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.

The study assessed the provision of brief alcohol interventions (BAIs) to trauma patients treated in hospitals and emergency departments (EDs). The comparator was no screening and no intervention.

Secondary prevention.

Cost-effectiveness analysis.

The study population comprised adult patients who were hospitalised (including those treated in EDs) because of injury and who had a blood alcohol concentration (BAC) of at least 100 mg/dL or a positive result on a standard brief alcohol disorder screening questionnaire. Patients with serious coexisting psychiatric disorders were excluded from the study, as were those incapable of participating in a brief intervention because of severe disability.

The setting was secondary care. The economic study was carried in the USA.

The effectiveness data were derived from studies published between 1969 and 2002. Most of the costs were derived from sources published between 1999 and 2002. All of the cost data were reported for the price year 2000.

The effectiveness data were derived from a review and synthesis of completed studies, augmented by authors' assumptions.

A decision analytic model was employed to determine the possible outcomes and costs of the adoption of BAIs, including the screening of BAC or the screening for drinking problems (using a standard screening questionnaire). In addition, to compare them to a no screening and no intervention scenario. The decision analytic model incorporated a Markov model that accounted for transitions between different health states. The health states included in the Markov model were well, injured requiring hospital admission and injured requiring ED care. The length of each cycle was one year and the cost of each health state mounted up during the cycle. The model was run for 3 years. The software used to proceed with the analysis was Data 4.0 (TreeAge Software, Williamstown). The model was based on the assumption that transition probabilities and the costs associated with each health state were time independent.

The following parameters were used in the model:

the number of adults (>/= 18 years of age) treated for injuries in EDs in the USA during 2000;

the proportion of injured patients or ED patients who required hospitalisation;

the proportion of injured patients with psychiatric illness who were excluded from the intervention;

the prevalence of intoxicated injured adult patients treated in the ED (with BAC >/= 100 mg/dL);

the prevalence of non-intoxicated trauma patients with a positive substance abuse screening questionnaire;

the consent rate to participate in a brief intervention study;

the annual proportion of injured adult patients with intoxication or alcohol problems readmitted to an ED for trauma;

the relative risk of readmission in injured patients with alcohol dependency in comparison with injured patients without dependency;

the hazard ratio when defined as the effectiveness of the intervention in reducing annual ED injury recidivism; and

the hazard ratio when defined as the effectiveness of the intervention in reducing annual injury recidivism requiring hospitalisation.

Primarily, the authors included studies that reported alcohol intoxication (BAC) or a drinking problem documented by positive standard screening questionnaires (CAGE, Michigan Alcohol Screening Test, or Alcohol Disorders Identification Test). They also included studies that reported BAIs for patients treated because of acute injury and had utilisation of health care resources as an outcome. The authors did not particularly report the study designs included in the review, but they mentioned that a randomised controlled trial was used to estimate the effectiveness of the intervention.

Not reported.

Not reported.

Not reported.

Overall, 28 primary studies provided effectiveness evidence.

The authors reported that it was not possible to conduct a meta-analysis. However, to obtain a baseline proportion of screen-positive patients who get treated in EDs on account of injury, the proportion of patients treated from each study was weighted by the study sample size. In addition, to obtain a screen-positive prevalence, the equivalent proportion from each study was weighted by the sample size.

The authors reported that the screen-positive rate varied between primary studies, owing to different proportions of all patients screened and to the different methods employed to categorise a screen as positive. It was because of these differences that a meta-analysis could not be undertaken.

The number of adults (>/= 18 years of age) treated for injuries in EDs in the USA during 2000 was 20,507,601 (range: 18,407,275 - 22,607,928).

The proportion of injured ED patients who required hospitalisation was 6% (range: 4.61 - 7.38).

The proportion of injured patients excluded from the intervention because of psychiatric illness was 2.4%.

The prevalence of intoxicated injured adult patients treated in EDs (with BAC >/= 100 mg/dL) was 7.4% (range: 3 - 12).

The prevalence of non-intoxicated trauma patients with a positive substance abuse screening questionnaire was 19.6% (range: 11.7 - 45.8).

The consent rate to participate in a brief intervention trial was 76% (range: 57 - 94).

The annual proportion of injured adult patients with intoxication or alcohol problems readmitted to an ED for trauma was 28% (range: 5 - 50).

The relative risk of readmission in injured patients with alcohol dependency in comparison with injured patients without dependency was 2.2 (range: 1.4 - 3.5).

The hazard ratio was 0.53 (range: 0.26 - 1.07) when defined as the effectiveness of the intervention in reducing annual ED injury recidivism, and 0.52 (range: 0.21 - 1.29) when defined as the effectiveness of the intervention in reducing annual injury recidivism requiring hospitalisation.

Some estimates of effectiveness were based on authors' assumptions.

The authors assumed that the proportion of patients who required admission to the hospital was equivalent to national estimates.

The authors did not derive a summary measure of benefit in the economic analysis. They only calculated savings (i.e. the costs averted). Therefore, the analysis is classified as a cost-consequences analysis.

The health service costs included in the analysis were for screening for alcohol use, the BAI and health care. Screening costs covered a blood alcohol test and a brief alcohol-disorder screening questionnaire, including the cost of materials (e.g. paper, photocopies). The costs of the BAI covered the professional's (psychologist) time attributed directly to the intervention and administrative time (e.g. follow-up and documentation). The health care costs covered the ED visit on account of injury, and hospitalisation. As the time horizon of the model was 3 years, discounting was appropriately carried out at a rate of 3%. The costs and quantities of resources used were all derived from published sources. All of the costs were appropriately adjusted to reflect 2000 prices, using the Consumer Price Index for wage data and the Consumer Price Index for Medical Care for medical expenditures. The costs of post discharge medical care and of patients who died were not included in the analysis.

No statistical analysis of the costs was undertaken.

The indirect costs were not included in the analysis.

US dollars ($).

A one-way sensitivity analysis was undertaken to assess the robustness of the results to variability in the data. Numerous parameters of the model were tested. For example, the costs of hospitalisation, an ED visit, a blood alcohol test and screening materials, the discount rate, and BAI provider wage and provider time. Also tested were the effect of the BAI on ED and hospitalisation trauma recidivism, the proportion of injured ED patients (non intoxicated, alcohol-dependent), the probability of ED recidivism in positive screened patients, the probability of accepting a brief intervention, the proportion of ED injured patients who are intoxicated, and the relative risk of repeated hospitalisation in alcohol-dependent and positively screened patients. All of the ranges used were derived from the literature. The authors constructed a "tornado plot" to compare the relative influence of each model parameter on the final costs.

A multivariate sensitivity analysis, where key parameters of the model were varied simultaneously, was also undertaken using Monte Carlo analyses. The authors used Data 4.0 software. All distributions were sampled 50,000 times. Probability density functions for key variables roughly matched the high and low ranges derived from published studies. Further, the costs of an ED and of hospitalisation were modelled using a gamma distribution, while all parameters that estimated the effectiveness of the intervention at reducing ED and hospital recidivism were modelled following a log normal distribution.

See the 'Effectiveness Results' section.

The cost results were reported per patient.

In the brief intervention group (screening and intervention), the weighted average cost over a 3-year period (including ED visits and hospitalisation) was $600 per trauma patient. In the scenario without screening and intervention, the equivalent cost was $689 per trauma patient.

The authors also reported the cost-savings. The screening and intervention option resulted in cost-savings of $89 per trauma patient screened, or $330 per patient receiving the intervention.

The intervention strategy resulted in cost-savings in health care of $3.81 for every $1.00 spent on screening and intervention.

The costs and benefits were not combined.

The sensitivity analysis demonstrated that the results were robust under most assumptions. The most influential parameters were the cost of hospitalisation, the hazard ratio for trauma recidivism after receiving the brief intervention (varied between 0.21 and 1.29) and the probability of subsequent injury that necessitates ED care (varied between 0.26 and 1,07). The Monte Carlo analyses demonstrated that the intervention would result in a decrease in health care expenditure in 91.5% of the simulations, or in cost-savings or a cost of less than $24 per patient screened in 95% of the simulations.

A screening and brief alcohol intervention (BAI) strategy to address alcohol problems in trauma patients was cost-effective.

The authors compared a screening and BAI strategy in trauma patients with the option of no screening and no intervention. The choice of the comparators was explicitly justified but the BAI was not described in detail.

A systematic review of the literature was not undertaken. Although this is common practice with models, it does not always ensure that the best data available are used in the model. The methods used to derive the estimates of effectiveness were not reported, but the authors did adopt a weighting scheme to reflect differences in sample size. There was little detail of the quality of the retrieved studies, making it difficult to comment on the quality of the efficacy estimates. However, the authors carried out several sensitivity analyses relating to the efficacy estimates. These analyses improve both the internal validity and the generalisability of the study by demonstrating the robustness of the results to changes in the base-case estimates. The estimates of effectiveness that were based on authors' assumptions were provided with reference to the medical literature.

Although the authors reported that a cost-benefit analysis was conducted, no measure of benefit was used in the economic analysis. Therefore, the study should be classified as a cost-consequences analysis.

The analysis was conducted from the perspective of the party (e.g. insurers, government or hospital) who incurs direct health service (medical) costs. It seems that all the relevant costs have been included in the analysis. It was reported that the costs of post discharge care for more severely injured patients and of death were excluded from the analysis, but their omission is unlikely to have affected the authors' conclusions. Overall, the cost analysis was well constructed as the costs and the quantities were reported separately, inflation adjustments and discounting were appropriately conducted, and the price year was reported. In addition, the authors conducted extensive sensitivity analyses to assess the robustness of the results and their generalisability to other settings.

The authors compared their findings with other published results, showing consistency in their findings. The issue of the generalisability of the results to other settings was not directly addressed. However, sensitivity analyses were carried out to assess the robustness of the results and their generalisability to other settings. The authors do not appear to have presented their results selectively. The study enrolled adult patients who had been injured and were found to be alcohol screen positive, and this was reflected in the authors' conclusions.

The authors reported a number of limitations to their study. First, that the estimates of the effectiveness of the intervention in reducing ED injury recidivism was based on a randomised study that reported the impact of trauma-centre interventions on the annual rate of ED injury-related visits; this was because of variation in the methodology of other, related published studies. Second, since the study did not include the costs of follow-up care, the medical care costs are probably an underestimate. Third, the economic analysis did not include any indirect costs (e.g. productivity losses) and did not evaluate quality of life. However, the authors felt that if a societal perspective had been adopted in the economic analysis, the screening and intervention option would have resulted in greater cost-savings.

The authors felt that legislative changes that forbid insurance companies to deny payments on a claim when an individual was injured and found to be alcohol screen positive should be implemented in all US states. They also recommended the development of a policy of routine-trauma centre screening and provision of brief interventions to those positively screened. The authors did not make any recommendations for future research. However, their discussion indicated some areas where further information could prove useful.

Supported by a grant from the Robert Wood Johnson Foundation.