Hospital delirium signals years of added risk for falls, sepsis, stroke and more
In 14,909 matched pairs, in-hospital delirium predicted higher risk of 12 of 15 adverse outcomes later, even after accounting for frailty and dementia.
Summary
Delirium, a sudden state of confusion during a hospital stay, is common in older adults. Little is known about what follows beyond cognitive decline, functional loss and death. Using UK Biobank data linked to hospital records, researchers matched 14,909 people who had delirium to hospitalised people without it, based on age, sex, frailty score, diagnosis and length of stay. Over up to 26 years of follow-up, delirium was linked to higher risk of 12 of 15 adverse outcomes in later hospital stays. These included urinary incontinence (about double the risk), falls, pressure injuries, kidney injury, sepsis, hip fracture and stroke. Each additional delirium episode raised risk by 6-17%. Results held after excluding people with dementia. The authors suggest delirium may be a warning sign of broader, long-term vulnerability, though an observational design cannot prove cause.
Detailed Summary
Delirium affects up to one in four older adults admitted to hospital. Its known consequences are cognitive impairment, functional decline and death. Whether it also marks wider, multisystem vulnerability that leads to other complications later has been unclear.
This matched cohort study used UK Biobank participants with linked hospital inpatient records from 1997 to 2022. Researchers identified 14,909 people with delirium, defined by ICD-10 codes, and matched each 1:1 to a hospitalised control without delirium. Matching used age, sex, Hospital Frailty Risk Score, primary diagnosis, length of stay and ICU stay. They then tracked 15 adverse events recorded during later hospitalisations. Fine-Gray models treated death as a competing risk and adjusted for socioeconomic factors.
Delirium was associated with higher risk of 12 of the 15 outcomes. The largest associations were for urinary incontinence (subdistribution HR 2.01), falls (1.96), pressure injury (1.72), acute kidney injury (1.71), sepsis (1.67), hip fracture (1.66) and stroke (1.62). Pneumonia, faecal incontinence, overall fractures, heart failure and gastrointestinal bleeding were also higher. Risk rose with each additional delirium episode in the 12 months after the index admission, by 6-17% per episode. Results were consistent when excluding people with short follow-up, prevalent dementia and the least well-matched pairs.
The authors propose that delirium be treated as a sentinel event signalling longer-term vulnerability, independent of frailty and dementia. This could support closer follow-up, falls and pressure-injury prevention, and review after discharge.
These are observational associations and cannot show that delirium causes the later events. Delirium identified through ICD-10 codes is likely under-recorded. Outcomes were captured only during later hospitalisations, so detection bias is possible. UK Biobank volunteers are also generally healthier than the wider population. This summary is based on the abstract only.
Key Findings
- Delirium was associated with higher risk of 12 of 15 adverse outcomes in later hospitalisations compared with matched controls.
- Urinary incontinence (SHR 2.01) and falls (SHR 1.96) showed the strongest associations, followed by pressure injury, kidney injury and sepsis.
- Hip fracture (SHR 1.66) and stroke (SHR 1.62) risks were substantially higher after delirium.
- Each additional delirium episode within 12 months was associated with a 6-17% higher risk, suggesting a dose-response pattern.
- Findings persisted after excluding people with prevalent dementia and short follow-up, and after matching on frailty.
Methodology
Matched cohort study of UK Biobank participants with linked hospital inpatient records (1997-2022). 14,909 patients with ICD-10-coded delirium were matched 1:1 to hospitalised controls on age, sex, Hospital Frailty Risk Score, primary diagnosis, length of stay and ICU stay. Fine-Gray subdistribution hazard models accounted for death as a competing risk and adjusted for socioeconomic covariates.
Study Limitations
This is an observational study, so causality cannot be inferred, and residual confounding is possible despite detailed matching. Delirium coded by ICD-10 is often under-recorded, and outcomes were captured only during later hospitalisations, which may introduce detection bias. UK Biobank participants tend to be healthier and less diverse than the general population, which may limit generalisability. This summary is based on the abstract only.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: