Issue: 60 Page: 48-53
A Rational Perspective on Adverse Events Reports on Herbs: Misinterpretation of Adverse Reactions Tabulated in the TESS Annual Report of the American Association of Poison Control Centers as They Relate to Ephedra Dietary Supplements
by Mark Blumenthal, Richard Kingston
HerbalGram. 2003; 60:48-53 American Botanical Council
A Rational Perspective on Adverse Events Reports on Herbs:
One of the persistent challenges facing professionals and regulators
alike is that of obtaining information regarding adverse events potentially associated
with herbs and then, sharing this information in a meaningful way with the public.
Despite the fact that many of these substances have been used for centuries and
a substantial body of historical data exists to support their relative safety,
there is still a need to ensure that current usage in terms of types of preparations,
strength and concentration, doses, indications and use patterns do not change
a given safety profile. For these reasons there is still a need to monitor the
clinical experience related to the use of these substances, especially as it pertains
to adverse effects.
There is a variety of venues whereby adverse events involving
herbs might be discussed. Health professionals often rely on information contained
in scientific reports that appear in the medical literature. In these instances,
adverse effect information may be included in the results of a clinical trial
or represent the experience of a random sampling of patients taking a given herb.
Another more common scientific reporting method involves that of anecdotal or
case reports. In these cases, independent practitioners evaluate and summarize
information regarding a given adverse effect or experience involving a specific
patient and the clinical observations that may suggest a temporal relationship
between that patient’s condition and other events or external factors which
might include the taking of an herbal product.
In addition to these types of scientific reports, there are
also situations whereby incidents involving adverse effects or injury involving
herbs may be "spontaneously reported" to a variety of entities. These
reports might be received by individual health professionals, regulators or other
governmental officials, the media, organized medical information services such
as poison centers, and manufacturers or distributors of dietary supplements. These
reports are often tabulated into a database and further subjected to analysis
and interpretation. The nature of these reports is often quite different from
other scientifically documented incidents. Although this type of information is
vital in monitoring the safety landscape, it must be recognized that the reports
are typically not the same as those contained in prospective clinical trials or
even isolated anecdotal reports from independent practitioners where readers may
examine all (or at least most) of the available facts regarding a given case and
draw their own conclusions. Spontaneous reports have both strengths and weaknesses,
both of which must be acknowledged and considered when attempting to evaluate
the significance of individually reported incidents or aggregated incidents.
Ephedra Under Scrutiny
Recently, the herb ephedra (Ephedra sinica Stapf, Ephedraceae)
has received substantial scrutiny regarding its adverse effect profile. While
some of the concern has been generated by legitimate clinical trials where certain
types of adverse effects have been documented, much of the information comes from
data contained in spontaneously reported incident data collection systems. In
these instances, aggregated spontaneously reported incident data has been interpreted
to suggest that a safety concern exists. As an example, in March of 2003 a paper
was published in the Annals of Internal Medicine1 that reviewed
the relative incidence of adverse effects "related" to ephedra that
were reported in the American Association of Poison Control Centers (AAPCC) incident
database known as the Toxic Exposure Surveillance System (TESS). The article received
relatively widespread media attention and was cited as a basis for the need for
increased warnings or a ban on ephedra dietary supplements. Before looking more
closely at this article, a brief review of TESS as well as the nature of spontaneously
reported incident data should be considered.
The TESS Database
Since 1984, the experience of poison centers has been reported
through the publication of an annual report of TESS.2-19 Currently,
these reports appear annually in the September issue of the American Journal
of Emergency Medicine. These data have been quoted in numerous publications
addressing issues of toxicology and the epidemiology of poisoning in the United
States.
The focus of certified regional poison center operations is
telephone-based poison information. There is no pre-registration, payment or any
other requirement of the callers who use the service and calls may be made anonymously
although phone numbers may be identified by the "Caller ID" service
provided by local phone companies. Calls or reports to the center are voluntary,
as there are no local or national requirements that any given incident must be
reported. The center’s phone lines are considered "hotlines" and
some centers align themselves or co-locate service with nurse "advice line"
programs, drug information centers, industry product surveillance services, or
community 911 emergency phone operations.
"Information" calls are separated from "exposure"
calls and have not been tabulated in TESS annual reports. That is, there does
not have to be an actual case of "poisoning" for an individual to contact
a poison center. An individual need only perceive that an adverse effect or poisoning
related threat may exist or has occurred. The poison center specialist assesses
the incident and determines the most appropriate method to mitigate injury, if
injury is likely. This may include advising the caller on appropriate treatment
options or referring a patient to a local healthcare facility for further medical
evaluation. Recognizing the limitations of telephone patient assessment, it is
usually the practice of poison specialists to consider a worst-case scenario regarding
the incident in question. This may result in referral of individuals without the
presence of any adverse effects in questionably toxic situations so that healthcare
providers can provide a more in-depth hands-on patient assessment.
It is important to realize that not all calls represent exposures
and not all exposures represent poisoning. In more than 72% of exposure related
inquiries, no adverse effects resulting from the exposure are reported. When including
those cases in which no more than minor effects are reported, the total represents
more than 87% of the cases.19 It is also important to recognize that
not all reported incidents are created equal, especially those calls classified
as "adverse effects" secondary to drugs, chemicals, or other agents.
Some reported effects might be as minor as self-limiting brief irritation from
a simple irritant in the eye whereas other incidents may represent serious life-threatening
effects such as arrhythmias or respiratory arrest.
Each case is documented and typically includes:
• General demographics regarding the caller and the
exposed individual;
• Reason for exposure;
• Substance(s) involved in the incident;
• Reported effects and subjective assessment of relatedness
of each effect to the substance involved;
• Treatments recommended (by the poison center or others)
and/or rendered;
• Patient disposition;
• Subjective assessment of the relatedness of the incident
severity with the substance(s) involved.
Each case is assessed as to its overall outcome. The options
are "no effect"; "minor effect"; "moderate effect";
"major effect"; "death"; "no follow-up, nontoxic";
"no follow-up, minimal toxicity", "no follow-up, potentially toxic";
and "unrelated". Unless specifically coded as "unrelated",
the severity rating implies that the overall outcome is the result of the exposure
to the substance(s) identified in the incident. Although a "causal link"
is suggested with this type of classification system, many epidemiologists do
not believe that these reported incidents are amenable to determination of "cause-effect"
relationships and at best, may only suggest an association between a given substance
and a certain effect or outcome.
Another aspect of TESS data is that in the vast majority of
reported incidents, information is simply taken at face value from individuals
who may not be knowledgeable of all the exposure circumstances, clinical effects
and treatments that have been rendered. Often, patients may relay what was explained
to them by medical practitioners regarding diagnosis and treatments rendered.
The specialist must then translate this information into corresponding TESS data
fields in order to document the incident. There is no attempt or ability to "authenticate"
these incidents in terms of their validity or the integrity of the information
that is collected. Furthermore, in those few cases where there is some attempt
to do so, it is impossible to segregate those incidents from aggregately reported
data contained in the overall database. It is important to acknowledge the strengths
and limitations of this and other databases representing spontaneously reported
information. This is especially true when drawing safety conclusions from aggregated
data represented in the database.
Problems with the Recent Report on Ephedra Adverse Event Reports
In the Bent et al. article1 which compared the number
of reported incidents involving ephedra-containing products to other herbs, a
number of concerns have been raised over the scientific methodology used and unsupported
conclusions regarding safety being drawn.20
The article was written by Stephen Bent, MD (University of
California San Francisco), Thomas N. Tiedt, PhD (Med-Tox Group, Longboat Key,
FL), and Michael D. Shlipak, MD, MPH (San Francisco Medical Affairs Medical Center,
San Francisco, CA). First, the authors in this report imply that the reported
incidents coded as "adverse reaction" and contained in TESS are adjudicated
as attributable to a given substance. As mentioned earlier, spontaneously reported
incidents contained in the TESS database are not investigated or validated in
such a way so as to allow a cause and effect determination. Many of these reports
are made anonymously, and all are taken at face value without the benefit of independent
confirmation by a licensed healthcare practitioner.
Another significant limitation in reviewing these data, is
that calls classified as adverse reaction are infrequently reported to poison
centers as compared to incidents involving single, acute exposures to potentially
toxic substances. Calls classified as representing "adverse reaction"
represent only 2.2% of the 1,931,841 calls received by poison centers nationwide.19
[And, with respect to all botanical calls, cases classified as "adverse
reaction" represent only 2,115 incidents or 0.1% of all cases reported in
the 2001 TESS annual report.] This suggests that calling poison centers is either
an infrequent option for consumers and healthcare providers for the reporting
of adverse reactions involving botanicals, or that the relative frequency of these
incidents is extremely small. This may further suggest that the "sensitivity"
of this database may not be sufficient to identify any significant trends one
way or the other.
An additional and substantial flaw in the methodology used
by Bent et al. pertains to the assumption of outcome and adverse clinical effects
regarding all incidents listed in the "adverse reaction" column for
ephedra-containing substances. It was assumed that all of the 1,178 incidents
represented cases in which adverse effects were either reported or believed to
be associated with these botanical substances. The authors were apparently unaware
that of all incidents involving ephedra-containing products, approximately
58% were coded in a number of outcome categories that either suggested there were
no documented adverse effects or that the effects were believed to be unrelated
to the substance in question. These outcome categories are not listed in the
table that was relied upon by the authors and not readily apparent from the tables
contained in the TESS annual report.* Furthermore, it is assumed that for those
cases where adverse clinical effects are reported, all of these effects are deemed
to be of equal concern and severity. Again, there is no way of determining how
the individual cases were coded according to the nine available options.
It is also curious as to why Bent et al. chose to compare the
effects of ephedra-containing substances to the wide variety of other botanicals,
the effects of which are not only diverse, but often times subtle and noted over
an extensive period of time. Ephedra, on the other hand, demonstrates clinical
effects that may be readily detected and quite noticeable to the user within a
relatively short period of time. These expected effects, even when minor in nature,
may be interpreted as unpleasant to one user as opposed to another. If the authors
were attempting to compare the relative incidence of reported adverse effects
of these substances, a more meaningful comparison might have been to compare against
other agents that possess either similar pharmacology or indications.
Lastly, the authors used inaccurate sales data to suggest that
ephedra-containing dietary supplements are disproportionately responsible for
many botanical adverse event reports (AERs): they erroneously stated that ephedra
sales constitute less than 1% of total supplement product sales in the botanical
market. More accurate sales data suggests that ephedra-containing dietary supplements
comprise an estimated 20-30% of the botanical supplement sales.** Considering
its shortcomings, for the Bent et al. paper to have received such considerable
attention with so little knowledgeable peer review and scrutiny, objective scientists
might reasonably question whether or not ephedra and other dietary supplement
botanicals may be the subject of a double standard (or perhaps inexact scientific
scrutiny) as compared to the evaluation of the safety of pharmaceutical drugs.
As an example of the importance given to this paper, the erroneous conclusions
from the Bent et al. article were cited by the U.S. Food and Drug Administration
in the Federal Register in February when the Agency made reference to it
as part of its rationale for proposing strong warnings for the labels of ephedra
supplements.21 [Note: The authors of the present article are
not attempting to diminish the safety concerns concerning ephedra by making these
comments. However, as demonstrated in this article, there appears to be a problem
of potentially significant proportions in attempting to rationally interpret and
draw meaningful conclusions surrounding AER data on herbal products from the existing
reporting mechanisms.]
In defense of their article, upon publication of letters raising
questions about their methodology by the senior author of this article and his
colleague Stephen W. Borron MD, MS (International Toxicology Consultants, Washington,
D.C.),20 Bent et al. respond that, "The data from the national
system of Poison Control Centers provide a critical safety valve for consumers.
Health researchers, government agencies, and product manufacturers rely on these
data to warn the public about product dangers and to design safer products."23
There is no question that such groups do utilize the TESS database. As noted in
the present article, misinterpretations can, and do, occur even when well-intentioned
or authoritative entities analyze information contained in the TESS annual report.
Bent et al. also responded with another calculation without grounds. They respond
to the point that an average of 15.4% of all AERs in the database (herb and non-herb)
are later determined to be unrelated to the substance in question by using this
figure to once again compare ephedra to other "herbs." The salient point
that was brought to their attention was that it is impossible to calculate this
relative incidence, because the numbers are not present in the data that Bent
et al. had available for evaluation. The percentage of AERs implicating ephedra,
and then subsequently determined to be unrelated, could theoretically range from
0—100%, rendering their calculation meaningless.
With respect to a letter from Annette Dickinson, PhD, president
of the Council for Responsible Nutrition, that questions the market statistics
used by the authors to conclude that ephedra supplements constitute only 1% of
the total herbal supplement market but in actuality constitute approximately 33%
of the market in 2001,24 Bent et al., respond, "We are perplexed
at Dr. Dickinson’s statistic of ephedra sales from the Nutrition Business
Journal. We did not rely on this journal’s data because the journal uses
an unspecified combination of sales data and information from surveys of herbal
manufacturers and also uses dollar sales rather than unit sales
(the relatively expensive cost of ephedra products biases the dollar sales estimates)."23
Despite this response, it is the opinion of the authors of this article that the
data suggesting that ephedra supplements are only 1% of the total herbal market
upon which Bent et al. relied is misleading in relation to the actual volume that
ephedra previously enjoyed in the U.S. market.
Monitoring of adverse effects potentially associated with herbs
and other natural medicinals will continue to be a challenge. While the use of
multiple methods to track and report incidents is important, the strengths and
limitations of all available reporting options must be considered. A significant
strength of monitoring spontaneously reported data is that it is well-suited to
generate hypotheses on safety or toxicity. It is also extremely useful in defining
a particular product’s safety profile, especially when a given method of
adverse event reporting is well positioned to be utilized by those who may experience
an adverse effect and report it to a given reporting system.
As the future of herbs and other dietary supplements unfolds,
there must also be a concerted effort on the part of healthcare providers and
other professionals, regulators, manufacturers, and the public at large to participate
in the process of dietary supplement post-market surveillance. Collection of spontaneously
reported botanical AERs in a systematic and consistent manner is a vital part
of product stewardship and safety assurance. Defining and monitoring the safety
of dietary supplements is a dynamic process, and cooperation among multiple stakeholders
using reliable methods of surveillance and analyzing collected data in proper
context will aid in this process.
Richard (Rick) Kingston, PharmD, is Co-Founder, Vice President
and Senior Clinical Toxicologist with the PROSAR International Poison Center and
an Associate Professor in the Department of Experimental and Clinical Pharmacology,
University of Minnesota, College of Pharmacy. He has over 26 years experience
in poison control and clinical toxicology, previously serving as co-founder and
former Director of the Minnesota Poison Control System. At the University of Minnesota,
Dr. Kingston serves as Course Director for "Therapeutics of Herbal and other
Natural Medicinals," a course dedicated to the clinical application of science
related to dietary supplement use. The PROSAR International Poison Center
is the largest product safety focused poison center in the United States, serving
over 200 corporate, governmental, and NGOs in the areas of health and safety call
center support, post-market surveillance, incident investigation, and health information
management.
Mark Blumenthal is founder and executive director of the American
Botanical Council and editor and publisher of HerbalGram.
* It is not possible to determine the actual statistical
significance of ephedra-containing botanical supplements in the TESS database
from the information used by Bent et al. from the tables in the annual report.
Although there is a breakdown for how all calls are distributed by the
five outcomes listed in the table, there is not a breakdown for each type of call
classification. Thus, it is unknown how calls classified as involving an "adverse
reaction" question are ultimately coded in terms of their medical outcome.
There are likely "adverse reaction" calls coded with outcomes indicating
that either "no effects" or "unrelated effects" occurred but
the number of cases coded in this manner is simply unknown. Unless a researcher
purchases that data from AAPCC and individually separates them out, one cannot
draw any meaningful and statistically documentable conclusions. The primary author
of this paper (RK) checked the numbers that were reported in the Bent et al. paper
and they match those from the annual TESS report tables; the authors stated in
their methodology that they simply relied on the annual report, not a purchased
subset containing the relevant details. The significance of this limited research
and its erroneous conclusions can be seen in the following analysis of the data:
If one were to view the table in the 2001 TESS report (page 439) which reports
7,115 exposures to "Multi-botanicals with ma huang" (i.e., ephedra),
only 42% of the 7,115 exposures were coded with outcomes representing clinical
effects (minor, moderate, major, and death) and these 7,115 exposures include
all reasons such as intentional overdose, intentional abuse/misuse, unintentional
misuse, and adverse reaction. One must ask how the other 58% of the cases were
coded for outcome. There are no data presented in the published annual report
to answer this. For "Ma huang/ephedra (single ingredient)" again, only
43% had outcomes reportedly associated with clinical effects. Again, it is not
readily apparent how the other 57% were coded for outcome.
** From public comments made by the American Botanical
Council to FDA in response to FDA’s proposed label warning on ephedra: "Bent
et al. conclude that there is a high (78-fold) ratio of ephedra-related AERs (a
reported 64% of all herbal AERs) in relation to ephedra market share of 0.82%.
The calculation of a relative risk factor of 78 times for ephedra compared to
other herbs listed in the report is erroneous, based on inaccurate market statistics.
That is, the denominator of 0.82 is not a reasonable statistic and has no realistic
basis in market reporting. ABC recognizes that it is difficult to adequately and
accurately measure the total retail sales levels of herb products sold as dietary
supplements because solid econometric data are not available for all market channels
(i.e., health and natural food stores, multi-level marketing companies, mail order
and internet sales, sales via healthcare professionals, etc.). Accordingly, only
the mainstream market (grocery stores, drugstores, mass market retailers) produces
reliable data on herb sales. Current estimates of herb supplement sales in 2001
range from $1.3 to 1.7 billion, according to Nutrition Business Journal and
ABC. Compared to a relatively conservative estimate of about $4 billion in total
retail sales, depending on how the herb market is defined and measured, the current
percentage of ephedra supplements should be in the range of 32.5 to 42.5%. If
the total herb market is more liberally estimated at $5 billion, then the ephedra
percentages drop to a range of 26 to 34%. Thus, given a more accurate and comprehensive
view of the total market, and based on the method of calculation employed by Bent
et al. of the TESS data, the range of the ratios of ephedra AERs to other herbs
may more likely be in the range of 2.5-fold (64% divided by 26%) to 1.5-fold (64%
divided by 42.5%) of relative risk, assuming, for the present purposes, the accuracy
of the 64% numerator (which as noted above, is probably not reliable).
Consequently, we believe that the study in question does not appear to provide
valid or reliable conclusions upon which the agency can contemplate current or
future regulatory policy on ephedra."23
References:
1. Bent S, Tiedt TN, Odden MC, Shlipak MG. The relative
safety of ephedra compared with other herbal products. Ann Intern Med 2003
Mar 18;138(6):468-71.
2. Litovitz T, Veltri J. 1984 Annual Report of the American
Association of Poison Control Centers National Data Collection System. Am J
Emerg Med 1985;3:423-50.
3. Litovitz T, Normann S, Veltri J. 1985 Annual Report of
the American Association of Poison Control Centers National Data Collection System.
Am J Emerg Med 1986;4:427-58.
4. Litovitz T, Martin T, Schmitz B. 1986 Annual Report of
the American Association of Poison Control Centers National Data Collection System.
Am J Emerg Med 1987;5:405-45.
5. Litovitz T, Schmitz B, Matyunas N, et al. 1987 Annual
Report of the American Association of Poison Control Centers National Data Collection
System. Am J Emerg Med 1988;6:479-515.
6. Litovitz T, Schmitz B, Holm K. 1988 Annual Report of
the American Association of Poison Control Centers National Data Collection System.
Am M Emerg Med 1989;7:495-545.
7. Litovitz T, Schmitz B, Bailey K. 1989 Annual Report of
the American Association of Poison Control Centers National Data Collection System.
Am J Emerg Med 1990;8:394-442.
8. Litovitz T, Bailey K, Schmitz B, et al. 1990 Annual Report
of the American Association of Poison Control Centers National Data Collection
System. Am J Emerg Med 1991;9:461-509.
9. Litovitz T, Holm K, Bailey K, et al. 1991 Annual Report
of the American Association of Poison Control Centers National Data Collection
System. Am J Emerg Med 1992;10:452-505.
10. Litovitz T, Holm K, Clancy C, et al. 1992 Annual Report
of the American Association of Poison Control Centers National Data Collection
System. 1993 September;11(5):494-555.
11. Litovitz T, Clark L, Soloway R. 1993 Annual Report of
the American Association of Poison Control Centers National Data Collection System.
Am J Emerg Med 1994;12:546-84.
12. Litovitz T, Felberg L, Soloway R, et al. 1994 Annual
Report of the American Association of Poison Control Centers National Data Collection
System. Am J Emerg Med 1995;13:551-97.
13. Litovitz T, Felberg MA, et al. 1995 Annual Report of
the American Association of Poison Control Centers Toxic Exposure Surveillance
System. Am J Emerg Med 1996;14:487-537.
14. Litovitz T, Smilkstein M, et al. 1996 Annual Report
of the American Association of Poison Control Centers Toxic Exposure Surveillance
System. Am J Emerg Med 1997;15:447-500.
15. Litovitz T, Klein-Schwartz W, et al. 1997 Annual Report
of the American Association of Poison Control Centers Toxic Exposure Surveillance
System. Am J Emerg Med 1998;16:443-97.
16. Litovitz T, Klein-Schwartz W, et al. 1998 Annual Report
of the American Association of Poison Control Centers Toxic Exposure Surveillance
System. Am J Emerg Med 1999;17:435-87.
17. Litovitz T, Klein-Schwartz W, et al. 1999 Annual Report
of the American Association of Poison Control Centers Toxic Exposure Surveillance
System. Am J Emerg Med 2000;18:517-74.
18. Litovitz T, Klein-Schwartz W, et al. 2000 Annual Report
of the American Association of Poison Control Centers Toxic Exposure Surveillance
System. Am J Emerg Med 2001;19:337-95.
19. Litovitz T, Klein-Schwartz W, et al. 2001 Annual Report
of the American Association of Poison Control Centers Toxic Exposure Surveillance
System. Am J Emerg Med 2002;20:391-452.
20. Kingston RL, Borron SW. The relative safety of ephedra
compared with other herbal products (letter). Ann Intern Med 2003 Sept
3;139(5)part I:385.
21. U.S. Food and Drug Administration. Dietary supplements
containing ephedrine alkaloids; Reopening of the comment period. Federal Register
2003 March 5;68(43):100417-20.
22. Blumenthal M. Letter to Food and Drug Administration
in response to Docket No. 95N-0304: Proposed Rule: Dietary Supplements Containing
Ephedrine Alkaloids; Reopening of Comment Period; 68 Fed. Reg. 10417 (March
5, 2003). Apr 7, 2003.
23. Bent S, Tiedt TN, Shlipak NG. The relative safety of
ephedra compared with other herbal products (letter). Ann Intern Med 2003
Sept 3;139(5)part I:386.
24. Dickinson A. The relative safety of ephedra compared
with other herbal products (letter). Ann Intern Med 2003 Sept 3;139(5)part
I:385-6.
|