Current Trends in the Medical Products Stability Industry

We stabilitarians share and discuss our knowledge and experience because we care about the industry and have a special bond with our peers at other companies—we want them to succeed. Do you remember how the community came together to discuss test initiation timing when it was a huge topic (not that it’s less important now), or when monitoring systems and probe placement were the key questions of the day? This article will identify five current regulatory trends and five logistical trends of the medical products stability industry seen as significant over the past year based on experienced program leaders’ insights. But, unlike a clickbait blog posts I won’t string you along till the end of the article—here they are:

Granted, these are what I’ve selected as the five, but there are many other things going on that are not mentioned, so my apologies if I didn’t include your latest regulatory mandated project. But, let’s delve into each of these that I’ve selected to learn why they are trending and what industry is saying about them. These trends were gathered in part through interview questions submitted to ten stability managers across the industry, though specifics of their responses are not disclosed to ensure confidentiality.

Stability testing is one of the most heavily scrutinized areas for data integrity during an inspection since it’s such a critical component of patient safety, as per 21 CFR 211.166 and ICH Q1A(R2). Due to the extended time data is gathered in the stability program of a product, regulatory bodies—as outlined in the FDA Data Integrity Guidance and PIC/S PI 041—heavily scrutinize stability data to ensure a continuous, traceable chain of custody and compliance with ALCOA+ and ++ principles from sample generation to final reports.

ALCOA is the acronym that just won’t go away, and like other contemporary acronyms it keeps growing, so we’re at ALCOA++ where investigators are looking at raw data and audit trails to capture things that for years were considered “not worth looking at”. At the beginning, data integrity focused on the point of data entry, but with technology advances, more and more systems went paperless and it became clear that changes made enroute to the final entry were not being captured like they were when we lined out our changes in a paper notebook (the good old days).

Where does this impact stability? We’re ultimately responsible for our data and as an inspector is looking through your data they could easily want to know why a peak was resolved at 3MO but is no longer there at 6MO. Were HPLC integration parameters modified, and if so, what were the decisions leading to that? If you have an in-house laboratory they will probably stand in to answer that for you, but if your data comes from a contract lab you have to be prepared to give the answer. This is linked to two other trends, as we’ll see shortly.

Meanwhile, the uneasy question is how much do inspectors expect the same traceability for decisions made regarding the stability study design and execution? As one stability manager stated, “Inspectors seem to be moving from did you do the work to can you prove the work, the system, and the decisions behind it are sound? They want to know if they can trust the data from start to finish and want proof that the entire electronic system behind those records is reliable and secure.” In a warning letter from September 2025 the FDA cited a company for exactly that, discontinuing a long-term stability study instead of conducting a thorough root-cause investigation. The FDA noted that the analytical software lacked basic controls that allowed failing stability pulls to be handled this way without a traceable digital footprint.²

Any stability LIMS system that’s up to agency standards will have a robust audit trail so entries and changes, like non-routine sample extractions or inserted test intervals, are traceable and defendable. So now, for example, an investigator could discover that all annual lots for the product-of-interest were historically stored in Chamber A but are now in Chamber B and they might want to look more deeply to see if there was a failure to properly investigate an excursion in Chamber A. It should all be right there at their fingertips in the LIMS raw data or in the maintenance system’s raw data, or both.

I’ve kept this persistent trend near the top of the list because the scope of the agency’s expectations regarding data review are not uniformly understood and the industry seems rather divided in implementing this. Some programs pour through scans of raw data notebooks and review electronic files of chromatograms to the point of almost repeating the analysis themselves, while others review some quantity less than 100%. As noted in the recent Spring 2026 PSDG meeting, “Companies adjust their data review processes based on the trust level with the specific lab. Trusted labs might only require basic quality agreements and high-level reviews, whereas labs with previous compliance issues require a “fine-tooth comb” review of all raw data.” The key is the quality agreement and rationale for the various decisions being included in the change history of that document.

The guiding principle of this trend is that the agency clearly identifies that file holder is the owner of the data generated by any contract labs. In relation to a significant uptick in regulatory observations in 2025 and continuing in 2026, IntuitionLabs writes that “A substantial portion of these actions target failures in manufacturing quality systems and data integrity at contract facilities. For example, FDA investigators recently emphasized that contract laboratories and manufacturers are “extensions of the manufacturer’s own facility,” underscoring that any CGMP lapse in a contract site “may affect the quality, safety, and efficacy of the drugs” made for clients.”³

This impacts stability very closely, and in fact some recent observations were directly stability-related. As such, an auditor will not go back to the CDMO to question why they didn’t detect an OOT result, they’ll question the owner of the data. Part of data review needs to include some measure intended to detect if a trend is developing or if a data point is inconsistent with the historical trends, but more importantly some documentation that the review and analysis was performed—not just filing their CoAs in with the rest of the data for that study.

It’s pertinent that this was one of the topics covered in the Sunday afternoon presentation session regarding “Stability Samples from Inception to Destruction” at the Spring 2026 PSDG meeting. In that presentation Kenley Joseph from Q1 Scientific cited a presentation at PDA 2025 in Ireland noting that regulatory data from agencies like the HPRA (Ireland) show transportation as a source of major industry deficiencies.⁴ These data were related to transportation compliance in general, but the trend developing is not just use of stability programs and data for validation of shipping lanes, but validation of stability sample shipment routes to labs and to and from stability storage facilities as well.

In relation to impact of shipping lanes on product shelf life, another pertinent topic covered in the discussions at the Spring PSDG meeting was about stability budgets. Not the financial budget, but a budget of time and intensity of thermal exposure a sample can experience and still meet shelf life. Such a budget is developed using long term stability data, often in conjunction with developmental stability data showing product robustness at elevated temperatures. If insufficient data is available at development to justify a reasonable stability budget to address expected stresses from normal shipping lanes, then special stability studies will be needed to fill in the gaps.

Of continued concern when presenting data to an auditor that involves samples shipped over large distances is the question of how transport might have impacted the samples, and for that concern this remains a trending topic among stabilitarians, even though the issue itself has drawn few, if any, direct regulatory deficiencies. Under 21 CFR 211.160(b), laboratory controls must include the establishment of scientifically sound and appropriate specifications, so if you don’t monitor the temperature of a stability sample in transit to a testing lab, you cannot really prove that the sample at the lab is representative of the sample that left the chamber. Therefore, the continuing trend is to find optimal ways to track and monitor sample shipments to prevent OOS or OOT events that could indicate poor sample integrity.

It’s amazing to me that even after 30 or more years since the BARR decision, inadequate OOS and OOT investigations are still areas for improvement, trending in our industry. To be fair though, the current level of improvement is clearly on a higher plane. The focus of concern now is the scrutiny that auditors are giving stability failure investigations to ensure that the root cause is not inappropriately pinned on laboratory error. Inspectors are increasingly wary of laboratory investigations that lack a thoroughly concluded scientifically sound root-cause.

A stability manager with 30 years of experience noted what she sees as an uptick in regulators who are “paying closer attention to repeated stability failures and wanting to know if companies really solved the problem or just closed the paperwork”. Some companies are taking on reviews of their stability OOS to ensure they’re meeting these expectations. I’d like to say “new” expectations, but they’re not, we’re just seeing audit attention focused there.

As recently as March of this year a FDA warning letter was delivered to a company for this issue after inspectors found that 12MO and 24MO stability data showed a peak not present during batch release testing. It was a degradation impurity that they neither monitored nor reported in their stability summaries.⁵ A few other recent warning letters indicate there were likely multiple observations that didn’t get that far.

In conjunction with the data integrity trend addressed, the ability for lab technicians to manually integrate chromatograms and potentially hide peaks, poses a risk to patients, but now with the expectation to view raw data and audit trails, there is such a high likelihood of detection that it should effectively quash the practice. That is where the trend is heading, with companies locking down integration methods and other testing loopholes that allow subjective analyses to impact results and ensure detection, quantitation and reporting of unknown and unexpected impurities.

It was hard to limit this selection to only 5 regulatory trends, and even harder since several of them seem to keep recurring. One theme emerges from these trends and that is that as computer analytical capabilities increase, so do regulatory expectations that those capabilities be used for tighter quality control. I’ve often joked (though serious) that I could never perform testing in the same labs where I once excelled as a technician—the documentation expectations are so far advanced, and verification now includes checking audit trails for illicit data changes. Even the ability to track and monitor shipments has become commonplace with several small devices available on the market. And finally, regarding CDMOs, where once the only means of data transfer was via PDF attachments in an e-mail, now direct access to their systems is possible. We will continue to see technology up the ante on where to improve our stability programs.

From the same interviews of stability managers, I also asked about innovations, future plans and regrets they have about the physical aspects of their programs—chambers, monitoring systems, LIMS, etc. Those responses as well as input from the participants at the recent Spring 2026 PSDG meeting held in Wilson, NC were compiled into this somewhat subjective selection of what is heavy on the minds of stabilitarians today. Again, forgive me if I’ve excluded or demoted to a lower ranking than 5th place some key trend or improvement that you could never do without.

The potential that AI has for use in stability programs is heavily muted by the necessity to be able to validate its responses. There are, however, several administrative tasks that AI can handle that save considerable time even if the final results must be verified prior to use in any quality decisions. One source referred to this as “using “human-in-the-loop” AI validation to slash the administrative bottleneck of data management.”

One of the administrative tasks that was brought up by several stabilitarians is the ability to have AI transcribe data from a PDF (such as one submitted from a CDMO), parse that data into Excel, or into a statistical program (like JUMP) and even have it generate charts, trendlines and such. The transcription would have to be verified while presumably the statistical calculations are already validated or can be verified by a qualified statistician.

This same transcription task could prove exceedingly valuable for moving old paper archived data into a new LIMS system. In fact, archived studies and samples could conceivably be created and study meta data and historical results populated into the LIMS system. Once the fields are manually verified and archived samples approved, the data could join the current population to build more robust trend charts through a linked statistical system—assuming your LIMS is fully digitalized.

In addition to integration of contract lab testing results into their systems using AI, one stability manager envisions their program applying AI to automate study status monitoring in the coming years. Again, with the “human-in-the-loop” model, AI could be used to monitor studies continuously while human review of the studies could be mandated every 6 months or so. Those human-conducted reviews would be the official routine evaluation but if AI detected some potential anomalous data trend in the meantime, that would also be verified by human eyes, thus having a much tighter safety net than is possible with the limited human resources available.

With the recent rapid growth of mRNA therapies, cell and gene therapies (CGTs), and advanced biologics, stability programs have been forced to scale up -70°C, -80°C, and liquid nitrogen storage volumes. Along with that it has become a priority to validate sample retrieval windows and maintain continuous sensor logs.

At the same time, as one manager points out, companies are using more advanced stability chambers that are easier on energy use, can be monitored wirelessly, can detect problems on their own, and can be checked remotely without someone standing in front of the unit. Having the ability to forecast if your -80°C chamber is about to die can be extremely valuable to transfer material before a temperature excursion actually occurs.

This concept, though already being utilized by a few companies represented at the Spring 2026 PSDG meeting, is relatively new as an emerging trend. The key is that historically, chamber mapping during validation has not adequately addressed the capability of the chamber to maintain uniformity when filled beyond the “normal” full level, meaning with extra carts in the aisles. The novel approach of “maximum full”, is to perform an additional “full” leg of the chamber mapping with material placed on carts loaded with the “test material” (like cardboard boxes) within the aisles, thus achieving confirmation that the ultra-worst-case scenario does not impact environmental performance.

Some questions were raised as to whether this “maximum full” leg of the mapping process could replace the typical “full” leg altogether, but that decision would need to be justified by individual companies. Still, whichever way it’s done, this approach can solve a frequently asked question during audits and avoid special mapping runs when unplanned volumes require overloading a chamber for a time.

We’re talking about digitalization of the environmental monitoring process. I touched on this a bit in the discussion on the increase in ultra-low chamber needs but here is a bit more. By having monitor data so readily available for analysis, companies can automatically generate charts, layer them atop data from other chambers or atop historical data and quickly detect shifts in a chamber’s performance. Shucks, AI could essentially perform this analysis continuously, and again, with a “human-in-the-loop” , what had been weekly human conducted reviews could conceivably be reduced to a monthly basis—until someone figures out how to validate the AI analyses (then we’re all out of a job).

When asked what physical improvements stability managers see being made to local stability programs industrywide, backup systems were part of every response. One manager noted that, “As we are replacing old walk-in rooms, we are building redundancy into the new rooms so we do not have to move product. If one system fails, the other will automatically turn on.” Another manager noted that, “Companies are also adding more backup space and organizing storage more carefully because mistakes or equipment problems have become too costly.”

A site could reserve space at a nearby contract storage facility, but it remains incredibly difficult to convince corporate business units to pay for that “insurance policy” on empty chamber space. To get beyond just relying on more and smarter storage space, contract stability storage companies are starting to offer flexible backup chamber space options, so clients don’t have to pay full price for empty space. But those same contract facilities caution clients that the storage facility may also be subject to the same disaster that necessitated the need for backup storage.

With that in mind, some companies are going so far as to maintain a second set of stability study materials out of the country that are only sampled for testing in a disaster or other emergency situation.

Just as with the regulatory trends, as technology speeds ahead, corporate management will expect stability programs to realize cost savings—primarily by reducing resources. One manager interviewed expects that in the future they will continue “Doing more and more with less resources, and [that they are being] told to utilize AI” to accomplish that. But it’s not just technology that will achieve these cost savings. Good old out-of-the-box thinking as stabilitarians and vendors help us derive better ways to find protection from disasters or over-loaded chambers. As one manager summarized, “Funding is always a challenge. . . [and] if it was not budgeted, we have to wait until next year’s budget to purchase.” Oh well, that’s the wonderful life of a stabilitarian.

1. Data Integrity in Labs: Key FDA Guidelines, Data Integrity in Labs: Key FDA Guidelines – Allan Chemical Corporation | allanchem.com, Allan Chemical Corporation,
2. FDA Warning Letter to Person & Covey, Inc., Persōn & Covey, Inc. – 711191 – 09/23/2025 | FDA, FDA, September 23, 2025
3. Contract Manufacturing Oversight: 2026 FDA Enforcement Data, Contract Manufacturing Oversight: 2026 FDA Enforcement Data | IntuitionLabs, InstitutionLabs, Update 6/2/26.
4. McCormack, E. & Bergin, J. (Health Products Regulatory Authority). Common GDP Inspection Deficiencies, Expectations for Addressing Them, and Tips for Maintaining Compliance & Getting Through a HPRA GDP Inspection. Presented at the PDA Ireland Event: Good Distribution Practice – Supply Challenges & Use of Future Technologies, 12 June 2025
5. FDA Warning Letter to Sato Pharmaceutical Company, Sato Pharmaceutical Co., Ltd. – 723059 – 05/18/2026 | FDA, FDA, March 18, 2026