Preventive Medicine 39 (2004) 1243 – 1248

www.elsevier.com/locate/ypmed

Prescription weight loss pill use among Americans: patterns of pill use
and lessons learned from the fen-phen market withdrawal
Heidi Michels Blanck, Ph.D., * Laura Kettel Khan, Ph.D., and Mary K. Serdula, M.D.
Division of Nutrition and Physical Activity, National Center for Chronic Disease Prevention and Health Promotion,
Centers for Disease Control and Prevention, Atlanta, GA 30341-3717, USA

Available online 15 June 2004

Abstract

Background. Despite the popularity of antiobesity medications, there is a lack of population-based data on their use. In addition, response
(termination of pill use and receipt of an echocardiogram) to the fenfluramine and dexfenfluramine market withdrawal among the public has
not been described. Lessons learned from this event have implications for future withdrawals.
Methods. We used data from the Behavioral Risk Factor Surveillance System (BRFSS) a random-digit telephone survey. In 1998, six
states included detailed questions about the use of prescription weight loss pills in the previous 2 years, n = 16,460 noninstitutionalized adults
aged 18 years or older.
Results. Almost one third of prescription weight loss pills users were not obese before taking pills. Family and friends and other
nonphysicians were reported as sources of medication by one in ten users. One third of users also reported taking nonprescription diet
products. Among fenfluramine or dexfenfluramine users, one third continued pill use after the market withdrawal and only one quarter
received echocardiograms.
Conclusions. Despite enormous publicity, many persons continued to use fen-phen after the market withdrawal and most did not receive
follow-up echocardiograms. Our study raises issues regarding the effectiveness of withdrawal warnings in a small but significant subset.
Additional means of communicating risk to individuals are needed for future product withdrawals including special strategies for those
lacking healthcare coverage.
Published by The Institute For Cancer Prevention and Elsevier Inc.

Keywords: Obesity; Prescriptions (drug); Behavioral Risk Factor Surveillance System

Introduction in significant weight loss when used over an extended

period of time [3,4]. After these publications, some physi-
Approximately 44% of women and 29% of men are cians prescribed fenfluramine or dexfenfluramine in combi-
trying to lose weight [1]. To aid in the battle of weight nation with phentermine, ‘‘fen-phen,’’ for use in weight loss
control, many turn to prescription weight loss drugs [2]. programs [5]. Besides physicians’ offices, the use of fen-
During the mid-1990s, three prescription drugs were avail- phen appeared to be popular at commercial diet clinics.
able for the management of obesity. These drugs suppressed Following reports of cardiac valvulopathy in persons
appetite (anorectics) through central nervous system mech- taking fenfluramine or dexfenfluramine [6], the FDA
anisms. Phentermine, a noradrenergic anorectic, was ap- issued a public health advisory on July 8, 1997, that led
proved in 1959. Fenfluramine, a serotonergic anorectic, was to the September 15, 1997, voluntary withdrawal of these
approved in 1973; and dexfenfluramine, another serotoner- two drugs from the U.S. markets [7,8]. Two months later,
gic anorectic, was approved for use in 1996. the U.S. Department of Health and Human Services
In 1992, articles were published suggesting that the recommended that all users of these pills undergo a
combined use of phentermine and fenfluramine could result medical history and cardiovascular examination; in addi-
tion, practitioners were encouraged to obtain an echocar-
diogram on all patients who had used these drugs before
* Corresponding author. Division of Nutrition and Physical Activity,
Chronic Disease Nutrition Branch, 4770 Buford Highway NE MS K-26,
undergoing any invasive procedure for which antimicrobial
Atlanta, GA 30341-3717. Fax: +1-770-488-6027. prophylaxis for endocarditis is recommended [9]. Because
E-mail address: [email protected] (H.M. Blanck). such prophylaxis is recommended even for common dental

0091-7435/$ - see front matter. Published by The Institute For Cancer Prevention and Elsevier Inc.
doi:10.1016/j.ypmed.2004.04.040
1244 H.M. Blanck et al. / Preventive Medicine 39 (2004) 1243–1248

and oral procedures such as the cleaning of teeth and the taking prescription weight loss pills for the first time? How
filling of dental caries [10], the recommendation applied to much weight did you lose in total in the past 2 years while
most users. using prescription weight loss pills, even if you gained some
To assess use of weight loss pills in wake of the fen-phen of it back?’’ At the end of the survey, respondents were
withdrawal, six states in the 1998 Behavioral Risk Factor asked their current height and weight. Using reported weight
Surveillance System (BRFSS) survey asked detailed ques- lost in the past 2 years while using pills, we calculated the
tions about use of prescription weight loss pills and non- proportion of persons who lost at least 10% of their prepill
prescription (including over-the-counter and dietary (weight before taking prescription weight loss pills for the
supplement) weight loss products. We used this unique data first time) body weight. We also calculated the difference
to assess usage patterns following the market withdrawal. between weight at the time of the survey and reported
Among fenfluramine or dexfenfluramine users, we analyzed prepill weight and determined the proportion of persons
patterns of discontinuation and receipt of recommended who maintained a loss of at least 10% of their prepill body
echocardiograms. Additionally, we assessed prescription pill weight. Respondents in five states were also asked about
usage by type, duration, source, whether pills were used in any use of over-the-counter (nonprescription) weight loss
accordance with recommended weight guidelines, weight products in the previous 2 years including use of ephedra-
loss, and use of other nonprescription weight loss products. based products [12]. Body mass index (BMI) was calculated
as weight in kilograms divided by height in meters squared
and was categorized as < 25, normal weight; 25 – 29.9,
Methods overweight; z 30, obesity [13]. According to national
guidelines, pharmacotherapy may be considered for persons
The data were obtained from adults aged 18 years or who are clinically obese (BMI z 30) or have a BMI of 27–
older who participated in the 1998 BRFSS in Arizona, 29.9 plus an obesity-related comorbid condition (for exam-
Florida, Iowa, Michigan, West Virginia, or Wisconsin. The ple, hyperlipidemia, hypertension, diabetes, or cardiovascu-
BRFSS uses a multistage cluster design based on random lar disease) [13]. Because the BRFSS survey did not collect
digit dialing methods to select a representative sample from complete information on comorbid conditions (only diabe-
each state’s noninstitutionalized residents. Further details tes diagnosis was available on the entire sample), we
about the BRFSS have been published [11]. The average conservatively defined inappropriate pill use as pill use in
cooperation rate [completed interviews / (refusals + termi- persons whose BMI was < 27 [2].
nations + completed interviews)] for the six states was We excluded all respondents for whom certain data were
72.2% (range 45.4 – 95.4%). missing: weight loss product use (n = 29); weight or height
Respondents were asked the following question: ‘‘In the (n = 619); and sociodemographic factors, including age, race,
past two years, have you taken any weight loss pill pre- and education (n = 142). We also excluded all pregnant
scribed by a doctor? Do not include water pills or thyroid women (n = 197). Five respondents were excluded because
medications.’’ Respondents who answered ‘‘yes’’ were their reported weight, height, or BMI was outside the sex-
asked, ‘‘What is the name of the prescription weight loss specific minimum and maximum reference values from the
pill you used most often during the past 2 years? Are you Third National Health and Nutrition Examination Survey
currently taking this pill? In what month and year did you (NHANES III), 1988– 1994, which used measured heights
stop taking this pill? In your lifetime, what is the total and weights [14]. The final analytical sample was n = 16,460.
number of months or years that you have taken this pill?’’ For the purpose of examining the use of specific types of
Respondents were asked these same questions for the prescription weight loss pills, we classified users in five
prescription weight loss pill used next most often in the categories: (1) fen-phen (n = 164, the combination of
past 2 years. In addition, respondents were asked the name fenfluramine or dexfenfluramine with phentermine hydro-
of the prescription weight loss pill they used third most chloride or resin), (2) fenfluramine or dexfenfluramine (n =
often but no further questions about its usage. If the 71, not in combination with phentermine hydrochloride or
respondent said the pill they used most or second most resin), (3) phentermine (n = 51), (4) other prescription
often included fenfluramine or dexfenfluramine, they were weight loss pill (n = 15, persons who did not take fenflur-
asked, ‘‘Since taking these weight loss pills, have you amine, dexfenfluramine, or phentermine), and (5) do not
received an echocardiogram? An echocardiogram is a spe- know or refused (n = 53; 45 did not know, eight refused).
cial picture of your beating heart similar to an ultrasound.’’ Data on termination of fenfluramine or dexfenfluramine use
In addition, all participants were asked, ‘‘In the past 2 were ‘‘do not know’’ or ‘‘refused’’ for 58 participants. There
years, did you usually obtain your prescription weight loss was no statistical difference in sex, age, race, education, or
pills from: commercial diet clinic, Health Maintenance BMI among those with pill use termination data versus
Organization (HMO), any other health care provider, friend those without.
or family member, other source, don’t know/not sure, Data are weighted for age, sex, and race or ethnicity
refused.’’ Individuals were also asked about their weight before data analysis. This weighting allows for inference
including: How much did you weigh just before you started to each state population. To account for the complex
 H.M. Blanck et al. / Preventive Medicine 39 (2004) 1243–1248 1245

Table 1 Table 1 (continued)

Prevalence of selected characteristics by sex among prescription weight loss (n = 354) (n = 301) (n = 34) P value
pill usersa Total Female Male
(n = 354) (n = 301) (n = 34) P value % (SE) % (SE) % (SE)
Total Female Male
Lost >10% of prepill body
% (SE) % (SE) % (SE)
weight while using
Age group prescription weight loss
18 – 34 years 33.3 (3.5) 35.4 (3.6) 21.6 (11.9) pills in the past 2 years
35 – 54 years 53.8 (3.5) 52.0 (3.6) 63.4 (11.4) Yes 45.5 (3.5) 46.1 (3.7) 42.5 (10.5)
z 55 years 13.0 (2.2) 12.6 (2.4) 15.0 (6.3) 0.50 Do not know or refused 22.3 (3.0) 25.2 (3.2) 15.6 (7.3) 0.55
Race Maintained a loss >10%
Non-Hispanic White 77.0 (3.3) 76.4 (3.3) 79.8 (12.0) of prepill body weight
Non-Hispanic Black 9.7 (2.1) 11.5 (2.4) 0.0 (0.0) Yes 26.9 (3.0) 28.6 (3.2) 17.9 (7.9)
Hispanic 13.0 (2.9) 11.7 (2.6) 20.2 (12.0) Do not know or refused 4.6 (1.3) 4.5 (1.3) 5.1 (4.3) 0.48
Other 0.3 (0.3) 0.3 (0.3) 0.0 (0.0) 0.003 Nonprescription weight
Education loss product use in the
< High school 10.4 (2.4) 10.5 (2.5) 10.0 (6.7) past 2 yearsb
High school 33.7 (3.4) 33.6 (3.5) 34.4 (9.8) Yes 34.9 (3.4) 39.6 (3.7) 10.6 (5.5) 0.05
Some college 38.4 (3.4) 40.4 (3.4) 27.3 (11.7) SE indicates standard error.
College graduate 17.5 (2.3) 15.5 (2.3) 28.3 (8.7) 0.48 a
Data by self-report.
Smoking status b
Data available for n = 336.
Current 27.1 (3.3) 24.2 (3.1) 42.9 (11.4)
Former 29.2 (3.3) 28.0 (3.5) 35.2 (9.8)
Never 43.7 (3.5) 47.7 (3.7) 21.8 (7.9) 0.05 sampling design used by BRFSS, we used SAS and
Pill type SUDAAN to perform statistical analyses [15,16]. Chi-
Fen-phen 45.9 (3.5) 44.6 (3.7) 53.3 (10.8) square analysis was used to determine whether categorical
Fenfluramine or 22.3 (2.8) 20.6 (2.9) 31.9 (9.8) variables were statistically independent, at a significance
dexfenfluramine
Phentermine 13.4 (2.2) 15.0 (2.5) 4.5 (2.6)
level of a = 0.05.
Other 3.5 (1.1) 4.1 (1.2) 0.0 (0.0)
Do not know or refused 14.9 (2.5) 15.7 (2.7) 10.3 (6.1) 0.02
Body mass index (kg/m2)
before taking Results
prescription weight
loss pills In our sample of 16,460 individuals, 50.9% were women
< 27 13.3 (2.3) 15.7 (2.6) 0.0 (0)
and 29.4% were aged between 18 and 34 years, 37.4% between
27 – 29 16.8 (2.7) 19.5 (3.0) 2.4 (2.2)
z 30 65.3 (3.2) 60.3 (3.5) 92.5 (4.8) 35 and 54 years, and 33.2% 55 years of age and older; 80.9%
Do not know or refused 4.6 (1.3) 4.5 (1.3) 5.1 (4.3) 0.0001 were non-Hispanic White, 7.4% were non-Hispanic Black,
Source of prescription 9.3% were Hispanic, and 2.4% were classified as other race or
weight loss pills ethnicity; 35.4% of the sample reported a high school educa-
Diet clinic MD 10.7 (2.1) 10.5 (2.1) 12.5 (6.3)
tion and 51.9% reported at least some college education. The
HMO MD 15.3 (2.5) 14.1 (2.4) 22.2 (8.4)
Any other health 39.8 (3.2) 42.0 (3.3) 30.7 (3.5) mean age for the sample was 48.5 years.
care provider Two percent of participants in the six states used at least
Friend or family 7.9 (2.7) 6.2 (2.0) 14.2 (11.9) one prescription weight loss pill during 1996– 1998 versus
Other 7.8 (1.9) 8.4 (2.0) 10.2 (5.9) 2.5% nationally (50 state data from 1998 BRFSS) [3].
Do not know or refused 18.5 (2.9) 18.8 (3.0) 10.3 (6.1) 0.69
Among users, almost half used fen-phen during 1996–
Duration, months
1 27.0 (3.8) 25.5 (3.6) 34.5 (12.7) 1998; one quarter fenfluramine or dexfenfluramine; and
2 12.0 (2.2) 12.2 (2.4) 10.9 (5.6) about one seventh phentermine (Table 1). A small percent-
3 15.6 (2.6) 16.3 (2.8) 11.8 (6.9) age (3.5%) of users took other prescription pills. In this
4–6 17.0 (2.8) 16.4 (2.8) 20.8 (8.9) latter group were < 10 individuals who reported use of
7 – 11 8.1 (2.0) 8.9 (2.3) 4.9 (4.2)
sibutramine (available as of November 1998). Because of
> 12 17.1 (2.7) 17.2 (2.9) 16.3 (6.7)
Do not know or refused 3.2 (1.1) 3.7 (1.3) 4.8 (0.9) 0.76 this small proportion ( < 3% of the sample), we were not
Weight lost while using able to present data on use of this drug nor on the newer
prescription pills in the antiobesity drug orlistat (available April 1999).
past 2 years, pounds Most prescription weight loss pill users were between 34
<5 7.3 (1.6) 7.6 (1.7) 5.9 (4.4)
and 54 years of age, were non-Hispanic White, and had either
5 – 20 33.7 (3.4) 33.7 (3.4) 33.5 (11.6)
21 – 40 22.7 (2.8) 23.2 (3.0) 20.0 (7.6) a high school or some college education. The majority of
>40 16.9 (2.5) 15.3 (2.5) 25.7 (8.9) users had a BMI of at least 30 when they started taking pills;
Do not know or refused 19.5 (2.9) 20.3 (3.1) 14.8 (7.3) 0.80 however, 13% had a BMI < 27 (Table 1). Seventeen percent
of users had a prepill BMI between 27 and 29.9. Half of users
received their pills from a traditional health care provider or
1246 H.M. Blanck et al. / Preventive Medicine 39 (2004) 1243–1248

HMO: 11% a diet clinic; 8% a friend or family member; and BMIs < 27 versus only 4% of phentermine users (Table 2).
8% some other source. One third of users also reported taking The usual provider of pills differed by pill type, with 10% of
a nonprescription weight loss product in the past 2 years. fen-phen and fenfluramine or dexfenfluramine users receiv-
Women were five times more likely as men to have taken a ing their pills from a friend or family member versus only 1%
prescription weight loss pill (3.1% versus 0.6%, P < 0.001). of phentermine users. Duration of pill use was similar by pill
Nearly all male users were obese (92.5%) compared to only type. Mean and median weight loss was similar by pill type:
three fifths (60.3%) of women. No men with a BMI < 27 were 30.7 and 27.5 pounds for fen-phen, 27.6 and 20.0 pounds for
users as compared to almost one sixth of women. Median fenfluramine or dexfenfluramine, 30.8 and 30.0 pounds for
duration of use was 3 months for both sexes. In both sexes, phentermine, and 30.2 and 17.0 pounds for other pill. Neither
over 40% of users had at some point achieved at least a 10% weight loss of at least 10% prepill body weight or weight
weight loss after having started using prescription diet pills. maintenance (maintaining a loss z 10% of prepill body
Over half the users of each pill type were obese at the time weight) differed by pill type. Overall use of nonprescription
they started taking the pills, but approximately 11% of users weight loss products in the past 2 years did not differ
of fenfluramine or dexfenfluramine or fen-phen took pills had statistically by pill type nor did use of ephedra-based non-

Table 2
Prevalence of selected weight and lifestyle characteristics by prescription weight loss pill typea
Total Fen-phen Dexfenfluramine Phentermine P value
(n = 354) (n = 164) or fenfluramine (n = 51)
% (SE) % (SE) (n = 71) % (SE) % (SE)
Body mass index (kg/m2) before taking prescription weight loss pills
< 27 13.3 (2.3) 12.3 (3.2) 11.3 (4.5) 4.4 (2.7)
27 – 29 16.8 (2.7) 16.8 (4.4) 10.1 (4.4) 25.1 (7.5)
z 30 65.3 (3.2) 67.6 (5.1) 75.7 (6.2) 66.9 (8.0)
Do not know or refused 4.6 (1.3) 3.3 (1.8) 2.9 (1.7) 3.6 (2.2) 0.37

Source of prescription weight loss pills

Diet clinic MD 10.7 (2.1) 16.1 (3.6) 0.2 (0.2) 20.7 (6.9)
HMO MD 15.3 (2.4) 18.5 (3.9) 23.7 (6.4) 9.3 (5.4)
Any other health care provider 39.8 (3.2) 43.5 (5.2) 51.9 (7.5) 56.5 (8.2)
Friend or family 7.9 (2.7) 9.9 (4.7) 10.0 (5.6) 1.1 (1.1)
Other 8.7 (1.9) 7.9 (2.8) 12.9 (5.0) 8.7 (4.0)
Do not know or refused 18.5 (2.8) 4.1 (3.0) 1.3 (1.1) 3.6 (3.5) 0.0002
Duration, months
1 27.0 (3.8) 25.7 (5.5) 32.6 (7.0) 22.8 (8.0)
2 12.0 (2.2) 13.8 (3.3) 11.3 (4.6) 9.3 (4.4)
3 15.6 (2.6) 14.0 (3.5) 17.8 (5.5) 19.3 (6.6)
4–6 17.0 (2.8) 18.9 (3.9) 11.9 (5.2) 20.8 (7.1)
7 – 11 8.1 (2.0) 9.5 (3.1) 8.4 (3.9) 5.0 (3.8)
>12 17.1 (2.7) 17.6 (3.7) 12.4 (5.7) 19.0 (6.6)
Do not know or refused 3.2 (1.1) 0.5 (0.6) 5.7 (3.1) 3.7 (2.5) 0.70

Weight lost while taking prescription pills in the past 2 years, pounds
<5 7.3 (1.6) 5.8 (2.0) 9.8 (4.0) 14.9 (6.3)
5 – 20 33.7 (3.4) 38.5 (5.4) 49.7 (7.5) 24.1 (7.1)
21 – 40 22.7 (2.8) 28.9 (4.6) 15.2 (5.1) 41.3 (8.5)
>40 16.9 (2.5) 20.8 (4.0) 21.1 (6.2) 17.5 (6.6)
Do not know or refused 19.5 (2.9) 5.9 (3.3) 4.3 (3.2) 2.3 (2.3) 0.11

Lost at least 10% of prepill body weight while using prescription weight loss pills in the past 2 years
Yes 48.1 (3.5) 59.4 (3.7) 42.9 (7.4) 69.9 (7.8)
Do not know or refused 4.6 (1.3) 3.3 (1.8) 2.9 (1.7) 3.6 (2.2) 0.16

Maintained a loss >10% of prepill body weight

Yes 26.9 (3.0) 19.3 (3.9) 30.4 (7.2) 41.5 (8.3)
Do not know or refused 4.6 (1.3) 3.3 (1.8) 2.9 (1.7) 3.6 (2.2) 0.17

Nonprescription weight loss

Product useb 34.9 (3.4) 35.4 (5.1) 33.9 (8.0) 34.7 (5.3) 0.65
*SE indicates standard error.
a
Data by self-report.
b
Data available for n = 336.
 H.M. Blanck et al. / Preventive Medicine 39 (2004) 1243–1248 1247

prescription weight loss products. We found that 10.2% of were willing to risk a cardiac valvulopathy for the real or
phentermine users who also reported using a nonprescription perceived efficacy of the pharmacotherapy.
diet product reported use of an ephedra-based product Our study indicates that despite government recommen-
compared to 16.9% and 11.1% of fen-phen or fenfluramine dations, only one quarter of fenfluramine or dexfenflur-
or dexfenfluramine users, respectively. amine users reported receiving an echocardiogram. It is
conceivable that some users underwent less definitive car-
Response to fen-phen withdrawal diovascular evaluation or saw their physicians but failed to
tell them about their use of these pills. Although we were
As of May 1997, 92 of the participants (n = 187) reported unable to determine reasons for the low proportion of
still taking the fenfluramine or dexfenfluramine pills. After testing, we did find that a lack of health coverage was one
the FDA public health advisory (July and August 1997) but factor. This suggests that individuals with transient coverage
before the market withdrawal (September 1997), 32 (35%) or no health coverage may fall through the cracks. Although
of the 92 participants stopped. Of the remaining 60 users, 27 there is potentially greater caution in the use of currently
(45%) stopped taking the pills in September and October. available prescription weight loss pills due to the fen-phen
The last 33 persons stopped taking the pills over the course withdrawal, in the case of future withdrawals where follow-
of the next 13 months. A comparison of early (July and up is indicated, communication strategies should reach
August 1997), moderate (September and October 1997), and outside the healthcare system to alert uninsured patients.
late (November 1997 – December 1998) termination of use These strategies could include use of posted notices at diet
found a significant trend by BMI and a suggestion of a sex clinics and pharmacies as well as Internet sites.
difference (data not shown). Those who were obese were Although fenfluramine and dexfenfluramine are no lon-
significantly more likely to be late terminators than those ger available, physicians can still prescribe phentermine,
who were not obese [BMI: z 30 (n = 70) 45.0%; BMI 27 – which was the most common antiobesity medication in
29.9 (n = 14) 16.2%; < 27 (n = 8) 20.2%], P = 0.04. Women 2001 and early 2002 [17]. In our survey, phentermine users
were evenly divided among the three groups, with one third had mean weight loss of 14.0 kg compared to 6.3 kg (range,
in the late terminator group; in contrast, over three fourths of 3.6 – 8.8 kg) in trials of phentermine (n = 6) reviewed by
men were in the late group, P = 0.07. We found no statistical Haddock et al. [18]. This difference might be due to
difference in termination of use by age, race or ethnicity, pill duration, as over one quarter of phentermine users in our
source, duration, weight loss of at least 10%, healthcare survey used the drug more than 6 months while the clinical
coverage, nonprescription diet product use, or receipt of trials of phentermine were 6 months or less. Forty-two
echocardiogram. percent of phentermine users in our sample reported taking
Among users of fenfluramine or dexfenfluramine, about the pill for more than 12 weeks, beyond the duration of
one quarter (26.7%, SE = 5.0) had an echocardiogram with which it is approved for use by the FDA [19]. About 70% of
the proportion similar for fen-phen (27.1%) and fenfluramine phentermine users in our sample had lost 10% of their body
or dexfenfluramine users (26.0%). None of the sociodemo- weight and about 40% were 10% below their prepill body
graphic variables (sex, race, age, income, education) were weight at the time of our survey. That about 10% of
significantly related to having had an echocardiogram nor phentermine users received their pills from either a friend
were duration of use, source, or prepill BMI (data not or family member or other nonphysician source is of
shown); however, users who reported not having health potential concern. Future research is needed to ascertain
insurance at any time in the past 12 months were signifi- nonphysician sources and also use of the Internet in obtain-
cantly less likely to have an echocardiogram (6.3%) than ing prescription weight loss pills.
those who had not gone without insurance (26.0%), P = 0.02. There are a number of limitations to our survey. Some
analyses had small sample sizes, particularly for specific pill
type, and standard errors are provided to help readers
Discussion interpret our results. As use in the previous 2 years was
asked about there is potential for recall bias. That one in
In this population-based study of adults in six states who seven users did not know the name of the prescription
took prescription weight loss pills between 1996 and 1998, weight loss pill they had taken adds a small but significant
we found that among those pill users who were taking either degree of uncertainty to our results. Because we do not
fenfluramine or dexfenfluramine, one third stopped pill know when the pills were started, we are not able to
usage immediately after the public health advisory and attribute weight loss solely to prescription drugs; it is
another third stopped during the time of the market with- possible that other behaviors impacted weight loss. We were
drawal (September and October 1997). That a remaining not able to verify the actual product(s) taken, weight lost, or
one third reported continued use for a substantial time after the actual height and weight of participants. Research has
the market withdrawal is of concern especially in light of the shown that overweight persons tend to underestimate their
extensive media attention surrounding the potential cardiac weight [20,21]. BRFSS data are cross sectional in nature not
dangers. We speculate that those struggling with obesity allowing for causal relationships; and because BRFSS is a
1248 H.M. Blanck et al. / Preventive Medicine 39 (2004) 1243–1248

telephone survey, coverage bias is a possibility. However, [3] Weintraub M, Sundaresan PR, Schuster B, et al. Long-term weight
control study II (weeks 34 to 104). An open-label study of contin-
one study that examined common health characteristics of
uous fenfluramine plus phentermine versus targeted intermittent med-
persons with and without telephones found limited bias in ication as adjuncts to behavior modification, caloric restriction, and
populations with high telephone coverage however [22]. exercise. Clin Pharmacol Ther 1992;51:595 – 601.
We found that over one third of prescription pill users [4] Weintraub M, Sundaresan PR, Schuster B, Moscucci M, Stein EC.
had also taken nonprescription weight loss products in the Long-term weight control study. III (weeks 104 to 156). An open-
past 2 years. Use of nonprescription weight loss products label study of dose adjustment of fenfluramine and phentermine. Clin
Pharmacol Ther 1992;51:602 – 7.
was common among women. Our survey did not collect [5] IMS America. National Prescription Audit. Plymouth meeting, PA:
information as to whether the products were taken simul- IMS America, 1997.
taneously or whether users told their physicians about their [6] Connolly HM, Crary JL, McGoon MD, et al. Valvular heart disease
use. Physicians need to know whether nonprescription associated with fenfluramine – phentermine. N Engl J Med
weight loss products (including herbal products) are being 1997;337:581 – 8.
[7] Lumpkin MM. FDA health advisory. FDA Bull 1997;27:2.
used when they prescribe a prescription drug because of [8] Food and Drug Administration. FDA Public Health Advisory. Reports
the possibility for herb –drug interactions. In addition, they of Valvular Heart Disease in Patients Receiving Concomitant Fenflur-
need to notify the FDA about any adverse events in amine and Phentermine. July 8, 1997. Available at: http://www.fda.
patients taking both prescription and nonprescription diet gov/cder/news/phen/phenfen.htm. Accessed March 3, 2003.
products to assist in monitoring the safety of available [9] Valvulopathy associated with exposure to fenfluramine or dexfenflur-
amine: U.S. Department of Health and Human Services interim public
products [23]. health recommendations, November 1997. Morb Mortal Wkly Rep
As noted earlier, national guidelines state that pharma- 1997;46:1061 – 6.
cotherapy may be considered for persons who are clinically [10] Dajani AS, Taubert KA, Wilson W, et al. Prevention of bacterial
obese (BMI z 30) or have a BMI of 27 –29.9 plus an endocarditis: recommendations by the American Heart Association.
obesity-related comorbid condition [13]. Per our survey, JAMA 1997;277:1794 – 801.
[11] Mokdad AH, Stroup DF, Giles WH. Public health surveillance for
only a small percentage (4%) of persons who used phen- behavioral risk factors in a changing environment: recommendations
termine alone had a prepill BMI < 27. However, one quarter from the behavioral risk factor surveillance team. MMWR
of phentermine users had BMIs between 27 and 29.9. Usage 2003;52:1 – 9.
by this group could be due to having risk factors for obesity- [12] Blanck HM, Khan LK, Serdula MK. Use of nonprescription weight
loss products: results from a multistate survey. JAMA
related diseases, which we were unable to assess; however,
2001;286:930 – 5.
it could also be due to the demand of patients without [13] Clinical guidelines on the identification, evaluation, and treatment of
obesity-related diseases who want to try obesity medications overweight and obesity in adults. National Heart, Lung, and Blood
to help in their battle with weight. Institute (NHLBI), Clinical guidelines for obesity, 1998.
Our study is the first population-based survey to document [14] U.S. Department of Health and Human Services. Third National
the public’s response to the fen-phen withdrawal. Despite the Health and Nutrition Examination Survey, 1988 – 1994, NHANES
III Examination Data File. Public use data file documentation number
enormous publicity about the risk of cardiac valvulopathy, 76200. Hyattsville, MD: Centers for Disease Control and Prevention,
one third of persons continued to use fenfluramine or dexfen- 1996.
fluramine after market withdrawal. Most persons, especially [15] SAS Institute MK. SAS User’s Guide: Statistics, Version 8. Cary, NC:
those without health insurance, did not receive the follow-up SAS Institute, 1999.
echocardiogram. Discontinuation of nonprescription obesity [16] Shah BV, Barnwell BG, Bieler GS. SUDAAN Users’ Manual, Re-
lease 7.5. Research Triangle Park, NC: Research Triangle Institute,
products, such as ephedra-based products and other herbs 1997.
shown to be unsafe, could prove to be more precarious [17] Stafford RS, Radley DC. National trends in antiobesity medication
because of the lack of oversight by health care professionals. use. Arch Intern Med 2003;163:1046 – 50.
[18] Haddock CK, Poston WS, Dill PL, Foreyt JP, Ericsson M. Pharma-
cotherapy for obesity: a quantitative analysis of four decades of pub-
Acknowledgments lished randomized clinical trials. Int J Obes Relat Metab Disord
2002;26:262 – 73.
[19] PDR Electronic Libraryk Online: n 2002 – 2004. Thomson PDR.
We would like to thank the BRFSS State Coordinators
MICROMEDEX. Medical Economics, 2002.
and staff from Arizona, Florida, Iowa, Michigan, West [20] Jeffery RW, French SA. Socioeconomic status and weight control
Virginia, and Wisconsin. practices among 20- to 45-year-old women. Am J Public Health
1996;86:1005 – 10.
[21] Cairney J, Wade TJ. Correlates of body weight in the 1994 National
References Population Health Survey. Int J Obes Relat Metab Disord
1998;22:584 – 91.
[1] Serdula MK, Mokdad AH, Williamson DF, Galuska DA, Mendlein [22] Ford ES. Characteristics of survey participants with and without a
JM, Heath GW. Prevalence of attempting weight loss and strategies telephone: findings from the third National Health and Nutrition
for controlling weight. JAMA 1999;282:1353 – 8. Examination Survey. J Clin Epidemiol 1998;51:55 – 60.
[2] Khan LK, Serdula MK, Bowman BA, Williamson DF. Use of pre- [23] Peng CC, Glassman PA, Trilli LE, Hayes-Hunter J, Good CB. Inci-
scription weight loss pills among U.S. adults in 1996 – 1998. Ann dence and severity of potential drug-dietary supplement interactions
Intern Med 2001;134:282 – 6. in primary care patients. Arch Intern Med 2004;164:630 – 6.