In medicine, randomized controlled studies have long been the gold standard in defining the “best” care. In these types of studies, usually two large groups of patients, the larger the better, receive the same care except for one intervention that is different for each group and the results are compared. This defines the best care medical science has to offer. No one disputes the usefulness of these types of studies, however, are they enough? Do they really tell us all we need to know about the care of an individual?
A recent article in the journal “Medical Decision Making” by Bruce Barrett M.D. entitled Sufficiently Important Difference: Concepts, Caveats, and Challenges questions whether our approach to using and interpreting these studies is adequate. The article looks at randomized controlled studies and the facts that they define in light of a new reality in health care. The new reality is that the value from the patient’s point of view is the key factor that must be taken into consideration when doctors and patients make decisions about care. That patient point of view has rarely been used as an end point in randomized controlled studies and the evidence based medicine that results. Dr. Barrett defines this focus on value from the patient’s point of view for clinical research design purposes as the “sufficiently important difference” (SID) or the “smallest worthwhile effect” which he defines as “the smallest amount of patient-valued benefit that an intervention would require to justify associated costs, risks and other harms.” He then adds, “SID is understood in the context of existing evidence and is measured at the individual level.” Measuring at the individual level is earthshaking for clinical researchers and for those who apply clinical research to the practice of medicine because when randomized controlled studies are done, the gold standard for results is always population based. While individual decisions must be informed by population statistics, the real impact of a particular intervention on an individual may not perfectly fit a population fact found by study. By telling us that the way to do clinical research is by measuring at the individual level and that it be “patient-valued” Dr. Barrett is telling us that we have to look anew at the outcomes of studies and as a result, many of our favorite “facts” and “goals” in population health and in the diagnosis and treatment of disease. A recent case in the news illustrates this point. As I write this an 11 year old girl is going home from the hospital after receiving a double lung transplant. The standard of care based on population evidence and expert consensus is that this lung transplant should not have been allowed to happen since this child was considered too young to receive adult lungs. A persistent legal and public relations fight by her parents forced the circumvention of this rule and this child is going home now with new lungs. The population facts were not disproved and remain helpful; they just did not perfectly apply to this young girl.
Dr. Barrett in this article points out that evidence based decisions, which are usually based on randomized controlled studies are not perfect in their design and suffer from their own limitations including “1) difficulties in forecasting individual outcomes from observed group effects; 2) the fact that negative outcomes are underassessed and underrepresented; and 3) the high degree of variability in how individuals value and weigh various positive and negative outcomes.”
He does not say this in a policy journal or in a popular magazine. He makes these statements in a journal that only a statistician can really love. The science of medical decision making and the articles written in this journal are more geared to statistical researchers than to clinicians. In this same issue, there is an article entitled, “Development of a Framework for Cohort Simulation in Cost-Effectiveness Analyses Using a Multistep Ordinary Differential Equation Solver Algorithm in R”. For the journal Medical Decision Making, that is a relatively common type of title. I mention that only to make the point that Dr. Barrett is making these pronouncements as a way to improve analytics and as someone who believes in scientific and even mathematical approaches rather than as a moral argument.
However he does realize that there is a moral dimension to his argument. After making a convincing case that “Patient-oriented evidence that matters (POEM) is superior to disease-oriented evidence (DOE), such as biomarkers or surrogate markers” on analytic grounds he goes on to say,
“Given this background, I would hazard the contention that the current system may to some extent be both unethical and irrational. Difficult questions must be asked: Is it rational to implicitly value benefits more than comparable harms? Should we continue to design and interpret trials based on benefit effect size only, ignoring harms? Is it ethical to take decision making away from individuals (guided by their clinicians) and to instead give that power to medical scientists, insurance companies, technocrats, and policy makers who set guidelines and formularies that determine care? I don’t believe that any of these questions should be answered in the affirmative.”
Thus he defines the challenge to the good, ethical clinical researcher and to the practitioner trying to follow evidence based medicine and best medical care. The challenge is at the least, to interpret randomized controlled studies with outcomes that are defined by patient values not only population metrics. It may mean moving to a whole new paradigm of clinical study that builds upon randomized controlled studies in new ways to measure these important patient-oriented outcomes that matter. He argues that until we understand how individuals value various benefits and harms we cannot really say what the best course is for a patient who is ill. For the practitioner that means knowing who your patient is as a person and not only knowing the biology of their disease. Ultimately, his scholarly analytic approach supports the contention that each individual is unique and autonomous with his or her own values that must be supported and respected.