Recently I was asked to think about the ethical aspects of health information technology (HIT), which I confess, is not a subject I had given much thought to before. After all, what could be unethical about such an obvious example of manifest destiny?
Of course, it turns out there are now a few ethical issues related to HIT and associated risks for unethical behavior, and I subsequently decided to examine the issue from the traditional ethical perspectives of beneficence, avoidance of malfeasance, autonomy, and justice.
Unfortunately, those interested in hearing further will need to wait for the 2013 ASCO Annual Meeting Educational Session. But, I will offer three observations that seem to me to be unique and whose impact I am still pondering.
The first is the validity of the separation of clinical research and routine clinical care. Most of biomedical ethics that resides in the consciousness of oncologists relates to our training in the conduct of clinical trials, and particularly with obtaining informed consent. Routine clinical care that is not part of a clinical trial does not have active oversight for ethical behavior and generally is assumed to be ethical if it follows standards of care. However, this bright line of distinction may be blurred by newer clinical research activities such as comparative effectiveness research and by data mining.
Second, the increasing knowledge about patient genetic material, even to the extent of whole gene sequencing, complicates the de-identification of patient data, which is one of the foundations upon which data aggregation without obtaining patient consent is based. The HIPAA Privacy Rule regarding research is based on presumption that de-identification can be achieved at a level where it is statistically unlikely that the patient can be re-identified.
Third is the concept of data ownership—and whether we should forgo it and substitute it instead with a more limited concept of data rights. Ownership seems to imply sole possession of an object as well as control over that possession. Health data that resides in a large data farm accessible only though proprietary software does not seem to be something that a patient owns. Electronic Health Records can easily cost $100,000,000 plus the expense of health care providers to document and enter that data, a cost that the patient did not bear. Although raw data has intrinsic value, that value is not realized without significant investment of time and effort that will increasingly be dependent upon machine learning such as IBM's Watson or ASCO's CancerLinQ™
. The world as it now exists has not yet established ethical constraints on the aggregation, use, and monetization of health information.
I would argue that the sooner we begin a dialogue on these three issues the better off we will be. Otherwise, we risk slowing the pace by which rapid learning can transform healthcare, once these issues surface on their own volition.