The usefulness of the community pharmacy blood pressure (CPBP) method in the diagnosis or treatment of hypertension has not been adequately addressed in controlled studies. The authors’ aim was to assess the agreement between awake ambulatory blood pressure (ABP), home blood pressure (HBP), and CPBP in treated hypertensive patients. This was a cross-sectional study carried out in 169 patients in which blood pressure (BP) was measured at the pharmacy (4 visits), at home (4 days), and by 24-hour ABP monitoring. Lin correlation-concordance coefficient (CCC) and Bland-Altman plots were used to evaluate quantitative agreement. The qualitative agreement to establish the degree of BP control was evaluated using κ coefficient. The agreement was acceptable between HBP and CPBP (CCC=0.80 for systolic BP [SBP] and 0.80 for diastolic BP [DBP]; κ=0.62) and moderate between awake ABP and CPBP (CCC=0.74/0.67, respectively; κ=0.56). The Bland-Altman plots also showed lowest mean differences (0.5/0.3 for SBP and DBP, respectively) for the comparison between CPBP and HBP. The CPBP has a better agreement with HBP than with awake ABP. Thus, the CPBP measurement method could be a good alternative to HBP monitoring, whereas it cannot be used as a screening test to assess the degree of BP control by awake ABP.
An optimal approach to blood pressure (BP) measurement in the hypertensive patient requires the use of devices and methods with the lowest possible error. Ambulatory BP (ABP) monitoring (ABPM) is the current reference method, whereas home BP (HBP) monitoring (HBPM) represents an acceptable alternative.1,2 Compared with office BP measurements, both ABPM and HBPM lack significant white-coat effect and show a better correlation with target organ damage and cardiovascular risk.3–6
In addition to the methods mentioned above, the community pharmacy BP (CPBP) measurement method is an interesting alternative when HBPM and/or ABPM are not available or are not indicated. This method is widely demanded by patients,7 readily accessible, and recommended by several scientific hypertension societies, including some Spanish and Canadian societies.8–10 However, the usefulness of this method in the diagnosis or treatment of hypertension has not been adequately addressed in previous studies.11 In order to better assess its usefulness, both systematic and random errors with respect to other BP measurements obtained by the reference methods in the management of hypertension (ABPM or HBPM) should be analyzed through agreement studies.12–14 Recently, we reported that the white-coat effect in the community pharmacy was negligible and significantly lower than that observed at the physician office.15,16 Thus, it is possible that the absence of a white-coat effect in the community pharmacy favors the agreement with BP measurement methods outside the clinical setting.
The aim of the present work was to assess the agreement between CPBP, awake ABP, and HBP in treated hypertensive patients. The prevalence of community pharmacy–isolated hypertension and masked hypertension were also determined.
The investigation of the clinical usefulness of the CPBP (the MEPAFAR study) was a cross-sectional study in treated hypertensive patients older than 18 years from 8 community pharmacies in Gran Canaria, Spain, between June 2008 and June 2009. Patients were excluded if any of the following criteria were met: systolic BP (SBP) ≥200 mm Hg and/or diastolic BP (DBP) ≥110 mm Hg on the initial visit to the pharmacy, arm circumference >42 cm, atrial fibrillation, physical or mental impairment, inability to perform HBPM, changes in the antihypertensive treatment schedule during the previous 4 weeks, history of cardiovascular disease <6 months, or pregnancy.
Selection and Size of the Sample
The sample size was based on the agreement between awake ABP and CPBP previously reported in a study carried out in the community pharmacy setting.17 Specifically, the data to calculate the sample size were expected κ coefficient (0.36), proportion of patients with CPBP ≥140 mm Hg and/or 90 mm Hg (34.0%), proportion of patients with ABP ≥135 mm Hg and/or 85 mm Hg (30.0%), confidence level (95%), and precision (0.15). Thus, the estimated sample size was 167 patients. Additionally, we added 20% to compensate for possible incomplete datasets from patients who might withdraw from or fail to complete the study (final sample size: 200 patients). In each community pharmacy, patients were identified and recruited consecutively during the dispensing process of antihypertensive medications until a minimum of 25 patients were recruited per pharmacy.
BP Measurement Methods
CPBP was measured by the same pharmacist at each pharmacy. A clinically validated OMRON M10-IT automatic electronic device (Omron Corp, Tokyo, Japan) was used, with a cuff adaptable to large (32–42 cm), medium (23–31 cm), and small (17–22 cm) arm circumferences. This device is equivalent to the clinically validated OMRON M618–20 (see http://www.dableducational.org). The CPBP was recorded during 4 visits to the pharmacy, over a 4-week period. At each visit, 3 measurements were taken (2 or 3 minutes apart) on the control arm (the arm on which the CPBP was higher on the first visit). The patient visits were scheduled at the same time for all 4 time points (±1 hour). All pharmacists were previously instructed on how to properly perform BP measurements according to international guidelines.21 Generally, CPBP measurements were taken after 5 minutes of sitting/rest and when the patient had assured the pharmacist that he had not consumed coffee or tea, smoked, or exercised in the 30 minutes prior to the measurement. The mean CPBP was calculated, discarding the data from the first visit and the first BP measurement from each visit. CPBP control was defined as SBP <140 mm Hg and DBP <90 mm Hg.
At home, the same OMRON-M10-IT model was used. Every patient was instructed on the HBPM technique at a 20-minute training session by his/her pharmacist. At the end of the session, the HBPM technique was tested by 3 consecutive self-measurements made in the presence of the pharmacist. Patients were also provided written guidelines to reinforce the training provided. Patients monitored their HBP during a 4-day period, taking 3 measurements in the morning (each measurement 2 minutes apart, between 6 am and 9 am) and 3 in the evening (between 6 pm and 9 pm) on the nondominant arm. The HBP readings were stored in the device’s memory. The mean HBP was calculated, discarding values obtained on the first day and the first measurement obtained each morning and each evening. HBP control was defined as SBP <135 mm Hg and DBP <85 mm Hg.22
The ABPM was performed on a working day (24 hours), using the nondominant arm. Usually, the ABPM started on the third visit to the community pharmacy. The clinically validated Spacelabs Medical 90207-5Q monitor (Spacelabs Inc, Redmond, WA) was used.23 Monitors were programmed to measure BP at 20-minute intervals between 7 am and 10 pm and at 30-minute intervals between 10 pm and 7 am. Patients were instructed to follow their usual daily activities (albeit avoiding vigorous exercise) but remain still with the forearm extended during each reading. Furthermore, they were asked to keep a sleep diary specifying the time that they went to bed and awoke, including daytime sleep if present. A large cuff was used if the arm perimeter was between 32 cm and 42 cm and a medium cuff was used if it was between 23 cm and 31 cm. The average awake ABP was used, which was calculated according to the record kept by each patient. Awake ABP control was defined as SBP <135 mm Hg and DBP <85 mm Hg.22
To characterize the study sample, the following variables were collected by the pharmacists: age, sex, heart rate (community pharmacy, daytime, home), smoking status, weight and height, number of antihypertensive drugs used, history of previous cardiovascular disease (stroke, myocardial infarction, angina, or peripheral artery disease), and presence of diabetes or dyslipidemia (documented diagnosis or previously prescribed drug treatment). The MEPAFAR study was approved by the Research Ethics Committee of the University of Granada (Spain). The patients’ participation was voluntary and all gave informed consent for study participation. To process and manage the patient information, the resources of the Spanish ABPM Registry (CARDIORISC-MAPAPRES project) were used.24,25
The SPSS version 15.0 (SPSS Inc, Chicago, IL), R version 2.10 (The R Foundation for Statistical Computing) and Epidat version 3.1 statistical packages were used to store and analyze the data. Patients were excluded from the analysis if: (1) they did not have all the CPBP measurements (4 visits), (2) the ABPM lasted <24 hours or provided <25% of the scheduled readings during that period, or (3) they monitored HBP for <4 days or provided <12 valid HBP readings in the last 3 days of the HBPM.
The mean and standard deviation (SD) were used to summarize the quantitative variables and frequencies and percentages for qualitative variables. Differences between CPBP, HBP, and awake ABP were assessed by paired t tests: Student t test for paired samples and repeated measures analysis of variance, applying the Bonferroni correction.
The Lin correlation-concordance coefficient (CCC)26,27 and the Bland-Altman method28 were used to evaluate the quantitative agreement between BP measurements. Fleiss29 proposed CCC limits that were used for agreement interpretation: very good (CCC>0.9), acceptable (0.71≤CCC≤0.9), moderate (0.51≤CCC≤0.7), poor (0.31≤CCC≤0.5), or no agreement (CCC<0.31). The Bland-Altman method is based on a graphical representation where the y axis represents the difference between 2 BP measurements and the x axis is the average of the 2 measurements. In each Bland-Altman plot, the aspects considered to evaluate the quantitative agreement were: (1) the mean of the differences between the 2 BP measurements (showing the systematic error between the measurements) and the SD (dispersion of random error or imprecision), (2) concordance limits (mean of the differences ±2 SD) (limits between which 95% of the differences between the 2 measurements should oscillate), (3) the amplitude of the explained interval within the concordance limits, (4) percentage of differences outside the concordance limits, and (5) the percentage of differences <5 mm Hg between the 2 BP measurements (considered differences clinically nonrelevant). Additionally, the qualitative agreement between BP measurement methods to establish the degree of BP control (controlled/not controlled) was evaluated using the κ coefficient.30 As a function of the κ value, the agreement was determined as31: very good (0.81≤κ≤1), acceptable (0.61≤κ≤0.80), moderate (0.41≤κ≤0.60), weak (0.21≤κ≤0.40), poor (0≤κ≤0.20), or no agreement (κ <0).
Community pharmacy–masked hypertension was defined as a CPBP (SBP/DBP) <140/90 mm Hg in the presence of HBP or awake ABP ≥135 mm Hg and/or 85 mm Hg; whereas community pharmacy–isolated hypertension was defined as CPBP ≥140 mm Hg and/or 90 mm Hg in the presence of a normal HBP and awake ABP (<135/85 mm Hg). Using awake ABP or HBP as references, the sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratio for the CPBP measurement method were calculated. The same figures were calculated for HBP, using awake ABP as the reference. The test was considered positive when the SBP, DBP, or both were >140 mm Hg and/or 90 mm Hg for CPBP and >135 mm Hg and/or 85 mm Hg for HBP or awake ABP. The 95% confidence intervals (CIs) were obtained and a P value <.05 was considered statistically significant. Using awake ABP and HBP as references, the receiver operating characteristic (ROC) curves for SBP and DBP in the community pharmacy were obtained.
A total of 213 individuals agreed to participate in the study, with 22 patients excluded for the following reasons: atrial fibrillation (n=5), changes in the antihypertensive treatment <1 month ago (n=1), SBP >200 mm Hg at the initial visit to the pharmacy (n=1), arm circumference >42 cm (n=1), and unable to perform HBPM technique (n=14). Furthermore, 14 additional patients were excluded from the analysis due to insufficient number of BP readings (<75% of ABP readings) and 8 patients withdrew informed consent. Thus, 169 patients were finally included in the analysis.
The mean age was 56.4 (SD: 10.6) years and 59.8% were women. The average time taken by each patient to obtain the 4 CPBP measurements was 21.1 (SD: 7.7) days. The general characteristics of the patients and the mean BP obtained by each method are shown in Table I.
Table I. General Characteristics of the Sample (N=169)
Abbreviations: BP, blood pressure; bpm, beats per minute. Values are expressed as mean (standard deviation) except where indicated.
Women, No. (%)
Body mass index
Normal weight, No. (%)
Overweight, No. (%)
Obese, No. (%)
Smokers, No. (%)
Dyslipidemia, No. (%)
Diabetes, No. (%)
History of cardiovascular disease, No. (%)
One drug, No. (%)
Two drugs, No. (%)
Three drugs, No. (%)
Four drugs, No. (%)
Community pharmacy BP
Systolic BP, mm Hg
Diastolic BP, mm Hg
Heart rate, bpm
Systolic BP, mm Hg
Diastolic BP, mm Hg
Heart rate, bpm
Systolic BP, mm Hg
Diastolic BP, mm Hg
Heart rate, bpm
The CCCs showed an acceptable agreement between CPBP and HBP (0.8 for SBP and DBP) and a moderate agreement between CPBP and awake ABP (0.74 for SBP and 0.67 for DBP) (Table II). Figure 1 shows the Bland-Altman plots to assess the agreement between BP measurements. The mean (SD) of the differences between CPBP and HBP were 0.5 (9.3)/−0.3 (6.0) mm Hg for SBP and DBP, respectively (Table II). The percentages of clinically nonrelevant differences (<5 mm Hg) were 46.7% for SBP and 58.0% for DBP. The mean (SD) of the differences between CPBP and awake ABP were −1.1 (9.8)/3.0 (7.5) mm Hg for SBP and DBP, respectively. The percentages of clinically nonrelevant differences were 40.8% for SBP and 48.5% for DBP.
Table II. Lin Correlation-Concordance Coefficients and Summary of the Bland-Altman Method to Test the Agreement Between the 3 Blood Pressure Measurements Used in the Study
Concordance Limits (MDM ±2 SD)
Percentage of Differences <5 mm Hg
CCC (95% CI)
Abbreviations: ABP, ambulatory blood pressure; BP, blood pressure; CCC, Lin correlation-concordance coefficient; CI, confidence interval; CPBP, community pharmacy blood pressure; HBP, home blood pressure; MDM, mean of the differences between measurements; PDTL, percentage of differences outside the concordance limits; SD, standard deviation. aAmplitude of the explained interval within the concordance limits. bStatistically significant differences.
−18.1 to 19.1
−20.7 to 18.5
−23.2 to 20.0
−12.3 to 11.7
−12.0 to 18.0
−12.9 to 19.5
Figure 2 shows the number of patients in each category of BP control according to the 3 different measured BPs. Overall, 117 patients (69.2%) were classified in the same way for all 3 methods.
The CPBP measurement method and awake ABP classified 137 patients (81.1%) in the same category. The κ coefficient showed a moderate agreement: 0.56 (95% CI, 0.42–0.69). Prevalence rates of isolated and masked hypertension in the community pharmacy were 5.3% (95% CI, 1.6–9.0) and 13.6% (95% CI, 8.1–19.1), respectively. Using awake ABP as the reference, the sensitivity and specificity of the CPBP measurement method were 61.0% (95% CI, 47.7–74.3) and 91.8% (95% CI, 86.2–97.4), respectively (Table III). Figure 3a shows the ROC curves for SBP and DBP in the community pharmacy using awake ABP as the reference; the optimal cut-off points (CPBP values with greater overall sensitivity and specificity) were found at 132/83 mm Hg.
Table III. Sensitivity, Specificity, Positive and Negative Predictive Values of CPBP and HBP
Abbreviations: CI, confidence interval; CPBP, community pharmacy blood pressure; HBP, home blood pressure; NLR, negative likelihood ratio; NPV, negative predictive value; PLR, positive likelihood ratio; PPV, positive predictive value. aAwake ambulatory blood pressure as the reference. bHBP as the reference.
The CPBP measurement method and HBP classified 139 patients (82.2%) in the same category; the κ coefficient showed an acceptable agreement between methods: 0.62 (95% CI, 0.50–0.73). Using HBP as the reference, prevalence rates of isolated and masked hypertension in the community pharmacy were 1.2% (95% CI, 0.1–4.2) and 16.6% (95% CI, 10.6–22.5), respectively. The sensitivity and specificity of the CPBP measurement method were 60.5% (95% CI, 48.5–72.6) and 98.0% (95% CI, 94.6–100), respectively (Table III). Figure 3b shows the ROC curves for SBP and DBP in the community pharmacy using HBP as the reference; the optimal cut-off points were also 132/83 mm Hg.
HBP and awake ABP classified 127 patients (75.1%) in the same category; the κ coefficient showed a moderate agreement: 0.48 (95% CI, 0.34–0.61). The sensitivity and specificity of HBPM were 74.6% (95% CI, 62.6–86.5) and 75.4% (95% CI, 67.0–83.9), respectively (Table III).
Finally, as normal values for CPBP are not totally established and CPBP lacks the significant white-coat effect, we repeated the agreement analyses by using 135/85 mm Hg as the cut-off value for CPBP. Table S1 shows the qualitative agreement analysis using this cut-off value. Results obtained were similar, and agreements between the CPBP and awake ABP or HBP were also moderate (κ=0.53) and acceptable (κ=0.66), respectively. Moreover, measuring BP in the community pharmacy was found sufficiently reliable to confirm the presence of both controlled and uncontrolled BP at home (sensitivity, 80.3%; specificity, 85.6%).
The present analysis of the MEPAFAR study offers new information on the error and the usefulness of the CPBP measurement method in treated hypertensive patients. Previously, a single study by Botomino and colleagues17 provided information in a sample of 22 individuals: the qualitative agreement between awake ABP and CPBP was moderate (simple agreement: 72.7%; κ: 0.45) and prevalence rates of isolated and masked hypertension in the community pharmacy were 22.7% and 4.5%, respectively. However, the study had several methodological limitations, such as small sample size, observer bias and digit preference (CPBP measurements were taken manually), and CPBP measured at one single visit; therefore, these results should be reviewed.
In the present study, agreement between CPBP and HBP was acceptable and greater than other comparisons (CPBP vs awake ABP and HBP vs awake ABP). This agreement is probably due to similar conditions when BP is measured in the community pharmacy or at home. Moreover, CPBP lacks white-coat effect.15,16 In this view, CPBP can be an alternative of HBPM to establish the degree of BP control in treated hypertensive patients. On the contrary, the agreement between CPBP and awake ABP was only moderate, and using the CPBP measurement method as a screening test to determine the degree of BP control by awake ABP cannot be recommended from the present results.
We found a very low prevalence of community pharmacy–isolated hypertension (1.2% using HBP as the reference or 5.3% using awake ABP as the reference), which can be explained by the high specificity and positive predictive values of the CPBP (high reliability to confirm the presence of lack of BP control). This may support the usefulness of the CPBP measurement method to detect patients who need to be referred to the physician and/or require up-titration of their antihypertensive treatment. On the other hand, sensitivity and negative predictive values of CPBP were lower (low ability to confirm the presence of BP control) and, therefore, the prevalence of masked hypertension was high (16.6% using HBP as the reference or 13.6% using awake ABP as the reference). This limits its usefulness in controlled patients, as some may require treatment intensification (based on high awake ABP and/or HBP figures) despite normal BP values at the community pharmacy.32
It should be noted that the upper limit for normal HBP and awake ABP used in this study are still under discussion, as there is no evidence showing which HBP or awake ABP should be considered as the optimal target for drug treatment.22,33,34 Furthermore, there is no recommendation that indicates which normal BP values should be used for the CPBP. We have assumed the thresholds used in the clinical setting as reasonable points of reference. Therefore, this matter should be studied in depth, as it is possible that the normal CPBP values may be different from those defined in the clinical setting; even similar to those established for HBP or awake ABP. In order to go in depth into the agreement between CPBP and both awake ABP and HBP, we also performed an analysis using 135/85 mm Hg as the cut-off for the CPBP (Table S1). Overall, the results of this analysis also reinforced the usefulness of the CPBP measurement method as an alternative to HBPM.
From a clinical viewpoint, the MEPAFAR study provides the first evidence to establish recommendations for community pharmacists and physicians when interpreting CPBP measurements in treated hypertensive patients. These recommendations are of particular interest due to the involvement of the community pharmacist in the follow-up of treated hypertensive patients and the need to achieve a better shared management of these patients (physician/pharmacist collaborative management).35–37 On the other hand, it is possible that in some circumstances, HBPM and ABPM may not be available or cannot be properly used. Then, the CPBP measurement method can be a valuable alternative to assess the effectiveness of treatment.
Limitations and Strengths
It is important to note that the present investigation was a short-term study and its results were limited to a specific sample of treated hypertensive patients. Therefore, further research to assess the long-term agreement between methods is needed. Also, the impact of changing the schedule of CPBP measurements (ie, twice a day) is still unknown. Caution should be exercised in interpreting these results more broadly, as the CPBP measurements were measured by the same pharmacist in each community pharmacy. This is distinct to a patient measuring his/her BP in the community pharmacy without pharmacist supervision, ie, using self-attended BP monitors.38,39 Also different pharmacists or pharmacy technicians who take CPBP measurements in the same pharmacy may impact the results. Finally, our results could be affected by specific characteristics of the MEPAFAR study, such as using trained pharmacists and trained patients or using validated equipment. Therefore, it is not possible to assure that using a different methodology will lead to the same results. We are aware that, due to different reasons (eg, community pharmacy’s business model, staff training, reimbursement for pharmacy services or staff availability), our methods are not used in all Spanish-community pharmacies nor in others from abroad (eg, United States). However, we believe that our methods are relatively simple and could be exported to any pharmacy in the world. Moreover, this approach meets the quality requirements that pharmaceutical care services should provide. Hence, it should be implemented in the community pharmacy setting in order to achieve a proper physician/pharmacist collaborative patient management.
CPBP showed an acceptable agreement with HBP and a moderate agreement with awake ABP. As a consequence, the CPBP measurement method could be a good alternative to HBPM, when the latter lacks suitability. On the other hand, CPBP measurements cannot be used as a screening test to assess the degree of BP control by awake ABP in treated hypertensive patients. Community pharmacy–masked hypertension is a relatively common condition, whereas community pharmacy–isolated hypertension shows a low prevalence.
Acknowledgments and disclosures: We thank Eleonora Feletto for assistance in editing the text. This work has been partially supported by a grant from LACER Laboratories (Spain). The authors do not have potential conflict of interest.
Members of the MEPAFAR study workgroup: Pedro Amariles Muñoz, María Isabel Baena Parejo, Nirma Esperanza Hernández Peña, José María Sabater Díaz, Antonio Artiles Campelo, María Elena Jorge Rodríguez, María Fernanda García Morales, Nayra Díaz Merino, María Esther Artiles Ruano, Ana María Contardi Lista, Ana Rosa García Rodríguez.