What is Screening? Hui Jin Department of Epidemiology and Health Statistics School of Public Health Southeast University
SCREENING: DEFINITION “The PRESUMPTIVE identification of UNRECOGNIZED disease or defect by the application of tests, exams or other procedures which can be applied RAPIDLY to sort out apparently well persons who PROBABLY have a disease from those who PROBABLY do not”* Key Elements: disease/disorder/defect screening test population *Commission on Chronic Illness, 1957
Issues in Screening Disease -Disease/disorder should be an important public health problem High prevalence Serious outcome -Early Detection in asymptomatic (pre-clinical) individuals is possible -Early detection and treatment can affect the course of disease (or affect the public health problem?)
Screening Test Concerned with a Functional Definition of Normality versus Abnormality Screening Test Normal Abnormal
Screening Spectrum Risk factor Recognized symptomatic disease Presymptomatic disease Unrecognized symptomatic disease è Fewer people è Easier to demonstrate benefit è Less potential for harm to exceed benefit
Issues in Screening for Risk Factors Risk factor treatment disease Does risk factor predict disease? Does treatment reduce risk factor? Does identification/treatment of risk factor reduce disease? Potential for harm exceeding benefit greatest when screening for risk factors! Caution: risk factors as surrogate outcomes
Cardiac Arrhythmia Suppression Trial (CAST) Are PVC’s after MI a risk factor for sudden death? Yes Do encainide and flecainide decrease PVCs? Yes Do these drugs save lives? NO! RCT showed total mortality after 10 months higher in treated group vs placebo: 8.3% vs. 3.5% (P <0.0001) Echt DS et al. N Engl J Med. 1991;324:781-8 Moore TJ. Deadly Medicine. NY: Simon and Schuster, 1995
Lipid screening for kids: Does screening detect risk factor? Yes Benefits to screening? Not studied Possible risks to children/society? Cost, testing, distraction from other priorities
Goals of Screening for Presymptomatic Disease Detect disease in earlier stage than would be detected by symptoms Only possible if an early detectable phase is present (latent phase) Begin treatment earlier Only beneficial if earlier treatment is more effective than later treatment Do this without greater harm than benefit
Special Case: Screening for Cancer Natural history heterogeneous Screening test may pick up slower growing or less aggressive cancers Not all patients diagnosed with cancer will become symptomatic “Pseudodisease” Diagnosis is subjective There is no gold standard
Malignant Benign
Interobserver Agreement Among Pathologists for Malignant Melanoma: 24 disagreements Malignant Can’t tell Benign
Why Not?
Possible harms from screening To those with a negative result To those with a positive result To all
Is this test sensitive enough? The general teaching: Maximize sensitivity for screening tests This is true IF Goal is not to miss anyone with the disease HOWEVER…. NPV already good in low- prevalence population
False Positives vs Pseudodisease
9/10/2002Natural history; population screening19 1. Suitable disease 2. Suitable test 3. Suitable program 4. Good use of resources Requirements for a screening program
Serious consequences if untreated Detectable before symptoms appear Better outcomes if treatment begins before clinical diagnosis 1. Suitable disease
Detect during pre-symptomatic phase Safe Accurate Acceptable, cost-effective 2. Suitable test
Reaches appropriate target population Quality control of testing Good follow-up of positives Efficient 3. Suitable program
Cost of screening tests Cost of follow-up diagnostic tests Cost of treatment Benefits versus alternatives 4. Good use of resources
U.S. Preventive Services Task Force December 4, 2009
U.S. Preventive Services Task Force December 4, 2009
David Shabtai Faculty Peer Reviewed In a bold move, the U.S. Preventive Services Task Force recently changed their breast cancer screening guidelines – recommending beginning screening at age 50 and even then only every other year until age 75. Bold, because the Task Force members are certainly aware of the media circus that ensued when in 1997, an NIH group issued similar guidelines, prompting comparisons to Alice in Wonderland. Revisiting the USPSTF Breast Cancer Screening Guidelines: Ethics, and Patient Responsibilities
September 10, 2010 Recommended Weekend Reading By NATASHA SINGER “Can we trust doctors’ recommendations on cancer screening, given that the medical profession has a vested financial interest in treating patients? That is one of the questions posed in a provocative article this week in The New England Journal of Medicine that looks at the fallout last year after a government panel recommended that women start having mammograms later in life and less frequently.” Mammography Wars
September 29, 2010 Mammogram Benefit Seen for Women in Their 40s By GINA KOLATA Researchers reported Wednesday that mammograms can cut the breast cancer death rate by 26 percent for women in their 40s. But their results were greeted with skepticism by some experts who say they may have overestimated the benefit. Who should get a mammogram?
Newsweek The Mammogram Hustle There is no evidence digital mammograms improve cancer detection in older women. But thanks to political pressure, Medicare pays 65 percent more for them. This story was reported and written by Center for Public Integrity. What should we pay for?
By Julie Steenhuysen CHICAGO | Wed Jan 26, :26pm EST (Reuters) - A new analysis of evidence used by a U.S. advisory panel to roll back breast cancer screening guidelines suggests it may have ignored evidence that more frequent mammograms save more lives, U.S. researchers said on Tuesday. New U.S. analysis backs annual breast screening
“The U.S. Preventive Services Task Force (USPSTF) “chose to ignore the science available to them” and brought about “potential damage to women’s health” in its 2009 recommendations for more limited mammography screening, costing an estimated 6,500 deaths in women each year, a study published in the February issue of the American Journal of Roentgenology concluded.” AJR: USPSTF mammo recommendations could cost 6,500 lives yearly
Survival time after diagnosis – lead time Pre-detectable Detectable, preclinical Clinical Disability or death Possible detection via screening Clinical detection Age: Lead time
Survival time must increase > lead time Pre-detectable Undetected (no screening) Clinical diagnosis & treatment Disability or death Age: Pre-detectable Early detect, diagnosis, & treatment Monitoring for recurrence ? Survival time after diagnosis Lead time
Slowly progressing diseases are easier to detect by screening Pre- detectable Clinical diagnosis, treatment Disability or death Age: Pre-detectable Detectable, pre-clinical Clinical diagnosis & treatment Disability or death Survival time after diagnosis
Early detection may over-diagnose Pre-detectable Undetected (no screening) Mild or no symptoms Favorable outcome Age: Pre-detectable Early detect, diagnosis, & treatment Monitoring for recurrence Favorable outcome Survival time after diagnosis Survival time after dx
Criteria for Evaluating a Screening Test Validity : provide a good indication of who does and does not have disease -Sensitivity of the test -Specificity of the test Reliability : (precision): gives consistent results when given to sameperson under the same conditions Yield : Amount of disease detected in the population, relative to the effort-Prevalence of disease/predictive value
Screening test Reliable – get same result each time Validity – get the correct result Sensitive – correctly classify cases Specificity – correctly classify non-cases [screening and diagnosis are not identical]
Reliability Repeatability – get same result Each time From each instrument From each rater If don’t know correct result, then can examine reliability only.
Validity versus Reliability of Screening Test Examiner 1Examiner 2Examiner 3 True cases Good Reliability Low Validity
Reliability Percent agreement is inflated due to agreement by chance Kappa statistic considers agreement beyond that expected by chance Reliability does not ensure validity, but lack of reliability constrains validity
Validity: 1) Sensitivity Probability (proportion) of correct classification of cases Cases found / all cases
Validity: 2) Specificity Probability (proportion) of correct classification of noncases Noncases identified / all noncases
Consider: -The impact of high number of false positives: anxiety, cost of further testing -Importance of not missing a case: seriousness of disease, likelihood of re-screening Where do we set the cut-off for a screening test?
Sensitivity of a screening test Probability (proportion) of correct classification of detectable, pre- clinical cases
Specificity of a screening test Probability (proportion) of correct classification of noncases Noncases identified / all noncases
True positive True negative False positive False negative Sensitivity = True positives All cases a + c b + d = a a + c Specificity = True negatives All non-cases = d b + d a + b c + d True Disease Status Cases Non-cases Positive Negative Screening Test Results a d b c
True Disease Status Cases Non-cases Positive Negative Screening Test Results a d 1,000 b c 60 Sensitivity = True positives All cases ,000 = Specificity = True negatives All non-cases = 19,000 20,000 1,140 19, ,000 = = 70% 95%
Yield from a Screening Test for Disease X Predictive Value X X Screening Test Negatives Positives X X X X
Yield from the Screening Test: Predictive Value Relationship between Sensitivity, Specificity, and Prevalence of Disease Prevalence is low, even a highly specific test will give large numbers of False Positives Predictive Value of a Positive Test (PPV): Likelihood that a person with a positive test has the disease Predictive Value of a Negative Test (NPV): Likelihood that a person with a negative test does not have the disease
Interpreting test results: predictive value Probability (proportion) of those tested who are correctly classified Cases identified / all positive tests Noncases identified / all negative tests
True positive True negative False positive False negative PPV = True positives All positives a + c b + d = a a + b NPV = True negatives All negatives = d c + d a + b c + d True Disease Status CasesNon-cases Positive Negative Screening Test Results a d b c
True Disease Status Cases Non-cases Positive Negative Screening Test Results a d 1,000 b c 60 PPV = True positives All positives ,000 = 140 1,140 NPV = True negatives All negatives = 19,000 19,060 1,140 19, ,000 = = 12.3% 99.7%
Positive predictive value, Sensitivity, specificity, and prevalence Prevalence (%) PV+ (%) Se (%) Sp (%)
Example: Mammography screening of unselected women Disease status Cancer No cancer Total Positive ,117 Negative 47 62,295 62,342 Total ,280 63,459 Prevalence = 0.3% (179 / 63,459) Se = 73.7% Sp = 98.4% PV+ = 11.8% PV– = 99.9% Source: Shapiro S et al., Periodic Screening for Breast Cancer
What is used as a “gold standard” 1. Most definitive diagnostic procedure e.g. microscopic examination of a tissue specimen 2. Best available laboratory test e.g. polymerase chain reaction (PCR) for HIV virus 3. Comprehensive clinical evaluation e.g. clinical assessment of arthritis
Principles for Screening Programs 1.Condition should be an important health problem 2.There should be a recognizable early or latent stage 3.There should be an accepted treatment for persons with condition 4.The screening test is valid, reliable, with acceptable yield 5.The test should be acceptable to the population to be screened 6. The cost of screening and case finding should be economically balanced in relation to medical care as a whole
Question? Assigned readings, session 6 Topic: Interpretation of screening tests Grimes DA, Schultz KF. Uses and abuses of screening tests. Lancet 2002;359:881-4.