This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.

An anti-microbial control programme (ACP) introduced into a teaching hospital in the USA. The programme was organised by an anti-microbial advisory group (AAG), which consisted of infectious disease physicians, the microbiology laboratory director and the pharmacist specialising in infectious diseases. The latter was given primary responsibility to approve the use of restricted and non-formulary antibiotics. The AAG set up a protocol for each antibiotic that was influenced partly by cost considerations and partly by considerations of effectiveness. The AAG also changed procedures for restricted or non-formulary antibiotics and the clinical specialist became involved in the treatment of all patients on these medications. The clinical specialist was responsible for educating the doctors treating the latter category of patients. When the cases were very complicated, or were outside the approved protocols, the clinical specialist was authorised to ask the team treating the patient to consult the infectious diseases section.

Treatment and secondary prevention.

Cost-effectiveness analysis.

Patients admitted to a teaching hospital who had an infection. Patients from the mental health and substance abuse units were not included.

Secondary care. The economic study was carried out in the USA.

The dates of the effectiveness evidence were 1992-6. The dates of the resource use evidence were 1993-6. No price year was used.

Effectiveness data were derived from a single study.

The costing was undertaken on the same patient sample as that used in the effectiveness study, but it is not clear whether it was carried out prospectively or retrospectively.

There was no sample selection as the study was a time series analysis of all patients in the hospital with infection (excluding those in the mental health and substance abuse units), before and after the intervention. The sample was appropriate for the study. There was no discussion of power calculations. There were 3,649 patients in the base-line period and 3,570 patients in the post-intervention period.

This was a non-randomised study with historical controls. The follow-up was for the time spent in hospital and 30 days after leaving hospital if readmitted. This means that there was a loss to follow up as no information was provided on any of the other patients after they left hospital.

The study was based on intention to treat. The primary health outcomes used were length of stay in hospital (LOS), mortality and readmission after 30 days. The groups were assessed for comparability on the basis of their health status on admission and that controlling for confounding factors did not change the result that the ACP had improved health outcomes.

The average (SD) LOS decreased from 13.2 (15.3) days to 10.8 days (12.7), (p<0.0001). Mortality decreased from 8.28% to 6.61%, (p=0.007). Readmission changed from 10.96% to 10.42%, (p=0.46).

The ACP led to an improvement in the health of patients with infection, mortality decreased and the length of stay in hospital also decreased. The rate of readmission within 30 days stayed constant.

The outcome measures used were the reduction in mortality, the reduction in LOS and the readmission rate. There was no valuation of health states.

The costs that were calculated were pharmaceutical costs, which were measured for the years 1993-6. Quantities and costs were not analysed separately. The cost of a hospital introducing new procedures was not measured. The costs of employing the clinical specialist in infectious diseases in the pharmacy were not included. The costs were not reflated or discounted. Discounting was not relevant, as the time interval for any one patient was not long enough. No difference was reported between marginal and average costs. The hospital was the source of the cost data. Adjustment to a common price year was not made.

No indirect costs were included.

US dollars ($).

No sensitivity analysis was carried out.

The ACP was accompanied by a reduction in mortality from 8.28% to 6.61% and a reduction in LOS from 13.2 days to 10.8 days. The ACP caused part of that improvement. The duration of the benefit took place during the hospital stay and 30 days afterwards, if the patient was readmitted.

The drug costs that were recorded decreased from $473,113 to $350,888 in the first year of the programme and then fell to $303,453 in the second year of the programme. This was not the total cost of the programme, as the cost of employing the specialist pharmacist was not included.

The recorded costs went down after the programme was introduced and health outcomes improved partly as a result of the programme. There was no systematic comparison of costs and benefits, but this was not necessary as the new programme was shown to be dominant.

The ACP introduced into the Oklahoma City Veterans Affairs Medical Centre at the end of 1994 improved clinical outcomes measured by mortality and LOS. It led to reduced expenditure on antimicrobials and other drugs.

This study attempts to show that costs went down after a clinical specialist in infectious diseases was employed in the pharmacy of the Oklahoma Veterans Medical Center and that health outcomes improved. Therefore the antimicrobial control programme introduced by the pharmacist was an obvious improvement for the hospital. The comparator was the practice before the introduction of the ACP in the Oklahoma hospital which may not be the practice in other hospitals, or outside the USA.

The study was of non-randomised design, which, although prone to some biases, appears to have been appropriate for the study question. No power calculations were reported, but the sample size suggests sufficient power for the study. Mortality, LOS in the hospital and readmission within 30 days are useful but not complete as there is a lack of information on those discharged patients who were not readmitted to the hospital within 30 days.

Validity of estimate of benefit:

No attempt was made to make an estimate of benefit.

Validity of estimates of costs:

The costs of employing the specialist were not included nor were the educational costs incurred by other medical staff who were affected by the programme. The reduction in costs resulting from the reduction in LOS was not systematically analysed and incorporated into the cost data. However if these costs were included they would have supported the authors' conclusions, whereas the earlier cost omissions would have weakened the authors' conclusions. The information given on costs of patient days in hospital suggests that the overall effect on costs of the omitted factors would support the authors' conclusions. The costs that were included, drug costs, were not broken down into prices and quantities, and were not adjusted by a price index to make them directly comparable. It would have been interesting to know how much the total amount of medication per patient changed during the programme.

Other Issues:

The authors acknowledge that the study would have been more useful if they had included the employment costs of a specialist pharmacist. This is a relevant consideration for any other hospital wanting to set up a similar programme. The authors do not address the issue of generalisability: whether their hospital had special characteristics that would influence the results. Also they do not discuss whether it would be possible for hospitals to introduce some of the procedures in the ACP without employing a specialist pharmacist, who could benefit from the experience of the Oklahoma Veterans Medical Center. The authors do not discuss the data on readmission which does not show an improvement. That omission highlights the fact that we do not know the medical history of most of the discharged patients, as only mortality in hospital was included in the data. However, the authors do highlight the reduction in methicillin-resistant staphylococcus aureus isolates after the programme was started which is another indication of an improvement in conditions in the hospital.

None stated