Fluidless Underwriting Solution: How to Handle Borderline Risk Cases

Every fluidless underwriting solution eventually runs into the same problem: what happens when an applicant doesn't clearly fall into standard or decline? The borderline risk case--the 45-year-old with controlled hypertension and a family history of cardiac events, or the applicant whose prescription history suggests a condition they didn't disclose--is where most accelerated programs either break down or quietly revert to traditional requirements. For chief underwriting officers and actuarial teams designing or expanding fluidless programs, these borderline cases aren't edge cases. They represent somewhere between 15% and 30% of all applications, depending on the program's eligibility criteria and triage thresholds.

Munich Re's retrospective study of 525 accelerated underwriting cases found that incorporating electronic health records increased decision rates from 68% to 79%, with roughly 12% of cases receiving changed underwriting decisions based on medical information the applicant had not disclosed.

How borderline risk cases emerge in fluidless programs

The fundamental tension in fluidless underwriting is straightforward: you're removing the most objective data source (blood and urine testing) while trying to maintain mortality outcomes comparable to fully underwritten business. For applicants who clearly qualify--young, healthy, no red flags in MIB, prescription history, or motor vehicle records--the program works well. For applicants who clearly don't qualify--active cancer treatment, recent cardiac events--knock-out rules handle the triage. The borderline population sits between these two groups.

Several data patterns create borderline cases in practice:

• Prescription histories that suggest a condition without confirming it. An applicant taking metformin could have Type 2 diabetes or could be using it off-label for weight management. Without labs, the underwriter can't distinguish between these scenarios.
• Conflicting signals across data sources. Clean MIB results paired with a prescription history showing three cardiovascular medications. Or a favorable credit-based mortality score combined with an unfavorable medical claims history.
• Age and face amount combinations that push actuarial confidence intervals wider. A $1.5 million policy on a 58-year-old requires more precision than a $250,000 policy on a 32-year-old, and the same data gaps carry different financial consequences.
• BMI readings at the margins of program thresholds. Most accelerated programs set BMI cutoffs, but the applicant at 31.2 versus 30.9 represents an arbitrary boundary that doesn't reflect actual mortality risk with any precision.

The NAIC's 2018 Delphi study on emerging underwriting methodologies noted that triage systems vary considerably in how they handle these ambiguous profiles, with some carriers routing borderline cases through different underwriting experiences based on initial risk signals rather than applying uniform requirements.

Triage architecture: how carriers resolve borderline cases

The most effective fluidless underwriting solutions don't treat triage as binary (accept into accelerated track or reject to full underwriting). They build graduated pathways that match the level of additional evidence to the level of uncertainty in the risk profile.

Triage scoring and threshold design

Munich Re's research on triage impact analyzed a fully underwritten sample of approximately 500,000 lives triaged at various Life and Non-medical Risk Classifier (LNRC) score thresholds. At a score of 600, the risk class distribution above and below the threshold was "extremely similar," meaning the triage boundary wasn't creating meaningful adverse selection. But this finding depends heavily on the scoring model's calibration and the data sources feeding it.

Milliman IntelliScript's work on the Irix Risk Score demonstrated that adding credit data to prescription and medical data increased straight-through processing rates from 71% to 82% while maintaining or improving mortality outcomes for the triaged population. The relative mortality for lives triaged into accelerated underwriting was lower under the combined score model. For borderline cases specifically, the additional data dimension often provides the discriminating signal that prescription data alone cannot.

The graduated pathway approach

Rather than a single pass/fail decision, carriers with mature fluidless programs typically operate three or four tiers:

Triage tier	Population share	Data requirements	Decision speed	Borderline handling
Instant issue	25-35% of applicants	Application + MIB + prescription check + risk score	Minutes	Not applicable--only clearly favorable profiles
Accelerated (light touch)	30-40% of applicants	Above + EHR retrieval + claims data	1-3 days	Minor ambiguities resolved by EHR
Accelerated (enhanced review)	15-25% of applicants	Above + targeted APS or phone interview	3-7 days	Borderline cases requiring one additional data point
Full underwriting referral	10-20% of applicants	Traditional paramedical exam + labs	2-4 weeks	High-uncertainty cases where data gaps are too wide

The Munich Re EHR retrospective study showed that electronic health records resolved a substantial portion of borderline cases without requiring fluid testing. EHRs were able to confirm or rule out conditions that prescription data only hinted at, reducing the "refer to underwriter" population and improving instant decision rates.

EHR as the borderline case resolver

Munich Re's lead EHR study examined whether introducing electronic health records early in the accelerated process--rather than reserving them for post-issue audits--could expand program eligibility. Their findings suggest it can. EHR data provided clinical context that prescription databases lack: actual diagnosis codes, lab values from clinical encounters, physician notes on condition severity and management.

For borderline cases, this clinical context matters enormously. The difference between "taking an ACE inhibitor" (which could indicate anything from mild hypertension to post-MI management) and "taking an ACE inhibitor, last blood pressure reading 128/82, no cardiac history per physician notes" is the difference between a referral to full underwriting and a same-day decision.

EHR hit rates remain a practical consideration, though. Not every applicant has accessible electronic records, and hit rates vary by geography, age, and healthcare system participation. Munich Re's research noted that many carriers still start with EHRs in the post-issue audit environment rather than the front-end workflow, partly because of inconsistent hit rates and partly because of cost-per-hit economics.

The role of biometric data in borderline resolution

Traditional fluidless underwriting relies on administrative and claims-based data--information about what an applicant has done (filled prescriptions, filed claims, visited doctors) rather than what their current physiological state actually is. For borderline cases, this backward-looking data has an inherent limitation: it describes historical events, not present-day health.

Biometric data from technologies like remote photoplethysmography (rPPG) introduces a different evidence category. A 30-second camera-based scan can capture heart rate, heart rate variability (HRV), respiratory rate, and blood pressure estimates. A 2025 review published in Frontiers in Digital Health documented rPPG accuracy of 99.1% for heart rate compared to pulse oximetry and a 0.9 Pearson correlation for HRV measurements.

For borderline underwriting cases, these physiological signals serve a specific function: they provide a real-time health snapshot that can confirm or challenge what administrative data suggests. An applicant flagged as borderline due to cardiovascular medication history but showing normal resting heart rate, healthy HRV patterns, and normal blood pressure estimates may warrant a more favorable risk classification than administrative data alone would support. Conversely, an applicant with a clean prescription history but poor cardiovascular biomarkers may warrant additional scrutiny.

The UK Biobank study--a collaboration between RGA and the University of Leicester analyzing 407,569 participants--found that non-traditional biometric factors including resting heart rate "dramatically improved the ability to differentiate mortality and morbidity risks." This is exactly the kind of discriminating power that borderline cases need.

Where biometric data fits in the triage stack

Biometric screening works best as a supplemental data layer rather than a replacement for existing data sources. In the triage architecture described above, it slots into the "accelerated enhanced review" tier--applied selectively to borderline cases where a physiological snapshot could resolve ambiguity without the cost and delay of a full paramedical exam.

The cost calculus supports this targeted approach. A smartphone-based rPPG scan costs a fraction of a paramedical exam and adds seconds rather than weeks to the process. For the 15-25% of applicants sitting in the borderline zone, this represents a potentially significant reduction in both refer-to-underwriter rates and applicant abandonment.

Mortality outcomes: what the data shows so far

The central actuarial question with any fluidless underwriting solution is whether mortality outcomes for the accelerated population match expectations. For borderline cases that are kept in the accelerated track rather than referred to full underwriting, this question is even more pointed.

RGA's analysis of accelerated underwriting programs identified seven factors carriers should evaluate, with mortality experience monitoring being the most critical for long-term program sustainability. The challenge is that mortality studies require years of exposure data before credible conclusions can be drawn, and many expanded fluidless programs are still in their early years.

Munich Re's EHR retrospective offers indirect evidence. By showing that EHR data changed approximately 12% of underwriting decisions in their study sample--and that these changes were directionally correct (identifying risks that applicants had not disclosed)--the research suggests that better data inputs do translate to better risk classification, even without fluid testing.

The Milliman IntelliScript findings on combined scoring models point in the same direction. When multiple data sources are layered--prescription history, medical claims, credit data, and eventually biometric signals--the composite risk picture approaches the discrimination power of traditional underwriting for all but the highest-risk applicants.

Practical implementation: building a borderline case workflow

For carriers designing or refining their fluidless underwriting solution, handling borderline cases requires specific operational decisions:

Define borderline explicitly. Establish quantitative thresholds for what constitutes a borderline case in your triage model. Vague routing rules create inconsistency. Munich Re's LNRC score analysis provides a framework: identify the score ranges where risk class distributions begin to diverge between your accelerated and fully underwritten populations.

Layer data sources progressively. Don't request all available data on every applicant. Start with the cheapest and fastest sources (MIB, prescription check, risk score), and only pull EHRs, APS records, or biometric assessments when the initial data creates ambiguity. This keeps costs proportional to uncertainty.

Set face-amount-specific triage rules. The financial exposure on a borderline $300,000 case is fundamentally different from a borderline $2 million case. Your triage thresholds should reflect this. Some carriers run entirely different accelerated programs for different face amount bands.

Track referral-to-decision conversion rates. If 25% of your applicants are being routed out of the accelerated track, investigate whether all of them truly need full underwriting. High referral rates often indicate triage rules that are too conservative, which harms both cycle time and applicant experience.

Monitor applicant abandonment at each tier. Borderline applicants who get kicked to full underwriting often never complete the process. According to industry data, application abandonment increases substantially with each additional requirement added. If your borderline handling process loses 40% of applicants, you're solving a risk classification problem by creating a distribution problem.

Current research and evidence

The research base for fluidless underwriting in borderline populations is growing but still relatively thin compared to traditional underwriting evidence. The most relevant work includes:

Munich Re's series of EHR retrospective studies (2022-2025) remains the most comprehensive examination of how electronic health records change underwriting decisions in accelerated programs. Their finding that EHRs increased decision rates from 68% to 79% in a 525-case study provides directional guidance, though the sample size limits actuarial credibility.

The NAIC's 2018 Delphi study brought together industry experts to project how emerging underwriting methodologies would affect mortality outcomes. While the study predated many current fluidless programs, its framework for evaluating triage systems and data source combinations remains relevant.

RGA's ongoing work on accelerated underwriting program design and the UK Biobank mortality study (with the University of Leicester) provides evidence that non-traditional data sources can match or exceed traditional biomarkers for mortality prediction.

Milliman IntelliScript's scoring model research demonstrates the incremental value of layering additional data sources, with each new data dimension narrowing the confidence interval on borderline risk classifications.

The Frontiers in Digital Health review (2025) on remote photoplethysmography provides the most current assessment of rPPG measurement accuracy across multiple vital sign parameters, establishing the technical foundation for using camera-based biometrics in underwriting workflows.

The future of borderline risk resolution

The borderline case problem in fluidless underwriting is essentially a data sufficiency problem. Each new data source that enters the underwriting workflow--EHRs, wearable data, rPPG biometrics, continuous health monitoring--narrows the population that truly requires fluid testing. The trajectory is clear: the "borderline" zone will keep shrinking as data layers accumulate.

What's less clear is the timeline. EHR hit rates need to improve before front-end EHR retrieval becomes universal. Biometric data from rPPG and similar technologies needs more actuarial validation studies before reinsurers fully credit it in their mortality assumptions. And regulators--particularly state insurance departments following the NAIC's lead on algorithmic underwriting--will shape what data sources carriers can use and how.

For carriers operating fluidless programs today, the practical path forward is incremental: add data sources selectively for borderline cases, track mortality outcomes rigorously, and adjust triage thresholds as evidence accumulates. The carriers who build flexible triage architectures now--systems that can incorporate new data inputs without wholesale redesign--will be best positioned as the data landscape continues to expand.

Companies like Circadify are working on camera-based biometric solutions that could add a real-time physiological layer to the underwriting data stack, giving carriers one more tool for resolving borderline cases without reverting to traditional exams.

Frequently asked questions

What percentage of fluidless underwriting applicants fall into the borderline category?

Depending on program eligibility criteria and triage model design, between 15% and 30% of applicants typically land in a borderline zone where initial data sources don't provide enough clarity for an automated decision. Munich Re's research suggests that incorporating EHRs can resolve a meaningful portion of these cases, reducing the percentage that requires full underwriting referral.

Can fluidless underwriting maintain mortality outcomes comparable to fully underwritten business?

Early evidence is encouraging but not conclusive. Munich Re's EHR studies and Milliman's scoring model research both show that layered data sources can approximate the discriminating power of traditional underwriting for most applicant profiles. The remaining question is whether this holds for the highest face amounts and oldest age bands, where mortality risk and financial exposure are both elevated.

How do reinsurers view borderline cases handled through accelerated programs?

Reinsurer confidence depends heavily on the carrier's data and monitoring practices. Programs that can demonstrate robust triage rules, ongoing mortality experience tracking, and post-issue audit results tend to receive more favorable reinsurance terms. Munich Re and RGA have both published frameworks for evaluating accelerated programs that carriers can use to align with reinsurer expectations.

What role will biometric data play in future fluidless programs?

Biometric data from technologies like rPPG is expected to serve as a supplemental resolution layer for borderline cases--providing real-time physiological signals that administrative data cannot capture. The technology's value is highest in the borderline population, where a quick health snapshot can confirm or challenge what prescription and claims data suggest. Several fluidless underwriting programs are already exploring these integrations.