Adoption of FMEA for Microbiological Contamination Risk Assessment to Implement USP Chapter <1115>

Abstract

In December 2014 the new chapter <1115> of the U.S. Pharmacopeial Convention came into effect. The new chapter deals with microbiological contamination risk control for nonsterile product manufacturing. To follow the recommended structured risk analyses for microbiological contamination control, a firm rating system for the three FMEA parameters severity, occurrence and detectability was developed and applied over several nonsterile drug production areas. The rating system provides an objective risk evaluation for every process step. In a first step, an Ishikawa diagram was prepared and process charts were created for each individual process. The single process steps were then assessed using the FMEA approach. The application of the FMEA provides a quantitative risk assessment and therefore the advantage of risk prioritization. Limits were defined rationally to classify process steps in risk categories and to determine whether risk-reducing measures should be implemented. Furthermore, the use of an Improvement Index and a Cost-Benefit Ratio of pre-evaluated measures provide a good aid for decision-making in terms of which measure should be implemented when reducing contamination risk while taking cost factors into consideration. The template developed here was used to assess the microbiological contamination risk for the production of a highly active drug product.

Introduction

In December 2014 a new chapter of the U.S. Pharmacopeial Convention came into effect. USP chapter <1115> covers microbiological contamination risk control for nonsterile product manufacturing. Riskbased microbiological contamination management is demanded in order to implement adequate controls in nonsterile drug production. The risk to patients based on product attributes, route of administration and target patient population should be considered as well as the manufacturing process itself and materials affected.

USP chapter <1115> provides a list of nonsterile pharmaceutical product categories ordered according to their risk of microbiological contamination. Inhalants and nasal sprays are named as the highest risk product categories. Oral tablets and powder-filled capsules hold the lowest risk of microbiological contamination.

The main sources of microbiological contamination in nonsterile drug products are water (as an ingredient), the pharmaceutical ingredients, the process equipment, the manufacturing personnel and the manufacturing environment.

For nonsterile drug product manufacturing, USP chapter <1115> recommends good microbiological quality of pharmaceutical ingredients and excipients and the corresponding identification of suitable suppliers. Furthermore a microbiological risk assessment of the manufacturing process and packaging system should result in the implementation of a risk based monitoring and control system. Also the case-to-case identification of contamination risk for final drug product and harm to patient should be included.

In the current article, the application of failure mode and effect analysis (FMEA) is presented to evaluate the microbiological contamination risk for nonsterile manufacturing according to USP <1115>. The aim was to develop a system that can be used for different manufacturing processes and in particular to define clear criteria to rate the different parameters objectively. The result of this approach is presented herein and its application when manufacturing a highly active film coated tablet.

Implementation of Microbiological Contamination Risk Assessment

To implement a risk-based microbiological contamination control program in nonsterile drug product manufacturing, the failure mode and effect analysis (FMEA) approach was applied. The considerable advantages of FMEA include that it is easy to carry out and that the quantitative risk assessment of detailed steps allows precise prioritization and pre-evaluation of contamination risk mitigation measures.¹

Preliminary Work (Ishikawa Diagram, Process Charts)

Before evaluating the risk factors on the process with an FMEA it is necessary to list extensively all potential microbiological contamination risks. To this end, an Ishikawa diagram, invented by Kaoru Ishikawa, was generated with the outcome titled: “Microbiological contamination risk of drug product” and all the influencing factors collected from brainstorming and research.² In the Ishikawa diagram, factors like Water, Facility, Lock in Process were sorted by the main areas “Personnel”, “Production Process”, “Environment”, “Material & Equipment”, “Cleaning & Disinfection” and “Airlocks & Room Concept” (Figure 1).

Figure 1. Ishikawa diagram to summarize all the possible factors that can affect the microbiological contamination risk of a drug product.

This structured overview of all the influencing factors is helpful to break down the processes and factors into detailed steps. Process charts for each factor help to clearly represent the process factors and therefore provide an easy transition to the next stages,³ which is the FMEA in this case. In Figure 2 the process charts developed are summarized.

Figure 2. Summary of the process charts derived from the Ishikawa of Figure 1. Abbreviations: pharm. = pharmaceutical, C&D = Cleaning & Disinfection, Interm. = Intermediate

In the following paragraph, three different factors are described in more detail as an example of the process charts.

When manufacturing drug products a major aspect of the contamination risk is human-related. So the first process chart “Personnel” gives personnel-related contamination risk factors (Figure 2, violet process chart). In manufacturing areas the topics listed are of importance. Training of personnel and their hygiene-related behavior must be evaluated. This applies to gowning, i.e. what kind of gowning material is used and the gowning procedure itself. Finally the type of tasks that a person executes (i.e. this relates to movement and the risk of microbial contamination that increases as a result) and the number of people present also have an influence on the microbiological contamination risk for the manufacturing process. It must be mentioned that there is no overall hierarchical order for the factors listed; therefore a risk analysis gives more information on individual risk.

Another important factor is “Cleaning & Disinfection”. Due to the different levels of interaction with the drug product, this item was broken down into cleaning and disinfection of equipment and cleaning and disinfection of rooms. Because of the higher microbiological contamination risk involved, only “Cleaning & Disinfection of Equipment” is further discussed (Figure 2, orange process chart). The most important aspect of the cleaning procedures of equipment and facilities is cleaning validation, which should show the reproducible appropriate cleaning of the equipment. In Figure 2 the cleaning process is separated into single steps which can then be assessed. Finally, the reassembling of the cleaned equipment and its storage or hold time can have an important effect on the microbiological contamination risk and must be taken into consideration.

For “Production and Packaging”, every process step from dispensing ingredients, mixing and process-related steps such as compressing, to packaging steps are important for microbiological product quality (see Figure 2, red process chart). In the overall risk assessment, intermediate storage or holding times, in-process control and campaign duration must be considered as additional factors.

Preparation of FMEA

The actual preparation of the microbiological FMEA was performed with a table template using Microsoft Excel software (Figure 3). This table consists of nine basic columns for process step, potential failure, effect (influence of failure), severity SEV, potential root cause, occurrence OCC, current control, detectability DET and risk priority number RPN. For each risk component (severity, occurrence and detectability), adding a column with enough space for a description of the component to provide better traceability of the assessment is recommended, especially if the FMEA is to be reviewed after some years or presented at an audit or inspection. For a better overview it could be started with an additional column for the area of the individual process steps.

Figure 3. FMEA for risk assessment showing the different columns. Abbreviations: SEV = Severity, OCC = Occurrence, DET = Detectability, RPN = Risk Priority Number

To start the FMEA, all the process steps defined from the process charts are added to the Excel table. The potential failures are described for each process step (column “Potential failure”). The next step is to describe the influence(s) of the listed failures. Based on this, the severity is assessed. To assess the risk component occurrence, the potential root cause and its likelihood of occurring must be evaluated next. As a third step, detectability is assessed based on the current controls.

In FMEA the risk components severity, occurrence and detectability are considered to quantify the risk of a specific failure and its effect. The risk priority number RPN functions as quantification of the risk and it is calculated by multiplying the severity SEV, occurrence OCC and detectability DET values. There are plenty of variants, but in this approach, single values are used as in Zimmermann 2011.⁴ The Risk Priority Number is calculated by multiplying each of the three risk factors: RPN = SEV*OCC*DET without additional factors.

Figure 4. Assessment for Product & Patient Risk to be added to the severity.

Figure 5. Assessment for Severity for Process Steps.

The assessment and quantitative rating of the components increases with higher risk relevance. In the present approach, values between 1 and 10 were used for rating. It is challenging to rate risk components objectively without a fixed pattern. Therefore firm assessment tables on the basis of the main influences on the risk components were created to enable an objective and overall consistent assessment of the process steps.

Calculation of Severity

To start the FMEA, the severity of a potential failure is assessed. In the case presented this calculation was broken down into two steps. Because the patient risk of microbiological contamination in nonsterile drug products is more differentiated than in sterile drugs,⁵ product dependent factors must be considered. First, the risk for the product and patient was calculated (Product & Patient Risk, Figure 4) by examining the three main characteristics of the drug product: water activity, which has a strong influence on the microbiological proliferation in drug product, dosage form and route of administration, which both have an important impact on the risk of infection for the patient. The sum of these three values is formed and could be in the range from 0 to 6. This number is then integrated in the following severity assessment (Figure 5 and 6). Second, to assess severity, the potential failure was split into two factors: Process Step (Figure 5) and Ingredients (Figure 6). For both assessments the Product & Patient Risk from table 1 that was previously rated (Figure 4) is included in the first row. The second block represents the basic assessment. For Process Steps the product contact is considered, for Ingredients the amount of drug product. Finally, to get higher values for a higher microbiological risk, influencing factors such as humidity of product/ingredient respectively of the process step or environment and other growth promoting properties can be added to the severity rating in the third block. The sum of all the values represents the SEV value, which is entered in the FMEA table mentioned above. For values under 1, the rating is set to 1 (lowest risk), for values over 10, the rating is set to 10 (highest risk).

Figure 6. Assessment for Severity for Ingredients.

Figure 7. Assessment of Occurence.

Figure 8. Assessment of Detectability.

Calculation of Occurrence (OCC) and Detectability (DET)

After defining the potential root cause, its likelihood of occurring is rated under occurrence (OCC) with the assessment shown in Figure 7. As for the severity assessment (SEV), a basic assessment must also be done for occurrence in the first block (Figure 8) that reflects the probability of its occurrence. Similarly, additional contributing factors such as human or technical errors are added and the final OCC rating is calculated.

In the next step, the detectability (DET), i.e. probability to detect a failure, is assessed with the assessment Figure 8 based on the current controls. Again a basic and an additional assessment are performed. The rating increases with a decreasing probability of detecting the failure. The basic assessment reviews the product testing frequency, while the additional factors increase the probability of detection and therefore reduce the detectability value by subtraction.

For both risk components OCC and DET, similar to SEV, the rating is the sum of all the values and must be between 1 and 10.

Risk Ranking by Risk Priority Number (RPN)

As mentioned above, all the values calculated for SEV, OCC and DET are added to the FMEA Excel table and the Risk Priority Number RPN is calculated. To classify the risk by the RPN value, limits must be set. The present approach distinguishes between critical control points, moderate critical and non-critical.

Due to the multiplication of SEV, OCC and DET without additional coefficients, the RPN range is between 1 and 1000. It was pointed out by Shebl et al. ⁶ that, especially for higher rates, not all values in the ranking could be obtained. For example, the next lower value to 1000 is 900. Thus there is no linear correlation of the values and therefore an arithmetic mean cannot be used. A mean rating of all the risk components (in the current case this would result in an RPN of 5 x 5 x 5 = 125) could be used but there are no regulations on the limit definition.⁷ By increasing or decreasing one of the risk components by 1, the limits were set to 5 x 5 x 4 = 100 and 5 x 5 x 6 = 150 and an appropriate classification was created (Figure 9).

According to the fragmentation described, all RPN are classified as critical control point, moderate critical or non-critical. For critical control points, where RPN is above or equal to 150, mitigation actions will be undertaken and implemented. The aim is to reduce the risk of all critical control points so that after implementation of actions the RPNs are lower than 100, but maximum 149. RPN between 100 and 149 are classified as moderate critical. For all moderate critical steps mitigation actions are recommended, but the prospectively analysis of risk reduction and costs (see next section) is important to decide for implementation or to accept risk without actions. In this case it must be considered, that risk based monitoring activities are good choices of mitigation actions concerning moderate critical steps by decrease of detectability. RPN fewer than 100 are classified as noncritical. In these cases no mitigation actions must be considered. There is no need of risk reduction for non-critical steps. For some non-critical steps it is even possible with prospective recalculation to reduce monitoring activity.

Figure 9. Summary of the three classes for evaluating the criticality of the RPN (Risk Priority Number).

Choice of Good Measures

As mentioned above, for critical control points risk-reducing measures or actions must be initiated. To identify the potential of improvement, a pre-evaluation of the planned measures by recalculating the RPN after implementation is important. The documentation should be entered directly into the FMEA table (Figure 10).

Figure 10. FMEA for the evaluation or risk reducing measures in addition to the columns in Figure 3. Abbreviations: SEV = Severity, OCC = Occurrence, DET = Detectability, RPN = Risk Priority Number, Imp I = Improvement Index, CBR = Cost- Benefit Ratio

When choosing between measures, the Improvement Index (Imp I) is helpful, which shows the improvement after implementation of the planned measures.⁸ The following formula is used: Imp I = RPN_before : RPN_after

Based on the Improvement Index, the measure with the higher rate should be preferred. In many cases it is not sufficient only to calculate the Improvement Index for measures under consideration. By only considering Improvement Index, the best measure would be nearly aseptic production even for oral dosage forms. But it is neither in terms of patient nor of economic feasibility to initiate measures where the ratio of the costs or investment is disproportionate to the improvement. Especially for moderate critical process steps or for measures under consideration that demand a substantial investment, the costs as opposed to the benefits must be considered. Therefore, in this application the Cost-Benefit Ratio (CBR) was calculated using the following formula: CBR = Imp I : Investment

The Investment used was valued between 1 and 3 for an easy assessment and includes all the necessary investment factors, such as working time and costs. The value 1 would be a low cost, 2 moderate and 3 high cost. It is possible to use a bigger range, but in the approach presented here this easy range was effective when deciding between measures.

Pilot Project for Implementation of FMEA in Nonsterile Drug Production

To implement the FMEA in nonsterile drug production, a pilot project was chosen within a production area for a highly active film coated tablet. Because of security priorities like negative pressure of production rooms compared to the environment to keep the dust from active ingredients inside the area, there was conflict with preventive measures for microbiological environment control such as overpressure in the production area to the adjacent environment.

Using the FMEA approach described above, a total of 113 single process steps were evaluated. On the basis of the quantitative evaluation of microbiological contamination risk, all the process steps could be prioritized by their RPN. Overall 12 critical control points, 12 moderate critical and 89 non-critical process steps were identified (Figure 11). After implementing appropriate risk-reducing measures, all critical control points were reduced in risk and resulted in non-critical process steps. For 9 of the moderate critical process steps, risk-reducing measures were implemented, for 3 moderate critical process steps no measures were initiated due to the inappropriately high investment for the potential improvement.

Figure 11. Overview of evaluated RPN (Risk Priority Number) for the manufacturing of a high potent drug product.

To illustrate each of these three scenarios, one example is given:

Reduction of critical control point to non-critical process step:

No written procedure of disinfection was recorded for material entering the production area. In line with the worst-case scenario that incoming material is never disinfected, the RPN value was 320. The disinfection procedure was integrated in the SOP for the production area and the personnel was trained accordingly. The implementation of the measure resulted in a new RPN with a value of 64, simply by reducing the possibility of the failure to occur and thereby reducing the OCC risk factor. With the Imp I value 5 and little effort (Value: 1), this resulted in a CBR of 5.

Reduction of moderate critical to non-critical process step:

A potential failure that was evaluated as moderate critical was the inadequate disinfection of material in contact with the product, resulting in a risk of contamination through water-borne microorganisms. This potential failure was assessed with a RPN value of 100. Thanks to a low effort measure, the RPN could be reduced to 50 (non-critical), after a reduction in the OCC. This was made possible by classroom training for all production employees on microbiological topics overall and especially about the disinfection of surfaces in contact with the product. In addition a QA-Oversight was implemented which increases the efficacy of the training and supports continuous improvement. The Imp I, which at 2 was moderate, could be compensated with very little effort with a value of 1. The resulting CBR at 2 was at a good level.

Retain moderate critical process step without actions or measures:

One of three moderate critical process steps, where the risk was accepted and therefore no action or measure was initiated, was weighing of the excipients. The excipients were sucked in with a lance out of an open barrel. The microbiological contamination risk was evaluated with a RPN of 100. The air in the production room was already under microbiological monitoring; therefore a problem with microbiological contamination of the air would lead to further product examination. Because this RPN is at the lower end of moderate critical process steps, and because of the inappropriate high cost of risk reducing measures (i.e. a high CBR was found), the risk was accepted.

Summary and Outlook

In USP <1115> no method for microbiological contamination risk analysis is dictated but the HACCP is suggested. Thanks to the numerical ranking to support risk prioritization and the possibility to integrate the product and patient risk directly and easily into the assessment, the FMEA approach was chosen instead of the suggested HACCP. As mentioned by Sandle⁷ and Shebl et al., ⁶ both methods are commonly used and have a lot of similarities (e.g. process charts, failure analysis). In contrast to the HACCP, the FMEA has a clear numerical ranking which supports the prioritization of failures.⁴

For the present study, a highly objective and well-documented FMEA approach was developed. After summarizing all the production steps, their severity, occurrence and detectability was calculated using clearly pre-defined assessment tables. With this – regardless of who is performing the risk assessment or which production process is being assessed – a very objective RPN number is calculated and prioritization of risk-reducing steps can be implemented.

With the help of the FMEA presented here, the potential failures of process steps in a highly active production area could be tackled in a risk prioritized manner. Overall, a big improvement in the microbiological contamination risk control in this pilot production area was achieved and the sampling plans for microbiological environmental monitoring were adjusted based on the microbiological contamination risk values evaluated by the FMEA. By doing so, at least certain sampling points could be reduced.

The resulting FMEA can be used to retrospectively analyze the microbiological contamination risk in existing drug production areas or to evaluate the microbiological contamination risk of drug production areas in advance (e.g. design phase). This is also mentioned by Shebl et al., 6 where the clear advantage of using FMEA in advance to reduce any later contamination risks can be seen. Furthermore, the system developed here can easily be adapted for each nonsterile production process, especially since the process is highly objective, resulting in the possibility of comparing different production processes or even different production sites.

Finally, applying a risk-based approach, forces the user to consider the entire process at a high level of detail. This allows a much better understanding of the process to be gained and further improvements can be implemented. In the present study, we found, among others, weak points in regard to personnel security that could be improved. Since an FMEA needs to be performed in a multidisciplinary team, the common purpose brings with it many advantages such as better and easier communication and better understanding of the problems and point of view of the other field of activity.

In compliance with USP chapter <1115> after numeric evaluation of microbiological contamination risk and prioritization, the FMEA permits practicable management of risk-reducing actions or measures. The suggested actions or measures could be integrated in the assessment table and then by initiating the measures the reduction in risk could be evaluated.

The developed firm assessment charts provide an objective evaluation of the risk factors of potential failures and are applicable to all nonsterile drug production areas and processes. By defining RPN limits for critical control points, moderate critical and non-critical process steps, riskbased conditions for measures were provided.

By evaluating the RPN with measures, before initiating the riskreducing measures and evaluating the costs involved, the Cost Benefit Ratio (CBR) could be used for management of risk-reducing measures. This proactive evaluation provides the possibility of the appropriate choice of measures in the case of critical control points and furthermore simplifies whether or not to initiate a measure in the case of moderate critical process steps. In the example presented here (see example C above), the contamination risk during transfer of the excipient to the mixing process in the isolator could be reduced with new equipment. The equipment is expensive and qualification and validation require a great deal of work. In this case, the financial investment outweighs the microbial risk reduction, i.e. the CBR is low.

Although the FMEA approach for the evaluation of the microbiological risk of nonsterile production process presented here is very objective and detailed, it is also quite time consuming. The development of the different risk assessment charts (Figures 4 – 8) in particular needed a lot of discussion. However, once the tables are developed they can be used for different (maybe all) other processes. Furthermore, the required level of detail should be defined before carrying out the FMEA. For general risk evaluation of various production areas, it is recommended to combine similar process steps. For a retrospective risk evaluation in a production area with microbiological problems, a higher level of detail could help to identify potential weak points and improve the microbiological situation based on the individual risk of the single process steps.

Conclusion

Carrying out an FMEA is a very labor-intensive but also very objective way to assess microbiological contamination risks. However, it must be decided case by case which level of detail is to be addressed. In certain cases, e.g. where microbiological problems can be expected, a higher degree of detail makes sense. Whereas in very easy processes (e.g. non-highly active, oral dosage forms with low water activity and antimicrobial process steps) a less detailed level can be chosen. The most important aspect for us was the objectivity of the rating. For this we developed assessment charts where the risk evaluation for microbiological contaminations in nonsterile drug production is objective for a consistent assessment of all the process steps over different production areas. By pre-evaluating risk-reducing measures, the benefit can be calculated using the Imp I Improvement Index. The Cost-Benefit Ratio CBR allows decision-making from an economic point of view. Finally the application of the FMEA shown here is very well documented and therefore the decisions taken can be followed even after an interval of some years. Thus it provides complete traceability and is especially suitable for inspection or audit presentations.

Acknowledgments

We thank Ylber Qusaj, Sascha Gasser and Alexander Schröder for their technical support, Ania Dardas for proofreading.

Conflict of Interest Declaration

The authors declare that they have no competing interests.

References

1. ICH Q9, 2005. Harmonised Tripartite Guideline: Quality Risk Management. Available on http://www.ich.org.
2. Wong, K. C., 2011. Using an Ishikawa diagram as a tool to assist memory and retrieval of relevant medical cases from the medical literature. Journal of Medical Case Reports 5, 1-3.
3. Carducci, A., Davini, G. and Ceccanti, S., 2013. The Application of Quality Risk Management to the Bacterial Endotoxins Test: Use of Hazard Analysis and Critical Control Points. PDA Journal of Pharmaceutical Science and Technology 67, 553-567.
4. Zimmermann, H. F., Hentschel, N., 2011. Proposal on How To Conduct a Biopharmaceutical Process Failure Mode and Effect Analysis (FMEA) as Risk Assessment Tool. PDA Journal of Pharmaceutical Science and Technology 65, 506-512.
5. Guilfoyle, D.E., Friedman, R.L., Hughes, P.F. Hussong, D., Rosenberg, A.S., Brorson, K., 2013, Microbial Risk in Pharmaceutical Manufacturing and ICH Q9. PDA Journal of Pharmaceutical Science and Technology 67, 79-80.
6. Shebl, N.A., F., Franklin, B.D., Barber, N., 2012. Failure mode and effects analysis outputs: are they valid? BMC Health Services Research 12, 150.
7. Sandle, T., 2006. Environmental Monitoring Risk Assessment. Journal of GXP Compliance 54-73.
8. Van Leeuwen, J.F., Nauta, M.J., de Kaste, D., Odekerken-Rombouts, Y.M.C.F., Oldenhof, M.T., Vredenbregt, M.J., Barends, D.M., 2009. Risk analysis by FMEA as an element of analytical validation. Journal of Pharmaceutical and Biomedical Analysis 50, 1085-1087.