Meta-analysis

08 February 2013

Finally, I have made up my mind to speak about meta-analysis. I’ve been evading talking about it since after my report before the team. The 116-page journal (in PDF format) that I made a report on is a meta-analysis. The research associate who assigned that to me said the day after the team meeting through email to the team that I made a good report. I felt he was just being polite or he was just building up on my self-esteem, making me feel at ease with the team. My own opinion of my presentation is on the negative. I haven’t finished my visual presentation, and there was no order in the way I presented it. I wasn’t even able to make it clear before the listeners what the journal was about. Because the team statistician was not there, I emailed him a softcopy of the journal to ask for his opinion. His reply was so technical but he made his point clear, that the meta-analysis wasn’t credible - internal and external validity both fell short.

I was trying not to impose on my audience then how I feel about that meta-analysis. I wanted them to decide for themselves if they would credit that study. I just wanted to present to them in plain terms how the authors managed all those RCTs. I failed on that objective; the basic assumption I hinged on to empower them to make enlightened decisions wasn’t invincible.

As for the statistical opinion, of course meta-analysis would fall short of statistical validity requirements, internal or external. Meta-analysis is in its early conception to pre-emergence stage, it is not yet perfect. I akin that to the time when the knowledgeable ones thought people came from fish, those times of people like Anaximander and Anaximenes. Their ideas seemed very profound then and we can laugh at their theories now because we can claim that we know better. But if not for those answers being formulated, the existence of a question to be addressed wouldn’t be made known to everyone. Those answers (however silly they may sound), confirm that the origins of humans should be known; and that knowing the answer to that question is an endeavor that needs to be undertaken by people.

Meta-analysis is only budding. It is a very exciting concept; I’m so glad that people have already thought of it in my time. As of now it is flawed and very vague, but the fact that it has made its way to the consciousness of more people who would later on be able to think more about it and contribute to its improvement, could not be overlooked. Its current framework (being prone to bias because the authors have complete power to decide for the study – which to include, remove; what should be taken into consideration; what to dismiss), methods of analysis and, attribution of effects from each included RCT to the generalized result are all dubitable but it still exists. That suffices to ensure it will be improved upon.

The tasks meta-analysis sets are gargantuan. It tries to remove the confounding effect of each study, standardizing each study until they are about equal in characteristics before they are all weighed out against each other and the combination of their results based on their weight (on a certain matter decided upon by the authors i.e., how long the RCT ran, how large their target population was, etc.) become the final results. The development of this concept is indisputable, especially now in our age where technology made advancements occur in exponentially shorter periods of time. Not convinced? Remember again ENIAC and EDVAC antiques and the ginormous floppy disks and how faster from innovation to innovation we had escalated. We are surmounting the paradigms of time and location with technology. That manner of progress is bound to spill out to other branches of learning and processes. I already foresee the structured meta-analysis that is yet to come.