The research team, led by Markus J. Haapanen of the University of Helsinki and the University of Queensland’s Australian Frailty Network, drew on the UK Biobank, one of the world’s largest prospective health databases. They examined inpatient records from England, Scotland and Wales spanning 1997 to 2022, with follow-up extending up to 26 years. From this vast resource they identified 14,909 adults whose records contained a first hospital episode of delirium coded under the International Classification of Diseases, tenth revision. Each of these individuals was then matched one-to-one with a hospitalised control patient who had no recorded delirium. The matching was deliberately stringent: controls were identical in age, sex, primary diagnosis, length of stay, intensive care unit admission and, crucially, the Hospital Frailty Risk Score, a measure of baseline vulnerability. This design ensured that the comparison isolated the effect of delirium itself rather than the underlying severity of illness or frailty that often accompanies it.

The scale of the matched cohort, 29,818 participants with a mean age of 72.1 years and 57 percent men, allowed the investigators to detect even moderate differences in long-term psychiatric outcomes. New-onset psychotic, bipolar, depressive and anxiety disorders were identified from subsequent inpatient diagnostic codes, and personality disorders served as a negative control, a condition that theory predicted should not be linked to delirium. The statistical approach was equally careful. Because older hospitalised patients face a high competing risk of dying before a psychiatric diagnosis could ever appear, the team used Fine–Gray subdistribution hazard models, which account for death as an alternative endpoint. Models were further adjusted for socioeconomic factors and for polygenic risk scores for schizophrenia and bipolar disorder, helping to separate the effect of delirium from inherited susceptibility to severe mental illness.

The headline result is striking. Among hospitalised patients without delirium, the incidence of any new psychiatric disorder was 4.9 per 100 person-years; among those who had experienced delirium, it rose to 6.7 per 100 person-years. In adjusted analyses, delirium was associated with a 44 percent higher incidence of any psychiatric disorder, corresponding to a subdistribution hazard ratio of 1.44 with a 95 percent confidence interval of 1.31 to 1.59. Given that the cohort was matched on so many clinical variables and adjusted for genetic risk, the association points toward something specific about the delirium episode itself, or the brain states it reflects, as a driver of later psychiatric morbidity.

The gradient across diagnostic categories tells a more intriguing story. The association was by far the strongest for psychotic disorders, with a hazard ratio of 3.79, meaning affected patients faced nearly four times the incidence of conditions such as delusional thinking and hallucinations that persist beyond the acute episode. Bipolar disorders followed closely at 3.44. Depressive disorders showed a more moderate elevation at 1.52, and anxiety disorders at 1.37. Tellingly, no association was observed for personality disorders, the prespecified negative control. That pattern is exactly what one would expect if delirium genuinely precipitates illness in vulnerable brains, rather than simply reflecting a general tendency toward hospitalisation or poor health. If the observed associations were artifacts of detection bias or reverse causation, one might expect them to appear across the diagnostic board, including for conditions with no plausible mechanistic link.

Perhaps the most provocative finding concerns repeated episodes. Only psychotic disorders showed a dose–response relationship with recurrent delirium: each additional delirium episode within a 12-month window conferred a 23 percent higher risk of subsequent psychosis, a hazard ratio of 1.23 with a confidence interval of 1.09 to 1.39. Dose–response relationships are a classic criterion for inferring causation in observational research. The more often the brain is plunged into acute confusion, the greater the likelihood that a psychotic disorder eventually emerges. This graded pattern fits with the idea that each delirium episode imposes physiological stress, perhaps involving neuroinflammation, neurotransmitter disruption or blood–brain barrier dysfunction, that can accumulate and unmask latent vulnerability in individuals already carrying genetic or environmental risk.

The mechanistic plausibility of such a link is supported by a growing literature on delirium biology. Delirium is associated with surges in inflammatory cytokines, disturbances in acetylcholine and dopamine signalling, and structural changes visible on neuroimaging in some patients. Many of the same pathways have been implicated in schizophrenia and bipolar disorder, conditions whose onset in later life has long puzzled psychiatrists. One interpretation of the new findings is that delirium acts as a kind of physiological accelerant, converting a predisposition into a manifest disorder years or decades before it might otherwise have declared itself. Another is that both delirium and later psychosis are downstream consequences of the same underlying brain fragility. The study cannot fully adjudicate between these possibilities, but the careful matching and genetic adjustment make a simple confounding explanation less likely.

For clinicians and hospital systems, the practical implications are considerable. Delirium is already considered potentially preventable in a meaningful fraction of cases, and multicomponent interventions such as early mobilisation, sleep protection, hydration, sensory aids and careful medication review have been shown in some settings to reduce its occurrence. If delirium is indeed a modifiable stepping stone toward severe mental illness, then preventing it takes on an urgency that extends far beyond the hospital stay itself. The authors suggest that delirium might warrant a place among the recognised risk factors for subsequent psychiatric disorders, alongside better-known candidates such as traumatic brain injury. Patients and families should also be alert to post-discharge changes in mood, perception or behaviour, since early psychiatric review after a delirium episode may allow timely intervention.

There are, of course, limits to what hospital records can reveal. Delirium is notoriously under-coded, and milder episodes in community settings escape detection entirely, which could bias estimates in unpredictable directions. Inpatient diagnostic codes may also miss psychiatric disorders treated exclusively in primary care. The study population, drawn from UK Biobank participants, skews somewhat healthier than the general population, though this would more plausibly attenuate than inflate the observed associations. Residual confounding by unmeasured illness severity, despite the meticulous matching, cannot be excluded. Nonetheless, the size of the cohort, the length of follow-up, the stringency of the control group and the internal consistency of the dose–response signal combine to make this one of the most compelling observational datasets yet assembled on the psychiatric aftermath of delirium.

What emerges is a reframing of a familiar hospital hazard. Delirium has traditionally been framed as an acute, reversible complication, distressing in the moment but forgotten once orientation returns. This study suggests the forgetting may be superficial, and that the episode can echo through the remaining years of life as elevated risk of depression, anxiety, psychosis or mania. For a condition affecting up to half of frail older people during admission, even a 44 percent relative increase carries enormous public health weight. The next step, the researchers argue, is to test whether aggressive delirium prevention and post-delirium monitoring genuinely reduce the long-term burden of mental illness. If they do, one of geriatric medicine’s most common complications could become one of psychiatry’s most unexpected targets for prevention.

Subject of Research: Association between in-hospital delirium and subsequent new-onset psychiatric disorders in older adults

Article Title: Delirium and subsequent psychiatric disorders in hospitalised older adults

Article References: Delirium and subsequent psychiatric disorders in hospitalised older adults. (n.d.). https://doi.org/10.1186/s12916-026-05314-9

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05314-9

Keywords: delirium, psychiatric disorders, psychosis, bipolar disorder, depression, anxiety, older adults, hospitalisation, UK Biobank, geriatric psychiatry, epidemiology, dementia risk