A pragmatic cluster randomised controlled trial of a tailored intervention to improve the initial management of suspected encephalitis

Objective To determine whether a tailored multifaceted implementation strategy improves the initial management of patients with suspected encephalitis. Design Pragmatic two arm cluster randomised controlled trial. Setting Hospitals within the United Kingdom. Participants Twenty-four hospitals nested within 12 postgraduate deaneries. Patients were identified retrospectively by searching discharge, microbiology, radiology and pharmacy records and included if they met clinical criteria or had a recorded suspicion of encephalitis. Intervention An implementation strategy designed to overcome barriers to change, comprising local action planning, education and training, feedback on performance, a lumbar puncture pack and a range of optional components. Outcomes The primary outcome was the proportion of patients with suspected encephalitis undergoing diagnostic lumbar puncture within 12 hours of admission and starting aciclovir treatment within six hours. Secondary outcomes included the proportions of adults and children who had a lumbar puncture, who had appropriate cerebrospinal fluid investigations, and who had appropriate radiological imaging within 24 hours of admission. Data were collected from patient records for 12 months before and 12 months during the intervention period, and analysed blind to allocation. Results 13 hospitals were randomised to intervention and 11 to control (no intervention), with 266 and 223 patients with suspected encephalitis identified respectively. There was no significant difference in primary outcome between intervention and control hospitals (13.5% and 14.8% respectively, p = 0.619; treatment effect -0.188, 95% confidence interval -0.927 to 0.552), but both had improved compared to pre-intervention (8.5%). Conclusion The improvement in both intervention and control arms may reflect overall progress in management of encephalitis through wider awareness and education. Trial registration Controlled Trials: ISRCTN06886935.


PROTOCOL SUMMARY
Title: Development and evaluation of an intervention based around the National guidelines on the management of suspected encephalitis, and its evaluation through a cost effectiveness analysis.
Population: Secondary care hospitals in the UK will be included. Hospital sites already participating in the ENCEPH-UK study will be excluded so as to reduce the risk of attention bias.

Number of Sites:
Up to twenty eight secondary care hospitals will be recruited across the UK.

Study Duration:
Fifteen months per site (to include delivery and evaluation of an intervention at 12 months).

Description of Intervention:
The intervention will target clinicians (and other healthcare workers) within the hospitals.
Outcome data will be collected on patients (adults and children) with suspected or confirmed encephalitis.

Objectives:
1. To evaluate the effects of an intervention to improve the clinical management of suspected encephalitis in secondary care.
2. To evaluate the cost-effectiveness of the intervention.

Primary outcome
The proportion of patients with suspected encephalitis* whose care met all of the following criteria:  Aciclovir given within 6 hours from admission to hospital at the appropriate dose unless there was an alternative diagnosis; and  A lumbar puncture (LP) was performed within 12 hours of hospital arrival unless clinically contraindicated.
The secondary outcomes will be:  The proportion of all adults given intra venous (IV) aciclovir for a neurological presentation who met the definition of suspected encephalitis.
 The proportion of all children given IV aciclovir 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 hours unless there was a clinical contraindication.
 The proportion of patients with suspected encephalitis who had a lumbar puncture after resolution of a clinical contraindication to that LP.

Design
Two arm cluster randomised controlled trial comparing a guideline implementation intervention with routine practice.

BACKGROUND INFORMATION Overview
This study will evaluate the clinical effectiveness and cost-effectiveness of an intervention to support the implementation of national encephalitis guidelines.
Intervention development will be informed by existing evidence on professional behaviour change and sub-studies within the ENCEPH UK programme. The intervention will target clinicians responsible for the diagnosis and initial management of suspected encephalitis. The use of a cluster randomised controlled trial (RCT) in up to twenty eight hospitals with an evaluation period of 12 months will allow both cost effectiveness and clinical outcomes to be assessed.

Introduction
Encephalitis (inflammation of the brain tissue) is one of the most important types of brain infection because of the devastating impact on those affected. It has a disproportionately large burden on the NHS and community through high health and economic costs, but this has never been quantified. In terms of research, encephalitis is relatively neglected because it is not easily funded from other sources, e.g. major charities.
HSV is the most common viral cause (1) and is treatable with aciclovir, but delays in treatment are associated with poor outcome. There has been little research into how and why delays occur. There are not economic studies in the UK that have attempted to estimate the economic impact of encephalitis. However, settlements for negligent management of a single encephalitis case are around £1-2 million.

Objectives
It has been found that many patients with suspected encephalitis are started late with aciclovir treatment. The barriers and enablers for the diagnosis and management of encephalitis patients will be assessed as part of a sub-study in the ENCEPH UK programme. This consists of semi structured interviews with health care professionals who have a variety of roles within several hospitals. From this, the major barriers and enablers to using the encephalitis guidelines will be deduced and a novel intervention will be developed and implemented in the hospitals randomised to the intervention arm over twelve months . This intervention will then be assessed for both clinical and cost effectiveness.

SELECTION OF CENTRES/CLINICIANS
Study centres will be initiated once all global (e.g. local research and development (R&D) approval) and study specific conditions (e.g. training requirements) have been met, and all necessary documents have been returned to the Brain Infections UK coordinating centre.
All centres will have a consultant and all centres must be equipped with the ability to perform, or have access to, CT/MRI scans and have aseptic conditions for conducting a LP to be performed. Hospitals will be randomised as a unit to either the standard care or intervention arm.

TRIAL DESIGN
A cluster RCT design has been chosen to compare routine practice with an intervention promoting 'best practice' within up to 28 UK hospitals. The intervention taking place in those randomised hospitals will target clinicians. Data on patient care will be collected to assess clinical practice. It is not envisaged that there will be any study related risks to the patients as data collection methods will be similar to the clinical audit that have been performed over the last ten years within our group (2,3).
Identifiable data will only be seen by a member of NHS staff currently working at the site and will not be sent to Brain Infections UK. The number of cases, management and cost effectiveness of the intervention will be assessed in all sites.

Primary outcome
The proportion of patients with suspected encephalitis* whose care met all of the following criteria:  Aciclovir given within 6 hours of admission to hospital at the appropriate dose unless there was an alternative diagnosis; and  A LP was performed within 12 hours of hospital arrival unless clinically contraindicated.
The secondary outcomes will be:  The proportion of all adults given intra venous (IV) aciclovir for a neurological presentation who met the definition of suspected encephalitis. There will be a time to event analysis for all of the primary and secondary outcomes listed above within a full time series analysis.

Inclusion Criteria
Within this study, any patients who have suspected encephalitis within the twenty eight sites will be included. The inclusion criteria for suspected encephalitis are: Patients with suspected encephalitis

Patient Transfer and Withdrawal
There is a possibility that patients could be transferred to and from the sites within this study. Efforts will be made to establish if any centres regularly receive patient transfers and appropriate recording mechanisms will be put in place. The first 24 hours of a patient's management are the most critical for the primary outcome.
Therefore, the patient will be counted at the site of their first admission if initial clinical management takes place within this time period. Should a patient be transferred to a different site within 12 hours, there should still be sufficient data to count the patient within that original study site.

INTERVENTION DEVELOPMENT AND DELIVERY
Intervention development involves a number of steps, some of which have been undertaken during earlier studies in the programme. These steps comprise:  Defining priorities for implementation, i.e. key guideline recommendations where there is scope for improved clinical performance;  Identifying perceived barriers to and enablers of recommended practice;  Defining candidate intervention components; and  Constructing a coherent intervention package.
Defining priorities for implementation. We already have a good understanding of hospital based performance from studies recently conducted (4,5), as well as preliminary analysis of the Health Protection Agency (HPA) cohort study. During the ENCEPH-UK programme, we will continue to collect data on the timing of the various diagnostic and treatment decisions.
One priority will be to increase the proportion of patients with suspected encephalitis appropriately and promptly started on treatment with aciclovir. This is chosen because of compelling evidence that not starting treatment for 48 hours or more is associated with a poor outcome in patients with HSV encephalitis.
Other priorities for attention will include the diagnostic tests performed, their sequence and timing, and the grade of clinicians involved in decision making, looking in particular for major deviations away from the guidelines. These are defined as those serious enough to cause appreciable morbidity and mortality (6), and are based on published guidelines on the management of suspected CNS infections (7-

11):
 Performing a LP, without performing imaging first, in a patient with contraindications to an immediate LP (8, 10);  Delaying a LP in order to perform imaging, when there is no contraindication to an immediate LP (8,10,11);  Failing to perform a diagnostic LP in a patient with a suspected CNS infection, and no contraindications (7,8,10,11); and  Failure to investigate CSF for cell count, protein, and CSF to plasma glucose ratio when performing a LP.
Priorities for implementation (i.e. key guideline recommendations) will be derived from the Association of British Neurologists and British Infection Association National Guidelines for Management of Suspected Encephalitis in Adults (12) and Paediatrics (13).

Data collection
Data will be collected by a trained member of hospital staff (eg research nurses, the principal investigator, the co-investigator or a member of their team). The data will be pseudo-anonymised during the completion of the case report form (CRF) with the study number acting as the encryption key (please see below for details). Data will be collected for the previous 12 months and then again in 12 months' time.

Cost-effectiveness analysis
Methods. Quantities of resource utilisation will be collected from trial records, while unit costs will be obtained from publicly available data published routinely. Costs related to inputs associated with health care services as used in the trial will be estimated at its conclusion. Costs and health benefits occurring after 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 SF-36 and EQ-5D in the Quality of Life substudy, following Brazier's methodology (14). Estimates of costs and health outcome (utilities) will be used to populate a model of encephalitis patient management under the status quo and the alternative under new guidelines being adopted. The model will cover the remaining patient lifetime within which the (probability of) costeffectiveness of the alternative will be analysed (15).
Statistical Methods. Cost data will be analysed using regression methods for handling censored cost data. Costs and benefits will be analysed jointly using a bivariate probability distribution. Sample uncertainty in estimated cost difference and Incremental Cost-Effectiveness Ratios (ICERs) between arms groups will be described using bootstrapped confidence intervals (16). Data analysis will be performed using a suitable statistical package such as R or STATA.
Uncertainty Analysis of Decision Model Results. Parameter uncertainty will be accounted for through univariate sensitivity analysis and probabilistic sensitivity analysis.

Trial Closure
The site shall be deemed as closed after the data at 12 months has been completed and all data queries resolved. When this has happened at all sites the database will be locked.

Randomisation
To ensure the best quality data within this study control and intervention sites must not have any routine crossover of staff between them, for example, trainee doctors moving as part of the standard rotation pattern within their training deanery.
Therefore, we will take deanery boundaries into account during recruitment and randomise using the deaneries as randomisation units. We intend to include a range of types of hospital in the trial to improve generalisability. The mix of these should broadly represent national provision. We will use deanery as the unit of randomisation, to guard against leakage of information between hospitals within the same deanery as trainee doctors move between hospitals. We will define two blocks of deaneries, a block of 6 in which research teams are currently actively involved in the ENCEPH UK programme and a block of 6 in which research teams are not currently actively involved in the ENCEPH UK programme. An independent statistician will randomise equal numbers within each block to the intervention and routine arms.

Outcome Measures
The primary endpoint will be aciclovir administration within 6 hours and an LP performed within 12 hours of admission to hospital for patients with a suspected CNS infection.
The outcome measures will assess all primary and secondary outcomes as detailed above and a time to event analysis will be performed for each of these. We will also conduct an analysis of outcome measures between adults and paediatric populations.

Sample Size and Analysis
Using pilot data from 315 patients across 26 hospitals in 4 deaneries we have estimated the standard deviations of the deanery and hospital random effects to be 0.244 and 1.108, respectively, and the current proportion of compliance under treatment as usual as 0.05. Using these estimates, the table below shows the power of the likelihood ratio test for a significant difference between standard care and the intervention as a function of m, the number of patients recruited per hospital, and p, the compliance proportion under the intervention ( Table 2). To allow for possible under-recruitment in some of the smaller hospitals, we will therefore seek to recruit up to 30 patients per hospital, per notes review (two reviews within this study) with a mix of adult and paediatric cases to be included. This will result in 840 cases in each of the two reviews, giving a total of 1680.
We will analyse the results in STATA or R using a generalised linear mixed model (17) with binomial errors, logistic link, fixed effects for blocks and treatments, random effects for deanery and for hospital.

Ethical Considerations
This study is encouraging 'best' healthcare in both the intervention and the control arms. The study will abide by the principles of the Research Governance Framework and ICH GCP. The study intervention will target clinicians and pseudo-anonymised patient data will be used to assess changes in clinical practice. Patients' records will be viewed by a member of NHS staff working at that site with medical information extracted using a CRF which will record a unique study number and patient date of birth but no other identifiers will be recorded.

Ethical Approval
The trial protocol has received the favourable opinion from Preston North West Multicentre Research Ethics Committee (MREC) and will undergo site specific assessment (SSA) by completing section C of the REC application form.

Regulatory Approval
R&D approvals will be sought from each research sites before commencement of the study. Furthermore, an International Standard Randomised Controlled Trial Number (ISRCTN) number will be in place before the first site is actively opened.

TRIAL MONITORING
This study will be monitored via a central risk based approach as per the agreed monitoring plan.

Risk Assessment
An appropriate risk assessment will be completed and its resulting procedures followed during the study. As this study does not directly involve patients, or affect their care, this study will have low to zero risk attached to it.

Source Documents
Source data is all information, original records of clinical findings, observations, or other activities in a clinical trial necessary for the reconstruction and evaluation of the trial. Each participating site will maintain appropriate medical and research records for this trial, in compliance with ICH E6 GCP, Section 4.9 and regulatory and institutional requirements for the protection of confidentiality of subjects.

Data collection
As we have done previously, a combination of methods to identify patients with suspected encephalitis involving screening of electronic sources (2, 3):  Hospital discharge codes;  Laboratory records for patients who have had a CSF sample taken;  Pharmacy records for patients who have received intravenous aciclovir;  Radiology records where 'suspected encephalitis' or other potential marker terms are used.
The first two methods will be applied across all sites to minimize identification biases associated with any one method.
Trained members of NHS staff (eg nurses, the principal investigator, the coinvestigator or a member of their team) will undertake retrospective case identification and data collection at each hospital site. The records of suspected and confirmed cases will be reviewed and data collected using a standardised form to document processes and selected outcomes of care. Data will be collected for the masked allocation to intervention or control hospital sites will be feasible during data collection but the feasibility of this will be reviewed.
The date and time from initial presentation to hospital, either via accident and emergency or medical admissions, will be taken as the starting point for data extraction and outcome measurement. As patients may be transferred between centres, patients within both groups will be characterised and the data analysed as per an intention to treat analysis.
Clinical data collection will be based upon the audit tool developed for the National Guideline. Apart from basic demographic data (for example date of birth and sex), no patient identifiers will be used, so that only sufficiently anonymised case records are transferred out of hospital settings for analysis. No one outside of the direct healthcare team will have access to any patient identifiable data as this will be pseudo-anonymised at site.

Case Report Forms
The study CRF is the primary data collection instrument for the study. All data requested on the CRF must be recorded. Data should be entered directly into OpenClinica ™ with paper CRFs to be used only in exceptional circumstances. All missing data must be explained with "N/D" for procedure not done, "UNK" for unknown and "N/A" for not applicable. If using a paper CRF, all entries should be printed legibly in black ink. If any entry error has been made, to correct such an error, draw a single straight line through the incorrect entry and enter the correct data above it. All such changes must be initialled and dated. Do not erase of white out errors. For clarification of illegible or uncertain entries, print the clarification above the item, then initial and date it.

Data checks
Data stored at Brain Infections UK coordinating centre will be checked for missing and unusual values (range checks) and checked for consistency within participants over time. Any suspect data will be returned to the site in the form of data queries.
Data query forms will be produced at the Brain Infections UK coordinating centre from a password protected central web based database (OpenClinica TM ) and sent either electronically or through the post to a named individual (as listed on the site delegation log). Sites will respond to the queries providing an explanation/resolution ENCEPH UK Intervention RCT protocol V4.0 091213 to the discrepancies and return the data query forms to Brain Infections UK coordinating centre. The forms will then be filed and the appropriate corrections made on the database. There are a number of monitoring features in place at the Brain Infections UK coordinating centre to ensure reliability and validity of study data.

Direct access to data
In order to perform their role effectively, monitors and persons involved in Quality Assurance and Inspection may need direct access to primary subject data, eg patient records, laboratory reports, appointment books, etc. As this affects the patient's confidentiality, permission will be given by the principal investigator (PI) at the site.

Confidentiality
Individual participant medical information obtained as a result of this study is considered confidential and disclosure to third parties is prohibited. CRFs will be labelled with the patient's initials and unique study screening and/or subject identity (ID) number before leaving site, however, data should be entered directly into OpenClinica TM and paper CRFs should be used as a last resort only. The Brain Infections UK coordinating centre will preserve the confidentiality of participants taking part in the study (only study number, sex and date of birth will be held) and The University of Liverpool is registered as a Data Controller with the Information Commissioners Office and this study is in full compliance of the Data Protection Act.

Quality Assurance and Quality Control of Data
This study has undergone a risk assessment, the outcome of which indicates this to be a low risk study. As such, site visits will be conducted and source data verification performed if indicated as a result of central monitoring processes. To this end:  The Principal Investigator, and designated staff from each centre will attend the study initiation meeting, coordinated by Brain Infections UK coordinating centre in conjunction with co-lead investigators.
 The trial coordinator (or delegated person), is to verify appropriate approvals are in place prior to the initiation of a site, and that the relevant personnel have attended study specific training.
 The trial coordinator (or delegated person), is to monitor screening, recruitment and drop-out rates between centres.
 The trial coordinator (or delegated person), is to conduct data entry consistency check and follow up data queries.
Local quality control (QC) will include:  Data will be evaluated for compliance with protocol and accuracy in relation to source documents if required  The study will be conducted in accordance with procedures identified in the protocol.

Records Retention
The Investigator at each investigational site must make arrangements to store the essential trial documents, (as defined in Essential Documents for the Conduct of a

FINANCIAL ARRANGEMENTS
This study is funded by National Institute for Health Research (NIHR) Programme Grant for Applied Research.
Sites will be reimbursed £150 for each of the two data collection periods and an additional £15 per CRF completed. This shall be paid at the end of each data collection period after data queries have been resolved.

TRIAL COMMITTEES
With this study being part of a larger programme the below meeting structure has been devised to facilitate effective communication and support.

Programme Steering Group
The Programme Steering Group will consist of an independent chairperson, an expert in the field of medical neurology, a biostatistician with expertise in randomised cluster trials, an epidemiologist, co-applicants and key collaborators. The role is to ENCEPH UK Intervention RCT protocol V4.0 091213 provide overall supervision for the overall programme and provide advice through its independent Chairman. The ultimate decision for the continuation of the trial lies with the Programme Steering Group.

Programme Management Group
The Programme Management Group will bring together the leads of the study management group to provide a monthly update on progress and to provide support and advice to any issues that may occur.

Study Management Group
The individual study management group will be responsible for the day to day running of the trial and will meet on a fortnightly basis throughout the duration of the study.

PUBLICATION
The results from different centres will be analysed together and published as soon as possible. Individual clinicians must undertake not to submit any part of their individual data for publication without the prior consent of the Programme Management Group.
A detailed publication policy will be drawn up and agreed with members of the programme steering group.