In our last article, we stated that health care data is not a solution to the need for complete, correct medical information when estimating life expectancy. This does not mean health care data cannot be useful to some degree, as a means to filtering out unqualified insureds or identifying potential health issues that may prove meaningful in an underwriting context.
There are all kinds of health care data stored in a vast universe of databases and systems, and the U.S. health care industry continues to collect more and more of this raw information. All sorts of companies have this data, and many of them sell it to aggregators, known as Health Information Exchanges (HIEs), who then resell for all sorts of reasons.
Complexities of Gathering Health Care Data
The networks that hold health care data are numerous, complex, and in many cases, proprietary and disconnected from each other. It is not uncommon to request a health data report that has significant gaps in the data provided. This can be caused by a number of issues, including the provider simply not having access to a source for some kinds of data, or an inability of the source’s system and the HIE’s system to exchange some kinds of data. In other words, you can often get a few pieces of the puzzle that give you a sense of what the insured’s health care “picture” looks like, but it’s not unusual for many pieces to be missing.
Keep in mind that “raw” data is not information. It is the raw material from which information is derived. Through various forms of processing, like formatting, organization, filtering, sorting, cross-referencing and other techniques, data can be “normalized” to make it more useful. However, even normalized data, cannot provide the context necessary for life expectancy underwriting.
Piecing Together the Data Puzzle
So, what can be done with some forms of normalized data? Well, imagine you have a list of medications prescribed for the insured over a period of time. You cannot know if the insured took the medication as prescribed, or if they took it at all (a concept known as “medication compliance”), but you can assume they are being advised to take it. If you then assume they are like most people and they do as directed by their health care providers, you may be able to infer things about their health from the medications you assume they are taking.
For example, many medications have a single or at least primary use, to treat heart disease for example. If an insured is taking statins, and the data indicates they have been taking them for some time, at some level, you might infer that they have high cholesterol. Similarly, other medications may suggest other health conditions that could be meaningful from a life expectancy underwriting perspective.
Finding the Meaningful Data for Life Underwriting
How can such inferences be useful? Using some forms of normalized health care data, a properly trained underwriter can try to get a sense of the health conditions that could be affecting the insured. In turn, underwriters can focus on where and from whom to get more complete information about the insured to further focus on developing an accurate assessment of their life expectancy. In short, the clues found in some normalized health care data may be helpful to the process of filtering out the unqualified, and focusing on selected insureds; however, it is not a proxy for the contextual information found in comprehensive medical records that is critical to assessing micro-longevity risk.
If you’re interested in learning how ISC Services develops accurate life expectancy assessments with decades of medical records and data from all over, contact us today. You can also download a free sample life expectancy report.