Association between different methods of assessing blood pressure variability and incident cardiovascular disease, cardiovascular mortality and all-cause mortality: a systematic review.

BACKGROUND: Blood pressure variability (BPV) is a possible risk factor for adverse cardiovascular outcomes and mortality. There is uncertainty as to whether BPV is related to differences in populations studied, measurement methods or both. We systematically reviewed the evidence for different methods to assess blood pressure variability (BPV) and their association with future cardiovascular events, cardiovascular mortality and all-cause mortality. METHODS: Literature databases were searched to June 2019. Observational studies were eligible if they measured short-term BPV, defined as variability in blood pressure measurements acquired either over a 24-hour period or several days. Data were extracted on method of BPV and reported association (or not) on future cardiovascular events, cardiovascular mortality and all-cause mortality. Methodological quality was assessed using the CASP observational study tool and data narratively synthesised. RESULTS: Sixty-one studies including 3,333,801 individuals were eligible. BPV has been assessed by various methods including ambulatory and home-based BP monitors assessing 24-hour, "day-by-day" and "week-to-week" variability. There was moderate quality evidence of an association between BPV and cardiovascular events (43 studies analysed) or all-cause mortality (26 studies analysed) irrespective of the measurement method in the short- to longer-term. There was moderate quality evidence reporting inconsistent findings on the potential association between cardiovascular mortality, irrespective of methods of BPV assessment (17 studies analysed). CONCLUSION: An association between BPV, cardiovascular mortality and cardiovascular events and/or all-cause mortality were reported by the majority of studies irrespective of method of measurement. Direct comparisons between studies and reporting of pooled effect sizes were not possible.