By Emma Kampouraki
Research has been constantly criticised in recent years. It seems rather odd that while increasing efforts have been made to upgrade the regulatory framework as well as the level of research, we are facing more than ever research outputs and publications of low quality or even results of trials that still remain unpublished. However, there are some simple steps that could improve the published research outputs.
1. Reducing publication bias
First and foremost, research results, positive or negative, should be published without reservation. I often hear early-career scientists complaining about being constantly rejected by journals for their negative results that lack significance. Rarely will I find a well-known researcher publishing studies that failed to prove the hypothesis. Whole books have indeed been written about the publication bias.
Equally, protocols that failed because of not easily predictable parameters should also be reported so that similar attempts are avoided. One reason for this is failure to critically assess the prior literature and another is the unspecified statistical assumptions in the analysis of studies. A statistician should be consulted to calculate sample sizes that are required for the target power of study and to set the relevant assumptions from the beginning.
Negative results and unsuccessful protocols should be seen as equally important and we should always allow them to influence our decisions to conduct further research based on previous failed attempts, the same way as positive results urge further study.
2. Maintaining transparency in research publishing
Peer-review is powerful precisely because it is made by peers; scientists that know how to recognise high-quality research and well documented research results. Most journals today publish work that has been peer-reviewed by at least two reviewers. Selecting a journal in which low quality studies with obvious pitfalls have been published is all but good practice.
Moreover, transparency is well maintained when study protocols and data analysis plans are published well in advance. These should be in accordance with the published results when the study is completed and any reasons for deviation from the initial plan should be well justified. Most researchers should be happy to make the complete set of data publicly available, for the purposes of not only transparency but also meta-analyses. Study funders should grant access to detailed clinical data in response to legitimate requests from both researchers and regulators. These data-sharing initiatives are increasing more and more lately and should be supported.
3. Clinical trial results
Randomised clinical trials are a special category, as they are considered the gold standard in biomedical research. In reality, not all questions are answered with a clinical trial that includes an intervention. Observational studies are very powerful especially when they are well designed and bias is reduced. It is required by law that all trial protocols are pre-published at clinicaltrials.gov. Other details such as recruitment goals, data analysis plan and sources of funding are recorded as well. Funders and researchers should stick to their commitment to publish the (positive or negative) results within a timeframe after completion to inform next steps that might already be in progress (e.g. research funding applications for similar study).
4. Systematic reviews
Last but not least, systematic reviews are an incredibly powerful tool to assess the quality of existing evidence and identify gaps in current knowledge. Every large trial should be supported by a systematic review that justify its planning and of course its cost. Otherwise, there is a great risk that the particular study may not add much to the problem and therefore it won’t be cost-effective. Systematic reviews should be reproducible, peer-reviewed and according to the Cochrane standards.
Each and every new generation of researchers should feel the responsibility to maintain the quality of research that the scientific community demands. It is now more essential than ever that we provide powerful and undoubtable evidence simply because we rely a lot on it to make informed decisions in clinical practice and patient management. We are all involved so we should care!