The Oversight Duty: What a Clinical-Stage Board Is Required to Watch
Most of what a board does is decide. It approves the budget, authorizes the financing, chooses whether to start the next study. Fiduciary law is forgiving of those decisions. A board that informs itself, acts without conflicts and makes a reasonable call will almost never be second-guessed by a court, however badly the call turns out.
There is a second part of the job that gets far less attention and carries a different kind of exposure: watching. Directors are expected to make sure the company has a way of telling them when something important is going wrong, and to pay attention when it does. Delaware law calls this the duty of oversight, and it is usually referred to by the name of the case that first described it, Caremark.
For most of its history, the oversight duty was regarded as close to theoretical. Claims under it were famously hard to win. That changed in 2019, when the Delaware courts allowed two oversight claims to proceed within a few months of each other. The second involved a clinical-stage oncology company and the way its board monitored a single trial. Every biopharma director should understand that case, because it describes, with uncomfortable precision, the oversight failure clinical-stage boards are most exposed to.
(This article is governance guidance, not legal advice. It describes Delaware law, which governs most US venture-backed biotechs; companies incorporated elsewhere should check how their own jurisdiction treats the question.)
Where the Duty Comes From
The duty was first set out in 1996 by the Delaware Court of Chancery in In re Caremark International. The core idea was modest. Directors cannot be expected to know everything that happens inside a company, but they can be expected to make a good-faith effort to ensure that a reporting system exists, so that material information about the company's operations and legal compliance reaches them.
A decade later, in Stone v. Ritter (2006), the Delaware Supreme Court adopted that standard and placed it within the duty of loyalty rather than the duty of care. That classification matters more than it sounds. Liability for an oversight failure requires bad faith, meaning that directors knew they were not doing their job and failed to act anyway. Simple carelessness is not enough.
The court described two ways a board can fail. The first is to make no serious attempt to set up any system for reporting on the risks that matter. The second is to have such a system and then consciously ignore what it reports, the warnings usually described as "red flags."
The loyalty classification has a practical consequence that every director should register. Most Delaware companies' charters exculpate directors from personal monetary liability for breaches of the duty of care. Those provisions cannot reach breaches of loyalty or acts in bad faith. An oversight claim that survives is therefore one of the few routes by which an outside director's personal exposure becomes real.
The bar remains high. Most oversight claims are still dismissed, and the courts have repeatedly said these are among the hardest claims in corporate law to plead. But "hard to plead" is not a governance standard. A board that relies on that difficulty is managing litigation risk rather than doing the job the duty describes.
2019: Mission-Critical Risk
The first of the 2019 cases had nothing to do with drugs. In Marchand v. Barnhill, the Delaware Supreme Court revived an oversight claim against the board of Blue Bell Creameries, an ice cream maker whose products were linked to a listeria outbreak in 2015. Blue Bell made a single product, and the court treated food safety as essential and mission critical to a company of that kind.
The board's defense was that the company had food safety programs and complied with FDA regulations. The court's answer is the sentence that matters for biopharma: nominal regulatory compliance at the management level did not show that the board had put any system in place to monitor that risk itself. The complaint alleged that there was no board committee covering food safety, no regular board-level process for hearing about it and no protocol for escalating problems to directors. That was enough, at the pleading stage, to support the inference that the board had made no good-faith effort at all.
The principle that came out of Marchand is that oversight is proportionate to consequence. A board does not need a reporting line for everything. It does need one for whatever risk, if realized, could destroy the company. Where that risk is also governed by regulation, the expectation is higher.
The Biotech Case
A few months later, on October 1, 2019, the Court of Chancery applied Marchand to a clinical-stage biopharma board in In re Clovis Oncology, Inc. Derivative Litigation.
Clovis was developing rociletinib, a lung cancer drug that the court treated as the company's mission-critical product. The pivotal question was efficacy, and the trial measured it the standard way for solid tumors: objective response rate under the RECIST criteria. RECIST distinguishes between responses that have been confirmed by a follow-up scan and those that have not. According to the complaint, the board knew that the FDA would judge the drug on confirmed responses and that investors would treat unconfirmed figures as unreliable.
The company publicly described the drug's response rate as about 60 percent. The plaintiffs alleged that this figure included unconfirmed responses, that the board received information showing the confirmed rate was materially lower, and that directors allowed the higher number to keep circulating in public statements anyway. In November 2015 the company disclosed a far lower response rate, and its share price fell by roughly 70 percent. The drug's application was withdrawn in 2016. In a separate proceeding in 2018, the company and two of its executives agreed to pay more than $20 million in penalties to settle SEC charges that investors had been misled about the drug's efficacy.
The derivative case was brought under the second of the two failure modes. Clovis had a reporting system; the board received detailed information about the trial at its meetings. The allegation was that the directors saw the warnings and did not act on them. The court held that the claim could proceed, reasoning, in line with Marchand, that when a company's mission-critical operation is subject to regulatory requirements, the board's oversight must be exercised more rigorously. The court also noted that the board was made up of directors with substantial pharmaceutical industry experience. The board later formed a special litigation committee, and the case went to mediation in 2022.
It is worth being exact about what the decision was. It was a ruling that the allegations were strong enough to proceed, not a finding that the directors had acted in bad faith, and the defendants disputed the allegations vigorously, including how RECIST should be read. The lesson for other boards does not depend on who was right. It depends on how ordinary the facts were.
Why This Is the Clinical-Stage Board's Risk
Nothing about the Clovis facts is exotic. A promising drug, one key endpoint, a technical distinction in how that endpoint is counted, strong interim data, an active financing environment and a management team that believes in the product. Most clinical-stage boards will recognize that situation as an ordinary quarter.
The structure of a clinical-stage company concentrates oversight risk in a way that a diversified business does not. Several features stand out.
The trial is the company. In a single-asset or lead-asset company, the conduct of the pivotal study, the integrity of its data and the accuracy of what is said about it are the mission-critical operation in exactly the sense Marchand described. There is no revenue stream elsewhere to absorb the damage when it goes wrong.
The risk is regulated at every point. Good clinical practice, protocol adherence, safety reporting, FDA correspondence and the securities rules governing what is disclosed about data all apply at the same time. That combination is the one Clovis singled out as demanding more rigorous oversight.
The data arrives through people with incentives. Interim results move through a CRO, a biostatistics group, the clinical team and the CEO before they reach the board. Each step is a chance for a methodological choice, made in good faith, to tilt the picture. Directors rarely see raw data and cannot be expected to; that is why the reporting system matters.
The distance between data and disclosure is short. A clinical-stage company describes its data publicly far more often than most businesses describe their operations, in conference presentations, press releases, investor decks and, for a public company, financing documents. Every one of those statements is an opportunity for the gap between what the protocol defines and what the company says to widen.
And the board is often expert. That is usually a strength. But an experienced board that receives a warning in its own field has a harder time explaining why it did not understand what it was being told.
Building the System
The first form of oversight failure, having no system at all, is the easiest to avoid. What a court looks for is evidence that the board itself, not just management, made an effort to make sure it would hear about mission-critical risk. In a clinical-stage company, that system has a few components.
Name the risks explicitly. The board should record, in a resolution or committee charter, which risks it regards as mission critical. For most clinical-stage companies the list is short: patient safety, clinical data integrity, regulatory compliance and interactions with the agency, and the accuracy of public statements about data. Writing it down converts an assumption into a responsibility someone owns.
Assign them to a forum. Small boards often have no committee that owns clinical and regulatory risk, and the earlier article on committees in a five-seat board explains why a separate one may not be practical. But someone has to own it. That can be the full board with a standing agenda item, or a director with relevant expertise asked to lead the topic. What matters is that the assignment exists and is visible in the board's records, which was exactly what Blue Bell's board could not show.
Define what gets reported, and when. The board should agree with management on what reaches it automatically: serious adverse events and safety signals, data monitoring committee recommendations, significant protocol deviations, CRO audit findings, clinical holds and material FDA correspondence, and any change in how an endpoint is calculated or described. Some items belong at every meeting. Others should reach the board within days, not at the next scheduled meeting.
Create a direct line. The board should hear from the chief medical officer, the head of clinical operations or the person responsible for clinical quality directly, at least periodically, and at times without the CEO present. The earlier article on board meetings described this as one of the executive session formats worth making routine. It is also the mechanism most likely to surface the disagreement inside a company before it surfaces outside.
Reconcile disclosure with the protocol. Every public statement of efficacy should be checked against the endpoint as the protocol and statistical analysis plan define it. Whoever reviews press releases and investor materials should be required to confirm that headline figures use the protocol-defined measure, and to state plainly when they do not, for example when a figure is interim, unconfirmed or from a subgroup. This one discipline would have addressed the core of the allegations in Clovis.
Responding to Red Flags
The second form of failure is harder, because it happens to boards that have done the first part right. A reporting system that works will, by design, sometimes deliver bad news. The duty is to respond to it.
A red flag in a clinical-stage company rarely looks dramatic when it first appears. It can be a footnote saying responses are "pending confirmation," a gap between the response rate in the investor deck and the one in the clinical update, a CRO audit finding that management describes as routine, a slowdown in enrollment at a single high-performing site, or an unexplained change in how an analysis is described from one meeting to the next. Each can have an innocent explanation. The board's obligation is to find out which.
Three habits separate boards that respond well from those that do not.
The first is asking the specific question rather than accepting the general reassurance. "Are we comfortable with the data?" invites a yes. "Is the 60 percent figure confirmed responses as the protocol defines them, and if not, what is the confirmed number?" does not.
The second is following through. A question asked at one meeting and never returned to is, in hindsight, a warning the board saw and did not pursue. The chair should keep track of open questions on mission-critical topics and make sure each is closed with an answer the board has examined.
The third is escalating when the answer is unsatisfactory. That can mean an executive session, a request for the underlying data, an independent statistician or regulatory consultant reporting to the board, or a pause on a planned public statement until the question is settled. Each of these is uncomfortable, and each is far less costly than the alternative.
The Record
Oversight claims are increasingly built from a company's own documents. Stockholders can demand access to a company's books and records before they sue, and they use those demands to obtain board minutes and materials. The board's record is therefore the evidence of whether it monitored mission-critical risk.
This is where the article on board meetings and this one meet. Minutes that show a standing agenda item on clinical and regulatory risk, the questions directors asked, the information they received and what was decided are the board's demonstration that the system existed and that it listened. Minutes that say "the board received a clinical update" show neither. A board that monitored carefully but recorded nothing will find it hard to prove it.
The record also cuts the other way. Board materials that contain the warning, with nothing in the minutes to show the board engaged with it, are close to the worst position a director can be in. That is not an argument for keeping warnings out of board materials. It is an argument for making sure that every warning in the materials has a visible response in the record.
Officers, Too
In 2023, in a case involving McDonald's, the Court of Chancery held for the first time that corporate officers, not only directors, owe a duty of oversight, applied in the context of each officer's area of responsibility. For a clinical-stage company, the implication is direct. The chief medical officer is responsible for oversight of clinical risk, the chief financial officer for financial reporting, and so on.
This matters for boards in two ways. It means the officers the board relies on are themselves accountable for raising red flags, which strengthens the reporting system. And it gives the board a basis for setting that expectation explicitly: officers responsible for mission-critical areas should know that escalating problems to the board is part of their job, not a disloyal act toward the CEO.
Private Companies Are Not Exempt
It is tempting for directors of private, venture-backed companies to treat this as a public-company problem. Most of the reported cases involve public companies, because that is where stockholder derivative suits usually come from.
But the fiduciary duties are the same in a private company, and the situations in which an oversight failure surfaces in a private company are common: a financing in which new investors discover a data problem in diligence, an acquisition where the buyer's review finds that the clinical record does not match what was presented, a dispute among investors after a failed trial, or an insolvency in which a trustee examines what the board knew. The habits also take years to build. A company that plans to go public, or to be acquired, will be judged on a clinical record that was created while it was private.
The Underlying Principle
The oversight duty does not ask directors to be clinicians, statisticians or regulatory specialists. It asks them to make sure the company has a way of telling them when its most important work is going wrong, and to act in good faith when it does.
For a clinical-stage biopharma, the most important work is almost always the trial, and the most consequential failures are usually not fraud. They are a gap between what the data shows and what the company says about it, which grows quietly over months while everyone involved believes in the product. The board is often the only body positioned to notice that gap, because it is the only body that sees both the clinical updates and the public statements.
The practical response is unglamorous. Name the risks that could end the company. Make sure someone on the board owns them. Agree on what gets reported and how fast. Hear directly from the people closest to the data. Check every public efficacy claim against the protocol. Ask the specific question, follow it through and write down what was done. A board that does those things is very unlikely to face an oversight claim. More importantly, it is much more likely to catch the problem while it can still be fixed.
Sources
- In re Caremark International Inc. Derivative Litigation, 698 A.2d 959 (Del. Ch. 1996).
- Stone v. Ritter, 911 A.2d 362 (Del. 2006).
- Marchand v. Barnhill, 212 A.3d 805 (Del. 2019).
- In re Clovis Oncology, Inc. Derivative Litigation, C.A. No. 2017-0222-JRS (Del. Ch. Oct. 1, 2019).
- U.S. Securities and Exchange Commission, Clovis Oncology, Inc., Patrick J. Mahaffy, and Erle T. Mast, Litigation Release No. 24273 (Sept. 18, 2018).
- In re McDonald's Corporation Stockholder Derivative Litigation, C.A. No. 2021-0324-JTL (Del. Ch. Jan. 25, 2023).