Healthcare, like finance, is a huge and highly regulated sector of society with many different types of institutions that have to exchange data. Both sectors have defined standards in the pursuit of seamless data exchange, and both now have a modern standard (FHIR in the case of healthcare; FIX in the case of finance) that meets the expectations of today’s programmers.
(To get technical, these expectations involve RESTful interfaces that exchange structured XML or JSON formats over HTTP and that can be programmed in the language of the developer’s choice.)
Finance has reached the promised land of seamless data exchange. But healthcare, as we all know, has not. What are the differences between these two major sectors that might explain healthcare’s lag?
Incentives
Cynics have long claimed that health institutions hold patient data hostage in order to prevent patients from switching to other providers. Finance, in contrast, depends on being able to transfer funds and information with competitors. I am willing to entertain this notion, but I don’t think it’s enough to explain the stubborn resistance of data to sharing in healthcare.
Human variation
Innovation in finance is suspect, but in health care every patient and environment is different. The financiers are free to exclude unconventional forms of money handling. Thus, there is nothing in the FIX standard about halawa. In contrast, doctors are reguarly confronted with unusual symptoms and behaviors.
Human variation provides a strong justification for free-text clinical notes. Doctors struggle to fit a patient’s diagnosis and treatment into the conceptual slots required for reimbursement. Reporting for the purposes of public health and regulatory compliance is yet another burden, lacking even the reward of getting a check in the mail.
Now that the healthcare field is finally confronting behavior change as a goal, it seems impossible to neatly classify clinical interventions. How do classify spending more time with your children?
Lack of hierarchy
Human variety is a challange to classification not just because each person is unique, but because illnesses and health conditions can’t be represented hierarchically. I have explored this problem in the article “International Classification of Diseases hampers the use of analytics to improve health care.”
All the popular technologies I mentioned at the beginning of this article—REST, XML, JSON—assume that information is arranged in a hierarchy or tree-like structure. ICD is also hierarchical in an odd, awkward way.
But healthcare is too complicated for a hierarchy. How do you represent, for instance, that someone’s heart condition was worsened by COVID-19? Diagnostic relationships are better represented by a graph with vertices and connecting lines.
Complexity
A glance over the FIX standard can convince you that it’s huge and complicated. It includes 16 standards. Like finance, healthcare consists of many different institutions doing very different tasks, regulated by many different laws in different jurisdictions. Like FHIR, FIX allows extensions—called specific components—for particular types of messages.
But in some ways, healthcare is even more complex than finance. Think of all the different medical disciplines, each with its own concerns, measurements, and terminology.
In conclusion, the field of healthcare has to do a lot more work than other fields to accurately represent patient information. This is hard to do for an industry that has tended in the past to underplay the role of information. But now that big players from the computer industry have gotten involved, we might be able to make more progress in data sharing.
