Table 2.
Subgroup analysis for studies included in the analysis.
| Subgroup analysis | Pooled OR (95% CI), P-value for the heterogeneity Q test, I2 statistics (%), number of estimates in included studies (n) |
|
|---|---|---|
| n | Risk estimates of cognitive impairment and dementia | |
| Mean age | ||
| ≥65 years | 12 | 0.80 (0.71–0.91); I2 = 54.7, P = 0.012 |
| < years | 3 | 0.79 (0.51–1.21); I2 = 70.8, P = 0.032 |
| Sex | ||
| Female | 4 | 0.82 (0.64–1.06); I2 = 71.9, P = 0.014 |
| Male | 2 | 1.18 (0.85–1.63); I2 = 0.0, P = 0.571 |
| Combined | 9 | 0.71 (0.66–0.78); I2 = 0.0, P = 0.567 |
| Geographic location | ||
| Europe | 7 | 0.84 (0.70–0.99); I2 = 60.8, P = 0.018 |
| United States | 2 | 0.78 (0.58–1.04); I2 = 21.9, P = 0.258 |
| China | 6 | 0.71 (0.64–0.78); I2 = 38.0, P = 0.153 |
| Study design | ||
| Cohort | 9 | 0.80 (0.69–0.92); I2 = 67.5, P = 0.002 |
| Cross-sectional | 6 | 0.80 (0.67–0.95); I2 = 23.5, P = 0.257 |
| Study quality scorea | ||
| Full marks | 7 | 0.91 (0.81–1.02); I2 = 28.1, P = 0.214 |
| Not full marks | 8 | 0.71 (0.65–0.77); I2 = 40.9, P = 0.106 |
| Disease type | ||
| Cognitive impairment | 10 | 0.82 (0.72–0.93); I2 = 57.8, P = 0.011 |
| Dementia | 5 | 0.73 (0.54–0.83); I2 = 56.9, P = 0.054 |
| Dietary assessment method | ||
| FFQ | 3 | 0.79 (0.65–0.96); I2 = 30.2, P = 0.239 |
| Others | 12 | 0.79 (0.69–0.91); I2 = 61.2, P = 0.003 |
OR, odds ratio; CI, confidence interval; FFQ, food frequency questionnaire.
aFull mark was defined as nine points for cohort studies and five points for cross-sectional studies based on the Newcastle–Ottawa Scale.