An expert witness for the makers of Zicam argues that Zicam does not and can not cause the loss of smell.
Smell disorders have been associated with exposure to over 200 diseases, toxins, or medications [Doty, 2004]. Since many of these contributing factors are commonly occurring, the cause of smell loss in any individual may be multiply confounded. Where zinc nasal gel is implicated in anosmia, confounding factors may include virally mediated olfactory dysfunction, and severe inflammation.
The analysis of AE reports directed to Matrixx regarding its Zicam product -- based on the most aggressive presumptions of exposure and outcome -- indicate that smell disorders rates are equal to or less than that diagnosed in managed care organizations.
3. Dr Jafek's contention that Zicam causes permanent anosmia is not based upon sound scientific data.
Based upon my review of Dr. Jafek's previous testimony and his basis for it, I cannot conclude that he has any scientific basis for his contention that Zicam causes anosmia. Specifically, he bases his allegation on spontaneous reports of complaints of anosmia in some patients who were also exposed to Zicam for treatment of their URI. Based upon my analysis, it appears that his review of these reports and his subsequent poster presented at the American Rhinologic Society meeting represents an anecdotal description of poorly documented clinical findings, with no scientific support or rigorous study. Spontaneous reports of a suspected cause of an event can only at best represent a signal or hypothesis of a possible association. Further, his data is very problematic because it is based primarily upon an internet survey; patients were not always examined by a physician, and there was no objective smell test.
By way of background for the opinions offered, after review of the materials relating to this case, below is a brief summary of my training and qualifications and documents reviewed:
I. Training and Experience
After graduation from Baylor College of Medicine, I trained and obtained boards in Internal Medicine and also obtained a Ph.D. in Pharmacology at the University of California, San Francisco. In addition to earlier clinical and academic practice in Internal Medicine, I began working with the FDA in 1973. My interaction with the FDA began as a member of one of the OTC Advisory Committees that reviewed safety, efficacy and labeling of OTC ingredients. I served in this capacity until 1978. From August 1978 to November 1983,1 served as the Director, Division of Drug Experience (which became the Division of Drug and Biological Evaluation, in 1982) in the Bureau of Drugs (now the Center for Drug Evaluation and Research). From then until June 1985, I served as part-time Special Assistant to the Director of the Office of Biometrics & Epidemiology in the Bureau.
During my tenure as Director of the Division of Drug & Biological Experience at FDA, I was responsible for administering the FDA's Drug Postmarketing Surveillance program, which included spontaneous reports processing, monitoring of drug use, and administering an intra-and extra-mural pharmacoepidemiology program. These activities were for the purpose of gathering data to determine if the marketed drugs continued to be safe as labeled. Accordingly, a major task was to recommend changes in product labeling and to work with the reviewing Divisions to effect labeling updates.
More recently, reflecting a long-standing career interest in drug and biologic adverse effects, I formed and serve as President of a research company, The Degge Group, Ltd. (Degge). This company specializes in epidemiologic studies and regulatory evaluation of drug and biologic safety issues. Degge also advises manufacturers on appropriate management of the safety issues of their products, including regulatory actions and product labeling. I am also President of the Pharmaceutical Education and Research Institute (PERI), an organization that educates personnel in the pharmaceutical and other medical product industry on the appropriate development, regulatory management, marketing and postmarketing surveillance of medical products.
While acting as President of Degge and PERI, I have also served as a member of the Matrixx Scientific Advisory Panel. Under contract to Matrixx Initiatives, Inc. (Matrixx), my colleagues and I conducted an epidemiologic study of smell and taste disorders in administrative healthcare claims databases and prepared a paper which has been submitted for publication. The results from this paper were presented at the University of Pennsylvania Anosmia Symposium, November 8, 2004. This data has also been presented to a research planning panel at the National Institutes of Health's National Institute for Deafness and Communicative Disorders (NIDCD) on March 7-8, 2005. This research forms the basis for the information and data presented in this report.
Further details on my background and qualifications are provided in my curriculum vitae, Attachment A.
II. Documents Reviewed
In preparation for this report, I am listing materials considered in Attachment B to this report.
III. Compensation
My hourly rate for time spent on this matter for consultation and related services is $500 per hour ($500 per hour for testimony at deposition or in court).
IV. Expert Testimony in Previous Four Years
Below is a list of all my deposition testimony during the previous four years. I have not testified at trial during this time.
TABLE1. What is Epidemiology?
A. General
Scientific studies are usually designed as (1) experimental or (2) observational studies. Experimental studies include carefully designed animal studies with controls, and clinical studies that include randomization into exposed or unexposed groups, and ideally blinding of all participants (physicians, patients) as to exposure. These clinical studies, or randomized double-blinded controlled clinical trials, represent the most rigorous tests of causation. However, there are many circumstances where such experimental trials will not provide sufficient information on a question. One of these instances is a rare event, where a question must be studied in large populations. In these cases, observational or epidemiological studies are the primary methods for study.
Epidemiology is the branch of medicine that studies the causes, distribution, and control of disease in populations. Pharmacoepidemiology is a subspecialty of epidemiology that is concerned with the use, benefits and risks of drugs in populations. Pharmacoepidemiologists study data from a wide variety of sources including clinical trials, databases, medical literature and others to detect and evaluate drug related adverse events and assess the risk/benefit ratio of pharmaceuticals scientifically.
The results of epidemiological studies are expressed as a comparison of the outcome or risks between two or more groups. In case reports, rarely are there any fair comparisons made to gauge the risk of the outcome. Statistical descriptions arising from epidemiology studies include an estimation of the “relative risk” that an event is seen in one exposure versus another or no exposure, along with 95% confidence intervals on this estimate based upon the size of the population and the degree of association. Results of epidemiologic (observational) studies may be affected by different biases or confounding in addition to true differences in risk. Biases (e.g., ascertainment, selection) can cause systematic deviations from the true differences due to study aspects such as selection of subjects or exposure to influential information by subjects or researchers. Confounding factors relate to both the exposure and the outcome (e.g., red cars are associated with more auto accidents, but the confounding factor is those who drive red cars). If not identified, they can result in misleading associations.
Good epidemiology dictates that statistical associations precede or support hypotheses of causation. In 1965 Sir Austin Bradford Hill published “The Environment and Disease: Association or Causation” [Hill, 1965]. In this paper, he describes nine factors that need to be assessed to distinguish between association and causation. They are now one of the most frequently invoked guidelines for assessing causation. The criteria are:
1. Strength (is the risk so large that we can easily rule out other factors?)
2. Consistency (have the results been replicated by different researchers and under different conditions?)
3. Specificity (is the exposure associated with a very specific disease as opposed to a wide range of diseases?)
4. Temporality (did the exposure precede the disease?)
5. Biological gradient (are increasing exposures associated with increasing risks of disease?)
6. Plausibility (is there a credible scientific mechanism that can explain the association?)
7. Coherence (is the association consistent with the natural history of the disease?)
8. Experimental evidence (does a physical intervention show results consistent with the association?)
9. Analogy (is there a similar result to which we can draw a relationship?)
B. Design issues in epidemiology
In the claims made for Zicam-induced anosmia, a number of biases and confounders exist that severely limit the ability to conclude true causation of this event. Some of these factors include:
• Biases due to likelihood of seeking medical care
• Biases related to physician recording a medical claim and/or reporting
• Bias due to media and information sources that postulate an untested association
• Confounding by indication (e.g., the reason for using the drug, upper respiratory infection (URI) is also one of the most common causes)
• Confounding by comorbidity or concomitant habits (e.g., smoking)
C. Challenges in designing an epidemiological study of OTC drug-induced anosmia
Rigorous controlled studies to determine any relationship of Zicam to the development of irreversible anosmia, a rare condition, face tremendous challenges in design and logistics. Before designing a study, epidemiologic estimates of the outcome must be made to aid in selecting subjects and identifying confounding factors. Overall, smell disorders in general are relatively common in the population. Between 2% and 5% of adults in the population may develop a smell disorder at various degrees of severity and character. This ranges from mild transient reduction in odor detection to complete and permanent absence of odor detection or identification. Also included are disturbances in the quality of smell such as detection of odors that are not there, or misidentification of odors. Research in patients with smell loss has connected the disorder to over 200 medical conditions, drug, and environmental exposures. Advancing age is most commonly associated with diminished smell. Acute conditions such as rhinitis, sinusitis, upper respiratory infections, and head trauma have been named as associated conditions in case series reports [Doty, 1997, 2001].
To my knowledge, the population incidence and prevalence rate of permanent anosmia -- the complete loss of smell -- has not been researched or published. Based on the overall estimates of the frequency of smell disorders and an understanding of olfactory function, permanent anosmia comprises an extremely small fraction of the population. The olfactory nervous system is unique in that its stem cells regenerate even after toxic or physical injury in a great majority of cases [Doty, 2001; Duncan, 1995]. This is a good physiologic feature since this is the only part of the human nervous system directly exposed to the environment.
People with the common cold, a viral upper respiratory infection, are typical Zicam users. An aggressive estimate of the proportion of total smell disorders is that 25% of all smell disorders are post-viral. That amounts to about 1.25% of the overall population. (In the claims study performed by Degge, upper respiratory infections accounted for 1 in 6 cases or
17% of all smell disturbance diagnoses.) Since a vast majority of patients recover their sense of smell, about 1 in 10 could be postulated to have specifically developed permanent anosmia, or 0.125% of the population. In a clinical trial of post-viral anosmia, investigators would have to recruit at least 24,000 subjects. Adding Zicam exposure as part of this study would require an even greater inflation of the sample size.
Thus, a very large number of subjects would be needed to assess if an increased risk of developing permanent anosmia truly exists after using Zicam. This sort of study is not feasible, and even if such a study were performed, the risk estimate would be confounded by all the diseases and exposures previously cited. When case reports arise claiming a drug-induced adverse event, it is common to ask if a study should be performed to estimate the risk of the event.
2. Study of Smell Disorders in Two Insurance Claims Databases
To estimate the incidence and prevalence of smell disorders, an epidemiologic study was conducted in two large insurance claims databases. A second objective was to examine any statistically significant associations to comorbid conditions and medication exposure. Subjects in this study had medical and pharmacy coverage over a three year period.
Descriptive and case-control analysis of data from the medical claims within IMS LifeLink (IMS) and Constella Managed Care Organization (MCO) medical insurance claims databases containing information on approximately 4.5 million covered lives in the United States from 3-year intervals: January 2000 to December 2002 (IMS), and October 1999 to September 2002 (MCO). Prevalent cases were defined from identification of the International Classification of Diseases - Clinical Modifications Version 9, coding for Disturbances of Smell and Taste (781.1) or Disorders of the First Cranial Nerve (951.8). The overall managed care population was determined based on medical coverage eligibility within the designated time interval. Incident patients were defined as cases found in the 2nd or 3rd year of the study period that did not have a diagnosis in the prior year. For the case-control analysis, three age- and gender-matched controls for each incident case were selected from the same population. Comorbid conditions and concurrent medications were analyzed at 30-, 90-, and 180-days prior to the date of the first case diagnosis. Univariate and multivariate analyses were performed on preselected drug and disease groupings.
Prevalence and incidence of disturbances of smell were calculated by dividing the number of cases found over the mean annual population of eligible persons in the databases. This proportion was normalized to the number of cases per 100,000 people. Odds ratios and 95% confidence intervals of associated conditions and medications previously noted in the medical literature, with other significant associations, were described.
After eligibility exclusions, 727 patients in the IMS database and 1352 in the MCO database met the selection criteria. The mean age was 61.3 and 55.7 years for IMS and MCO, respectively. A majority of patients with SD were aged 50 years or older: 60.4% of IMS cases and 66.1% of MCO cases (Table 1). There were more women than men in each age band starting at 20 years or older. Overall, women outnumbered men by approximately 3 to 2 in both cohorts. This proportion of women to men remained constant for age 50 years or older.
Prevalence. For the 3-year period between January 1, 2000 and December 31, 2002, the mean annual prevalence of SD in IMS was 26.2 cases per 100,000 (95% Cl: 23.1-29.6, eligible population: 2,911,547). Likewise for the period between October 1, 1999 and September 30, 2002, the mean annual prevalence for MCO was 17.2 cases per 100,000 (95% Cl: 15.6-18.7, eligible population: 2,782,145).
Incidence. For the 3-year period between January 1, 2000 and December 31, 2002, the mean annualized incidence rate of SD for IMS was 26.3 cases per 100,000 (95% Cl: 23.1-29.8, eligible population: 941,362). Likewise for the period between October 1, 1999 and September 30, 2002, the mean annualized rate for MCO was 15.9 cases per 100,000 (95% Cl: 14.5-17.5, eligible population: 2,782,145).
Acute sinusitis was significantly associated with smell and taste disturbance in both databases appearing within the 30 days prior to diagnosis (IMS: OR=16.61, 95% Cl: 9.16-30.14; MCO, 2.76, 95% Cl: 1.95-3.90). Also significant in both databases were cerebrovascular disease, chronic sinusitis, and allergic rhinitis. Not predicted prior to the analysis, hypertension and gastroesophageal reflux disease were also found to be significantly associated with smell and taste disturbance. Among medications prescribed 0-30 days prior to diagnosis, a number of antibiotics, antihistamines, and nasal steroids were associated with smell disturbance. Proton-pump inhibitors emerged as a significant predictor, but were not identified prior to the analysis. Use of insulin products was greater in the control group, suggesting an hypothesis of a protective effect on the occurrence of smell and taste disturbance.
In multivariate analyses, a number of predicted associations were found to be significant in both databases. In the IMS database, 15 of 31 pre-selected disease groupings were retained in the multivariate model along with 2 of 12 drug groups. The five strongest predictors of SD were chronic sinusitis, oropharyngeal inflammatory disorders, other upper respiratory disease not including sinusitis, cerebrovascular disease, and systemic viral disease (Table 5). In the MCO group, 12 of 31 disease groupings and 2 of 12 drug groups were retained in the multivariate model. The strongest drug or disease predictors were chronic sinusitis, cerebral degeneration, allergic rhinitis, and acute sinusitis (Table 5). The drug group consisting of nasal corticosteroids, antihistamines, and decongestants were among the significant predictors in both IMS (RR=3.5, 95% Cl: 2.1-6.1) and MCO (RR= 3.9, 95% Cl: 1.6-9.3). Systemic corticosteroids were also twice as likely to be use in patients with SD in the IMS group (RR=2.0, 95% Cl: 1.1-3.9). In the MCO group, calcium channel blocker antihypertensives were found to be significantly associated with SD. Other disease groups found in both databases were disorders of lipid metabolism, cerebral degenerative conditions, and other cardio-pulmonary diseases (excluding heart disease and hypertension).
Skovde Population-Based Study of Olfactory Dysfunction
Bramerson and colleagues investigated the prevalence of olfactory dysfunction in the adult population of Skovde, Sweden [Bramerson, 2004]. The researchers tested the relationship of olfactory dysfunction to age, gender, diabetes mellitus, nasal polyps, and smoking habits. In their cross-sectional, population-based epidemiologic study, a random sample of 1900 adult inhabitants were identified from the municipal population registry of Skovde, Sweden. Subjects
(N=1387, 73% yield) were stratified for age and gender. Subjects were called to clinical visits that included questions about olfaction, diabetes, and smoking habits. Examination was performed with a smell identification test and nasal endoscopy.
The overall prevalence of olfactory dysfunction was 19.1%, composed of 13.3% with hyposmia and 5.8% with anosmia. A logistic regression analysis showed a significant relationship between impaired olfaction and aging, male gender, and nasal polyps, but not diabetes or smoking. In an analysis of a group composed entirely of individuals with anosmia, diabetes mellitus and nasal polyps were found to be risk factors, and gender and smoking were not.
These results are very different from our epidemiologic study for several reasons:
1. The Skovde population was specifically tested for smell loss. Most studies of anosmia show that many are unaware of loss unless tested.
2. Our study only enumerated those patients who:
a. Went to a physician
b. Likely complained of smell loss (e.g., it is unlikely that the physician did routine testing for smell loss)
c. Had their complaint recorded as a medical claim (not all events in a clinical visit are billed).
Thus, the cases in our study likely represent only the most severe and symptomatic variants of anosmia, and our results are not directly comparable to the Skovde study.
Zicam Adverse Event Reporting Rates Relative to Epidemiology Studies
We conducted an unadjusted comparison of the Zicam adverse-event reporting rate to the rate of smell disorders related to upper respiratory tract infections (URI) found in the insurance claims study described above. URI-related events were selected because Zicam is used by patients with the common cold. A population estimate of Zicam-related adverse events rates was calculated using data from Matrixx and annual unit of sales for Zicam Nasal Gel.
All adverse event reports to Matrixx for the years 2000 to 2003 were included and presumed to be related to olfactory dysfunction. For each year, unit sales data were obtained and were converted into persons exposed. Three exposure levels were developed based on unit sales that allowed for three variations computing the “at-risk” population, presuming that every unit sold is used. At the standard level, each unit sold represented one person exposed. A low exposure level presumed that every two units sold represented one person (halving the population, thus increasing the rate). A high exposure level presumed that each unit sold represented two persons exposed (doubling the population, decreasing the rate). Event rates were calculated by dividing the number of reports each year by each of three “at-risk” populations calculated above. These event rates were compared to rates found in the claims study.
As shown in Attachment C, in the IMS and MCO managed care databases, the annual rate of smell disturbance associated with URI was 3.72 and 3.95 cases per 100,000 persons, respectively (Table). As expected, the low exposure group (2 units = 1 exposed person) had the highest rates of adverse event reports from 4.07 to 7.37 cases per 100,000 persons. In this group, only the rate from the year 2000 (7.37/100k) appeared significantly different from the rates found in the claims study. All other rates were similar to or lower than those of the claims study for each year and each exposure level.
Conclusions Based on Epidemiologic Evidence
Smell disorders have been associated with exposure to over 200 diseases, toxins, or medications [Doty, 2004]. Since many of these contributing factors are commonly occurring, the cause of smell loss in any individual may be multiply confounded. Where zinc nasal gel is implicated in anosmia, confounding factors may include virally mediated olfactory dysfunction, and severe inflammation.
The analysis of AE reports directed to Matrixx regarding its Zicam product -- based on the most aggressive presumptions of exposure and outcome -- indicate that smell disorders rates are equal to or less than that diagnosed in managed care organizations.
3. Dr Jafek's contention that Zicam causes permanent anosmia is not based upon sound scientific data.
Based upon my review of Dr. Jafek's previous testimony and his basis for it, I cannot conclude that he has any scientific basis for his contention that Zicam causes anosmia. Specifically, he bases his allegation on spontaneous reports of complaints of anosmia in some patients who were also exposed to Zicam for treatment of their URI. Based upon my analysis, it appears that his review of these reports and his subsequent poster presented at the American Rhinologic Society meeting represents an anecdotal description of poorly documented clinical findings, with no scientific support or rigorous study. Spontaneous reports of a suspected cause of an event can only at best represent a signal or hypothesis of a possible association. Further, his data is very problematic because it is based primarily upon an internet survey; patients were not always examined by a physician, and there was no objective smell test.
4. Other studies relied on by Dr. Jafek are also inadequate.
One basis for Dr. Jafek's contention is cited as the experience with use of zinc sulfate many years ago for prophylaxis of polio in children [Peet, 1937]. Specifically, children were required to assume a supine head-down position to allow administration of the zinc sulfate to their olfactory surface. This position and the instrument used for application cannot be replicated in usual use of a nasal product. The results of this experience revealed that to the extent children did experience anosmia, in almost all cases this was transient, and consistent with the fact that this surface and smell function regenerates due to the stem cells. Thus, even the most radical attempt to eliminate smell is not possible. Further, the substance used in this experiment was zinc sulfate; this zinc salt is not chemically the same as the zinc gluconate.found in Zicam. Thus, for both of these reasons this 1937 data represent very inadequate support for Dr. Jafek's hypothesis.
5. Dr. Jafek's Reports in Zicam litigation cases
In his various reports issued on behalf of an assortment of plaintiffs dating from August 9, 2005 through January 13, 2006, Dr. Jafek repeated his allegations that Zicam causes anosmia. To support this contention, Dr. Jafek cites his own “epidemiology” study - the poster he presented at the American Rhinologic Society meeting which has now been published in that society's journal [Jafek, et al, 2004]. Dr. Jafek's paper cannot be considered an epidemiological study as it does not meet the criteria for epidemiology studies outlined previously in this report. Instead Dr. Jafek presents a series of case reports, which, as noted, cannot constitute scientific evidence of an association.
Dr. Jafek continues to selectively rely on the experiments conducted in the 1930s by Dr. Peet. As previously stated, in those experiments, Dr. Peet placed patients in the Proetz position, with the patient on his or her back with the head hanging upside down over the edge of the table. The instrument used to deliver the zinc sulfate, an atomizer with a long metal speculum, was inserted into the nose and applied the zinc sulfate directly to the olfactory area. Peet, however states the following:
The actual application of the zinc sulfate solution to the olfactory area has been found more difficult than was anticipated.... Direct nasal examination after spraying a large number of children with methylene blue showed that in practically all instances the solution did not go above the middle turbinate if an ordinary atomizer was used with the tip of the spray introduced only slightly within the nostril.... From our experiments in which radiopaque substances and certain dyes were used, it is evident that the spray must be applied directly to the olfactory area. Such application can be made under direct vision with an atomizer with a long, narrow metal tip. (emphasis added)
Furthermore, Dr. Peet stated that in those patients whose sense was lost or impaired by the treatment, their sense of smell “ always returns to normal in a week to two weeks”.
Dr. Jafek continues to conflate the “zinc sulfate” used in the Peet experiments with the use of “zinc gluconate”, the active ingredient in Zicam. They are different chemical compounds and are not absorbed by the body the same way. Zinc gluconate, as contained in Zicam, has not been determined to be toxic to humans in the amount delivered with Zicam use. Furthermore, since Dr. Jafek's claim that Zicam is toxic to the human olfactory epithelium presumes that Zicam reaches it, the remainder of his claim is moot.
6. Conclusions
Based on the results of my epidemiologic study, my knowledge of epidemiology, pharmacology and the pharmacokinetics of Zicam, it is my professional opinion that there is no scientific evidence that use of Zicam causes anosmia.
