Prevention: When Do Benefits Start and End? Providence Internal Medicine Spring Symposium April 14, 2016 Mari Kai, MD What Do They Have in Common? 82 year old female 62 year old male Estimated Life Expectancy is about 11-13 years 1
Overview Framework for evaluating when screening interventions benefit Life expectancy estimates Age is only crude marker Prediction models for mortality (eprognosis) Lagtime to benefit When will it help? Dynamic effects of co-morbidities on harms and benefits Potential barriers to incorporate into daily practice For Sister Madonna Would you recommend a mammogram? Would you recommend colon cancer screening? Should she get a bone density? According to her ASCVD risk calculator, her 10 year risk of a CV event is 19.6% (79 yo)- would you start a statin? 2
Problem with Preventative Guidelines Many preventative guidelines use age as a crude marker to stratify groups into specific recommendations Most conclude that there is insufficient evidence for patients over age 75 Many pts under age 75 with multiple co-morbidities and reduced life-expectancy get screening, not likely to be beneficial One study reported that over 120,000 mammograms done on women with severe dementia in 2002 13% of ESRD patients who died within 5 yrs had mammograms Need a framework to guide patient-specific recommendations Decision Framework Step 1: Estimate life expectancy (LE) Will my patient live long enough to benefit? Step 2: Estimate the lagtime to benefit of the preventative intervention -When will it help? Step 3: If LE > LtB, then consider if benefits outweigh upfront risks and harms* Lee et al, JAMA 2013 dec 25:310(24) 3
Estimating Life Expectancy Physicians poor at estimating prognosis For terminal ill, tend to over-estimate Many pts get referred to hospice too late Moderately accurate on 10 yr thresholds More focus on intervention and diagnostics have reduced training, discussion on estimating prognosis No calculators for life expectancy Life Expectancy 4
eprognosis Website, app developed at UCSF Multiple calculators that can estimate different mortality rates (6mos, 1yr, 4 yr, 10 yr) Need different calculators for different decisions Living at home, nursing home, hospital Lee-Schonberg index Distilled down to 15 key questions that had the highest influence on mortality Estimate a 4 yr and a 10 yr mortality rate Yourman, L et al. Prognostic Indices for Older Adults JAMA January 11, 2012 vol 307, No. 2 Lee-Schonberg Index Questions 1-9 5
Lee-Schonberg Index Questions 10-15 Point Score No. who Died/ %mortality No. at risk 0 8/354 2.3 1 25/489 5.1 2 62/889 7.0 3 100/971 10 4 147/986 15 5 195/842 23 6 258/758 34 7 272/637 43 8 260/498 52 9 234/401 58 10 216/308 70 11 189/232 82 12 159/192 83 13 144/159 91 >14 239/257 93 C-statistic 0.834 Validation of the Lee Index for 10-year Mortality A C-statistic > 0.8 is considered an excellent accuracy of a prediction rule Cruz, M et al. JAMA. 2013 Mar 6; 309(9): 874 876. 6
Decision Framework Step 1: Estimate life expectancy (LE) Will my patient live long enough to benefit? Step 2: Estimate the lagtime to benefit of the preventative intervention -When will it help? Step 3: If LE > LtB, then consider if benefits outweigh upfront risks and harms Lee et al, JAMA 2013 dec 25:310(24) Step 2: When Will It Help? Lagtime to Benefit- time between the preventative intervention and when improved health outcomes are seen Most studies don t report this, so it has to be extrapolated from survival curves The time-point at which the curves last separate provides a qualitative estimate Not many clinical trials define the timing, safety, or risks of discontinuing interventions 7
Lagtime to Benefit for Interventions Prevention Outcome Lagtime to benefit Bisphosphonates for Fracture 3 to 6 months osteoporosis Primary preventionstatins CV events 1-2 years Secondary preventionstatins CV events 6 months-1 yr Colon cancer screening Colon CA mortality (1/1000) 10.3 years Breast cancer screening Breast CA mortality (1/1000) 10.7 years Aspirin for CV prevention Non-fatal MI (Not mortality) 5 years * Aspirin for colon CA Colon CA mortality 10 to 19 years prevention Figure 1. Cumulative incidence of cardiovascular events in the JUPITER trial, according to study group. Paul M Ridker Circ Cardiovasc Qual Outcomes. 2009;2:279-285 Copyright American Heart Association, Inc. All rights reserved. 8
WOSCOPS Trial of Primary Prevention Ford I et al. N Engl J Med 2007;357:1477-1486. 9
RCT of Discontinuing Statin Tx Kutner, J et al. JAMA Intern Med, March 2015 N=381, RCT, multicenter, unblinded, pragmatic trial Inclusion criteria: estimated life expectancy of between 1 month and 1 year statin tx of 3 months or more for primary or secondary prevention Recent decline in functional status Exclusion criteria: Active cardiovascular disease Other reasons to stop statins Intervention: Stopping statins Outcomes: Death within 60 days (primary), survival, CV events, performance status, QOL, symptoms, # of nonstatin medications, and cost savings No Survival Advantage for Statin 10
Some Quality of Life Markers Better Step 3: Benefit to Harm Ratio Non-linearity of benefits, harms With age, co-morbidities the risks of disease may increase >>> improved benefit But, the risks of harms also increase with the intervention, so upfront costs higher 11
Life expectancy 13 yrs Colon Cancer Screening LtB =10 yrs Breast Cancer Screening LtB=10 yrs -1/1000 would not die of colon or breast cancer because of screening -1/100 will have a harm from the screening intervention -potential to save 3 years of life (instead of living to 95, would live to 92) Bisphosphonate treatment LtB=3-6 months - RRR 30% Statin Primary Prevention LtB=1-2 years -RRR 30% -Baseline risk of 20% reduced to 14% -ARR 6%(6/100 would not have a CV event for taking a statin) What about colon cancer screening for Charles? Stroke- decreased function BMI 30 Diabetes COPD Smoking Lee-Schonberg Index 12 points 46%, 4 yr mortality 6.5 yr life expectancy LtB of 10 yrs > LE 6.5 yrs, so would not benefit from screening Statin tx - yes 12
Barriers to Using this Framework Time- Incorporate into Medicare Annual Wellness visit? Physicians have harder time stopping interventions or stopping offering prevention-need to find the right language when talking with patients AVOID- You re not going to live long enough to need this test or medication CONSIDER- Every patient is different, and what is best for one person may not be best for another. Chances are that you are more likely to be harmed, or will not benefit from this test or medication Difficult to discuss prognosis with patients- Need to understand how to describe risk It may be easier to discuss risk of mortality rather than life expectancy An average person with a 10 year mortality of 50%, will have a life expectancy of at least 10 years Conclusions Become familiar with predicting prognosis and using some of the prognostic indices on eprognosis Considering Lagtime to benefit (When will it help?) is as important for interventions as Does it help? Using a framework can be helpful, but it is still a very complicated decision that requires time and understanding patient preferences Need to demand future trials to include reporting specifically about lagtime to benefit 13