The Pernicious Fiction of Accurate Medical Information
Lately, in the life insurance world, the buzz is that electronic health care data will revolutionize underwriting. Someday...maybe, but not today. Why? Because much of what’s taking place now is the “digitization” of existing information, that is the transfer of information in documents into data, or bits and bytes, of information stored in databases and other software formats. However, turning information into data does nothing to address the inaccuracies in the data itself, and that is definitely not happening – at least not in any way that is apparent to an underwriting business.
Errors with Electronic Health Records
It will come as no surprise to life expectancy underwriters, that medical records, including digitized records, are literally rife with errors of all sorts. Furthermore, some of these errors are literally inscrutable gibberish that cannot be assessed by anyone (or anything for you AI mavens out there) with any degree of certitude. Bad, erroneous, conflicting, inaccurate, misplaced information is everywhere in health care records, and it’s being transformed into bad data, plain and simple. Take for example, this information from just one case: “Tobacco non User, Current non-smoker. Social History: 7.5 pkyr, no smoker since 1985; drinks socially. One cigar per day to stop after he passed out with his. Cigars 2-3 per week. 2010 actually 7-10 per week. 2011 not smoking cigars....” Later in other records from the same Health Care Provider (HCP) there is this: “Smoked Tobacco use: Current someday smoker” and “Pipes: Yes”.
These statements were made by the same HCP about the same insured (i.e., patient) within the same year. Now imagine this degree of variability and imprecision multiplied over years and throughout hundreds or thousands of pages of medical records – and this just goes to whether the insured smokes, nothing more complex than that. Will the transformation of this obviously conflicting information into bits and bytes improve the correctness of this information? Nope, not at all. At the very least, the information is confusing and, at its worst, it is so contradictory as to be useless. Yet this is what life expectancy underwriters see day in and day out.
HCPs are doing a poor job at best of recording correct, clear, precise information, regardless of how that information is stored, and yet, the Insurtech world is all aflutter over the digital transformations it claims will improve outcomes tomorrow. Again, eventually, this may be the case (we certainly hope so), but today the information derived from this data in virtually printed formats (e.g., PDF pages numbering in the millions each year) is simply festering with errors that have to be evaluated and reconciled before it can be used to determine a life expectancy.
Trust but Verify
What is needed is a focus on verification of the information regardless of its form, which as we've discussed previously is the second step in our DIKA process toward knowledge which can serve as the basis for action (i.e., decision-making in the case of life expectancy assessment). Verification of this sort requires judgment and judgment, good judgment at least, comes from experience, expertise and real intelligence.
Once the information is verified through a thorough, consistent, and objective underwriting process, then and only then can it become knowledge and thereby made actionable. As the old adage holds: “Trust, but verify.” When it comes to health care data, digital or paper, “traditional” or electronic, form is still far less important than substance, and substance must be viewed with skilled and diligent caution. To learn more about the impact of medical information accuracy in life expectancy assessments, contact ISC Services.