Information is increasingly recognized as a clinical intervention and there is a growing demand to provide controlled data on the clinical effect of information technology. Unfortunately, it is difficult to access the results of information service trials and quantitative reviews (meta- analyses) of the available evidence have not been published in the area of medical informatics. The purpose of this project is to assess the clinical value of computer- assisted and conventional information services by synthesizing evidence from randomized controlled clinical trials. A series of meta-analyses is proposed to evaluate the effect of major information interventions on the behavior of providers and patients. These analyses will evaluate the effect of (1) periodic delayed information feedback on clinical practice patterns, (2) alerts and reminders on medical decision-making, and (3) interactive patient education, instruction, and self-administered therapy on health status. In addition, cumulative meta- analyses will focus on several other promising information interventions including computer-calculated prediction, summaries of patient data, and computer-assisted planning of drug treatment. When appropriate, computer systems will be compared with conventional information services. Extensive and systematic searches will be performed to collect all relevant trial reports: bibliographic database retrievals (e.g., MEDLINE, CINAHL, HEALTH), manual searches (unindexed publications, reference lists), extended intervention searches, and informal contacts (experts, reviewers). Through the evaluation of the initial pool of trial reports (eligibility check, quality rating) studies will be selected for meta-analysis. Blinded extraction method will be applied to obtain the necessary data from the reports. If necessary, the authors will be contacted form more information. Binary and continuous effect (end-point) variables will be transformed and summarized in tables (odds ratio, rate difference, mean value, t value). Point and .95 interval estimates will be calculated for each study and for the overall effect. Statistical hypothesis testing will check homogeneity of the pooled studies, effect of information intervention in comparison to the control, and influence of various factors on the clinical effect of the intervention. Sensitivity analysis will assess the potential modifying effect of unpublished studies on the validity of our conclusions.
| Balas, E A; Boren, S A; Brown, G D et al. (1996) Effect of physician profiling on utilization. Meta-analysis of randomized clinical trials. J Gen Intern Med 11:584-90 |
