Written Statement for the January 30, 2009 Joint Meeting of the Anesthetic and Life Support Drugs Advisory Committee and the Drug Safety and Risk Management Advisory Committee
Statement of Sidney M. Wolfe, M.D.
Director, Health Research Group at Public Citizen
Hearing on Propoxyphene
Before FDA’s Anesthetic, Analgesic and Rheumatologic Drugs and Drug Safety and Risk Management Advisory Committees
There is little doubt that were propoxyphene and propoxyphene-containing products to come before these committees today for approval, based on what is now known, they would be rejected because of one of the most unfavorable benefit-to-risk ratios ever seen for a drug. This is not to say that there was insufficient evidence for a ban 30 years ago when we first petitioned the FDA to withdraw the approval. But the forceful and successful war then waged by Lilly in opposition to such a ban tended to drown out evidence of minimal benefit and rapidly growing evidence of life-threatening, often lethal harm.
Propoxyphene-related deaths: DAWN
The Federally-funded Drug Abuse Warning Network (DAWN) collects data from both emergency rooms and from medical examiners concerning drugs which have been determined to be related to emergency room visits or deaths. This medical examiner data has changed, over the years, in that the number of entities reporting has increased and definitions such as accidental death, have been modified or combined. In 2006 and 2007, the same definitions were used. DAWN Reporters are trained to report only those drugs related to the death.
SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network 2008 (9/2008)
It can be seen that there was an increase in propoxyphene-related death cases from 446 in 2006 to 503 in 2007 but there was a concomitant increase in the number of jurisdictions reporting. The cases in which the death was categorized as accidental outnumbered those in which the manner of death was stated as suicide in both years. The DAWN data do not imply causality but rather that the death was related to the drug.
It must be noted that these data represent only a fraction of the U.S. population.
In 2007, the population covered by DAWN was approximately 109,048,573. That’s about 36% of the total population for 2007 which the U.S. Census estimated at 301,290,332. However, the population covered by the DAWN mortality component is primarily from metropolitan areas, and is not based on a statistical sample. Thus the DAWN data must not be used to extrapolate to the Nation. As of 2007, there was complete state reporting from only 10 states, not including Florida.
Propoxyphene-caused deaths: Florida
The system of collecting and recording Medical Examiner data in Florida provides more details than the DAWN data. The presence of a drug in a decedent is categorized as “cause” if the Medical Examiner concluded that there was enough of the drug present to have been either the sole cause or a contributory cause of the death. Other drugs are listed as merely “present” if the Medical Examiner did not conclude that the drug played a role in the patient’s death. In their words: “The state’s medical examiners were asked to distinguish between the drugs being the “cause” of death or merely “present” in the body at the time of death.”
SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network 2008: (9/2008)
From: Drugs Identified in Deceased Persons by Florida Medical Examiners: 2007. Report by the Florida Department of Law Enforcement
The chart below shows the total propoxyphene-related deaths in Florida and the subset in which the drug was found to be a cause, for the years 2003 through 2007.
It can be seen that in 2007, for example, of a total of 314 propoxyphene-related deaths reported in Florida, 85, or 25%, were cases in which the medical examiner concluded that the drug was a “cause of death”.
The next table examines, for 2007, the categories into which the 85 propoxyphene-caused deaths were divided.
Of the total 85 deaths in which propoxyphene was listed as a cause, in 25 cases it was the only causal drug and in another 60 cases, there was one or more other drugs also judged to be causal to the deaths, along with propoxyphene. Of these 85 deaths, 66 were judged to be accidental and 16, suicidal. The further breakdowns of these data can be seen in the chart above. It should be noted that with 18.7 million people, Florida constitutes 1/16 of US population.
FDA Review of Efficacy
The conclusion of the FDA’s extremely comprehensive and carefully done review of propoxyphene efficacy states:
“there is evidence that propoxyphene possesses weak analgesic effects in patients with acute pain compared to placebo….While most of the studies show that in combination with acetaminophen, the propoxyphene component appears to contribute little or no additional analgesic effect beyond the efficacy of the acetaminophen when studied in patients with acute pain, there is at least one study that does support the contribution of propoxyphene to the efficacy of the combination.”
The FDA review not only included the NDA data submitted to the agency but a series of meta-analyses and individual published randomized trials.
One of the published studies reviewed by the FDA (Hopkinson-1973) found that:
“A global evaluation at the end of treatment (4 hours) showed no difference between the PPX/APAP combination and the single-ingredient APAP in the percentage of patients reporting “effectiveness:”
• 64% in the combination group
• 62% in the single-ingredient APAP group
• 34% in the single-ingredient propoxyphene group
• 30% in the placebo group”
An additional finding here, noted in some other studies, was that single ingredient APAP was more effective than single ingredient propoxyphene which, itself, was not distinguishable from a placebo. In some other studies, propoxyphene was superior to a placebo.
The FDA review stands on its own merits and, in essence, finds that a/ the addition of propoxyphene to acetaminophen does not result in a statistically significant improvement in pain relief compared with acetaminophen alone and
b/ that propoxyphene alone has only “weak analgesic effects”.
Statement by Dr. Steven B. Karch
I am a former assistant medical examiner in San Francisco where my practice was confined mainly to the investigation of deaths involving drug toxicity. I am a member of the Royal Academy of Physicians, Faculty of Forensic and Legal Medicine (London). My textbook, Karch’s Pathology of Drug Abuse (4th Edition published December 15, 2008), is widely used by pathologists and medical examiners in the United States and Europe. It is generally considered authoritative. This textbook contains a subsection specifically devoted to propoxyphene-related deaths.
I have been asked to confine my analysis to the cardiologic and toxicological issues of the effects of dextropropoxyphene (DPX) in clinical practice from the viewpoint of a death-investigator interested in evidenced-based medicine.
Dextropropoxyphene received FDA approval more than 50 year ago, at a time when it was impossible to measure propoxyphene metabolite directly, at a time when the family of CYP450 polymorphisms had not yet been discovered, when the ion channels in the heart had not yet been characterized (or, in some cases even discovered), and at a time when unintended drug reactions were difficult to identify and often went unnoticed.
The principal oxidation product of DPX is norpropoxyphene (NPX); it has a much longer half-life than DPX and tends to accumulate. In itself, that might not be a problem, but NPX accumulates in the heart where it exerts disruptive toxicity.
Use of this drug involves important safety issues. It is widely used, around the world, in suicidal gestures and for frank suicide. The physical properties of this drug make it particularly well fitted since it is widely available, rapidly absorbed, can exert toxicity even when the plasma concentration is within the “therapeutic range.” Indeed, most who attempt suicide with DPX die before they reach the hospital.
Simple DPX poisoning is straightforward and easy to understand. Even though DPX is a very weak mu agonist, when present in sufficient quantities, or taken with alcohol, it can cause respiratory depression.
However, the most dangerous aspect of DPX is that its cardio toxic oxidative metabolite, norpropoxyphene (NPX), is very long acting, and is also 2.5 times more potent than its parent compound in producing cardiac depression. It has a half-life (time before ½ of the substance is cleared from the body) of approximately 36 hours. In other words, the half-life of NPX is nearly three times longer than that of propoxyphene, which means that it will accumulate.
DPX accumulates particularly in the heart and liver. Because NPX also accumulates in the liver, it can cause disruption in the metabolism of many of the most important drugs now in use.
In the heart, NPX blocks both the IK and hERG currents. Blockade of the former can cause conduction delay and even heart block. Blockade of hERG (slow rapid depolarizing K channel) may cause QT interval prolongation leading to torsades des pointes and sudden death.
Patients who have overdosed with DPX have widening of the QRS complex, and the degree of widening is dose dependent. These findings have clinical relevance to the management of patients with DPX poisoning; heart block must be anticipated.
Disruption of liver function may be even more profound. CYP3A4 is the major CYP enzyme catalyzing DPX metabolism. The variability in pharmacodynamic and pain relieving effectiveness of DPX is likely due to large inter-subject genetic variability in hepatic CYP3A4 expression and/or drug-drug interactions.
DPX is also a competitive inhibitor of CYP3A4 and this enzyme oxidizes a large number of other important drugs. These include, calcium channel blocking agents, macrolide antibiotics, isonazid, and proton pump inhibitors. Perhaps most importantly, if the breakdown of carbamazepine, a commonly used anticonvulsant, is slowed because of propoxyphene, toxic levels may accumulate.
Strong evidence suggests that DPX is also an inhibitor of another drug-metabolizing enzyme, CYP2D6. This opens up the possibility for other types of drug interactions. Most beta blockers are metabolized by CYP2D6. A report of bradycardia in a user of metoprolol (a beta-blocker) suggest that symptomatic drug interactions are, in fact, occurring.
DPX toxicity increases in the presence of alcohol. When DPX is co-administered with Etoh, first pass hepatic metabolism is decreased, which means that DPX concentrations increases. Etoh is frequently present in DPX-related deaths. In a study of 123 DPX-related suicides in the UK, alcohol was found to be involved in 58.5% of the cases. In addition, these individuals generally had lower blood propoxyphene levels, and consumed fewer tablets.
Unless DPX is the only drug present, determining the cause of death can be problematic. When confronted with a decedent who has taken multiple drugs, including DPX, it is often impossible to tell which, if any of the drug was the cause of death, and the diagnosis is usually polypharmacy. As a consequence, many DPX-related deaths go unreported, where death is simply listed as a consequence of “polypharmacy,” and the risk posed by use of this drug goes under-estimated.
There are numerous reasons for banning this drug:
(1) It is a dangerous drug: Large amounts of DPX are rapidly absorbed from the GI tract very quickly, making attempted suicide difficult to treat.
(2) Even modest amounts of this drug can cause lethal cardiac arrhythmias in any individual with an undiagnosed hERG genetic polymorphism.
(3) Use of DPX can lead to toxic levels of antibiotics and anticonvulsants
(4 The drug is not particularly effective. For far less money, patients would get more pain relief if they took aspirin or acetaminophen.
Clinical Pharmacology Data
Because there are very little clinical data on patients found with the unexpected deaths typical of coroners’ cases, it is useful to examine clinical data from patients who lived after their overdoses of propoxyphene. A very unique series of 222 consecutive patients admitted over a six-year period to just one Danish hospital provides extremely useful data confirming what the FDA has referred to in the pre-clinical studies they described.
The first figure below describes the findings on admission of these patients.
As can be seen, 48% of patients had heart failure/impaired circulation, 15% had a cardiac arrest (asystole), 9% had abnormally slow pulse and 41% had an abnormal electrocardiogram, including 19 patients with a ventricular arrhythmia.
The authors commented on experimental evidence of a negative chronotropic effect (slower pulse) and negative inotropic effect (weaker heart contraction) with propoxyphene that would explain some of these clinical findings, including the fact that only a few of the patients with circulatory failure exhibited a compensatory tachycardia (faster pulse) to make up for the decreased circulation because of the negative chronotropic effect..
The chart below illustrates further life-threatening clinical findings on admission:
44% of patients were in acute respiratory failure and had to be placed on a ventilator, 10% of patients had convulsions and 73% of patients were in a stupor or in a coma.
The final slide from this study concerns the 17 patients (8% of all 222 patients) who died despite apparently excellent care in the ICU.
Nine (53%) of the deaths were from heart failure, a total of 13 (76%) of deaths were from all cardiovascular causes and 4 deaths (24%) were from brain damage.
As we discussed in our petition, there is a very narrow margin of safety with propoxyphene, partly because of the accumulation, even at normal doses, of the cardio toxic metabolite, norpropoxyphene. This is borne out by pharmacokinetic studies, especially the findings with multiple doses over time and more so in older patients (age 70-79).
The median maximum blood levels (and ranges) for a single dose in healthy elderly people were 156 (39-366) micrograms per liter for propoxyphene and 193 (112-283) for norpropoxyphene. For multiple sequential doses (the equivalent of just three 65 milligram doses of propoxyphene a day) for one week, the median level for propoxyphene was 239 (151-509) micrograms per liter and, for norpropoxyphene, it was 1100 (775-1500). Thus, although blood levels of the shorter half-life propoxyphene increased from 156 to 239, from the single to multiple doses (an increase of 1.5 times), levels of the longer half-life cardio toxic norpropoxyphene increased from a median value of 193 to 1100, an increase of 5.7 times when the drug was used over time in elderly people..
In our petition, we had cited data from patients being given chronic doses of propoxyphene having similar blood levels, in many cases using doses such as those in the data above.
Type of Subject
Duration of Drug Use
Maximum Blood Concentration (μg/l)
Mortality Compared to codeine-containing combinations with acetaminophen: Effect of UK Ban
A recent study estimated the frequency of overdose and death for the three most popular acetaminophen-opioid compound analgesics: propoxyphene and acetaminophen and two different codeine preparations and acetaminophen. Adjusting for relative amounts of prescriptions of the three drugs, overdoses involving propoxyphene and acetaminophen Scotland were 10 times more likely to be fatal (24.6 (19.7, 30.4)) when compared with cocodamol (2.0 (0.88, 4.0)) or co-dydramol (2.4 (0.5, 7.2). The authors estimated from this study that withdrawal of propoxyphene and acetaminophen would prevent 39 excess deaths per annum in Scotland alone.
A very recently published study measured the effect of the UK announcement of the ban of propoxyphene and acetaminophen (called co-proxamol in the UK) on suicides in Scotland. The study showed early evidence of a reduction even before the ban was finalized in January, 2008. In the 5 years pre-legislation (2000–2004), there was a mean of 37 co-proxamol deaths per year. In the first year after the legislation was announced, 2006, the number had fallen to 10 per year in Scotland. The average number of total Scottish drug poisoning deaths in the 2000-2004 interval was 171 per year. Rather than any compensation for the decrease in co-proxamol use by increases in other drugs which could result in suicide, there was an actual decrease in all poisoning deaths to 126 in 2006.
In a personal communication with Dr. Nick Bateman, of the Royal Infirmary of Edinburgh, an author of both of the studies described above, he informed me that in England, unpublished data show that there has been a 90% decrease in co-proxamol deaths in that country since the ban was announced. He also stated that he supports the efforts to ban these products in the U.S.
Current Prescribing of Propoxyphene in the U.S.
Despite the evidence for the serious dangers of this drug and its marginal effectiveness, for most of the past 35 years or more, propoxyphene-containing drugs, now mainly the combination of propoxyphene and acetaminophen, have been among the top 25-selling drugs in the U.S. The data for the most recent eight years shows the continuation in top-25 status for the generic combination of propoxyphene and acetaminophen (similar to brand-name Darvocet): In the most recent year for which data are available, 2007, there were 21.3 million prescriptions filled for the generic combination of propoxyphene and acetaminophen, making it the 21st most-prescribed generic drug in the country.
Other top 25 drugs with vastly different benefit risk ratios
Two other top-25 generic drugs, oxycodone (16th) and warfarin (22nd), are worth mentioning because although they, too, have significant risks, they have clear and significant benefits, in the case of warfarin, a unique benefit. Because of proven, important efficacy, it would not be sensible to even consider removing either oxycodone or warfarin from the market. Risk mitigation strategies for such drugs, including the criminal prosecution of Perdue for mis-promoting Oxycontin and proper cautions for doctors and patients for both drug are necessary. For propoxyphene, market withdrawal is the only rational alternative.
Recent overview articles concerning propoxyphene
In a review entitled, Propoxyphene (Dextropropoxyphene): A Critical Review of a Weak Opioid Analgesic That Should Remain in Antiquity, the authors, all members of the Barkins family, concluded that “propoxyphene offers no therapeutic advantages over any other opioid. The propoxyphene-induced iatrogenic events and risk outweigh any perceived benefits that could be achieved. Any therapeutic benefit from this drug has long been overdue for disuse. The time has arrived for its imminent disposal into antiquity. More suitable therapeutic agents are available with a better risk-benefit ratio and less end organ damage such as those with CNS, cardiac, and pulmonary systems.”
In another review, focusing on efficacy, Sachs found that “propoxyphene has poor efficacy and significant side effects. A meta-analysis of 26 trials involving 2,231 patients compared the combination of acetaminophen and propoxyphene with acetaminophen alone or placebo. The narcotic combination offered little benefit over acetaminophen alone. Another systematic review found that the NNT for a single 65-mg dose of propoxyphene to achieve at least 50 percent pain relief
was 7.7 (95 percent CI, 4.6 to 22) when compared with placebo. For the combination of propoxyphene and acetaminophen (650 mg), the NNT was 4.4 (95 percent CI, 3.5 to 5.6) when compared with placebo, similar to the NNT for acetaminophen alone….Thus, propoxyphene provides minimal if any additional analgesia to acetaminophen alone and is associated with significant adverse effects. It cannot be recommended for routine use.”
When we first petitioned the FDA to ban propoxyphene-containing drugs 30+ years ago, our petition for a ban was supported by a number of medical examiners from around the country. Our more recent petition was prompted by the conclusion of the UK government, after a commission reviewed all of the evidence, that efficacy of this product “is poorly established and the risk of toxicity in overdose, both accidental and deliberate, is unacceptable.” They further said that “It has not been possible to identify any patient group in whom the risk-benefit [ratio] may be positive.”
In response to an article by former Mayo Clinic Chief of Oncology, Dr. Charles Moertel, attacking the use of propoxyphene 36 years ago by stating that “It appears that factors other than intrinsic therapeutic value are responsible for the commercial success of propoxyphene,” a representative of Lilly wrote: “Darvon products have won a remarkable acceptance by patients and physicians since their introduction.” In a letter in response to the Lilly letter, Dr. Moertel replied,
“The implication that general acceptance of a therapeutic procedure by physicians in a given era constitutes obligate proof for effectiveness is not tenable. If this were true, we would still be bound to the mummy dust, unicorn’s horn, leeching, purgatives and mustard plasters universally endorsed by our forebears. We must constantly offer challenge to all our sacred cows, so that our patients may afforded the highest care a the most reasonable cost.”
Finally, I contacted Dr. Donald Kennedy who was FDA Commissioner when we originally filed our petition to ban propoxyphene in 1978. When I asked him, after sending him our 2006 petition to ban the drug, if he would now support our petition to ban the drug, he replied, “You can sign me up!!”
As I said in the beginning, propoxyphene has one of the most unfavorable benefit-to-risk ratios I have ever seen for a drug. This committee will hopefully agree with this and recommend the beginning of a two-year phased withdrawal of these products.
Thank you for taking the time to listen to my presentation.
 FDA Efficacy Briefing Document, page 18.
 Acta Anaesthesiol. Scand 1884;28:661-665
 Br. J. Clin. Pharmac.1989; 28:463-469
 Br J Clin Pharmacol 2005;60:444–447
 Br J Clin Pharmacol 2008;66:290-293
 American Journal of Therapeutics 2006’13:534–542
 A comparative evaluation of marketed analgesic drugs. N Eng J Med 1972;286:813-815
 N Eng J Med 1972;286:1158 (letter)
 N Eng J Med 1972;286:1158(letter)