TERN Top Papers February 2022

Outcomes in children advised to attend the Emergency Department by a healthcare telephone advice service (NHS111) ¹

This research letter describes a single centre prospective cohort study comparing demographics and outcomes for children who self-present to the Emergency Department (ED) against those advised to attend by the health service telephone advice line (NHS 111).

2574 consecutive children attending an urban paediatric ED in a paediatric hospital were enrolled and data was collected and validated against that held by the North of England Commissioning support team.

The majority (72%) of children referred by NHS 111 had no investigations other than urine dipstick or blood glucose and no treatment beyond simple dressings, whilst 46% received only advice and just 3% were admitted. Between children referred by NHS 111 and those that self-presented there was no significant difference in the proportion assigned to higher triage categories or those that were deemed suitable for primary care.

As a single centre study based in an urban ED located in the only paediatric hospital in the region the caseload may not be representative of less specialist centres. Data was only analysed on those children who were referred to ED. It would be interesting to know what proportion of the patients who NHS 111 deemed did not require ED care did ultimately require an ED attendance.

As the researchers suggest, there may be scope for the NHS 111 system to triage more children to non-emergency care. However, further research is needed to identify the sensitivity of the NHS 111 decision tool for children requiring ED care to ensure that a change to the triage system would not result in unacceptably high numbers of unwell patients failing to be referred to ED.

Bottom Line

The majority of children advised to attend ED by NHS 111 did not require anything more than basic investigations or treatment

Reference

Ramlakhan S, Nathan K, Eferakaya S.. Outcomes in children advised to attend the emergency department by a healthcare telephone advice service (NHS111). European Journal of Emergency Medicine. 2021; 28(4): 315316. doi:10.1097/MEJ.0000000000000799

Characteristics of frequent paediatric users of emergency departments in England: an observational study using routine national data ²

This study was an observational retrospective study, using data from Hospital Episode Statistics to look at all <16s attending type 1 EDs between April 2014 and March 2017. They removed duplicates (those with <25 mins between two attendances for the same patient) and deaths. They then broke the total down by age group, socio-economic status and ethnic group.

There was discussion around the definition of what makes a frequent attender, with this study using 4 or more attendances in any 12-month period as the definition.

They then used logistic regression to calculate the odds of being a frequent attender for each group.

There was a total of 3,272,864 attendances. 9.1% were classed as frequent attenders. Those most likely to attend frequently were more likely to be under-1 (highest odds ratio), male, and from deprived areas. They were most likely to be either Asian or of mixed ethnic background, less likely to be white and least likely to be black or other.

A major strength of this study is that it included all type-one EDs throughout England (which hasnt been done before) thus increasing its power and external validity. Its approach to defining frequent attenders is probably better than simply counting the number of attendances in a chosen year.

It didnt, however, make full use of its sample by making any comparisons between different areas. It was limited by being reliance upon HES data, and there was no information about comorbidities – this is clearly important. The 25-minute cutoff may not have eradicated all duplicates.

The paper highlighted the need to support parents to reduce avoidable frequent attendances. It shows the need for further studies to look at factors such as education, housing, cultural factors, geographical variations, proximity to the ED and access to primary care and comorbidities to point towards possible solutions.

Bottom Line

Paediatric frequent attenders were more likely to be under-1, male, from a deprived background and either of Asian or mixed ethnic background.

Reference

Greenfield G, Blair M, Aylin PP, Saxena S, Majeed A, Bottle A. Characteristics of frequent paediatric users of emergency departments in England: an observational study using routine national data. Emergency Medicine Journal. 2021; 38(2):146-150. doi:10.1136/emermed-2019-209122

Injured adolescents should they be treated as big kids or little adults? ³

This paper was a review of Trauma Audit Research Network (TARN) data between 2012-2018 aiming to compare the mortality rates between adolescents presenting to the different types of Major Trauma Centres (MTCs) at 30 days.

Trauma networks were established in England in 2012 with the designation of 11 adults, 5 paediatric and 11 mixed (adult and children) MTCs. Severely injured patients under the age of 16 years are triaged to paediatrics and those older than 16 years to adult MTCs.

According to the World Health Organisation and UNICEF, adolescents are traditionally considered to be between the ages of 10-19 years. However, acknowledging that physical and psychological development occurs up to the mid-20s and factoring in social changes such as delayed partnering, parenting and financial independence a more modern and appropriate definition may include the ages of 10-24 years.

Based on these age ranges, it is evident that adolescents can be seen across all 3 MTC designations.

From the sample selected, 30,321 patients were identified as being between the ages of 10-24.99 who had a primary transfer to a MTC between 2012-2018. This can be further divided to show that 54% attended a mixed MTC, 38% to adult and 8% attended a childrens MTC.

After adjusting for confounders such as Injury Severity Score (ISS), mechanism of injury and patient physiology, it was noted that the 30-day mortality was higher in the mixed (4.4%) and adult (4.9%) MTCs compared with the paediatric MTCs (2.5%, p <0.0001)

It should be noted however, that those attending a paediatric MTC had lower ISS and comorbidities than those attending other centres. When patients were deemed unstable, they were conveyed to the nearest hospital (more likely to be an adult/mixed centre) than a paediatric MTC.

Bottom Line

Adolescents are challenging to manage for multiple reasons. On scene and transfer medical assistance to the correct centres may reduce mortality, but further work needs to be done in this field.

Reference

Leech C, Jenner R. Injured adolescents should they be treated as big kids or little adults? Emerg Med J. 2021; 38(7): 486-487. doi:10.1136/emermed-2020-211105.

Utility of specific laboratory biomarkers to predict severe sepsis in paediatric patients with SIRS ⁴

This was a retrospective cohort study that aimed to establish the correlation between defined laboratory biomarkers and children with and without severe sepsis.

It was completed in the Emergency Department of a US specialist childrens hospital over June 2018 to June 2019.

Patients were identified via the hospital database using Systemic Inflammatory Response Syndrome (SIRS) alerts in their records. Biomarkers studied were CRP, ESR, lactic acid and procalcitonin.

4349 patients triggered a sepsis alert. 1207 had blood tests including at least one of the relevant biomarkers sent within 6 hours of their ED arrival.

Of the 1207 patients who had blood tests, 100 achieved criteria for severe sepsis and 41 of these 100 had positive blood cultures.

The authors found that of the 100 patients with severe sepsis, patients were more likely to have positive cultures, to be admitted, to be younger and to have other co-morbidities.

Procalcitonin was the best biomarker for differentiating severe sepsis from SIRS. It demonstrated a 90% specificity for development of severe sepsis when over 2.72 ng/mL but less than 90% sensitivity when procalcitonin levels were below 0.08 ng/mL. The other biomarkers performed more poorly.

The area under the curve (AUC) values for all biomarkers in severe sepsis and culture-positive sepsis were 0.5-0.7.

Limitations included a small procalcitonin sample size of 370. The authors acknowledged this may have explained why their study showed no difference in AUC values for severe sepsis and culture-positive sepsis. This contrasts with the findings of previous studies in which procalcitonin was found to distinguish bacterial from viral infection.

Other limitations included potentially missed patients, as some patients with abnormal observations who fit SIRS criteria did not have blood tests. Additionally, the authors noted that few patients had all the relevant biomarkers sent.

Bottom Line

If higher sensitivity and specificity could be achieved, there is the potential for a decision tree that involves biomarker combination to allow more accuracy in identifying sepsis early. However, a larger study size to try and remove any confounding factors would likely be necessary.

Reference

Byler S, Baker A, Freiman E, Herigon J, Eisenberg M. Utility of specific laboratory biomarkers to predict severe sepsis in pediatric patients with SIRS. The American Journal of Emergency Medicine. 2021; 50: 778-783. doi:10.1016/j.ajem.2021.09.081

Evaluation of a novel approach to recognising community-acquired paediatric sepsis at ED triage by combining an electronic screening algorithm with clinician assessment ⁵

Sepsis in children can be difficult to identify, despite its high morbidity and mortality. Current UK paediatric sepsis screening tools are unvalidated and not widely used. This study aimed to develop a new triage-based screening tool that could be used in combination with clinical acumen.

The study was performed retrospectively at 2 UK paediatric emergency departments between 2016 and 2018. An Electronic Sepsis Recognition Pathway was used. This scored 6 areas:

1. General well-being
2. Temperature <36 C or >38 C
3. Capillary refill time >3s or HR above reference cut-off
4. Respiratory distress or breathing above reference cut-off
5. Irritability or confusion
6. Risk factors e.g. rigors or non- blanching rash

Each of these areas was scored either 0 or 1. A score >2 was classified as positive. If positive, the patient was immediately assessed by the most senior clinician available. Patients were then categorised as sepsis-confirmed or sepsis-excluded. Sepsis-confirmed patients received IV antibiotics whereas sepsis-excluded patients were seen in time order as per usual ED protocol.

All children >16 years who had been triaged were included in the study. The outcome of the Sepsis Recognition Pathway and the discharge diagnosis (taken from the ED database) were compared. The gold standard diagnosis of sepsis was children attending ED who received sepsis treatment and had a discharge diagnosis of sepsis.

99% of all eligible patients were screened and 19,912 children were included. 90 children had a discharge diagnosis of sepsis. The electronic Sepsis Recognition Tool identified 78 of these; a 86.7% sensitivity. It over-triaged 2573 patients providing a 87.0% specificity. This improved to a sensitivity of 90.0% and specificity of 99.4 with combined clinical assessment. The negative predictive value for the tool combined with clinical assessment was 99.9%.

The Sepsis Recognition Tool showed good accuracy to recognise sepsis at triage in the ED, which was improved further with combined clinician assessment. Further refinement is needed to avoid missing any sepsis cases as this can prove fatal in children.

This study was performed in 2 large secondary care hospitals with mixed rural and urban populations. Therefore, the results can be extrapolated to EDs with similar demographics.

There are a few limitations of the study. The utility of the screening tool relies on a senior clinician being available rapidly for assessment, but this is not always possible. 8 patients were missed and reattended with sepsis. The actual number might be higher as only 2 EDs were included so children may have reattended at other healthcare facilities. The study was underpowered as the sample size was not large enough. In order to have a sensitivity of 95%, a sample size of 104,000 was needed.

It is a retrospective study which means bias could be introduced when recording the data. In addition, all patients who were brought directly to resus were excluded as they had not been triaged. This could account for a lot of children with sepsis. Finally, the diagnosis of sepsis at discharge was a clinical decision therefore may not be consistent. Children may have improved after receiving intravenous antibiotics and therefore not considered septic at discharge, artificially reducing the specificity of the screening tool.

Bottom Line

The electronic Sepsis Recognition Tool in combination with clinical assessment was useful in identifying children with potential sepsis. It needs refining in order to improve its sensitivity.

Reference

Gomes S, Wood D, Ayis S, Haliasos N, Roland D. Evaluation of a novel approach to recognising community-acquired paediatric sepsis at ED triage by combining an electronic screening algorithm with clinician assessment. Emergency Medical Journal. 2021; 38(2): 132138. doi:10.1136/emermed-2019-208746

Early prediction of serious infections in febrile infants incorporating heart rate variability in an emergency department: a pilot study ⁶

Febrile infants (<3 months) present a diagnostic challenge in terms of differentiating those with simple viral illness from those with serious infection (SI).

Diagnostic difficulty has led to clinical prediction tools being developed, however their reliance on invasive testing means they do not help in triaging patients. This pilot study aimed to compare vital signs and heart rate variability (HRV) in babies with serious infection versus those without, assess performance of vital signs and HRV against current triage tools and compare HRV and vital signs to vital signs and blood biomarkers, in predicting SI.

This was a prospective pilot observational study in a tertiary paediatric centre in Singapore from December 2018 to November 2019 where standard practice is to hospitalise all febrile infants under 3 months of age.

They recruited a convenience series of infants under 3 months with fever over 38 at triage, during office hours to enable research support, and excluded those with arrhythmias, heart block and infants under 35 weeks gestation due to their increased risk of serious infection.

Severity Index Score (SIS), vital signs and a single lead ECG measured for 5 minutes were performed at triage. Additional information including length of fever, GBS at birth, lethargy, prolonged capillary refill time, respiratory distress, blood results (excluding procalcitonin as this was not routinely measured in the unit conducting the study) were collected, and patients were followed up by a research coordinator to confirm presence of serious infection.

Data analysis, comparing vital signs, HRV (time frequency and non-linear domains) and blood results between the 2 groups were done using appropriate statistical tests. They aimed to create a new prediction model for use at triage using multivariable regression.

There were significant differences in all 3 domains of HRV parameters between infants with versus those without SIs. In predicting SIs, gender, temperature and the HRV non-linear parameter Poincar plot SD2 (AUC 0.78, 95%CI 0.71 to 0.84) performed better than SIS alone (AUC 0.61, 95%CI 0.53 to 0.68). Model performance improved with the addition of absolute neutrophil count and C reactive protein.

There were some limitations to this study. It was a single centre so may not be widely applicable, or reproducible, and patient recruitment was non-linear which could introduce bias. There were 18 patients who did not have their HRV analysed due to artefact on the ECG and to mitigate this, they used imputation data which could introduce bias as well. They recognise this and are working on an improved prototype to reduce artefact from ECGs.

Bottom Line

It appears that HRV can be used to predict which infants may develop serious infection when incorporated into a risk-prediction model. Further work is needed to validate these results before such predictive tools can be used clinically.

Reference

Chong S, Ong GY, Allen JC, Lee JH, Piragasam R, Koh GZX et al. Early prediction of serious infections in febrile infants incorporating heart rate variability in an emergency department: a pilot study. Emergency Medicine Journal. 2021; 38(8): 607-612. doi:10.1136/emermed-2020-210675