We describe here the results of a continuous quality improvement (CQI) project, the Delayed Discharge Project, in a general medicine service in a New Zealand teaching hospital. Average length of stay (ALOS) dropped by 2.6 days (6.5 to 3.9), readmission rates did not rise, costs of service delivery dropped by $2.4 million, patient numbers increased by 145 (2445 to 2590), while bed numbers reduced from 56 to 32 and ward outliers all but disappeared, suggesting success. However, 2 years after the successful cost-saving measures were introduced the new system crashed as a result of additional bed closures and A RECENT COCHRANE REVIEW of eight selected randomised and controlled studies found some evidence that discharge planning reduced hospital length of stay, especially for elderly patients, but no evidence was found that it reduced costs. 1 The problem of basing policy changes on randomised controlled trials is that the methodology, by its very nature, is unsuitable for assessing the outcome of comprehensive change.In this paper we show how the Delayed Discharge Project solved a bed crisis, and controlled expenditure in a general medical department in a New Zealand teaching hospital. We then consider why subsequent closure of medical beds and relocation of geriatric medicine created a second bed crisis that still needs to be resolved.In 1964, Feldstein 2 reported that medical staff with bed shortages, under pressure to admit, did not change discharge behaviour and created waiting lists. By analysing the distribution of length of stay during two bed crises, we show that bed borrowing by physicians is the modern equivalent of Feldstein' s paradox and consider the implications of this for the planning of bed allocation and service delivery in general medical departments. What is known about the topic?Little is known about the relationship between staff learnt behaviour and bed borrowing because current methods of reporting inpatient activity mask the day-to-day reality of care. What does this paper add?Using 12 years of data from a New Zealand acute medical service we reveal how comprehensive change in the process of medical care in a general medical and geriatric medical service solved a bed crisis by establishing a new constant style of discharge behaviour. What are the implications for practitioners?Under pressure to admit, staff behaviour does not change. Bed crises occur rapidly when the supply of home-ward beds is insufficient to meet the need (demand) for inpatient care. Fast-tracking homeward acute medical care depends on establishing a new stable state of staff behaviour within and without acute medical care. Thereafter, changes made in bed allocation and use within or without the acute hospital can precipitate a bed crisis in acute medical care.

No abstract