Skip to main content

Table 1 Overview of the applied methods to investigate publication bias

From: Investigation of publication bias in meta-analyses of diagnostic test accuracy: a meta-epidemiological study

Reference

Funnel plot

Results of the funnel plot

Test

Results of the test

Remarks

 

x-axis

y-axis

    

Chang 2011 [23]

-

-

-

Egger

3/7

 

Chang 2012 [24]

Sensitivity Specificity

SE

Not considered

Begg Egger

1/2 1/2

 

Cheng 2012 [25]

lnDOR

1/root(ESS)

No publication bias

Not specified

0/2

 

Descatha 2012 [26]

lnDOR

1/root(ESS)

No publication bias

Deeks

0/2

 

Dong 2011 [27]

-

-

-

Begg Egger

0/1 0/1

Results for a second diagnostic tool were not presented.

Dym 2011 [28]

Sensitivity Specificity

1/SE

Inconclusive 2/2

-

-

 

Gao 2011 [29]

lnDOR

SE(lnDOR)

1/2

Begg

1/2

 

Gargiulo 2011 [30]

lnDOR

1/root(ESS)

Not considered

Deeks

1/2

 

Glasgow 2012 [31]

lnDOR

1/Var(lnDOR)

0/2

-

-

 

Gong 2011 [32]

Sensitivity

Sample size

Inconclusive 2/2

-

-

Plots had too low power.

Hernaez 2011 [33]

-

-

-

Deeks

0/1

 

Inaba 2012 [34]

lnDOR RR1

SE(lnDOR) SE(RR)

1/2

Egger2

1/2

Level of significance p-value <0.10

Kobayashi 2012 [35]

DOR

SE(DOR)

2/2

Begg

0/2

Both plots indicated publication though the tests were not significant.

Li 2011 [36]

-

-

-

Egger

1/1

Publication bias was detected for a subgroup by the test.

Li 2012 [37]

-

-

-

Egger

1/1

 

Lu 2011 [38]

lnDOR

1/root(ESS)

Not considered

Deeks

0/1

 

Lundstrom 2011 [39]

-

-

-

Egger

0/1

 

Luo 2011 [40]

lnDOR

1/root (ESS)

Not considered

Egger

0/3

 

Manea 2012 [41]

-

-

?

Begg

?

Results were not presented

Mao 2012 [42]

-

-

-

Egger

1/1

 

Marton 2012 [43]

Not specified

Not specified

Not considered

Egger

1

One plot and test to investigate two diagnostic tools

Mathews 2011 [44]

AUC(ROC)3

SE(AUC(ROC))

0/2

Egger

0/2

 

McInnes 2011 [45]

lnDOR

SE(lnDOR)

-

Egger

0/1

 

Meader 2011 [46]

-

-

-

Egger

?

Results were not presented.

Mitchell 2011 [47]

-

-

-

Begg

?

Results were not presented.

Onishi 2012 [48]

-

-

-

Egger

2/2

 

Papathanasiou 2012 [49]

lnDOR

SE(lnDOR)

Not considered

Begg

1/1

 

Plana 2012 [50]

lnDOR

1/root(ESS)

Not considered

Deeks

0/2

Not identified by tests Plots was not used to draw conclusions.

Qu 2011 [51]

logDOR

Sample size

?/2

-

-

Results of funnel plots were inconclusive, too low power.

Sadeghi 2012 [52]

logDetectionRate4 logSensitivity

SE(logDetect Rate) SE(logSens)

0/2

Egger

0/2

 

Sadigh 2011 [53]

-

-

-

Deeks

0/1

 

Summah 2011 [54]

lnDOR

SE(lnDOR)

1/1

Egger

1/1

 

Sun 2011 [55]

-

-

-

Deeks

0/1

No publication bias was detected by the test.

Takakuwa 2011 [56]

lnDOR

1/root (ESS)

1/1

Deeks

0/1

Identified by plot though not by test.

Thosani 2012 [57]

lnDOR

SE(lnDOR)

Not considered

Egger

2/2

Plots were not used to draw conclusion.

Tomasson 2012 [58]

Difference in arcsine5

Precision(Dif. in arcsine)

2/2

Egger

0/2

Identified by plots though not by tests.

Trallero-Araguas 2012 [59]

-

-

-

Deeks

0/1

 

Wang 2011 [60]

-

-

-

Begg Egger

0/2 0/2

 

Wang 2012 [61]

lnDOR

SE(lnDOR)

7/7

Egger

3/7

 

Wang 2012 [62]

lnDOR

SE(lnDOR)

0/2

Begg Egger

0/2

 

Wang 2012 [63]

lnDOR

SE(lnDOR)

0/2

-

-

 

Wu 2012 [64]

lnDOR

1/root(ESS)

0/1

Deeks

0/1

 

Xu 2011 [65]

-

-

-

Egger

0/1

 

Xu 2011 [66]

lnDOR Standardized effect6

SE(lnDOR) Precision(St. effect)

0/2

Begg-Mazumdar Harbord-Egger

0/2

 

Ying 2011[67]

lnDOR

1/root(ESS)

0/2

Deeks

0/2

 

Yu 2012 [68]

lnDOR

SE(lnDOR)

1/1

-

-

 

Zhang 2011[69]

lnDOR

1/root(ESS)

0/1

Deeks

0/1

 
  1. 1RR = Relative Risk; It is unclear which estimates were used to calculate the RR.
  2. 2The methods section specifies that the Egger test has been used though the text of the figures specified the Begg test.
  3. 3AUC(ROC) = Area Under the Curve (AUC) of the Receiving Operating Characteristic (ROC).
  4. 4There was no definition for Detection Rate specified in the article.
  5. 5Difference in arcsine = Transformed ratios of arcsine for those with rise in Anti-Neutrophil Cytoplasmic Antibody (ANCA) and persistent ANCA among subjects who had relapse and those who did not.
  6. 6Standardized effect was explained as differentiating benign and malignant lymph nodes.
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us