Clinical epidemiology in the era of big data: new opportunities, familiar challenges

Socioeconomic disparities in first stroke incidence, quality of care, and survival: a nationwide registry-based cohort study of 44 million adults in England

BACKGROUND

We aimed to estimate socioeconomic disparities in the incidence of hospitalisation for first-ever stroke, quality of care, and post-stroke survival for the adult population of England.

METHODS

In this cohort study, we obtained data collected by a nationwide register on patients aged 18 years or older hospitalised for first-ever acute ischaemic stroke or primary intracerebral haemorrhage in England from July 1, 2013, to March 31, 2016. We classified socioeconomic status at the level of Lower Super Output Areas using the Index of Multiple Deprivation, a neighbourhood measure of deprivation. Multivariable models were fitted to estimate the incidence of hospitalisation for first stroke (negative binomial), quality of care using 12 quality metrics (multilevel logistic), and all-cause 1 year case fatality (Cox proportional hazards).

FINDINGS

Of the 43·8 million adults in England, 145 324 were admitted to hospital with their first-ever stroke: 126 640 (87%) with ischaemic stroke, 17 233 (12%) with intracerebral haemorrhage, and 1451 (1%) with undetermined stroke type. We observed a socioeconomic gradient in the incidence of hospitalisation for ischaemic stroke (adjusted incidence rate ratio 2·0, 95% CI 1·7-2·3 for the most vs least deprived deciles) and intracerebral haemorrhage (1·6, 1·3-1·9). Patients from the lowest socioeconomic groups had first stroke a median of 7 years earlier than those from the highest (p<0·0001), and had a higher prevalence of pre-stroke disability and diabetes. Patients from lower socioeconomic groups were less likely to receive five of 12 care processes but were more likely to receive early supported discharge (adjusted odds ratio 1·14, 95% CI 1·07-1·22). Low socioeconomic status was associated with a 26% higher adjusted risk of 1-year mortality (adjusted hazard ratio 1·26, 95% CI 1·20-1·33, for highest vs lowest deprivation decile), but this gradient was largely attenuated after adjustment for the presence of pre-stroke diabetes, hypertension, and atrial fibrillation (1·11, 1·05-1·17).

INTERPRETATION

Wide socioeconomic disparities exist in the burden of ischaemic stroke and intracerebral haemorrhage in England, most notably in stroke hospitalisation risk and case fatality and, to a lesser extent, in the quality of health care. Reducing these disparities requires interventions to improve the quality of acute stroke care and address disparities in cardiovascular risk factors present before stroke.

FUNDING

NHS England and the Welsh Government.

Large-Scale Genomic Biobanks and Cardiovascular Disease

PURPOSE OF REVIEW

Cardiovascular disease is a leading cause of morbidity and mortality worldwide and is the focus of extensive biomedical research. Large genetic consortia combining data from many traditional prospective cohort and ascertained case-control study designs have facilitated the discovery of genetic associations for a variety of cardiovascular diseases including diabetes, coronary artery disease, and hypertension. Biobank-based genetic studies offer an alternative whereby large populations are genotyped and linked to electronic health records.

RECENT FINDINGS

Biobank sample sizes worldwide have surpassed even the largest genetic consortia and have yielded key insights into the genetic determinants of both common and rare cardiovascular phenotypes. Herein, we provide an overview of the largest genomic biobanks and discuss the relevant advantages and challenges inherent to the biobank model of cohort generation and genomic study design.

Evidence-based data and rare cancers: The need for a new methodological approach in research and investigation

Rare cancers are not so rare, their incidence is increasing and, as a group, they have worse survival than the common cancers. These factors emphasise the societal need to ensure sufficient focus on research into their biological basis, aetiological factors, new more effective therapies and organisation of healthcare to improve access to best practice and innovation. Accuracy of diagnosis is one of the first hurdles to be overcome, with around one third of tumours being reclassified - by type or risk group - when subject to a centralised pathology review process. Timely access to appropriate expert knowledge is a second challenge for patients - in Europe this is being addressed by the establishment of European Reference Networks (ERNs) as part of the EU cross border healthcare initiative. There are ERNs for adult solid and haematological cancers and childhood cancers, all of which are individually rare. These ERNs will facilitate creation of large databases of rare tumours that will incorporate knowledge of their molecular features and build an evidence base for the effectiveness of innovative, biology-directed therapies. With an increasing focus on 'real world' outcome data, research methodologies are evolving, to include randomised registry trials and data linkage approaches that exploit the ever-richer information held on patients in routine health care data. The inclusion of genomic analysis into cancer diagnosis, treatment and risk prediction raises many issues for the conduct of clinical research and cohort studies and personal data sharing. Sophisticated means of pseudonymisation, together with full involvement of affected and 'at risk' patients, are supporting novel research designs and access to data that will continue to build the evidence base to improve outcomes for patients with rare cancers.

Population Neuroscience: Dementia Epidemiology Serving Precision Medicine and Population Health

Over recent decades, epidemiology has made significant contributions to our understanding of dementia, translating scientific discoveries into population health. Here, we propose reframing dementia epidemiology as "population neuroscience," blending techniques and models from contemporary neuroscience with those of epidemiology and biostatistics. On the basis of emerging evidence and newer paradigms and methods, population neuroscience will minimize the bias typical of traditional clinical research, identify the relatively homogenous subgroups that comprise the general population, and investigate broader and denser phenotypes of dementia and cognitive impairment. Long-term follow-up of sufficiently large study cohorts will allow the identification of cohort effects and critical windows of exposure. Molecular epidemiology and omics will allow us to unravel the key distinctions within and among subgroups and better understand individuals' risk profiles. Interventional epidemiology will allow us to identify the different subgroups that respond to different treatment/prevention strategies. These strategies will inform precision medicine. In addition, insights into interactions between disease biology, personal and environmental factors, and social determinants of health will allow us to measure and track disease in communities and improve population health. By placing neuroscience within a real-world context, population neuroscience can fulfill its potential to serve both precision medicine and population health.

An e-health interactive self-assessment website (Sanoia®) in rheumatoid arthritis. A 12-month randomized controlled trial in 320 patients

INTRODUCTION

Sanoia is an online interactive electronic e-health platform developed to allow patient self-assessment and self-monitoring. The objective was to assess in rheumatoid arthritis (RA) patients, the efficacy on patient-physician interactions, of giving access to Sanoia.

METHODS

In this French, multi-center, 12-months randomized controlled trial (CarNET: NCT02200068), patients with RA and internet access were randomized to: access without incentives to the Sanoia platform after minimal training, or usual care. The primary outcome was the change from baseline in patient-physician interactions, by the patient-reported Perceived Efficacy in Patient-Physician Interactions (PEPPI-5) questionnaire. The number of accesses to Sanoia was recorded and satisfaction with the platform was assessed through a 0-10 numeric rating scale. Analyses were in intention to treat (ITT), on SAS.

RESULTS

Of 320 RA patients (159 Sanoia versus 161 usual care), mean (standard deviation) age was 57.0 (12.7) years, mean (SD) disease duration was 14.6 (11.1) years, 216 (67.5%) were taking a biologic and 253 (79.1%) were female. Mean (SD) PEPPI scores at baseline and 12 months were 38.6 (8.2) and 39.2 (8.0) (delta=+0.60 [5.52]) versus 39.7 (7.3) and 38.8 (8.0) (delta=-0.91 [6.08]) in the Sanoia and control group, respectively (P=0.01). Although mean satisfaction with the platform was very high (1.46 [1.52]), 41 patients (25.7%) never accessed Sanoia.

CONCLUSION

Giving RA patients access to the interactive Sanoia e-health platform led to a small improvement in patient-perceived patient-physician interactions. A disjunction between patient satisfaction and access to the platform was noted. E-Health platforms are promising in RA.

Series: Pragmatic trials and real world evidence: Paper 8. Data collection and management

Pragmatic trials can improve our understanding of how treatments will perform in routine practice. In a series of eight papers, the GetReal Consortium has evaluated the challenges in designing and conducting pragmatic trials and their specific methodological, operational, regulatory, and ethical implications. The present final paper of the series discusses the operational and methodological challenges of data collection in pragmatic trials. A more pragmatic data collection needs to balance the delivery of highly accurate and complete data with minimizing the level of interference that data entry and verification induce with clinical practice. Furthermore, it should allow for the involvement of a representative sample of practices, physicians, and patients who prescribe/receive treatment in routine care. This paper discusses challenges that are related to the different methods of data collection and presents potential solutions where possible. No one-size-fits-all recommendation can be given for the collection of data in pragmatic trials, although in general the application of existing routinely used data-collection systems and processes seems to best suit the pragmatic approach. However, data access and privacy, the time points of data collection, the level of detail in the data, and the lack of a clear understanding of the data-collection process were identified as main challenges for the usage of routinely collected data in pragmatic trials. A first step should be to determine to what extent existing health care databases provide the necessary study data and can accommodate data collection and management. When more elaborate or detailed data collection or more structured follow-up is required, data collection in a pragmatic trial will have to be tailor-made, often using a hybrid approach using a dedicated electronic case report form (eCRF). In this case, the eCRF should be kept as simple as possible to reduce the burden for practitioners and minimize influence on routine clinical practice.