Packing-Line Improvement Based on a Fault-Tree Analysis Approach

By Arturo Toledo Rivero,Nelson Sierra Prado,Yohann Pérez Molina,Ian Toledo de Zayas

This article focuses on upgrading and improving a packing process to comply with current good manufacturing practices. The authors sought to maintain proper quality assurance for finished products.

Packaging is the final step in obtaining a finished pharmaceutical product. It is important because it ensures product integrity, identification, and presentation, including primary information for the patient (1).
For this reason, the packaging process should be analyzed, and manufacturers should take into account the occurrence of deviations such as equipment and component failures, human error, operational errors, and other deviations that negatively affect the final process result (2–4). Risk analysis (RA) methods are valuable tools for mitigating fault events by focusing on the cause–effect interrelations that create them. RA methods thus facilitate fault elimination or, when faults cannot be eliminated, fault reduction.
In today's pharmaceutical industry, most processes are conceived and designed with a high level of automation to minimize operator intervention. Nevertheless, economic considerations lead companies to retain manual operations in certain situations. The packing process that is the subject of this study is one example. In such a process, a fault-tree analysis (FTA) approach can greatly help determine critical process points. The approach also helps manufacturers introduce barriers against equipment and operator failure and minimize their probability of occurring (5, 6). Following these considerations, the packing line was arranged, bearing in mind the potential failures to which the final result could be exposed. The arrangement was validated by challenging the process to demonstrate its effectiveness.
Packing process description and considerations

Figure 1: The packing process represented in a block diagram. QA is quality assurance.

The packing process in general can be summarized in a block diagram (see Figure 1). According to in-house procedures, the quality assurance (QA) department must verify the information from quarantined product and the corresponding packing material for approval. A product-release document is therefore indispensable to starting any product packaging activity. A packing-order document is created upon product release. This document matches the product vials and packing material that are involved in the process. Refrigerated product vials must pass a room-temperature acclimatization stage to eliminate external moisture on the vials before they are labeled. All other product vials are directly transferred together with label rolls for automatic labeling and imprinting. Printed packing material such as labels for multiple unit boxes (MUBs) and shipping boxes, as well as cartons and leaflets, are semiautomatically imprinted, folded, identified, and prepared as needed. These items are then temporarily stored together with the rest of the materials until vial labeling and manual packaging begins.

Figure 2: Packing area layout. MUB is multiple unit box.

The original packing-area distribution was modified to physically separate the preparation of packing material from the labeling, imprinting, and manual packaging operations of the main packing line (see Figure 2). Note that the contiguous packing line included a Sensitive 350 labeling machine (Libra Pharmaceutical Technologies, Fairfield, CT) followed by a 6.65-m long conveyor with speed control to transport labeled vials to manual packaging operators. According to the original arrangement, one of the operators near the labeling machine (No. 8) acts as a buffer by collecting vials and returning them to the conveyor in groups of five. The other operator next to the labeling machine (No. 1) simultaneously configures and distributes empty MUBs using the same conveyor. The rest of the operators (Nos. 2–7) package product vials in individual cartons, including leaflets, placing the packed units into the MUBs in groups of 10.

Figure 3: Packing-process fault tree showing the main start section and the major unwanted event.

A BP 4100 technical balance (Sartorius, Göttingen, Germany) with a 0–4100-g range and a resolution of ± 0.1 g is located at the end of the line. An operator places each MUB and reads its weight to verify that it has been filled. The MUB is then sealed, and the corresponding label is affixed. Shipping boxes are also checked in the same way using a QC1500NNP scale (Sartorius ) with 0–1500-kg range and ± 200-g resolution. Shipping boxes are then sealed and labeled accordingly.

Figure 4: Packing-process fault tree showing the branch that follows from product temperature-requirement failure.

This arrangement is designed to ensure good coordination among all operators regarding manual performance. The time each packing operator took to complete an MUB, including visual inspection, was measured as 160–170 s, with a margin of 10 s for recovery. The conveyor was therefore set to a linear velocity of 900 cm/min. The labeling machine had an optimum working speed of 45–47 vials/min. Hence, the buffer operator has an important task, which is to control vial distribution through the conveyor at intervals of 12–13 s. The buffer operator avoids bottlenecks by gathering the overflow of labeled vials to operators, and releasing the collected vials when the labeling machine is interrupted. Fault-tree analysis of the packing process

Figure 5: Packing-process fault tree showing the branch following from failure in primary-information imprinting.

An FTA model was specially developed for the packing process by representing failure events with logical operators to form a logical information flow chart (see Figure 3). The special symbols employed in this flow chart are explained in the literature (7). The major unwanted event was defined as "Finished-product quality noncompliance resulting from packing process." All cause–effect fault events flow downward. Any of the following failures can provoke the major unwanted event:

• Failure in product temperature requirements during the packing process (which exclusively applies to products that require refrigeration)
• Failure of packing material to display the corresponding primary product information (e.g., batch manufacturing date, batch number or identification code, and batch expiration date)
• Lack of correspondence between product specifications, printed packing material (e.g., vial labels, cartons, leaflets, MUBs, and shipping labels), and documentation (e.g., QA release authorizations, packing order, and records created during packaging)
• Failure to comply with cleaning and line-clearance requirements for areas involved in the packing process, including labeling machine and all workstations, to avoid product-batch cross contamination.

Figure 6: Packing-process fault tree showing the branch following from correspondence failure. QA is quality assurance. MUB is multiple unit box.

The fault events that cause these four failures were identified, and their cause–effect connections were traced until they reached the basic fault events (see Figures 4–7). Note the use of conditional events, which represent process controls that restrict fault-event occurrence. Neglecting process controls exposes the process to failures stemming from basic events.

Figure 7: Packing-process fault tree showing the branch following from cleaning and line-clearance failure.

Noncompliance with product-temperature requirements is considered the most critical of all possible process failures because it can cause premature product deterioration. Because process history shows that these failures occur with a relatively low frequency, the following actions should be sufficiently effective:

• Ensure cold-chamber qualification on at least a yearly basis to demonstrate consistency in keeping product temperature between 2 and 8 °C. Perform a similar study in the acclimatization room.
• Establish a metrological assurance plan, created by the QA department, to guarantee verification and adjustment of cold-chamber temperature controllers on at least a yearly basis.
• Monitor cold chambers daily. Monitor acclimatization- and packing-room temperatures linked to standby and transfer follow-up of each product batch. Standard operational procedures (SOPs) and supervision must be implemented accordingly.
• Implement an SOP to mitigate possible interruptions during the packaging of refrigerated products, and ensure that the quality of these products is preserved.

Failures in the printing of primary information and in the correspondence between products, printed packing material, and documentation are considered less critical because they have less impact on final product quality. Insufficient workstation cleaning and line clearance are also considered less critical failures. All of these faults occur more frequently, however, because of human error. The authors decided to create barriers in the packing-line configuration illustrated in Figure 2 for detecting and eliminating faults throughout the process. The barriers are as follows:

• A first level to detect and eliminate possible faults in product-vial labeling and primary-information imprinting in which the labeling-machine automatically senses and rejects unlabeled vials or nonimprinted vial labels.
• A second level to detect and eliminate possible faults in product-vial labeling and primary-information imprinting in which a buffer operator visually inspects and rejects unlabeled vials, nonimprinted vials, badly pasted or wrinkled labels, and illegible, misplaced, and erroneous imprints.
• A third level to detect and eliminate possible faults in product-vial labeling and primary-information imprinting (also for previously prepared printed packing material) in which packing operators visually inspect and reject unlabeled vials, nonimprinted vials, badly pasted or wrinkled labels, and labeled vials or cartons with illegible, misplaced, or erroneous imprints.
• A fourth level to check for MUB completion in which the first end-of-line operator looks for missing packed units, leaflets, and vials by weighing. The second end-of-line operator performs visual inspection to verify proper MUB manual sealing and correct primary-information printing on the corresponding label.
• A fifth level to check for shipping-box completion in which a handler operator looks for missing multiple boxes by weighing and performs visual inspection to verify proper shipping-box sealing and correct primary information printing on the corresponding label.
• A documented, in-process control implemented at intervals by a supervisor who visually inspects printed packing materials and their primary information. The supervisor also checks the appearance and completion of packed units by taking MUB samples from the end of the line. A similar in-process control is also implemented for previously printed packing material during the imprinting of cartons, MUBs, and shipping-box labels.
• Documented control performed by the supervisor to ensure the cleaning and line clearance for all involved areas at the beginning and end of each process.

Validation of proposed packing-line configuration and organization

Table I: Technical balance qualification for checking multiple-unit boxes (MUB) completion by weighing.

Qualifications of key process components were carried out. The vial-labeling machine was qualified according to the protocol designed for that purpose (8). The technical balance used to checking MUB completion by weighing was calibrated and verified according to in-house metrological procedures. In addition, the balance's ability to detect missing components was also qualified, taking into account MUB weight variations caused by the weights of individual components. The technical balance's sensitivity for reproducibly detecting single leaflets missing from MUBs leaving the packaging line was tested. The results of the testing are shown in Table I. Six MUBs were taken from the end-of-line at different moments (i.e., the start, middle, and end times) during the packing process. Weight differences were determined by removing a single leaflet from each MUB. When the weight differences were compared statistically with the corresponding leaflet weights, they did not vary significantly (9). The packing line was then validated to determine the buffer and packing operators' ability to detect and correct the failures described previously. Three consecutive runs were performed for this purpose by simulating automatic labeling and manual packaging operations in each run on 1440 placebo vials with printed packing material (i.e., vial labels, cartons, and leaflets for testing). Labels for MUB and shipping boxes were not used. To challenge the operators, vials were introduced during each run that had been deliberately prepared with defects. Among them were 40 unlabeled vials, 40 vials with badly pasted or wrinkled labels, and 45 vials with primary information missing from their labels, making a total of 125 faulty vials to detect.
To establish an acceptance criterion in each run, a single sampling plan for normal inspection was employed based on a general inspection level II and 0.10 acceptance-quality limit index (10). It was established that if any of the deliberately defective vials were not detected during a run, the packing-line performance would be considered unsatisfactory. The test also required the detection of all other faulty vials that resulted from errors in labeling-machine operation.
The packing operators' ability to ensure MUB completion was simultaneously verified by weighing MUBs at the end-of-line workstation. Based on the above qualification, an operational procedure determined weight limits at the beginning of each packing operation by averaging the weight of the first three complete MUBs leaving the line. A tolerance of ±1.8 g (i.e., the weight of the lightest leaflet) was established.
Results and analysis

Table II: Results from packing-line simulation runs: Visual checking of defects by operators.

All runs demonstrated that operator performance was good and that the barriers implemented against failure were effective. All 125 of the faulty vials deliberately prepared and introduced into the packing line as failures were detected (see Table II). The buffer operator rejected 73.6–94.4% of the faulty vials. The packing operators rejected the rest. In the worst case, 33 faulty vials reached them. All of the nonprogrammed faulty vials were also detected and rejected. The vials' defects (e.g., labels with tiny wrinkles and labels slightly out of their intended position) were almost imperceptible to the untrained eye. These defects were evidently difficult for the buffer operator to perceive. This operator had intervals of 13 s maximum to inspect vials between each distribution to the packing operators. The labeling machine functioned at a pace of 46 vials/min. The six packing operators had no problem perceiving and eliminating the faulty vials that the buffer operator had missed.

Table III: Results from packing-line simulation runs: Checking multiple-unit box (MUB) completion by weighing.

The packing operator's performance ensuring MUB completion was verified by reading the weight of all MUBs at the end-of-line station. Only one MUB with a weight equal to the lower limit was detected and set aside during the first run. Further visual inspection by the supervisor showed that no components were missing. No out-of-limit MUBs were found during the other runs (see Table III). This result was corroborated by the in-process control documented by the supervisor using the inspected MUBs taken from the end of the line. All packaging runs, including shipping-box completion for market distribution, were thus considered successful.

Figure 8: Diagram of a workstation fitted out for packing operators.

A further operational improvement of the above results can be achieved by reducing the time invested in manual packaging of vials, based on the principle of economy of operator movement. Economy of movement is increased when operators use both hands to pack two vials simultaneously. The time needed to complete MUBs could be reduced accordingly, thereby improving the balance between the manual processing rate of the overall packing line and the optimal working regime of the labeling machine. The amount of vials gathered and assimilated by the buffer operator could be minimized. For this purpose, packing-operator workstations should be fitted out according to Figure 8. The conditions shown can facilitate the use of both hands for packaging manipulation. Conclusion
The experiment demonstrated that a risk-analysis approach based on a fault-tree analysis model of the packing process achieved upgrades and improvements. Packaging operations were arranged according to the analysis. The new arrangement resulted in a product that was consistently packed and identified according to established requirements and current regulations.
Validation demonstrated the effectiveness of barriers against failures, which were implemented throughout the packing line, in detecting and rejecting faulty vials. The barriers detected faulty vials deliberately included in the tests and also the nonprogrammed faulty vials resulting from labeling-machine operation failures. The barriers did not compromise the correct completion of multiple boxes, which reached 100%. The system therefore ensures an exceptionally low probability of product failure and unacceptable defects at the process outlet.
This level of assurance can, in principle, significantly reduce customer complaints. It can also reduce the need for product recalls in the worst case, which normally cause great losses for each batch withdrawn from the market.
Authors' note
A colleague of the authors suggested a test of the end-of-line weighing station during packing-process validation by moving several MUBs along the line with one leaflet missing from each. The authors stand by the original concept followed in this study. Key process components must be qualified before the packing line is validated as a whole. The process is validated by integrating all qualified components and simulating normal process function as closely as possible without compromising the regular course of the process by adding an extra challenge. The authors believe that the technical balance's sensitivity for detecting missing components should be determined in advance. Then, one can focus on the objective of process validation, which is to evaluate the packing operator's ability to deliver completed MUBs throughout the simulation runs. Nevertheless, the above suggestion could be analyzed in a further revalidation of justified in-process changes.
Arturo Toledo Rivero* is the head of the research and development department, Nelson Sierra Prado is the head of the validation group, and Yohann Pérez Molina is a quality engineering specialist at LIORAD Laboratories, Ave. 27A No. 26402, La Lisa, Havana, Cuba, tel. 1537 2717935, fax 1537 2717899, atliorad@infomed.sld.cu [atliorad@infomed.sld.cu]
Ian Toledo de Zayas is a logistics specialist at DUJO Business Group.
*To whom all correspondence should be addressed.
Submitted: Aug. 8, 2007. Accepted: Nov. 1, 2007.
References
1. European Commission, "Production", in Volume 4—Medicinal Products for Human and Veterinary Use: Good Manufacturing Practice, (European Commission, Brussels, Belgium, 2005), pp. 47–49.
2. ISPE, "ISPE Baseline Guide—Packaging, Labeling, and Warehousing," (ISPE, Tampa, FL, vol. 7, rev. B, 2005).

Figure 1: The packing process represented in a block diagram. QA is quality assurance.
Figure 2: Packing area layout. MUB is multiple unit box.
Figure 3: Packing-process fault tree showing the main start section and the major unwanted event.
Figure 6: Packing-process fault tree showing the branch following from correspondence failure. QA is quality assurance. MUB is multiple unit box.
Figure 4: Packing-process fault tree showing the branch that follows from product temperature-requirement failure.
Figure 5: Packing-process fault tree showing the branch following from failure in primary-information imprinting.
Figure 7: Packing-process fault tree showing the branch following from cleaning and line-clearance failure.
Table I: Technical balance qualification for checking multiple-unit boxes (MUB) completion by weighing.
Table II: Results from packing-line simulation runs: Visual checking of defects by operators.
Table III: Results from packing-line simulation runs: Checking multiple-unit box (MUB) completion by weighing.
Figure 8: Diagram of a workstation fitted out for packing operators.