Background

There is evidence that the current clinical management of serious acute neurological infections is suboptimal [1-3]. Encephalitis, inflammation of the brain tissue, is most commonly caused by herpes simplex virus in the United Kingdom (UK) [4,5]. When herpes simplex virus encephalitis is treated promptly with aciclovir, there is a significant improvement in patient outcomes [6,7]. Encephalitis affects between five and eight people per 100,000 per year [8]. Sequelae after hospital discharge can include significant morbidities such as epilepsy, memory loss, and speech and behavioural disorders [9,10], which also impair patients return to work [10].

Whilst herpes simplex virus encephalitis is relatively rare [5], clinical presentations including features consistent with suspected encephalitis occur relatively frequently but in different ways to other brain injuries. Encephalitis typically presents with one or more of headache, fever, new-onset seizures, altered consciousness, and behavioural disturbances [11]. This variable and non-specific presentation often results in delayed diagnosis, especially in children who may only present with fever and irritability [12]. Furthermore, delays in using the main diagnostic technique, lumbar puncture, may further delay treatment [13-16].

Clinical guidelines have been developed in response to these problems [1-3,5,17]. However, simple dissemination of clinical guidelines is often unlikely to bring about significant changes in clinical practice [18-20]. Furthermore, interventions to implement clinical guidelines should ideally be based upon a diagnosis of barriers to change, preferably focusing on those most amenable to change [21].

Aims

We developed a multifaceted intervention package including core and flexible components with embedded behaviour change techniques selected on the basis of identified needs and barriers to change (Backman, submitted). We will evaluate the effectiveness and cost-effectiveness of a tailored intervention to improve the secondary care management of suspected encephalitis.

Methods

Study design

Participating sites will be randomly allocated to intervention or control (no intervention) arms in a cluster randomised controlled trial.

Participants

Hospitals

This trial takes place in the context of ENCEPH UK—Understanding and Improving the Outcome of Encephalitis, an ongoing research programme assessing the epidemiology and clinical outcomes of encephalitis. In order to reduce the likelihood of any unintended co-intervention effects we sought hospitals not directly participating in other ENCEPH UK studies. Sites had to have facilities to perform lumbar punctures and neuroimaging and willing to be randomised to intervention or control arms. We aimed to recruit a range of types of hospital, providing secondary, tertiary, (specialist) and paediatric care, to broadly represent national provision and improve the generalizability of subsequent findings.

We were aware that trainee doctors, one key target intervention group, work and rotate between different hospitals within postgraduate deaneries. If we randomised hospitals to intervention and control arms within the same deanery, there would be a risk of contamination. We therefore used deaneries as the unit of randomisation to minimise contamination.

We assessed all 266 acute trusts in England, Wales and Scotland for eligibility (Figure 1 and Additional file 1 detail a full CONSORT checklist). After excluding 47 participating in other ENCEPH UK studies and 10 specialist hospitals not usually providing routine care for suspected encephalitis patients, e.g. orthopaedic hospitals, we invited 209 hospitals to participate via senior medical members of staff.

Figure 1
figure 1

CONSORT 2010 checklist of information to include when reporting a cluster randomised trial.

Full size image

Patients

We will identify records of patients with features suggestive of suspected encephalitis using three sets of criteria adapted from previous studies [1,22].

Method 1

Mandatory

  • Acute or sub-acute (<4 weeks) alteration in consciousness, cognition, personality or behaviour persisting for more than 24 h. Personality/behaviour change includes: agitation, psychosis, somnolence, insomnia, catatonia, mood liability, altered sleep pattern and (in children) new-onset enuresis or irritability.

Plus any two of:

  • Fever (≥38°C) or prodromal illness—acute or sub-acute

  • New-onset seizures

  • Focal neurological signs of acute or sub-acute onset, including focal weakness, oromotor dysfunction, movement disorders (chorea, athetosis, dystonia, hemiballismus, stereotypies, orolingual dyskinesia and tics) including Parkinsonism (bradykinesia, tremor, rigidity and postural instability) and amnesia

  • Pleocytosis: cerebrospinal fluid white cell count of more than four cells per microlitre

  • Neuroimaging compatible with encephalitis

  • Electroencephalogram (EEG) compatible with encephalitis

Method 2

There was an initial clinical suspicion of encephalitis.

Method 3

There was a clinical suspicion of encephalitis, and the patient died before investigations were completed.

Intervention package

Using theoretically informed semi-structured interviews based upon the Theoretical Domains Framework [23,24], we explored barriers and enablers to diagnosing and managing patients with suspected encephalitis, specifically performing lumbar punctures and initiating antiviral therapy within 6 h (Backman, submitted). We mapped identified barriers and enablers to the patient pathway. We matched behaviour change techniques targeting clinicians to the most salient barriers and enablers and embedded them within an intervention package [25].

The intervention package comprised ‘core’ interventions and, to allow for local flexibility, ‘optional’ interventions (Table 1). We defined ‘core’ interventions as those which we anticipated all hospitals being able to use. These included educational and action planning meetings, feedback of pre-intervention audit data and provision of lumbar puncture kits within refillable boxes. We defined ‘optional’ interventions which hospitals could use depending upon local resources and skills. These included decision support via phone apps and algorithms, an online quiz, prompts and posters, personalised invitation letters to attend educational meetings and a quality improvement cycle pack. Table 1 gives full details of the package following the Template for Intervention Description and Replication (TIDieR) reporting guidance [26].

Table 1 Intervention description using the TIDieR guidelines for intervention reporting [26]
Full size table

We presented the package to a 1-day meeting of senior doctors and nurses from intervention hospitals. We emphasised their roles in directly delivering the various intervention components locally and recommended that they each convene an action planning meeting on return to their hospitals.

Outcomes

The primary outcome is a composite measure of the proportion of patients with suspected encephalitis whose care meets both of the following criteria: aciclovir given within 6 h from admission to hospital and a lumbar puncture performed within 12 h of hospital arrival unless clinically contraindicated.

Secondary outcomes comprise:

  • The proportion of all adults started on intravenous aciclovir within an appropriate dosage range for a neurological presentation who met the definition of suspected encephalitis

  • The proportion of all children started on intravenous aciclovir within an appropriate dosage range for a neurological presentation who met the definition of suspected encephalitis

  • The proportion of patients with suspected encephalitis who had a lumbar puncture performed within 12 h unless there was a clinical contraindication

  • The proportion of patients with suspected encephalitis who had a lumbar puncture at any point during the index presentation

  • The proportion of patients with suspected encephalitis who had either magnetic resonance imaging (MRI) or computer tomography (CT) scan within 24 h of admission

  • The proportion of patients with suspected encephalitis having had a lumbar puncture, who had the following cerebrospinal fluid (CSF) investigations performed: ratio of glucose within the cerebrospinal fluid and the serum calculated and having herpes simplex virus polymerase chain reaction (PCR) performed

We will also compare outcomes between adults and children.

Data collection

We recognise that suspected cases would go through one or more of several types of hospital department but mainly paediatrics, neurology, infectious diseases, medical assessment unit, accident and emergency and microbiology. Staff in participating hospitals will identify suspected cases retrospectively by performing two mandatory searches, of discharge codes for encephalitis in the preceding 12 months and of all patients undergoing a lumbar puncture within the same time period. Hospitals can also find eligible patients via records of intravenous aciclovir prescriptions or of orders for cranial CT scan or MRI within the last 12 months. We are using these combined approaches to maximise likelihood of case identification and reduce differences in ascertainment between hospitals. We included both adult and paediatric cases.

Pre-intervention, we collected data to pilot outcome measures and provide data for the feedback intervention. We aimed for 30 adult and paediatric cases per hospital to allow for possible under-recruitment in some of the smaller hospitals; however, we later sought some additional cases from higher recruiting hospitals to better inform exploratory analyses of the baseline data and to compensate for lower recruiting sites. We plan to ask for a further 30 cases per hospital in the final trial data collection, thus the total patient sample size will be a maximum of 720. We anticipate a degree of imbalance between hospitals to achieve the total required sample size, limited by the upper limit of 40 cases per hospital.

Data are collected using structured case review forms. No patient identifiable information is sent to the central trial team. We trained data collectors, mainly nurses and trainee doctors, via face-to-face meetings and/or written briefing materials. We emphasised the need for a systematic approach to case identification to reduce the likelihood of selection bias. The trial research fellow (RB) is monitoring fidelity to the intervention via regular telephone and email contact with relevant hospitals. She is also collecting qualitative data from staff interviews and observational field notes in a sub-sample of four intervention hospitals data as part of a process evaluation.

Sample size

Using pilot data from 315 patients across 26 hospitals in four deaneries, we estimated the standard deviations of the deanery and hospital random effects to be 0.244 and 1.108, respectively, and the current proportion of adherence to the primary outcome to be 5%. Table 2 shows the power of the likelihood ratio test for a significant difference between intervention and control arms as a function of m, the number of eligible patients recruited per hospital, and p, the proportion meeting the primary outcome criteria under the intervention. Using these estimates and based upon a total of 24 hospitals, recruiting 20 patients per hospital should achieve a power of at least 0.8 when the compliance proportion under the intervention is 0.20.

Table 2 Sample size estimates for the trial
Full size table

Randomisation

As explained above, we used deaneries as the unit of randomisation to minimise contamination between hospitals within the same deanery. We defined two blocks of deaneries, a block of six including hospitals (outside of the trial) where research teams were already actively involved in other ENCEPH UK studies and a block of six where there were no such ongoing studies (Figure 1). An independent statistician randomised equal numbers of clusters within each block to the intervention and routine arms, blinded to hospital identity.

Statistical analysis

We will analyse the results in R (http://www.r-project.org) using a generalised linear mixed model [28] with binomial errors, logistic link, fixed effects for blocks and treatments, random effects for deanery and for hospital. If the intervention is effective, it will raise awareness of suspected encephalitis and lead to increased documentation in the clinical records of encephalitis as a differential diagnosis. In our baseline (pre-intervention analysis), we found that cases mainly included on this basis (method two) tended to be less likely to comply with the primary outcome criteria. One explanation is that these may represent clinical presentations at lower risk of having encephalitis and may therefore tend to be investigated and treated less thoroughly compared with more strongly suspected cases. It is therefore likely that the intervention will result in differential case mixes between the intervention and control arms, in this scenario diluting any real intervention effect. In our analysis of the outcome data, we will adjust for any relationship between the method of inclusion and compliance with the primary outcomes and analysis will be undertaken by a statistician blind to hospital assignment.

Cost-effectiveness analysis

The economic evaluation will take the perspective of the UK National Health Service (NHS). We will use trial records and observations to estimate the costs of intervention delivery. Data on health care resource utilisation will be gathered from case record reviews. Unit costs will be obtained from publicly available routine data [29,30]. Costs and health benefits occurring over 12 months will be discounted at 3.5% per annum. Therefore, results will include, in addition to incremental costs and benefits, an estimate of total cost per patient for each arm. In order to estimate quality-adjusted life years (QALYs), utility values will be estimated from the results of administering the EQ-5D, primary economic outcome, and the SF-6D utility scores derived from the SF-36, following Brazier’s methodology [31].

We will analyse cost data using regression methods for handling censored cost data and accounting for clustered nature of the data [32]. We will analyse costs and benefits jointly using a bivariate probability distribution. Sample uncertainty in estimated cost difference and incremental cost-effectiveness ratios between arm groups will be described using bootstrapped confidence intervals [33] as well as with cost-effectiveness acceptability curves. Parameter uncertainty and robustness of findings will be accounted for through univariate sensitivity analysis and probabilistic sensitivity analysis.

Estimates of costs and health outcome (utilities) will be used to populate a model of suspected encephalitis patient management in both intervention and control conditions. The model will cover the remaining patient lifetime within which the (probability of) cost-effectiveness of the alternative will be analysed [34]. Data analysis will be performed using Stata [35].

Ethical review

The study was reviewed by Preston North West Research Ethics Committee (13/NW/0279) (Additional file 2).

Trial status

The trial is ongoing with pre-intervention data collection completed; we are yet to undertake post-intervention outcome data collection and analysis.