HR (95% CI) for self-harm during 6 months’ follow-up in users of varenicline versus NRT, adjusted for an unmeasured binary confounder with a HR of 3
| P0/ P1 | 0.0 | 0.1 | 0.2 | 0.3 | 0.4 | 0.5 | 0.6 | 0.7 | 0.8 | 0.9 | 1.0 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.0 | 0.78 (0.37 to 1.63) | 0.94 (0.44 to 1.96) | 1.09 (0.52 to 2.28) | 1.25 (0.59 to 2.61) | 1.4 (0.67 to 2.93) | 1.56 (0.74 to 3.26) | 1.72 (0.81 to 3.59) | 1.87 (0.89 to 3.91) | 2.03 (0.96 to 4.24) | 2.18 (1.04 to 4.56) | 2.34 (1.11 to 4.89) |
| 0.1 | 0.65 (0.31 to 1.36) | 0.78 (0.37 to 1.63) | 0.91 (0.43 to 1.9) | 1.04 (0.49 to 2.17) | 1.17 (0.56 to 2.45) | 1.3 (0.62 to 2.72) | 1.43 (0.68 to 2.99) | 1.56 (0.74 to 3.26) | 1.69 (0.8 to 3.53) | 1.82 (0.86 to 3.8) | 1.95 (0.93 to 4.08) |
| 0.2 | 0.56 (0.26 to 1.16) | 0.67 (0.32 to 1.4) | 0.78 (0.37 to 1.63) | 0.89 (0.42 to 1.86) | 1 (0.48 to 2.1) | 1.11 (0.53 to 2.33) | 1.23 (0.58 to 2.56) | 1.34 (0.63 to 2.79) | 1.45 (0.69 to 3.03) | 1.56 (0.74 to 3.26) | 1.67 (0.79 to 3.49) |
| 0.3 | 0.49 (0.23 to 1.02) | 0.59 (0.28 to 1.22) | 0.68 (0.32 to 1.43) | 0.78 (0.37 to 1.63) | 0.88 (0.42 to 1.83) | 0.98 (0.46 to 2.04) | 1.07 (0.51 to 2.24) | 1.17 (0.56 to 2.45) | 1.27 (0.6 to 2.65) | 1.37 (0.65 to 2.85) | 1.46 (0.69 to 3.06) |
| 0.4 | 0.43 (0.21 to 0.91) | 0.52 (0.25 to 1.09) | 0.61 (0.29 to 1.27) | 0.69 (0.33 to 1.45) | 0.78 (0.37 to 1.63) | 0.87 (0.41 to 1.81) | 0.95 (0.45 to 1.99) | 1.04 (0.49 to 2.17) | 1.13 (0.53 to 2.35) | 1.21 (0.58 to 2.54) | 1.3 (0.62 to 2.72) |
| 0.5 | 0.39 (0.19 to 0.82) | 0.47 (0.22 to 0.98) | 0.55 (0.26 to 1.14) | 0.62 (0.3 to 1.3) | 0.7 (0.33 to 1.47) | 0.78 (0.37 to 1.63) | 0.86 (0.41 to 1.79) | 0.94 (0.44 to 1.96) | 1.01 (0.48 to 2.12) | 1.09 (0.52 to 2.28) | 1.17 (0.56 to 2.45) |
| 0.6 | 0.35 (0.17 to 0.74) | 0.43 (0.2 to 0.89) | 0.5 (0.24 to 1.04) | 0.57 (0.27 to 1.19) | 0.64 (0.3 to 1.33) | 0.71 (0.34 to 1.48) | 0.78 (0.37 to 1.63) | 0.85 (0.4 to 1.78) | 0.92 (0.44 to 1.93) | 0.99 (0.47 to 2.07) | 1.06 (0.5 to 2.22) |
| 0.7 | 0.33 (0.15 to 0.68) | 0.39 (0.19 to 0.82) | 0.46 (0.22 to 0.95) | 0.52 (0.25 to 1.09) | 0.59 (0.28 to 1.22) | 0.65 (0.31 to 1.36) | 0.72 (0.34 to 1.49) | 0.78 (0.37 to 1.63) | 0.85 (0.4 to 1.77) | 0.91 (0.43 to 1.9) | 0.98 (0.46 to 2.04) |
| 0.8 | 0.3 (0.14 to 0.63) | 0.36 (0.17 to 0.75) | 0.42 (0.2 to 0.88) | 0.48 (0.23 to 1) | 0.54 (0.26 to 1.13) | 0.6 (0.28 to 1.25) | 0.66 (0.31 to 1.38) | 0.72 (0.34 to 1.5) | 0.78 (0.37 to 1.63) | 0.84 (0.4 to 1.76) | 0.9 (0.43 to 1.88) |
| 0.9 | 0.28 (0.13 to 0.58) | 0.33 (0.16 to 0.7) | 0.39 (0.19 to 0.82) | 0.45 (0.21 to 0.93) | 0.5 (0.24 to 1.05) | 0.56 (0.26 to 1.16) | 0.61 (0.29 to 1.28) | 0.67 (0.32 to 1.4) | 0.72 (0.34 to 1.51) | 0.78 (0.37 to 1.63) | 0.84 (0.4 to 1.75) |
| 1.0 | 0.26 (0.12 to 0.54) | 0.31 (0.15 to 0.65) | 0.36 (0.17 to 0.76) | 0.42 (0.2 to 0.87) | 0.47 (0.22 to 0.98) | 0.52 (0.25 to 1.09) | 0.57 (0.27 to 1.2) | 0.62 (0.3 to 1.3) | 0.68 (0.32 to 1.41) | 0.73 (0.35 to 1.52) | 0.78 (0.37 to 1.63) |
This tables shows how the observed HR (diagonal line of cells with dotted border) would change in the presence of an unmeasured confounder with a HR of three and different combinations of prevalence rates among the user groups. P1/P0=prevalence of the unmeasured confounder among users of varenicline (P1) and NRT (P0). The cells filled with grey mark the situations in which varenicline would be associated with a statistically significant increased hazard of the event. These calculations are based on: Lin et al. Biometrics 1998;54:948–63 (equation 2.9).
NRT, nicotine replacement therapy.