• Allocation of Health Care Resources
• Distributive Justice
• Minorities
• Resource Allocation
• Statistics

R. Scott Braithwaite, Elizabeth R. Stevens and Arthur Caplan argue that some risk stratifications—that is, “employing patient characteristics to reduce the uncertainty that a future event will occur”—amount to profiling and, thus, invidious discrimination.1 These are forms of risk stratification “in which there is concern that ethical harms exceed likely or proven benefits for a group, and in the case of health care, involves any differential treatment in response to a personal characteristic that may cause an unwanted consequence for that person or for other persons with that characteristic”. Braithwaite et al recognise the potential benefits of (increasingly fine-grained) risk stratification: “It can make the provision of therapies safer…[and] improve diagnostic accuracy… Additionally, it can promote the efficient utilization of resources”. However, risk stratification also involves ‘ethical harms’, which must be weighed against the benefits, that is, it can (1) stigmatise groups; (2) violate privacy; (3) increase distributive injustice, for example, by making an already unjustly disadvantaged group suffer further disadvantages relative to a ‘counterfactual situation of no risk stratification’; and (4) imperil autonomy.

Roughly, their position on the moral permissibility of risk stratification is captured by the risk stratification principle:Risk stratification is morally permissible if, …