Informed consent is a key concept in research ethics, and it has a central place in the Helsinki Declaration. In the randomized clinical trial, participants need to be informed about the potential risks and benefits of the treatment under study in order to make an informed decision before saying yes or no to the invitation to take part in the trial.
It is more complicated in observational studies. People are not invited to take part in any experiment. They are invited to provide data or agree to have their data used in research to benefit others, in some situations also themselves. In some cases, informed consent is impossible because the participants are no longer alive or capable of making a decision. In some cases, the specific aims of the study are difficult to explain, like in the GWAS studies. In some cases, it may be impractical because the study involves thousands or perhaps millions of people, using already existing data.
In the randomized clinical trial, the informed consent is often written and signed with a personal signature, but some people feel uncomfortable about signing documents they cannot fully understand. Many observational studies accept oral consent or a de facto consent. By responding to a questionnaire following proper information they accept being part of the study by filling out the questionnaire (implicit consent).
In many situations data already exist as part of health care treatment, or from previous studies. Data on gene variants are, for example, often circulated between several consortia, making it difficult not only for data providers but also for the primary data collectors to keep track of where they are and what they are being used for. Maybe participants provided a very broad consent at the time of data collection, but years later that will be forgotten, and their consent cannot any longer be classified as “informed”. Some will not accept a written informed consent to be valid for more than the time period where they still remember the conditions. Many kinds of homemade “house rules” exist together with the legal formal procedures making it very difficult to navigate complex projects.
In life course epidemiology we take an interest in data collected and stored decades ago, because understanding disease causation often requires decades, sometimes even lifelong, observations for thousands of people, and useful data are often used from existing registers that collected data with a completely different purpose. Even when personal interviews or measurements have been carried out in the past, informed consent reflects what was acceptable at that time, not necessarily complying with current ethical guidelines that are changing rapidly. Renewing a consent every year, or every few years, will in most cases be close to impossible. In principle one should assume that someone who gave full informed consent in the past would also provide such consent today, even if the ethical demands became more stringent.
The EU Directive on Data Protection labels all health data as sensitive data that can only be used after informed consent with some exceptions that are used more or less strictly across different countries, making intervention collaboration difficult and time consuming.
Handling personal data in research
Fortunately, technology has made it much simpler and safer to handle personal data on a large scale. Data can be stored on centralized servers where users have strict password protected access to data with encrypted identifiers. All data are analyzed on these servers, and only tables and statistics go back to the user. Gone are the days (hopefully) where copies of research data were widely distributed to several PCs that were used at work, at home, or on travels. Using proper data protection tools only leaves a few options of violating privacy. One is the misuse of data to stigmatize or harm segments of populations. Several governments have done this, also in recent time. Available personal data may also be provided to the police after a court decision in criminal cases, and data may be available for insurance companies according to the contract between the company and the individual seeking insurance. These risks relate to the existence of personal data, not to their use in research.
New technologies provide good protection of private information used in research. Our interest in research relates to populations, not individuals. We have no problems with only having access to personally identifiable data when we do data checking and data linkage, and we often leave that job to the data managers.
Given these conditions, the principle of informed consent makes little sense in many observational studies, and the right to privacy must also be balanced with the right to advance knowledge for the common good.
Asking for informed consent in all situations will slow down progress and may do harm. It must be possible to leave the decision of using already existing data to an IRB, ethics committee or data protection agency, depending on the regulative practice in the country. This access should be granted freely as long as the research serves the public by providing more information on what causes diseases, which treatments work and for which diseases, and which treatments have severe side effects. We should also be able to analyze how we best organize health care and prevention, and how political decisions affect public health. For example, what are the health consequences of the severe economic crises that exist in many countries?
People have a right to know what their data are being used for, and they usually have a right to withdraw their data from a given research database. Most would agree that an opt-out option should be available. A general opt-out statement for all research would be easiest to administer. In some countries, you can now refuse to receive advertisements in your mailbox or telemarketing calls, and a similar system could be arranged for research. Such a general opt-out declaration would not be “informed” because of its general nature. An informed opt-out procedure would require access to a website with information on all the research projects where your personal data are being used. Then the opt-out procedure can be explicit, similar to the explicit nature of the informed consent. Such an arrangement is rather simple to set up.
The implicit consent is, for example, being used when people return the complete questionnaire. We must assume that they know their data will be analyzed, and they trust this will be done in a fair manner. If research is considered a natural part of medical treatment one may argue that patients have accepted this if they did not use their opt-out right. By not using an opt-out option, is can be assumed that they accept that data can be used for research to benefit others, and they leave the permission process to an independent committee or board.
As illustrated, personal data can be provided by a “trusted third party”. A statistical department that maintains data linkage and quality control and provides coded data for the researchers. Data (like GWAS data) cannot be considered anonymous because their content is so rich that by combining information we make it possible to identify individuals.
This risk must be taken seriously and would be best handled by researchers themselves, those who treat patients or collect data. Leaving sensitive data to be managed by a “trusted third party” implies that epidemiologists cannot be trusted. We hope (believe) this is not true, and we find no documentation for this. Misuse of personal data has been done by governments, not by researchers.
What is certain is that our practice will change, and we need to rethink how we best secure privacy and people’s right to know. Letting informed consent be the sine qua non principle will severely limit what we can do to document determinants of health.
The idea of informed consent being a carrying principle of health care science, not just as a legal construct but as a de facto state of mind, is impossible to live up to in modern health science. Research has become a necessary integrated part of modern health care that involves thousands of people over decades of their lives. Part of this research is purely inductive without specific hypotheses to inform participants. Being informed may only mean being informed about participation in research.
People have a right to expect the best possible health care and to live in a safe environment. For that we identify health hazards as soon as possible. We need to tell them in a better way that this requires research at a much larger scale than what we have at present. We will all have to participate in these activities, and it is unrealistic to expect all to be fully informed. We have to find other means of securing people’s right to opt out of research activities they do not want to take part in, especially if the research involves risks for the participants.
Jørn Olsen, Cesar Victora, Neil Pearce, Patricia Buffler, Shah Ebrahim