Understanding why dementia is underdiagnosed: Recording practices of cognitive impairment in the elderly

Talk Code: 
Shanu Sadhwani
Elizabeth Ford, Flis Henwood, Helen Smith
Author institutions: 
Brighton & Sussex Medical School, University of Brighton, Lee Kong Chian School of Medicine


Early diagnosis of dementia has proven benefits to patients and carers, yet estimates suggest almost 50% of people living with dementia do not have a formal diagnosis in their electronic health record (EHR). Policymakers have tried to encourage early diagnosis through national strategies and financial incentives, believing that a lack of diagnosis is an indication of substandard care. The study aimed to investigate reasons why GPs might not record a diagnosis of dementia in a typical case of cognitive decline in an elderly patient.


A cross-sectional within subjects, experimental design using an online questionnaire presented fictional clinical vignettes to GPs in the UK. Three vignettes, with accompanying image of the patient, represented sequential consultations of an 83-year-old white woman with symptoms indicative of dementia, over a 16-month period. The aim of the vignettes was to illustrate a presentation of the dementia continuum, from earliest signs such as mild memory impairment, to latter overt and complex symptoms such as paranoia and behavioural changes. The vignettes were designed with the following features: a typical presentation with memory symptoms, a white woman, approximately 80 years old, poor insight, without a life-threatening illness, and with involvement from a daughter in latter consultations.Participants were asked to indicate their clinical response to this imaginary consultation, including decisions made, motivations, and how they would record each consultation in Read codes and free-text.


60 GPs participated in the online survey. A sensitivity analysis indicated that all participants identified cognitive impairment in the first vignette, with 80% pursuing a further investigation to identify its underlying cause. Whilst 68% of participants suspected dementia by vignette 3, only 34% recorded suspicions in the EHR as a diagnostic Read code, such as ‘Worsening dementia’, ‘Suspected dementia’, or ‘Dementia’. 16% of participants recorded suspicions of dementia in free text, without using dementia Read codes, indicating that the underdiagnosis of dementia is in part explained by researchers’ and auditors’ inability to examine the full EHR. Over 30% of participants indicated a reluctance to diagnose dementia without confirmation from specialists, suggesting that patients who are reluctant or unable to attend such services are less likely to have a diagnosis of dementia in their record, even if GPs are highly suspect.


This study indicates that GPs prioritised patient safety and carer wellbeing over diagnoses. Cognitive impairment was formally investigated when memory symptoms impacted on the patients’ daily functioning.Therefore, a lack of diagnostic dementia Read code in an EHR does not necessarily indicate that dementia is unidentified or that care is substandard. Whilst a patient living with dementia may not have a formal diagnosis in their EHR, this study indicates that GPs are actively ensuring that patients’ safety and wellbeing needs are prioritised.

Submitted by: 
Shanu Sadhwani
Funding acknowledgement: