Hospital delirium: when someone you love is suddenly not themselves

Delirium is a sudden change in attention and thinking, arriving over hours to days, that waxes and wanes. The person cannot hold focus. They may be disoriented, see things that are not there, sleep all day and rove all night, or say things nothing like themselves. The hallmark is the swing: lucid at lunch, lost by dinner, better again in the morning. Families call it good days and bad days, sometimes good hours and bad hours. When the confusion reliably worsens in the late afternoon and evening, nurses call it sundowning, and sundowning that is new or sharply worse in a hospitalized person is delirium until proven otherwise.

The staff seem calm because they see it every week. You are alarmed because you are seeing it for the first time, in someone whose baseline you know better than anyone in the building. Both reactions are correct. Delirium is common and usually temporary, which is exactly how the UK's national clinical guideline tells hospitals to describe it to families (NICE CG103). It is also serious, linked to longer stays, falls, and worse long-term outcomes, which is why it deserves more attention than it usually gets.

Both things are true at once, and this guide will not pretend otherwise. Most people get better. The process is slower and messier than anyone warns you, and a minority carry lasting effects. The next sections are about tilting the odds.

This is the fear underneath everything, so here is the distinction plainly. Delirium arrives suddenly, over hours to days, and it fluctuates, sometimes swinging within a single afternoon. Dementia arrives gradually, over months to years, and holds fairly steady from day to day. A person who was managing her own checkbook two weeks ago and is now profoundly confused does not have new dementia. Dementia does not move that fast. Something acute is going on, and acute things can be treated.

The two are tangled in one important way. Existing dementia, even mild and undiagnosed, is the single biggest risk factor for delirium. A hospital stay sometimes unmasks a decline the family had half-noticed and explained away. So the honest answer to "is this dementia now" is usually: the confusion you are looking at today is delirium, and whether there is something underneath it is a question that can only be answered later, when the delirium has cleared.

That has one hard practical edge. Nobody can diagnose dementia during a delirium episode, and cognitive testing done in the hospital mid-episode is measuring the delirium, not the person. Families in caregiver forums describe fighting to keep a dementia label off the chart that was written while their parent was acutely confused, and they are right to fight. If a clinician wants to conclude dementia from how your person looks this week, it is fair to say: she was not like this a month ago, and we would like the cognitive assessment repeated after she has recovered.

When people picture delirium they picture the agitated version: pulling at lines, shouting, trying to climb out of bed. That form gets noticed, because it is impossible not to notice. But it is the minority. In the Marcantonio review, only about a quarter of delirium is the hyperactive kind. The rest is the quiet, hypoactive form or a mix of the two: drowsy, withdrawn, slow to answer, barely eating, "pleasantly confused," sleeping most of the day. Many people cycle between the two, which clinicians call mixed delirium.

The quiet form is the dangerous one, partly because it hides. In the nurse-recognition study, the odds of hypoactive delirium going unrecognized were about seven times higher than for the agitated kind. In one large ICU treatment trial, 89 percent of the delirium researchers found was hypoactive (Girard and colleagues, New England Journal of Medicine, 2018). A hospital floor full of quiet, sleepy older patients contains more delirium than anyone has charted.

So the sentence worth saying out loud, to yourself and to the nurse, is: sleepy is not the same as fine. If your mother is suddenly unrousable at 2 pm, answers in single words when she is usually chatty, or has stopped tracking conversation, that is the same emergency as shouting, wearing a quieter coat. Report it exactly the way you would report agitation. The withdrawn patients are the ones nobody reports.

There is no blood test or brain scan that shows delirium. A CT of a delirious brain usually looks normal, which surprises families every time. The diagnosis is made by observation, most often with a bedside tool called the Confusion Assessment Method, or CAM, developed by Dr. Sharon Inouye in 1990 and still the standard (intensive care units use a version called the CAM-ICU). It looks for four things: a sudden change from baseline that fluctuates, inattention, meaning the person cannot hold focus, disorganized thinking, and an altered level of alertness, either agitated or abnormally drowsy.

Notice what the first feature requires: someone who knows the baseline. The doctors meeting your father this week have no idea what he was like last month. You do. That is why the UK national guideline explicitly lists changes "reported by the person at risk, or a carer or relative" as grounds for assessment. The same guideline instructs clinicians to "think delirium" for anyone over 65, anyone with cognitive impairment, anyone with a hip fracture, and anyone severely ill. You are not an anxious bystander. You are the measurement instrument the test depends on.

The way to use that power is one precise sentence to the nurse or doctor: "This is an abrupt change from his baseline. Ten days ago he was driving and managing his own medications. I would like him assessed for delirium." The word delirium does real work in that sentence. It converts a vague complaint into a clinical finding with a protocol attached. Vague worry gets reassurance. A named condition gets a workup.

Delirium almost always has triggers, usually several stacked together on a vulnerable brain. The big ones, per the NEJM review and the NICE guideline, are: medications, especially sedatives, opioids, and anticholinergic drugs, a class that includes diphenhydramine, the antihistamine in most over-the-counter sleep aids; infection, including urinary tract infections and pneumonia; surgery and anesthesia; dehydration and electrolyte problems; uncontrolled pain; constipation; low oxygen; and the hospital environment itself, with its noise, lights, and shattered sleep.

Caregiver forums have taught families to ask about UTIs, and it is a fair question, since a UTI in an older adult can show up as confusion with no fever and no complaint. Ask it. Then keep going, because the checklist is longer than one infection: Has anything been checked for infection? What medications were started, stopped, or raised this week? Is he getting anything for sleep, and is it diphenhydramine? When did he last have a bowel movement? Is he drinking, and are his kidneys keeping up? Is pain controlled? Does he have his glasses and hearing aids?

The medication question is where families most often hold the missing clue, because home medication lists and hospital records famously disagree. The sleep aid nobody mentioned, the dose that changed last week, the drug that was stopped abruptly at admission: those are exactly the details that surface a cause. A current medication list with recent changes visible turns "we think he takes something for his heart" into an answer, and it is the single most useful piece of paper you can bring to the bedside.

The most effective delirium treatment ever tested is not a drug. It is a bundle of ordinary-sounding things: reorientation, sleep protection, getting the person up and moving, glasses and hearing aids on and working, and fluids. Delivered systematically as the Hospital Elder Life Program, that bundle cut new delirium episodes by about a third in the original trial, from 15 to 9.9 percent of at-risk patients (Inouye and colleagues, NEJM, 1999), and a 2015 meta-analysis across 14 studies found roughly half the odds of delirium and fewer falls (Hshieh and colleagues, JAMA Internal Medicine).

Here is the part that should change your week: families can deliver it. In a randomized trial of 287 at-risk hospital patients, researchers taught family members to do the bundle themselves, with a clock and calendar in the room, familiar objects from home, extended family presence, and gentle daily reorientation. Delirium fell from 13.3 percent to 5.6 percent, a 59 percent relative reduction (Martinez and colleagues, Age and Ageing, 2012). Your presence at that bedside is not moral support. It is the intervention.

The practical version: be there as much as you can, and spread visits across the day rather than clustering everyone at 2 pm. Bring the glasses and both hearing aids, working, with batteries. Bring photos and a familiar blanket. Open the blinds in the morning and push for lights off and clustered care at night, so sleep can happen. Say who you are each time, mention the day and place casually rather than as a quiz, and keep one calm voice going instead of three.

Two things not to do. Do not argue with the confusion. If he says he was at the office this morning, correcting him tends to inflame things; answer the feeling instead, with something like "you are safe, I am here, we are at the hospital while you get over the infection." And do not take what they say personally. Delirious people accuse, insult, and wound, sometimes with startling precision. It is the illness talking, and delirious brains aim at whoever is closest, which is usually the person who loves them most.

On restraints: if he keeps pulling at lines, the reflex is wrist restraints, and it usually backfires. Being tied down tends to escalate the fear and agitation it was meant to contain, which is why most delirium protocols treat restraints as a last resort.

Hospitals have a better tool, a bedside sitter, and family presence often substitutes. If restraints come up, it is reasonable to ask: can we try a sitter first, can family cover more hours, and can any lines and tubes that are no longer essential come out, since every tether is something to pull against.

Families arrive at this from both directions. Some ask why nobody will give him something to calm down. Others are alarmed that their mother is suddenly on an antipsychotic. The evidence answers both.

Antipsychotics like haloperidol (Haldol) do not treat delirium itself. In the largest ICU trial, 566 delirious patients got haloperidol, ziprasidone, or placebo, and the drugs made no difference in days alive without delirium or coma: 8.5 days on placebo, 7.9 and 8.7 on the drugs (Girard and colleagues, NEJM, 2018). A systematic review the same era concluded flatly that current evidence does not support antipsychotics for prevention or treatment of delirium (Neufeld and colleagues, Journal of the American Geriatrics Society, 2016). What the drugs can do is blunt dangerous agitation. The NICE guideline reserves them for people who are severely distressed or a danger to themselves, at the lowest dose, usually for a week or less, and warns against them in Parkinson's disease and Lewy body dementia, where they can cause serious harm.

Sedatives are worse. Benzodiazepines like lorazepam tend to deepen and prolong confusion in older adults, and outside alcohol withdrawal they are generally something to question, not request. The same goes for the over-the-counter sleep aid diphenhydramine, which is anticholinergic and delirium fuel. If sleep is the problem, the fixes that work are the unglamorous ones from the last section: darkness, quiet, fewer overnight interruptions.

So the bedside summary: there is no pill that switches delirium off. Treatment is finding and fixing the triggers while the drugs, when used at all, manage safety. If your person is on an antipsychotic, the questions worth asking are what it is for, what the plan is to stop it, and whether it is still needed today. Started in a crisis, these drugs have a way of riding along to discharge and beyond, and someone has to be the person who asks. A few well-aimed questions do more here than any amount of hovering.

Here are the numbers nobody on the ward will volunteer. Delirium is slower to clear than families are led to expect. In a systematic review of 1,322 older patients who developed it, delirium was still present at discharge in about 45 percent of cases. At one month, about a third were still affected. At three months, about a quarter, and at six months, roughly one in five (Cole and colleagues, Age and Ageing, 2009). Read those numbers in both directions. Most people do recover. And recovery is routinely measured in weeks, sometimes months, not the days the word "temporary" implies.

Recovery is also ragged rather than linear. Expect good days and bad days, sharp mornings and lost evenings, progress that retreats and returns. Families often describe sleep settling first, attention next, and the subtle things coming back last: initiative, humor, complex tasks like finances. Going home genuinely helps many people, and family stories of a parent who "came back to herself" at home are common and real. Home is not a guaranteed cure, but familiar surroundings, better sleep, and fewer drugs are all pulling the right direction.

Discharge itself deserves care, because hospitals discharge people who are still delirious every day. Before you leave, ask three things: what triggered this, as best anyone can tell; which new medications he is leaving on, and which of those can stop; and who is rechecking his thinking, and when. If rehab is proposed, our guide to skilled nursing and rehab stays covers what Medicare actually pays for. And do not let anyone finalize permanent decisions, a memory-care placement, a home sold, a license surrendered, based on how a person looks mid-delirium. Those decisions should wait for the recovered version of the person, assessed weeks later.

The long-term picture cuts both ways. An episode of delirium is a stress test the brain did not pass gracefully, and it is a risk marker: in pooled studies, older patients who had delirium went on to die, be institutionalized, and develop dementia at meaningfully higher rates than similar patients who did not (Witlox and colleagues, JAMA, 2010). In people who already have Alzheimer's, decline after a delirium episode ran about twice as fast in one long-running cohort study (Gross and colleagues, 2012).

That is a reason for follow-up, not fatalism. A cognitive check with the primary doctor around three months out, once the acute episode is well past, tells you what is really underneath. Many families get to that visit and find the answer is: nothing. She came back.