Antibiotic resistance is a major problem for clinicians, but computer systems can help by automating testing and reporting procedures, writes Peter Mitchell

Concerned about increasing antibiotic-resistant strains of common bacteria like Staphylococcus, the House of Lords select committee on science and technology last year published a warning report.1

The Department of Health's standing medical advisory committee subsequently added its comments in The Path of Least Resistance, and the government has now published its own response to the Lords' report.2,3

Several witnesses to the select committee said an important way of encouraging more prudent use of antibiotics would be to test strains for their susceptibility as quickly as possible. Normally, however, this data is not available until 48 hours after taking a sample, and this is when doctors often prescribe inappropriate antibiotics, the committee said.

The SMAC report calls for 'good local information' - antibiotic resistance patterns vary considerably from location to location. It suggests both GPs and hospital clinicians use computerised decision-support systems to guide prescribing.

Proper surveillance - identifying micro-organisms down to the genus level - is vital to obtain an accurate picture of changing antibiotic resistance. But most hospital microbiology laboratories do not identify many samples this fully.

University College Hospital in London is using 'expert' software in conjunction with a special fast antibiotic susceptibility tester to gain a clearer picture of in vivo bacterial resistance. 'This shows the local incidences of resistance and gives us a picture of the bacterial ecology of each ward,' says consultant microbiologist Dr Nandini Shetty. The biggest threat is an invasion of the intensive care unit by multi-drug resistant MRSA, but Dr Shetty says constant surveillance of all quarters is vital. 'You have to have your radar scanner on all the time, or you might not see the next attack coming.'

Her department has installed a Vitek semi-automated identification and susceptibility tester that can deal with nearly all relevant specimens. 'The reports generated by the system make it far easier for us to predict how a micro-organism will react to antibiotics in vivo. If we based our prescribing practice solely on the in vitro data, it would be easy to make a mistake about likely resistance to other antibiotics.'

The expert software applies a set of rules, based on published scientific data, to the newly-generated laboratory data. By referring to a library of bacterial phenotypes, known patterns of resistance can be translated within the reports provided.

'We will be able to pinpoint problem areas for hospital-acquired infection and resistant organisms, and advise wards to commence infection control procedures,' says Dr Shetty.

At present she and her colleagues phone through a large number of results as soon as they become available. But this will be much simplified when the laboratory starts using its new Sunquest pathology information management system this summer.

'Once this is up and running, we will be able to relay all results straight back to the requesting ward.' That is, if the hospital clinicians can get used to checking their terminals regularly for new reports.

Dr Shetty hopes to add patient demographics from the hospital patient administration system into the results, thus generating local antibiotic resistance data for the hospital and community. The laboratory is preparing to deploy special epidemiology software, called bioLiaison, to generate and distribute its resistance data to hospital staff and local GPs. 'I want to use this to advise clinicians about which antibiotics are likely to work best and avoid inappropriate prescribing,' she says.

For example, guidelines could be issued quickly in a local outbreak to switch clinicians to effective narrow-spectrum antibiotics instead of the usual recommended frontline drug.

This is in line with proposals in The Path of Least Resistance regarding the dissemination of local information on prevalence and resistance. The data should also help staff decide how to rotate the antibiotic armoury so as to reduce the selective pressure for resistance to develop.

Several other trusts are developing similar reporting systems: Ninewells Hospital and medical school in Dundee; Edinburgh Royal Infirmary and Raigmore in Inverness; and SmithKline Beecham Clinical Labs, which performs West Middlesex Hospital's microbiological testing.


1 Resistance to Antibiotics and Other Antimicrobial Agents. Stationery Office, 1998.

2 The Path of Least Resistance. Standing Medical Advisory Committee, 1998. Also at

3 Government Response to the House of Lords Report. Stationery Office, 1998.