Table 3. Overview of Methodological Study Biases.
| Estimate | Types of methodological study biases | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Bias from handling of potential confounders | Diagnostic bias | Competing risks, estimated cumulative risks | Survival bias and related biases | ||||||
| Missing adjustment for age, sex, or educational level | Adjusted for factors influenced by cancer | Cognitively impaired individuals not excluded at baseline | Cancer status might influence AD diagnosis | Prevalent cancers not separated from incident cancers | Cancer type that raises subsequent mortality risk | High % of missing data | Restrictive inclusion and exclusion criteria | ||
| Studies with bias, No. | 12 | 4 | 5 | 15 | 6 | 6 | 20 | 3 | 1 |
| Metaregression estimatesa | |||||||||
| Pooled ln HR (95% CI) in studies without the bias | −0.15 (−0.34 to 0.04) | −0.15 (−0.28 to −0.02) | −0.09 (−0.22 to 0.03) | −0.32 (−0.54 to −0.10) | −0.13 (−0.26 to 0.00) | −0.09 (−0.20 to 0.02) | −0.19 (−0.57 to 0.20) | −0.10 (−0.22 to 0.01) | −0.12 (−0.24 to 0.00) |
| Difference in ln HR (95% CI) for studies with the bias | 0.04 (−0.20 to 0.29) | 0.13 (−0.13 to 0.39) | −0.14 (−0.45 to 0.16) | 0.26 (0.01 to 0.52) | 0.09 (−0.32 to 0.50) | −0.34 (−0.71 to 0.03) | 0.07 (−0.33 to 0.48) | −0.46 (−1.13 to 0.22) | −0.01 (−0.92 to 0.90) |
| R2, % | 1.6 | 32.4 | 22.1 | 16.7 | 6.7 | 21.1 | 5.5 | 16.3 | 6.2 |
Abbreviations: AD, Alzheimer disease; ln HR, natural log hazard ratio.
a
Metaregressions adjusted for study design (case-control vs cohort) as a covariate.