Sensitivity and specifi city in the diagnosis of hypertension with different methods

OBJECTIVE: To evaluate sensitivity and specifi city of different protocols for blood pressure measurement for the diagnosis of hypertension in adults. METHODS: Cross-sectional study conducted in a non-probabilistic sample of 250 public servants of both sexes aged 35 to 74 years in Vitória, southeastern Brazil, between 2008 and 2010. The participants had their blood pressure measured using three different methods: clinic measurement, self-measured and 24-hour ambulatory measurement. They were all interviewed to obtain sociodemographic information and had their anthropometric data (weight, height, waist circumference) collected. Clinic measurement and self-measured were analyzed against the gold standard ambulatory measurement. Measures of diagnostic performance (sensitivity, specifi city, accuracy and positive and negative predictive values) were calculated. The Bland & Altman method was used to evaluate agreement between ambulatory measurement (standard deviation for daytime measurements) and self-measured (standard deviation of four measurements). A 5% signifi cance level was used for all analyses. RESULTS: Self-measured blood pressure showed higher sensitivity (S=84%, 95%CI 75;93) and overall accuracy (0.817, p<0.001) in the diagnosis of hypertension than clinic measurement (S=79%, 95%CI 73;86, and overall accuracy=0.815, p<0.001). Despite the strong correlation with daytime ambulatory measurement values (r=0.843, p<0.001), self-measured values did not show good agreement with daytime systolic ambulatory values (bias=5.82, 95%CI 4.49;7.15). Seven (2.8%) cases of white coat hypertension, 26 (10.4%) of masked hypertension and 46 (18.4%) of white-coat effect were identifi ed. CONCLUSIONS: The study shows that self-measured blood pressure has higher sensitivity than clinic measurement to identify true hypertension. The negative predictive values found confi rm the superiority of self-measured when compared to clinic in identifying truly normotensive individuals. However, clinic measurement cannot be replaced with self-measured, as it is still the most reliable method for the diagnosis of hypertension. DESCRIPTORS: Hypertension, diagnosis. Diagnostic Techniques and Procedures. Sensitivity and Specifi city. Cross-Sectional Studies. 2 Methods for the diagnosis of hypertension Molina MCB et al


INTRODUCTION
The present study aimed to evaluate sensitivity and specifi city of different protocols for blood pressure measurement for the diagnosis of hypertension in adults.

METHODS
This cross-sectional study was developed as part of the Longitudinal Study of Adult Health (ELSA Brazil).The ELSA Brasil study was designed to investigate chronic disease determinants in the Brazilian population with main focus on cardiovascular diseases and diabetes.b The study participants were active or retired public servants of the Universidade Federal do Espírito Santo aged 35 to 74 years.The sample size was estimated to detect a 3-mmHg difference in mean blood pressure using different measurement methods, an alpha error of 5% and a statistical power of 90%.Assuming a loss of 20%, the estimated sample size was 248 individuals.ELSA Brasil participants who had completed all previous stages of the study project were invited to participate in this substudy.Those participants whose anthropometric characteristics (left arm circumference greater than 50 cm or lower than 17 cm) were not suitable for AMBP and SMBP were excluded from the study.
Data was collected at the ELSA-ES Research Center.The study sample comprised 255 individuals but four were excluded from the analysis because they did not complete the set number of AMBP (16 valid measurements during daytime and eight during sleep). 15Another participant was excluded for not having taken SMBP as established by the protocol.The fi nal sample consisted of data from 250 individuals studied between 2008 and 2010.
The body mass index (BMI) was calculated as the ratio between weight (kg) and height (m 2 ) and the recommended World Health Organization (WHO) 20 cutoffs were used.Anthropometric measurements were made according to international recommendations. 20Weight was measured using an electronic scale (Toledo®) with capacity of 200 kg and height was measured with a stadiometer (Seca®) with precision of 0.1 cm and bulb level.
Measures of waist circumference (WC), hip circumference (HC) and mid-upper arm circumference (MUAC) were obtained using an inelastic, fl exible tape placed in contact with the skin but without compressing tissues.WC was taken at the midpoint between the lower Hypertension is a major public health concern worldwide due to its high prevalence in the adult population and strong impact on cardiovascular morbidity and mortality. 4,15Although there is no cure for this condition, blood pressure control can be achieved in most cases with general management associated or not with drug therapy.a An almost normal survival can be achieved in hypertensive patients 18,19 when management is based on an accurate diagnosis as antihypertensive drugs produce major side effects, especially when chronically used. 16Epidemiologically speaking, reducing false-positive and false-negative rates should be a priority goal while approaching hypertension as it can optimize management, prevents the effects of unnecessary drug use and thus reduce health care costs and improve quality of life. 7,10,11though the diagnosis of hypertension is apparently straightforward, it may be affected by factors related to the examiner, instrument used, environment and the very patient. 7Some patients may experience stress that is strongly infl uenced by the presence of a medical examiner and have unintentionally elevated blood pressure (BP) levels during assessment resulting in false positive (white coat hypertension) 17 or false negative (white-coat effect) diagnosis. 16re recently the use of oscillometric devices has improved blood pressure assessment and allowed standardizing home monitoring of blood pressure and self-measured blood pressure (SMBP). 8SMBP allows patients to measuring their BP in an environment where they spend most of their time, for example, at work. 1 The appropriate use of these devices could reduce false positives as elevated blood pressure due to stress would be minimized with blood pressure assessment in environments other than hospitals and medical offi ces (clinic measurement of blood pressure, CMBP). 1,9owever, ambulatory measurement of blood pressure (AMBP) is still superior to SMBP and CMBP because it provides an automated measurement and allows prolonged (e.g., 24-h) monitoring as well. 12though SMBP is a more affordable method than AMBP 6 and can provide a larger number of measures compared to CMBP, 1 few studies have compared blood pressure measurements obtained with these different methods. 14,17There are numerous issues related to blood pressure assessment and it is therefore critical to establish the actual effectiveness of the currently used methods.margin of the rib and the iliac crest in the midaxillary line; HC was taken by positioning the tape at the greatest protuberance of the buttocks.MUAC was measured as follows: the participant was in a standing position and his/her arm was positioned with the elbow fl exed at 90° and palm facing up.The measure was taken from the lateral aspect of the acromion (bony edge of the shoulder) to the olecranon (elbow tip).c All participants were interviewed using a questionnaire on sociodemographic and health-related information.The variable race/skin color was self-reported.
Participants were asked to empty their bladder and do not eat, drink (including coffee) and smoke 30 minutes before the assessment.With the participant in a sitting position after resting for at least fi ve minutes, three readings were made concomitantly in each arm according to the Brazilian Society of Hypertension criteria. 16BP in each arm was calculated as the arithmetic average of the two last readings and classifi ed according to the readings obtained in the arm with the highest BP. 16llowing CMBP an ambulatory blood pressure monitoring device (Spacelabs 2000®) was placed on left arm and programmed to perform automatic measurements every 15 minutes during daytime and every 30 minutes during sleep (10 p.m. to 6 a.m.).Mean systolic and diastolic blood pressure and heart rate were automatically recorded for each participant for a 24-hour period, during daytime and nighttime.Each participant was then given a validated oscillometric device (Omron 705CP-Intelissense®) for home measurement of blood pressure.They were instructed on measurement procedures and to keep a "monitoring log."Four BP readings were programmed to be made between 11 a.m. and 12 p.m., 5 p.m. and 6 p.m., 9 p.m. and 10 p.m. and 7 a.m. and 8 a.m. the next morning.The readings on the instrument's display were to be recorded in the participant's log, but they were automatically saved on the instrument as well.Each participant was instructed to perform one reading at each time point and only repeat it fi ve minutes later if there was an error.They were also advised to make the reading in a quiet environment, comfortably sitting and resting their arm and feet.
All instructions for SMBP were given by the same nurse during a 45-minute session.Participants were asked to perform the entire procedure to show they felt comfortable with it.This was intended to reduce the participant's anxiety related to the instrument's use 1 -a determinant factor to prevent false-positive results.It should be noted that the study sample included individuals with all levels of schooling.Only one set of readings was not validated due to inconsistent procedure and was excluded from the study.SMBP was considered adequate when four valid readings were recorded in the instrument's memory at the preset time points.The participant's SMBP was calculated as the average of these four readings.Hypertension was diagnosed based on each BP measurement method used according to the VI Brazilian Guideline on Hypertension. 18For daytime SMBP and AMBP hypertension was defi ned as systolic blood pressure (SBP) ≥135 or diastolic blood pressure (DBP) ≥85.For 24-h AMBP and AMBP hypertension was defi ned as SBP ≥130 or DBP ≥ 80 and SBP ≥140 or DBP ≥90, respectively.
For identifying white coat hypertension (false positives), masked hypertension and white-coat effect (false negatives), the following combinations of methods were used: SMBP and CMBP as the method tested; AMBP and CMBP as the gold standard for the three conditions described above.
In the statistical analysis categorical variables were described as percentages.The chi-square test (χ²) was used to test the hypothesis of homogeneity of proportions.The Kolmogorov-Smirnov test was used to assess the normality of continuous variables.Measures of diagnostic performance (sensitivity, specifi city, positive and negative predictive values, and accuracy) were calculated using 24-h AMBP as the gold standard.The Bland & Altman methodology 3,5 was used to assess the agreement between AMBP (daytime standard deviation) and SMBP (standard deviation of four measures).The level of signifi cance for all tests was set at 5%.All statistical analyses were performed with SPSS for Windows (version 17.0) and Epidat 3.0.
The study was approved by the Research Ethics Committee of the Center for Health Sciences at Universidade Federal do Espírito Santo (protocol no.140/08).All participants signed an informed consent form.1.There were signifi cant differences in gender, race/skin color, age, height, weight and diagnosis of hypertension (p<0.05).Hypertensive participants showed higher mean age (55.9, SD = 8.7) compared to normotensive (53.2, SD = 9.7).There were no signifi cant differences in WC (p = 0.891) and HC (p = 0.425) among hypertensive women and BMI in hypertensive men (p = 0.099).

Sociodemographic characteristics of participants according to BP classifi cation determined by 24-h AMBP are shown in Table
Table 2 shows that SMBP and CMBP showed similar sensitivity whereas CMBP had higher specifi city.Also, both methods showed similar accuracy.
Table 2 shows a comparison of the performance of two BP measurement methods to identify white coat hypertension, masked hypertension and white-coat effect.There were identifi ed seven (2.8%) participants with white coat hypertension, 26 (10.4%) with masked hypertension and 46 (18.4%) with white-coat effect.The sensitivity, specifi city, positive and negative predictive values and accuracy for identifying a white-coat effect were 59%, 84%, 45%, 90% and 79%, respectively.
Figure 1 presents a comparison of mean BP obtained using SMBP and daytime AMBP. Figure 1a shows that BP measures had a strong correlation (r = 0.843, p<0.001) but they did not show good agreement as the Bland & Altman plot (Figure 1b) indicates statistically signifi cant bias (d) (d = 5.82, 95%CI: 4.49, 7.15).a Diagnosis defi ned using a combination of self-measured blood pressure and method and clinic measurement of blood pressure.SMBP: self-measured blood pressure; CMBP: clinic measurement of blood pressure; WCH: white coat hypertension; MH: masked hypertension; WCE: white-coat effect.

DISCUSSION
The study results suggest that SMBP has higher sensitivity than CMBP to identify true hypertension in the general population as this is a highly prevalent condition.The NPVs found support the fi nding that SMBP is superior to CMBP to identify truly normotensive individuals.
The method tested for identifying white coat hypertension showed greater specifi city than sensitivity, and good diagnostic ability evidences through high accuracy and agreement, and a NPV indicating low false-negative rates.The results with the same combination (SMBP and CMBP) used for identifying masked hypertension and white coat effect suggest that a similar performance to that seen for white coat hypertension.
Since SMBP do not follow a standard protocol time points of BP measures and number of steps required to establish the most accurate diagnosis are set at the physician's and patient's discretion. 1,15Thus, a wide range of methodological approaches have been SMBP: Self-measured blood pressure; AMBP: ambulatory measurement of blood pressure investigated, making it difficult comparison with other studies.However, the present study allows a comparison of performance of the three methods studied.One limitation of this study is the high number of participants taking antihypertensive drugs, which makes it diffi cult to know their actual BP but does not preclude a comparison of methods for the diagnosis of hypertension.
The superior results found refers only to aspects related to a method's sensitivity and accuracy and should not be interpreted as an option to replace one method for another as they are described in the literature as being complementary for diagnosing hypertension. 13It is not intended here to conclude that one method is superior to another, but rather to point to different methods available that can be used to prevent misdiagnosis.This is supported by the Bland & Altman method results 3 showing that SMBP and AMBP while having a strong correlation did not show good agreement as to suggest that one method should replace the other. 3,5ile there have been efforts to reduce the patient's anxiety regarding blood pressure measurement, our study found signifi cant BP differences between the methods tested compared to the gold standard, especially for SBP.SMBP measures were higher than those obtained by the nurse at the medical offi ce (CMBP) and AMBP, showing that although CMBP is less sensitive it remains crucial for the clinical diagnosis of hypertension.
Since white coat hypertension, masked hypertension and white coat effect have low prevalence, a test's specifi city is the best marker of diagnostic quality.We did not compare a test's superiority because a single test was used to identify different conditions.We found that combining SMBP and CMBP was more effective for the diagnosis of white coat hypertension than masked hypertension and white coat effect.
The study results suggest that the anxiety experienced during SMBP can be as important as that due to CMBP.It also showed that when appropriately performed SMBP is a feasible, safe and low-cost approach that can help the clinical diagnosis of hypertension.
In conclusion, SMBP can be used to help the diagnosis when hypertension is suspected but it cannot replace CMBP, which is still the most reliable method.The indiscriminate use of SMBP should be discouraged in the general population because it is only helpful when the protocol is followed consistently but most people do not know the correct procedure.Also, some people may carry out a BP measure at times when they feel their "BP is high," increasing their anxiety.Therefore, the management of hypertension based solely on SMBP is not recommended.

Figure 1 .Figure 2 .
Figure 1.Dispersion measures of correlation and agreement of systolic blood pressure measures between self-measured blood pressure and daytime ambulatory measurement of blood pressure.Vitória, Southeastern Brazil, 2008-2010.

Table 1 .
Sociodemographic and anthropometric characteristics of the sample studied according to the diagnosis of hypertension using 24-hour ambulatory measure of blood pressure.Vitória, Southeastern Brazil, 2008-2010.
BMI: Body mass index; WC: waist circumference, WHR: waist-to-hip ratio.* Refer to χ 2 statistics for gender and race/skin color (% and N) and Student's t-test for the other variables presented (mean and standard deviation).

Table 2 .
Estimated sensitivity, specifi city, positive and negative predictive values and accuracy of measures of diagnostic performance and agreement among the methods studied to predict hypertension.Vitória, Southeastern Brazil, 2008-2010.