Eight per cent of a £24m claims spend identified as potential FWA — and the biggest risk wasn’t fraud. It was chronic over-treatment hiding in mental health.

Case 7 · Employer Level · United Kingdom

WHERE WE STARTED

A large UK employer — a household name — provides private medical insurance to employees through a self-funded arrangement administered by a major insurer and brokered by a leading employee benefits consultancy. Healthcare costs had been rising year on year. The question the employer and their broker wanted answered was specific: how much of that spend was genuinely necessary, and how much might be FWA — fraud, waste or abuse — that had never been identified?

In 2024, Kirontech was brought in to run a deep-dive pilot analysis on the full claims history: January 2016 to mid-June 2025.

THE CHALLENGE

Self-funded employer schemes face a particular blind spot. Unlike insurers, who see claims patterns across thousands of clients, an employer’s benefits team sees only their own population — and only what the administrator chooses to surface. The employer had no independent analytical capability to interrogate what was actually being claimed, by whom, and whether it reflected genuine clinical need.

What had already been tried:

The scheme was managed on a trust basis. Standard claims adjudication processes were in place, and there was no indication anything was systematically wrong. The problem wasn’t that anyone was failing to follow procedure — it was that the procedure had no mechanism to detect the patterns that sit beneath individual claims.

HOW KIRONTECH FOUND IT

HIP ingested the full claims database — over 120,000 lines of information covering approximately 80,000 claims — and ran its full suite of analyses across six finding types: chronic condition patterns, practitioner service utilisation, provider service utilisation, missing preceding services, service conflicts, and subject service replication.

The headline finding was consistent and significant across the entire period:

In every year from 2016 to 2025, potential FWA represented between 5% and 10% of total paid claims — reaching 10% in 2021, 2023 and 2024, the highest years on record.
600+ findings and 2,000+ individual alerts were generated across the dataset, spanning all major condition categories.
Mental health was identified as by far the highest-risk category, accounting for approximately one-third of all FWA findings. Up to 65% of total mental health claims carried potential FWA indicators — meaning one or more signals consistent with over-treatment, billing irregularity or inappropriate care. These are indicators flagged for investigation, not confirmed fraud.
MSK findings revealed a separate pattern: 45 of 120 knee arthroscopies were performed without prior physiotherapy — procedures going ahead without the conservative treatment that clinical guidelines require as a prerequisite.
Practitioner-level analysis flagged outliers billing at double the volume of their peers, with low patient volume but disproportionately high claim counts.
Service conflict analysis uncovered unbundling of procedures — disc removal billed separately from disc replacement, calcium assays billed separately from bone profiles — each a billing practice that inflates cost without clinical justification.

Client challenge

Outcome & impact

No independent analytical lens on claims — the employer relied entirely on the administrator’s own processes. 9.5 years of data had never been reviewed for systemic anomalies at this level of granularity. The mental health claims rise post-pandemic appeared to reflect genuine need; no one had tested whether treatment volumes and durations were clinically appropriate. MSK surgical pathways were being followed without the prior conservative treatment steps that guidelines require.

£1.9m in potential FWA identified across the full dataset period. Six specific case types referred for targeted investigation and claims audit. Mental health pathway and benefit design flagged for urgent review — evidence-based recommendations for session limits, case management protocols and chronic condition handling. MSK pathway redesign recommended to enforce physiotherapy-first protocols before surgical authorisation. Findings to be used as the basis for a formal clinical claims audit prioritising highest-risk areas.

WHAT CHANGED

Before this analysis, the employer and their broker had rising costs and no mechanism to explain them. After it, they had a specific, evidence-based picture of where money was going and why — and a set of recommendations grounded in nine years of their own data.

The mental health finding was the most significant shift in understanding. What had appeared to be a natural post-pandemic increase in employee mental health need was in part something more systematic: chronic conditions being managed through extended therapy without clear clinical justification, diagnoses shifting in ways that extended benefit eligibility, and treatment durations running at double the peer average. The data flagged these patterns for investigation — it did not render a verdict, but it raised questions that the employer is now equipped to pursue.

This engagement also established something more durable: the model for how a self-funded employer scheme should be reviewed. Not through trust, not through administrator-reported metrics, but through independent analytical scrutiny of the full claims history — applied consistently, and acted on.

THE RESULT

9.5 years of claims data analysed across 80,000+ claims and 7,500+ practitioners and providers. £1.9m in potential FWA identified — 8% of total paid claims. Mental health emerged as the primary risk category: up to 65% of post-pandemic mental health claims carrying potential FWA indicators. Findings referred for targeted investigation and claims management action.

Eight per cent of a £24m claims spend identified as potential FWA — and the biggest risk wasn’t fraud. It was chronic over-treatment hiding in mental health.

Expertise and insight

Start your medical claims and payment risk intelligence journey