Diagnostic accuracy of attended versus unattended automated blood pressure measurements compared to ambulatory blood pressure monitoring conducted by community health workers in rural sub-Saharan Africa

Abstract

Community Health Workers (CHWs) are increasingly being involved in screening for hypertension in rural sub-Saharan Africa. However the diagnostic accuracy of CHW-led attended (CHW present during procedure) and unattended (CHW absent) automated blood pressure (BP) in home settings has not been compared. We conducted a cross-sectional, population-based diagnostic accuracy study in two health and demographic surveillance systems in Kenya and The Gambia. A random age-stratified sample of individuals aged ≥ 30 years were included. The index tests were attended and unattended automated BP measurements conducted by CHWs at participants’ homes (AHBP). The reference test, 24-hour ambulatory BP monitoring (ABPM), was also conducted by the CHWs. We applied BP cutoffs derived from the European Society of Hypertension to determine sensitivity, specificity, predictive values and area under receiver operating characteristic curve (AUC). We included 1220 participants (659 in Kenya and 561 in The Gambia). The median age was 54 years (interquartile range 43 years, 66 years) and 775 (63.5%) participants were women. Age standardized prevalence of hypertension on 24-hour BP was 25.1% (95% confidence interval 21.7, 28.4) in Kenya and 32.7% (28.8, 36.6) in The Gambia. There was no difference in diagnostic accuracy of attended and unattended measurements. In Kenya, the age-standardized sensitivity of attended AHBP was 59.5% (52.1, 66.9) and 56.1% (48.4, 63.8) for unattended AHBP. In The Gambia, these sensitivities were 42.2% (34.9, 49.5) and 42.2% (35.2, 49.2) respectively. The respective age-standardized specificities were 92.5% (90.2, 94.7) and 92.6% (90.2, 95.0) in Kenya and 93.2% (90.6, 95.9) and 93.3% (90.8, 95.9) in The Gambia. The AUC was 0.76 (0.72, 0.80) and 0.74 (0.70, 0.78) for attended and unattended AHBP in Kenya and, 0.68 (0.64, 0.72) and 0.68 (0.64, 0.71) respectively in The Gambia. Attended and unattended AHBP conducted by CHWs had comparable diagnostic accuracy. Both approaches showed high specificity but low sensitivity compared with 24-hour ABPM, suggesting that CHW-led AHBP may serve as an initial community-based screening but should not replace confirmatory BP assessment.