Journal of Hypertension | 2021
Blood pressure and outcome in patients with atrial fibrillation: floating in uncharted waters.
Abstract
W e read with interest the retrospective observational study by Minhas et al. [1] published in the Journal of Hypertension, which investigated the optimal office blood pressure (OBP) level in patients with atrial fibrillation treated with direct oral anticoagulants (DOACs). This study included 9051 atrial fibrillation patients in primary care in England who were registered to the Clinical Practice Research Datalink (CPRD), an anonymized primary care electronic health records database. Data were also obtained by Hospital Episode Statistics (HES) Admitted Patient Care database and Office of National Statistics (ONS). A composite of major events was used, including first or recurrent stroke, myocardial infarction, symptomatic intracranial bleed and significant gastrointestinal bleed. The results suggested that systolic OBP 160 mmHg or less was associated with higher all-cause mortality, but lower event risk. Clinical trials in hypertension have consistently excluded atrial fibrillation patients because of uncertainty in measuring blood pressure (BP) because of inherently increased beat-to-beat BP variability in these patients [2]. Thus, despite the fact that atrial fibrillation and hypertension often coexist, particularly in the elderly, the evidence on the prognostic value of BP and the optimal BP levels in hypertensive patients with atrial fibrillation is scarce, and mostly derived as secondary data from trials of DOACs in atrial fibrillation [3]. However, these trials did not have BP as primary or even secondary endpoint, resulting in poorly standardized BP measurement methodology (devices and protocol) and, thereby, questionable conclusions. Despite these methodological deficiencies, OBP measurements in atrial fibrillation patients have been shown to be relatively accurate and significant in terms of prognosis [2]. A recent meta-analysis of nine studies including 65 637 atrial fibrillation patients receiving DOACs, showed that elevated OBP (usually defined as systolic OBP at least 140 mmHg) and hypertension diagnosis predict stroke or systemic embolism, with follow-up OBP control having even stronger predictive ability than baseline OBP [3]. However, no association was found with major haemorrhagic events or all-cause mortality, which might be because of U-shaped association of BP with these outcomes [3]. These findings are generally in line with the respective data in sinus rhythm individuals and are reassuring for the clinical relevance of OBP measurements in atrial fibrillation patients. Although the study by Minhas et al. [1] addressed an inadequately investigated research question, its findings should be interpreted by considering several methodological issues, some of which are acknowledged by the authors. First, OBP measurement, which was the primary measure of the analysis, was not standardized and inadequately described (measurement method, device type and validation, number of readings, etc.). Second, history of hypertension, use of antihypertensive drugs and follow-up OBP levels were not considered in the analysis. Third, only 5% of the study participants had stage 2 hypertension (systolic OBP>160 mmHg). Fourth, this analysis was based on three national databases and conclusions were derived by taking into consideration timepoints of electronic prescriptions and disease coding (e.g. ICD-10). Such data do not have the quality of prospective research trials. Fifth, the discrepancy between morbidity and mortality trends (major events increased and total/cardiovascular mortality decreased with increasing OBP) was unexpected. In addition, the composite event endpoint included both thrombotic and bleeding events, which may behave differently. In the supplementary material, bleeding events were analysed separately, and as their hazard ratios did not change across OBP categories, it can be assumed that the rate of thrombotic events was increased with increased OBP levels. Unfortunately, such an analysis was not available. Moreover, whenever using linear models, systolic OBP level of 210 mmHg appeared to carry the lowest risk for all-cause mortality, which is questionable, given the observed high rate of events at this BP level. This association was not confirmed by alternative analysis (restricted cubic splines approach) [1]. Hypertension and atrial fibrillation are closely associated and often coexist, particularly in the elderly. Unfortunately, atrial fibrillation patients have been consistently excluded from clinical research trials in hypertension, because of issues with BP measurement accuracy. Thus, the evidence on the prognostic ability of BP measurements in atrial fibrillation is scarce and mainly derived from secondary analyses of clinical trials designed for other purposes. Studies such as that by Minhas et al. provide new data on this important but unexplored field, yet their methodological flaws must be considered. Randomized hypertension trials including atrial fibrillation patients and with appropriate BP measurement methodology are needed. Until such data become available, indirect data support the use of standard BP measurement methods and BP thresholds in treating patients with hypertension and atrial fibrillation.