Separate meta-analyses by study characteristics estimating relative risk reduction (RRR) in tobacco use from reduction of tobacco retailer density and proximity, k=285, n=27
| Characteristics | n | k | I2 | Estimated RRR (95% CI) |
| Outcome variable | ||||
| All behavioural outcomes | 26 | 255 | 99.82 | 2.48 (1.95 to 3.02) |
| Smoking | 10 | 68 | 99.92 | 2.06 (1.26 to 2.85) |
| Quitting (reverse coded) | 15 | 125 | 99.75 | 6.50 (4.64 to 8.37) |
| Relapse | 3 | 12 | 0 | −0.04 (−0.24 to 0.17) |
| Young adult initiation | 3 | 19 | 0 | −0.25 (−1.09 to 0.60) |
| Other (eg, stage of change) | 5 | 31 | 99.72 | 3.28 (1.54 to 5.02) |
| Health outcomes | 2 | 30 | 41.84 | 0.01 (−0.01 to 0.03) |
| Predictor variable* | ||||
| Proximity (reverse coded) | 14 | 100 | 87.88 | 2.38 (1.39 to 3.37) |
| Density—all types | 24 | 155 | 99.89 | 2.55 (1.91 to 3.19) |
| Count | 8 | 53 | 99.99 | 5.29 (2.49 to 8.10) |
| Per land area | 10 | 45 | 99.94 | 3.30 (2.19 to 4.42) |
| Per persons | 1 | 6 | 0 | 0.07 (−0.88 to 1.03) |
| Other (eg, proportion of smaller area units within area unit with retailers) | 6 | 51 | 74.85 | 3.84 (2.24 to 5.44) |
| Design and statistics* | ||||
| Use of activity space | 2 | 12 | 33.01 | 19.85 (12.44 to 27.26) |
| Longitudinal | 10 | 96 | 99.80 | 4.91 (3.03 to 6.79) |
| Cross-sectional | 18 | 159 | 99.83 | 2.22 (1.65 to 2.79) |
| Adjustment for confounders or covariates | ||||
| Adjusted: no controls (unadjusted) | 10 | 73 | 0 | −0.01 (−0.03 to 0.01) |
| Adjusted: individual characteristics only | 13 | 89 | 99.81 | 4.04 (2.95 to 5.14) |
| Adjusted: neighbourhood characteristics only | 2 | 2 | – | – |
| Adjusted: both individual and neighbourhood characteristics | 14 | 91 | 96.45 | 2.96 (1.70 to 4.23) |
| No delinearisation of predictor | 15 | 81 | 3.56 | −0.0003 (−0.02 to 0.02) |
| Used delinearisation of predictor | 16 | 174 | 77.89 | 4.33 (3.32 to 5.33) |
| Used ordinal or nominal outcome variable | 3 | 24 | 0 | 0.15 (−0.25 to 0.56) |
| No ordinal or nominal outcome variable | 23 | 231 | 99.87 | 2.87 (2.25 to 3.50) |
| Results with men and women combined | 24 | 215 | 99.77 | 2.04 (1.57 to 2.50) |
| Women-specific effect sizes | 3 | 22 | 68.99 | 0.17 (−4.25 to 4.58) |
| Men-specific effect sizes | 2 | 18 | 53.65 | 15.30 (8.28 to 22.31) |
| Results with incomes combined | 26 | 244 | 99.83 | 2.59 (2.04 to 3.14) |
| Low income-specific effect sizes | 2 | 5 | 36.72 | 4.31 (0.19 to 8.43) |
| High income-specific effect sizes | 2 | 6 | 32.04 | −1.01 (−2.48 to 0.46) |
| Aggregated tobacco use intensity | 25 | 223 | 99.81 | 2.39 (1.88 to 2.90) |
| Low intensity | 1 | 16 | 28.63 | −2.61 (−6.94 to 1.72) |
| High intensity | 1 | 16 | 73.22 | 7.84 (−1.09 to 16.77) |
| Context* | ||||
| Country | ||||
| Australia | 2 | 16 | 37.79 | 0.001 (−0.03 to 0.04) |
| Canada | 10 | 104 | 99.14 | 3.10 (2.06 to 4.14) |
| Finland | 2 | 54 | 91.84 | 5.84 (2.06 to 9.62) |
| New Zealand | 1 | 12 | 16.39 | 1.28 (0.15 to 2.40) |
| UK | 4 | 22 | 91.60 | 4.48 (2.97 to 5.98) |
| USA | 8 | 47 | 99.82 | 1.61 (0.72 to 2.49) |
| Retail point-of-sale advertising† | ||||
| Without retail advertising restrictions | 14 | 137 | 99.93 | 2.86 (2.04 to 3.69) |
| With retail advertising restrictions | 12 | 118 | 86.00 | 0.86 (0.49 to 1.24) |
| Retail display of tobacco products‡ | ||||
| Without retail display restrictions | 17 | 155 | 99.87 | 2.32 (1.68 to 2.96) |
| With retail display restrictions | 9 | 100 | 99.41 | 3.80 (2.60 to 5.00) |
Each row represents a separate meta-analysis of the effect sizes in that row; n is the number of studies include and k is the number of effect sizes included. I2 is a measure of heterogeneity. Bold indicates a statistically significant estimate at the historical p<0.05 threshold. We suppressed results with fewer than five effect sizes.
*Includes only behavioural effect sizes (k=255).
†Presence of an advertising policy includes results from Australia, Canada and New Zealand.
‡Presence of a display policy includes only results from Canada.
NR, not reported.