Skip to main content
. Author manuscript; available in PMC: 2019 Jun 1.
Published in final edited form as: Res Synth Methods. 2018 Mar 24;9(2):261–272. doi: 10.1002/jrsm.1293

Table 3.

The estimated overall log odds ratios (95% CrI) of the 8 risk factors in the pterygium data obtained by Model H (accounting for both between- and within-study correlations), Model M0 (only accounting for between-study correlations), and Model U (univariate model ignoring both between- and within-study correlations) using the Bayesian method.

Risk factor No. of
studies
Estimated overall log odds ratio

Model H Model M0 Model U
(1) Occupation type 23 0.65 (0.40, 0.92) 0.65 (0.41, 0.93) 0.66 (0.42, 0.91)
(2) Smoking status 16 0.10 (−0.07, 0.29) 0.10 (−0.07, 0.29) 0.08 (−0.07, 0.25)
(3) Education attainment 10 0.73 (0.40, 1.10) 0.74 (0.42, 1.07) 0.74 (0.46, 1.06)
(4) Use of hat 6 0.47 (−0.87, 1.62) 0.44 (−0.77, 1.55) 0.32 (−0.84, 1.41)
(5) Use of spectacles 6 −0.59 (−1.02, −0.09) −0.58 (−1.06, −0.06) −0.60 (−1.26, 0.05)
(6) Area of residence 5 1.04 (0.29, 1.78) 1.10 (0.22, 1.97) 1.24 (0.39, 2.11)
(7) Use of sunglasses 5 −0.36 (−1.53, 0.74) −0.51 (−1.64, 0.62) −0.57 (−1.76, 0.63)
(8) Latitude of residence 4 0.92 (−0.74, 2.72) 1.04 (−0.98, 3.10) 1.14 (−0.70, 3.37)