Environmental Monitoring in ISO 5/Grade A Aseptic Areas: Common FDA Warning Letter and Form 483 Citations

Environmental monitoring in ISO 5 / Grade A aseptic areas: recurring deficiencies in FDA warning letters and 483s

FDA's enforcement record on environmental monitoring (EM) in aseptic processing repeatedly returns to six failure modes. They appear in warning letters to sterile drug manufacturers (US and foreign) and in Form 483 observations at sterile compounders, outsourcing facilities and radiopharmacies:

Alert and action limits that are missing, not justified, too permissive for ISO 5, or not based on facility history.
Excursion investigations that were never opened, left open, or closed without root cause, product impact assessment, batch scope, or effective CAPA.
Personnel monitoring that skips operators, runs too infrequently, uses sampling technique that hides contamination, or does not respond to glove and gown recoveries.
Program design: sampling locations away from the critical zone, too little sampling, and sampling under static rather than dynamic conditions.
Isolate identification and trending: organisms not identified to genus or species, and adverse trends not detected or acted on.
EM data integrity: desiccated plates, misread or unrecorded CFU counts, discarded plates, and fabricated results.

The sections below set out the FDA benchmark that investigators apply, then give representative citations for each theme.

The benchmark FDA applies

Investigators and warning-letter authors measure firms against FDA's guidance Sterile Drug Products Produced by Aseptic Processing — Current Good Manufacturing Practice. Its key expectations for ISO 5 areas:

No growth is the normal state. "Air monitoring samples of critical areas should normally yield no microbiological contaminants," and contamination there warrants investigation ⁵²². The guidance's table states the same expectation for Class 100 (ISO 5) ⁵²³.
Recommended action levels. For ISO 5 the guidance lists an active-air action level of 1 cfu/m³ and a settling-plate action level of 1 cfu/4 hours ⁵²⁴. For comparison, ISO 7 is 10 cfu/m³ and 5 cfu/4 hours ⁵²⁶. Alternate action levels may be appropriate depending on the operation or method, and settling plates are optional ⁵²⁸.
Alert vs. action. An alert level is an early warning of drift that calls for scrutiny and follow-up. Exceeding an action level should trigger investigation and corrective action based on what the investigation finds ³⁷⁵.
Personnel monitoring. Sample each operator's gloves daily or with each lot, and sample strategically selected gown locations at an appropriate frequency. Investigate promptly when an operator exceeds levels or shows an adverse trend. Gloves should not be sanitized immediately before sampling ³⁷⁶.
Isolate identification. Monitoring of critical areas, surrounding clean areas, and personnel should include routine identification to species (or, where appropriate, genus) level ⁵³⁰. Ancillary areas should be identified at frequent intervals so the firm keeps a current contaminant database ⁵³¹.

These expectations recur in the warning letters discussed below.

1. Alert and action limits

ISO 5 limits set above "normally no growth"

FDA repeatedly cites ISO 5 limits that let recoveries go uninvestigated.

Nephron SC (October 11, 2022): FDA called the firm's ISO 5 action levels inappropriate because critical-area samples should normally yield no contaminants. Under those levels, the firm did not investigate three of six viable samples above 1 CFU on its blow-fill-seal lines. FDA also said USP <797>, which the firm had relied on, was not the appropriate standard for conventional sterile drug manufacturing ²¹⁴²¹⁵.
Akorn (February 4, 2019): Personnel intervening in an ISO 5 cabinet had a 2 CFU glove action limit, no alert limit, and no scientific justification. Repeated 1 CFU results drew no appropriate response. FDA said glove growth during any ISO 5 activity should trigger an alert or action condition, not only growth after interventions designated "critical" ²¹⁶²¹⁷.
Samchundang Pharm (May 13, 2020): Personnel monitoring limits for aseptic staff were 3 CFU/plate (alert) and 4 CFU/plate (action). FDA said growth on samples from personnel who can work in ISO 5 should trigger an investigation, and asked the firm to justify its limits ²¹⁸²¹⁹.
AllerQuest (June 24, 2019): FDA said fingertip monitoring should have a 1 CFU action limit ³⁰⁸.

Limits not justified or not based on facility data

Form 483s often cite limits that were copied from a reference instead of derived from site performance.

Claris Injectables (August 2017): Alert and action limits were not based on documented historical EM performance, and actual counts ran well below the limits ¹⁰⁷¹⁰⁸.
Jubilant HollisterStier, Kirkland (June 2024): Alert and action levels relied on guidance values rather than facility history and trends ¹¹⁰¹¹¹.
Auro Pharmacies (2016 and 2017): The firm could not justify its EM limits. On reinspection, FDA found that settle-plate limits used during ISO 5 aseptic filling were set incorrectly ¹⁰¹¹⁰⁴¹⁰⁵.
Shilpa Medicare (November 2025): The firm had no scientific justification for line-specific particle thresholds that triggered impact assessments, and Grade A particle excursions during an injection batch had no documented impact assessment or corrective action ¹²³¹²⁴.

Excursions above action limits that were never recorded or acted on

Taiwan Biotech (May 31, 2018): ISO 5/Class A and surrounding ISO 8 plates exceeded action limits with no investigation. One ISO 5 sample yielded 140 CFU once FDA had the plates enumerated, although the firm had reported no out-of-limit results for over a year ²²⁰.
Amman Pharmaceutical Industries (February 14, 2024): The firm reported no alert or action findings for two years. FDA then observed more than 50 microbial excursions, including ISO 5 recoveries ²²¹²²².
Kilitch Healthcare India (483, October 2023): Samples read October 16 to 18, 2023 produced 39 excursions, 15 of them Grade A action-level results. Staff described replacing over-limit results with results below the alert limit ¹¹³¹¹⁴¹¹⁵.

2. Excursion investigations

FDA expects an exceeded action level to trigger investigation and corrective action ³⁷⁵. The deficiencies cited range from no investigation at all to investigations that stop short of root cause, product impact, and batch scope.

No investigation or organism identification. At Brookfield Medical/Surgical Supplies, FDA cited nine ISO 5 microbial excursions between May 2025 and June 2026. Only one was identified to an organism, and all compounded products were released ⁷⁶⁷⁷. Stemell (August 28, 2019) did not investigate product impact, root cause, or CAPA after Kocuria salsicia growth in a Class 100 area, and its procedure had no alert/action limits or provisions for investigating exceedances ⁴⁸⁴⁹.
Investigations left open. At Fagron Compounding Services (July 2025), more than 64 EM excursion investigations were open. One concerned Penicillium citrinum recovered in an ISO 5 hood during a vancomycin IV bag lot ⁷⁸.
No root cause, impact assessment, or CAPA, with product released. At Allergy Laboratories, an out-of-limit Class 100 stopper-bowl surface sample produced an investigation with no documented root cause, no product contamination assessment, and no CAPA. The vials were released to inventory ⁷⁹⁸⁰. Smart Surgical dba Burst Biologics (February 2, 2022) distributed product linked to personnel monitoring excursions above its action limit, including gloves entering an ISO 5 area, without investigating them ⁵⁵⁰⁵⁵¹.
Root cause asserted without evidence. At Maiva Pharma (January 2026), the firm blamed a Grade B action-limit result on increased personnel movement without comparing traffic data or reviewing entry/exit history ⁸¹.
Scope too narrow. Cipla (February 25, 2020) rejected about 18 batches after ISO 5 excursions but did not explain root causes, the impact on other products made under the same conditions, or all CAPA implemented ⁵¹⁵². Nephron later investigated the air recoveries but did not conduct a comprehensive retrospective review of historical actionable results ⁵⁰. Brassica Pharma (July 11, 2024) had 37 ISO 5 action-level excursions during the inspection. Its proposed investigation left out a retrospective review of all recorded data ⁵⁵⁵⁶.
Recurring contamination treated as isolated events. PETNET Solutions (October 17, 2019) had adverse ISO 5 trends for roughly three years and Bacillus on multiple ISO 5 cabinet surfaces without an adequate response ⁵³⁵⁴. Sterling Pharmaceutical Services (September 27, 2022) investigated individual incidents but did not address a persistent ISO 5 trend that included spore-formers and fungi ²⁰². At Biocon (February 2019), CAPAs from EM excursion investigations were ineffective, and a 1 CFU Grade A recovery was not treated as a significant excursion ⁸⁴³⁶⁷.
Remediation without verification. At PETNET (483, February 2026), ISO 5 glove action-limit events of 15 and 38 CFU were followed by delayed remediation and no resampling after sporicidal cleaning ¹²⁰¹²¹.

3. Personnel monitoring

Personnel monitoring is among the most frequently cited EM topics in both warning letters and 483s. The citations fall into four groups.

Operators not monitored, or monitored too rarely

At CP Pharmaceuticals (November 16, 2016), most operators on the ISO 5 filling line were not monitored ²⁹⁵²⁹⁶.
Becton Dickinson's Franklin, Wisconsin site (September 14, 2018) did not require daily gloved-finger sampling of every operator performing aseptic syringe filling ²⁹⁷²⁹⁸.
Optikem International (June 20, 2024) monitored gloves and gowns only at a limited frequency, and operators chose who was sampled. FDA said personnel should be monitored on each day of aseptic processing ²⁹⁹³⁰¹³⁰².
At SCA Pharmaceuticals (2023), aseptic assistants who reached into ISO 5 hoods were never included in routine fingertip monitoring ²⁴⁶²⁴⁷.
Several 483s cite fingertip sampling done only at media fills or operator qualification, not after routine production ²⁴⁸²⁵⁰²⁵¹²⁷⁵.
Amman Pharmaceutical had no personnel monitoring tied to aseptic connections or complex manual interventions ²⁵⁸²⁵⁹.

Sampling technique that masks contamination

FDA consistently cites glove sanitization or glove changes immediately before sampling, which conflicts with the guidance ³⁷⁶.

Akorn staff sanitized gloved hands with isopropyl alcohol just before sampling ³⁰⁶.
At Med-Pharmex (May 17, 2017), one operator changed gloves before exit monitoring and another sprayed gloves before sampling ³⁰⁵.
483s at Boothwyn Pharmacy and Central Admixture Pharmacy Services cite glove spraying or sanitization before sampling, light contact that touched only the fingertips, and self-sampling by quality staff ²⁶⁰²⁶¹²⁶²²⁶³²⁶⁵.
An operator at Customceutical Compounding wore two pairs of gloves for end-of-production sampling but only one pair during production ²⁶⁶²⁶⁷.

Gown coverage and exit monitoring

California Pharmacy & Compounding Center did not sample arms, chest, or other gown sites ²⁴⁴²⁴⁵.
Toyobo (August 19, 2021) did not consistently make contact plates touch gown surfaces before operators left the Grade B area ³⁰³.
At Mylan Laboratories (August 6, 2015), an operator touched RABS gloves without sanitizing first, and operators were not monitored on exit ³⁰⁷.

Inadequate response to glove and gown recoveries

Cord for Life (March 29, 2019) had 23 positive fingertip samples in four months with no investigation or CAPA ³⁰⁹.
QuVa Pharma (January 26, 2024) did not thoroughly investigate every ISO 5 microbial recovery, including glove recoveries ³¹⁰.
At SCA Pharmaceuticals, the sleeve action limit was greater than 10 CFU. Thirty sleeve samples with spore-formers below that limit drew no investigation or product impact assessment ²⁷¹.
At Customceutical, an over-limit glove result was investigated without identifying the organism or assessing the products made that day ²⁷⁰.
Med-Pharmex had not assessed individual operator trends or whether operators needed requalification ³⁰⁵³¹².

4. Program design: locations, frequency, and dynamic conditions

FDA cites EM programs whose samples cannot represent conditions where product is exposed.

Locations away from the critical zone. Mylan's nonviable particle probes sat away from the ISO 5 working area and could not detect particles where sterile drug was exposed ¹³⁸. Emcure (March 3, 2016) did not define RABS sampling locations, let operators choose swab sites, and did not sample equipment used daily in aseptic operations ¹³⁹. 483s cite a settle plate placed under a wall-mounted laptop instead of near aseptic manipulations, and a viable air sampler more than 12 inches from the critical filling zone ³³⁰³³¹³³²³³³³³⁴.
Frequency. Amman did not monitor air in ISO 5 filling areas as often as required, and monitored ISO 5 surfaces infrequently despite ongoing production ¹⁴⁰. Several compounder 483s cite ISO 5 active air and surface sampling that was not done on every production day ³³⁵³³⁶³³⁷³³⁸.
Static instead of dynamic conditions. Viable air samples were taken with no operators or production at ISO 5 workstations ³⁴¹, or only during static requalification ³⁴²³⁴³. At Bausch & Lomb (2016), all five Class A nonviable locations on an aseptic fill line were sampled with equipment idle, and changed probe locations had no justification ³⁴⁸. Amman's airflow studies were run only under static conditions ¹⁴².
Particle probe placement. Staska Pharmaceuticals had a probe that was not oriented into the airflow ³⁴⁹. Central Admixture's nonviable monitoring was not continuous during processing and did not represent the highest-risk product area ³⁴⁷.

5. Isolate identification and trending

Identification

FDA expects routine species- or genus-level identification for ISO 5, surrounding areas, and personnel ⁵³⁰. Common deficiencies:

Panacea Biotec (September 24, 2020) identified only one of 6,360 microorganisms isolated from an ISO 7 area in 2019, relying partly on an incomplete photographic library ¹⁹³¹⁹⁴.
Samchundang grouped isolates by colony morphology and identified only one isolate per group ¹⁹⁵.
Amman did not routinely identify EM isolates. FDA said routine identification should cover ISO 5, surrounding cleanrooms, and personnel ¹⁹⁷¹⁹⁸.
483s at Dr. Reddy's (2019) and Amneal (2022) cite reliance on photographic "albums" without further identification ³⁶²³⁶³. Sun Pharmaceutical, Halol (2016) left 107 of 233 excursion isolates unidentified ³⁶¹.

Mylan did not trend repeated out-of-action-level results, including ISO 5 excursions in early 2015 ¹⁹⁹²⁰⁰.
Med-Pharmex did not investigate site-level recoveries visible in its own trending data, and one trend report recorded zero for an operator whose batch record showed CFU ²⁰¹.
Biocon Biologics did not investigate adverse trends that exceeded its own recovery-rate criteria, and did not evaluate how fungi and gram-negative organisms entered Grade A/B areas ⁸².
Biocon Sdn Bhd (2024) recorded 31 Bacillus cereus group personnel recoveries in Grade B but raised no adverse trend because alert limits had not been reached ³⁶⁸.
Hospira, Rocky Mount did not define how aseptic trend data should be presented, so a negative ISO 5 nonviable trend was missed at management review ³⁶⁴³⁶⁵.
Claris Injectables did not include personnel monitoring data in its EM summary report ³⁶⁶.

6. EM data integrity and sample handling

The most serious EM warning letters often involve data that cannot be trusted. FDA treats unreliable EM data as a sterility assurance problem because it hides contamination.

Desiccated or damaged media. Mylan used dried, previously incubated plates for surface and personnel monitoring ¹⁶⁷. Emcure had 36 desiccated plates ¹⁶⁸. Corden Pharma Latina had 61 damaged plates, which FDA said risked false negatives ¹⁶⁹.
Misread or misrecorded counts. Hanlim Pharm recorded settle-plate results before reading the plates ¹⁷⁰. Intas counted two merged colonies as one CFU ¹⁷¹. Liveyon Labs entered "too numerous to count" as 1 CFU ¹⁷². Altaire recorded "0" for a plate that was too numerous to count, and entered "0" for personnel monitoring that had not been done for up to a year ¹⁷⁴. 483s at Cantrell Drug Company and Exela Pharma Sciences cite investigators finding higher counts than the firm recorded, and no second-person verification ²⁵²⁷²⁸.
Discarded or unrecorded plates. Amman and Kilitch put Grade A plates showing growth in the waste before reading was due, with no counts recorded ²⁹³⁰.
Fabrication. Emcure labeled plates as exposed when no samples had been collected ¹⁷⁵. At Kilitch and Brassica, FDA reported staff statements that results were recorded for samples never taken ¹⁷⁶¹⁷⁷. Toyobo manipulated contact plate and particle measurements ¹⁷⁸. At Teva Pharmaceutical Works, settle plates were exposed unlabeled and EM paperwork was completed after the operator left the cleanroom ³¹.
Incubation control. 483s cite inconsistent incubation times, missing incubator temperature records, plates read days late or after 14 days, and uncalibrated thermometers ¹⁹²⁰²²²³.

What FDA asks for in warning letter responses

Warning letters citing EM deficiencies usually ask for more than a corrected SOP. Common requests:

A comprehensive assessment of the EM and personnel monitoring program with a CAPA plan covering sampling locations and frequency, alert and action limits, sampling technique, and trending (Hanlim, Samson Pharmaceuticals) ⁵³⁹⁵⁴⁰.
Independent retrospective review of EM and personnel data and past excursions, including falsified, misreported, or invalidated results (Samchundang, Altaire) ²¹⁹⁵⁴¹.
Sample accountability and isolate identification, along with investigation of action-limit deviations and risk assessments for affected U.S. product (Taiwan Biotech) ⁵⁴².
Independent assessment of ISO 5 data against equipment failures, with an action plan for distributed product (Cipla) ¹²⁹.
An operational program with review of daily results and trends, plus a personnel monitoring program for aseptic operators (Viatrexx, Sun Pharmaceutical) ⁵⁴⁴⁵⁴⁵⁵⁴⁶.

Practical implications for sterile manufacturers and compounders

The citations above translate into a set of questions an internal auditor can test against site practice:

Do ISO 5 alert and action limits reflect FDA's expectation that critical-area samples normally show no growth, and does every ISO 5 recovery, including any glove recovery, get a documented response ⁵²²²¹⁶³¹⁰?
Are limits justified with site historical data rather than copied from guidance or USP <797> ¹⁰⁷¹¹⁰²¹⁴?
Does every action-level excursion lead to a documented root cause, product impact assessment covering all potentially affected batches, and a CAPA that is checked for effectiveness ⁵¹⁷⁹⁸⁴?
Is every operator who performs or supports ISO 5 work sampled on gloves each day or each lot, with representative gown sites, no glove sanitization beforehand, and exit sampling ³⁷⁶²⁴⁶³⁰⁵?
Are viable and nonviable samples taken under dynamic conditions at locations that represent product exposure ³⁴¹³⁴⁷¹³⁸?
Are ISO 5, surrounding area, and personnel isolates identified to species or genus, and do trending rules catch recurring organisms before alert limits are reached ⁵³⁰³⁶⁸?
Can every plate be traced from exposure to reading, with verified counts, controlled incubation, and no discards before reading ²⁹²⁷³¹?