Microbial Testing of Botanical Products

Microbial testing: what the label doesn't tell you

A certificate of analysis for a botanical extract can run to three pages and still say nothing about what's living in it. Most CoAs lead with cannabinoid potency (cannabidiol, tetrahydrocannabinol), maybe a terpene panel listing myrcene or beta-caryophyllene, and then a single line at the bottom: "microbial: pass". One word. No method, no limit cited, no reference standard. That tells you almost nothing, and microbial testing is exactly where a vague line does the most damage.

Microbial contamination in botanical products is not a fringe concern. Plants grow in soil. They're harvested, dried, handled, and processed by people and machinery. Every one of those steps is a contamination opportunity. And unlike a synthetic pharmaceutical that starts as a purified precursor in a sterile environment, a botanical extract begins its life as an agricultural product, with all the microbiological complexity that implies.

If you want the wider picture of what a certificate should contain, the library's piece on certificates of analysis and lab testing covers the structure. This article stays with the microbiology.

Why botanicals are higher risk than synthetics

The risk profile of a botanical product depends on several intersecting factors: the plant's growth substrate, post-harvest handling conditions, moisture content at drying, storage humidity, and the processing steps used to produce the final extract or preparation. A weakness at any one of them can undo careful work at the others.

Soil-dwelling fungi, particularly Aspergillus species, colonise plant material readily. Some Aspergillus strains produce aflatoxins, a class of mycotoxins that are acutely hepatotoxic at high exposures and classified as Group 1 human carcinogens by the International Agency for Research on Cancer. Aspergillus fumigatus is a particular concern for immunocompromised patients, because inhaling contaminated material can precipitate invasive aspergillosis. This is not a hypothetical. A 2017 case series published in Clinical Microbiology and Infection documented pulmonary infections in cancer patients that were linked to contaminated cannabis products. The mechanism of harm is microbiological, not phytochemical.

That distinction matters. A product can be accurately labelled for its active constituents and still be unsafe for a vulnerable person, because the hazard comes from organisms or their toxins rather than from the plant's own chemistry.

Bacterial contamination is a parallel issue. Escherichia coli, Salmonella, and coliforms broadly indicate faecal contamination, from water, animal contact, or handling. Some finished botanical preparations are consumed orally; a contaminated extract entering a compromised GI tract is not a trivial risk. Good Manufacturing Practice (GMP) requirements exist precisely because these failure modes are real and recurring.

What the Therapeutic Goods Administration actually requires

In Australia, the TGA sets microbiological limits for therapeutic goods through the Australian Pharmacopoeia and through TGO 92, the Therapeutic Goods (Standard for Tablets, Capsules and Pills) Order; along with the broader Therapeutic Goods Orders applicable to different product categories. For herbal medicinal products specifically, TGO 93 and the TGA's guidance on manufacturing controls under the Therapeutic Goods Act 1989 are the operative frameworks. For a plain-language orientation to the regulator itself, see the TGA explained.

Products listed on the Australian Register of Therapeutic Goods (ARTG) must comply with specified microbiological acceptance criteria. These typically draw from the International Council for Harmonisation (ICH) Q6A guidelines and from the United States Pharmacopeia (USP) chapters <61> and <62>, which Australia's manufacturing standards reference. The limits vary by product category and intended route of administration. An oral herbal liquid has different microbial acceptance criteria to a preparation intended for inhalation, and the latter's limits are substantially stricter, for obvious reasons.

The short version is: if a product is an ARTG-listed therapeutic good manufactured under GMP licence, it has been through microbiological testing with defined method validation. If it's sold as a food, cosmetic, or in an unregistered grey-market channel, different story entirely. How licensing works in practice is set out in the article on TGA Good Manufacturing Practice licensing.

Testing methods: what "pass" actually means

The two foundational methods in microbiological quality control are Total Aerobic Microbial Count (TAMC) and Total Yeast and Mould Count (TYMC). Both are culture-based, meaning a sample is plated on selective or non-selective agar, incubated under defined conditions, and colony-forming units (CFU) are counted after a specified period. TAMC uses Soybean-Casein Digest Agar (SCDA); TYMC uses Sabouraud Dextrose Agar (SDA). Limits are expressed as CFU per gram or per millilitre.

The medium and the incubation conditions are not trivia. Different agars favour different organisms, so a count is only meaningful when the method that produced it is stated. A "pass" without a method is like a distance without a unit.

Beyond those counts, testing for specified organisms (Escherichia coli, Salmonella spp., Pseudomonas aeruginosa, Staphylococcus aureus, bile-tolerant gram-negative bacteria) is standard for oral and inhalable preparations. The specified organisms panel varies depending on product type and regulatory context.

Increasingly, quantitative PCR (qPCR) and next-generation sequencing (NGS) methods have entered quality-control discussions. PCR-based assays are faster than culture (hours versus days) and can detect organisms that are difficult to culture. But they can't distinguish live organisms from dead ones; a sterilised preparation might still return a positive PCR signal from residual DNA. This matters for regulatory purposes, because the question isn't just "is the DNA present?" but "is the organism viable and capable of causing harm?" Culture methods, for all their slowness, remain the regulatory standard because of this distinction. I find that a reasonable pragmatic position, even if it does mean waiting four to five days for a result.

Mycotoxins and the problem of invisible contamination

Mould counts tell you whether fungal growth is occurring, but they don't tell you whether mycotoxins have already been deposited. A product can test negative on TYMC if the mould has been killed by post-harvest processing; but if aflatoxin was produced before that kill step, the toxin persists. This is why mycotoxin testing (typically for aflatoxins B1, B2, G1, G2, and for ochratoxin A) is a separate analytical panel, usually conducted by HPLC or LC-MS/MS. The two panels address different risk windows.

Those instruments are covered elsewhere in the library, in HPLC testing of cannabinoids and mass spectrometry in cannabis testing. The principle is the same whether the target is a cannabinoid or a toxin: separate the compounds, then identify and quantify them against a reference.

The European Pharmacopoeia has long included mycotoxin limits for herbal medicinal products. Australia's TGA guidance aligns with international best practice here, and manufacturers operating under GMP licence who export to Europe must meet those limits explicitly.

Heavy metals, pesticides, and the wider contamination picture

Microbial testing sits within a broader quality-control frame. Botanical raw materials also require testing for heavy metals (lead, arsenic, cadmium, mercury) and pesticide residues. Cannabis-derived products attract particular scrutiny for pesticide carry-through, since extraction can concentrate residues from the plant matrix into the final extract. None of this is microbial analysis strictly speaking, but any honest treatment of botanical product quality has to acknowledge that the contamination risks are multi-dimensional. A product can pass microbial testing and still carry concerning levels of organochlorine pesticides from the growth substrate.

This is one area where I'd push back, mildly, on the way some product quality claims are framed: "third-party tested" as a headline claim means very little without knowing which parameters were actually tested and to what limits. Third-party testing is not a standard; it's a description of who paid for the testing. What matters is the scope of the panel, the validated methods used, and whether the limits applied correspond to a recognised pharmacopoeial standard.

A useful habit is to ask what is missing. If a certificate covers potency and microbes but is silent on metals, pesticides or mycotoxins, the reader has a partial picture and should treat it as one.

How this connects to formulation decisions

Microbial limits constrain formulation choices in real ways. Water activity (aw) is a key variable: products with higher water activity support microbial growth more readily, which is why anhydrous oil-based preparations have lower baseline microbial risk than water-containing emulsions or aqueous extracts. The choice of base matters too; the article on carrier oils in cannabinoid products looks at that side. Preservative selection in multi-dose liquid preparations has to account for both antimicrobial efficacy and regulatory acceptability. Packaging choices (airtight seals, moisture-control sachets, UV-protective glass) all interact with the product's microbial stability profile over its intended shelf life.

That last point links microbiology to time. A preparation that meets its limits on the day of release has not necessarily been shown to meet them later, which is why stability testing and shelf-life studies and shelf life and storage of cannabinoid products belong in the same conversation.

The standardised extract model is partly a quality-control tool. By targeting a defined concentration of key phytoconstituents through validated extraction methods, manufacturers can also apply more consistent processing parameters, including any decontamination steps that affect the microbial load. A well-characterised extraction process with defined kill steps is a more defensible position than a variable post-harvest decontamination applied to inconsistent raw material. The library's article on standardisation of botanical extracts goes further on this.

Reading a microbial claim on a certificate

The gap between what the TGA's published GMP guidance for herbal medicines requires and what appears on a standard market CoA is worth paying attention to. Not every botanical product on the market is manufactured under a TGA GMP licence. Some come through import pathways; others sit in food or supplement categories where the regulatory overlay is thinner. Knowing which category a product falls into tells you a great deal about the standard of microbiological testing that underpins any quality claim.

A few questions sort a meaningful result from a decorative one. Is the method named? Are the limits stated, and do they correspond to a recognised pharmacopoeial standard? Does the panel include specified organisms as well as TAMC and TYMC? Is there a separate mycotoxin result? Does the testing laboratory's scope match what it is reporting? A certificate that answers none of these is asking for trust it hasn't earned.

If you're reading into the broader phytomedicine literature, the microbial quality chapter is one that often gets skipped. It probably shouldn't be. The wider context is in the re-emergence of botanical medicine.

Sources

• Good Manufacturing Practice for herbal medicines; Therapeutic Goods Administration (TGA)
• Contamination of Cannabis by Pathogenic Microorganisms, National Center for Biotechnology Information (NCBI/NIH)
• Microbiological quality of non-sterile products, Therapeutic Goods Administration (TGA)
• WHO guidelines on good agricultural and collection practices (GACP) for medicinal plants; World Health Organization

Tom Whitfield, Health librarian, references & fact-checking