Heart rate and systolic blood pressure in patients with minor to moderate, non-haemorrhagic injury versus normal controls
- 1Division of Emergency Medicine, University of Cape Town, Cape Town, South Africa
- 2Department of Emergency Medicine, Derriford Hospital, Plymouth, UK
- 3Trauma Audit Research Network, Manchester Medical Academic Health Sciences Centre, University of Manchester, Salford Royal NHS Foundation Trust, Salford, UK
- Correspondence to Dr Stevan Raynier Bruijns, Division of Emergency Medicine, University of Cape Town, Karl Bremer Hospital, Mike Pienaar Blvd, Bellville, 7535, South Africa;
- Accepted 22 October 2012
- Published Online First 26 November 2012
Background Raised blood pressure (and heart rate (HR)) due to anxiety in a clinical situation is well described and is called the white coat effect (WCE). It is not known whether the pain and anxiety that results from trauma causes a measurable WCE.
Methods A sample of patients with a non-haemorrhagic injury from the Trauma Audit and Research Network (TARN) was compared with a healthy, non-injury sample from the Health Survey for England (HSE) databases. Two-way analysis of variance with rank transformation of data was used to compare systolic blood pressure (SBP) and HR between the groups at different ages. In the injured group, the SBP and HR were also compared between spinally immobilised and non-immobilised patients.
Results There was a statistically significant difference between the groups for both HR and SBP (p<0.001). Median HR remained approximately 10 bpm higher in the TARN set when compared to the HSE set, irrespective of age. The difference for SBP was not considered clinically relevant (the highest was 5 mm Hg). There was no significant difference between immobilised and non-immobilised patients, for either HR or SBP (p=0.07 and 0.3, respectively).
Discussion Median HR remained approximately 10 bpm higher in the TARN (injury) set compared to the HSE (non-injury, control) set, irrespective of age. Understanding that HR reacts in this way for mild to moderately injured patients is important as it will affect clinical interpretation during the initial assessment.
Evaluation of an injury includes assessment of a patient's vital signs in order to make an early judgement of the patient's condition.1 However, this is not straightforward, as vital signs in trauma are affected by raised intracranial pressure, spinal cord injury, tissue damage, haemorrhage, pain and anxiety.2 Despite widespread use of Advanced Trauma Life Support (ATLS) principles, their guidance relating to haemorrhage and vital signs lacks evidence.3 ,4 In particular, ATLS does not appear to acknowledge either the vagal response to pure haemorrhage which results in a bradycardia, or the effect of tissue injury.4 Tissue injury alters the cardiovascular response to haemorrhage, by attenuating the vagal response (resulting in slightly higher blood pressure (BP) and heart rate (HR).2 ,5 This is largely due to a prevailing sympathetic outflow as the result of decreased baroreflex receptor sensitivity induced through the tissue injury. Pain appears to affect baroreceptor sensitivity, but through mitigation of the vagal component of the receptor (the exact mechanism still eludes scientists).6 However, only severe pain induces any clinically distinguishable effect and even then the result can be variable.6–8 Haemorrhage, injury and pain (each exerting its own effect) all contribute to the changes in vital signs which result from trauma.2
Anxiety can induce an increased BP and HR without the presence of injury, haemorrhage or pain.9 ,10 The white coat effect (WCE), associated with anxiety,11 ,12 is well described and refers to a raised BP (and HR) observed when patients present in a clinical situation.11 ,12 A comprehensive search of the literature (including the British Nursing Index, EMBASE, CINAHL, MEDLINE and Google Scholar) revealed no studies on WCE in a trauma setting. It is thus not known whether a WCE is observed in the trauma setting, and if it is, whether it can be related to simple trauma adjuncts such as spinal immobilisation. If the WCE in a trauma setting was known, this could be factored in during the evaluation, allowing distinction from what is less relevant and what is likely to be of clinical concern.
The primary aim of this study was to compare the relationship between HR and systolic BP (SBP) in non-haemorrhagic, minimally injured patients with that of a non-injured control group. A secondary objective was to compare within the injured group those that required spinal immobilisation and those that did not. The null hypothesis for all objectives was that no difference existed.
Permission was obtained from the Trauma Audit and Research Network (TARN) and the Health Survey for England (HSE) to use data from their respective databases to perform a retrospective case–control study. The TARN database is the largest trauma database in Europe, collecting data related to trauma patients from a group of collaborative hospitals in England and Wales since1989. The HSE is an annual national health survey representing people of different age, sex, geographic area and socio-demographic backgrounds in England.13 This study was approved by the Research Ethics Committee of the University of Cape Town, South Africa (reference 014/2010). The HSE data would be representative of a physiologically unstressed population such as might attend a UK emergency department (ED) following trauma. The TARN data was used as the test group and the HSE data as the control.
Subjects from either database were included if they were older than 16 and had SBP and HR recorded. Respiratory rate was not included as this is not part of the HSE dataset. Both datasets drew their respective samples from the period 1996 to 2006. The HSE sample included subjects who had eaten, drank alcohol, smoked or exercised vigorously within the 30 min preceding measurement in addition to subjects who did none of these. Specific inclusion criteria for the TARN sample were patients with upper extremity, or below-knee, lower extremity injury. In order to ensure an injury sample that did not include haemorrhage as a cause for deranged vital signs, subjects in the TARN database with the following were excluded:
Amputation, crush, degloving, penetrating, laceration, skin avulsion or vessel injuries present
Open fractures, scapular injuries, femur or pelvic injuries
Additional injuries other than minor injury
Glasgow Come Score less than 15, intubated or cardiopulmonary resuscitation required
More than 1 litre fluid used (prehospital and/or ED)
Only partial spinal immobilisation used
Median, IQR, mean, SD and 95% CIs were used to describe different datasets. Although not typically reported for non-parametric data, mean and SD were included to compare to the reference range for HR given in a study mentioned in the discussion section.14
An age variable consisted of seven age categories grouped as follows: 16–25, 26–35, etc, with all those aged over 75 grouped together. A further two variables included an injury variable (HSE or TARN) and an immobilisation variable (full immobilisation or no immobilisation). A two-way analysis of variance (ANOVA) was done to compare the mean SBP and HR between the injury variable, and the immobilisation variable, respectively, using the age variable as the additional factor for each analysis. Rank transformation of SBP and HR was required since the real data violated the assumption of normality. Use of rank transformed data in parametric tests (including a two-way model) generally compares well with non-parametric test results in terms of robustness and power. This approach provides a novel solution where non-parametric alternatives are not readily available, such as with two-way ANOVA.15
A p value of less than 0.05 was considered statistically significant.16 Outcomes measured were the difference for HR and SBP between the TARN and HSE datasets, and the fully immobilised and non-immobilised datasets, respectively. The literature suggests that a clinically relevant WCE effect is likely to cause SBP to increase by 25–30 mm Hg and the HR to increase by 10–15 bpm.11 ,12 Values of 25 mm Hg for SBP and 10 bpm for HR were used as outcome measures for this study.
Table 1 presents the different datasets used. The disparity in data between the TARN sample and the full/non-immobilised samples relates to 1930 TARN cases where immobilisation data were not available. These data points were included for analysis in the larger TARN sample, but not the immobilisation variable.
Two-way ANOVA of rank transformed data showed that there was a significant main effect of injury on both the HR and SBP (p<0.001). The main effect of age was also significant for both HR and SBP (p<0.001). There was a significant interaction effect between the injury and age variables, for both HR and SBP (p<0.001). This result indicates that the vitals in the no injury (HSE) and injury (TARN) groups were affected differently by age. These differences are shown in figure 1A,B. In real values, median HR remained approximately 10 bpm higher in the TARN set when compared to the HSE set, irrespective of age, and this value satisfied the outcome measures (figure 2). SBP in the TARN set was higher than the SBP in the HSE set at younger ages, but this difference reduced dramatically in the older age groups. The difference (although significant in some groups) did not satisfy the outcome measures at any point. Tables S2 and S3 (data supplement) show the descriptive statistics for HR and SBP, respectively, for the injury–age variables’ interaction.
Two-way ANOVA of rank transformed data showed that the effect of immobilisation on the HR was not significant (p=0.36). The effect of immobilisation on SBP was significant (p<0.001). There was no significant interaction effect between immobilisation and age, for either the HR or SBP (p=0.07 and 0.3, respectively), indicating that the vital signs in the no immobilisation and full immobilisation groups were not affected differently by age. These differences are shown in figure 1C,D. Although the SBP in the fully immobilised set was higher than the SBP in the non-immobilised set throughout the age groups, this difference was not clinically significant. This is shown in table S4 (data supplement). As the difference in HR was not significant for either immobilisation or its interaction with age, descriptive data are not reported.
The most pertinent finding was that median HR remained approximately 10 bpm higher in the TARN (injury) set compared to the HSE (non-injury, control) set, irrespective of age. SBP was slightly higher in younger age groups (<56 years), but although statistically significant, even the highest mean difference (5 mm Hg) was considered clinically irrelevant when compared with a BP rise expected due to a WCE (25–30 mm Hg). No statistical differences were found between the immobilised and non-immobilised groups for HR, and the difference found with SBP was not considered clinically relevant.
These data suggest that the upper limit for the ‘normal’ HR in injured patients may need to be reconsidered. In order to determine this, one has to consider the calculations that derived the reference range of the current accepted standard (60–100 bpm). This reference originates from a publication of the New York Heart Association in 1928.17 Interestingly, its basis was to a large extent due to the fact that 60 and 100 bpm, respectively, represented five and three small blocks on an ECG recorded at 25 mm/s.15 ,18 A statistical solution followed using the mean±2 SD (included in table 1).14 This approach drew criticism since HR is known not to be normally distributed.18 The mean is therefore not robust enough to describe central tendency in this skewed sample. An alternative approach suggested was to use IQR.18 This approach was based on the observations made in several studies showing increased cardiovascular risk when resting HRs were found to be in the upper quintile of a sample's distribution.19 ,20 ,21
Interestingly in our sample, the 25th and 75th HR centiles increased, similar to the median, by approximately 10 bpm from the HSE to the TARN dataset. Mean HRs maintained the 10 bpm difference (with very minimal variation) throughout all age groups evaluated. It is not known whether this increase in HR increases cardiovascular risk as well, or how long this relative tachycardia will be sustained following injury. As no haemorrhage occurred in our sample population and it is known that less than severe pain has a very limited effect on vital signs,7 8 this relative tachycardia is more likely to be the result of anxiety, injury or a combination of anxiety and injury.6 Whether anxiety was the result of a WCE or directly related to the trauma event is debatable. Longitudinal studies have previously looked at HR in the ED as a predictor of post-traumatic stress disorder (PTSD) and found that the resting HR (at 1 month post-ED attendance) was lower than that recorded in the ED in non-PTSD groups (PTSD groups had even higher ED HRs, with resting HRs similar to the non-PTSD group).22 ,23 The difference ranged between 6 and 11 bpm depending on the study. The difference in the PTSD groups was higher (0.5 and 13 bpm, respectively). The difference seen in the non-PTSD subjects were put down to subjects ‘expressing a stress-response’.24 It seems plausible that the 10 bpm difference seen in our study was due to a similar stress-response and that this stress-response was not affected by spinal immobilisation. Younger patients in this study tended to have higher SBPs, although this was not clinically relevant. Previously it has been noted that in blunt injured children, SBP was higher than was seen in similarly aged children at rest. This difference appeared to be unrelated to injury severity.25 HR, although higher than seen in the at-rest group, was not significantly different from a resting baseline HR.25 In our sample, SBP was higher in the injured population in lower age groups, although this difference was not considered clinically relevant. It is likely that the SBP difference increases with decreasing age.
The authors are aware of the inherent information and selection biases associated with case–control studies from unrelated databases and results should be interpreted with these in mind. The large sample sizes yielded statistically significant results despite low clinical relevance. Arguably, deciding what is clinically relevant can be tricky as it is often open to interpretation. Ensuring that clear outcome measures were included in the design was a way to define clinical relevance and turned out to be a useful guide to interpret results.
Changes seen in the vital signs of patients with injuries have complex mechanisms and include haemorrhage, injury, fear and pain. This study has shown that as far as mild to moderate injury is concerned, HR tends to be higher than that expected in an uninjured person. Differences seen in this study due to spinal immobilisation are also not likely to be clinically relevant. Understanding that HR reacts in this way for mild to moderately injured patients is important as it will affect the way we interpret the HR during the initial assessment. Clinicians should be aware of this occurrence so as to not confuse mild to moderately injured patients with a more severe cohort. More work is required to evaluate this phenomenon further.
The authors would like to acknowledge Dr Jennifer Mindell from HSE for her help with the study's ethics application and assistance in obtaining the HSE data sample. They would also like to acknowledge the Plymouth Hospitals Research and Development service for their role as sponsors. The service had no further involvement and in particular, no funding involvement.
Contributors All authors made a significant contribution to the conception and design of the study, acquisition of data or analysis and interpretation of data; drafting and revising the article; and approval of the final version submitted.
Competing interests None.
Ethics approval Research Ethics Committee of the University of Cape Town, South Africa (reference 014/2010).
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All results from analysed data have been presented in this paper. Raw data are available from the TARN and the HSE to bona fide researchers on request.