Article Text

Download PDFPDF
Critical appraisal for emergency medicine: 4 Evaluation of service organisation and delivery
  1. S Goodacre
  1. Medical Care Research Unit, University of Sheffield, Sheffield, UK
  1. Dr S Goodacre, Medical Care Research Unit, University of Sheffield, Regent Court, 30 Regent Street, Sheffield S1 4DA, UK; s.goodacre{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Emergency physicians should use scientific evidence to decide how to organise emergency services. Triage, staffing changes, educational interventions and short stay facilities are examples of organisational interventions that require evaluation. Applying rigorous research methods to address these issues is challenging and it may be difficult to find robust data. Nevertheless, this should not be used as an excuse for basing organisational decisions upon hunches or anecdote, rather than scientific evidence.

Appraisal of studies evaluating changes to service delivery have not traditionally been covered in great detail in texts on evidence-based medicine. However, their importance in emergency care means that it is worthwhile for us to be familiar with the key issues. If we do not consider these studies as a separate group then there is a risk that we may either attempt to apply appraisal methods used to assess clinical trials inappropriately, or accept at face value claims made on the basis of very weak methods, such as simple before-and-after intervention comparisons.


The advantages of randomisation described in the previous article in this series also apply to the evaluations of service organisation and delivery, although the practicalities of using randomisation are much more challenging. If patients, carers or researchers can select which service the patients receive in a comparison of two services, then the findings are very likely to be subject to bias. Randomly selecting patients to receive one service or another provides powerful protection against bias. However, this requires us to provide two services simultaneously, which is often not feasible. Furthermore, there are some interventions, such as triage methods, which are inevitably applied to groups of patients rather than individuals.

In these circumstances cluster randomisation may be used. Groups of patients are randomly selected instead of individual patients. For example, periods of time (such as days of the week), …

View Full Text


  • Competing interests: None.