Table 8. Observed outcomes by treatment arm and by benefit/harm subgroup for the SPRINT trial (derivation cohort) and ACCORD-BP trial (validation cohort) when applying models fit by traditional backwards selection.
| Cohort | Benefit/harm subgroup | Number of patients | Number of events (%) | Expected absolute risk difference (95% CI) | Observed absolute risk difference, (95% CI) P value | |||
|---|---|---|---|---|---|---|---|---|
| Intensive treatment | Standard treatment | All patients | Intensive treatment | Standard treatment | ||||
| SPRINT | CVD events/deaths | |||||||
| 1 (lowest predicted benefit) | 990 | 982 | 76 (3.9) | 38 (3.8) | 38 (3.9) | 0.000 (−0.010, 0.009) | 0.000 (−0.017, 0.017), P = 0.971 | |
| 2 (middle predicted benefit) | 2,188 | 2,184 | 193 (4.4) | 76 (3.5) | 117 (5.4) | −0.019 (−0.029, −0.010) | −0.019 (−0.031, −0.007), P = 0.002 | |
| 3 (highest predicted benefit) | 1,181 | 1,139 | 211 (9.1) | 86 (7.3) | 125 (11.0) | −0.051 (−0.112, −0.030) | −0.037 (−0.060, −0.013), P = 0.002 | |
| Serious adverse events | ||||||||
| 1 (lowest predicted harm) | 23 | 14 | 4 (10.8) | 3 (13.0) | 1 (7.1) | 0.000 (−0.009, 0.005) | 0.059 (−0.134, 0.252), P = 0.575 | |
| 2 (middle predicted harm) | 3,512 | 3,485 | 576 (8.2) | 329 (9.4) | 247 (7.1) | 0.022 (0.009, 0.038) | 0.023 (0.010, 0.036), P = 0.001 | |
| 3 (highest predicted harm) | 824 | 806 | 298 (18.3) | 164 (19.9) | 134 (16.6) | 0.053 (0.040, 0.088) | 0.033 (−0.005, 0.070), P = 0.087 | |
| ACCORD-BP | CVD events/deaths | |||||||
| 1 (lowest predicted benefit) | 986 | 986 | 240 (12.2) | 121 (12.3) | 119 (12.1) | 0.001 (−0.010, 0.027) | 0.002 (−0.027, 0.031), P = 0.890 | |
| 2 (middle predicted benefit) | 748 | 765 | 208 (13.7) | 95 (12.7) | 113 (14.8) | −0.018 (−0.010, 0.029) | −0.021 (−0.055, 0.014), P = 0.242 | |
| 3 (highest predicted benefit) | 421 | 415 | 196 (23.4) | 85 (20.2) | 111 (26.7) | −0.052 (−0.030, −0.115) | −0.066 (−0.123, −0.008), P = 0.025 | |
| Serious adverse events | ||||||||
| 1 (lowest predicted harm) | 17 | 13 | 9 (30.0) | 8 (47.1) | 1 (7.7) | −0.003 (−0.019, 0.005) | 0.394 (0.116, 0.672), P = 0.020 | |
| 2 (middle predicted harm) | 1,699 | 1,712 | 346 (10.1) | 216 (12.7) | 130 (7.6) | 0.025 (0.011, 0.039) | 0.051 (0.031, 0.071), P < 0.001 | |
| 3 (highest predicted harm) | 439 | 441 | 115 (13.1) | 88 (20.0) | 27 (6.1) | 0.053 (0.040, 0.087) | 0.139 (0.096, 0.183), P < 0.001 | |
SPRINT cohort: N = 9,664 (intensive treatment, N = 4,359; standard treatment, N = 4,305). ACCORD-BP cohort: N = 4,321 (intensive treatment, N = 2,155; standard treatment, N = 2,166). The lowest predicted benefit subgroup had a <1-percentage-point predicted absolute risk reduction in CVD events/deaths, while the highest predicted benefit subgroup had a >3-percentage-point predicted absolute risk reduction. The lowest predicted harm subgroup had a <0.5-percentage-point predicted absolute risk increase in serious adverse events, while the highest predicted harm subgroup had a >4-percentage-point predicted absolute risk increase. Cut points were chosen to correspond to the tertiles of the distribution of predicted benefit and harm for the combined data from SPRINT and ACCORD-BP.
CVD, cardiovascular disease.