In the emergency department, triage is a formal process of immediate assessment of all patients seeking emergency care. A number of triage scales are used internationally, the most salient of which are the Manchester Triage System (MTS),1 the Emergency Severity Index (ESI) used predominantly in the USA2 and the Canadian Emergency Department Triage and Acuity Scale (CTAS).3 The MTS is a 5-point triage scale which was developed using a consensus approach by the Manchester Triage Group, a multidisciplinary group of emergency care experts, to triage patients coming into the emergency department.1 The MTS was introduced in the UK in 1996 and is now used universally across the UK and many hospitals in other EU countries.4 Table 1 shows the five MTS categories.

Despite its widespread use, to our knowledge there are no published studies validating the MTS, although it has been recently compared against the ESI and was found to have better triage accuracy.2 In terms of triage practice, this means that the consensus opinion of the Manchester Triage Group has not been verified in relation to clinical urgency,3 with the exception of its predictive value for entry into intensive care units by Cooke and Jinks.4

Triage assessment findings are used to prioritise patients on the basis of illness severity and the need for medical care.5 An effective triage system thus aims to ensure that patients in A&E departments get the urgent care in the right priority and also that their subsequent care is allocated appropriately to the degree of illness,5 which often can imply being admitted into hospital as an inpatient for further diagnosis and treatment. It is easy to calculate the chance of this outcome for the total number of patients coming into the emergency department, but no study has investigated the frequency distribution for subsets of different patients according to the severity of their presenting symptoms. Intuitively it seems there would be an association between the severity (regardless of disease group specificity) and hospital admission. To our knowledge this has never been shown. Evidence from a study on a Canadian eTRIAGE system seems to suggest that it “demonstrates excellent predictive validity for resource utilization and ED and hospital costs”,3 but the authors did not focus on admission rates.

MTS was first introduced in Portugal in 2000 in two large hospitals, of which one is our institution. The aim of this study was to determine whether the subgroups created by the application of the Manchester Triage Protocol (ie, illness severity at presentation) were associated with different propensities for indirect triage outcomes such as death in the A&E department or being admitted into hospital.

METHODS

Data were collected from the adult emergency department at Hospital Fernando Fonseca, Lisbon. This excluded patients entering directly into the obstetrics and gynaecology sections as well as paediatric patients defined as those below the age of 16 years. Our department has been using the MTS since 2000, applied by nurses using a computerised protocol. The database covers about 128 000 patient visits per year from a population of about 700 000. This means our A&E department has a significantly higher throughput than most UK and non-UK hospitals, of which 91.7% are self-referrals.

Patients coming into the department undergo triage, with a small proportion being assigned a “White” code if in possession of a reference letter from an in-house doctor referring them to a certain colleague or area for observation (for methodological consistency, these patients were excluded from the study). All others undergo the classic Manchester Triage Protocol before being sent to an observation area. After being triaged, patients can be sent for observation to medical specialities (general medicine, psychiatry and neurology) or surgical specialities (general surgery, minor surgery, ophthalmology, ear, nose and throat and orthopaedics).

We excluded the first few years of data as the learning phase for the professionals and the adjustment phase for the organisation could have influenced the results. Data were considered from January 2004 to the end of June 2007, a 30-month period during which admission practice was stable, the use of the computer system and the MTS was a daily routine. This corresponded to a total of 321 539 patients triaged. Data from the department database was transferred into Microsolf Excel spreadsheets and analysed using EpiInfo and SPSS software. A univariate analysis was preformed. For analysis purposes, codes “Red” and “Orange” were grouped together into a “high priority” cluster and codes “Yellow”, “Green “ and “Blue” were grouped together into a “low priority” cluster. Table analysis used the χ² test with a confidence interval of 0.05 for one-tailed analysis and 0.025 for two-tailed analysis.

Admissions to the emergency department 24–72 h observation room, hospital floors and intensive care units were considered for inclusion in the study. One group of patients was triaged alive and admitted to die in the A&E department. They either entered the A&E resuscitation room directly or were sent to the resuscitation room from the location where they were waiting to be observed due to deterioration in their clinical condition. These patients were considered as admissions in their respective initial MTS colour categories and they would almost certainly have been admitted if resuscitation had been successful. They represented a very small number of patients; only in the “Red” category did this number represent more than 2.5%.

The following hypotheses were investigated:

• Hypothesis 1: The different subgroups created by applying a triage system like the MTS are associated with different propensities for being admitted into an acute care hospital and staying in hospital for more than 24 h. As such, patients triaged as higher priority (Red + Orange) are admitted into hospital in a higher proportion than those triaged as lower priority.

• Hypothesis 2: The associated degree of priority at triage is directly associated with short-term mortality (defined as death occurring within the emergency department, usually during the first 24–48 h).

• Hypothesis 3: There are differences in the MTS categories admitted by medical and by surgical specialities.

RESULTS

Table 2 summarises the findings of the study.

Descriptive statistics

Of a total of 321 539 patients who presented to the department, 4917 (1.53%) came with a reference note and were assigned the “White” code and did not go through the MTS. Of these, 615 (12.5%) were admitted into hospital, representing 2% of total admissions. The remaining 316 622 patients where triaged according to the MTS into “Red” (0.74%), “Orange” (24.78%); “Yellow” (50.65%); “Green” (20.28%) and “Blue” (2.00%) categories. For each category, table 1 shows the number of patients admitted, broken down into those who died in the department, those admitted by medical specialities and those admitted by surgical specialities. The percentage of patients admitted in each MTS code was 31.8%, 22.0%, 6.9%, 1.8% and 1.4%, respectively. The means for each variable were all smaller than or equal to the standard deviation, so the median values were used for analysis. The median number of patients in each MTS code was 65 226, the median number of deaths was 6 and the mean percentage of admissions was about 7%. Figure 1 shows the contribution of the different MTS codes to the total number of hospital admissions. The weight of each code in the total of admissions is very different from the weight of each code in the number of patients being triaged. Patients triaged by the MTS accounted for 98% of the total admissions, with only 2% of elective admission (“White” code) via the emergency department.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Admissions to an acute hospital stratified according to the Manchester Triage System (MTS) code on arrival in the emergency department.

Statistical analysis

A preliminary χ² test of all isolated MTS codes and the proportion of admissions showed that clustering for priority would be meaningful. “Red” and “Orange” codes were clustered into a “higher priority” group and “Yellow”, “Green” and “Blue” were clustered into a “lower priority” group. Application of the χ² test then showed that there was a clear association between the priority group and the number of deaths (χ² = 756.67; p<0.001), whereby the higher the priority the higher the number of deaths (difference of about 39-fold). Equally, there was a clear association between the priority group and the proportion of patients admitted into hospital (χ² = 15320.41; p<0.001). This ranged from 31.8% for patients triaged as “Red” down to 1.4% for those triaged as “Blue”. On average, patients in the high priority group were four times more likely to be admitted than those triaged into the low priority group.

Medical specialities admitted about 3.4 times more patients than surgical specialities. It was also evident that patients in the high priority group were twice as likely to be admitted by medical specialities as those in the low priority group, whereas the opposite was true for surgical specialities which admitted 1.6 times more patients in the low priority group than those triaged as high priority.

DISCUSSION

The findings of this study clearly support the hypothesis that the different MTS subgroups are associated with different propensities for admission, with the higher priority groups being admitted up to five times more often. Although these findings could be reduced by the additive effects of other variables such as age and gender that are not part of the MTS, the magnitude of the χ² results is so high that these effects could be negligible. This is not a surprising finding, but one that has not yet been stated in the literature. This result is even more interesting since the decision to admit a patient is taken by the physician who sees him/her after triage. The practice in our department involves a joint discussion with a physician staffing the inpatient area and rarely is the MTS code mentioned.

The number of short-term deaths occurring inside the A&E department is clearly associated with the degree of severity as triaged by applying the MTS, with almost 10% of patients triaged as having an unadjusted risk of death. This risk also seems to increase exponentially as one moves from lower to higher priority groups. Although this was theoretically predictable and, in part, one of the reasons for implementing the MTS in hospitals with large numbers of patients coming into A&E, these findings provide evidence that this is the case and identify a 39-fold difference in the risk of death between patients triaged as high versus low priority.

Findings relating to the hypothesis that surgeons and physicians might admit patients differently perhaps provide the most puzzling and thought-provoking body of evidence. On first impression it seems that the MTS is associated more strongly and directly with admission via medical specialities than surgical specialities. Also, surgical specialities seemed to admit more low priority patients than high priority patients. This could partly be explained by the fact that our department is not a major trauma centre and thus potential “surgical” “reds” are never brought to us; however, it also seems there may be a bias in the MTS towards “medical” signs and symptoms away from those used by surgeons to guide their decision-making about admission. Another possible explanation is that surgical/orthopaedic problems are more painful but can be dealt with in the emergency department and the patient is then discharged; conversely, non-specific abdominal pain can be hard to diagnose, especially in elderly patients, who would therefore be admitted for investigation.

This study has several limitations related to the nature of the available data which is currently being changed so that multivariable analysis could not be applied. The fact that we cannot suggest some influence due to the knowledge about the pre-assigned MTS code by those admitting the patient weakens the argument for an independent association; however, this was not the aim of the paper as we were seeking to determine whether or not admission rates differ and, if so, whether this is associated with the MTS. While the large numbers in our sample help to dissipate statistical concerns, they make it almost impossible for individual case analysis until the hospital is fully computerised.

CONCLUSIONS

MTS provides information that extends beyond its immediate usefulness as a prioritisation mechanism in the emergency department. It appears to be a very powerful tool for distinguishing between patients with high and low unadjusted risk of short-term death as well as those who will stay in hospital for at least 24 h before being discharged. It therefore correlates with short-term mortality as well as showing a propensity for hospital admission. MTS is not an equally powerful discriminator when considering admissions via medical and surgical specialities. This may be due to the nature of its inbuilt discriminators. Further studies are needed to confirm these findings and to determine the effects of age, gender and social status. The relationship between MTS and length of stay, total in-hospital all-cause mortality and the probability of the patient requiring intensive care later during the course of his/her hospital stay rather than straight after the emergency admission is under investigation.