Arterial blood pressure monitoring in stroke cohorts: the impact of reduced sampling rates to optimise remote patient monitoring.

Objective

Remote patient monitoring (RPM) beat-to-beat blood pressure (BP) provides an opportunity to measure poststroke BP variability (BPV), which is associated with clinical stroke outcomes. BP sampling interval (SI) influences ambulatory BPV, but RPM BP SI optimisation research is limited. SI and RPM device capabilities require compromises, meaning SI impact requires investigation. Therefore, this study assessed healthy and stroke subtype BPV via optimised BP sampling, aiding sudden BP change identification and potentially assisting cardiovascular event (recurrent stroke) prediction.

Methods

Leicester Cerebral Haemodynamic Database ischaemic [acute ischaemic stroke (AIS), n  = 68] and haemorrhagic stroke (intracerebral haemorrhage, n  = 12) patient and healthy control (HC, n  = 40) baseline BP data were analysed. Intrasubject and interpatient SD (SD i /SD p ) represented individual/population variability with synthetically altered SIs. Matched-filter approaches using cross-correlation function detected sudden BP changes.

Results

At SIs between 1 and 180 s, SBP and DBP SD i staticised while SD p increased at SI < 30 s. Mean BP and HR SD i and SD p increased at SI < 60s. AIS BPV, normalised to SI1s, increased at SI30s (26%-131%) and SI120s (1%-274%). BPV increased concomitantly with SI. Cross-correlation analysis showed HC and AIS BP sudden change detection accuracy reductions with increasing SI. Positive BP deviation detection fell 48.48% (SI10s) to 78.79% (SI75s) in HC and 67.5% (SI10s) to 100% (SI75s) in AIS. Negative BP deviation detection fell 50% (SI10s) to 82.35% (SI75s) in HC and 52.27% (SI10s) to 95.45% (SI75s) in AIS.

Conclusion

Sudden BP change detection and BPV are relatively robust to SI increases within certain limits, but accuracy reductions generate unacceptable estimates, considerable within RPM device design. This research warrants further SI optimisation.