The human risks of bias in medical and rehabilitation research and practice: the eight Is

Antti Malmivaara

The human mind is liable to make biased interpretations. The biomedical paradigm supposes that an objective truth can be reached with rigorous scientific methods. Human risk of bias is defined in this paper as threats to validity of study results that cannot be controlled even by rigorous scientific methods. To identify categories of potential human risks of biases in quantitative medical and rehabilitation practice and research; and to review the available evidence of human risks of biases in each of these categories. A narrative review. Eight potential categories of human risk of bias (HRoB) in clinical and rehabilitation practice and research were identified: identity, integrity, independence, intelligence, ideology, interest, incentive and inequity. There is evidence that conflict of interest, deficient integrity and inequity contribute to biased scientific interpretations; and that due to inequity the evidence of effectiveness of interventions for disadvantaged patient groups is scarce. There is also evidence that biased decisions can occur in health care practice due to the state of inequity. As all categories of HRoB's start with the letter 'I', these are named the 8 I's. The categories overlap with each other. Studies assessing the impacts of biases related to conflict of interests, deficient integrity and inequity were identified. However, other potential human risks of biases have been studied very little or not at all. The human beings' liability to biased thinking and all categories of HRoB should be recognized in clinical and rehabilitation practice and research, and use the best verified means to reduce the HRoB. More research is needed, particularly on how to reduce HRoB in medical practice and quantitative clinical research. These actions should be considered congruent to those aiming to increase the validity of the scientific method. The human risk of bias should be considered in assessment of evidence from clinical epidemiology as well as in clinical praxis.