Home 
Contact 
Profile 
Videos 
Presentation
by Professor Omar Hasan Kasule Sr. at the Medical Grand Round King Fahad
Medical City on November 18, 2015
OVERVIEW OF DIAGNOSTIC TESTS
·
Tests are an
extension of clinical examination for signs.
·
Tests are not
100% accurate or reliable.
·
Tests are not
necessarily more accurate or more objective than physical signs for example
clinical signs of brain stem death are more reliable than laboratory and
radiological tests.
·
Test results
cannot stand alone; they are interpreted in light of other relevant clinical
data.
·
The criteria of
diagnostic tests must be standardized to enable comparability.
·
Problem of
relying on foreign standards for the range of the normal
TYPES OF DIAGNOSTIC TESTS
·
Assays of body
fluids (blood, urine, and cerebrospinal fluid)
·
Radiological
(x-ray, CAT scan, sonar)
·
Tissue biopsy
·
Function tests
(e.g. LFT, RFT).
MEASURES
OF TEST VALIDITY
·
Validity is
when a test measures what it is supposed to measure.
·
Sensitivity,
specificity, and predictive value are measures of validity (accuracy).
·
Sensitivity is
a measure of the strength of association.
·
Specificity
measures the uniqueness of association.
COMMON SENSE UNDERSTANDING OF SENSITIVITY AND SPECIFICITY
·
Use of the
‘skirt’ and ‘trouser’ as measures of validity in determining gender
·
Picture of a
man wearing trousers
·
Picture of a
woman wearing trousers
·
Picture of a
woman wearing a skirt
·
Picture of a
man wearing a skirt (Scottish)
STATISTICAL UNDERSTANDING
OF SENSITIVITY and SPECIFICITY
|
|
Test +
|
Test -
|
|
Diagnosis+
|
a
|
b
|
|
Diagnosis-
|
c
|
d
|
·
True positives
(TP) = a;
·
True negatives
(TN)=d;
·
False
negative(FN)=b;
·
False positives
(FP)=c;
·
Sensitivity=a/a+b;
·
Specificity=d/c+d.
RELATION BETWEEN SENSITIVITY and SPECIFICITY
·
There is a
trade-off between specificity and sensitivity.
·
High
sensitivity is associated with low specificity & vice versa.
·
High
specificity is associated with low sensitivity & vice versa.
·
These
relationships can be seen on a receiver operating curve that shows variation of
sensitivity with specificity.
·
True/correct
diagnosis is based on high specificity (with low or high sensitivity).
·
Pathognomonic
signs have high specificity usually 85-95%.
PREDICTIVE
VALUE OF DIAGNOSTIC TESTS
|
|
Diagnosis+
|
Diagnosis-
|
|
Test+
|
a
|
b
|
|
Test-
|
c
|
d
|
·
Predictive
value of positive test (PV+ve)= a/a+b.
·
Predictive
value of negative test (PV-ve)=d/c+d.
·
Predictive
value can be related to sensitivity and specificity using Baye’s theorem
PREDICTIVE VALUE OF A POSITIVE TEST
·
positive
predictive value = detection of disease
·
PV+ve indicates
the proportion of those with disease among those who are test-positive and can
be alternatively expressed as PV+ = TP / {TP + FP}
·
High prevalence
of disease increases PV+ve. Stated in other words this means that common
diseases are more likely to be picked up by the diagnostic test.
PREDICTIVE VALUE OF A NEGATIVE TEST
·
Negative
predictive value = correct indication of absence of disease
·
PV-ve is the
proportion with no disease among those who are test-negative.
RELATION OF PREDICTIVE VALUE TO SENSITIVITY and SPECIFICITY
·
PV+ =
[{prevalence)(Sensitivity)}] / [(prevalence)(sensitivity) +
1-specificity)(1-prevalence).
·
PV-=[{1-Prevalence)(Specificity)]
/ [{1-Prevalence) (specificity) + (1-sensitivity) (Prevalence)].
MEASURES
OF TEST REPRODUCIBILITY
·
Repeatability
·
Consistency
·
Reliability
·
Stability
REPEATABILITY
FOR CONTINUOUS DATA
·
Repeatability
is a measure of the ability of a test to give the same answer when the subject
is re-examined.
·
The measures of
repeatability used for continuous data are the standard deviation of replicate
measurements and the coefficient of determination which is the standard
deviation divided by the mean.
REPEATABILITY FOR DISCRETE DATA
|
|
Rater 1
|
Total
|
|
|
Rater 2
|
+
|
-
|
|
|
+
|
a
|
b
|
p1
|
|
-
|
c
|
d
|
q1
|
|
Total
|
p2
|
q2
|
1.0
|
·
For discrete
data two measures are used: overall agreement and the kappa coefficient of
inter-rater reliability.
·
The overall
agreement is computed as the total of the diagonal cells in a contingency table
of the scores of one rater against those of another one divided by the total
number of tests as shown below:
·
The kappa
statistic measures the observer test bias and is defined as k = {2(ad – bc)} /
{p1q2 + p2q1} where a, b, c, and d are
actual counts and p and q are proportions. A kappa statistic >0.75 indicates
excellent agreement. A kappa statistic of 0.4 to <0.75 indicates fair to
good agreement. A kappa statistic <0.4 indicates poor agreement.
RAPID MALARIA DIAGNOSTIC TEST vs GOLD STANDARD
OF ELECTRON MICROSCOPY…1
·
Light
microscopy of blood smears for diagnosis of malaria in the field has several
limitations, notably delays in diagnosis.
·
This study in
Ahmedabad in Gujarat State, India, evaluated
the diagnostic performance of a rapid diagnostic test for
malaria (SD Bioline Malaria Ag P.f/Pan) versus blood smear examination as the
gold standard.
·
All fever cases
presenting at 13 urban health centres were subjected to
rapid diagnostic testing and thick and thin blood smears.
RAPID MALARIA DIAGNOSTIC TEST vs GOLD STANDARD
OF ELECTRON MICROSCOPY…2
·
A total of 677
cases with fever were examined; 135 (20.0%) tested positive by
rapid diagnostic test and 86 (12.7%) by blood smear.
·
The sensitivity
of the rapid diagnostic test for malaria was 98.8%, specificity
was 91.5%, positive predictive value 63.0% and negative predictive value 99.8%.
·
For detection
of Plasmodium falciparum the sensitivity of
rapid diagnostic test was 100% and specificity was 97.3%.
·
The results
show the acceptability of the rapid test as an alternative to light
microscopy in the field setting.
Source:
Vyas S, Puwar B, Patel V, Bhatt G, Kulkarni S, Fancy M. Study on validity of a
rapid diagnostic test kit versus light microscopy for malaria
diagnosis in Ahmedabad city, India. East Mediterr Health J. 2014 May 1;20(4):236-41.
GLYCATED HEMOGLOBIN (A1C) vs GOLD STANDARD OF ORAL GTT FOR DIABETES
AMONG ARABS..1
·
A
population-based representative sample of 482 randomly selected adult Arabs
without known diabetes was studied.
·
A1C testing
correctly identified 5% of individuals diagnosed with diabetes by oral glucose
tolerance test, 13% by fasting plasma glucose, and 41% by both criteria.
·
A1C alone
identified 14% of individuals diagnosed with impaired glucose tolerance, 9%
with impaired fasting glucose, and 33% with both abnormalities.
GLYCATED HEMOGLOBIN (A1C) vs
GOLD STANDARD OF ORAL GTT FOR DIABETES AMONG ARABS..2
·
Sensitivity, specificity
were 19% (16-23%) and 100% (99-100%) for diabetes A1C cutpoint
·
Sensitivity, specificity
were 14% (11-17%) and 91% (89-94%) for prediabetes A1C range.
·
A1C cutpoint of
6.2% for diabetes and 5.1% for prediabetes yielded the highest accuracy but
still missed 73% of those with diabetes and 31% with prediabetes.
·
Agreement
between A1C and diabetes (κ = 0.2835) or prediabetes (κ = 0.0530) was low.
GLYCATED HEMOGLOBIN (A1C) vs GOLD STANDARD OF ORAL GTT FOR DIABETES
AMONG ARABS..3
·
Conclusion 1: A1C-based
criteria yield a high proportion of false-negative tests for diabetes and
prediabetes in Arabs.
·
Conclusion 2: Racial/ethnic
differences in A1C performance for diagnosis and prediction of diabetes exist.
Source:
Pinelli NR, Jantz AS, Martin ET, Jaber LA.
Sensitivity and specificity of
glycated hemoglobin as a diagnostic test for diabetes and
prediabetes in Arabs. J Clin Endocrinol Metab. 2011
Oct;96(10):E1680-3.