Day et al's study of thiamine prescribing ^{Reference Day, Callaghan, Kuruvilla, George, Webb and Bentham1} was interesting and valuable. The results given in the abstract report only the small positive change in the post-intervention group, rather than reflecting the mixed picture of positive and negative change in the appropriateness of prescribing which are outlined more fully in the body of the paper. It is concerning and disappointing that such a clear and ostensibly easy-to-use flowchart did not produce the degree of change in practice that one might reasonably have hoped for, and still left the significant majority of patients apparently receiving suboptimal treatment.

The authors highlight the role of clinician-dependent factors, such as incomplete history taking on admission, lack of knowledge and disproportionate concern with rare adverse reactions. I would argue that the relative failure of an information-giving intervention to produce real improvements in clinical practice should encourage us to look more deeply at the patient-related factors which may act as barriers to the delivery of ‘optimal’ treatment.

From my own clinical experience, I would suggest that factors such as patient concordance, cooperativeness and capacity are major determinants of the feasibility of delivering what, on paper, would be best practice. Patients with chronic alcohol misuse not uncommonly have comorbid psychiatric conditions or personality styles which affect their adherence to the relatively unpleasant treatments of cannulation and intramuscular injection. Acute confusion, noted in around a third of the sample in Day et al's study, ^{Reference Day, Callaghan, Kuruvilla, George, Webb and Bentham1} would often impair the capacity to consent to treatment. The risks, to staff and the patient alike, of attempting to administer thiamine parenterally to an uncooperative individual are considerable, and must be evaluated in any best-interests decision-making process. Such patient-related factors may explain the preference among treating professionals to take the route of oral medication despite advice to the contrary, particularly in less clearly defined cases.

I look forward to seeing further exploration of factors that bear influence on the delivery of treatment in future studies.