Table 2

Meta-regression of the risk of all-cause dementia according to marital status, stratified by study time period, case ascertainment methodology, study type and study quality

 Widowed versus married Divorced versus married Lifelong single versus married Stratified analysis: Relative risk (95% CI) Number of studies Meta-regression coefficient (95% CI) p value Stratified analysis: Relative risk (95% CI) Number of studies Meta-regression coefficient (95% CI) p-value Stratified analysis: Relative risk (95% CI) Number of studies Meta-regression coefficient (95% CI) p value Method of case ascertainment Clinical assessment 1.20 (1.02 to 1.41) n=9 b=−0.06 (−0.18 to 0.05) p=0.29 0.99 (0.71 to 1.37) n=7 b=0.34 (0.06 to 0.62) p=0.02 1.42 (1.07 to 1.90) n=6 b=−0.27 (−0.55 to 0.01) p=0.06 Clinical registers 1.12 (1.07 to 1.18) n=2 1.11 (0.52 to 2.38) n=2 1.23 (1.17 to 1.29) n=2 Study type Cohort 1.10 (1.05 to 1.28) n=7 b=0.28 (0.09 to 0.46) p=0.004 1.16 (0.87 to 1.55) n=6 b=−0.83 (−1.69 to 0.03) p=0.06 1.24 (1.17 to 1.30) n=5 b=0.08 (−0.45 to 0.62) p=0.76 Case–control/cross-sectional 1.39 (1.16 to 1.67) n=4 0.55 (0.23 to 1.31) n=3 1.21 (0.67 to 2.18) n=3 Global quality score Higher quality ≥6 1.13 (1.02 to 1.31) n=4 b=0.08 (-0.06 to 0.23) p=0.27 1.16 (0.83 to 1.62) n=4 b=−0.40 (-0.88 to 0.08) p=0.10 1.26 (1.09 to 1.45) n=3 b=0.20 (-0.17 to 0.57) p=0.29 Lower quality <6 1.22 (0.96 to 1.54) n=7 0.88 (0.54 to 1.44) n=5 1.33 (0.92 to 1.92) n=5 Increase in quality by one point b=−0.04 (−0.08 to −0.002) p=0.04 n=11 b=0.12 (0.01 to 0.24) p=0.04 n=9 b=−0.05 (−0.13 to 0.03) p=0.21 n=8 Time period Mean DoB before 1927 1.11 (0.93 to 1.31) n=6 b=0.15 (−0.14 to 0.43) p=0.32 0.98 (0.71 to 1.37) n=6 b=0.35 (0.08 to 0.63) p=0.01 1.40 (1.06 to 1.85) n=5 b=−0.22 (−0.50 to 0.06) p=0.13 Mean DoB after 1927 1.23 (1.06 to 1.43) n=5 1.08 (0.50 to 2.35) n=3 1.24 (0.94 to 1.62) n=3 Mean year of birth 10 years later b=0.08 (−0.08 to 0.23) p=0.34 n=11 b=0.24 (0.01 to 0.47) p=0.04 n=9 b=−0.15 (−0.33 to 0.02) p=0.09 n=8
• Figures are based on random-effects meta-analysis.

• DoB, date of birth.