The Application of Manufacturing Management Systems to Aseptic Filling Operations and Pharmaceutical Water Systems

Tony Cundell, Ph.D., Principal Consultant, Microbiological Consulting, LLC

Introduction

In my recent article in the September/October 2025 issue of American Pharmaceutical Review on risk assessment in aseptic processing evaluated in cleanroom technology, risk assessment tools, and regulatory guidance as drivers for risk mitigation (Cundell, 2025), but the important role of information management was overlooked for the data collection of facility parameters and air monitoring results, to demonstrate environmental control. This is corrected in this article.

Membership of the Modern Microbial Methods (M3) Industry Group, pursuing the implementation of the biofluorescent particle counting (BFPC) technology to monitor the air for viable microorganisms during aseptic processing gave me new insights (Martindale et al, 2025). Continuous monitoring during manufacturing generates copious data that requires data management systems to interpret and identify adverse changes in classified areas. The days of control charting data with a pencil and adjusting the manufacturing process are long over. Advanced statistical processing enables real-time identification of excursions outside the alert and action levels, highlights adverse trends, and provides alarms in response to a loss of environmental control. Furthermore, this refinement can be enhanced by the application of machine learning to the data. A holistic approach to information management is required, viable particulate counts and operating parameters including air velocity, number of air changes, space pressurization, and non-viable particulate counts providing a clear understanding of manufacturing areas to detect anomalous results, determine correlations and optimized operating parameters.

Benefits of Holistic Aseptic Monitoring Systems

Real-time monitoring during aseptic operations provides sterility assurance to understand changes in environmental controls, or a human intervention with proactive line clearance will largely eliminate protracted failure investigations. Chronic issues related to aseptic processing would be highlighted and corrected, strengthening the compliance posture of the pharmaceutical manufacturer. Another advantage is the ability to demonstrate an area has returned to baseline after a shutdown, increasing facility utilization. Clearly, the objective is not only limited to the automation of data collection but includes the prediction of loss of environmental control and the optimization of the aseptic processing.

Pharmaceutical manufacturers, aseptic processing suppliers, and monitoring equipment vendors need to jointly develop user requirement specifications and ensure that data management systems are fit for purpose. Integration of hardware like sensors, networks, dashboards, and software to process, display, and store the data similar to that process analytical technology (PAT) has been doing for decades. The business justification will depend on whether sterile drug product manufacturers can be convinced that the investment is necessary to improve manufacturing (productivity), confirm product quality, and reduce product contamination.

Aseptic Environmental and Water Monitoring Data Integration

The automation of complex data collection with real-time detection supports the optimization of aseptic filling operations. Critical operating parameters and environmental controls like air velocity, air changes, and particulate contamination levels are measured continuously and using baseline data to determine the alert and action levels of an adverse trend. Changes in classified manufacturing areas before an excursion prevents product contamination and enhancing productivity. Excursions from predetermined action levels would be alarmed and would trigger appreciate responses, this may include line clearance, stoppage, or shutdown. Multivariate analysis would he conducted to determine if the parameters are statistically correlated.

A recently published study (Larsson et al, 2025) on the correlation between a real-time bioparticle detection device (BFPC) and a traditional microbiological active air sampler monitoring air quality in an operating room reported a correlation coefficient of 0.7 (95% confidence level 0.57 to 0.79). It would be important to extend these studies to determine the correlation of viable microorganisms to other operating parameters. Sandle (2025) has highlighted concerns about BFPC false positive. Any monitoring tests, e.g., HIV screening of blood donation, designed to limit false negatives will result in a measurable false positive rate, which is evaluated by repeat testing and clinical investigation of the donor. With BFPC in a Grade A (ISO 5) aseptic processing area the detection algorithm is adjusted to differentiate non-viable auto-fluorescent particle from viable microorganisms by size, shape, and fluorescence wavelength and intensity, the sources of non-viable particles are identified and eliminated and continuous monitoring, i.e. repeat testing will isolate an excursion.

The parameters, typical requirement, criticality, frequency of measurement, alarm status, and probable cause of failure associated with cleanrooms are found in Table 1.

Table 1. Recommended Measurement of Cleanroom Operating and Environmental Parameters - Grade A areas
Parameters	Typical Requirement	Criticality	Frequency of Measurement	Alarmed	Probable Cause of Failure
Air Velocity	0.45 m/min (90 ft/min) (± 20%)	Critical	Continuous	Yes	HVAC System Failure
Air Changes	120-240 air change/hr.	Critical	Continuously calculated from air velocity & room dimensions	Yes	HVAC System Failure
Airflow Pattern	Laminar	Critical	Periodic Smoke Studies	No	HVAC System Failure
Space Pressurization	12 Pa (0.05-inch water gauge)	Major	Continuous	Yes, with a time delay	HVAC System Failure Open Doors
Temperature	18 to 22°C (64.4 to 71.7°F)	Minor	Continuous	No	HVAC System Failure Climatic Season
Relative Humidity	20 to 50% RH	Minor	Continuous	No	HVAC System Failure Climatic Season
Non-viable Particulate Levels (0.5 and 5 micron)	<3500 and <1 particles/m3 respectively (in operation)	Major	Continuous	Yes	Breaches of HEPA Filters and Housings Particulate Generation by Equipment, Packaging and People
Viable Particulate Levels	<1 CFU/m3 (<1 AFU/m3)	Critical	Continuous	Yes	Human Interventions
CCTV Camera Images	Interventions per hr.	Critical	Motion Activated	No	Human Interventions
Intervention Downtime	Frequency and duration	Major	Line Stoppage Activated	No	Filling Operation Events
Units Filled	Accumulative total	Minor	Continuous	No	Poor Line Set-up Maintenance Issues

This discussion can be extended from air to water (Table 2). An important plant utility is the water system that produces pharmaceutical-grade water, which is often used as ingredient water. Heat sanitization of purified water distribution systems is energy consuming and takes the system out of service until the water approaches ambient temperature, which disrupts production. Traditional microbial monitoring leads to delayed results. Scheduling sanitization based on need mitigates risk and may lengthen the time between sanitizations.

Other parameter that could be added to the analysis, for information only, may include pH, and non-viable particulate measurement. Note: important parameters like Assimilable Organic Carbon (a subset of TOC) better related to microbial growth and bacterial endotoxins currently are not measurable online.

Table 2. Recommended Measurement Parameters of Pharmaceutical-Grade Water
Parameters	Typical Requirement	Criticality	Frequency of Measurement	Alarmed	Probable Cause of Failure
Total Organic Carbon (TOC)	NMT 500 ppb	Critical	Continuous	Yes	System Unit Operation Failure
Conductivity	< 1.3S/cm at 25°C	Critical	Continuous	Yes	System Unit Operation Failure
Bio-fluorescent Particle Count	Limits established by control charting	Critical	Continuous	Yes	Dead legs Infrequent Sanitization
Temperature	Ambient (PW) >60°C (WFI)	Major	Continuous	Yes	Heating system
Flow rate	1.2 m/s (approx. 3.9 ft/s)	Major	Continuous	Yes	Insufficient Circulation
Reynolds’s Number	>10,000	Major	Periodic (Calculated from the distribution velocity)	No	Lack of turbulence due low flow rate

Calculated Parameters

Most parameters can be measured directly, whereas some must be calculated from the collected data, i.e., derived parameters. The number of air changes per hour are calculated from the air velocity and dimension of the cleanroom using the following equation (Whyte, 2001):

Air changes per hour = Average air flow velocity x room area x 60 min/hr/Room volume

The Reynolds Number (Re) is calculated using the formula:

Re = ρvLμ where ρ is the fluid’s density, v is the fluid velocity, L is a characteristic length (like a pipe’s diameter), and μ is the dynamic viscosity. This dimensionless number predicts the type of flow, with low numbers indicating laminar flow and high numbers indicating turbulent flow (Saldana et al, 2024).

Multivariate Data Analysis to Statistical Process Control

Collecting data is simple and has limited value, whereas analysis and providing actionable information provides a golden opportunity in aseptic manufacturing. Univariate control charts track only a single parameter over time, which is insufficient for complex aseptic processes with multiple operating parameters and environmental monitoring results that are interactive. Most parameters can be measured directly where specific attributes can be monitored, whereas some must be calculated from the collected data, i.e., derived parameters. Multivariate Statistical Process Control (MSPC) is defined as the application of multivariate statistical techniques to analyze complex process data. MSPC in combination with automated data collection and analysis may be used to generate control charts based on a multivariate (chemometric) model. These charts can then be used to control and improve manufacturing processes including environmental control in aseptic processing areas and report summaries on a dashboard.

In a pioneering publication (Guh, 2003) on integrating artificial intelligence/machine learning into online statistical process control (SPC), the author pointed out that SPC applications involve three major objectives, which are: 1) monitoring the process, 2) detecting deviations from the process control, and 3) taking corrective action in response to the alarm signal and related deviations. Since the 2003publication the literature on the application of machine learning to automated has become voluminous. The application of multivariate statistical techniques to analyze complex process data associated with drug manufacture with potentially correlated variables was recently reviewed in a book chapter (Manzano and Whitford, 2023). Notably, the PDA has an active Interest Group on Advanced Manufacturing and Applied Process Digitalization chaired by Peter Makowenskyj and Toni Manzano.

Other industry groups with strong digitalization programs include the annual ISPE Pharmaceuticals 4.0 Conferences, the BioPhorum Digital Technology Roadmap, the ASTM standards E1130subcommittee, and the University College London Future Targeted Healthcare Manufacturing Hub.

With respect to environmental monitoring in an aseptic filling area exceeding an action limit, an increased frequency of exceeding an alert limit, or the recognition of a potential adverse trend must be detected, and the appropriate warning issued. This can be viewed as a rule-based decision system for detecting an abnormal situation (anomaly). To avoid being disruptive, false alarms must be limited to significant deviations, and subject to system qualification. In practice, four kinds of information about the status of the environmental control must be provided: 1) the current status of the process, i.e., in-control or out-of-control, with an alarm if the process is out-of- control, 2) the probable cause of an out-of-control situation often related other environmental monitoring systems in the cleanroom, e.g., opening of a barrier system, HEPA filter failure, reduction in laminar air flow velocity, etc., 3) the effective action to be taken in response to the out-of-control situation, e.g., limiting an ongoing human intervention, stopping the line and clear vials from the line, or to abort the filling operation, and 4) documentation of the event for investigation purposes. To maintain a higher level of sterility assurance, this feedback must be conveyed directly to the cleanroom operators and not just the area managers, so the operators may take immediate action.

Reporting Deviations

To optimization of the aseptic processing data must processed in real-time with feedback loops that alert operators and control key manufacturing steps. These proposed requirements are listed above:

• “Traffic light” alarm reporting- Displaying flashing red, orange, and green lights on a centralized pole visible to the cleanroom operators.
• Dashboard reporting (Current) – Visible to the operators, area supervisors and quality assurance. Continuously updated for critical parameters.
• Dashboard reporting (Retrospective) - Benchmarking all parameters against operational history.
• Multivariant Analysis – Determination of which the parameters show a low, medium or high level of correlation.
• Data retention – There are GMP issues around the retention of raw data and batch related summaries. Cloud storage will be necessary with sunset and exception-based retention rules to control and diminish storage space.
• Sensor and software development – This should be vendor-driven with pharmaceutical companies selecting only those which meet their accuracy and data management requirements.
• Deviation investigation – Time-stamped data and event-activated video for review during failure investigation.

Inclusion of all parameters in the investigations.

Regulatory Agencies Guidance

The 2022 Revision of the EU GMPs Annex 1 Section 2.1 states “Facility, equipment and process should be appropriately designed, qualified and/or validated and where applicable, subjected to ongoing verification according to the relevant sections of the Good Manufacturing Practices (GMP) guidelines. The use of appropriate technologies (e.g., Restricted Access Barriers Systems (RABS), isolators, robotic systems, rapid/alternative methods and continuous monitoring systems) should be considered to increase the protection of the product from potential extraneous sources of endotoxin/pyrogen, particulate and microbial contamination such as personnel, materials and the surrounding environment, and assist in the rapid detection of potential contaminants in the environment and the product.” Our industry supports these recommendations.

Regulators see continuous process validation (CPV) that would be assembled by a Manufacturing Information System as evidence of sustained control of a validated manufacturing process throughout commercial production (FDA Guidance for Industry: Process Validation: General Principles and Practices 2011 and EMA Guideline on Process Validation for Finished Products – Information and data to be Provided in Regulatory Submissions 2016). The EMA have followed up with the publication of drafts of EU GMP Guideline Annex 11 Computerized Systems, Annex 22 Artificial Intelligence and Chapter 4 Documentation in November 2025.

The European Pharmacopeia published a technical chapter 5.28 Multivariate Statistical Process Control in October 2020, whereas the USP cited the possible application of MSPC in their 2020 general information chapter <1039> Chemometrics. This is a good start.

Conclusions

This article is meant to be forward looking. In sterile product manufacturing sites, environmental monitoring represents the largest volume of microbial testing. Collecting data and not converting it to meaningful information is a wasteful loss of an opportunity. The data collection and analysis tools are available and not using them can be viewed as a failure in our obligation to the patient. The objective of this review is to encourage the implementation of the so-called 4.0 pharmaceutical manufacturing as it related to microbiology. As with most microbiologists, the author lacks a mathematical mindset to discuss multivariant statistical process control at any depth, so this area will be left to others in statistics, instrumentation and process engineering to develop.

Acknowledgements

Thanks to Jeffrey Weber from Verista Consulting for reviewing the article and making useful comments.

Subscribe to our e-Newsletters
Stay up to date with the latest news, articles, and events. Plus, get special
offers from American Pharmaceutical Review delivered to your inbox!
Sign up now!