Cancer Screening in the Elderly

1. Cancer Screening in the Elderly Louise C. Walter, MD Associate Professor of Medicine, Division of Geriatrics San Francisco VA Medical Center University of California, San Francisco October 25, 2009

2. Objectives • Develop a systematic way to think about benefits and harms of cancer screening in older adults • Consider life expectancy when making screening recommendations • Understand the importance of factoring patient preferences into screening decisions • Identify problems with “quality indicators” that equate high screening rates with better quality

3. Cases • Mrs. A 70 y/o woman with Alzheimer’s dementia (MMSE=10/30) and functionally dependent in many ADLs. She lives with her daughter who brings her in for a routine check-up. She has no history of any cancer screening tests. • Mrs. B 80 y/o woman with a history of osteoarthritis. She walks 2 miles a day and cares for her older sister. She has not seen a doctor in several years but decides to come in for a routine check-up. She has no history of any cancer screening tests.

4. Uncertainty Uncertainty about when to screen asymptomatic elderly patients for cancer • Most trials of cancer screening tests have excluded patients over age 75 • Extrapolate data to older patients • Data from randomized trials not always applicable to an individual patient • Trials do not address individual characteristics that may change the likelihood of benefit vs. harm

5. What to Do? • Guidelines used to be based on age cutoffs and were conflicting Mammography Guidelines (until 2002) • USPSTF: Stop mammography at age 70 • American College of Physicians: Stop at age 75 • American Geriatrics Society: Stop at age 85 • American Cancer Society: Never stop • Now guidelines agree that screening: • Should continue if an older person is healthy • Should stop if an older person is unhealthy—has a limited life expectancy

6. Screening Decisions • Age should not be the most important factor in screening decisions • “One-size-fits-all” approach to medical care does not work in the diverse elderly population • Variation in health status/life expectancy/preferences • Present a framework to guide how to think through cancer screening decisions in elderly • Incorporates individual characteristics (health/life expectancy) and preferences into decisions

7. Framework for Individualized Decisions • Estimate life expectancy • Determine potential benefits of screening • Determine potential harms of screening • Weigh potential benefits and harms according to an individual’s values and preferences

8. Life Expectancy for Women Years Age (years) Walter LC. JAMA 2001;285:2750-2756

9. Life Expectancy • Possible to estimate if a person is likely to live substantially longer or shorter than average • Number/severity of comorbid conditions and functional impairments stronger predictors than age • Life expectancy substantially below average • CHF (Class III, IV), ESRD, Severe COPD (home O2), Severe dementia (MMSE < 10), Dependency in many ADL • Life expectancy substantially above average • No comorbid conditions or functional impairments

10. Lag-Time to Benefit • Benefit of screening does NOT occur immediately • Screening results in benefit by finding cancers at an early stage, which would have caused symptoms or killed a person years later • A life expectancy of > 5 yrs is required to have some chance of survival benefit from screening • RCTs of mammography and FOBT show survival curves of screened vs. unscreened do not separate significantly until > 5 years after start of screening

11. Screening Mammography RCTs • RCTs in women aged 50-69 years Cumulative Breast Cancer Mortality by Study Group Screening Control 500 400 300 Mortality/ 100,000 200 100 0 1 2 3 4 5 6 7 8 9 10 Time (years) Nystrom L. Lancet. 2002:909-919

12. Benefits of Screening • Unlikely to benefit if life expectancy < 5 years • If life expectancy > 5 years benefit should be considered in terms of absolute risk reduction (ARR) rather than relative risk reduction (RRR) • RRR usually reported in studies or advertisements: “TEST REDUCES RISK OF CANCER BY 25%!” • A 25% RRR describes a reduction from 40% to 30% or 4% to 3%

13. Absolute Benefit • More benefit from test that reduces risk from 40% to 30%  ARR = 10% rather than 4% to 3%  ARR = 1% • Absolute benefit presented as Number Needed to Screen (NNS) = 1/ARR 40% to 30%  NNS = 1/0.1 = 10 4% to 3%  NNS = 1/0.01 = 100 • Absolute benefit is fair way to present benefit because accounts for baseline risk of disease

14. Number Needed to Screen NNS over remaining lifetime to prevent one cancer death for women at selected ages and life expectancies AGE 70 AGE 80 AGE 90 Life Expectancy Life Expectancy Life Expectancy Screening Test RRR U M L U M L U M L Mammography 26% 142 242 642 240 533 - 1,066 - - FOBT 18% 178 340 1,046 262 581 - 1,163 - - Pap Smear 60% 934 1,521 4,070 1,694 3,764 - 7,528 - - - indicates life expectancy < 5 years Walter LC. JAMA 2001;285:2750-2756

15. Harms of Screening • Complications from additional diagnostic procedures due to inaccurate test results • Identification and treatment of clinically unimportant disease that would not have progressed to symptoms in patient’s lifetime • Psychological distress

16. Conducted study at On Lok (PACE): Health program for frail elderly State auditors insisted mammograms be performed in all women despite poor health and advanced age (per American Cancer Society) 50% of patients have dementia Median life expectancy is 4 years We assessed harms of this universal screening mammography policy in frail older women Harms of Screening Mammography

17. Screening mammography in frail elderly women frequently led to harm 38 with abnormal result 32 agree to work-up 4 diagnosed with cancer or DCIS 216 women had screening mammogram 1/95-1/97 4 treated but died of other causes within 4yrs 178 with normal result 6 refused work-up 28 with false- positive results Walter LC. J Gen Intern Med 2001:779-784

18. Harm of Finding Clinically Unimportant Cancers 80 y/o woman with severe dementia from multiple strokes who underwent screening mammography on enrollment • Abnormal mammogram • 2 biopsy attempts—inconclusive results • Underwent surgery—ductal carcinoma in situ • Developed wound infection—daily trips to wound clinic for 3 months • Second surgery to close wound • Died of large stroke 9 months later Walter LC. J Gen Intern Med 2001:779-784

19. Harms of Colorectal and Cervical Cancer Screening • Colorectal Cancer Screening • False-positives: ~1 in 10 for FOBT • Complications: 3 per 1000 screening colonoscopies (perforation, bleeding, stroke) • Cervical Cancer Screening • Of ~2,500 women 44-79 yrs screened  110 had abnormal Pap  33 colposcopies, 35 endocervical curettages, 39 biopsies, 4 dilation & curettages  1 woman with mild-moderate cervical dysplasia Lieberman DA. NEJM 2000;343:162-8 Sawaya G. Ann Intern Med. 2000:942-950

20. Psychological Harm • Many important burdens not measured in RCTs • Emotional pain of cancer diagnosis in people whose lives were not extended by screening • Alarm of false-positive results • Stress of undergoing the screening tests • Magnitude of psychological harm is individual • Cognitive or sensory problems may make tests and follow-up procedures more difficult, painful or frightening Sox. JGIM 1998;13:424-5

21. Preferences • Assess how patients view potential harms/benefits and integrate values/preferences into decisions • Different from public health strategy in which experts weigh benefits/risks and decide what is best for a population • Since many decisions in older patients are “close calls,” need to consider values/preferences • Harms look larger to some people • Non-mortality benefits considered more substantial to some people (e.g., “peace of mind”)

22. Other Considerations • Perhaps older adults would have less need for reassurance from continued screening if: • Used less alarming language about cancer risk • Ex: Breast cancer responsible for ~1% of deaths in women > 80 yrs • Talked about screening as a choice (not obligation) • 41% of Americans labeled an 80 y/o woman “irresponsible” if she did not have mammography • Need to encourage informed discussions: Screening is a “double-edged sword” Welch HG. Ann Intern Med. 2004;140:754-5

23. Discuss Preferences for Work-up • Prior to screening, discuss possible procedures or treatments required to work-up abnormal result • Patients who would not want work-up/treatment for an abnormal result should NOT be screened

24. Cases Mrs. A – 70 y/o woman with severe dementia Mrs. B – 80 y/o woman with osteoarthritis • Estimate life expectancy • Mrs. A is younger but has severe dementia and functional dependency, so life expectancy < 5 years • Mrs. B is probably in upper quartile of life expectancy for her age, so likely to live > 13 years • Probability of benefit • Mrs. A unlikely to benefit since life expectancy < 5 yrs • Mrs. B has reasonable likelihood of benefit (NNS) • 240 for mammography; 262 for FOBT; (1,694 for Paps)

25. Cases • Probability of harm • Mrs. A has severe dementia so tests may cause distress and if cancer identified likely is unimportant • Mrs. B understands and accepts risks of tests • Values and preferences • Mrs. A has avoided doctors and becomes agitated if anything interrupts her daily routine • Mrs. B worries about her health and wants a mammogram, FOBT, and Pap smear • Screening recommendations • Recommend AGAINST screening Mrs. A • Recommend screening Mrs. B

26. General Recommendations • Recommend against cancer screening if estimated life expectancy < 5 yrs • If life expectancy 5-10 yrs decision is a “close call” and preferences play major role in decision • If estimated life expectancy >10 yrs recommend • Mammography Q2yrs • Yearly FOBT +/- Flex Sig Q5yrs; colonoscopy Q10yrs • Pap smears (2 if never had Pap; stop at 65 yrs) USPSTF 2002: www.ahrq.gov/clinic/uspstfix.htm

27. PSA Screening • Guidelines: Insufficient evidence • USPSTF: Do not screen men > 75 years old • Benefits uncertain—RCT’s with mixed results • RCT’s ongoing (none included men > 75 years old) • If life expectancy < 10 yrs very unlikely to benefit • Harms are significant • Finds insignificant disease  unnecessary distress, incontinence, impotence, etc. • Recommend: • If life expectancy > 10 yrs, preferences drive decision • Avoid screening in men with life expectancy < 10yrs Andriole GL and Schroder FH. N Engl J Med. 2009

28. PSA Screening Across the VA System • To determine if PSA screening is being avoided in elderly men with limited life expectancies • Conducted cohort study of 597,642 men > 70 years with an outpatient visit at 104 VA’s in 2002 and 2003 • Excluded men who had a PSA test for non-screening reasons (e.g., prostate cancer, elevated PSA) • Outcome was receipt of a PSA screening during 2003 • Rates of PSA screening were high • 56% of veterans > 70 years had a PSA test in 2003

29. Even Older Men in Poor Health Had Substantial PSA Screening Rates N= 597,642 Veterans Charlson Score Walter LC. JAMA 2006;296:2336-42

30. Study Conclusions • Screening often performed in elderly veterans in poor health, even those 85+ who have less than a 10% chance of living 10 years • High rates not because patients requested test: Audited 100 random charts: 4% requested test • Many quality indicators used by VA encourage cancer screening regardless of health/life expectancy so this enthusiasm may have spread to PSA screening

31. Quality Indicators • Quality indicators often based rigidly on rewarding high cancer screening rates rather than encouraging individualized decisions • Do not account for when risks > benefits • Do not account for patient preferences • Rewarding high screening rates challenges clinicians who wish to individualize decisions • Target screening to those for whom benefits > risks

32. VA Colorectal Cancer Screening Quality Indicator • In 2002, SF VA clinicians were frustrated • Our screening rate of 58% failed to meet target 65% • Clinicians told to screen more patients • Failure to increase rates  financial penalties • We decided to assess whether this “quality indicator” reflected quality

33. Background: VA Quality Indicators • VA quality indicator based on chart review by external auditors who compute % eligible patients who received timely colorectal cancer screening • Eligible patients: > 52 yrs who did NOT have cancer of liver, pancreas, esophagus, or life expectancy < 6 mos • Over-sampled patients with specific comorbid illnesses, such as CHF and ischemic heart disease • We reviewed medical charts of the 229 patients previously reviewed by auditors to compute the 2002 colorectal cancer screening quality indicator

34. 229 Patients Not Tested N=81 Tested N=148 Diagnostic N=57 Screening N=91 Results of Our “Audit of the Audit” Patient too Sick to Benefit N=25 Refused Test N=38 Failed to Complete Test N=10 No Valid Reason For Not Tested N=8 Walter LC. JAMA 2004;291:2466-2470

35. Lessons Learned • 90% of patients classified as receiving poor quality care for not being screened had valid reasons (e.g., poor prognosis, preferences) • Minimizes importance of individualized decisions • Findings resulted in VA making several improvements to the performance measure • Auditors no longer focus on screening in ill patients • Need to develop measures that go beyond equating good care with high screening rates • Measures whether screening is individualized Walter LC. JAMA 2004;291:2466-2470

36. Summary Do Screen Don’t Screen Likelihood of Benefit Likelihood of Harm Patient Preferences (moveable fulcrum)

37. Conclusions • Understanding potential benefits and harms of medical interventions and being aware of patient wishes are core principles of good medicine and should be applied to screening decisions