Pharmaceutical and Medical Device Manufacturing Case Studies
Industrial5 Case StudiesUpdated Sep 2026
The five cases below come from several fictional sites that make sterile injectables, assembled medical devices, and packaged diagnostic components. Each centers on a decision that comes up on the floor and in the quality office: reading a signal before it becomes a failure, passing work across a shift boundary, rationing scarce capacity, explaining uneven outcomes, and deciding what to tell a customer when the answer is incomplete. Every case is examined through the same lens of design constraints, operating data, quality tolerances, and maintenance records. Keep the site's own procedures open while reading, because the distance between written rules and daily practice is usually where the hard choices sit.
Case Study 1
Rising particle counts and a late escalation decision
Ferncliff Sterile Works, a contract fill-finish site producing single-use parenteral vials across two aseptic suites.
The situation
Ferncliff runs two aseptic filling suites that share one air handling unit. In the second week of a campaign for a parenteral product, the environmental monitoring team began logging non-viable particle counts in Suite B that sat just under the action limit during each fill. Over five consecutive batches the counts climbed slightly, and the differential pressure across the final HEPA filter drifted downward. Ruth, the shift supervisor, read the trend as possible filter loading. Asif, the validation lead, wanted to wait for the next scheduled sampling window. The SOP requires documented assessment when a trend approaches an action level, but it does not say how many consecutive upward points count as a trigger.
The challenge
The team had to decide whether the signal justified stopping the line, pulling a filter for integrity testing, or continuing under added monitoring. Stopping would consume the planned changeover window and push the campaign past its supply commitment. Continuing without action risked a batch failing sterility assurance review after product had already left the building. Nobody could easily see the maintenance history of the filter, which had been replaced during a different maintenance cycle and recorded in a separate system.
Analysis
The first step was to separate the design constraints from the operating picture. Suite B's room classification sets the limit, and the action level sits well inside the alert level, so a value near the action line is a real signal even when no single reading fails. The operating data showed three things at once: airflow velocity stayed within specification, the pressure drop across the final filter rose slowly, and the particle counts tracked the fill head that had run longest in each shift. Taken together, those patterns pointed toward filter loading or a seal bypass rather than a people problem. Reading the signal this way kept the team from treating five small movements as five unrelated events.
Prioritizing investigations meant ranking each action by how quickly it could confirm or rule out a cause. A review of the pressure log took minutes, a smoke study took most of a shift, and an integrity test required the suite to be shut down. The maintenance records added a complication. The last filter change had been performed by a contractor using a leak test method that differed from the site standard, so the earlier baseline was hard to compare. Escalation was not a single yes or no decision. Ruth documented the trend, placed a hold on the next batch release pending assessment, and notified the quality unit the same day, which is what the SOP requires once a trend is judged to approach an action level.
Actions taken
- Pull the differential pressure and particle logs for the last twelve batches and plot them against fill head and shift, so every reviewer sees the same pattern.
- Schedule an integrity test on the final HEPA filter during the planned changeover, and compare the result with the contractor's leak test record before accepting either baseline.
- Add a defined trigger to the SOP, such as three consecutive upward points inside a rolling window, that requires documented assessment before the next fill starts.
- Require the filter maintenance history and last change date to appear on every environmental trend review, so the baseline is never rebuilt from memory.
Results
The integrity test found a bypass at one filter frame seal rather than general loading. The seal was re-gasketed and the suite re-qualified within the planned window, so the campaign shipped on schedule. The batch on hold was released after post-repair monitoring showed counts returning to their prior range over the next two runs. The SOP was revised with an explicit trigger, and in the following quarter two other trends were caught before reaching action levels. The site also found that the contractor's leak test method had never been reconciled with the internal standard, which became its own corrective action.
Questions for discussion
- Which of the four data streams, airflow, pressure drop, particle counts, or maintenance history, would you have weighted most heavily in the second week, and why?
- How should a site decide how many consecutive upward points count as a trigger without creating so many alerts that operators stop responding?
- If the integrity test had been delayed until after the campaign shipped, what evidence would the quality unit need to defend the release decision?
Case Study 2
An open sterilization hold lost at shift change
Tallow Creek Medical Assembly, a contract plant that assembles, seals, and packs single-use catheter kits for two device companies.
The situation
Tallow Creek runs three shifts in its sterile packaging area, where catheter kits are assembled, sealed, and staged for ethylene oxide sterilization. Near the end of a long afternoon, the day packaging lead recorded a seal inspection failure on one pouch lot in a paper log and placed the lot on quality hold. The hold was not entered in the electronic batch record before shift change. The verbal handover covered line speed and a torque issue on one sealer, but not the hold. At 2,400 units, the lot was the largest of the week, and the night team staged it for the next sterilizer load from the batch record status alone.
The challenge
A staging check on the sterilizer loading sheet required a second signature, but that check confirmed lot count and load configuration, not quality status, so it did not catch the hold. The team had to establish how close the lot came to sterilization, whether any units from the same sealer run had already left the building, and why a paper hold could escape the electronic system entirely. Leadership wanted a fix that did not simply tell staff to communicate better.
Analysis
Tracing the handover began with the design of the record system. The batch record showed a lot's status as of its last update, but it had no mandatory field that stopped a lot from moving while a hold was open. Paper logs filled that gap in practice, and no automated check read them. Operating data showed the handover averaged about twelve minutes, covered about six items, and that open holds were the item most often left out when the outgoing lead was tired. Teams grounded in Pharmaceutical and Medical Device Manufacturing tend to map a handover as a chain of information transfers, then ask which link had no second check at all.
Quality tolerances matter here because a hold is a binary state, and a missing hold is not a small error. Maintenance records added a second layer. The sealer that produced the failed pouches had a torque adjustment logged two days earlier, and that entry had never been linked to the seal inspection trend. Each safeguard on paper depended on a person noticing something the system did not show, which made the protection weaker than it appeared. Measurable improvement needed targets that could be audited, not reminders to be careful.
Actions taken
- Make the quality hold a required field in the electronic batch record, so a lot cannot be scheduled for sterilization while a hold is open, with the hold reason visible on the staging screen.
- Adopt a structured handover script with a fixed list of open holds, pending verifications, and equipment adjustments, read aloud and signed by both leads before the outgoing lead leaves.
- Measure handover quality by auditing a sample of twenty handovers each month for omitted holds and adjustments, with a target of no omissions across three consecutive months.
- Link every sealer adjustment in the maintenance log to the seal inspection trend view, so a torque change appears beside any shift in failure rate before the next lot is staged.
Results
The electronic hold field went live in the third week after the event, and the structured handover script was piloted on two lines before wider rollout. In the first month of audits, twenty handovers were sampled and two omitted a sealer adjustment, but none omitted an open hold. The lot was dispositioned as nonconforming and destroyed, with no units shipped. A review of the previous quarter found 14 paper-only holds, which were entered retroactively and reviewed. Handover time rose by about four minutes, which the operations manager accepted as the cost of a complete record.
Questions for discussion
- Which link in the handover chain would you test first, and what result would tell you the paper log was not the main weakness?
- How would you set an audit sample size that is large enough to detect omissions without pulling the outgoing leads away from the line?
- Should a missing hold ever be treated as a personnel matter, or only as a gap in the system, and how would your answer change the corrective action?
Case Study 3
Lyophilizer demand exceeds the cycle schedule
Brackenridge Biologics Fill, a mid-sized contract site running freeze-drying and aseptic fill for biologic drug products.
The situation
Brackenridge operates four lyophilizers on a rotating schedule, each with a fixed number of cycle slots per week. Within three weeks, a commercial partner moved a launch forward and added roughly 9,000 vials to the monthly plan. Two lyophilizers were running at full utilization, one was in a planned condenser service window, and the fourth had a validated recipe that the QC lab could not yet release because of a stability sample backlog. Operators warned that compressing cycles would raise the risk of product temperature excursions. The planning group had about 180 cycle slots available for the month.
The challenge
The planning group needed to choose between extending cycles, adding a shift for the fourth unit, outsourcing part of the freeze-drying to a partner site, or re-sequencing the fill calendar. Each option touched a different constraint: cycle validation, staffing, chain of custody, or the stability program. The decision had to protect the product's critical quality attributes, not simply meet the volume target.
Analysis
Start with the constraints that do not move. Each lyophilizer cycle has a validated load pattern, minimum hold times for primary and secondary drying, and a maximum product temperature the process can tolerate. Operating data showed the average cycle ran about 63 hours including loading and unloading, and that the condenser set the pace when the chamber was fully loaded. Any shortening of a step would need data behind it, not an assumption about throughput. Several options that looked available were in fact unvalidated, and others would have required a change control that could not be finished within the window.
Quality tolerances set the outer boundary. Stability data showed that vials held at the upper storage temperature beyond the validated limit lost potency faster than the specification allowed, so any option that lengthened hold times was closed off. Maintenance records added a practical limit. The condenser serviced recently showed a rising vacuum recovery time, so the planners could not assume it would hold full rate across an extended campaign. Adding a shift for the fourth unit depended on whether QC release could keep pace, which in turn depended on the stability backlog. Each option was tested against all three limits before it was ranked.
- 1Demand signal receivedLaunch moved forward, adding roughly 9,000 vials to plan.
- 2Constraint checkValidation, stability, and maintenance limits applied before any option.
- 3Option screeningExtend cycles, add shift, outsource, or re-sequence the calendar.
- 4Quality gate reviewAny option raising product temperature exposure is rejected outright.
- 5Schedule commitmentValidated runs confirmed; overflow placed with partner site.
Actions taken
- Re-sequence the fill calendar so the three validated lyophilizer recipes run first, and move the fourth recipe into formal change control with a fixed review date.
- Add a second shift to the fourth unit only for recipes already validated, and reject any plan that extends primary drying beyond the approved range.
- Place the overflow batches with a qualified partner under a technical agreement that covers chain of custody, cycle records, and release testing, and keep those lots out of the local stability program.
- Assign one planner and one QC analyst to review slot utilization each day, with any forecast below release capacity triggering a schedule review within the same shift.
Results
The site moved two batches to the partner site and re-sequenced the fill calendar so validated recipes ran continuously. The first two shipments of the launch held their dates, and the third moved by four days. No cycle exceeded its approved product temperature limit, and no batch ran on an unvalidated recipe. Overtime came to about 340 hours across the window, within the budgeted contingency. The stability backlog shrank once launch-linked samples were prioritized, but the fourth recipe's change control would not finish before the next planning cycle.
Questions for discussion
- Which constraint would you test first in this case, and what result would have made you reconsider the partner site option?
- How would you decide when overtime on a validated process stops being a staffing cost and becomes a quality risk?
- What data would you need before deciding whether the condenser's slower recovery justified a planned service rather than continued running?
Case Study 4
Inconsistent fill weight rejects across two shifts
Larkspur Device Packaging, a plant that fills and packs prefilled syringe kits and single-use diagnostic cartridges.
The situation
Larkspur reviews rejection rates for each filling and inspection station every month, anonymized by station code rather than operator name. Across the most recent four months, one prefilled syringe fill station ran a rejection rate about double that of its sister stations, but only on the second shift. The pattern did not follow product lot or material supplier. Operators on both shifts worked to the same instruction, and the process owner had already tightened the weight check frequency once without a sustained change.
The challenge
The review had to explain an uneven outcome without turning it into a question of blame, and then design a cycle that would show whether a change worked. The indicators were limited to a rejection count by station, a weight deviation log, a changeover record, and environmental readings from the filling room. The team had to judge which sources were reliable enough to support a conclusion and which needed to be confirmed or cleaned first.
Analysis
The station design sets the first constraint. The filling head uses a peristaltic pump with a tubing set changed on a fixed interval, and the fill weight tolerance is narrow enough that small shifts in pump speed or product viscosity push units out of limit. The weight deviation log showed rejects clustering in the first hour after a tubing change rather than later in the run. The second shift changed tubing more often, because a line lead had shortened the interval to match a different product. That practice was not in any written procedure, and it explained most of the variation.
Maintenance records were the second source. Pump calibration checks were logged on both shifts, but second-shift entries had been made at the end of the shift rather than the start. The calibration therefore appeared to drift when it had in fact been set on a different day. Before drawing any conclusion, the team reconciled record timestamps with the tubing change log. They also checked whether humidity in the filling room, which the quality tolerances do not reference directly, correlated with rejects. It did not, which removed one plausible explanation and kept attention on the tubing interval.
- Four months earlierSecond-shift rejections begin running about double sister stations.
- Review monthAnonymized station data flagged; tubing log not yet checked.
- Week one of cycleInterval standardized; calibration moved to shift start.
- Week threeSecond-shift rejects reach sister station range.
- End of cycleStop rule not triggered; cleaned baseline reported to process owner.
Actions taken
- Standardize the tubing change interval in the work instruction, set by tubing set hours rather than by shift or product, and require each change to be signed in the station log.
- Move pump calibration checks to the start of each shift and reconcile calibration timestamps against tubing change records in the weekly review before any trend is reported.
- Run a four-week monitored cycle with a weekly rejection rate per station, a target of matching sister stations within one percentage point, and a stop rule if the target is missed twice.
- Flag historical entries with mismatched timestamps, recalculate the baseline without them, and use only the cleaned figures for the before and after comparison.
Results
After the interval change and calibration move, the second shift's rejection rate came within the range of the sister stations by the end of the third week. The target was not met in the first week, and the stop rule was not triggered, because the team held the change for two more weeks before judging it. The cleaned baseline showed that about a third of the earlier variation had been a recording artifact. The process owner now reviews tubing hours and calibration timing in the same report, and two other stations have adopted that format.
Questions for discussion
- Which of the indicators in this review would you distrust first, and what check would you run before using it?
- How should a team respond when the written procedure and floor practice differ, but the floor practice produces better results on one shift?
- What stop rule would you set for a monitored improvement cycle, and who should be allowed to invoke it?
Case Study 5
A release request with an open sterility read
Oriel Bank Sterile Products, a contract manufacturer of sterile injectables with one investigational device accessory line.
The situation
Oriel Bank supplies a sterile injectable to several hospital pharmacies and fills one investigational lot for a sponsor under a separate agreement. A hospital pharmacy director called the customer quality lead and asked for early release of a lot still inside its final sterility read window. The request supported a compassionate supply for one patient, and the treating physician had written that delay could worsen the patient's condition. The lot's environmental record showed one excursion on the same filling line, traced to another customer's run two days earlier, and the quality agreement treats that customer's details as confidential.
The challenge
The site had to weigh a patient's clinical need against an unresolved sterility question, the confidentiality owed to another customer, and the obligation to tell the requesting pharmacy what it could and could not know. It also had to decide whether the excursion needed reporting before any release decision, and who should own that judgment. Each choice carried a different duty, to the patient, the customer, the sponsor, and the regulator, and the team could not satisfy all of them through the same message.
Analysis
The design constraints come first. The release procedure requires a complete sterility result and an environmental review before a lot can be certified, and the quality agreement contains no exception for clinical urgency. The same documents define who may approve a deviation from release timing, which is the qualified release authority rather than the customer quality lead. Operating data supported a narrow reading of the excursion. The particle count and the one positive environmental plate had been traced to the other customer's run, and the line passed a requalification check afterward. That reading did not close the sterility question for this lot, and the team kept the two separate in every discussion.
The communication choices followed from those duties. The pharmacy was entitled to know that a sterility result was pending, when it was expected, and that release timing was under review. It was not entitled to the other customer's excursion details, which the confidentiality clause protects, though the site could describe the category of event in general terms if regulatory guidance required it. The treating physician could receive the same status through the pharmacy, with the patient's consent to compassionate supply documented first. Escalation went to the release authority and the quality director the same day, and regulatory affairs assessed reportability before any shipment decision.
Actions taken
- Keep the lot in quarantine and tell the pharmacy in writing that release is under review, the expected sterility read date, and that the decision will come from the release authority.
- Refer the excursion to regulatory affairs for a reportability assessment before the release decision, and record the reasoning, including why the other customer's details were withheld.
- Provide the treating team a clinical status summary through the pharmacy, limited to product status and expected timing, once the patient's consent to compassionate supply is recorded.
- Brief the sponsor on the delay and the confidentiality limit in one call, and record each party's agreed next step so no group assumes another has already informed the others.
Results
The lot remained in quarantine through the sterility read. The result came back negative on the scheduled day, and the release authority certified the lot after a documented review of the environmental record. Regulatory affairs concluded the excursion did not meet the reporting threshold for this product, and that conclusion was recorded with its basis. The patient received the supply about two days later than requested. The pharmacy director accepted the timing but asked for a standing procedure for compassionate requests, which the site adopted as a written pathway with defined escalation points.
Questions for discussion
- Which duty would you rank highest here, to the patient, the requesting customer, the sponsor, or the regulator, and what would change your ranking?
- What is the least a customer quality lead should be allowed to tell a clinician about a pending release, and who should confirm that wording?
- When the information that protects one party's confidentiality leaves another party unable to act safely, how would you decide what can be shared and by whom?
Learning Resources
Industrial CASE STUDIES
China, India, United States, Indonesia, Pakistan, Nigeria, Brazil, Bangladesh, Russia, Japan, Mexico, Philippines, Vietnam, Egypt, Turkey, Germany, United Kingdom, France, Italy, Canada, South Korea, Australia, Spain.