The cost shows up in productivity, not just claims
The largest share of substance-related cost to employers isn't medical, it's lost work. CDC researchers estimated in December 2025 that substance use disorders cost the U.S. economy roughly $93 billion a year in lost productivity, with the biggest pieces coming from inability to work and absenteeism. The research team noted that productivity losses are routinely underestimated next to medical costs, which is exactly the line item HR and benefits leaders tend to overlook.
The medical cost is real too, and it lands squarely on employer plans. A study in JAMA Network Open found that substance use disorders cost employer-sponsored insurance $35.3 billion, about $15,640 per affected enrollee, with alcohol-related disorders the single costliest category. For a workforce quietly self-medicating with after-work drinks, the alcohol breakout is the number that should get attention.
Drill into alcohol specifically and the per-person cost is stark. Avidon's analysis of national cost data puts excessive alcohol use at up to $5,106 in healthcare costs and $3,247 in lost productivity per affected employee each year, with roughly one in nine workers drinking at risky or excessive levels. For a workforce quietly self-medicating after-hours, that's not a rounding error; it's a line item hiding in plain sight.
Job insecurity is a documented, rising stressor
People don't reach for unhealthy coping in a vacuum. Stress drives it, and right now job security is one of the loudest stressors in the workplace. The American Psychological Association's 2025 Work in America survey found that 54% of workers say job insecurity significantly raises their stress.
That anxiety is increasingly pointed at automation. Gallup's February 2026 survey of more than 23,000 workers found that 18% believe their own job is very or somewhat likely to be eliminated by AI or automation within five years, up from 15% the year before. Pew Research found a broader unease, with 52% of U.S. workers worried about future AI use in the workplace.
The magnitude varies by how you ask the question, but the direction is consistent: more workers are carrying low-grade fear about their livelihood into the workday, and that AI anxiety is reshaping workplace mental health. Chronic stress without a healthy outlet tends to find an unhealthy one.
What workers reach for to cope
Here's where the picture gets uncomfortable. A 2026 survey by Modern Health of 1,000 full-time employees found that 63% reported using alcohol, THC, or unprescribed medications after work to cope, and 52% reported doing so during the workday. The same survey found that 69% expected AI-driven layoffs at their company within three years.
Those numbers are striking, and they come with an important caveat. This is a single survey from a workplace mental-health vendor, and Modern Health's own Chief Clinical Officer cautioned that the data can't identify ongoing patterns of use, only that respondents report using substances to cope. In other words, it's a real signal worth paying attention to, not a diagnosed-disorder rate, and not proof that AI fear causes substance use.
The honest read: stress and job insecurity are high, a meaningful share of workers say they're self-medicating, and the cost of that coping eventually shows up on the employer's books. You don't need a perfect causal chain to see why substance use in the workplace matters to a benefits buyer.
Why the default fix usually misses
When an employer notices stressed, struggling employees, the standard move is to point them toward the Employee Assistance Program. The trouble is that almost no one walks through that door. Industry data puts average EAP utilization at roughly 4 to 5% of eligible employees, held down by stigma, low awareness, confidentiality fears, and the friction of having to self-refer.
The effectiveness evidence is thin on top of the engagement problem. An NCBI review noted that definitive studies of EAP effectiveness have yet to be conducted. That's not the same as saying EAPs don't work, but it does mean the most common response to this problem is a low-engagement referral whose impact was never well established. A resource 95% of employees never use can't change anyone's behavior.
This is the real gap. It usually isn't a lack of coverage; most workers report they have adequate benefits. It's a lack of support that's engaging enough to actually shift what people do day to day.
It's worth being honest about why this is easy to get wrong. Plenty of employers have already bought something that promised to engage employees, whether an app, a portal, or a points program, and watched it gather dust the same way the EAP does. The lesson isn't that wellness support doesn't work. It's that two things have to be true at once: the support has to be built to be used, and it has to be built on an approach that actually changes behavior. Most tools deliver one or neither.
What does change behavior: structured, evidence-backed coaching
The approaches that actually move the needle on unhealthy drinking are built around active behavior change, not a passive referral. Avidon's own outcomes bear this out. Across an alcohol behavior-change program with more than 2,000 participants, 93% met their alcohol-related goals and 52% reduced their drinking at six-month follow-up. A separate stress program with over 7,500 participants saw 96% say it helped them handle stress more effectively, and 47% report lower stress levels. These aren't lab conditions; they're real employees working through digital, self-paced programs.
The reason it works comes down to engagement, and there's a clean number for it. In a controlled study of 300 non-incentivized participants, Avidon compared three groups: no coaching, live coaching alone, and live coaching paired with its digital platform. Completion ran 17% with no coaching, 28% with coaching alone, and 36% with coaching plus technology, a 112% improvement over no coaching. Set that against the 4 to 5% of employees who ever touch an EAP, and the gap isn't subtle. People engage with support that's built to be used.
Coaching and behavior-change tools work because people actually use them, repeatedly, with feedback that adapts to where they are. That's the opposite of a referral that sits unopened.
That difference is easier to see in the numbers than to argue in the abstract. The challenge was seven brief two-minute quizzes delivered by email with no incentive attached. The format was low-stakes enough that people actually used it.
And what some of them took away maps directly onto the problem this piece started with. One participant described the stress portion this way: "Not only did it bring to light the unhelpful things I do to cope with stress, it showed me replacement behaviors that are healthier for me." That's the whole point in a single sentence: not a referral that sits unopened, but a small, repeated touchpoint that helps someone trade an unhealthy coping habit for a better one. For the drinking piece specifically, there's a self-guided resource you can share with employees who want to cut back without quitting.
For employers watching stress and self-medicating climb, the more useful question isn't whether you offer support, it's whether your support is built to be used.
