Microbial Identification, Excursions, and the EM Data Trend

Environmental monitoring (EM) tells you if your contamination control is holding. The colony-forming units on a settle plate, the organism on a fingertip dab, the bioburden in a water sample: each is a data point that either confirms control or warns you it is slipping. The work that turns those data points into a defensible quality decision is microbial identification, excursion investigation, limit-setting, and trending. This article walks through all four, in the sequence you actually do them, with the regulatory basis, the procedures, the acceptance criteria, and the mistakes that show up in inspection findings.

The audience here is the QC microbiologist, the micro SME, the QA reviewer who dispositions excursions, and the sterility assurance lead who owns the EM program. If you can do this job well, you can keep a fill line running without releasing a contaminated batch, and you can survive an inspection where the investigator pulls your EM trend report and asks why a recurring Gram-positive rod was never identified to species.

Why this matters and where the requirements come from

Environmental monitoring is not a standalone activity. It is part of the contamination control strategy (CCS) required by the EU GMP Annex 1, “Manufacture of Sterile Medicinal Products” (2022, effective August 2023). Annex 1 makes EM, personnel monitoring, and the response to excursions explicit, and it ties them to the CCS as a holistic design. The US framework lives in 21 CFR 211.42 (design and control of aseptic processing areas) and 21 CFR 211.113 (control of microbiological contamination), reinforced by the FDA guidance “Sterile Drug Products Produced by Aseptic Processing, Current Good Manufacturing Practice” (2004), usually called the aseptic processing guidance.

The compendial chapters give you the methods. USP <1116>, “Microbiological Control and Monitoring of Aseptic Processing Environments,” is the key reference for EM program design, contamination recovery rates, and the philosophy of limits. USP <61> and <62> cover microbial enumeration and tests for specified microorganisms in nonsterile products. USP <1113>, “Microbial Characterization, Identification, and Strain Typing,” is the chapter that governs the identification work itself. For sterile products, USP <71> sterility testing sits downstream, and the water chapters USP <1231> and the monographs <643>/<645> govern utilities. PIC/S guidance (PI 011, the GxP data integrity guide, and the GMP guide PE 009) mirrors the EU expectations.

The risk rationale is simple. A sterile product cannot be tested into compliance because sterility testing samples a tiny fraction of the batch. Assurance comes from the process and from the monitoring data that demonstrate the process stayed in control. When an EM result exceeds a limit, that excursion is evidence that contamination control may have failed, and the investigation has to decide whether the product is at risk. Identification tells you what the organism is, where it likely came from, and whether it is objectionable. Trending tells you whether a single result is noise or the leading edge of a real problem.

Microbial identification: what, why, and how

Why you identify, and to what level

Identification answers three questions: where did this organism come from, is it objectionable, and is the same organism showing up repeatedly. Annex 1 expects organisms recovered from Grade A and Grade B areas to be identified, and it expects identification to support investigation and trending. The aseptic processing guidance is explicit that microbiological monitoring data should be characterized, and that organisms from critical areas warrant identification to support investigations.

Identification has levels, and you choose the level by risk:

• Genus level is often enough for routine Grade C/D recoveries and for trending the general flora of a controlled-not-classified area.
• Species level is expected for Grade A and Grade B recoveries, for sterility test isolates, for any excursion investigation, and for organisms used to challenge media fills and disinfectant efficacy.
• Strain level (typing) is reserved for investigations where you must prove or disprove a link, for example matching a sterility test positive to an EM isolate, or tracking a persistent contaminant through a cleanroom over time. Strain typing methods include rep-PCR, ribotyping, multilocus sequence typing, and whole-genome sequencing.

USP <1113> frames this tiered approach and warns that no single method is adequate for every situation. The level of identification should be justified in your SOP and tied to the risk of the area and the purpose of the sample.

Phenotypic methods

Phenotypic identification reads what the organism does or what it is made of.

• Colony and cell morphology plus Gram stain is the first read on any isolate. It is cheap, fast, and tells you if the isolate is a Gram-positive coccus, a Gram-positive rod, a Gram-negative rod, a yeast, or a mold. It is not a final identification on its own, but it directs the next step and catches gross mismatches.
• Biochemical and substrate-utilization panels (commercial automated systems that read carbon-source utilization or enzymatic reactions) give a genus and often species with a confidence score. They depend on a healthy, pure, correctly-aged culture and on the organism being in the system’s database. Slow-growers, fastidious organisms, and environmental species are common failure points.
• Fatty acid methyl ester (FAME) analysis profiles the cell membrane lipid composition against a library. It is sensitive to growth medium, incubation temperature, and culture age, so it requires tightly standardized culturing.

MALDI-TOF mass spectrometry

Matrix-assisted laser desorption ionization time-of-flight mass spectrometry has become the workhorse for routine micro ID in pharma QC. It reads the ribosomal protein fingerprint of the organism and matches it against a reference library, returning a genus or species call with a score.

What makes it attractive: results in minutes from a single colony, low consumable cost per sample, and broad coverage of bacteria and many yeasts. What you have to manage:

• Library coverage. Environmental and pharma-relevant organisms are not always well represented in the manufacturer’s default library. Many labs build supplemental in-house libraries from characterized isolates, which is itself a controlled, validated activity.
• Sample preparation. Direct smear works for many bacteria; tube extraction (formic acid) is often needed for Gram-positives, yeasts, and molds. Molds frequently need extended extraction or remain hard to call.
• Score thresholds. The instrument reports a confidence score. Your SOP must define the score cutoffs for a species-level call versus a genus-level call versus “no reliable identification,” and what you do when the score is borderline (typically re-prep and re-run, then escalate to genotypic).

MALDI-TOF used in a GMP lab is a qualified instrument running validated software, so it falls under the same controls as any other computerized system: see analytical instrument qualification and the data integrity expectations in chromatography data system integrity and audit trail design and review. The library is a controlled record; changes to it go through change control.

Genotypic methods (sequencing)

When phenotypic and MALDI-TOF results conflict, when the score is below threshold, or when you need species certainty for an investigation, you sequence. The standard routine method is 16S rRNA gene sequencing for bacteria and the internal transcribed spacer (ITS) region or D2 region of the large-subunit rRNA gene for fungi. Sequencing reads the actual gene, compares it against a curated database, and is the most discriminating routine method for genus and species. It is slower and more expensive than MALDI-TOF, which is why most labs use it as the escalation step rather than the front line.

Genotypic identification is what you reach for when an inspector asks, “You called this Bacillus by MALDI with a low score and never confirmed it. How do you know it was not an objectionable spore-former in a sterile area?” Sequencing closes that gap.

Choosing a method: a comparison

No single method is adequate for every situation, which is why a real lab runs a tiered workflow rather than picking one tool. The trade-offs:

The pattern most labs settle on: Gram stain to direct, MALDI-TOF as the workhorse, sequencing as the escalation, and strain typing reserved for link investigations. The escalation decision is rule-based:

A worked identification decision

A Grade A glove print yields one colony. Workflow:

A Grade A recovery of a typical human skin organism still triggers a full investigation because of where it was found, not because the organism is exotic. The ID supports the conclusion that the source was personnel, which points the corrective action at gowning and aseptic technique.

Acceptance criteria for identification

• The method is qualified and the database/library is version-controlled and current.
• The score or match threshold for each call level is defined in the SOP and met.
• Pure culture is confirmed before identification; mixed cultures are subcultured and re-isolated.
• The identification is recorded with the raw result, the score, the database version, the analyst, and the date, with a reviewer signature.
• Escalation to a higher-discrimination method is documented when thresholds are not met.

Alert and action limits: what they are and how you set them

Definitions and basis

An alert limit is a preset count that flags the first sign of drift away from your normal operating state. Crossing it does not by itself mean control is lost, but it does call for attention and a look at what changed. An action limit is the count at which you must investigate and take corrective action, because being above it suggests the environment may no longer be operating within an acceptable band. Both concepts trace to USP <1116> and line up with how Annex 1 and the aseptic processing guidance treat limits.

A critical point that trips people up: in Grade A / ISO 5 critical zones, the expectation is essentially zero recovery. Under Annex 1 the viable limit for Grade A is no growth, so any organism recovered there is treated as significant and investigated. You do not get to set a comfortable nonzero action limit for the critical zone. Lower grades (B, C, D) carry the limits laid out in Annex 1 tables and your own qualification data.

Limits are not specifications in the release sense. USP <1116> stresses that EM data describe the state of control of the environment, and that contamination recovery rates and trends matter more than treating each limit as a pass/fail acceptance criterion. The cleanroom grade definitions and particle limits come from ISO 14644-1 (see cleanroom classification ISO 14644); the microbial limits come from Annex 1 and your validation.

How limits are set

1. Start from the regulatory grade limits. For Grade A/B/C/D, Annex 1 provides recommended microbial limits for air, settle plates, contact plates, and gloves. These are ceilings, not targets.
2. Generate baseline data. During qualification of the cleanroom and during routine operation, collect enough data to characterize the normal recovery distribution for each location and sample type.
3. Derive alert limits from the data, capped by the regulatory limit. Because EM counts are mostly zeros with occasional low counts, simple mean-and-standard-deviation statistics behave poorly. USP <1116> recommends a contamination recovery rate approach: track the percentage of samples showing any recovery and the percentage exceeding limits, rather than treating counts as a normal distribution. Many programs set alert limits using a percentile of historical data (for example a high percentile of nonzero counts) or a count just below the action limit, with the action limit set at the regulatory level.
4. Document the rationale. The limit-setting basis goes into the EM program document and is justified, version-controlled, and periodically reviewed.

A worked limit table

Illustrative limits for a settle-plate program (90 mm, 4-hour exposure). The Grade values follow the Annex 1 recommended levels; alert limits are facility-derived and must be supported by your own data.

The Grade A row is the one to internalize: there is no alert/action split where the action limit is a nonzero number you are allowed to drift up to. Any colony in Grade A is investigated.

Common limit-setting mistakes

• Setting alert limits equal to action limits, so you lose the early-warning function entirely.
• Treating the EM limit as a release specification and dispositioning product purely on the EM number rather than on the investigation.
• Applying mean-plus-3SD statistics to count data dominated by zeros, producing nonsensical limits.
• Failing to revisit limits when the recovery pattern changes, so the limits no longer reflect the actual state of control.

Excursion investigations: the procedure

An EM excursion is a result that exceeds an alert or action limit, or any recovery in a zero-tolerance zone such as Grade A. The investigation is a deviation handled under your quality system; see deviation management and root cause analysis techniques for the mechanics. Here is the EM-specific flow.

Step by step

1. Detect and secure. The result is read, recorded, and flagged against its limit. Preserve the plate or isolate; do not discard it. The isolate is evidence and you will need it for identification and possibly strain typing.
2. Open the event and assess immediate impact. Within the timeframe your SOP defines, open a deviation. Identify the affected area, sample type, date/time, and any product manufactured in that area during the relevant window. This batch-impact question is the one that matters most: is there product on the line that this excursion calls into question?
3. Verify the result. Rule out a laboratory or sampling artifact: media growth promotion was current, the plate was not damaged or contaminated during handling, the incubation conditions were correct, the sampling technique was sound. A confirmed sampling/lab error is a legitimate root cause, but it must be evidenced, not assumed. Do not invalidate a result without documented justification; unjustified invalidation is a classic inspection finding.
4. Identify the organism. Take it to the level the area demands (species for Grade A/B). The identity drives the investigation: a Gram-negative rod from water flora points one direction, a mold points to HVAC or a breach, a human skin organism points to personnel and gowning.
5. Investigate the source and scope. Review concurrent data: other EM locations on the same day, personnel monitoring, the same location’s recent history, HVAC and pressure differentials, cleaning and disinfection records, interventions logged during the session, and the media fill history. Look for whether this is isolated or part of a pattern. Pull the trend for the location and the organism.
6. Determine root cause. Use a structured tool (fishbone, 5-whys) tied to the evidence. Resist the reflex to blame “operator error” without a mechanism; see human error in deviations.
7. Assess product impact and disposition. With the ID, source, and scope in hand, the QA dispositioner decides whether affected batches are at risk. This feeds batch disposition decisions. The decision is documented and justified, not defaulted.
8. Define CAPA. Correct the immediate issue and prevent recurrence. Effectiveness is verified later; see CAPA effectiveness verification and what is a CAPA.
9. Close with QA approval. The investigation, conclusion, impact assessment, and CAPA are reviewed and approved by QA. The isolate disposition and trending update are recorded.

A worked excursion narrative

Event: Grade B active air sample returns 7 CFU against an action limit of 10 and an alert limit of 5. The alert limit is exceeded.

Immediate impact: An aseptic fill of Batch ABC was in progress in the adjacent Grade A zone during the sampling window. Batch quarantined pending investigation.

Verification: Media growth promotion current, plate intact, incubation within range, sampler calibrated. Result confirmed real.

Identification: MALDI-TOF after extraction gives Micrococcus luteus, species-level score. A common human-associated environmental organism, not classically objectionable, not a spore-former, not a Gram-negative waterborne organism.

Source and scope: Grade A settle plates and glove prints for the session were all no-growth. Personnel monitoring for the two operators showed no recovery. The location’s 6-month trend showed two prior alert-level recoveries of skin flora, both during gowning-intensive sessions. HVAC and pressure differentials were in range.

Root cause: Probable transient personnel-shed organism in Grade B during a high-activity setup; no evidence of Grade A breach.

Product impact: Grade A monitoring was clean throughout the fill; the excursion was confined to Grade B and the organism is not objectionable. Batch ABC assessed as not impacted, supported by clean Grade A data and the media fill record. Disposition: release pending all other tests.

CAPA: Refresher on Grade B gowning movement minimization; add the location to enhanced trending watch for one quarter.

That narrative is what a good closed investigation reads like: each conclusion is tied to evidence, the impact decision rests on the right data (Grade A cleanliness for a Grade A fill), and the CAPA matches the root cause.

Acceptance criteria for an excursion investigation

• Opened and progressed within SOP timeframes.
• Batch-impact question addressed explicitly and early.
• Result verified before any invalidation; invalidation justified by evidence.
• Organism identified to the required level.
• Scope assessed across concurrent and historical data, not just the single sample.
• Root cause supported by evidence, not assumed.
• Product disposition documented and justified.
• CAPA defined, with effectiveness verification planned.
• QA approval recorded.

The sterility-positive linkage: when identification has to prove a connection

The highest-stakes version of this work is a positive sterility test. The question the investigation must answer is whether the positive reflects genuine product contamination or laboratory contamination introduced during the test, because that determination drives whether a batch can be considered for any path other than rejection. Identification and strain typing are central to it, and the rules for invalidating a sterility test are strict (see USP <71> and the aseptic processing guidance; an invalidation needs demonstrable evidence of a lab cause, not a plausible story).

The disciplined workflow:

1. Preserve and identify the sterility isolate to species. A genus call is not enough here.
2. Pull every environmental and personnel isolate from the relevant window (the fill session, the test session, the suite) and identify the candidates to species.
3. Compare species first. If the sterility isolate species appears nowhere in the EM or personnel data, that absence is informative but not exonerating. If it matches an EM or personnel isolate, go to strain typing.
4. Strain type the matched isolates (rep-PCR, MLST, or whole-genome sequencing) to determine whether they are the same strain. Same species, different strain weakens an environmental-source argument; same strain strengthens it.
5. Reach an evidence-based conclusion. A strain match between the sterility positive and a Grade A EM isolate points to a genuine contamination event and against invalidation. A match to a microbiology-lab environmental isolate, with a documented lab breach, can support a lab-error conclusion. A clean Grade A EM record through the fill, with the positive matching a known lab organism, is the classic lab-contamination pattern, but it still needs the evidence, not just the pattern.

A short worked outcome: a sterility positive identifies as Ralstonia pickettii. The microbiology lab’s water-bath environmental monitoring recovered the same species the same week. Whole-genome sequencing shows the two isolates are the same strain, and the fill suite’s Grade A EM was no-growth throughout the fill. The weight of evidence supports laboratory contamination from the water bath, documented and corrected; the batch disposition proceeds on that evidence-based conclusion rather than on a convenient assumption. Reverse the facts (the strain matches a Grade A isolate from the fill), and the same rigor forces the opposite conclusion.

Objectionable organisms

What makes an organism objectionable

There is no fixed universal list. An objectionable organism is one that, given the product, its route of administration, the patient population, and the manufacturing context, poses a risk to product quality or patient safety. The concept comes from 21 CFR 211.113 (control of objectionable microorganisms) and is developed in USP <1111>, “Microbiological Examination of Nonsterile Products: Acceptance Criteria for Pharmaceutical Preparations and Substances for Pharmaceutical Use,” and in USP <62>, which tests for specified microorganisms. The determination is risk-based and product-specific.

Drivers that make an organism objectionable in context:

• Route and population. An organism tolerable in a topical product can be unacceptable in a sterile injectable or an inhalation product, and far more serious for immunocompromised patients. Sterile products: any recovered organism in a sterile-area critical zone is significant.
• Pathogenicity and toxin production. Gram-negative organisms that produce endotoxin, recognized pathogens, and spore-formers in a sterile context raise the stakes; see bioburden and endotoxin testing.
• Product survival and proliferation. An organism that can grow in the product, survive its preservative system, or persist in a water system is more objectionable than one that cannot.
• Source significance. A waterborne Gram-negative organism recovered near a sterile fill suggests a utility or barrier breach, which is more concerning than a transient skin organism.

For sterile products manufactured aseptically, the practical rule is that any organism recovered in the critical zone is treated as significant and is fully investigated, regardless of whether it appears on a “pathogen” list.

How to operationalize it

1. Define, in an SOP, the framework for classifying organisms as objectionable, the data inputs (product, route, population, preservative system, manufacturing context), and the decision authority (micro SME plus QA).
2. When an organism is identified, assess it against that framework in the investigation record.
3. Maintain a facility flora library and a watch list of organisms of concern for your products, and update it from trending.
4. Tie objectionable-organism findings to product disposition and, where relevant, to recall or field-action evaluation; see recall management and field actions.

Worked objectionable assessment

A nonsterile oral liquid bioburden sample recovers Burkholderia cepacia complex. This organism is a recognized concern for nonsterile aqueous products because it survives in water and resists some preservatives, and it has been associated with recalls of contaminated aqueous products. Even at a count within a numeric bioburden limit, recovering it triggers an objectionable-organism assessment, batch impact review, and a hard look at the water system and preservative efficacy. The numeric limit being met does not end the inquiry; the identity does.

EM data trending

Why trend, and what the regulators expect

A single in-limit result tells you little. The signal is in the pattern over time: a location creeping up, an organism recurring, seasonal shifts, a step change after a maintenance event. Annex 1 requires trending of EM and personnel monitoring data and expects trends to feed the contamination control strategy and the periodic review of the program. USP <1116> centers on contamination recovery rates and trend analysis rather than single-point pass/fail. The aseptic processing guidance expects monitoring data to be reviewed for adverse trends. Out-of-trend handling connects to out-of-trend investigations and the statistical tools in statistics in quality cpk and control charts.

What to trend and how

• By location and sample type. Track each EM point and each sample type (active air, settle plate, contact plate, glove, garment, surface) so a single drifting location is not hidden in a facility average.
• By contamination recovery rate. Per USP <1116>, track the percentage of samples with any recovery and the percentage exceeding alert/action limits, by grade and by location. This is more meaningful for low-count data than averaging CFU.
• By organism. Track recovered genera/species over time. A recurring single organism is a stronger signal than a one-off, and it can reveal a persistent resident contaminant or a niche.
• By personnel. Trend operator gowning and glove results; a specific operator or a shift pattern can emerge.
• Frequency. Routine trend reports are commonly monthly or quarterly, with an annual review feeding the product/quality review; see annual product review pqr and continued process verification cpv.

Defining an adverse trend

Set objective rules in advance so an adverse trend is detected consistently, not by eye. Common rules:

• A defined number of consecutive results above the alert limit at the same location.
• A rising contamination recovery rate over consecutive periods.
• Repeated recovery of the same organism at the same or related locations within a defined window.
• A step change in recovery rate after a facility or equipment event.

When a rule fires, open an out-of-trend investigation even if no single result breached an action limit. That is the point of trending: catch the drift before it becomes an excursion over the action limit.

Worked trend snapshot

Grade B contamination recovery rate for active air, one location, by month:

No single April result breached the action limit, but the recovery rate tripled over the quarter and alert exceedances are climbing. The pre-defined rule (rising recovery rate over three consecutive periods) fires. This is an adverse trend warranting an out-of-trend investigation into that location: HVAC performance, disinfection efficacy at that point, and the organisms recovered. Waiting for an action-limit breach would mean acting after control was already lost.

Acceptance criteria for trending

• All EM and personnel data captured in a controlled system, with audit trail and defined data review; see audit trail design and review.
• Trending performed by location, sample type, recovery rate, and organism.
• Adverse-trend rules pre-defined and applied consistently.
• Trend reports reviewed and approved on a defined cadence.
• Findings feed the contamination control strategy and the periodic program review.

Roles and responsibilities

A frequent organizational failure is letting micro own the data and QA rubber-stamp the conclusion. QA has to engage with the impact assessment and the disposition, not just sign it.

Common mistakes and inspection-finding patterns

• Organisms not identified to an adequate level. Critical-area recoveries identified only to genus, or “Gram-positive rod” recorded with no further work. Inspectors cite the inability to assess whether an organism was objectionable or whether it matched a sterility positive.
• Excursions closed without a real root cause. “Probable operator error” with no mechanism, no scope assessment, no link to trend. Recurrence then shows the closure was hollow.
• Unjustified invalidation of results. Calling a recovery a “lab error” or “sampling artifact” without evidence to retest and disposition the batch. This is a recurring data integrity and investigation finding; see data integrity foundations.
• Trending that exists on paper but is not acted on. A rising recovery rate visible in the trend report with no out-of-trend investigation opened, so the program detected the drift and ignored it.
• Alert limits set equal to action limits, eliminating early warning.
• Treating Grade A recoveries as acceptable because a number was set instead of “no growth,” or not investigating any Grade A recovery as significant.
• Isolate not retained, so identification and strain matching cannot be done after the fact.
• EM and personnel data not integrated in the investigation, so a personnel source is missed.
• Media growth promotion or incubation conditions not demonstrated, undermining every negative result the program relies on.
• Objectionable-organism assessment skipped for nonsterile product because the count was within the numeric limit, missing a waterborne pathogen.

Interview-ready questions and answers

What is the difference between an alert limit and an action limit? An alert limit is an early-warning level; exceeding it is not necessarily a loss of control but requires assessment. An action limit, when exceeded, signals the process may be out of control and requires investigation and corrective action. EM limits describe the state of control of the environment, not a product release specification.

What is the action limit for microbial recovery in Grade A? No growth. Any recovery of any organism in the Grade A critical zone is significant and investigated. Annex 1 sets the Grade A recommendation as no growth, so there is no nonzero action limit you may drift up to.

When do you identify to species versus genus? Genus is often acceptable for routine lower-grade trending. Species is expected for Grade A/B recoveries, sterility test isolates, and excursion investigations. Strain typing is reserved for investigations where you must prove or disprove a link, such as matching a sterility positive to an EM isolate.

Walk me through an EM excursion investigation. Secure the isolate, open a deviation, assess product impact immediately (what was on the line during the window), verify the result is not a lab/sampling artifact, identify the organism to the right level, assess source and scope using concurrent and historical data, determine an evidence-based root cause, disposition affected product with justification, define and later verify CAPA, and close with QA approval.

How do you set EM limits? Start from the regulatory grade limits as ceilings, generate baseline data during qualification and routine operation, derive alert limits from facility data using a contamination-recovery-rate or percentile approach because the data are mostly zeros, document the rationale, and review periodically. Mean-plus-3SD on count data is inappropriate.

What is an objectionable organism? One that poses a risk to product quality or patient safety given the product, route, population, and context. It is risk-based and product-specific, not a fixed list. Drivers include route and patient population, pathogenicity and endotoxin, ability to survive or grow in the product, and source significance. For sterile critical zones, any recovered organism is treated as significant.

Why does MALDI-TOF sometimes need confirmation by sequencing? MALDI matches a protein fingerprint against a library and returns a score. If the score is below the species threshold, if the organism is poorly represented in the library, or if the result is challenged in an investigation, you escalate to 16S (bacteria) or ITS/D2 (fungi) sequencing for a definitive call.

How do you handle trending, and what is an adverse trend? Trend by location, sample type, contamination recovery rate, and organism, on a defined cadence. An adverse trend is detected by pre-defined rules: consecutive alert exceedances at a location, a rising recovery rate over periods, or repeated recovery of the same organism. When a rule fires you open an out-of-trend investigation even without an action-limit breach.

A sterility test is positive and you recovered an organism from EM the same week. What do you do? Identify both isolates to species, and if they match, use strain typing to determine whether they are the same strain. A match supports a genuine contamination event and informs whether the sterility failure is attributable to the environment; a mismatch helps argue laboratory contamination but does not by itself prove it. Either way it is a full investigation with QA-led disposition.

Practical tips

• Keep every isolate from a critical area until the investigation is closed. You cannot strain-type a plate you threw away.
• Build and qualify a facility flora library. Knowing your normal residents makes an unusual organism obvious and speeds investigations.
• Version-control your ID database/library and put changes through change control; an inspector will ask what library version produced a given call.
• Write the batch-impact assessment first, not last. The whole point of the urgency is the product on the line.
• Track contamination recovery rates, not just CFU averages. Low-count data hide drift when averaged.
• Pre-define your adverse-trend rules so detection is objective and consistent.
• Tie disinfectant efficacy and media fills to your EM organisms; the organisms you recover should be the ones your disinfectant program and media fill challenges actually address; see aseptic processing and media fills.

Related articles

• environmental monitoring program
• aseptic processing and media fills
• annex 1 contamination control strategy
• cleanroom classification iso 14644
• bioburden and endotoxin testing
• sterility testing usp 71
• out-of-trend investigations
• deviation management
• root cause analysis techniques
• batch disposition decisions
• water system validation usp 1231