Article Text

Download PDFPDF

Validation of the Taiwan triage and acuity scale: a new computerised five-level triage system
  1. Chip-Jin Ng1,
  2. Zui-Shen Yen2,
  3. Jeffrey Che-Hung Tsai3,
  4. Li Chin Chen4,
  5. Shou Ju Lin5,
  6. Yiing Yiing Sang6,
  7. Jih-Chang Chen1,
  8. TTAS national working group*
  1. 1Department of Emergency Medicine, Chang Gung Memorial Hospital at Linkou, Chang Gung University College of Medicine, Taoyuan, Taiwan
  2. 2Department of Emergency Medicine, National Taiwan University, Taipei, Taiwan
  3. 3Department of Emergency Medicine, Cheng Ching General Hospital, Taichung, Taiwan
  4. 4Department of Nursing, Chang Gung Memorial Hospital at Linkou, Taoyuan, Taiwan
  5. 5Department of Nursing, National Taiwan University Hospital, Taipei, Taiwan
  6. 6Department of Nursing, Taipei Veterans General Hospital, Taipei, Taiwan
  1. Correspondence to Dr Jih-Chang Chen, Department of Emergency Medicine, Chang Gung Memorial Hospital, 5 Fu-Hsing Street, Gueishan, Taoyuan 33333, Taiwan; a9999999{at}


Objective An ideal emergency department (ED) triage system accurately prioritises patients on the basis of the urgency of interventions required to avoid under- or over-triage. The objective of this study was to develop and validate a five-level Taiwan triage and acuity scale (TTAS) with an electronic decision support tool.

Methods This prospective, multicentre, observational study included 10533 patients triaged at 11 academic medical centres, 18 regional and four district hospitals. Adult patients presenting to the ED were independently triaged by the duty triage nurse in the usual way and trained research nurses using TTAS with a computerised decision support system. Weighted κ statistics were used to assess the reproducibility. Hospitalisation, length of stay, and medical resource consumption were analysed by TTAS acuity levels.

Results Most cases were stratified into levels 2 to 3 by the existing four-level triage system, whereas the TTAS stratified most patients to levels 3 (41.4%) and 4 (25.0%), and only a small number to level 1 (3.9%) (resuscitation; most urgent). Weighted κ for TTAS assignment was 0.87 (95% CI 0.85 to 0.89). The decrease in mean medical resource consumption and hospitalisation rate was statistically significant with each decrease in the TTAS triage acuity level. The length of stay also decreased significantly as the TTAS level acuity fell from levels 2 to 5.

Conclusions The TTAS was found to be a reliable triage system that accurately prioritises the treatment needed to avoid overtriage, more efficiently deploying the appropriate resources to ED patients.

  • Canadian triage acuity scale
  • clinical assessment
  • competence
  • five-level triage system
  • interobserver agreement
  • reliability
  • Taiwan triage and acuity scale
  • validity

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Since the implementation of the National Health Insurance (NHI) programme in Taiwan, emergency department (ED) visits have increased by approximately 27.7% over the past decade, while over the same period the number of hospitals has decreased by 24%.1 Most hospital ED are facing the problems of increasing patient volumes, increasing acuity, and enormous pressure to control costs as NHI resources are constrained.2 3 Triage, the assignment of ED patient acuity, is a process that helps predict the intensity and nature of treatment likely to be required. An ideal triage system accurately prioritises patients based on urgency of need and avoids undertriage (ie, assigning a triage rating lower than actual acuity) or overtriage (ie, assigning a triage rating higher than actual acuity), helps ED personnel to understand better the daily operations and acuity case mix,4 and streams patients to appropriate care areas within the ED to mitigate some of the negative impacts caused by overcrowding. These concepts have led to extensive research and validation of five-level triage systems, such as the Australasian triage scale,5 the Manchester triage scale,6 the emergency severity index (ESI)7 and the Canadian triage and acuity scale (CTAS).8 9 The use of five-level triage scales has been shown to be superior in reliability and validity in the era of ED overcrowding.4–9Recently, it has been advocated that a five-level international triage scale be developed and implemented to support collaborative research, comparative analysis and evaluation, with a goal to support international best practices.10

Among five-level triage systems, the CTAS has been shown to have good to very good reliability.8 9 11 The CTAS was developed in the late 1990s by the Canadian Association of Emergency Physicians and National Emergency Nurses' Affiliation, and revised in 2004 and 2008.12 13 The CTAS classifies patients in descending order of acuity, ie, level 1, resuscitation; level 2, emergent; level 3, urgent; level 4, less urgent; and level 5, non-urgent. The original CTAS guidelines recommend a time to physician assessment based on triage acuity level. However, the emphasis on the time to physician assessment and the lack of understanding of fractile response rates for system performance was felt to have led to both over and underestimates of the triage level. In the 2004 revision, the time to physician assessment was revised to allow for the timely nurse reassessment of patients waiting to be seen as a safe substitute for ‘time to physician’ to ensure that unavoidable delays do not put patients at risk. The recommended reassessment time intervals are as follows: level 1 patients, continuous nursing care; level 2, every 15 min; level 3, every 30 min; level 4, every 60 min; and level 5, every 120 min.12

Complaint-oriented triage followed the integration of the Canadian Emergency Department Information System presenting complaint list14 15 into CTAS12 and was developed to simplify and better standardise the assignment of a triage level. After ensuring that the patient does not need immediate resuscitation, the triage nurse selects the most appropriate presenting complaint. Each complaint has a specific set of first-order and second-order modifiers the nurse can use to assign the appropriate acuity score. First-order modifiers include vital sign criteria comprised of respiratory distress, haemodynamic stability, level of consciousness and temperature, as well as pain severity, bleeding disorder and mechanism of injury. Stable trauma patients presenting following a high-risk mechanism injury are assigned a CTAS level II. Second-order modifiers are applied when the first-order modifiers are inadequate to assign an appropriate acuity level, and are specific to one or a few complaints. For example, a chemical injury to the eye will not affect vital signs; however, untreated could lead to blindness. Using the ‘chemical exposure, eye’ second-order modifier, this patient is assigned a CTAS level II.12 A web-based triage decision support tool (eTRIAGE), based on CTAS, was developed in Canada and was shown to be reliable and valid.16 17

In Taiwan, the Department of Health and NHI have mandated the use of the Taiwan Triage System (TTS) since 1998. TTS is a four-level triage system that classifies patients using a combination of variables including vital signs, chief complaints and or diagnoses. The accuracy of TTS has been questioned in previous studies and has shown poor reliability and validity.18 19 TTS is reported to lack predictive ability regarding hospitalisation and intensity of medical resource consumption, and exhibits a propensity for overtriage.19

Given the limitations of the TTS, we modified the CTAS, with permission from the CTAS national working group, based on Taiwan ED needs and system context, and then undertook to validate the new five-level Taiwan triage and acuity scale (TTAS). We simultaneously created an electronic clinical decision TTAS support tool (eTTAS) for Taiwan ED use. The study objectives were to measure the interrater reliability of the new triage system, and to validate the new triage system according to ED resource utilisation and hospitalisation rates.


Settings and subjects

This prospective observational study was conducted from 10 October 2006 to 30 July 2007 at 33 study sites in Taiwan: 11 large urban academic medical centres with an average annual ED census of approximately 95 912 patients, 18 regional hospitals with 47 056 annual ED visits, and four district hospitals with 23 789 ED patients per year. We used a population-based convenience sample of patients triaged at the study sites between 12:00 and 22:00 hours within a 2-month period from March 2007 to May 2007. Patients under 17 years of age were excluded. The study was conducted in accordance with the requirements and approval of the institutional review boards of all the participating hospitals.

We first trained 33 senior triage nurses from each participating hospital to become TTAS instructors. These instructors were later involved in training and coordinating the study process in each hospital. Then, we trained another 142 nurses who were later involved in patient triage and data collection. All research nurses were required to triage at least 80 cases within the 2-month period according to the study protocol. In order to maximise the number of patient observations per study shift, we specifically chose the time interval between 12:00 and 22:00 hours for this study because this is when the ED is the busiest.

Study protocol

The new TTAS criteria and chief complaint list was developed by an 11-member emergency medicine expert panel (seven members from the Taiwan Society of Emergency Medicine and four from the Taiwan Association of Critical Care Nurses) that formed the TTAS national working group, using the CTAS and a literature review. Guidelines were adapted and modified by consensus whenever required to be pertinent to ED conditions in Taiwan. The content validity index for approved TTAS changes was between 0.80 and 1.0.

The TTAS retains most features of the CTAS, but includes some major modifications such as: (1) Shortened time to reassessment intervals (10/30/60/120 min). (2) Dividing the chief complaint list into two domains, non-trauma and trauma (the non-trauma domain is similar to that of the CTAS, and the trauma chief complaint list was categorised using anatomical region and environmental injury) with a total of 179 presenting complaints. The non-trauma system included 14 categories and 132 chief complaints and the trauma system included 15 categories and 47 chief complaints. (3) Explicit threshold levels for haemodynamic stability (eg, tachycardia/bradycardia (140 bpm/50 bpm) with/without symptoms of shock or a blood pressure <70 mm Hg). (4) Revised pain severity determination, with the deletion of chronic pain to accommodate ED conditions specific to Taiwan.

A computerised triage decision support tool, eTTAS, was then developed, linking a standardised list of national presenting complaints to preferred TTAS acuity levels based on TTAS criteria. The eTTAS was designed for point-of-care use to assist the triage nurse. The eTTAS provides the nurse with rapid access to the standardised presenting complaint list, which the user selects from. The application then displays a complaint-specific TTAS-based template with all relevant level 1–5 discriminators (modifiers) to assist the nurse in assigning the appropriate triage score. The eTTAS was designed so that the user can ‘override’ the system if the computer-generated triage does not reflect his/her clinical impressions; however, the reason for the override must be recorded and can be used as a quality assurance measure. The TTAS and eTTAS were then assessed by the team of 33 senior triage nurses (mean clinical experience 7.64 years; mean ED experience 5.73 years), each of whom underwent an 8-h workshop of training and critiquing, and later served as approved instructors. Content validity of the TTAS and eTTAS were assessed, and the expert panel then reconvened to reconcile new recommendations and criticism.

The TTAS and eTTAS were first pilot tested with 142 triage research nurses who received 4 h of TTAS education including 20 written triage scenarios. After the training, we then utilised patients presenting to the ED to test the interrater reliability of the eTTAS. Both instructors and research nurses were requested to use the eTTAS to triage 10 continuous patients who appeared in the ED. On ED arrival, patients were simultaneously triaged by the duty triage nurse using the TTS, and by both the instructor and research nurse using the eTTAS. Both the instructor and research nurse were blinded to each other with respect to triage assignment and triage scores assigned. One of the investigators monitored the paired triage assessment to ensure compliance and blinding. The pair/blinded TTAS ratings were then used to evaluate interobserver agreement.

When data collection began, all patients over 17 years of age presenting to the ED were triaged by an on-duty triage nurse using the current TTS, while research nurses used a laptop linked eTTAS and triaged simultaneously using the five-level TTAS. Both nurses were blinded to each other with respect to triage assessment and level assignment. During the study period, neither physicians nor the duty nurses were aware of the nature of the study and the study outcome parameters. All patients were managed according to their TTS triage assignment under routine ED practice.

Outcome measures

Patient demographics and study endpoints for each patient including length of stay (LOS), medical resource consumption and admission status were collected for analysis. Admission was defined as inpatient admission, died in ED, or critically ill transfer. LOS was defined as the total LOS in the ED. The ESI resources categories and criteria were used for scoring the medical resource consumption.20 Medical resource consumption was defined as a medical consumption frequency score, with 0 indicating no resources used and 12 indicating that there were 12 instances in which resources were used. To avoid the overestimation of resources used due to prolonged stay in the ED, all estimations of medical resource consumption in the present study were based on the number of medical procedures and examinations received by patients in the first 6 h in the ED.

Statistical analysis

The weighted κ (κw) statistic was used to assess interobserver agreement. The weighted κ adjusts for degrees of discordance in an ordinal scale. A weighted κ value of 0 corresponds to agreement no more than that expected by chance, whereas a weighted κ of 1 shows perfect agreement. Analysis of variance and the χ2 test were used to compare intersite differences. Continuous data were presented as mean±SD, while categorical data were presented as a number (percentage). To compare differences in LOS, resource utilisation and hospitalisation rates, the Kruskal–Wallis test was used, and data were given as median (IQR). Multiple comparisons were performed using the Bonferroni procedure with type I error adjustment when significant differences were observed. All statistical assessments were two-sided and considered significant if p<0.05. Statistical analyses were performed using SPSS 15.0 statistics software.


Interrater reliability

Table 1 shows the interobserver agreement between the instructors and research nurses in assigning a TTAS triage score. A total of 1420 combination was collected; however, 17 were excluded because of incomplete data. This left a total of 1403 combinations to be included in the analysis. Among the 1403 combinations, 1222 combinations (87%) were in perfect agreement, 157 combinations (11.2%) exhibited a one-level difference and 24 combinations (1.7%) exhibited a two-level difference. The analysis results indicate that the differences in determination of triage level between the instructors and the training nurses were not significant. Overall, the weighted κ of determination of triage level was 0.87 (95% CI 0.85 to 0.89), with a high percentage of agreement in category.

Table 1

TTAS interobserver agreement between triage instructors and study nurses

Comparison of basic characteristics of ED patients between hospital levels

A total of 12 300 adult patients was enrolled in this study; however, 1767 were excluded because of incomplete data. Therefore, 10 533 patients were included in the analysis: 5618 patients (53.3%) from academic medical centres, 4252 (40.4%) from regional hospitals and 663 patients (6.3%) from district hospitals. The patient characteristics, triage scores, outcomes and medical resource consumption profiles are shown in table 2. Of the patients, 26.3% were categorised as trauma and 73.7% as non-trauma. The overall median LOS and medical resource consumption were 2.2 h and 4, respectively. The overall distribution of patients in TTS levels 1, 2, 3 and 4 was 7.8% (826/10 533), 44.3% (4663/10 533), 45.8% (4828/10 533) and 2.1% (216/10 533), respectively, whereas 3.9%, 17.2%, 41.4%, 25.0% and 12.6% of patients were assigned into TTAS levels 1, 2, 3, 4 and 5, respectively. Level 3 was the most common TTAS and TTS level. As shown in table 2, 51 patients died (0.5%); 30 deaths occurred in academic medical centres, 19 in regional hospitals and two in district hospitals. There were significant differences in the distribution of patient categories (non-trauma, trauma), TTS and TTAS levels, disposition, LOS and medical resource consumption among the academic medical centres, regional hospitals and district hospitals (all p<0.001). Patients in the academic medical centres had higher admission rates, longer LOS and greater medical resource consumption than those in the other hospitals.

Table 2

Patient characteristics between hospital levels (N=10 533)

Redistribution of TTS to TTAS

Table 3 compares the distribution of patients by four-level TTS to five-level TTAS acuity score. Many of the TTS level 1, 2 and 3 category patients were widely redistributed using the five-level TTAS. The overall distribution of patients applying the four-level TTS was heavily weighted in favour of levels 2 (44.3%; 4663/10 533) and 3 (45.8%; 4828/10 533), with levels 1 and 4 only comprising 7.8% (826/10 533) and 2.1% (216/10 533), respectively, of patients. Applying the five-level TTAS led to a much more normative distribution from levels 1 to 5, ie, 3.9%, 17.2%, 41.4%, 25.0% and 12.6%, respectively.

Table 3

Triage distribution comparing TTS to TTAS

Relationship between TTAS level and LOS, medical resource consumption and hospitalisation rate

Table 4 shows the LOS, medical resource consumption and hospitalisation rate associated with each TTAS level. Regardless of hospital type, the same LOS, medical resource consumption and hospitalisation rate trends, based on TTAS acuity level, were appreciated (figure 1A–C). These differences reached statistical significance between all TTAS categories (all p<0.001). TTAS level 2 patients had the longest LOS (median 3.2 h; IQR 1.8–8.4); however, TTAS level 1 patients had the highest medical resource consumption (median 8; IQR 6–12) and hospitalisation rates (92.4%; 375/406).

Table 4

The relationship between TTAS level and LOS, medical resource consumption and hospitalisation rate (N=10 533)

Figure 1

(A) Length of stay, (B) medical resource consumption, and (C) hospitalisation rates stratified by Taiwan triage and acuity scale (TTAS) level and hospital type.


Triage accuracy is important to the safe and efficient operation of the ED by allowing for prioritisation of treatment based on acuity. With significant overcrowding in ED, the need for accurate triage is even more important.4–8 In order for triage to be effective, the triage scale must be reliable, valid and easy to use. One problem with traditional triage methods is their reliance on memory, which is often flawed by a lack of time and the inability to recall the guidelines. In a busy ED, and especially during a busy shift, triage nurses cannot be expected to refer to paper documents for reference, or accurately recall the entire contents from memory, leading to subjectivity and inconsistency in the triage process.9 In Taiwan, computerisation in the ED is becoming prevalent and programs continue to evolve, thus during the development of the TTAS criteria, the national working group had also decided to design a decision support tool in order to standardise the triage process and assist triage nurses in assigning the appropriate triage score. These decision support tools are not intended to replace clinical judgement and the triage nurses are not expected to become totally dependent on these tools. The goal is to develop convenient and valid tools for users, and any feedback could be used for future modifications.

The study results support the premise that computerised TTAS can facilitate the assignment of reliable five-level triage. The levels of TTAS agreement between the triage instructors and research nurses were excellent (κw 0.87), consistent with rates observed in CTAS studies using written standardised case-based reliability.8 11 21 It also compares favourably with CTAS studies also using electronic decision support. Grafstein et al,22 using the same paired observer methodology as our TTAS study, reported a κw agreement of 0.75. The studies by Dong et al23 using eTRIAGE and two independent blinded ED triage nurse assessments achieved lower agreement κw 0.65; however, in applying the tool to written cases achieved very high agreement κw 0.91.24 Differences may be due to the training of research and triage nurses, study methodology and reporting differences; however, overall there is a consistent level of reliability, which our study equalled or surpassed. We believe that the high κ agreement achieved with CTAS was maintained or exceeded because the TTAS content changes were based on a high level of agreement by the members of an expert panel. In addition, the complaint-oriented triage process permits the development of a highly functional electronic decision support system, supporting both education and triage at the point of care.

The triage criteria in the TTAS contrast with the lack of explicit clinical descriptors and the use of vital signs for each triage level in the TTS, which may help explain the overtriage of non-urgent patients by the TTAS (only 2.1% non-urgent triaged by TTS vs 12.6% non-urgent triaged by TTAS). The TTAS exhibited better discrimination and triaged patients more accurately, as shown in table 3. The effect of the TTAS was to discriminate more accurately high, intermediate and low acuity patients, and direct the more seriously ill to receive prompt treatment, thus improving patient safety. With better knowledge of triage acuity level-based volumes stratified by presentation type, it is possible to optimise patient placement and nurse and physician utilisation more efficiently to deliver the most appropriate care to each group of patients based on available resources. This does not mean that treatment for TTAS level 4 and 5 patients will be unnecessarily delayed, but rather streamed to the most appropriate care area and provider.

The validation of a triage system requires that it not only be consistent with medical needs, but also that it leads to predictable outcomes, including morbidity, mortality, resource utilisation and cost.7 15 17 25 Studies of the Australasian triage scale have demonstrated correlation with admission rates and ED costs.26 27 One validity study of the Manchester triage scale demonstrated the ability of that scale to detect critical patients on arrival.28 In validating the CTAS, several studies have shown that the CTAS is a valid instrument for predicting admission rates, hospital LOS and diagnostic utilisation.17 29 30 Dong et al17 have demonstrated that eTRIAGE (a web-based triage decision support tool) has excellent predictive validity for ED resource utilisation and hospital costs. These results are important. To our knowledge, the current study is the first to evaluate the validity of an electronic triage decision support system in multiple and different level hospitals.

Our study confirmed a significant correlation between TTAS acuity level, hospitalisation rate, LOS and medical resource consumption. The trend was observed regardless of the hospital level. Both hospitalisation and LOS were chosen because they are unambiguous and were readily available from the hospitals' information system. Because we had difficulty in obtaining the total cost of all patients in each participating hospital, we used the ESI resources category to record the medical resource consumption in each TTAS triage category, because ESI has a clear definition and had been proved to be useful in resource estimation.20

It is worth noting that these are all surrogate markers, because there is currently no ‘reference standard’ measure of triage scale validity.17 25 31 Higher acuity patients generally receive more interventions and have higher hospitalisation rates, justifying the more rapid placement into treatment areas and more rapid physician assessment. There are some instances in which high acuity patients were discharged with a shorter LOS, and low acuity patients were admitted with a longer LOS after treatment. For example, a patient with a drug overdose and severe respiratory distress would be categorised as TTAS 1 because of the need for immediate respiratory support; however, after treatment and observation, the patient may be discharged with limited investigations. By contrast, a patient with vague abdominal pain, TTAS 3 or 4 (based on pain severity), may prove to be appendicitis. The patient may require a prolonged ED stay that includes repeat laboratory and imaging studies (eg, CT scan), consultations, and ultimately hospital admission for surgery, despite the less acute TTAS triage score. The ability to identify and measure the relationship between patient acuity level and outcome depends not only on the measurement of the surrogate outcome marker, but also very importantly on confounding factors such as patient type and complexity, availability of inpatient beds, patient volume, rate and surges of patient presentation, and efficiency of care provided. These may all affect the surrogate outcome marker.31 Therefore, when evaluating acuity and resource utilisation, it is important to compare patients based on their type of presentation (taking advantage of the TTAS complaint list), and also using the data to compare individual sites on a yearly basis to look for changes or outliers. This recognises local case mix variation (looking not only at acuity but also complexity), and also local medical practice variation on resource intensity. Taking advantage of the presenting complaint granularity, however, it may be valuable to study complaint-specific similarities and variations by regions, hospital types and even nationally. as an opportunity to identify operational efficiencies based on outcomes.

The use of eTRIAGE has been shown to exhibit good interrater agreement, a reduction of the subjective down or uptriage of patients based on current ED workload, and facilitates accurate triaging by preventing the subjective recall of criteria of paper-based triage systems.9 16 In addition, acuity, as determined by eTRIAGE, has been shown to demonstrate excellent predictive ability for resource utilisation and hospital and ED costs.17 In the current study, we found the eTTAS to have the same advantages observed in studies of eTRIAGE. From a quality improvement perspective, an electronic triage tool will allow monitoring of TTAS guidelines and facilitate changes and their implementation. All sites can be updated simultaneously, maintaining standardisation. Furthermore, a standardised triage decision support tool allows for site-to-site and region-to-region validation and benchmarking.

The present study has a number of limitations that should be considered. The study cohort was non-randomised and only included adult patients. The study nurses took samples between 12:00 and 22:00 hours during a 2-month period, thus the category distribution in the present study cannot represent the distribution of triage categories in every hospital during a 24-h period. Because of convenience sampling, the nurses may have chosen patients in whom it was easy to perform triage or data collection. This may have altered the distribution of chief complaints compared with what may actually be encountered in the ED. The estimation of resource consumption in the present study was based on the number of medical procedures and examinations received by patients in the first 6 h in the ED. We did not know the total medical expenditure of the patients, and we did not estimate the time and manpower used for administering care to patients in the ED. The study result cannot thus entirely represent the actual medical resource consumption, but can only be considered as a reference. Further validation studies comparing the TTAS on inhospital outcomes such as mortality, length of inpatient stay, time in the intensive care unit, and also validation studies such as benchmarking with other measures of severity (eg, injury severity score) as related to the TTAS are warranted. Paediatric patients were not included in the present study. Therefore, further study is needed to determine if the TTAS is applicable to paediatric patients.


This study demonstrated that the computerised TTAS appears to be a reliable and valid triage system applicable to any ED setting in Taiwan. The eTTAS can increase triage reliability and is a promising method of defining reliable ED case mix groups. The eTTAS also exhibits significant correlation between TTAS scores and admission rate, LOS and resource utilisation among ED patients. Further research on validity, as it relates to complaint-specific medical outcomes, is recommended during the implementation of this novel five-level computerised triage system in Taiwan.


The authors want to thank the CTAS national working group for granting permission to make modifications to the CTAS, permitting them to create the TTAS. The authors also want to express their gratitude to Professor MJ Bullard for his advice and help in editing the manuscript.The authors also acknowledge the hard work of the other members of the TTAS national working group, Wai- Mau Choi (TSEM), Hung-Jung Lin (TSEM), Yu- Che Chang (TSEM), Fong-Dee Huang (TSEM) and Chin Yuan Tsan (TACCN), in developing, supporting and implementing the TTAS.



  • * Includes representatives from the Taiwan Society of Emergency Medicine and Taiwan Association of Critical Care Nurses.

  • Funding This study was supported by a grant from the Taiwan Department of Health.

  • Competing interests None.

  • Ethics approval The study was conducted in accordance with the requirements and approval of the institutional review boards of all the participating hospitals.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • Primary survey
    Malcolm Woollard