文章基本信息

标题：Exploring a novel linked dataset and building linked data analytics skills in Public Health Intelligence teams in Sussex.
本地全文：下载
作者：Natalie Johnston ; Elizabeth Ford ; Richard Tyler 等
期刊名称：International Journal of Population Data Science
电子版ISSN：2399-4908
出版年度：2022
卷号：7
期号：3
页码：1-1
DOI：10.23889/ijpds.v7i3.1937
语种：English
出版社：Swansea University
摘要：Objectives Public health intelligence teams in Sussex wanted to use newly linked health and social care data, to gain insights into local patterns of multi-morbidity, service use, service provision and socio-demographic data. In this study we report initial exploration of this new linked dataset, in a partnership between university and local authority analysts. Approach The Sussex Integrated Dataset (SID) comprises person-level health and social care data on residents and services users across Sussex. During a 6-month secondment, two analysts evaluated the number of data sources available for each individual, evaluated data quality for identifying long-term conditions, developed presentation methods to compare SID outputs on demographics and condition prevalence with open source or expected distributions, and identified the skills-mix and infrastructure required in local authorities for future work. They worked alongside the SID data processing team to inform and improve data quality; and with university data-scientists to learn prediction modelling. Results Analysts established an efficient querying system to investigate the breadth of data available, more thoroughly focusing on encounters and demographic data in all sources. Long-term conditions were identified through code lists in a range of NHS data sources, to enable consideration of multi-morbidity by demographic. A range of quality issues were identified, such as non-current patients being uploaded into the SID, distorting prevalence estimates, and GP practice populations that did not match expected figures published by NHS digital. Results were represented in multi-morbidity plots, maps, and theographs. Through this data exploration, we have been able to identify the skills-mix needed for local Public Health Intelligence teams to maximise the use of linked data to achieve Public Health objectives. Conclusion We have made many conceptual breakthroughs, particularly in understanding data quality, however still need a more complete understanding of quality issues in SID for public health outputs to have meaningful use. Further investigation into the patterns of service use, as well as modelling of multi-morbidity to make predictions and target interventions, will be key next steps.