Search
Close
Search
Advanced Search
Search Menu
AI Discovery Assistant

Extract

This commentary refers to ‘Blood pressure variability and risk of cardiovascular events and death in patients with hypertension and different baseline risks’, by M.H. Mehlum et al., 2018;39:2243–2251.

Mehlum et al.1 provided a remarkable paper about blood pressure variability (BPV) and risk of cardiovascular events or death in patients with hypertension. Using a Cox model, they quantified the association between BPV and several cardiovascular endpoints in the VALUE trial. Individual BPV was defined as the standard deviation of blood pressure (BP) measurements from all visits from 6 months to 60 months in order to increase the precision of BPV.1

We would like to challenge this strategy as it violates the fundamental rule of survival analysis, which is to never condition on the future.2 Using BP measurements over the entire follow-up implies that, at each time point, the BPV value of each subject at risk is derived not only from his past observed measurements, but also from future measurements, including those after a cardiovascular event. They performed sensitivity analysis restricted to patients who had no cardiovascular events during the first 24 months, their BPV being defined from 4 months to 24 months,1 which is more appropriate since it does not condition on the future. However, this analysis on a smaller sample has a reduced statistical power and it is also exposed to a selection bias of the sub-population. The results were indeed quite different compared with the main analysis in terms of hazard ratios, confidence intervals, and P-values (Table 2 vs. Table S5).1

Topic:
Issue Section:
Discussion Forum
You do not currently have access to this article.
Download all slides