Table 4.
Secondary and sensitivity analyses for pooled analysis of the association between propranolol and non-selective beta-blocker use and breast cancer-specific and all-cause mortality
| Medication usage | Deaths | Patients | Person-years | Unadjusted | Fully adjusted | ||||
|---|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P | Hetero I 2 (P) | Adjusted HR (95% CI) | P | Hetero I 2 (P) | ||||
| Breast cancer-specific mortality | |||||||||
| Medication use after diagnosis | |||||||||
| Main time-varying covariate analysis in stage 1–3 breast cancer patients | |||||||||
| Propranolol in stages 1–3 | 3389 | 44,376 | 112,450 | 1.02 (0.82, 1.26) | 0.88 | 0% (0.84) | 1.06 (0.85, 1.33) | 0.62 | 0% (0.60) |
| Main time-varying covariate analysis in all breast cancer patients | |||||||||
| Any beta-blocker | 5419 | 55,252 | 181,714 | 1.25 (1.11, 1.40) | <0.001 | 63% (0.03) | 1.07 (0.99, 1.16) | 0.10 | 0% (0.82) |
| Analysis based upon use in year after diagnosisa | |||||||||
| Propranolol | 5426 | 55,252 | 181,959 | 0.94 (0.72, 1.21) | 0.61 | 35% (0.19) | 1.07 (0.72, 1.60) | 0.72 | 65% (0.02) |
| Non-selective beta-blocker | 5426 | 55,252 | 181,959 | 1.10 (0.87, 1.39) | 0.43 | 51% (0.09) | 1.15 (0.85, 1.56) | 0.35 | 60% (0.04) |
| Medication use before diagnosisb | |||||||||
| Propranolol | 6883 | 53,870 | 215,978 | 0.97 (0.82, 1.15) | 0.73 | 0% (0.51) | 1.03 (0.86, 1.22) | 0.78 | 3% (0.39) |
| Non-selective beta-blocker | 6883 | 53,870 | 215,978 | 1.09 (0.95, 1.25) | 0.22 | 0% (0.47) | 1.05 (0.92, 1.21) | 0.45 | 0% (0.68) |
| All-cause mortality | |||||||||
| Medication use after diagnosis | |||||||||
| Main time-varying covariate analysis in stage 1–3 breast cancer patients | |||||||||
| Propranolol in stages 1–3c | 17,219 | 96,097 | 382,1512 | 1.07 (0.93, 1.24) | 0.32 | 29% (0.21) | 1.13 (1.02, 1.24) | 0.02 | 0% (0.46) |
| Main time-varying covariate analysis in all breast cancer patients | |||||||||
| Any beta-blocker | 25,472 | 133,251 | 570,968 | 1.57 (1.41, 1.75) | <0.001 | 92% (<0.001) | 1.12 (1.05, 1.20) | <0.001 | 65% (0.006) |
| Analysis based upon use in year after diagnosisa | |||||||||
| Propranolol | 25,487 | 133,251 | 571,213 | 1.02 (0.89, 1.16) | 0.82 | 48% (0.06) | 1.04 (0.89, 1.21) | 0.62 | 48% (0.06) |
| Non-selective beta-blocker | 25,487 | 133,251 | 571,213 | 1.35 (1.17, 1.55) | <0.001 | 78% (<0.001) | 1.14 (0.99, 1.30) | 0.06 | 68% (0.003) |
| Medication use before diagnosisb | |||||||||
| Propranolol | 31,556 | 139,760 | 664,448 | 0.97 (0.86, 1.09) | 0.60 | 44% (0.09) | 1.02 (0.94, 1.10) | 0.68 | 0% (0.54) |
| Non-selective beta-blocker | 31,556 | 139,760 | 664,448 | 1.30 (1.14, 1.49) | <0.001 | 80% (<0.001) | 1.13 (1.06, 1.21) | <0.001 | 27% (0.21) |
aSimplified analysis, not requiring time-varying covariate use, comparing medication users with non-users in the first year after diagnosis in individuals living more than 1 year after cancer diagnosis; fully adjusted column adjusted for age, year, stage and all confounders presented in Table 1
bBased on use in the year prior to diagnosis, restricted to individuals with 1 year of records prior to diagnosis; fully adjusted column only adjusted for age at diagnosis and year of diagnosis
cExcludes the Belgian cohort
CI confidence interval, HR hazard ratio