Searching from Medline, Scopus, CENTRAL and WHO ICTRP databases identified 1,104, 1,224, 267, and 59 articles, respectively. After deleting duplications, 1,945 studies were screened by titles and abstracts. A total of 233 full articles were reviewed. Fifty-eight studies met inclusion criteria and were included in the review. Among the included studies, 50[9, 11, 19-66], 4[67-70] , and 4[71-74] studies assessed CBPM, HBPM, and both CBPM and HBPM performances (see Figure 1). Their characteristics are described in Table 1. Among the 58 studies, 32 and 26 studies recruited participants from hospital and community settings, correspondingly. Most studies included general population, whereas 4 studies included specific populations, i.e. , white-collar workers[25, 66], male football players[40] and male military workers[37]. Twenty-one studies included participants who had not been prior diagnosed with HT, whereas 37 studies included both participant who had or had not been diagnosed with HT.
Risk of bias assessment
Results of risk of bias assessment are presented in Additional Table 1. Almost all CBPM studies (94.44%) were low risk in all domains of applicability. Eight[20, 31, 38, 39, 44, 50, 52, 57] (16.7%) and 7 (12.9%) studies[30, 34, 41, 46, 49, 55, 58] were high or unclear bias in selection of study subjects, accordingly. Fifty-two studies (96.3%) [9, 11, 19-31, 33-38, 40-66, 70, 71, 74] applied the index/study test before the reference standard but with unclear explanation of blinding. Thirty-eight studies (70.4%) were high or unclear risk of bias in flows and timing. This is due to a lack of reporting the time interval between the study test and the reference standard or the exclusion of subjects with invalid test results or those lost to follow up. All HBPM studies were low risk of bias in all domains of applicability. Six studies (75%) applied the index test before the reference standard without blinding information, and 4 studies (50%) were high risk of bias in their flow and timing.
Pooling CBPM diagnostic performances
Among 54 CBPM studies, 31 studies[9, 11, 19-34, 40, 58-66, 71-73] reported 2x2 table data which could be assessed for diagnostic performance, while 7[35-39, 41, 42] and 16[43-57, 70] studies reported data for only positive and negative CBPM respectively (see Table 1). The mean age ranged from 28 to 62 years and percent male ranged from 35% to 100%. The number of CBPM measurements per visit ranged from 1 to 5 times (Table 1).
Among the studies that reported 2x2 data (n = 66,767), 29 studies[9, 11, 20-34, 40, 58-63, 66, 72-74] used a CBPM cutoff threshold as ≥140/90 for diagnosis of HT, while one study[19] used the threshold of DBP>95 mmHg and one study did not reported the threshold. The 24-hour ABPM had a cut-off of ≥130/80 mmHg for 12 studies[11, 26, 31-33, 58, 59, 61-63, 74] and daytime ABPM had a cut-off of ≥135/85mmHg for 16 studies[20-25, 28-30, 34, 60, 65, 66, 73]. Two[27, 72] and one[19] studies applied daytime ABPM with cut-offs of ≥140/85 and DBP≥95 mmHg, respectively (see Additional Table 2). When using the 24-hour ABPM with the cut-off of ≥130/80 mmHg as the reference standard, the diagnostic performance of CBPM were 0.74 (95% CI 0.65-0.82; I2=99.4%), 0.79 (95% CI: 0.69, 0.87; I2=99.65%), 3.6 (95% CI: 2.4, 5.3; I2=99.67%) and 0.32 (95% CI: 0.24, 0.44; I2=99.58%) for sensitivity, specificity, LR+ and LR-, respectively (see Figure 2a and Additional Figure 1). These diagnostic characteristics all require setting a threshold and trading off sensitivity for specificity or LR+ for LR- hence they must be judged in pairs. For example, given a pretest probability of HT of 44%, the post-test probability was increased to 74% if CBPM was positive or reduced to 20% if CBPM was negative (see Fagan’s plot Figure 3a). Alternatively, a single measure of diagnostic performance, i.e., the DOR was 11.11 (95% CI: 6.44, 19.160; I2 = 100%), see Additional Figure 1c. The HSROC reflects diagnostic performance across the entire range of possible threshold values; in this case, the pooled HSROC was 0.83 (95% CI: 0.82, 0.85) indicating moderately good discrimination for judging presence of HT (Additional Figure 2a).
When using daytime ABPM with cut-off of 135/85 mmHg as the reference standard, the pooled sensitivity and specificity were 68% (95% CI: 57, 77; I2=97.36%) and 82% (95%CI: 70, 90; I2=99.13%), see Figure 2b. In addition, LR+, LR- and DOR of CBPM were 3.7 (95% CI: 2.3, 6.0; I2=98.57%), 0.39 (95%CI: 0.30, 0.52; I2=94.54%) and 9.46 (95% CI: 5.39, 16.60; I2=100%), accordingly (see Additional Figure 3). When all ABPMs with no restriction on the cutoffs as the reference standards were used, the pooled sensitivity and specificity were 70% (95% CI: 63%, 76%; I2 = 98.56%) and 81% (95% CI: 73%, 87%; I2 = 99.47%) and pooled LR+ and LR- were 3.67 (95% CI: 2.69, 5.00; I2 = 99.35%) and 0.37 (95% CI: 0.31-0.44; I2 = 98.57%). For publication, Deeks’ funnel plot showed no evidence of publication bias (Additional Figure 4a).
Subgroup analysis
Subgroup analyses were performed by results of risk of bias assessment, age group (< 50 and ≥50 years), percent males (< 50% and ≥50%), number of repeated measurements of CBPM (1, 2-5 times), setting of studies (community and hospital-based) and type of patients (no HT, mixed HT with non-HT). When considering only studies with low risk of bias in the domain of flow and timing (N=7), pooled sensitivity, specificity, LR+ and DOR of CBPM were 73% (95% CI: 60, 83), 75% (95% CI: 51, 89), 2.9 (95% CI: 1.5, 5.3), and 8 (95% CI: 5, 14), respectively. The degrees of heterogeneity (I2) did not decrease for each sub-group of these factors (Additional Table 3), but performances of CBPM improved in some sub-groups including age group ≤50 year, percent male ≤50% and community-based setting with the LR+ of 5.1 (95% CI: 3.0, 8.7), 5.8 (95% CI: 3.5, 9.8), and 6.0 (95% CI: 3.9, 9.3), respectively.
Pooling HBPM diagnostic performances
Eight HBPM studies[67-74] reported 2x2 data (Additional Table 4) with cutoff threshold of 135/85[67-71, 74] (N=7) and 140/90[73] (N=1) mmHg and measurement duration of about 3 to 7 days. The number of measurements per day ranged from 2 to 12 times (see Table 1). Mean age and percent male ranged from 48.1 to 51.8 years and 46.5% to 54.9% respectively. Among them, five and three studies applied 24-hour and daytime ABPM, respectively.
The pooled sensitivity, specificity, DOR, LR+ and LR- of HBPM were respectively 0.71 (95% CI: 0.61, 080; I2 = 97.17%), 0.82 (95% CI: 0.77, 0.87; I2 = 85.48), 11.60 (95% CI: 8.98, 15.13; I2 = 100%), 4.02 (95% CI: 3.38, 4.78; I2 = 19.39%) and 0.35 (95% CI: 0.26, 0.46; I2 = 95.69%), when the 24-hour ABPM with cut-off of ≥130/80 mmHg was applied as the reference standard (see Figure 4a and Additional Figure 5). In addition, among persons having HBPM positive, 14% had normotension from 24-hour ABPM. In contrast, 40% of those having negative HBPM were diagnosed with hypertension from 24-hour ABPM. Again, given a pretest-probability of 44%, a positive HBPM would result in a post-test probability of 76%, while a negative HBPM would reduce the probability to 21% (Figure 3b). The pooled HSROC was 0.85 (95% CI: 0.82, 0.88), reiterating moderately good discrimination, see Additional Figure 2b.
When all types ABPM with all cut-offs as the reference standard were applied, the overall pooled sensitivity and specificity were respectively 74% (95% CI: 66%, 80%; I2 = 95.52%) and 83% (95% CI: 76%, 89%; I2 = 90.20%), see Figure 4b. The pooled DOR, LR+ and LR-were 13.73 (95% CI: 8.55, 22.03; I2 = 99.99%), 4.36 (95% CI: 3.04, 6.27; I2 = 75.06%), and 0.32 (95% CI: 0.25, 0.41; I2 = 94.34%). Analysis using daytime ABPM as the reference standard and subgroup analysis of HBPM could not be performed due to the small number of included studies. Deeks’ funnel plot indicated no evidence of publication bias, see Additional Figure 4b.
Pooling prevalence of WCHT and MHT by CBPM
Seven[35-39, 41, 42] and 16 studies[43-57, 64] reported only data of WCHT and MHT, see Table 1. These studies were then combined with 31 CBPM studies with 2x2 data above yielding a total of 38 and 47 studies for pooling proportions of WCHT and MHT, respectively. Among the 38 studies with WCHT, time of ABPM measures were 24-hour (N=14) and daytime (N=24). Among the 47 studies with MHT, ABPM measurements were 24-hour ABPM (N=20) and daytime ABPM (N=27). Four studies compared the performance of CBPM with both HBPM and ABPM but only two studies provided the number of people who had negative CBPM but had high blood pressure from either ABPM or HBPM. Therefore, most studies (N=45) used only ABPM as the reference standard for pooling the prevalence of MHT.
Using the 24-hour ABPM with a cut-off of 130/80 mmHg as the reference standard (N=23), the pooled prevalence of WCHT and MHT were 0.24 (95% CI: 0.19, 0.29; I2=97.96%) and 0.29 (95% CI: 0.20, 0.38; I2=99.64%), see Figure 5a and 5b. If daytime ABPM with cut-off of 135/85 mmHg was applied as the reference standard, the pooled prevalence of WCHT (N=21) and MHT (N=20) would be 0.29 (95% CI: 0.22, 0.36; I2=97.47%) and 0.24 (95% CI: 0.20, 0.27; I2=96.09%), see Figure 5c and 5d.
When all types of ABPM were applied with any cut-offs as the reference standard, the pooled prevalence of WCHT (N=38; n=32685) and MHT (N=47; n=47713) were 0.28 (95% CI: 0.25, 0.32) and 0.27 (95% CI: 0.22, 0.32). Subgroup analyses were performed, but none of the co-variables could decrease the degree of heterogeneity (see Additional Table 7). However, subgroup of repeated measures of CBPM 4-5 times and 24-hr ABPM could respectively reduce the pooled WCHT from 0.28 to 0.23 (95% CI: 0.16, 0.31) and 0.23 (95% CI: 0.18, 0.28). Likewise, repeated CBPM measure could reduce the pooled MHT from 0.27 to 0.15 (95% CI: 0.10, 0.19) whereas the 24-hr ABPM conversely increased the prevalence to 0.33 (95% CI: 0.22, 0.43).