Feasibility Study of N-of-1 Trials With Blood Pressure Self-Monitoring in Hypertension
Abstract The objective of this study was to assess individual responses to antihypertensive treatment by N-of-1 trials using blood pressure self-monitoring in 79 patients of both sexes referred to a hypertension clinic. Thirty-five patients who remained untreated (study 1) and 44 N-of-1 trial participants (study 2) were consecutively selected if their clinic blood pressure was between 160/95 and 220/115 mm Hg and there were no hypertensive complications. Blood pressure was measured daily at home for 21 days (three consecutive measures, morning and evening). Each N-of-1 trial was a single-blind treatment consisting of two successive 10-day treatment pairs, each pair comprising 5 days of placebo followed by 5 days of 20 mg enalapril once daily in the morning. Study 1 showed no significant blood pressure regression toward the mean over 20 days and justified the choice of 5-day treatment periods in study 2. In study 2, blood pressure fell significantly 12 hours after the first administration of enalapril and rose within 24 hours of the end of the 5-day active treatment period. Using evening blood pressure values (12 hours after enalapril intake) from the first treatment pair, 33 patients were classified as responders (diastolic blood pressure fall ≥6 mm Hg). In 16 of these 33 patients, the fall in blood pressure above 6 mm Hg was not maintained in the morning, 24 hours after drug intake. Response reproducibility was tested by comparison with the second treatment pair: the observed agreement was only 0.71 (chance-corrected agreement: 0.34) when defined according to both evening and morning values. N-of-1 trial methodology can be useful for decision making in the care of individual patients. It can be based on the use of home blood pressure determinations, provided that there are at least 30 readings in each 5-day trial period. The individual agreement between the antihypertensive responses to two successive enalapril treatment periods is only moderate and is not sufficient to justify using this design for performing in individual patients randomized comparison of an angiotensin-converting enzyme inhibitor and another antihypertensive drug.
- randomized controlled trials
- antihypertensive therapy
- angiotensin-converting enzyme inhibitors
- self care
- monitoring, physiologic
- blood pressure monitors
The increasing number of effective and well-tolerated antihypertensive drugs has not eradicated the problem of poor blood pressure (BP) control.1 Among the various strategies that could improve the quality of hypertension control, sequential monotherapy2 and N-of-1 trials3 4 have been proposed as alternatives to the usual stepped-care programs. These strategies have not yet been tested, even though their basic design, the crossover trial, has been applied in antihypertensive drug studies. This design has the advantage of minimizing the number of subjects needed to compare the BP effect of two different drugs.5 It also has two limitations: time and carry-over effects. When these effects are present, the advantages of this design are lost because data analysis must be restricted to the first period only.6 7 The successive exposure of the same patient to two different drugs has the advantage of constituting sequential monotherapy: ideally, at the end of a crossover trial, one could select for each participant the most effective or best-tolerated drug. This choice of therapy should allow better BP control than the systematic prescription of one of the drugs to all patients.8
Even if the five major first-line antihypertensive drug classes have comparable efficacies at the group level, they have greatly varying efficacies at the individual level. This variability in patient response explains why in most trials the standard deviation of the drug-induced BP fall is similar to the average fall itself.9 10 Consequently, in practice, a clinician is unable to predict which first-line drug will have the best efficacy in most cases of mild to moderate uncomplicated essential hypertension. Moreover, he or she has major difficulties analyzing an individual BP response because of the placebo effect and BP variability. The “N-of-1” methodology is appealing in this context because it combines individualization of treatment choice and unbiased evaluation of treatment efficacy.
In this report, we explore the possibility of applying the method of randomized trials in individual patients (N-of-1 trials), proposed for use in various diseases by Guyatt et al,3 4 to hypertensive patients. The purpose of such N-of-1 trials in hypertension management would be to allow selection of first-step treatment in an individual hypertensive patient by testing precisely two drugs likely to have contrasting effects on BP, eg, diuretics and β-blockers, or angiotensin-converting enzyme (ACE) inhibitors and calcium channel blockers.5
To explore the feasibility of this approach, we designed two pilot studies aiming first to determine patient acceptability of such methodology and the minimal duration of treatment periods, and second to assess the reproducibility of treatment effect in the individual patient. To achieve this last goal, patients underwent two successive treatment pairs, each including a 5-day placebo period and 5-day enalapril period. We selected enalapril as the test drug because it had been shown to be immediately effective, without reinforcement of or escape from its initial effect.11 We chose BP self-measurement at home as the appropriate method for increasing the number of BP determinations available for statistical analysis.12 13
This investigation included two successive studies. In study 1, the variance components of BP and the choice of treatment period duration were determined in 35 hypertensive patients who remained untreated during the measurements. In study 2, a series of individual trials of identical design was performed after study 1 completion in 44 other consecutive patients.
For both studies, patients of both sexes were selected from referrals to the Hypertension Clinic at Broussais Hospital between April 1991 and April 1992 according to the following criteria: age between 20 and 80 years; elevated auscultatory BP levels (systolic BP 160 to 219 mm Hg or diastolic BP 95 to 114 mm Hg), with or without treatment, at the first consultation in the clinic; absence of cardiovascular complications (stroke, coronary heart disease), plasma creatinine level lower than 175 μmol/L, and absence of arrhythmia; absence of reasons to suspect potential compliance problems (depression, alcoholism, poor comprehension of the French language); attendance at a 1-hour teaching session on hypertension and BP self-measurement; willingness to return to the clinic 3 weeks later; no antihypertensive drug treatment at the start of the trial (no calcium channel blockers in the preceding 48 hours and no other antihypertensive medication in the preceding 2 weeks); and acceptance of the individual trial program for patients included in study 2.
Patients were eventually included in the study when they satisfied two home BP measurement criteria: (1) More than 80% of the scheduled BP self-measurements at home had to be obtained and valid. Measurements were considered invalid when heart rate was less than 40 beats per minute or greater than 120 beats per minute and/or systolic and diastolic BP readings were not compatible (less than 20 mm Hg difference); (2) Mean systolic pressure was greater than or equal to 130 mm Hg and/or mean diastolic pressure was greater than or equal to 80 mm Hg to take into account the generally lower BP values obtained at home14 15 during the first 5 days of measurement (days 2 through 6).
Table 1⇓ shows the general characteristics of all patients included in this report.
Each patient attended a 1-hour teaching session18 that included explanations concerning the basic principles of BP self-measurement, the meaning of BP values, and the use of the monitoring device. Each patient measured his or her BP at least three times in the presence of the physician, who verified in each case the correct use and functioning of the device. Patients were instructed to measure their BP three consecutive times in the sitting position in the morning before breakfast and in the evening after dinner. In addition, it was explained to each study 2 participant that he or she would take one tablet per day at breakfast, after the morning BP measurements, from a pill box containing one tablet per day: either 20 mg enalapril or its placebo. Patients were given clear explanations and reassurances concerning this medication, which has been widely prescribed since 1985. The patients were aware that they would not be actively treated every day of the 3 weeks, but they had no indication of which days corresponded to active treatment and which days corresponded to placebo. Each patient was provided with a typed pamphlet including general information about the use of the device, enalapril, and the timing of drug intake and BP measurements.
Design of the N-of-1 Trial
Each single-blind individual trial consisted successively of 6 days of placebo-enalapril (allowing 1 day of adaptation to the BP measurement method plus the standard period of 5 days) designated P1, 5 days of 20 mg enalapril (E1), 5 days of placebo-enalapril (P2), and 5 days of 20 mg enalapril (E2). The choice of 5-day periods, two pairs, and enalapril were made for the following reasons, respectively: (1) The 5-day period was a compromise between acceptability of the length of the individual trial and the reduction of variance allowed by the repetition of measurements with time (see “Results”). (2) The second identical pair (P2/E2) was included to test the reproducibility of the BP response obtained during the first treatment pair (P1/E1) and to check that BP returned immediately to a similar baseline state (P2 versus P1) following abrupt enalapril cessation after 5 days of administration. (3) The pharmacological characteristics of 20 mg enalapril, ie, rapid onset and offset of action and no serious side effects, made it a suitable candidate drug. Placebo-enalapril tablets were necessarily identical in appearance to the 20-mg enalapril tablets. All tablets and pill boxes were kindly provided by Merck, Sharp & DohmeChibret.
Active renin was measured with an immunoradiometric assay in blood obtained after patients had stood for 1 hour just before starting the study.19
Patients returned their automatically printed BP measurements for analysis after 3 weeks of measurements. Data were entered on an Apple Macintosh microcomputer into a previously prepared program identical for each patient. Day 1 measurements were considered as training measurements and were not included in any calculations. Global means (±1 SD) were automatically calculated for systolic and diastolic BP values for each of the four 5-day periods (30 measurements). Means (±1 SD) were also calculated for the morning and evening values for each period in each patient (15 measurements), allowing investigation of drug efficacy over 24 hours in study 2.
Data are shown for both systolic and diastolic BP; statistical calculations are only presented for diastolic BP because results were similar for both pressures.
Data are mean±1 SD. Means were compared using a paired or unpaired Student’s t test as appropriate. Percentages were compared using a χ2 test. Standard statistical procedures and ANOVA were performed with SAS statistical software.20
ANOVA was used to test the effect of different sources of variation on BP level. In study 1, the following effects and their first-order interactions were considered: period: a period was a 5-day interval; day: a day was the 24-hour interval during which BP was measured morning and evening; there were 5 days for each period; time of day: time of day corresponded to either the morning or evening BP values.
For the series of N-of-1 trials (study 2), pair, period (four periods for each patient, during which either enalapril or placebo was taken), day, and time of day effects and their first-order interactions were considered. A pair was the 10-day interval during which a patient was given first placebo and then enalapril; there were therefore two pairs in each individual trial.
A diastolic BP fall greater than or equal to 6 mm Hg was selected as the threshold for differentiating responders from non-responders. Within a Gaussian distribution of BP responses, such a choice is arbitrary but corresponds to the mean difference in office measurements between placebo and active treatment in large-scale therapeutic trials.21
The acceptability of our individual trials (study 2) was analyzed in a subsample of 51 consecutive patients to whom it was proposed. Forty-eight agreed to participate in such a trial and 39 (76%) completed it successfully (more than 80% of requested measurements obtained and valid). Of the 9 patients who were unable to complete the trial successfully, 2 abandoned, 1 because of hypotension when the first enalapril tablet was ingested after 2 days of debilitating gastroenteritis, and 1 whose high BP levels during the first days of the trial motivated a consultation with her general practitioner. The other 7 patients completed the trial, but less than 80% of the requested measurements were obtained and valid. Of these seven, 3 misunderstood the timing of BP measurement, 3 were unable to perform enough BP measurements because of extremely busy lifestyles and possibly questionable motivation, and the last patient did not return for a consultation after delivering his printed measurements.
Study 1: Analysis of Self-Measured BP Values at Home
Components of the Variance and Regression to the Mean Effect
ANOVA showed no period and no day effect. There was a significant time of day effect: BP was slightly higher (by 1.3 mm Hg) in the morning than in the evening. As shown in Table 2⇓, although no significant regression to the mean was observed, the evening BP fell slightly, mainly from period 1 to period 2. This small effect was further demonstrated by analyzing the components of variance. When we applied the model used for the series of individual trials (study 2) to the 35 subjects observed without treatment (study 1), 60.6% of the variance was attributable to between-subject variance. For the within-subject variance (39.4%), 0.25% was due to pair effect, 8.9% to period effect, 0.25% to day effect, 48.0% to time of day effect, and 42.3% to model error.
Selection of the Number of Days and Measurements Necessary To Minimize the Variability of Difference Between Two BP Values in the Absence of Treatment
To determine the influence of the number of measurements on the variability of the difference between two mean BP values, we calculated the standard deviation of the difference between two means derived from increasing numbers of evening BP measurements over two 10-day intervals in these untreated patients. A first difference was calculated between the first BP taken on the second day and the first BP taken on the 12th day. A second difference was calculated between the mean of the first two BP measurements taken on the second day and the mean of the first two BP measurements taken on the 12th day. This procedure was repeated until a last (30th) difference calculated between the mean of the 30 BP measurements taken from the second to the 11th day and the mean of the 30 BP measurements taken from the 12th day to the 21st day. Results are shown in Fig 1⇓. The maximal reduction in the standard deviation of the difference between two mean BP values was obtained when each mean was defined by the maximal 30 measurements (3 measurements for 10 consecutive days). Fig 1⇓ shows that 80% of this maximal reduction was already obtained with mean values defined by 15 measurements (3 measurements collected over 5 days). Therefore, a period of 5 consecutive days of BP measurement was considered sufficient to accurately detect a drug-induced fall in BP in a single patient.
Study 2: Series of Individual Trials
The objective of an N-of-1 trial is to allow an individualized decision to be made based on the data collected in a single patient. Therefore, each individual trial was analyzed separately. A typical group analysis was then performed to verify that enalapril had its expected antihypertensive effect.
Analysis at the Individual Level
Analysis of the first pair. Fig 2⇓ shows the distribution of the enalapril-induced diastolic BP falls in the evening and morning. For practical purposes, it was necessary to define responders and non-responders, although the distribution of BP falls was continuous. We classified the patients on their diastolic BP values measured in the evening, 12 hours after drug intake. Thirty-three patients (75%) had a fall in diastolic BP greater than or equal to 6 mm Hg and were classified as responders to enalapril. Table 3⇓ compares responder characteristics with those of non-responders. Responders had higher clinic diastolic BP (P=.059) and slightly higher mean diastolic BP during the 5 days of placebo (P=.095). The only statistically significant difference concerned supine active renin level (after logarithmic transformation), which was significantly higher in responders than in non-responders (P<.01).
Duration of effect. The morning/evening effect was studied in the first treatment pair. Only 17 patients among 44 were classified as both morning and evening responders, 16 were responders only in the evening, 2 were responders only in the morning, and 9 were both morning and evening non-responders. Forty-nine percent of the patients whose BP responded to ACE inhibition (75% of the initial 44 patients) were not covered for 24 hours by 20 mg enalapril once daily.
Reproducibility of BP response between two successive treatment pairs. The reproducibility of the BP response between two successive treatment pairs in the same individual was assessed by applying the same criteria to both treatment pairs (ie, an enalapril-induced diastolic BP fall ≥6 mm Hg). Using evening diastolic BP, 21 patients were classified as responders during the second enalapril period, compared with 33 during the first enalapril period. Table 4A⇓ shows the concordance between the two classifications. The observed agreement between the two methods was 0.59. The κ value was 0.20, meaning that there was only poor agreement beyond that due to chance alone between the responses observed in both pairs.22
Using a more stringent criterion, ie, diastolic BP fall greater than or equal to 6 mm Hg for both evening and morning diastolic BP values, 12 patients were classified as responders during the second enalapril period, compared with 17 during the first enalapril period. Table 4B⇑ shows the concordance between the two classifications. The observed agreement between the two methods was 0.71. The κ value was 0.34, meaning that there was moderate agreement beyond that due to chance alone between the responses observed in the two pairs.22
Analysis at the Group Level
Overall effect. Fig 3⇓ shows the day-by-day overall treatment effect. For each patient, a day was defined by the mean of six BP measurements, and Fig 3⇓ was obtained by averaging individual means. Means obtained during the four periods are shown in Table 2⇑. Enalapril lowered systolic and diastolic BP values significantly, and its effect was detectable as soon as 12 hours after the first drug intake.
For diastolic BP, ANOVA showed no pair (P=.42) and no day (P=.79) effects but a period effect (P<.0001), and a time of day effect (P<.0001). A treatment by time of day interaction was present (P<.001), indicating that the difference between morning and evening BPs was not the same during placebo and enalapril treatment. During placebo, morning and evening BP values were comparable in these 44 patients (morning minus evening BP: 0.3 [5.7] mm Hg), whereas during enalapril, morning BP was higher than evening BP (morning minus evening BP: 3.8 [6.6] mm Hg), which is explained by the fact that the effect of enalapril did not last for 24 hours in all responders.
Reproducibility of enalapril effect. The second enalapril period also decreased BP (Fig 4⇓), and the effect was again detectable 12 hours after administration of the first active tablet. ANOVA showed a period-by-pair interaction (P=.014), indicating that the same dose of enalapril did not have the same hypotensive effect in the two treatment pairs. The diastolic BP fall (P1−E1) in the first pair was 9.1 (6.0) mm Hg, whereas that measured during the second pair (P2−E2) was 5.1 (5.8) mm Hg. This difference in the BP fall between the two pairs was attributable to both a lower initial BP level during the second placebo period (P2) and a smaller hypotensive effect during the second enalapril period (E2). The magnitude of the difference between the two identical periods of each pair did not depend on the BP level, neither for enalapril (E2−E1) nor for its placebo (P2−P1).
As pointed out by Guyatt et al,3 even when randomized controlled trials provide a valid answer to a therapeutic question, the overall result may not apply to an individual patient: a precise and individual assessment of the effect of an intervention is always necessary. The response to a given antihypertensive drug varies greatly among patients,2 23 and the signal of the drug-induced fall in BP remains difficult to extract from the noise caused by random BP variations.12 24 It is therefore worthwhile to look for new techniques that might improve the quality of hypertension management at the individual level, by improving the signal-to-noise ratio. We aimed to determine whether N-of-1 methodology, supported by a precise measurement method, could be used to achieve this. The key issues of our series of individual trials were (1) the choice of BP self-measurement at home as the measurement instrument, (2) the choice of the ACE inhibitor enalapril as the intervention drug, (3) the study design involving a 3-week protocol with two successive pairs of placebo-active treatment, and (4) the individual enalapril response reproducibility in these conditions.
BP Self-Measurement at Home as the Measurement Instrument
An appropriate measurement method must be sensitive (responsive), valid, and precise. BP self-measurement at home has been shown to be a sensitive tool, ie, able to detect small changes.25 26 However, the BP fall detected by this method is smaller than that observed with clinic measurements, where BP is generally higher.13 The precision of the measurement method has a major influence on the quantification of the hypotensive effect and consequently on the physician’s treatment decisions. Several reports have pointed out that increasing the number of BP values by continuous monitoring allows reduction in the number of subjects necessary for meaningful results in drug trials.12 27 Precision improvement is not only a valuable goal in drug research, it should also be considered a key element in managing the individual hypertensive patient. Study 1 confirmed that the steady state necessary for performing an N-of-1 trial was achieved during a 3-week period in patients with mild to moderate hypertension after exclusion of the first day of measurement. We therefore conclude that daily home BP measurements for 3 weeks fulfill the necessary methodological requirements for assessing an individual BP response to drug treatment.
Enalapril as the Study Drug
Enalapril was chosen because the pharmacokinetic and pharmacodynamic heterogeneity of BP response to this drug class has been well documented. For most antihypertensive drugs, there are no precise data regarding the timing of onset and of maximal effect and the duration of persistent action after therapy cessation. Meredith et al11 demonstrated that the first dose of enalapril was effective and produced effects similar to those measured after 7 days and 1 month of treatment. The pooling of our individual trials allowed the analysis of BP changes every 12 hours in a group of 44 patients over 20 consecutive days. This provided a complete picture of the time course of an enalapril-induced fall in BP when prescribed once a day. The maximal hypotensive effect is already detected in the evening, 12 hours after enalapril intake, and BP returns toward its initial level within 24 to 36 hours when enalapril is abruptly withdrawn after 5 days of therapy. Treatment cessation after such a short treatment course has a different effect from cessation after several weeks or months of treatment, where a slow return of BP toward initial values is frequently observed, at least with the classic methodology of BP monitoring during clinic visits scheduled at regular intervals.28
Study Design and Analysis
Our individual trials were 3 weeks long and included two 10-day treatment pairs. The relatively short duration of treatment periods and the rapid shift from placebo to active drug were chosen to maximize the likelihood that all patients would participate in the entire trial. When a patient complied with a 3-week program, the number of measurements did not decrease with time, and when a patient did not comply with the program, this was evident from early on. This methodology was successfully used by 75% of the patients selected for the present work, a proportion similar to that observed in a previous study in patients referred to the same hypertension clinic.29 These results support the feasibility of this design.
In our patients, the distribution of the hypotensive effect of enalapril was unimodal (Fig 4⇑). Therefore, the distinction between responders and non-responders appears as arbitrary as the distinction between hypertensive and normotensive subjects. In both cases, a binary classification is necessary to decide in practice whether a patient is hypertensive and whether he or she benefits from a particular drug. The 6 mm Hg threshold was selected because it was the average BP fall observed in treated patients in an overview of large-scale trials19 and identified 33 responders. The presence of the expected difference in initial renin levels between responders and non-responders supports the validity of this classification method.23 A classification based on a Bayesian approach30 resulted in comparable results (data not shown).
In 16 of the 33 responders, the enalapril-induced BP reduction did not last 24 hours. These results confirm those of Meredith et al,11 who pointed out that twice-daily administration of a lower enalapril dose might be preferable to once-daily administration of a higher dose, an observation also made with other ACE inhibitors.31 These results suggest that both the choice of drug and its daily administration schedule should be tailored for each patient, even with drugs labeled as once-daily drugs. Home BP self-monitoring is clearly superior in this respect to clinic BP measurement, which cannot be routinely performed just before drug intake or precisely at the peak effect. This could provide each patient with an optimal therapeutic schedule and improve the prevention of early rises in morning BP.
Individual BP Response Reproducibility
The last goal of the present study was to test the reproducibility of short-term individual BP responses to a given drug (absence of period effect), a prerequisite before performing classic N-of-1 trials involving the successive administration of two different drugs. The efficacy of enalapril was easy to demonstrate for both pairs at the group level, but the individual agreement between the two responses to enalapril was only moderate at best. However, this should be interpreted in the context of usual clinical practice where clinical decisions (to treat or not, to alter drug treatment or not, etc) are made on the basis of relatively few measurements or even sometimes a single measurement performed at the physician’s office. The lack of reproducibility of BP response to enalapril is partly due to BP fluctuation around the decisional threshold32 and partly to a treatment-period interaction, ie, a significantly smaller BP response to enalapril in the second period, a well-known limitation of crossover trials.6 7 Consequently, in an N-of-1 trial over this time interval, in which two drugs with different mechanisms of action would be successively tested against their respective placebos, the BP response to the second drug would be at risk of being systematically underestimated. Can this limitation be overcome? Our trial design included a very short second placebo period (5 days), during which the BP level did not return completely to the level observed during the first placebo period. This is a first factor that reduced the observed BP fall with the second enalapril administration. Since adherence to the protocol was undiminished with time in compliant patients and home BP monitoring has already been reported to be feasible over 9 consecutive weeks,33 it certainly would be informative to design trials with longer placebo or active treatment periods. The patients’ task could be facilitated if BP measurements were restricted to the last 5 days of 2-week or 4-week pairs. Second, our choice of an ACE inhibitor may have unmasked a specific drug-related phenomenon that decreased the effectiveness of its second administration. During repeated ACE inhibitor administration, ACE induction34 and the rise in renin and angiotensin I35 can cause drug escape, with reappearance of plasma angiotensin II and minimization of the fall in BP, as already reported with captopril36 and spirapril.37 Increasing the treatment duration would allow estimation of the BP fall at a new steady state. Along with the use of longer treatment pairs, studies of other antihypertensive drug classes are necessary to fully investigate the usefulness of this technique in individualized hypertension management.
This first attempt to examine whether randomized trials in individual patients could be used for drug selection in the individual hypertensive patient has highlighted the potential advantages and limitations of this methodology. (1) BP self-measurement at home over several days in the morning and evening can provide evidence of the steady state necessary before a drug-induced BP fall can be assessed precisely. Independent of the performance of an N-of-1 trial, this measurement method is able to improve the appropriateness of the choice of the first-line treatment and also the drug administration schedule. (2) To perform an N-of-1 trial comparing two drugs in the same patient, a prolongation of treatment periods or of both treatment and placebo periods is likely to be necessary. This would have the advantages of increasing the washout period after the first active drug administration period and the number of measurements and also of minimizing the period effect that would bias the comparison of the BP effects of two different drugs prescribed successively. However, these advantages must be weighed against the limits of patient acceptability, when periods of treatment are too long. (3) Enalapril, 20 mg once a day, was selected for testing the feasibility of the N-of-1 trial in hypertension and was effective in 79% of the patients. It did not control BP for 24 hours in all patients. The smaller hypotensive response of the second course of treatment may be the equivalent of the time-dependent decline previously observed during the first days of captopril treatment37 and has contributed to the unsatisfactory reproducibility of two successive identical treatment periods. (4) Other drugs, especially some of the calcium blockers, could be tested in this kind of trial. Drug pharmacokinetic and pharmacodynamic characteristics should be taken into account when the test drugs, their daily dose, and the most appropriate duration of the treatment pairs are selected.
This work was supported by grants from Association Claude Bernard. The authors thank Dr Gordon Guyatt (McMaster University, Hamilton, Canada) for his constant support during this investigation.
Reprint requests to Pr Joël Ménard, Centre de Médecine Préventive Cardio-Vasculaire, Hôpital Broussais, 96 rue Didot, 75674 Paris Cedex 14, France.
- Received July 5, 1994.
- Revision received August 3, 1994.
- Accepted October 3, 1994.
Guyatt GH, Keller JL, Jaeschke R, Rosenbloom D, Adachi JD, Newhouse MT. The N-of-1 randomized controlled trial: clinical usefulness. Our three-year experience. Ann Intern Med. 1990;112:293-299.
Chalmers J, Wing L, Taylor S. Analysis of time trends, individual subject responses and background variation in crossover factorial studies with antihypertensive drugs. J Hypertens. 1990;8(suppl 4):S27-S35.
Ménard J, Brunner HR, Waeber B, Plouin P-F, Burnier M. Individualization of antihypertensive therapy. Hypertension. 1988;12:526-528.
Materson BJ, Reda DJ, Cushman WC, Massie BM, Freis ED, Kochar MS, Hamburger RJ, Fye C, Lashman R, Gottdiener J, et al, for the department of Veterans Affairs Cooperative Study Group on Antihypertensive Agents. Single drug therapy for hypertension in men: a comparison of six antihypertensive agents with placebo. N Engl J Med. 1993;328:914-921.
Ménard J, Serrurier D, Bautier P, Plouin PF, Corvol P. Crossover design to test antihypertensive drugs with self-recorded blood pressure. Hypertension. 1988;11:153-159.
Cesana G, De Vito G, Ferrario M, Libretti A, Mancia G, Mocarelli P, Sega R, Valagussa F, Zanchetti A. Ambulatory blood pressure normalcy: the Pamela study. J Hypertens. 1991;9(suppl 3):S17-S23.
Asmar R, Sassano P, Demolis P, Ménard J, Safar M. Validation de l’appareil d’automesure de la pression artérielle ‘Oscillomat.’ Presse Med. 1991;20:551-555.
Ménard J, Guyenne TT, Corvol P, Pau B, Simon D, Roncucci R. Direct immunometric assay of active renin in human plasma. J Hypertens. 1985;3(suppl 3):S275-S278.
SAS Institute Inc. SAS Stat User Guide, Version 6. Cary, NC: SAS Institute Inc; 1989:891-896.
Collins R, Peto R, MacMahon S, Hebert P, Fiebach NH, Eberlein KA, Godwin J, Qizilbash N, Taylor JO, Hennekens CH. Blood pressure, stroke and coronary heart disease, part 2: short-term reductions in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet. 1990;335:827-838.
Brennan P, Silman A. Statistical methods for assessing observer variability in clinical measures. BMJ. 1992;304:1491-1494.
Laragh JH, Lamport B, Sealey J, Alderman MH. Diagnosis exjuvantibus: individual response patterns to drugs reveal hypertension mechanisms and simplify treatment. Hypertension. 1988;12:223-226.
Julius S. Home blood pressure monitoring: advantages and limitations. J Hypertens. 1991;9(suppl 3):S41-S47.
Medical Research Council Working Party on Mild Hypertension. Course of blood pressure in mild hypertension after withdrawal of long term antihypertensive treatment. BMJ. 1986;293:988-992.
Bobrie G, Day M, Tugayé A, Chatellier G, Ménard J. Self blood pressure measurement at home. Clin Exp Hypertens. 1993;15:1109-1119.
O’Brien E, O’Malley K, Fitzgerald D. The role of home and ambulatory blood pressure recording in the management of hypertension. J Hypertens. 1985;3(suppl 1):35-39.
Juillerat L, Nussberger J, Ménard J, Mooser V, Christen Y, Waeber B, Graf P, Bellet M, Brunner HR. Determinants of angiotensin II generation during converting enzyme inhibition. Hypertension. 1990;16:564-572.
Van Den Meiracker AH, Man In’t Veld AJ, Admiraal PJJ, Ritsema Van Eck HJ, Boomsma F, Derkx FHM, Schalekamp MADH. Partial escape of angiotensin converting enzyme (ACE) inhibition during prolonged ACE inhibitor treatment: does it exist and does it affect the antihypertensive response? J Hypertens. 1992;10:803-812.