Results
The most common machines in pharmacies were Vita-Stat models 90550 (n=11), 8000 (n=9), and 9000 (n=3). Two stores had Health Clinic Cardio-Analysis devices. Pharmacy staff were unable to state when each machine was last calibrated or had received maintenance, indicating that responsibility for maintaining these machines fell solely on the manufacturer’s representative. Calls to the information numbers on the machines were met with nonspecific answers that the machines are calibrated regularly.
The Table 1 shows each subject’s average systolic and diastolic pressure as determined by the machine and the clinician and the variance between readings using the same method. For the subject with the small arm size, store machines reported systolic values that averaged 10 mm Hg higher than those from the mercury manometer (P <.001) and diastolic pressures that were 9 mm Hg higher (P <.001). The mean systolic pressures for the subject with the medium-sized arm were not significantly different between store and clinician methods and were only modestly different for diastolic readings. For the subject with the large arm size, store machines reported diastolic pressures that averaged 8.3 mm Hg lower than those obtained by the clinician P <.001). The difference in systolic readings between the 2 methods for this subject was not significant. Standard deviations of the store mercury blood pressure differences were greater than the 8.0 mm Hg allowable limit set by the AAMI for systolic readings for the subject with a medium arm and for the diastolic readings for the subject a large arm.7
To determine if there is more variability in readings from store machines than with those from a clinician, we also compared the variance of readings for each method . For the subject with medium arm size, there was much greater variance with store machines, compared with that from the mercury manometer, for both systolic (P=.001) and diastolic readings (P=.002). For the subject with the large arm size, there was significantly more variance in diastolic pressures obtained by store machines than with the clinician’s readings (P=.03).
Discussion
We found that automated blood pressure machines from a representative community-based sample of pharmacies did not meet accepted standards of accuracy and reliability of measurement. Depending on the subject’s arm size, automated blood pressure machines significantly underestimated or overestimated the subject’s blood pressure compared with the reading obtained by a clinician with a standard mercury manometer. For the subject with the medium arm size, the automated machines were generally more accurate but less reliable (higher variance). For the subject with the large arm size, automated machines tended to both underestimate diastolic pressure and show an unacceptably large variance, and for the subject with a small arm size, automated machines overestimated both systolic and diastolic blood pressure.
Although previous limited studies have found automated blood pressure machines less reliable than manometers, our study was unique in several ways.2-6 Unlike previous studies, the same 3 volunteers were used for all readings, rather than random passersby. Each subject also had a different arm size to assess any potential effect on readings. Also, our study used a random selection of the machines that are available and widely used in the community, rather than a few specific machines and models.
The magnitude of difference between the 2 methods of blood pressure measurement was substantial and clinically meaningful. Chronic underestimation of blood pressure will result in inadequate treatment, while overestimation can lead to unnecessary concern by the patient and overtreatment by the physician.
Limitations
Our study has a few limitations that should be considered. The reason for the variation of automated blood pressure machine accuracy with arm size is uncertain and deserves further study. During the study it was discovered that the subject with the large arm size was hypertensive, and the findings attributed to large arm size may be due to a difference in accuracy for subjects with and without hypertension. Previous studies of these machines showed no differences in accuracy between patients with and without hypertension, however.2,6 Although the clinician was blinded to the machine readings, there is a possibility of recall bias when the same clinician measured the same subject’s blood pressure repeatedly. We were not able to assess the degree to which automated machines were maintained or calibrated, so we are unable to comment on the role that machine maintenance may have on the results. Nonetheless, these are the machines used frequently by patients in the real world and not the best-case scenario submitted by the manufacturer or factory. We were unable to compare the relative accuracy of specific brands or models because of the size of our sample, and that may merit further study.