Farr Institute Research Articles

A national initiative in data science for health: an evaluation of the UK Farr Institute.

ObjectiveTo evaluate the extent to which the inter-institutional, inter-disciplinary mobilisation of data and skills in the Farr Institute contributed to establishing the emerging field of data science for health in the UK.Design and Outcome measuresWe evaluated evidence of six domains characterising a new field of science:defining central scientific challenges,demonstrating how the central challenges might be solved,creating novel interactions among groups of scientists,training new types of experts,re-organising universities,demonstrating impacts in society.We carried out citation, network and time trend analyses of publications, and a narrative review of infrastructure, methods and tools.SettingFour UK centres in London, North England, Scotland and Wales (23 university partners), 2013-2018.Results1. The Farr Institute helped define a central scientific challenge publishing a research corpus, demonstrating insights from electronic health record (EHR) and administrative data at each stage of the translational cycle in 593 papers with at least one Farr Institute author affiliation on PubMed. 2. The Farr Institute offered some demonstrations of how these scientific challenges might be solved: it established the first four ISO27001 certified trusted research environments in the UK, and approved more than 1000 research users, published on 102 unique EHR and administrative data sources, although there was no clear evidence of an increase in novel, sustained record linkages. The Farr Institute established open platforms for the EHR phenotyping algorithms and validations (>70 diseases, CALIBER). Sample sizes showed some evidence of increase but remained less than 10% of the UK population in primary care-hospital care linked studies. 3.The Farr Institute created novel interactions among researchers: the co-author publication network expanded from 944 unique co-authors (based on 67 publications in the first 30 months) to 3839 unique co-authors (545 papers in the final 30 months). 4. Training expanded substantially with 3 new masters courses, training >400 people at masters, short-course and leadership level and 48 PhD students. 5. Universities reorganised with 4/5 Centres established 27 new faculty (tenured) positions, 3 new university institutes. 6. Emerging evidence of impacts included: > 3200 citations for the 10 most cited papers and Farr research informed eight practice-changing clinical guidelines and policies relevant to the health of millions of UK citizens.ConclusionThe Farr Institute played a major role in establishing and growing the field of data science for health in the UK, with some initial evidence of benefits for health and healthcare. The Farr Institute has now expanded into Health Data Research (HDR) UK but key challenges remain including, how to network such activities internationally.

Investigating the effect of England's smoke-free private vehicle regulation on changes in tobacco smoke exposure and respiratory disease in children: a quasi-experimental study

Comprehensive tobacco control policies can help to protect children from tobacco smoke exposure and associated adverse respiratory health consequences. We investigated the impact of England's 2015 regulation that prohibits smoking in a private vehicle with children present on changes in environmental tobacco smoke exposure and respiratory health in children. In this quasi-experimental study, we used repeated cross-sectional, nationally representative data from the Health Survey for England from Jan 1, 2008, to Dec 31, 2017, of children aged up to 15 years. We did interrupted time series logistic or ordinal regression analyses to assess changes in prevalence of self-reported respiratory conditions, prevalence of self-reported childhood tobacco smoke exposure (children aged 8-15 years only), and salivary cotinine levels (children aged 2 years or older) before and after implementation of the smoke-free private vehicle regulation on Oct 1, 2015. Children who were considered active smokers were excluded from the analyses of salivary cotinine levels. Our primary outcome of interest was self-reported current wheezing or asthma, defined as having medicines prescribed for these conditions. Analyses were adjusted for underlying time trends, quarter of year, sex, age, Index of Multiple Deprivation quintile, and urbanisation level. 21 096 children aged 0-15 years were included in our dataset. Implementation of the smoke-free private vehicle regulation was not associated with a demonstrable change in self-reported current wheezing or asthma (adjusted odds ratio 0·81, 95% CI 0·62-1·05; p=0·108; assessed in 13 369 children), respiratory conditions (1·02, 0·80-1·29; p=0·892; assessed in 17 006 children), or respiratory conditions probably affecting stamina, breathing, or fatigue (0·75, 0·47-1·19; p=0·220; assessed in 12 386 children). Self-reported tobacco smoke exposure and salivary cotinine levels generally decreased over the study period. There was no additional change in self-reported tobacco smoke exposure in cars among children aged 8-15 years following the legislation (0·77, 0·51-1·17; p=0·222; assessed in 5399 children). We observed a relative increase in the odds of children having detectable salivary cotinine levels post legislation (1·36, 1·09-1·71; p=0·0074; assessed in 7858 children) and levels were also higher (1·30, 1·04-1·62; p=0·020; ordinal variable). Despite introduction of the regulation, one in 20 children still reported being regularly exposed to tobacco smoke in cars and one in three still had detectable salivary cotinine levels. We found no demonstrable association between the implementation of England's smoke-free private vehicle regulation and changes in children's self-reported tobacco smoke exposure or respiratory health. There is an urgent need to develop more effective approaches to protect children from tobacco smoke in various places, including in private vehicles. Netherlands Lung Foundation, Erasmus MC, Farr Institute, Health Data Research UK, Asthma UK Centre for Applied Research, Academy of Medical Sciences, and Newton Fund.

Exploring barriers and solutions in advancing cross-centre population data science.

IntroductionIt is widely acknowledged that population health and administrative data, especially when linked at the individual level, hold great value for research. Cross-centre working between data centres providing access to such data has the potential to further increase this value by effectively expanding the data available for research. However, there is limited published information on how to address the challenges and achieve success. The aim of this paper is to explore perceived barriers and solutions to inform developments in cross-centre working across data centres.MethodsWe carried out a narrative literature review on data sharing and cross centre working. We used a mixed methods approach to assess the opinions of members of the public on cross-centre data sharing, and the views and experiences of among data centre staff connected with the UK Farr Institute for Health Informatics Research.ResultsThe literature review uncovered a myriad of practical and cultural issues. Our engagement with a public group suggested that cross-centre working involving anonymised data being moved between established centres is considered acceptable. The main themes emerging from discussions with data centre staff were dedicated resourcing, practical issues, information governance and culture.ConclusionIn seeking to advance cross-centre working between data centres, we conclude that there is a need for dedicated resourcing, indicators to recognise data reuse, collaboration to solve common issues, and balancing necessary barrier removal with incentivisation. This will require on-going commitment, engagement and an academic culture change.

Seasonal Influenza Vaccine Effectiveness for the Prevention of Laboratory-Confirmed Influenza in Asthma During the Influenza Seasons 2010-11 to 2015-16 in Scotland: A National Test-Negative Design Case-Control Study

Background: Influenza infection is an important trigger of asthma attacks. Influenza vaccination has the potential to reduce the incidence of influenza in people with asthma, but uptake remains persistently low, partially reflecting concerns about vaccine effectiveness (VE). Methods: We conducted a test-negative design case-control study to estimate the effectiveness of influenza vaccine in children and adults with asthma in Scotland over six influenza seasons (2010/11 to 2015/16). We used individual patient level data from 223 primary care practices which yielded 1,830,772 patient-years of data, which were linked with hospital and virological (n=5,910 swabs) data. Findings: Vaccination was associated with an overall 55.03% (95% confidence interval (CI): 45.84-62.66) reduction in the risk of a laboratory-confirmed influenza infection in people with asthma over the six seasons. There was substantial variation in VE between seasons, influenza strains and age groups. The highest VE (76.08%; 95% CI: 55.60-87.12) was found in 2010/11 season where the A(H1N1) strain dominated and there was a good antigenic vaccine match. High protection was observed against A(H1N1) (e.g. 2010/11: 70.68%; 95%CI: 32.46-87.27) and B strains (e.g. 2010/11: 83.17%; 95%CI: 44.31-94.92), but there was lower protection for the A(H3N2) strain (e.g. 2014/15: 26.40%; 95%CI: -12.00-51.64). The highest VE against all viral strains was observed in adults aged 18-54 years (56.98%; 95%CI: 42.25-67.95). Interpretation: Influenza vaccination gave meaningful protection against laboratory-confirmed influenza in people asthma across all six seasons. Strategies to boost influenza vaccine uptake has the potential to substantially reduce influenza triggered asthma attacks. Funding Statement: Chief Scientist's Office of the Scottish Government, National Institute for Health Research - Health Technology Assessment Programme (13/34/14), Asthma UK and the Medical Research Council's Farr Institute, Health Data Research UK (HDR UK). Declaration of Interests: The authors declare no competing interests. Ethics Approval Statement: The Privacy Advisory Committee of the Information Services Division, NSS, approved the linkage and the statistical analysis of the anonymised data used in this study.

A case study in distributed team science in research using electronic health records.

IntroductionDue to various regulatory barriers, it is increasingly difficult to move pseudonymised routine health data across platforms and among jurisdictions. To tackle this challenge, we summarized five approaches considered to support a scientific research project focused on the risk of the new non-vitamin K Target Specific Oral Anticoagulants (TSOACs) and collaborated between the Farr institute in Wales and Scotland.ApproachIn Wales, routinely collected health records held in the Secure Anonymous Information Linkage (SAIL) Databank were used to identify the study cohort. In Scotland, data was extracted from national dataset resources administered by the eData Research & Innovation Service (eDRIS) and stored in the Scottish National Data Safe Haven. We adopted a federated data and multiple analysts approach, but arranged simultaneous accesses for Welsh and Scottish analysts to generate study cohorts separately by implementing the same algorithm. Our study cohort across two countries was boosted to 6,829 patients towards risk analysis. Source datasets and data types applied to generate cohorts were reviewed and compared by analysts based on both sites to ensure the consistency and harmonised output.DiscussionThis project used a fusion of two approaches among five considered. The approach we adopted is a simple, yet efficient and cost-effective method to ensure consistency in analysis and coherence with multiple governance systems. It has limitations and potentials of extending and scaling. It can also be considered as an initialisation of a developing infrastructure to support a distributed team science approach to research using Electronic Health Records (EHRs) across the UK and more widely.

International Consensus Statement on Public Involvement and Engagement with Data-Intensive Health Research

IntroductionData-intensive health research is a fast moving field in which public involvement and engagement (PI\&E) is essential for developing socially acceptable and ethically robust processes and ensuring a social license for research practices. Nevertheless while some consensus is emerging around the importance of PI\&E, commitments and practices are varied.
 Objectives and ApproachOur objective was to identify a set of principles which would underpin international best practice in data-intensive health research.
 A one-day consensus workshop was held in Manchester, U.K. in April 2017 and attended by 31 international participants from six countries, with a range of expertise relating to PI&E with data-intensive health research. Participants took part in facilitated discussions to identify key principles and a consensus statement was written to reflect the discussions at the workshop, this developed through iterative drafts on which all workshop participants as well as the four Farr Institute public panels were consulted.
 ResultsThe consensus statement sets out key principles to establish a secure role for PI&E related to data-intensive health research internationally and to ensure best practice in its execution. While methods used may not vary greatly from PI&E with other types of research, or other policy areas, we have identified a number of particular features of data-intensive health research which make PI&E in this area worthy of special consideration. Often using data from multiple sources without explicit consent, PI&E in data-intensive health research is particularly important for raising awareness, for giving people a voice and for enabling people to participate in processes of research and governance. The consensus statement sets out the key principles to guide future best practice in PI&E with data-intensive health research.
 Conclusion/ImplicationsThe emergence of data intensive health research and the importance of the social contract upon which it relies, demands that we move beyond rhetorical commitments and engage anew with clearly stated principles to build PI\&E into data-intensive health research at all levels.

Advancing cross-centre research networks: learning from experience, looking to the future

IntroductionMany jurisdictions have programmes for the large-scale reuse of health and administrative data that would benefit from greater cross-centre working. The Advancing Cross centre Research Networks (ACoRN) project considered barriers and drivers for joint working and information sharing using the UK Farr Institute as a case study, and applicable widely.
 Objectives and ApproachACoRN collected information from researchers, analysts, academics and the public to gauge the acceptability of sharing data across institutions and jurisdictions. It considered international researcher experiences and evidence from a variety of cross centre projects to reveal barriers and potential solutions to joint working. It reviewed the legal and regulatory provisions that surround data sharing and cross-centre working, including issues of information governance to provide the context and backdrop. The emerging issues were grouped into five themes and used to propose a set of recommendations.
 ResultsThe five themes identified were: organisational structures and legal entities; people and culture; information governance; technology and infrastructure; and finance and strategic planning. Recommendations within these included: standardised terms and conditions including agreements and contractual templates; performance indicators for frequency of dataset sharing; communities of practice and virtual teams to develop cooperation; standardised policies and procedures to underpin data sharing; an accredited quality seal for organisations sharing data; a dashboard for data availability and sharing; and adequate resource to move towards greater uniformity and to drive data sharing initiatives.
 Conclusion/ImplicationsThe challenges posed by cross-centre information sharing are considerable but the public benefits associated with the greater use of health and administrative data are inestimable, particularly as novel and emerging data become increasingly available. The proposed recommendations will assist in achieving the benefits of cross-centre working.

A tool to improve the efficiency and reproducibility of research using electronic health record databases

BackgroundInterrogation of electronic health record databases often involves time-consuming, manual, repetitive work in developing database queries. We developed a tool to automate this process.
 MethodsWe identified elementary approaches to query primary care data from the Secure Anonymised Information Linkage databank of Wales. We designed a web-based query builder that allows using combinations of these approaches as ‘building blocks’ to query complex variables. We created an R programme to automatically generate and execute the corresponding Structured Query Language queries.
 ResultsThe tool allows data extraction using combinations of the following methods: event count (e.g., asthma prescriptions); code/date of earliest/latest event; code/date/value of the event of maximum/minimum value; and frequency of temporally constrained events. Query intervals could be fixed, dynamic, or individualised. The tool integrates with a codeset repository. Data extraction procedures and codesets are saved on a web server as versioned, shareable, and citable objects.
 ConclusionThis versatile tool allows rapid and complex data extraction with minimal to no programming skills, reduces human errors, and improves research transparency and reproducibility.
 Funding/SupportHealth and Care Research Wales, ABMU Health Board, AUKCAR (AUK-AC-2012-01), Farr Institute of Health Informatics Research (MR/K006525/1-MR/K007017/1).

Child mortality in England compared with Sweden: a birth cohort study

SummaryBackgroundChild mortality is almost twice as high in England compared with Sweden. We aimed to establish the extent to which adverse birth characteristics and socioeconomic factors explain this difference.MethodsWe developed nationally representative cohorts of singleton livebirths between Jan 1, 2003, and Dec 31, 2012, using the Hospital Episode Statistics in England, and the Swedish Medical Birth Register in Sweden, with longitudinal follow-up from linked hospital admissions and mortality records. We analysed mortality as the outcome, based on deaths from any cause at age 2–27 days, 28–364 days, and 1–4 years. We fitted Cox proportional hazard regression models to estimate the hazard ratios (HRs) for England compared with Sweden in all three age groups. The models were adjusted for birth characteristics (gestational age, birthweight, sex, and congenital anomalies), and for socioeconomic factors (maternal age and socioeconomic status).FindingsThe English cohort comprised 3 932 886 births and 11 392 deaths and the Swedish cohort comprised 1 013 360 births and 1927 deaths. The unadjusted HRs for England compared with Sweden were 1·66 (95% CI 1·53–1·81) at 2–27 days, 1·59 (1·47–1·71) at 28–364 days, and 1·27 (1·15–1·40) at 1–4 years. At 2–27 days, 77% of the excess risk of death in England was explained by birth characteristics and a further 3% by socioeconomic factors. At 28–364 days, 68% of the excess risk of death in England was explained by birth characteristics and a further 11% by socioeconomic factors. At 1–4 years, the adjusted HR did not indicate a significant difference between countries.InterpretationExcess child mortality in England compared with Sweden was largely explained by the unfavourable distribution of birth characteristics in England. Socioeconomic factors contributed to these differences through associations with adverse birth characteristics and increased mortality after 1 month of age. Policies to reduce child mortality in England could have most impact by reducing adverse birth characteristics through improving the health of women before and during pregnancy and reducing socioeconomic disadvantage.FundingThe Farr Institute of Health Informatics Research (through the Medical Research Council, Arthritis Research UK, British Heart Foundation, Cancer Research UK, Chief Scientist Office, Economic and Social Research Council, Engineering and Physical Sciences Research Council, National Institute for Health Research, National Institute for Social Care and Health Research, and the Wellcome Trust).

Assessment of health care, hospital admissions, and mortality by ethnicity: population-based cohort study of health-system performance in Scotland

SummaryBackgroundEthnic minorities often experience barriers to health care. We studied six established quality indicators of health-system performance across ethnic groups in Scotland.MethodsIn this population-based cohort study, we linked ethnicity from Scotland's Census 2001 (April 29, 2001) to hospital admissions and mortality records, with follow-up until April 30, 2013. Indicators of health-system performance included amenable deaths (ie, deaths avertable by effective treatment), preventable deaths (ie, deaths avertable by public health policy), avoidable deaths (combined amenable and preventable deaths), avoidable hospital admissions, unplanned readmissions, and length of stay. We calculated rate ratios and odds ratios (with 95% CIs) using Poisson and logistic regression, which we multiplied by 100, adjusting first for age-related covariates and then for socioeconomic-related and birthplace-related covariates. The white Scottish population was the reference (rate ratio [RR] 100).FindingsThe results are based on 4·61 million people. During the 50·5 million person-years of study, 1·17 million avoidable hospital admissions, 587 740 unplanned readmissions, and 166 245 avoidable deaths occurred. South Asian groups had higher avoidable hospital admissions than the white Scottish group, with the highest reported RRs in Pakistani groups (RR 140·6 [95% CI 131·9–150·0] in men; RR 141·0 [129·0–154·1] in women). There was little variation between ethnic groups in length of stay or unplanned readmission. Preventable and amenable mortality were higher in the white Scottish group than several ethnic minorities including other white British, other white, Indian, and Chinese groups. Such differences were partly diminished by adjustment for socioeconomic status, whereas adjustment for country of birth had little additional effect.InterpretationThese data suggest concerns about the access to and quality of primary care to prevent avoidable hospital admissions, especially for south Asians. Relatively high preventable and amenable deaths in white Scottish people, compared with several ethnic minority populations, were unexpected. Future studies should both corroborate and examine explanations for these patterns. Studies using several indicators simultaneously are also required internationally.FundingChief Scientist's Office, Medical Research Council, NHS Research Scotland, Farr Institute.

A National Initiative in Data Science for Health: An Evaluation of the UK Farr Institute

Objective: To evaluate the extent to which the inter-institutional, inter-disciplinary mobilisation of data and skills in the Farr Institute contributed to establishing the emerging field of data science for health in the UK. Design and Outcome measures: We evaluated evidence of six domains characterising a new field of science: defining central scientific challenges, demonstrating how the central challenges might be solved, creating novel interactions among groups of scientists, training new types of experts, re-organising universities, demonstrating impacts in society. We carried out citation, network and time trend analyses of publications, and a narrative review of infrastructure, methods and tools. Setting Four UK centres in London, North England, Scotland and Wales (23 university partners), 2013-2018. Population Subsets of the UK’s 65 million population with health records accessible for research. The Farr Institute published a research corpus around a central scientific challenge, demonstrating insights from electronic health record (EHR) and administrative data at each stage of the translational cycle in 593 papers with at least one Farr Institute author affiliation on PubMed. Sample sizes showed some evidence of increase but remained less than 10% of the UK population in primary care-hospital care linked studies. The Farr Institute established the first four ISO27001 certified trusted research environments in the UK, and approved more than 1000 approved research users, published on 102 unique EHR and administrative data sources and established open platforms for the scalable re-use of EHR phenotyping (>70 diseases, CALIBER). The co-author publication network expanded from 944 unique co-authors (based on 67 publications in the first 30 months) to 3839 unique co-authors (545 papers in the final 30 months). 4/5 Centres established 27 new faculty (tenured) positions, 3 new university institutes, and 3 new master’s courses, training >400 people at master’s, short-course and leadership level and 48 PhD students. There were over 2300 citations for the 10 most cited papers and Farr research informed eight practice-changing clinical guidelines and policies relevant to the health of millions of UK citizens. Conclusion: The Farr Institute played a major role in establishing and growing the field of data science for health in the UK, with some initial evidence of benefits for health and healthcare. The Farr Institute has now expanded into Health Data Research UK but a key challenge is to network such activities internationally. What is already known: • National research initiatives in data science for health are under way in several countries seeking to harness insights from electronic health record (EHR) and administrative data at regional and national scale for patient and public benefit. • One approach to grow this emerging field, adopted by the UK, is to establish a dedicated national research institute, the Farr Institute. • We do not know how effective such initiatives are: multi-centre, inter-disciplinary research initiatives are common, but there is a lack of research evaluating such initiatives (in general) and national research institutes (in particular). • The Farr Institute ran from 2013 until 2018 when its larger-scale successor, Health Data Research UK, was established. What this study adds: • We provide a framework of six domains relevant for evaluating new inter-institutional, inter-disciplinary initiatives seeking to establish and grow an emerging field of science: defining central scientific challenges, demonstrating how the central challenges might be solved, creating novel interactions among groups of scientists, training new types of experts, re-organising universities, demonstrating impacts in society. • We show that the Farr Institute created new activities in and across each of the six domains for developing a distinctive research field. • The Farr Institute demonstrated the ability of multiple UK health research funders and multiple universities to partner in mobilising data, methodology and expertise across disciplines, organisations and information governance domains, resulting in a larger scale of research and improved methodology. • We have demonstrated globally relevant challenges and opportunities for developing data science for health across institutional and disciplinary barriers, consistent with the need for big investigation not simply big data.

Risk factors for Escherichia coli bacteraemia: a population-based case-control study

BackgroundEscherichia coli bacteraemia is associated with high morbidity and mortality and is increasing in incidence. This study investigated putative risk factors associated with development of E coli bacteraemia, to inform prevention strategies. MethodsA population-based case-control study was conducted from de-identified linked data sources in Wales. Pseudonymised blood microbiology culture reported from all Welsh laboratories between Jan 1, 2005, and Dec 31, 2011, were included in the Secure Anonymised Information Linkage databank and linked to other routinely collected health records. Cases were individuals with E coli-positive blood cultures. 20 sets of controls were randomly selected from Welsh demographic service data. Records were linked with Patient Episode Database for Wales (inpatient) and Welsh general practice data to identify risk and potential confounding factors, including demographic characteristics, comorbidity score, and clinical factors. Both cases and controls needed to be residents of Wales on the date the E coli-positive blood sample of the case was received (reference date), and to have lived in Wales during the 91 days before the reference date. Conditional logistic regression modelling was used to identify factors associated with developing bacteraemia. FindingsThere were 10 815 cases over 7 years and 216 300 controls. Factors associated with a particularly high risk of bacteraemia included renal infection (odds ratio 145·7, 95% CI 72·2–294·0), laboratory-confirmed urinary tract infection (21·5, 19·6–23·7), conditions with likely hospital antibiotics prescription (16·5, 15·0–18·0), and high comorbidity score category (16·0, 14·7–17·5). 12 other factors had significantly elevated risks. InterpretationThis study has identified particular categories at substantially higher risk of developing bacteraemia. This information will be used in developing targeted interventions. Continuous or repeated analysis could be used to evaluate effectiveness of the interventions. The use of comprehensive linked records avoids selection biases and creates matched population based controls. Privacy-protecting data linkage allows for efficient retrospective follow-up of patients. However, in studies with e-records, there can be incomplete control of confounding, despite the large number of data and datasets used. The presumed hospital prescribing variable was defined from a diagnosis that should result in treatment due to absence of hospital prescription data. FundingPublic Health Wales NHS Trust, Farr Institute.

Identification of predictors of objectively measured physical activity in 12-month-old British infants: a machine learning driven study

BackgroundDespite physical activity playing an important part in enhancing and maintaining good health throughout the life course, little is known about it in young children. This study aimed to identify predictive factors associated with intensity of physical activity in 12-month-old infants. MethodsParents of participants in the Environments for Healthy Living birth cohort study were approached. We collected data on physical activity from 141 infants using ankle-mounted tri-axial accelerometers (GENEActiv, Kimbolton, UK) in the unit of signal vector magnitude (SVM) score over 7 consecutive days, and linked the data to abstracts of postnatal notes and electronic medical records. To identify factors associated with intensity of physical activity, a machine learning framework was used to process the complex data, in which a multivariate linear regression model was fitted on a set of selected predictors using the ReliefF feature ranking algorithm to distinguish lower and higher active groups. FindingsThe physical activity level of mean SVM score ranged from 0·677 to 4·932 in the lower active group (n=70) compared with 4·975 to 10·628 in the higher active group (n=71)]There were more boys than girls in the active group (49 vs 22) and more girls than boys in the inactive group (42 vs 28; r2=0·615). High intensity of physical activity in 12-month-old infants was associated with larger infant length (12 months; 95% CI 13·47 to 29·38), higher maternal blood pressure (12 months; 5·54 to 27·70), longer gestation period (13·75 to 34·47), higher breastfeeding duration (weeks; 2·33 to 12·84,), more movement during sleep (1·25 to 2·32), and high intake of vegetables (0·19 to 4·44) and juice (0·04 to 1·67) and less consumption of adult crisps (0·05 to 0·73). InterpretationThe findings suggest that at age 12 months boys are more active than girls and that both gestational and postnatal environment factors, such as preterm birth (<37 weeks), dietary habits, and longer breastfeeding duration, are important predictors of intensity of physical activity in 1 year olds. However, a limitation of this study is the requirement for infants to wear the accelerometer for 7 full days; all the children with fewer than 7 days' wear were excluded from the study. These findings can help inform interventions to promote healthier lifestyles in young families. FundingGrowing Up in Wales birth cohort, National Centre for Population Health and Wellbeing Research (CA02), Centre for Improvement in Population Health through E-records Research (CIPHER) within the Farr Institute of Health Informatics Research (MR/K006525/1).

Socioeconomic deprivation and inequalities in asthma care in Wales

BackgroundArea-based deprivation indices are widely used to study health inequalities. We explored whether inequality exists for asthma care across socioeconomic deprivation levels in Wales. MethodsUsing the Wales-wide Secure Anonymised Information Linkage databank, we accessed anonymised data on patients with a general practitioner (GP) diagnosis of asthma in or before 2009 and continuous GP registration between 2009 and 2014, and linked these data to the 2011 Welsh Index of Multiple Deprivation quintiles. The primary outcome variables were numbers of asthma-related GP visits, annual reviews, emergency department visits, and hospital admissions between 2010 and 2014. Statistical analysis included zero-inflated negative binomial regression adjusted for age and sex. No ethics approval was needed for use of anonymised data. FindingsWe analysed data from 209 887 patients with asthma. Over 5 years, there were 8·2% more asthma-related GP visits per patient in the most deprived areas than in the least deprived areas (4·0 vs 3·7; incidence rate ratio 1·082, 95% CI 1·071–1·094), 3·7% more asthma annual reviews (2·35 vs 2·26; 1·037, 1·026–1·049), 31·1% more asthma emergency department visits (0·021 vs 0·016; 1·311, 1·151–1·494), and 123·3% more asthma inpatient admissions (0·074 vs 0·033; 2·233, 1·983–2·515) (all p<0·0001). InterpretationHigher health-care utilisation in the most deprived group indicates worse asthma control. Despite an 8·2% excess of primary care contacts in this group, outcomes are poorer in terms of much higher emergency department visits and hospital admissions. Disease severity, quality of primary care, self-management, or wider non-health-care-related determinants may be contributing factors, indicating further potential to reduce inequality in the impact of disease. FundingHealth and Care Research Wales, Abertawe Bro Morgannwg University Health Board, Asthma UK Centre for Applied Research (AUK-AC-2012-01), Farr Institute of Health Informatics Research (CIPHER MR/K006525/1, Scotland MR/K007017/1).

Mining electronic health records to identify predictive factors associated with hospital admission for Campylobacter infections

BackgroundCampylobacter infection is one of the most common bacterial infections in human beings. Most cases of Campylobacter infection are not well explained by commonly recognised risk factors. Data-driven clinical rules offer patients an unprecedented measure of control over their own health information. This study sought to identify influential predictors from general practice (GP) data to predict the outcomes of Campylobacter infections—ie, being admitted to hospital or remaining with GP care. MethodsDiagnosis of Campylobacter infection was collected from GP and hospital data in Wales between Jan 1, 1990, and Dec 31, 2015, via the Secure Anonymised Information Linkage databank. A specially designed feature selection scheme was used to select the most influential factors from a large number of initial variables formed by GP read codes (a hierarchical thesaurus of clinical terms) and from other codes (sex, age, location, and Townsend deprivation scores). Information retrieval and decision tree techniques were used to construct predictive models. The linked data were split into training, validation, and testing subsets. The testing dataset was used to assess the predictive performance of the identified risk factors. FindingsHealth records of 20 376 patients with Campylobacter infections were used, of whom 12 324 were admitted to hospital at least once. Higher rates of infections occurred in male patients (10 793/20 376), but female patients had higher rates of being admitted to hospital (6207 of 9583 admissions). From an initial 33 471 variables, 31 most predictive variables were finally identified and aggregated, such as “second pneumococcal conjugated vaccination”, and “stool culture. 30 if–then clinical rules with these 31 variables were generated by decision tree (eg, if patients had “second pneumococcal conjugated vaccination”, then “not admitted”). These rules were applied to an independent test dataset and performed significantly above chance to predict the outcomes (sensitivity 0·642, 95% CI 0·638–0·644; specificity 0·62, 0·616–0·622; positive prediction value 0·724, 0·72–0·73). InterpretationOur findings suggest that construction of machine learning models from routine GP data enables the influential risk factors identified to predict outcomes of Campylobacter infections significantly above chance. To address some unexplained model variations, incorporation of more information about the infection and symptoms (strain, severity) is required. FundingNational Centre for Population Health and Wellbeing Research (CA02), Centre for Improvement in Population Health through E-records Research (CIPHER) within the Farr Institute of Health Informatics Research (MR/K006525/1).

Effect of tobacco control policies on perinatal and child health: a systematic review and meta-analysis

SummaryBackgroundTobacco smoking and smoke exposure during pregnancy and childhood cause considerable childhood morbidity and mortality. We did a systematic review and meta-analysis to investigate whether implementation of WHO's recommended tobacco control policies (MPOWER) was of benefit to perinatal and child health.MethodsWe searched 19 electronic databases, hand-searched references and citations, and consulted experts to identify studies assessing the association between implementation of MPOWER policies and child health. We did not apply any language restrictions, and searched the full time period available for each database, up to June 22, 2017. Our primary outcomes of interest were perinatal mortality, preterm birth, hospital attendance for asthma exacerbations, and hospital attendance for respiratory tract infections. Where possible and appropriate, we combined data from different studies in random-effects meta-analyses. This study is registered with PROSPERO, number CRD42015023448.FindingsWe identified 41 eligible studies (24 from North America, 16 from Europe, and one from China) that assessed combinations of the following MPOWER policies: smoke-free legislation (n=35), tobacco taxation (n=11), and smoking cessation services (n=3). Risk of bias was low in 23 studies, moderate in 16, and high in two. Implementation of smoke-free legislation was associated with reductions in rates of preterm birth (–3·77% [95% CI −6·37 to −1·16]; ten studies, 27 530 183 individuals), rates of hospital attendance for asthma exacerbations (–9·83% [–16·62 to −3·04]; five studies, 684 826 events), and rates of hospital attendance for all respiratory tract infections (–3·45% [–4·64 to −2·25]; two studies, 1 681 020 events) and for lower respiratory tract infections (–18·48% [–32·79 to −4·17]; three studies, 887 414 events). Associations appeared to be stronger when comprehensive smoke-free laws were implemented than when partial smoke-free laws were implemented. Among two studies assessing the association between smoke-free legislation and perinatal mortality, one showed significant reductions in stillbirth and neonatal mortality but did not report the overall effect on perinatal mortality, while the other showed no change in perinatal mortality. Meta-analysis of studies on other MPOWER policies was not possible; all four studies on increasing tobacco taxation and one of two on offering disadvantaged pregnant women help to quit smoking that reported on our primary outcomes had positive findings. Assessment of publication bias was only possible for studies assessing the association between smoke-free legislation and preterm birth, showing some degree of bias.InterpretationSmoke-free legislation is associated with substantial benefits to child health. The majority of studies on other MPOWER policies also indicated a positive effect. These findings provide strong support for implementation of such policies comprehensively across the world.FundingChief Scientist Office Scotland, Farr Institute, Netherlands Lung Foundation, Erasmus MC.

Tomorrow’s Health Data Scientists in the Making: 2 Years of Training &amp; Capacity Building Activities with the Farr Institute

ABSTRACT&#x0D; ObjectivesThe poster will showcase the Training &amp; Capacity Building Programme (TCBP) established by the Farr network that is available to PhD students and Early Career Researchers (ECRs) both from within and outside the network.&#x0D; ApproachThe authors are using a mixed-methods approach to present the current state of the art that has motivated the structure of the TCBP and how the programme has been received by students, academic and industry leaders.&#x0D; Since 2014, the Farr network education leads have worked on a unique education programme that aims to equip students with key professional and methodological skills required by the roles they are likely to take in their future careers.&#x0D; The content of the programme has been informed by current research in the fields of data linkage and analysis of big datasets as well as the immediate experience of researchers, and healthcare and industry leads working actively in the field.&#x0D; We have focussed on developing a community of practice through the provision of an environment that has enabled students to share experiences and knowledge and to start building their own networks of collaboration across the Farr centres.&#x0D; ResultsThe Farr network of PhD students comprises of approximately 66 students from a variety of backgrounds including bioinformatics, computer science, epidemiology, mathematics, psychology and statistics.&#x0D; The Farr network hosted 2 main events in 2015 that provided the foundation for 34 PhD students to come together to develop skills in written and oral communication, Public and Patient Involvement (PPI), problem-solving, requirements gathering, data visualisation and spatial data analysis.&#x0D; Formal feedback collected from each event suggests that the programme is received well by students who see it as a set of activities that complement their specialised PhD related education. Feedback from supervisors and future employers also suggests that this programme has facilitated the development of skills that are useful for their PhD and employment.&#x0D; ConclusionThe Farr network has laid the foundation for the community of practice around the analysis of real-world health datasets across the UK, and will continue to cultivate this through the enrichment of the programme.

Identifying Household Level Risk Factors for Unintentional House Fire Incidents, Injuries and Fatalities

ABSTRACTObjectivesUnintentional house fire incidents, injuries and deaths are a serious public health concern in the UK, which disproportionally affect certain groups in the population. Whilst house fires have decreased in recent years; growing financial pressures in the Fire and Rescue Services (FRSs) have resulted in funds dedicated to fire preventative activities becoming increasingly limited. To ensure ever limiting resources are targeted towards those households at greatest risk, it is essential the FRSs’ are accurately informed about the types of household at increased risk. The aim of this project is to undertake a large-scale case-control study, to identify the distinguishing household level risk factors associated with unintentional house fire incidents, injuries and deaths.&#x0D; ApproachUnintentional house fire incidents reported to the Welsh FRS between the years 2003-2008, were anonymised and incorporated into the Secure Anonymised Information Linkage (SAIL) Databank at the Farr Institute, Swansea University. 6943 case households (households which reported a fire to the FRS) were time-matched to 347,150 control households (case:control ratio 1:50). Individuals registered as living at these properties on the date of the fire were established using the Welsh Demographic Service (WDS) dataset. Household level variables will be created by linking case and control households to other demographic, health, educational and environmental datasets in SAIL. Conditional Logistic Regression will be used to estimate matched odds ratios and 95% confidence intervals.&#x0D; ResultsPotential risk factor variables were selected on the basis of a systematic review and theoretically plausible variables. Covariates include: household composition (e.g. age and gender of residents), socioeconomic status, educational attainment, smoking, alcohol consumption, mental health conditions, other health related conditions, mobility and sensory impairments and property related characteristics. Fire related circumstances (e.g. fire ignition source, presence of a smoke alarm) will also be investigated in logistic regression models exploring risk factors for injury and death. Results will be presented at the conference.&#x0D; ConclusionThis is the first large-scale analysis of risk factors for unintentional house fire incidents, injuries and deaths. The findings from this project will be translated into comprehensible infographics, designed to support the FRSs, other partner organisations and the general public, recognise high risk households in need of preventative interventions.

Looked after children and access to dental services and oral health in Scotland: a national data-linkage study (LAC-DENTAL)

ABSTRACTObjectivesChildren that are ‘looked after’ include those that are accommodated in foster, kinship and residential care placements, as well as those at home on compulsory supervision. They have poorer physical and mental health than their peers and there are concerns about the relatively high levels of untreated morbidity. Oral health and access to dental services among Looked After Children (LAC) has received limited attention to date. The objective of this study was to compare the oral health and access to dental services of children who are looked after by the state, with comparable children in the general population.&#x0D; ApproachSchool and Social Work datasets were able to be linked using the Scottish Exchange of Data (ScotXed) Unit. This in turn was linked with health data making use of the Scottish national record-linkage system provided by the NHS eDRIS team for the FARR Institute Scotland. All of the following datasets used in this study are complete national datasets for the time periods noted. School Pupil Census 2012: a census of children in local authority primary and secondary schools that provides each child’s age, sex and socioeconomic status as measured by the Scottish Index of Multiple Deprivation (SIMD). The other datasets cover the period 2008-2012: LAC- all children with social work referrals for various types of placement; registration with dentists; hospital discharge data for all episodes of tooth extraction; National Dental Inspection Programme (NDIP) data from Primary 1 and Primary 7 school years (dental decay). The LAC group were compared to their peers by logistic regression adjusted by SIMD, using remote access to the National Safe Haven.&#x0D; ResultsThere were 633204 subjects in the study group (10927 LAC, 622280 nonLAC). Ages ranged from four to 17 years (mean 12 LAC and 10 nonLAC); with 53% male for LAC and 51% nonLAC; and 42% in the most deprived SIMD level for LAC and 21% for non-LAC. The subjects in the LAC group were more likely to have dental decay at Primary 1, odds-ratio (OR) 2.93 (2.55, 3.38), and Primary 7, OR 1.81 (1.68, 1.94). LAC subjects were less likely to be registered with a dentist, OR 0.50 (0.43, 0.59), and more likely to have teeth extracted, OR 1.50 (1.40, 1.60). All tests p&lt;0.001.&#x0D; ConclusionLooked after children are more likely to have dental problems and less likely to use dental services than their peers, after adjustment for socioeconomic status.

Psychiatric research platform within SAIL - Linkage of prospectively ascertained phenotypically rich and genetic data to routinely collected anonymised records.

ABSTRACT&#x0D; ObjectivesWe have conducted a feasibility study linking clinically rich survey data to routine data to create a platform for psychosis research in Wales: K Lloyd et al (2015), A national population-based e-cohort of people with psychosis (PsyCymru) linking prospectively ascertained phenotypically rich and genetic data to routinely collected records: overview, recruitment and linkage, Schizophrenia Research. Now we expand upon this through the linkage of large clinically rich cohorts with a range of mental health diagnoses along with genetic data to conduct validation exercises, develop novel methodologies, assess genetic and environment interactions and outcomes and address hypothesis-driven research questions.&#x0D; ApproachThrough collaborations between the Farr Institute, Cardiff University based MRC centre for Neuropsychiatric Genetics and Genomics and the National Centre for Mental Health (NCMH) clinically rich data and genetic (CNVs, SNPs &amp; polygenic scores) data from around 6000+ participants recruited from a variety of mental health research studies including ‘PsyCymru’, ‘Genetic susceptibility to cognitive deficits study and NCMH amongst others will be loaded and linked to the datasets within SAIL. The analysis plan would firstly include validation exercises to compare the data between sources. Methodologies would be developed using this data to determine illness onset, relapse, chronicity, severity and response to treatment applied to large population-based mental health e-cohorts.&#x0D; ResultsBy pooling together health service data, genetic variants, environmental and lifestyle factors, phenotypic and endo-phenotypic (cognitive scores) along with the ability to ascertain temporal relationships afforded by the longitudinal perspective available in SAIL we may be able to evaluate potential risk factors, assess the complex GxE interactions that lead to disease progression, and assess outcomes such as prognosis, remission, relapse and premature mortality. The on-going routine updates provide us with the opportunity to follow-up these individuals across multiple health care settings in a cost effective and in-obtrusive manner and to carry out health services utilization/benefit and treatment surveillance in a naturalistic setting. This resource will continue to expand over the coming years in size, breadth and depth of data, with continued recruitment and additional measures planned.&#x0D; ConclusionTo advance mental health research by developing our understanding of the causes, course and outcomes of mental illness that may lead to the development of better diagnostic classification, predictive, preventative strategies and therapeutic approaches.