Blood pressure measurement is an essential physiological measurement for all critically ill patients. Previous work has shown that non-invasive blood pressure is not an accurate reflection of invasive blood pressure measurement. In a transport environment, the effects of motion and vibration may make non-invasive blood pressure less accurate. Consecutive critically ill patients transported by a dedicated aeromedical retrieval and critical care transfer service with simultaneous invasive and non-invasive blood pressure measurements were analysed. Two sets of measurements were recorded, first in a hospital environment before departure (pre-flight) and a second during aeromedical transport (in-flight). A total of 56 complete sets of data were analysed. Bland-Altman plots showed limits of agreement (precision) for pre-flight systolic blood pressure were −37.3 mmHg to 30.0 mmHg, and for pre-flight mean arterial pressure −20.5 mmHg to 25.0 mmHg. The limits of agreement for in-flight systolic blood pressure were −40.6 mmHg to 33.1 mmHg, while those for in-flight mean blood pressure in-flight were −23.6 mmHg to 24.6 mmHg. The bias for the four conditions ranged from 0.5 to −3.8 mmHg. There were no significant differences in values between pre-flight and in-flight blood pressure measurements for all categories of blood pressure measurement. Thus, our data show that non-invasive blood pressure is not a precise reflection of invasive intra-arterial blood pressure. Mean blood pressure measured non-invasively may be a better marker of invasive blood pressure than systolic blood pressure. Our data show no evidence of non-invasive blood pressures being less accurate in an aeromedical transport environment.
Blood pressure is a routinely measured physiological variable and forms part of the minimum requirement for monitoring patients during transfer in critical care [1, 2]. Values obtained for blood pressure are frequently used to make clinical decisions and are used in a variety of scoring systems. Therefore, the method of measurement and subsequent accuracy of blood pressure values is important. The gold standard for blood pressure measurement is by intra-arterial blood pressure (IABP) monitoring via a pressure transducer . However, insertion of an intra-arterial cannula is not desirable or necessary in all patients due to the risks associated with an intra-arterial catheter [4, 5]. Non-invasive blood pressure (NIBP) measurement using a limb cuff, usually around the upper arm, and an oscillometric technique is fast, repeatable and non-invasive. However, concerns exist as to the accuracy of the obtained measurements [6–9], when compared with IABP measurements, particularly in unstable or haemodynamically abnormal patients [10–12].
In a transport environment, it is recommended to have a lower threshold for invasive arterial monitoring , as the accuracy of oscillometric NIBP measurement devices is reduced by noise and vibration in the transport environment . Noise and vibration are significant in aircraft , and it has been suggested that this may make NIBP measurement less accurate in the aeromedical environment . If the effect of flight is to make NIBP measurements less accurate, then the threshold for establishing a patient on invasive monitoring should be lower for patients being transferred by aeromedical transport.
The Emergency Medical Retrieval Service (EMRS) provides a national aeromedical critical care service within Scotland. The EMRS mainly undertakes retrieval and transfer of critically ill adults from small healthcare facilities in remote and rural areas of Scotland to larger urban centres for definitive care. The service has a mixed population of trauma and medical illness of which 60% are level-3 patients and the remaining 40% level-2 patients.
The aim of this study was to assess the accuracy of NIBP management compared with IABP monitoring in an aeromedical transport environment, and to investigate the effect of flight on the accuracy of NIBP measurements.
A copy of the protocol was submitted to the Local Research Ethics Committee who reviewed the protocol and confirmed that ethical approval was not required.
Data were gathered over a 24-month period from April 2009 to April 2011. All EMRS transfers generate standardised patient data collection, which includes basic data including times for patient loading, take-off, landing and transfers as well as a printout from the default service monitor (Propaq Encore, Welch Alleyn, UK), which collects physiological data in real-time. Data were gathered retrospectively from these.
Measurements were considered eligible for inclusion if the transported patient had simultaneous measurements for NIBP and IABP for the entire duration of their transfer. If data were missing for part of the transfer, then the patient was not studied. Paired values for NIBP and IABP were recorded for both systolic and mean arterial pressure. Patient data collection forms were reviewed to categorise these measurements as ‘pre-flight’ or ‘in-flight’.
Pre-flight readings were defined as those taken after the patient had been established on the retrieval team monitoring and any additional treatment commenced, but before any transport. This reading was taken with the patient in a healthcare environment, such as a rural hospital or rural general practitioner’s surgery.
In-flight was defined as the mid-flight time between leaving the originating hospital and arrival at the destination hospital. Patients were all transported by aircraft, either civilian fixed-wing aircraft (Beech King Air 200, Wichita, KA, USA), civilian rotary-wing aircraft (Eurocopter 135, Marignane, France) or military search-and-rescue aircraft (Westland Sea King Mk V, Yeoviltown, UK or Sikorsky S92, Stratford, CT, USA).
Data were gathered by a single researcher and recorded on a Microsoft Excel (Microsoft, Richmond, WA, USA) spreadsheet. Statistical analysis was then performed using spss 15.0 (IBM, New York, NY, USA).
Bland-Altman plots  were constructed to assess the accuracy of NIBP against the gold standard of IABP. This statistical method allows assessment of how well one mode of measurement agrees with another and is particularly useful in a scenario where the true values remain unknown. It plots the difference of the two measured values against the mean of the values. Plots are presented with 95% limits of agreement (precision; ± 2 SD) and bias (the mean difference). This provides a visual guide to judge how well the two methods of measurement agree. The smaller the range between the two limits the better the level of correlation . Data were compared for systolic and mean pressures during the two separate time periods (‘pre-flight’ and ‘in-flight’) and analysed for evidence of a difference using paired t-tests.
A total of 251 patients had an intra-arterial catheter sited during the study period. Complete datasets were available for 56 patients during the study period. The limits of agreement for pre-flight systolic blood pressure were −37.3 mmHg to 30.0 mmHg (Fig. 1a). The limits of agreement for pre-flight mean arterial pressure were −20.5 mmHg to 25.0 mmHg (Fig. 1b).
The limits of agreement for in-flight systolic blood pressure were −40.6 mmHg to 33.1 mmHg (Fig. 1c). The limits of agreement for in-flight mean blood pressure in-flight were −23.6 mmHg to 24.6 mmHg (Fig.1d). The bias for the four conditions ranged from 0.6 to −3.8 mmHg (Fig. 1).
There were no differences between pre-flight and in-flight readings for NIBP systolic pressure, NIBP mean arterial pressure, IABP systolic pressure and IABP mean arterial pressure (data not shown).
Our data show that NIBP is imprecise as a measure of true IABP. The wide range of the limits of agreement markedly reduces the clinical reliability of using non-invasive pressure results as an estimation of blood pressure, even though the bias is small. The limits of agreement in this study suggest that mean blood pressure measured non-invasively is more precise than systolic blood pressure. This should be borne in mind by clinicians using systolic blood pressure in making clinical decisions, and should provoke caution when applying scoring systems that include systolic blood pressure. Indeed, the validity of relying on non-invasive measurements of systolic blood pressure to make clinical decisions could be questioned. Therefore, in services or clinical situations where IABP monitoring is not possible, NIBP is a reasonable surrogate, but caution should be exercised with regards the overall accuracy as previously discussed.
In hospital, measurements of NIBP have previously been shown to be inaccurate compared with IABP in a variety of patient groups [3, 6, 8–12]. Rutten et al.  demonstrated variation between methods of −29% to +40% in 1947 paired measurements in elective (aortic) surgical patients. Another operating theatre-based study  showed 95% limits of agreement of 25 mmHg for systolic, 18 mmHg for diastolic and 17% for mean arterial pressure. This would support our finding that mean arterial pressure is the more accurate measurement. In obese critically ill patients, limits of agreement have been shown to be −49.5 mmHg to +33.6 mmHg, with increasing BMI not affecting the accuracy of results . Muecke et al.  tested results obtained from different monitors and obtained limits of agreement for mean arterial pressure ranging between −19.6 mmHg and 13.7 mmHg. Another large emergency department-based study investigating various NIBP cuff sizes showed a bias of −2.4% ± 11.8 mmHg or −5.3 ± 11.6 mmHg in 1011 paired readings .
Previous reports have suggested that NIBP measurements may be less accurate, specifically in the in-flight environment, with differences of 33 mmHg for systolic and 15 mmHg for mean arterial pressure, with some cases of unrecordable NIBP, but acceptable IABP . Possible explanations are that aircraft vibration may make NIBP measurement less accurate . Our study did not show a significant difference in the accuracy of NIBP measurements in an aeromedical environment when compared with the measurements in ‘pre-flight’. This may reflect improved technology within NIBP monitoring devices or newer aircraft’s causing less vibration in the patient and monitoring equipment.
The use of IABP as the gold standard in critically ill adults, including in the transport environment, also raises issues about what should be done in the pre-hospital environment. Physician-staffed pre-hospital services are becoming more prevalent throughout the world due to the potential to improve outcomes, particularly for the critically injured [19–21]. The inaccuracy of NIBP has been shown to persist out of hospital as well. A French land-based pre-hospital medical team found 44% of systolic readings results differed by 20% or greater in one group , and another intensive care transport team concluded that NIBP underestimated systolic pressure by 13–21%, yet over estimated diastolic by 5–27% . Feasibility studies have suggested that IABP monitoring can be established in the pre-hospital environment [22, 23]. It remains to be seen whether the risks of this procedure in the pre-hospital environment, particularly infection risk and increased time to definitive care, are outweighed by clinical benefit.
The limitations of this study are the relatively small number of cases with complete results, despite a large database of cases that were eligible. This was due to limited use of NIBP monitoring simultaneously with IABP monitoring. Clinicians were allowed to turn off the NIBP monitoring if there was a clinical concern that it might affect the monitor battery adversely. The small study size also introduces the possibility of a type-2 error in stating there is no difference between pre-flight and in-flight readings. However, if this is the case, the numerical difference between pre-flight and in-flight readings is likely to be small and of no clinical significance.
Also, it was assumed, but not guaranteed, that the correct sized cuff was used to obtain the NIBP results for each patient. Incorrectly sized cuffs will reduce the accuracy of the measurements, and although we carry a range of sizes, the design of our study did not ensure that measurement and sizing were performed for every patient. Furthermore, these results are for single blood pressure measurements, not a trend in results. Trends in NIBP may be more reflective of true IABP changes, and therefore more clinically useful.
From our results, IABP monitoring should be used in any unwell patient in whom accurate blood pressure measurement is desirable. The general inaccuracy of the NIBP measurements obtained dictate that IABP monitoring should remain the accepted ‘gold standard’ of care in any critical care environment. Notwithstanding this, in our study, the aeromedical transport environment does not lead to less precise NIBP results than the non-transport environment. Thus, we conclude, where NIBP measurement is the only option, that the mean blood pressure should be used in preference to systolic measurements.
We are grateful for the assistance of our clinical colleagues in collecting the data for this study. We also acknowledge the help and advice of David Young of the University of Strathclyde Department of Medical Statistics.
No external funding or competing interests declared.