ESCAPE-ancillary blood pressure measurement study: end-digit preference in blood pressure measurement within a cluster-randomized trial Jean-Pierre Lebeau a , Denis Pouchain b , Dominique Huas a , Franck Wilmart c , Clarisse Dibao-Dina a and Florent Boutitie d ; The French National College of Teachers in General Practice Background In a cluster-randomized trial including 1832 hypertensive patients, all 126 general practitioners (GPs) in the intervention group (IG) used an oscillometric device that was provided for blood pressure (BP) measurements. Of the 131 GPs of the control group (CG), 24.6% used an oscillometric device (OCG), and 75.4% used a manual device (MCG). At baseline, patients in the IG and CG were comparable for all clinical and biological characteristics, except BP, which was higher in the IG (146/84 vs. 139/ 81 mmHg; P < 0.001). The purpose of this ancillary study was to assess whether these differences in BP values were related to the end-digit preference (EDP), selection bias, or both. Methods Analysis was carried out and comparison was made of 3629 BP measurements by 257 GPs. Statistical analysis used hierarchical mixed-effect models with random physician effect and fixed-effect covariables. Results The frequencies of 0 end digit were 16.7% in the IG, 32.4% in the OCG and 68.8% in the MCG for systolic BP (SBP; P <0.001 for all comparisons), and respectively 17.7, 38.1, and 74.1% for diastolic BP (DBP; P = 0.017 for all comparisons). SBP was higher in the IG than in OCG ( + 3.65 mmHg, P = 0.017). The same trend was observed for DBP, though not significant ( + 1.50 mmHg, P = 0.20). The EDP in the CG led to a mean underevaluation of 2.4 mmHg (P < 0.0005) of SBP and DBP. Conclusion The observed differences in BP between the groups are partly explained by the impact of EDP. Compared with the manual, oscillometric measurement may reduce EDP. Blood Press Monit 00:000–000  c 2011 Wolters Kluwer Health | Lippincott Williams & Wilkins. Blood Pressure Monitoring 2011, 00:000–000 Keywords: blood pressure measurement, end-digit preference, general practice, hypertension a Department of General Practice, Tours Medical University, b Department of General Practice, West Paris Medical University, c Department of General Practice, Paris 7 Medical University and d Department of Biostatistic, CNRS, UMR 558, Lyon Medical University, France Correspondence to Dr Jean-Pierre Lebeau, MD, Departement de Medecine Generale, Faculte de Medecine, 10, Bd Tonnelle ´, BP 3223, 37032 TOURS Cedex 1, France Tel: +33 60 7390999; fax: +33 25 4779787; e-mail: jean-pierre.lebeau@univ-tours.fr Received 25 August 2010 Revised 27 December 2010 Accepted 4 January 2011 Introduction End-digit preference End-digit preference (EDP) is defined as the tendency of physicians to round down the results of their blood pressure (BP) recordings to the lower multiple of 5 or, more often, of 10 [1]. Along with other observer-related biases during measurement, EDP contributes to the deviation between the patient’s real BP and the value on which physicians base their decision. EDP is frequently observed and well documented both in clinical practice [2–5] and during research studies [6], particularly, when physicians use conventional ausculta- tory methods such as mercury or aneroid sphygmoman- ometers [3]. Depending on the studies [2–6], the reported frequency of multiples of 10 in systolic BP (SBP) measurements ranges from 50 to 90% with a mean value of 65% [7]. The underestimation of BP values modifies the physi- cians’ behavior, who by rounding figures down to the multiple of 10, avoids initiating or reinforcing antihyper- tensive medication when this would be necessary [2]. Several studies have shown that EDP delays the insti- tution of a suitable treatment and substantially and artifi- cially raises the number of patients reaching the targets proposed in guidelines [2,8,9]. The difference between the real SBP and the one noted (and used) by the physician is also likely to have clinical consequences for patients. A meta-analysis [10] showed that a mean variation of 2 mmHg for SBP in a population aged from 40 to 89 years was associated with a 10% difference in mortality from stroke. In addition, a mean pharmacological reduction of 2.2 mmHg in SBP leads to a significant reduction of 19% in the relative risk (RR) of myocardial infarction and 15% in the RR of stroke in a treated hypertensive population [11,12]. An underevalua- tion of SBP may lead to the inappropriate transfer of a patient from a population that should receive a treatment (or its adjustment) to a population that should not [5–8]. Original article 1 1359-5237  c 2011 Wolters Kluwer Health | Lippincott Williams & Wilkins DOI: 10.1097/MBP.0b013e328344d067 Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.