The clear strength of Naidu et al's paper ^{Reference Naidu, Bolton and Smith1} is its attempt to map the development of liaison services in London over the past 8 years. The authors have also appraised the various models of liaison services. It was interesting to see which models have been adopted in Greater London as well as the variations that exist, including the absence of a liaison service in one trust.

When we were reviewing policy documents, ^{Reference Aitken2} it has caught our attention that recommended staff numbers have not changed since they were first proposed by the Royal College of Psychiatrists in 2007. The context for this observation is the continuing reduction in acute bed numbers as well as increased recognition of the need to promptly identify and treat psychiatric comorbidities in acute settings. These developments would have been expected to affect liaison psychiatry team sizes and/or structure. It may well be that these changes have balanced themselves, hence unchanged staff numbers recommendations.

Also, treatments which would normally be given in acute hospitals are being gradually moved into the community. One would have expected that there should be a corresponding development in community liaison services to facilitate good healthcare, but this has not materialised.

Evidence suggests that untreated mental illness is associated with an increase in hospital bed days. ^{Reference Saraway and Iavin3} Depression and anxiety, for example, are likely to increase the numbers of days spent in an acute hospital bed. ^{Reference Pollack and Alovis4} Hence, it would appear that benefits accrue to acute trusts where there is a liaison service on-site. This may be an impetus for acute trusts to fund the establishment of liaison services within their set-up, but this has generally not been the case, as Naidu et al's paper illustrates.

To bring the study up to current standards, it would have been interesting for London services to have been compared against the RAID liaison psychiatry model which is now accepted as effective and efficient. ^{Reference Aitken, Robens and Emmens5} It proposes three consultants, which is an increase from the Royal College's recommendation of only one consultant.

Naidu et al suggest that demographics could possibly have had an influence on the variation in the commissioning of liaison services. For example, there may have been greater need in certain areas for particular services for older adults.

We think Naidu et al's paper would be of interest to commissioners, as it illustrates how service models have developed, with funding but without corresponding investments in the community side of liaison services, to facilitate present government policy of moving care into the community.