A Comparison of the Diagnostic Accuracy of Common Office Blood Pressure Measurement Protocols

A Comparison of the Diagnostic Accuracy of Common Office Blood Pressure Measurement Protocols Abstract BACKGROUND The optimal approach to measuring office blood pressure (BP) is uncertain. We aimed to compare BP measurement protocols that differed based on numbers of readings within and between visits and by assessment method. METHODS We enrolled a sample of 707 employees without known hypertension or cardiovascular disease, and obtained 6 standardized BP readings during each of 3 office visits at least 1 week apart, using mercury sphygmomanometer and BpTRU oscillometric devices (18 readings per participant) for a total of 12,645 readings. We used confirmatory factor analysis to develop a model estimating “true” office BP that could be used to compare the probability of correctly classifying participants’ office BP status using differing numbers and types of office BP readings. RESULTS Averaging 2 systolic BP readings across 2 visits correctly classified participants as having BP below or above the 140 mm Hg threshold at least 95% of the time if the averaged reading was <134 or >149 mm Hg, respectively. Our model demonstrated that more confidence was gained by increasing the number of visits with readings than by increasing the number of readings within a visit. No clinically significant confidence was gained by dropping the first reading vs. averaging all readings, nor by measuring with a manual mercury device vs. with an automated oscillometric device. CONCLUSIONS Averaging 2 BP readings across 2 office visits appeared to best balance increased confidence in office BP status with efficiency of BP measurement, though the preferred measurement strategy may vary with the clinical context. blood pressure, hypertension, measurement, screening Hypertension is the most common chronic medical condition in the United States, affecting nearly one-third of the adult population,1,2 and a leading risk factor for cardiovascular disease and mortality.3,4 Guidelines recommend screening for hypertension by first measuring office blood pressure (BP), and then obtaining confirmatory out-of-office testing using either 24-hour ambulatory blood pressure monitoring or home blood pressure monitoring.5–7 Misclassifying a patient as having elevated office BP can lead to inappropriate diagnostic labeling, overtreatment with antihypertensive medications, and unnecessary health care and insurance costs.8–10 Conversely, misclassifying a patient as having non-elevated office BP may lead to undertreatment of hypertension. Despite the importance of its accurate measurement, there is no consensus among guidelines on the optimal number of readings or the best device for measuring office BP.11,12 Office BP measurement is used to approximate an individual’s average, or usual, resting office BP, referred to here as the “true office BP.” Confirmatory factor analysis (CFA) is a statistical method that can be used to model the relationships of multiple office BP readings with the “true office BP.”13 Thus, we developed a CFA model that enables us to compare the diagnostic accuracy of common approaches to measuring office BP including: (i) increasing the number of BP readings taken within a visit; (ii) increasing the number of office visits in which BP readings are taken; (iii) averaging 3 BP readings taken in a single visit vs. dropping the first of 3 readings and averaging the rest; and (iv) using manual vs. automatic BP devices.14 METHODS Study population We analyzed data from the 707 participants enrolled at the Stony Brook University site of the Masked Hypertension Study, a community-based study of the prevalence, predictors, and prognosis of masked hypertension.15 Participants were recruited by conducting BP screenings for employees of Stony Brook University, Stony Brook University Hospital, and a small financial firm. At the time of the screening, employees were briefly told about the study. Potentially eligible employees who were interested in participating were subsequently contacted by telephone to confirm their eligibility. Those who were eligible and chose to participate were formally consented at their first study visit. The institutional review boards of Stony Brook University and Columbia University Medical Center approved all procedures for this study. Employees were eligible if they were 18 years of age or older, spoke and read English, were employed more than 20 hours per week, worked on at least 2 consecutive days per week, were not taking a hypertension medication, and had a screening BP <160/105 mm Hg (average of second and third readings). Employees were ineligible if they self-reported a history of cardiovascular disease, or other major chronic medical condition (i.e., kidney, liver, adrenal, thyroid disorder; organ transplant; cancer not in remission more than 6 months); were prescribed antihypertensive or any other cardiovascular medications other than statins; or took any medications known to influence BP. Other exclusions included current or planned pregnancy, severe psychiatric disorder, active substance abuse, and unavailability for follow-up during the 3 months after screening. Smoking, hyperlipidemia, type II diabetes mellitus, and a past history of untreated hypertension were not exclusion criteria. Procedure Participants attended 4 study visits over a 4-week period. During the first 3 visits, each approximately 1 week apart, office BP was assessed using a standardized protocol in accordance with the American Heart Association’s guidelines.11 At each office visit, study personnel first confirmed that participants had not eaten, smoked, or consumed a caffeinated beverage during the prior 30 minutes. A trained nurse then measured the circumference of the participant’s non-dominant arm with a Gullick II tape measure and chose the appropriately sized cuff. After the participant had been seated comfortably with their feet on the floor for at least 5 minutes, a mercury sphygmomanometer (Baum, Copiague, NY) was used to take 3 manual BP readings. The nurse waited 1–2 minutes between each reading. After the third reading, the nurse waited another 1–2 minutes and then took another 3 readings, each 2 minutes apart, using the BpTRU oscillometric device (BpTRU Medical Devices, Coquitlam, BC) on the same non-dominant arm. The nurse remained in the same room as the patient at the time of the automated readings. This protocol generated 18 office BP readings for each participant distributed across 3 visits (3 manual followed by 3 oscillometric readings per visit). Participants also completed a questionnaire that included sociodemographic characteristics. At the fourth visit, participants completed a medical history interview and had their height and weight measured in standardized fashion for calculation of body mass index. Statistical analysis We used CFA to model the relationships among the 18 office BP readings obtained by the sphygmomanometer and the oscillometric device (Figure 1). CFA is a multivariable statistical technique used to evaluate the relationship of multiple measured variables to the underlying constructs they are presumed to represent.16 Our CFA accounted for 4 main sources of variance among the individual BP readings. First and foremost, individuals differ in their “true office BP.” Second, office BPs measured at different visits may vary systematically from the “true office BP” due to a variety of unspecified visit-specific factors (e.g., recent salt intake, stress, emotional factors). Third, office BP readings often differ from each other (and the visit-specific “true BP”) due to random “noise.” By averaging multiple office readings during a single visit, the random fluctuations between readings average out and one arrives at a better estimate of the person’s “true BP” for that visit. Last, a single office BP reading may vary from the “true office BP” due to systematic upward (or downward) biases in measurement. Specifically, an alerting response found in initial readings may lead to a difference between the first and subsequent measurements, and this was explicitly incorporated into our model.17–19 Our model yields estimates of these different sources of variance that determine the individual office BP readings. Our model also accounts for potential systematic differences between manual and automatic office BP in means and variance components. Figure 1. View largeDownload slide Confirmatory factor analysis model of “true blood pressure” based on observed readings. The circle containing the text “Visit 1 BPman” represents the “true” average of the person’s manual BP during the 5 minutes manual BP readings were being taken at a single office visit. The rectangular boxes (BPM11, BPM12, and BPM13) represent the 3 observed manual BP readings taken at Visit 1. These readings are expected to be similar to “Visit 1 BPman” as they each depend primarily on the “true” average BP during those few minutes, represented by the arrows from “Visit 1 BPman” to each box. The fact that the 3 readings are not identical is portrayed by the 3 “ε” terms representing the fluctuations of the individual readings around the “true” average, including any random measurement error in the taking of the reading. The same logic applies to repeat visits (i.e., “Visit 2 BPman” and “Visit 3 BPman”). While not agreeing perfectly, “Visit 1 BPman,” “Visit 2 BPman,” and “Visit 3 BPmal” are also expected to be highly correlated because they each depend on ‘“true” Manual BP; the visit-to-visit fluctuations are portrayed by λM1, λM2, and λM3. There is an exactly parallel model for the automatic (BpTRU) measurements. The 2 halves of the overall model are linked by curved arrows representing the following correlations: the ‘“true” Manual BP is very highly correlated with the “true” Automatic BP; and whatever it is that makes someone’s Visit 1 manual BP deviate from their “true” manual office BP will tend to make their Visit 1 Automatic BP similarly deviate from their “true” Automatic BP (λM1 is expected to be positively correlated with λA1, and similarly for λM2 with λA2 and λM3 with λA3). Abbreviations: auto, automatic; BP, blood pressure; man, manual. Figure 1. View largeDownload slide Confirmatory factor analysis model of “true blood pressure” based on observed readings. The circle containing the text “Visit 1 BPman” represents the “true” average of the person’s manual BP during the 5 minutes manual BP readings were being taken at a single office visit. The rectangular boxes (BPM11, BPM12, and BPM13) represent the 3 observed manual BP readings taken at Visit 1. These readings are expected to be similar to “Visit 1 BPman” as they each depend primarily on the “true” average BP during those few minutes, represented by the arrows from “Visit 1 BPman” to each box. The fact that the 3 readings are not identical is portrayed by the 3 “ε” terms representing the fluctuations of the individual readings around the “true” average, including any random measurement error in the taking of the reading. The same logic applies to repeat visits (i.e., “Visit 2 BPman” and “Visit 3 BPman”). While not agreeing perfectly, “Visit 1 BPman,” “Visit 2 BPman,” and “Visit 3 BPmal” are also expected to be highly correlated because they each depend on ‘“true” Manual BP; the visit-to-visit fluctuations are portrayed by λM1, λM2, and λM3. There is an exactly parallel model for the automatic (BpTRU) measurements. The 2 halves of the overall model are linked by curved arrows representing the following correlations: the ‘“true” Manual BP is very highly correlated with the “true” Automatic BP; and whatever it is that makes someone’s Visit 1 manual BP deviate from their “true” manual office BP will tend to make their Visit 1 Automatic BP similarly deviate from their “true” Automatic BP (λM1 is expected to be positively correlated with λA1, and similarly for λM2 with λA2 and λM3 with λA3). Abbreviations: auto, automatic; BP, blood pressure; man, manual. With this conceptual model in place, we initially specified a maximally parsimonious CFA model (Figure 1) in which only 6 parameters were used to model the 18 means, 18 SDs, and 153 correlations among the 9 manual and 9 automatic office BP readings. Constraints on the model were removed in a conceptually guided, step-wise manner—and the model was made increasingly complex—if lifting constraints substantially improved the fit of the model. This model was estimated by full information maximum likelihood using the M-Plus software (version 6.1; Muthén & Muthén, Los Angeles, CA) and all available data under the assumption that any missing data were missing at random. We used multiple model fit indices along with standard recommendations for defining an “acceptable” model to assess model fit (Supplementary Tables 1and2).20 After arriving at a parsimonious model that adequately fit the data, model estimates (Supplementary Table 3) were used to generate estimates of the precision with which commonly used BP measurement protocols would approximate the “true office BP.” These estimates were then used to determine the probability that office BP, measured by a given protocol, would correctly classify an individual’s “true office BP” status as elevated or non-elevated using a cutpoint of 130/80 mm Hg, and separately 140/90 mm Hg. These cutpoints represent the thresholds for diagnosing stage I and stage II hypertension, respectively, as per the 2017 American College of Cardiology/American Heart Association high BP guidelines.7 Separate CFA models were estimated for systolic and diastolic BP. SAS (version 9.4, Cary, NC) was used to generate descriptive statistics for the sample. RESULTS The mean (SD) age of participants was 45.4 (10.2) years. Sixty percent of participants were women, 6% were African-American, and 7% were Hispanic (Table 1). Of the 6,363 planned manual and automatic BP readings, there were only 0.05% and 1.23% missing data, respectively. The average manual BP at the initial visit (average of 3 manual readings) was 116/76 mm Hg. The overall average manual BP (average of 9 manual readings) was 116/75 mm Hg, and the overall average oscillometric BP was nearly identical, 115/75 mm Hg (average of 9 oscillometric readings). Table 1. Demographic and clinical characteristics of participantsa Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Abbreviation: BP, blood pressure. aBlood pressure characteristics were based on the average of 3 manual readings during Visit 1. Participants who withdrew from the study prior to the fifth visit (n = 44) were missing data pertaining to body mass index, smoking status, and diabetes, and some (n = 18) were missing years of education. One participant’s Visit 1 blood pressure readings were lost prior to data entry. View Large Table 1. Demographic and clinical characteristics of participantsa Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Abbreviation: BP, blood pressure. aBlood pressure characteristics were based on the average of 3 manual readings during Visit 1. Participants who withdrew from the study prior to the fifth visit (n = 44) were missing data pertaining to body mass index, smoking status, and diabetes, and some (n = 18) were missing years of education. One participant’s Visit 1 blood pressure readings were lost prior to data entry. View Large Increasing the number of readings within visits and/or the number of visits The gain in confidence obtained by increasing the number of BP measurements/visit and/or increasing the number of visits is portrayed in Figure 2. Overall, one gained more confidence that a measured systolic BP was within 5 mm Hg of the “true” office systolic BP by increasing the number of visits as compared to increasing the number of readings within a visit. For example, when increasing the number of systolic BP readings within a single visit from 1 to 2, the likelihood that the measured systolic BP was within 5 mm Hg of the “true” office systolic BP increased by 9% for manual measurement and 7% for oscillometric measurement. In contrast, adding a second visit, again with one reading, increased the confidence by 14% for manual measurement and 16% for oscillometric measurement. There were diminishing returns from increasing the number of systolic BP readings within a visit from 2 to 3, or from increasing the number of visits from 2 to 3. For example, increasing the number of manual measurements within a single visit from 2 to 3 only increased the likelihood that the measured systolic BP was within 5 mm Hg of the “true” systolic BP by 4% for manual measurement and 3% for oscillometric measurement. The same pattern was present for diastolic BP (Supplementary Figure 1). Figure 2. View largeDownload slide Probability that the average of systolic blood pressure readings is within 5 mm Hg of the “true” office systolic blood pressure, by number of readings per visit and number of visits. Abbreviations: Avg, Average; BpTRU, an automatic oscillometric blood pressure device. Figure 2. View largeDownload slide Probability that the average of systolic blood pressure readings is within 5 mm Hg of the “true” office systolic blood pressure, by number of readings per visit and number of visits. Abbreviations: Avg, Average; BpTRU, an automatic oscillometric blood pressure device. Dropping the first of 3 BP readings and averaging the second and third readings The first manual systolic BP reading was 1.3 mm Hg higher than the overall average systolic BP. Figure 2 shows the difference in probability of being within 5 mm Hg of the “true” office systolic BP for a given number of visits when dropping the first of 3 BP readings and averaging the latter 2 as compared to averaging all 3 readings. Dropping the first of 3 manual BP readings did not substantially alter the precision of classification, but there was reduced precision for oscillometric measurements. The same pattern was found for diastolic BP (see Supplementary Figure 1). Manual vs. oscillometric measurement Figure 3 shows a comparison of the probability of being within 5 or 10 mm Hg of the “true” office systolic BP when using standardized, guideline-concordant manual and oscillometric BP measurements (i.e., 2 visits, 3 readings per visit, removing the first reading of each visit and average the remaining 2 readings11). The research quality manual readings were more likely (6%) to correctly classify “true” systolic BP status than the automatic readings. There was no appreciable difference between the 2 methods for diastolic BP (Supplementary Figure 2). Figure 3. View largeDownload slide Probability that the average measured systolic blood pressure is within 5 or 10 mm Hg of the “true” office systolic blood pressure for Manual vs. Automatic BpTRU measurement methods. Systolic blood pressure is assessed as the average of blood pressure readings from 2 visits with 3 readings per visit and the first reading of each visit dropped. Figure 3. View largeDownload slide Probability that the average measured systolic blood pressure is within 5 or 10 mm Hg of the “true” office systolic blood pressure for Manual vs. Automatic BpTRU measurement methods. Systolic blood pressure is assessed as the average of blood pressure readings from 2 visits with 3 readings per visit and the first reading of each visit dropped. Comparison of the probability of correct office BP classification according to different office BP measurement protocols Figure 4 shows the probability that an average office systolic BP will correctly classify a patient’s office systolic BP as elevated or non-elevated using a cutpoint of 140/90 mm Hg depending on the measurement protocol (combination of number of visits and number of readings per visit). If the average BP based on the indicated protocol is in the green zone, then the patient and clinician can be >95% confident that a diagnosis based on this average will be correct; stated differently, there is <5% chance that the true office systolic BP is on the opposite side of 140 mm Hg. If the average BP is in the yellow zone, then one’s confidence that a diagnosis based on this BP is correct is 75–95%; it will be wrong 5–25% of the time. Finally, if the average is in the red zone, then a diagnosis based on the observed average will be incorrect >25% of the time. Although data for this analysis were taken from BP readings from 3 office visits, once specified, the model can provide probability estimates of correct classification based on any number of visits and readings. Figure 4. View largeDownload slide Probability of correct classification of office blood pressure as elevated or non-elevated according to observed average systolic blood pressures obtained with different measurement strategies. Probabilities are based on estimates derived from the confirmatory factor analysis model. Although the data for this analysis were taken from blood pressure (BP) readings from 3 office visits, once specified, the model can provide probability estimates of correct classification of office BP based on an even larger number of visits and readings. Accordingly, the figure shows estimates for the probability of correct classification of office BP for BP measurement protocols with more than 3 visits. Figure 4. View largeDownload slide Probability of correct classification of office blood pressure as elevated or non-elevated according to observed average systolic blood pressures obtained with different measurement strategies. Probabilities are based on estimates derived from the confirmatory factor analysis model. Although the data for this analysis were taken from blood pressure (BP) readings from 3 office visits, once specified, the model can provide probability estimates of correct classification of office BP based on an even larger number of visits and readings. Accordingly, the figure shows estimates for the probability of correct classification of office BP for BP measurement protocols with more than 3 visits. The figure shows that there is greater benefit, in terms of increasing the range of office systolic BPs with high confidence classifications (green zone), from increasing the number of visits than from increasing the number of readings per visits. For example, using a protocol with a single reading at a single visit, a systolic BP of 150 mm Hg would provide only intermediate confidence (yellow zone) that the true office systolic BP was elevated whereas using a protocol with a single reading per visit across 2 visits, a systolic BP of 150 mm Hg would provide >95% confidence that the true office systolic BP was elevated (green zone). A similar pattern was observed for office diastolic BP classification (Supplementary Figure 3). For patients with average measured systolic BPs closer to the 140 mm Hg cutpoint, roughly between 137 and 143 mm Hg, even a strategy of more than 3 readings across more than 3 visits would leave significant probability (>25%; red zone) that the average measured office systolic BP was on the opposite side of the cutpoint from their “true” office systolic BP. Thus, one cannot reliably classify the “true” office systolic BP status of patients with these readings, and alternative approaches to classifying BP status such as ambulatory blood pressure monitoring or home blood pressure monitoring may be needed. We also examined the probability of correct classification using a threshold systolic BP of 130 mm Hg (Supplementary Figure 4). Our model shows the same pattern for the association between the measurement protocol and confidence in correct BP classification, with a greater number of readings needed to achieve >95% confidence as the BP is closer to the 130 mm Hg threshold. Similar pattern was observed using a threshold diastolic BP of 80 mm Hg (Supplementary Figure 5). DISCUSSION Guidelines and scientific statements recommend measuring BP in the office setting to identify patients with elevated BP and to monitor response to antihypertensive medications.5–7,21–23 However, the optimal protocol for determining which patients have an elevated BP in the office setting remains unclear.23 Our CFA yielded some interesting findings that can inform recommendations for office BP measurement protocols in clinical practice. First, with respect to number of readings, our analysis revealed that clinicians gain more confidence in office measurements from increasing the number of visits with readings than from increasing the number of readings per visit, consistent with Rosner and Polk’s prior investigation of this question.24 Further, there are diminishing returns from increasing the number of BP measurements beyond 2 or 3 per visit, or for increasing the number of visits beyond 2. These findings reinforce the recommendations by the American Society of Hypertension to estimate office BP by averaging BP readings from 2 consecutive office visits.25 Our CFA model also showed that dropping the first manual BP reading within a visit, as is recommended by some guidelines, had no benefit in terms of increasing the confidence of BP classifications, and even slightly decreased the probability of the measured office systolic BP being close to the “true” office BP for oscillometric measurements.6 In the case of manual BP measurements, the benefit of preventing the upward bias attributable to the initial reading by dropping the first manual BP reading was counterbalanced by the loss of reliability from having fewer total readings with the first one excluded. The finding for oscillometric measurements may be due to the fact that they were taken after the manual measurements, and therefore there was no tendency for the first oscillometric reading to be elevated, and hence no benefit from dropping this reading. Our CFA also demonstrated that there were minimal differences in the confidence of BP estimates obtained by manual mercury vs. oscillometric devices when both are measured using rigorous, guideline-concordant standards. However, it is important to recognize that prior studies have shown that in routine clinical practice, oscillometric BP measurements are usually more accurate than manual measurements as the automated oscillometric devices are less susceptible to human errors such as digit rounding and rapid cuff deflation.26,27 Finally, we learned that one cannot gain high confidence that office systolic BP is non-elevated among individuals with average systolic BP just below the 140/90 mm Hg cutpoint (e.g., 137–139 mm Hg), even after averaging multiple readings across multiple visits. Our findings provide support for recommendations to measure BP at least twice across 2 visits to confidently diagnose hypertension.7 However, in healthy patients who have few reasons to return for office visits, measuring BP 2 or 3 times within a single visit may be the most efficient, patient-centered strategy for determining who should be referred for out-of-office BP testing, as even with return visits, a substantial number of patients with average BP readings near the cutpoint for classifying BP will still have an indication for out-of-office BP testing. Although one prior study has examined the number of BP readings needed to confidently classify BP status, to our knowledge our study is the first to empirically examine this question in patients being screened for hypertension.28 The finding that out-of-office BP readings may be indicated for patients slightly below the cutpoint used to diagnose hypertension represents a significant change from the usual recommendations for referral to out-of-office BP testing during hypertension screening. These findings must be interpreted in the context of several possible limitations. First, all BPs were measured by trained nurses carefully adhering to a measurement protocol under no time pressure. Hence, the application of these findings to BPs measured in usual practice must be made with caution. Nevertheless, guidelines recommend that clinicians strive for high-quality clinic assessments. Second, one of the aims of this study was to compare manual and oscillometric methods. In the study protocol, manual readings were always taken prior to automatic readings, and this prevented any impact of an alerting response during initial readings on automatic readings. Other limitations included the absence of participants with initial screening systolic BP >160 mm Hg, inclusion of only a limited number of elderly participants, and the exclusion of patients with known cardiovascular disease or other serious medical conditions, thereby limiting the ability to extrapolate our results to these groups. Also, we did not compare our office BP measurements with out-of-office BP measures as our goal was to compare protocols for measuring office BP. Hence, conclusions about the extent to which office BP readings correspond to gold-standard out-of-office assessments cannot be made from the present analysis.29 Nevertheless, office BP measurement approaches are still needed to determine who should be referred for out-of-office testing, and office BP measurement still represents the predominant approach to BP measurement and hypertension diagnosis in clinical practice.30,31 Finally, there are emerging data supporting the use of unattended office BP measurements, and this approach was not compared in our study.32 PERSPECTIVES In summary, we used a state-of-the-art modeling approach to compare the confidence gained in classifying office BP by different approaches to office BP measurement. We learned that averaging one BP reading across 2 visits may best balance maximizing accuracy with efficiency of measurement during hypertension screening, though the exact recommended protocol may vary with the clinical context. These findings can be used to inform the development of office BP measurement protocols and to guide the indications for referring patients for ambulatory or home BP monitoring as part of hypertension screening. SUPPLEMENTARY MATERIAL Supplementary data are available at American Journal of Hypertension online. ACKNOWLEDGMENTS Prior to his untimely death in May 2009, Dr Thomas G. Pickering was the Co-PI of the Masked Hypertension Study and the PI of the program project that funds it. He played a critical role in the design of this study and long looked forward to our ability to empirically address the issues examined in this manuscript with high-quality data. J.E.S. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. This study was supported by grant P01 HL047540 from the National Heart, Lung, and Blood Institute (NHLBI). I.M.K. received support from the Agency for Healthcare Research and Quality (R01 HS024262) and the National Center for Advancing Translational Sciences (NCATS; U01 TR001873). D.S. received support from NHLBI (K24 HL125704). Additional support was provided by NCATS through grants M01 RR10710 (Stony Brook University) and UL1 TR000040 (formerly, UL1-RR024156; Columbia University), and through grant 15SFRN23480000 from the American Heart Association. The funders had no role in the study design, collection, analysis, or writing of the report. DISCLOSURE The authors declared no conflict of interest. REFERENCES 1. Chobanian AV. Shattuck Lecture. The hypertension paradox—more uncontrolled disease despite improved therapy. N Engl J Med  2009; 361: 878– 887. Google Scholar CrossRef Search ADS PubMed  2. Nwankwo T, Yoon SS, Burt V, Gu Q. Hypertension among adults in the United States: National Health and Nutrition Examination Survey, 2011–2012. NCHS Data Brief  2013; 1– 8. 3. Neaton JD, Wentworth D. Serum cholesterol, blood pressure, cigarette smoking, and death from coronary heart disease. Overall findings and differences by age for 316,099 white men. Multiple Risk Factor Intervention Trial Research Group. Arch Intern Med  1992; 152: 56– 64. Google Scholar CrossRef Search ADS PubMed  4. Yang Q, Cogswell ME, Flanders WD, Hong Y, Zhang Z, Loustalot F, Gillespie C, Merritt R, Hu FB. Trends in cardiovascular health metrics and associations with all-cause and CVD mortality among US adults. JAMA  2012; 307: 1273– 1283. Google Scholar CrossRef Search ADS PubMed  5. Siu AL; U.S. Preventive Services Task Force. Screening for high blood pressure in adults: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med  2015; 163: 778– 786. Google Scholar CrossRef Search ADS PubMed  6. Krause T, Lovibond K, Caulfield M, McCormack T, Williams B; Guideline Development Group. Management of hypertension: summary of NICE guidance. BMJ  2011; 343: d4891. Google Scholar CrossRef Search ADS PubMed  7. Whelton PK, Carey RM, Aronow WS, Casey DE, Collins KJ, Dennison Himmelfarb C, DePalma SM, Gidding S, Jamerson KA, Jones DW, MacLaughlin EJ, Muntner P, Ovbiagele B, Smith SC, Spencer CC, Stafford RS, Taler SJ, Thomas RJ, Williams KA, Williamson JD, Wright JT. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol  2017; e-pub ahead of print. 8. Campbell NR, Culleton BW, McKay DW. Misclassification of blood pressure by usual measurement in ambulatory physician practices. Am J Hypertens  2005; 18: 1522– 1527. Google Scholar CrossRef Search ADS PubMed  9. Spruill TM, Gerber LM, Schwartz JE, Pickering TG, Ogedegbe G. Race differences in the physical and psychological impact of hypertension labeling. Am J Hypertens  2012; 25: 458– 463. Google Scholar CrossRef Search ADS PubMed  10. Gonçalves CB, Moreira LB, Gus M, Fuchs FD. Adverse events of blood-pressure-lowering drugs: evidence of high incidence in a clinical setting. Eur J Clin Pharmacol  2007; 63: 973– 978. Google Scholar CrossRef Search ADS PubMed  11. Pickering TG, Hall JE, Appel LJ, Falkner BE, Graves J, Hill MN, Jones DW, Kurtz T, Sheps SG, Roccella EJ. Recommendations for blood pressure measurement in humans and experimental animals: part 1: blood pressure measurement in humans: a statement for professionals from the Subcommittee of Professional and Public Education of the American Heart Association Council on High Blood Pressure Research. Circulation  2005; 111: 697– 716. Google Scholar CrossRef Search ADS PubMed  12. O’Brien E, Asmar R, Beilin L, Imai Y, Mancia G, Mengden T, Myers M, Padfield P, Palatini P, Parati G, Pickering T, Redon J, Staessen J, Stergiou G, Verdecchia P; European Society of Hypertension Working Group on Blood Pressure Monitoring. Practice guidelines of the European Society of Hypertension for clinic, ambulatory and self blood pressure measurement. J Hypertens  2005; 23: 697– 701. Google Scholar CrossRef Search ADS PubMed  13. Pickering TG. The ninth Sir George Pickering Memorial Lecture. Ambulatory monitoring and the definition of hypertension. J Hypertens  1992; 10: 401– 409. Google Scholar CrossRef Search ADS PubMed  14. Pickering TG. Measurement of blood pressure in and out of the office. J Clin Hypertens (Greenwich)  2005; 7: 123– 129. Google Scholar CrossRef Search ADS PubMed  15. Shimbo D, Newman JD, Schwartz JE. Masked hypertension and prehypertension: diagnostic overlap and interrelationships with left ventricular mass: the Masked Hypertension Study. Am J Hypertens  2012; 25: 664– 671. Google Scholar CrossRef Search ADS PubMed  16. Brown TA. Confirmatory Factor Analysis for Applied Research , 2nd edn. Guilford Press: New York, 2015. 17. Burnier M, Gasser UE. End-digit preference in general practice: a comparison of the conventional auscultatory and electronic oscillometric methods. Blood Press  2008; 17: 104– 109. Google Scholar CrossRef Search ADS PubMed  18. Myers MG, Valdivieso M, Kiss A. Use of automated office blood pressure measurement to reduce the white coat response. J Hypertens  2009; 27: 280– 286. Google Scholar CrossRef Search ADS PubMed  19. Mancia G, Bertinieri G, Grassi G, Parati G, Pomidossi G, Ferrari A, Gregorini L, Zanchetti A. Effects of blood-pressure measurement by the doctor on patient’s blood pressure and heart rate. Lancet  1983; 2: 695– 698. Google Scholar CrossRef Search ADS PubMed  20. Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Modeling  1999; 6: 1– 55. Google Scholar CrossRef Search ADS   21. Mancia G, Fagard R, Narkiewicz K, Redón J, Zanchetti A, Böhm M, Christiaens T, Cifkova R, De Backer G, Dominiczak A, Galderisi M, Grobbee DE, Jaarsma T, Kirchhof P, Kjeldsen SE, Laurent S, Manolis AJ, Nilsson PM, Ruilope LM, Schmieder RE, Sirnes PA, Sleight P, Viigimaa M, Waeber B, Zannad F; Task Force Members. 2013 ESH/ESC Guidelines for the management of arterial hypertension: the Task Force for the management of arterial hypertension of the European Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). J Hypertens  2013; 31: 1281– 1357. Google Scholar CrossRef Search ADS PubMed  22. Pickering TG, White WB; American Society of Hypertension Writing Group. ASH Position Paper: home and ambulatory blood pressure monitoring. When and how to use self (home) and ambulatory blood pressure monitoring. J Clin Hypertens (Greenwich)  2008; 10: 850– 855. Google Scholar CrossRef Search ADS PubMed  23. Piper MA, Evans CV, Burda BU, Margolis KL, O’Connor E, Whitlock EP. Diagnostic and predictive accuracy of blood pressure screening methods with consideration of rescreening intervals: a systematic review for the U.S. Preventive Services Task Force. Ann Intern Med  2015; 162: 192– 204. Google Scholar CrossRef Search ADS PubMed  24. Rosner B, Polk BF. The implications of blood pressure variability for clinical and screening purposes. J Chronic Dis  1979; 32: 451– 461. Google Scholar CrossRef Search ADS PubMed  25. Weber MA, Schiffrin EL, White WB, Mann S, Lindholm LH, Kenerson JG, Flack JM, Carter BL, Materson BJ, Ram CV, Cohen DL, Cadet JC, Jean-Charles RR, Taler S, Kountz D, Townsend RR, Chalmers J, Ramirez AJ, Bakris GL, Wang J, Schutte AE, Bisognano JD, Touyz RM, Sica D, Harrap SB. Clinical practice guidelines for the management of hypertension in the community: a statement by the American Society of Hypertension and the International Society of Hypertension. J Clin Hypertens (Greenwich)  2014; 16: 14– 26. Google Scholar CrossRef Search ADS PubMed  26. Beckett L, Godwin M. The BpTRU automatic blood pressure monitor compared to 24 hour ambulatory blood pressure monitoring in the assessment of blood pressure in patients with hypertension. BMC Cardiovasc Disord  2005; 5: 18. Google Scholar CrossRef Search ADS PubMed  27. Myers MG, Godwin M, Dawes M, Kiss A, Tobe SW, Grant FC, Kaczorowski J. Conventional versus automated measurement of blood pressure in primary care patients with systolic hypertension: randomised parallel design controlled trial. BMJ  2011; 342: d286. Google Scholar CrossRef Search ADS PubMed  28. Powers BJ, Olsen MK, Smith VA, Woolson RF, Bosworth HB, Oddone EZ. Measuring blood pressure for decision making and quality reporting: where and how many measures? Ann Intern Med  2011; 154: 781– 788. Google Scholar CrossRef Search ADS PubMed  29. Sheppard JP, Stevens R, Gill P, Martin U, Godwin M, Hanley J, Heneghan C, Hobbs FD, Mant J, McKinstry B, Myers M, Nunan D, Ward A, Williams B, McManus RJ. Predicting Out-of-Office Blood Pressure in the Clinic (PROOF-BP): derivation and validation of a tool to improve the accuracy of blood pressure measurement in clinical practice. Hypertension  2016; 67: 941– 950. Google Scholar CrossRef Search ADS PubMed  30. Shimbo D, Abdalla M, Falzon L, Townsend RR, Muntner P. Role of ambulatory and home blood pressure monitoring in clinical practice: a narrative review. Ann Intern Med  2015; 163: 691– 700. Google Scholar CrossRef Search ADS PubMed  31. Shimbo D, Kent ST, Diaz KM, Huang L, Viera AJ, Kilgore M, Oparil S, Muntner P. The use of ambulatory blood pressure monitoring among Medicare beneficiaries in 2007–2010. J Am Soc Hypertens  2014; 8: 891– 897. Google Scholar CrossRef Search ADS PubMed  32. Kjeldsen SE, Lund-Johansen P, Nilsson PM, Mancia G. Unattended blood pressure measurements in the systolic blood pressure intervention trial: implications for entry and achieved blood pressure values compared with other trials. Hypertension  2016; 67: 808– 812. Google Scholar CrossRef Search ADS PubMed  © American Journal of Hypertension, Ltd 2018. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Hypertension Oxford University Press

A Comparison of the Diagnostic Accuracy of Common Office Blood Pressure Measurement Protocols

Loading next page...
 
/lp/ou_press/a-comparison-of-the-diagnostic-accuracy-of-common-office-blood-GK6CW8K0Cy
Publisher
Oxford University Press
Copyright
© American Journal of Hypertension, Ltd 2018. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
0895-7061
eISSN
1941-7225
D.O.I.
10.1093/ajh/hpy053
Publisher site
See Article on Publisher Site

Abstract

Abstract BACKGROUND The optimal approach to measuring office blood pressure (BP) is uncertain. We aimed to compare BP measurement protocols that differed based on numbers of readings within and between visits and by assessment method. METHODS We enrolled a sample of 707 employees without known hypertension or cardiovascular disease, and obtained 6 standardized BP readings during each of 3 office visits at least 1 week apart, using mercury sphygmomanometer and BpTRU oscillometric devices (18 readings per participant) for a total of 12,645 readings. We used confirmatory factor analysis to develop a model estimating “true” office BP that could be used to compare the probability of correctly classifying participants’ office BP status using differing numbers and types of office BP readings. RESULTS Averaging 2 systolic BP readings across 2 visits correctly classified participants as having BP below or above the 140 mm Hg threshold at least 95% of the time if the averaged reading was <134 or >149 mm Hg, respectively. Our model demonstrated that more confidence was gained by increasing the number of visits with readings than by increasing the number of readings within a visit. No clinically significant confidence was gained by dropping the first reading vs. averaging all readings, nor by measuring with a manual mercury device vs. with an automated oscillometric device. CONCLUSIONS Averaging 2 BP readings across 2 office visits appeared to best balance increased confidence in office BP status with efficiency of BP measurement, though the preferred measurement strategy may vary with the clinical context. blood pressure, hypertension, measurement, screening Hypertension is the most common chronic medical condition in the United States, affecting nearly one-third of the adult population,1,2 and a leading risk factor for cardiovascular disease and mortality.3,4 Guidelines recommend screening for hypertension by first measuring office blood pressure (BP), and then obtaining confirmatory out-of-office testing using either 24-hour ambulatory blood pressure monitoring or home blood pressure monitoring.5–7 Misclassifying a patient as having elevated office BP can lead to inappropriate diagnostic labeling, overtreatment with antihypertensive medications, and unnecessary health care and insurance costs.8–10 Conversely, misclassifying a patient as having non-elevated office BP may lead to undertreatment of hypertension. Despite the importance of its accurate measurement, there is no consensus among guidelines on the optimal number of readings or the best device for measuring office BP.11,12 Office BP measurement is used to approximate an individual’s average, or usual, resting office BP, referred to here as the “true office BP.” Confirmatory factor analysis (CFA) is a statistical method that can be used to model the relationships of multiple office BP readings with the “true office BP.”13 Thus, we developed a CFA model that enables us to compare the diagnostic accuracy of common approaches to measuring office BP including: (i) increasing the number of BP readings taken within a visit; (ii) increasing the number of office visits in which BP readings are taken; (iii) averaging 3 BP readings taken in a single visit vs. dropping the first of 3 readings and averaging the rest; and (iv) using manual vs. automatic BP devices.14 METHODS Study population We analyzed data from the 707 participants enrolled at the Stony Brook University site of the Masked Hypertension Study, a community-based study of the prevalence, predictors, and prognosis of masked hypertension.15 Participants were recruited by conducting BP screenings for employees of Stony Brook University, Stony Brook University Hospital, and a small financial firm. At the time of the screening, employees were briefly told about the study. Potentially eligible employees who were interested in participating were subsequently contacted by telephone to confirm their eligibility. Those who were eligible and chose to participate were formally consented at their first study visit. The institutional review boards of Stony Brook University and Columbia University Medical Center approved all procedures for this study. Employees were eligible if they were 18 years of age or older, spoke and read English, were employed more than 20 hours per week, worked on at least 2 consecutive days per week, were not taking a hypertension medication, and had a screening BP <160/105 mm Hg (average of second and third readings). Employees were ineligible if they self-reported a history of cardiovascular disease, or other major chronic medical condition (i.e., kidney, liver, adrenal, thyroid disorder; organ transplant; cancer not in remission more than 6 months); were prescribed antihypertensive or any other cardiovascular medications other than statins; or took any medications known to influence BP. Other exclusions included current or planned pregnancy, severe psychiatric disorder, active substance abuse, and unavailability for follow-up during the 3 months after screening. Smoking, hyperlipidemia, type II diabetes mellitus, and a past history of untreated hypertension were not exclusion criteria. Procedure Participants attended 4 study visits over a 4-week period. During the first 3 visits, each approximately 1 week apart, office BP was assessed using a standardized protocol in accordance with the American Heart Association’s guidelines.11 At each office visit, study personnel first confirmed that participants had not eaten, smoked, or consumed a caffeinated beverage during the prior 30 minutes. A trained nurse then measured the circumference of the participant’s non-dominant arm with a Gullick II tape measure and chose the appropriately sized cuff. After the participant had been seated comfortably with their feet on the floor for at least 5 minutes, a mercury sphygmomanometer (Baum, Copiague, NY) was used to take 3 manual BP readings. The nurse waited 1–2 minutes between each reading. After the third reading, the nurse waited another 1–2 minutes and then took another 3 readings, each 2 minutes apart, using the BpTRU oscillometric device (BpTRU Medical Devices, Coquitlam, BC) on the same non-dominant arm. The nurse remained in the same room as the patient at the time of the automated readings. This protocol generated 18 office BP readings for each participant distributed across 3 visits (3 manual followed by 3 oscillometric readings per visit). Participants also completed a questionnaire that included sociodemographic characteristics. At the fourth visit, participants completed a medical history interview and had their height and weight measured in standardized fashion for calculation of body mass index. Statistical analysis We used CFA to model the relationships among the 18 office BP readings obtained by the sphygmomanometer and the oscillometric device (Figure 1). CFA is a multivariable statistical technique used to evaluate the relationship of multiple measured variables to the underlying constructs they are presumed to represent.16 Our CFA accounted for 4 main sources of variance among the individual BP readings. First and foremost, individuals differ in their “true office BP.” Second, office BPs measured at different visits may vary systematically from the “true office BP” due to a variety of unspecified visit-specific factors (e.g., recent salt intake, stress, emotional factors). Third, office BP readings often differ from each other (and the visit-specific “true BP”) due to random “noise.” By averaging multiple office readings during a single visit, the random fluctuations between readings average out and one arrives at a better estimate of the person’s “true BP” for that visit. Last, a single office BP reading may vary from the “true office BP” due to systematic upward (or downward) biases in measurement. Specifically, an alerting response found in initial readings may lead to a difference between the first and subsequent measurements, and this was explicitly incorporated into our model.17–19 Our model yields estimates of these different sources of variance that determine the individual office BP readings. Our model also accounts for potential systematic differences between manual and automatic office BP in means and variance components. Figure 1. View largeDownload slide Confirmatory factor analysis model of “true blood pressure” based on observed readings. The circle containing the text “Visit 1 BPman” represents the “true” average of the person’s manual BP during the 5 minutes manual BP readings were being taken at a single office visit. The rectangular boxes (BPM11, BPM12, and BPM13) represent the 3 observed manual BP readings taken at Visit 1. These readings are expected to be similar to “Visit 1 BPman” as they each depend primarily on the “true” average BP during those few minutes, represented by the arrows from “Visit 1 BPman” to each box. The fact that the 3 readings are not identical is portrayed by the 3 “ε” terms representing the fluctuations of the individual readings around the “true” average, including any random measurement error in the taking of the reading. The same logic applies to repeat visits (i.e., “Visit 2 BPman” and “Visit 3 BPman”). While not agreeing perfectly, “Visit 1 BPman,” “Visit 2 BPman,” and “Visit 3 BPmal” are also expected to be highly correlated because they each depend on ‘“true” Manual BP; the visit-to-visit fluctuations are portrayed by λM1, λM2, and λM3. There is an exactly parallel model for the automatic (BpTRU) measurements. The 2 halves of the overall model are linked by curved arrows representing the following correlations: the ‘“true” Manual BP is very highly correlated with the “true” Automatic BP; and whatever it is that makes someone’s Visit 1 manual BP deviate from their “true” manual office BP will tend to make their Visit 1 Automatic BP similarly deviate from their “true” Automatic BP (λM1 is expected to be positively correlated with λA1, and similarly for λM2 with λA2 and λM3 with λA3). Abbreviations: auto, automatic; BP, blood pressure; man, manual. Figure 1. View largeDownload slide Confirmatory factor analysis model of “true blood pressure” based on observed readings. The circle containing the text “Visit 1 BPman” represents the “true” average of the person’s manual BP during the 5 minutes manual BP readings were being taken at a single office visit. The rectangular boxes (BPM11, BPM12, and BPM13) represent the 3 observed manual BP readings taken at Visit 1. These readings are expected to be similar to “Visit 1 BPman” as they each depend primarily on the “true” average BP during those few minutes, represented by the arrows from “Visit 1 BPman” to each box. The fact that the 3 readings are not identical is portrayed by the 3 “ε” terms representing the fluctuations of the individual readings around the “true” average, including any random measurement error in the taking of the reading. The same logic applies to repeat visits (i.e., “Visit 2 BPman” and “Visit 3 BPman”). While not agreeing perfectly, “Visit 1 BPman,” “Visit 2 BPman,” and “Visit 3 BPmal” are also expected to be highly correlated because they each depend on ‘“true” Manual BP; the visit-to-visit fluctuations are portrayed by λM1, λM2, and λM3. There is an exactly parallel model for the automatic (BpTRU) measurements. The 2 halves of the overall model are linked by curved arrows representing the following correlations: the ‘“true” Manual BP is very highly correlated with the “true” Automatic BP; and whatever it is that makes someone’s Visit 1 manual BP deviate from their “true” manual office BP will tend to make their Visit 1 Automatic BP similarly deviate from their “true” Automatic BP (λM1 is expected to be positively correlated with λA1, and similarly for λM2 with λA2 and λM3 with λA3). Abbreviations: auto, automatic; BP, blood pressure; man, manual. With this conceptual model in place, we initially specified a maximally parsimonious CFA model (Figure 1) in which only 6 parameters were used to model the 18 means, 18 SDs, and 153 correlations among the 9 manual and 9 automatic office BP readings. Constraints on the model were removed in a conceptually guided, step-wise manner—and the model was made increasingly complex—if lifting constraints substantially improved the fit of the model. This model was estimated by full information maximum likelihood using the M-Plus software (version 6.1; Muthén & Muthén, Los Angeles, CA) and all available data under the assumption that any missing data were missing at random. We used multiple model fit indices along with standard recommendations for defining an “acceptable” model to assess model fit (Supplementary Tables 1and2).20 After arriving at a parsimonious model that adequately fit the data, model estimates (Supplementary Table 3) were used to generate estimates of the precision with which commonly used BP measurement protocols would approximate the “true office BP.” These estimates were then used to determine the probability that office BP, measured by a given protocol, would correctly classify an individual’s “true office BP” status as elevated or non-elevated using a cutpoint of 130/80 mm Hg, and separately 140/90 mm Hg. These cutpoints represent the thresholds for diagnosing stage I and stage II hypertension, respectively, as per the 2017 American College of Cardiology/American Heart Association high BP guidelines.7 Separate CFA models were estimated for systolic and diastolic BP. SAS (version 9.4, Cary, NC) was used to generate descriptive statistics for the sample. RESULTS The mean (SD) age of participants was 45.4 (10.2) years. Sixty percent of participants were women, 6% were African-American, and 7% were Hispanic (Table 1). Of the 6,363 planned manual and automatic BP readings, there were only 0.05% and 1.23% missing data, respectively. The average manual BP at the initial visit (average of 3 manual readings) was 116/76 mm Hg. The overall average manual BP (average of 9 manual readings) was 116/75 mm Hg, and the overall average oscillometric BP was nearly identical, 115/75 mm Hg (average of 9 oscillometric readings). Table 1. Demographic and clinical characteristics of participantsa Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Abbreviation: BP, blood pressure. aBlood pressure characteristics were based on the average of 3 manual readings during Visit 1. Participants who withdrew from the study prior to the fifth visit (n = 44) were missing data pertaining to body mass index, smoking status, and diabetes, and some (n = 18) were missing years of education. One participant’s Visit 1 blood pressure readings were lost prior to data entry. View Large Table 1. Demographic and clinical characteristics of participantsa Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Characteristic  N  Mean (SD) or %  Age  707  45.4 (10.2)   20–39 years    27.6%   40–59 years    66.6%   ≥60 years (max = 73)    5.8%  Female  707  60.1%  Hispanic  707  6.5%  Race  707     White    82.3%   Black    5.9%   Asian    7.8%   Other    4.0%  Years of education, mean (SD)  689  16.4 (3.0)   ≤12 years    8.9%   13–15 years    26.0%   16+ years    65.2%  Body mass index (kg/m2)  663  27.7 (5.3)   <25    34.8%   25–29    35.7%   ≥30    29.4%  Smoking status  663     Never smoked    66.4%   Past smoker    25.9%   Current smoker    7.7%  Diabetes  663  3.5%  Systolic BP  706  115.8 (12.5)   <120 mm Hg    63.2%   120–139 mm Hg    33.4%   ≥140 mm Hg    3.4%  Diastolic BP  706  75.6 (8.6)   <80 mm Hg    71.0%   80–89 mm Hg    23.4%   ≥90 mm Hg    5.7%  BP ≥ 140/90 mm Hg  706  7.2%  Abbreviation: BP, blood pressure. aBlood pressure characteristics were based on the average of 3 manual readings during Visit 1. Participants who withdrew from the study prior to the fifth visit (n = 44) were missing data pertaining to body mass index, smoking status, and diabetes, and some (n = 18) were missing years of education. One participant’s Visit 1 blood pressure readings were lost prior to data entry. View Large Increasing the number of readings within visits and/or the number of visits The gain in confidence obtained by increasing the number of BP measurements/visit and/or increasing the number of visits is portrayed in Figure 2. Overall, one gained more confidence that a measured systolic BP was within 5 mm Hg of the “true” office systolic BP by increasing the number of visits as compared to increasing the number of readings within a visit. For example, when increasing the number of systolic BP readings within a single visit from 1 to 2, the likelihood that the measured systolic BP was within 5 mm Hg of the “true” office systolic BP increased by 9% for manual measurement and 7% for oscillometric measurement. In contrast, adding a second visit, again with one reading, increased the confidence by 14% for manual measurement and 16% for oscillometric measurement. There were diminishing returns from increasing the number of systolic BP readings within a visit from 2 to 3, or from increasing the number of visits from 2 to 3. For example, increasing the number of manual measurements within a single visit from 2 to 3 only increased the likelihood that the measured systolic BP was within 5 mm Hg of the “true” systolic BP by 4% for manual measurement and 3% for oscillometric measurement. The same pattern was present for diastolic BP (Supplementary Figure 1). Figure 2. View largeDownload slide Probability that the average of systolic blood pressure readings is within 5 mm Hg of the “true” office systolic blood pressure, by number of readings per visit and number of visits. Abbreviations: Avg, Average; BpTRU, an automatic oscillometric blood pressure device. Figure 2. View largeDownload slide Probability that the average of systolic blood pressure readings is within 5 mm Hg of the “true” office systolic blood pressure, by number of readings per visit and number of visits. Abbreviations: Avg, Average; BpTRU, an automatic oscillometric blood pressure device. Dropping the first of 3 BP readings and averaging the second and third readings The first manual systolic BP reading was 1.3 mm Hg higher than the overall average systolic BP. Figure 2 shows the difference in probability of being within 5 mm Hg of the “true” office systolic BP for a given number of visits when dropping the first of 3 BP readings and averaging the latter 2 as compared to averaging all 3 readings. Dropping the first of 3 manual BP readings did not substantially alter the precision of classification, but there was reduced precision for oscillometric measurements. The same pattern was found for diastolic BP (see Supplementary Figure 1). Manual vs. oscillometric measurement Figure 3 shows a comparison of the probability of being within 5 or 10 mm Hg of the “true” office systolic BP when using standardized, guideline-concordant manual and oscillometric BP measurements (i.e., 2 visits, 3 readings per visit, removing the first reading of each visit and average the remaining 2 readings11). The research quality manual readings were more likely (6%) to correctly classify “true” systolic BP status than the automatic readings. There was no appreciable difference between the 2 methods for diastolic BP (Supplementary Figure 2). Figure 3. View largeDownload slide Probability that the average measured systolic blood pressure is within 5 or 10 mm Hg of the “true” office systolic blood pressure for Manual vs. Automatic BpTRU measurement methods. Systolic blood pressure is assessed as the average of blood pressure readings from 2 visits with 3 readings per visit and the first reading of each visit dropped. Figure 3. View largeDownload slide Probability that the average measured systolic blood pressure is within 5 or 10 mm Hg of the “true” office systolic blood pressure for Manual vs. Automatic BpTRU measurement methods. Systolic blood pressure is assessed as the average of blood pressure readings from 2 visits with 3 readings per visit and the first reading of each visit dropped. Comparison of the probability of correct office BP classification according to different office BP measurement protocols Figure 4 shows the probability that an average office systolic BP will correctly classify a patient’s office systolic BP as elevated or non-elevated using a cutpoint of 140/90 mm Hg depending on the measurement protocol (combination of number of visits and number of readings per visit). If the average BP based on the indicated protocol is in the green zone, then the patient and clinician can be >95% confident that a diagnosis based on this average will be correct; stated differently, there is <5% chance that the true office systolic BP is on the opposite side of 140 mm Hg. If the average BP is in the yellow zone, then one’s confidence that a diagnosis based on this BP is correct is 75–95%; it will be wrong 5–25% of the time. Finally, if the average is in the red zone, then a diagnosis based on the observed average will be incorrect >25% of the time. Although data for this analysis were taken from BP readings from 3 office visits, once specified, the model can provide probability estimates of correct classification based on any number of visits and readings. Figure 4. View largeDownload slide Probability of correct classification of office blood pressure as elevated or non-elevated according to observed average systolic blood pressures obtained with different measurement strategies. Probabilities are based on estimates derived from the confirmatory factor analysis model. Although the data for this analysis were taken from blood pressure (BP) readings from 3 office visits, once specified, the model can provide probability estimates of correct classification of office BP based on an even larger number of visits and readings. Accordingly, the figure shows estimates for the probability of correct classification of office BP for BP measurement protocols with more than 3 visits. Figure 4. View largeDownload slide Probability of correct classification of office blood pressure as elevated or non-elevated according to observed average systolic blood pressures obtained with different measurement strategies. Probabilities are based on estimates derived from the confirmatory factor analysis model. Although the data for this analysis were taken from blood pressure (BP) readings from 3 office visits, once specified, the model can provide probability estimates of correct classification of office BP based on an even larger number of visits and readings. Accordingly, the figure shows estimates for the probability of correct classification of office BP for BP measurement protocols with more than 3 visits. The figure shows that there is greater benefit, in terms of increasing the range of office systolic BPs with high confidence classifications (green zone), from increasing the number of visits than from increasing the number of readings per visits. For example, using a protocol with a single reading at a single visit, a systolic BP of 150 mm Hg would provide only intermediate confidence (yellow zone) that the true office systolic BP was elevated whereas using a protocol with a single reading per visit across 2 visits, a systolic BP of 150 mm Hg would provide >95% confidence that the true office systolic BP was elevated (green zone). A similar pattern was observed for office diastolic BP classification (Supplementary Figure 3). For patients with average measured systolic BPs closer to the 140 mm Hg cutpoint, roughly between 137 and 143 mm Hg, even a strategy of more than 3 readings across more than 3 visits would leave significant probability (>25%; red zone) that the average measured office systolic BP was on the opposite side of the cutpoint from their “true” office systolic BP. Thus, one cannot reliably classify the “true” office systolic BP status of patients with these readings, and alternative approaches to classifying BP status such as ambulatory blood pressure monitoring or home blood pressure monitoring may be needed. We also examined the probability of correct classification using a threshold systolic BP of 130 mm Hg (Supplementary Figure 4). Our model shows the same pattern for the association between the measurement protocol and confidence in correct BP classification, with a greater number of readings needed to achieve >95% confidence as the BP is closer to the 130 mm Hg threshold. Similar pattern was observed using a threshold diastolic BP of 80 mm Hg (Supplementary Figure 5). DISCUSSION Guidelines and scientific statements recommend measuring BP in the office setting to identify patients with elevated BP and to monitor response to antihypertensive medications.5–7,21–23 However, the optimal protocol for determining which patients have an elevated BP in the office setting remains unclear.23 Our CFA yielded some interesting findings that can inform recommendations for office BP measurement protocols in clinical practice. First, with respect to number of readings, our analysis revealed that clinicians gain more confidence in office measurements from increasing the number of visits with readings than from increasing the number of readings per visit, consistent with Rosner and Polk’s prior investigation of this question.24 Further, there are diminishing returns from increasing the number of BP measurements beyond 2 or 3 per visit, or for increasing the number of visits beyond 2. These findings reinforce the recommendations by the American Society of Hypertension to estimate office BP by averaging BP readings from 2 consecutive office visits.25 Our CFA model also showed that dropping the first manual BP reading within a visit, as is recommended by some guidelines, had no benefit in terms of increasing the confidence of BP classifications, and even slightly decreased the probability of the measured office systolic BP being close to the “true” office BP for oscillometric measurements.6 In the case of manual BP measurements, the benefit of preventing the upward bias attributable to the initial reading by dropping the first manual BP reading was counterbalanced by the loss of reliability from having fewer total readings with the first one excluded. The finding for oscillometric measurements may be due to the fact that they were taken after the manual measurements, and therefore there was no tendency for the first oscillometric reading to be elevated, and hence no benefit from dropping this reading. Our CFA also demonstrated that there were minimal differences in the confidence of BP estimates obtained by manual mercury vs. oscillometric devices when both are measured using rigorous, guideline-concordant standards. However, it is important to recognize that prior studies have shown that in routine clinical practice, oscillometric BP measurements are usually more accurate than manual measurements as the automated oscillometric devices are less susceptible to human errors such as digit rounding and rapid cuff deflation.26,27 Finally, we learned that one cannot gain high confidence that office systolic BP is non-elevated among individuals with average systolic BP just below the 140/90 mm Hg cutpoint (e.g., 137–139 mm Hg), even after averaging multiple readings across multiple visits. Our findings provide support for recommendations to measure BP at least twice across 2 visits to confidently diagnose hypertension.7 However, in healthy patients who have few reasons to return for office visits, measuring BP 2 or 3 times within a single visit may be the most efficient, patient-centered strategy for determining who should be referred for out-of-office BP testing, as even with return visits, a substantial number of patients with average BP readings near the cutpoint for classifying BP will still have an indication for out-of-office BP testing. Although one prior study has examined the number of BP readings needed to confidently classify BP status, to our knowledge our study is the first to empirically examine this question in patients being screened for hypertension.28 The finding that out-of-office BP readings may be indicated for patients slightly below the cutpoint used to diagnose hypertension represents a significant change from the usual recommendations for referral to out-of-office BP testing during hypertension screening. These findings must be interpreted in the context of several possible limitations. First, all BPs were measured by trained nurses carefully adhering to a measurement protocol under no time pressure. Hence, the application of these findings to BPs measured in usual practice must be made with caution. Nevertheless, guidelines recommend that clinicians strive for high-quality clinic assessments. Second, one of the aims of this study was to compare manual and oscillometric methods. In the study protocol, manual readings were always taken prior to automatic readings, and this prevented any impact of an alerting response during initial readings on automatic readings. Other limitations included the absence of participants with initial screening systolic BP >160 mm Hg, inclusion of only a limited number of elderly participants, and the exclusion of patients with known cardiovascular disease or other serious medical conditions, thereby limiting the ability to extrapolate our results to these groups. Also, we did not compare our office BP measurements with out-of-office BP measures as our goal was to compare protocols for measuring office BP. Hence, conclusions about the extent to which office BP readings correspond to gold-standard out-of-office assessments cannot be made from the present analysis.29 Nevertheless, office BP measurement approaches are still needed to determine who should be referred for out-of-office testing, and office BP measurement still represents the predominant approach to BP measurement and hypertension diagnosis in clinical practice.30,31 Finally, there are emerging data supporting the use of unattended office BP measurements, and this approach was not compared in our study.32 PERSPECTIVES In summary, we used a state-of-the-art modeling approach to compare the confidence gained in classifying office BP by different approaches to office BP measurement. We learned that averaging one BP reading across 2 visits may best balance maximizing accuracy with efficiency of measurement during hypertension screening, though the exact recommended protocol may vary with the clinical context. These findings can be used to inform the development of office BP measurement protocols and to guide the indications for referring patients for ambulatory or home BP monitoring as part of hypertension screening. SUPPLEMENTARY MATERIAL Supplementary data are available at American Journal of Hypertension online. ACKNOWLEDGMENTS Prior to his untimely death in May 2009, Dr Thomas G. Pickering was the Co-PI of the Masked Hypertension Study and the PI of the program project that funds it. He played a critical role in the design of this study and long looked forward to our ability to empirically address the issues examined in this manuscript with high-quality data. J.E.S. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. This study was supported by grant P01 HL047540 from the National Heart, Lung, and Blood Institute (NHLBI). I.M.K. received support from the Agency for Healthcare Research and Quality (R01 HS024262) and the National Center for Advancing Translational Sciences (NCATS; U01 TR001873). D.S. received support from NHLBI (K24 HL125704). Additional support was provided by NCATS through grants M01 RR10710 (Stony Brook University) and UL1 TR000040 (formerly, UL1-RR024156; Columbia University), and through grant 15SFRN23480000 from the American Heart Association. The funders had no role in the study design, collection, analysis, or writing of the report. DISCLOSURE The authors declared no conflict of interest. REFERENCES 1. Chobanian AV. Shattuck Lecture. The hypertension paradox—more uncontrolled disease despite improved therapy. N Engl J Med  2009; 361: 878– 887. Google Scholar CrossRef Search ADS PubMed  2. Nwankwo T, Yoon SS, Burt V, Gu Q. Hypertension among adults in the United States: National Health and Nutrition Examination Survey, 2011–2012. NCHS Data Brief  2013; 1– 8. 3. Neaton JD, Wentworth D. Serum cholesterol, blood pressure, cigarette smoking, and death from coronary heart disease. Overall findings and differences by age for 316,099 white men. Multiple Risk Factor Intervention Trial Research Group. Arch Intern Med  1992; 152: 56– 64. Google Scholar CrossRef Search ADS PubMed  4. Yang Q, Cogswell ME, Flanders WD, Hong Y, Zhang Z, Loustalot F, Gillespie C, Merritt R, Hu FB. Trends in cardiovascular health metrics and associations with all-cause and CVD mortality among US adults. JAMA  2012; 307: 1273– 1283. Google Scholar CrossRef Search ADS PubMed  5. Siu AL; U.S. Preventive Services Task Force. Screening for high blood pressure in adults: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med  2015; 163: 778– 786. Google Scholar CrossRef Search ADS PubMed  6. Krause T, Lovibond K, Caulfield M, McCormack T, Williams B; Guideline Development Group. Management of hypertension: summary of NICE guidance. BMJ  2011; 343: d4891. Google Scholar CrossRef Search ADS PubMed  7. Whelton PK, Carey RM, Aronow WS, Casey DE, Collins KJ, Dennison Himmelfarb C, DePalma SM, Gidding S, Jamerson KA, Jones DW, MacLaughlin EJ, Muntner P, Ovbiagele B, Smith SC, Spencer CC, Stafford RS, Taler SJ, Thomas RJ, Williams KA, Williamson JD, Wright JT. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol  2017; e-pub ahead of print. 8. Campbell NR, Culleton BW, McKay DW. Misclassification of blood pressure by usual measurement in ambulatory physician practices. Am J Hypertens  2005; 18: 1522– 1527. Google Scholar CrossRef Search ADS PubMed  9. Spruill TM, Gerber LM, Schwartz JE, Pickering TG, Ogedegbe G. Race differences in the physical and psychological impact of hypertension labeling. Am J Hypertens  2012; 25: 458– 463. Google Scholar CrossRef Search ADS PubMed  10. Gonçalves CB, Moreira LB, Gus M, Fuchs FD. Adverse events of blood-pressure-lowering drugs: evidence of high incidence in a clinical setting. Eur J Clin Pharmacol  2007; 63: 973– 978. Google Scholar CrossRef Search ADS PubMed  11. Pickering TG, Hall JE, Appel LJ, Falkner BE, Graves J, Hill MN, Jones DW, Kurtz T, Sheps SG, Roccella EJ. Recommendations for blood pressure measurement in humans and experimental animals: part 1: blood pressure measurement in humans: a statement for professionals from the Subcommittee of Professional and Public Education of the American Heart Association Council on High Blood Pressure Research. Circulation  2005; 111: 697– 716. Google Scholar CrossRef Search ADS PubMed  12. O’Brien E, Asmar R, Beilin L, Imai Y, Mancia G, Mengden T, Myers M, Padfield P, Palatini P, Parati G, Pickering T, Redon J, Staessen J, Stergiou G, Verdecchia P; European Society of Hypertension Working Group on Blood Pressure Monitoring. Practice guidelines of the European Society of Hypertension for clinic, ambulatory and self blood pressure measurement. J Hypertens  2005; 23: 697– 701. Google Scholar CrossRef Search ADS PubMed  13. Pickering TG. The ninth Sir George Pickering Memorial Lecture. Ambulatory monitoring and the definition of hypertension. J Hypertens  1992; 10: 401– 409. Google Scholar CrossRef Search ADS PubMed  14. Pickering TG. Measurement of blood pressure in and out of the office. J Clin Hypertens (Greenwich)  2005; 7: 123– 129. Google Scholar CrossRef Search ADS PubMed  15. Shimbo D, Newman JD, Schwartz JE. Masked hypertension and prehypertension: diagnostic overlap and interrelationships with left ventricular mass: the Masked Hypertension Study. Am J Hypertens  2012; 25: 664– 671. Google Scholar CrossRef Search ADS PubMed  16. Brown TA. Confirmatory Factor Analysis for Applied Research , 2nd edn. Guilford Press: New York, 2015. 17. Burnier M, Gasser UE. End-digit preference in general practice: a comparison of the conventional auscultatory and electronic oscillometric methods. Blood Press  2008; 17: 104– 109. Google Scholar CrossRef Search ADS PubMed  18. Myers MG, Valdivieso M, Kiss A. Use of automated office blood pressure measurement to reduce the white coat response. J Hypertens  2009; 27: 280– 286. Google Scholar CrossRef Search ADS PubMed  19. Mancia G, Bertinieri G, Grassi G, Parati G, Pomidossi G, Ferrari A, Gregorini L, Zanchetti A. Effects of blood-pressure measurement by the doctor on patient’s blood pressure and heart rate. Lancet  1983; 2: 695– 698. Google Scholar CrossRef Search ADS PubMed  20. Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Modeling  1999; 6: 1– 55. Google Scholar CrossRef Search ADS   21. Mancia G, Fagard R, Narkiewicz K, Redón J, Zanchetti A, Böhm M, Christiaens T, Cifkova R, De Backer G, Dominiczak A, Galderisi M, Grobbee DE, Jaarsma T, Kirchhof P, Kjeldsen SE, Laurent S, Manolis AJ, Nilsson PM, Ruilope LM, Schmieder RE, Sirnes PA, Sleight P, Viigimaa M, Waeber B, Zannad F; Task Force Members. 2013 ESH/ESC Guidelines for the management of arterial hypertension: the Task Force for the management of arterial hypertension of the European Society of Hypertension (ESH) and of the European Society of Cardiology (ESC). J Hypertens  2013; 31: 1281– 1357. Google Scholar CrossRef Search ADS PubMed  22. Pickering TG, White WB; American Society of Hypertension Writing Group. ASH Position Paper: home and ambulatory blood pressure monitoring. When and how to use self (home) and ambulatory blood pressure monitoring. J Clin Hypertens (Greenwich)  2008; 10: 850– 855. Google Scholar CrossRef Search ADS PubMed  23. Piper MA, Evans CV, Burda BU, Margolis KL, O’Connor E, Whitlock EP. Diagnostic and predictive accuracy of blood pressure screening methods with consideration of rescreening intervals: a systematic review for the U.S. Preventive Services Task Force. Ann Intern Med  2015; 162: 192– 204. Google Scholar CrossRef Search ADS PubMed  24. Rosner B, Polk BF. The implications of blood pressure variability for clinical and screening purposes. J Chronic Dis  1979; 32: 451– 461. Google Scholar CrossRef Search ADS PubMed  25. Weber MA, Schiffrin EL, White WB, Mann S, Lindholm LH, Kenerson JG, Flack JM, Carter BL, Materson BJ, Ram CV, Cohen DL, Cadet JC, Jean-Charles RR, Taler S, Kountz D, Townsend RR, Chalmers J, Ramirez AJ, Bakris GL, Wang J, Schutte AE, Bisognano JD, Touyz RM, Sica D, Harrap SB. Clinical practice guidelines for the management of hypertension in the community: a statement by the American Society of Hypertension and the International Society of Hypertension. J Clin Hypertens (Greenwich)  2014; 16: 14– 26. Google Scholar CrossRef Search ADS PubMed  26. Beckett L, Godwin M. The BpTRU automatic blood pressure monitor compared to 24 hour ambulatory blood pressure monitoring in the assessment of blood pressure in patients with hypertension. BMC Cardiovasc Disord  2005; 5: 18. Google Scholar CrossRef Search ADS PubMed  27. Myers MG, Godwin M, Dawes M, Kiss A, Tobe SW, Grant FC, Kaczorowski J. Conventional versus automated measurement of blood pressure in primary care patients with systolic hypertension: randomised parallel design controlled trial. BMJ  2011; 342: d286. Google Scholar CrossRef Search ADS PubMed  28. Powers BJ, Olsen MK, Smith VA, Woolson RF, Bosworth HB, Oddone EZ. Measuring blood pressure for decision making and quality reporting: where and how many measures? Ann Intern Med  2011; 154: 781– 788. Google Scholar CrossRef Search ADS PubMed  29. Sheppard JP, Stevens R, Gill P, Martin U, Godwin M, Hanley J, Heneghan C, Hobbs FD, Mant J, McKinstry B, Myers M, Nunan D, Ward A, Williams B, McManus RJ. Predicting Out-of-Office Blood Pressure in the Clinic (PROOF-BP): derivation and validation of a tool to improve the accuracy of blood pressure measurement in clinical practice. Hypertension  2016; 67: 941– 950. Google Scholar CrossRef Search ADS PubMed  30. Shimbo D, Abdalla M, Falzon L, Townsend RR, Muntner P. Role of ambulatory and home blood pressure monitoring in clinical practice: a narrative review. Ann Intern Med  2015; 163: 691– 700. Google Scholar CrossRef Search ADS PubMed  31. Shimbo D, Kent ST, Diaz KM, Huang L, Viera AJ, Kilgore M, Oparil S, Muntner P. The use of ambulatory blood pressure monitoring among Medicare beneficiaries in 2007–2010. J Am Soc Hypertens  2014; 8: 891– 897. Google Scholar CrossRef Search ADS PubMed  32. Kjeldsen SE, Lund-Johansen P, Nilsson PM, Mancia G. Unattended blood pressure measurements in the systolic blood pressure intervention trial: implications for entry and achieved blood pressure values compared with other trials. Hypertension  2016; 67: 808– 812. Google Scholar CrossRef Search ADS PubMed  © American Journal of Hypertension, Ltd 2018. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

American Journal of HypertensionOxford University Press

Published: Apr 20, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off