3.1 Million Eye Drops Recalled. The GMP Crisis Nobody Fixed.

On March 3, 2026, K.C. Pharmaceuticals, a private label contract manufacturer based in Pomona, California, initiated a voluntary recall of 3,111,072 bottles of over the counter eye drops. The reason: lack of assurance of sterility. Eight different ophthalmic formulations were affected, sold under store brand names at CVS, Walgreens, Rite Aid, Kroger, Publix, Harris Teeter, HEB, and Meijer. The FDA classified the recall as Class II on March 31, meaning the products may cause "temporary or medically reversible adverse health consequences."

K.C. Pharmaceuticals has been flagged by the FDA for sterility failures since 2009. Eight inspections before 2023 documented the same problems: validation lacking for sterile drug processing, inadequate sterilization processes, contamination prevention procedures not established or not followed. In August 2023, after 13 years of lenient Voluntary Action Indicated classifications, the FDA finally issued a formal Warning Letter citing aseptic processing design failures, pressure monitoring deficiencies, cleaning validation gaps, and an ineffective quality unit. Two and a half years later, 3.1 million bottles went to market without sterility assurance.

The recall did not happen in isolation. In the same weeks, Macleods Pharma recalled 1,315 bottles of Levothyroxine Sodium Tablets for sub potency, a narrow therapeutic index drug where dosing shortfalls directly harm patients. Appco Pharma recalled 175,061 bottles of Prazosin Hydrochloride capsules after detecting N nitroso prazosin, a probable human carcinogen, above acceptable limits. These are symptoms of a systemic gmp pharmaceutical compliance crisis that has been building for years and that the industry has not resolved.

The pattern should be familiar. In 2023, contaminated artificial tears from Global Pharma Healthcare killed four people, blinded fourteen others, and infected 81 patients across 18 states with an extensively drug resistant strain of Pseudomonas aeruginosa never before reported in the United States. In the years since, ophthalmic product recalls have continued: more than 20 eye drop products recalled in October 2023, over 3 million tubes of Allergan Refresh products recalled in September 2024 for pharmaceutical packaging seal breaches, Alcon Systane recalled in December 2024 for fungal contamination, nearly 1.82 million cartons of AvKARE eye care products recalled in May 2025 for sterility assurance failures.

Every one of these events traces back to the same root causes. The pharmaceutical engineering of sterile manufacturing has not kept pace with the scale and complexity of the products it supports.

The GMP Crisis by the Numbers

Over 60% of FDA warning letters are linked to documentation failures. Manufacturing quality issues account for 74% of all FDA drug application rejections.

FDA FY2024 Report on the State of Pharmaceutical Quality

The scale of gmp pharmaceutical noncompliance across the industry is visible in every enforcement metric the FDA publishes.

Warning letters from CDER jumped 50% in FY2025 over FY2024, confirmed by Jill Furman, Director of CDER's Office of Compliance. Total drug quality warning letters hit 105 in FY2024, a five year high. Foreign warning letters nearly doubled, rising from 35 to 64. Import alerts reached 126. Manufacturing GMP violations accounted for 35% of all warning letters.

Form 483 observations tell the same story. The FDA issued 779 Form 483s to drug facilities in FY2024, continuing the rebound from pandemic era lows. The most cited violations: contamination control failures in cleaning validation and microbiological controls, stability program deficiencies (17% of all 483s), SOPs not established or followed, documentation and investigation gaps, training deficiencies, and process validation failures, particularly for sterile products.

Drug recalls reached 260 events in FY2024, a 15% increase over FY2023. Recall events hit a 19 year high in 2023, rising 42.4% from 2022. Of the FY2024 recalls, 48% originated at U.S. facilities and 41% at Indian facilities. Contamination breakdowns: microbial contamination at 31%, sterility assurance issues at 28%, foreign material and particulates at 20%, product mix up and cross contamination at 17%. More than one third of all drug recalls between 2012 and 2023 were due to sterility failures. Failed specification recalls were driven by impurities above specification limits (46%) and dissolution failures (36%).

A telling data point sits in the FDA's newly published Complete Response Letter database. Of 202 CRLs issued between 2020 and 2024, 150 involved quality or manufacturing issues. That is 74% of all drug application rejections driven by manufacturing failures. Among the 89 CRLs from January 2024 through January 2025, 56% contained facility inspection related approvability issues. Roughly 70% of CRLs were issued to small or midsize sponsors. The most common deficiency types: pharmaceutical process validation gaps, inadequate stability data, incomplete control strategies, unvalidated analytical methods, container closure integrity failures, and insufficient extractables and leachables studies.

Manufacturing quality is the primary reason drugs fail to reach the market. The pharmaceutical companies submitting these applications pass on efficacy and safety in most cases. They fail on quality, on the ability to manufacture what they designed.

Why Sterility Keeps Failing

Aseptic manufacturing is among the most demanding processes in pharmaceutical engineering. Unlike terminally sterilized products, where the filled container is sterilized at the end of the line, aseptic processing requires that sterility be maintained throughout the entire manufacturing chain: from pharmaceutical raw materials preparation through compounding, filtration, filling, and container closure. Every surface, every airflow path, every human operator in the cleanroom is a potential contamination vector.

Three recurring themes emerge from the FDA's inspection record. First, poor personnel practices. Operators remain the primary contamination risk, and failures in gowning qualification, aseptic technique, and behavior monitoring appear in 483 after 483. Second, loss of environmental and process control. Inadequate air handling, pressure differential monitoring, and environmental monitoring programs allow contamination events that should have been caught in real time. Third, fundamentally flawed operational design. Facility layouts, equipment qualification, and process flows that were never adequate for the products they are asked to produce.

Multi dose ophthalmic products compound every one of these challenges. The preservative system must remain effective through repeated opening and closing by the patient. The pharmaceutical packaging must maintain container closure integrity through dozens of uses in uncontrolled home environments. Each use exposes the product to ambient microorganisms. When multiple formulations share the same aseptic filling line, as they did at K.C. Pharmaceuticals, pharmaceutical validation of cleaning procedures between products becomes critical. At K.C. Pharmaceuticals, that cleaning validation was never completed.

The regulatory framework for aseptic manufacturing is detailed and well established. The FDA's guidance on aseptic manufacturing practices provides comprehensive requirements that have been in place for years. The FDA's guidance on sterile drug products produced by aseptic processing specifies continuous nonviable particle monitoring, routine viable monitoring of air, surfaces, and personnel, airflow velocity requirements of 0.45 meters per second with 20% tolerance for unidirectional flow, media fill validation, sterile filtration validation, pre filtration bioburden monitoring, personnel gowning qualification, and equipment cleaning and sterilization validation. These requirements have been in place for decades. The failures at K.C. Pharmaceuticals, Global Pharma Healthcare, and the companies behind every other ophthalmic recall in the last three years were failures of execution, not of understanding.

The Inspection Gap

The FDA oversees nearly 280,000 regulated facilities worldwide. More than half are outside the United States. CDER tracks 4,619 pharmaceutical manufacturing sites, of which 59% are foreign. In FY2024, the agency conducted 989 inspections, a 27% increase from the prior year, with 62% of those inspections at foreign facilities, an all time high.

Those numbers mask a structural weakness. When the FDA suspended most inspections in March 2020, a backlog formed that has never been cleared. As of May 2024, approximately 2,000 pharmaceutical manufacturing firms had not been inspected since before the pandemic. That includes over 340 plants in India and China, the two largest sources of active pharmaceutical ingredients and finished dosage forms for the U.S. market. An Associated Press analysis found that 42% of registered laboratories remained uninspected after five years.

The staffing situation has deteriorated in parallel. As of June 2024, the FDA had 225 vacancies in its inspection workforce, nearly four times the pre COVID number. The inspector vacancy rate jumped from 9% to 16% between November 2021 and June 2024. Experienced inspectors earning government salaries are leaving for higher paying positions in the private sector.

Then came the April 2025 HHS workforce reductions. Approximately 3,500 full time FDA employees were laid off on April 1, 2025. The inspectors themselves were not directly eliminated, but the infrastructure supporting them was gutted. The travel operations division, responsible for booking flights and securing translators for foreign inspections, was eliminated. The unannounced foreign inspection pilot program was paused. Inspectors now handle both administrative and investigative work, creating bottlenecks that reduce the number of inspections each person can complete.

The result is a growing gap between the number of pharmaceutical companies requiring oversight and the FDA's capacity for meaningful inspections, particularly at aseptic manufacturing sites where contamination risks are highest. Companies that might be caught earlier are allowed to operate with minimal scrutiny. The K.C. Pharmaceuticals timeline, 13 years of lenient VAI inspections before a Warning Letter, is what the system produces when inspection capacity cannot match the scale of the industry.

The API supply chain compounds the problem. The FDA's FY2024 quality report found that compounding pharmacy API sites, representing only 18% of API manufacturers, accounted for 72% of regulatory actions. Of violative API sites, 51% were in China and 30% in India. China represented 39% of sites added to import alerts despite being only 17% of foreign sites. Pharmaceutical raw materials face additional pressures: U.S. tariffs on Chinese pharmaceutical raw materials, APIs, and finished products reached upward of 35% effective February 2025, creating supply chain disruptions and longer clearance times that further strain quality verification.

The Nitrosamine Crisis That Will Not End

In July 2018, NDMA, a probable human carcinogen, was discovered in valsartan manufactured by Zhejiang Huahai Pharmaceutical Company in China. That discovery triggered the largest pharmaceutical contamination investigation in modern history. Irbesartan was recalled for NDMA in October 2018. Losartan followed in November 2018 for both NDMA and NDEA. In September 2019, NDMA was found in ranitidine. By April 2020, the FDA requested withdrawal of all ranitidine products from the market. In the years since, nitrosamines have surfaced in metformin, pioglitazone, nizatidine, and multiple other drug products.

Eight years after the initial valsartan recall, the crisis continues. In March 2026, Appco Pharma recalled 175,061 bottles of Prazosin Hydrochloride capsules for N nitroso prazosin impurity above acceptable limits. The FDA database shows more than 1,400 product lots recalled due to carcinogenic N nitrosamine impurities exceeding the acceptable daily intake limit of 26.5 nanograms.

The root causes trace to specific manufacturing processes: solvents, reagents, recycled materials, and in the case of ranitidine, a degradation pathway where NDMA levels increased significantly under elevated temperatures. These are pharmaceutical process validation failures. The contamination was a predictable consequence of process conditions that were not adequately understood, monitored, or controlled. In every case, the issue was detectable before products reached patients. In every case, it was not detected until after distribution.

Pharma 4.0 and AI in Manufacturing Quality

The pharmaceutical technology to prevent the failures described in this briefing already exists, but adoption remains slow.

The FDA's Process Analytical Technology framework promotes integration of real time analytical measurements into manufacturing processes. The concept is straightforward: measure quality attributes during production rather than relying solely on end product testing. In January 2025, the FDA updated its guidance on 21 CFR 211.110 to explicitly support advanced technologies, real time quality monitoring, PAT, and continuous manufacturing systems. PAT integrated continuous manufacturing digital twins have demonstrated API consistency improvements to 99.95%.

53% of manufacturers are now using factory AI. McKinsey estimates $150 billion in annual pharma manufacturing efficiency improvement potential through digitization.

Gartner, 2025; McKinsey Global Institute

AI and machine learning applications in pharmaceutical manufacturing are moving from pilot programs to production. The applications fall into four categories: inferential measurements using soft sensors to estimate critical quality attributes from readily available process signals, abnormal event detection for fault identification and deviation triage, advanced process control for optimization and closed loop strategies, and computer vision for visual defect inspection.

The results from early adopters are substantial. Merck is using AI to identify process deviations and improve batch consistency, and has applied deep convolutional neural networks to detect defects in film coated tablets. Pfizer is deploying AI that detects anomalies and suggests real time corrective steps, targeting 10% improvement in product yield and 25% reduction in cycle time. Samsung Biologics positions digital automation, real time monitoring, and predictive optimization as competitive differentiators. Documented results across implementations: 50 to 60% reductions in investigation time, 50 to 70% reduction in root cause analysis duration, and meaningful improvements in deviation management.

Digital twins are one of the fastest growing segments in pharmaceutical technology, with the market projected to grow from USD 1.3 billion in 2025 to USD 8.5 billion by 2032. Pfizer has built digital twins for computational laboratory experiments, reducing preclinical testing times by 30% and deploying predictive quality models across multiple sites. Novartis has committed $23 billion for six new U.S. plants incorporating facility wide digital replicas. Demonstrated improvements from digital twin adoption: 55% reduction in process variability, 64% reduction in compliance incidents, 25 to 40% throughput gains, and 15 to 20% lead time reduction.

Continuous manufacturing is a major departure in pharmaceutical engineering from traditional batch processing. Production time reductions of approximately 90%, product quality improvements of approximately 40%, energy and water consumption reductions of 25 to 50%, and working space reductions of 30 to 50%. Over 15 pharmaceutical products are now manufactured via continuous processes with FDA approval. The FDA's Emerging Technology Program has reviewed over 160 proposals, with 22 leading to successful regulatory submissions. An FDA self audit found no higher risks for continuous manufacturing versus batch, and identified potential advantages in manufacturing scale up flexibility.

The Quality 4.0 business case is documented across multiple implementations. QC laboratory cost reductions of 25 to 45%. Batch review time compressed from 7 months to 5 days. Deviation reduction of 65 to 80%. Manual documentation elimination reaching 80% of tasks. Document cost per unit dropping from $20 to $4.82. Typical ROI payback within 1 to 3 years.

The pharmaceutical companies investing in pharma 4.0 are already seeing measurable returns in quality, throughput, and compliance, yet most of the industry has not deployed these tools.

What This Means for Formulation Science

The GMP crisis is a formulation science problem. Every recall in this briefing traces back to decisions made during product development and process design: the choice of preservative system in a multi dose ophthalmic product, the selection of pharmaceutical packaging and container closure, the design of cleaning validation protocols for shared aseptic lines, the control strategy for nitrosamine formation during synthesis. These are formulation decisions. They are made months or years before the first commercial batch runs.

The 74% CRL rejection rate for manufacturing and quality issues reflects a disconnect between how pharmaceutical companies develop formulations and how those formulations perform at manufacturing scale. Pharmaceutical process validation failures, inadequate stability data, container closure integrity issues, extractables and leachables gaps: each of these starts in development and compounds during scale up.

• Quality by Design: ICH Q8 principles connect formulation development decisions to manufacturing outcomes through systematic design space identification. AI driven Design of Experiments can map the relationship between formulation variables and critical quality attributes before a single batch runs at scale.
• Process Validation: ICH Q8, Q9, Q10, Q12, and Q13 form a lifecycle framework that ties pharmaceutical validation to formulation understanding. The companies receiving CRLs for process validation gaps are failing at the development stage where process understanding should have been established, not at manufacturing itself.
• Computational Quality: Predicting critical quality attributes, anticipating stability challenges, modeling extractables and leachables interactions with formulation components, and identifying process risks before scale up begins. These are the applications where AI can directly address the root causes behind the recall and CRL data.
• Raw Materials Intelligence: With 72% of compounding pharmacy API sites subject to regulatory actions and tariff driven supply disruptions affecting pharmaceutical raw materials from China and India, formulation scientists need systematic tools for raw material qualification, supplier assessment, and excipient compatibility analysis.

The FDA's Quality Management Maturity program, now in its third year, evaluates pharmaceutical companies beyond traditional compliance on dimensions like operational excellence, supply chain resilience, and quality culture. The distinction the program draws matters: compliance asks whether rules were followed, quality culture asks whether the right decisions were made for the patient. K.C. Pharmaceuticals passed VAI inspections for years while the same sterility failures persisted. That is the difference between compliance and quality, and it starts in formulation development.

DeepC is built for the work that prevents these outcomes. Its tools address the specific gap between formulation design and manufacturing quality: excipient selection and compatibility assessment using pharmaceutical machine learning models trained on real world data, critical quality attribute imputation for incomplete CQA datasets, spectral prediction from molecular structures to anticipate analytical challenges, and structured data extraction from CMC documents, batch records, stability reports, and regulatory filings. Every recommendation carries data lineage from FDA, EMA, and ICH regulatory sources. Every prediction is traceable.

The pharmaceutical manufacturing market is projected to reach $1.91 trillion by 2034. The pharmaceutical quality management software market is growing at roughly 13% annually. AI in pharma and biotech is projected to reach $154 billion by 2034. The investment is flowing toward manufacturing quality and pharmaceutical technology. The companies that connect formulation intelligence to manufacturing execution will be the ones that avoid the next recall, the next CRL, the next consent decree.