Common Disqualifying PHSP Structures and CRA Audit Patterns

A health and welfare trust for medical spending only keeps its tax-free status if it actually functions as insurance, not as a corporate pass-through for personal medical bills. The tax authority has spent the last two decades sharpening what "insurance" means in this context. The audit record shows a small set of structures failing repeatedly for the same reasons. Knowing those failure patterns before setting up a plan affects whether the plan survives audit in a way that knowing the rules in the abstract does not.

Section 248(1) of the Income Tax Act defines a PHSP as a contract of insurance for hospital or medical expenses, or a medical care plan, or some mix of the two. Provincial and federal government health plans are specifically carved out. The tax authority's own guidance builds a structured test out of that short definition. For a plan to count as "insurance," it needs: an undertaking by one party (the employer or an insurer), to indemnify another party (the employee and dependants), for a consideration (usually salary or services rendered), against a loss tied to a specific event (an eligible medical or hospital expense), where that event's happening is uncertain at the time the plan exists. That last part, the uncertainty, is where most disqualified plans go wrong. CRA also applies what's known as the "all or substantially all" rule: at least 90% of the premiums or benefits paid have to relate to expenses that qualify for the Medical Expense Tax Credit under section 118.2(2). Fall short of that threshold and the whole plan, not just the ineligible piece of it, is at risk.

The Insurance Risk Requirement and Where Plans Fail

CRA's central test asks a blunt question: did the employer take on any real risk, or was reimbursement basically guaranteed from day one? If there's little realistic chance the employee won't eventually get back the full amount set aside for them each year, there's no risk, and without risk there's no insurance.

Two plan features tend to give this away on audit. First, an annual allocation that functions less like coverage and more like a savings account the employee is certain to drain. Second, a plan the employer can cancel or rewrite at any time, with no notice and no binding terms, meant there was never a real commitment to indemnify anyone against anything.

Cost-plus self-insured Health Spending Accounts carry the highest audit risk of any PHSP structure because of this. Under a cost-plus arrangement, a third-party administrator pays out the employee's actual medical expenses, then bills the employer for that amount plus a service fee, typically somewhere between 5% and 15%. When the person filing the claim is also the person who owns and runs the company, the "insurer" is not spreading risk across a pool of people. It's just handing the owner's own money back to them, with an administrative fee stapled on. Under the 90% METC rule, insured PHSPs, where a licensed insurance company actually carries the risk, get treated differently than self-insured cost-plus setups do. That distinction runs through nearly every audit pattern that follows.

Sole shareholder / sole employee corporations: the most litigated disqualifying structure

Picture a small corporation with just one person on payroll, and that person also owns all of the shares. The company sets up a self-insured cost-plus HSA covering that individual, plus their spouse and kids. This is the single most contested PHSP structure in CRA's files, and the agency's position on it has moved steadily in one direction over time.

In a 2006 technical interpretation (2005-0163771E5), CRA allowed this kind of plan, subject to conditions, treating the employment-versus-shareholder question as a factual determination made case by case. By 2014, interpretation 2014-0521301E5 had hardened that stance considerably: a cost-plus plan covering a sole employee-shareholder and their household "would not likely constitute a plan in the nature of insurance," because the structure lacks the basic elements insurance requires. The 2022 CALU Roundtable reaffirmed and sharpened that 2014 position further. CRA's view now is straightforward: a sole employee-shareholder using this structure is paying their own medical bills through the corporation, and no risk is actually changing hands.

There's also a presumption built into how CRA reads these cases. Where someone is both the sole employee and the controlling shareholder, CRA starts from the assumption that any benefit they receive is a shareholder benefit, not an employment benefit. Overcoming that presumption means showing that a non-shareholder employee, doing similar work at a similarly sized company, would get a comparable benefit under a comparable plan. That's a hard bar to clear when there's no comparable employee to point to.

One carve-out applies here: it excludes a genuine one-member plan where the sole employee deals with the company at arm's length and is not also the controlling shareholder. CRA's stated view allows a genuine one-member plan where the sole employee deals with the company at arm's length and is not also the controlling shareholder. The disqualifying factor is the combination of sole employee and controlling owner in the same person. It's the combination of sole employee and controlling owner in the same person.

Shareholder-exclusive plans in multi-employee corporations

The clearest illustration of what happens when a plan tracks ownership instead of employment came out of a case involving one company. v. The Queen. The company's majority shareholder, who held 51% of the shares, ran a claim exceeding $38,000 through the corporation's cost-plus PHSP to cover a knee operation performed in the United States. The plan was not available to others at the company in any meaningful sense. It existed, in practice, for the majority shareholder alone.

The Tax Court found the payment was received in the person's capacity as shareholder, not as employee, because access to the plan tracked equity ownership rather than any defined employment class. The result: $35,996 of the claim was treated as a taxable shareholder benefit under subsection 15(1), taxed as income in the shareholder's hands, and the corporation couldn't deduct the expense because it wasn't incurred to earn business income.

That's the double-taxation trap built into a failed PHSP. The same dollars get taxed twice: once as income to the individual who received the benefit, and again because the corporation loses the deduction it would otherwise have claimed. Nobody comes out ahead.

For shareholder-employees at companies with other staff on payroll, CRA looks for a few conditions before it will treat the arrangement as employment income rather than a shareholder benefit. The shareholder needs to be actively working in the business. The benefits they receive need to be reasonable and roughly in line with what an arm's-length employee in a similar role would get. And there should be a formal employment contract on file. The common thread with the sole-shareholder scenario is the same in both cases: the plan has to be organized around job function, around employee classes, not around who owns what percentage of the company.

Unincorporated businesses and the arm's-length employee requirement

Unincorporated businesses are governed by a separate framework rather than the general corporate deduction rules for PHSPs. They're governed by section 20.01 of the Income Tax Act, a narrower framework with its own requirement: at least one "Qualified Employee." That means an employee dealing at arm's length with the owner and meeting the statute's qualifying conditions.

A sole proprietor with no employees who meet that bar can't run a self-insured cost-plus HSA as a PHSP. CRA's reasoning follows directly from the insurance test: without at least one other person in the plan, there's no pooling of risk, no indemnity relationship, nothing that functions like insurance. Just one person's own expenses running through their own business.

Sole proprietors without a qualifying arm's-length employee aren't entirely locked out. An insured PHSP through a licensed insurance company remains an option, but the contribution limits on that route are tightly capped compared to what an incorporated business with an HSA can offer its employees. The gap in flexibility between the two paths is significant, and it's often the deciding factor in whether incorporating makes sense before setting up a health plan.

Plans without spending limits, formal documentation, or employment contracts

A PHSP needs a written plan document. Without one, CRA has nothing to measure against the insurance test, and no way to confirm the plan was ever structured the way it needs to be structured.

Unlimited reimbursement is its own red flag. If the plan document sets no cap on what an employee can claim in a year, the arrangement starts to look less like insurance and more like an open-ended expense account, because there's no defined boundary around the risk being covered. Insurance, by definition, has limits. A policy with no ceiling isn't really a policy.

The plan document can set different annual limits for different groups of employees, executives getting a higher cap than general staff, for instance, and that's fine. What it can't do is restrict eligibility to shareholders specifically, unless there's a genuine employment-class reason behind the distinction. And where a shareholder is also drawing a salary as an employee, CRA looks for a formal employment contract on file as evidence the benefit is tied to the job, not the ownership stake.

Retroactive plan setup: covering expenses that predate the plan

The uncertainty requirement in the insurance test is a timing rule, and it's absolute. It's a timing rule, and it's absolute: the plan has to exist before the expense happens. A cost that's already been incurred isn't uncertain anymore, so it can't be insured against, no matter how the paperwork gets dated afterward.

The pattern appears often enough in audits to have a shape. A business owner has a major medical expense, surgery, a round of dental work, treatment from a specialist, and only after the bill arrives does the company set up an HSA, or expand an existing one, specifically to absorb that cost.

At the moment that plan gets created, there's no risk left with respect to an expense that already happened. It's just reimbursement of a known personal cost at that point, routed through a corporate structure to get favorable tax treatment on the way out. It's just reimbursement of a known personal cost, routed through a corporate structure to get favorable tax treatment on the way out. CRA's position on this is direct: a plan can't be built after the fact and applied backward to cover an event that already occurred. HSA status runs from the date the plan is actually and formally in place.

Ineligible expenses and the 90% METC threshold

The 90% rule isn't a rough guideline, it's the line CRA draws between a plan that qualifies and one that doesn't. At least 90% of premiums paid (for insured plans) or benefits paid out (for self-insured plans) need to relate to expenses eligible for the METC under section 118.2(2). If the share drops below that threshold, the entire plan is at risk of losing PHSP status, not just the specific claims that shouldn't have been covered.

That "entire plan" consequence is what makes ineligible expenses so costly. A handful of claims for things that never should have been covered can taint the whole arrangement, because CRA doesn't isolate the bad claims from the good ones when it assesses whether the plan itself qualifies.

Some categories keep appearing in the data as the source of trouble:

• Over-the-counter vitamins and supplements, which don't qualify even when a doctor recommends them
• Basic reading glasses bought without a formal vision exam behind the prescription
• Gym memberships and general personal development costs
• Cosmetic or lifestyle expenses with no underlying medical basis

Prescription drugs sit in their own category. To qualify, a medication has to meet the eligibility criteria set out under the METC rules. Having a prescription attached to something doesn't automatically make it eligible on its own, particularly for supplements that fall outside what those rules actually cover.

Self-adjudication and the absence of third-party claims oversight

Arm's-length adjudication means the person deciding whether a claim gets paid isn't the same person who incurred the expense, and isn't under the direct control of that person either. CRA expects a real separation between the employee filing a claim and whoever is reviewing it.

That separation breaks down fast in small, owner-run plans. When a shareholder-owner is the one reviewing and approving their own medical claims, with no independent administrator or claims system making an impartial call, the arrangement stops resembling insurance. It starts to look like what it functionally is: personal expense funding, moved through a corporate account, with the owner signing off on their own reimbursement. That absence of independent oversight is, on its own, enough to raise the same question every other structure in this piece runs into. Was there ever any actual risk here, or just money moving in a circle back to the person who put it in?