Download Download

Setting Standards for Access to
Radiotherapy
1
Mackillop WJ, 1Chen Z, 3Pearcey R, 1King W, 1Kerba M, 2 Shortt S,
1Feldman-Stewart D, 4Brouwers M, 1Brundage M, 5Coldman A
1
Cancer Research Institute, and 2Centre for Health Services Research and
Policy, Queen’s University, Kingston; 3Cross Cancer Research Institute,
Edmonton; 4Program in Evidence-Based Care, CCO/ McMaster; 5BC
Cancer Agency, Vancouver
In collaboration with Neil Berman, George Browman, Timothy Caulfield, Steve Hall, Mike
Milosevic, Tom Pickles, Brent Schacter,
and Simon Sutcliffe
Taming of the Queue III March 2006
Objectives of this talk
ƒ To report on our progress in setting “evidence-based
benchmarks” for waiting times for RT and discuss next
steps in this process
ƒ To demonstrate that waiting times alone are an
inadequate measure of access to care
ƒ To demonstrate that setting benchmarks and monitoring
compliance with them, is not useful unless it leads to
effective action to reduce waiting times
Radiotherapy (RT)
• RT = Cancer treatment using ionizing radiation
• It has diverse roles, e.g.
– It may be used alone to cure some localized cancers
(head and neck cancers, cervix)
– It may used after surgery to reduce the risk of cancer
recurrence (breast, rectum)
– It may be used to control symptoms in people with
incurable cancers
• Overall, RT is useful in about 50% of patients.
– Most common indications for RT are supported by
level 1 evidence
Objectives of this talk
ƒ To report on our progress in setting “evidence-based
benchmarks” for waiting times for RT and discuss next
steps in this process
ƒ To demonstrate that waiting times alone are an
inadequate measure of access to care
ƒ To remind ourselves that setting benchmarks and
monitoring compliance with them is worthless unless it
leads to effective corrective action
Guidelines or Benchmarks?
• Evidence-based guidelines: recommendations for
care based on a systematic review of the literature
• Benchmarks: benchmarking is “measuring
products or services against those provided by
companies recognized as industry leaders*”
• Evidence-based benchmarks!!!!????
*from Benchmarking for Best Practices: Winning through
Innovative Adaptation, Bogan and English 1994
Defining maximum acceptable WT’s for RT:
What kinds of information are available?
– Expert opinion
• Surveys of individual opinion, existing guidelines/standards
published by expert groups, clinical trial protocols
– Prevailing waiting times (benchmarks)
– Radiobiological Models based on:
• a) the relationship between the volume of tumor and risk of
recurrence,
• b) the observed rates of growth of human cancers
– Direct observations of associations between waiting
times and risk of recurrence in clinical practice
– Patients’ opinions
Defining maximum acceptable WT’s for RT:
What kinds of information are available?
– Expert opinion
• Surveys of opinions of individual experts, clinical trial
protocols, existing guidelines/standards published by expert
groups,
– Prevailing waiting times (benchmarks)
– Radiobiological Models based on:
• a) the relationship between the volume of tumor and risk of
recurrence,
• b) the observed rates of growth of human cancers
– Direct observations of associations between waiting
times and risk of recurrence in clinical practice
– Patients’ opinions
Radiation Oncologists’ opinions about the maximum
acceptable interval between referral and start of RT
for a T2,N0 Ca larynx
Maximum acceptable waiting times from surgery to post-operative RT
in protocols of current and recent clinical trials.
Prostate
Number of Trials
6
4
2
0
6
4
2
0
Rectum
Head & Neck
6
4
2
0
6
4
2
0
Cervix
Breast
6
4
2
0
NSCLC
0
5
10
Waiting Time (weeks)
15
6
4
2
0
Guidelines of the Canadian Association
for Radiation Oncology (CARO)
•
Created by an informal consensus-based process; grounded on expert opinion;
not explicitly evidence-based.
•
first adopted 1993, last reviewed and approved by members, Sept.2000
•
Waiting time from referral to consultation with a radiation oncologist should not
exceed 10 working days.
•
Waiting time for radiotherapy should not exceed 10 working days
–
–
For single modality treatment with RT, WT for RT is defined as the time from consultation to
start of RT, or from date that the RO makes the decision to treat, whichever is later.
For multi-modality treatments, WT for RT is defined as the interval between the target RT start
date and the actual RT start date.
•
Consistent with guidelines from the UK, NZ, and several European countries which are
also based entirely on expert opinion
•
http://www.caro-acro.ca/
Defining acceptable WT’s for RT: What kinds
of information are available?
– Expert opinion
• Surveys of individual opinion, existing guidelines/standards
published by expert groups, clinical trial protocols
– Prevailing waiting times (benchmarks)
– Radiobiological Models based on:
• a) the relationship between the volume of tumor and risk of
recurrence,
• b) the observed rates of growth of human cancers
– Direct observations of associations between waiting
times and risk of recurrence in clinical practice
– Patients’ opinions
WT’s from referral to start of RT for a T2, N0, Ca larynx at
Cancer Centres in Canada (n=27) and the US (n=65)
Defining acceptable WT’s for RT: What kinds
of information are available?
– Expert opinion
• Surveys of individual opinion, existing guidelines/standards
published by expert groups, clinical trial protocols
– Prevailing waiting times (benchmarks)
– Radiobiological Models based on:
• the relationship between the volume of tumor and risk of
recurrence
• the observed rates of growth of human cancers
– Direct observations of associations between waiting
times and risk of recurrence in clinical practice
– Patients’ opinions
Biological Background Information
• The probability that RT will eradicate a cancer is
inversely related to the volume of the tumor
– Experimental studies in animals
– Direct observation in human subjects
• Untreated cancers usually grow exponentially
– Experimental studies in animals
– Direct observations in human subjects
Tumour Doubling Time by Primary Cancer Site
260
104
52
26
30
22
21
18
8
12
13
6
4
4
1
Tumor Doubling Time (weeks)
520
Based on the results of 128 studies including ~10,000 cases
Prostate
Breast
Colorectal
Gastric
Lung
Hepatic
Pancreas Head/Neck
Work in progress: with thanks to Gavin Li and Zheng Chen
Distribution of Tumour Doubling Times
0.4
0.3
0.2
Breast
median 30 wks
Percent of Total
0.1
0.0
0.4
Lung
0.3
median 18 wks
0.2
0.1
0.0
0.4
0.3
Head & Neck
median 4 wks
0.2
0.1
0.0
7
30
90
360
Tumour Doubling Time (days)
% Local Control Rate
Predicted Decrease in Local Control with Increasing WT
100
for RT for Cancer of the Tonsil
80
T1
T2
60
40
20
0
T3
T4
mean Td = 82 days
mean Td = 58 days
20
40
Waiting Time (days)
Mackillop, Bates, O’Sullivan, Withers et al, Radiation Research,
60
Defining acceptable WT’s for RT: What kinds
of information are available?
– Expert opinion
• Surveys of individual opinion, existing guidelines/standards
published by expert groups, clinical trial protocols
– Prevailing waiting times (benchmarks)
– Radiobiological Models based on:
• a) the relationship between the volume of tumor and risk of
recurrence,
• b) the observed rates of growth of human cancers
– Direct observations of associations between waiting
times and risk of recurrence in clinical practice
– Patients’ opinions
A systematic review of the literature relating
WT’s to outcomes of RT: Study Questions
• Is delay in RT associated with an increased risk of local
recurrence?
•
Is there a threshold below which delay is free of risk?
• Does the impact of delay on the risk of recurrence differ
from one clinical situation to another?
•
Does delay affect other outcomes of RT?
Methods
• Search MEDLINE and CANCERLIT
databases: any language; Jan 1975 to
June 2005
• Manual search of Abstracts from
CARO and ASTRO
• Secondary search of reference lists
Flowchart for study identification
Reviewed
72 potentially
relevant studies
were reviewed in
detail
Entry
Criteria
36 studies met
basic entry
criteria1
Quality
Criteria
20 studies met
quality criteria2
Entry criteria and quality criteria
ƒ Entry criteria:
ƒ All patients were treated with RT;
ƒ WT from diagnosis or surgery to start of RT was reported;
ƒ Local recurrence rate was reported.
ƒ Quality criteria:
ƒ There were no RCT’s: Quality assessment was therefore designed for
observational studies. We asked 2 questions about each study:
ƒ Were potential confounding factors adequately described?
ƒ patient-related, disease-related, and treatment-related factors known to be
associated with risk of recurrence
ƒ Were the potential confounding factors either balanced between WT
groups or controlled for in a multivariate analysis?
ƒ If the answer to both questions was yes, the study was classified as
meeting quality criteria.
Characteristics of 36 studies that met the
basic entry criteria
• Year: 75-89, 1; 90-94, 9; 95-00, 17; 01-05, 9
• Origin: Canada, 7; Europe 9; US: 15; other, 5.
• Cancer site: Breast, 22; Head & Neck, 12; Sarcoma, 2
• Sample Size: <100, 6; 100-299, 10;
300-49, 7; 500-1000, 10; >1000, 3
• Waiting time measurement:
– continuous variable, 8; categorical variable 28
Statistical Methods
• Data transformation
– Results of studies which described WT as a
categorical variable were re-analyzed to
estimate relative risk per month of delay
• Median waiting times in each category were
attributed to all cases in that category
• Meta-analysis
– Fixed effect model
– Weighting factor (1/SE2)
Breast Cancer (met quality criteria)
Study
RR (95%CI)
Weight%
Postop RT Without Chemo
Froud,2000
0.91 (0.69, 1.22)
6.75%
Vujovic,1998
1.06 (0.80, 1.39)
7.37%
Whelan,1996
1.46 (0.88, 2.42)
2.20%
Clarke,1985
2.07 (1.19, 3.59)
1.83%
1.11 (0.94, 1.33)
Subtotal
Postop RT With Chemo
Hebert-C,2004
1.18 (1.00, 1.39)
20.32%
Benk,2004
1.13 (0.96, 1.33)
20.93%
Yock,2004
0.95 (0.81, 1.12)
20.34%
Wallgren*,1996
1.11 (0.62, 1.98)
1.67%
Wallgren,1996
1.12 (0.91, 1.37)
13.44%
Recht,1996
1.37 (0.92, 2.03)
3.55%
Slotman,1994
2.62 (1.45, 4.72)
1.61%
Subtotal
All Breast Studies
1.12 (1.03, 1.22)
1.12 (1.04, 1.21)
0.2
0.5
1.0
2.0
Relative Risk
5.0
10.0
Breast Cancer (All studies)
Study
RR (95%CI)
Weight%
Postop RT Without Chemo
3.29%
0.91 (0.69, 1.22)
Froud,2000
3.59%
1.06 (0.80, 1.39)
Vujovic,1998
3.51%
1.08 (0.82, 1.42)
Bahena,1998
1.07%
1.46 (0.88, 2.42)
Whelan,1996
4.92%
1.16 (0.92, 1.47)
Fourquet,1995
1.08%
0.42 (0.26, 0.70)
Nixon,1994
0.89%
2.07 (1.19, 3.59)
Clarke,1985
Subtotal
1.06 (0.94, 1.20)
Postop RT With Chemo
9.90%
1.18 (1.00, 1.39)
Hebert-C,2004
10.20%
1.13 (0.96, 1.33)
Benk,2004
9.91%
0.95 (0.81, 1.12)
Yock,2004
20.42%
0.99 (0.88, 1.11)
Metz,2000
0.81%
1.11 (0.62, 1.98)
Wallgren*,1996
6.55%
1.12 (0.91, 1.37)
Wallgren,1996
1.88%
0.84 (0.57, 1.23)
Meek,1996
1.73%
1.37 (0.92, 2.03)
Recht,1996
10.94%
McCormick,1996 1.25 (1.07, 1.47)
2.41%
0.92 (0.66, 1.28)
Leonard,1995
0.93%
1.59 (0.93, 2.74)
Hartsell,1995
0.78%
2.62 (1.45, 4.72)
Slotman,1994
0.53%
1.10 (0.54, 2.25)
Nguyen,1993
1.70%
1.64 (1.10, 2.45)
Buchholz,1993
2.95%
1.38 (1.02, 1.87)
Recht,1991
Subtotal
1.11 (1.05, 1.18)
1.10 (1.04, 1.16)
All Breast Studies
0.2
0.5
1.0
2.0
Relative Risk
5.0
10.0
Head & Neck Cancer (met quality criteria)
Study
RR (95%CI)
Weight%
Leon,2003
1.08 (0.94, 1.23)
50.80%
Fortin,2002
2.03 (1.24, 3.33)
3.84%
Brouha,2000
1.03 (0.68, 1.56)
5.49%
O'Sullivan,1998
1.52 (1.07, 2.17)
7.47%
Lee,1994
0.63 (0.23, 1.70)
0.93%
Definitive RT
Subtotal
1.15 (1.02, 1.29)
Postop RT
Marshak,2004
1.63 (0.67, 4.01)
1.16%
Suwinski,2003
1.22 (1.01, 1.48)
25.55%
Kajanti,1991
1.56 (1.01, 2.43)
4.77%
Subtotal
1.28 (1.08, 1.52)
All Head&Neck Studies 1.19 (1.08, 1.31)
0.2
0.5
1.0
2.0
Relative Risk
5.0
10.0
Head & Neck Cancer (All studies)
Study
RR (95%CI)
Weight%
Leon,2003
1.08 (0.94, 1.23)
45.64%
Fortin,2002
2.03 (1.24, 3.33)
3.45%
Brouha,2000
1.03 (0.68, 1.56)
4.93%
O'Sullivan,1998
1.52 (1.07, 2.17)
6.71%
Barton,1997
0.74 (0.47, 1.17)
3.97%
Lee,1994
0.63 (0.23, 1.70)
0.84%
Definitive RT
Subtotal
1.12 (1.00, 1.25)
Postop RT
Marshak,2004
1.63 (0.67, 4.01)
1.04%
Suwinski,2003
1.22 (1.01, 1.48)
22.95%
Dixit,1998
2.76 (1.39, 5.46)
1.79%
Ampil,1993
1.60 (0.88, 2.93)
2.29%
Kajanti,1991
1.56 (1.01, 2.43)
4.29%
Schiff,1990
1.73 (0.92, 3.24)
2.11%
Subtotal
1.38 (1.18, 1.61)
All Head&Neck Studies 1.20 (1.09, 1.31)
0.2
0.5
1.0
2.0
Relative Risk
5.0
10.0
0.2
0.3
0.4
0.5
0.6
Standard Error of Log RR
0.1
0.0
Funnel plot with pseudo 95% confidence limit
All studies
0.13
-1.0
-0.5
0.0
Log Relative Risk
0.5
1.0
Summary
High Quality Studies
RR/m 95%CI
All Studies
1.15 (1.09, 1.22)
All Breast Studies
1.12 (1.04, 1.21)
All Head&Neck Studies 1.19 (1.08, 1.31)
All Studies
RR/m 95%CI
All Studies
1.13 (1.09, 1.18)
All Breast Studies
1.10 (1.04, 1.16)
All Head&Neck Studies 1.20 (1.09, 1.31)
Is delay in RT associated with an
increased risk of local recurrence?
• Yes
•
Longer WT was significantly associated with a higher risk
of local recurrence in breast and H &N cancer.
•
There was no direct evidence that delay is associated with
increased recurrence rates in other cancer sites, but beware
of the problem of beta error “absence of evidence is not
evidence of absence”
Is there a threshold below which delay
is free of risk?
• No
• There is no empirical evidence that there is a threshold
below which delay has no adverse effect
• There is no theoretical reason to suspect that such a
threshold exists.
• Similar stochastic risks are common with environmental
hazards, hence the ALARA principle i.e. keep exposures
As Low As Reasonably Achievable.
Does the impact of delay on the risk of recurrence
differ from one clinical situation to another?
• Unanswerable!
• No significant differences in effect size in the different
situations studied. Study power was inadequate to address
this question.
• But, the RR/month delay in post-op RT was greater for
head and neck cancer than for breast cancer.
• The RR/month delay was greater for post-operative RT
than for definitive RT in the context of head and neck
cancer.
Does delay affect other outcomes of RT?
Data not presented. Analysis is incomplete
• There is no good empirical evidence that delay in RT is
associated with an increase in the risk of distant metastasis.
There are good theoretical reasons to believe that any such
effect would be small.
• There is some empirical evidence that delay in RT
decreases the probability of long term survival in situations
in which RT is used with curative intent, including head
and neck cancer and cervical cancer.
• There is almost no information about the effect of delay in
RT on symptoms or quality or life.
Defining acceptable waiting times
• What kinds of information are available?
– Expert opinion
• Surveys of individual opinion, existing guidelines/standards
published by expert groups, clinical trial protocols
– Prevailing waiting times (benchmarks)
– Models based on the observed relationship between the
volume of tumor and risk of recurrence, and the
observed rates of growth of human cancers
– Direct observations of associations between waiting
times and risk of recurrence in clinical practice
– Patients’ opinions
Patients Attitudes to Waiting for RT
• When offered the choice between delayed RT close to
home or immediate RT at distant centre, many patients
choose to accept significant delays rather than relocate.
–
Palda, VA., et al (1997). Journal of Clinical Oncology, 15, 3192-3200
• But, if patients are informed that delay is associated with
any decrease in the effectiveness of RT, they will choose
treatment at a distant centre to avoid even short delays
– Lehman, et la (2004). Radiotherapy and Oncology, 70, 283-289.
That’s the evidence! How should we
interpret it?
• The paradigm of “Evidence-based
medicine”
• The paradigm of “Risk assessment and risk
management”
The Paradigm of “Evidence-based Medicine”
• Context: widely used for evaluating the effectiveness of
clinical interventions and establishing CPGs
• Rules of Evidence: Sackett’s rules: The RCT is the gold
standard; observational data is of secondary value; the
results of animal studies are inadmissible
• Action Guide: the default position is not to recommend
treatment unless there is clear evidence of its effectiveness
(because unproven therapy may do more harm than good)
• Who interprets the data and makes the call? Expert
health professionals+ clinical epidemiologists
The Paradigm of “Risk Assessment”
• Context: evaluation of the risks of environmental
hazards and establishment of regulations for
human exposures
• Rules of Evidence: Hill’s criteria; observational
studies in humans and experimental studies in
animals are both admissible.
• Action Guide:, The evidentiary threshold for
action is lower because inaction may lead to harm
and there is no health risk associated with limiting
exposures to environmental hazards
• Who interprets the data and makes the call?
Experts AND those potentially affected
Our next step will be to engage
stakeholders in evaluation of the risks
Risk Assessment Framework from the Presidents Commission Report
In the meantime, we recommended
adoption of the CARO/WTA guidelines
•
•
•
•
•
•
They are clear and simple
They are known to be achievable
They are auditable
They are widely accepted by Canadian ROs
They are consistent with standards elsewhere
They are consistent with the available evidence
and the principle that WTs for RT should be As
Short As Reasonably Achievable
Were these guidelines accepted?
CARO/WTA
• Waiting time from referral to consultation with a
radiation oncologist should not exceed 10 working
days.
• Waiting time for radiotherapy should not exceed 10
working days
– For single modality treatment with RT, WT for RT is defined as
the time from consultation to start of RT, or from date that the
RO makes the decision to treat, whichever is later.
– For multi-modality treatments, WT for RT is defined as the
interval between the target RT start date and the actual RT start
date.
Objectives
ƒ To report on our progress in setting “evidence-based
benchmarks” for waiting times for RT and discuss next
steps in this process
ƒ To demonstrate that waiting times alone are an
inadequate measure of access to care
ƒ To remind ourselves that setting benchmarks and
monitoring compliance with them is worthless unless it
leads to effective corrective action
The Concept of Accessibility
in Health Care
– Accessibility embraces all factors that influence
the level of use of a service, given a particular
level of need…. “describes the degree of fit
between the system and the patients”
Aday, Penchansky
Components of Accessibility
• Availability* : total system capacity/ total needs
– Determined by total resources, efficiency, flexibility
• Spatial Accessibility*
– Distance and travel times
• Accommodation* (Convenience)
– hours of operation; lodges; transportation services
• Affordability*
– direct and indirect costs; ability and willingness to pay
• Awareness of services, and the indications for
their use, among doctors and patients
* terms used by Penchansky 1981
Limitations of Waiting Time as a
Measure of Access to RT
ƒ Waiting times are sensitive to supply-side
problems associated with availability.
ƒ BUT Waiting times are entirely insensitive to
demand-side problems associated with awareness,
spatial accessibility, affordability, and
accommodation
ƒ Therefore, the absence of a waiting list does not
imply that access is optimal.
Indicators of Access to RT
• Waiting times for RT
• Rates of use of RT
Measures of Use of RT
• Use of RT in initial treatment of cancer
– *defined as the proportion of incident cases treated
within one year of diagnosis
• Use of palliative RT among patients dying of
cancer
– **defined as the proportion of cases that die of cancer
who receive palliative RT in the last two years of life
*Mackillop et al, JCO, 1997, **Huang at al JCO, 2002
a) RT within 1 Year of Diagnosis
Northern Ontario
Ottaw a
Thunder Bay
0
25 0
Kingston
500 Miles
Quintile
17.5% - 24.3%
25.1% - 26.8%
27.2% - 29.1%
29.2% - 30.7%
30.8% - 35.2%
data not available
Toronto
Sudbury
Windsor
London
Ham ilton
0
100
200 Miles
Southern Ontario
Radiotherapy Centre
Zhang-Salomons 2005
b) Palliative RT in the Last 2 Years of Life
Northern Ontario
Ottaw a
Thunder Bay
250
0
Kingston
500 Miles
Quintile
10.9% - 23.5%
23.5% - 24.5%
24.5% - 26.5%
26.5% - 29.6%
29.6% - 35.5%
data not available
Toronto
Sudbury
Windsor
London
Ham ilton
0
100
200 Miles
Southern Ontario
Radiotherapy Centre
Zhang-Salomons 2005
Factors associated with the use of palliative
RT in Ontario*
Characteristic
OR
95% CI
Low
1.00
–
Medium
1.09
1.04-1.15
High
1.17
1.11-1.24
Diagnosing
Hospital
Not associated with an RT center
1.00
–
Associated with an RT center
1.35
1.30-1.40
Place of
residence
Counties with no cancer center
1.00
–
Counties with cancer center
1.24
1.21-1.27
Toronto
0.88
0.84-0.92
Ottawa
1.00
–
Hamilton
1.20
1.14-1.26
Median
household
income
Region
*controlling for primary site and age
Huang et al, JCO, 2002
RT Use in Older Patients
RT Rate or Functional State *
100%
Adjuvant RT for
resected Breast
Ca
80%
Palliative RT
within yr of death
from Prostate Ca
60%
40%
Radical RT for
Oropharynx Ca
20%
Estimated
Functional
Status
0%
45-54
55-64
65-74
>75
Age Groups
*Relative to 45-54 years
Tyldesley et al, 2000
The value of Rates of RT Utilization as a
Measure of Accessibility
• May identify demand-side problems relating to
spatial accessibility and awareness of services,
that would otherwise remain invisible
• May provide a quantitative measure of the
shortfall between supply of service and the need
for service.
• Public policy about access to care that is
informed only by WT’s is blind to half of
the problem.
• Waiting times and utilization rates both
require to be “benchmarked” and monitored
to ensure adequate access to RT.
Objectives
ƒ To report on our progress in setting “evidence-based
benchmarks” for waiting times for RT and discuss next
steps in this process
ƒ To demonstrate that waiting times alone are an
inadequate measure of access to care
ƒ To remind ourselves that setting benchmarks and
monitoring compliance with them is worthless unless it
leads to effective corrective action
Median Wait Time from Referral to Start of RT: CCO
10
8
Breast
GI
GU
Lung
6
4
2
19
98
19
99
20
00
20
01
20
02
20
03
20
04
19
97
19
96
19
95
19
94
19
93
0
19
92
Median Wait (weeks)
12
Calendar Year
Conclusion
• Our plans for setting WT standards and
monitoring access to care will be worthless
unless they are linked to effective strategies
to deal with problems that they identify.
5-year Survival (%)
Cause-specific Survival in Ontario and the
U.S.(SEER), by Income Group (1987-92)
60
All sites except prostate
55
50
Ontario
U.S.
45
1
2
Highest Quintile
3
4
5
Lowest Quintile
Median Household Income
Boyd et al, J Clin Oncol, 1999
Initial Rate of RT for Lung Cancer in Ontario by County
Estimated Appropriate Rate=44.6%
RT Rates
50%
40%
Estimated Shortfall=10.6%
30%
20%
Overall Observed Rate=34%
10%
1
5
9
13
17
21 25
29
33
37 41
45
Counties ordered by RT rates
Bars (-) represent the 90% CI
Tyldesley et al, IJROBP 2001; Barbera et al, Medical Care, 2003