Literature Search and Study Characteristics
In total, we used data from 21 articles (Table 1), including 897 participants with sample sizes ranging from 10 to 100 participants. The study that provided information on participants stated that there were 375 men (51.72%), with an average age of 55.07 years. Most studies (n = 13) were conducted in Asia, 6 studies were conducted in China(17, 24–26, 30, 32), 4 in Japan(21, 33–35), 3 in Kore(16, 23, 27); 1 in Egypt(15), Israel(20), ITALY(36), Australia(19), Thailand(28), Indonesia(18), and UK(22) and Netherlands(29) respectively. All studies reported SBP and DBP, except for 2 studies(15, 29) that reported only SBP. In 21 studies, wearable devices and cuff sphygmomanometers were measured at the same time.
Comparison of BP readings from wearable devices and cuff sphygmomanometers in different populations
Thirteen studies(15, 16, 19–21, 25, 27, 29, 30, 32–35) reported the mean age of participants. We divided these studies into two groups based on reported mean age: group 1, subjects less than 60 years old, and group 2, subjects older than or equal to 60 years old. The results showed that there was no difference in SBP and DBP readings between wearable devices and cuff sphygmomanometers in both age groups (pooled MD, 0.39 mmHg; 95% CI, -1.41 to 2.18; n = 13; I² = 0%, P = 0.67 for SBP; pooled MD, 0.15 mmHg; 95% CI, -1.09 to 1.39; n = 11; I² = 0%, P = 0.81 for DBP) (Supplemental Figure S3). The funnel plot for comparing the accuracy of measuring systolic BP and diastolic BP in the two age groups (Supplemental Figure S4).
Comparison of BP readings from wearable devices of different techniques and cuff sphygmomanometers.
Sixteen studies(15–18, 20–22, 25–27, 30, 32–36) reported on techniques used for BP measurements in wearable devices. We divided these studies into four groups: subgroup 1, photoplethysmography (PPG); subgroup 2, oscillometry; subgroup 3, electrocardiograph (ECG); and subgroup 4, PPG-ECG combination. Overall, there was no significant difference in BP readings between wearable devices and cuff sphygmomanometers, with the pooled MD in SBP was 0.87 mmHg (95% CI, -0.39 to 2.14; n = 16; I² = 0%, P = 0.18) and the pooled MD in DBP of 15 studies(15–18, 20–22, 25–27, 30, 33–36) was − 0.02 mmHg (95% CI, -1.45 to 1.50; n = 15; I² = 56%, P = 0.97). Except that ECG-based BP measurements have higher readings for systolic and diastolic BP compared to cuff sphygmomanometers, BP readings from PPG, oscillometry, and PPG-ECG combination-based measurements are similar to the readings from cuff sphygmomanometers (Supplemental Figure S5). The funnel plot for comparing the accuracy of measuring SBP and DBP with different techniques (Supplemental Figure S6).
Comparison of BP readings from wearable devices at different measurement sites and cuff sphygmomanometers.
In addition, we conducted a subgroup analysis of 17 studies(15–18, 20, 21, 23–25, 27–30, 32–35) that reported measurement sites, including subgroup 1, wrist and subgroup 2, fingers. The pooled MD in SBP between wearable devices and cuff sphygmomanometers was 0.33 mmHg (95% CI, -0.68 to 1.34; n = 16; I² = 0%, P = 0.53). There was no significant difference in SBP readings between wearable devices and cuff sphygmomanometers in each subgroup and overall level (Fig. 3A). However, the pooled MD in DBP between wearable devices and cuff sphygmomanometers was − 1.42 mmHg (95% CI, -2.33 to 0.51; n = 14; I² = 18%, P༜0.05), and the wrist MD in DBP of 11 studies(16–18, 20, 21, 23, 24, 27, 28, 34, 35) was − 1.85 mmHg (95% CI, -2.75 to -0.95; n = 11; I² = 10%, P༜0.05). The results showed that DBP readings of wearable devices at the wrist was lower than cuff sphygmomanometers, the difference was statistically significant (Fig. 3B). The funnel plot for comparing the accuracy of measuring SBP and DBP at different measurement sites (Supplemental Figure S7)
Based on differences in wrist DBP readings, we performed a further subgroup analysis of the wrist wearables devices with different techniques. Ten studies of wrist wearable devices that reported measurement techniques, subgroup 1, oscillometry(18, 21, 34, 35); subgroup 2, PPG(16, 20); and subgroup 3, other techniques, including PPG and ECG(17, 27), PTT(24), Hall device(23), LCPS and ECG(17). The analysis results show that there was no significant difference in DBP readings between cuff sphygmomanometers and the wrist wearable devices of oscillometry and PPG technology (Oscillometry MD, -0.75 mmHg (95% CI, -2.68 to 1.19; n = 4; I² = 0%, P = 0.45; PPG MD, 0.70 mmHg; 95% CI, -4.09 to 5.50; n = 2; I ²= 0%, P = 0.77) (Supplemental Figure S8). The funnel plot for comparing the accuracy of diastolic BP at the wrist with different techniques (Supplemental Figure S9).