Substance Mentions in Claims

Substance keywords appear in roughly 0.8% of the 74,339 claims on Loop's broker book in the latest twelve months — a figure that says almost nothing about how much the workforce actually drinks or smokes. The survey side of this report puts any-alcohol consumption near 37% and nicotine use near 17%. The gap between 37% and 0.8% is the entire reason this page exists as a sidebar. A claim records a substance only when a clinician noted it as material to the admission, or when an insurer queried whether an exclusion clause applied. The vast majority of drinkers and smokers in the covered population never generate a claim where the habit is written down. What surfaces in the claims data is the narrow, expensive, male-skewed tail where substance use turned into a hospital bed.

The claims signal is one part in a hundred of the survey signal#

The claims free-text is not a second, harder measurement of the same thing the survey measures. It is a different thing entirely. The survey asks the whole population a question. The claims data records only the admissions where the answer became relevant to a hospital stay.

Substance-flagged claims cost more than twice the book average#

The few claims that carry a substance mention are expensive ones.

A substance-flagged claim costs, on average, 2.15 times the typical claim on the book. That is not because alcohol or tobacco is itself expensive to treat. It is because the admissions where a clinician thought the substance worth recording are the severe ones — the trauma, the organ damage, the long stays. The mention and the severity travel together.

Trauma is sixteen times over-represented in flagged claims#

The clearest clinical pattern in the flagged group is trauma.

Trauma is 2.2% of all claims on the book. Among substance-flagged claims it is 36.1% — a sixteen-fold concentration, and the dominant category of the flagged group. The underlying illness text confirms what the category implies: head injuries, road accidents, ACL ruptures, humerus and ulna fractures, falls. This is the claims-side signature of alcohol's best-documented acute harm — not the slow organ damage of chronic use, but the sudden injury of a single night.

The gastrointestinal and liver group is the second concentration, at 18.3% against an 11.3% book share. Buried inside it is chronic pancreatitis, which appears repeatedly in the flagged illness text and is, in the absence of gallstones, one of the most specific markers of sustained heavy drinking in clinical medicine.

The flagged claims are overwhelmingly male#

The book is split roughly evenly between male and female claimants. The flagged group is 79.9% male — the sharpest gender skew of any cut in the claims data. By relationship, the flagged group is just over half employees and another fifth fathers; spouses and mothers together account for well under a fifth. This tracks the survey pages precisely: drinking and nicotine use in the covered population skew heavily male, and the severe, hospitalising tail of that behaviour skews more heavily still. The claims data does not contradict the survey here. It is the same population, seen at its most severe edge.

The mention is a query, not a diagnosis#

The most important methodological fact on this page is where the keywords actually live.

Ninety-nine percent of substance keywords appear in the insurer's workflow fields — deficiency, deduction, denial — and barely any in the clinical illness field, the same workflow-versus-diagnosis split the Claims Free-Text page finds when it mines those notes at scale. The substance is rarely the recorded diagnosis. It is the insurer asking a question: was alcohol involved, and if so, does the policy's intoxication or substance-abuse exclusion apply to this claim? The keyword is the paper trail of an adjudication, not the paper trail of a disease.

That distinction matters for how the number is read. The 0.8% of flagged claims are not admissions for alcoholism. They are admissions — mostly trauma — where someone in the claims chain raised the substance question. And in most cases the answer cleared the claim: the large majority of flagged claims were ultimately paid, with only a small minority repudiated outright. The exclusion was checked far more often than it was applied.

Methodology and limits#

The figures above come from a keyword scan across five claim fields — three insurer workflow fields (Deficiency Reason, Deduction Reason, Denial Reason) and two clinical fields (Illness, Ailment Group). Common false positives such as "alcohol swab" and "alcohol prep" were excluded. Each claim is counted once, against the first field in which a keyword appears.