There is a paucity of research on the appropriateness of prescribing in secondary care in the UK. Existing work has been based on implicit criteria, rather than robust and explicit indicators. Further, little is known about the application of such indicators or criteria in practice. This study aims to explore the operationalisation of explicit indicators of the appropriateness of long-term medication. This is the first part of a programme of work which will go on to look at hospital-led, "inappropriate" prescribing from the GPs' and patients' perspective. For illustrative purposes, the data presented here concentrate on one indicator, namely the indication for which the drug was prescribed.

A random sample of 50 patients were selected who, during their admission, were commenced on drugs intended for continued prescribing in primary care. Data from the patient records (comprising all medical and nursing notes, prescriptions and letters, whether paper or computer-based) were used in the application of the indicators. The indication was defined as the sign, symptom, or condition for which the drug was prescribed and was valid if upheld by the BNF. As the drugs were to be continued by the GP, this indication had to be recorded in the discharge prescription or letter.

The patients were aged 64.5+16.2 years and they had been in hospital for 8.2+7.0 days, prior to discharge on 111 eligible drugs (mean 2.2+1.3). A valid indication was recorded for 73 drugs (65.8%); an invalid indication for 6 drugs (5.4%) and no indication for 25 drugs (22.5%). There was ambiguity over the indication for a further 7 drugs (6.3%). Although no indication was given to the GP, for 9 drugs an indication was written in the in-patient medical or nursing notes. For 11 drugs, the indication could be surmised from other information provided in the discharge information, commonly due to drugs being prescribed to replace others. However, no explanation for the change was explicitly stated. This work was supplemented by interviews with hospital doctors which suggests that prescribing may be given a low priority within their overall responsibilities.

It is well recognised that hospital discharge letters are often incomplete. However, even access to the full patient record often requires assumptions to be made. It has been suggested that the medical record does 'not so much reveal what went on: it presupposes that the reader already knows what usually goes on, and if furnishes details as to this specific occasion¹'. Such suppositions may make the work of the hospital doctors less arduous, but it has profound implications for research such as this that uses explicit operational definitions. The next phase of this work will seek to explore the impact of poor information transfer on the GP's decision-making process.

1. 1. Berg M. Medical work and the computer-based patient record: a sociological perspective. Methods InfMed 1998:37:294-301