Table 3.

Summary of meta-analytic results

Exposure categoriesStudy design
ExposureHighest (min to max)Lowesta (min to max)CohortsCase–control studiesRCTPooled RR for gastric cancer (95% CI)PQI2 (%)Outlier studiesbPEgger's
Years of fertility>35 to >39<27 to <344400.74 (0.63–0.86)0.670None0.29
Age at menarche (y)>14 to >16<13 to <166501.0 (0.85–1.19)0.0350.4Frise and colleagues (35) and Persson and colleagues (38).0.25
Age at menopause (y)>50 to >55<45 to 50–546400.84 (0.67–1.05)0.1828.9None0.41
Parity (number of children)>2 to >40 to 1–275c00.94 (0.74–1.19)<0.00183.4La Vecchia and colleagues (26), Chung and colleagues (42), and Chang and colleagues (43).0.19
Age at first birth (y)>25 to >34<20 to 20–306400.99 (0.85–1.15)0.2323.1None0.92
OC useEverNever3301.15 (0.71–1.88)<0.00189.7Dorjgochoo and colleagues (40), Freedman and colleagues (41), and Chung and colleagues (42).0.10
HRTEverNever4300.77 (0.64–0.92)0.990None0.95
Tamoxifen treatmentTreatedUntreated5091.82d (1.39–2.38)0.410N/AN/A

Abbreviations: PQ, p from Q statistics; PEgger's, p from Egger's test; RCT, randomized clinical trials; N/A, nonapplicable.

  • aReference category.

  • bStudies identified based on visual inspection of Galbraith plots.

  • cExcludes the study by Inoue and colleagues (30) which reported a linear trend without stratum-specific estimates.

  • dBased on the marginal RR estimates for cohort studies and RCT.