A useful collateral history does not decide whether the patient or relative sounds more convincing. It establishes what each person observed, when it happened, how it differs from baseline and what remains uncertain. For cognitive and behavioural change, that timeline can matter more than the label an informant initially uses.
This article is published by iatroX and includes iatroX alongside SPMM and MRCPsych Mentor as preparation options. The cases and worksheets are original educational exercises, not College examination recalls. Clinical assessment principles are drawn from NICE dementia guidance and NICE delirium guidance, checked on 27 September 2026.
Start with two accounts, not two diagnoses
Consider the original fictional discrepancy: a patient says they are sleeping normally and attending work, while a sibling says the patient is awake all night and has stopped going in. Neither statement yet establishes the clinical picture. The sibling may be describing late messages, not direct observation of sleeplessness. The patient may be working remotely, taking agreed leave or using "normal" to describe a changed routine.
Ask which days each account covers. Establish whether the sibling lives with the patient and what contact they have had. Ask the patient what a typical night and working day actually involved. The resulting information may resolve the discrepancy without anyone having lied. Alternatively, it may leave a genuine disagreement that must remain visible in the summary.
Do not infer lack of insight from the sibling's frustration, or dismiss a concern because the patient gives a confident account. The task is to investigate a difference, not to conduct a courtroom cross-examination. A useful follow-up is: "Could you describe a particular example and when it happened?"
Clarify the informant, consent and the task
Identify the informant's relationship, frequency of contact, practical responsibilities and opportunity to observe the relevant behaviour. A neighbour who visits daily may have direct knowledge of meals and mobility that a distant relative lacks. A relative may know the longstanding baseline but have little recent observation. These are differences in access to information, not rankings of personal credibility.
Receiving information and disclosing confidential information are different activities. Listen to relevant concerns while checking what can appropriately be shared in return. Explain the purpose of the discussion and the limits of confidentiality rather than promising secrecy that cannot be maintained. The GMC's confidentiality guidance, checked on 27 September 2026, provides the professional framework for disclosure decisions.
Check the station instruction. A collateral-history task is not permission to announce a diagnosis, disclose unrelated information or deliver a complete management plan. Nevertheless, an immediate safety concern still needs recognition and appropriate escalation. Staying within the task does not mean ignoring a newly disclosed emergency.
Establish the previous baseline
For cognitive change, "previously independent" is too broad. Ask about managing appointments, medicines, finances, meals, transport and communication. Establish which tasks the person has always delegated and which they have recently stopped doing. A partner taking over banking after retirement is not automatically evidence of cognitive deterioration.
Explore cognition, behaviour and support together. Ask for examples of forgetting, repetition, getting lost, changed judgement or difficulty completing familiar activities. Clarify sensory impairment, language, education and usual communication style. The NICE dementia assessment recommendations, checked on 27 September 2026, support obtaining history from the person and, where possible, someone who knows them well, including the impact on daily life.
A useful baseline statement is specific: "Until the summer, she organised her own appointments and collected her medicines, although her daughter has always managed online banking." That gives a later change a meaningful comparator. It also avoids turning every supported activity into evidence of disease.
Build the timeline of change
Separate first noticed change, subsequent progression and the most recent deterioration. "Over the last year" and "since yesterday" may both be true. A gradual decline does not explain away an acute change superimposed on it.
Ask whether symptoms fluctuate, whether there are better periods, and what was happening when the change appeared. Establish recent illness, pain, disrupted sleep, medicine changes, missed medicines, substance use, hospital attendance and changes in support. Do not diagnose delirium from an informant's description alone, but recognise that an acute, fluctuating change warrants prompt assessment.
NICE's delirium guidance, reviewed on 27 September 2026, concerns hospital and long-term-care settings. Its emphasis on recent change and fluctuation is useful here, but the actual assessment pathway must fit the person's location and clinical condition. A conversational exercise is not a substitute for examining an acutely unwell patient.
Keep observation, quotation and interpretation apart
Use three different types of statement deliberately. "He sent messages at three in the morning" is an observation reported by the sibling. "He says he sleeps well" is the patient's account. "He may have reduced sleep" is an interpretation requiring clarification. Combining them into "He has not slept" manufactures certainty.
The same distinction applies to cognition. "She left the hob on twice this week" is more informative than "She is confused". Ask what was left on, who found it, whether this is new and what happened afterwards. Do not assume intent, diagnosis or severity from the initial label.
| Issue | Patient's account | Informant's account | Clarification needed |
|---|---|---|---|
| Time period | Describes the current week | Describes the previous fortnight | Compare the same dates |
| Sleep | Says sleep feels normal | Has received late-night messages | Actual sleep opportunity and duration |
| Work | Reports continuing employment | Has not seen the patient leave home | Remote work, leave or missed attendance |
| Baseline | Describes usual independence | Describes recent assistance | Which tasks have genuinely changed |
| Consequences | Reports little disruption | Describes missed commitments | Concrete examples and their impact |
This original worksheet should retain unresolved cells. An empty box is a prompt for further enquiry, not permission for an AI system or learner to fill it with a plausible answer.
Practise two older-person collateral cases
Case one: a slowly changing baseline. In this fictional role card, an older adult says that forgetfulness is no worse than their friends'. Their daughter describes repeated appointment reminders and several unpaid household bills over recent months. The daughter visits weekly; a neighbour sees the person most days. The patient still prepares meals and travels to familiar places. No acute deterioration is reported.
The candidate should establish dates, examples, consequences and the previous division of responsibilities. Ask whether bills were missed because of memory, changed banking arrangements, poor vision, financial difficulty or another explanation. A reasonable summary identifies concern about progressive cognitive or functional change and the need for direct assessment, without diagnosing a particular dementia from one account.
Case two: recent deterioration on a changing baseline. A fictional care-home resident has had gradual memory difficulties but usually recognises staff, eats with prompting and walks to the dining room. Since yesterday, staff report marked inattention, alternating drowsiness and agitation, and reduced intake. A relative says, "This is just the dementia getting worse."
The candidate should not accept that explanation without assessing the acute change. Ask about onset, fluctuation, observations, physical symptoms, pain, medicines and recent events, then communicate the need for prompt clinical assessment. The learning point is not to identify a single hidden cause from the role card. It is to recognise that the time course changes the task from routine history gathering to evaluating possible acute illness.
| Case | Baseline evidence | New information | Defensible next question |
|---|---|---|---|
| Gradual change | Familiar activities largely maintained | Repeated reminders and unpaid bills | What direct assessment and corroboration are needed? |
| Acute deterioration | Usual interaction and supported mobility | Abrupt fluctuation and reduced intake | What urgent assessment is required now? |
Summarise without erasing disagreement
An annotated handover for case one might read: "The patient reports occasional forgetfulness without major functional difficulty. Their daughter reports repeated reminders and unpaid bills over several months, but has weekly contact and cannot describe the daily pattern. There is no reported acute change. We need the patient's assessment, clarification of sensory and practical factors, and further corroboration of baseline function."
The summary attributes each claim, states its timeframe, identifies missing information and avoids a premature diagnosis. It also distinguishes "no acute change reported" from "acute illness excluded". The latter requires more than the history supplied.
For case two: "Staff describe a marked change since yesterday from the resident's usual interaction and supported function, with fluctuation, inattention and reduced intake. This needs prompt assessment for acute contributors rather than attribution to longstanding cognitive problems alone." The next handover should include actual observations and actions taken, not merely the phrase "possible delirium".
Rehearse without turning the interview into a contest
Use three people where possible: candidate, patient and informant. Give the two role players different cards and interview them sequentially. An observer records the questions that clarified timing, the statements that preserved attribution and any inference that became an unsupported fact.
Repeat with one controlled alteration. In one version, the accounts concern different dates. In another, they concern the same dates but genuinely conflict. In a third, the informant has direct observations of one issue and only second-hand information about another. The candidate must adapt rather than recite the same sequence.
Finish by asking each role player whether the summary represented what they actually said. That is a practical check on accuracy, not a popularity score or a requirement to make everyone agree. Disagreement can remain after a competent interview.
Choose preparation tools for the feedback needed
The earlier version of this comparison, published in September 2026, discussed SPMM's course and CASCBOT offering, MRCPsych Mentor's advertised AI simulator and iatroX's CASC-relevant practice. Those product descriptions should not be read as proof that a particular subscription reproduces a simultaneous multi-person consultation. Check the current demonstration and exact package before buying.
Per iatroX product information, September 2026, its simulations provide voice and text practice, transcript-linked feedback and Tutor-led remediation. A useful exercise is to rehearse a patient account and an informant account separately, then compare the resulting summary with the source statements. Do not claim that the interface directly observed a physical examination, independently verified a relative's account or established a diagnosis.
For a study group with willing role players, paired accounts and a careful observer may be sufficient. A faculty-led session is particularly useful when the difficulty is an interpersonal habit that needs expert observation. A question bank or Tutor can help when the learner does not understand which information matters. These are different needs, not a single product ranking.
As supplied in September 2026, iatroX's paid learning subscription combines question banks, Socratic Tutor, study planning, simulations and CPD tools for £99 paid upfront annually, equivalent to £8.25 a month billed annually, or £29 monthly. One complete simulation is free. The value is support for the candidate's psychiatry learning goal, not access to unrelated examinations.
Frequently asked questions
Should I accept the relative's account when it conflicts with the patient's?
Not automatically. Clarify direct observations, dates, baseline and opportunity to observe, then preserve any unresolved discrepancy in the assessment.
Do I need a multi-person AI simulator to practise collateral histories?
No. Separate role cards, sequential interviews and a source-attributed summary can provide a useful exercise with peers or suitable conversational tools.
Does a good collateral history have to produce an agreed diagnosis?
No. It should produce a clearer timeline, attributable evidence and an appropriate next assessment, including urgent action when an acute change requires it.
Practise turning separate accounts into a clear formulation →