Saturday, September 12, 2009

Probability Blindness & Fallacies of FDA REMS

While much speculation and angst continue to surround the U.S. FDA’s planned Risk Evaluation and Mitigation Strategies (REMS) for increasing the safety of opioids and other analgesics, it is possible that the initiative is being driven to begin with by misconstrued data and faulty reasoning. We need to take a closer look at how analgesic-related risk probabilities might be calculated and viewed. In the final analysis, risk-avoidance regulations may incur unintended consequences of making relatively minuscule but manageable problems pertaining to analgesics into major obstacles to patient care.

When it comes to risks of adverse events with medications, there is a tendency to think in black or white terms — a drug is either safe or unsafe — when risk actually is a shades-of-gray concept [Gardner 2008]. Such risk can be calculated as a probability, which is the ratio of the number of particular drug-related adverse events (the numerator) divided by the total number of potential risk-event exposures (eg, number of people taking the drug, total drug doses administered; the denominator) occurring during a period of time. What is an optimally low probability of risk? How high a risk demands remedial action? Answers to these questions are often subjective, and may involve a degree of what Harvard professor Cass Sunstein calls “probability blindness” [Sunstein 2007] whereby judgments are guided more by public sentiment and/or political pressures than by compelling objective evidence of risks overpowering benefits. This is what may be happening with recent initiatives by the FDA to mandate REMS programs for an array of analgesic products.

Analgesic Safety Clearly Established
Incidents of overdose poisonings and deaths associated with analgesics have garnered a great deal of public notice and governmental concern. Yet, when considering the millions of doses administered each year (the denominator in a risk-probability calculation) currently approved analgesic products have remarkably favorable safety records. For example, in a previous blogpost we calculated that propoxyphene-containing products have been used without fatality 99.999% of the time, and many of the deaths that did occur were related to suicides or multiple-drug misuse [see post 7/12/09]. Similarly, using a worst-case scenario, acetaminophen-containing products have been used without liver injury or death 99.86% of the time, and most of the harmful events were due to carelessness in exceeding recommended maximum daily doses [see 7/8/09]. Despite the diminutive severe adverse event probability rates of 0.001% and 0.14% for propoxyphene and acetaminophen products, respectively, both have been denigrated as threats to public health and the FDA (as of this writing) is considering rigorously restricting their future availability.

Methadone provides a further example. A recent concern has been alleged cardiotoxicity of methadone and there is serious consideration being given to requiring ECG monitoring in all patients prescribed the drug for pain or for addiction therapy [see prior posting 8/8/09]. However, in a recent report describing 7 years of observation in methadone maintenance treatment (MMT) programs, encompassing 6,450 patient years of treatment, no deaths were attributed to cardiac arrhythmia [Anchersen et al. 2009]. Still, 4 deaths were of unknown causes that might have been cardiac related; so, a possible rate of fatal arrhythmia per patient year was 0.06% (4/6450 x 100). However, the denominator is actually much larger: assuming it would take only one dose of methadone to trigger a fatal cardiac event and patients are dosed at least once daily, the hypothetical probability becomes a negligible 0.0002% (4/[6450 x 365] x 100). Such a low and speculative risk does not seem to justify routine ECG monitoring in ALL patients.

As with other analgesics, methadone-associated mortality has been a special concern. A recent and credible U.S. government report examining this issue noted that methadone overdose deaths increased roughly 5-fold (from 786 to 4,462) between 1999 and 2005 [GAO 2009], which sounds menacing. Yet, during approximately that same time period there was an 8-fold increase in methadone prescriptions for pain alone (from 531,000 to 4.1 million). Proportionately, it appears that deaths (the numerator) associated with methadone prescriptions (the denominator) were actually declining: the mortality-probability rate decreased from 0.15% to 0.11%. Or, put another way, in the most recent year (2005) methadone prescriptions were used safely without overdose death at least 99.89% of the time. At that, the government report concedes that methadone might be noted in forensic investigations as being present at death but that does not mean it was the cause; thus, the numerators may be skewed larger than they should be by assuming methadone was the culprit when it was only an “innocent bystander,” and incidence rates would be even smaller than noted above. Still, methadone is among the long-acting and extended-release opioid analgesics that the FDA believes need much tighter regulations under REMS for controlling safety risks.

What Are Acceptable Risks?
Similar to the above calculations, severe adverse event risk probabilities would most likely be extremely low for all other analgesics, and it is important to note that the majority of poisonings and fatalities with analgesics have resulted from improper prescribing and/or misuse of the products. In fact, the relatively minuscule incidences of severe adverse events with analgesic products is actually a testament to the effectiveness of the FDA in evaluating, approving, and monitoring medications that are essentially safe when properly prescribed and taken as directed on the product labeling.

Still, the relatively uncommon overdoses and deaths associated with analgesics make for sensational news headlines, especially when celebrities are involved, and the public or safety-advocacy groups clamor for action without knowledge of the already low risks or the ultimate costs of such action. The outcry often is, “It’s worth whatever it takes even if only one life is saved,” and policy makers and regulators listen. Risk-averse public officials take safety very seriously, as Sunstein notes [2007, p 24], “Even when ordinary people can reasonably treat tiny risks as if they were zero, the analysis is altogether different for those whose business is to reduce risks.” However, when it comes to analgesics, the risk-probability threshold at which the FDA decides that remedial regulatory actions are absolutely necessary is unknown, and costs in terms of unnecessary human suffering can be extremely high if such actions serve to deny access to those medications for patients who might benefit from them. FDA representatives have stated that they want to preserve such access, but the recently-mandated REMS program for a newly approved fentanyl product, Onsolis®, demonstrates that the FDA is willing to invoke extremely robust restrictions on analgesic distribution and availability [see blogpost 7/17/09].

Education Not Regulation Needed
How close to zero risk (100% safe use) must an analgesic medication be to satisfy policy makers and the FDA? This question is especially vexing, considering that a large share, possibly most, of the severe adverse events result from misbehaviors (eg, misuse or abuse) related to the drugs. It is interesting to consider that, without much thought (a form of probability blindness) the American public is rather risk-tolerant when it comes to another potent drug — alcohol — which is ubiquitous (61% of all adults imbibe and 12% do so excessively) and hazardous when misused. According to our calculations, using data from the Centers for Disease Control and Prevention for 2006, the risk of death directly related to alcohol use in the U.S. was 0.18% that year, excluding alcohol-involved homicides and non-driving accidents [driving data here, other death data here]. In this perspective, alcohol incurs a higher mortality risk than acetaminophen, propoxyphene, methadone, and probably any other analgesic. And, as was learned years ago, restrictive-distribution regulations (eg, prohibition) do not eliminate alcohol-risk behaviors. Similarly, it seems logical that responsible medication-use behaviors that might serve to lower the risk probabilities for analgesics cannot be achieved by increasingly restrictive regulations.

In sum, it appears that fallacies of REMS result from two misconceptions:
    1. erroneously considering that risks of severe adverse events associated with analgesic medications are at such dangerously high levels that an immediate and rigorous response is necessary, and

    2. wrongly assuming that regulations potentially restricting access to the medications will ameliorate either unintentional or volitional misbehaviors contributing to severe adverse events.
We believe a better approach would focus not on more regulation but on better education for healthcare providers, patients, and the public regarding safe analgesic use. Such education may not be the mission of the FDA, in which case they might step aside and turn the task over to other public and private agencies. Of further interest, the above mentioned Cass Sunstein has just been appointed America’s new “regulatory czar,” with broad-ranging responsibilities for reviewing and guiding federal regulations; perhaps, he will bring an enlightened vision for overcoming the probability blindness surrounding opioid REMS initiatives. Any opinions? Add a comment.

> Anchersen K, Clausen T, Gossop M, Hansteen V, Waal H. Prevalence and clinical significance of corrected QT interval prolongation during methadone and buprenorphine treatment: a mortality assessment study. Addiction. 2009;104(6):993-999 [see
> GAO (U.S. Government Accountability Office). Methadone-Associated Overdose Deaths. March 2009. GAO-09-341 [see
document PDF].
> Gardner D. The Science of Fear. Why We Fear the Things We Shouldn’t and Put Ourselves in Greater Danger. New York, NY: Dutton; 2008.
> Sunstein CR. Worst Case Scenarios. Boston, MA: Harvard University Press; 2007.