Can early markers of poor long-term outcomes in dementia be identified from primary care electronic health records?

Talk Code: 
Kelvin Jordan
Trishna Rathod-Mistry, Michelle Marshall, Paul Campbell, James Bailey, Carolyn A. Chew-Graham, Martin Frisher, Richard Hayward, Rashi Negi, Swaran Singh, Athula Sumathipala, Nwe Thein, Kate Walters, Scott Weich
Author institutions: 
Keele University, Midlands Partnership NHS Foundation Trust, University of Warwick, University College London, University of Sheffield


The prevalence of dementia is rising. The UK government has prioritised early recognition and treatment to prolong independence, reduce hospital admissions, and delay nursing home admission and mortality. Advances have been made in “case finding” and identifying risk factors for dementia using primary care Electronic Health Records (EHR). “Hard” long term outcomes such as hospital admissions and mortality are obtainable from EHR but less is known about the early course of the disease. This information could help clinicians spot patients at risk of quicker progression and enable shorter term research outcomes to be assessed. The objective was to assess whether early markers of dementia progression can be identified within information routinely recorded in primary care EHR.


Through reviewing research studies, consensus meetings with experts in dementia and EHR research, and analysis of a North Staffordshire EHR database, we have established a list of potential markers of dementia-related health that are feasible to assess in primary care EHR. Markers were grouped into domains. We analysed records of patients newly diagnosed with dementia between 1998-2017 from a UK-wide primary care EHR database (Clinical Practice Research Datalink). The incidence of recording of markers from domains up to three years after diagnosis were determined. We tested the validity of the domains as measures of dementia progression with the hypothesis that early recording of the domains should be associated with increased risk of long-term hospital admission, palliative care, and mortality.


Sixty-three markers were mapped within thirteen domains. The domains included severe neuropsychiatric (example marker: psychosis), other neuropsychiatric (e.g. depression), cognition (e.g. memory loss), daily functioning (e.g. walking limitation), selected other illnesses (e.g. heart problem), symptoms (e.g. hearing loss), increased polypharmacy. There were 30,463 people with newly diagnosed dementia. Common first recorded markers after diagnosis included those in the comorbidity (incidence 300/1000 person-years), symptoms (230/1000), and other neuropsychiatric (186/1000) domains. Increased polypharmacy was also common (469/1000). Risk of long-term outcomes increased with increasing number of new markers recorded in the first year after diagnosis. For example, patients with four or more domains in which a new marker was recorded (18% of patients) had greater risk of long term hospital admission (adjusted hazard ratio 1.24 versus no domains; 95% CI 1.16, 1.32), palliative care (1.86; 1.61, 2.16), and mortality (1.57; 95% CI 1.47, 1.67). The individual domains were also associated with long term outcomes, with the neuropsychiatric domains and increased polypharmacy, in particular, having stronger associations.


Primary care EHR capture measures of dementia-related health associated with long-term outcomes. These may facilitate better understanding of progression, make it easier for GPs to recognise people on a more severe course of dementia, and provide early outcome measures for research studies including clinical trials.

Submitted by: 
Kelvin Jordan
Funding acknowledgement: 
This work is supported by a grant from the Dunhill Medical Trust (RPGF 1711/11). The views and opinions expressed within this abstract are those of the authors and not necessarily the views of the Dunhill Medical Trust. This study is based in part on data from the Clinical Practice Research Datalink obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. The data is provided by patients and collected by the NHS as part of their care and support. The Office for National Statistics (ONS) is the provider of the ONS data contained within the linked CPRD data used for this study. ONS Data and Hospital Episode Statistics (HES) Data: copyright © (2019), re-used with the permission of The Health and Social Care Information Centre; all rights reserved. The interpretation and conclusions contained in this study are those of the authors alone.