Table 2.
Multivariable Analysis for the Prognostic Values of Various Parameters Describing Night‐to‐Morning SBP Changes With All‐Cause and Cardiovascular Mortality
| Parameters | Thresholdsa | All‐Cause Mortality (Event=607) | Cardiovascular Mortality (Event=182) | ||||||
|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P Value | Statistical Power for All‐Cause Deathb | HR | 95% CI | P Value | Statistical Power for Cardiovascular Deathb | ||
| Amplitude of STMS, mm Hg | 43.67 | 1.245 | 0.911–1.701 | 0.16 | 0.25 | 0.966 | 0.535–1.747 | 0.9101 | 0.05 |
| Rate of STMS, mm Hg/h | 11.33 | 1.666 | 1.185–2.341 | 0.0033c | 0.84c | 2.608 | 1.554–4.375 | 0.0003c | 0.86c |
| Preawakening surge, mm Hg | 32.75 | 0.826 | 0.534–1.277 | 0.39 | 0.2 | 1.380 | 0.3048–2.555 | 0.3048 | 0.18 |
| ME difference, mm Hg | 25.73 | 1.24 | 0.916–1.679 | 0.16 | 0.24 | 1.183 | 0.664–2.108 | 0.5692 | 0.08 |
| MN difference, mm Hg | 25.7 | 1.114 | 0.767–1.620 | 0.57 | 0.1 | 0.728 | 0.343–1.545 | 0.4086 | 0.17 |
Each morning blood pressure surge parameter was 1‐by‐1 tested in a multivariable Cox proportional hazard model. Models were adjusted for sex, age, antihypertensive treatment, body mass index, low‐density lipoprotein cholesterol, smoking, alcohol drinking, 24‐hour SBP, and the SBP night:day ratio. CI indicates confidence interval; HR, hazard ratio; ME difference, difference between morning and evening SBP; MN difference, difference between morning and night SBP; SBP, systolic blood pressure; and STMS, sleep‐trough morning blood pressure surge.
The thresholds were determined by the 95th percentile of parameters describing the night‐to‐morning SBP change.
For both all‐cause and cardiovascular deaths, a significant increase of risk for normal vs high STMS rate is practically meaningful. There is 84% power to detect a 66% increase of risk for all‐cause death, given a sample size of 2020 patients, a hazard ratio of 1.666, an overall probability of death of 0.3475, a ratio of patients in the group of high STMS rate as 5%, and a type I error rate of 5%. Similarly, for the power for cardiovascular death, with the previously described conditions, except for hazard ratio of 2.608, the overall probability of cardiovascular death is 0.1042 and there is 86% power to detect a 160% increase of risk.
Given the present sample size, the explanation of the p values and statistical power for all‐cause and cardiovascular mortalities were illustrated as above.