Remote Monitoring of Asthma in Children and Young People

Overview

The objective of this study is to determine whether healthcare data and remotely collected patient data can accurately predict asthma attacks in children and young people aged 5-17 years. The main outcome is: when using this new system, is there a reduction in asthma attacks compared with a historic average. The whole population of children and young people with asthma will have routine healthcare data monitored, with a subset of people with high risk asthma asked to participate in a more detail study involving remotely monitored data.

Asthma affects approximately 1 in 11 people in the UK. It is one of the few chronic health conditions that affects CYP more than older people. Asthma is a fluctuating condition with periods of stability and periods of poor control. Overall cost to the NHS in the UK is greater than £1 billion per year. Predicting an asthma attack, where there is loss of symptom control, is often associated with family or life changes and poor adherence. Viral infections, environmental triggers, seasonal changes etc. also upset symptom control and can lead to an attack. This attack, which will need an increase in therapy and may result in a hospital admission, is the end point of a period of loss of symptom control which may be up to a month in development. While adherence to treatment is a major factor influencing risk of attack, there are many underlying complex issues that prevent CYP from adhering to their medication. Asthma attacks are dangerous. They represent a failure of management and can result in loss of school, lost earnings, loss of confidence, loss of sleep, admission to hospital, requirement for intensive care or, in the worst cases, permanent neurological impairment or death due to lack of oxygen to the brain. Severe outcomes are rare but tragically are considered avoidable. Severe attacks can result in a significant aftermath of anxiety, avoidance of normal life activities and general chronic family worry. There is also a high healthcare cost due to poor asthma control. The national review of asthma deaths, published in 2014 showed several areas where improvements could be made. One of the key recommendations was the introduction of personalised asthma action plans. These have been used more commonly recently but tend to be static printed documents which do not respond to changes in circumstances. In addition, current healthcare systems are poorly responsive to changes in individuals. Around 50% of those who had fatal asthma attacks were classed as having mild asthma. Those classed as severe (\<10%) tend to be looked after in specialised hospital clinics with the rest looked after in primary care. In time some who were classed as severe may become more stable, and some who are considered "mild" may still have severe attacks. According the NRAD data, detecting which patients are at risk of an attack at any point in time cannot be limited to whoever is in the tertiary hospital clinic. A whole population approach is needed to reduce the risk of asthma attacks in all CYP. There is lack of evidence so far that the findings of the NRAD report have resulted in change. A Nuffield report into the health of young people found that the UK lagged far behind comparable countries in Europe in terms of asthma deaths. While the precise reasons for this are not well understood, the report highlights lack of basic access to healthcare and poor understanding of asthma risks by young people as two most likely factors. Asthma is currently managed predominantly in primary care, which is very appropriate for almost all patients. Some patients have difficult asthma and are at risk of an asthma attack, but predicting which patients out of a large population is difficult without better data. Mostly asthma can be managed with inhaled treatment (inhaled corticosteroid and a bronchodilator). There are many ways of delivering this medication and numerous approaches. There are also some additional treatments that are used less commonly. Patients currently attend regular clinic appointments in primary care or hospital clinic for review. This helps to assess the health of the child and level of symptom control. There is also an opportunity to discuss factors that make asthma worse and to educate on the ways to reduce the risk of an asthma attack. Patients and families will be given a plan for if there is an asthma attack to keep the patient safe. Those who are at high risk of an attack are currently only identified after that attack has taken place. This project aims to predict which patients are at high risk of an asthma attack and to intervene in a way that prevents that attack happening.Predicting which individuals are heading for an asthma attack and intervening to prevent this would have a significant impact on the health of this population and on the acute services that look after these cases: primary care, out of hours assessment, emergency departments, inpatient wards and critical care. A recent systematic review of studies in this area showed that the following factors (amongst others) are important in predicting asthma attacks in CYP: Previous asthma attack Persistent symptoms Sub-optimal preventor inhaler use Increased short term reliever use. Associated atopy/allergy Poverty Exhaled tobacco smoke exposure Most of these factors can be collected from existing health care data, however current systems do not generally provide risk stratification based on a pre-defined algorithm. Key to this challenge is the utilisation of existing healthcare and other relevant data for the risk stratification of individual CYP of asthma, and the subsequent measures that will modify this risk. In conjunction with a company experienced in healthcare data (redstar.ai), we propose building a clinical pathway that will utilise healthcare data and other remote patient data to stratify individuals into risk categories, and intervene in the highest risk patients to prevent an asthma attack. This system will utilise healthcare data such as primary care prescriptions, hospital episodes, blood test results etc along with patient symptom reporting and overnight cough monitoring to understand the individual's risk of asthma attack at any point in time. Thus, the clinician responsible for a population of people with asthma will be able to easily see which individuals are at higher risk and intervene appropriately. The intervention may be a phone call, text message, clinic appointment or even hospital admission depending on the situation. This contrasts with the current system of routine reviews and responding to asthma attacks after the event. The system will be a combination of a clinician dashboard and a patient-facing app. The dashboard will allow the responsible clinician to see which patients are at high risk of attack. The patient facing app will show the individual's asthma action plan, provide information on asthma, reporting of symptoms and cough monitoring at night (using the Hyfe cough detection software ). The clinicians using the dashboard will be based at the children's hospital asthma clinic and will communicate clinical information with primary care colleagues and patients and families. This project is a healthcare innovation project but also seeks to assess the effectiveness of this approach for patients in NHS Lothian. This will then help to justify this as a product to be used in Lothian. This approach may then be evaluated in a similar way in other Health Boards for the benefit of CYP across Scotland. We hope to use a new approach to monitoring asthma in a population of CYP with asthma in NHS Lothian. This pilot will be based around a single GP cluster. By including the whole population within a specified are a , rather than seeking to recruit individuals one by one, it is most representative of a normal NHS Lothian population. This is a non-randomised interventional study of a new risk stratification tool with associated clinical pathway and patient facing app. There is no randomisation of patients as this is an observation of the effects of this intervention on the whole population with asthma. Initially, all patients within a pre-defined population will have a novel risk algorithm applied. This system will monitor continuously to detect which individuals are at increased risk of an asthma attack. Identified individuals will be flagged to the responsible clinical team and contacted directly. The outcome of this contact will be determined by the clinician making the call. Where there is repeated flagging of high risk, poor adherence, or poor response to treatment an appointment will be offered in the children's hospital asthma clinic within 2 weeks for an MDT appointment. Following the initial clinical intervention, those flagged as high risk will be offered a patient facing app that provides information on asthma and a display of the standard asthma action plan. In addition, the app will monitor night time cough, step count and actively entered symptom score. Patients recruited to this stage of the study will provide consent given the additional data being collected. A GP cluster has been identified who are willing to be involved in using the new system. It is expected that the system will be piloted for 6 months to assess whether it is working and to troubleshoot problems. In this time we will be able to evaluate the above endpoints. This study seeks to assess a risk algorithm in the whole population of CYP with asthma. This approach, avoiding selection bias of limiting to those who will proactively sign up to the study, will be able to capture those at highest risk of attack - those less likely to seek medical help and attend the GP practice. Technology involved Redstar patient-facing app This is developed to allow patients and carers to enter information about their asthma. Available on play store and iOS app store and includes: Question about whether asthma is good or not good Asthma control test with graphical display of results Educational video links (asthma+lungUK content) Display of asthma action plan Two-way asynchronous messaging between individual and clinician The version applicable to the study will link information above with a clinician app (see below). In addition to the features above, there are additional features to be studied: Cough module - developed by Hyfe. Pedometer (internal personal device sensor) - step counter. GPS location (internal personal device sensor) - to provide environmental information e.g. pollen count, weather Information from the app is sent to the Redstar secure data storage. This system has been approved for use in NHS Lothian through a data governance process (Digital Protection Impact Assessment). This is not classed as a medical device as it does not provide management information other than a fixed version of the normal asthma action plan, written by the responsible clinician. Hyfe cough monitor This company have developed an app previously (CoughPro). While the monitoring within the patient app is similar it is designed specifically for this app. This feature is part of the Redstart app. Cough is monitored at night without compromising privacy. A short period of sound is analysed to determine whether it is a cough or not. This is not long enough to capture any information. Only the timestamp of the cough is sent from the app to the Redstar secure data storage. Clinician Dashboard Information from the patient app will be used to flag those at increased risk of asthma attack, based on pre-defined criteria. In addition, healthcare data will be used to flag those at increased risk of asthma attack, based on a data-driven algorithm. This algorithm has been developed using historic patient data (Dataloch) in a separate study. Patients who meet criteria for increased risk of asthma attack based on either patient app or healthcare data algorithm will be flagged on the clinician app, so a clinician led decision can be made about the response. This novel approach means patients will receive clinical care prior to having a severe asthma attack. The dashboard stratifies patients based on likelihood of asthma attack but does not offer advice or input on the management response. The responsibility for care lies with the clinician using the dashboard. Study assessments Initial screening: Eligible patients will be identified from a data query of coded diagnosis and prescription data for that GP cluster Initial enrolment - whole population: Eligible patients will be set up on the clinician dashboard with an algorithm applied to continually assess risk of asthma attack and report weekly by listing patients in order of most to least high risk.. If identified as high risk: Clinical pathway will intervene in those considered high risk according to the data algorithm analysis. Either primary care or hospital review will be arranged. The clinical response to increased risk will be within 1 week and is likely to be one of: Phone call from primary care to family to assess asthma control and determine whether primary care review is needed Phone call from asthma nurse specialist to assess asthma control Patients in this group will be offered an appointment to set up access to an enhanced version of the patient facing app with night time cough and daytime step counting. In addition, there will be a daily symptom questionnaire to complete. Data will be stored both on secure internal NHS systems (clinical dashboard for initial screening ) - and in an external cloud-based system (remotely collected patient data). Analysis of aggregated data to understand the benefits of both a whole population risk stratification and remotely collected data will be analysed within the NHS secure file environment. No biological samples will be collected. The app data will be collected through the patient app and held on secure cloud environment which will be penetration tested prior to deployment. DATA COLLECTION All CYP with asthma or managed as asthma will have specific data collected as detailed below to be viewed on the clinician dashboard. A clinician will view this and make a clinical decision on how to respond. The dashboard will show which individuals are at highest risk of attack, based on pre-defined criteria. Routine asthma prescriptions, emergency prescriptions and hospitalisations will be recorded via the clinician dashboard. This will be built on the NHS Innovation server with no patient identifiable data leaving this location. Downloading of non-identifiable data for purposes of analysis will be to NHS Servers (shared drive). A system of pseudo-anonymisation with a study number will further limit the identifiability of downloaded data. Downloading of data will be only by a clinician involved in the study. Data generated using the patient facing app will be securely stored within the Redstar.ai cloud based system. Linkage of records with app data will be using the study number. Identifiable data from the secure cloud will not be available outside this system. There is no ability for NHS data to be made available to the Redstar Cloud server. A record of CHI numbers and linked study numbers will be on the NHS drive viewable by the clinician only to allow cross checking and data validation. Data to be collected: Clinician dashboard with healthcare generated data: CHI Name Date of birth Postcode Medication prescriptions - asthma related only (inhalers, prednisolone, montelukast, antibiotics, antihistamines, steroid nasal spray) Blood tests related to asthma if available - eosinophil count, IgE and specific IgE, Skin prick testing if available Lung function if available Hospital admissions - emergency department, ward, critical care Interventions from the clinical team - e.g. phone call, clinic appointment, change in treatment Patient facing app Name Date of birth Medication - asthma related only. As entered by the patient and carer Medication adherence - patient generated values Cough analysis data Activity analysis - step count Symptom score - Asthma control test Source data will be Healthcare data used to assess asthma attack risk score will exist within NHS Lothian clinical systems. These are routine clinical data and would not be extracted. Specific clinical data will be used to build a picture of individuals to display to clinicians looking after them e.g. blood eosinophil count, medication list, date of clinical episodes. The source of such clinical data is within NHS Lothian clinical systems. All data entered in the patient facing app when recruited to the increased risk section of the study - symptom score, medication taken, sleep cough data, activity data (step count). Source data are held in the patient app, and then transferred to the secure cloud location Results of patient and carer questionnaires to evaluate the usefulness of the app will be completed digitally, with source data stored in a secure online location. Interactions with patients and families based on increased risk or change in circumstances will be recorded as normal in NHS Lothian clinical systems. DATA MANAGEMENT PERSONAL DATA The following personal data will be collected as part of the research: Whole population personal data: CHI number Name Age/Date of birth Postcode Selected high risk individuals additional information: Symptom score, night cough rate and step count Data from the patient facing app will be stored securely in a cloud based system Cough data Hyfe's cough monitoring technology (Hyfe - Detect \& Quantify Cough) has been IRB-approved in more than 50 studies around the world. Below we provide examples of content that may be useful as you build your own IRB proposal for studies involving Hyfe's AI. The cough monitoring technology on the RedStar's app is an app running on a study participant's smartphone. The app quantifies coughing by passively detecting and timestamping cough. Cough data (timestamps) is uploaded to RedStar's Cloud via available cellular network or wi-fi, and transmitted to researchers via a web-based dashboard in near real time. Privacy: Hyfe takes user privacy extremely seriously. The cough monitoring technology monitors sound level through the device's microphone. The first "peak detection" algorithm identifies 0.5 second explosive sounds that are similar to a cough sound. All processing happens on-device, that is, no acoustic data ever leaves the device. Then a second "cough recognition" algorithm is used to classify these "explosive sounds" between coughs and non-coughs. Only timestamps of the recognized cough sounds are sent to RedStar's Cloud. Cough data is transmitted to researchers via a password- protected research dashboard in near real time. DATA INFORMATION FLOW Patient information will be stored in a password protected Microsoft Excel file, to allow clinical data to be accessed from the clinical record. Extracted selected data will be viewed on the clinical facing dashboard, viewable only by the study clinicians. DATA STORAGE Personal identifiable data will be digitally stored by the research team using NHS Lothian secure shared drive. Personal data (pseudanonymised) will be physically stored by the research team at a digital cloud location managed by RedStar.ai Anonymised data will be physically stored by the research team on the NHS Lothian shared drive. Any paper will be immediately scanned in and stored within NHS Lothian server. DATA RETENTION All study documentation will be kept for a minimum of 3 years from the protocol defined end of study point. When the minimum retention period has elapsed, study documentation will be destroyed with permission from the Sponsor. Personal identifiable data will be stored for 6 months. Personal data (pseudonymised will be stored for 12 months. Anonymous data will be stored for 5 years.