Chronic eye disease:Role and Overview of
ScreeningRichard Wormald
London School of Hygiene and Tropical Medicine
acknowledgements
Richard Wormald is funded in the UK by financial support from the Department of Health through the award made by the National Institute for Health Research to Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology for a Specialist Biomedical Research Centre for Ophthalmology.
The views expressed in this presentation are those of the author and not necessarily those of the Department of Health.”
No commercial conflicts of interest or financial gain exist to the presenter in the context and contents of this talk
In this talk
What is screening?
What is its relevance to chronic eye disease?Children
ROP, Amblyopia, refractive error etcOlder people
vision impairmentPrevention of blindness
• Diabetic retinopathy• Glaucoma
Open angle glaucoma
Angle closure
Evidence for effectivenessLead time bias
Length bias
Overdiagnosis and disease mongering
Confusion due to loose use of poorly defined terms
• Screening• Active identification and enumeration, invitation and testing
an entire and defined population at risk• Case finding
• Any method for finding people who may be affected by disease before they present to health services
• Indiscriminate testing of participants at health fairs and exhibitions etc, is often more about marketing a product than a public health intervention
• Opportunistic surveillance• Testing people in a clinical setting when they attend for other
concerns – such as testing for glaucoma in a person needing reading glasses
• Screening tests may be used in all of the above as well as in population sample surveys and trials
screening
• Is a specific public health intervention intended to reduce the burden (morbidity/mortality/disability, economic impact) of a specific disease or condition in a defined population (at risk)
• It is a totally different model of health care delivery to the traditional one where the patient actively seeks help
• In screening, we impose a health intervention on a population so evidence of safety and effectiveness is even more important
• Implicit is that early detection in a presymptomatic phase of the condition has a strong beneficial influence on outcome
• Only relevant for chronic conditons – hence wet ARMD not suitable• Screening is harmful: false negatives, false positives and wastage of
scarce resources• Ultimately, the only way to determine effectiveness of screening is
the impact on the burden of disease in the population
Measuring the impact of screening with reliable outcomes
• What is the intended effect?• For Breast Cancer
• reduced mortality from the disease
• For glaucoma, diabetic retinopathy• Prevention of blindness
• Squint and amblyopia?• Number of cases detected is a process measure not
an outcome• Measuring the effect can only be achieved by ..
• Randomised controlled trials of screening• Economic modelling
UK National screening committee criteria
Based on 10 commandments of Wilson and Jungner WHO 1968
http://www.screening.nhs.uk/criteria#fileid9287
The condition – 1..4
The test – 5…9
The treatment – 10….12
The programme – 13….22
The condition
1. The condition should be an important health problem
2. The epidemiology and natural history of the condition, including development from latent to declared disease, should be adequately understood and there should be a detectable risk factor, disease marker, latent period or early symptomatic stage.
3. All the cost-effective primary prevention interventions should have been implemented as far as practicable.
4. If the carriers of a mutation are identified as a result of screening the natural history of people with this status should be understood, including the psychological implications.
The test
5. There should be a simple, safe, precise and validated screening test.
6. The distribution of test values in the target population should be known and a suitable cut-off level defined and agreed.
7. The test should be acceptable to the population.
8. There should be an agreed policy on the further diagnostic investigation of individuals with a positive test result and on the choices available to those individuals.
9. If the test is for mutations the criteria used to select the subset of mutations to be covered by screening, if all possible mutations are not being tested, should be clearly set out.
The treatment
10. There should be an effective treatment or intervention for patients identified through early detection, with evidence of early treatment leading to better outcomes than late treatment.
11. There should be agreed evidence based policies covering which individuals should be offered treatment and the appropriate treatment to be offered.
12. Clinical management of the condition and patient outcomes should be optimised in all health care providers prior to participation in a screening programme.
The screening programme13. There should be evidence from high quality Randomised Controlled Trials that the screening programme is effective in reducing mortality or morbidity. Where screening is aimed solely at providing information to allow the person being screened to make an “informed choice” (eg. Down’s syndrome, cystic fibrosis carrier screening), there must be evidence from high quality trials that the test accurately measures risk. The information that is provided about the test and its outcome must be of value and readily understood by the individual being screened.
14. There should be evidence that the complete screening programme (test, diagnostic procedures, treatment/ intervention) is clinically, socially and ethically acceptable to health professionals and the public.
15. The benefit from the screening programme should outweigh the physical and psychological harm (caused by the test, diagnostic procedures and treatment).
16. The opportunity cost of the screening programme (including testing, diagnosis and treatment, administration, training and quality assurance) should be economically balanced in relation to expenditure on medical care as a whole (ie. value for money). Assessment against this criteria should have regard to evidence from cost benefit and/or cost effectiveness analyses and have regard to the effective use of available resource.
The screening programme
17. All other options for managing the condition should have been considered (eg. improving treatment, providing other services), to ensure that no more cost effective intervention could be introduced or current interventions increased within the resources available.
18. There should be a plan for managing and monitoring the screening programme and an agreed set of quality assurance standards.
19. Adequate staffing and facilities for testing, diagnosis, treatment and programme management should be available prior to the commencement of the screeningprogramme.
20. Evidence-based information, explaining the consequences of testing, investigation and treatment, should be made available to potential participants to assist them in making an informed choice.
21. Public pressure for widening the eligibility criteria for reducing the screening interval, and for increasing the sensitivity of the testing process, should be anticipated. Decisions about these parameters should be scientifically justifiable to the public.
22. If screening is for a mutation the programme should be acceptable to people identified as carriers and to other family members.
Screening for vision impairment in the elderly
Systematic review and economic analysis
Exhaustive review of the literature on prevalence, accuracy of screening tests and economic modeling of glaucoma screening
Large team of reviewers, clinicians and economists working over 18 months
Less than £200k
Results – bottom line only
• Most screening tests had an estimated specificity of 85% or higher
• None were clearly superior and good quality evidence was scarce
• Prevalence would have to be 3-4% in 40 year olds with a screening interval of 10 years to approach cost effectiveness
• Selective screening of higher risk groups with a higher risk might be worthwhile although this would cover only 6% of the population
• Screening using a test with initial automated classification followed by assessment by specialised optometrist for test positives was more cost-effective than initial assessment by optometrists
More bottom line
• The cost-effectiveness of the screening programme was highly sensitive to the perspective on costs (NHS or societal).
• In the base-case model, the NHS costs of visual impairment were estimated as £669. If annual societal costs were £8800, then screening might be considered cost-effective for a 40-year-old cohort with 1% OAG prevalence assuming a willingness to pay of £30,000 per quality-adjusted life-year.
• Of lesser importance were changes to estimates of attendance for sight tests, incidence of OAG, rate of progression and utility values for each stage of OAG severity.
More bottom line
• Cost-effectiveness was not particularly sensitive to the accuracy of screening tests within the ranges observed.
• However, a highly specific test is required to reduce large numbers of false-positive referrals.
• The findings that population screening is unlikely to be cost-effective are based on an economic model whose parameter estimates have considerable uncertainty.
• In particular, if rate of progression and/or costs of visual impairment are higher than estimated then screening could be cost-effective.
Screening tests
Vs diagnostic testsWhat’s the difference?
Hierarchy of tests “gold” or reference standard,
• Maybe composite – requiring more than one test or outcome• Maybe longitudinal – does early detection lead to better later
outcome?
Diagnostic test• Maybe one component of the reference standard
Screening test• Positive result increases the probability of a diagnosis justifying
further investigation – colon cancer, prostate cancer, negative result rules it out (almost)
Validity of diagnostic tests
SeNsitivity and sPecificityN for false negatives, P for false positives
1- N, Sensitivity = proportion of true positives of all positive cases
1 – P, = proportion of true negatives out of all negative cases
+PPV – predictive value of a positive test, probability of a positive test being a true positive
-PPV – predictive value of a negative test, probability of negative test being a true negative
disease (or truth) always on top
Disease positive Disease negative
Test positive True positive False positive +PPV
Test negative False negative True negative -PPV
sensitivity specificity Prevalence
Worked example - POAG
Disease positive Disease negative
Test positive 18 98 116 (18.4%)
Test negative 2 882 884 (0.2%)
20 (90%) 980 (90%) 1000 (2%)
Other important process measures
CoverageThe proportion of the target population tested in
one screening round
A screening test with 100% sensitivity will be much less sensitive in reality if coverage is only 50% especially when responders are the worried well and non-responders, the careless sick
yield
How many tested to reveal one true positive
1000/18 = 56
EconomicsCost per case detected
(Cost per test x 1000) / 56
Cost of false positive evaluation
Against cost of not detecting or preventing morbidity/mortality
Opportunity cost
Other statistics
ROC (receiver operating characteristic) curvePlot sensitivity by 1 – specificity for different test
cut-off values, calculate the area under the curve
Likelihood ratio statistics
Diagnostic odds ratio
Bayesian methods
ROC curves
Positive predictive value and prevalence
Disease positive
Disease negative
total
Test positive
Test negative
total
Positive predictive value and prevalence
Disease positive
Disease negative
total
Test positive
Test negative
total 20 1000
Positive predictive value and prevalence
Disease positive
Disease negative
total
Test positive
18
Test negative
2
total 20 1000
Positive predictive value and prevalence
Disease positive
Disease negative
total
Test positive
18
Test negative
2
total 20 980 1000
Positive predictive value and prevalence
Disease positive
Disease negative
total
Test positive
18 98 116
Test negative
2 882 884
total 20 980 1000
Positive predictive value and prevalence 90/90/2
PPV = 18/116 = 15.5%
Between one in six or seven people who screen positive have the disease
Positive predictive value and prevalence – 99%
sensitivityDisease positive
Disease negative
total
Test positive
198 980 1160
Test negative
2 8820 8840
total 200 9800 10,000
Positive predictive value and prevalence 99/90/2
PPV = 198/1160 = 0.17
Between one in five or six people who screen positive have the disease
Positive predictive value and prevalence – 99%
specificityDisease positive
Disease negative
total
Test positive
180 98 278
Test negative
20 9702 9722
total 200 9800 10000
Positive predictive value and prevalence 90/99/2
PPV = 180/278 = 0.65%
One in every one or two people with positive test have the disease
Positive predictive value and prevalence – 20%
prevalenceDisease positive
Disease negative
total
Test positive
180 80 260
Test negative
20 720 740
total 200 800 1000
Positive predictive value and prevalence 90/90/20
PPV = 180/260 = 0.69%
One in every one or less than two people who test positive have the disease
Where and when used
The validity of a screening test can only be evaluated in the setting in which it is to be used on a sample of the population to be screened.
The ability of the test to discriminate between disease and non-disease states will depend on the prevalence of the disease in the sample population and the case mix
Discriminating between known obvious cases and known obvious normals will not reflect its performance in the real world.
Angle closure Glaucoma
• Definable at risk population by a simple noninvasive test?
• Single quick safe and effective intervention?
• Potential long term benefit• But also harms?• Is lens extraction better than PI?• We await ZAP and EAGLE
Lead time and length bias
Lead time biasDetecting a disease earlier in it course may
create the illusion of longer survival where in fact all that has occurred is a prolonged awareness (and anxiety) due to the illness
Length biasDisease with slow natural history and more
benign cause more likely to be detected by interval screening than rapidly progressive aggressive disease
Lead time bias
AFMC Primer on population health
Length bias
AFMC Primer on Population Health
Pros and Cons
Prevention better than cure
Early diagnosis is key to good outcome
Health awareness and surveillance
Overdiagnosis and overtreatment
Disease mongering
The worried well – half the population are sick
Preventing overdiagnosis
Clear case definition
setting the threshold for disease status where
1: the diagnosis is not in doubt
2: intervention has a clear margin of benefit over harm
3: and is cost-effective (including opportunity cost)
IN ANCIENT CHINA, THE DOCTOR WAS PAID ONLY WHEN THE PATIENT WAS WELL?
Is it true?