@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix dc: . @prefix skos: . vivo:departmentOrSchool "Medicine, Faculty of"@en, "Anesthesiology, Pharmacology and Therapeutics, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Mattu, Gurdial Singh"@en ; dcterms:issued "2009-10-28T22:51:29Z"@en, "2003"@en ; vivo:relatedDegree "Master of Science - MSc"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """Hypertension is one of the most common reasons why North Americans visit a physician's office and its measurement technique has been reasonably standard since the earliest accepted description of an indirect method of measuring blood pressure by Riva Rocci in 1896. Korotkoff later modified the traditional sphygmomanometer in 1905. However, since then the mercury sphygmomanometer has become not only the gold standard, but also an essential diagnostic tool in everyday medical care. Recently however, there has been growing concern over the accuracy of the measurements obtained and the potential biohazard with standard mercury sphygmomanometers. Many cities and countries in Europe and North America, are in the process of phasing them out. Newer instruments such as the auscultatory aneroid, and the automated electronic devices of which the majority are of the oscillometric type, are beginning to appear. All new devices have the option of undergoing a validation process to assess the accuracy according to the relevant governing bodies. The BpTRU™ is an automated oscillometric electronic blood pressure device developed by VSM Medtech Ltd of Vancouver. Connected in parallel by means of a T-tube, to the current gold standard mercury sphygmomanometer, allowed blood pressures to be measured simultaneously. Two observers blinded from each other and the device, recorded blood pressures individually. The mean observer average was compared to the device mean blood pressures for agreement. This was done according to the American National Standard for Electronic or Automated Sphygmomanometers-1992 and the British Hypertension Society protocol-1993. The validation process was initially conducted in adults and agreement was within the above standards. However the device tended to underestimate at higher systolic blood pressures, so the algorithm was modified and the device re-validated using embedded raw data. After running the validation study in children, which again satisfied the above criteria, the raw data was combined in a final study. The BpTRU™ is the only automated blood pressure measuring device that has been independently validated and has attained a high level of accuracy over this broad age and blood pressure range."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/14252?expand=metadata"@en ; dcterms:extent "3465530 bytes"@en ; dc:format "application/pdf"@en ; skos:note "V A L I D A T I O N O F O S C I L L O M E T R Y B L O O D P R E S S U R E M E A S U R I N G D E V I C E S ; A C A S E S T U D Y O F T H E B p T R U ™ by D R . G U R D I A L S I N G H M A T T U M B . BCh. , T h e Universi ty of W a l e s Col lege of Medic ine, 1993 C.C.F.P., T h e Universi ty of Brit ish Co lumb ia , 1999 A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R S IN S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Depar tment of Pharmaco logy and Therapeut ics ; Facul ty of Medic ine) W e accept this thesis as con fo rming to the required s tandard T H E U N I V E R S I T Y O F BRIT ISH C O L U M B I A July 2003 © Gurdial Singh Mat tu , 2003 In present ing this thesis in partial fu l f i lment of the requ i rements for an advanced degree at the Universi ty of Brit ish Co lumbia , I agree that the Library shal l m a k e it f reely avai lable for re ference and study. I fur ther agree that permiss ion for ex tens ive copying of this thesis for scholar ly purposes m a y be granted by the head of m y depar tment or by his or her representat ives. It is unders tood that copying or publ icat ion of this thesis for f inancial gain shall not be a l lowed wi thout m y wri t ten permiss ion. Depar tment of Pharmaco logy and Therapeut ics Facul ty of Medic ine T h e Universi ty of Brit ish Co lumbia Vancouver , C a n a d a Date / t' -faHr £oo3 ABSTRACT Hypertension is one of the most common reasons why North Americans visit a physician's office and its measurement technique has been reasonably standard since the earliest accepted description of an indirect method of measuring blood pressure by Riva Rocci in 1896. Korotkoff later modified the traditional sphygmomanometer in 1905. However, since then the mercury sphygmomanometer has become not only the gold standard, but also an essential diagnostic tool in everyday medical care. Recently however, there has been growing concern over the accuracy of the measurements obtained and the potential biohazard with standard mercury sphygmomanometers. Many cities and countries in Europe and North America, are in the process of phasing them out. Newer instruments such as the auscultatory aneroid, and the automated electronic devices of which the majority are of the oscillometric type, are beginning to appear. All new devices have the option of undergoing a validation process to assess the accuracy according to the relevant governing bodies. The BpTRU™ is an automated oscillometric electronic blood pressure device developed by VSM Medtech Ltd of Vancouver. Connected in parallel by means of a T-tube, to the current gold standard mercury sphygmomanometer, allowed blood pressures to be measured simultaneously. Two observers blinded from each other and the device, recorded blood pressures individually. The mean observer average was compared to the device mean blood pressures for agreement. This was done according to the American National Standard for Electronic or Automated Sphygmomanometers-1992 and the British Hypertension Society protocol-1993. The validation process was initially conducted in adults and agreement was within the above standards. However the device tended to underestimate at higher systolic blood pressures, so the algorithm was modified and the device re-validated using embedded raw data. After running the validation study in children, which again satisfied the above criteria, the raw data was combined in a final study. The BpTRU™ is the only automated blood pressure measuring device that has been independently validated and has attained a high level of accuracy over this broad age and blood pressure range. ii TABLE OF CONTENTS Declared Author izat ion Abstract ii Tab le of Contents 111 A c k n o w l e d g e m e n t s v C H A P T E R I Overal l Introduct ion 1 Object ives 5 C H A P T E R II Compar i son of the osci l lometr ic blood pressure moni tor ( B P M - 1 0 0 B e t a ) wi th the auscul ta tory mercury s p h y g m o m a n o m e t e r 6 Abst ract 7 Introduct ion 8 Methods 10 Resul ts 14 Discuss ion 17 Tab les 20 Figures 22 References 24 C H A P T E R III Val idat ion of a new algor i thm for the BPM-100 electronic osci l lometr ic off ice blood pressure moni tor 26 Abst ract 27 Introduct ion 29 Methods 30 Resul ts 32 Discuss ion 34 Tab les 36 Figures 38 References 40 C H A P T E R IV Compar i son of the au tomated non- invasive osci l lometr ic blood pressure moni tor ( B p T R U ™ ) with the auscul tatory mercury s p h y g m o m a n o m e t e r in the paediatr ic populat ion 41 Abst ract 42 Introduct ion 43 Methods 45 Resul ts 51 Discuss ion 55 Tab les 57 Figures 59 References 61 i i i C H A P T E R V Accu racy of the B p T R U ™ , an au tomated electronic b lood pressure dev ice 62 Abst ract 63 Introduct ion 64 Methods 66 Resul ts 68 Discuss ion 71 Tab les 74 Figures 77 References 79 C H A P T E R VI Overal l D iscuss ion 81 C H A P T E R VII Conc lus ions 87 C H A P T E R VIII References 90 iv ACKNOWLEDGEMENTS Many thanks to Dr. J a m e s M. Wr igh t for providing inspirat ion and suppor t to enab le the complet ion of this Masters thesis. A special thank you to my suppor t ive and unders tand ing fami ly, especial ly Parveen and Charan Mat tu CHAPTER I OVERALL INTRODUCTION Measuremen t of blood pressure accurately is crucial ; as based on these measu remen ts , pat ients are labelled and non-drug and drug therapies r e c o m m e n d e d [1][2]. It is therefore essent ia l that the method of measur ing blood pressure is accurate and reproducible. T h e current s tandard for blood pressure measu remen t is the auscul tatory method using the mercury s p h y g m o m a n o m e t e r . However, there are pressures to replace that s tandard with other methods. One of the concerns is about mercury poisoning. As a result of this many States in the US, other cit ies in North A m e r i c a and s o m e European countr ies have banned or are prepar ing to ban mercury in both t he rmomete rs and blood pressure s p h y g m o m a n o m e t e r s [3][4]. It is predicted by s o m e that it is only a matter of t ime before al ternat ives to the 100-year-old mercury s p h y g m o m a n o m e t e r are required [5]. Ano ther pressure on the mercury s p h y g m o m a n o m e t e r is that it is subject to unacceptab le inter and int ra-observer variabil ity. This is a sett ing in wh ich a computer , wh ich uses a reproducible and repeti t ive method has obvious advantages. Another pressure on the mercury s p h y g m o m a n o m e t e r c o m e s f rom the requ i rements of clinical trials. Despi te these requi rements, bias can be demonst ra ted to c o m m o n l y occur dur ing the select ion procedure. In the clinical trial sett ing it is critical to min imize bias at all s tages of the trial. During a systemat ic rev iew of a large number of trials it w a s found that diastol ic blood pressure variabil ity at basel ine w a s substant ial ly less than systolic variabil ity at basel ine and substant ial ly less than diastolic variabil i ty at the end of the study [6]. The most likely explanat ion for this decreased variabil i ty at basel ine is that many subjects must have had a diastolic blood pressure equal to or just above the entry criteria. This ref lects observer bias in order to enhance recrui tment of pat ients into the study. T h e use of an object ive blood pressure measur ing device to identify pat ients at basel ine wou ld prevent this bias. Another reason for developing new blood pressure measur ing dev ices are the practicali t ies of most physician's off ices or clinics, the place where most of our blood pressure m e a s u r e m e n t s take place. Most physic ians do not have the t ime, knowledge or skill to careful ly measu re blood pressure in the s a m e manner that they were measured in the clinical trials [7]. Thus if they cannot reproduce the sett ing that the ev idence is based upon they cannot m a k e decis ions for their pat ients based on the best avai lable ev idence. A method that would replace the mercury s p h y g m o m a n o m e t e r must deal with the practical t ime constra ints in a physician's off ice and mimic the blood pressure m e a s u r e m e n t s ach ieved in a clinical trial sett ing as m u c h as possible. 2 Most of the new methods for measur ing blood pressure at the present t ime use the osci l lometr ic technique. Th is method refers to the measu remen t of the osci l lat ions, caused by the arterial pulse pressure. T h e s e osci l lat ions are the result of the cuff caus ing the occ lus ion of the artery (usual ly the brachial) , that it over l ies. T h e method does not use sound so m ic rophones are not needed and external noise is not a p rob lem. However , the method is sensi t ive to pat ient m o v e m e n t . T h e cuff, w rapped around the patient 's upper arm is automat ical ly inflated with air to occ lude the brachial pulse. On def lat ion, pressure data is recorded by the device in wave fo rm (as shown in f igure 1). Fig 1: pulse wave fo rm against t ime T h e m a x i m u m ampl i tude of the pulse w a v e is taken as the m e a n arterial p ressure ( M A P ) and the systolic and diastol ic blood pressures are calculated f rom this plus the pattern of the w a v e f o r m s . T h e sof tware a lgor i thms used by the many manufac turers of blood pressure measur ing dev ices are proprietary and therefore not publ ished in format ion. W h e n a local company , V S M M e d T e c h Ltd. of Vancouver , C a n a d a approached T h e U B C High Blood Pressure Clinic wi th a new and novel approach to measur ing and record ing blood pressure in the off ice, it was o f cons iderab le interest to m e . I w a s interested in learning h o w s u c h n e w dev ices w e r e val idated and w a s therefore enthusiast ic to be involved in the des ign, modi f icat ion, execut ion and analysis of clinical trials for that purpose. This involved learning the r igorous uni form cri teria as set by the Amer ican National S tandard for Electronic or Au tomated S p h y g m o m a n o m e t e r s , A N S I / A A M I SP10-1992 and the Brit ish Hyper tens ion Society protocol , the s tandard sett ing bodies for these types of dev ices [8,9]. T h e process of part ic ipat ion in 2 clinical val idat ion trials of the B p T R U ™ , manu fac tu red by V S M Medtech Limited plus re-val idat ion using stored electronic data permit ted m e to learn a lot about the val idat ion process and s tandards . A s a result of this learning process I have been able to m a k e recommenda t ions 3 as to how the val idat ion process and s tandards can be improved and how the B p T R U ™ should be further tested in the future. The results of this thesis are descr ibed in four chapters . Chapter two represents a publ ished paper that summar izes the val idat ion trial in adults [10]. Chapter three descr ibes how the compute r algor i thm was modi f ied using stored electronic data f rom the trial outl ined in Chapter two [11]. Chapter four descr ibes a second trial to de termine the accuracy of the dev ice in chi ldren (age range 3 to 18 years) . T h e Chapter is ready to be submit ted for publ icat ion. Finally, to summar i ze the overall accuracy of the B p T R U ™ the raw data for each subject in the adult and paediatr ic trials was combined and presented in Chapter f ive. This study is also now ready to be submi t ted for publ icat ion. Chapter six represents the overal l d iscuss ion. It includes recommenda t ions to the regulatory bodies as to how to improve the val idat ion process and recommendat ions to the c o m p a n y that m a k e s the B p T R U ™ as to future trials to fur ther demonst ra te the usefu lness of the device. 4 Objectives: • T o c o m p a r e the B p T R U ™ osci l lometr ic blood p ressure-measur ing dev ice wi th the auscul tatory mercury s p h y g m o m a n o m e t e r in adul ts. • T o c o m p a r e the blood pressure device against the s tandard auscul ta tory mercury s p h y g m o m a n o m e t e r in the paediatr ic populat ion. • T o comb ine the adult and paediatr ic data in the fo rm of an individual data meta-analys is , to assess the overal l accuracy of the device. • T o cri t ique the regulatory requ i rements for blood pressure-measur ing dev ices and m a k e suggest ions for improvement . • T o r e c o m m e n d and develop new clinical trials to demons t ra te the usefu lness in the pr imary care sett ing. 5 CHAPTER II COMPARISION of the OSCILLOMETRIC BLOOD PRESSURE MONITOR (BPM-100 B e t a) WITH THE AUSCULATORY MERCURY SPHYGMOMANOMETER Gurdial S. Mat tu , MB BCh, C C F P a , T h o m a s L. Perry, Jr. M D , F R C P ( C ) 3 0 J a m e s M Wr igh t , M D , PhD, F R C P ( C ) a b , Depar tment of Pharmaco logy & Therapeu t i cs 3 and Depar tment of Medic ine , Universi ty of Brit ish Co lumb ia , Vancouver , BC, C a n a d a . \"Cor respond ing Author : J a m e s M. Wr ight , M D , PhD, F R C P ( C ) , Depar tmen t of Pharmaco logy & Therapeu t i cs 3 and Depar tment of Med ic ine b , T h e Universi ty of Brit ish Co lumb ia , 2176 Health Sc iences Mall , Vancouver , BC, Canada , V 6 T 1Z3, te lephone: (604) 822-4270 , fax: (604) 8 2 2 - 0 7 0 1 , emai l : jmwr igh t@in te rchange.ubc .ca Source of Fund ing : V S M - M e d T e c h Short Tit le: Val idat ion of the B P M - 1 0 0 B e t a monitor. 6 ABSTRACT Background: T o c o m p a r e directly the accuracy of the B P M - 1 0 0 B e t a moni tor (an au tomated osci l lometr ic blood pressure device) wi th s tandard auscul tatory mercury sphygmomanomet ry . Design: T h e B P M - 1 0 0 B e t a w a s connected in parallel v ia a T- tube to a mercury s p h y g m o m a n o m e t e r . T h e B P M -1 0 0 B e t a and two trained observers (bl inded f rom each other, and the B P M - 1 0 0 B e t a ) measu red the sitt ing blood pressures (BP) s imul taneously . Methods: Means, s tandard deviat ion and range were calculated for all demograph ic da ta : age, a rm size, heart rate and BP. A g r e e m e n t between the B P M - 1 0 0 B e t a and the m e a n of 2 observers ( re ference) w a s determined and expressed as the m e a n ± SD, plus the % of d i f ferences within 5, 10 and 15 m m H g . Results: Of 92 recrui ted subjects , 85 (92.4%) met the inclusion cri ter ia, and 391 sets of sitt ing BP and heart rate m e a s u r e m e n t s w e r e avai lable for analysis. T h e m e a n d i f ference be tween the B P M - 1 0 0 B e t a moni tor and the re ference w a s - 0 . 6 2 ± 6.96 m m H g for systolic BP, - 1 . 4 8 ± 4 .80 m m H g for diastol ic BP and 0.14 ± 1.86 bpm for heart rate. T h e only l imitation of the device w a s a tendency for the dev ice to underest imate higher systol ic blood pressures. This l imitation has been addressed by a minor c h a n g e in the algor i thm (see fol lowing c o m p a n i o n publ icat ion). [8] Conclusion: T h e B P M - 1 0 0 B e t a is an accurate BP moni tor for the off ice sett ing, wh ich mee ts all requ i rements of the Assoc ia t ion for the A d v a n c e m e n t of Medical Inst rumentat ion and ach ieves an \"A\" g rade accord ing to the Brit ish Hyper tens ion Society s tandard . Key Words: b lood pressure, measurement , monitor, osci l lometr ic, val idat ion, automat ic . 7 INTRODUCTION Sir George Picker ing descr ibed blood pressure as a dynamic process, such that a single blood pressure reading in the physician's off ice is like watch ing a single f r a m e of a mov ie [1]. T o pract ice rational ev idence-based medic ine w h e n manag ing elevated blood pressure, one should documen t blood pressures in a manner at least as good as that employed in the large randomized control led trials (RCTs) . T h e latter general ly require that the patient rest comfor tab ly for at least 5 minutes before a technic ian or nurse measures the blood pressure repeatedly (at least 3 t imes) fo l lowing approved standards. [2] This m a y be ach ievab le in s o m e special ty cl inics, but is se ldom fo l lowed in the fami ly physic ian sett ing, where most pat ients wi th high blood pressure are m a n a g e d . It is a pract ical reality that accurate reproducib le documenta t ion of rest ing b lood pressure is diff icult for the average outpat ient physic ian or nurse to achieve. C o m m o n errors in blood pressure readings include: insuff icient or no t ime for the patient to rest, unsuppor ted a r m , unsuppor ted back, a rm not at heart level, talk ing at the t ime of the measurement , distract ing background noise, inappropr iate cuff and bladder size, fai lure to record m e a s u r e m e n t immediate ly, rounding of numbers to nearest 5 or 10 instead of to 2 m m H g , too rapid def lat ion of cuf f pressure, and using a non-cal ibrated ins t rument [3]. Any method that might improve blood pressure m e a s u r e m e n t must sat isfy e c o n o m i c and practical realit ies of of f ice pract ice. It must generate measu remen ts that are an accura te ref lect ion of the present s tandard (auscul tatory measu remen ts with a mercury s p h y g m o m a n o m e t e r ) f r om a techn ique that is equal to or better than usual pract ice. In addit ion it should reduce or not add to phys ic ian/nurse effort and t ime. T h e B P M - 1 0 0 B e t a . p roduced by V S M M e d T e c h Ltd. of Vancouver , C a n a d a , is an au tomated , non- invasive blood pressure moni tor des igned to generate repeated measures of the blood pressure and pulse rate of pat ients wi thout requir ing the presence of a physician or nurse. T h e dev ice uses s tandard blood pressure cuf fs to measure the blood pressure in the upper arm using the osci l lometr ic techn ique. T h e purpose of this s tudy w a s to direct ly c o m p a r e the accuracy of the B P M - 1 0 0 B e t a moni tor wi th s tandard auscul tatory m e a s u r e m e n t s using a mercury s p h y g m o m a n o m e t e r in accordance wi th guidel ines provided by the 8 Assoc ia t ion for the A d v a n c e m e n t of Medical Inst rumentat ion (AAMI) SP10 :1992 , the s tandard sett ing body for both these devices. [4] 9 METHODS Subject Enrolment Subjects w e r e recrui ted using public not ices in a Universi ty set t ing, th rough the Blood Pressure Clinic at the Universi ty of Brit ish Co lumbia , and through fami ly physic ian pract ices assoc ia ted with the University. Screening w a s des igned to enrol at least 85 subjects with a m i n i m u m of 3 acceptab le pairs of blood pressure m e a s u r e m e n t s and meet ing the required target populat ion object ives def ined below. Ethical Approval Ethical approval of the s tudy and consent fo rm w a s obta ined f rom the Clinical Screen ing Commi t tee for H u m a n Exper imenta t ion of the Universi ty of Brit ish Co lumbia prior to init iation of any of the study procedures. Inclusion criteria All subjects had to be hemodynamica l l y stable and at least 18 years o ld. Subjects wi th rhythm irregularit ies such as atrial f ibri l lation or with an auscul tatory gap were el igible. All sub jects s igned the consent fo rm prior to enro lment . T o be included screening m e a s u r e m e n t cri teria had to be met, and a m i n i m u m of 3 valid pairs of m e a s u r e m e n t s obta ined. Exclusion criteria Subjects wi th unstable, accelerated or mal ignant hypertension w e r e exc luded , as w e r e patients with severe per ipheral vascular d isease or other exist ing condi t ion that the invest igators felt wou ld not al low for safe or accura te non- invasive blood pressure m e a s u r e m e n t s . The re w e r e also pre-speci f ied m e a s u r e m e n t exc lus ion cri teria for both subjects and individual m e a s u r e m e n t s : 1) Subjects in w h o m the inter-observer ag reemen t w a s greater than 10 m m H g in either the systol ic or diastol ic blood pressure m e a s u r e m e n t s at screen ing. 2) Subjects w h o had less than 3 valid pairs of read ings. 3) Subjects w h o had such w e a k Korotkof f sounds , that either of the observers d e e m e d acceptab le auscul tat ion impossib le. 4) Any m e a s u r e m e n t in wh ich the inter-observer ag reement w a s greater than 10 m m H g for either the systolic or diastol ic reading. 5) Any measu remen t in wh ich the B P M - 1 0 0 B e t a did not record or recorded an error. 10 Target population objectives: 1) A n equa l n u m b e r of ma les and f e m a l e sub jects . 2) Range of a rm sizes (measured at mid-b iceps) . • At least 1 0 % greater than 35 cm c i rcumference ( large arm cuf f ) . • At least 1 0 % less than 25 cm c i rcumference (smal l a rm cuf f ) . 3) Range of systol ic blood pressures ( m e a n of 2 observers) . • At least 1 0 % greater 180 m m H g systol ic. • At least 1 0 % less than 100 m m H g systolic. 4) Range of diastol ic blood pressures (mean of 2 observers) . • At least 1 0 % greater 100 m m H g diastol ic. • At least 1 0 % less than 6 0 m m H g diastol ic. Study design Subjects w e r e recrui ted dur ing Sep tember and October 1999. All sub jec ts provided s igned in formed consent prior to enro lment . A s igned copy of the consent fo rm w a s provided to each subject . All subjects were entered into the subject log and ass igned a subject code. Basic demograph ic data w e r e col lected on each subject enrol led into the trial including age, sex, pre-exist ing heal th condi t ions and medicat ions, and a rm c i rcumferences (measured at mid-b iceps on the arm of blood pressure record ings) . Subjects w e r e then seated in a quiet but well i l luminated room on a chair wi th comfor tab le back support . T h e arm f rom wh ich blood pressure m e a s u r e m e n t s were recorded w a s p laced in a comfor tab le posit ion on an adjustable tab le at heart level. T h e appropr ia te-s ized cuff (based o n cuf f mark ings) w a s selected f rom four sizes provided with the B P M - 1 0 0 B e t a and appl ied to the subject 's bare upper arm with the indicator over the brachial artery. T h e B P M - 1 0 0 B e t a w a s connec ted in paral lel wi th a mercury s p h y g m o m a n o m e t e r by m e a n s of a T- tube wi th a rms of equal length. A Tr iml ine precision mercury s p h y g m o m a n o m e t e r ( range 0 to 300 m m H g , 11 gradat ions of 1 m m H g , accuracy ± 0.5 m m H g ) w a s used as the s tandard s p h y g m o m a n o m e t e r to provide the auscul tatory re ference readings. A heart rate moni tor , the Nonin f inger pulse ox imeter , w a s connec ted to the subject 's other a rm or al ternate body locat ion as appropr iate; this device has a range of 18 to 300 beats per minute and accuracy of ± 3%. One of the two observers located the d iaphragm of the s te thoscope over the brachial artery and determined opt imal posit ion for adequate auscul tat ion. Auscul tat ion for b lood pressures w a s per formed wi th a dua l -headed teaching s te thoscope. T h e observers w e r e seated on ad justab le seats opposi te the subject so that they could read the s tandard mercury s p h y g m o m a n o m e t e r wi thout introducing parallax errors. A n interposed curtain bl inded observers f rom each other. T h e two observers w e r e exper ienced registered nurses w h o s e blood pressure recording skil ls were con f i rmed by the invest igators prior to any trial subject enro lment . T h e B P M - 1 0 0 B e t a dev ice automat ical ly inf lates and def lates the cuff, and then uses the osci l lometr ic techn ique to calculate systol ic and diastol ic blood pressure. In this techn ique, the m e a n arterial pressure (MAP) is measu red directly f rom the cuff pressure dur ing def lat ion, and the systol ic and diastol ic BP points are calculated as ratios of the M A P pulse ampl i tude. T h e B P M - 1 0 0 B e t a automat ical ly inflates the cuff after initiation to preset pressures, and automat ical ly ad justs to h igher pressures if necessary to ensure comp le te capture of the pulse waves . T h e cuff s lowly def lates automat ica l ly at a rate of 4 m m H g / s e c o n d in a t rue l inear fash ion. T h e B P M - 1 0 0 B e t a w a s set to cycle every 2 minutes and the observers (b l inded f rom each other, and the B P M - 1 0 0 B e t a ) s imul taneous ly visual ly recorded the systolic and diastol ic b lood pressures by l istening to the Korotkof f sounds and watch ing the s tandard mercury s p h y g m o m a n o m e t e r . Korotkof f sound phase I (the beginning of the compress ion sound) w a s taken as the systol ic blood pressure; and phase V (the point of comple te d isappearance of the sound) w a s taken as the diastol ic blood pressure. T h e s e m e a s u r e m e n t s were independent ly recorded by each of the observers . T h e B P M - 1 0 0 B e t a readings and 12 pulse rate were recorded by one of us ( G S M ) independent ly and bl inded f rom the two observers . A total of six m e a s u r e m e n t s w e r e recorded by this me thod . T h e first m e a s u r e m e n t w a s cons idered the screening m e a s u r e m e n t and was not used fur ther in the val idat ion s tudy ( the f irst reading w a s used to establ ish observer ag reemen t and sat isfactory auscul tat ion techn ique) . O n e of us ( G S M ) also recorded the heart rate re ference va lue via the s tandard Nonin f inger pulse ox imeter at approx imate ly the mid-point of automat ic def lat ion, wi thout knowledge of the B P M - 1 0 0 B e t a est imate. After six m e a s u r e m e n t s w h e n all readings and record ings w e r e proper ly documen ted the cuff was removed f rom the subjects ' a r m , and the subject was a l lowed to leave. Data Analysis M e a n , s tandard deviat ion and range were calculated for all the demograph ic da ta : age, a rm size, heart rate and blood pressures . T h e first blood pressure m e a s u r e m e n t w a s a screen ing m e a s u r e m e n t only and w a s not used in the analysis. Each subsequent m e a s u r e m e n t w a s rev iewed for the exclus ion criteria out l ined above for both the B P M - 1 0 0 B e t a and the observers . For each included systol ic and diastol ic blood pressure m e a s u r e m e n t the m e a n of the 2 observer values w a s calculated and def ined as the re ference s tandard . Di f ferences be tween the 2 observers were expressed as the m e a n d i f ference ± s tandard deviat ion, and as the % < 5 m m H g . Di f ferences between the B P M - 1 0 0 B e t a and the re ference s tandard were also determined for e a c h systol ic and diastol ic blood pressure m e a s u r e m e n t and expressed as the m e a n di f ference ± s tandard deviat ion accord ing to the A A M I s tandard [4], as wel l as the % of inter- technique d i f ferences wi th in 5 m m H g , 10 m m H g and 15 m m H g accord ing to the B H S standard [5]. Inter- technique d i f ferences be tween the B P M - 1 0 0 B e t a heart rates and the re ference heart rates obta ined f rom the Nonin pulse ox imeter w e r e also calculated similarly. T o assess how the d i f ferences related to the absolute blood pressure va lue the B P M - 1 0 0 minus reference s tandard d i f ference w a s plotted against the average of the 2 BP va lues us ing a Bland A l tman display for both systol ic and diastol ic BP [6]. 1 3 RESULTS Enrolled subjects Of the 92 subjects enrol led, 85 (92 .4%) w e r e included and 7 (7 .6%) w e r e exc luded . Reasons for exc lus ion are detai led below. Pre-exist ing medical condi t ions of enrol led subjects included s o m e with hyper tension, renal d isease ( including subjects on dialysis), obesity, s t roke, ang ina and Crohn 's d isease. Subjects w e r e also tak ing var ious ant i -hypertensive and other medicat ions. No adverse side effects were reported dur ing or for up to 30 days after the trial. Th ree individuals w h o had e levated blood pressures and possibly unrecogn ized hypertension w e r e advised to fo l low up wi th their regular physic ian. Excluded Subjects T h e f irst 3 of the 92 volunteers were exc luded because of the inability to obta in at least 3 sat isfactory sets of BP readings. Th is w a s found to be due to a technical prob lem with the B P M - 1 0 0 B e t a moni tor def lat ion m e c h a n i s m , wh ich w a s rectif ied prior to any fur ther subjects being s tud ied. T w o subjects were exc luded because of a screen ing BP inter-observer d i f ference of greater than 10 m m H g , and 2 because of inability to record accurate ly the diastol ic blood pressures. Target population and demographic data All target populat ion object ives were met . Of the 85 included subjects 44 (51 .8%) were male. T h e m e a n age w a s 43.1 ± 15.6 years ( range of 18 - 83 years) . The re w e r e 10 subjects (11 .8%) with arm c i rcumference exceed ing 35 c m , and 9 subjects (10.6%) subjects wi th a rm c i rcumference less than 25 c m . Included blood pressure measurements Included data total led 391 sets of s imul taneously recorded m e a s u r e m e n t s . In 41 (10 .5%) systolic BP exceeded 180 m m H g , and in 53 (13 .6%) systol ic w a s less than 100 m m H g . In 43 (11 .0%) diastol ic BP exceeded 100 m m H g , and in 54 (13.8%) diastol ic was less than 60 m m H g . 14 Excluded blood pressure measurements Thir ty- four (8 .0%) systol ic and diastol ic BP measuremen ts w e r e exc luded . Four (0 .9%) were due to a B P M - 1 0 0 B e t a technical error; the B P M - 1 0 0 B e t a either recorded an error or did not d isplay the reading. Four were exc luded because one observer forgot the n u m b e r before record ing the mercury s p h y g m o m a n o m e t e r read ing. Twenty -one w e r e exc luded because one of the observers did not feel that a sat isfactory auscul tat ion could be obtained (15 systolic, 6 diastol ic) . Five (1 .3%) were exc luded because the observers did not agree within 10 m m H g for the systol ic or diastol ic BP. Sphygmomanometer readings Mean d i f ference between the observers (observer 1 - observer 2) for the 391 systol ic m e a s u r e m e n t s w a s - 0 . 6 4 ± 1.94 m m H g . T h e major i ty of the di f ferences, 97 .7%, were < 5 m m H g . All were < 10 m m H g as the protocol prescr ibed exclus ion of d i f ferences exceed ing 10 m m H g . M e a n d i f ference between the 2 observers for diastol ic BP w a s - 1 . 0 8 + 2.46 m m H g . Most (95 .7%) w e r e < 5 m m H g and all were < 10 m m H g . Range and distribution of reference standard measurements T h e m e a n systol ic BP for the 391 systol ic re ference s tandard m e a s u r e m e n t s (average of the two observers) w a s 128.7 ± 30.7 m m H g ( range 81.5 to 223.5 m m H g ) . T h e m e a n re ference standard diastol ic BP w a s 77.4 ± 16.7 m m H g ( range 45.5 to 120.5 m m H g ) . M e a n heart rate using the Nonin f inger pulse ox imeter w a s 70.2 + 12.3 beats per minute ( range 42 to 104 beats per minute) . Accuracy of BPM-100Beta as compared with the reference standard measurements T h e m e a n d i f ference be tween the B P M - 1 0 0 B e t a and the reference s tandard systol ic and diastol ic BP ( B P M - 1 0 0 B e t a - re ference) is wel l within the A A M I s tandard [4] and is s h o w n in Tab le 1 . Tab le 2 shows the proport ion of systol ic and diastol ic d i f ferences within 5, 10, and 15 m m H g and how this con fo rms to the BHS s tandard [5]. T h e BPM-100 dev ice is des igned to de termine the average of up to f ive b lood pressure readings in individual sub jects . W e therefore calculated the subject m e a n B P M - 1 0 0 B e t a m inus the subject m e a n reference s tandard systol ic and diastol ic BP m e a s u r e m e n t s as a m e a s u r e of accuracy in the practical 15 clinical set t ing. Th is gave a slightly di f ferent m e a n di f ference and a lower s tandard deviat ion for the 85 subjects as s h o w n in Tab le 1. It w a s also assoc ia ted wi th a sl ight improvemen t in the proport ion of systolic and diastol ic d i f ferences within 5, 10 and 15 m m H g (Table 2) . In Fig 1 the Bland A l tman display of m e a s u r e m e n t s for systol ic B P shows the d i f ference of each B P M -100 B e ta and re ference s tandard systolic BP plotted against the average of the B P M - 1 0 0 B e t a and the re ference s tandard systolic BP. Th is f igure demons t ra tes that the B P M - 1 0 0 B e t a tends to underest imate the systol ic BP (negat ive va lues) for systolic measu remen ts >150 m m H g . In Fig 2 the Bland A l tman display of m e a s u r e m e n t s for diastol ic blood pressure shows the d i f ference of each re ference and B P M diastol ic blood pressure plotted against the average of the B P M - 1 0 0 B e t a and re ference s tandard diastolic blood pressure. In this case the d i f ferences are c lustered around a d i f ference of 0 over the who le range of pressures. T h e m e a n d i f ference of the B P M - 1 0 0 B e t a and the re ference heart rate w a s 0.14 ± 1.86 beats per minute. 16 DISCUSSION T h e most c o m m o n reason for patient off ice visits to a physic ian in C a n a d a and the United States is hypertension. [6] For d iagnosis and t reatment to be consistent , it is impor tant that blood pressure m e a s u r e m e n t s in the physic ian's off ice be as s tandard ized and accura te as possib le. A n accurate, convenient and af fordable automat ic blood pressure measur ing dev ice wou ld improve cons is tency and accuracy of m e a s u r e m e n t s in dif ferent pract ice set t ings. T h e Assoc ia t ion for the A d v a n c e m e n t of Medical Inst rumentat ion (AAMI) , is the s tandard sett ing body for blood pressure m e a s u r e m e n t dev ices. Its guidel ines require that the m e a n d i f ference of blood pressure m e a s u r e m e n t s be tween a new device and the mercury s tandard must be wi th in ± 5 m m H g with a standard deviat ion of <8 m m H g . It can be seen that in this s tudy the B P M - 1 0 0 B e t a moni tor easi ly met this s tandard . It can also be seen that the m e a n d i f ference be tween the BPM-100 B e t a and the reference m e a s u r e m e n t s w a s similar to that for the inter-observer m e a s u r e m e n t s . T h e s tandard deviat ion of the di f ference be tween the B P M - 1 0 0 B e t a measu remen ts and the re ference s tandard exceeded that of the inter-observer m e a s u r e m e n t s , but was well wi thin the A A M I requ i rement of a s tandard deviat ion of less than 8 m m H g . T h e low variabil i ty of the inter-observer m e a s u r e m e n t s in this s tudy is not an accurate ref lect ion of this measure , as the protocol dictated that inter-observer d i f ferences exceed ing 10 m m H g be exc luded (N = 5 in this s tudy) . T h e major potential l imitation of the B P M - 1 0 0 B e t a detected in this s tudy is the underest imat ion of systolic m e a s u r e m e n t s > 150 m m H g as demonst ra ted by the Bland A l tman display (Fig 1). It is known that the osci l lometr ic technique, w h e n compared with s tandard auscul tatory me thods tends to both underest imate systolic BP and to yield high standard deviat ions for higher systol ic pressures. [7 ] A s a result of this potential l imitat ion the manufac turer undertook a fur ther deve lopmenta l s tudy using the stored raw electronic data and the auscul tated blood pressure data obta ined dur ing this study. T h e methods and result of this subsequen t study are presented in the fol lowing c o m p a n i o n paper [8]. 17 Another potential l imitat ion is the relatively high variabil i ty of the d i f ferences be tween the B P M device and the re ference s tandard individual measuremen ts . However , the re ference s tandard is an indirect est imate of the true intra-arterial p ressure and studies compar ing auscul tat ion and intra-arterial pressures are assoc ia ted with a higher degree of variabil i ty than that seen here. [9-12] In other words w h e n a large d i f ference be tween 2 m e a s u r e m e n t s is seen one cannot be certa in wh ich of the m e a s u r e m e n t s is the best es t imate of intra-arterial b lood pressure. T h e B P M - 1 0 0 B e t a uses the osci l lometr ic me thod , wh ich is the one used by mos t ambu la to ry and h o m e blood pressure measur ing devices. T h e B P M - 1 0 0 B e t a is des igned to m e a s u r e 6 blood pressures and discard the first. T h e first is d iscarded, as it is normal ly done wi th the physic ian or nurse present in the r o o m wi th the pat ient. T h e dev ice is des igned to m e a s u r e a n d s a v e the last 5 m e a s u r e m e n t s plus the average of these 5 m e a s u r e m e n t s automat ical ly whi le the phys ic ian/nurse is involved in other activit ies. T h e m e a n BP as wel l as the individual m e a s u r e m e n t s can be recorded in the pat ient 's chart . T h e m e a n B P is the best es t imate of the rest ing pressure of the pat ient for that visit. The re is therefore s o m e value in calculat ing the d i f ferences between the m e a n va lues for all individual subjects as measured by the B P M - 1 0 0 B e t a and by auscul tat ion. This data probably better ref lects the clinical reliability of this moni tor in a pract ice sett ing. As can be seen in Tab le 1 and 2 this analysis is assoc ia ted with ei ther no c h a n g e or a n improvement in the pe r fo rmance of the B P M - 1 0 0 B e t a moni tor . T h e impor tance of measur ing blood pressure proper ly cannot be underest imated. [11-13] Diagnosing hyper tension in pat ients w h o are truly normotens ive and v ice ve rsa has potential far reaching c o n s e q u e n c e s fo r the pat ient involved. B lood pressure m e a s u r e m e n t s there fore shou ld not only be accurate and consis tent but reproducib le [13,14] . Our s tudy sugges ts this is possib le using the B P M -1 0 0 B e t a moni tor . Langlois [3], descr ibes c o m m o n errors in measur ing and recording b lood pressure m e a s u r e m e n t s . Using a well val idated automat ic blood pressure moni tor could rectify m a n y of these errors. 'No wait ing t imes ' can be prevented by sett ing the t ime between m e a s u r e m e n t s f rom 1 to 5 minutes (cycle t ime) , depend ing 18 on how long the patient has been seated at rest. S ince the first m e a s u r e m e n t is not used in calculat ing the m e a n blood pressure, it g ives t ime for pat ients to relax and get a c c u s t o m e d to the method. 'Distract ing noise or ta lk ing' is not an issue because pat ients can be left a lone in a quiet r o o m . 'Rounding off numbers ' is avo ided because the B P M - 1 0 0 B e t a moni tor records m e a s u r e m e n t s to the nearest 1 m m H g wi thout any bias. 'Fast def lat ion ' is not a factor because def lat ion is uni form and preset at 4 m m H g / s e c o n d . A s m a r and Zanchet t i state \"as w e approach the end of the 2 0 t h century, w e are assist ing the birth of a new era in blood pressure measurement . \" [15 ] By util izing a good au tomat ic BP moni tor , one can el iminate the \" h u m a n error\" associated with both mercury and anero id dev ices, whi le per forming mult iple, accura te BP readings wh ich are in accordance wi th current clinical guidel ines. In conc lus ion, this s tudy demons t ra tes that the B P M - 1 0 0 B e t a osci l lometr ic dev ice provides an accurate measure of blood pressure that c losely approx imates that ach ieved by the indirect auscul tatory method using a mercury s p h y g m o m a n o m e t e r . Th is val idat ion study sat isf ies the A A M I and B H S standards for new dev ices. W e bel ieve that the B P M - 1 0 0 B e t a of fers pract ical potential advan tages to improve the accuracy, reproducibi l i ty and eff ic iency of blood pressure m e a s u r e m e n t by physic ians or nurses. 19 21 xi co •o c CO 0 0 co o o o o Q_ CO 0 0 o c co 0 Q . 0 CO o _ l O I-< Q x> c \"i-S E JS 0 E o oo | V o oo p +1 co o cd | v CD CD cd p I i n +1 CM CD CO CN CL CO 0 o c 0 0 I K CO o o CD CM CN CO CN CO CO CD oo IK z: o o I CD c CO 0 E o 0 ' ¥ l w c o c 0 E \"55 _c 15 o X I 0 0 E 0 o c co > X I < 0 c g 'o o W < o CN V I CD co co CN co Fig 1 Bland-Al tman Display of Systo l ic Measurements , All Cuf fs , 391 measurements 50 75 100 125 150 175 2 0 0 2 2 5 250 Mean of BPM and Obs Avg (mmHg) Bland-Al tman display of all systol ic measu remen ts for all cuf fs (n = 391) . B P M , B P M - 1 0 0 B e t a monitor; O b s Avg , observer average. 22 Fig 2 T3 C ^ i i m E O o) 0) > o < (/) Bland-Altman Display of Diastolic Measurements, All Cuffs, 391 measurements 2 0 15 10 5 0 -5 -I - 1 0 - 1 5 - 2 0 -I -25 2 0 OOo O C D 4 0 6 0 8 0 1 0 0 1 2 0 Mean of BPM and Obs Avg (mmHg) mean +2SD mean -2SD 140 B land-A l tman display of all diastol ic m e a s u r e m e n t s for all cuf fs (n = 391) . B P M , B P M - 1 0 0 B e t a moni tor ; Obs Avg , observer average. 23 References 1. Picker ing G. Hyper tens ion: causes , consequences and managemen t . 2 n d Edi t ion. Churchi l l L iv ingstone. Edinburgh and London, 1974. 2. Systol ic Hyper tens ion in the Elderly P rog ram. Prevent ion of s t roke by ant i -hypertensive drug t rea tment in older persons wi th isolated systolic hyper tens ion. Final resul ts of the Systol ic Hyper tension in the Elderly P rog ram. T h e Journal of the Amer i can Medical Associat ion 1 9 9 1 ; 265: 3255-3264 . 3. Langlois S. Measur ing b lood pressure: H o w competen t a re w e ? Perspect ives in Card io logy 1999; 29-39. 4 . Amer i can Nat ional Standard for Electronic or Au tomated S p h y g m o m a n o m e t e r s : ANSI /AAMI SP10 . Ar l ington, VA : Assoc ia t ion for the A d v a n c e m e n t of Medical Ins t rumentat ion. 1992. 5. O'Br ien E, Petr ie J , Littler W , et a l . T h e Brit ish Hyper tens ion Society protocol for the evaluat ion of blood pressure measur ing devices. Journal of Hyper tens ion 1993; 11 (suppl 2) : S43-S62 . 6. Bland J M , A l tman D G . Statistical me thods for assess ing agreement between two methods of cl inical measuremen t . Lancet 1 9 8 6 ; 1 : 3 0 7 - 3 1 0 . 7. Woodwe l l DA. Nat ional ambu la to ry medical care survey: 1995 summary . A d v a n c e Data 1997; 295: 1-25. 8. O'Br ien E, Atk ins N, M e e F, O'Mal ley K: Evaluat ion of blood pressure measur ing devices. Clinical & Exper imenta l Hyper tension 1993; 15: 1087-1097. 9. Wr igh t J M , Mat tu GS, Perry T L Jr, Gelfer M, et a l . Val idat ion of a new algor i thm for the BPM-100 electronic osci l lometr ic off ice blood pressure monitor. Blood Pressure Moni tor ing 2 0 0 1 ; 6: 161-165. 10. Berl iner K, Fujiy H, Lee D H , Yildiz M, G a m i e r B. Blood pressure m e a s u r e m e n t s in obese persons: Compar i son of intra-arterial and auscul tatory m e a s u r e m e n t s . Amer i can Journal of Cardio logy 1 9 6 1 ; July: 10-17. 1 1 . S p e n c e J D , Sibbald W J , Cape RD. Direct, indirect and m e a n blood pressures in hypertensive pat ients: T h e prob lem of cuf f artefact due to arterial wal l st i f fness and a partial solut ion. Clinical & Invest igat ive Medic ine 1979; 2 : 165-173. 12. Campbe l l N R . Hyper tension m a n a g e m e n t in clinical pract ice. Canad ian Journal of Card io logy 2000 ; 16: 574-576. 13. Ca rew M et al. Hyper tens ion in C a n a d a . Canad ian Fami ly Physic ian 1999; 45 : 1756-1758. 14. Handa SP. Measur ing blood pressure properly. Perspect ives in Card io logy 1999; 1-2. 24 15. Campbe l l NR, McKay D W . Accura te blood pressure measuremen t : W h y does it mat ter? Canad ian Medical Assoc ia t ion Journal 1999; 1 6 1 : 277-278. 16. A s m a r R, Zanchet t i A. Guidel ines for the use of sel f -blood pressure moni tor ing: A s u m m a r y report of the first international consensus conference. Journal of Hyper tens ion 2000 , 18: 493-508 . 25 CHAPTER III Validation of a New Algorithm for the BPM-100 Electronic Oscillometric Office Blood Pressure Monitor J a m e s M. Wr igh t a b * , M D PhD FRCP(C) , Gurdial S. M a t t i / , M B B C h C C F P , T h o m a s L. P e r r y a b Jr. M D FRCP(C) , Mark E. Gel fer 0 , M D , Kevin D. S t range 0 , M A S c , An ton Zorn°, Dipl T c h , Y u n q u a n C h e n 0 PhD Depar tment of Pharmaco logy & Therapeu t i cs 3 and Depar tment of Med ic ine b , Universi ty of British Co lumb ia , V S M M e d T e c h 0 Ltd., Vancouver , BC, C a n a d a ^Corresponding Author : J a m e s M. Wr ight , MD, PhD, FRCP(C) , Depar tmen t of Pharmaco logy & Therapeu t i cs 3 and Depar tment of Med ic ine 0 , T h e Universi ty of Brit ish Co lumb ia , 2176 Heal th Sc iences Mall , Vancouver , BC, Canada , V 6 T 1Z3, te lephone: (604) 822-4270 , fax: (604) 8 2 2 - 0 7 0 1 , emai l : jmwr igh t@in te rchange.ubc .ca Source of Fund ing: V S M - M e d T e c h Short Tit le: BPM-100 algor i thm val idat ion. 26 ABSTRACT Background: T o test the accuracy of a new algor i thm for the BPM-100 , an au tomated osci l lometr ic blood pressure (BP) monitor, using stored data f rom an independent ly conduc ted val idat ion trial compar ing the B P M - 1 0 0 B e t a with a mercury s p h y g m o m a n o m e t e r . Design: Raw pulse w a v e and cuff pressure data were stored electronical ly using e m b e d d e d sof tware in the B P M -100 B e ta , dur ing the val idat ion trial. T h e 391 sets of m e a s u r e m e n t s w e r e separa ted object ively into two subsets . A subset of 136 measu remen ts was used to develop a new algor i thm to e n h a n c e the accuracy of the device w h e n reading higher systolic pressures. T h e larger subset of 255 m e a s u r e m e n t s (3 readings for 85 subjects) w a s used as test data to val idate the accuracy of the new a lgor i thm. Methods: Dif ferences be tween the new algor i thm BPM-100 and the re ference ( m e a n of two observers) were determined and expressed as the m e a n di f ference + SD, plus the % of m e a s u r e m e n t s within 5 m m H g , 10 m m H g , and 15 m m H g . Results: T h e m e a n d i f ference be tween the BPM-100 and reference systol ic BP w a s - 0 . 1 6 ± 5.13 m m H g , with 7 3 . 7 % < 5 m m H g , 9 4 . 9 % < 10 m m H g and 9 8 . 8 % < 15 m m H g . T h e m e a n d i f ference between the B P M -100 and re ference diastol ic BP w a s - 1 . 4 1 ± 4.67 m m H g , with 7 8 . 4 % < 5 m m H g , 9 2 . 5 % < 10 m m H g , and 9 9 . 2 % < 15 m m H g . T h e s e data improve upon that of the B P M - 1 0 0 B e t a and pass the A A M I s tandard, and \"A\" g rade B H S protocol . 27 Conclusion: This s tudy i l lustrates a new method for developing and test ing a c h a n g e in an algor i thm for an osci l lometr ic BP moni tor util izing col lected and stored electronic data and demons t ra tes that the new algor i thm mee ts the A A M I s tandard and BHS protocol . Key Words: b lood pressure, measurement , monitor, a lgor i thm, osci l lometr ic 28 INTRODUCTION T h e use of electronic blood pressure measur ing inst ruments in d iagnos ing hyper tens ion and moni tor ing blood pressure (BP) has increased dramat ical ly over the past several years . The re have been many art icles wr i t ten suppor t ing the use of such devices, especial ly in the h o m e sett ing (self) [1] and for ambula tory use (24 hour) [2]. Most of these dev ices use the osci l lometr ic techn ique, wh ich measures the m e a n arterial BP direct ly f rom cuff pressure, then calculates the systol ic and diastol ic BP's accord ing to an a lgor i thm that is unique to each device or manufacturer . T h e s e ins t ruments can be val idated by test ing them against auscul tatory measu remen t using a mercury s p h y g m o m a n o m e t e r , accord ing to establ ished protocols set by ei ther the Assoc ia t ion for the A d v a n c e m e n t of Medical Ins t rumentat ion (AAMI) [3] or the Brit ish Hyper tens ion Society (BHS) [4]. It is known that most osci l lometr ic moni tors , though accurate, tend to underes t imate and give a higher s tandard of deviat ion for higher systol ic p ressures w h e n compared wi th s tandard auscul ta tory methods [5]. T h e B P M - 1 0 0 is an osci l lometr ic blood pressure measur ing ins t rument that has been deve loped and manufac tu red by V S M M e d T e c h Ltd. of Vancouver , C a n a d a . Th is dev ice w a s des igned specif ical ly for the pr imary care sett ing, to aid the cl inician in d iagnosing hyper tension and in moni tor ing the patient 's course. It w a s tested in an independent ly conducted val idat ion study accord ing to the A A M I s tandard ; the detai ls of this s tudy are reported separate ly in the previous article [6]. T h e dev ice per fo rmed wel l , passing both the A A M I s tandard and BHS protocol , but tended to underes t imate the higher systol ic blood pressures as detected wi th the B land-A l tman plot. In this art icle, w e report a me thod for the deve lopment and test ing of a modi f icat ion to the algor i thm for est imat ing systolic BP using stored e lectronic da ta . 29 METHODS Subjects w e r e recrui ted and tested at the Blood Pressure Clinic of the Universi ty of Brit ish Co lumbia accord ing to the method reported in the compan ion paper and included ethical approva l , explicit inclusion and exc lus ion cri ter ia, target populat ion object ives, s tudy des ign, data analysis and results [6]. In the original B P M - 1 0 0 B e t a cl inical trial [6], 85 patients w e r e tested and accep ted for statistical analysis. Each patient w a s measured a total of six t imes. T h e first reading w a s used for subject screening purposes, and not used fur ther in the analysis. T h e remain ing f ive readings w e r e fur ther sc reened for exclus ion accord ing to pre-determined cri ter ia, result ing in a m i n i m u m of th ree and m a x i m u m of f ive readings for each subject . T h e total number of readings included for statist ical analysis w a s 3 9 1 , 136 readings m o r e than the m in imum A A M I requ i rement of 255 readings (85 sub jec ts t imes 3 readings for each subject ) . Dur ing the original study, the osci l lometr ic algor i thm of the B P M - 1 0 0 B e t a col lected the cuff pressure and pulse informat ion f rom the blood pressure cuff dur ing def lat ion and calculated the systol ic and diastolic blood pressures as wel l as the pulse rate. T h e e m b e d d e d sof tware in the B P M - 1 0 0 B e t a per formed the data col lect ion and BP calculat ions. Af ter each measurement , the cuff p ressure and pulse informat ion w a s expor ted to another compute r to archive the raw data and al low for analysis of the a lgor i thm. T h e original B P M - 1 0 0 clinical trial data set of 391 m e a s u r e m e n t s w a s object ively separa ted into two data subsets as descr ibed below. T h e first subset included a total of 255 m e a s u r e m e n t s , wh ich included 3 m e a s u r e m e n t s f rom each of the 85 subjects, and represented the m i n i m u m requ i rements for A A M I cri ter ia. T h e second subset of 136 measu remen ts was avai lable to assist in the deve lopment of a modi f icat ion to the B P M - 1 0 0 B e t a a lgor i thm. O n c e the new algor i thm w a s deve loped the f irst subset of 255 m e a s u r e m e n t s w a s used to object ively evaluate the per fo rmance of the a lgor i thm on \"new\" data . T o avoid bias towards early, middle or late data points in the s e q u e n c e of up to f ive measu remen ts col lected f rom each subject in the original clinical data [6], the f irst subset w a s selected using a rotating 30 select ion pointer. T h e procedure involved select ing the first th ree avai lable m e a s u r e m e n t s f rom the first subject and then the second , th ird, and four th m e a s u r e m e n t f rom the second subject and so on . S o m e subjects had only three or four measuremen ts , as s o m e w e r e exc luded in the original clinical trial data col lect ion, but the pointer sk ipped to the next avai lable m e a s u r e m e n t until e a c h subject had exact ly 3 m e a s u r e m e n t s . W h e n the end of the m e a s u r e m e n t s w a s reached for any subject wi thout select ing 3 m e a s u r e m e n t s , the select ion pointer rotated back to the start of the m e a s u r e m e n t s for that subject. T h e smal ler subset w a s used to develop a new algor i thm that gave a better es t imate for higher systolic m e a s u r e m e n t s . Th is did not require changing the algor i thm for the diastol ic BP. Af ter the algor i thm modif icat ion w a s successfu l ly deve loped using the smal ler data subset , it w a s then ready for val idat ion. T h e external a lgor i thm w a s formal ly tested on the first subset of 255 m e a s u r e m e n t s f rom the original clinical trial populat ion (85 subjects t imes 3 m e a s u r e m e n t s per subject ) . T h e pe r fo rmance results of the new algor i thm were analyzed accord ing to the A A M I s tandard for all populat ion, observer , and accuracy requ i rements . T h e new algor i thm w a s then incorporated into the e m b e d d e d sof tware to create the BPM-100 . A n NIBP s imulator (BP P u m p , Bio-Tek Inst ruments, Inc., W i n o o s k i , V e r m o n t ) w a s used to provide osci l lometr ic input data to the BPM-100 device. T h e BPM-100 d isplayed the resul ts and expor ted the cuff pressure and osci l lometr ic data to the PC. T h e new algor i thm deve loped and tested in the PC now used the new data f rom the s imulator via the BPM-100 , and calculated and d isplayed the resul ts. A regression test c o m p a r e d the 2 sets of data over a w ide range of parameters (BP 80-220 /50-130 m m H g and heart rate 50 to 130 bpm) ; acceptance cri teria w e r e that the d i f ference be tween each reading must be within +/-1 m m H g . Th is test ensured that the a lgor i thms imp lemented in the 2 di f ferent so f tware p lat forms operated identical ly and val idated the implementat ion of the algor i thm in the e m b e d d e d sof tware. 31 RESULTS Enrolled subjects See c o m p a n i o n publ icat ion for detai ls of enrol led, included and exc luded sub jec ts [6]. All target populat ion object ives and requi rements of the A A M I protocol w e r e met in this study as they included the s a m e 85 subjects included in the compan ion study [6]. Included blood pressure measurements T h e total n u m b e r of included measu remen ts amoun ted to 255 . Twen ty -seven blood pressure m e a s u r e m e n t s (10 .6%) had a systol ic of greater than 180 m m H g , and 37 m e a s u r e m e n t s (14.5%) had a systol ic of less than 100 m m H g . T h e remain ing 191 systol ic m e a s u r e m e n t s (74 .9%) w e r e between these ex t remes . Twenty- f ive blood pressure measu remen ts (9 .8%) had a diastol ic of greater than 100 m m H g , and 39 m e a s u r e m e n t s (15 .3%) had a diastol ic of less than 60 m m H g . T h e remain ing 191 diastolic m e a s u r e m e n t s (74.9 %) were between these ex t remes . Excluded blood pressure measurements For detai ls of the 34 exc luded blood pressure m e a s u r e m e n t s see c o m p a n i o n s tudy [6]. In addi t ion, a second subset of 136 measu remen ts used to deve lop the new algor i thm was exc luded as descr ibed in the me thods sect ion. Reference standard measurements (Inter-observer) Overal l m e a n d i f ference between the observers (observer 1 - observer 2) for the 255 systol ic blood pressure m e a s u r e m e n t s w a s - 0 . 6 5 + 1.99 m m H g , ( range - 1 0 to +9 m m H g ) . Most of the systolic m e a s u r e m e n t d i f ferences, 98 .0%, were within 5 m m H g . All w e r e within 10 m m H g as any outs ide this range w e r e exc luded as one of the exclus ion cr i ter ia. T h e overal l m e a n d i f ference between the 2 observers for diastol ic blood pressure w a s - 1 . 0 5 ± 2.46 m m H g , ( range, - 8 m m H g to + 7 m m H g ) . Most (95 .3%) w e r e within 5 m m H g and all were within 10 m m H g as descr ibed above . 32 Range and distribution of reference measurements T h e overal l m e a n systol ic blood pressure for the 255 systol ic blood p ressure m e a s u r e m e n t s recorded (average of the two observers) w a s 128.5 ± 30.9 m m H g ( range 81.5 to 223.5 m m H g ) . T h e overal l m e a n diastol ic blood pressure was 77.3 ± 16.6 m m H g ( range 48 to 119.5 m m H g ) . T h e m e a n heart rate using the Nonin f inger pulse ox imeter (Onyx, Nonin Medical Inc., P lymouth , Minnesota) w a s 70.1 + 12.4 beats per minute ( range 4 2 to 104 beats per minute) . Accuracy of BPM-100 as compared to the reference standard measurements. T h e overal l m e a n d i f ference between the re ference s tandard systol ic and diastol ic blood pressure and the BPM-100 ( re ference - BPM-100) is well within the A A M I s tandard and is shown in Tab le 1 [3]. Tab le 2 shows the proport ion of systol ic and diastol ic d i f ferences within 5, 10, and 15 m m H g and how this con fo rms to the B H S protocol . [4] T h e BPM-100 dev ice is des igned to de termine the average of up to f ive b lood pressure readings in individual sub jects . W e therefore compared the m e a n of the 3 re ference s tandard systol ic blood pressure m e a s u r e m e n t s wi th the m e a n of the 3 BPM-100 m e a s u r e m e n t s for individual subjects so as to observe the accuracy in the pract ical clinical set t ing. This gave the s a m e m e a n d i f ference and a lower s tandard deviat ion as can be seen in Tab le 1. It a lso led to an improvement in the proport ion of systolic and diastol ic d i f ferences within 5, 10 and 15 m m H g accord ing to the B H S protocol (Table 2) . In Fig 1 the B land-A l tman display [7] of individual m e a s u r e m e n t s for systol ic blood pressure (N = 255) shows that the d i f ferences of the reference s tandard and BPM-100 blood pressures are c lustered around 0 over the who le range of systol ic readings. In Fig 2 the B land-A l tman display of individual measu remen ts for diastol ic blood pressure (N = 255) shows that the d i f ferences of the re ference s tandard and BPM-100 blood pressures gives the s a m e picture as for the 391 m e a s u r e m e n t s in the prev ious paper. [6] T h e m e a n d i f ference be tween the re ference and measured heart rate ( m e a s u r e d minus reference) was 0.16 ± 2 . 0 2 b p m , wi th a range o f - 1 6 to +6 b p m . 33 DISCUSSION T h e BPM-100 electronic off ice blood pressure inst rument wi th the new algor i thm meets the protocol and s tandard of the Assoc ia t ion for the A d v a n c e m e n t of Medical Inst rumentat ion for accuracy w h e n compared to tradit ional mercury BP measu remen t per fo rmed by trained nurses using a research-grade precision mercury s p h y g m o m a n o m e t e r . A s wel l , it earned an \"A\" g rade for both systol ic and diastol ic readings, accord ing to the Brit ish Hyper tension Society protocol . It improves on the accuracy of est imat ion of higher systol ic blood pressures as compared to the previous vers ion ( B P M - 1 0 0 B e t a ) [6]; the BPM-100 is the only vers ion that is avai lable for sale in North Amer i ca . T h e me thods used in this s tudy demonst ra te the unique character is t ics of us ing the osci l lometr ic method for detect ing blood pressure. T h e readings were obta ined and recorded us ing s tandard val idat ion methods . However , because the osci l lometr ic technique util izes electronic analysis of the pulse waves of each blood pressure measurement , w e were able to utilize the previously recorded test readings to develop a new algor i thm to improve upon the est imate of systol ic blood pressure and subsequent ly to val idate the new algor i thm using recognized protocols and s tandards . T h e use of osci l lometr ic dev ices for measur ing blood pressure is rapidly increasing, and O'Br ien and others have wr i t ten about the need for better methods to test and val idate these dev ices, [8] [9]. They are aware that m a n y manufac tu rers change the algor i thm in their dev ices and do not a lways report or re-val idate the dev ices because the cost of re-test ing is prohibit ive. Presumably , as in the case of the B P M -100, the c h a n g e s in the a lgor i thm are to improve the accuracy of the dev ice. T h e me thod w e descr ibe is one w a y in wh ich c o m p a n i e s can val idate and report these changes , whi le sav ing the t ime and money associated wi th repeat ing the entire val idat ion process. T h e original cl inical data is col lected in the prescr ibed manner , and is recorded and stored both manual ly and electronical ly. Th is data is then avai lable to \"re-test\" any changes in the a lgor i thms used to es t imate the systol ic and diastol ic blood pressures. W e bel ieve that the method presented could represent a precedent for improv ing the eff ic iency of val idat ion of changes to the e m b e d d e d sof tware of osci l lometr ic blood p ressure measur ing dev ices. In 34 this part icular case the method created an algor i thm that improved the accuracy of the BPM-100 osci l lometr ic dev ice for higher systol ic measuremen ts . W h e n this a lgor i thm w a s tes ted, it readily satisf ied the A A M I s tandard and BHS \"A\" protocol . 35 co x> c 03 55 0 X> CD co Q . E o o o o Q. CQ CD o 0 O c co E o x: CD Q . 0 0 CO f— O _i o I— CO < Q O _ i o co >-co x> c CO ^ 3 \"o co S 0 (O X ) E E CO Q) - 1 -J ) \" c !*= E XI c . c o ^ ^ c ~ E 5 5 £ o — 0 = 03 CO <$ CO < 0 1 X I _ u E E I in m CD oo CO CD O O VI m I oo Al O i l CD CN CD oo oo CD CO in CD CN 0 -Q CD 03 I X E E I m VI Al o o o o CL CO x CD 0 lo m in CN oo 0-|co o o 1/3 03 O c CD CD H— co | c o o 0 . CQ 0 . CQ o o 03 eg -a CD o c: CD k_ CD H— CD DH 1/3 co in co 0-|CQ \"3, 03 > ! IDC co I 3 C 0-00 -#—< 03 CD X3 4 — 0 CC co o o Q-c CD 0 E o 0 ' ¥ IW O O O o Q . CO X CQ 0 —^' c T i 0 TO g. CO CO CD 03 c >, 0 0 E o 0 1_ o CO CO CD c 0 sio E irten dual 0 Q- > >* T3 X . c — CO CD 03 m de i CO o X CD Fig 1 Bland-Altman Display of Systolic Measurements, All Cuffs, 255 measurements 30 T3 C 2 0 -re O) X 10 -Q. E CQ E 0 -O a> > o < -10 -ren .a o o - 2 0 -a -.30 --V ^ o 0' o < £ .o jg ° 100 m m H g but <180 m m H g and those >180 m m H g . T h e s e percentages were also expressed for diastol ic measuremen ts , <60 m m H g , >60 m m H g but <100 m m H g and those >100 m m H g . T h e m e a n d i f ference ± s tandard deviat ion (d i f ference be tween the B P M - 1 0 0 and the observer reference) w a s also de termined for both the systol ic and diastol ic m e a s u r e m e n t s . T h e a g r e e m e n t between the two techn iques w a s also expressed as the percentage agreement within 5 m m H g , 10 m m H g and 15 m m H g for each of the systol ic and diastol ic blood pressure m e a s u r e m e n t s . T h e m e a n d i f ference ± standard 49 deviat ion w a s also determined for the d i f ference between the Nonin f inger pulse ox imeter and the observer re ference va lues. W e also calculated the individual subject m e a n s ± s tandard deviat ion (d i f ference be tween the BPM-100 and the observer re ference) for both systol ic and diastol ic blood pressure m e a s u r e m e n t s and expressed the ag reemen t as percentages within 5, 10 and 15 m m H g as above . Th is gave us a better es t imate of individual subject m e a s u r e m e n t s as opposed to total m e a s u r e m e n t s in the g roup. For all exc luded subjects the age, sex and arm size w a s also recorded . T h e reason for exclus ion w a s also l isted and all sub jects w e r e ult imately accounted for. T h e magn i tude of d i f ference in blood pressure m e a s u r e m e n t s by the B P M - 1 0 0 and the auscul tatory re ference va lues by the observers , w a s plotted against , the m e a n s of b lood p ressure measu remen ts by the two me thods for each of the systol ic and diastol ic m e a s u r e m e n t s . Th is is an accepted method as sugges ted by Bland and A l tman [4]. 50 Results Enrolled subjects There w e r e 4 6 subjects enrol led. Of these, 36 (78 .3%) subjects w e r e inc luded and 10 (21.7%) subjects were exc luded. Of the included subjects average age w a s 9.4 years wi th a s tandard deviat ion of 4 .2. Enrol led subjects inc luded those with hyper tension, renal artery s tenosis , k idney fai lure and congenital co-arctat ion of the heart. S o m e of these subjects w e r e tak ing m a n y di f ferent medicat ions including ant i -hyper tensives, steroids and immunosuppressan ts . Excluded subjects T e n (21 .7%) subjects were exc luded; all were pos t -measurement exc lus ions. T h e average age was 6.9 with a s tandard deviat ion of 4.2 years. T h e age range of this g roup w a s be tween 3 and 15 years. There w e r e 3 (30 .0%) ma le subjects and 7 (70.0%) fema le subjects a m o n g those exc luded . Six (60%) of those exc luded w e r e because observer ag reement on the first or screen ing m e a s u r e m e n t did not meet the pre-speci f ied cr i ter ia. Of these, one was because one of the observers did not hear the screening measuremen t . Th ree subjects were exc luded because of technical errors p roduced by the BPM-100 that did not a l low three sat isfactory blood pressure measu remen ts to be ach ieved . O n e subject had readings that because of technical reasons and observer ag reemen t on individual data points did not al low for three sat isfactory m e a s u r e m e n t s to be col lected. Target populat ion results All of the target populat ion object ives were met, except for the hyper tens ive subjects . There were 23 (63.9%) male subjects and 13 (36.1%) fema le subjects inc luded. T h e ages ranged f rom 3 to 18 years. There w e r e 6 subjects aged 3 to 5 years, 10 subjects aged 6 to 8 years, 9 subjects aged 9 to 11 years, 5 subjects aged 12 to 14 years and 6 subjects aged 15 to 18 years . Of these 28 (60 .9%) subjects were between the ages of 3 and 12 years, this met all age requ i rements speci f ied in the me thods sect ion and fulf i l led the A A M I Draft SP 10 criteria [d]. There were 5 hyper tensive sub jec ts that enrol led and all were inc luded. Th is did not meet the enro lment criteria speci f ied in the me thods sect ion and will be d iscussed later. Seven (15 .2%) of the enrol led subjects, all of w h o m w e r e inc luded, w e r e tested wi th the infant cuff (13 to 18 c m c i rcumference at mid-b iceps wi th the cuff, as speci f ied in the target populat ion sect ion) and 5 1 met the A A M I Draft S P 10 criteria [5]. There were 16 (44.4%) subjects wi th a rm sizes less than or equal to 18 c m , and the remain ing 20 (55.6%) had arm sizes greater than 18 c m m e a s u r e d at mid-b iceps. T h e m e a n populat ion a rm size was 20.8 ± 5.2 cm with a range f rom 15 - 33.5 c m . Included blood pressure measurements Of the included 162 sets of readings; 62 (38.3%) m e a s u r e m e n t s were less than 100 m m H g for systolic blood pressure, 100 (61.7%) measu remen ts were greater than or equal to l O O m m H g but less than 180 m m H g . There were no readings greater than 180 m m H g for the systol ic blood pressures . For diastol ic blood pressure 13 (8 .0%) measu remen ts w e r e below 60 m m H g , 149 (92 .0%) of m e a s u r e m e n t s were greater than or equal to 60 m m H g wi th no readings greater than 100 m m H g . In the included populat ion 5 or 13 .9% subjects were previously d iagnosed as having hypertension by their Pr imary Care Physic ians. Th is did not meet the target criteria of 6 sub jec ts as speci f ied in the methods sect ion. Th is pre-speci f ied target criteria w a s intended to ref lect the A A M I S P 10-1992 standard [6], wh ich requires 1 0 % of the systolic m e a s u r e m e n t s to be greater than 180 m m H g (or s tage 3 hyper tension) , and 1 0 % of the diastol ic measu remen ts should be greater than 100 m m H g (or s tage 2 hyper tension) [7]. It should be noted that the determinat ion of hyper tens ion in chi ldren is based on the chi ld's height, and therefore a range of blood pressures are cons idered hyper tensive [1]. However regardless of age or height of the child a systol ic blood pressure greater than 140 m m H g for boys or 132 m m H g for girls, or diastol ic pressures greater than 89 m m H g for boys or 86 m m H g for girls would be cons idered hyper tensive [8]. T h e 23 measu remen ts col lected f rom these sub jec ts accounted for 14 .2% of the total 162; a l though of these only 9 (5 .6%) systol ic m e a s u r e m e n t s exceeded the 140 m m H g for boys or 132 m m H g for girls and 1 (0 .6%) diastol ic m e a s u r e m e n t s exceeded the 89 m m H g for boys or 86 m m H g for gir ls. Excluded blood pressure measurements From the 10 subjects exc luded, thirty-f ive measu remen ts were recorded that w e r e not used in the data analysis for the study. Their m e a n reference systolic and diastol ic blood pressures were 98.6 ± 9.5 m m H g ( range 82.5 - 118.5 m m H g ) and 63.5 ± 7.0 m m H g ( range 48.5 - 76.0 m m H g ) respect ively, with heart rates f rom the Nonin f inger pulse ox imeter of 85.3 ± 20.2 bpm ( range 54 - 1 1 3 b p m ) . There were 22 52 (62 .9%) systol ic blood pressure measu remen ts less than 100 m m H g and 13 (37 .1%) were greater than or equal to 100 m m H g but less than 180 m m H g . The re w e r e no systol ic m e a s u r e m e n t s greater than 180 m m H g . The re were 8 (22 .9%) diastol ic pressure m e a s u r e m e n t s less than 60 m m H g and 28 (77.1%) were greater than or equal to 60 m m H g but less than 100 m m H g , wi th no m e a s u r e m e n t s greater than 100 m m H g . Reference standard blood pressure readings T h e re ference s tandard ( m e a n of the two observer 's m e a s u r e m e n t s ) inc luded 162 sets of readings. T h e m e a n systol ic blood pressure (average of the two observers) was 100.9 wi th a s tandard deviat ion of 15.3. T h e m e a s u r e m e n t s ranged f rom 78.0 to 151.5 m m H g . T h e m e a n re fe rence s tandard diastol ic b lood pressure m e a s u r e m e n t was 62.3 ± 9.3 m m H g with a range of 45 .0 to 89.5 m m H g . Similar ly the m e a n reference heart rate using the f inger pulse ox imeter w a s 88.2 ± 13.0 b p m with a range of pulse rates f rom 59 to 116 b p m . Observer ag reemen t (observer 1 - observer 2) for systolic m e a s u r e m e n t s showed a m e a n di f ference of 0.4 ± 2.8 m m H g ( range -10 to 10 m m H g ) . Of these 154 (95.1%) m e a s u r e m e n t s w e r e less than or equal to 5 m m H g . For diastol ic measurements , observer ag reemen t m e a n d i f ference w a s 0.8 ± 4.0 m m H g ( range -10 to 10 m m H g ) wi th 135 (83 .3%) m e a s u r e m e n t s less than or equa l to 5 m m H g . Th is did not meet the 8 5 % agreemen t level l isted in the A A M I SP10-1992 , a l though this s tandard w a s establ ished f rom studies in adult populat ions with much higher blood pressures [6]. Measurements o f B p T R U ™ (model BPM-100) There were also 162 sets of included measu remen ts col lected by the B P M - 1 0 0 . T h e m e a n systolic blood pressure m e a s u r e m e n t was 102.3 wi th a s tandard deviat ion of 14.6 m m H g and a range of 77 to 145 m m H g . M e a n diastol ic blood pressure m e a s u r e m e n t s were 59.1 ± 8.6 m m H g ( range of 40 to 83 m m H g ) . T h e pu lse rates col lected by the B P M - 1 0 0 had a m e a n o f 88.4 ± 12.8 b p m a n d s h o w e d ranges f rom 63 to 116 b p m . 53 Accuracy T h e d i f ference be tween the m e a n B P M - 1 0 0 ' readings and the re ference s tandard measurements , de termined by the observers , as d isplayed in Tab le 1 for systol ic b lood pressures w a s 1.45 ± 5.67 m m H g ( range -15 to 2 2 . 5 m m H g ) and for diastol ic measu remen ts w a s -3.24 ± 7.39 m m H g ( range -22.5 to 17 m m H g ) , wh ich met the cri teria set by the A A M I s tandard [c]. T h e accuracy of the B P M - 1 0 0 against the heart rate re ference showed a m e a n di f ference of 0.20 ± 2.47 bpm ( range -8 to 7 b p m ) . Tab le 2 shows the proport ion of individual and then subject m e a s u r e m e n t s of both systol ic and diastol ic blood pressure d i f ferences within 5, 10 and 15 m m H g and how this c o m p a r e s to the BHS protocol [9]. A lso the m e a n systol ic blood pressure between the B P M - 1 0 0 and the observer re ference s tandards w a s 101.6 ± 14.7 m m H g and the diastol ic blood pressure w a s 60.7 ± 8.2 m m H g . 54 Discussion Most International bodies agree that all au tomated blood pressure dev ices should have independent val idat ion per fo rmed. However this is not the case for most dev ices [10]. Recent ly the work ing group on blood pressure moni tor ing of the European Society of Hyper tens ion has c o m e up wi th a revised BHS protocol that will be known as the ' International Protocol ' in order to encou rage blood pressure device manufac turers to val idate their dev ices and m a k e the process easier and potential ly less costly. However , it is des igned for adul ts greater than 30 years o f a g e a n d d o e s not m a k e recommenda t i ons for special ages or c i rcumstances including chi ldren [11]. As d iscussed above, the hypertensive cri terion w a s intended to ref lect the A A M I S P 1 0 - 1 9 9 2 standard and ul t imately ensure the B P M - 1 0 0 to be accurate in subjects with e levated b lood pressures . Hyper tension in chi ldren is not a c o m m o n occur rence and, therefore, despi te n u m e r o u s a t tempts to locate this populat ion subset , V S M Medtech Ltd. chose to stop enro lment at 5 subjects . A l though this s tudy only included 5 pre-determined hyper tensive subjects , e levated blood pressures were tested as part of the main (original) clinical study [2], T h e B P M - 1 0 0 per fo rmed well in both original studies [2] [3]. In the c o m p a n i o n paper that comb ines the adult and chi ld data as a meta-analys is it can be seen that all hyper tensive criteria are clearly met . Hypotens ion w a s not directly tested in this study because the A A M I s tandard requests that in adults 1 0 % of systol ic m e a s u r e m e n t s to be less than 100 m m H g and 1 0 % of diastol ic m e a s u r e m e n t s to be less than 60 m m H g . In our s tudy 6 1 . 7 % (n = 100) of the systolic readings w e r e <100 m m H g and 4 2 . 6 % (n = 69) of diastol ic read ings w e r e <60 m m H g . T h e inter-observer ag reemen t required 8 5 % of readings to be within ± 5 m m H g . In our study 8 3 . 3 % of diastol ic readings w e r e within this requi rement . This is not total ly unexpec ted as auscul tat ing diastolic blood pressure in chi ldren wi th low blood pressures and rapid heart rates is m o r e difficult than in adults [8] [12]. 55 T h e 8 5 % agreemen t descr ibed in the A A M I SP10-1992 s tandard w a s estab l ished based on adult studies for adul ts in a samp le size of 85. Th is requ i rement m a y be unreal ist ic w h e n looking at a smal l subset of subjects m a d e up of only chi ldren. Ger in et al descr ibed in their paper on , 'How w e should be measur ing b lood pressure in the doctor 's of f ice?' that \"Rout ine blood pressure measu remen t would be m o r e representat ive of daily ambulatory pressures if an au tomated device, wi thout doctor or nurse present, were u s e d \" [13]. T h e B p T R U ™ ( B P M -100 mode l ) is not only a safe and accurate device for determin ing b lood pressure in chi ldren in the pr imary care set t ing, but a lso meets the requ i rements of the A N S I / A A M I S P 1 0 - 1 9 9 2 and achieved a 'B ' g rade accord ing to a modi f ied B H S protocol . 56 o 'c o o 0 LU co CZ ro co ro c o ro Z ro o ' \\ 0 E < 0 .c: CD i_ co CL E o o w CD E CD i_ ZZ CO CO CD CO \"O c ro \"oo CD o c CD L _ CD H— CD v_ CD sz -«—I \"O c ro o o 0_ CO CD sz cz CD 0 £ 0 -O 0 o cz 0 CO 1 0 0 E o c ro E o E CD _ CL 0 C O T> Q ® T - ro S £ ro 3 I— < c o £ £ c | E iS 0 E (0 \"0 •—' O O -4—< CO ro 0 ^ c £ °> ro CD -L. 0 S- p cr 0 c -E 3= E 1-a E JS 0 E CO T3 O O \"(0 >> co 0 ^ CO Q) - L . 0 i - P c 0 £ \" E st E T3 ' ' o o CO oo 1^ cd V CN o CN i n i n CO co\" + i 1 • CO o CD CN c o i n V i n o O CD i n t — + i CN CL C O C O CN CD CD CO CO TJ C ro \"co 0 o cz 0 a 0 k_ CO c 2 E o o C L co ro x> c ro to 0 o c 0 1_ 0 H— 0 CO c o o C L C O cz ro 0 E o 0 ic? C O X E E LO VI ^8 co o O cd CO o O) 03 o CO Al D) X E E o VI CD CN T— i n o CO CO Al CD X E E CO VI ^8 CN CN o O CN a j o LO 1^ en CD in CN CN CD CD CD CD CO CO 0 £ -a i» co •o c 3 \"to C O X C O C O CD T3 2 o 3 to to 0 T3 O O X I O ~o \"to >< to 0 o c 0 L _ 0 M — 0 l_ CO c o o Q . C Q 0 to to 0 l_ a. T 3 o o o o \"to co T3 0 O c 0 0 i in 13 c E E o o Q . C O 13 to to 0 T3 O o O O 0 3 \" CO c o o 0_ C O c co 0 o 0 CO 3 0 o c 0 0 0 c: c co co 0 0 E E o 0 la\" to w c E E o o Q . C D c co 0 E o 0 D C O oo Fig. 2 U) CO > «s <» I 1 0 ° o £ c E iH £ -10 Q -20 Bland-Altman Display of Systolic Measurements, All Cuffs, 162 measurements M e a n D i f f e r e n c e : 1 . 4 5 m m H g S D o f D i f f e r e n c e s : 5 . 6 7 m m H g 30 20 0 • * # • • I • • — w 3 i , • +2SD Mean -2SD 50 75 100 125 150 175 Mean of B P M and O b s A v g (mmHg) Bland A l tman display of all systol ic blood pressure m e a s u r e m e n t s (n = 162). BPM, B p T R U ™ (model BPM-100) monitor; Obs Avg , observer ave rage ( re ference s tandard) 59 Fig. 3 Bland Altman Display of Diastolic Measurements, All Cuffs, 162 measurements Mean Difference: -3.23 mmHg SDof Differences: 7.39 mmHg & 1 20 ° ^ 10 <1> CO £. O is re 0 -10 -20 -30 30 2 2 : • V '«» 4 « ~ • • * • +2SD Mean -2SD 45 60 75 90 105 Mean of B P M a n d O b s A v g (mmHg) Bland Altman display of all diastolic blood pressure measurements (n = 162). BPM, BpTRU™ (model BPM-100) monitor; Obs Avg, observer average (reference standard) 60 References 1. National Heart, Lung and Blood Institute. Report of the Second Task Force on Blood Pressure Control in Children-1987. Paediatrics 1987; 79: 1-25. 2. Mattu GS, Perry TL Jr, Wright JM. Comparison of the oscillometric blood pressure monitor (BPM-100 B e ta) with the auscultatory mercury sphygmomanometer. Blood Pressure Monitoring 2001; 6: 153-159. 3. Wright JM, Mattu GS, Perry TL Jr, Gelfer M, Strange KD, Zorn A, Chen Y. Validation of a new algorithm for the BPM-100 electronic oscillometric office blood pressure monitor. •* Blood Pressure Monitoring 2001; 6: 161.165. 4. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-310. 5. Manual, electronic or automated sphygmomanometers, 3ed (proposed AAMI / American National Standard. Draft ID: AAMI / CDV-1 SP10-2001. Arlington, VA. 6. American National Standard for Electronic or Automated Sphygmomanometers. ANSI / AAMI SP10-1992. Arlington, VA: AAMI, 1993. 7. The Sixth Report on the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. NIH Publication No. 98-4080, November 1997. 8. Update on the Task Force Report (1987) on High Blood Pressure in Children and Adolescents: A working Group Report from the National High Blood Pressure Education Program. NIH Publication No. 96-3790 September 1996. Paediatrics 1996; 98: 649-658. 9. O'Brien E, Petrie J, Littler W, de Swiet M, Padfield PI, Altman DG, et al. The British Hypertension Society protocol for the evaluation of blood pressure measuring devices. Journal of Hypertension 1993; 11(suppl 2): S43-S62. 10. O'Brien E. State of the market in 2001 for blood pressure measuring devices. Blood Pressure Monitoring 2001; 6: 171-176. 11. O'Brien E, Pickering T, Asmar R, Myers M, Parati G, Staessen J, Mengden T, Yutaka I, Waeber B, Palatini P, and with statistical assistance of Atkins N, Gerin W. Working Group on Blood Pressure Monitoring of the European Society of Hypertension International Protocol for validation of blood pressure measuring devices in adults. Blood Pressure Monitoring 2002; 7: 3-17. 12. Nehal US, Ingelfinger JR. Pediatric hypertension: recent literature. Current Opinion in Pediatrics 2002; 14: 189-196. 13. Gerin W, Marion RM, Friedman R, James GD, Bovbjerg DH, Pickering TG. How should we measure blood pressure in the doctor's office? Blood Pressure Monitoring 2001; 6: 257-262. 61 ABSTRACT Background To comb ine the data f rom an earl ier adult s tudy with the data f rom the subsequen t paediatr ic study using the B p T R U ™ (BPM-100 model ) in order to determine the overal l accuracy against recognised standard auscul tatory mercury s p h y g m o m a n o m e t e r in the general populat ion. Design The individual blood pressure points recorded for both adult and paediatr ic studies w a s compared directly to its cor respond ing observer re ference measu remen ts f rom data col lected and stored f rom the two separate studies. There were 255 sets of readings in the adult s tudy and 162 sets f rom the paediatr ic study, wh ich were comb ined to m a k e 4 1 7 pairs of blood pressure readings for this study. Methods The overal l observer s tandard reference m e a n for 417 measu remen ts w a s calculated and the di f ference between this and the overall m e a n BPM-100 w a s calculated with S D and ranges. Measuremen ts within 5, 10 and 15 m m H g agreement were expressed as percentages. Results A total of 121 subjects were included for this study (85 f rom the adult s tudy and 36 f rom the paediatr ic study). From these, 417 paired measu remen ts were recorded. T h e m e a n d i f ference between the B p T R U ™ and the reference standard systol ic BP w a s 0.47 + 5.40 m m H g wi th 8 9 . 2 % measuremen ts within 5 m m H g , 9 6 . 4 % within 10 m m H g and 99 .3% within 15 m m H g . Comparat ive ly w e found a mean di f ference between the B p T R U ™ and reference diastolic BP w a s -2.12 ± 5.93 m m H g with 8 1 . 1 % within 5 m m H g , 9 2 . 1 % within 10 m m H g and 97 .6% within 15 m m H g agreement . Conclusion The B p T R U ™ has been shown to be an accurate non- invasive blood pressure moni tor ing device in the general populat ion f rom age 3 up to the elderly. This comb ined study meets all requ i rements of the Assoc ia t ion of A d v a n c e m e n t of Medical Instrumentat ion and achieved a g rade 'A' in the BHS protocol. 63 Introduction T h e B p T R U ™ (mode l BPM-100) is a non- invasive au tomated electronic b lood pressure moni tor that uses the osci l lometr ic techn ique to de termine the m e a n arterial p ressure and then calculate the systolic and diastol ic blood pressures f rom this. It records six m e a s u r e m e n t s automat ical ly , separa ted by a t ime period of 1 to 5 minutes set as required by the operator . It uses the first reading to acc l imat ise to the patient 's p ressures and then genera tes the average of the remain ing f ive read ings. H o m e blood pressure readings wi th an au tomated device has been s h o w n to have greater predict ive value than off ice blood pressures for mortal i ty in a recent popula t ion-based observat ional s tudy [1]. Hyper tension has a major impact on our health serv ices; it is the major risk fac tor for heart d isease, s t roke and renal d isease. Heart d isease is the leading cause of death in the US, wi th cerebrovascu lar d isease being the third leading cause of death [2]. Jof f res et al d iscuss d i f ferences in hyper tens ion awareness , t reatment and control of hypertension between the US and C a n a d a yet p lace little impor tance on the two very di f ferent sets of observers taking the blood pressures, saying that these should have minimal impact. In contrast Ger in et al clearly emphas ized the di f ferences and conc luded that an au tomated device would be better than both methods [3] [4]. The most accura te techn iques for measur ing blood pressure (eg: intra-arterial catheter izat ion) are not practical or funct ional for the pr imary care physic ian w h o is the front l ine for d iagnos ing, moni tor ing and manag ing hyper tens ion. If per fo rmed correctly, off ice blood pressures wi th a mercury s p h y g m o m a n o m e t e r m a y be accurate; however, it has been shown that at least 2 0 % of hypertension is misd iagnosed this w a y [5]. In addit ion the diff icult ies of manag ing mercu ry spil ls is a growing concern to a degree that m a n y European and US cities have a l ready banned mercu ry and others are set to fo l low suit [6] [7]. It has been shown m a n y t imes that blood pressures recorded in the physic ians of f ice by the physician can be mis leading and readings per formed by the off ice nurse or the pat ient are m o r e representat ive. In fact 64 such measu remen ts would result in less t reatment initiation, change in current t reatment or escalat ion of dosage [8]. The impor tance of correct ly d iagnosing and manag ing blood pressure clearly has a potential for impact on mortality, morbid i ty and health care costs. However, these benef i ts can only be real ized if w e have a method of accurate ly record ing blood pressures in a manner similar to that seen in the clinical trial sett ing. The purpose of this study w a s to comb ine the data for the general populat ion f rom 3 years of age to the elderly in order to determine the overall accuracy of the au tomated blood pressure monitor, B p T R U ™ against the gold s tandard mercury sphygmomanomete r . To ach ieve this, the study w a s assessed in accordance with the s tandard guidel ines of the AAMI SP10-1992 provided by the Associat ion for the A d v a n c e m e n t of Medical Instrumentat ion (AAMI) and in accordance with the Brit ish Hyper tension Society protocol -1993; the regulat ing bodies for val idat ing these types of moni tors . 65 Methods Adul t subjects w e r e recrui ted dur ing Sep tember and October 1999 at the Universi ty of Brit ish Co lumbia using subjects f rom the Hyper tension clinic and aff i l iated pr imary care pract ices. T h e paediatr ic subjects were recrui ted f rom June to August 2001 at the Research of f ice of V S M M e d T e c h Ltd f rom B.C.'s Chi ldren's Hospital and also fami ly pract ice off ices. Detai ls of all ethical approval , consent acqu i red, inclusion, exc lus ion cri teria and target populat ion requ i rements can be obta ined f rom the compan ion paper and the two original adult publ icat ions [9] [10] [11]. In the adult populat ion two exper ienced nurses were used w h o s e blood pressure reading skills were tested prior to the initiation of any data col lect ion. Similar ly in the paediatr ic populat ion three exper ienced nurses rotated as the two observers . T h e set up w a s the s a m e for both studies with the B p T R U ™ being connec ted in parallel wi th the Tr iml ine precis ion s tandard mercury s p h y g m o m a n o m e t e r and the Nonin f inger pu lse ox imeter being connected to the other a rm or al ternate body part. T h e observers , w h o were seated oppos i te the s tandard mercury s p h y g m o m a n o m e t e r , were bl inded by a curtain between them and also bl inded to the B p T R U ™ device. O n e of them wou ld palpate the brachial pulse and put the d iaphragm of the dua l -headed s tethoscope over it. T h e only d i f ference between the two populat ion sub-groups w a s that the paediatr ic subjects were al lowed to sit on the laps of their accompany ing parent or guard ian. All data recorded by the B p T R U ™ dev ice w a s s imul taneously saved onto a connec ted laptop f rom which subsequent analysis could be per fo rmed. From the 121 subjects recrui ted for this study, 417 data points were obta ined, all of wh ich w e r e used in our s tudy having a l ready con fo rmed to the inclusion and exc lus ion cri teria speci f ied in the previous studies. 66 Data Analysis After combin ing the 85 adult subjects (at least 18 years of age) and the 36 paediatr ic subjects ( f rom 3 to 18 years of age) , a total of 121 subjects w e r e inc luded. For these subjects the sex and age distr ibut ions were de te rmined . For each of the inc luded measuremen ts , distr ibut ion of blood pressures w a s de te rmined and expressed as the n u m b e r and percentages in the ranges 180 m m H g < systol ic < 100 m m H g and 100 m m H g < diastol ic < 60 m m H g . Included in this w a s an express ion of the m e a n , s tandard deviat ion and range of blood pressures in the populat ion group. From the 4 1 7 m e a s u r e m e n t s the m e a n , s tandard deviat ions and range of blood pressure and heart rate was also determined for observer 1 and observer 2. Subsequent ly , the m e a n of the two observers w a s also calculated and b e c a m e known as the re ference s tandard m e a s u r e m e n t s . A g r e e m e n t of readings within 5 and 10 m m H g w a s also expressed . Similarly, the m e a n , s tandard deviat ion and range of blood pressure w e r e also determined for the B p T R U ™ device. T o assess the accuracy of the device, the m e a n di f ference and s tandard deviat ion, be tween the B p T R U ™ and the re ference s tandard w a s determined as per the A N S I / A A M I S P 1 0 - 1 9 9 2 s tandard [12]. Subsequent ly , in order to reflect this accuracy in the clinical set t ing, the sub ject m e a n di f ference with s tandard deviat ion w a s also determined (subject m e a n B p T R U ™ m e a s u r e m e n t s minus the subject m e a n reference measu remen ts ) . T h e percentage agreement of m e a s u r e m e n t s wi th in 5, 10 and 15 m m H g w a s also expressed in order to represent the requ i rements of the Brit ish Hyper tens ion Society protocol [13]. 67 Results Included subjects There w a s a total of 121 subjects included f rom the two studies (85 f rom the adult s tudy [10] and 36 f rom the paediatr ic s tudy [11] - see compan ion paper) . Of these 67 (55 .4%) w e r e male subjects and 54 (44.6%) w e r e fema le subjects . T h e average age of included subjects w a s 33.1 years of age, ranging f rom 3 to 83 years . Detai ls of included and exc luded subjects can be seen in the original adult study [10] and the paediatr ic compan ion paper [11]. All of the target populat ion requ i rements for each study were met, except the hyper tensive criteria speci f ied in the paediatr ic study, this w a s d iscussed at s o m e length in the paediatr ic paper. Included blood pressure measurements From the 121 subjects included there w a s a total of 417 sets of readings ob ta ined . Of these, 27 (6.5%) had systol ic m e a s u r e m e n t s greater than 180 m m H g and 137 (32 .8%) had systol ic measu remen ts less than 100 m m H g . T h e remain ing or 253 (60.7%) had blood pressure m e a s u r e m e n t s equal to or greater than 100 m m H g and equal to or less than 180 m m H g as accord ing to the m e a n reference observer s tandard m e a s u r e m e n t s . Twenty- f ive (6 .0%)^ blood pressure readings had diastol ic measuremen ts greater than 100 m m H g and 108 (25.9%) had diastol ic blood pressures less than 60 m m H g with remain ing 284 (68 .1%) be tween these. T h e overal l average systol ic b lood pressure w a s 117.7 ± 29.23 m m H g ( range 78 to 223.5 m m H g ) and the overal l average diastol ic blood pressure w a s 71.5 ± 15.95 m m H g ( range 45 to 119.5 m m H g ) . T h e overal l m e a n heart rate w a s 77.1 + 15.41 bpm (range 42 to 116 bpm) using the Nonin f inger pulse ox imeter . T h e s e overal l va lues w e r e de termined f rom the m e a n observer re ference m e a s u r e m e n t s . Excluded blood pressure measurements There w e r e six subjects exc luded because of technical errors m a d e by the B p T R U ™ device that did not a l low for a m i n i m u m of three blood pressure readings to be recorded. More detai led explanat ions of all exc luded subjects and m e a s u r e m e n t s can be obta ined f rom the previous publ icat ions [9] [10] [11]. 68 Reference standard blood pressure readings Observer ag reemen t (or the m e a n di f ference between observer o n e and two) for the 417 systolic m e a s u r e m e n t s w a s -0.26 ± 2.38 m m H g ( range -10 to 10 m m H g ) as recorded using a Tr iml ine mercury s p h y g m o m a n o m e t e r . Of these 404 (96 .9%) readings were within 5 m m H g . Similar ly the observer ag reement for the 4 1 7 diastol ic blood pressure readings w a s -0.35 ± 3.27 m m H g ( range -10 to 10 m m H g ) with 378 or 9 0 . 6 % being within 5 m m H g , as seen in Tab le 1. A l though mos t of the readings were within 5 m m H g ag reemen t (as m e a n values greater than 10 m m H g w e r e exc luded as one of the post-m e a s u r e m e n t exc lus ion cr i ter ia), the diastol ic m e a s u r e m e n t s had a smal le r n u m b e r wi th in this ag reemen t range. Th is highl ights the diff iculty of accurate ly measur ing diastol ic b lood pressure, part icularly in the paediatr ic populat ions or young fema le adults w h e n reading va lues f rom as low as 45 m m H g . Measurements of BpTRU™ (model BPM-100) Table 1 a lso shows is that the overal l m e a n systol ic blood pressure recorded by the B p T R U ™ was 118.2 ± 28.47 m m H g with a range of 77 to 221 m m H g , and diastol ic blood pressure of 69.4 ± 16.3 m m H g ( range 40 to 120 m m H g ) for 417 m e a s u r e m e n t s . T h e overal l m e a n heart rate recorded by the B p T R U ™ was 77.3 + 15.33 b p m with a range of 116 to 417 bpm for the 417 read ings. Accuracy T h e accuracy of the B p T R U ™ , or the m e a n di f ference between the B p T R U ™ and the observer reference s tandard systol ic b lood pressure w a s 0.47 ± 5.40 m m H g which is wel l wi thin the A N S I / A A M I SP10-1992 s tandard . Simi lar ly the m e a n d i f ference be tween the B p T R U ™ a n d the re fe rence diastol ic b lood pressure was 2.12 ± 5.93 m m H g which is also within the A N S I / A A M I S P 1 0 - 1 9 9 2 s tandard , as seen in Tab le 2. T h e m e a n d i f ference of the observer re ference and the B p T R U ™ for the heart rate w a s 0.18 + 2.20 b p m . In Tab le 3 it can be seen that f rom these values 372 (89 .2%) systol ic m e a s u r e m e n t s were less than or equal to 5 m m H g , 402 (96.4%) w e r e within 10 m m H g and 414 (99 .3%) w e r e within 15 m m H g agreement . For the diastol ic m e a s u r e m e n t s 338 (81 .1%) were within 5 m m H g , 384 (92 .1%) w e r e within 10 m m H g and 407 (97 .6%) w e r e within 15 m m H g . 69 T h e s e determinants all fall within the Brit ish Hyper tension Society (BHS) g rade 'A' protocol . It can also be seen that the percentage of systol ic measu remen ts in ag reemen t w a s better than that of the diastol ic. As stated previously the B p T R U ™ is an au tomated device that records and ca lcu lates the average of f ive blood pressure m e a s u r e m e n t s . It is intended to be used in the physic ian's of f ice in this manner . Thus the subject m e a n d i f ference w a s determined in order to reflect the accuracy of the dev ice in the clinical set t ing. T h e subject m e a n di f ference for systol ic blood pressures w a s 0.4 ± 4.3 m m H g , diastol ic m e a s u r e m e n t s p roduced -2.0 + 5.1 m m H g and the subject m e a n d i f ference for heart rate was 0.2 + 1.1 b p m . From these, for systol ic blood pressure 91 (75 .2%) subjects w e r e wi th in 5 m m H g , 119 (98.3%) w e r e within 10 m m H g and 121 (100%) w e r e within 15 m m H g . For diastol ic blood pressure, 86 (71.1%) subjects w e r e within 5 m m H g , 112 (92.6%) w e r e within 10 m m H g and 119 (98 .3%) w e r e within 15 m m H g ag reemen t be tween the B p T R U ™ and the m e a n observer re ference s tandard , as s h o w n in Tab le 3. Bland A l tman plots w e r e also displayed as the d i f ference of the B P M and the observer average against the m e a n of the B p T R U ™ and the observer average for both systol ic and diastol ic pressures (see f igures 1 and 2) [14]. 70 Discussion There have been several publ icat ions on the impor tance of tak ing accura te and reproducib le blood pressures. Campbe l l and M c K a y have i l lustrated the potential dif f icult ies wi th inaccurate readings; demonst ra t ing that consis tent overest imat ion of diastol ic blood pressure by 5 m m H g wou ld increase the hypertension in a physic ians pract ice by 100% and conversely, underes t imat ing the diastol ic blood pressure by just 5 m m H g wou ld reduce by 6 2 % the number of pat ients seen to be hyper tensive [15]. Blood pressure should be recorded bear ing in mind the fol lowing fac tors : - [4] [16] [17] [18] pat ients should refrain f rom smok ing or ingest ing caf fe ine at least 30 minutes before having their b lood pressures recorded blood pressure should be recorded in a quiet and calm env i ronment pat ients should be seated with their backs wel l suppor ted , feet f lat on the g round with the mid-point of the bare upper a rm suppor ted at heart level an appropr ia te sized cuff ( the bladder inside the cuff should be at least 8 0 % of the c i rcumference) should be placed on the upper arm of the patient 2 to 2.5 c m above the antecubi ta l fossa blood pressure m e a s u r e m e n t s should only be recorded after at least 5 minutes of rest in this posit ion m e a s u r e m e n t s should be taken either with an au tomated blood pressure dev ice or a regularly cal ibrated mercury s p h y g m o m a n o m e t e r inflate the cuff to at least 30 m m H g above the d isappearance of the radial pulse of the arm f rom which pressures are being recorded al low a l inear def lat ion of 2 m m H g / sec Korotkof f V or the d isappearance of sounds should be recorded as the diastol ic blood pressure the average of at least two readings, at least two minutes apart should be taken at each visit at least two visits are required to m a k e the d iagnosis of hyper tens ion unless there is ev idence of end-organ d a m a g e wi th hypertension at one visit 71 T h e au tomated dev ice assessed here saves t ime and m a k e s it more likely that the a b o v e criteria will be fo l lowed: -pat ients can be placed in a quiet room by themse lves avoid ing the ef fects of wh i te coat or off ice hyper tension di f ferent cuf fs suppl ied wi th the device are clearly marked to al low for the mos t appropr ia te cuff to be used accord ing to a rm size the dev ice automat ica l ly cal ibrates to a tmospher ic pressure and amb ien t tempera tu re before each m e a s u r e m e n t reducing errors of cal ibrat ion it inf lates to 35 m m H g the level of occlus ion of the radial pulse so as not to miss auscul tatory gaps it has l inear def lat ion to reduce errors of detect ing pulse w a v e s the device has reproducib le smoo th inflation and def lat ion that can only be obta ined using s tandard ised compute rs or a device and not by any observers the dev ice is cal ibrated to agree with Korotkof f V generat ing accura te diastol ic b lood pressures the f irst m e a s u r e m e n t is ignored s ince the person start ing the dev ice m a y still be in the room f ive m e a s u r e m e n t s can be obta ined at 1 to 5 minute intervals wi th the average calculated at the end giving a m o r e representat ive recording of their b lood pressure status Wi th the advent of newer blood pressure measur ing devices, there has been renewed call for more val idat ion trials and standardisat ion of the compar isons of these dev ices . [19]. Recent ly T h e European Society of Hyper tens ion W o r k i n g Group has used publ ished data to dev ise a new International protocol that is hoped wou ld encourage all device manufac turers to per form and adhere to m i n i m u m val idat ion and subsequent ly approval protocols [20]. T h e dev ice tested here met the requ i rements set by the govern ing bodies ment ioned , wh ich are far more r igorous than the new International Protocol . However , it w a s desi rable to see if a s tudy in all age groups would a lso meet these strict requ i rements . Due to t ime, and f inancial constra ints it w a s more appropr iate to comb ine the s tudy data if reasonable to do so. It w a s felt that because w e were combin ing raw individual data rather than m e a n s and imputed s tandard deviat ions that this wou ld be appropr iate to do. In reality w e w e r e do ing m u c h the s a m e as if w e were combin ing data for a meta-analys is ; however w e were better off in that w e had original data do all analyses f rom scra tch . 72 T h e B p T R U ™ is the only non- invasive au tomated osci l lometr ic blood pressure dev ice current ly avai lable that has been tested in a populat ion group aged f rom 3 years to the elderly; and that has satisf ied the cri teria set out by the Amer i can National Standard for Electronic or A u t o m a t e d S p h y g m o m a n o m e t e r s , A N S I / A A M I S P 1 0 - 1 9 9 2 [12]. Overal l it has also achieved a g rade 'A' accord ing to the British Hyper tension Society protocol for the evaluat ion of blood pressure measur ing dev ices-1993 [13]. 73 =3 h-Q. CO 0 T 3 c CD CO i CD £ 0 10 JD O o 0 0 E 0 L _ CO CO 0 0 -*—• co 0 \"co •c CO 0 X o o \"GO CO 0 c CD E C Q. CO J3 Q E CO Q. +1 -O co E 0 Q. E -o 0 .£? i i co Q X co E c ™ co X 0 F 0 °> c n X 1 P co co C O T— T— T— ^— CNI C N C M T f T f T f T f T f TT i r i i r i i r i T— C D o a> T - C M T— T— T f co L O T f T f T f •>- o un T f CO CO CN CM CM CM CNI CM 00 co CO 1^ CO C M T f o C M o T f C M C M i i C O o C M C M 00 o C N C N C N 00 L O C N L O C N o C O C O C D L O C D i n C O C D in\" C O C O i n i n C O co C M d I ai co CM\" LO CN CN i i n oo co 0 _c T3 c CO 0 CO CO 0 o o .o o o \"oo co o o \"to CO co E +1 c C O T f C O co O T — co C N C O T f T f o i ar> C D csi C O L O C M C M C N C M CZ CO CO X co CD 0 E E E T f T f T f C D C N 1^ T f C N 00 T f T f d T f U o to co 0 CO 1 -o 0 c o i_ 0 0 to o 0 c-0 CO O c co 0 0 0 CO .a o C M c 0 E 0 0 L— CO co .a O to CO c T3 CO 0 3 OH \\— CL GO 0 §| CO X 3 C ^ 1-2 E 03 CD CO T 3 O O -«—» CO ro o ro § I E E • D C ^ c | £ iS 0 E CO T 3 ^ O o -*—' CO > » CO 0 3 ro CD X Q) i l C X I ^ CO in o o oo iri iri V C N O o ^~ o in C N C N +i i i o o c o c d in V o i r i +1 C N C 3 0 3 0. CO 3 co o 1^ -3-o C N ro X I c ro \"to 03 o c 0 CD H— CD L co c o o 3 DC I— C L CQ ro x> c ro CD o c CD CD i CO c DC H C L CQ c ro E o 0 !£? CO r-CO X E E L O VI 00 co o co L O C D O cd CO C D C D o C D Al X E E o 8^ T - CO co L O cb C M 00 csi 00 C D C D C D C D VI o o o o a. co X L O VI C M C M O C D L O co co oo 1^ Al 0 T J CD CO X m •3\" -3\" C N C M CD £ CD 'o O CO c o co c 0 tr 0 CL >, X co 0 0 CD CL O O 3 Cd \\— CL CQ O 0 O CZ CD o t : 0 0 . oo 0 CD CD X ) cz CD CO CO X CQ 0 T J CD l_ o 3 CO CO 0 T J o o - Q C J o 00 0 o cz 0 l _ 0 M — 0 1 CO c £ s Qi I — CL CQ 3 CO CO 0 T J o o O o -•—' CO CD T J 0 O CZ 0 £ 0 1 CO c 3 DC r -Q. CQ 3 00 00 0 I CL T J O O J3 O 0 O CZ 0 0 c: CD 0 E o 0 !c? CO CO 3 c E t-Q£ H CL CQ cz CD 0 E o 0 Ic? CO 3 CO 00 0 l_ CL T J O O J D O 03 03 O CO CD T J 0 O CZ 0 0 c CD 0 o 0 Ic? 3 CO CO 3 CZ E s I-3 DH h-CL CQ c: CD 0 E o 0 Ic? 3 CO Fig. 1 C « a? S X ^ E £ E ° D) <1> > = 2 £ O Bland-Altman Display of Systolic Measurements, All Cuffs, 417 measurements M e a n D i f f e r e n c e : 0 . 4 7 m m H g S D o f D i f f e r e n c e s : 5 . 4 0 m m H g 30 20 10 0 -10 -20 -30 4 j ^ i i H i 4 a • • * • • ^ • • — * • * — * z — *w.y -is .—-50 75 100 125 150 175 200 Mean of BPM and Obs Avg (mmHg) +2SD Mean -2SD 225 Bland A l tman display of all systol ic blood pressure m e a s u r e m e n t s (n = 417 ) . BPM, B p T R U ™ (model BPM-100) moni tor ; Obs A v g , observer average ( re ference s tandard) 77 Fig. 2 Bland Altman Display of Diastolic Measurements, All Cuffs, 417 measurements Mean Difference: -2.12 mmHg SD of Differences: 5.93 mmHg tf) A O TJ C « o5 2 x ° u> 0) > J2 < a 3 0 2 0 1 0 4 0 - 1 0 - 2 0 - 3 0 3 0 * 5 * — * • 1 : 4 5 6 0 7 5 9 0 Mean of BPM and Obs Avg (mmHg) 1 0 5 Bland A l tman display of all diastol ic blood pressure m e a s u r e m e n t s (n = 417 ) . B P M , B p T R U ™ (model BPM-100) monitor; Obs Avg , observer ave rage ( re ference s tandard) 78 References 1. O h k u b o T, Imai Y, Tsuj i I, Nagai K, Kato J , Kikuchi N, N ish iyama A, A ihara A, Sekh ino M, Kikuya M, Ito S, Satoh H, Hisamichi S. H o m e blood pressure m e a s u r e m e n t has a s t ronger predict ive power for mortal i ty than does screening blood pressure: a populat ion-based observat ion in O h a t a m a , J a p a n . Journal of Hyper tens ion 1998; 16: 971-975. 2. Nat ional Cent re for Health Stat ist ics. A d v a n c e report of f inal mortal i ty stat ist ics, 1986. Monthly vital stat ist ics report. [Suppl ] , vol 37, no 6. Hyattsvi l le, Md. : Public Heal th Serv ice, 1988. (Publ icat ion no D H S S (PHS) 88-1120) . 3. Jof f res M R , Harnet P, Mac lean DR, L'italien G J , Fodor G. Distr ibut ion of Blood Pressure and Hypertension in C a n a d a and the United States. Amer i can Journal of Hyper tension 2 0 0 1 ; 14: 1099-1105. 4 . Ger in W , Mar ion R M , F r iedman R, J a m e s G D , Bovbjerg D H , P icker ing T G . H o w should w e measure blood pressure in the doctor 's of f ice? Blood Pressure Moni tor ing 2 0 0 1 ; 6: 257-262. 5. Picker ing T G , J a m e s G D , Boddie C, Harshf ield GA, Blank S, Laragh J H . How c o m m o n is whi te coat hyper tension? T h e Journal of the Amer i can Medical Associat ion 1988; 259 : 225-228 . 6. Markandu N D , Whi tcher F, Arnold A, Carney C. T h e mercury s p h y g m o m a n o m e t e r should be abandoned before it is proscr ibed. Journal H u m a n Hyper tension 2000 ; 14: 31-36. 7. O'Br ien E. Replac ing the mercu ry s p h y g m o m a n o m e t e r . Brit ish Medical Journal 2000 ; 320: 815-816. 8. Little P, Barnet t J , Barnsley L, Mar joram J , F i tzgerald-Barron A, Mant D. Compar i son of ag reemen t between dif ferent measures of b lood pressure in pr imary care and dayt ime ambula tory blood pressure. Brit ish Medical Journal 2002 ; 325: 254. 9. Mattu G S , Perry T L Jr, Wr igh t J M . Compar i son of the osci l lometr ic blood pressure moni tor ( B P M - 1 0 0 B e t a ) with the auscul tatory mercury s p h y g m o m a n o m e t e r . Blood Pressure Moni tor ing 2 0 0 1 ; 6: 153-159. 10. Wr igh t J M , Mat tu G S , Perry T L Jr, Gel fer M, St range K D , Z o r n A, C h e n Y. Val idat ion of a new algor i thm for the BPM-100 electronic osci l lometr ic off ice blood pressure moni tor . Blood Pressure Moni tor ing 2 0 0 1 ; 6: 161-165. 1 1 . Mat tu GS, Heran BS, Wr igh t J M . Compar i son of the au tomated non- invasive osci l lometr ic blood pressure moni tor ( B p T R U ™ ) with the auscul tatory mercury s p h y g m o m a n o m e t e r in the paediatr ic populat ion. Ready for submiss ion 2003. 12. Amer i can Nat ional Standard for Electronic or Au tomated S p h y g m o m a n o m e t e r s . A N S I / A A M I SP10-1992 . Ar l ington, VA: A A M I , 1993. 13. O'Br ien E, Petr ie J , Littler W , de Swiet M, Padfield PI, A l tman D G , et a l . T h e Brit ish Hyper tension Society protocol for the evaluat ion of b lood pressure measur ing devices. Journal of Hyper tens ion 1993; 11 (suppl 2) : S43-S62 . 79 14. Bland J M , A l tman D G . Statist ical me thods for assess ing agreement between two me thods of cl inical measuremen t . Lancet 1986; 1 :307-310 . 15. Campbe l l N R C , M c K a y D W . Accura te blood pressure measurement : W h y does it mat ter? Canad ian Medical Assoc ia t ion Journal 1999; 1 6 1 : 277-278 . 16. McAl ister FA, St raus SE. M e a s u r e m e n t of blood pressure: an ev idence based review. Brit ish Medical Journal 2 0 0 1 ; 322: 9 0 8 - 9 1 1 . 17. Abbot t D, Campbe l l N R C , Carru thers-Czyzewsk i P. Guidel ines for measu remen t of blood pressure, fo l low-up and l ifestyle counsel l ing. Canad ian Journa l o f Publ ic Heal th 1994; 8 5 : s29-s43. 18. Beevers G, Lip G Y H , O'Br ien E. Blood pressure measuremen t : Part 1 - S p h y g m o m a n o m e t r y : fac tors c o m m o n to all techn iques. Brit ish medica l Journal 2 0 0 1 ; 322: 981-985. 19. O'Br ien E. State of the marke t in 2001 for blood pressure measur ing dev ices . Blood Pressure Moni tor ing 2 0 0 1 ; 6: 171-176. f 20. O'Br ien E, Picker ing T, A s m a r R, Myers M, Parati G, S taessen J , M e n g d e n T, Yu taka I, W a e b e r B, Palatini P, and wi th statistical ass is tance of Atk ins N, Ger in W . W o r k i n g Group on Blood Pressure Moni tor ing of the European Society of Hypertension International Protocol for val idat ion of blood pressure measur ing dev ices in adul ts . Blood Pressure Moni tor ing 2002 ; 7: 3-17. 80 CHAPTER VI OVERALL DISCUSSION In North Amer i ca the most c o m m o n reason for pat ients to visit their physic ian 's of f ices is hypertension, and one of the c o m m o n e s t d iagnost ic procedures to be per fo rmed is the m e a s u r e m e n t of blood pressure [14]. T h e screening, d iagnosis and ongoing m a n a g e m e n t of hyper tens ion depend extensively on an accurate and s tandard ised method of blood pressure measuremen t . In the l i terature there are numerous examp les of the potential diff icult ies with inaccurate readings. T h e c o n s e q u e n c e s of inaccuracy have been es t imated: the consis tent overest imat ion of diastol ic blood pressure by 5 m m H g wou ld increase the d iagnosis of hyper tension in a physic ians pract ice by 100% and converse ly underest imat ing the diastol ic blood pressure by 5 m m Hg wou ld reduce by 62% the number of pat ients d iagnosed to be hypertensive. In other words the potential for large numbers of fa lse posit ive and fa lse negat ive blood pressure m e a s u r e m e n t s is signif icant, with possible adverse consequences for the patient [15]. T h e comple t ion of these two trials has demons t ra ted the diff iculty in enrol l ing the n u m b e r of subjects required and having them meet the required parameters such as b lood p ressure range, a rm size and age range. However , after complet ing these requi rements and obtaining the resul ts seen here, it is possible to gain a fair amoun t of conf idence in the accuracy and reproducibi l i ty of the dev ice under quest ion. Th is conf idence takes cogn izance of the prob lems associated wi th the gold s tandard , the auscul tatory mercury s p h y g m o m a n o m e t e r . Th is is partly ref lected in the somewha t surpr is ing in ter-observer variabil i ty seen in these tr ials. If t ra ined research nurses under research s tandards measur ing blood pressure s imul taneously get var iable results, one can imagine the m u c h larger variabi l i ty of blood pressures measured in a busy pr imary care off ice. T h e variabil i ty in that usual sett ing is probably so large that most d iagnost ic and t rea tment dec is ions cannot be m a d e with any degree of conf idence. S o m e ex amp les of the m a n y dif ferent c i rcumstances that can affect blood pressure are: • talk ing can increase the systolic blood pressure by up to 17 m m H g • having pat ients arm 10 cm below the heart level can increase the blood pressure by 8 m m H g • measur ing blood pressure with a cuff that is too smal l can increase the systol ic blood pressure by up to 8 m m H g . All potential errors have to be recognised and steps put into p lace to avoid these errors [16]. An au tomated blood pressure measur ing device can be an important s tep towards avoid ing s o m e of these errors. 82 Modern blood pressure-measur ing dev ices have to be accurate, reproducib le, and easy to operate, convenient , t ime saving and af fordable. Many avai lable dev ices however , do not have the most basic of mer i ts such as accuracy and reproducibi l i ty demons t ra ted in a publ ished report . Dev ices that have publ ished val idat ion data , and fulfill s o m e of the other criteria speci f ied have signi f icant advantage over those that do not [17]. T h e B p T R U ™ is an electronic au tomated blood pressure dev ice that has been shown to be of similar accuracy and reproducibi l i ty to the current gold s tandard of auscul ta tory mercu ry s p h y g m o m a n o m e t r y in a w ide range of pat ients. In the original s tudy in adul ts it w a s seen that the earl ier vers ion of the device the B P M - 1 0 0 B e t a w a s s h o w n to underest imate systol ic blood pressures over 150 m m H g . Th is only became evident w h e n the data w a s plotted on a B land-Al tman plot. Th is vers ion of the dev ice met the s tandards of both regulatory bodies and could have been marke ted as such . However , both the invest igators and the manufac tu rers found this l imitation unacceptab le . As a result a random s a m p l e of the data was used to create a modi f ied a lgor i thm, wh ich corrected the prob lem. T h e rest of the data w a s then avai lable to test the modi f ied a lgor i thm and it was clear using the B land-Al tman plot that it had corrected the prob lem. T h e methods descr ibed could set a precedent al lowing manufac tu rers to re-val idate al tered a lgor i thms using data a l ready col lected by their e m b e d d e d sof tware. Th is could save t ime and resources and al low test ing of m o r e dev ices especial ly those that have modi f ied a lgor i thms. I bel ieve the present s tandards for val idat ion of blood pressure measur ing dev ices are not suff iciently str ingent and should be revised in the fol lowing ways . • T h e regulat ions should require val idat ion using s imul taneous m e a s u r e m e n t s as w a s done in our protocol . T h e new International protocol and previous s tandards al low or r e c o m m e n d the use of sequent ia l b lood pressure m e a s u r e m e n t s rather than s imul taneous m e a s u r e m e n t s . T h e sequent ia l m e a s u r e m e n t s are recorded at least 30 seconds apart , but not m o r e than 60 seconds apart . Th is has the signif icant d isadvantage of adding the variabil i ty of blood pressure over t ime and does not al low enough t ime between measu remen ts for resolut ion of venous conges t ion , wh ich can also affect the m e a s u r e m e n t . 83 • T h e dev ice blood pressure is compared to pressures measured by re ference s tandard observers both before and after. T h e di f ference between the observer and the dev ice is de te rmined and the smal ler of the two values used . This invariably creates a bias to m a k e the dev ice look m o r e accurate than it is [18]. • T h e regulat ions should a lso require that a device meets both the A A M I s tandard and the BHS s tandard . T h e A A M I s tandard appears to be the more robust of the two, however the requ i rement for a s tandard deviat ion of the d i f ference of as m u c h as 8 m m H g is relat ively lenient. • T h e requ i rement for a Bland A l tman plot is essent ial as demons t ra ted by the f i rst tr ial. A device could underes t imate blood pressure for high va lues and overest imate for low va lues and pass the standard in te rms of a m e a n di f ference f rom the mercury measuremen ts . S tandards need to be establ ished as to wha t deviat ions seen on the Bland A l tman plot would be acceptab le . [20] . It w a s clearly seen in the first adult paper that the mach ine w a s under-est imat ing higher systol ic blood pressures wh ich was recti f ied in the reval idat ion paper wi th the modi f ied a lgor i thm. In addit ion the new International protocol int roduces addit ional p rob lems. It requires that if the inter-observer d i f ference for systolic or diastol ic pressures is not within 4 m m H g ag reemen t it must be repeated. Th is leads to lesser inter-observer variabil i ty and could lead to b ias. If a Sphygmocorder is used to record the sounds and the observers do not agree, they are requi red to re -assess until ag reement is reached . Aga in this appears to int roduce s o m e bias. Since the B p T R U ™ has met the more str ingent requ i rements set out by the A m e r i c a n National Standard for Electronic or Au tomated S p h y g m o m a n o m e t e r s and the Brit ish Hyper tens ion Society, it can be a s s u m e d that it wou ld a lso meet the more recent and less r igorous Internat ional Protocol for val idat ion of blood pressure measur ing devices in adults as set out by the W o r k i n g Group on Blood Pressure Moni tor ing of the European Society of Hyper tension [18]. O n e of the a ims of this International protocol was to encourage m o r e device manufac turers to undergo val idat ion before marke t ing . I 84 If cl inical dec is ions are going to m a d e based on the best avai lable ev idence then blood pressures need to be measured in a manner that is similar to that wh ich was used in the tr ials. It is c lear that single blood pressures measured by a physician in the off ice represent a signif icant overes t imate of the rest ing blood pressure m e a s u r e m e n t s by a nurse in a clinical tr ial. A n y new blood p ressure measur ing methodology must be shown to be a better es t imate of the research s tandard than casua l of f ice measu remen ts by a physic ian. My recommenda t i on to the manufac turer is that they fund a trial compar ing B p T R U ™ with the research s tandard and casual off ice measu remen ts by a physic ian. T h e best des ign for this trial would be to assess 50 pat ients in a pr imary care clinic set t ing. T h e best des ign wou ld be a cross-over des ign where all m e a s u r e m e n t s are taken on the s a m e day, but the order of m e a s u r e m e n t s is randomized . In the clinic sett ing w h e r e in one room the B p T R U ™ mach ine is set up, the pat ient fo l lows the s tandard routine of rest ing for 5 minutes; the first pressure is taken wi th the invest igator in the room and is d iscarded, the blood pressure recorded is the m e a n of the remain ing 5 m e a s u r e m e n t s wi th nobody in the room. T h e research s tandard wou ld be rest ing 5 minutes fo l lowed by 3 m e a s u r e m e n t s taken by a nurse in the correct s tandard ised way. T h e third wou ld be one or two m e a s u r e m e n t s by a physic ian as part of his or her busy pract ice day. T h e chal lenge in the trial wou ld be to not a l low the research des ign to affect how the pressures are taken by the physic ian. T h e trial should answer the quest ion as to whether physician m e a s u r e m e n t s or B p T R U ™ measuremen ts most closely mimics b lood pressures as taken in the clinical trial sett ing (ie: those m e a s u r e m e n t s recorded by the nurse reading wi th research s tandards) . W i th the w a v e of newer dev ices coming to the market and the end of an era with mercury s p h y g m o m a n o m e t e r s gett ing closer, w e can be conf ident that the B p T R U ™ is one of those devices that has at tained the accuracy that is required of it by the current govern ing bodies on blood pressure measur ing ins t ruments . T h e B p T R U ™ can be used wi th conf idence and unders tand ing that patients in the physic ian's off ice will have appropr ia te decis ions based on accurate , reproducib le and complete ly object ive blood pressure m e a s u r e m e n t s . It can thus not only reduce inappropr ia te prescr ib ing to those with 'white coat hyper tens ion ' and consequent ly reduce iatrogenic d isease e lements , but also appropr iate ly identify that 3 0 % of the populat ion unaware of their hyper tens ion and consequent ly monitor 85 their b lood pressure to help avoid or reduce the huge n u m b e r of dea ths direct ly related to being uncontrol led [19]. Hencefor th w h e n the B p T R U ™ w a s tested in chi ldren (aged 3 to 18 years wi th a m e a n of 9.4 years) it also met the requ i rements of the A N S I / A A M I SP10-1992 and ach ieved a 'B ' g rade accord ing to a modi f ied BHS protocol . It is one of the f ew dev ices to have been val idated in chi ldren and to have ach ieved signif icant levels of accuracy. W h e n the raw paediatr ic data w a s c o m b i n e d wi th the raw adult data , the B p T R U ™ met the A N S I / A A M I SP10-1992 criteria and ach ieved an overal l g rade 'A' f rom the Brit ish Hyper tens ion Society protocol . 86 CHAPTER VII CONCLUSIONS f 87 Regard ing the B p T R U ™ : • T h e B p T R U ™ , an electronic non- invasive au toma ted osci l lometr ic b lood p ressure dev ice has been shown to be as accurate as the current gold s tandard of auscul ta tory mercu ry s p h y g m o m a n o m e t r y . • W h e n tested in adults (ages 18 to elderly - oldest included subject w a s 83 years o ld, m e a n age of 43.1 years) it met all requ i rements of the Associat ion for the A d v a n c e m e n t of Medical Instrumentat ion and ach ieved an \"A\" g rade accord ing to the Brit ish Hyper tens ion Society s tandard . • W h e n tested in chi ldren (aged 3 to 18 years with a m e a n 9.4 years) it a lso met the requ i rements of the A N S I / A A M I S P 1 0 - 1 9 9 2 and achieved a 'B ' g rade accord ing to a modi f ied B H S protocol . • W h e n the raw data f rom both populat ion groups was c o m b i n e d to de te rmine the overal l accuracy of this blood pressure device over an age range of 3 years to 83 years ( m e a n age 33.1 years) it sat isf ied the cri teria set out by the Amer i can Nat ional S tandard for Electronic or Au tomated S p h y g m o m a n o m e t e r s , A N S I / A A M I SP10-1992 . • Overal l it has also ach ieved a g rade 'A' accord ing to the Brit ish Hyper tens ion Society protocol for the evaluat ion of blood pressure measur ing dev ices-1993. • T h e B p T R U ™ is the only au tomated blood pressure measur ing dev ice that has been val idated and has at tained this high level of accuracy over this broad age range. 88 Regard ing the val idat ion process, my opinion is such that: • All b lood pressure devices i r respect ive of the target aud ience ( h o m e or of f ice use) should undergo val idat ion. • Dev ice manufac tu rers should be more t ransparent about their a lgor i thms. • T h e methods and requ i rements for cal ibrat ion should be clearly s ta ted. • In order to val idate an au tomated blood pressure measur ing dev ice, s imu l taneous measu remen ts f rom the s a m e a rm should be directly compared to the re ference s tandard . • T h e m e a n of two observer 's m e a s u r e m e n t s should be regarded as the gold or re ference s tandard . • Dev ices should be m a d e to clearly demonst ra te the accuracy across a full range of blood pressures and cuff s izes using the Bland A l tman plot. • All dev ices should meet the requ i rements set for th by both the A N S I / A A M I S P 1 0 - 1 9 9 2 and the Brit ish Hyper tens ion Society protocol . • Newer val idat ion guidel ines, such as the International Protocol that appear to encourage device manufac tu rers to under take the val idat ion process, but have less rigid cr i ter ia are unacceptab le . 89 REFERENCES 1. Wr igh t J M , Lee C, C h a m b e r s GK. Systemat ic rev iew of ant ihypertensive therapies: Does the ev idence assist in choos ing a f irst-l ine drug? Canad ian Medical Associat ion Journal 1999; 1 6 1 : 25-32 . 2. Haj jar I, Ko tchen TA. T rends in prevalence. Awareness , t reatment , and control of hyper tens ion in the United States, 1988-2000. T h e Journal of the Amer i can Medical Assoc ia t ion 2003; 290 : 199-206. 3. Markandu ND, Wh i t cher F, Arnold A, Carney C. T h e mercury s p h y g m o m a n o m e t e r should be abandoned before it is p roscr ibed. Journal H u m a n Hyper tension 2000 ; 14: 31-36. 4 . O'Br ien E. Replac ing the mercury s p h y g m o m a n o m e t e r . Brit ish Medical Journal 2000 ; 320: 815-816. 5. O'Br ien E. D e m i s e of the mercury s p h y g m o m a n o m e t e r and the dawn ing of a new era in blood pressure m e a s u r e m e n t . Blood pressure moni tor ing 2003; 8: 1 9 - 2 1 . 6. Little P, Barnet t J , Barnsley L, Mar joram J , F i tzgerald-Barron A, Mant D. Compar i son of ag reement between dif ferent measures of blood pressure in pr imary care and dayt ime ambula tory blood pressure. f» Brit ish Medical Journal 2002 ; 325: 254. 7. Amer i can Nat ional S tandard for Electronic or Au tomated S p h y g m o m a n o m e t e r s . A N S I / A A M I SP10-1992 . Ar l ington, VA: A A M I , 1993. 8. O'Br ien E, Petr ie J , Littler W , de Swiet M, Padfield PI, A l tman D G , et a l . T h e Brit ish Hyper tens ion Society protocol for the evaluat ion of blood pressure measur ing devices. Journal of Hyper tens ion 1993; 11 (suppl 2) : S43-S62 . 9. T h e Sixth Report on the Joint Nat ional Commi t tee on Prevent ion, Detect ion, Evaluat ion, and T rea tmen t of High Blood Pressure. NIH Publ icat ion No. 98-4080, November 1997. 10. Mattu G S , Perry T L Jr, Wr igh t J M . Compar i son of the osci l lometr ic blood pressure moni tor ( B P M - 1 0 0 B e t a ) wi th the auscul tatory mercury s p h y g m o m a n o m e t e r . Blood Pressure Moni tor ing 2 0 0 1 ; 6: 153-159. 1 1 . Wr igh t J M , Mattu GS, Perry T L Jr, Gelfer M, St range KD, Zorn A, C h e n Y. Val idat ion of a new algor i thm for the BPM-100 electronic osci l lometr ic off ice blood pressure moni tor . Blood Pressure Moni tor ing 2 0 0 1 ; 6: 161-165. 12. Mattu G S , Heran BS, Wr igh t J M . Compar i son of the au tomated non- invasive osci l lometr ic blood pressure moni tor ( B p T R U ™ ) with the auscul tatory mercury s p h y g m o m a n o m e t e r in the paediatr ic populat ion Ready for submiss ion 2003. 13. Mat tu G S , Heran BS, Wr igh t J M . Accuracy of the B p T R U ™ , an au tomated electronic blood pressure dev ice. Ready for submiss ion 2003. 90 14. W o o d w e l l DA. Nat ional ambu la to ry medical care survey: 1995 summary . A d v a n c e Data 1997; 295: 1-25. 15. Campbe l l N R C , McKay D W . Accura te blood pressure measurement : W h y does it mat ter? Canad ian Medical Assoc ia t ion Journal 1999; 1 6 1 : 277-278. 16. McAl is ter FA, St raus SE. M e a s u r e m e n t of blood pressure: an ev idence based review. Brit ish Medical Journal 2 0 0 1 ; 322: 9 0 8 - 9 1 1 . 17. O'Br ien E. State of the marke t in 2001 for blood pressure measur ing dev ices . Blood Pressure Moni tor ing 2 0 0 1 ; 6: 171-176. 18. O'Br ien E, Picker ing T, A s m a r R, Myers M, Parati G, S taessen J , M e n g d e n T, Yu taka I, W a e b e r B, Palatini P, and wi th statistical ass is tance of Atk ins N, Ger in W . W o r k i n g Group on Blood Pressure Moni tor ing of the European Society of Hypertension International Protocol for val idat ion of blood pressure measur ing dev ices in adul ts . Blood Pressure Moni tor ing 2002 ; 7: 3-17. 19. Jof f res MR, Hamet P, MacLean DR, L'italien G J , Fodor G. Distr ibut ion of blood pressure and hyper tension in C a n a d a and the Uni ted States. Amer i can Journal of Hyper tension 2 0 0 1 ; 14: 1099-1105. 20 . Bland J M , A l tman D G . Statist ical me thods for assess ing agreement between two m e t h o d s of cl inical measurement . Lancet 1986; 1: 307-310. 91 "@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "2003-11"@en ; edm:isShownAt "10.14288/1.0091081"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Pharmacology"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Validation of oscillometric blood pressure measuring devices : a case study of the BpTRUtm"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/14252"@en .