Mukaetova-Ladinska & Scully’s comments on the challenges of conducting clinical research involving people with dementia on acute hospital wards ^{Reference Sampson, White, Leurent, Scott, Lord and Round1} raise the issue of whether BPSD may present differently in the acute hospital.

In our study, those with more severe dementia were unwell, often bed-bound, and may have been less able to display BPSD (e.g. wandering or pacing). Additionally, 12% of our cohort was taking an antipsychotic (details available from the authors on request), and many of these patients had more severe dementia. We note the problem of recall bias in staff who may report problems that are harder to manage; for example, depression in those with severe dementia may not have been recognised. Although BPSD were not associated with severity of dementia assessed by the FAST, ^{Reference Reisberg2} we notice a possible difference between those of stages 3–5 (mean total BEHAVE-AD score over admission: 2.4) and the more severe stages (means 3.6, 3.4 and 3.7, respectively, for stages 6a-c, 6d-e and 7a-f).

We agree the BEHAVE-AD scale has shortcomings; for example, it misses apathy and disinhibition. ^{Reference Reisberg, Borenstein, Salob, Ferris, Franssen and Georgotas3} Our choice was pragmatic, based on ease of administration and available staff time. The Neuropsychiatric Inventory has more detailed items on agitation and aggression, but we also used the Cohen-Mansfield Agitation Inventory to characterise agitated behaviour (details available from the authors on request) and wished to avoid duplicating data collection. We would like to highlight that most of our cohort did not come from residential or nursing care; 67% were admitted from their own home (Table 2). ^{Reference Sampson, White, Leurent, Scott, Lord and Round1}

Although admission is overall a negative experience, the precipitating illness may require hospital treatment. We had no data on BPSD prior to admission or how they would have evolved in another setting. Teasing out which elements of the admission have the strongest influence on poor outcomes, or whether the physical illness causing the admission produces negative effects, would require further investigation. Unfortunately, the answers to these questions will be complex and methodologically challenging to define properly.

There is recent evidence that improving the hospital environment for people with dementia is worthwhile. ^{Reference Harwood, Goldberg, Whittamore, Russell, Gladman and Jones4} We hope our paper provides information to inform more effective interventions.