-
3
1. (Introduction) 4 2. (Asking answerable question) 5 3.
(Acquiring the evidence) 7 4. (Appraising the evidence) 13
(Etiology / Harm) 13 (Prognosis) 14 (Diagnosis) 15
/ (Therapy / Prevention) 16 (Systematic reviews) 17 (Economic
analysis) 18 5. (Applying the evidence) 19 6. (Assessing outcome)
21 7. (Glossary) 23
-
4
1
(EBM) EBM = (Evidence-based medicine is the conscientious,
explicit and judicious use of current best evidence in making
decisions about the care of individual patients, David L Sackett)
EBM (information explosion) EBM (Etiology/Harm) (Diagnosis)
(Prognosis) / (Therapy/Prevention) (Health economics) EBM (The 5 As
of EBM) 1. A sk question : 2. A cquire evidence : 3. A ppraise
evidence : 4. A pply evidence : 5. A ssess outcome :
-
5
2
EBM (Asking clinical question) P I C O
1. (P roblem or P atient) 2. (I ntervention) 3. (C omparison
intervention) 4. (O utcome)
(P)
.....
(I)
........................ estrogen .......
(C)
............... estrogen
........................................
(O)
.....
(Etiology/Harm) (Prognosis) (Diagnostic test) (Therapy or
prevention) (Health economics)
-
6
PICO (asking clinical question)
Domain Patient (P) Intervention (I) Comparison (C)
Outcome (O)
Therapy
acute
M.I.
aspirin
placebo
recurrent M.I.
Prognosis
hemodialysis
hemodialysis
.
life expectancy
Diagnosis
coronary disease
exercise ECHO
exercise EKG
coronary artery disease
Etiology / Harm
hormone
replacement therapy (HRT)
HRT
CA breast
Economic
chronic renal
failure
renal
transplant
hemodialysis
(cost- effective)
-
7
3
EBM (Acquiring evidence) 1. (key words) (P) (I) (C) (O) domain
(etiology, diagnosis, prognosis therapy, prevention, etc.) (cohort
studies, case-control studies, randomized controlled trial,
systematic review, meta-analysis) 2.
Cochrane Library http://gateway.ovid.co
m website
Bibliographic database (MEDLINE)
www.pubmed.com
CATs (Critically appraised topics)
www.ebem.org/cats/
appraise
appraise
TRIP (Turning Research Into Practice)
www.tripdatabase.com
-
8
3. internet websites
PUBMED OVID MEDLINE 1. PUBMED National Library of Medicine
www.pubmed.com address web browser (Internet Explorer) 1
1
Click Clinical Queries ( 1) 2
2
-
9
category ( 2) therapy diagnosis etiology prognosis Enter subject
search click Go
2. OVID MEDLINE : PUBMED OVID MEDLINE OVID server .
(http://library.ra.mahidol.ac.th) OVID 3
3
14 1 click
select more than one database to search 5
-
10
click click to begin search 4
4
(Medical subject Heading, MeSH) Enter Keyword or
phrase check box Limit to click PICO prognosis cohort studies,
prognosis, survival analysis
-
11
Ovid Medline Filters for Evidence-based Clinical Queries
Enter Keyword or phrase click
Domain : Therapeutics/Interventions Domain : Diagnosis1 exp
research design/ 2 exp clinical trials/ 3 comparative study/ or
placebos/ 4 multicenter study.pt. 5 clinical trial$1.pt. 6
random$.ti,ab. 7 (double blind$ or triple blind$3).ti,ab. 8
placebo$.ti,ab. 9 (clinical adj trial$1).ti,ab. 10 exp
epidemiologic research design/ 11 (controlled clinical trial or
randomized controlled trial).pt. 12 practice guideline.pt. 13
feasibility studies/ 14 clinical protocols/ 15 exp treatment
outcome/ 16 or/1-15
1 exp "sensitivity and specificity"/ 2 false negative reactions/
or false positive reactions/ 3 (sensitivity or specificity).ti,ab.
4 (predictive adj value$1).ti,ab. 5 (likelihood adj ratio$1).ti,ab.
6 (false adj (negative$1 or positive$1)).ti,ab. 7 (randomized
controlled trial or controlled clinical trial).pt. 8 double blind
method/ or single blind method/ 9 practice guideline.pt. 10
consensus development conference$.pt. 11 random$.ti,ab. 12 random
allocation/ 13 (single blind$3 or double blind$3 or triple
blind$3).ti,ab. 14 or/1-13
-
12
Domain : Etiology Domain : Prognosis 1 random$.ti,ab. 2 exp
epidemiologic studies/ 3 odds ratio/ 4 cohort$.ti,ab. 5 (case$1 adj
control$).ti,ab. 6 risk$.ti,ab. 7 (odds adj ratio$1).ti,ab. 8
causa$.ti,ab. 9 (relative$1 adj risk$).ti,ab. 10 predispos$.ti,ab.
11 (randomized controlled trial or controlled clinical trial).pt.
12 exp risk/ 13 practice guideline.pt. 14 case-control studies/ 15
or/1-14
1 exp cohort studies/ 2 prognosis/ 3 exp mortality/ 4 exp
morbidity/ 5 (natural adj history).ti,ab. 6 prognos$.ti,ab. 7
course.ti,ab. 8 predict$.ti,ab. 9 exp "outcome assessment (health
care)"/ 10 outcomes$1.ti,ab. 11 (inception adj cohort$1).ti,ab. 12
disease progression/ 13 exp survival analysis/ 14 or/1-13
key word Enter Keyword or phrase click
16+17 search strategy save option save username password ()
-
13
4
(Etiology) (Harm) 1. (valid)
1.1 1.2 1.3 2 1.4 80
1.5
2. (importance) relative risk (R.R.) cohort study odds ratio
(O.R.) case-control study (confidence interval or C.I.) 1 () NNH
(number needed to harm) 1
NNH = [{PEER (OR 1) } +1] / [PEER (OR 1) x (1 PEER)] [PEER =
patients expected event rate ]
-
14
(Prognosis) 1. (valid)
1.1
1.2
1.3 (adjust)
1.4
1.5 80 1.6 blind
2. (importance)
2.1. (risk of outcome) 3
- - 50 - 2.2 (precision)
(confidence interval)
-
15
(Diagnosis) 1. (valid)
1.1 (gold or reference standard) 1.2 1.1 blind (/
) 1.3 1.4 (reproducible)
2. (importance) accuracy test 2.1 sensitivity :
2.2 specificity : 2.3 predictive values : 2.4 likelihood ratio
(L.R) :
Disease present (gold std) Disease absent (gold std) Test + a b
Test - c d Sensitivity = a/(a+c) Specificity = d/(b+d) Positive
predictive value = a/(a+b) Negative predictive value = c/(c+d) L.R.
for positive test = {a/(a+c)}/{b/(b+d)} L.R. for negative test =
{c/(a+c)}/{d/(b+d)} Prevalence (or pre-test probability) = (a+c) /
(a+b+c+d)
-
16
/(Therapy/Prevention) 1. (valid)
1.1 1.2 1.3 (randomized list concealed) 1.4 80 1.5
intention-to-treat (analyze as randomized) 1.6 1.7
2. (importance) 2.1 p-value (
-
17
(Systematic reviews) 1. (valid)
1.1 1.2 1.3 (validity) 1.4
2. (importance)
2.1 2.2 2.3 (precision) confidence interval
2.4 3. (Applicability) 3.1 3.2 3.3 3.4
-
18
(Economic analysis) 1. (valid)
1.1 2 1.2 cost-effectiveness, cost-benefit cost-utility 1.3 cost
outcome 1.4 sensitivity analysis
2. (importance)
2.1 cost outcome 2.2 cost outcome (subgroup) 2.3 (sensitivity
analysis)
3. (Applicability) 3.1 3.2 3.3
-
19
5
6. 6.
Diagnostic test pre-test probability
Pre-test odds = (pre-test probability) / (1 pre-test
probability) Post-test odds = pre-test odds x LR Post-test
probability = post-test odds (ost-test odds + 1) Therapy PEER
(Patients Expected Event Rate) NNT () = 1 / (PEER x RRR) 3. 4. 5.
6.
-
20
6
- - - outcome -
David Sacket. Evidence-Based Medicine : How to Practice and
Teach EBM ( New York : Churchill Livingstone, 2000) www.cebm.net
www.cebm.utoronto.cawww.cochrane.org
-
21
Glossary of Evidence-based Medicine
Absolute risk reduction (ARR) treatment effects Accuracy (true
positive + true negative) / (true positive + true negative + false
positive + false negative) Age standardization () Alpha error error
Type I Alternative hypothesis hypothesis Analytic study
cross-sectional, cohort case-control Anectdotal evidence 1 (case
report) Association 2 (statistical association) (causal
association) Attributable fraction (exposed, A.R.) exposure (IE -
IU ) / IU (RR - 1) / RR IE incidence in the exposed group, IU
incidence in the unexposed group, RR = relative risk Attributable
fraction (population, P.A.R.) () exposure (IT - IU ) / IU B(RR - 1)
/ {B(RR-1) + 1} IT incidence in the total population, IU incidence
in the unexposed group, RR = relative risk, B = (exposed
proportion)
-
22
Attributable risk (exposure, E) IE - IU excess risk Beta error
error Type II Bias (systematic error) (non-systematic or random
error) Blinding single-blind double-blind intervention Case-control
study (case) (control) (exposure) odds ratio Case series () ()
Cause ( Hills criteria) -temporality : -strength : -experiment :
-consistency : 1 -coherence : -specificity : -dose-response
relationship -biologic plausibility : Chi-square test (discrete or
categorical variables)
-
23
Clinical decision analysis intervention 3 (choice) (chance)
(values) intervention Clinical epidemiology classical epidemiology
Clinical practice guideline (CPG) Clinical significance magnitude
of effect Clinical trial interventions interventions Cluster
sampling Cohort Cohort study 2 (exposure) 2 Co-intervention
intervention intervention Confidence interval (CI) 95 95% CI = 5 -
15 90 99 Confidence limits (CL) confidence interval
-
24
Confounding 2 3 Confounding variable, confounder ( 3) 2
Contamination control intervention Control group, controls Control
event rate (CER) event rate Correlation coefficient 2 continuous 1
+1 1 1 (linear relationship) Correlation (ecological) study
Cost-benefit analysis () Cost-effectiveness analysis () ()
intervention Cost-minimization analysis () (health effect)
Cost-utility analysis (utility) quality gain quality-adjusted life
year (QALY) 1 Critical appraisal (validity) (application)
-
25
Crossover study design 2 Cross-sectional study (exposure)
(outcome) Cumulative incidence average risk Decision analysis (or
clinical decision analysis) Deduction Descriptive epidemiology
Differential misclassification misclassification Dose-response
relationship outcome exposure Double-blind blinding Dropout
Ecological fallacy Ecological survey (aggregated data) (risk
factor) (outcome) Effectiveness efficacy Efficacy effectiveness
Efficiency
-
26
Error, type 1 (alpha error) null hypothesis Error, type 2 (beta
error) null hypothesis Event rate EER (experimental event rate) CER
(control event rate) PEER (patient expected event rate) treatment
effects Evidence-based health care (evidence-based medicine)
Evidence-based medicine Experimental event rate (EER) event rate
Exposure External validity validity False negative False positive
Gold standard Hawthorne effect Hierarchy of evidence I: randomized
controlled trial 1 study II-1 controlled trial randomization
-
27
II-2 cohort case-control studies 1 II-3 uncontrolled studies III
(case report / series) Historical control () Hypothesis -null
hypothesis (H0) -alternative hypothesis (HA) Inception cohort ()
Indirect standardizeation/adjustment standardization Inference
inferential statistic Information bias exposure outcome Informed
consent Intention-to-treat analysis Internal validity validity
Kappa 2 2 (P0 Pc) / (1 Pc) P0 = observed agreement, Pc = chance
agreement Lead-time bias ()
-
28
Length bias ( ) Likelihood ratio (LR) Positive LR = Sensitivity
/ (1 Specificity) Negative LR = (1 Sensitiivity ) / Specificity
Meta-analysis Misclassifaction
-differential misclassification -non-differential
misclassification Negative predictive value (NPV) Null hypothesis
hypothesis Number needed to treat (NNT) 1 1 / ARR Observational
study intervention Odds p / (1 - p) Odds ratio 2 odds odds of case
( exposure exposure ) odds of control ( exposure exposure )
-
29
Outbreak epidemic p-vale (probability value) null hypothesis
Patient expected event rate (PEER) event rate Power () 1- beta
error Type II Positive predictive value (PPV) Post-test odds
pre-test odds x likelihood ratio Post-test probability post-test
odds / 1 + post-test odds Precision (reliability, repeatability,
reproducibility) (valid) Pre-test odds pre-test probability / (1 +
pretest probability) Pre-test probability or prevalence (point
prevalence) (period prevalence) Publication bias Random error
Randomization (or random allocation) () outcome
-
30
Randomized controlled clinical trial (RCT) randomize Recall bias
Relative risk reduction (RRR) treatment effects Research Risk Risk
ratio (RR) () () RR = EER / CER Screening Sensitivity a / a + c
SnNout sensitivity (Sn) (N) rule out Specificity d / b + d SpPin
specificity (Sp) (P) rule in Standardization 2 -direct
standardization -indirect standardization Statistical significance
p-value 0.05 Survival analysis Systematic error bias Systematic
review (critical appraisal)
-
31
Treatment effects ARR (absolute risk reduction) = EER - CER RRR
(relative risk reduction) = (EER - CER) / CER good outcome RRI
(relative risk increase) = (EER - CER) / CER bad outcome ABI
(absolute benefit increase) = ARR RBI (relative benefit increase) =
RRR NNT (number needed to treat) = 1 / ARR good outcome NNH (number
needed to harm) = 1 / ARR bad outcome EER = experimental event
rate; CER = control event rate Validity ( bias) Validity, study
-internal validity -external validity (generalizability)
(EBM) 2
3
1 2Domain : Etiology(Etiology) NNH = [{PEER (OR 1) } +1] / [PEER
(OR 1) x (1 PEER)](Prognosis) (Diagnosis)a
/(Therapy/PrevenARR=CEREER; RRR=(CER EER) / CER
(Systematic re (Economic ana