——> Follow us on TWITTER

Mega studies – the new trend?

Epidemiology has never been a static science in its more than 100 years of existence. Methods change over time, and research, focus, funding and infrastructures do not remain as they once were. Still, the changes we see now overshadow anything we have seen in the past. Large-scale cohorts are being planned, have started and will start soon all over the world. Funding exceeds anything we have seen before with hundreds of millions of dollars, even billions. All is driven by new technologies; the ability to examine millions of gene variations, epigenetic changes, copy variants and soon total gene sequencing at low costs using small volumes of blood, saliva or other bio materials. Most of this is based on cohort studies, because the research ideas are explorative or lacking. Strong deduction based on a prior hypothesis stating predictions that can be refuted is so much yesterday’s thinking and no one remembers Popper anymore. Get information, the more the better and hope for the best. Epidemiologists are key in the movement; they have experience in collecting data, but maybe not in setting the research agendas in this new area.

Since genes may have pleiotropic effects, and there are millions of gene variants to study, and strong hypotheses are limited, the cohort approach is a natural choice. We have “exposures” (genetic) looking for a disease. These cohorts have to be large, 100,000 or more, better with more, since some of the endpoints are rare. Even then, rare outcomes will often require collaboration from several cohorts, and meta analyses are conducted in coordination centers that pay collaborators with co-authorships with dozens, even hundreds of names, producing difficulties for bibliometric analyses and research appraisal for government funding and institutional rankings. Authorship and authorship position may be proportional to the amount of data provided, be alphabetic, and be rationed by study. The terms “joint first” authors and “joint senior” authors have arisen to deal with the “sharing” involved and the need to remain collaborative and competitive.

We have seen plans for large-scale birth cohorts in many countries; the National Children’s Study is the most ambitious with a budget that exceeds 5 billion dollars. Now come the Life-Gene study, national cohorts in Germany and large-scale regional cohorts in China (e.g. Kadoorie) and Latin America. We already have Framingham, Nurses Health, the UK Birth Cohorts, the EPIC studies, Monica and others. We will get many more cohorts in the years to come, and the amount of data we will have for each participant will outnumber anything we have worked with before. New developments in computer science make it possible, perhaps even useful.

Epidemiologists should be excited and welcome this trend, if they think as scientists more than taxpayers. And they cannot do anything about it but to take responsibility for adding social and environmental data to the cohorts. With these data sources we will get an infrastructure that makes epidemiology a very attractive research career. It used to take time to start a research project from scratch in epidemiology, often much more time than in biologic research. Furthermore, non-responding populations and the time consuming process to get permission to do research make use of already existing data sources attractive, although these sources will often lack information relevant to specific hypotheses, since they were established to be data sources for many hypotheses. A cohort study with prospective data but limited data on confounders and perhaps also on the exposure may still be better than the alternatives; an ad hoc study with low response rates and poor funding.  Epidemiologists could even be forgiven for fitting their hypotheses to the data actually available. Research is driven as much by opportunities as by research ideas.

The problem is that these cohorts are very expensive and will take research resources away from other research activities – particularly large scale randomized trials relevant to prevention and control of common diseases that are capable of making policy. It is therefore important to make sure that these cohort data will be available for all with good quality study plans.

Epidemiologists now have the opportunity to be (or remain) a leading discipline in both public health and clinical research. New analytical models, large resources and new lab technologies will provide opportunities and challenges never seen before. We have to establish new research training programs, and we have to scale up the number of PhD students we produce.

We also have to make sure that not all epidemiologists are carried away by all of these opportunities. The big avoidable determinants of ill health and diseases are to be found in our social environment and not in our genes. Hopefully, some will remain responsible for the big picture.

Jørn Olsen, Cesar Victora, Neil Pearce, Shah Ebrahim

Share Button