Esti Noviyanti, S.Kep.
Trainer Insight: Medical VA Insurance Support
This article is contributed by Azmi Tanjung
Health Insurance Claims Specialist | Registered Nurse (RN) | 10+ Years in Medical Claims Analysis, Utilization Review & Fraud Detection | Cashless Claims, Case Management, Claims Adjudication | Remote — Globa
When people hear "insurance fraud," they picture something dramatic — a staged accident, a fabricated hospital stay, an organized ring. In ten years of reviewing medical claims, those cases were the rare exception. Most of what erodes a health insurer's integrity is far quieter: small mismatches, padded bills, and pre-existing conditions that quietly went undeclared. The damage is real, but it rarely announces itself.
Here is what a decade of reading claims — first at a hospital helpdesk, then as a claim analyst — taught me about spotting it.
The clearest signal is mismatch
If I had to teach one thing to a new reviewer, it would be a single question: does the treatment fit the diagnosis?
A patient admitted for dengue does not usually need a thyroid panel. When a request for a test like TSH or FT4 turns up on a dengue claim, that is not fraud by itself — but it is a flag. Either the documentation is incomplete, or someone is rounding out the bill with services the diagnosis doesn't justify. Most of the leakage I saw was not invented illness; it was real admissions with extra items bolted on. Train yourself to notice what doesn't belong, and you catch the majority of it.
Timing tells a story
Waiting periods exist for a reason, and that reason is human nature. A cancer that surfaces within weeks of a policy being issued, a chronic condition like diabetes or hypertension diagnosed inside the first year, a serious illness appearing right as a waiting period closes — these patterns repeat because conditions that "appear" the moment coverage starts were often present long before it.
This is not about assuming bad faith. It is about recognizing that the timing of a diagnosis is itself information. A claim that lands suspiciously early deserves a closer look at when the condition first began, not just when it was reported.
Watch the detail that doesn't belong
Some of the most useful findings were incidental. A patient comes in with a fever, but the admission workup shows a blood pressure of 180/100 and elevated blood sugar. The fever may be entirely real — yet those readings hint at an undisclosed chronic condition sitting quietly behind an acute claim.
Clinical training is what makes this visible. A checklist sees "fever, covered." A clinician sees a number that doesn't fit the picture and asks why. That instinct — pulling the thread on the one data point that's out of place — is where a nurse's background becomes an investigative advantage.
Validate the narrative, not just the numbers
For accident claims especially, the story has to hold together. A fracture from slipping in the bathroom, or from a motorcycle fall where the rider had a valid license and wore a helmet, fits a clean clinical picture. When the mechanism of injury, the radiology, and the rest of the examination don't line up, that gap is exactly where you investigate.
Fraud detection is as much about coherence as it is about figures. If the clinical evidence, the chronology, and the documentation tell three different stories, one of them is wrong.
Declining is not the goal — verifying is
This is the lesson I most want honest reviewers to hear: an anomaly is a reason to verify, not to reject. The large majority of unusual claims turn out to be legitimate once you have the full picture.
So when something looked off, the right move was rarely a flat denial. It was to request the medical records, ask for an attending physician statement, or run a deeper investigation — and then to reopen or refund the moment the evidence cleared the patient. Protecting the fund and treating policyholders fairly are not opposing goals. Done well, fraud detection defends the honest majority, because every rupiah lost to leakage is a rupiah that should have reached a real claim.
The real lesson
Ten years in, I stopped thinking of fraud detection as catching bad actors. It is closer to disciplined pattern recognition layered on top of clinical knowledge: does the treatment match the diagnosis, does the timing make sense, does the story cohere — and have I verified before I decided?
The rules engines and AI tools we use are excellent at flagging the obvious. But the subtle cases, the ones that matter most, still come down to a trained eye asking a simple question: does this make medical sense?
I'm always glad to compare notes with others working in claims, utilization review, and medical underwriting. The patterns are remarkably consistent across markets — and the clinical lens travels well.
Baca Selengkapnya
