In reply: Our intervention1 was developed following concerns with the quality of test ordering in our department, and resulted in a significant improvement in the checking and documentation of test results. Quality of care is also influenced by test over- and underutilisation. The study demonstrated a marked reduction in test ordering and accorded with the current estimates for test overutilisation.2 Test underutilisation was monitored using established mechanisms for reporting critical incidents (including missed or incorrect diagnoses) and patient complaints, as well as feedback from staff, general practitioners and other departments and hospitals. Patient outcome factors (eg, readmission rates, length of stay) were confounded by the dramatic onset of access block during the intervention period, making retrospective comparisons unreliable.

The intervention is unlikely to have increased GP and outpatient referrals, as the previous practice of the department had been to refer patients having non-urgent tests (where the result was not immediately available) to a GP or outpatient clinic for follow-up of the test result. The department did not have the resources to ensure all patients attended for follow-up of the test result, raising medicolegal concerns and quality-of-care issues. The intervention, by deferring the ordering of non-urgent tests until after review, has the potential to reduce test duplication and the ordering of inappropriate (specialised) tests by junior emergency department staff. In addition, this process allows the patient's condition to be reviewed to determine whether further or alternative testing is required.

Gosbell and colleagues speculate on the potential adverse outcomes that may follow from a reduction in the routine ordering of blood cultures. Blood cultures change patient management in only a fraction of cases, and the clinical situations where this occurs have been defined.3 A major concern is the high rate of false positive results and the consequent economic and social cost of additional unnecessary testing, treatment and prolonged hospital stay.4 As with any clinical tool, the use of blood cultures must be supported by evidence-based guidelines rather than based on expert opinion.

The widespread use of blood cultures to limit the prescribing of broad-spectrum antimicrobial agents and development of disease resistance needs to be subjected to scientific examination and a cost–benefit analysis. If, as Gosbell et al seem to argue, the value of blood cultures lies predominantly in their public health role, the public have a right to be informed of the evidence used to substantiate the "test-all" approach being advocated to allow proper debate on the opportunity costs to public health.