Antihypertensive treatment and compliance

Author�s reply

We appreciate the many responses to our article and wish to respond to some of the points that have been made.⁽¹⁾

Bland finds that categorising the response to treatment by continuous ambulatory blood pressure readings is significantly different from categorising by clinical blood pressure readings if McNemar�s test is used. His analysis supports the design of our study¾ that is, the use of continuous ambulatory blood pressure to assess the response to treatment rather than clinical blood pressure alone. We strongly believe that the use of clinical blood pressure readings to categorise patients as being resistant to treatment leads to overestimation of resistance and thus to substantial bias. Furthermore, improving compliance in patients who are are resistant to treatment in terms of clinical blood pressure readings but not continuous blood pressure readings might lead to significant overtreatment and thus adverse effects of drugs due to hypotension.

Parienti and Burnier and Brunner ask about differential effects of compliance monitoring on blood pressure in subgroups in our study. This point is well taken, and blood pressure improves more significantly in patients who are resistant to treatment than in those who are responsive. Schroeder et al, Cramer, and Burnier and Brunner mention a paper by Burnier et al that was published after our manuscript was finalised for publication. This paper describes how clinical blood pressure readings improved with compliance monitoring in some of the 41 patients who were resistant to treatment.⁽²⁾ Yet, in our and Parienti�s additional analysis and the study by Burnier et al, blood pressure readings of most patients remain unchanged or resistant to treatment even after compliance monitoring.

Bland and Burnier and Brunner write that using binary data and corresponding tests reduces the statistical power of our analysis. We agree that use of binary data may reduce statistical power. Using binary data is general clinical practice and was therefore presented in our study. Nevertheless, an analysis not presented in our paper shows that blood pressures broken down as a continuous variable versus response to or compliance with treatment does not change our results or the conclusions of our paper.

Burnier and Brunner wonder why patients categorised as non-compliant could have had normal blood pressures. They err in including patients with white coat hypertension among those classified as resistant to treatment in our study. Continuous blood pressure monitoring was performed before and after compliance assessment in all our patients, and thus no patient with white coat hypertension was classified as resistant to treatment. However, most patients categorised as non-compliant were partially compliant, similar to nearly all patients in studies of this issue. We agree with Schroeder et al that partial compliance may induce sufficient therapeutic responses depending on the pharmacological properties of the antihypertensive drug. Therefore a cut-off point of 80% for compliance has been widely used in the hypertension literature and was used in our study.⁽³⁾ Furthermore, normalisation of blood pressures is common with placebo treatment or after discontinuation of longstanding treatment.⁽⁴⁾ ⁽⁵⁾ Regression to the mean occurs in any observational collective and therefore requires a control group. As Schroeder et al mention, the study by Burnier et al did not include a control group.⁽²⁾

We entirely agree with Burnier and Brunner that antihypertensive treatment and adequate compliance is pivotal for blood pressure control. They are, however, wrong in assuming that our paper is misleading. Our main point is that compliance rates in hypertensive patients who are resistant to treatment are better than often assumed. In fact, Burnier et al observed better compliance rates than we did¾ above 90%¾ yet a majority of their patients remained resistant to treatment despite compliance monitoring, becoming responsive only after treatment was adapted.⁽²⁾

Jacobs asks for a power analysis. The power analysis has been given in the discussion section of our paper.

We agree with Parienti, Schroeder et al, and Burnier and Brunner that compliance and response to treatment fluctuate. Therefore we selected patients receiving stable antihypertensive treatment and performed continuous blood pressure monitoring twice to ascertain adequate classification of the response to treatment. Obviously, our study, like any other study, glimpses only a specific period of the patients� lives. We agree with all respondents that this subject needs further creative and unprejudiced research.

Edouard Battegay
assistant professor of internal medicine
ebattegay{at}uhbs.ch

Reto Nüesch
senior registrar

Benedict Martina
head of medical emergencies

Thomas Dieterle
senior fellow

Outpatient Department and Hypertension Clinic of Internal Medicine, University Hospital, CH-4031 Basel, Switzerland

Electronic responses. Relation between insuficient response to antihypertensive treatment and poor compliance with treatment. bmj.com 2001;323 (www.bmj.com/cgi/eletters/323/7305/142; accessed 1 Nov 2001).
Burnier M, Schneider MP, Chioléro A, Fallab Stubi CL, Brunner HR. Electronic compliance monitoring in resistant hypertension: the basis for rational therapeutic decisions. J Hypertens 2001;19:335-41.
Hill MN. Adherence to antihypertensive therapy. In: Izzo JL, Black HR, eds. Hypertension primer. The essentials of high blood pressure (American Heart Association). Baltimore: Lippincott, Williams and Wilkins, 1999:348-51.
Preston RA, Materson BJ, Reda DJ, Williams DW. Placebo-associated blood pressure response and adverse effects in the treatment of hypertension. Arch Intern Med 2000;160:1449-54.
Aylett M, Creighton P, Jachuck S, Newrick D, Evans A. Stopping drug treatment of hypertension: experience in 18 British general practices. Br J Gen Pract 1999;449:977-80.