Scaling up and improvement: To implement and evaluate a programme of Shared Haemodialysis care (dialysis self-management support). SHARE-HD

The School of Health and Related Research (ScHARR) at the University of Sheffield requires Hospital Episode Statistics (HES_ admitted patient care (APC) and accident and emergency (AE) data for use in the evaluation work stream of the scaling-up and improvement implementation and evaluation of shared haemodialysis care (dialysis self-management support) project, known as SHAREHD.

There is evidence from a range of long-term conditions of a relationship between patient involvement in aspects of their own care and improved outcomes. Patients who receive dialysis treatment at centres do so in an environment that generally discourages patient involvement. Shared Haemodialysis Care (SHC) describes an intervention where patients treated with in-centre haemodialysis are given the opportunity to learn tasks relating to their own dialysis treatment. This study is a quality improvement collaboration, which is led by Sheffield Teaching Hospitals (STH) to scale up shared haemodialysis care for patients on centre-based haemodialysis. The study is a 24-month research study, nested within a 30-month quality improvement project, which aims to assess the effectiveness and economic impact of a structured programme to encourage patient involvement in centre-based haemodialysis.

STH are the sponsor of the main study and are responsible for the overall project management and recruited patients. ScHARR, University of Sheffield are responsible for the 24-month research study to evaluate the effectiveness of SHARE-HD.

Sheffield Teaching Hospital in collaboration with the NIHR Collaborations for Leadership In Applied Health Research and Care (CLAHRC) based at ScHARR at the University of Sheffield were successful in responding to a request from the Health Foundation Scaling Up Improvement Programme for taking successful health care interventions and delivering them to scale, and duly won the grant-funding Award from the Health Foundation. STH contributed towards the writing of the protocol for the evaluation study. The evaluation, which includes the design, analysis, and writing up of the evaluation is the sole responsibility of ScHARR at the University of Sheffield. The statistical and health economic analysis plan (SHEAP) was developed independently from the rest of the project by ScHARR at the University of Sheffield. The aims and direction of the evaluation study are the sole responsibility of the University of Sheffield. The data requested will be analysed by ScHARR, presented to the wider team and in the same way will be presented for reports and publications in aggregated summaries, complying with HES analysis guides requirements for small numbers.

This study, which began in January 2016, aims to evaluate the scaling up of shared haemodialysis care in 12 dialysis centres across England using an open pragmatic stepped wedge randomised controlled trial study design, with three steps each lasting 6 months. Recruitment of participants took place across the 12 sites, ranging from 27 at the lowest site to 80 at the largest. Following a 6-month baseline data collection period or step; the 12 clusters (hospitals) are randomised in two steps, of 6 hospitals at a time to either the SHAREHD programme or usual care. After the first six-month step 6 out the 12 hospitals will be delivering the SHAREHD intervention; and after the second six month step all 12 hospitals will be delivering the SHAREHD intervention.

The intervention at the centre of the programme is to give centre based haemodialysis patients the choice to undertake up to 14 treatment related tasks. These range from performing observations, preparatory hygiene, setting up the dialysis machine, securing dialysis access, programming the machine, initiating and completing treatment.

The primary aim of the study is to test the clinical effectiveness of the intervention (SHAREHD) compared to usual care on the ability of sampled haemodialysis patients to complete 5 or more out of 14 tasks at the 6 assessment points.

The secondary aims are set out below:

1. To test the impact of the intervention on the secondary outcomes (number of patients who choose to go home on HD or perform in centre dialysis independently; Patient activation, quality of life and renal symptom score; hospitalisation, infection, vascular access intervention, safety and adverse events).

2. To determine the resource use and costs of providing the SHAREHD intervention.

3. To examine the cost-effectiveness/cost-utility of the intervention

4. To examine the interaction between the effectiveness of the intervention and a range of patient-level characteristics, including age, ethnicity and socioeconomic deprivation.

5. To model the incremental cost-effectiveness of the SHAREHD if it was rolled out across England and Wales

The data requested is to identify the co-morbidities of patients which will be adjusted for in all analyses, to identify hospitalisations (secondary aim 1) and to derive resource utilisation information for the economic evaluation (secondary aims 2 and 5). Data will only be accessed by staff of ScHARR, University of Sheffield.

The primary analysis will compare those completing five or more tasks and the information to analyse this has been collected directly as part of the study. A multi-level generalised linear mixed (random-effects) logistic regression model will be fitted to the longitudinal data to answer this question.

The HES data will be used to identify the presence of co-morbid conditions previously proven to be prognostically significant in patients receiving renal replacement therapy. By identifying individual ICD10 and OPCS codes from the diagnosis and procedure code fields in HES from admitted patient care episodes from the two years prior to the start of the study, the presence of these co-morbid conditions will be assigned to patients at the start of the intervention. Previous applications of this method in this setting have found it to

a. reduce variation in outcomes

b. have similar accuracy to clinician coded co-morbidities (Fotheringham et al, 2014)

Sub group analysis will be performed regardless of the statistical significance on the overall intervention effect.

Multilevel logistic regression will be applied to explore the impact of serious adverse events, including hospitalisations (Secondary aim 1) Continuous secondary outcomes including quality of life and change in renal symptoms scores will be analysed using random effects linear regression. Other secondary outcomes including hospitalisations will be analysed using random effects logistic regression.

The base case cost-effectiveness analysis will compare SHAREHD with usual care. The primary cost-effectiveness analysis will be reported as incremental costs per quality adjusted life year (QALY) over the study period from an NHS and social care perspective. Secondary analysis will examine the cost-effectiveness from wider perspective (including participant and companion travel time and time away from usual activities), incremental costs per competency achieved/person achieving five or more tasks, the cost per home HD case achieved and the cost-effectiveness over an extended time horizon.

Data on resource use will be collected using; Hospital episode statistics (HES data), Your Health Survey and ethnographic observation. Each specific resource used will be valued using unit costs based on NHS reference costs and other national averages, to generate a total cost of national generalisability. The elements of total cost will include:

• Set up cost of the SHAREHD programme: including the cost of learning events (staff time (trainers)), printing materials, time (trainees),

• Cost of delivering the intervention: including the time spent delivering the SHAREHD programme to participants (ethnographic study)

• Impact of the intervention on the NHS: including hospital visits, treatments and investigations (HES data)

The cost per cluster / unit will be generated across the period of the trial (24 months post randomisation). This will be constructed from resource use aggregated at the unit level with the relevant unit cost applied. Costs (excluding research related costs) that do not vary by trial arm (e.g. recruitment of clusters) or by cluster within a trial arm (e.g. costs setting up SHAREHD) will be apportioned and added to each cluster. The cost per participant will also be generated and presented as incremental costs per quality adjusted life year.

The beyond-trial modelling of longer-term expectations for cost-effectiveness will reflect the health service use consequences of the SHAREHD programme, mean cost and QALYs will be modelled over a lifetime horizon for the intervention compared with usual care and will take the form of a cost-utility analysis (CUA) to estimate the incremental cost and the incremental effect of intervention compared with usual care. The model will have four states:

• Share HD/HD

• Home HD

• Transplant

• Death

For each state there will be distributions around the expected costs and utilities and time spent in each state, where costs and utilities will be a function of patient age at the point in the model. For the SHAREHD state patients the number of tasks a patient is likely to achieve independence on will also be generated and a patient’s utilities and costs will be estimated as function of this based on data observed in the SHAREHD study. Model outcome will be the represented as an incremental cost-effectiveness ratio (ICER).

Further details of the cost-effectiveness analysis can be found in the statistical and health economic analysis plan (SHEAP) and the study protocol, which have been included with this application.

This research is performed in the public interest for scientific research and statistical purposes, in compliance with Article 6(1)(e) and 9(2)(j) of the GDPR. Ethics approval has been granted for the purposes as described in the study protocol with which this agreement falls in line.

If it is proven that changing health behaviours leads to improvements in patient reported outcomes while being cost neutral and without demonstrable harm then this needs to be robustly demonstrated to gain traction for a wider implementation. The impact of this health intervention will be determined in the social care setting through the data collection instruments that have been developed. The wider cost implications of this will be modelled. All of the above statements are measurable from the analyses proposed. Sustainability and spread are being maximised through the programme and with the proposed supporting evidence these benefits will be realised within the lifecycle of the service specification documents for haemodialysis: authored by clinical reference groups, these mandate the specification of a dialysis unit and the services it offers, including shared haemodialysis care.

Sheffield Teaching Hospitals provided NHS digital with SHAREHD study ID, NHS number, date of birth, surname, forename and gender for the participants enrolled in the SHAREHD study who gave consent for the study team to request HES data from NHS digital. Data was uploaded over the secure system provided by NHS digital.

NHS digital returned linked APC and AE data including the unique study ID and no other identifiers. No subsequent flows of data will take place.

ScHARR store the data on a secure drive at the University of Sheffield on a (securely housed) networked virtual machine accessible only from within the campus network. This data is stored in a separate location to the participant identifiers. The two datasets will not be re-linked and the data will remain pseudonymised.

Data is only accessed by six employees within ScHARR who have authorisation from SHAREHD study team to access the data for the purposes described, all of whom are substantive employees of the University of Sheffield. The employees who will have access to the data include: data-manager (1 senior and 1 junior staff member) who are responsible for cleaning the data, statistician team (1 senior and 1 junior staff member) who carry out the statistical analysis, and health economist team ((1 senior and 1 junior staff member) who are responsible for the economic modelling. No record-level data will be shared outside of these substantive employees.

Data will be verified by confirming number of records provided matches that are specified in the HES extract covering letter provided by NHS Digital, data ranges will also be checked. Best-practice HES data cleaning scripts will be applied to remove duplicate and erroneous records.

A report will be produced and shared with the Health Foundation showing the results and will include a series of tables and figures summarising the case mix characteristics, primary and secondary outcomes (completion of five or more tasks, whether patients start home haemodialysis, patient activation, health related quality of life, renal symptoms and cost-effectiveness) information will be presented for baseline, and step 1 and step 2 of the stepped wedge design (see SHEAP). An open access journal article will be prepared and submitted to peer review journals reporting the effectiveness and the cost-effectiveness of SHAREHD.

Descriptive statistics for the cohort will be presented at summary level, for example mean with standard deviation for continuous variables and numbers with proportions for categorical data. Results will be presented in an aggregate form: for tasks performed this would be proportions achieving 5 or more tasks. Multilevel models will be used to account for the clustered nature of the data; multi-level generalised linear mixed/random effects models will be used for binary data and multi-level lineal mixed/random effects models will be used for continuous data. Results and figures will again be presented at an aggregate level. The economic analysis will follow guidance on cost-effectiveness analysis set out by the National Institute for Health and Care Excellence (NICE, 2013). Results will be presented as total costs to the NHS over the study period and modelled over a patient's lifetime using beyond-trial modelling. The costs of SHAREHD will be compared with costs of usual care for in-centre haemodialysis patients as a cost per quality adjusted life year gained. See Appendix C of SHEAP for examples of summary tables.

The data will not be linked with any record level data.

There will be no requirement nor attempt to re-identify individuals from the data.

The data will not be made available to any third parties except in the form of aggregated outputs with small numbers suppressed in line with the HES Analysis Guide.

SHAREHD holds HES APC and HES A&E data for analysis. The data is for the period 2014/15 to present, this data is for two years prior to the study commencing in order to identify existing patient co-morbidities as described in Objectives for Processing above. The remaining period is for the observed period of the study. Data held under this agreement is only for those participants who have consented to take part in the SHAREHD study and who have consented for researchers to access this HES data.

The statistical package STATA will be used for all analysis.

The data will be analysed to establish the effectiveness of the SHAREHD programme and for estimating its cost-effectiveness (see specific outcomes for further details). All outputs are aggregated with small number suppression in line with the HES Analysis Guide. There will be no sharing of record-level data with third parties. No attempt will be made to re-identify the patients from the data. The data will not be linked or used for commercial purposes.

Descriptive statistics for the cohort will be presented at summary level, for example mean with standard deviation for continuous variables and numbers with proportions for categorical data. Results will be presented in an aggregate form, for number and proportion of tasks performed, and graphical figures of this data will be presented in the same format. The time series analysis of this data is looking at trends over time before and after the implementation of the SHAREHD programme. Multi-level generalised linear models will be applied to analysis, whether a participant can complete five or more tasks, the number of tasks each patient can complete (supervised or independent), the number of tasks a patient can complete independently and whether a patient has home dialysis. Results and figures will again be presented at an aggregate level.

The economic analysis will follow guidance on cost-effectiveness analysis set out by the National Institute for Health and Care Excellence (NICE, 2013). Results will be presented as total costs to the NHS and the costs of the SHAREHD programme will be compared with costs of usual care as costs per QALY and cost per competency achieved and cost per home HD case achieved.

A draft report was provided to the Health Foundation in December 2018. An article on the effectiveness and cost-effectiveness of SHAREHD will be prepared for an open access peer review journal, possible journals to be submitted to include: Nephrology Dialysis Transplantation or Medical Decision Making. A simplified version of findings will be published on SHAREHD study website and issued to study participants.

All outputs are aggregated with small number suppression in line with the HES Analysis Guide.

Because of the particular need for user engagement, newsletters are issued regularly and the patient facing website, https://www.shareddialysis-care.org.uk, is constantly updated.

May 2020: The extension is to complete the analysis for the health economic publication where the data controllers still aim to submit to either Journal of Nephrology or Medical Decision Making or an appropriate clinical journal. The additional time will be used to run some further sensitivity analysis using alternative costs for home dialysis.

SHAREHD is designed to empower patients to take more control over their chronic health condition. Kidney disease costs the NHS approximately 1.4 billion pounds a year, with the 57,000 renal replacement therapy patients consuming over half of this. Like many other interventions, there are benefits and risks, costs and savings, which need to be determined in order that the wide range of stakeholders can appraise and invest in SHAREHD. This study is able to achieve this through linkage to HES.

Quantifying the reductions in adverse events, hospitalisation and healthcare usage are powerful motivators: patients need this information to help them decide to undertake tasks relating to their dialysis. Clinical staff and policy makers can evaluate shared haemodialysis care and its benefits alongside other interventions which may compete for similar resources. Understanding the cost of implementing shared haemodialysis care and potential downstream savings, or cost-neutrality in the presence of increased quality is essential for managers, policy makers and clinical reference groups. Many evaluations of health interventions are not subjected to such rigorous cost-effectiveness evaluation, and the reporting of these analyses both informs evaluations being designed by others looking to improve healthcare, and also more widely argues the benefits of such interventions. It also supports evaluation using linked datasets including those offered by NHS Digital.

The hazard of devolving aspects of haemodialysis care to patients should not be overlooked. However, demonstrating more broadly that patient involvement in their own care, not by choice or motivation, but through health improvement interventions is essential knowledge for the wider health and social care community. Much of the existing evidence has been conducted cross-sectionally (e.g. associative not causative) as articulated in the Health Affairs vol 34, no 3 article Green Hibbard et al: "When Patient Activation Levels Change Health Outcomes and Costs Change Too", or is based on individual testimonies.

The report that the data controllers submitted to the funders showed that although there were no significant benefits in terms of quality of life from shared HD there were potential cost savings. It is important that the data controllers publish these findings in a peer review journal and run some additional sensitivity analysis to ensure that the findings are robust. These additional analyses include using alternative unit costs for home haemodialysis.