Article Text
PDF
Abstract
Background: The modified early warning score (MEWS) is a useful tool for identifying hospitalised patients in need of a higher level of care and those at risk of inhospital death. Use of the MEWS as a triage tool to identify patients needing hospital admission and those at increased risk of inhospital death has been evaluated only to a limited extent.
Aim: To evaluate the use of the MEWS as a triage tool to identify medical patients presenting to the emergency department who require admission to hospital and are at increased risk of inhospital death.
Methods: Physiological parameters were collected from 790 medical patients presenting to the emergency department of a public hospital in Cape Town, South Africa. MEW scores were calculated from the data and multivariate regression analysis was performed to identify independent predictors of hospital admission and inhospital mortality.
Results: The proportion of patients admitted and those who died in hospital increased significantly as the MEW score increased (p<0.001). Multivariate regression analysis identified five independent predictors of hospital admission: systolic blood pressure ⩽100 mm Hg, pulse rate ⩾130 beats per minute, respiratory rate ⩾30 breaths per minute, temperature ⩾38.5°C and an impaired level of consciousness. Independent predictors of inhospital death were: abnormal systolic blood pressure (⩽100 or ⩾200 mm Hg), respiratory rate ⩾30 breaths per minute and an impaired level of consciousness.
Conclusion: The MEWS, specifically five selected parameters, may be used as a rapid, simple triage method to identify medical patients in need of hospital admission and those at increased risk of inhospital death.
Statistics from Altmetric.com
The modified early warning score (MEWS), which is calculated using simple-to-measure physiological parameters (table 1), is a useful tool for the bedside evaluation of ill medical patients. This tool has previously been shown to identify hospital inpatients at risk of catastrophic deterioration and at increased risk of inhospital death.1 A raised MEW score in medical admissions correlates with an increased need for transfer to a higher level of care and a score of 5 or more is associated with a fivefold increased risk of inhospital death.1 Similar scoring criteria, which include additional parameters such as oxygen saturation, urine output and seizures, have also been found to predict mortality in hospitalised patients with an increasing number of abnormal parameters.2 3
In busy emergency departments (ED) there is a clear need for the early identification of ill medical patients who require admission. This is particularly true of ED where all emergencies, including trauma cases, are seen in a single system. Trauma cases may, because of their dramatic presentation, be preferentially seen, whereas apparently “stable” medical patients may wait for extended periods of time before being evaluated. To date, four reliable ordinal ED triage scales have been researched and published: the Manchester triage scale,4 the Canadian triage acuity scale,5 the Australasian triage scale6 and the emergency severity index.7 Unfortunately these complex scoring methods are of limited use in resource-constrained settings or circumstances in which junior clinical staff, who have limited training and experience, practice. This may be particularly true of ED in developing countries.8 A simple tool that facilitates the early identification of ill medical patients who require admission to hospital would be of great value in resource-constrained settings. The utility of the MEWS as a triage tool in the ED setting was evaluated in a pilot study in which it was shown that MEW scores of 2 or more were associated with an increased likelihood of hospital admission.9 More recently, the use of physiological scoring systems, including the MEWS, has been validated in the accident and emergency department setting.10 Further evaluation of the MEWS as a triage tool to identify medical patients requiring admission to hospital and at risk of catastrophic deterioration (inhospital death) has not been undertaken.
METHODS
Design and setting
A prospective, observational study was conducted in the ED of GF Jooste Hospital, an urban hospital in Cape Town, South Africa. The hospital is a 210-bed facility providing general medical, surgical, trauma, orthopaedic and limited gynaecological services to a population of approximately 1.2 million people.11 The ED evaluates approximately 54 000 patients per year, of which approximately 15 500 are admitted.7 9 The hospital operates at a bed occupancy rate in excess of 100% most days of the year (personal communication).
Study population
The study was conducted over 3 months and patient data were captured for all ED visits every sixth day (24-h cycle, including weekends) throughout the study period. Medical patients were defined as patients presenting with a clinical problem that was not surgical, orthopaedic, gynaecological or trauma related, as determined by the attending ED doctor.
Data capture
Before commencing the study, ED nursing staff were trained to record the following information in the nursing notes: gender, age, systolic blood pressure (SBP), pulse rate, temperature, respiratory rate and AVPU (A, alert, V, responding to voice, P, responding to painful stimuli, U, unresponsive) score upon presentation to the ED. Blood pressure and pulse were measured electronically (DINAMAP, GE Healthcare, St Giles, UK), temperature was measured using a tympanic thermometer (First Temp, Genius; Sherwood-Davis and Geck, St Louis, Missouri, USA), respiratory rate was counted over one minute and AVPU scores were scored according to the best response obtained at the time of measuring the blood pressure and pulse. Admission MEW scores (table 1) were calculated from raw data, as recorded in the ED notes, after patients were discharged from the ED or after admission to hospital. ED doctors did not have access to MEW scores at the time when patients were evaluated in the ED. The primary outcomes of the study were admission to hospital and death during hospital stay.
Statistical analysis
Data were analysed using Stata 8.0. Frequency distributions and univariate descriptive statistics were used for preliminary analyses. Bivariate associations were described using Student’s t-tests (for means), Wilcoxon rank-sum tests (for medians) and risk ratios and χ2 tests (for proportions), as appropriate. A non-parametric test for trend across ordered groups was used to assess the relationship between the MEW score and outcomes (admission and death). Variables were entered into logistic regression models to assess independent effects of the MEWS parameters on outcomes. One model was developed for admission to hospital and one for death during hospital stay.
RESULTS
During the study period it was estimated that approximately 13 500 patients were seen in the ED.8 At least 50% of all ED visits were medical patients (personal communication)—approximately 6750 patients. In order to achieve a sample of approximately 20% of all medical patients seen, data were captured every sixth day (24-h cycle, weekends included) resulting in an expected study cohort of approximately 1125 patients. Complete data were collected for 790 patients; approximately 70.2% of the potential study cohort. Fifty-five per cent of patients were women and the median age was 43 years (range 14–89). Fifty-nine per cent of patients (n = 469) were admitted to hospital after evaluation in the ED. Of these, 24% (n = 113) died during their hospital stay. The mortality for all patients included in the study was 14.3%. MEW scores recorded in the ED ranged from 0 to 11, with a median score of 3 (fig 1). A MEW score of 5 or more was recorded for 204 patients (26%) seen in the ED.
Admission to hospital
The proportion of patients admitted increased significantly as the MEW score increased (p value for trend <0.001) (fig 2). Forty-five per cent of patients with a MEW score of 0–2 (reference category for respiratory rate), 59% with a MEW score of 3–4 (risk ratio 1.3; 95% CI 1.1 to 1.6) and 79% with a MEW score of 5 or more (risk ratio 1.7; 95% CI 1.5 to 2.0) were admitted, respectively. Admitted patients had a significantly higher MEW score, lower SBP, higher pulse rate and higher respiratory rate than patients discharged from the ED (table 2). Admitted patients were not significantly older than patients discharged from the ED.
Inhospital mortality
The proportion of patients who died in hospital increased significantly as the MEW score increased (p value for trend <0.001) (fig 2). Approximately 5% of patients with a MEW score of 0–2 (reference category for respiratory rate), 16% with a MEW score of 3–4 (risk ratio 2.8; 95% CI 1.7 to 4.8) and 26% with a MEW score of 5 or more (risk ratio 4.6; 95% CI 2.7 to 7.8) died, respectively. Inhospital mortality increased significantly with an increasing number of abnormal parameters recorded in the ED (p value for trend <0.001). Comparison of inhospital deaths with admissions discharged alive showed that the mean MEWS was significantly higher among those who died (4.5 versus 3.8, p = 0.001), but there was no significant difference between the two groups in terms of mean age, SBP, pulse rate, respiratory rate or temperature.
Risk of admission and inhospital mortality
Multivariate regression analysis identified MEWS parameters independently associated with an increased risk of hospital admission: SBP of 100 mm Hg or less, pulse rate of 130 beats per minute or greater, respiratory rate of 30 breaths per minute or greater, temperature of 38.5°C or greater and an impaired level of consciousness. Three of these parameters, SBP (⩽100 or ⩾200 mm Hg), respiratory rate of 30 breaths per minute or greater and an impaired level of consciousness were also independent predictors of inhospital death (table 3).
Impact of age on admission and inhospital death
Subgroup analysis (age less than 40 years, age 40–65 years, age more than 65 years) showed that there were no significant differences in age for admission or inhospital mortality.
Abbreviated MEWS
Using the five parameters identified by multiple regression analysis, we calculated an abbreviated MEWS for all patients studied. This was done by assigning a score of 0 or 1 to each one of the five parameters, thereby producing a score ranging from 0 to 5. Fig 3 shows the outcome of patients according to the abbreviated MEWS. What can be seen is that 45.5% of patients with a score of 0 required admission; this figure steadily increased to 100% by the time a score of 4 was reached. The mortality data show that 5.2% of patients with a score of 0 died during their hospital stay; this increased dramatically as the score rose to 4, with an inhospital mortality in excess of 50%.
DISCUSSION
This study shows that the MEWS can be used as an ED triage tool to identify ill medical patients requiring admission to hospital and those at increased risk of inhospital death. There was a significant linear relationship between the ED MEWS and admission to hospital. Seventy-nine per cent of ED medical patients with a MEW score of 5 or more were admitted to hospital. These findings support those of a recent study, which found that patients admitted to hospital had a higher MEWS than those discharged from the ED.9
In addition to identifying patients requiring hospital admission, there was also a significant linear relationship between the ED MEWS and inhospital mortality. The ability of deranged physiological parameters to predict the risk of inhospital death has previously been demonstrated for medical inpatients.1–3 12 The use of the MEWS to predict inhospital death before admission is, however, a new finding that further supports the use of the MEWS as a triage tool for medical patients presenting to busy hospital ED. In this study we did not evaluate 30-day mortality and so we may have underestimated the longer-term mortality rate in this cohort.
The most important advantage of using the MEWS as a triage tool, compared with other international triage systems, is its simplicity. The MEWS is especially suitable for use in resource-limited environments and settings where clinical staff have limited training and experience. In our study the MEWS parameters were recorded by trained enrolled nursing assistants; nursing staff with very limited training in South Africa.
In this study we identified five abnormal physiological parameters that independently predicted hospital admission—SBP of 100 mm Hg or less, pulse rate of 130 beats per minute or more, respiratory rate of 30 breaths per minute or more, temperature of 38.5°C or more and an impaired level of consciousness. Furthermore, three of these parameters were also independent predictors of inhospital death—SBP, respiratory rate and an impaired level of consciousness. Our data are supported by two other studies identifying these abnormal observations as independent predictors of inhospital mortality.2 3
Using these five independent predictors of outcome to calculate an abbreviated MEWS we showed that a score of 1 or more was associated with an inhospital mortality of at least 20%. It therefore seems reasonable to propose the use of an abbreviated MEWS to identify medical patients rapidly in the ED who are at risk of catastrophic deterioration and are in need of urgent evaluation and admission.
In contrast to other published data, age was not an independent predictor of admission or death in this study. In three studies from the United Kingdom the mean age of the study population was 60 years or older,1 3 12 and in two of the studies older age was associated with an increased risk of inhospital death.1 3 The median age of our study population was 43 years and there was no significant difference in age between those admitted to hospital and those discharged from the ED or admitted survivors and inhospital deaths. This difference between our findings and the findings of other studies is likely to be due, at least partly, to the high burden of HIV/AIDS in South Africa.13–15 The HIV infection prevalence in the communities served by our hospital is approximately 30%.14 Approximately 43% of medical admissions are HIV positive and more than 85% of cases have advanced disease.11 As a consequence of this epidemic the average life expectancy of South Africans is currently approximately 45 years.16
Having demonstrated the value of the MEWS as a triage tool in a high HIV prevalence setting is important given that developing countries, where simple triage tools may be of greatest benefit, are currently also world regions with high HIV prevalence rates, as is the case in South Africa.
Although the utility of the MEWS as a triage tool is apparent, there is an important limitation that needs to be recognised when using the MEWS. In our study, as shown in the literature,10 45% of patients having low MEW scores (0–2) required admission to hospital. This may be due to clinical conditions requiring admission that often do not have significant derangement of selected physiological parameters, for example, acute coronary syndrome or cerebrovascular injury. This is clearly cause for concern, and attempts to address this limitation of the MEWS have been made by the Cape Triage Group in South Africa. They propose use of the MEWS in conjunction with a list of clinical conditions that may present without significant physiological parameter derangement.8 This system, currently being evaluated in South Africa, may significantly improve the case detection rate of the MEWS.
In our study we also observed that patients with high MEW scores, 5 or more, were not all admitted to hospital. There are two possible explanations for this observation. First, chronically unwell patients may present with high MEW scores that do not change dramatically despite optimal therapy. Given the high prevalence of advanced HIV infection among medical admissions to our hospital, at least 43% of admissions,11 it is likely that a similar percentage of medical patients in our ED have concurrent advanced HIV disease and are therefore likely to present with a higher MEW score than would be expected of a healthy population presenting with acute illness. The impact of chronic HIV infection on ED MEW scores, particularly the trend of these scores after initial evaluation, is not known and deserves further evaluation given the magnitude of the pandemic in the developing world, where a simple physiological scoring system such as the MEWS is likely to be of greatest utility.
Second, as a result of a lack of bed space in our hospital, up to 10% of patients requiring admission are routinely managed in the ED where they receive treatment for acutely reversible conditions such a pulmonary oedema, bronchospasm or dehydration caused by vomiting or diarrhoea (personal communication). These patients may present with a MEW score of 5 or more but rapidly settle on appropriate therapy in the ED. This means that the study may have underestimated the number of patients requiring acute treatment for 24 h or less. We did not serially evaluate the MEWS of medical patients kept overnight in the ED and so we are unable to substantiate this possible explanation.
Another limitation of the use of the MEWS in the ED setting is that it has not been validated as a triage tool for non-medical and trauma emergencies. This issue has also recently been addressed by the Cape Triage Group in South Africa. They suggest the addition of a mobility parameter to the original MEWS as well as the inclusion of a list of surgical and trauma conditions that may not present with sufficiently deranged physiological parameters to be identified by the MEWS.8 This tool, the Triage Early Warning Score (TEWS), is currently being evaluated in ED settings that deal with medical, non-medical and trauma emergencies.
It is important to remember that the MEWS was originally designed to detect critically ill patients at risk of catastrophic deterioration. In this paper we have shown that an abbreviated MEWS is able to detect such patients in the ED setting. This has significant resource implications because it has been shown that the admission of critically ill patients directly to intensive care units (ICU) from the ED, compared with the admission of critically ill patients to a general ward before transfer to the ICU, results in a shorter duration of stay in the ICU and an overall shorter duration of stay in hospital.10 This may serve as an incentive for using the abbreviated MEWS to identify rapidly critically ill ED patients who require urgent attention and possible direct admission to a high care unit or ICU.
CONCLUSION
This study demonstrates the utility of the MEWS as a triage tool for medical emergencies seen in ED settings where resource and personnel constraints limit the use of more complex triage systems used in developed countries. Further studies are required to evaluate the impact of an abbreviated MEWS, as described in this paper, on the outcome of critically ill patients in the ED setting. Further studies are also needed to validate other simple-to-use clinical parameters that may facilitate the rapid triaging of medical, non-medical and trauma emergencies, for example, the TEWS as proposed by the Cape Triage Group in South Africa.8
Acknowledgments
The authors would like to thank all the nursing staff at GF Jooste Hospital who collected the data for this study.
Footnotes
Contributors: VCB had the original idea and wrote the first draft of the paper. GT collected all the data and contributed to the first draft of the paper. CM and VCB performed the data analysis. All authors contributed to the final draft of the paper. VCB is the guarantor of the paper.
Competing interests: None.
Ethics approval: The data used for this manuscript were collected as part of a routine hospital practice audit for quality control purposes. All data were anonymously collected and routine patient care was not affected by the data collection process. The hospital board approved the use of the data for the preparation of this manuscript and deemed that patient consent was not required.
Linked Articles
- Primary survey