CFU/g vs MPN/g vs Not Detected in 25 g: Reading Frozen-Produce Test Results

The name of the organism or group is the first field to read. "Aerobic plate count," "Enterobacteriaceae," a named pathogen and a generic "bacteria" label do not describe the same target. An indicator count can help monitor process hygiene or product condition under a defined method. A pathogen detection test asks whether a named organism is found in the specified analytical portion by the method used. Some targets can be both detected qualitatively and enumerated quantitatively, but the tests then answer different questions. Do not compare two numbers or labels just because they appear under a COA heading called "microbiology."

Next read the purpose of the result. A routine supplier trend record, a contractual receiving specification and a legal criterion may use different sample plans and decision language. A result prepared for a bulk ingredient destined for further cooking may not be assessed under the same requirement as a ready-to-eat product. The relevant rule must be checked for the actual matrix, use and destination. The FDA Bacteriological Analytical Manual lists separate methods and chapters, while ISO lists distinct food-chain methods for detection and enumeration. These resources establish why method identity matters; they do not provide a single release rule for every frozen vegetable.

Product form and handling can affect what is sampled. Whole green beans, cut beans and mixed vegetables expose different surfaces and may arrive in different pack sizes. A test portion from one bag can differ from another because microorganisms are not always distributed uniformly. Freezing limits growth while the product remains frozen, but it is not a general kill step that guarantees absence. Sample collection, transport, thawing and homogenization therefore need a laboratory protocol appropriate to the organism and food. The report should connect the result to the correct lot, date, sublot or carton identifier and describe any composite if one was used.

Start with the named target and the purpose of the test.

The line between indicator and pathogen testing should stay clear in buyer communication. A low routine plate count may be favorable evidence about a process under its established baseline, but it does not establish that a specific pathogen was absent. A "not detected" pathogen line does not replace general hygiene monitoring. If a pathogen is detected, stop any casual spreadsheet comparison and involve qualified food-safety personnel under the applicable disposition process. The exact response depends on organism, product, intended use, legal obligations and sampling circumstances. GreenLand-food can provide product and traceability information to support that investigation, but a blog article cannot declare a shipment safe or unsafe from one generic result phrase.

The actual product form belongs in the laboratory request.

For an RFQ, ask whether the buyer wants a process indicator, a release criterion or a verification test for a named hazard. That choice determines the laboratory request. Writing "micro test pass" on a purchase order leaves the target, method and sample plan unresolved. A more useful instruction specifies organism or group, applicable method or validated equivalent, units or qualitative portion, number of samples, acceptance wording and reporting qualifiers. The supplier and laboratory can then confirm what can be tested for the frozen matrix within the required lead time.

CFU stands for colony-forming units. In an appropriate plate-count method, a prepared and diluted food sample is placed on or in growth medium; visible colonies that develop under specified conditions are counted and expressed relative to the original sample mass. A colony may arise from one viable cell or from a clump, so CFU is the method's countable-unit expression, not a direct microscope count of every individual cell in the bag. Incubation conditions, medium, dilution and counting rules are part of the method. The FDA's current BAM aerobic plate-count chapter is one example of a documented route; a COA should identify the method actually used.

MPN means most probable number. It usually starts with multiple test units at defined dilutions. After incubation or another specified detection step, the pattern of positive and negative units is used to estimate the concentration most consistent with that pattern. FDA BAM Appendix 2 gives the statistical route and tables or calculations for serial dilutions. An MPN/g result is therefore an estimate from a particular design, not a direct plate-colony count. Its confidence or uncertainty information can be wide, especially near the ends of the method range. The number of tubes or wells and the dilution series affect the reported estimate and lower or upper reportable bounds.

The word "number" in both results can tempt a buyer to perform arithmetic. Multiplying MPN/g by an assumed factor to obtain CFU/g has no general scientific basis. Even when two methods target the same organism, differences in detection, media, recovery and statistical design can influence results. If a specification names CFU/g and a supplier submits MPN/g, ask whether the method is accepted as equivalent under the agreed requirement and how its uncertainty and decision limits are handled. A qualified laboratory can assess method comparability for that organism and matrix. The buyer should not create a conversion table from unrelated COAs.

CFU follows a defined colony-count method.

Units should remain attached to every value. CFU/g and CFU/25 g are not equivalent report forms simply because both contain "CFU." MPN/g and MPN/100 g differ by a mass basis, but a mathematical scaling of a reported estimate may still be inappropriate if the method and rounding rule are not considered. The test request should name the expected unit and reporting range. If a report uses a different unit, ask the laboratory for a validated expression of the same result where possible. The purpose is to maintain the method's meaning, not to make a spreadsheet look uniform.

MPN uses a positive and negative unit pattern, not a CFU conversion.

For a frozen-bean COA, we would record the exact row label, method, organism group, test portion, dilution scheme where relevant and reportable range. We would compare results over time only when the method and sampling protocol remain suitable and stable. A trend can show a process change worth investigating, but routine counts can vary across samples. The prior GreenLand article on microbiological standards for frozen vegetables discusses broader criteria; this page keeps attention on what the three report expressions mean.

"Less than 10 CFU/g" is a censored quantitative result under a stated method and reporting threshold. It says the laboratory did not obtain a count it could report at or above that threshold by its procedure. It does not mean the measured product has exactly zero colony-forming units per gram. "Less than 3 MPN/g" similarly reflects an MPN method's lower reporting bound or pattern, which depends on its design. Record the less-than sign and number together. Replacing either result with zero in a procurement spreadsheet changes its meaning and can distort averages, trends or compliance decisions.

"Not detected in 25 g" is different in form. It applies to a named target tested qualitatively in a defined 25 g portion using a specified procedure, often including enrichment and confirmation steps. The test result is an outcome under the method's detection capability. It does not say that every 25 g portion of the lot would produce the same outcome, nor does it automatically correspond to "<0.04 CFU/g" or another invented numerical conversion. A negative result has meaning only alongside the sampling plan, matrix, method and number of units tested. The laboratory may use phrases such as "absence in 25 g" under a particular standard; the buyer should preserve the exact wording and method context.

The tested mass is consequential. A report of "not detected in 10 g" cannot be silently changed to "not detected in 25 g." Increasing or combining tested portions may change the detection question, but a buyer should not infer that two separate 10 g negatives equal one 20 g or 25 g result. The laboratory's validated protocol determines how much material enters the test and how it is handled. If a purchase specification requires 25 g, request that exact method and portion before testing. A post hoc change to the COA sentence cannot enlarge what was actually examined.

A less-than count retains its threshold and qualifier.

Qualifiers also require attention. A detected target may be below a quantitative enumeration limit; that is not equivalent to a qualitative negative. A result may be reported as estimated, outside a countable range, or subject to confirmation. For MPN, a pattern at an endpoint may yield a "greater than" or "less than" result rather than a central estimate. Each qualifier should travel with the value into the buyer's evaluation. Software that strips symbols and stores only a number erases information the laboratory deliberately supplied.

A negative outcome applies to the portion and method tested.

When an unfamiliar phrase appears, ask the laboratory for the full report legend and method reference. Its explanation should identify what the finding rules out within the test's validated scope and what it does not. In food control, there is no value in making a negative result sound stronger than the actual tested evidence. The supplier's role is to connect sample identity and process records to the laboratory document, while the responsible food-safety team applies the appropriate criterion.

Consider two fictional COAs for the same type of IQF cut green bean. Report A says "E. coli <10 CFU/g" from a named enumeration method on one composite sample. Report B says "E. coli 3.6 MPN/g" but omits the dilution design and sample code. These invented entries do not prove that one lot is cleaner than the other. The methods use different estimation routes, Report B lacks context, and the samples may not represent the same production unit. A buyer's first request is the full method, target definition, sample plan, date, lot connection, reporting range and any confidence information. Only then can a qualified specialist decide whether comparison is useful.

Add a third line: "Salmonella not detected in 25 g." It cannot be placed in the same numeric column as either E. coli result. It names a different organism and uses a qualitative portion-based outcome. If a dashboard must summarize the records, keep separate fields for target, method, result text, units, portion, qualifiers and decision criterion. A single "microbial score" obscures the questions the tests answered. The same principle applies when comparing supplier, inspection and receiving reports. A matching sample code is helpful; it does not make unlike methods equivalent.

Sampling explains another apparent conflict. Report A may test a production composite collected before packing, while Report B tests one finished carton after distribution. Even with the same method, the results describe different material unless the sampling design links them. Note carton and pallet selection, production time, retention conditions and chain of custody. Frozen product should be shipped to the lab in a way that protects its state and avoids uncontrolled thawing. The microbiologist should evaluate whether the handling could influence the target organism's recovery. We would keep these details with the batch traceability record rather than reducing the dispute to which COA has the lower number.

A genuine method comparison uses paired representative portions and a preplanned analysis, not an informal collection of COAs from different lots. Such a study needs the organism, matrix, concentration range and relevant performance criteria. The FDA's newer validation guidance and ISO food microbiology standards show that method performance is assessed under defined conditions. A supplier should not declare two methods interchangeable merely because their result labels look similar. A buyer can request an equivalence justification from the laboratory or choose the specified method for both sides of a transaction.

The physical GreenLand green-bean product photos in this article illustrate the lot material and forms a buyer might sample. They are not laboratory evidence and carry no microbial result. For the broader frozen vegetable range, specify the exact product rather than applying a generic matrix label across beans, broccoli and mixed vegetables without method review. Commercially, this prevents a report from being attached to the wrong product or portion of a shipment.

A tested portion cannot prove every piece in the lot has the same status.

A workable laboratory request begins with the named target and purpose. State whether the buyer needs a quantitative indicator count, an MPN estimate or a qualitative detection result. Give the full product description: IQF frozen green beans, whole or cut, pack size, any seasoning or added ingredients, lot identifier and intended use. Ask the laboratory to confirm the method's suitability for that matrix. The request should specify the sampling plan or refer to the governing standard and identify who will collect, seal and transport the material. These fields let the laboratory say whether the requested test and reporting format are feasible.

Then set the expected report language. For a count, ask for units per gram, reportable range, treatment of results below or above that range and any uncertainty information relevant to the decision. For MPN, ask for the design and the confidence or reporting bounds where the method provides them. For a qualitative test, ask that the target, tested mass and detected/not-detected expression appear together. State the acceptance criterion and whether a formal conformity statement is required. If it is, provide the decision rule or ask the laboratory to agree one before work begins. The laboratory must apply the method and reporting conventions it can defend; a purchasing request cannot authorize a misleading conversion.

The purchasing agreement should also define how an unexpected result will be handled. Name the receiving quality contact, the retained-sample policy, the review of lot and temperature records, and the path for a qualified food-safety investigation. A positive pathogen result should trigger that established response rather than a casual retest until a negative appears. If a report has a missing field, request a corrected or expanded laboratory document with an audit trail. Do not silently edit the COA in a local spreadsheet or remove qualifiers in a translated summary.

A usable request keeps the target, method, portion and units together.

At GreenLand-food, we can connect frozen vegetable form, packing, quantity, destination and batch identity to the documents a buyer requests. A quote discussion can include the target specification and test plan before production or shipment. The laboratory and food-safety team remain responsible for the analytical method and disposition judgment. Clear report language gives both buyer and supplier a common record: exactly what was tested, by which route, in what amount, and under which acceptance rule. That is the basis for comparing evidence without making the three result expressions mean more than they do.

Where a buyer receives a multi-row COA, keep the original laboratory wording beside any translated procurement summary. A dashboard can group rows by organism and purpose, but it should preserve the result qualifier, tested portion and report revision. If a laboratory later corrects a method code or a sample identifier, the buyer needs to know which version supported the original decision. An unversioned spreadsheet can quietly combine a revised value with an old sampling note. Keeping the signed report, sample chain and decision record linked is a simple way to prevent that loss of context.

For ongoing supply, compare trends only within a stable method and sampling design. A shift in plate count may warrant review of raw-material condition, sanitation or handling, but the team should first confirm that the laboratory and counting protocol did not change. An MPN trend should retain its estimation bounds, especially when values sit near the method's reporting limit. Qualitative results require their own sequence of test units and sample plan; a run of negative portions is useful evidence under that plan, not proof of a zero concentration in every pack. Buyers gain more from a well-defined series than from isolated results copied into a single unlabeled column.

Communication across teams should use the same vocabulary. A purchasing officer might say "negative" when the laboratory wrote "not detected in 25 g," while a receiving inspector might call "<10 CFU/g" a zero. Each shortcut loses the tested amount or the reporting threshold. A short COA legend can define the symbols and units once for the order. It should direct questions about method capability and pathogen results to qualified QA personnel. That does not make the report longer for its own sake; it prevents a decision based on a translation or shorthand that changed the actual evidence.