Real-time stability for natural health products: what GMP version 4 asks for

A shelf life on a bottle of capsules is a claim: on the last day before expiry, each capsule still contains what the label says. Health Canada’s GMP guide for natural health products, version 4, spells out how that claim must be supported.

The guide was issued on September 4, 2025 and implemented on March 4, 2026 (GUI-0158 v4, cover page). Its stability section runs from page 83 to 91, and this piece goes through it from the lab side.

What every manufacturer and importer must have

The guide is direct (p. 83). Manufacturers and importers must:

• have a completed stability study for each NHP marketed in Canada
• implement and make available a stability protocol for each study
• maintain a stability report summarizing the data, the evaluation and the conclusions

It does not matter who does the testing: the manufacturer, the licence holder, a foreign site or a contract lab. The manufacturer or importer is responsible for the records. For imported products, foreign stability studies can be accepted, but the importer must check the protocol meets Canadian requirements and obtain any missing tests (p. 83).

There must also be a written stability programme procedure covering the initial expiry date, real-time studies, any additional data, protocol design and the report (p. 84).

Accelerated data: a start, not the answer

For a new product, the initial expiry date can be estimated from accelerated or real-time studies, or from similar products in similar packaging (p. 84). The guide then adds a caution:

While accelerated testing generates preliminary shelf-life data in a relatively short timeframe, it may be unreliable when used to assign an expiry date.

It names the products where this is most likely: those that “separate, melt, soften, crack or degrade under high temperatures and humidity”. Gummies, softgels and creams come to mind. And it says plainly: use real-time testing to confirm data from accelerated testing.

On page 85 the requirement is firmer: real-time stability studies must be conducted and used to verify any shelf life that was proposed based on extrapolation, so that each product meets its label claims at expiry when stored as labelled.

The lots and the storage must be real

Real-time studies follow a protocol, and the guide sets conditions on the lots (p. 85):

• same formulation and the same container closure system as the commercial product (or a scientifically demonstrated equivalent)
• the same manufacturing process as commercial lots, meeting the same specifications

Stability chambers or rooms must be monitored for temperature, humidity and, where applicable, light, with records kept (p. 86).

The clock starts at manufacture

This is the detail that catches people. The expiry date is assigned from the date of original batch manufacture, not the date of packaging. The guide’s example: a 24-month shelf life means an expiry 24 months after manufacture, “even if the product was held in bulk for 6 months before being packaged” (pp. 85–86).

It also suggests including batches stored at the limits of extended bulk hold times, for example more than one month, as a worst case (p. 85).

Time points and trends

The regulations do not prescribe stability time points (p. 86). The guide still encourages several, because trends show up before failures: preservative strength, microbial contamination and medicinal ingredient strength can drift toward an out-of-specification result before expiry. Out-of-trend results should be investigated, and an OOS result can lead to a recall (p. 86).

From a chromatography point of view, this is where the method matters. A trend is only visible if the method is precise enough to see it. A potency method with a wide spread will show noise, not a trend, until the product fails.

When the ingredient cannot be assayed

Some NHPs list ingredients “quantified by input”, where there is no assay for the finished product. For those, the guide suggests watching other changes over shelf life, such as colour, odour, viscosity, disintegration and microbial contamination, and recommends a shorter shelf life when the medicinal ingredients cannot be quantified (p. 86). It also asks manufacturers to update the programme when a suitable test method becomes available.

What this means for the testing calendar

1. Every marketed product needs a study and a report, so list your products and check which ones have only accelerated data or data borrowed from a similar product.
2. Plan real-time pulls at time points that let you see a trend, not only at release and expiry.
3. Count from the manufacture date, and include long bulk holds as a worst case.
4. Use stability-indicating methods with enough precision to see drift, and investigate out-of-trend results.
5. Re-evaluate shelf life after significant changes to formulation, process or packaging (p. 87).
6. Consider bracketing or matrixing for multiple package types. The guide points to ICH Q1D, while noting NHPs are outside the scope of ICH guidelines (p. 84).

The specification the stability results are compared with matters too. I wrote about what changed for NHP finished product specifications in June 2026.

Sources

• Health Canada. Good manufacturing practices guide for natural health products (GUI-0158), version 4, issued 4 September 2025, implemented 4 March 2026, pages 83–91 (stability). Replaces version 3.0.

This is a summary of Health Canada’s published guidance, not regulatory advice. I did natural health product QC testing by LC-MS earlier in my career. If you are building a stability programme and want a second opinion on the methods or the time points, tell me the product type on LinkedIn.