DIAGNOSTICS AT WORK:
From the clinic to the community
Nitika Pant Pai, MD, MPH., PhD
Madhukar Pai, MD, PhD
WHY SCREENING, DIAGNOSTIC AND
PROGNOSTIC TESTS MATTER
o Diagnosis is the first and important step in the pathway to correct
treatment
• “Companion diagnostics” are emerging as a critical theme
o Early and rapid diagnosis can reduce morbidity, improve patient
outcomes, and reduce cost of care
o Tests can
o identify risk factors
o predict prognosis
o monitor therapy over time
o promote healthy behaviours
o tailor therapies (i.e. personalized medicine)
IN THE PAST, DOCTORS RELIED ON THEIR
CLINICAL INTUITION…
Source: The Hindu
AND SOME BASIC TESTS LIKE MICROSCOPE
Antonie van Leeuwenhoek
(1632–1723), the father of
microbiology, is best known
for his work to improve the
microscope and use it for
microbiology
Source: Wikipedia
STETHOSCOPE AND X-RAYS
Rene Laënnec in Paris invented
the first stethoscope in 1816.
Wilhelm Rontgen detected
x-rays in 1895.
Source: Wikipedia
TODAY, WE HAVE COME A LONG WAY…
www.nature.com
THE OLD STETHOSCOPE COMES OF AGE!
http://www.thinklabs.com/
TESTS FOR TUBERCULOSIS: FROM MICROSCOPE
TO RAPID MOLECULAR TESTS (<2 HOURS)
TESTS FOR HIV:
FROM THE LAB TO OUR HOMES
ULTRASONOGRAPHY:
FROM SPECIALIST TO FAMILY DOC
Source: www.insideradiology.com
Source: www.siemens.co.uk
YOUR MOBILE SMARTPHONE HAS EMERGED AS
A POWERFUL NEW DIAGNOSTIC TOOL!
Retina
imaging
http://www.healthcanal.com/eyes-vision/41915-eye-phone-that-could-help-prevent-blindness.html
Vital signs, oxygen
saturation
http://lgtmedical.com/global-health/pneumonia.html
YOUR TABLET IS NOW A HEALTH
TECHNOLOGY ASSISTANT
This Slate is a book-sized
machine; it wirelessly
communicates with an
Android tablet and includes a
bag of plug-and-play sensors
that measure blood pressure
and levels of blood sugar and
hemoglobin, conduct
electrocardiography (EKG)
tests, etc.
http://www.the-scientist.com/?articles.view/articleNo/33761/title/A-Dime-a-Dozen/
WEARABLE HEALTH TECHNOLOGIES:
Fitbit, iWatch, OMwear RECORD SLEEP,
VITAL SIGNS
http://www.nuubo.com/
http://www.cnet.com/news/wearable-tech-most-important-race-turning-heartbeats-into-cash/
WEARABLE HEALTH
TECHNOLOGIES
Real-time glucose monitoring
http://www.medtronicdiabetes.com/res/img/misc/guardian-introducing.png
WE ARE SO READY FOR THE TRICORDER!
http://blogs-images.forbes.com/markpmills/files/2012/01/tricorder-spock1.jpg
INNOVATION: HIVSmart!:
INTEGRATED SMARTPHONE APP AND
INTERNET BASED HIV SELF TESTING
STRATEGY
PROBLEM:
HEALTH FACILITY BASED TESTING
o Barriers:
o Stigma and Discrimination
o Social Embarrassment
o Fear of visibility
o Lack of Confidentiality
o Judgemental Attitudes
o
About 50% people WORLDWIDE and 25% DOMESTICALLY don’t know their
HIV status!!
19
NEED
a. Confidential, Anonymous, personalized,
b. convenient, non invasive, cost effective
c. Easy linkages, easy conduct
d. Saves time and money
HIV SELF-TEST APP: ANDROID, iPHONE
Copyright protected. HIVSmart! is copyright protected authored by Dr Pant Pai and trainees (Roni Deli-Houssein,
Sushmita Shivkumar, Caroline Vadnais) and owned by McGill University
21
SMARTER CONDUCT &
LINKAGES
© Pant Pai and McGill University 2013
22
SO, WE HAVE NOW ENTERED THE BRAVE
NEW WORLD OF DIAGNOSTICS!
o Faster, Easier, time savings
o Real-time
o Personalized
o Connectivity (from your
mobile to your doctor’s
laptop)
o At point of clinical care, at
home, any place,
anywhere!
And the world of big data
1. Data:
1. Social, medical, genetic, daily caloric intake, daily sleep, vital signs. Step
count, all will be available (in real time)
2. Anonymized, compliant with confidentiality guidelines
2. Sources:
1. from electronic medical records available on your phone, to your family
physician and in your health facility
2. from smart wear and Fitbits- on the internet and phones
3. from personalized genomics testing are available on the cloud
3. What for?
1. For a personalized, plan for health promotion, targeted medication intake,
monitoring, and better management.
This is all exciting,
but…
HOW DO WE KNOW THESE NEW TESTS ARE
ACCURATE? THAT THEY ACTUALLY MAKE A
DIFFERENCE??
o This is where clinical epidemiologists enter the
picture.
o Diagnostic tests, just like drugs and vaccines, need
adequate validation before they can be used on
people.
o Just like drug trials, we do diagnostic trials.
DIAGNOSIS VS SCREENING:
THEY ARE DIFFERENT!
o A diagnostic test is done on sick people
• patient presents with symptoms
• pre-test probability of disease
prevalence is high)
is
high
(i.e.
disease
o A screening test is usually done on asymptomatic,
apparently healthy people
• healthy people are encouraged to get screened
• pre-test probability of disease is low (i.e. disease prevalence
is low)
29
PROCESS OF DIAGNOSIS: ALL ABOUT
PROBABILITY!
Test
Treatment
Threshold
Threshold
0%
No Tests
100%
Probability of
Diagnosis
Need to Test
Treat
30
THE PERFECT DIAGNOSTIC TEST
☺
X
No Disease
☻
Y
Disease
31
VARIATIONS IN DIAGNOSTIC TESTS
☺
☻
Overlap
Range of Variation in Disease free
Range of Variation in Diseased
32
SO, CUT-POINTS MATTER A LOT!
o Many tests produce continuous numbers, and
doctors tend to use cut-points to make decisions
o Cut-points are a compromise – they are not
perfect
o Same test can produce different results, based
on cut-points used
o Cut-points can change over time
There is no perfect test!
All we can hope to do is increase
or decrease probabilities, and
Bayes’ theorem helps with this
process
34
BAYES' THEORY
What you thought before + New information = What you think now
pre-test
probability
post-test
probability
Test
Post-test odds = Pre-test odds x Likelihood ratio
35
Bayesian approach to
diagnosis
o An accurate test will help reduce
uncertainty
o The pre-test probability is revised using
test result to get the post-test
probability
o Tests that produce the biggest changes
from pretest to post-test probabilities
are most useful in clinical practice [very
large or very small likelihood ratios]
post-test
probability
HIGH
pre-test
probability
LOW
Test
pre-test
probability
HIGH
Test
post-test
probability
LOW
36
Steps in evaluating a diagnostic test
o Define gold standard
o Recruit consecutive patients in whom the test is indicated (in
whom the disease is suspected)
o Perform gold standard and separate diseased and disease free
groups
o Perform test on all and classify them as test positives or negatives
o Set up 2 x 2 table and compute:
•
•
•
•
Sensitivity
Specificity
Predictive values
Likelihood ratios
37
Evaluating a diagnostic test
•Diagnostic 2 X 2 table*:
Disease + Disease Test +
True
Positive
False
Positive
Test -
False
Negative
True
Negative
38
Sensitivity
[true positive rate]
Disease
present
Disease
absent
Test
positive
True
positives
False
positives
Test
negative
False
negative
True
negatives
The proportion of patients with disease who test
positive = P(T+|D+) = TP / (TP+FN)
39
Specificity
[true negative rate]
Disease
present
Disease
absent
Test
positive
True
positives
False
positives
Test
negative
False
negative
True
negatives
The proportion of patients without disease who test
negative: P(T-|D-) = TN / (TN + FP).
40
Predictive value of a positive test
Disease
present
Disease
absent
Test
positive
True
positives
False
positives
Test
negative
False
negative
True
negatives
Proportion of patients with positive tests who have
disease = P(D+|T+) = TP / (TP+FP)
41
Predictive value of a negative test
Disease
present
Disease
absent
Test
positive
True
positives
False
positives
Test
negative
False
negative
True
negatives
Proportion of patients with negative tests who do not have
disease = P(D-|T-) = TN / (TN+FN)
42
Imagine a hypothetical population
(some with disease and others without)
Loong T BMJ 2003;327:716-719
©2003 by British Medical Journal Publishing Group
If a test was positive in everyone, what would
you make of this test?
Loong T BMJ 2003;327:716-719
©2003 by British Medical Journal Publishing Group
What about this scenario?
Loong T BMJ 2003;327:716-719
©2003 by British Medical Journal Publishing Group
In reality, most tests will produce these sorts of results
Loong T BMJ 2003;327:716-719
©2003 by British Medical Journal Publishing Group
Example: Ultrasonography for Down Syndrome
NUCHAL FOLD & DOWN SYNDROME
Down Syndrome
Yes
No
21
4
25
7
188
195
28
192
220
Nuchal fold Positive
Negative
Sensitivity = 75%
Specificity = 98%
N Engl J Med 1987;317:1371
LR+ = 36
LR- = 0.26
The interpretation of this test will depend a
lot on the context (i.e. prior probability)
o Scenario 1:
• Mrs. B, a 20-year old woman with a previous normal pregnancy,
seen at a community hospital
o Scenario 2:
• Mrs. A, a 37-year old woman with a previous affected
pregnancy, seen at a high-risk clinic in a tertiary, referral hospital
o Who has a higher pretest probability of Down syndrome?
TAKE HOME MESSAGES
o Diagnostic tests are important and can help all of us
o But, they are not perfect
o Diagnostic and screening tests are very different and
should not be confused
o Doctors and patients need to understand that all tests
have their inherent error (i.e. false positives and false
negatives)
o Tests should always be interpreted in context
o Tests should be avoided unless there is a clear indication
THANK YOU!!!
We are grateful to:
• Prof Jim Hanley
• Prof Abe Fuks
• Mini-Med team:
• Elise, Gloria, Ibby & Kappy
• Department of Epidemiology,
Biostatistics & Occupational Health
Slide credits: Sehar Manji, Executive Assistant