Evidence-Based Physical Diagnosis_Lect. 1_ What is Evidence-Based Physical Diagnosis

Lecture 1:
What Is Evidence-Based Physical Diagnosis?
Marc Imhotep Cray, M.D.
Evidence-Based
Physical Diagnosis

Goals
2
The goals of this presentation are
 To elucidate the term evidence-based physical diagnosis.
 To provide the learner with a first-layer understanding
modern-day physical diagnosis.
 To demonstrate how concepts in basic epidemiology,
biostatistics and probability serve as requisites to
applying clinical epidemiology, evidence-based medicine
and thus, evidence-based physical diagnosis in core
clerkships.

“Icebreaker Admonition”
3
“Read with two objectives: First to acquaint yourself with
the current knowledge on the subject and the steps by
which it has been reached; and secondly, and more
important, read to understand and analyze your cases.”
From: LeBlond RF, et al. DeGowin’s Diagnostic Examination, 10th Ed. New York:
McGraw-Hill Education, 2015; xxxi.
Originally: Sir William Osler
“The Student Life”

Why is Diagnosis Important?
4
Medical history and physical examination (H&P) are
basis for diagnostic hypothesis generation the first
step in the diagnostic process
Accurate Dx precedes three tasks central to healing
professions: explanation, prognostication & therapy
 These three tasks provide answers to patient’s three
fundamental questions:
1. What is happening to me and why?
2. What does this mean or my future?
3. What can be done about it, how will that change my future?

Why is Diagnosis Important cont’d.
5
 Failure to pursue a Dx  may permit a disease to
progress from curable to incurable
 Contrastly, for many complaints, in otherwise healthy
people w no alarm Sx or Sn a good prognosis can be
determined w/o knowing exact cause of complaint
 For example, an upper respiratory infection (URI)
o An experienced clinician can reassure pt. further testing is
unnecessary and will not change Px or Tx

Why is Diagnosis Important cont’d.
6
 It takes…
 Experience,
 Knowledge of the medical literature,
 Good judgment, and
 Understanding of fundamentals of clinical
epidemiology and decision making
…to determine when pursuit of specific Sx & Sn is
warranted
Note: For a first-rate review of principles of epidemiology, see Fletcher et al.
[Fletcher RH, Fletcher SW, Fletcher GS. Clinical Epidemiology, the Essentials. 5th
ed. Baltimore, MD: Lippincott, Williams & Wilkens, 2012].

Review of Diagnostic process:
(9 Sequential Steps)
7
Step 1: Take a History: Elicit symptoms and a timeline; begin a
problem list.
Step 2: Develop Hypotheses: Generate a mental list of anatomic sites
of disease, pathophysiologic processes, and diseases that might
produce the symptoms.
Step 3: Perform a Physical Examination: Look for signs of physiologic
processes and diseases suggested by history, and identify new
findings for problem list.
Step 4: Make a Problem List: List ALL problems found during history
and PE that require an explanation.

Diagnostic process steps cont’d.
8
Step 5: Generate a Differential Diagnosis: List most probable
diagnostic hypotheses with an estimate of their pretest
probabilities.
Step 6: Test the Hypotheses: Select laboratory tests, imaging
studies, and other procedures with appropriate likelihood ratios to
evaluate your hypotheses.
Step 7: Modify Your Differential Diagnosis: Use test results to
evaluate your hypotheses, eliminating some, adding others, and
adjusting probabilities.
Step 8: Repeat Steps 1 to 7: Reiterate your process until you have
reached a diagnosis or decided that a definite diagnosis is neither
likely nor necessary.

Diagnostic process steps cont’d.
9
Step 9: Make the Diagnosis or Diagnoses: When tests of your
hypotheses are of sufficient certainty that they meet your stopping
rule you have reached a diagnosis.
 If uncertain, consider a provisional diagnosis or watchful
waiting.
 Decide whether more investigation (return to Step l),
consultation, treatment, or watchful observation is best course
based upon severity of illness, prognosis, and comorbidities.
To learn more: LeBlond RF, et al. Part 1: The Diagnostic Framework. In: DeGowin’s
Diagnostic Examination, 9th Ed. New York: McGraw-Hill Education, 2009; 1-47.

Diagnosis
10
When clinicians diagnose disease, their intent is to place patient’s
experience into a particular category (or diagnosis) a process
implying specific pathogenesis, prognosis, and treatment
 This procedure allows clinicians to explain what is happening to patients
and to identify best way to restore pt’s health
 A century ago, such categorization of disease rested almost
entirely on empiric observation—what clinicians saw, heard, and
felt at patient’s bedside
 almost all diagnoses were based on traditional physical examination

Diagnosis cont’d.
11
For example, if patients presented a century ago w
complaints of fever and cough,  Dx of lobar
pneumonia rested on presence of characteristic PE
findings of pneumonia =
 fever, tachycardia, tachypnea, grunting respirations, cyanosis,
↓ excursion of affected side, dullness to percussion, ↑ tactile
fremitus, ↓ breath sounds (later bronchial breath sounds),
abnormalities of vocal resonance (bronchophony, pectoriloquy,
and egophony), and crackles
o If findings were absent pt. did not have pneumonia
• Chest radiography played no role in diagnosis b/c it was not widely
available until early 1900s

Diagnosis cont’d.
12
 Modern medicine, of course, relies on technology much more
than medicine did a century ago (to our patients’ advantage)
and for many modern categories of disease, diagnostic standard
is a technologic test
 For example, if patients present today with fever and cough
Dx of pneumonia is based on presence of an infiltrate on chest
radiograph
 Similarly, Dx of systolic murmurs depends on echocardiography
and that of ascites on abdominal ultrasonography
o In these disorders, clinician’s principal interest is result of technologic
test and decisions about Tx depend much more on tech result
than on,
o whether pt. exhibits egophony, radiation of murmur into neck, or
shifting dullness

Dx a Century Ago
13
McGee S, Steven R. Evidence-based Physical Diagnosis, 4th Ed. Philadelphia, PA: Elsevier, 2018; Fig..1.1, 2.
“EVOLUTION OF THE DIAGNOSTIC STANDARD”…

Dx in Modern Times
14
McGee S, Steven R. Evidence-based Physical Diagnosis, 4th Ed. Philadelphia,
PA: Elsevier, 2018; Fig..1.1, 2.
 One century ago, most Dx were
defined by bedside observation,
whereas
 Today technologic standards have a
much greater diagnostic role
 Nonetheless, there are many
examples today of Dx based solely
on bedside findings (Ex. appear in
large gray shaded box)
 Evidence-based physical diagnosis, on
other hand, principally addresses
those Dx defined by technologic
standards b/c it identifies those
traditional findings that accurately
predict result of technologic test
…“EVOLUTION OF THE DIAGNOSTIC STANDARD”

Evidence-based medicine
15
 EBM is a modern term for application of clinical
epidemiology to care of patients it includes:
 Formulating specific “answerable” clinical questions,
 Finding best available research evidence bearing on those
questions,
 Judging evidence for its validity, and
 Integrating critical appraisal w clinician’s expertise & patient’s
situation and values
See previous lecture: Introduction to Evidence Based Medicine (EBM) .Ppt

Dx in Modern Times cont’d.
16
Reliance on technology creates tension for medical
students* b/c they spend hours mastering
traditional exam yet later learn (when first appearing
wards) traditional exam pales in importance compared
to technology
 A realization prompting a fundamental question:
o What is true diagnostic value of traditional physical
examination? Is it outdated and best discarded? Is it completely
accurate and underutilized? Is the truth somewhere between
these two extremes?
* This tension applies most in Western and other
highly technologically reliant and wealthy societies.

Dx in Modern Times cont’d.
17
 “EVOLUTION OF THE DIAGNOSTIC STANDARD” discussed above indicates
Dx today is split into two parts
 For some categories of disease diagnostic standard still
remains empiric observation— what clinician sees, hears, and
feels—just as it was for all diagnoses a century ago
 For example, how does a clinician know pt. has cellulitis?
Ans: Only way is to go to patient’s bedside and observe fever
and localized bright erythema, warmth, swelling, and
tenderness on leg
o There is no other way to make this diagnosis (technologic or not)

Marc Imhotep Cray, M.D. 18
NB: The purpose of the follow snippet of concepts
in basic epidemiology, biostatistics and probability
are entirely select, and only intended to support
understanding of how these concepts are applied
in clinical epidemiology, evidence-based medicine
and thus, evidence-based physical diagnosis.
To learn more student is referred to Public Health Sciences lectures
and the Sociology, Epidemiology/Population Health (SPH) &
Interpretation of the Medical Literature (EBM) cloud folders.
“Understanding the Evidence”

Biostatistics Definitions
19
Incidence: The number of new cases of a disease in a population
over a specific period of time (=longitudinal)
Prevalence: The total number of people in a population affected
by a condition at one point in time (=cross-sectional)
Duration relates incidence to prevalence
For example:
 Upper respiratory infections (URIs) have a high incidence every year
during winter months but a low prevalence b/c most URIs resolve
quickly
Contrastly,
 Diabetes mellitus (DM) has a relatively low incidence but high
prevalence b/c a patient who has diabetes has it for life

Statistics for Diagnostic Tests
20
 True positive (Tp): Disease is present and diagnostic test is
positive  a correct result
 True negative (Tn): Disease is absent and diagnostic test is
negative  a correct result
 False positive (Fp): Disease is absent and diagnostic test is
positive  an incorrect result
 False negative (Fn): Disease is present and diagnostic test is
negative  an incorrect result

Statistics for Diagnostic Tests:
Sensitivity
21
 Sensitivity: Given disease is present, probability that test will be
positive
 Stated another way, sensitivity is ability of a test to become
positive in presence of the disease
 It is defined as Tp/(Tp + Fn) or Tp/(total number of people
with disease)
 Sensitive tests are useful for screening b/c there are few false
negatives
 A highly sensitive test can, therefore, rule out the disease
o Consider the mnemonic SN-N-OUT= for a test that is SeNsitive, a
Negative result rules OUT a disease

Sensitivity cont’d.
22
 Example: An HIV test with 98% sensitivity means that, when a
disease is present, it will be detected 98% of time
 Example: Consider a test that was positive 100% of time
regardless of presence or absence of disease
 It would technically have 100% sensitivity b/c it would be
positive in all patients with disease (but would be clinically
useless b/c it would be positive in all patients without
disease, too)
o Therefore, sensitivity is not whole picture when it comes to test
characteristics specificity is also important

Statistics for Diagnostic Tests
23
 Specificity: Given disease is absent, probability that test will be
negative
 Stated another way specificity is ability of a test to remain
negative in absence of disease
 It is defined as Tn/(Tn + Fp) or Tn/(total number of people
without disease)
 Tests with high specificity are useful to confirm a diagnosis b/c
there are few false positives
 A highly specific test can, therefore, rule in the disease
o Consider the mnemonic SP-P-IN—for a test that is SPecific, a Positive
result rules IN a disease
For Example: An HIV test with 98% specificity means that, when a disease is
absent, test will be negative 98% of time

PPV & NPV
24
Positive Predictive Value (PPV): Given test is positive, probability
that disease is present
 PPV = TP/(TP + FP) or TP/(total number of positive tests)
o For example, if a computed tomography (CT) scan has 98%
specificity for appendicitis, then given a positive finding of
appendicitis, patient will truly have disease 98% of time
 Negative Predictive Value (NPV): Given test is negative,
probability that the disease is absent
 NPV = TN/(TN + FN) or TN/(total number of negative tests)
o For example, if an HIV test has 98% NPV, then given a negative test,
patient will truly be HIV negative 98% of time

Evaluation of diagnostic tests
25
 Uses 2 × 2 table comparing test results w actual presence of
disease TP, FP, TN, FN
 Sensitivity and specificity are fixed properties of a test
 PPV and NPV vary depending on disease prevalence in
population being tested

Pre- and post-test probability
26
Pre-test probability and post-test probability (pretest and
posttest probability) are probabilities of presence of a condition
(such as a disease) before and after a diagnostic test, respectively
Post-test probability, in turn, can be positive or negative
depending on whether test falls out as a positive test or a
negative test, respectively
Ability to make a difference betw. pre- and post-test
probabilities of various conditions is a major factor in indication
of medical tests

cont’d.
27
Estimation of post-test probability: In clinical practice,
post-test probabilities are often just roughly estimated
 This is acceptable in finding of a pathognomonic Sn or Sx in
which case it is almost certain target condition is present;
or
 In absence of finding a sine qua non Sn or Sx in which case
it is almost certain target condition is absent

cont’d.
28
In reality, subjective probability of presence of a
condition is never exactly 0 or 100%
 Yet, there are several systematic methods to estimate that
probability (eg. likelihood ratios [next slide]) methods are
based on previously having performed test on a reference
group in which presence or absence of condition is known (a
test that is considered highly accurate= "Gold standard")
o These data are used to interpret test result of any individual tested
by method

Likelihood ratios (LRs) in
diagnostic testing
29
 In EBM, likelihood ratios are used for assessing value of
performing a diagnostic test (=PE, Lab or other Dx study)
 Use sensitivity and specificity of test to determine whether a test
result usefully changes probability that a condition (such as a
disease state) exists
Application
 A likelihood ratio of greater than 1 indicates test result is associated
with disease
 A likelihood ratio less than 1  indicates test result is assoc. w absence
of disease
 Tests where likelihood ratios lie close to 1 have little practical
significance as post-test probability (odds) is little different from pre-
test probability

Calculation of likelihood ratio
30
Two versions of likelihood ratio exist
 one for positive and one for negative test results
respectively, known as
o positive likelihood ratio (LR+, likelihood ratio positive, likelihood
ratio for positive results) and
o negative likelihood ratio (LR–, likelihood ratio negative, likelihood
ratio for negative results)
Positive likelihood ratio is calculated as
which is equivalent to
 Or " probability of a person who has disease testing positive divided by probability
of a person who does not have disease testing positive“
o "T+" or "T−" denote that result of test is positive or negative, respectively
o "D+" or "D−" denote that disease is present or absent, respectively
 "true positives" are those that test positive (T+) and have disease (D+), and
 "false positives" are those that test positive (T+) but do not have disease (D−)

Calculation of likelihood ratio
cont’d.
31
Negative likelihood ratio is calculated as
which is equivalent to
 or "probability of a person who has disease testing negative divided by
probability of a person who does not have disease testing negative."
Pretest odds of a particular diagnosis X multiplied by
likelihood ratio= determines post-test odds (calculation is
based on Bayes' theorem )
 MKSAP Audio 1-07 – General Medicine_The Bayes Theorem (Offline)
Note : Odds can be calculated from, and then converted to, probability

Likelihood ratios: Key Points
32
Likelihood ratios (LRs) are diagnostic weights= numbers
that quickly convey to clinicians how much a physical
sign argues for or against disease
 LRs have possible values between 0 and ∞
o Values greater than 1 ↑probability of disease (greater value of LR,
greater ↑in probability)
 LRs less than 1 decrease probability of disease (closer
number is to zero, more probability of disease ↓)
 LRs that equal 1 do not change probability of disease at all

LRs Key Points cont’d.
33
 LRs of 2, 5, and 10 increase probability of disease about 15%,
30%, and 45%, respectively (in absolute terms)
 LRs of 0.5, 0.2, and 0.1 (i.e., reciprocals of 2, 5, and 10)
decrease probability 15%, 30%, and 45%, respectively
 Tables comparing LRs of different physical signs quickly inform
clinicians about which findings have greatest diagnostic value
See: Medical Likelihood Ratio Repository
The Likelihood Ratio Database

Easy Estimation Table
34
 Use this table to estimate how likelihood ratio changes probability without
needing a calculator
Likelihood ratios in diagnostic testing. Article is issued from WikiMed Medical Encyclopedia - version 9/28/2016.

Estimation Example
35
1.Pre-test probability: For example, if about 2 out of every 5
patients with abdominal distension have ascites, then pretest
probability is 40%
2.Likelihood Ratio: An example "test" is that the physical exam
finding of bulging flanks has a positive likelihood ratio of 2.0 for
ascites
3.Estimated change in probability: Based on table of previous slide,
a likelihood ratio of 2.0 corresponds to an approximately + 15%
increase in probability
4.Final (post-test) probability: Therefore, bulging flanks increases
probability of ascites from 40% to about 55% (i.e., 40% + 15% =
55%, which is within 2% off exact probability of 57%)

Calculation Example
36
 An example is likelihood that a given test result would be expected in a
patient with a certain disorder compared to likelihood that same result
would occur in a patient without target disorder
 A worked example: A diagnostic test w sensitivity 67% and specificity 91% is applied
to 2, 030 people to look for a disorder with a population prevalence of 1.48%
Likelihood ratios in diagnostic testing. Article is issued from WikiMed Medical Encyclopedia - version 9/28/2016.

37
THE END
See next slide for further study tools.

Further study:
38
Recommended textbook reading
 McGee S, Steven R. Evidence-based Physical Diagnosis, 4th Ed.
Philadelphia, PA: Elsevier, 2018; 1-18.
Cloud folders
 Sociology, Epidemiology/Population Health (SPH)
 Interpretation of the Medical Literature (EBM)

Evidence-Based Physical Diagnosis_Lect. 1_ What is Evidence-Based Physical Diagnosis

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Evidence-Based Physical Diagnosis_Lect. 1_ What is Evidence-Based Physical Diagnosis

Similar to Evidence-Based Physical Diagnosis_Lect. 1_ What is Evidence-Based Physical Diagnosis (20)

More from Imhotep Virtual Medical School

More from Imhotep Virtual Medical School (20)

Recently uploaded

Recently uploaded (20)

Evidence-Based Physical Diagnosis_Lect. 1_ What is Evidence-Based Physical Diagnosis