Meta-analysis is the quantitative process of retrieving and subsequent combining of the results from separate, yet related studies. With increased implementation of clinical information systems and particularly clinical decision support, it will be interesting to see how meta-analysis is utilized by these systems.
Meta-Analysis in Clinical Information Systems
Karl Pearson, an English lawyer, mathematician, statistician (as well as staunch eugenicist) is, by most accounts, the first person to use meta-analysis in medical research. In 1904, he integrated information found in a number of different studies which had assessed the effectiveness of possible serum vaccines against typhoid fever.
Perhaps the first published look at meta-analysis for therapeutic intervention was in 1955, which looked at the effectiveness of using a treatment for postoperative wound pain, cough, as well as chest pain. Ironically, the “intervention” was the placebo, and the study was aptly named “The Powerful Placebo.”
Meta-analysis provides a number of benefits for clinicians. For instance, it lessens the time and cost required to accumulate enough patients from a particular office, hospital, or system in order to help make informed treatment decisions. With meta-analysis, larger patient outcomes are able to be evaluated, additional data is obtained, and more is learned from existing treatments.
Determining differences between therapies is another area meta-analysis can influence. Often there can be little, if any distinction seen between medications, procedures, etc. in clinical trials alone. Sometimes meta-analysis can more easily allow for us to observe these differences, if in fact they exist.
In addition, meta-analysis can help demonstrate statistical significance more easily due to the vast amounts of data that can be combined. Certain endpoints just cannot be established without either large-scale studies performed, or meta-analysis of existing studies. Due to the already discussed issues of time, cost, and number of patients needed restrictions, meta-analysis can often be a much more viable option.
Meta-analysis can provide credence to the generality of study results. Studies sometimes contain mainly certain patient populations. For example, most of the patients evaluated may be male or female, young or old, healthy or sick, wealthy or poor, etc. If different studies investigating certain patient populations are coming to the same conclusions, it can be surmised that the results of the treatment being analyzed may have some degree of generality.
There are, however, restraints with meta-analysis and not everyone is a proponent. For example, bias of publications has been one issue cited, along with some individuals questioning whether it is truly a statistical science. Bias is currently one of the areas of emphasis being addressed in meta-analysis.
- Egger M, Ebrahim S, Smith GD. Where now for meta-analysis? International Journal of Epidemiology 2002 Feb;31:1-5.
- Beecher H K. The powerful placebo. JAMA 1955;159:1602-6.
- Egger M, Smith GD. Meta-analysis Potentials and promise. British Medical Journal 1997 Nov;315(7119):1371-4.
- Sutton AJ, Higgins JP. Recent developments in meta-analysis. Statistics in Medicine 2008 Feb;27(5):625-50.
Submitted by Justin Fossum