Breaking Evidence: The Rod Little prize 2019

“Development of a Clinical Decision Rule for the Early Safe Discharge of Patients with Mild Traumatic Brain Injury and Findings on Computed Tomography Brain Scan: A Retrospective Cohort Study”
By Carl Marincowitz et al. (ST4 in Emergency Medicine and Clinical Lecturer, University of Sheffields).

This work is now published in the Journal of Neurotrauma, DOI: 10.1089/neu.2019.6652

Thinking back over the last month, you’ve probably seen at least one person with a head injury presenting to your Emergency Department (ED). Head injuries are really common but the vast majority of patients we see are alert and well when we review them. Despite this, a large proportion will earn themselves a CT scan and around 7% will have a skull fracture or intracranial injury found on neuroimaging¹, NICE guidance currently recommend that any patients with “new, clinically significant abnormalities on imaging” need admission for observation². However there’s ongoing debate internationally about this. Do we really need to admit all patients with pathology on neuroimaging if they’re well and unlikely to need intervention, or are there groups of patients we can safely discharge home?

This study aimed to develop a clinical decision rule to identify patients with abnormalities on neuroimaging that could be safely discharged from the ED. The authors conducted a retrospective case note review of all patients presenting over a 7 year period to 3 major trauma centres in England. Patients were included if they had an initial Glasgow Coma Scale (GCS) score of >13 and had acute injuries on imaging. The notes were then reviewed to see if the patient developed any clinically important deterioration or required admission to a neurosurgical unit within 30 days. Overall 1,699 patients were included in the analysis. Of these, 27.7% experienced a clinical deterioration and 13.1% required admission to a neurosurgical unit. A risk model was derived using multi-variable logistic regression which suggested that patients who are fully conscious with either simple skull fractures or intracranial bleeding <5mm could be safely discharged from the ED. The sensitivity and specificity of the model was good, with sensitivity of 99.5% (95% CI, 98.199.9) and specificity of 7.4% (95% CI, 6.09.1) for clinically important deterioration. This needs validation in a larger cohort but shows excellent promise for the early discharge of low risk patients in the future.

Carl’s take home message:

The prevalence of clinically important deterioration in this this mild traumatic brain injury population was higher than previously reported, with over a quarter of patients deteriorating.

It was possible to develop a highly sensitive decision rule to select patients to be safely discharged from the ED, but at the cost of specificity. Future inclusion of novel predictive factors such as biomarkers may improve the performance of the derived model. Further external evaluation and implementations studies are needed before it is used in the ED.

Contact Carl about this work:

Contact cm444@doctors.org.uk if you are interested in this research and would like to know more.

“Computer beats doctor? Estimating the probability of Acute Coronary Syndrome for individual patients”
By Govind Oliver et al. (ST4 in Emergency Medicine at Royal Preston Hospital and Academic Research Fellow at the University of Manchester)

One of the big worries with patients presenting with chest pain is whether or not they have an acute coronary syndrome. Refer on or discuss a patient with chest pain and the question often asked is “and do you think it’s cardiac in nature?” But how good are we at actually predicting that? This study compared clinician gestalt to the Troponin-only Manchester Acute Coronary Syndrome (T-MACS) risk prediction model. T-MACS is a validated tool that uses 7 variables to calculate the risk of ACS, generating “rule out” and “rule in” cohorts of very low risk and high risk patients respectively³. Clinician gestalt was taken from the Bedside Evaluation of Sensitive Troponin (BEST) study, a large multi-centre study in U.K. Emergency Departments, where clinicians were asked to record an estimated percentage risk of ACS after reviewing their patient presenting with potential ACS. This study compared the two and evaluated the diagnostic accuracy for predicting major adverse cardiac events within 30 days.

In total, 782 cases were included, of which just under 15% experienced a major adverse cardiac event within 30 days. Compared to T-MACS, clinicians tended to overestimate the probability of ACS and avoided putting down extremes of possibility. This was most pronounced in the two areas where decision-making makes a difference: in the low risk patients potentially suitable for early discharge, and the high-risk patients where treatment could be expedited. For example 49% patients were calculated as being very low risk (<2% likelihood of ACS) by T-MACS, compared to only 9% using clinician gestalt. Similarly, T-MACS identified over 6% of patients as very high risk (95% likelihood of ACS), compared to only 2% of patients using clinician estimates. The overall diagnostic accuracy for T-MACS was significantly higher than gestalt. The C-statistic describes the overall discriminatory accuracy of a test where 0.5 would represent a test no better than chance and 1.00 a model with perfect discrimination. Gestalt performed with a C-statistic of 0.76 (considered fair) and T-MACS 0.93 (considered very good).

Govind’s take home message:

Clinician gestalt has inferior diagnostic accuracy to T-MACS but the appropriate application of T-MACS requires clinician skill. The conclusion from this work should therefore not be that computers are better, but that offloading risk estimation to evidence-based decision tools can augment our thinking and performance.

Contact Govind about this work:

Contact govind.oliver@manchester.ac.uk if you are interested in this research and would like to know more.

“A review of reviews of Emergency Department interventions for older people: outcomes, costs and implementation factors”
By James Van Oppen (ST3 in Emergency Medicine at Leicester Royal Infirmary and Academic Clinical Fellow at the University of Leicester).

The care of older people in our Emergency Departments (ED) is coming under closer scrutiny and is of particular importance given that we are all living longer, with increased rates of chronic disease and increasingly complex care needs.

The authors of this study conducted a review examining published review papers of interventions for older people delivered or initiated within the ED. They found 15 review articles in total, one of which was a meta-analysis. They found that the quality of the reviews was limited, with variability in reporting standards and inconsistent descriptions of the interventions used. The interventions that had been looked at were generally fairly broad in their nature (e.g. discharge planning, follow-up, composition of the multi-disciplinary team) but the outcomes used to evaluate these interventions were often based on service metrics rather patient-centred outcomes. There was evidence that interventions which continued into the community had better outcomes for older patients and we should consider this when thinking about implementing new pathways of care for our older patients.

James’ take home message:

Research into ED interventions for older people is predominantly reported using service metrics rather than person-centred outcomes. The evidence supports those interventions which are holistic and which are continued into the community rather than delivered solely within the emergency department

Contact James about this work:

Contact jvo3@leicester.ac.uk if you are interested in this research and would like to know more.

“Introduction of a trauma care Patient Reported Experience Measure (PREM) during weekly governance meetings”
By Blair Graham (RCEM Doctoral Research Fellow, Lecturer in Urgent & Emergency Care, University of Plymouth and Registrar in Emergency Medicine, Derriford Hospital, Plymouth) and Matthew Owen (Intercalated Medical Student in Urgent & Emergency Care, University of Plymouth)

We’ve all sat in governance meetings where discussion has focussed on the processes and pathways of care. Blair and Matt found this to be the case in their local trauma-governance meeting; they came away wondering whether better understanding each patient’s experience might offer some different insights into the strengths and vulnerabilities of current care, and better highlight where improvements could be achieved.

Patient experience can be best described as the incidence of those components of care that should happen during a healthcare encounter actually happening. This is clearly different from patient satisfaction, which is an individual’s perception of a care episode. Evidence, including a previous meta-analysis⁴, shows us that patient experience is a fundamentally important component to high quality care and is linked with, and likely impacts upon, patient safety and clinical effectiveness.

How did Blair and Matt look at patient experience in trauma care?

They undertook face-to-face administration of a validated questionnaire (Quality of Trauma Care Patient Reported Experience Measure (QTAC- PREM)⁵. Blair and Matt included TARN eligible trauma patients presenting to their MTC who were scheduled to be presented at weekly governance meetings. The study, conducted over a five-month period, also aimed to see if routinely capturing patient experience could be performed within a week of injury, whilst still an in-patient. A total of 27 patients were recruited in this convenience sample with an average age of 50.1 (8 – 89) years, a sex breakdown of 66% male and an average Injury Severity Score of 17.2 (9-36).

In order to identify areas of strength and vulnerability, the questionnaire items were ranked by the proportion of ‘positive’ and ‘critical’ responses. The pre-agreed threshold for ‘acceptable performance’ was defined as the number of positive responses exceeding 90%; proportions of positive responses between 65-90% were deemed ‘borderline’; and less than 65% termed ‘unsatisfactory’.

What were the results? (see Fig. 1):

The questionnaire was straightforward to administer to patients during the acute phase of their in-patient admission and all disciplines welcomed the inclusion of patient experience data within weekly trauma meetings.

Data demonstrated that patients reported a mostly positive experience of receiving trauma care, providing positive feedback for the team. In particular, proportions of positive responses for staff introducing themselves and patients’ perceptions of feeling safe were above the predetermined 90% threshold. An additional two items concerning the provision of emotional and mental support, and pain management fell below 65%. A subsequent round of data collection will assess effectiveness of increased attention to these areas.

Blair and Matt’s Take Home Messages

1. Ensuring optimal patient experience is fundamental the delivery of high quality care. Patient experience should be prioritised alongside patient safety and clinical effectiveness.
2. Providers should ensure that patients’ experiences and perspectives are represented during clinical governance events such as major trauma meetings and event reviews.
3. It is acceptable to ask patients about their emergency care experiences soon after major injury. This may be desirable in order to reduce recall bias and maximise the validity of data.
4. Finally – specific to trauma care, it is important to recognise patients’ psychological and mental health needs early on in their patient journey.

What are the next steps?

A Quality Improvement Project to improve immediate psychological aftercare for Trauma Patients is currently being designed. The effectiveness of this project will be measured using QTAC-PREM.

This work is also being used to inform the design of a Patient Reported Experience Measure targeted at older adults attending the ED. You can find out more about this project at @prem_ed_study

Contact Blair about this work:

Contact blair.graham@plymouth.ac.uk , or find him on twitter @timecritical, if you are interested in this research and would like to know more.