Skip navigation

Reducing the length of stay for patients with upper gastrointestinal tract haemorrhage


Upper gastrointestinal tract haemorrhage is a common and potentially life-threatening disorder. It leads to high levels of hospitalisation. Resource use varies markedly, with no corresponding effect of patient outcome. More does not always mean better. Development of scoring systems to predict risk has been perceived as a way to reduce hospital costs without any adverse effect on patient outcomes. Could length of stay be reduced while the quality of care for patients was maintained or improved? A study [1] in California has shown that a clinical guideline can safely reduce the length of stay for low-risk patients with upper gastrointestinal tract haemorrhage.


Jeffrey A. Hay and colleagues. Prospective evaluation of a clinical guideline recommending hospital length of stay in upper gastrointestinal tract haemorrhage. JAMA 1997 278: 2151-2156.

How was the question tackled?

The key stages were the formulation of a clinical guideline followed by a controlled time-series study with alternate month design.

  1. A clinical practice guideline was derived from a structured literature review. Recommendations were based on specific risk groups using a comprehensive scoring system using both clinical and endoscopic predictions of outcome. Points were assigned based on initial clinical data. Computer software assisted physicians with risk calculation. Identified high-risk patients were excluded.
  2. The study was based at the Cedars-Sinai Medical Centre a large teaching hospital in Los Angeles with about 1000 operational beds. No special hospital policies existed for the management of patients with upper gastrointestinal tract haemorrhage before the study.
  3. Patients admitted to the hospital between May 1994 and July 1995 were eligible for enrolment in the study. Staff were trained to identify and enrol patients who met the eligibility criteria and who achieved low-risk status within 7 days of admission. Criteria were explicit to minimise patient selection bias. Patients were enrolled in the study in 7 alternative months with patients in the intervening 7 months acting as controls.
  4. Feedback of validated risk information was provided to managing physicians (in notes in the patients' records or by telephone). They were asked to consider the guideline recommendations for discharge as they made their patient care decisions. The guideline and risk assessment tool was not available to managing physicians during the 'control' months.
  5. Two main outcomes for the study were patient safety measured by complication rates and resource utilisation measured by hospital length of stay. A series of explicit complications were defined prior to the start of the study. Patients were also surveyed by telephone one month after discharge about the general health using the short form SF-36. A series of statistical tests were used to analyse and validate the data collected during the study.
  6. A combination of methods was used to change physician behaviour: education, concurrent feedback, and the involvement of (known, local) opinion leaders

What did the study find?

During the 14-month study 299 patients were admitted to the hospital with upper gastrointestinal tract haemorrhage. About 200 (70%) met the inclusion criteria as defined by the guideline. The main factors why patients were not accorded low risk status were because of anticoagulation therapy (11%) and persistent unstable comorbidity (8%). The overall mean age of patients in the study was about 64 years: the majority had pre-existing comorbid disease. 101 patients were allocated to the intervention group and 108 to the control group.


What are the lessons from this study?

The results from the study were encouraging: resource use (as measured by length of stay) was reduced while patient safety and satisfaction for low -risk patients was maintained. Physicians welcomed the guideline and its use resulted in more standardised high quality practice.

This is another successful study that explored how to change clinical behaviour Specifically, to implement an evidence-based guideline, which defined the optimal duration of hospitalisation for a particular condition. It used the proven 'multi-faceted' approach - tackling an evidence-based intervention and bringing together an educational programme, structured feedback of information about progress and respected local physicians (opinion leaders).

It's another example that shows that change can be achieved if it's tackled the 'right' way. The challenge as ever is to make sure that the improvement in the quality of care for patients endures after the study ends: moving from project to mainstream.