US Preventive Services Task Force

1
US Preventive Services Task Force

• Diana Petitti, MD, MPH
• Arizona State University

2
Todays outline

• Background on the USPSTF
• USPSTF Analysis and Recommendation on Breast
  Cancer Screening

3
U.S. Preventive Services Task Force

• 16 member independent, volunteer panel convened
  by AHRQ
• Non-Federal experts in clinical prevention and
  primary care
• Use evidence to create new and updated
  recommendations on screening, counseling, and
  medications to prevent illness

4
USPSTF

• Relevant to practice of primary care for
  asymptomatic persons AND average risk persons
• Uses systematic, unbiased evidence reviews to
  gather data on both benefits and harms

5
USPSTF

• The USPSTF does not use cost or cost
  effectiveness data in making recommendations
• The USPSTF does not make insurance coverage or
  policy determinations

6
USPSTF

• New member nominations are sought each year from
  the public and from partner organizations through
  a Federal Register notice
• Requirements for nominees
• Expertise in prevention and primary care
• Strong experience in critical appraisal of
  evidence
• Primary care experience
• New members are named by the AHRQ Director

7
The making of a recommendation

• Each systematic review starts with an analytic
  framework and key questions
• Project at this stage is informed by
• Previous evidence review and recommendation (if
  an update)
• Topic Prioritization workgroup of the USPSTF
• 3-4 member topic workgroup of the USPSTF
• Evidence-based practice center (EPC) Principal
  Investigator and team

8
Analytic Framework on Screening for a Disease
9

• USPSTF Recommendations
• On Breast Cancer Screening

10
Breast Cancer Screening Recommendation

• Update of 2002 recommendation begun in 2007
• Two reports commissioned by AHRQ
• An updated systematic review and meta-analysis of
  trial data, including new information from large
  databases
• A collaborative modeling study from the Cancer
  Information and Surveillance Network (CISNET)

11

• Systematic Evidence Review

12
Updated Systematic Evidence Review- (SER)

• PI Heidi Nelson, Oregon Evidence-based Practice
  Center (EPC)
• Trials of screening with breast cancer mortality
  as outcome
• New trial from UK, updates from older trials
• Harms of screening radiation exposure, pain,
  adverse psychosocial responses, overdiagnosis,
  false positive mammograms, additional imaging,
  biopsies
• Using primary data from the Breast Cancer
  Surveillance Consortium (BCSC).

13
SER Results Film Mammography

• 8 screening trials for age 39 49 year olds
  indicate reduced breast cancer mortality in
  screened women
• 1 screening trial for ages 70-74 years indicates
  no mortality reduction

14
SER Results New evidence for women 40-49

• Age Trial, in United Kingdom
• Annual mammography to age 48 yrs vs. usual care
• Results
• Breast cancer mortality RR 0.83 (0.66- 1.04)
• Number needed to invite 2,512 (1,1,49- 13,544)

15
SER results New Evidence for Women 40-49

• Additional follow-up for the Gothenburg trial
• RCT of mammography among women aged 39-59 in
  Gothenburg, Sweden in 1982
• Results
• Breast cancer mortality RR 0.69 (0.45-1.05)

16
SER Results
Meta-analysis of Screening Trials of Women Age
39 to 49 Years
Screened
Control
Relative Risk for Breast Cancer Death (95 CI)
Study (yr)
Cases/N
Cases/N
Gothenburg (2003) Kopparberg (1995) Malmo
(2002) HIP (1986) Age (2006) CNBSS-1
(2002) Ostergotland (2002) Stockholm
(2002) Total
0.69 (0.45, 1.05) 0.72 (0.38, 1.37) 0.73 (0.51,
1.04) 0.78 (0.56, 1.08) 0.83 (0.66, 1.04) 0.97
(0.74, 1.27) 1.05 (0.64, 1.73) 1.47 (0.77,
2.78) 0.85 (0.75, 0.96)
34/11,724 59/14,217 22/9,582
16/5,031 53/13,568
66/12,279 64/13,740 82/13,740
105/53,884 251/106,956 105/25,214
108/25,216 31/10,285 30/10,459
34/14,303 13/8,021
437/152,300 615/195,919

0.2
0.5
1
2
5
Favors screening
Favors control

16
17
SER Results
Summary of Meta-analyses of Screening Trials For All Age Groups Summary of Meta-analyses of Screening Trials For All Age Groups Summary of Meta-analyses of Screening Trials For All Age Groups Summary of Meta-analyses of Screening Trials For All Age Groups Summary of Meta-analyses of Screening Trials For All Age Groups

Age Groups Number of trials RR for Breast Cancer Death (95 CI) NNI to Prevent 1 Breast Cancer Death (95 CI)

39-49 8 0.85 (0.75-0.96) 1,904 (929-6,378)
50-59 6 0.86 (0.75-0.99) 1,339 (322-7,455)
60-69 2 0.68 (0.54-0.87) 377 (230-1,050)
70-74 1 1.12 (0.73-1.72) Not available

Trials and their acronyms are discussed in the text. Trials and their acronyms are discussed in the text. Trials and their acronyms are discussed in the text.
Abbreviations RR, relative risk CI, confidence interval NNI, number needed to invite to screening. Abbreviations RR, relative risk CI, confidence interval NNI, number needed to invite to screening. Abbreviations RR, relative risk CI, confidence interval NNI, number needed to invite to screening. Abbreviations RR, relative risk CI, confidence interval NNI, number needed to invite to screening. Abbreviations RR, relative risk CI, confidence interval NNI, number needed to invite to screening.
17
18
SER ResultsHarms of Screening Mammography

• Radiation per study very low
• Pain common, transient
• Adverse psychosocial responses anxiety,
  distress, worry
• Overdiagnosis
• estimates vary - 9 European studies from 1 to
  10

19
SER ResultsIllustration of Overdiagnosis Rates
of Invasive Cancer and DCIS
Invasive Cancer
Number per 1000 Women Screened
DCIS
Age (yrs)

Invasive cancer 2.7 4.5 6.3 7.8
DCIS 0.9 1.4 1.6 1.6
19
20
SER Results Harms of Screening- Rates of False
Positive and False Negative Mammograms
12 10 8 6 4 2 0
False Positive
Rates ()
False Negative
Age (yrs)

False Pos 9.8 8.7 7.9 6.6
False Neg 0.1 0.1 0.1 0.2
20
21
SER results Breast Self Examination

• Benefits Two trials conducted in countries
  (China, Russia) without mass mammography
  screening
• No mortality reduction in either trial
• Harms
• Increased benign biopsy rates in the BSE group
  compared to controls

22
SER Results Clinical Breast Examination

• RCTs in countries without mass mammography
  screening (one discontinued, two underway)
• Canadian trial from the 1980s compared
  mammography plus CBE plus BSE versus CBE plus BSE
  and found no difference in mortality between
  groups.
• Harms- inconclusive data, potential harms include
  false positives, anxiety, excess imaging and
  benign biopsies

23
SER Results Digital Mammography and MRI

• No studies of MRI screening in average risk women
• No trials of digital mammography for screening
  average risk women. Studies of diagnostic
  accuracy suggest similar to film mammography and
  more accurate in younger women and those with
  dense breasts.

24

• CISNET Modeling Data

25
Advantages of (Collaborative) Modeling

• Models can test strategies not feasible in the
  population
• Models can test strategies in large samples
• Models can ask what if questions
• Multiple models can use common data
  (experimental conditions) and
• Replicate experiments
• Control the experimental conditions
• Provide sense of qualitative ranking
• Provide range of plausible quantitative effects
• Results can inform practice and policy debates

26
Overview of CISNET Breast Cancer Models
Original Objective Assess Impact of Screening
and/or Adjuvant Therapy on Breast Cancer Mortality
Population Inputs (Common to all models)

• Dissemination of Adjuvant Therapy
• Dissemination of Mammography
• Change in Background Risk
• Mortality from Other Causes

Model Specific Inputs and Assumptions
Predicted Mortality

• Efficacy of Treatment
• Tumor Growth Rates MetastaticSpread
• Operating Characteristics of Screening (e.g.,
  sensitivity, lead time)
• Consequences of Screening (e.g., stage
• shift, over diagnosis)
• Post Diagnosis Survival by Tumor Characteristics

• For
• Treatment Alone
• Screening Alone
• Treatment and Screening

27
Outcome Measures -Benefits

• Two primary measures of benefit of screening (vs.
  no screening)
• reduction in breast cancer mortality
• Life years gained (per 1000 women)
• Secondary metrics
• Additional change in effect for screening at ages
  younger or older than 50 to 69.
• .

28
Outcome Measures-Resources and Harms

• Resources required
• Number of screening mammograms
• Exposure to harms
• False positive screens
• Number of un-necessary biopsies
• Detection of tumors never destined to cause
  breast cancer death (over diagnosis)
• (NO measure of morbidity or decrement in QOL)

29
Benefit Maintained Moving from Annual to
Biennial Screening by Strategy and Model
Screening Strategy Models Models Models Models Models Models
Screening Strategy W M G D S E
50-69 68 93 85 75 74 75
40-69 67 97 86 75 73 73
45-69 70 96 91 78 78 74
40-79 70 98 87 78 76 75
40-84 71 97 88 81 77 75
55-69 71 92 91 80 80 75
60-69 70 93 86 74 74 73
50-74 72 94 89 80 79 76
50-79 70 94 88 78 85 75
50-84 73 95 89 81 79 76
70 to 98 of benefit maintained screening
biennial
30
Efficiency Frontier Non-dominated Strategies (
Mortality Decline) Exemplar Model
Model S
A 40-79
A 50-79
B 40-79
B 50-74
B 50-69
31
Harms Screen Detection of Invasive Tumors Never
Destined to Cause Cancer Death by Age
Annual Screening Ages 40-84

• Model assumes that all invasive cancers progress
  with different age-specific lead times
• Percent dying in lead time increases steeply in
  older age due to
• High rate of death from other illnesses
• Longer lead time in older age

Model D
32
Harms Screen Diagnosis of Tumors Never Destined
to Cause Cancer Death

• Two models (E, W) include
• Some DCIS/small local tumors that never progress
  (low malignant potential)
• Screen detection of progressive invasive cancers
  where death occurs in the lead time from other
  illness
• These models project over-diagnosis rates
  several orders of magnitude higher than models
  without low malignant potential tumors
• Overall, there is uncertainty for this potential
  harm due to limited primary data upon which to
  base models

33
Potential Harms False Positive Results,
Unnecessary Biopsies
Based on published age-specific specificity in
BCSC

• False positives increase in linear fashion with
  number of mammograms performed (8.3 rate
  varies by age)
• If 9 screens ?0.8 false per woman
• If 18 screens ?1.5 false
• If 36 screens ?3.0 false
• Adding 10 years screening in younger women adds gt
  2x as many false positives as adding 10 years at
  older ages.
• 7 of false positives lead to unnecessary
  biopsy

34
Balance Sheet of Potential Benefits Harms
Starting Ages
Model S Potential Benefits (vs. no screening) Potential Benefits (vs. no screening) Potential Benefits (vs. no screening) Potential Harms Potential Harms Potential Harms
Strategy Average Screens per 1000 Mortality Reduction Breast Cancer Deaths Averted per 1000 Life Years Gained per 1000 False positives per 1000 of unnecessary biopsies per 1000 Over diagnosis of invasive cancer
Biennial
B 40-69 13700 16 6.1 120 1250 88 0.8
B 45-69 11800 17 6.2 116 1050 74 0.8
B 50-69 8900 15 5.4 99 780 55 0.7
B 55-69 6900 13 4.9 80 590 41 0.8
B 60-69 4200 9 3.4 52 340 24 0.6
Annual
A 40-69 27600 22 8.3 164 2250 158 1.0
A 45-69 22600 22 8.0 152 1800 126 0.9
A 50-69 17800 20 7.3 132 1350 95 0.9
A 55-69 13000 16 6.1 102 950 67 0.9
A 60-69 8400 12 4.6 69 600 42 0.8
Shaded dominated by other strategies
over-diagnosed invasive cancers within the
strategy divided by all cancer cases occurring
over life time from age 40. Probability of
over-diagnosis is 10 times higher in models E
and W with explicit LMP
35
Balance Sheet of Potential Benefits Harms
Stopping Ages
Model S Potential Benefits (vs. no screening) Potential Benefits (vs. no screening) Potential Benefits (vs. no screening) Potential Harms Potential Harms Potential Harms
Strategy Average Screens per 1000 Mortality Reduction Breast Cancer Deaths Averted per 1000 Life Years Gained per 1000 False positives per 1000 of unnecessary biopsies per 1000 Over diagnosis- invasive
Biennial
B 50-69 8900 15 5.4 99 780 55 0.7
B 50-74 11100 20 7.5 121 940 66 1.5
B 50-79 12300 25 9.4 130 1020 71 2.1
B 50-84 13800 26 9.6 138 1130 79 3.3
Annual
A 50-69 17800 20 7.3 132 1350 95 0.9
A 50-74 21400 26 9.5 156 1570 110 1.7
A 50-79 24400 30 11.1 170 1740 122 2.6
A 50-84 26900 33 12.2 178 1880 132 3.7
over-diagnosed invasive cancers within the
strategy divided by all cancer cases occurring
over life time from age 40. Probability of
over-diagnosis is 10 times higher in models E
and W with explicit LMP. Shadeddominated by
other strategies
36
USPSTF Assessment- Grades
Net Benefit (Benefit Harms)
Certainty Substantial Moderate Small Zero or negative
High A B C D
Moderate B B C D
Low Insufficient Insufficient Insufficient Insufficient
37
Summary of New USPSTF recommendations

• Biennial screening mammography between 50 and 74
  years (B grade)
• The decision to start regular screening before
  the age of 50 should be an individual one and
  take into account patient context, including
  values regarding specific benefits and harms (C
  grade)
• Previous recommendation was to screen women 40
  and older every 1 to 2 years

38
Summary of New USPSTF recommendations

• The USPSTF concludes evidence is insufficient to
  assess the additional benefits and harms of
  screening mammography in women 75 years or older
  (I statement)
• Previous recommendation had no ending date
  (applied to women 40 and older)
• Insufficient evidence on the additional benefits
  and harms of clinical breast examination beyond
  mammography in women 40 or older (I statement)
• This is unchanged from previous

39
Summary of New USPSTF recommendations

• USPSTF recommends against clinicians teaching
  women how to perform breast self-examination (D
  grade)
• Previous recommendation teaching BSE was given
  an Insufficient Evidence rating
• Insufficient evidence to assess additional
  benefits and harms for
• digital mammography or
• magnetic resonance imaging
• (I statement)
• These new modalities were not mentioned in the
  2002 recommendation

40

• Questions and Discussion