Home Self Blood Pressure Measurement in General Practice

Transcription

1 A ; :-5 Home Self Blood Pressure Measurement in General Practice The SMART Study Gilles Chatellier, Claire Dutrey-Dupagne, Laurent Vaur, Faiez Zannad, Nathalie Gem%,Fran~ois Elkik, and Jokl M&zard The SMART study (Self-Measurement for the Assessment of the Response to Trandolapril) was performed in general practice and enrolled patients in order to assess on a large scale the feasibility and informative value of selfmeasurement of blood pressure at home (SMBP), define home blood pressure (BP) levels in comparison to office readings, and determine the number of home measurements necessary to provide an accurate and precise BP value After a -week washout period, patients with office diastolic blood pressure within the range 5 to mm Hg received mg trandolapril once daily in the morning for weeks Four days of SMBP were performed both at the end of the washout period and the end of the treatment period, with an automatic printer-equipped oscillometric device (A&D UA5) The first day values were not analyzed Thus, the maximum number of BP measurements obtained per patient and per period was 8 Four hundred and twentyfour patients (5% ) did not perform any measurements One thousand one hundred and nine patients (5% ) performed at least measurements Among them, ( ) correctly performed all 8 measurements A preference for digits Oand 5 was detected in physicians measurements (three consecutive values, during a single office visit) This digit preference was not found with the semiautomatic device When the number of measurements selected for analysis was increased from to 8, in the patients who provided all recordings and fulfilled all protocol criteria, the standard deviation of the mean BP of the cohort was reduced by for SBP and by % for DBP Eighty-five percent of this reduction was already achieved by six home measurements taken at random BP was significantly lower at home than at the office by S mm Hg for systolic BP (SBP), and 8 ~ mm Hg for diastolic BP (DBP) This difference was independent of age, more marked in women (P < for SBP and P <5 for DBP), and had a Gaussian distribution Under treatment, office SBP/DBP decreased from ~ 8/ ~ 5 mm Hg to * /8 * 8 mm Hg, while SMBP decreased from 5 * 8/8 t mm Hg to * /85 * 5 mm Hg (all P < ) A major aim in research studies and individual care is to reduce BP measurements variability This study demonstrates the ability to evaluate baseline SMBP level in two-thirds of patients previously unfamiliar with the method, the ability to evaluate treatment effect in about one-half of the patients, the improvement in the measurement precision obtained with the repetition of measures (at least six home measurements), and the absence of bias of SMBP as compared to office measurements Am J Hypertens ; :-5 KEY WORDS: Self-measurement of blood pressure, office blood pressure, hypertension Received August, 5 Accepted January, From the Clinical Investigation Center, Broussais Hospital, Paris, France (GC, JM), Roussel, Paris, France (CD-D, LV, NG, FE), and Service de Pharmacologic Clinique, CHRU de Nancy, Nancy, France (FZ) The study was supported by Laboratoires Roussel, rue de Vaugirard, 5 Paris, France Address correspondence and reprints reauests to Dr Gilles Chatellier, Centre d Investigations Cliniques, H ipitalbroussais, rue Didot, 5 Paris Cedex, France Downloaded from by the American Journal of Hypertension, Ltd on November Published by Elsevier Science, Inc 85-//$5 PII 85-()8-

2 AJH-JULY -VOL, NO SELFBP MEASUREMENT IN GENERALPRACTICE 5 The optimal method for measuring blood pressure (BP) remains controversial Office blood pressure (OBP) recorded by physicians with a mercury sphyg~omanometer is still considered as the reference because it has been used to describe the relationship between BP and cardiovascular morbidity and mortality* and to establish the efficacy of antihypertensive drug therapy in long term trialsz Nevertheless, (BP has some important limitations including unprecise BP level estimate, inaccurate diagnosis of hypertension and poor prediction of target organ damage, compared to ambulatory blood pressure Among the new emergent methods to evaluate BP, home selfmeasured blood pressure (SMBP) may offer some advantages both in therapeutic evaluation and in patient management Previous reports have pointed out that SMBPis better correlatedto leftventricularmass than OBP, 5and provides a more accurate evaluation of the effect of a treatmentthe use of SMBP may also be costeffective by avoiding the need for frequent office visits and by helptig patients to participate more actively in the treatment of their disease, The SMART study (Self-Measurementfor the Assessment of the Response to Trandolapril)was carried out in order to assess on a large scale populationthe antihypertensive effect of a month treatment with trandolapril Anotherobjectiveof this studywas to assessthe feasibility and the informativevalue of home SMBP in generalpractice Results presentedin this report will focus on feasibility, reproducibilityand BP level estimate using SMBP as compared to OBP PATIENTS AND METHODS Patients Patients of both sexes having mild to moderate essential hypertension were included in the study according to the following criteria Patients were included if they were between 8 and 5 years of age, had the presence of at least two additional cardiovascular risk factors (smoking habit, hypercholesterolemia, diabetes mellitus, prior cardiovascular disease), and had office diastolic blood pressure (DBP) in the range 5to mm Hg at the end of a -weekperiodwithout antihypertensive treatment Patients were excluded if they had any serious chronic disease, a known hypersensibility to angiotensin converting enzyme inhibitors, or a need for an antihypertensive drug other than the study drug during the study Downloaded from on November Study Design and Setting This was an open study performed in primary practices In both previously treated and untreated patients, the first phase of the study consisted of a -week period without treatment At the end of this period, eligible patients were given mg trandolapril once daily in the morning for weeks Patients were instructed to take the medication at 8: AM The protocol was approved by the Comit de Protection des Personnes se Pr&ant a la Recherche Biomdicale Written informed consent was obtained from each patient Blood Pressure Measurements OfficeBlood Pressure At each consultation, three consecutive measurements were performed by the physician using a mercury sphygmomanometer with the patientin the sitting position after a 5-rein rest The mean of these measurements defined the blood pressure of each patient at the end of the washout period and at the end of the treatment period Systolic BP (SBP) was taken at phase of Korotkoff sounds and diastolic BP at phase 5 of Korotkoff sounds The participating physicians were asked to perform BP measurements between 8: AM and ; AM SelfBlood Pressure Measurement Self-measurement of blood pressure (SMBP) was performed during the last week of both the washout and the treatment periods For each period, SMPwas planned for days chosen at the patients convenience Every day, a series of three consecutive measurements was requested in the morning before drug intake (8: AM) and repeated in the evening (8: PM), Measurements were performed in the sitting position, after a 5 min rest, using the A & D UA 5 device (A & D Engineering, Milpitas, CA), which is a printer-equipped, semiautomatic, digitized devicebased on the oscillometric method that has been previously validated by comparison to a random zero mercury sphygmomanometer Each patient was asked to write results of measurements in a booklet designed for the study and store all printouts in a sleeve of the booklet SMBP Data Management The first day of each period was considered as training and the corresponding values were not analyzed Aberrant values were deleted according to the following rules: ) DBP s or DBP = 5 mm Hg, SBP s or SBP = 5 mm Hg, pulse pressure s mm Hg; ) measurements performed out of predefine morning and evening time (outside the : to : AM range or : to : PM range); and ) missing systolic or diastolic value To assess the feasibility of SMBP in general practice, we used as an index the number of correct measurements available at the end of the washout period For days, the maximum number of expected measurements was 8, ie, three series of three morning measurements and three series of three evening measurements Patients were classified in four subgroups, defined as follows: ) excellentmeasurers: performanceof all 8measurements; ) good measurers: performance of at least two morning and two evening series comprising at least one measurement, ie, from to measurements; ) poor measurers: from one to four measurements; and ) bad measurers: absence of measurements For each patient, the mean of all the available home

3 CHATELLIERETAL AJH-JULY -VOL, NO measurements was taken as the home blood pressure valueand used for comparisonof home and officelevels Statistical Analysis Data are summarized as means * one standard deviation (SD) The difference between two quantitative variables was tested by the Student s paired or unpaired t test, as appropriate An analysis of variance was performed to compare more than two means Correlation coefficients were obtained by the least squares method A P < 5 was considered significant Calculations were performed with the SAS statistical package (SAS Institute Inc, Cary, NC) RESULTS Feasibilityof SelfBloodPressureMeasurementin General Practice One thousand seven hundredand ten patients( men and women), recruitedby general practitioners were enrolled into the study Their general characteristics are shown in Table Among these patients, 8 -ZOhadatleasttwo associatedcardiovascularrisk factors: personalor familialhistory of cardiovasculardisease (Yo), hypercholesterolernia(y), tobacco consumption (%) or diabetesmellitus (5) During the washout period, (%) patients did not perform any measurement, () were poor measurers, (8 ) were good measurers, and (%) performed all the 8 expected measurements The distribution of the number of measurementsperformed in the evening and in the morning during the washout period is shown in Table Among the 8 patients who performed at least one measurement, 5 (%) performed a multiple of three measurements both in the morning and in the evening Most of the patients (8/ 8, %) performed the same number of measurements in the evening andin the morning, (%) had more measurements in the morning than in the evening, and 8 (5Y) had more measurements in the evening than in the morning During the treatment period, performance was TABLE GENERALCHARACTERISTICSOF THE PATIENTSFIGURESARE MEAN (SD) OR NUMBEROF PATIENTS(PERCENTAGES) Age (years) No (%)of males Knowndurationof hypertension(years)* No (%)nevertreated Bodymassindex(kg/m ) No (%)of smokers OfficeSBP (mmhg) OfficeDBP(mmHg) * Values are median (range) Meanor Number 5 8 SBP: systolic blood pressure; DBP: diastolic blood pressure SD or Percentage 55% o- 5% % 5 Downloaded from on November slightly improved, as shown in Table A total of (%) patients did not perform any measurement, and 5 (Y) were excellent measurers Prediction of Compliance to Home Self Blood Pressure Measurement Patients were classified into four groups according to the quality of measurements during the washout period, as defined in the Methods section As shown in Table, gender, obesity, smoking habits and diastolic blood pressure were not related to the quality of measurements Age and systolic blood pressure were slightly higher in bad measurers compared to other groups The proportion of patients with newly diagnosed hypertension ( < year) was lower in the group of bad measurers than in the other groups Relationship Between Office and Self Measurement Blood Pressure A total of patients (5%) performed at least two morning and two evening series during the washout period (excellent and good measurers) Among these patients, had either office or SMBP measurements performed after the beginning of the treatment or an insufficient washout period They were excludedfrom correlation analysis Theremaining patients (%) performed on average 55 t 8 measurements Digit Preference During the washout period,,5 SMBP measurements and office BP measurements were available for analysis (Table ) A digit preference for O and 5 was evident in physician measurements The digit 5 was more prevalent in DBP than in SBP values (Y v?z ) During the treatment period, a digit preference for Oand 5 was still observed Nevertheless, the prevalence of digit 5 was the same (%) for SBP and DBP This digit preferencewas not observed with the semiautomatic device However, there was a tendency to provide more frequently evenfigures (55% and % for systolic and diastolic BP, respectively) than odd figures (P < for both) Comparison of Ofjice and Home Blood Pressure Measurements at Baseline Mean BP levels were t 5/ t mm Hg and 5 t 8/ * mm Hg for office BP and SMBP, respectively The mean differences were Y 5 mm Hg and 8 ~ mm Hg for systolic and diastolic BP, respectively Women exhibited a larger difference than men both on SBF (5 ~ mm Hg v f 5 mm Hg, P < ) and DBP (8 * mm Hg v Y ~ $) mm Hg, P < 5) Compared to nonobese patients, obese patients had a smaller SBP difference ( t 5 v 5 t mm Hg, P < 5) but a comparable DBP difference (8 * v f mm Hg, P = NS) Age, height, smoking habit, or duration of hypertension were not related to the extent of difference The correlation between the two methods was obtained by averaging the three measurements taken by the physician, and all the available SMBP measure-

4 AJH-JULY -VOL, NO SELFBP MEASUREMENT IN GENERAL PRACTICE TABLE CLASSIFICATIONOF PATIENTS(N = ) ACCORDINGTO THE NUMBEROF CORRECT MORNINGAND EVENINGSMBMEASUREMENTSAVAILABLEDURING THE WASH-OUT PERIOD AND THE TREATMENTPERIOD Washoutperiod No of measurements Treatmentperiod MorningSMBP o 5 8 MorningSMBP o 5 8 EveningSMBP o 5 8 Mean numberof measurementsforwashout period:?, median: Mean number of measurements for the treatment period: 5?, median: ments The relationship is described by the following equations: SBP (home) = SBP (office) + (r = 5, P < ) and DBP (home) = DBP (office) + (r = P < ) A comparison of BP recorded by physicians in the office and by patients at home was done according to the method of Bland and Altman8 In this method, the average of SMBP and physician measurements was plotted against their difference for both SBP and DBP Large differences were observed for some individuals, as shown in Figure A (SBP) and B (DBP) Larger differences at low BP values were observed for both SBP and DBP Influence on the Precision of BP Estimate ofvarious Strategies ofcalculating a Patient s Average SMBP This analysis was done on the patientswho correctlyperformed all the 8 SMBP measurements required in the protocol TABLE CHARACTERISTICSOF PATIENTSACCORDINGTO THE QUALITYOF MEASUREMENTDURING THE WASHOUT PERIOD ( PATIENTS) Characteristics Excellent Good Poor Bad F No of patients No (%)of men Age (years) No (%)of obesesubjects+ No (%)of smokers No with durationof hypertension< year InitialofficeSBP(mmHg) InitialofficeDBP(mmHE) (55) 5 ~ (5) 5() (8) (55) 5 k () 8 () () * P value is that obtained with a Xtest forcategorical variables or an F test for continuous variables 88 () 5 ~ (5) () (5) 8 t tobesity was defined by a body weight above % of the ideal body weight determined according to Lorentz SBP: systolic blood pressure; DBP: diastolic blood pressure Downloaded from on November (5) 58 ~ (88) 8() 8() t 5 * 8

5 8 CHATELLIERET AL AJH-JULY -VOL, NO TABLE DIGIT PREFERENCEOF PHYSICIANMEASUREMENTCOMPAREDTO MEASUREMENTDURING THE WASHOUTPERIOD SELF BLOOD PRESSURE Office Blood Pressure Self-Measured Blood Pressure (n = ) (n = 5) SBP DBP SBP DBP Digit* n % n % n % n % o * Last figure of blood pressure value duringthe washout period To determinethe minimum number of home available measurements necessary to provide an accuratebp evaluation,we used the following procedure ) For each patient, we randomly selected among the 8 measurements,,,,, or 5 values, or all 8 measurements) The individualblood pressure level was then defined by averaging these,,,,, 5, or 8 values ) We then compared the group mean andsd obtainedwith thesevariousdefinitionsof individual blood pressure level Results are shown in Table 5 The number of measurements averaged had no effect on the estimate of the mean home blood pressure of the whole group As expected, the lowest standard deviation was obtained when all 8 measurements were taken into account to estimate individual BP levels Compared to the SD obtained with only one measurement, the one obtained with 8 measurements was reduced by % and Y for DBP and SBP, respectively However, the gain in precision was modest when more than six measurements are taken into account When either morning or evening measurements were chosen, the results were not altered (data not shown) The SD of the blood pressure defined on the basis of a single physician measurement was lower than the one obtained by SMBP, whatever the number of measurements used for its definition This is partly explained by the use of aninclusion criterion based on officedbp This is shown by the comparison of coefficients of variation (SD /mean) of SMBP and physician BP values For one SMBP measurement, the SBP and DBP coefficients of variation were comparable: ~and 8YBy contrast, for one clinical measurement, the coefficient of variation was lower for DBP than for SBP (Yo v %), reflecting the narrow range of DBP (used as inclusion criterion) Downloaded from on November Office and SMBP Responses to Therapy A total of 8 patients who correctly performed SMBP during both periods and followed the protocol schedule were used for efficacy analysis Office and home BP levels during both periods are shown in Table Office SBP / DBP lowering was 8 f / 5 f 85 mm Hg (P < for both) Home SBP/DBP lowering was 8 t 8/8 * mm Hg (P < for both) These differences can be expressed as a percentage of baseline BP values For SBP, we observed a % decrease with office and a O %decrease with home measurements For DBP, a 5% decrease with office and a % decrease with home measurements DISCUSSION This large study shows that, among patients and general practitioners previously unfamiliar with the method, home measurement of blood pressure can be used in the management of hypertensive patients However, among patients recruited on an ambulatory basis, only / succeeded in providing at least two morning and two evening series of measurements In another study performed in our hypertension clinic, patients out of were able to provide more than 8% of required measurements As expected, results are slightly worse in this large scale study than in a research unit, where doctors were involved in a research program on SMBP and patients were selected and educated through a h teaching course We did not find strong predictors of the compliance to the method Patients with known hypertension and older patients tended to be bad measurers more frequently than the other patients Sex, obesity, and smoking habits were not associated with SMBP performance Systolic, but not diastolic, blood pressure was higher in bad measurers than in the other pa-

6 AJH-JULY -VOL, AJO SELFBP MEASUREMENT IN GENERAL PRACTICE m D ndo~ ra5 +mml y=x*5,e SJ - 5 E i DneSD -, % s - al m m w IW lm Jm a m tin DkWk moodpiwuwoimiihg) - : :; > - -: r-o, P=oaool Y-5X+5 CneSO -~k%---% Mem CneSD FIGURE Comparisonof home and office blood pressure according to method of Bland and Altman Mean blood pressure was obtained by averaging pressure measured by physician and by device and difference was calculated as pressure measured by physician minus pressure measured by device A Systolic blood pressure B Diastolic blood pressure -w - do ,,,-,*,W,W,ml,IIJ la km S@OlloBhod~m(ml+g) zoo A tients All these differences, although statistically significant, were small and of minor clinical relevance Usually, compliance with appointments tends to be better in older than in younger patients,lofllas is compliance to antihypertensive treatmentpatients with newly discovered hypertension tended to be less compliant with appointmentsl or treatment Male gender, smoking habit, and obesity are generally markers of poor compliancel Thus, markers of compliance with treatment and appointments may be different from those of a self-assessment method such as SMBP As usually reported, we found lower blood pressure values by SMBP than by physician measurements This may be due to either a problem in measurements or a true difference of blood pressure The validity of SBMP devices has been questioned and many devices have been shown to be inaccu- Downloaded from on November rate, *-lbwhich explains the recent publication of recommendations for validation Results obtained in by Jamieson et alb with the A & D would have qualified this device in the A category of the BHS protocollb The mean difference between simultaneous measurements of four observers and the device (from to -) and its SD (from to 8) was within the range ( * 5 mm Hg for the mean and t 8 mm Hg for SD) stated by the AAMI recommendations8 Besides the problems of precision and accuracy, we also found that the device tended to provide more frequently even figures than odd figures This was not previously mentioned for automatic devices These results should be put in perspective with the quality of the physician s measurements, which concerns both the device and the observer For example, Burke et al found that % of aneroid manometers

7 5 CHATELLIERET AL AJH-JULY -VOL, NO TABLE 5 NUMBER OF MEASUREMENTSAND CONSEQUENCESON THE PRECISION OF HOME BLOOD PRESSURE Mean SD Mean SD No of Measurements DBP SBP Home BP Office BP 8 55 had an error greater than ~ mm Hg when compared to mercury sphygmomanometers and that Yhad a cuff shorter than the length recommendedeven if the device and the conditions of measurement are appropriate, observer bias is still a problem In an epidemiological survey, regular and frequent retraining of technicians was necessary to decrease observer biaszo However, this training is not given to general practitioners In the present study, in addition to usual measurement bias, we found that digit preference was a major problem, showing that practitioners were used to measure blood pressure with a precision of 5 mm Hg Many studies have found, as the SMART study has, that SMBP was lower than office blood pressure The main characteristics of the largest studies are summarized in Table The variability of office-smbp differences may have many sources, such as various clinic blood pressure levels, differences in population selection, and devices accuracies One can expect greater differences in hypertensive patients compared to normotensives, since hypertensives are selected on the basis of elevated clinic pressures James et a5compared home-clinic differences in a small series of patients: the difference was 8/ in hypertensives versus 8 / in normotensives We found a lower difference in men than in women as found by Imai et al, but not in the Duebendorf study or in the study by Laughlin et algthe individual differences show a continuous distribution This does not allow a clear partition of the patients between true and white coat hypertensives It also shows how arbitrary would be the definition of white coat hypertension based on a threshold value for the difference between office and home blood pressure When multiplying the number of measurements used to define individual BP level, we observed a decrease of the population SD This decrease in total variability is due to the decrease of intraindividual variability Six measurements were sufficient to decrease total variability, independently of the time repartition of the measurements However, we observed a higher standard deviation using SMBP (one measure or even all 8 measurements) compared to casual measurements (Table 5) Several explanations can be proposed The first one is statistical When calculating the parameters of the distribution (mean and SD of the population) of blood pressure, one compares an open range of vaiues (SMBP) to a truncated distribution (office BP), as a consequence of patient inclusion criteria Consequently, the variance, which is a summary measurement of the spread of the distribution, is higher by definition with SMBP, which is not as constrained as office measurement Evidence is provided by comparing variabilities in studies with and without truncature In the study by James et al,5 the SD of diastolic blood pressure in hypertensive subjects was mm Hg at clinic and mm Hg at home, compared to mm Hg and mm Hg, respectively, in normotensive subjects In a substudy of the SYST-EUR trial based on ambulatory the SD Of systolic blood pressure measurement, diastolic blood pressure of patients taking placebo was / mm Hg at clinic versus 5/ mm Hg using h ambulatory blood pressure at baseline During follow-up, in the same group of patients treated by placebo, the SD increased to / mm Hg in the clinic, whereas it remained stable (5/ mm Hg) using ambulatory pressure We observed the same results (Table ) The SD of home BP slightly decreased between baseline and treatment period For office SBP, which was not an inclusion criterion, it remained stable For office DBP, it increased by ZO The tnmcated distribution could partly explain the lowercorrelationcoefficientbetweencasualbp andsmbp observedin the SMART study when compared to previous studies in unselected subjects sumrnarized in Table Anotherexplanationis probablydueto the poorerqual- TABLE OFFICE AND SMBP RESPONSES TO THERAPY SBP (mmhg) DBP (mmhg) Office Home Office Home Baseline * 8 5f 8 * 5 8 ~ Treatment * t 8* 8 85? 5 Downloaded from on November

8 AJH-JULY -VOL, NO SELFBP MEASUREMENT IN GENERALPRACTICE ity of physicians measurements compared to that obtained in studies conducted by research teams, An open study, such as the SMARTstudy, underlinesthe preferentialchoiceof DBP valuesto complywith inclusioncriteria: digit 5 was preferedfor DBP (inclusion criteria: 5 mm Hg) more frequentlythan for SBP (no inclusion criteria) duringthe washout periodbut not during the treatment period Inconclusion, improvement of accuracyby repetition of measurements and minimization of biases should prove to be useful for the assessment of blood pressure levels in the current management of hypertensive patients, as well as during therapeutic trials Compared to a single measurement, six measurements provide both increased precision and reasonable acceptabili~ (ie, six measurements are performed by the majority of patients) In cross-over studies, improvement in precision has clearly been shown to decrease the number of subjects to be includedzssmbp appears to be a cheap alternative to ambulatory blood pressure measurement for obtaining a precise and unbiased evaluation of a treatment effect in open studies Unfortunately, this method cannot be applied to all the patients, and bad measurers cannot be prospectively identified Downloaded from on November 5 8 REFERENCES MachfahonS, Peto R, Cutler J, et al: Blood pressure, stroke, and coronary heart disease Part Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias Lancet ;5:5 Collins R, Peto R, MacMahon S, et al: Blood pressure, stroke, and coronary heart disease Part Short term reduction in blood pressure: overview of randomized drug trials in their epidemiological context Lancet ;5:8-88 Pickering TG: The ninth Sir George Pickering memorial lecture Ambulatory monitoring and the definition of hypertension J Hypertens ; :- Abe H, Yokouchi M, Nagata S, et al: Relation of office and home blood pressure to left ventricular hypertrophy and performance in patients with hypertension High Blood Press ;:-85 James GD, Pickering TG, Yee LS, et al: The reproducibility of average ambulatory, home and clinic pressures Hypertension 88;:55-5 Chatellier G, Day M, Bobrie G, Mnard J: Feasibility study of N-of- trials with blood pressure self-monitoring in hypertension Hypertension 5;5: Jamieson MJ, Webster J, Witte K, et al: An evaluation of the A&D UA-5 semi-automated cuff-oscillometric sphygmomanometer J Hypertens ;8: 8 Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of measurement Lancet 8;i: Bobrie G, Day M, Tugay A, et al: Self blood pressure measurement at home Clin Exper Hypertens [A] ; 5():- Degoulet P, Mnard J, Vu HA, et al: Factors predictive

9 5 CHATELLIERETAL AJH-JULY -VOL, NO 5 8 of attendance at clinic and blood pressure control in hypertensive patients Br Med J 8;8:88- McClellan WM, Hall WD, Brogan D, et al: Continuity of care in hypertension Arch Intern Med 88;8:55 58 Consoli SM, Safar ME: Valeur predictive du profil psychologique du patient et du style de relation medecinmalade vis-h-vis de l observance d un traitement antihypertenseur (in French) Arch Mal Coeur 88; 8 (SUp@ HTA) :5-5 Stewart M,: The validity of an interview to assess a patient s drug taking Am J Prev Med 8;():5- O Brien E, Mee F, Atkins N, O Malley K: Inaccuracy of seven popular sphygmomanometers for home measurement of blood pressure J Hypertens ;8: Van Egmond J, Lenders JWM, Weernink E, Thien T: Accuracy and reproducibility of devices for selfmeasurement of arterial blood pressure Am J Hypertens ;() :8 8 Evans CE, Haynes RB, Goldsmith CH, Hewson SA: Home blood pressure-measuring devices: a comparative study of accuracy J Hypertens 8;():- O Brien E, Petrie J, Littler W, et al: The British Hypertension Society protocol for the evaluation of blood pressure measuring devices J Hypertens ; (SU@ ):S-S White WB, Berson AS, Robbins C, et al: National standard for measurement of resting and ambulatory blood pressures with automated sphygmomanometers Hypertension ; ():5-5 Burke MJ, Towers HM, O Malley K, et al: Sphygmoma- 5 8 nometers in hospital and family practice: problems and recommendations Br Med J 8;85:- Bruce NG, Shaper AG, Walker M, Wannamethee G: Observer bias in blood pressure studies J Hypertens 88;:5-8 Mancia G, Sega R, Bravi C, et al: Ambulatory blood pressure normalcy: results from the PAMELA Study J Hypertens 5;:- Weisser B, Griine S, Burger R, et al: The Diibendorf study: a population-based investigation on normal values of blood pressure self-measurement J Hum Hypertens ;8:- Mejia AD, Julius S, Jones KA, et al: The Tecumseh blood pressure study Arch Intern Med ;5:- Imai Y, Satoh H, Nagai K, et al: Characteristics of a community-based distribution of home blood pressure in Ohasama in northern Japan J Hypertens ; :- De Gaudemaris R, Chau NP, Mallion JM, for the Groupe de la Mesure, French Society of Hypertension: Home blood pressure: variability, comparison with office readings and proposal for reference values J Hypertens ;:8-8 Laughlin KD, Sherrard DJ, Fisher L: Comparison of clinic and home blood pressure levels in essential hypertension and variables associated with clinic-home differences J Chron Dis 8;:- Staessen JA, Thijs L, Mancia G, et al: Clinical trials with ambulatory blood pressure monitoring: fewer patients needed? Lancet ;:55 55 Conway J, Coats A: Value of ambulatory blood pressure monitoring in clinical pharmacology J Hypertens 8;(supp):S-S Downloaded from on November