powerpoint presentation
DESCRIPTION
TRANSCRIPT
![Page 1: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/1.jpg)
![Page 2: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/2.jpg)
Ben A. Dwamena, MDDepartment of Radiology, University of Michigan Medical
School
Nuclear Medicine Service, VA Ann Arbor Health Care System
Ann Arbor, Michigan
METAGRAPHITI
![Page 3: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/3.jpg)
Statistical Graphics For Interpretation, Exploration And Presentation Of Meta-analysis Data
METAGRAPHITI
![Page 4: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/4.jpg)
VISUOGRAPHIC FRAMEWORK FOR
Exploring distributional assumptions
Testing and correcting for
publication bias
Investigating heterogeneity
Summary of Data and Sensitivity
Analyses
METAGRAPHITI
![Page 5: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/5.jpg)
Avoid potential misrepresentation by faulty distributional and other statistical assumptions.
Facilitates greater interaction between the researcher and the data by highlighting interesting and unusual aspects of the quantitative data.
METAGRAPHITI
![Page 6: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/6.jpg)
User-friendlier summaries of large, complicated quantitative data sets
Preliminary exploration before definite data synthesis
Effective emphasis of important features rather than details of data
METAGRAPHITI
![Page 7: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/7.jpg)
CONTINGENCY TABLE FOR SINGLE STUDY
![Page 8: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/8.jpg)
True Positives =Experimental Group With Outcome Present (a).
False Positives = Control Group With Outcome Present (b).
False Negatives=Experimental Group With Outcome Absent (c).
True Negatives= Control Group With Outcome Absent (d).
DIAGNOSTIC VERSUS TREATMENT TRIAL
![Page 9: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/9.jpg)
Odds Ratio (OR) =(a x d)/(b x c).
Relative risk in experimental group {[a/(a +
c)]/[b/(b+ d)]} =Likelihood Ratio for a
Positive Test.
Relative Risk in Control Group = Likelihood
Ratio for a Negative Test.
DIAGNOSTIC VERSUS TREATMENT TRIAL
![Page 10: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/10.jpg)
Box plots
Normal quantile plots
Stem-and-Leaf plots
DISTRIBUTION PLOTS
![Page 11: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/11.jpg)
Displays important characteristics of the dataset based on the five-number summary of the data.
“Box” covers inter-quartile range. “Beltline” of box represents the
median value. “Whiskers” include all but outlier
observations.
BOX AND WHISKER PLOT
![Page 12: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/12.jpg)
12
34
56
BOX AND WHISKER PLOT
![Page 13: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/13.jpg)
1** | 47 1** | 81,93 2** | 20,48 2** | 51,86 3** | 04,22 3** | 59,81,85 4** | 10 4** | 59,67,67,68 5** | 24,34,48 5** | 57,58,67 6** | 06rounded to nearest multiple of .01 plot in units of .01
STEM-AND-LEAF PLOT
![Page 14: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/14.jpg)
Plot of standardized effect size, Ei/Vi vs. normal distribution.
Deviations from linearity deviations from normality.
Slope of regression line =standard deviation of data= 1 for effect size if the studies from a single population and have large samples.
The y-intercept of the regression =the
mean.
NORMAL QUANTILE PLOT
![Page 15: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/15.jpg)
NORMAL QUANTILE PLOT
![Page 16: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/16.jpg)
NORMAL QUANTILE PLOT
![Page 17: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/17.jpg)
NORMAL QUANTILE PLOT
![Page 18: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/18.jpg)
NORMAL QUANTILE PLOTS
![Page 19: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/19.jpg)
Selective publication of articles showing certain types of results over those of showing other types of results
Commonly, tendency to publish only studies with statistical significant results
PUBLICATION BIAS
![Page 20: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/20.jpg)
Published studies do not represent all studies on a specific topic.
Trend towards publishing statistically significant (p < 0.05) or clinically relevant results.
Publication bias assessed by examining asymmetry of funnel plots of estimates of odds ratios vs. precision.
INVESTIGATING PUBLICATION BIAS
![Page 21: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/21.jpg)
Funnel plot
Begg’s rank correlation plot
Egger’s regression plot
Harbord’s modified radial
plot
INVESTIGATING PUBLICATION BIAS
![Page 22: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/22.jpg)
A funnel diagram (a.k.a. funnel
plot, funnel graph, bias plot)
Special type of scatter plot with
an estimate of sample size on
one axis vs. effect-size estimate
on the other axis
FUNNEL PLOT
![Page 23: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/23.jpg)
Based on statistical principle that sampling error decreases as sample size increases
Used to search for publication bias and to test whether all studies come from a single population
FUNNEL PLOT
![Page 24: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/24.jpg)
FUNNEL PLOTS
![Page 25: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/25.jpg)
metafunnel ldor seldor, xlab(0(2)8) xtitle (Log odds ratio) ytitle(Standard error of log OR) saving(zfunnel, replace)
metafunnel ldor seldor, xlab(0(2)8) xtitle(Log odds ratio) ytitle (Standard error of log OR) egger saving (eggerfunnel, replace)
FUNNEL PLOT: STATA SYNTAX
![Page 26: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/26.jpg)
FUNNEL PLOT: STATA DIALOG
![Page 27: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/27.jpg)
0.5
11.5
2S
tand
ard
err
or
of lo
g O
R
0 2 4 6 8Log odds ratio
Funnel plot with pseudo 95% confidence limits
FUNNEL PLOT: EXAMPLE
![Page 28: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/28.jpg)
0.5
11.5
2S
tand
ard
err
or
of lo
g O
R
0 2 4 6 8Log odds ratio
Funnel plot with pseudo 95% confidence limits
FUNNEL PLOT: WITH REGRESSION LINE
![Page 29: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/29.jpg)
An adjusted rank correlation method to assess the correlation between effect estimates and their variances.
Deviation of Spearman's rho from zero=estimate of funnel plot asymmetry.
Positive values=a trend towards higher levels of effect sizes in studies with smaller sample sizes
BEGG’S BIAS TEST
![Page 30: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/30.jpg)
metabias LogOR seLogOR, graph(b) saving(beggplot, replace)
BEGG’S BIAS TEST: STATA SYNTAX
![Page 31: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/31.jpg)
BEGG’S BIAS TEST: STATA DIALOG
![Page 32: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/32.jpg)
Begg's funnel plot with pseudo 95% confidence limits
logOR
s.e. of: logOR0 .5 1 1.5 2
0
5
10
BEGG’S BIAS PLOT
![Page 33: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/33.jpg)
Adjusted Kendall's Score (P-Q) = 26 Std. Dev. of Score = 40.32 Number of Studies = 24 z = 0.64 Pr > |z| = 0.519 z = 0.62 (continuity corrected) Pr >|z| = 0.53(continuity corrected)
BEGG’S BIAS TEST: STATISTICS
![Page 34: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/34.jpg)
Assesses potential association b/n effect size and precision.
Regression equation: SND = A + B x SE(d)-1.
1. SND=standard normal deviate (effect, d divided by its standard error SE(d));
2. A =intercept
3. B=slope. .
EGGER’S REGRESSION TEST
![Page 35: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/35.jpg)
The intercept value (A) = estimate of asymmetry of funnel plot
Positive values (A > 0) indicate higher levels of effect size in studies with smaller sample sizes.
EGGER’S REGRESSION METHOD
![Page 36: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/36.jpg)
EGGER’S BIAS PLOT
![Page 37: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/37.jpg)
metabias logOR selogOR, graph(e) saving(eggerplot, replace)
EGGER’S BIAS TEST: STATA SYNTAX
![Page 38: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/38.jpg)
EGGER’S BIAS TEST: STATA DIALOG
![Page 39: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/39.jpg)
Egger's publication bias plot
standardized effect
precision0 1 2 3 4
0
2
4
6
8
EGGER’S BIAS PLOT: EXAMPLE
![Page 40: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/40.jpg)
------------------------------------------------------------- Std_Eff | Coef. P>|t| [95% CI]
-------------+-----------------------------------------------
slope | 1.737492 0.001 .8528166 2.622168
bias | 1.796411 0.002 .7487423 2.84408
-------------------------------------------------------------
EGGER’S BIAS TEST: STATISTICS
![Page 41: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/41.jpg)
Test for funnel-plot asymmetry Regresses Z/sqrt(V) vs. sqrt (V),
where Z is the efficient score and V is Fisher's information (the variance of Z under the null hypothesis).
Modified Galbraith plot of Z/sqrt(V) vs. sqrt(V) with the fitted regression line and a confidence interval around the intercept.
MODIFIED BIAS TEST(HARBORD)
![Page 42: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/42.jpg)
metamodbias tp fn fp tn, graph z(Z) v(V) mlabel(index) saving(HarbordPlot, replace)
MODIFIED BIAS TEST: STATA SYNTAX
![Page 43: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/43.jpg)
8
18
213
137
15
64
92211
10
12
24
17
1
2
1416 19
5
23
20
05
10
15
Z/sqrt(V)
0 1 2 3 4sqrt(V)
Study regression line 90% CI for intercept
MODIFIED BIAS PLOT
![Page 44: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/44.jpg)
----------------------------------------------------------------------------- ZoversqrtV | Coef. Std. Err. P>|t| [90% Conf. Interval]
--+--------------------------------------------------------------------------
sqrtV| 2.406756 .3464027 0.000 1.811933 3.00158
bias| .9965934 .6383554 0.133 -.0995549 2.092742-----------------------------------------------------------------------------
MODIFIED BIAS TEST: STATISTICS
![Page 45: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/45.jpg)
A rank-based data augmentation technique
used to estimate the number of missing studies and to produce an adjusted estimate of test accuracy by imputing suspected missing studies.
Both random and fixed effect models may be used to assess the impact of model choice on publication bias.
TRIM-AND-FILL METHOD
![Page 46: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/46.jpg)
•metatrim LogOR seLogOR, eform funnel print graph id(author)saving(tweedieplot, replace)
TRIM-AND-FILL TEST: STATA SYNTAX
![Page 47: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/47.jpg)
TRIM AND FILL: STATA DIALOG
![Page 48: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/48.jpg)
TRIM-AND-FILL BIAS PLOTFilled funnel plot with pseudo 95% confidence limits
th
eta
, fil
led
s.e. of: theta, filled0 .5 1 1.5 2
-2
0
2
4
6
![Page 49: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/49.jpg)
When effect sizes differences are attributable to only sampling error, studies are homogeneous.
Heterogeneity means that there is between-study variation and variability in effect sizes exceeds that expected from sampling error.
HETEROGENEITY
![Page 50: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/50.jpg)
Potential sources of heterogeneity:
1. Characteristics of study population
2. Variation in study design
3. Statistical methods
4. Covariates adjusted for (if relevant)
HETEROGENEITY
![Page 51: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/51.jpg)
DEALING WITH HETEROGENEITY
Use analysis of variance with
the log odds ratio as dependent
variable and categorical
variables for subgroups as
factors to look for differences
among subgroups
![Page 52: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/52.jpg)
DEALING WITH HETEROGENEITY
Repeat analysis after excluding
outliers
Conduct analysis with predefined
subgroups
Construct multivariate models to
search for the independent effect of
study characteristics
![Page 53: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/53.jpg)
Standardized effect vs. reciprocal of the standard error.
Small studies/less precise results appear on the left side and the largest trials on the right end .
GALBRAITH’S PLOT
![Page 54: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/54.jpg)
A regression line , through the origin, represents the overall log-odds ratio.
Lines +/- 2 above regression line =95 per cent boundaries of the overall log-odds ratio.
The majority of points within area of +/- 2 in the absence of heterogeneity.
GALBRAITH’S PLOT
![Page 55: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/55.jpg)
galbr LogOR seLogOR, id(index) yline(0) saving(gallplot, replace)
GALBRAITH’S PLOT: STATA SYNTAX
![Page 56: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/56.jpg)
GALBRAITH PLOT: STATA DIALOG
![Page 57: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/57.jpg)
b/se(b)
1/se(b)
b/se(b) Fitted values
0 3.80729
-2-2
0
2
13.1045
.
7 8131511
18321
1222
1419
17
69
2
24
1
4
10
23
16
5
20
GALBRAITH PLOT: EXAMPLE
![Page 58: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/58.jpg)
This plots the event rate in the experimental (intervention) group against the event rate in the control group
Visual aid to exploring the heterogeneity of effect estimates within a meta-analysis.
L’ABBE PLOT
![Page 59: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/59.jpg)
labbe tp fn fp tn, s(O) xlab(0,0.25,0.50,0.75,1) ylab(0,0.25,0.50,0.75,1) l1("TPR) b2("FPR") saving(flabbeplot, replace)
L’ABBE PLOT: STATA SYNTAX
![Page 60: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/60.jpg)
L’ABBE PLOT: STATA DIALOG
![Page 61: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/61.jpg)
TPR
FPR0 .25 .5 .75 1
0
.25
.5
.75
1
L’ABBE PLOT: EXAMPLE
![Page 62: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/62.jpg)
twoway (rcap dorlo dorhi Study, horizontal blpattern(dash))(scatter Study dor, ms(O)msize(medium) mcolor(black))(scatter DOR_with_CIs eb_dor, yaxis(2) msymbol(i) msize(large) mcolor(black))(scatteri 26 83, msymbol(diamond) msize(large)), ylabel(1(1)25 26 "OVERALL", valuelabels angle(horizontal)) xlabel(0 10 100 1000 10000) xscale(log) ylabel(1(1)25 26 "Pooled Estimate", valuelabels angle(horizontal) axis(2)) legend(off) xtitle(Odds Ratio) xline(83, lstyle(foreground)) saving(OddsForest, replace)
DATA SUMMARY: STATA SYNTAX
![Page 63: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/63.jpg)
metan tp fn fp tn, or random nowt sortby(year) label(namevar=author, yearvar=year) t1(Summary DOR, Random Effects) b2(Diagnostic Odds Ratio) saving(SDORRE, replace) force xlabel(0,1,10,100,1000)
DATA SYNTHESIS: RANDOM EFFECTS
![Page 64: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/64.jpg)
• metan tp fn fp tn, or fixed nowt sortby(year) label(namevar=author, yearvar=year) t1(Summary DOR, Fixed Effects) b2(Diagnostic Odds Ratio) saving(SDORFE, replace) force xlabel(0,1,10,100,1000)
DATA SYNTHESIS: FIXED EFFECTS
![Page 65: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/65.jpg)
106.33(6.40 - 1765.75107.23(40.31 - 285.28107.67(6.57 - 1763.9912.33(2.04 - 74.41)17.50(2.65 - 115.66)175.00(28.22 - 1085.2189.00(11.36 - 3143.3191.67(12.02 - 3055.921.00(1.43 - 308.21)240.88(68.75 - 843.9625.00(1.72 - 364.11)262.66(24.39 - 2829.1266.00(33.46 - 2114.732.82(5.43 - 198.37)33.00(8.83 - 123.26)4.33(1.05 - 17.90)429.00(14.65 - 12562.45.00(3.75 - 539.38)46.93(17.48 - 125.96)56.08(4.62 - 680.33)6.10(3.95 - 9.42)6.86(1.81 - 26.01)7.09(1.06 - 47.42)9.00(0.58 - 140.71)98.80(15.24 - 640.50)Pooled Estimate
DO
R_w
ith_C
Is
Adler 1997Avril 1996
Bassa 1996Danforth 2002
Greco 2001Guller 2002
Hubner 2000Inoue 2004
Lin 2002Nakamoto(a) 2002Nakamoto(b) 2002
Noh 1998Ohta 2000
Palmedo 1997Rostom 1999
Scheidhauer 1996Schirrmeister 2001
Smith 1998Tse 1992
Utech 1996Wahl 2004Yang 2001
Yutani 2001Zornoza 2004
van Hoeven 2002OVERALL
0 10 100 1000 10000Odds Ratio
FOREST PLOT: STATA GRAPHICS
![Page 66: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/66.jpg)
Assumes homogeneity of effects across the studies being combined.
There is a common true effect size for all studies.
In the summary estimate, only the variance of each study is taken into account.
FIXED EFFECTS META-ANALYSIS
![Page 67: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/67.jpg)
Summary DOR, Fixed Effects
Odds ratio.1 1 100 1000 10000
Study
Odds ratio (95% CI)
36.27 (4.27,308.02) Adler (1997)
98.80 (10.66,916.11) Avril (1996)
21.00 (0.86,515.50) Bassa (1996)
4.33 (0.80,23.49) Danforth (2002)
107.23 (33.42,344.07) Greco (2001)
12.00 (1.23,117.41) Guller (2002)
429.00 (7.67,23982.81) Hubner (2000)
189.00 (6.63,5384.60) Lin (2002)
17.50 (1.84,166.04) Nakamoto (2002)
6.86 (1.40,33.57) Nakamoto (2002)
208.33 (7.72,5621.57) Noh (1998)
60.23 (3.09,1174.51) Ohta (2000)
106.33 (3.74,3023.90) Palmedo (1997)
289.00 (15.83,5276.04) Rostom (1999)
107.67 (3.85,3013.13) Scheidhauer (1996)
46.93 (14.47,152.17) Schirrmeister (2001)
266.00 (22.50,3145.19) Smith (1998)
9.00 (0.34,238.21) Tse (1992)
262.66 (15.47,4459.70) Utech (1996)
6.10 (3.64,10.24) Wahl (2004)
25.00 (1.03,608.09) Yang (2001)
45.00 (2.33,867.81) Yutani (2001)
240.88 (54.08,1072.94) Zornoza (2004)
12.33 (1.45,104.97) van Hoeven (2002)
19.85 (14.16,27.82) Overall (95% CI)
FIXED EFECTS META-ANALYSIS
![Page 68: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/68.jpg)
Heterogeneity is incorporated into the
pooled estimate by including a between study component of variance.
Assumes sample of studies included in the analysis is drawn from a population of studies.
Each sample of studies has a true effect size.
RANDOM EFFECTS META-ANALYSIS
![Page 69: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/69.jpg)
Summary DOR, Random Effects
Odds ratio.1 1 100 1000 10000
Study
Odds ratio (95% CI)
36.27 (4.27,308.02) Adler (1997)
98.80 (10.66,916.11) Avril (1996)
21.00 (0.86,515.50) Bassa (1996)
4.33 (0.80,23.49) Danforth (2002)
107.23 (33.42,344.07) Greco (2001)
12.00 (1.23,117.41) Guller (2002)
429.00 (7.67,23982.81) Hubner (2000)
189.00 (6.63,5384.60) Lin (2002)
17.50 (1.84,166.04) Nakamoto (2002)
6.86 (1.40,33.57) Nakamoto (2002)
208.33 (7.72,5621.57) Noh (1998)
60.23 (3.09,1174.51) Ohta (2000)
106.33 (3.74,3023.90) Palmedo (1997)
289.00 (15.83,5276.04) Rostom (1999)
107.67 (3.85,3013.13) Scheidhauer (1996)
46.93 (14.47,152.17) Schirrmeister (2001)
266.00 (22.50,3145.19) Smith (1998)
9.00 (0.34,238.21) Tse (1992)
262.66 (15.47,4459.70) Utech (1996)
6.10 (3.64,10.24) Wahl (2004)
25.00 (1.03,608.09) Yang (2001)
45.00 (2.33,867.81) Yutani (2001)
240.88 (54.08,1072.94) Zornoza (2004)
12.33 (1.45,104.97) van Hoeven (2002)
42.54 (20.88,86.68) Overall (95% CI)
RANDOM EFFECTS META-ANALYSIS
![Page 70: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/70.jpg)
Process of “prospectively” performing a new or updated analysis every time another trial is published
Provides answers regarding effectiveness of an intervention at the earliest possible date in time
CUMULATIVE META-ANALYSIS
![Page 71: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/71.jpg)
Studies are sequentially pooled by adding each time one new study according to an ordered variable.
For instance, the year of publication; then, a pooling analysis will be done every time a new article appears.
CUMULATIVE META-ANALYSIS
![Page 72: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/72.jpg)
In theory, the effect of any continuous or ordinal study-related variable can be assessed
Ex: sample size, study quality score, baseline risk etc
CUMULATIVE META-ANALYSIS
![Page 73: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/73.jpg)
metacum LogOR seLogOR, eform id(author) effect(f) graph cline saving(year_fcummplot, replace)
CUMULATIVE META-ANALYSIS: SYNTAX
![Page 74: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/74.jpg)
CUMULATIVE META-ANALYSIS: DIALOG 1
![Page 75: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/75.jpg)
CUMULATIVE META-ANALYSIS: DIALOG 2
![Page 76: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/76.jpg)
LogOR7.67387 23982.8
Wahl 2004
Yang 2001
Yutani 2001
Zornoza 2004
Inoue 2004
Nakamoto(b) 2002
Nakamoto(a) 2002
van Hoeven 2002
Adler 1997
Rostom 1999
Avril 1996
Guller 2002
Schirrmeister 2001
Smith 1998
Utech 1996
Greco 2001
Ohta 2000
Hubner 2000
Bassa 1996
Tse 1992
Noh 1998
Scheidhauer 1996
Palmedo 1997
Lin 2002
Danforth 2002
CUMULATIVE META-ANALYSIS: PLOT
![Page 77: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/77.jpg)
Studies are pooled according to influence of a trial on overall effect defined as the difference between the effect estimated with and without the trial
INFLUENCE ANALYSIS
![Page 78: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/78.jpg)
metaninf tp fn fp tn, id(author) saving(influplot, replace) save(infcoeff, replace)
INFLUENCE ANALYSIS: STATA SYNTAX
![Page 79: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/79.jpg)
INFLUENCE ANALYSIS: STATA DIALOG
![Page 80: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/80.jpg)
INFLUENCE ANALYSIS: STATA DIALOG
![Page 81: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/81.jpg)
INFLUENCE ANALYSIS: PLOT
4.75 6.53 5.41 7.88 10.30
1 2 3 4 5 6 7 8 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Lower CI Limit Estimate Upper CI Limit Meta-analysis estimates, given named study is omitted
![Page 82: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/82.jpg)
INFLUENCE ANALYSIS: STATA DIALOG
![Page 83: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/83.jpg)
A scatter plot of true positive fraction (sensitivity) vs. false positive fraction (1-specificity)
Aids in visualization of range of results from primary studies
ROC PLANE
![Page 84: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/84.jpg)
twoway (scatter TPF FPF, sort ) (lfit uTPR FPF, sort range(0 1) clcolor(black) clpat(dash) clwidth(vthin) connect(direct)) (lfit sTPR FPF, sort range(0 1) clcolor(black) clpat(dot) clwidth(vthin) connect(direct)), ytitle(Sensitivity) ylabel(0(.1)1, grid) xtitle(1-Specificity) xlabel(0(.1)1, grid) title(ROC Plot of SENSITIVITY vs. 1-SPECIFICITY, size(medium)) legend(pos(3) col(1) lab(1 "Observed Data") lab(2 "Uninformative Test") lab(3 "Symmetry Line")) saving(ROCplot, replace) plotregion(margin(zero))
ROC PLANE: STATA SYNTAX
![Page 85: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/85.jpg)
0.1
.2.3
.4.5
.6.7
.8.9
1S
en
sitiv
ity
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 11-Specificity
Observed Data
Uninformative TestSymmetry Line
ROC Plot of SENSITIVITY vs. 1-SPECIFICITY
ROC PLANE: PLOT
![Page 86: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/86.jpg)
ORDINARY LEAST SQUARES METHOD:
Studies are weighted equally
WEIGHTED LEAST SQUARES METHOD:
Weighted by the inverse variance weights of the
odds ratio, or simply the sample size
ROBUST-RESISTANT METHOD:
Minimizes the influence of outliers
SROC: LINEAR REGRESSION MODELS
![Page 87: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/87.jpg)
Logit transformations of the TP rate (sensitivity)
and FP rate (1 - specificity).
D=ln(DOR) =logit(TPR) – logit(FPR)
Differences in logit transformations, D, regressed
on sums of logit transformations, S.
S=logit(TPR)+logit(FPR)
Logit(TPR)=natural log odds of a TP result and
logit(FPR) =natural log of the odds of a FP test
result.
SROC: LOGIT TRANSFORMATION
![Page 88: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/88.jpg)
twoway (scatter D S, sort msymbol(circle)) (lfit tfitted S, clcolor(black) clpat(solid) clwidth(thin) connect(direct))(lfit wfitted S, clcolor(black) clpat(dash) clwidth(thin) connect(direct)), ytitle(Discriminatory Power/D) xtitle(Diagnostic Threshold/S) title(REGRESSION PLOT) legend(lab(1 "Observed Data")lab(2 "EWLSR")lab(3 "VWLSR"))saving(regplot, replace) xline(0) yscale(noline)
ACCURACY/THRESHOLD PLOT
![Page 89: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/89.jpg)
12
34
56
Dis
crim
inato
ry P
ow
er/
D
-4 -2 0 2 4Diagnostic Threshold/S
Observed Data EWLSRVWLSR
REGRESSION PLOT
ACCURACY/THRESHOLD PLOT
![Page 90: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/90.jpg)
Back transformation of logistic regression
to conventional axes of sensitivity [TPR]
vs. (1 – specificity) [FPR]) with the equation
TPR = 1/{1 + exp[- a/(1 - b )]} [(1 -
FPR)/(FPR)](1 + b )/(1 - b ).
Slope b and intercept a are obtained from
the linear regression analyses
SUMMARY ROC CURVE
![Page 91: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/91.jpg)
twoway (scatter TPF FPF, sort msymbol(circle) msize(medium) mcolor(black))(fpfit tTPR FPF, clpat(dash)clwidth(medium) connect(direct ))(fpfit wTPR FPF, clpat(solid)clwidth(medium) connect(direct ))(lfit uTPR FPF, sort range(0 1) clcolor(black) clpat(dash) clwidth(thin) connect(direct)) (lfit sTPR FPF, sort range(0 1) clcolor(black) clpat(dot) clwidth(medium) connect(direct)), ytitle(Sensitivity/TPF) yscale(range(0 1)) ylabel( 0(.2)1,grid ) xtitle(1-Specificity/FPF) xscale(range(0 1)) xlabel(0(.2)1, grid) legend(lab(1 "Observed Data")lab(2 "EWLSR")lab(3 "VWLSR")lab(4 "RRLSR")lab(5 "Uninformative Test") lab(6 "Symmetry Line") pos(3) col(1)) title(SUMMARY ROC CURVES) graphregion(margin(zero)) saving(aSROCplot, replace)
SUMMARY ROC CURVE: STATA SYNTAX
![Page 92: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/92.jpg)
0.2
.4.6
.81
Sen
sitiv
ity/T
PF
0 .2 .4 .6 .8 11-Specificity/FPF
Observed Data
EWLSRVWLSRRRLSRUninformative Test
SUMMARY ROC CURVES
SUMMARY ROC CURVE: EXAMPLE
![Page 93: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/93.jpg)
SUMMARY ROC : SUBGROUP ANALYSIS
![Page 94: PowerPoint Presentation](https://reader035.vdocuments.net/reader035/viewer/2022062613/54555f09af795974408b90bc/html5/thumbnails/94.jpg)
STATA 8.2 (Stata Corp, College Station, Texas,
USA)
SOFTWARE