1- Evidence For Specific Types Of Clinical Quest

Front 1

Why do we do diagnostic tests?

Back 1

-TO REDUCE DIAGNOSITC UNCERTAINTY
-impt quest. to ask:
1. what do we think the likelihood is that our pt has the disease before we do the test (pre-tst probability)- often based on clinical judgement
2. what is the likelihood that out pt has the disease after the test (post-test probability)
-tests that produce a big change from pre-test to post-test probability are likely to be important an useful in practice

Front 2

search cascade for diagnosis articles

Back 2

Systems, synopses, syntheses, studies
-cochrane reviews of diagnositc test accuracy

Front 3

Tools for Article Appraisal: Users’ Guides Series in JAMAEBM Working Group

Back 3

1. Are the Results VALID?
2. Are the Results IMPORTANT?
3. Will the Results Help Me Care for My Patients?

Front 4

Is this evidence about diagnosis valid?

Back 4

1. was there an independent blind comparison with a reference ("gold") standard of diagnosis?
2. was the diagnostic test evaluated in an appropriate spectrum of pts?
3. Was the reference standard applied regardless of the diagnostic test result?
4. was the cluster of tests validated in a second, independent group of pts
-Does this valid evidence demonstrate an important ability of this test to accurately distinguish between patients who do and don’t have a specific disorder?

Front 5

sensitivity

Back 5

-Ability of the test to identify people who DO have the disease
-Proportion of patients with the condition who test positive for the condition.
a (true positive test results)/ a + c (all pts with disease
-Highly Sensitive tests don’t miss many of the people who have the disease (few false-negatives)
-used for screening or ruling out disease (SnOUT)

Front 6

specificity

Back 6

-ability of the test to identify people who do not have the disease
-proportion of pts w/out the condition whose test results are - for the condition
d (TN) / d + b (all pts without disease)
-Highly Specific tests very rarely label people with a diagnosis if they don’t have the disease (few false-positives)
-used for confirming or ruling in disease (SpPIN)

Front 7

Sensitivity and Specificity

Back 7

-Tell us how good a test is at identifying the presence or absence of disease
-Doesn’t tell us what is more useful to us as clinicians (and what our patients want to know):

Front 8

Positive Predictive Value

Back 8

-proportion of pts with a positive test who have the condition
a (TP)/ a + b (all positives)
-changes with prevalence

Front 9

Negatives Predictive Values

Back 9

-proportion of pts with a negative test who do not have the condition
d (TN) / d + c (all negatives)
-changes with prevalence

Front 10

Likelihood ratios (LR)

Back 10

-Like sensitivity and specificity, LR describe the discriminatory power of diagnostic tests

Front 11

Positive Likelihood raito

Back 11

-Shows how much the odds of disease are increased if the test result is positive.

Sensitivity/ (1 - specificity)

Front 12

Negative Likelihood ratio

Back 12

-Shows how much the odds of disease are decreased if the test result is negative.

(1 - sensitivity) / specificity

Front 13

what does LR mean?

Back 13

1. LR of 2 increases probability by 15%
2. LR of 5 increases probability by 30%
3. LR of 10 increases probability by 45%
4. LR of 0.5 decreases probability by 15%
5. LR of 0.2 decreases probability by 30%
6. LR of 0.1 decreases probability by 45%

Front 14

how to interpret likelihood ratios

Back 14

-LR >10 or <0.1 Cause LARGE changes in disease likelihood- will make a difference it what you do for patients
-LRs ≥10, with pre-test probabilities ≥ 33% generate post-test probabilities ≥ 83% (more likely to start doing something for the patient)
-LRs ≤ 0.1, with pre-test probabilities ≤ 33% generate post-test probabilities ≤ 5%
-LR=1 Causes no change at all

Front 15

post-test probability =

Back 15

pre-test probability (%) + LR (%)

Front 16

Determining the Pre-test Probability

Back 16

1. Available evidence (prevalence)
2. Published estimates
3. Your personal experience
4. Expert opinion

Front 17

Is this evidence about the pre-test probability valid?

Back 17

1. did the study pts represent the full spectrum of those who present with this clinical problem?
2. Were the criteria for each final diagnosis explicit and credible?
3. was the diagnostic work-up comprehensive and consistently applied?
4. for initially undiagnosed patients, was follow-up sufficiently long and complete?

Front 18

in this valid evidence about pre-test probability important?

Back 18

1. what were the diagnoses and their probabilities?
2. how precise were these estimates of disease probability?

Front 19

Precision

Back 19

-how closely individual measurements agree with each other

Front 20

screening and case finding

Back 20

Screening: early diagnosis of disease among asymptomatic individuals
Case finding: making early diagnosis of disease among patients who have come to us for some other unrelated disorder

Front 21

Potential harm from screening

Back 21

1. risks of work-up and treatment, if any
2. "labeling"
3. True positives: "healthy" time can turin into "sick" time
4. False positives: only experience harm

Front 22

Lead-time between screening and usual detection

Back 22

-Earlier detection will almost always appear to improve survival!!
-Why? –b/c if you measure survival time from when it was detected but it was detected before the people would have experienced symptoms

Front 23

Observational Cohort Study: Prognosis

Back 23

pts at risk of target event --> prognostic factor --> time --> suffer target outcome or do not suffer target outcome

Front 24

Is this evidence about prognosis valid?

Back 24

1. Was a defined, representative sample of pts assembled at a common point in the course of their disease?
2. was f/u of study pts sufficiently long and complete?
3. were objective outcome criteria applied in a "blind" fashion? i.e. death vs. cause of death
if subgroups with different prognoses are identified?
-was there adjustment for impt prognostic factors?
-was there validation in an independent group of "test-set" pts?

Front 25

Kaplan-Meier Survival Curves

Back 25

-The “curve” is a step function, with sudden changes in the estimated probability corresponding to times at which
an event was observed. The times of the censored data
are indicated by short vertical lines.

Front 26

Relative Risk

Back 26

-Risk of an event is the # of subjects who have the event in the group divided by # subjects
-Can be compared to risk of same event in another group (Relative Risk or Risk Ratio; 1 = no difference between groups)

Front 27

Hazard Ratio (HR)

Back 27

-a specific type of relative risk: “treatment” hazard rate, divided by “placebo” hazard rate
-calculated using "survival analysis:" how many subjects have NOT experienced the event at any given time (i.e. disease free); descending curve
-accounts for subjects dropping out before the event of interest happens or the study ending before all of the subjects experience the event
-HR =1 means no difference between groups

Front 28

How precise are the prognostic estimates?

Back 28

-Look at the Confidence Interval!
-if the 95% CI crosses 1.0, the difference is not significant
-If relative risk was the same in both groups it would be 1.0

Front 29

Can we apply this valid, important evidence about prognosis to our patient?

Back 29

1. Is our patient so different from those in the study that its results cannot apply?
2. Will this evidence make a clinically important impact on our conclusions about what to offer or tell our patient?