Article Text

Safety of the Manchester Triage System to identify less urgent patients in paediatric emergence care: a prospective observational study
  1. Mirjam van Veen1,
  2. Ewout W Steyerberg2,
  3. Lizanne Lettinga1,
  4. Madelon Ruige3,
  5. Alfred H J van Meurs3,
  6. Johan van der Lei4,
  7. Henriëtte A Moll1
  1. 1Department of General Paediatrics, Erasmus MC–Sophia Children's Hospital, Rotterdam, The Netherlands
  2. 2Center for Medical Decision Making, Public Health, Erasmus MC, Rotterdam, The Netherlands
  3. 3Department of Paediatrics, Haga Hospital, Juliana Children's Hospital, The Hague, The Netherlands
  4. 4Department of Medical Informatics, Erasmus MC, Rotterdam, The Netherlands
  1. Correspondence to Henriëtte A Moll, Department of General Paediatrics, Room Sp 1540, Erasmus MC–Sophia Children's Hospital, PO Box 2060, 3000 CB Rotterdam, The Netherlands; h.a.moll{at}erasmusmc.nl

Abstract

Objective To assess hospitalisation rate as a proxy for the ability of the Manchester Triage System (MTS) to identify less urgent paediatric patients. We also evaluated general practitioner (GP) services to determine if they met patients' needs compared to emergency department care.

Methods Self-referred children triaged as less urgent by the MTS in two emergency departments in the Netherlands were included in a prospective observational study. Therapeutic interventions during emergency department consultation, hospitalisation after consultation and determinants for hospitalisation were assessed using logistic regression analysis.

Results During emergency department consultation, extensive therapeutic interventions were performed more often in patients with extremity problems (n=175, 19%) and dyspnoea (n=30, 15%). 191 (3.5%) of 5425 patients were hospitalised. Age and presenting problem remained statistically significant in multivariable logistic analysis, predicting hospitalisation with ORs of 3.0 (95% CI 2.2 to 4.1) for age <1 year, 2.5 (1.5 to 4.1) for dyspnoea, 3.5 (2.5 to 4.9) for gastrointestinal problems and 2.8 (1.1 to 7.2) for patients with fever without identified source compared to all other patients. 3975 (76%) of 5234 patients were contacted for follow-up after discharge. Six (0.15%) patients were hospitalised after emergency department discharge.

Conclusion In the MTS less urgent categories, overall hospitalisation is low, although children <1 year of age or with dyspnoea, gastrointestinal problems or fever without identified source have an increased risk for hospitalisation. Except for these patient groups, the MTS identifies less urgent patients safely. It may not be optimal for GP services to treat patients with extremity problems.

Statistics from Altmetric.com

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.

Introduction

Emergency departments are increasingly visited by patients without referral by a physician or with non-urgent problems.1 2 This contributes to high costs, increased use of diagnostic services, longer waiting times, full waiting rooms and higher work pressure for hospital personnel.3,,5

Triage aims to easily and quickly identify urgent patients and prioritise them to be seen by a physician. Triage systems can be also used to identify patients with less urgent problems who can safely wait for a longer time or can be seen by another caregiver such as a general practitioner (GP).6

What is already known on this topic

  • The Manchester Triage System is commonly used in Europe and shows moderate validity in children.

  • Safety in identifying less urgent patients in paediatric emergency care using the Manchester Triage System has not been previously studied.

What this study adds

  • Self-referred patients triaged as less urgent are rarely hospitalised except for children less than 1 year of age or presenting with specific problems.

  • The Manchester Triage System can identify less urgent patients safely, except for children less than 1 year of age or presenting with specific problems.

The Manchester Triage System (MTS) is a five-level triage system. The MTS suggests patients triaged as ‘standard’ (level 4) or ‘non-urgent’ (level 5) with non-traumatic injuries should be referred to primary care. This referral guideline was developed using expert opinion and the authors state that the efficacy of these processes must be proved rather than assumed.6

Referring patients to a GP can help solve the problem of increasing numbers of patients who present on their own initiative to emergency departments with less urgent problems.7 Consultation for less urgent patients in emergency departments is also costly since more diagnostic tests are ordered and prescriptions issued compared to GP consultations.3

The safety of identifying less urgent patients in paediatric emergency care has not previously been studied. The aim of the study was to assess hospitalisation rate as a proxy for the ability of the MTS to identify less urgent paediatric patients. We further assessed predictors for hospitalisation and also evaluated the ability of GP services to meet patient needs compared to emergency department care.

Methods

Study design

We carried out a prospective observational study. We assessed interventions performed in the emergency department, hospitalisation and determinants for hospitalisation in less urgent triaged children who visited the emergency department on their own initiative and were seen by a physician. Discharged patients received a routine telephone follow-up call 2–4 days later to assess hospitalisation after discharge. This study is part of ongoing research on the validity of the MTS in children8 9 and was approved by the institutional medical ethics committee.

Patients

All children between 0 and 15 years of age who visited the emergency departments of the Haga Hospital–Juliana Children's Hospital in The Hague between August 2007 and May 2008 and the Erasmus MC–Sophia Children's Hospital in Rotterdam between May 2007 and August 2008 were triaged using the MTS.10 We included patients triaged as MTS level 4 or 5 who presented on their own initiative to the emergency department (self-referred). Data on therapeutic interventions in the emergency department of the Haga hospital were only available for August 2007 to December 2007.

The emergency department of the Haga Hospital–Juliana Children's hospital is a mixed paediatric–adult emergency department in a large teaching hospital visited by nearly 30 000 patients annually of whom about 15 000 are children. The Erasmus MC–Sophia Children's Hospital is a university hospital with a specific paediatric emergency department visited by nearly 9000 patients each year. Both hospitals are in the southwest of the Netherlands, which has a population of approximately 4 million people and an annual birth rate of 47 000.11

Hospitalisation refers to inpatient bed hospitalisation. Patients who were observed for a period of less than 6 h were considered as discharged after emergency department consultation.

Manchester Triage System

Emergency department nurses performed a short assessment and triaged patients according to the MTS.6 10 The system is a flowchart-based algorithm and consists of 52 flowchart diagrams, of which 49 are suitable for children. These flowcharts are specific for the patient's presenting problem. They contain six key discriminators (life threat, pain, haemorrhage, acuteness of onset, consciousness level and temperature) as well as specific discriminators relevant to the presenting problem. Selection of a discriminator leads to one out of five urgency categories.

We used a adapted version of the MTS with specific modifications for children. The modifications were developed for patient groups which showed low validity in our earlier study. In this study, we evaluated MTS validity compared to a predefined reference standard of urgency.8

The modified MTS was used in 87% of cases to triage patients. The modification led to a shift of patients to less urgent categories. Patients were more often triaged into the ‘standard’ and ‘non-urgent’ categories (48%, n=5347/11 210) compared to the original MTS (41%, n=5552/13 554). Modifications were shown to improve the validity of the MTS in paediatric emergency care compared to a predefined reference standard for urgency. The specificity improved from 79% (95% CI 79% to 80%) to 87% (86% to 87%), but sensitivity did not improved significantly (63% (95% CI 59% to 66%) vs 64% (95% CI 60% to 68%)). Details of the modifications (M van Veen et al, submitted) are given in the online appendix.

Data collection

Nurses recorded patient characteristics on electronic forms when patients presented at the emergency department. Data on triage, interventions in the emergency department and hospitalisation were registered using the triage software package. Details of hospitalisation were extracted from medical files.

Telephone follow-up

A nurse and medical students followed-up patients using a standardised telephone questionnaire. Patients without a GP referral who were triaged as level 4 or 5, received a telephone follow-up call 2–4 days after their emergency department visit.

We attempted to contact parents each day until 96 h after their emergency department visit. Language barriers were overcome using a conference call with an interpreter. Patients were sent a written questionnaire if they could not be reached by telephone.

Data analysis

We categorised the selected MTS flowchart into nine different categories according to the present problems: skin problems (flowchart ‘rashes’), dyspnoea (flowchart ‘asthma’, ‘shortness of breath in children’), upper respiratory tract infection (flowchart ‘sore throat’, ‘nasal problems’, ‘ear problems’), gastrointestinal problems (flowchart ‘vomiting’, ‘diarrhoea’, ‘abdominal pain in children’, ‘gastrointestinal bleeding’), head injury (flowchart ‘head injury’), extremity problems or wounds (flowcharts ‘limb problems’, ‘wounds’ and ‘limping child’), other traumatic (flowcharts ‘falls’, ‘assault’, ‘torso injury’ and ‘major trauma’) and other medical. If the MTS ‘general’ flowchart or the MTS ‘worried parent’ flowchart was selected, we used the presenting problem as registered by the nurse and categorised these into one of the categories or to ‘fever without identified source’.6 10

Predictors for hospitalisation were data easily obtainable for triage, such as age, gender, MTS urgency level and presenting problem. Uni- and multivariable logistic regression analyses were performed. Since the relationship between age and hospitalisation may not be linear, we used a restricted cubic spline (RCS) function to model the relationship between age and hospitalisation.12 RCSs contain cubic (X3) terms which are restricted to be linear in the tails.13 In order to calculate clinical interpretable ORs, age was divided into categories and ORs were shown compared to patients aged 8–15 years.

The presenting problem was shown with ORs for the different categories compared to the category ‘extremity problems or wounds’.

The presenting problem categories and age groups with the highest ORs were selected to achieve the final model. ORs were shown for the different presenting problem categories and age groups compared to all other categories. SPSS 15.0 and R v 2.9.1 using the Design library (http://www.r-project.org) were used for statistical analysis.

Results

A total of 5425 children triaged as less urgent attended the emergency departments on their own initiative during the studied period. Of these patients, 191 (3.5%) were hospitalised (figure 1). Information on interventions performed during emergency department consultation were available for a selected period during which 3536 patients were seen.

Figure 1

Flow of patients seen in the emergency departments during the study period.

Interventions during emergency department consultation

Overall, extensive therapeutic interventions were performed in 11% (n=178) of patients with traumatic injuries and in 4% (n=83) of patients with medical problems.

Extensive therapeutic interventions were performed more often in patients with extremity problems (n=175, 19%) and in patients with dyspnoea (n=30, 15%) (table 1).

Table 1

Therapeutic interventions in the emergency department by age group and presenting problem, N=3536

Hospitalisation

Hospitalised patients had a median age of 1.5 years (IQR 0.4–4.4 years), 45% (n=86) were female, 91% (n=173) were triaged as urgency level 4 and 9% (n=17) as urgency level 5.

One patient (1%) was admitted to the intensive care unit. She was a 12-year-old girl with a history of a catecholaminergic polymorphic ventricular tachycardia, who presented with a syncope and an irregular cardiac rhythm but was haemodynamically stable.

Details of hospitalisation could be retrieved for 172 patients (90%). The median length of admission was 2 days (IQR 1–3 days). Overall, 45% (n=78) had a length of stay of less than 24 h, 20% (n=34) of between 24 and 48 h and 35% (n=60) of longer than 48 h. The main reasons for hospitalisation were risk of dehydration in 6% (n=31), head injury in 12% (n=20) and acute life threatening event in 9% (n=15).

Interventions were needed in 63% (n=109) and observation without intervention in 37% (n=63). The interventions included oral medication (n=34), intravenous therapy (n=24), urgent surgery (n=2), non-urgent surgery (n=28), oxygen (n=1), rehydration by nasogastric tube with oral rehydration solution (n=14) and inhaled medication (n=5). In 97 (56%) patients an urgent intervention (defined as intravenous therapy, oxygen or urgent surgery) was required or the length of stay was longer than 24 h.

Determinants of hospitalisation

Gender and urgency were not associated with hospitalisation (ORfemale 0.98, 95% CI 0.73 to 1.31, pWald=0.89; ORurgency 4, 0.96, 95% CI 0.57 to 1.61, pWald=0.87).

Patients with gastrointestinal problems (8%), dyspnoea (8%) and fever without identified source (6%) were more often hospitalised. Hospitalisation was also more likely for young patients (0–2 months (14%), 3–11 months (8%)) (table 2).

Table 2

Determinants for hospitalisation in less urgent, self-referred patients, N=5407*/5425

Figure 2 and table 2 shows the multivariate regression model with age and presenting problems as discriminators and hospitalisation as outcome. These predictors remained statistically significant in a final multivariable analysis with adjusted ORs of 3.0 (95% CI 2.2 to 4.1) for age less than 1 year, 2.5 (95% CI 1.5 to 4.1) for dyspnoea, 3.5 (95% CI 2.5 to 4.9) for gastrointestinal problems and 2.8 (95% CI 1.1 to 7.2) for patients with fever without symptoms compared to all other patients.

Figure 2

Probability of hospitalisation for patients presenting with skin problems (1), dyspnoea (2), upper respiratory tract infections (3), gastrointestinal problems (4), other (5), fever without identified source (6), head injury (7) and wounds or extremity problems (8), depending on age. The scatter diagram on the x-axis represents the density of age data.

In patients over 1 year old without dyspnoea, gastrointestinal problems or fever without identified source, 54 of 3738 (1%) were hospitalised after emergency department consultation. Details of these hospitalisations were available for 52 of the 54. In 19 out of the 52 (36%), an urgent intervention was required (intravenous therapy, oxygen or urgent surgery) or the length of stay was longer than 24 h.

Follow-up after emergency department discharge

Telephone follow-up contact was made for 76% of patients (figure 1).

Patients who could not be reached or who did not want to participate did not differ regarding median age (no follow-up, 4.2 years (IQR 1.9–8.7); follow-up performed, 4.4 years (IQR 1.7–9.1); Mann–Whitney U, p=0.75) but differed in presenting problem (χ2 test, p<0.001).

Patients who could not be contacted more often had gastrointestinal problems (17%, n=218/1247 vs 14%, n=558/3974) and fever without specific complaints (3.8%, n=48 vs 0.7%, n=29) and less often ‘other’ problems (19%, n=233 vs 23%, n=934).

In total, 301/3975 patients (8%) had an unscheduled follow-up visit, of which 65% were to the GP surgery and 34% to the emergency department. Six patients out of 3975 were subsequently admitted (0.15%). Details of these six hospitalisations after emergency department discharge are provided in box 1.

Box 1

Patients hospitalised after emergency department discharge (6 out of 3975 less urgent patients)

  1. Female 11 months, admitted because of vomiting, diarrhoea and dehydration (4%), rehydration with oral rehydration solution by nasogastric tube, discharge after 5 days, follow-up visit scheduled

  2. Male 10 months, upper respiratory tract infection with otitis, not drinking, not able to swallow antibiotics*

  3. Male 5 years old, heavy abdominal pain after fall on abdomen, observation, discharged after 1 day, follow-up visit scheduled

  4. Female 5 months, suspicion of dehydration, follow-up visit was planned (1 day), but patient presented again on own initiative on the same day and was admitted because of rehydration, administered oral rehydration solution and discharged after 1 day, follow-up visit scheduled

  5. Male 2 years old, admitted because of pneumonia and inability to take oral antibiotics, antibiotics by nasogastric tube, discharge after 2 days, follow-up visit scheduled

  6. Female 6 years old, fever and rash, some petechiae, intravenous antibiotics, observation during admittance, discharge after 3 days, no scheduled follow-up

  • * Details of hospitalisation were unknown for this patient.

Discussion

Self-referred patients triaged as less urgent are rarely hospitalised except for children less than 1 year of age (10%) or presenting with dyspnoea (8%), gastrointestinal problems (8%) or fever without identified source (6%).

Follow-up after emergency department discharge showed that only 0.15% (n=6/3975) of patients were hospitalised. Referring less urgent children to another caregiver may be implemented safely excluding these specific patient and age groups.

One less urgent patient presenting with an arrhythmia was admitted to the intensive care unit. The MTS does not contain a specific discriminator for patients with arrhythmia,14 but we argue that these children should be considered to be very urgent.

The MTS is a commonly used triage system in and outside Europe6 which demonstrated good reproducibility in children and moderate reproducibility in children and adults.9 15 Our earlier study showed moderate validity in paediatric emergency care. We developed and implemented modifications resulting in improved validity of the MTS, with better specificity in particular and more patients triaged as less urgent (M van Veen et al, submitted).

How safely less urgent patients can be identified has only previously been studied for one triage system. The study found that 4.4% of less urgent (level 4 or 5) patients (n=7116) identified using the Canadian Triage and Acuity Scale were hospitalised and 3.1% of children (n=383).16 The authors concluded that the system is not valid to identify less urgent patients to free up emergency departments. We found a lower hospitalisation rate of 1% for self-referred, less urgent patients with selected problems.

Several solutions have been developed to safely reduce the numbers of patients attending emergency departments. In the USA, out-of-hours call centres function as gatekeepers and screen patients who want to attend the emergency department, treating less urgent patients in a fast track area.17,,19 In the UK, GPs are located in emergency departments to attend to less urgent patients.20 21

Guidelines advising which patients should be to referred to GPs should be further evaluated in future research.

In light of our results, when the MTS is used to refer less urgent patients to GPs, we are mostly concerned about the 1% (n=54/3738) hospitalised after an emergency department visit, especially the 0.51% (n=19/3738) who require interventions or a length of stay of more than 24 h. This hospitalisation rate is relatively low compared to the higher MTS urgency categories. The proportion of hospitalisations for self-referred patients was 44% (n=28/64) for MTS level 1, 31% (n=171/558) for MTS level 2 and 10% (n=178/1836) for MTS level 3, respectively (based on data from our previous validation study of the modified MTS) (M van Veen et al, submitted).

If less urgent patients are actually referred to a GP, the 1% of patients hospitalised in this study will probably be referred back to the emergency department after GP consultation. To minimise any possible delay in treatment, patients should be referred to a GP on the same day. However, the time window in which patients should be seen by a physician should be evaluated in future research.

In patients with extremity problems, extensive therapeutic interventions which are not available in the GP setting were frequently performed. Since hospitalisation is rare in these patients, we suggest they should be referred to a fast track area within the emergency department for treatment.

At integrated emergency care services in the Netherlands where a GP out of office centre is combined with an emergency department at the same location, patients are triaged and referred to either the GP or the emergency department. The introduction of a triage system resulted 10–53% of patients being transferred from the emergency department to the GP.22,,24 Hospitalisations in adults and children decreased by 34% and after discharge fewer patients needed follow-up, compared to the situation before the introduction of the integrated emergency care services. However, that study did not evaluate the effect on unscheduled follow-up visits or hospitalisations after discharge to identify adverse effects, neither was the focus on children.24

Limitations

In this study we focused on the safety of referring less urgent children to primary care. We studied those patients who could have been referred but were actually treated in the emergency department and showed that referral might be safe for a large group of patients. However, in order to evaluate the actual effect of referral, patients should be referred and the effects on safety then evaluated.

Telephone follow-up was used to identify any adverse events due to emergency department discharge. Overall, 76% of patients (n=3975/5234) could be contacted for a telephone questionnaire. Failure to contact patients was due to incorrect phone numbers or absence of the patients from home when we phoned. It is possible that the selection of contacted patient might have influenced the results. Relatively fewer patients with gastrointestinal problems could be contacted which could have led to a slight underestimation of the percentage of unscheduled hospitalisations.

On the other hand, by using a telephone call instead of the hospital information system to track revisits, we identified all revisits to the same or other emergency departments and to the GP. Compared to one other study with a telephone follow-up (response 46%) our response rate was high.18

To assess therapeutic interventions we selected a time period from the total database and assume that selection of these patients did not cause selection bias; therefore the results on therapeutic interventions can be generalised to the total study population.

Conclusion

The overall hospitalisation rate is low in the MTS less urgent categories, however children below 1 year of age or with dyspnoea, gastrointestinal problems or fever without identified source have an increased risk for hospitalisation. Therefore, except for these patient groups, the MTS can identify less urgent patients safely. It may not be optimal for GP services to treat patients with extremity problems.

Acknowledgments

The authors thank Badies Manai, MSc, research nurse for organisation of telephone follow-up and making most calls and Saskia de Vries, emergency department nurse, Marieke van Hemert and Csila de Knijff, both medical students, for their cooperation in making the follow-up phone calls.

References

Supplementary materials

  • Web Only Data adc.2010.199018

    Files in this Data Supplement:

Footnotes

  • Funding This study was funded by the Netherlands Organization for Health Research and Development (ZonMw) and Erasmus MC, Rotterdam, The Netherlands.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of Erasmus MC, Rotterdam, The Netherlands.

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