Health Professionals

Summary of Incidence of Seizures, Strokes, and Myocardial Infarctions in the Population and Estimations of Risk in the Population from Ephedra Products

I am an Assistant Professor of Medicine in the Cardiovascular Division at the University of Pennsylvania School of Medicine. I am also an Assistant Professor of Epidemiology in the University of Pennsylvania's Department of Biostatistics and Epidemiology and a Senior Scholar in the Center for Clinical Epidemiology and Biostatistics. I am board certified in both cardiovascular medicine and internal medicine. I received my undergraduate degree from Princeton University, my medical degree from the New York University School of Medicine, and a Master of Science in Clinical Epidemiology from the University of Pennsylvania, with a focus in pharmacoepidemiology. I did my internal medicine training at Harvard's Brigham and Women's Hospital and my cardiology fellowship at the Hospital of the University of Pennsylvania. My research focuses on the cardiac effects of non-cardiac drugs.

OVERVIEW:

The purpose of this report is twofold: (1) to present data on the incidence rate (also referred to as "incidence" and defined below) of seizures, strokes, and myocardial infarctions (MIs) in the population; and (2) to provide estimates of the risk of these events from ephedra products and compare them with population-based estimates of risk of the same events.

For Aim 1, the data on incidence are, when possible, "population incidences," reflecting the occurrence of these events in the general population. These incidence rates therefore include people with risk factors for the events and those without risk factors. This is not a critique of the literature cited. Each study may have inherent strengths and weaknesses. For example, studies of Health Maintenance Organization (HMO) patients may underestimate incidence because of the healthy worker effect (people enrolled in HMOs are employed and therefore able to work and more likely to be healthy than the general population that includes both healthy and ill people). Studies may also underestimate incidence if people who die prior to hospitalization are not identified by the study. Incidence may be overestimated if patients are incorrectly diagnosed with the disease under study. (Of course, misdiagnosis can occur in clinical practice done outside of a strict research protocol, and therefore studies relying on clinical judgment may better reflect adverse event reporting.)

No attempt was made to appraise these, and other, strengths and weaknesses. When there were specific aspects of the study that made them less "population-based," these are noted in the tables.

The second Aim, to estimate the incidence of adverse events among users of ephedra products, provides only rough estimates, given the limitations of currently available data. However, care was taken to use conservative assumptions in order to try to overestimate, rather than underestimate, the incidence of adverse events in users of ephedra products.

This type of analysis cannot prove or disprove an association between ephedra and adverse events. However, by estimating actual event rates, it provides superior evidence compared with simple adverse event reports, which do not provide estimates of the number of events per person.

This report illustrates the need to understand the background risk, described in the first section, in evaluating any adverse effect of any exposure in the population. As the FDA's Center for Drug Evaluation and Research pointed out in its February 10, 2000 memorandum, "it is possible that the reported serious adverse events are reflective of coincidental background spontaneous occurrences in the population and are not necessarily causally related to [the use of dietary supplements containing ephedrine alkaloids]."

The limitations of adverse event reporting preclude determination of an association between an exposure and an outcome. This analysis does not attempt to overcome these limitations, but rather it provides a range of possibilities to put the adverse event reporting into perspective. However, by using conservative assumptions to attempt to overestimate risk from ephedra products and by using a range of assumptions, the analysis in this report provides a broad range of possible results that allow one to assess the possibility that an association exists between ephedra products and the risk of seizures, strokes or myocardial infarctions.

Back to top

METHODS:

Aim 1 ‚ Incidence of Events in the Population

Inclusion Criteria

A search of the 1966-2000 Medline database was performed for all English-language articles using the following search terms:

For seizures: exp seizures/ep

For strokes: exp cerebrovascular accident/ep

For MI: exp myocardial infarction/ep AND incidence.tw.;

exp myocardial infarction/ep AND age factors AND female

Articles published from 1985 to the present were eligible. Studies were included only if they presented data from 1980 or later, in order to reflect more recent incidence rates. When separate incidence rates for multiple years were given in a study, the most recent years' data were used. Only studies presenting data on first events (i.e., those able to separate out a first-time event from a recurrence) were included.

No study was included or excluded on the basis of the incidence rates that it reported.

All populations were eligible. Although there may be variability in rates by country, these differences may be, at least in part, due to genetic factors. Therefore, people living in the U.S. whose families originate from other countries may have similar rates of events to people living in these other countries (albeit environmental and dietary differences among countries may diminish these similarities).

Exclusion Criteria

Studies were excluded if they:

1. presented data only from years prior to 1980
2. did not separate first events from repeat events
3. focussed solely on identifying risk factors for events and did not provide population-based incidence
4. studied only patients over 65 years-old
5. studied patients under and over 65 years-old but did not provide separate incidence rates for those under 65
6. were randomized trials of new therapies
7. did not or could not report or calculate incidence (e.g., case-control studies).

Specific Conditions

Seizures

Infantile seizures were excluded when possible. For studies presenting data on infants, the youngest age range is presented separately. This lowest age range would be the one that includes infantile seizures. No attempt was made to distinguish seizure type (e.g., simple, complex, etc.). Studies of only epilepsy (recurrent seizures) were not included.

Strokes

No attempt was made to distinguish stroke type (e.g., hemorrhagic, thrombotic) unless data were presented only in this manner. Many studies did not ascertain fatal strokes leading to death prior to hospitalization, so many may underestimate the true rate of strokes.

Myocardial Infarction

It is often difficult to distinguish a definite MI from a possible MI. There is extensive literature examining this issue. The definition used for each study is the one presented here. If studies presented rates that included out-of-hospital cardiac deaths, these are the rates presented because they represent the possible rate of all MI-events, including the approximately 25% that result in sudden death.^1-4

Data Presentation

All data are presented as incidence rates. An incidence rate is the number of events divided by the amount of person-time at risk to have the event. For example, if 1000 people used ephedra for one year each, this would constitute 1000 person-years. If there was one event among them, the rate of events would be one per 1000 person-years. If 1000 people used ephedra for 2 years, this would constitute 2000 person-years, and one event would be equal to 1 event per 2000 person-years (or approximately 50 events per 100,000 person-years).

Each incidence is presented as the number of events per 100,000 people per year. Overall (i.e., not age-specific or gender-specific) results are usually age-standardized and often standardized to gender. When available, the 95% confidence interval is presented for each incidence. When data were presented graphically without exact numbers, an estimate of incidence rate was obtained from the graph and this was noted in the tables.

Aim 2 ‚ Estimates of Incidence of Events Among Ephedra Users

Estimate of Incidence

The incidence rate of adverse events among ephedra users is defined as: the number of adverse events occurring among all users divided by the number of person-years of ephedra use (and presented as events per 100,000 person-years, the same way they are presented for the background population rates). Because the true number of people exposed to ephedra and the true number of adverse events among users is not known, only a range of estimates can be provided by estimating each of these parameters. The a priori goal was to provide conservative estimates (i.e., ones that, if anything, might overestimate the rate of events among ephedra users). Incidence rate estimates are derived separately for each of the three outcomes: seizure, stroke, and MI.

Estimate of Number of Adverse Events

Source of Data and Assumptions

The number of adverse events is estimated by dividing the number of events reported to the FDA adverse event reporting system by the estimated proportion of all events that are reported to the FDA. Both of these (the number of events and the reporting proportion) will be discussed in turn.

Number of Events

The number of adverse events reported is derived from the report by the Center for Food Safety and Applied Nutrition (CFSAN) of the FDA of adverse events from June 1, 1997 through March 31, 1999. Adverse events from these dates have been chosen for several reasons: (1) Most important, these events have been carefully organized, reviewed, and enumerated by FDA. Therefore, the data provide the number of each type of adverse event (seizure, stroke, MI) occurring during this time period as determined by FDA. (2) Because these are recent data, they are likely to reflect current usage of ephedra alkaloid-containing products. (3) The reports follow the FDA's proposed regulation on ephedrine alkaloid-containing dietary supplements, and therefore represent a time period during which public awareness might have been heightened and reporting increased. (4) The reports are most likely to include events that have actually occurred during this time period (as opposed to earlier reports which might have included events occurring in years prior to the actual year reported). In fact, according to CFSAN, 8.3% of events in the June 1997 to March 1999 reports actually occurred in 1996 or earlier. These events are included in the estimates for this analysis because the specific cases are not identified in the CFSAN report and because it is possible that a similar proportion of events occurring during the June 1997 to March 1999 time period are not included in the June 1997 to March 1999 report. (5) The reports occur in a time period during which estimates of ephedra consumption can be made. Of course, despite these justifications, the selection of this time period is somewhat arbitrary, and different time periods may produce different estimates (as discussed below).

It is important to note that the assumption for this analysis is that all reported adverse events occurred during exposure to ephedra products at recommended serving sizes. This includes the "attributable," "supporting," and "insufficient data" groups. This provides for an "upper-limit" of reported adverse events for two reasons. First, it is possible that some of the acute events occurred without appropriate temporal exposure to ephedra. Second, as stated in the FDA report, the "insufficient data" group could have included "intentional misuse or abuse of the product." The purpose of this analysis is to estimate risk with labeled use. Including events occurring from misuse will artificially elevate this estimated risk. (As an extreme example, if all people with adverse events misused the product, there would, in fact, be no adverse events associated with proper use.)

It is also important to emphasize that no attempt has been made to assess causality in this analysis. In fact, this is not necessary because the risk of adverse events among ephedra-exposed subjects is being compared with the risk in the general population in order to estimate whether the events that occurred in ephedra users could represent background events that would have happened irrespective of ephedra use. If, in fact, there is no causal relationship between ephedra use and the events studied, the events occurring among the ephedra-exposed should equal the number expected in the general, non-ephedra exposed population. Eliminating people with events felt to be causally unrelated to ephedra would falsely reduce the estimate of risk that this analysis is attempting to measure.

The number of events reported to FDA and the distribution by age and gender are shown in Table 1 below. In order to be all inclusive, the MI category includes any unstable angina and all reported MIs, plus any sudden deaths in which an MI could not be excluded as a possible cause. The stroke category includes any stroke, and also includes central nervous system hemorrhagic events (including subarachnoid hemorrhage) and transient ischemic attacks (TIAs).

Table 1 ‚ Reported Events Used in Analysis June 1997 ‚ March 1999
	MI	Stroke	Seizure
Number of events	10	14	7
Mean Age	35.9	42.9	36.0
Age Range	15-59	18-64	21-51
Gender (% men)	50%	42.9%	57.1%

 

All of these inclusions will, if anything, tend to overestimate the risk of events (relative to population statistics which tend to exclude these diagnoses).

The MI category includes the following adverse event reports: 12851, 12485, 12452, 13009, 13463, 13096, 12721, 13127, 13021, and 12843. The stroke category includes: 12483, 13062, 12713, 12733, 12861, 12980, 12888, 13336, 13418, 12460, 12487, 12871, 12921, and 13380. The seizures include: 12948, 12975, 13001, 13405, 13344, 11919, and 12927.

Reporting Rate

The true number of events among persons using ephedra alkaloid-containing products is not known. The number of reported events is likely to represent a fraction of the number of events occurring among persons using ephedra products because of under-reporting of events. The value of this fraction is unknown and numerous factors may increase or decrease this value.

Factors that might increase reporting include the severity of events, the acuity of events, the rarity of the event in the exposed population, close temporal relationship between exposure and the event, and publicity surrounding the exposure.⁵ The events evaluated in this report (seizure, stroke, and MI) are all acute, dramatic, usually severe, and often life-threatening. In addition, it is expected that ephedra users are relatively young, and therefore, these events are rare in this population (as demonstrated in the first aim of this report). Further, other traditional causes of these events (especially older age) are less likely to be present and therefore to be attributed as the cause of the events. Also, these are not events that have long latency periods; MIs, strokes, and seizures are acute events (albeit often occurring on a pre-existing substrate such as atherosclerotic or congenital cardiovascular disease). These events are not long-term, delayed consequences of exposure that would be less likely to be reported. Finally, there has been substantial publicity of adverse events and ephedra alkaloid-containing products during the period studied. All of these factors might increase reporting of adverse events.

On the other hand, several factors may reduce reporting of adverse events. Reporting for dietary supplements may be less than for over-the-counter (OTC) or prescription medications for several reasons. Although, to my knowledge, there have not been direct comparisons between actual reporting rates for dietary supplements and traditional medications, both dietary supplement users and physicians may be less likely to associate and report adverse events with dietary supplements. People also may be less likely to tell their "traditional medicine" doctors about consumption of dietary supplements, and doctors may be less likely to ask patients about dietary supplement use.⁶ In a recent British study, 51% of patients using herbal products stated that they would probably consult a healthcare provider or pharmacist if they had a serious adverse event after taking the product compared with 67% for OTC medications.⁷ In addition, an important source of adverse event reporting, pharmaceutical companies, is not available to report adverse events from dietary supplements.

Several studies have been performed to try to estimate adverse event reporting (although none, to my knowledge, have directly examined dietary supplements). In one British study, 2.4% of MIs following numerous medication exposures were reported.⁸ From this same study, cardiac arrest was reported at a rate of 18%.⁹ In a study of seizures following vaccines (a time of heightened awareness of the possibility of an adverse event and using a system that promoted reporting), the reporting rate was 24-42%.¹⁰ Two studies performed by the same group of researchers tried to estimate reporting rates using an indirect method of calculation, but did not actually measure reporting rates. One study estimated reporting of serious events in 1 in 4,610 events (95% confidence interval from 2,514 to 8,454) while the other estimated 1 in 605 (95% confidence interval from 0 to 1,444). Interpretation of these two studies' findings is difficult because of the severely limited study design, because the same investigators obtained such disparate results using the same methodology, and because of imprecision (one of the confidence intervals actually included 0, meaning no under-reporting).^11,12 Although not a direct study of reporting rates, a report from FDA estimated that 25-50% of cases of eosinophilia-myalgia syndrome from dietary supplements containing L-Tryptophan were reported following a media campaign (in some ways, not dissimilar to the ephedra experience).¹³

FDA estimated that 10% of events were reported for the 1997 proposed rule on dietary supplements containing ephedra alkaloids.¹⁴ Whether this applies to the serious adverse events discussed here or whether rates are higher or lower (for the reasons described above) is unknown. Therefore, a range will be presented, from 1% to 20%. This provides a reasonable range of possible results.

Estimate of Person-Time Exposure to Ephedra

Source of Data

Similar to many pharmaceutical products, there is no direct measurement of the number of individuals exposed to ephedra. Therefore, in order to estimate exposure, data are used from a recent survey of ephedra sales conducted by Arthur Andersen for the American Herbal Products Association (AHPA) (hereinafter AHPA Survey). This survey is attached as Exhibit 1. Briefly, Arthur Andersen mailed surveys to 42 companies that distribute dietary supplements containing ephedra alkaloids and whose products were listed in FDA's proposed rule. Of the 42 companies solicited, 13 (31%) reported on annual sales of products containing ephedra alkaloids for each year of the period of 1995 through 1999. All reports were anonymous. The number of servings (defined as the maximum amount to be consumed each use) sold for the years of interest is depicted in Table 2 below:

Table 2 ‚ Number of Servings Sold Based on Reports of 13 Companies in Survey by Arthur Andersen
Year	Number of Servings* Sold
1997	976,466,984
1998	1,751,381,254
1999	3,086,041,072
* The maximum amount to be consumed each use

 

Estimate of Total Sales

The total number of servings for the years included in the adverse event reports (June 1997-March 1999) is estimated to be 3,011,125,014 [(976,466,984/2 = one-half of 1997) + (1,751,381,254) + (3,086,041,072/4 = one-quarter of 1999)]. Because sales may not be distributed evenly across a year, this is just an estimate: sales in second half 1997 may represent more than half of all annual sales in 1997, and sales in the first quarter of 1999 may represent less than one quarter of all sales in 1999.

Only 13 out of 42 companies (31%) provided data on ephedra supplement sales, and not all companies who make supplements containing ephedra alkaloids were included in the 42 surveyed. Therefore, the numbers presented in Table 2 are clearly an underestimate of total sales of ephedra alkaloid-containing dietary supplements. However, to be as conservative as possible, the primary calculations assume that these data represent the total amount of ephedra sales. This will underestimate the number of users and overestimate the risk of events among users. As a secondary estimate of use (see below), it will be assumed, arbitrarily, that the 13 companies reporting represent 75% of all ephedra-containing supplement sales. This is simply to estimate how sensitive the results are to the assumption of sales.

Estimate of Person-Years Exposure

In order to estimate the amount of person-years of ephedra use, two other pieces of data from the AHPA survey were used.

First, the maximum amount, in milligrams (mg) of ephedrine alkaloids, recommended for each serving was ascertained in the AHPA survey (Table 3).

 

Table 3 ‚ Maximum Amount (mg) of Ephedrine Alkaloids Recommended Per Serving (From AHPA Survey)
Recommended Serving Size	Aggregate Number of Products
Less than 10 mg	28
10 - 15 mg	22
16 - 25 mg	33
26 mg or more	1
TOTAL:	84 Products

 

From these results, an estimate of the average serving size was calculated as follows: Table 4 displays the average serving size per category from Table 3 and the proportion of products that are within that serving size. The > 25 mg group was assumed

Table 4 ‚ Average Serving Size and Proportion of Products Within That Serving Size
Recommended Serving Size	Average Serving Size	Proportion of Products (%)
< 10 mg	5 mg	28/84 (33.3%)
10 - 15 mg	12.5 mg	22/84 (26.2%)
16 - 25 mg	20.5 mg	33/84 (39.3%)
> 25 mg	30 mg	1/84 (1.2%)

 

to be an average of 30 mg. Although this is arbitrary, an unstructured review of dietary supplements containing ephedra alkaloids revealed no products that list > 30 mg as the maximum serving size. Using the data above, the estimated average serving size is:

5*(33.3%) + 12.5*(26.2%) + 20.5*(39.3%) + 30*(1.2%) = 13.35 mg

It is possible that the average serving size consumed is greater than 13.35 mg, especially since the most common serving size is 16-25 mg. If this were so, the estimate of person-years exposure in these calculations will be falsely low (as discussed below), and the risk among ephedra users will be falsely high. It is also possible that consumers use less than 13.35 mg per serving. As stated previously, the variability in use cannot be determined from this study, and the use of an average serving size seems the most reasonable.

Second, the average daily maximum use was calculated from the following data from the AHPA survey.

 

Table 5 ‚ Maximum Amount of Ephedrine Alkaloids Recommended to be Consumed in a 24-hour Period (from AHPA Survey)
Maximum Dosage in a 24-Hour Period	Aggregate Number of Products
25 ‚ 50 mg	63
50 ‚ 100 mg	60
More than 100 mg	0
TOTAL:	123 Products

 

Using the average dosage from each category, the average 24-hour maximum dosage is therefore:

37.5*(63/123) + 75.5*(60/123) + >100 mg%*(0%)= 56.0 mg

From the above two estimates (average serving size of 13.35 mg and average maximum daily dose of 56.0 mg), the average servings/day can be calculated:

Average Max Dose (56 mg) ¸ Average servings/day (13.35 mg) = 4.2 servings/day

This estimate is likely to be conservative (i.e., will underestimate number of person-years) because most users probably consume fewer than four servings per day. Although the reports from the AHPA survey were anonymous and identification of companies and products is not possible, a review of many of the web sites listed in the FDA market analysis of ephedra-containing products revealed that many products are marketed for less than four servings per day, particularly those marketed for energy or performance enhancement. Similarly, people who use products for episodic effects, such as products promoted for energy (which often state to take prior to a workout), performance enhancement, cough or colds are unlikely to consume four servings per day. Importantly, although some products may be marketed with instructions for use up to more than four tablets per day (e.g., "take 1 or 2 tablets three times a day"), none were identified that recommended use more than four servings a day (albeit all products were not surveyed). Because the definition of "servings" in the survey was the "maximum amount to be consumed each use," even products suggesting multiple tablets per use would still not be used more than four times a day. Even if this was misunderstood by the survey respondents as "servings" rather than "episodes of use," a review of many of the web sites listed by FDA in their market review suggests that most products are labeled for no more than four "servings" per day. In the AHPA survey itself, 48% of products were marketed for energy or athletic performance. Therefore, as discussed above, many people were likely to use the product less than four times a day.

In addition, the estimate of use is likely to be low because the average serving is probably greater than 13.35 mg (as discussed above). If so, the calculation of 4.2 servings/day, above, is an overestimate (because the denominator should be greater than 13.35), and the number of person years (total sales divided by servings/day) will be an underestimate.

Using the conservative estimate of 4.2 servings/day and the total number of servings sold in the time period of interest of 3,011,125,014, the estimated person-days of exposure is 717,426,006 (3,011,125,014 servings ¸ 4.2 servings/day) or 1,965,551 person-years of ephedra use. This is the primary estimate used for the calculations in this analysis. If only 75% of all ephedra products were represented by the 13 companies responding (who represented only 31% of all companies surveyed), the estimate of person-years of ephedra use would be 2,620,734 person-years. This number was used as a secondary estimate of person-years of exposure. As mentioned previously, the purpose of this secondary analysis is simply to determine how sensitive the assumptions are to the estimate of person-time of exposure. If the average use was less than 4 servings/day, the above estimates of person-years exposure will be an underestimate.

In fact, some recent data suggest that usage is greater than the estimated 2 million person-years. In a recent survey of patients presenting to anesthesiologists in the preoperative period, investigators in Texas found that over 11% of patients ages 18-80 years old were taking products containing ma huang. See Exhibit 2. Because this was a selected patient population, it is difficult to generalize these findings to the rest of the population. Whether the need for surgery reflected a population less likely to use supplements (e.g., the use of ephedra for weight training may be less in this population) or more likely to use them (e.g., if obesity-related disease was the reason for some surgeries, this could lead to more use) is unknown. Also, because underreporting of use of dietary supplements may occur,⁶ this estimate may be low. With these caveats: even assuming that this estimate applied only to adults ages 20 to 64 in the U.S. in 1997 (approximately 156 million [Population Estimates Program, Population Division, U.S. Census Bureau, Washington, D.C. 20233]), there would be an estimated 17 million ma huang (ephedra) users. Even if the duration of use were only three months, this would represent over five million person-years of use.

Estimate of Baseline Risk in the Population

The estimates of risk among ephedra users (estimated number of users ¸ estimated number of person-years) was then compared with the population background risk. The estimated risk of seizures, strokes, and MIs in the general population who are the same age and gender as the cases among ephedra users cannot be known exactly. Therefore, a range of probable risk is presented, based on the results of Aim 1 of this report. The risk in the U.S., under-65 population derived from Aim 1 was used. The upper age of 65 is used because no events were reported in individuals over 65. When possible, the lower age limit is 15, corresponding to the youngest reported case in this series. Because different results do not produce exactly the same estimates, a range is given. Because all U.S. data derive from prior to 1995 (and most derive from the 1980s), it is extremely unlikely that these rates are influenced by users of ephedra alkaloid-containing dietary supplements in the population. That is, the rates are likely to be truly representative of the background risk of events in the absence of ephedra-containing supplement use.

Presentation of Results

For each outcome (seizure, stroke, MI), the estimated rate of events per 100,000 person-years of ephedra use is presented graphically over a range of reporting rates of adverse events. The primary exposure time is 1,965,551 person-years, and the secondary time is 2,620,734 person-years. The range of background risk in the population (per 100,000 person-years) is displayed as a gray box over the graph to show the range of possible risk of events in the general population.

Back to top

RESULTS:

Aim 1 ‚ Incidence of Events in the Population

A total of 383 articles were identified for seizures. Review of the abstracts of these articles revealed that 11 were studies of population-based incidence. All 11 were reviewed and 4 met criteria to be included.

Of the 374 papers on stroke, 44 were selected on the basis of review of the abstracts and 40 were found. (Four were not available at our University library.) An additional 8 were identified on the basis of a review paper, and 7 of these were available in our library. Of the 47 papers reviewed, 24 met criteria for this review.

For MI, 654 papers were identified and, of these, 51 abstracts appeared to meet criteria for inclusion. Of these, 45 were reviewed and 6 were not available. Of the 45 reviewed, 11 met criteria to be included.

Seizures

Table 6 presents the incidence rates of seizures. The range of incidence rates are from about 20 per 100,000 to 76 per 100,000. Excluding infants and adults over 65 years old, the rates are approximately 11 to 60 per 100,000. In the U.S., 15-65 age group, the incidence of events in these studies ranges from about 20 per 100,000 to about 60 per 100,000.

Strokes

Table 7 presents the incidence rates of strokes. The incidence of stroke increases with age. However, strokes do occur in the younger population. In the less than 65-year-old population, the stroke incidence ranges from 0 to 363 per 100,000, depending on the age group. In the U.S., under-65 adult population, the incidence of strokes in these studies ranges from about 3 per 100,000 to about 363 per 100,000, again depending on the age group.

Myocardial Infarctions

Table 8 presents the incidence rates of MIs. The incidence of MI increases with age, but MIs do occur in the younger population, both men and women. In the less than 65-year-old population, the MI incidence ranges from 0 to 1939 per 100,000, depending on the specific age group included. In the U.S., under-65 adult population, the incidence of MIs in these studies ranges from about 0.2 per 100,000 to about 337 per 100,000, again depending on the specific age group.

 

Table 6 ‚ Incidence Estimates of Seizures*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
Swedish15	1992-1994	17+	76 (59-93)	UnprovokedÜ
		Men
		17-29	29
		30-39	11
		40-49	44
		50-59	47
		60-69	108
		Women
		17-29	40
		30-39	36
		40-49	38
		50-59	47
		60-69	36
U.S.16	1935-1984	All ages	61	UnprovokedÜ
		0-20	Range ~50-80	Idiopathicá; estimated from graph
		21-40	Range ~25-35
		41-60	Range ~20-35
		60-80	Range ~40-80
Sweden17	1985-87	³ 17	33.6 (25.4-41.8)	Included those secondary to other cause, but 51% had unknown cause
U.S.18	1988-1994	All ages		HMO patients only, ~80% had unknown cause
		0-14	50.9
		15-24	59.6
		25-34	23.3
		35-44	20.0
		45-54	24.8
		55-64	36.5

* CI=confidence interval

Ü No identifiable cause to incite event at the time of event

á No identifiable cause to incite event at the time of event and no prior, remote central nervous system insult to explain the event

 

Table 7 ‚Incidence Estimates of Strokes*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
U.S. Managed Care members19	1991-1994	15-44	10.7 (9.9-12.1)	Women only
		15-19	1.5 (0.6-2.9)
		20-24	4.0 (2.5-6.2)
		25-29	5.8 (3.9-7.9)
		30-34	8.5 (6.4-11.0)
		35-39	14.7 (12.1-17.7)
		40-44	26.6 (23.0-12.1)
Finland20	1966-68; 72	³ 15	300	Includes fatal strokes
		15-44	40
		45-54	120
		55-64	340
		65-74	730
		75+	4,060
Italy21	1983-85	15-44	8.8 (6.5-11.7)
		Men	9.0 (5.8-13.4)
		Women	8.7 (5.5-13.0)
U.S.22	1960-84	All ages	13	Lacunar stroke only (12% of all strokes in this study)
Italy23	1992-93	All ages	224 (204-245)
		0-34	0
		35-44	46
		45-54	114
		55-64	242
		65-74	720
		75-84	1,317
		85+	3,413

 

Table 7 ‚Incidence Estimates of Strokes (cont.)*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
Libya24	1983-84	All ages	63
Australia25	1989-1990	All ages
		Men
		15-24	11 (0-29)
		25-34	5 (0-11)
		35-44	45 (4-86)
		45-54	110 (34-186)
		55-64	351 (215-488)
		Women
		15-24	10 (0-27)
		25-34	17 (0-30)
		35-44	22 (0-52)
		45-54	77 (12-142)
		55-64	98 (24-173)
U.S.26	1945-1984	All ages	135	Estimated from graph
		1980-84
		<45	10
		45-54	100
		55-64	200
		65-74	800
		75-84	1500
		85+	>2000
Japan27	1989-1993	³ 35	212 (192-235)
Sweden28	1983-19885 1993-1995	³ 15 ³ 15	134 (126-143) 158 (149-168)	Rate adjusted to European population
		15-34	2.0 (0.5-5.2)
		35-54	52 (42-63)
		55-64	286 (245-331)
		65-74	701 (629-779)

 

 

Table 7 ‚Incidence Estimates of Strokes (cont.)*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
Australia29	1986, 1989, 1995	All ages		Estimated from graph
		1995
		Men
		35-44	~ 50
		45-55	~ 100
		55-64	~ 200
		65-74	~ 500
		Women
		35-44	~0	Upper limit of confidence interval not provided
		45-55	~ 50
		55-64	~ 100
		65-74	~ 300
Sweden30	1969-1983	Women, all ages		Women only; estimated from graph
		1983
		40-44	~ 50
		45-49	~ 100
		50-54	~ 100
		55-59	~ 300
		60-64	~ 400
		65-69	~ 800
France31	1985-89	All ages	13.9 ‚ men 12.3 ‚ women	Cerebral hemorrhage only
Norway32	1994-96	³ 15	221	Rate adjusted to European population
Japan33	1977-1992	³ 40
		Men	522
		Women	436

 

Table 7 ‚Incidence Estimates of Strokes (cont.)*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
U.S.34	1955-1989	All ages
		1985-1989	145
		Men
		0-34	4
		34-44	36
		45-54	61
		55-64	363
		Women
		0-34	3
		34-44	29
		45-54	64
		55-64	195
Israel35	1992-1993	17-49	10.4
		17-19	1.0
		20-24	2.3
		25-29	2.4
		30-34	3.3
		35-39	7.9
		40-44	18.5
		45-49	47.2
Denmark36	1982-1991	³ 25
		1990-1991
		Men	201 (182-221)
		Women	176 (159-195)
New Zealand37	1981, 1991	³ 15
		1991
		Men
		15-24	5 (1-10)
		25-34	14 (6-22)
		35-44	44 (23-64)
		45-54	85 (56-113)
		55-64	276 (202-349)
		65-74	759 (601-916)

 

Table 7 ‚Incidence Estimates of Strokes (cont.)*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
		Women
		15-24	4 (0-8)
		25-34	7 (1-13)
		35-44	37 (23-52)
		45-54	76 (48-103)
		55-64	205 (142-268)
		65-74	515 (405-626)
Poland38	1991-92	All ages
		Men
		<30	4
		30-44	44
		45-54	88
		55-64	324
		Women
		<30	2
		30-44	19
		45-54	71
		55-64	134
U.S.39	1988	15-44
		Black Men		Cerebral infarction only
		15-39	12.5
		40-44	99.3
		White Men
		15-39	6.0
		40-44	36.5
		Black Women
		15-39	16.6
		40-44	48.6
		White Women
		15-39	6.1
		40-44	38.5
		Black Men		Intracerebral hemorrhage only
		15-39	9.8
		40-44	46.3
Table 7 ‚Incidence Estimates of Strokes (cont.)*
Study Population	Years	Age	Incidence per 100,000 (95% CI)		Notes
		White Men
		15-39	3.6
		40-44	11.0
		Black Women
		15-39	3.9
		40-44	10.8
		White Women
		15-39	1.2
		40-44	3.5
Lithuania40	1986-88	25-64
		Men	230 (208-252)
		Women	131 (117-145)
		Men			Estimated from graph
		25-44	~ 50
		45-54	~ 200
		55-64	~ 500
		Women			Estimated from graph
		25-44	~ 40
		45-54	~ 100
		55-64	~ 250
Italy41	1987-89	15-44	13.6 (9.1-19.6)
		Men	14.0 (7.9-23.2)
		15-24	0 (0-4.2)
		25-34	11.1 (3.0-28.4)
		35-44	31.4 (15.7-56.3)
		Women	13.1 (7.2-22.2)
		15-24	2.9 (0.1-16.0)
		25-34	11.1 (6.3-37.1)
		35-44	19.6 (7.9-40.3)

* CI=confidence interval

 

Table 8 ‚Incidence Estimates of Myocardial Infarction*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
U.S.19	1991-1994	15-44	5.0 (4.4-5.9)	Managed care patients; women only
		15-19	0.6 (0.1-1.6)
		20-24	0.2 (0.0-0.7)
		25-29	1.6 (0.7-3.1)
		30-34	1.5 (0.7-2.8)
		35-39	5.3 (3.8-7.3)
		40-44	18.4 (15.5-21.8)
Estonia42	1990-91	25-64
		Men
		25-34	3
		35-44	28
		45-54	61
		55-64	145
		Women
		25-34	0	No confidence interval given
		35-44	2
		45-54	16
		55-64	54
Sweden30	1969-1983	Women, All ages		Women only; estimated from graph
		1983
		40-44	~ 20
		45-49	~ 50
		50-54	~ 90
		55-59	~ 100
		60-64	~ 500
		65-69	~ 700
France43	1985-1989	30-59	103	Men only
England44	1994-95	£ 74
		Men
		35-64	189 (162-215)
		Women
		35-64	58 (44-73)
Table 8 ‚Incidence Estimates of Myocardial Infarction (cont.)*
Study Population	Years	Age	Incidence per 100,000 (95% CI)		Notes
Sweden45	1985-1994	45-64
		Men	Range ~ 380-590		Depends on socioeconomic status (range shown)
		Women	Range ~ 80-200
Estonia46	1991-94	35-64
		1994
		Men	383 (340-426)
		Women	92 (74-110)
Denmark47	1988-91	£ 74
		Men
		30-34	8 (2-22)
		35-39	34 (20-56)
		40-44	118 (91-150)
		45-49	169 (135-209)
		50-54	277 (229-333)
		55-59	442 (376-516)
		60-64	651 (564-748)
		Women
		30-34	8 (2-22)
		35-39	2 (0-11)
		40-44	24 (13-40)
		45-49	48 (31-70)
		50-54	56 (36-83)
		55-59	128 (94-170)
		60-64	261 (208-322)
U.S.48	1975-1981	³ 25
		1981
		25-44	41
		45-54	234
		55-64	337

 

Table 8 ‚Incidence Estimates of Myocardial Infarction (cont.)*
Study Population	Years	Age	Incidence per 100,000 (95% CI)	Notes
Finland49	1983-85	25-64		Rates vary by geographic region (range shown)
		Men
		25-34	16-35
		35-44	140-245
		45-54	500-1083
		55-64	1069-1939
		Women
		25-34	2-10
		35-44	24-37
		45-54	91-335
		55-64	367-852
Canada50	1974-1985	>25
		1985		Estimated from graph
		Men
		35-39	~ 100
		40-44	~ 250
		45-49	~ 400
		50-54	~500
		55-59	~ 1000
		60-64	~1100
		Women
		35-39	Cannot estimate from graph
		40-44	~ 50
		45-49	~ 100
		50-54	~ 200
		55-59	~ 250
		60-64	~ 400

* CI=confidence interval

Summary of Aim 1:

In summary, there is a wide range of estimated incidences of seizures, strokes, and MIs. Although strokes and MIs are more likely to occur in the older population, they occur in all age groups. Unprovoked seizures can occur with equally high frequency among all age groups.

Aim 2 ‚ Estimates of Incidence of Events Among Ephedra Users

Seizures

Figure 1 presents the range of estimated risk of seizures per 100,000 person-years of ephedra use. The age range of seizures reported to FDA is 21-51 years old. The background risk of seizure in the U.S. population in approximately this age range is derived from the two U.S. studies presented in Table 6. In the study of Hauser et al.,¹⁶ the rates in the 21-40 and 41-60 year age groups are approximately 25-35 and 20-35 per 100,000, respectively. In the study by Annegers et al.,¹⁸ the rates per 100,000 in the 15-24, 25-34, 35-44, and 45-54 age groups are, respectively: 50.9, 59.6, 23.3, 20.0, and 24.8. Therefore, Figure 1 uses a range of 20 to 60 per 100,000.

Using the above estimates, Figure 1 demonstrates that the estimated risk of seizures among ephedra users is unlikely to exceed the underlying risk in the population, even under the most extreme assumptions of 1% reporting of events and the most conservative assumption of ephedra use. Although most of the risk estimates are lower than the population range, this should not be interpreted as a protective effect of ephedra. It is possible that reporting rates are, in fact, towards the lower limits shown in the graph because seizures are more common in the younger population and often not life-threatening; therefore, these events may be less likely to be attributed and reported to FDA.

Stroke

Figure 2 presents the range of estimated risk of stroke per 100,000 person-years of ephedra use. The age range of strokes reported to FDA is 18-64. Sixty-nine percent of reports with known age involved people over the age of 34 and almost half were in people older than 45.

The background risk of stroke in a similar age range is derived from several U.S. studies presented in Table 7. In Petitti et al.,¹⁹ the risk of stroke in women 15-44 is 10.7 per 100,000. The study of Broderick²⁶ estimated the stroke incidence in people under 45 years old to be about 10 per 100,000; in those 45-54, about 100 per 100,000; and in those 55-64, about 200 per 100,000. Brown et al.³⁴ estimated the risk in men and women 0-34 years old to be only 4 and 3 per 100,000 respectively, but the risk increased substantially in those over 34: in men the incidence ranged from 36 to 363 per 100,000 and in women, from 29 to 195 per 100,000. In Kittner et al.,³⁹ the incidence of stroke is separated into infarction and hemorrhage. Combining these (as they are combined in the estimates of strokes among reports to FDA), the incidence in the 15-39 age group varies (depending on gender and ethnicity) from about 7 to about 20 per 100,000. In the 40-44 year age group, the range is about 42 to 146 per 100,000. Given the wide range of estimates in the literature, and to be conservative, Figure 2 uses a range of 3 to 60 per 100,000. The upper range of 60 was used because it represents the lowest rate in people 45-54 in the U.S. studies. This upper limit may be too low because almost half of all reported strokes occurred in individuals older than 45 years, and over 20% were in people older than 54; the incidence in the 45-64 year old population is almost uniformly over 60 per 100,000 and often substantially higher (e.g., around 100-200^{23,25,26,29,30,40}) .

For almost all of the assumptions made, the incidence of strokes among ephedra users is within the estimated range for background risk in the population. For many of the assumptions made, the risk is no greater than even the lower end. Only for the most conservative assumption of use and the assumption of 1% reporting is the estimate above the range. However, it is not strikingly higher and is certainly consistent with the higher stroke rates of a population that includes people over the age of 45.

Myocardial Infarction

Figure 3 presents the range of estimated risk of MI per 100,000 ephedra users. The age range of MIs reported is 15-59, and 30% of reported events occurred in people over the age of 44. Among MIs in those younger than 44, 57% were in men. Based on the two U.S. studies presented in Table 6, the range of MI incidence in the population ages 15-44 is about 5 per 100,000 (women only, ages 15-44)¹⁹ to 41 per 100,000 (ages 25-44).⁴⁸ Because 30% of MIs occurred in people older than 44, these latter estimates do not reflect the higher incidence in the > 44 population. For example, in the latter study, the incidence was 234 per 100,000 in those 45-54. However, a summary incidence for all people ages 15 to 59 is not available from the U.S. It is certainly higher than the estimates for those younger than 45. For example, in one French study, the incidence in men 30-59 was 103 per 100,000.⁴³ Nonetheless, a conservative range of 5-41 per 100,000 person-years was used.

For almost all of the assumptions made, the incidence of MI among ephedra users is within the estimated range for background risk in the population. For many of the assumptions made, the risk is no greater than even the lower end. Only for the most extreme assumptions of under-reporting and usage was the incidence higher than the population-range. As for stroke, this estimate was not strikingly higher. Furthermore, because the population range is for people less than 45 years old, the comparison is biased towards showing more events among the older ephedra MI cases.

Limitations

As discussed previously, there are numerous limitations to the analysis for Aim 2. This is due to the many limitations of adverse event reporting data and the limited data available on the exposed population. First, the true number of adverse events is not known, for the reasons previously described. An attempt was made to provide a range of estimates that, if anything, would overestimate the number of events. For example, all events, including those in the "insufficient data" group and those that may not be included in background estimates (e.g., unstable angina and transient ischemic attacks), were included. Because the reporting rate of adverse events is not known, a range of estimates was used with the lowest estimate (1%) likely representing a true lower estimate given the severity and nature of the events studied (although even this cannot be known for sure). Second, the number of exposed individuals was not known. Again, attempts were made to, if anything, underestimate this number. For example, it was assumed for the primary estimates that, even though only 13 companies reported on ephedra sales (31% of those surveyed), this represented total sales of ephedra. The average number of servings per day used for the calculations (four) was also probably an overestimate, thus underestimating the number of ephedra users. The data from the Texas study suggest that, in fact, the person-time estimate used was indeed an underestimate. Third, it is likely that not all ephedra sold was actually consumed. This is a limitation of all analyses using sales data and is similar for studies of drugs that use prescription data (because not all patients take the drugs that are prescribed). However, in randomized trials of OTC ephedrine combined with caffeine, the drop-out rate from side-effects is about 4-16%.^51,52 Even assuming that 16% of all ephedra sold is not consumed, the results do not substantially change. Fourth, and most importantly, a direct comparison of risk of events among users of ephedra alkaloid-containing products versus non-users cannot be made. Such a comparison is necessary to determine if there is or is not an association between ephedra use and the outcomes examined.

Of course, different assumptions of all parameters and different time periods of reporting may have led to somewhat different results. This simply highlights the known limitation of using adverse event reporting and sales data to answer questions about cause-and-effect relationships.

Summary of Aim 2

This analysis demonstrates the need to consider background risk in evaluating adverse event reports. Although seizures, strokes, and MIs are relatively rare in the younger population, they do occur. Given the substantial number of people who are likely to use dietary supplements containing ephedra alkaloids, the number of events reported may not be inconsistent with the expected number of events in the absence of ephedra use. Only under the most extreme assumptions was the estimated risk of events near the upper limit of estimated risk in the population. In no instance was the risk extremely elevated, as it was, for example, with fenfluramine. When fenfluramine and dexfenfluramine were removed from the market the estimated prevalence of valve abnormalities based on case-series was over 30-fold higher than the estimated population prevalence.

However, this analysis was not designed to rule in or out a possible cause-and-effect relationship between ephedra and the outcomes evaluated, nor could it do so. Only formal comparative studies can be used to compare the risk of events in ephedra users compared with non-users. In their absence, this analysis suggests that the adverse events reported among ephedra users may very well represent simply the background rate of events expected among such a large number of people, unrelated to ephedra use.

 

References

1. Uusküla M, Lamp K, Väli M. An age-related difference in the ratio of sudden coronary death over acute myocardial infarction in Estonian males. J Clin Epidemiol 1998;51:577-80.

2. Wilhelmsen L, Rosengren A, Johansson S, Lappas G. Coronary heart disease attack rate, incidence and mortality 1975-1994 in Göteborg, Sweden. Eur Heart J 1997;18:572-81.

3. White AD, Rosamond WD, Chambless LE, Thomas N, Conwill D, Cooper LS, Folsom AR. Sex and race differences in short-term prognosis after acute coronary heart disease events: the Atherosclerosis Risk In Communities (ARIC) study. Am Heart J 1999;138:540-8.

4. American Heart Association. 2000 Heart and Stroke Statistical Update. 1999; Dallas, Texas: American Heart Association.

5. Baum C, Kweder SL, Anello C. The Spontaneous Reporting System in the United States. In: Strom BL, editor. Pharmacoepidemiology. Second ed. New York, NY: John Wiley & Sons, 1994:125-137.

6. Hensrud DD, Engle DD, Scheitel SM. Underreporting the use of dietary supplements and nonprescription medications among patients undergoing a periodic health examination. Mayo Clin Proc 1999;74:443-7.

7. Barnes J, Mills SY, Abbot NC, Willoughby M, Ernst E. Different standards for reporting ADRs to herbal remedies and conventional OTC medicines: face-to-face interviews with 515 users of herbal remedies. Br J Clin Pharmacol 1998;45:496-500.

8. Fletcher AP. Spontaneous adverse drug reaction reporting vs event monitoring: a comparison. Journal of the Royal Society of Medicine 1991;84:341-4.

9. Fletcher AP. An appraisal of spontaneous adverse event monitoring. Adverse Drug Reactions & Toxicological Reviews 1992;11:213-27.

10. Rosenthal S, Chen R. The reporting sensitivities of two passive surveillance systems for vaccine adverse events. Am J Public Health 1995;85:1706-9.

11. Moride Y, Haramburu F, Requejo AA, Begaud B. Under-reporting of adverse drug reactions in general practice. Br J Clin Pharmacol 1997;43:177-81.

12. Alvarez-Requejo A, Carvajal A, Begaud B, Moride Y, Vega T, Arias LH. Under-reporting of adverse drug reactions. Estimate based on a spontaneous reporting scheme and a sentinel system. Eur J Clin Pharmacol 1998;54:483-8.

13. Nightingale SL. From the Food and Drug Administration. JAMA 1992;268:1828

14. Department of Health and Human Services, Food and Drug Administration. Dietary supplements containing ephedrine alkaloids. Fed Regist 1997;62:30678-724.

15. Forsgren L, Bucht G, Eriksson S, Bergmark L. Incidence and clinical characterization of unprovoked seizures in adults: a prospective population-based study. Epilepsia 1996;37:224-9.

16. Hauser WA, Annegers JF, Kurland LT. Incidence of epilepsy and unprovoked seizures in Rochester, Minnesota: 1935-1984. Epilepsia 1993;34:453-68.

17. Forsgren L. Prospective incidence study and clinical characterization of seizures in newly referred adults. Epilepsia 1990;31:292-301.

18. Annegers JF, Dubinsky S, Coan SP, Newmark ME, Roht L. The incidence of epilepsy and unprovoked seizures in multiethnic, urban health maintenance organizations. Epilepsia 1999;40:502-6.

19. Petitti DB, Sidney S, Quesenberry CP, Jr., Bernstein A. Incidence of stroke and myocardial infarction in women of reproductive age. Stroke 1997;28:280-3.

20. Reunanen A, Aho K, Aromaa A, Knekt P. Incidence of stroke in a Finnish prospective population study. Stroke 1986;17:675-81.

21. Nencini P, Inzitari D, Baruffi MC, Fratiglioni L, Gagliardi R, Benvenuti L, Buccheri AM, Cecchi L, Passigli A, Rosselli A. Incidence of stroke in young adults in Florence, Italy. Stroke 1988;19:977-81.

22. Sacco SE, Whisnant JP, Broderick JP, Phillips SJ, O'Fallon WM. Epidemiological characteristics of lacunar infarcts in a population. Stroke 1991;22:1236-41.

23. Lauria G, Gentile M, Fassetta G, Casetta I, Agnoli F, Andreotta G, Barp C, Caneve G, Cavallaro A, Cielo R. Incidence and prognosis of stroke in the Belluno province, Italy. First-year results of a community-based study. Stroke 1995;26:1787-93.

24. Ashok PP, Radhakrishnan K, Sridharan R, el-Mangoush MA. Incidence and pattern of cerebrovascular diseases in Benghazi, Libya. J Neurol Neurosurg Psychiatry 1986;49:519-23.

25. Anderson CS, Jamrozik KD, Burvill PW, Chakera TM, Johnson GA, Stewart-Wynne EG. Ascertaining the true incidence of stroke: experience from the Perth Community Stroke Study, 1989-1990. Med J Aust 1993;158:80-4.

26. Broderick JP. Stroke trends in Rochester, Minnesota, during 1945 to 1984. Ann Epidemiol 1993;3:476-9.

27. Kita Y, Okayama A, Ueshima H, Wada M, Nozaki A, Choudhury SR, Bonita, Inamoto Y, Kasamatsu T. Stroke incidence and case fatality in Shiga, Japan 1989-1993. Int J Epidemiol 1999;28:1059-65.

28. Johansson B, Norrving B, Lindgren A. Increased stroke incidence in Lund-Orup, Sweden, between 1983 to 1985 and 1993 to 1995. Stroke 2000;31:481-6.

29. Jamrozik K, Broadhurst RJ, Lai N, Hankey GJ, Burvill PW, Anderson CS. Trends in the incidence, severity, and short-term outcome of stroke in perth, Western Australia. Stroke 1999;30:2105-11.

30. Falkeborn M, Persson I, Terent A, Bergstrom R, Lithell H, Naessen T. Long-term trends in incidence of and mortality from acute myocardial infarction and stroke in women: Analyses of total first events and of deaths in the Uppsala Health Care Region, Sweden. Epidemiology 1996;7:67-74.

31. Giroud M, Gras P, Chadan N, Beuriat P, Milan C, Arveux P, Dumas R. Cerebral haemorrhage in a French prospective population study. J Neurol Neurosurg Psychiatry 1991;54:595-8.

32. Ellekjaer H, Holmen J, Indredavik B, Terent A. Epidemiology of stroke in Innherred, Norway, 1994 to 1996. Incidence and 30-day case-fatality rate. Stroke 1997;28:2180-4.

33. Nakayama T, Date C, Yokoyama T, Yoshiike N, Yamaguchi M, Tanaka H. A 15.5-year follow-up study of stroke in a Japanese provincial city. The Shibata Study. Stroke 1997;28:45-52.

34. Brown RD, Whisnant JP, Sicks JD, O'Fallon WM, Wiebers DO. Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989. Stroke 1996;27:373-80.

35. Rozenthul-Sorokin N, Ronen R, Tamir A, Geva H, Eldar R. Stroke in the young in Israel. Incidence and outcomes. Stroke 1996;27:838-41.

36. Thorvaldsen P, Davidsen M, Bronnum-Hansen H, Schroll M. Stable stroke occurrence despite incidence reduction in an aging population: stroke trends in the danish monitoring trends and determinants in cardiovascular disease (MONICA) population. Stroke 1999;30:2529-34.

37. Bonita R, Broad JB, Beaglehole R. Changes in stroke incidence and case-fatality in Auckland, New Zealand, 1981-91. Lancet 1993;342:1470-3.

38. Czlonkowska A, Ryglewicz D, Weissbein T, Baranska-Gieruszczak M, Hier, DB. A prospective community-based study of stroke in Warsaw, Poland. Stroke 1994;25:547-51.

39. Kittner SJ, McCarter RJ, Sherwin RW, Sloan MA, Stern BJ, Johnson CJ, Buchholz D, Seipp MJ, Price TR. Black-white differences in stroke risk among young adults. Stroke 1993;24:I13-I15

40. Rastenyte D, Cepaitis Z, Sarti C, Bluzhas J, Tuomilehto J. Epidemiology of stroke in Kaunas, Lithuania. First results from the Kaunas Stroke Register. Stroke 1995;26:240-4.

41. Guidetti D, Baratti M, Zucco RG, Greco G, Terenziani S, Vescovini E, Sabadini R, Bondavalli M, Masini L, Salvarani C. Incidence of stroke in young adults in the Reggio Emilia area, northern Italy. Neuroepidemiology 1993;12:82-7.

42. Laks T, Tuomilehto J, Joeste E, Maeots E, Pullisaar O, Salomaa V, Viidebaum T, Palomaki P, Karu K, Torppa J. Acute myocardial infarction in Estonia: first results from the Tallinn AMI Register. J Clin Epidemiol 1996;49:573-9.

43. Lang T, Ducimetiere P, Arveiler D, Amouyel P, Cambou JP, Ruidavets, JB, Montaye M, Meyer V, Bingham A. Incidence, case fatality, risk factors of acute coronary heart disease and occupational categories in men aged 30-59 in France. Int J Epidemiol 1997;26:47-57.

44. Volmink JA, Newton JN, Hicks NR, Sleight P, Fowler GH, Neil HA. Coronary event and case fatality rates in an English population: results of the Oxford myocardial infarction incidence study. The Oxford Myocardial Infarction Incidence Study Group. Heart 1998;80:40-4.

45. Peltonen M, Rosen M, Lundberg V, Asplund K. Social patterning of myocardial infarction and stroke in Sweden: incidence and survival. Am J Epidemiol 2000;151:283-92.

46. Laks T, Tuomilehto J, Joeste E, Maeots E, Salomaa V, Palomaki P, Pullisaar O, Karu K, Torppa J. Alarmingly high occurrence and case fatality of acute coronary heart disease events in Estonia: results from the Tallinn AMI register 1991-94. J Intern Med 1999;246:53-60.

47. Kirchhoff M, Davidsen M, Bronnum-Hansen H, Hansen B, Schnack H, Eriksen LS, Madsen M, Schroll M. Incidence of myocardial infarction in the Danish MONICA population 1982-1991. Int J Epidemiol 1999;28:211-8.

48. Goldberg RJ, Gore JM, Alpert JS, Dalen JE. Recent changes in attack and survival rates of acute myocardial infarction (1975 through 1981). The Worcester Heart Attack Study. JAMA 1986;255:2774-9.

49. Tuomilehto J, Arstila M, Kaarsalo E, Kankaanpaa J, Ketonen M, Kuulasmaa K, Lehto S, Miettinen H, Mustaniemi H, Palomaki P. Acute myocardial infarction (AMI) in Finland--baseline data from the FINMONICA AMI register in 1983-1985. Eur Heart J 1992;13:577-87.

50. Trends in incidence and mortality from acute myocardial infarction in Nova Scotia and Saskatchewan 1974 to 1985. The Nova Scotia-Saskatchewan Cardiovascular Disease Epidemiology Group. Can J Cardiol 1992;8:253-8.

51. Pasquali R, Baraldi G, Cesari MP, Melchionda N, Zamboni M, Stefanini C, Raitano A. A controlled trial using ephedrine in the treatment of obesity. Int J Obes 1985;9:93-8.

52. Toubro S, Astrup AV, Breum L, Quaade F. Safety and efficacy of long-term treatment with ephedrine, caffeine and an ephedrine/caffeine mixture. Int J Obes 1993;17:S69-S72

________________________ _________________

Stephen E. Kimmel, MD, MS Date

Back to top