi rst round of the
National Bowel Cancer Screening Program were invited
to the next and subsequent rounds of screening. The
colonoscopy follow up procedure was performed in 82.3%,
86.7% and 81.7% of participants with positive FOBT results in
the Funen, Nottingham and Minnesota trials, respectively.3-19
This was reported to be lower in the Pilot program (55%),
although incomplete data were available to correctly
determine the follow up compliance rate.2
5.2 Cost-effectiveness of the National Bowel
Cancer Screening Program
The cost-effectiveness analysis indicated that implementation
of the National Bowel Cancer Screening Program represents
value for money for the Australian healthcare system. The
current model-based evaluation indicated incremental cost-
effectiveness of approximately $48,921 per life-year saved,
relative to no screening. This fi nding relates to a modelled
program in which people turning 55 or 65 years of age were
fi rst invited to participate, and invitations to re-screen were
repeated biennially thereafter until participants reached the
age of 75 years. The cohort’s base line age was distributed
between 50 and 74 years in accordance with the current
population data.
The implementation of screening was found to be generally
cost-effective and, importantly, likely to be so in the long
run. The current program is being gradually phased in by
targeting those who turn 55 or 65 years each year, which
would eventually achieve a complete screening coverage for
Australians aged between 55 and 74 years if the program
continues to be implemented. The long-term cost-
effectiveness of the current program was estimated to be
$41,321 per life year saved.
Other eligibility age scenarios – fi rst invitation at 45 and
50 years of age – were also investigated. The screening
program was shown to be cost-effective under
these scenarios.
A value of $50,000 per life-year saved is generally regarded
as an upper threshold for acceptable cost-effectiveness of
pharmaceutical treatments in the Australian healthcare
system. It should be acknowledged that factors other than
cost-effectiveness, such as quality of clinical and health
economic data and clinical needs, contribute to funding
decisions made by the Pharmaceutical Benefi ts Advisory
Committee (PBAC). The cost-effectiveness of drugs
considered for reimbursement by the PBAC, with decisions
regarding listing, is shown in Table 60.57
The current evaluation clearly indicated that screening
reduces mortality and thus, generated additional life
60
Health Economics Review of Bowel Cancer Screening in Australia
years among the screened population. Screening was
demonstrated to represent a cost-effective strategy in
the long run in all eligibility age groups considered. The
practicality and feasibility of expanding the eligibility age
should be however assessed in relation to the additional
healthcare resource requirements and associated
fi nancial costs.
Assumptions made to calculate the cost-effectiveness
were tested in sensitivity analyses, presented in Section 3.2.2. Cancer survival estimates were found to be a key
determinant of cost-effectiveness expressed in terms of
life-years saved. This is because the incremental survival
benefi ts provided by detecting cancers declines as the chance
of survival increases, and vice versa. The current model
suggests that the program would produce similar costs per
cancer detected regardless of the cancer survival rate used in
the model.
The base case analysis was conducted using Australian
estimates reported by McLeish et al (2002) – 90%, 80%,
35% and 0% for Dukes’ A to Dukes’ D stage bowel cancer,
respectively.25 Findings reported by McLeish et al (2002)
refl ected cancer treatment from the late 1980s to early
1990s. Recent estimates from the American Cancer Society
(2007)xvii may be interpreted to indicate more favourable
fi ve-year survival estimates than McLeish et al (2002)25
(Table 47). Overall, bowel cancer survival in the US was
reported to be about fi ve percentage points higher than
in Australia.xviii The current analysis should be revised as
more recent cancer survival data become available for the
Australian population.
Reliable comparison between the current analysis and
the available cost-effectiveness evidence in the literature
is diffi cult because of differences in assumptions made
and approaches used. Several Australian studies have
demonstrated the cost-effectiveness of bowel cancer
screening using FOBT (Table 61). The incremental cost-
effectiveness ratios of FOBT screening presented in the
literature varies, depending on underlying assumptions and
screening populations under consideration. Nonetheless,
most studies demonstrate FOBT screening to be
cost-effective.
The current estimate of the cost-effectiveness of FOBT
screening programs was within the range of values reported
by other Australian studies (Table 61). Key differences
between previous analyses and the current approach are also
presented in Table 61. Most previous studies analysed the
cost-effectiveness of older, guaiac FOBTs; immunochemical
FOBTs (iFOBTs) were used in the Pilot study and in the
current national program.
The Pilot evaluation report suggested that the national
program represented a cost-effective intervention, although
it reported lower incremental cost-effectiveness ratios.2
Although the models used to conduct the simulation were
similar to the current analysis and the Pilot study, some
data inputs differed. Nonetheless, results from the current
analysis further reinforced that national bowel cancer
screening is likely to offer value for money in the Australian
healthcare system.
Other tests, such as fl exible sigmoidoscopy or colonoscopy,
can be used to screen for bowel cancer. These methods
have greater sensitivity and specifi city to detect
abnormalities, and are therefore potentially more effective
than FOBT screening (Table 61). However, they are also
signifi cantly more expensive and would require far greater
healthcare resources to undertake a screening program
based on these methods. They are also invasive procedures,
and less acceptable to the population as a national screening
strategy, which would therefore compromise participation.
The cost-effectiveness fi gures for the Australian national
breast cancer and cervical cancer screening programs have
been reported as $7,879–13,132 and $36,749 per life-year
saved, respectively (Australian Health Ministers’ Advisory
Council [AHMAC] 1990 cited by O’Leary et al 2004 at
1995–1996 prices).22 Gyrd-Hansen (1999) estimated that the
overall cost-effectiveness of bowel cancer screening using
FOBT was superior to current (at that time) breast cancer or
cervical cancer screening programs in a Danish setting, based
on data from the Funen randomised FOBT trial.59
American Cancer Society. Detailed Guide: Colon and Rectum Cancer [Online]. 2007; xvii. URL: http://www.cancer.org/docroot/CRI/content/CRI_2_4_3X_How_is_colon_and_rectum_cancer_staged.asp?sitearea
AIHW 2001.xviii.
61
Table 60 Incremental cost per life-year saved for drugs considered by the PBAC for reimbursement under the
Pharmaceutical Benefi ts Scheme
Source: George et al (2001).57
Values reported in the submissions. All values were adjusted to 2004 prices.a.