Skip to main content

Table 2 Multivariable regression analysis of potential factors for reporting quality

From: Improving the quality of reporting of systematic reviews of dose-response meta-analyses: a cross-sectional survey

Influential factors

Estimated regression coefficients (95%CI)

Multivariable

P-value

Sensitivity analysis

P-value

No. of authors

  ≤ 4

Reference

 

Reference

 

 5 ~  6

0.78 (0.35, 1.20)

<  0.001

0.73 (0.28, 1.18)

0.002

 7~  8

0.86 (0.37, 1.36)

0.001

0.68 (0.18, 1.19)

0.008

  > 8

1.15 (0.61, 1.70)

<  0.001

0.99 (0.48, 1.49)

<  0.001

Year of publication

 2011

Reference

 

Reference

 

 2012

0.39 (−0.41, 1.20)

0.338

0.77 (− 0.05, 1.58)

0.066

 2013

1.23 (0.28, 2.18)

0.011

1.12 (0.18, 2.05)

0.020

 2014

0.93 (0.13, 1.74)

0.023

1.08 (0.20, 1.96)

0.016

 2015

1.35 (0.58, 2.11)

0.001

1.39 (0.54, 2.25)

0.001

 2016

2.01 (1.28, 2.75)

<  0.001

2.19 (1.38, 3.00)

<  0.001

 2017 (up to July-31)

2.39 (1.60, 3.18)

<  0.001

2.56 (1.62, 3.52)

<  0.001

 Linear trend test

0.38 (0.28, 0.47)

<  0.001

0.38 (0.27, 0.50)

<  0.001

Use of reporting guidance

 No

Reference

 

Reference

 

 Yes

0.98 (0.63, 1.32)

<  0.001

0.99 (0.61, 1.37)

<  0.001

Region

 European

Reference

 

Reference

 

 Asia Pacific

−0.21 (− 0.66, 0.23)

0.348

− 0.28 (− 0.74, 0.17)

0.224

 America

− 0.18 (−1.31, 0.95)

0.752

− 0.54 (−1.61, 0.53)

0.320

Methodologist involved

 No

Reference

 

Reference

 

 Yes

0.86 (0.42, 1.32)

<  0.001

0.78 (0.36, 1.19)

<  0.001

  1. The multivariable regression was based on weighted least square linear regression; the sensitivity analysis was based on generalized estimating equation (GEE); both the two methods with the variance estimation based on robust standard error