We read with interest the paper by Okorie et al, ^{Reference Okorie, McDonald and Dineen1} which studied the characteristics of patients who present frequently to emergency services. Knowing the local profile of emergency presentation is critical to targeting improvements in service provision, and we were glad to see the authors tackle this matter. We do raise two issues with the study, one methodological and one regarding applicability of the conclusions.

First, age and gender were matched as part of the study between the two groups. These are important measures of demographics that may predict frequent attendance, as has been concluded in the previous studies referenced. ^{Reference Arfken, Zeman, Yeager, White, Mischel and Amirsadri2,Reference Sullivan, Bulik, Forman and Mezzich3} We presume the authors were aiming to reduce confounding by these variables using a case-control design. Matching is used ‘to ensure that controls and cases are similar in variables which may be related to the variable we are studying but are not of interest in themselves’. ^{Reference Bland and Altman4} We think age and gender are of interest, and wonder whether controlling for these factors makes it less easy to decide the target group for community-oriented strategies. The previous studies were in different health systems, and it may be an unwarranted assumption that there will be similar gender and age relationships in an Irish population. It is unsurprising that the mean age and gender of the two groups are equal, as this was matched for at the start of the study. This led to an incorrect conclusion being made in the first paragraph of the Discussion - ‘frequent attenders… had equal gender distribution as compared with single attenders’. From the data, it appears that one can only draw the conclusion that gender distribution was equal within the frequent attender group.

Second, it would be useful to know what other services are available in Galway. If no early intervention in psychosis team was present, then perhaps this is why people with schizophrenia present more often at the accident and emergency department according to this study data. Without this information, it would be difficult to apply the conclusions to other localities. We would like to know about the structure and nature of community teams in the area and the provision of substance misuse services.