In the Health Affairs article, “Era of Faster Drug Approval Has Also Seen Increased Black-Box Warnings and Market Withdrawals,” published in the August issue, Cassie Frank and coauthors compare the number of approved prescription drugs that received black-box warnings or were withdrawn from the market for safety-related reasons prior to the 1992 Prescription Drug User Fee Act (PDUFA) with black-box warnings and safety-related withdrawals in the post-PDUFA era.
PDUFA for the first time authorized FDA to collect user fees from brand-name manufacturers that submitted New Drug Applications, with the funds being earmarked for more review staff (not until 2007 were funds also permitted to be used to expand post-approval safety surveillance capacity).
As a quid pro quo, the FDA was required to act on all new drugs within a fixed deadline: drugs given priority review designations because they were particularly promising therapies offering substantial improvements in treating serious conditions were to be reviewed within 6 months and standard review drugs were to be reviewed within 12 months (later shortened to 10 months in 2002). By all accounts, PDUFA substantially expedited the review process. The review times for new molecular entities decreased from an average of 33.6 months between 1978 and 1986 to about 10 months for drugs approved between 2001-2010.
Frank and coauthors found that drugs approved in the 16 years after PDUFA were more likely to have a black-box warning or be withdrawn from the market for safety reasons than drugs approved prior to PDUFA between 1979-1986 (26.7 percent vs. 21.2 percent; odds ratio 1.35, p <0.05). They concluded that, all else being equal, clinicians should favor older drugs over newer ones as older drugs’ risk profiles are better understood due to longer time on the market.
Frank et al. also recommend that recently approved drugs be specifically designated as “new to the market” in their labels and advertising. As the authors note, such a strategy already exists in the UK, where a “black triangle” is added to the label of new medications for at least its first 5 years in the market, signaling its being under greater monitoring by the UK’s Medicines and Healthcare products Regulatory Agency (MHRA).
Black-Box Warnings: FDA’s Strongest Risk Communication
Black-box warnings, officially known as boxed warnings, are the most serious safety warnings that the FDA can impose on a drug. First instituted in 1979, black-box warnings are intended to prominently highlight a serious adverse reaction that may be life-threatening, fatal, or disabling, or to provide enhanced notice about a serious adverse reaction that could be prevented or reduced in frequency by appropriate limitations on use of the drug.
The intent of a black-box warning is to provide a clarion signal to physicians and patients to consider the adverse event and its magnitude in their benefit-risk analysis prior to prescribing the drug to which it is attached. Black-box warnings may be drug-specific, as in the case of transdermal fentanyl, which has a black-box warning addressing respiratory depression. Black-box warnings can also be added to entire classes of drugs. For example, in 2008, a black-box warning was added to all fluoroquinolone antibiotics, warning of increased tendon injuries in older patients, those on steroids, and transplant recipients.
Adverse event data leading to a black-box warning come from numerous sources, including clinical trials submitted to the FDA prior to approval (e.g., QT prolongation or Stevens-Johnson syndrome), post-approval studies, clinical trials submitted to the FDA after approval to support a new indication of the drug, meta-analyses, and other publications in the academic literature. Accumulating data from MedWatch and the FDA Adverse Event Reporting System may also inform a warning.
Finally, litigation over drug safety and its resulting media attention have influenced addition of black-box warnings in the past. For example, litigation over the association between long-term use of the antiemetic medication metoclopramide (Reglan) and a severe neurological side effect called tardive dyskinesia—that ultimately reached the Supreme Court—led the FDA to add a black-box warning to the drug’s label in 2009. A definitive causal relationship between a drug and adverse event is not required to support a black-box warning.
Maintaining Black-Box Warning Value, Risk Threshold, and Data Availability
Black-box warnings can have substantial effects on physician prescribing practices. For example, in the early 2000s, an investigation from New York Attorney General Eliot Spitzer and litigation from individual plaintiffs led to the uncovering of numerous trials of brand-name selective serotonin reuptake inhibitor antidepressants in pediatric patients that had been withheld from the public by the sponsoring manufacturers even as they widely promoted the drugs for these purposes. Many of these trials showed no benefit of taking the drug but did show an increased risk of suicidal thoughts or behaviors in adolescents taking drugs like paroxetine (Paxil).
In response, the FDA imposed black-box warnings on SSRIs including paroxetine about these risks of their use in adolescent patients, additionally pointing to the only drug (fluoxetine [Prozac]) that had been approved in this population. Studies since that time have shown a marked reduction in SSRI antidepressant use among adolescents, though how much the trend is due to the publicity surrounding these cases or the black-box warning itself cannot be known.
Other studies have pointed to a rise in adolescent suicides during this time as well, further reinforcing the fact that despite the association between the black-box warning and a reduction in prescribing, the optimal treatment of adolescent patients with mental health issues remains a complex issue with multiple contributors.
Because of their substantial potential impact on prescribing, the question of whether a drug deserves a black-box warning can be quite controversial. In the past, some pharmaceutical manufacturers have strongly lobbied the FDA against imposing black-box warnings on their drugs. One motivation may be that the inclusion of a black-box warning can greatly reduce expected sales.
For example, litigation over the diet drug dexfenfluramine (Redux) revealed that the manufacturer actively campaigned against addition of a black-box warning concerning the risk of pulmonary hypertension. The warning was never included. However, the drug was withdrawn from the market a year after this controversy for safety reasons based on a risk of cardiac valve disease and pulmonary hypertension that outweighed its modest effect on weight loss.
The dexfenfluramine case highlights the contentiousness over imposition of a black-box warning and the key role of the FDA as arbiter. If the FDA has too low of a threshold for imposing a black-box warning, then the number of warnings will proliferate, which may dilute the significance of the warning mechanism in the eyes of consumers. Alternatively, if the FDA has too high of a threshold, then potentially important safety information may be missed because it is buried in less prominent parts of the label. A good example of this tension can be seen in the case of the COX-2 non-steroidal anti-inflammatory drug (NSAID) rofecoxib (Vioxx) and its aftermath.
Evidence of its cardiovascular toxicity dated back to before its FDA approval in 1999, and notably included the 2001 Vioxx Gastrointestinal Outcomes Research (VIGOR) trial, which indicated a four-fold increase in likelihood of myocardial infarction with Vioxx versus naproxen, a non-selective anti-inflammatory drug. With Merck actively downplaying the cardiovascular side effects, the FDA never imposed a black-box warning on the drug before its withdrawal from the market in 2004; in retrospect, such a move may have reduced the 20 million patients who were exposed to the drug.
Shortly after rofecoxib’s withdrawal, a black-box warning was added to the U.S. label of the remaining available COX-2 inhibitor, celecoxib (Celebrex), reporting that “Celebrex may cause an increased risk [of cardiovascular disease]… All NSAIDs may have a similar risk. … Patients with cardiovascular disease or risk factors for cardiovascular disease may be at greater risk.” At the same time, a black-box warning with the same wording was added to all non-selective NSAIDs, even though the actual association with cardiovascular disease among the many members of the non-selective NSAID class varies considerably from high (diclofenac) to low (naproxen).
Though since that time, prescribing of NSAIDs may have fallen among some patients with gastrointestinal and cardiovascular diseases, the FDA’s approach to the NSAID black-box warning case was arguably over-broad, perhaps as a reaction to the rofecoxib public health disaster. By contrast, in Europe, COX-2 inhibitors were contraindicated in all patients with ischemic vascular disease, while no warnings were imposed on non-selective NSAIDs.
The article by Frank et al. provides important data about the increased popularity of black-box warnings in the U.S. The authors contend that one reason that such warnings are growing in prevalence is that PDUFA and other expedited approval programs have led to the approval of more new drugs before they could be rigorously evaluated in pre-approval clinical trials. When data about these drugs’ safety then emerges after approval, one response has been to add black-box warnings or, in extreme cases, withdraw the drug for safety reasons.
Experiences with dexfenfluramine, rofecoxib, and other similar cases may have also affected the FDA’s threshold for seeking black-box warnings. But to preserve the value of this drug safety communication tool, black-box warnings must be data-driven, appropriately nuanced, applied in a timely fashion despite predictable pushback from the manufacturer, and not over-used.
To help achieve these goals, we need additional research on the impact of black-box warnings, and even the wording within a black-box warning, and on physician and patient perceptions of drug risk. Warnings like the one on celecoxib that the drug “may cause” a side effect with additional context that all NSAIDs cause the side effect are likely confusing to patients and physicians considering whether to use the product.
Since such work would necessarily involve patients in the research process, it is a natural fit for a funding agency like the Patient-Centered Outcomes Research Institute. One positive outcome of research in this area would be adoption of more specific standards for how to communicate risk within a black-box warning that may promote safer prescription drug use.