evidence based dentistryjokstad.no/singbiostat.pdf · 21 jan 2003 1 evidence based dentistry...
TRANSCRIPT
![Page 1: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/1.jpg)
21 Jan 2003 1
Evidence Based Dentistry
BiostatisticsAsbjorn Jokstad
Institute of Clinical Dentistry,University of Oslo, Norway
![Page 2: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/2.jpg)
2
1.Clinical Trials
2.Metanalysis
3.Prognosis
4.Diagnostic tests
![Page 3: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/3.jpg)
21 Jan 2003 3
Clinical trials
![Page 4: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/4.jpg)
4
A is better
than B
A is no better
than B
A is better
than B
A is no better
than B
The truth
What the
trial shows
x
x
What can you show with a trial?
![Page 5: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/5.jpg)
5
A is better
than B
A is no better
than B
A is better
than B
A is no better
than B
The truth
What the
trial shows
x
x
What can you show with a trial?
Type 1 error
Alfa error
Optimism error
![Page 6: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/6.jpg)
6
Fallacies of observed clinical success• Spontaneous remission
• Placebo response
• Multiple variables in treatment
• Radical versus conservative treatment
• Over-treatment
• Long-term failure
• Side effects and sequelae of treatment
Type 1 error
![Page 7: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/7.jpg)
7
A is better
than B
A is no better
than B
A is better
than B
A is no better
than B
The truth
What the
trial shows
x
x
What can you show with a trial?
Type 2 error
Beta error
Pessimism error
![Page 8: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/8.jpg)
8
1. Underpowered study2. Fallacies of observed clinical failures• Wrong diagnosis
• Incorrect cause-effect correlations
• Multifactorial problems
• Lack of cooperation
• Improper execution of treatment
• Premature evaluation of treatment
• Limited success of treatment
• Psychological barriers to success
Type 2 error
![Page 9: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/9.jpg)
9
Meta-analysis
![Page 10: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/10.jpg)
10
Meta-analysis
An overview with a specific statistical technique which summarizes the results of several studies into a single estimate
![Page 11: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/11.jpg)
11
Meta-analysis/Systematic Review
• Systematic Review
– Exhaustive exploration, critical evaluation and synthesis of all the unbiased evidence
• Meta-analysis
– Exhaustive exploration, critical evaluation and quantitative synthesis of all the unbiased evidence
– Combination of the results of a number of related randomised trials
![Page 12: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/12.jpg)
12
Odds ratio = (A/B)/(C/D)
Relative risk (RR) = [A/(A+B)]/[C/(C+D)]
Adverse outcome
+ -
Treatment
+ A B
- C D
![Page 13: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/13.jpg)
13
Odds ratio = (A/B)/(C/D)
Relative risk(RR)= [A/(A+B)]/[C(C+D)]
Relative risk reduction (RRR) = 1 – RR
Absolute risk reduction(ARR)=A/(A+B)-C/(C+D)
Number needed to treat = 1/ARR
Adverse outcome
+ -
Treatment
+ A B
- C D
![Page 14: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/14.jpg)
14
Odds Ratio
1Less than 1 More than 1
Line of no difference
Favours treatment Favours control
Therapeutic gain
![Page 15: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/15.jpg)
15
Odds Ratio
If you want more of something to happen, such as greater reduction in new cavities and the experimental intervention is successful
the results will show in the right-hand side
1Less than 1 More than 1
moRe
I
g
h
t
![Page 16: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/16.jpg)
16
Odds Ratio
If you want less of something to happen, e.g less swelling following a minor surgical procedure if you prescribe a particular tablet and the experimental intervention is successful
the results will show in the left-hand side
1Less than 1 More than 1
Less
e
f
t
![Page 17: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/17.jpg)
17
Clarkson I, Worthington H. Prevention and treatment of oral mucositis and oral candidiasis for patients with cancer
![Page 18: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/18.jpg)
18
Effect of study methodology on validity
Studies of lower methodological quality, particularly those including non-representative patients or applying different reference standards, tend to overestimate the diagnostic performance of a test. Lijmer et al. JAMA, 1999; 282: 15.
Diagnostic
”gain”
![Page 19: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/19.jpg)
19
Streptokinase for infarction
![Page 20: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/20.jpg)
20
Effects of inadequate study design on results
Jüni et al.Methodological quality of controlled trials and effect estimates. BMJ 2001.
Favours treatment Favours control
![Page 21: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/21.jpg)
21
Prognosis
![Page 22: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/22.jpg)
22
• Proportion of survival or success according to some specific criteria after a given temporal interval, e.g. after 1 or 5 years
• Median time of survival (in years), where 50% of the study unit, e.g. the patient, prosthesis, restorations or tooth, have failed, or
• Survival curves – describe for each time unit along a horizontal axis estimates of the proportion of the study unit that remain intact according to survival or success according to some specific criteria
Prognosis – likelihood estimates
![Page 23: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/23.jpg)
23
Survival Curves
50
60
70
80
90
100
0 2 4 6 8 10 12 14Years
%
![Page 24: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/24.jpg)
24
Intraoral location
McLaren & White. J Prosthet Dent 2000
![Page 25: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/25.jpg)
25Napankangas et al. J Oral Rehabil, 2000
Hemmings et al. J Prosthet Dent 2000
![Page 26: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/26.jpg)
26
Aquilino et al. J Prosthet Dent 2001
Erpensten et al. J Prosthet Dent 2001
![Page 27: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/27.jpg)
27
Sjögren et al. J Prosth Dent 1999
Malament et al. J Prosth Dent 1999
![Page 28: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/28.jpg)
28
Implants freestand vs connected
Naert et al., Clinical Oral Implants Research, 2001
![Page 29: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/29.jpg)
29
Creugers et al. J Dent 2001
Etch bridges
![Page 30: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/30.jpg)
30
• All good clinical prognosis studies include measures of confidence intervals for prognosis-estimates
• A 95% confidence interval consists of two values that indicating an interval where we can be 95% certain that the true value lies
• A narrow confidence interval is an indication of a precise estimate of the true value
Prognosis - Precision of the likelihood estimates
![Page 31: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/31.jpg)
31
Sample size and confidence interval
Malament et al. Survival of Dicor glass-ceramic dental restorations over 14 years. J Prosth Dent 1999
![Page 32: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/32.jpg)
32
Diagnostic tests
![Page 33: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/33.jpg)
33
Assessment of the efficacy of a diagnostic test
Parameter Description, e.g.
Sensitivity Ability to identify patients in a patient population
Specificity Ability to identify non-patients in an asymptomatic population
Positive predictive value Ability of a diagnostic test to identify a patient correctly, given that the
test is positive
Negative predictive value Ability of a diagnostic test to identify a non-patient correctly, given that the test is
negative
Measurement validity The accuracy of a measurement technique when compared with a known
standard
Measurement reliability The variability of the measurements over time and in different envirorunents
Diagnostic validity The ability to separate those with the disease from those without the
disease
![Page 34: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/34.jpg)
34
Sensitivity and Specificity
• Sensitivity
– Probability that a subject with the disease will screen positive
• Specificity
– Probability that a subject who is disease free will screen negative
![Page 35: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/35.jpg)
35
Test
Positive
Test
Negative
Disease
Present
Disease
Absent
a
c
b
d
a+b
c+d
a+c b+d a+b+c+d
2 x 2 Tables
![Page 36: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/36.jpg)
36
Test
Positive
Test
Negative
Disease
Present
Disease
Absent
215
15
16
114
231
129
230 130
Sensitivity
215
230= 93%
Sensitivity
= a
a+c
![Page 37: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/37.jpg)
37
Test
Positive
Test
Negative
Disease
Present
Disease
Absent
215
15
16
114
231
129
230 130
Specificity
114
130= 87%
Specificity
= d
b+d
![Page 38: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/38.jpg)
38
Positive and Negative Predictive Values
• Positive Predictive Value– probability of those testing/screening positive actually
having the disease
• Negative Predictive Value– probability of those testing/screening negative NOT
actually having the disease
Relevant when you know the prevalence of the disease in the population.
![Page 39: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/39.jpg)
39
Test
Positive
Test
Negative
Disease
Present
Disease
Absent
215
15
16
114
231
129
230 130
Positive Predictive Value
Positive predictive value = a / a+b
215
231
= 93%
![Page 40: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/40.jpg)
40
Test
Positive
Test
Negative
Disease
Present
Disease
Absent
215
15
16
114
231
129
230 130
Negative Predictive Value
114
129
= 88%
Negative predictive value = d/c+d
![Page 41: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/41.jpg)
41
Likelihood Ratio
Indicates the value of the test for increasing certainty about a positive diagnosis
Sensitivity
1 - Specificity= 215/230 = 8
1- 114/130
![Page 42: Evidence Based Dentistryjokstad.no/singbiostat.pdf · 21 Jan 2003 1 Evidence Based Dentistry Biostatistics Asbjorn Jokstad Institute of Clinical Dentistry, University of Oslo, Norway](https://reader031.vdocuments.mx/reader031/viewer/2022011808/5c8d639209d3f218598b6996/html5/thumbnails/42.jpg)
42
Likelihood ratio nomogram