A Month of Kitchen-Table Notes Walks Into the Memory Clinic

Forty caregiver-patient pairs in Rome kept a purpose-built symptom diary for a month, then sat the standard clinical interview. The diary took less time to turn into a score, held up between two independent raters, and bent less to caregiver stress. It is small, single-centre and cross-sectional, and it settles nothing on its own.

Almost every family living with dementia ends up keeping some version of a diary. A notebook by the kettle. A running note on a phone. The scribbled observation that the shouting starts when the television news comes on, that it passes if someone sits down and puts the kettle on instead, that the worst hour is the one before dinner. Then the clinic appointment arrives, and a month of that accumulated knowledge has to be compressed into a structured interview in which a clinician asks whether there has been agitation, and how often, and how bad, and the caregiver, who slept badly, answers from memory. The question a group at Sapienza University of Rome set out to test is whether the notebook could be used directly instead.

Their study, published online on 23 April 2026 in the Journal of Alzheimer's Disease, worked like this. Consecutive outpatients with dementia and neuropsychiatric symptoms at the Center for Cognitive Disorders and Dementia in Rome were considered for enrolment. Caregivers were given a specially designed symptom diary and kept it for one month. Afterwards the standard Neuropsychiatric Inventory (sNPI) was administered in the usual way, as a caregiver interview, and separately two trained raters independently reconstructed NPI scores from the diary alone, without seeing each other's work. Forty patient-caregiver dyads completed the procedures. Scores were compared with repeated-measures ANOVA and Cochran's Q test, regression models looked at what else was driving the numbers, and the diary entries themselves were analysed qualitatively, with caregivers giving structured feedback. This is a cross-sectional comparison of measurement tools in one clinic. No treatment was given and no outcomes over time were measured.

Three results carry the story. The standard interview produced higher scores than the diary-derived versions, and took longer to administer. The two raters working from the same diary agreed closely with each other, which is the technical claim behind the word reliability here. And the diary-based scores were less influenced by caregiver burden, meaning that how strained the caregiver was mattered less to the number that came out. The authors describe the diary as "ecological, caregiver stress-resistant, reliable, time-efficient, and feasible", and are careful to frame it as complementary: integration with standard evaluation, not replacement of it. That caution is earned, because a lower score is not automatically a truer one. With no independent observational yardstick, this design cannot tell you whether the interview inflates symptoms under stress or the diary quietly misses some of them, and 40 dyads at one Italian memory clinic is a long way from a settled answer.

What the diary did add is the part the standard instrument was never built to hold. The Neuropsychiatric Inventory, published by Cummings and colleagues in Neurology in December 1994, scores a set of behavioural domains by frequency and severity as reported by the person who knows the patient. It is a good tool, and it is the reason behavioural symptoms are counted at all in trials. But frequency times severity does not record that the agitation always follows the same trigger, that it resolves when a particular person arrives, that it clusters in the late afternoon. The Rome diary captured temporal fluctuation, triggers and resolution strategies, and that is the difference between a number that goes in a chart and a description a clinician can act on. The idea has precedent: the paper's own reference list points back to an earlier Italian adaptation of the NPI into a daily diary format, published in 2018.

This also lands squarely on the puzzle left by the trial we reported earlier. In the randomised comparison of two online caregiver platforms, GamePlan4Care and Resources4Care, published in Alzheimer's & Dementia on 14 October 2025, a six-month exploratory trial with 120 caregivers enrolled per platform, caregiver distress on the NPI-Q improved in the GamePlan4Care group (effect size 0.30, p = 0.010) while the severity of the behaviours themselves did not shift. One reading is that the intervention helped the caregiver and not the behaviour. Another, which the Rome study makes harder to dismiss, is that when the measure of behaviour runs through a stressed caregiver's recall in a single sitting, you are partly measuring the caregiver. If diary-derived scores really are less sensitive to burden, trials that use them might separate those two things more cleanly. Might. That hypothesis has not been tested in a trial.

For a family reading this, the honest translation is narrow and still useful. Nothing here says a diary should replace an appointment, and nothing here is a reason to change what anyone is taking or doing. What it says is that the informal log many caregivers already keep is not merely sentimental record-keeping; in a structured format, two strangers reading it independently arrived at closely matching clinical scores, faster than the interview took. Whether a given clinic wants that information, and in what format, is a conversation to have at the next appointment rather than a decision to make alone.

The next useful study is the unglamorous one: more sites, more dyads, diary scores checked against direct observation rather than against another report, and a hard look at whether having the diary in front of them changes what clinicians actually do.