People are more likely to avoid loss than to seek gains. HIPAA creates a framework where it rewards risk adverse behavior for data sharing even when data sharing would ultimately be beneficial to the enterprise, the mission, and the patients. This is a general issue at the heart of making progress in health care regarding data sharing and interoperability. I have some new thoughts on how to bridge this divide.
Recently I read the book ‘Thinking, Fast and Slow’ by the Nobel Prize winning economist Daniel Kahneman. This book discusses the concept of Prospect Theory. In reading through it I could see a hint of why our industry has so much trouble trying to share medical records and in general has trouble sharing almost anything among trading partners and competitors. If you haven’t read about Prospect Theory, the following tests provide some of the basics into how humans make decisions about risk.
Decision 1: Which do you choose? Get $900 for sure OR 90% chance to get $1,000
Decision 2: Which do you choose? Lose $900 for sure OR 90% chance to lose $1,000″1
The common answer to #1 is to take the $900. The common answer to #2 is to take the 90% chance to avoid the loss. As a result, we take risks to avoid danger but avoid risks when we see certain rewards. This behavior is relevant to data sharing and access to PHI and can be instructive on how people will approach risk.
Kahneman defines this approach:
Category 1: In mixed gambles, where both a gain and a loss are possible, loss aversion causes extremely risk-averse choices.
Category 2: In bad choices, where a sure loss is compared to a larger loss that is merely probable, diminishing sensitivity causes risk seeking.2
Here is the challenge. Sharing data is squarely in category 1; a mixed gamble for any covered entity within the current model of HIPAA. The sharing of data may provide a chance at increased benefit, but the existing condition of not sharing allows for gains without the consequence of risks. Furthermore, the risks are difficult to really quantify and can include ransoms, reputation risk, competitive exposure in negotiating payment, and potential unknown fines from regulators. This mixed gamble scenario is accurate for any data sharing that includes PHI either in an interoperability exchange or data sharing agreement for analytics with a partner, trading partner, or market competitor.
Now for the patient who has a severe illness or is in a family with a severe illness, they will almost always fall into the category 2. If sharing their data has a remote chance of improving their prognosis, it isn’t a major gamble, as long as there isn’t a big risk to their livelihood, since they already have a major health condition. So it is no surprise that a patient with ALS would gladly tell everything in an open online community.
But not all patients have a severe illness. Patients with non-life threatening illnesses would be in the mixed gamble category if they were to think of themselves as healthy/well and unlikely to benefit from sharing. So when asked, they will often shy away from it given the unknown risks and almost zero benefit.
So the health system or covered entity is now in a bit of a bind. Some of their patients are likely to be very supportive in the current frame of sharing. In fact, the ones who likely are the highest utilizers will favor open sharing. But others who are the highest volume of patients, are supportive of privacy and low to zero risk. Since democracy tends to rule and the natural inclination in the first place is to not share, it is thus the dominant response and behaviors follow suit. The US government has put some rules in place to try to balance this by calling for interoperability and patient rights to acquire and share their data directly. The groups with data are still begrudgingly complying since they can work to avoid losses but won’t go beyond the letter of a government mandate.
This is a challenging ethical outcome for a dispassionate researcher who sees the mission of the health system to address the unmet needs of the suffering patients. And in the case of not sharing data, it is slow, ineffective, and can cause real harm to patients in need because of the lack of a functional decision by leaders to have integrated data sets for evaluating quality, evidence of outcomes, and patient satisfaction. What it looks like is that without extreme incentives to overcome this bias, the ‘data mining’ philosophy of ‘this data is mine,’ prevails.
If all of this makes sense and we want to get out of the current trap of the data mining problem then we have an urgent need to frame the data sharing and data aggregation debate not in the light of the mixed gamble but in the light of bad choices. This may only be solvable with new legal and regulatory frameworks that generate the switch, since natural market forces don’t seem to have achieved a desirable current state. It also is likely that leaders such as politicians, religious figures, and CEOs would need to help citizens to reframe their point of view on these topics to move from status quo to a model of these concepts where change can succeed.
1Kahneman, Daniel, Thinking Fast and Slow, Farrar, Straus and Giroux, 2011.
This post originally appeared on The Health Care Blog.