Accuracy of the Dinamap 1846 XT automated blood pressure ... - Nature

Journal of Human Hypertension (2002) 16, 647–652  2002 Nature Publishing Group All rights reserved 0950-9240/02 $25.00 www.nature.com/jhh

ORIGINAL ARTICLE

Accuracy of the Dinamap 1846 XT automated blood pressure monitor ER Beaubien1,3, CM Card2,3, SE Card3, HJ Biem3 and TW Wilson3,4 1

Division of Nephrology, McGill University, Montreal, Quebec, Canada; 2Division of Oncology, University of Toronto, Toronto, Ontario, Canada; 3 Division of General Internal Medicine, University of Saskatchewan, Saskatoon, Saskatchewan, Canada; 4Division of Clinical Pharmacology, University of Saskatchewan, Saskatoon, Saskatchewan

Accurate blood pressure (BP) measurement is important for the detection and treatment of hypertension. Despite widespread use of automated devices, there is limited published evidence for their reliability and accuracy. To determine the reliability and accuracy of the Dinamap 1846XT (Critikon Corporation, Tampa, FL, USA), a commonly used non-invasive oscillometric BP monitor The Dinamap was evaluated against the mercury manometer in 70 randomly selected adult hospitalised medical patients. Each individual underwent three sets of standardised BP measurement by automated method and three sets by mercury manometer by two independent observers. Reliability of BP measurement was assessed by repeated measures analysis. Dinamap accuracy was evaluated according to the American Association of Medical Instrumentation (AAMI) and British Hypertension Society (BHS) guidelines. Most patients were either normotensive or had stage I hypertension. The Dinamap tended to overesti-

mate lower diastolic BP, and displayed poor reliability (P ⬍ 0.05). Despite meeting AAMI guidelines, only 59% of systolic and 56% of diastolic Dinamap readings were within 5 mm Hg of the mercury manometer and 84% of systolic and 80% of diastolic readings were within 10 mm Hg (BHS grade C). Systolic and diastolic accuracy were worse with pressures ⬎160/90 mm Hg (grade D) although these measures were based on a smaller sample of subjects. In conclusion the Dinamap yields inaccurate estimates of both systolic and diastolic BP even under standardised, and thus optimal conditions. This inaccuracy is exaggerated at higher BP (⬎160/90 mm Hg), although the number of measurements at higher pressures was small. We recommend that this device not be used when accurate BP measurement is needed for therapeutic decision-making. Journal of Human Hypertension (2002) 16, 647–652. doi:10.1038/sj.jhh.1001463

Keywords: Dinamap; blood pressure; oscillometric monitor; accuracy

Introduction Hypertension is one of the leading causes of cardiovascular morbidity and mortality in Western society.1–4 It is associated with increased risk of stroke, myocardial infarction, congestive heart failure, peripheral vascular disease and renal dysfunction. The benefits of antihypertensive therapy in patients with sustained hypertension (defined as ⬎140/90 mm Hg) have been demonstrated in a number of large randomised controlled trials.4,5 The HOT study5 showed benefits is lowering diastolic blood pressure (BP) down to 82.6 mmHg, and in those patients with diabetes mellitus there was a marked reduction (51%) in major cardiovascular events in Correspondence: ER Beaubien, Department of Nephrology, Suite #202, 849 Alexander Court, Peterborough, Ontario, Canada K9J 7H8. E-mail: eliot beaubien얀hotmail.com Received 20 December 2001; revised 20 December 2001 and 4 June 2002; accepted 21 June 2002

those with a target ⱕ80 mm Hg compared with those with a target ⱕ90 mm Hg.5 Accurate BP measurement is therefore essential for hypertension detection and follow-up, as a change of 5 mm Hg in either systolic or diastolic pressure can alter decisions regarding hypertensive therapy. Several automated devices are currently used for measuring BP non-invasively. The most common of these rely on oscillometric testing, in which a pizoelectric crystal measures arterial waveforms distal to a deflating cuff to obtain systolic and diastolic pressures. The cuff is inflated to higher than systolic pressure then deflated at a predetermined rate (usually 2– 4 mm Hg/sec). Systolic BP is taken at the appearance of a pulsatile signal, while diastolic BP is calculated using a proprietary formula.6 Although used frequently in hospital wards, emergency departments, operating theatres, intensive care facilities and outpatient settings, many automated devices have not undergone rigorous testing for accuracy or reliability. Both the American Associ-

Dinamap 1846 XT ER Beaubien et al

648

ation of Medical Instrumentation (AAMI) and British Hypertension Society (BHS) have put forth standards for validating semi-automated and automated BP devices7,8 (Table 1). Many devices have been found to be inaccurate on formal testing.9–14 The AAMI suggests that, for systolic and diastolic BP each considered separately, the mean difference between test devices (ie, Dinamp compared with mercury sphygmomanometer) be ±5 mm Hg or less, with a standard deviation of 8 mm Hg or less. The BHS describes a different method of evaluating automated BP devices,15,16 that uses a grading scheme based on repeated measures analysis. Only those devices that achieve a grade A or B for both systolic and diastolic measures are recommended for clinical use. Studies of other automated BP devices have raised concerns about accuracy. In 1993, O’Brien et al9 evaluated the Dinamap 8100 in 86 patients with a wide range of BP. The device received a BHS B grading for systolic BP, but only a D grading for diastolic BP (AAMI criteria were met only for systolic pressures). Accuracy deteriorated at higher pressures (⬎160/90 mm Hg). The authors therefore recommended that the Dinamap 8100 not be used for clinical circumstances where accuracy is desired. In 1996, Wattigney et al10 evaluated the Dinamap 845 XT and Dinamap 8100 in 417 children as part of the Bogalusa Heart Study. They found that both oscillometric devices, and especially the 8100 model, underestimated diastolic BP. Accuracy was not described according to the BHS grading scheme. In 1998, Shuler et al11 evaluated the IVAC 4200 model under standardised conditions and as routinely used by ward staff. Under standardised conditions, only 59% of systolic and 54% of diastolic readings were within 5 mm Hg of a mercury manometer standard (BHS grade C rating). When obtained by routine ward personnel only 40% of systolic and 50% of diastolic readings were within 5 mm Hg (BHS grades D and C respectively). The authors concluded that the IVAC 4200 yielded substandard estimates of systolic and diastolic BP even under standardised, thus optimal, conditions. The current reference standard for non-invasive BP measurement is the mercury manometer. This device has been used extensively in therapeutic trials and health prevention studies, and has served Table 1 British Hypertension Society blood pressure grading system (based on cumulative percentage of readings) Grade

A B C D

Difference between standard and test device (mm Hg) ⬍5 ⬍10 ⬍15 80 65 45 *

90 85 75 *

* Asterisk indicates worse than grade C. Journal of Human Hypertension

95 95 90 *

in previous studies of other blood pressure monitors as the reference standard.17,18 We therefore decided to evaluate the accuracy and reliability of the Dinamap 1846 XT (Critikon Corporation, Tampa, FL, USA), a non-invasive oscillometric BP monitor common in our health district, using the mercury manometer reference standard. Measurements were obtained under standardised conditions, in medically stable hospital patients, by staff trained in BP measurement.17,19

Methods Subjects Seventy-two patients admitted to a general medicine ward at Royal University Hospital were randomly selected for BP measurement, through blocked sampling of patient admission lists over a 16-week period. Patients were provided a brief explanation of the study, and gave verbal consent. Patients were excluded if they: (1) were clinically unstable, or (2) had an arm circumference outside of the available cuff sizes (14 –45 cm). Of the 72 patients approached, two were unable to complete all BP measurements because of cuff discomfort. Patient height, weight, arm circumference measured at the mid-humerus, pulse rate and regularity were recorded prior to BP determination. Admission ECGs were evaluated for arrhythmias. Protocol Six systolic/diastolic BP measurements were recorded for each patient, alternating between the Dinamap and the mercury manometer at 90-sec intervals. The initial device alternated consecutively with each patient. Therefore, a total of 210 measurements were obtained for each device. Manual blood pressure measurements Manual readings were simultaneously measured by two observers (EB, CC) trained in BP measurement. A standard portable mercury manometer and dual headed stethoscope were utilised. Results were recorded independently, in a blinded manner. Equipment was examined for defects in the outlet valve, tubing, bladder and cuff, as well as mercury column height. For all BP measurements, patients were supine with legs uncrossed, and resting for 5 min prior to examination. They could not eat, drink or smoke during this period or during BP measurement. Measurements were taken from the right arm (unless a medical condition precluded this — ie, A-V fistula, venous thrombosis, etc) with the relaxed arm slightly flexed and placed at heart level. Cuff size was selected according to arm circumference (small ⬍22 cm, medium 22–33 cm, large ⬎33 cm). Systolic BP was recorded at the appearance of two consecutive beats (Korotkoff phase I)


and diastolic BP with disappearance of all auscultatory sounds (Korotkoff phase 5). Korotkoff phase 4 was recorded if phase 5 was ⬍30 mm Hg. Automated blood pressure measurements Two separate Dinamap monitors were used throughout the study. Both were calibrated prior to the study by hospital technical staff and examined for defects in the cuff, bladder and tubing. The measurement protocol was similar to that used in manual BP measurement. Statistical analysis SPSS 8.0 (SPSS Inc. Chicago, IL, USA) was used for data analysis. Inter-observer reliability of paired manometer readings was evaluated using frequency distribution of difference20 for all systolic and diastolic measures as described by the American National Standards Institute.7,8 Device reliability was assessed using repeated measures analysis of variance on the three sequentially paired mercury manometer and Dinamap measurements to assess absolute change over time. A graph of Dinamap error (difference between Dinamap and mercury manometer) vs an averaged Dinamap and mercury manometer pressure was created for all systolic and diastolic readings. The Dinamap reliability was evaluated using repeated measures analysis for all systolic and diastolic BP as well as mean and standard deviation of difference between each Dinamap reading and the mercury manometer reading. For the determination of Dinimap accuracy, true BP was assumed to be an average of both observers’ mercury manometric readings. Pre-specified subgroup analyses were also performed for patients with high BP (either systolic ⬎160 mm Hg or diastolic ⬎90 mm Hg), low (systolic ⬍120 mm Hg and diastolic ⬍70 mm Hg), or intermediate BP. Results were classified according to the BHS standards.15,16 The impact of patient size (arm circumference, weight, body mass index), heart rate, arrhythmia, and cuff size on device accuracy was assessed using analysis of variance, while independent sample t-test compared Dinamap accuracy within groups.

Results Patient characteristics A total of 70 patients were evaluated. Group characteristics are outlined in Table 2. Mean age was 62 years (19–90, s.d. 18.5) and weight 75.3 kg (41–125, s.d. 19.2). Fifty-four per cent were males, with the majority of patients being Caucasian (86% white, 10% native, 4% Asian). A history of hypertension was present in 49% of patients, coronary artery disease in 36%, diabetes in 24%, renal impairment in

Table 2 Clinical characteristics of study group

649

Characteristics Mean ± s.d. age (year) 61.7 ± 18.5 (range) (19–90) Sex, Male/Female (%) 38/32 (54%/46%) Race, No (%) White 60 (85.7) Nativea 7 (10) Asian 3 (4.3) Diagnosis (%) (Admitting ± co-morbid conditions) Hypertension 48.6 Diabetes 24.3 CAD 35.7 Renal Failure 18.6 Hypercholesterolaemia 5.7 Smoker 22.9 CVA 21.4 CHF 22.9 Arrythmia 7.1 Mean ± s.d. Weight (kg) 75.3 ± 19.2 BMI 26.2 ± 5.9 Systolic BP (mm Hg) 131 ± 26.2 (range) (82–251) Diastolic BP (mm Hg) 74.5 ± 11.7 (range) (45–107) Arm circumference, cm 27.3 ± 3.7 (range) (20–36) a Native refers to the aboriginal people of Canada. BP = blood pressure.

19%, and prior cerebral vascular accident in 21% as determined by chart review. Mean systolic BP (measured by mercury manometer) in the study group was 131 mm Hg (s.d. 26.2, range 82–251 mm Hg) and diastolic pressure 75 mm Hg (s.d. 11.7, range 45–107 mm Hg). Reliability of reference standard For the mercury manometer paired readings, 98% of systolic and 97% of diastolic were within 5 mm Hg of each other, while 100% of systolic and diastolic readings were within 10 mm Hg. Dinamap reliability Repeated measures analysis between values showed greater variance among Dinamap readings. The percentage of readings differing by more than 5 mm Hg between first and third systolic readings was 54% (95% confidence interval (CI), 42–66%) for the Dinamap compared to 28% (95% CI, 18–39%) for the mercury manometer (P ⬍ 0.05) while the percentage of readings differing by more than 10 mm Hg was 22% (CI, 13–32%) for the Dinamap and 10% (95% CI, 2.7–17%) for the mercury manometer (nonsignificant). The percentage of diastolic readings differing by more than 5 mm Hg between first and third readings was 34% (95% CI, 23– 45%) for the Dinamap compared to 132% (95% CI, 5.1–20%) for the mercury manometer (P ⬍ 0.05), while the percentJournal of Human Hypertension


650

Figure 1 Syotolic measures.

age of readings differing by more than 10 mm Hg was 26% (95% CI, 16–36%) for the Dinamap and 4.3% (95% CI, −0.4 to 9%) for the mercury manometer (P ⬍ 0.05). There was no evidence of a systematic increase or decrease in readings over time. Dinamap accuracy To assess accuracy of the Dinamap, each oscillometric systolic and diastolic reading was compared with the corresponding averaged mercury manometer reading. The difference (Figures 1 and 2) between the Dinamap and averaged mercury manometer was then plotted against the mean of the two devices (ie, [Dinamap + mercury manometer] / 2). Across the measured blood pressures there was a slight upward trend for systolic values at higher blood pressures whereas diastolic values did not seem to vary. AAMI criteria were achieved for both systolic and diastolic values by the Dinamap (systolic: mean difference = 0 mm Hg, s.d. = 8 mm Hg, diastolic; mean difference = −3 mm Hg, s.d. 7.7 = mm Hg). However,

for BHS criteria only 59% of systolic readings values were within 5 mm Hg, 84% within 10 mm Hg and 93% within 15 mm Hg of the mercury sphygmomanometer, for a grade C rating (Table 3). For diastolic readings, 56% fell within 5 mm Hg, 80% within 10 mm Hg and 93% within 15 mm Hg. When accuracy was evaluated in low, intermediate and high BP subgroups the readings of both the low BP and the intermediate BP subgroups received grade C ratings, and the high BP subgroup readings merited only a D rating. The ratings did not change when we examined the first, second and third dinamap measurement separately. For systolic hypertension (⬎140 mm Hg), the Dinamap false-negative rate was 6% (6/139) and false-positive rate 13% (9/71). For diastolic hypertension (⬎90 mm Hg), the false-negative rate was 3% (6/195) and false-positive rate 53% (8/15). We used analysis of variance and independent sample t-tests to assess factors associated with Dinamap accuracy. Systolic error (absolute value) was associated with increased weight and body mass index [r = 0.168, ␤ = 0.05, P = 0.015 where ⌬ = 0 at weight = 35.4 kg and r = 0.124, ␤ = 0.157, P = 0.012, where ⌬ = 0 at BMI = 13.6 kg/m2]. However, these associations were weak. For instance a considerable increase of 10 kg/m2 in body mass index would equate to an increase in absolute systolic error of only 1.6 mm Hg (ie, 10 × ␤ (0.157) = 1.57 mm Hg). In patients with pulse irregularity, mean systolic difference (± s.e.) between Dinamap and mercury manometer was −3.1 (±1.3) mm Hg vs 0.28 (± 0.59) mm Hg (P = 0.03). Analysis of variance between other factors and systolic error demonstrated no significant associations. For diastolic readings, the Dinamap tended to have lower values at increased heart rate [r −0.237, ␤ = −0.125, P = 0.001, where ⌬ = 0 at heart rate = 50 bmp], arm circumference [r −0.175, ␤ = −0.365, P = 0.011, where ⌬ = 0 at arm Table 3 British Hypertension Society blood pressure grading chart Difference between Dinamap and mercury sphygmomanometer (mm Hg)

Figure 2 Diastolic measures. Journal of Human Hypertension

Reading

⬍5

⬍10

⬍15

All BP values (No.) Systolic (210) Diastolic (210)

59 56

84 80

93 93

BP ⬎160/90 mm Hg Systolic (27) Diastolic (15)

44 40

67 80

82 87

BP: 120/70–160/90 mm Hg Systolic (113) Diastolic (125)

58 55

86 82

96 94

BP ⬍120/70 mm Hg Systolic (74) Diastolic (75)

62 61

89 80

95 95

BP = blood pressure.


circumference = 18.6 cm], increased body mass index [r −0.163, ␤ = −0.214, P = 0.018, where ⌬ = 0 at 0 at BMI = 10 kg/m2] and younger age [r −0.164, ␤ = 0.068, P = 0.001, where ⌬ = −6 mm Hg at 20 years of age]. In patients with pulse irregularity mean diastolic difference (± s.e.) between Dinamap and mercury manometer was 3.2 (± 2.54) mm Hg vs −3.7 (±0.52) mm Hg (P = 0.01), and in patients with atherosclerotic heart disease −1.5 (±0.92) mm Hg vs −4.2 (±0.64) mm Hg (P = 0.015). There were no associations between diastolic error and other patient characteristics.

Discussion In summary, we have examined the reliability and accuracy of the Dinamap 1846 XT automated BP monitor, using standard non-invasive BP measurement by mercury sphygmomanometer. We found the Dinamap 1846 XT to yield substandard estimates of both systolic and diastolic BP (BHS grade C), despite achieving AAMI standards. This inaccuracy was worse at high BP and in patients with increased body weight and body mass index. We found 11 of 210 (4%) of systolic readings were 16 or more mm Hg higher than true readings, while five (2%) were 16 or more mm Hg lower (Figure 1). We evaluated the Dinamap under standardised conditions only. Testing under ‘real world’ conditions may well have shown the Dinamap to be even less accurate and reliable; automated BP monitors have been shown to have even less accuracy when used by general ward personnel.9 Under standardised conditions in this study, we used several readings over a short period of time, whereas repeated readings would be unusual in routine practice. We studied a range of medical patients, many of whom had hypertension, and/or complications of macrovascular disease, with broad ranges in both systolic and diastolic BP. We found that a sizeable number of diastolic measurements were misclassified as hypertensive, likely a result of the Dinamap overestimating diastolic pressures in the lower BP range. Several factors may explain the inaccuracy of some oscillometric monitors. Some devices have cuff deflation rates exceeding 2 mmHg/sec, which tends to underestimate blood pressures and different proprietary algorithms are used by manufacturers in calculating systolic and diastolic BP. There may be errors in cuff pressure measurement and oscillometry, which may not accurately reflect the Korotkoff sounds heard by manual method. A limitation of our study was the lack of a true gold standard such as intra-arterial BP measurement. It is known that cuff BP recordings tend to underestimate intra-arterial systolic pressure and overestimate diastolic pressure.21 We found no evidence of such a systematic error between the Dinamap and mercury devices. Although our sample size was similar to that rec-

ommended by the BHS, a larger sample size would be useful for subgroup analysis. However, we believe that a larger sample size would be unlikely to change our conclusions. In conclusion, we would recommend that the Dinamap 1846 XT not be used in clinical settings when accurate BP measurement is required. Future studies should assess the reliability and accuracy of automated BP devices in other settings such as outpatient clinics, intensive care units, and obstetrical wards. Our health district, like many others, is phasing out mercury manometers because of concerns about mercury toxicity. Moreover, nursing staff have turned to automatic systems to improve their efficiency. In our district, there is no scheduled maintenance for these devices and they are recalibrated only if a member of staff alerts the clinical engineering department. Therefore, it is likely that even greater measurement errors than we found occur on a regular basis. We recommend that all automated Dinamap readings be confirmed by a mercury manometer, or a recently calibrated aneroid instrument, for therapeutic decision making

651

References 1 Neaton JD, Wentworth D, MRFIT Research Group. Serum cholesterol, blood pressure, cigarette smoking, and death from coronary heart disease. Overall findings and differences by age for 316,099 white men. Arch Intern Med 1992; 152: 56–64. 2 Brophy JM. The epidemiology of acute myocardial infarction and ischemic heart disease in Canada: data from 1976 to 1991. Can J Cardiol 1997; 13: 474 –478. 3 Grundy SM et al. Guide to Primary Prevention of Cardiovascular Diseases. A statement for healtheare professionals from the Task Force on Risk Reduction. Circulation 1997; 95: 2329–2331. 4 Hansson L. The benefits of lowering elevated blood pressure: a critical review of studies of cardiovascular morbidity and mortality in hypertension. J Hypertens 1996; 14: 537–544. 5 Hansson L et al. Effects of intensive blood pressure lowering and low-dose aspirin in patients with hypertension: principal results of the Hypertension Optimal Treatment (HOT) randomized trial. Lancet 1998; 351: 1755–1762. 6 Wonka F. Oscillometric blood pressure measurement: description of the method used. Z Kardiol 1996; 85 (S3): 1–7. 7 White WB et al. National standard for measurement of resting and ambulatory blood pressures with automated sphygmomanometers. Hypertension 1993; 21: 504 –509. 8 AAMI. American National Standard for Electronic or Automated Sphygmomanometers. Arlington, VA, Association for the Advancement of Medical Instrumentation, 1987. 9 O’Brien E, Mee F, Atkins N, O’Malley K. Short Report: accuracy of the Dinamap portable monitor, model 8100 determined by the British Hypertension Society protocol. J Hypertens 1993; 11: 761–763. 10 Wattigney WA, Webber LS, Lawrence MD, Berenson GS. Utility of an automatic instrument for blood pressJournal of Human Hypertension


652

11 12

13

14 15

ure measurement in children. The Bogalusa Heart Study. Am J Hypertens 1996; 9: 256–262. Shuler CL et al. Accuracy of an automated blood pressure device in stable inpatients: optimum vs routine use. Arch Intern Med 1998; 158: 714 –721. Sun M, Tien J, Jones R, Ward R. A new approach to reproducibility assessment: clinical evaluation of SpaceLabs Medical oscillometric blood pressure monitor. Biomed Instrum Technol 1996; 30: 439– 448. Ling J et al. Clinical evaluation of the oscillometric blood pressure monitor in adults and children based on the 1992 AAMI SP-10 standards. J Clin Monit 1995; 11: 123–130. Fitzgeralds D et al. Accuracy and reliability of the Del Mar Avionics pressurometer III. J Hypertens Suppl 1985; S3: S359–S361. O’Brien E et al. The British Hypertension Society protocol for the evaluation of automated and semi-automated blood pressure measuring devices with special reference to ambulatory systems. J Hypertens 1990; 8: 607–619.

Journal of Human Hypertension

16 O’Brien E et al. An outline of the revised British Hypertension Society protocol for the evaluation of blood pressure measuring devices. J Hypertens 1993; 11: 677–679. 17 CM Grim, CE Grim. A curriculum for the training and certification of blood pressure measurement for health care providers. Can J Cardiol 1995; 11 (Suppl H): 38H– 42H. 18 Haynes RB et al. Report of the Canadian Hypertension Society Consensus Conference: 2. Diagnosis of hypertension in adults. Can Med Assoc J 1993; 149: 409– 418. 19 Curb JD et al. Training and certification of blood pressure observers. Hypertension 1983; 5: 610–614. 20 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 307–310. 21 Stolt M, Sjonell G, Astrom H, Hansson L. Factors affecting the validity of the standard blood pressure cuff. Clin Physiol 1993; 13: 611–620.