Skip to main content

Table 2 The mean difference of effect estimates and 95% limits of agreement for the meta-analytic immediate level-change (top triangle, difference calculated as column method – row method) and slope-change per month (bottom triangle, difference calculated as row method – column method) (n = 17)

From: Comparison of statistical methods used to meta-analyse results from interrupted time series studies: an empirical study

 

Level-change

Slope-change per month

OLS ITS

Fixed MA

-0.04 (-0.48,0.40)

-0.034 (-0.50,0.43)

0.15 (-0.85,1.15)

-0.01 (-0.65,0.63)

0.00 (-0.66,0.66)

0.01 (-0.05,0.06)

OLS ITS

DL MA

0.00 (-0.04,0.04)

0.19 (-0.90,1.27)

0.03 (-0.30,0.36)

0.04 (-0.29,0.36)

0.01 (-0.08,0.10)

0.00 (-0.04,0.04)

OLS ITS

REML MA

0.19 (-0.92,1.29)

0.03 (-0.31,0.37)

0.03 (-0.29,0.36)

0.02 (-0.06,0.10)

0.01 (-0.07,0.10)

0.01 (-0.10,0.12)

REML ITS

Fixed MA

-0.16 (-1.28,0.96)

-0.15 (-1.27,0.96)

0.04 (-0.11,0.18)

0.03 (-0.08,0.14)

0.03 (-0.06,0.11)

0.01 (-0.10,0.13)

REML ITS

DL MA

0.01 (-0.05,0.07)

0.04 (-0.17,0.26)

0.04 (-0.14,0.21)

0.03 (-0.11,0.18)

0.02 (-0.17,0.21)

0.01 (-0.07,0.09)

REML ITS

REML MA

  1. The top row of the label indicates the ITS analysis methods, bottom row indicates the meta-analysis method, e.g., OLS Fixed is OLS ITS analysis and fixed-effect meta-analysis. For example, the mean meta-analytic level-change yielded by REML ITS analysis with fixed-effect meta-analysis was 0.15 higher than that yielded by OLS ITS analysis with fixed-effect meta-analysis (column 4, row 1). The mean meta-analytic slope-change per month yielded by REML ITS analysis with fixed-effect meta-analysis was 0.02 higher than that yielded by OLS ITS analysis with fixed-effect meta-analysis (column 1, row 4)
  2. DL DerSimonian and Laird, ITS interrupted time series, MA meta-analysis, OLS ordinary least squares, REML restricted maximum likelihood
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us