Chris Dibben

The Role of Administrative Data in the Big Data Revolution in Social Science Research

The term big data is currently a buzzword in social science, however its precise meaning is ambiguous. In this paper we focus on administrative data which is a distinctive form of big data. Exciting new opportunities for social science research will be afforded by new administrative data resources, but these are currently under appreciated by the research community. The central aim of this paper is to discuss the challenges associated with administrative data. We emphasise that it is critical for researchers to carefully consider how administrative data has been produced. We conclude that administrative datasets have the potential to contribute to the development of high-quality and impactful social science research, and should not be overlooked in the emerging field of big data.

Area deprivation, individual factors and low birth weight in England: is there evidence of an "area effect"?

Objective: To explore the relationship between low and very low birth weights, mother’s age, individual socioeconomic status and area deprivation. Design: Analysis of the incidence of low and very low birth weights by area deprivation, maternal age, social class of household and estimated income. Setting: England 1996–2000. Subjects: 2 894 440 singleton live births and the 10% sample of these births for which parents’ individual-level socioeconomic measures were coded. Results: Social class, estimated household income, lone-parenthood and mother’s age were all associated with the risk of low and very low birth weight. Even when controlling for these individual level factors, area income deprivation was significantly associated with low and very low birth weight (p<0.00). For low birth weight there was a significant interaction between area income deprivation and mother’s age. For very young mothers, the area effect was non-significant (p<0.37). For older mothers, particularly those aged 30–34 years, it was stronger (p<0.00). As a result, mothers aged <18 years, although at relatively high risk of low birth weight irrespective of area income deprivation, were actually at slightly lower risk than mothers aged >40 years in the most deprived areas. Conclusions: For all but very young mothers, there seems to be a negative effect on birth weight from living in areas of income deprivation, whatever their individual circumstances.

Changes in mortality inequalities over two decades: register based study of European countries

Objective To determine whether government efforts in reducing inequalities in health in European countries have actually made a difference to mortality inequalities by socioeconomic group. Design Register based study. Data source Mortality data by level of education and occupational class in the period 1990-2010, usually collected in a census linked longitudinal study design. We compared changes in mortality between the lowest and highest socioeconomic groups, and calculated their effect on absolute and relative inequalities in mortality (measured as rate differences and rate ratios, respectively). Setting All European countries for which data on socioeconomic inequalities in mortality were available for the approximate period between years 1990 and 2010. These included Finland, Norway, Sweden, Scotland, England and Wales (data applied to both together), France, Switzerland, Spain (Barcelona), Italy (Turin), Slovenia, and Lithuania. Results Substantial mortality declines occurred in lower socioeconomic groups in most European countries covered by this study. Relative inequalities in mortality widened almost universally, because percentage declines were usually smaller in lower socioeconomic groups. However, as absolute declines were often smaller in higher socioeconomic groups, absolute inequalities narrowed by up to 35%, particularly among men. Narrowing was partly driven by ischaemic heart disease, smoking related causes, and causes amenable to medical intervention. Progress in reducing absolute inequalities was greatest in Spain (Barcelona), Scotland, England and Wales, and Italy (Turin), and absent in Finland and Norway. More detailed studies preferably using individual level data are necessary to identify the causes of these variations. Conclusions Over the past two decades, trends in inequalities in mortality have been more favourable in most European countries than is commonly assumed. Absolute inequalities have decreased in several countries, probably more as a side effect of population wide behavioural changes and improvements in prevention and treatment, than as an effect of policies explicitly aimed at reducing health inequalities.

DataSHIELD: taking the analysis to the data, not the data to the analysis

Background: Research in modern biomedicine and social science requires sample sizes so large that they can often only be achieved through a pooled co-analysis of data from several studies. But the pooling of information from individuals in a central database that may be queried by researchers raises important ethico-legal questions and can be controversial. In the UK this has been highlighted by recent debate and controversy relating to the UK’s proposed ‘care.data’ initiative, and these issues reflect important societal and professional concerns about privacy, confidentiality and intellectual property. DataSHIELD provides a novel technological solution that can circumvent some of the most basic challenges in facilitating the access of researchers and other healthcare professionals to individual-level data. Methods: Commands are sent from a central analysis computer (AC) to several data computers (DCs) storing the data to be co-analysed. The data sets are analysed simultaneously but in parallel. The separate parallelized analyses are linked by non-disclosive summary statistics and commands transmitted back and forth between the DCs and the AC. This paper describes the technical implementation of DataSHIELD using a modified R statistical environment linked to an Opal database deployed behind the computer firewall of each DC. Analysis is controlled through a standard R environment at the AC. Results: Based on this Opal/R implementation, DataSHIELD is currently used by the Healthy Obese Project and the Environmental Core Project (BioSHaRE-EU) for the federated analysis of 10 data sets across eight European countries, and this illustrates the opportunities and challenges presented by the DataSHIELD approach. Conclusions: DataSHIELD facilitates important research in settings where: (i) a co-analysis of individual-level data from several studies is scientifically necessary but governance restrictions prohibit the release or sharing of some of the required data, and/or render data access unacceptably slow; (ii) a research group (e.g. in a developing nation) is particularly vulnerable to loss of intellectual property—the researchers want to fully share the information held in their data with national and international collaborators, but do not wish to hand over the physical data themselves; and (iii) a data set is to be included in an individual-level co-analysis but the physical size of the data precludes direct transfer to a new site for analysis.

Methodological Developments in Data Linkage

The increasing availability of large administrative databases has led to a dramatic rise in the use of data linkage, yet the standard texts on linkage are still those which describe the seminal work from the 1950-60s, with some updates. Linkage and analysis of data across sources remains problematic due to lack of discriminatory and accurate identifiers, missing data and regulatory issues. Recent developments in data linkage methodology have concentrated on bias and analysis of linked data, novel approaches to organising relationships between databases and privacy-preserving linkage.

Dog ownership and physical activity in later life: A cross-sectional observational study

OBJECTIVE To examine whether dog ownership amongst community dwelling older adults (≥ 65 years) is associated with objectively measured physical activity (PA). METHODS We used data from the Physical Activity Cohort Scotland (PACS) which consists of 547 people aged 65 and over, resident in the community in Tayside, Scotland. The data was collected in 2009-2011. We assessed whether dog ownership is associated with objectively measured physical activity (accelerometry counts). RESULTS The physical activity (PA) counts of 547 older people (mean age 79 (standard deviation (SD) 8 years, 54% female) were analysed. Linear mixed models showed that dog ownership was positively related to higher PA levels. This positive relationship remained after controlling for a large number of individual and contextual variables, including attitude towards exercise, physical activity intention and history of physical activity. Dog owners were found to be 12% more active (21,875 counts, 95% Confidence Interval (CI): 2810 to 40,939, p<0.05) than non-dog owners. CONCLUSION Dog ownership is associated with physical activity in later life. Interventions to increase activity amongst older people might usefully attempt to replicate elements of the dog ownership experience.

Place of work and residential exposure to ambient air pollution and birth outcomes in Scotland, using geographically fine pollution climate mapping estimates.

Objectives A relationship between ambient air pollution and adverse birth outcomes has been found in a large number of studies that have mainly used a nearest monitor methodology. Recent research has suggested that the effect size may have been underestimated in these studies. This paper examines associations between birth outcomes and ambient levels of residential and workplace sulphur dioxide, particulates and Nitrogen Dioxide estimated using an alternative method – pollution climate mapping. Methods Risk of low birthweight and mean birthweight (for n=21,843 term births) and risk of preterm birth (for n=23,086 births) were modelled against small area annual mean ambient air pollution concentrations at work and residence location adjusting for potential confounding factors for singleton live births (1994–2008) across Scotland. Results Odds ratios of low birthweight of 1.02 (95% CI, 1.01–1.03) and 1.07 (95% CI, 1.01–1.12) with concentration increases of 1 µg/m3 for NO2 and PM10 respectively. Raised but insignificant risks of very preterm birth were found with PM10 (relative risk ratio=1.08; 95% CI, 1.00 to 1.17 per 1 µg/m3) and NO2 (relative risk ratio=1.01; 95% CI, 1.00 to 1.03 per 1 µg/m3). An inverse association between mean birthweight and mean annual NO2(−1.24 g; 95% CI, −2.02 to −0.46 per 1 µg/m3) and PM10 (−5.67 g; 95% CI, −9.47 to −1.87 per 1 µg/m3). SO2 showed no significant associations. Conclusions This study highlights the association between air pollution exposure and reduced newborn size at birth. Together with other recent work it also suggests that exposure estimation based on the nearest monitor method may have led to an under-estimation of the effect size of pollutants on birth outcomes.

Evaluating the health inequalities impact of area-based initiatives across the socioeconomic spectrum: a controlled intervention study of the New Deal for Communities, 2002–2008

Background Previous evaluations of area-based initiatives have not compared intervention areas with the full range of areas from top to bottom of the social spectrum to evaluate their health inequalities impact. Setting Deprived areas subject to the New Deal for Communities (NDC) intervention, local deprivation-matched comparator areas, and areas drawn from across the socioeconomic spectrum (representing high, medium and low deprivation) in England between 2002 and 2008. Data Secondary analysis of biannual repeat cross-sectional surveys collected for the NDC National Evaluation Team and the Health Survey for England (HSE). Methods Following data harmonisation, baseline and time trends in six health and social determinants of health outcomes were compared. Individual-level data were modelled using regression to adjust for age, sex, ethnic and socioeconomic differences among respondents. Results Compared with respondents in HSE low deprivation areas, those in NDC intervention areas experienced a significantly steeper improvement in education, a trend towards a steeper improvement in self-rated health, and a significantly less steep reduction in smoking between 2002 and 2008. In HSE high deprivation areas, significantly less steep improvements in five out of six outcomes were seen compared with HSE low deprivation areas. Conclusions Although unable to consider prior trends and previous initiatives, our findings provide cautious optimism that well-resourced and constructed area-based initiatives can reduce, or at least prevent the widening of, social inequalities for selected outcomes between the most and least deprived groups of areas.

Estimating the incidence, prevalence and true cost of asthma in the UK: secondary analysis of national stand-alone and linked databases in England, Northern Ireland, Scotland and Wales—a study protocol

Introduction Asthma is now one of the most common long-term conditions in the UK. It is therefore important to develop a comprehensive appreciation of the healthcare and societal costs in order to inform decisions on care provision and planning. We plan to build on our earlier estimates of national prevalence and costs from asthma by filling the data gaps previously identified in relation to healthcare and broadening the field of enquiry to include societal costs. This work will provide the first UK-wide estimates of the costs of asthma. In the context of asthma for the UK and its member countries (ie, England, Northern Ireland, Scotland and Wales), we seek to: (1) produce a detailed overview of estimates of incidence, prevalence and healthcare utilisation; (2) estimate health and societal costs; (3) identify any remaining information gaps and explore the feasibility of filling these and (4) provide insights into future research that has the potential to inform changes in policy leading to the provision of more cost-effective care. Methods and analysis Secondary analyses of data from national health surveys, primary care, prescribing, emergency care, hospital, mortality and administrative data sources will be undertaken to estimate prevalence, healthcare utilisation and outcomes from asthma. Data linkages and economic modelling will be undertaken in an attempt to populate data gaps and estimate costs. Separate prevalence and cost estimates will be calculated for each of the UK-member countries and these will then be aggregated to generate UK-wide estimates. Ethics and dissemination Approvals have been obtained from the NHS Scotland Information Services Divisions Privacy Advisory Committee, the Secure Anonymised Information Linkage Collaboration Review System, the NHS South-East Scotland Research Ethics Service and The University of Edinburghs Centre for Population Health Sciences Research Ethics Committee. We will produce a report for Asthma-UK, submit papers to peer-reviewed journals and construct an interactive map.

Differences in 5-year survival after a ‘homeless’ or ‘housed’ drugs-related hospital admission: a study of 15–30-year olds in Scotland

Background Young drug misusers and the homeless both have a greater risk of death than their peers. This study sought to estimate the additional impact of homelessness on the risk of death for young drugs misusers. Methods From all admissions to NHS hospitals in Scotland between 1986 and 2001, those that were: drug misuse related, for people born between 1970 and 1986 and aged over 15 years (n=13 303), were selected. All subsequent admissions and registrations of death were linked to this dataset. Each admission was coded as homeless if the health board of residence was coded as ‘no fixed abode’. 5-year survival after an admission was modelled using (1) life table and (2) proportional hazard models and then (3) differences in causes of deaths were explored. Results Immediately after a drugs-related hospital admission there was no difference in survival between the homeless and those with a ‘fixed address’. However, over a 3-year period the risk for those who were homeless was 3.5 times greater (CI 95% 1.2 to 12.8). This elevated risk seemed to be particularly focused on the second year after an admission. The causes of death were similar for the two groups. Conclusion Although a homeless hospital admission is associated with a greater risk of death for young drug users, it is also a point in time when a young person is in contact with public services. An attempt to link their discharge with housing services would seem a potentially productive policy.