Systematic reviews in animal research
Is there scope for more widespread use?
"A systematic review involves the application of scientific strategies, in ways that limit bias, to the assembly, critical appraisal, and synthesis of all relevant studies that address a specific clinical question"
Systematic reviews are used to study a specific question in depth and provide a basis for making clinical decisions. They usually analyse and compare the human research data generated by clinical trials. To be effective systematic reviews require that all the studies included in the review have similar methods and objectives, so that there is a sound basis for comparing their data. The literature databases must be searched extensively so that all relevant information is included, and the studies selected must be included according to the criteria of the review. This approach helps to minimise bias.
Meta-analysis is a type of systematic review which uses statistical methods to combine the results from several studies addressing the same problem. Clinical studies often suffer from a lack of statistical power due to a high degree of variation among subjects and small sample size, and publication bias is common because negative studies often do not get published. The meta-analysis attempts to minimise these effects, producing “summary” result which represents the outcome of several different studies. This increases statistical precision, making the result more clinically relevant.
Not all systematic reviews are suitable for meta-analysis, which requires that the studies use similar methods and assess the same outcomes, so that the data can be analysed in the same way without introducing errors.
In contrast, a narrative review, is often broader in scope, does not usually specify methods of selecting material, does not specifically attempt to avoid bias and usually has a more qualitative outcome. It gives an overall view of the available literature and is often a subjective account.
While narrative reviews are common in animal research, systematic reviews and meta-analyses are rare. Only 103 such reviews, comprising 57 systematic reviews, 29 systematic reviews accompanied by a meta-analysis and 17 meta-analyses only were identified up to 2005.
The value of systematic reviews of animal research would appear to be self-evident. However, there are problems.There is often no way of getting negative results published and whereas clinical trials are often reasonably similar experiments, studying the effect of an intervention (drug or treatment) on clinical outcome, animal experiments are likely to have different methodologies and objectives. Some studies are designed to study mechanisms rather than outcomes, there may be important differences in species, strain and methods, and the repetition of experiments, which is often crucial to generating sufficient data in clinical trials, is strongly discouraged both by regulators and by the publication process. These differences in approach mean there may be relatively few identical experiments which can be combined in the systematic review, and meta-analysis is unlikely to be possible in most cases.
When animals are used as models of humans, the study results must answer two questions to effectively inform clinical outcome.
- Does a systematic review of results for a particular treatment or medicine in an animal model lead to a clear conclusion in animals?
- Is the animal model a good model of humans?
No systematic review can answer the second question unless the response in humans is already known. On the other hand, the more certain outcomes of animal studies that can be delivered by a systematic review would help to increase confidence in the results.
Recently, several systematic reviews and meta-analyses of animal research have been used to question whether animal research is useful in predicting human responses. In a study of 44 randomised controlled studies of fluid resuscitation in animals there was substantial heterogeneity in the results due to the method of inducing the bleeding, none of the experiments had sufficient power to detect reliably a halving in the risk of death, only two said how the animals had been allocated to treatments, blinding was rare, there was almost certainly publication bias and as a result the odds ratios were impossible to interpret. The authors questioned whether the animal research was helpful in determining human health care.In a larger systematic review of six interventions with known outcomes in humans in only three cases was there a clear agreement between the animal and human results.
Systematic reviews have proved most useful for assessing the efficacy of drugs and other medical interventions in animal models. While systematic reviews have highlighted problems with the quality of some animal research, this does not necessarily justify their widespread use. There are clear advantages to well-reported, unbiased studies which use consistent methods, but training animal research workers in better experimental design may have greater advantages than systematic reviews of the data. They are potentially a useful tool for analysis of animal data, but care needs to be taken that systematic reviews are limited to appropriate areas and are used with consideration for the 3Rs.
- Cook, D. J., Mulrow, C. D., and Haynes, R. B. (1-3-1997) Systematic reviews: synthesis of best evidence for clinical decisions. Ann Intern Med 126: 376-380.
- Peters, J. L., Sutton, A. J., Jones, D. R., Rushton, L., and Abrams, K. R. (2006) A systematic review of systematic reviews and meta-analyses of animal experiments with guidelines for reporting. J Environ Sci Health B 41: 1245-1258.
- Lemon, R. and Dunnett, S. B. (30-4-2005) Surveying the literature from animal experiments. BMJ 330: 977-978.
- Roberts, 1., Kwan, I., Evans, P., and Haig, S. (2002) Does animal experimentation inform human healthcare? Observations from a systematic review of international animal experiments on fluid resuscitation. British Medical Journal 324: 474-476.
- Perel, P., Roberts, I., Sena, E., Wheble, P., Briscoe, C, Sandercock, P., Macleodn M., Mignini, L. E., Jayaram, P., and Khan, K. S. (15-12-2006) Comparison of treatment effects between animal experiments and clinical trials: systematic review. BMJ.