When comparing the breakouts, there are notable differences in RHO vs. PIA misclassification severities. Keep in mind that the vast majority of the historical PIA data reflected in this study comes from APS data.

1. RHOs have more reverse misclassification (9x), contributing to more negative slippage, which lowers the overall RHO slippage results.
2. RHOs have more low-severity misclassifications of 1-2 preferred class differences compared to PIA audits. These smaller misclassifications have less of an impact on overall slippage.
3. RHOs are more effective at uncovering tobacco misrepresentation versus PIAs.
4. PIAs have a higher prevalence of severely misclassified cases (rated, postpones, and declines) compared to RHOs, which are most impactful on mortality slippage.

Why do these two monitoring methods have such a different makeup of misclassified cases? This can be explained by the monitoring process itself, as well as the additional data being evaluated in the audit process.

1. Monitoring process: RHOs are conducted pre-issue and once selected, an applicant must be willing to participate in the paramedical exam and insurance labs. If the applicant is knowingly misrepresenting their health status, it is more likely that they will withdraw from the process before a severe misrepresentation can be discovered. PIAs are typically designed to look for misrepresentation or reasons to adversely change an underwriting assessment, whereas RHOs are intended to target the FUW risk class decision, explaining the higher prevalence of low severity and reverse misclassifications.
2. Data being evaluated: APS audit assessments are intended to serve as a proxy for FUW, but are not an apples-to-apples comparison with the traditional insurance labs and paramedical exams used in RHOs. Medical records can uncover different conditions, as evidenced by the provided reasons for misclassification.

• Build/BMI misclassification is the leading misclassification reason and is equally prevalent in both monitoring methods. Despite being the most common, it is more likely to lead to many of the preferred classification differences and not have a significant impact on overall slippage on a per-case basis. 
• Tobacco use is the second leading misclassification reason uncovered in RHOs, and the fourth leading reason in PIAs. Routine cotinine testing in insurance labs (although not foolproof) is more effective at uncovering tobacco use than relying on an APS, as clinical labs do not routinely test for cotinine and many physicians do not document (or may be unaware of) their patients’ tobacco use. As demonstrated in our prior study, tobacco misrepresentation has been trending up over time as fluidless underwriting programs gain awareness in the industry.
• Cholesterol/lipids is another leading reason that contributes to primarily 1-2 class differences and little to no severe misclassification on its own. These are uncovered in random holdouts at twice the rate of post-issue audits, as recent clinical laboratory results are not always available in APS records depending on the specialty of the physician listed in the application.
• Blood pressure is a common misclassification reason in both monitoring types that contributes to a wide range of severities from preferred class differences to declines. Contrary to cholesterol, high blood pressure is uncovered in PIAs more frequently than RHOs, likely stemming from the presence of elevated blood pressure readings as well as hypertension diagnoses indicated in the medical records.
• Undisclosed medical history and substance use are more often uncovered in medical records (e.g., EHRs) than in insurance labs. The rise in substance abuse deaths among college-educated individuals is of particular concern to the life insurance industry. We most frequently see undisclosed medical history and substance abuse attributed to decline or postpone audit decisions.

While the majority of the historical post-issue audit data in our study is based on APS data, we are seeing more carriers adopt electronic health records (EHRs) in both front and back-end underwriting. Although APS and EHR records can contain similar information, an APS is typically a targeted query to a specific physician while EHR queries return data from any medical provider(s) participating in the insurance use case. Therefore, EHRs may uncover additional medical history from other healthcare providers not disclosed by the applicant. Given the undisclosed medical history seen to date in post-issue audit results, the obvious question becomes, “Why not eliminate AUW slippage altogether by using the same tools for underwriting applicants on the front end?”  While the lag time in receiving APSs defeats a primary purpose of accelerated underwriting – speed to issue – EHRs can now provide carriers with similar comprehensive medical information on their applicants in near real time.

Munich Re Life US recently completed a series of comprehensive EHR retro studies that evaluated the protective and operational value of EHRs in various underwriting scenarios, including an accelerated underwriting environment. These studies included a sample of over 800 underwriting files consisting of business from multiple carriers across diverse distribution channels. We tasked underwriters with making risk assessments based on varying levels of underwriting evidence in the file and documented any differences in assessments when an EHR was added. We found that adding EHRs to AUW at the point of sale uncovered additional medical information, enabling more accurate risk class assessments and preventing potential mortality slippage. In fact, the results suggest that it may be possible to get back to fully underwritten mortality by adding EHRs to accelerated underwriting evidence or replacing fluids and parameds with EHRs in fully underwritten environments.  When factoring in the average cost of EHRs per policy while accounting for the cost of obtaining multiple EHR data sources per life, the protective value of the EHRs far outweighed the cost and provided a clear net positive financial impact.

In addition, we found that EHRs uncovered similar medical histories as APSs; however, the prevalence and impact differed. In particular, build/weight changes were the top reason for variances in risk class assessments, followed by medical histories, including diabetes, and substance use. Similar to what we observe with APSs, medical histories found in EHRs had a higher impact on mortality savings. On the other hand, while they were lower in prevalence, EHR lab results often indicated a higher mortality impact than what we typically see in PIA reviews with APSs. EHRs also have the potential to uncover additional tobacco misrepresentation, as we observed that 76% of EHRs in the study contain tobacco use details.

On top of providing protective value, our retro study found that EHRs may add operational value by driving a net increase in the number of risk class decisions that can be made without requesting traditional underwriting evidence, as demonstrated in Table 2.