Caffeine Dependence Syndrome

Evidence From Case Histories and Experimental Evaluations

Eric C. Strain, MD; Geoffrey K. Mumford, PhD; Kenneth Silverman, PhD; Roland R. Griffiths, PhD

Objective.\p=m-\Theextent to which daily caffeine use is associated with a drinking coffee, tea, maté, soft drinks;
substance dependence syndrome similar to that associated with other psychoac- chewing kola nuts; consuming cocoa and
tive drugs is unknown. The purpose of this study was to assess volunteers who re- guaranà products) and in widely differ¬
ported problems with their use of caffeine for evidence suggesting a diagnosis of ent, but culturally well-integrated, so¬
cial contexts (eg, the coffee break in the
caffeine dependence based on the generic criteria for substance dependence from United States, teatime in the United
the Diagnostic and Statistical Manual ofMental Disorders, Fourth Edition (DSM-IV).
Kingdom, kola nut chewing in Nigeria).
Design.\p=m-\Case-seriesevaluations. The wide generality of caffeine consump¬
Setting.\p=m-\Anacademic research center. tion is also reflected in the high preva¬
Participants.\p=m-\Self-identifiedadults who believed they were psychologically or lence of its use in the United States,
physically dependent on caffeine. where more than 80% of adults regu¬
Main Outcome Measure.\p=m-\Diagnosesmade by a psychiatrist using a structured larly consume behaviorally active doses
clinical interview that included a section on caffeine dependence based on generic of caffeine3·4 and the average daily con¬
criteria for DSM-IV substance dependence. sumption of caffeine is estimated to be
280 mg per adult consumer.5
Secondary Outcome Measure.\p=m-\Double-blindcaffeine-withdrawal evaluation.
Results.\p=m-\Ninety-ninesubjects were screened for the study, and 16 were iden-
tified as having a diagnosis of caffeine dependence. Median daily caffeine intake For editorial comment see 1065.
was 357 mg, and 19% of subjects consumed less than the national (US) daily av-
erage of caffeine. Criteria used for making diagnoses (and rates of their prevalence) Caffeine tends to produce a pattern of
were as follows: withdrawal (94%), use continued despite knowledge of a persis- subjective effects that varies as a func¬
tent or recurrent physical or psychological problem that is likely to have been caused tion of dose. Although low doses, in the
or exacerbated by caffeine use (94%), persistent desire or unsuccessful efforts to range of 20 to 200 mg, generally produce
cut down or control use (81%), and tolerance (75%). Eleven subjects underwent mild positive subjective effects (eg, in¬
the double-blind caffeine-withdrawal evaluation portion of the study, and nine (82%) creased feelings of well-being, alertness,
of the 11 showed objective evidence of caffeine withdrawal, including eight of 11 energy),6·7 higher doses, in the range of
with functional impairment. 200 to 800 mg, can produce negative
effects (eg, nervousness, anxiety), es¬
Conclusions.\p=m-\Theseresults, together with other experimental evidence, sug-
pecially in volunteers who are usually
gest that caffeine exhibits the features of a typical psychoactive substance of de- caffeine abstinent.8'10 Consistent with the
pendence. It is valuable to recognize caffeine dependence as a clinical syndrome, mild positive subjective effects of caf¬
since some people feel compelled to continue caffeine use despite desires and feine observed at low doses, human stud¬
recommendations to the contrary. ies have also shown that caffeine can
(JAMA. 1994;272:1043-1048) function as a reinforcer (ie, it maintains
self-administration or is preferentially
chosen over placebo), and studies in ani¬
From the Department of Psychiatry and Behavioral CAFFEINE has been consumed by hu¬ mals have also demonstrated that caf¬
Sciences, The Johns Hopkins University School of mans for hundreds if not thousands of feine can function as a reinforcer under
Medicine, Baltimore, Md.
Presented at the American College of Neuropsycho- years1 and is currently the most widely certain experimental conditions.11 Stud¬
pharmacology, Honolulu, Hawaii, December 15,1993. used psychoactive substance in the ies examining the relationship between
Reprint requests to Department of Psychiatry, Be- world.2 Throughout the world, the pre¬ caffeine and various illnesses (eg, car¬
havioral Pharmacology Research Unit, The Johns Hop-
kins University School of Medicine, 5510 Nathan Shock ferred mode for consuming caffeine oc¬ diovascular disease, cancer, increased
Dr, Baltimore, MD 21224 (Dr Griffiths). curs in markedly different forms (eg, cholesterol concentration, low birth

Downloaded From: http://jama.jamanetwork.com/ by a New York University User on 06/05/2015

 weight) generally have failed to find evi¬ paper notices that sought study volun¬ DSM-III-R (SCID),18 with a modified
dence suggesting that typical daily doses teers who believed they were psycho¬ -module (that section on psychoactive
of caffeine are etiologically related to logically or physically dependent on caf¬ substance use disorders) that followed
these conditions or have yielded ambigu¬ feine (contained in coffee, tea, soda, or the format of the original SCID E-mod-
ous and contradictory results.12 Thus, tablets). Subjects were included in the ule and included questions regarding caf¬
the wide use and cultural acceptance of study if they were 18 to 50 years old; feine dependence. Interviews were con¬
caffeine can be understood in the con¬ had at least a high school diploma or ducted by the same psychiatrist (E.C.S.),
text of this combination of positive sub¬ equivalent; had a normal blood pressure, and the DSM-III-R criteria were coded
jective and reinforcing effects with rela¬ heart rate, and electrocardiogram to allow all diagnoses to be made using
tively few adverse effects. (ECG); had no physical condition con- DSM-IV. Three of seven criteria must
While caffeine is consumed by a large traindicatingthe consumption of caffeine be present for a DSM-IV diagnosis of
segment of the population, it is not (eg, palpitations, arrhythmias); had not substance dependence. While all diag¬
known whether some consumers have a used illicit drugs in the past 6 months; nostic criteria in the modified E-module
were not pregnant; consumed caffeine were probed, only four criteria were
pattern of caffeine use that would qualify
them for a diagnosis of abuse or depen¬ on a daily basis; and reported problems considered when making a diagnosis
dence as defined by the American Psy¬ associated with their caffeine use, based of dependence, and participants were
chiatric Association's Diagnostic and on screening questions derived from the required to fulfill three of these four
Statistical Manual ofMental Disorders, DSM-III-R {Diagnostic and Statistical criteria to qualify for a diagnosis of caf¬
Fourth Edition (DSM-IV).™M These di¬ Manual of Mental Disorders, Revised feine dependence. These four criteria
agnoses in DSM-IV are based on a set Third Edition") diagnosis of psychoac¬ were chosen to represent clinically mean¬
of generic criteria, with substance de¬ tive substance dependence (criteria as ingful aspects of pathological use of a
pendence a more severe disorder than described in the "Diagnostic Interview" substance that is widely available and
abuse. Although physical dependence, section below). The study was approved culturally accepted. The four DSM-IV
as evidenced by a withdrawal syndrome, by the local institutional review board. criteria were tolerance (criterion 1);
is sometimes erroneously equated with withdrawal (criterion 2); persistent de¬
the diagnosis of dependence, in fact, Study Procedures sire or unsuccessful efforts to cut down
withdrawal is only one component of the Subjects were initially screened for or control use (criterion 4); and use con¬
DSM-IV diagnosis of dependence. While suitability by telephone, using a ques¬ tinued despite knowledge of a persis¬
evidence suggests that abruptly stop¬ tionnaire that reviewed their medical tent or recurrent physical or psycho¬
ping caffeine consumption sometimes and psychiatric history, including use of logical problem that is likely to have
produces a distinct clinical syndrome alcohol, nicotine, and caffeine, as well been caused or exacerbated by substance
characterized by headache, lethargy, and as both prescription and illicit drugs. use (criterion 7). The remaining three

depression,11·15,16 the presence of a with¬ Screening questionnaires were reviewed criteria were excluded because of con¬
drawal syndrome is only one of the cri¬ by one of the investigators (E.C.S), and cern that these criteria would trivialize
teria used for a diagnosis of substance eligible participants were then requested the diagnosis or did not apply to a sub¬
dependence, and it is neither necessary to come to the research unit for further stance widely available and culturally
nor sufficient for making the diagnosis. evaluation. A total of 99 applicants were accepted. The three excluded criteria
The primary purpose of the current screened by telephone. Subjects were were substance often taken in larger

study was to assess volunteers self-iden¬ told they were participating in a study amounts or over a longer period than
tified as being caffeine dependent—that evaluating the effects of food compo¬ intended (criterion 3); a great deal of
is, reporting problems associated with nents on mood and behavior. There were time spent in activities necessary to ob¬
their use of caffeine—for evidence sug¬ two phases to the study, and 27 ofthe 99 tain, use, or recover from the effects of
gesting a diagnosis of caffeine depen¬ applicants were eligible and willing to the substance (criterion 5); and impor¬
dence based on the criteria from DSM- participate in the first phase, during tant social, occupational, or recreational
IV as applied by a psychiatrist employ¬ which subjects reported to the labora¬ activities given up or reduced because
ing a standardized structured interview. tory on two or more occasions, signed of substance use (criterion 6).
A secondary purpose of the study was consent, completed a history question¬ Double-Blind Caffeine-Withdrawal
to subsequently evaluate these volun¬ naire, underwent a physical examina¬
teers for evidence ofcaffeine withdrawal. tion including an ECG and screening Evaluation
This withdrawal assessment provided a blood tests, and completed 1 week of In the second phase of the study, par¬
means for objectively testing one of the food diaries. They also underwent a stan¬ ticipants were required to complete a
common, but not necessary, features of dardized psychiatric interview (de¬ battery of assessments on three occa¬
dependence. The identification of se¬ scribed below) that included an assess¬ sions, once as a set of practice tests while
lected volunteers with problematic caf¬ ment of caffeine dependence based on following their normal eating patterns,
feine use consistent with a DSM-IV di¬ DSM-IV criteria for substance depen¬ and then again at the end of each of two
agnosis of substance dependence would dence. Sixteen of the 27 were diagnosed 2-day study periods. Subjects adhered
provide valuable clinical support for the as caffeine dependent. One of these 16 to a caffeine-free diet during these 2-day
establishment of a distinct syndrome of was medically disqualified from further periods, which were generally separated
caffeine dependence, and would provide participation because of newly diagnosed by 1 week and occurred on the same
an opportunity to assess the clinical fea¬ hypertension. Eleven of the remaining weekdays.
tures of people with a substance depen¬ 15 were willing to participate in the sec¬ To achieve a blind caffeine-free diet,
dence syndrome for caffeine. ond phase of the study, a double-blind participants were instructed, both ver¬
caffeine-withdrawal evaluation that is bally and in writing, to maintain certain
METHODS
described below. dietary restrictions during the 2-day
study periods and were further in¬
Subjects Diagnostic Interview structed that the purpose of these re¬
Participants were a self-identified Subjects were interviewed utilizing strictions was to examine the effects on
group of adults recruited through news- the Structured Clinical Interview for mood and behavior of compounds nor-

Downloaded From: http://jama.jamanetwork.com/ by a New York University User on 06/05/2015

 Table 1.—Participants With Caffeine Dependence (n=16)*
Other Psychiatric
Diagnoses^ Caffeine Dependence
Race/ Marital Currently Caffeine Intake, Primary (DSM-IV Criterion
Subject Gender Age, y Status Smokingf Past Present mg/d§ Vehicle§ Numbers)!
502 WM 33 Yes None 2548 Coffee 1,2,4,7
506 WF 43 No 2,3 231 Soft drink 1,2,4,7
509 WF Yes 1,2 None 642 Coffee 1,2,4,7
WF 48 Yes 1,2,3,4 None 1038 Coffee 2,4,7
525 32 No 1 Coffee 1,4,7H
WM No 1,2,3 None 302 Soft drink 1,2,4,7
532 WF 33 No None None 430 Soft drink 2,4,7
535 WF 36 No None None 342 Soft drink 1,2,4,7H
542 BF Yes 1,3 None 589 Coffee 1,2,7
543 WF Yes None 295 Coffee 1,2,7
544 WF 31 No 1,2,4 371 Soft drink 1,2,4
545 WF No None None 320 Tea 1,2,7H
548 WF 42 No None 270 Soft drink 2, 4, 7H
549 WF 31 No 1,2 None 129 Soft drink 1,2,4,7H
550 WF No None None 516 Coffee 2, 4, 7H
551 BF 42 No None None 300 Coffee 1,2,4,7H
*
DSM-IV indicates Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; D, divorced; S, single; and M, married.
tNicotine dependence is not included in the Structured Clinical Interview for DSM-III-R (SCID)." Current smoking status was obtained during the medical history.
^Diagnoses in remission (past) or current (present) based on the SCID Interview: 1 indicates substance use disorder; 2, mood disorder; 3, anxiety disorder; and 4, eating
disorder.
§Based on the 7-day food diary. The primary vehicle, which was determined from the food diaries, was defined as the substance that accounted for the majority of that subject's
caffeine intake.
| All subjects qualified for a provisional diagnosis of caffeine dependence based on assessment of four generic DSM-IV criteria for substance use disorder'3: 1 indicates
tolerance; 2, withdrawal; 4, persistent desire or unsuccessful efforts to cut down or control use; and 7, use continued despite knowledge of a persistent or recurrent physical
or psychological problem that Is likely to have been caused or exacerbated by substance use.
1jlndicat.es subjects who were advised by their physician to reduce or eliminate caffeine consumption but who failed to do so.

mally found in the foods and beverages der double-blind conditions. On each day garding the use of any medications dur¬
of their daily diet. These restrictions of the placebo and caffeine periods, cap¬ ing the time of the dietary restrictions.
were given without reference to caffeine. sule administration times were spaced Participants completed the BDI, POMS,
The only beverages allowed were milk, throughout the day to match the pat¬ and Study Questionnaire based on how
fruit juices, and water; chocolate prod¬ tern ofthe individual's reported caffeine they had felt that day and during the
ucts were prohibited. To divert atten¬ consumption. The maximum amount of previous day. After completing the ques¬
tion from caffeine, food items without caffeine in a single capsule was 200 mg, tionnaires, the subjects completed a tap¬
caffeine were also restricted, including and the maximum dose of caffeine ad¬ ping task in which they were instructed
shellfish and all foods containing sac¬ ministered at one time was 400 mg (two to press a button 200 times as fast as
charin or aspartame (Nutrasweet). Fi¬ capsules). Subjects typically ingested they could. Three consecutive tapping
nally, because subjects having with¬ capsules at three administration times trials, separated from one another by
drawal symptoms might be tempted to during the day; as many as seven ad¬ approximately 10 seconds, were con¬
take analgesic drugs, subjects were told ministration times were used to accom¬ ducted. This task has been shown to be
not to take any medications without first modate the dosing of subjects using very sensitive to the effects of caffeine with¬
contacting one of the investigators. high doses of caffeine. Subjects came to drawal.15 Subjects were then inter¬
During the two 2-day study periods, the laboratory for the first administra¬ viewed by an investigator blind to the
subjects received capsules containing, in tion each day, which was done under order of the study conditions, who re¬
random order, either caffeine in an observation, and then were given pack¬ viewed the subjects' experiences dur¬
amount equal to their individual average ets of capsules with instructions regard¬ ing the study period, including any evi¬
daily caffeine consumption or placebo. ing the timing of subsequent adminis¬ dence of functional impairment.
The average daily caffeine consumption trations. In addition, participants were
was calculated from each subject's food given emergency contact cards with in¬ Analysis of Salivary Caffeine
diaries, using standard caffeine amounts vestigators' telephone numbers, in case Five-milliliter samples of saliva were
contained in the specific food items con¬ questions or problems arose during the collected at each laboratory visit during
sumed.8 Subjects were told that their study. the second phase of the study to assess
capsules would contain placebo or one of Assessments, which were adminis¬ compliance with the dietary restrictions.
several compounds (chlorogenic acids, di- tered on the second day of each 2-day Salivary caffeine concentrations were
terpines, caffeine, tannin, sugar, or the- study period, included the Beck Depres¬ measured as previously described.7,22 No
ophylline) commonly found in foods and sion Inventory (BDI), a 21-item ques¬ subjects showed evidence of violation of
beverages. Assessments occurred on the tionnaire designed to assess depressive the dietary restrictions during the 2-day
second day of each of the 2-day study symptoms19·20; the Profile of Mood States placebo dosing study period.
periods and occurred a minimum of 3 (POMS), a 65-item questionnaire de¬
hours after the administration of the last signed to assess mood states21; and the RESULTS
capsule (usually between 4 PM and 6 PM Study Questionnaire, a 33-item check¬ Results From the
on the second day). list that assessed symptoms related to
Placebo (powdered lactose) and caf¬ caffeine withdrawal (eg, headache, Diagnostic Interview
feine (anhydrous) were administered in drowsy/sleepy).15 The end of the Study After telephone screening, 27 subj ects
opaque, hard, size 0 gelatin capsules un- Questionnaire included a question re- participated in the first phase of the

Downloaded From: http://jama.jamanetwork.com/ by a New York University User on 06/05/2015

 Table 2.—Double-Blind Caffeine-Withdrawal Evaluation Results (n=11)*
Fatigue Vigor Depression Tapping Analgesic
Subject Headache (POMS) (POMS) (BDI) (Mean) Use Functional Impairment
502 0/2 5/28 23/1 5/1 ß 281/201 No Moderate (screaming at his children)
506 1/3 16/20 25/12 5/23 284/262 Yes Severe (missed work; emesis)
2/2 9/5 9/11 4/5 300/283 No None
525 1/3 3/23 13/2 3/7 421/290 No Severe (multiple costly mistakes at work; left work
early; went to bed early)
531 1/3 0/4 17/10 8/14 240/259 Yes Mild (unable to complete schoolwork)
532 0/3 1/27 30/2 5/24 294/291 Yes Severe (canceled son's birthday party; called
spouse home early because of inability to care
for children)
535 1/3 5/13 15/2 0/6 377/354 Yes Severe (could not perform work responsibilities, ie,
sat in office awake with lights off and head
down; went to bed 2 hours early; needed spouse
to care for children)
543 0/M 3/2 19/21 0/3 345/346 No None
544 0/3 0/28 28/5 2/15 303/283 No Severe (data-entry errors at work; went to bed 4.5
hours early; unable to do recreational reading)
548 1/3 7/17 13/2 11/12 340/391 Yes Severe (stopped doing errands; spent time
napping; failed to do household chores, ¡e,
making child's lunch, preparing for child's school
activities; did not exercise)
550 1/0 25/7 4/21 1/0 408/411 No None

*Withdrawal test data are presented as caffeine score/placebo score.

Î Functional impairment was defined as a disruption of usual work or social behavior. Descriptions refer to functional Impairment during the placebo study periods. One subject
(506) also reported severe functional Impairment during the caffeine study period (stayed home from work, broke two glasses).
Designates scores showing significant caffeine-withdrawal symptoms during the placebo trial: headache (rating of 3 on a scale ranging from 0 [not at all] to 3 [severe] from
the Study Questionnaire15); fatigue and vigor (2 SD above and below, respectively, the norm for college students on the Profile of Mood States [POMS] questionnaire21); depression
(15 or above on the Beck Depression Inventory [BDI]20); tapping (no overlap between three caffeine trials and three placebo trials); and functional impairment (based on the
subject's verbal description of changes in work and behaviors). M indicates missing data.

study, which included food diaries and a (either abuse or dependence—10 sub¬ amination; four of the subjects were not
psychiatric interview. Sixteen of these jects [63%]), followed by mood disor¬ willing to participate in the second phase
participants fulfilled criteria for a cur¬ ders (seven subjects [44%]), anxiety dis¬ of the study. Results for the 11 subjects
rent diagnosis of caffeine dependence as orders (four subjects [25%]), or eating who participated in this phase of the
determined by the SCID (Table 1). The disorders (three subjects [19%]). The study are presented in Table 2. Nine
16 subjects had a mean age of 38 years most common class of substance use dis¬ (82%) of the subjects showed evidence
and a mean of 16 years of education; 14 orders in remission was alcohol; nine of caffeine withdrawal during the pla¬
(88%) were women, and 12 (75%) were subjects (57%) had a diagnosis of alcohol cebo period. Seven (64%) of the parti¬
employed. Their median daily consump¬ abuse or dependence. The mean length cipants reported maximal ratings of
tion of caffeine was 357 mg (range, 129 of time in remission per alcohol diagno¬ headache (from the Study Questionnaire)
to 2548 mg), and they primarily con¬ sis was 9.8 years, and the mean length during the days on which they received
sumed either coffee or soft drinks (50% of time in remission for all substance use placebo, and seven (64%) showed sig¬
and 44% of subjects, respectively). Sub¬ disorders was 7.9 years. Besides caf¬ nificant elevations in ratings of fatigue
jects fulfilled a mean of 3.4 of the four feine, according to the SCID none of the or depression, or decreases in ratings of
criteria for caffeine dependence: 12 (75%) subjects fulfilled diagnostic criteria for vigor (from the BDI and POMS). Five
met criterion 1 (tolerance), 15 (94%) met any substance use disorder in the year subjects (45%) used an analgesic (eg,
criterion 2 (withdrawal), 13 (81%) met prior to study participation. However, acetaminophen), although they had been
criterion 4 (persistent desire or unsuc¬ subjects were not assessed for a diag¬ discouraged from doing so. In the in¬
cessful efforts to cut down or control nosis of nicotine dependence, since it is terview following the caffeine and pla¬
use), and 15 (94%) met criterion 7 (use not included in the SCID; five subjects cebo periods, eight (73%) of the subjects
continued despite knowledge of a per¬ (502, 509, 517, 542, 543) were currently reported functional impairment in nor¬
sistent or recurrent physical or psycho¬ daily cigarette smokers as determined mal daily activities during the placebo
logical problem that is likely to have from the medical history (Table 1). These (caffeine-withdrawal) period. In con¬
been caused or exacerbated by substance five smokers tended to have a higher trast, only one subject reported func¬
use). Seven of the subjects fulfilling cri¬ mean daily caffeine consumption than tional impairment during the caffeine
terion 7 reported a history of physical the 11 subjects who did not smoke (1022 period.
conditions such as acne rosacea, preg¬ mg vs 385 mg, respectively).
nancy, palpitations, and gastrointesti¬
COMMENT
nal problems that had led physicians to Results From the Double-Blind This study identified the character¬
recommend that they reduce or elimi¬ Caffeine-Withdrawal Evaluation istics of caffeine use in a population of
nate caffeine consumption; all seven had Fifteen of the 16 subjects given a di¬ volunteers self-identified as having prob¬
failed to comply with the physicians' rec¬ agnosis of caffeine dependence were eli¬ lems with caffeine, using a structured
ommendations. gible to participate in the second phase interview and DSM-IV criteria for a di¬
Only two subjects (13%) had an ad¬ of the study (the double-blind caffeine- agnosis of substance dependence, and
ditional current psychiatric diagnosis be¬ withdrawal evaluation, which was iden¬ found 16 volunteers with a diagnosis of
sides caffeine dependence (Table 1), and tified to subjects as an assessment of caffeine dependence. Since evidence of
both had diagnoses of anxiety disorders. compounds commonly found in foods and withdrawal is one of the criteria for a
Eleven (69%) of the subjects had a psy¬ beverages). One subject was ineligible diagnosis of dependence (although it is
chiatric diagnosis in remission, most com¬ to participate because of hypertension not necessary for the diagnosis), these
monly another substance use disorder newly diagnosed during the physical ex- volunteers were then challenged with a

Downloaded From: http://jama.jamanetwork.com/ by a New York University User on 06/05/2015

 double-blind caffeine-withdrawal evalu¬ The primary purpose of this study remission for these mood disorders was
ation as a means for objectively testing was to determine if there were caffeine 5.7 years. This rate of mood disorders
one aspect of this diagnosis; nine (82%) consumers who fulfilled the criteria for (44%) is higher than expected for the gen¬
of the subjects who participated in the a diagnosis of caffeine dependence as eral population, as determined by the Na¬
challenge phase showed evidence of caf¬ determined by a standardized psychi¬ tional Comorbidity Survey.26 The higher-
feine withdrawal. These results suggest atric interview. The inclusion of the caf¬ than-expected rate of mood disorders
that caffeine can produce a clinical de¬ feine-withdrawal evaluation in this study found in these subjects with a diagnosis
pendence syndrome similar to those pro¬ was an attempt to objectively test one of caffeine dependence is similar to ear¬
duced by other psychoactive substances. criterion used in the diagnosis of depen¬ lier findings of an association between
Participants in this study reported a dence, but evidence of caffeine with¬ one mood disorder (major depression) and
wide range in daily caffeine consump¬ drawal is neither necessary nor suffi¬ the diagnosis of nicotine dependence.27"29
tion, from 129 to 2548 mg per day. The cient to make a diagnosis of caffeine de¬ The wide use of caffeine, its cultural
diagnosis of caffeine dependence was not pendence. The presence of a withdrawal acceptance, and the absence of signifi¬
simply related to a high daily dose of syndrome suggests that a patient is cant medical problems associated with
caffeine; three subjects with a diagnosis physically dependent on a substance, but its use12 may lead to questions regarding
of caffeine dependence had a daily con¬ the presence of physical dependence (ie, the need for advancing a formal diagno¬
sumption less than the average daily a withdrawal syndrome) does not mean sis of caffeine dependence analogous to
consumption of caffeine in the United that the person fulfills diagnostic crite¬ those for other drugs with more clear
States (280 mg per adult consumer5). ria for a dependence syndrome. For ex¬ morbidity (such as alcohol, nicotine, co¬
However, this study did not assess caf¬ ample, it is possible for a pattern of sub¬ caine, and opioids). Establishing the di¬
feine blood levels, and it is known that stance use to qualify for a diagnosis of agnosis of caffeine dependence is not
caffeine elimination (and thus actual caf¬ substance dependence without evidence meant to detract from the general con¬
feine exposure) can vary widely among of physical dependence—ie, a with¬ cept of substance dependence, but is
individuals and can be influenced by fac¬ drawal syndrome (eg, hallucinogens, meant to demonstrate the common fea¬
tors such as cigarette smoking, preg¬ short-term binge alcohol use); it is also tures of substance dependence across
nancy, and liver disease.23 Future stud¬ possible for substance use to produce a psychoactive substance classes, and also
ies should determine the relationship withdrawal syndrome without fulfilling to serve the clinically useful purpose of
between a diagnosis of caffeine depen¬ criteria for a diagnosis of substance de¬ identifying people previously unrecog¬
dence and actual caffeine exposure. pendence (eg, chronic opioid use in the nized as having problematic caffeine use.
Over 80% of the subjects fulfilled cri¬ treatment of pain). We have previously The volunteers for this study reported a
teria 2 (withdrawal), 4 (persistent de¬ shown that caffeine withdrawal can oc¬ variety of problems associated with their
sire or unsuccessful efforts to cut down cur in consumers of typical daily doses caffeine use, including arguments with
or control use), or 7 (use continued de¬ of caffeine,15 although the relationship family members and friends over their
spite knowledge of a persistent or re¬ between caffeine withdrawal and a di¬ use, going to extremes to obtain caffeine-
current physical or psychological prob¬ agnosis of caffeine dependence was not containing products, using them in po¬
lem that is likely to have been caused or determined in that study. The present tentially dangerous situations, and con¬
exacerbated by substance use). In ad¬ results suggest that caffeine withdrawal tinuing to use them despite being told
dition, 75% met criterion 1 (tolerance). is common in volunteers with a diagno¬ not to by physicians. Several subjects in
This profile of criteria demonstrates that sis of caffeine dependence. the study were interested in learning
the diagnosis of caffeine dependence was While there were few concurrent psy¬ about how to stop using caffeinated prod¬
not simply the result of participants' chiatric disorders in this population, ucts, since they had been unsuccessful in
awareness of being physically depen¬ there were high rates of past psychiat¬ doing so on their own. Thus, it is valuable
dent on or tolerant to caffeine (that is, ric disorders (Table 1). The most com¬ to recognize caffeine dependence as a
endorsing criteria 1 and 2). The high mon psychiatric disorders in remission distinct clinical syndrome because there
prevalence of criteria 4 and 7 suggests were other substance use disorders (10 are people who feel compelled to con¬
that the use of caffeine, like use of other subjects [63%], excluding nicotine de¬ tinue to use caffeine, despite a strong
psychoactive substances, can be diffi¬ pendence), with the most prevalent drug desire to the contrary.
cult to stop for some people. class being alcohol. Nine subjects (57%) This study did not attempt to deter¬
Interestingly, the two subjects (543 had a past diagnosis of alcohol abuse or mine the prevalence of a diagnosis of
and 550) who did not show evidence of dependence. In addition, five subjects caffeine dependence. In a survey that
caffeine withdrawal during the experi¬ smoked tobacco cigarettes daily, and four used DSM-IH-R criteria modified to in¬
mental withdrawal phase of the study of the five had a past diagnosis of alcohol clude a diagnosis of caffeine dependence,
reported a history of having symptoms abuse or dependence. (Subject 509 had Hughes et al4 reported that 17% of 166
of a caffeine-withdrawal syndrome dur¬ a past diagnosis of stimulant depen¬ respondents fulfilled criteria for mod¬
ing the SCID interview. Analysis of sa¬ dence.) This tendency for caffeine, al¬ erate or severe caffeine dependence in
liva samples showed both subjects were cohol, and nicotine disorders to cluster the past year. Notably, the study by
compliant with the caffeine dietary re¬ has been previously reported.26 While Hughes et al was a telephone survey,
strictions. The absence of caffeine with¬ these five subjects from the current restricted to a relatively small epide¬
drawal during the placebo period in these study are a limited sample, the finding miologie sample in Vermont, and used
two subjects is consistent with results that almost all these smokers with a all nine criteria from DSM-IH-R. The
from a previous study that showed, within diagnosis of caffeine dependence had a current study was not an attempt to
an individual, a single episode of experi¬ history of alcohol abuse or dependence determine the prevalence of a diagnosis
mental caffeine cessation may underes¬ is an intriguing observation that should of caffeine dependence, it employed a
timate that subject's vulnerability to be further characterized. face-to-face standardized psychiatric in¬
showing withdrawal, since there is con¬ Seven subjects had a past diagnosis of terview, and it used only four of the
siderable within-subject variability in the a mood disorder, either bipolar disorder seven DSM-IV criteria. The more re¬
withdrawal effects produced across re¬ (one subject) or major depressive disor¬ strictive diagnostic approach used in this
peated episodes of caffeine cessation.24 der (six subjects), and the mean time in study, and the recruitment of partici-

Downloaded From: http://jama.jamanetwork.com/ by a New York University User on 06/05/2015

 pants self-identified as having problems studies have shown that subjects can be behavioral and physiological factors that
with caffeine, may have resulted in the intoxicated with the excessive use of may potentiate the development of caf¬
set of extreme cases of caffeine depen¬ caffeine and that caffeine can produce a feine dependence. In addition to pro¬
dence reported herein. It would be valu¬ withdrawal syndrome when subjects viding valuable data about the descrip¬
able to determine the prevalence of a stop habitual use.30 The results of this tive features and clinical importance of
diagnosis of caffeine dependence in a study provide evidence that subjects also caffeine dependence, further character¬
large sample of the general population, can become clinically dependent on caf¬ ization of the dependence syndrome of
especially with concurrent assessments feine. The recognition of syndromes of the most widely used psychoactive drug
for other psychiatric disorders, such as intoxication, withdrawal, and depen¬ in the world may also serve as a useful
other substance dependence disorders. dence suggests that caffeine is like other model for understanding the dependence
CONCLUSION psychoactive drugs. The identification syndromes of other drugs.
of a caffeine dependence syndrome sug¬
This study was supported in part by US Public
This study provides clinical evidence gests several areas deserving further Health Service grants K20 DA 00166 and R01 DA
supporting a caffeine dependence syn¬ exploration, including investigation of 03890.
drome similar to substance dependence the prevalence of the syndrome, the oc¬ The authors wish to thank Kim Puhala for her
syndromes for other drugs. Previous currence of comorbid disorders, and the assistance in conducting this study.

References
1. Graham HN. Tea: the plant and its manufacture; response to caffeine. Behav Pharmacol. 1992;3:219\x=req-\ Calif: Educational and Industrial Testing Service;
chemistry and consumption of the beverage. In: 228. 1971.
Spiller GA, ed. The Methylxanthine Beverages and 11. Griffiths RR, Mumford GK. Caffeine\p=m-\adrug 22. Jacob P III, Wilson M, Benowitz NL. Improved
Foods: Chemistry, Consumption, and Health Ef- of abuse? In: Bloom FE, Kupfer DJ, eds. Psycho- gas chromatographic method for the determination
fects. New York, NY: Alan R Liss Inc; 1984:29-74. pharmacology: The Fourth Generation ofProgress. of nicotine and cotinine in biologic fluids. J Chro-
2. Gilbert RM. Caffeine consumption. In: Spiller New York, NY: Raven Press Ltd. In press. matogr B Biomed Appl. 1981;222:61-70.
GA, ed. The Methylxanthine Beverages and Foods: 12. James JE. Caffeine and Health. London, En- 23. Yesair DW, Branfman AR, Callahan MM. Hu-
Chemistry, Consumption, and Health Effects. New gland: Academic Press Ltd; 1991. man disposition and some biochemical aspects of

York, NY: Alan R Liss Inc; 1984:185-213. 13. American Psychiatric Association. Diagnostic methylxanthines. In: Spiller GA, ed. The Methyl-
3. Graham DM. Caffeine\p=m-\itsidentity, dietary and Statistical Manual of Mental Disorders, Fourth xanthine Beverages and Foods: Chemistry, Con-
sources, intake and biological effects. Nutr Rev. Edition. Washington, DC: American Psychiatric sumption, and Health Effects. New York, NY: Alan
1978;36:97-102. Association; 1994. R Liss Inc; 1984:215-233.
4. Hughes JR, Oliveto AH, Helzer JE, Bickel WK, 14. Hughes JR, Oliveto AH, Helzer JE, Higgins 24. Griffiths RR, Evans SM, Heishman SJ, et al.
Higgins ST. Indications of caffeine dependence in ST, Bickel WK. Should caffeine abuse, dependence, Low-dose caffeine physical dependence in humans.
a population-based sample. In: Harris L, ed. Prob- or withdrawal be added to DSM-IV and ICD-10? J Pharmacol Exp Ther. 1990;255:1123-1132.
lems of Drug Dependence, 1992: Proceeding of the Am J Psychiatry. 1992;149:33-40. 25. Kozlowski LT, Henningfield JE, Keenan RM,
54th Annual Scientific Meeting, The College on 15. Silverman K, Evans SM, Strain EC, Griffiths et al. Patterns of alcohol, cigarette, and caffeine and
Problems of Drug Dependence, Inc. Rockville, Md: RR. Withdrawal syndrome after the double-blind other drug use in two drug abusing populations.
US Dept of Health and Human Services, National cessation of caffeine consumption. N Engl J Med. J Subst Abuse Treat. 1993;10:171-179.
Institute on Drug Abuse; 1993:194. NIDA Research 1992;327:1109-1114. 26. Kessler RC, McGonagle KA, Zhao S, et al. Life-
Monograph 132. 16. Griffiths RR, Woodson PP. Caffeine physical time and 12-month prevalence of DSM-III-R psy-
5. Barone JJ, Roberts H. Human consumption of dependence: a review of human and laboratory ani- chiatric disorders in the United States: results from
caffeine. In: Dews PB, ed. Caffeine: Perspectives mal studies. Psychopharmacology (Berl). 1988;94: the National Comorbidity Study. Arch Gen Psy-
From Recent Research. New York, NY: Springer\x=req-\ 437-451. chiatry. 1994;51:8-19.
Verlag New York Inc; 1984:59-73. 17. American Psychiatric Association. Diagnostic 27. Glassman AH, Helzer JE, Covey LS, et al. Smok-
6. Griffiths RR, Evans SM, Heishman SJ, et al. and Statistical Manual of Mental Disorders, Re- ing, smoking cessation, and major depression.
Low-dose caffeine discrimination in humans. J Phar- vised Third Edition. Washington, DC: American JAMA. 1990;264:1546-1549.
macol Exp Ther. 1990;252:970-978. Psychiatric Association; 1987. 28. Breslau N, Kilbey MM, Andreski P. Nicotine
7. Silverman K, Griffiths RR. Low-dose caffeine 18. Spitzer RL, Williams JBW. Structured Clini- dependence and major depression: new evidence
discrimination and self-reported mood effects in cal Interview for DSM-III-R. New York, NY: New from a prospective investigation. Arch Gen Psy-
normal volunteers. J Exp Anal Behav. 1992;57:91\x=req-\ York State Psychiatric Institute, Biometrics Re- chiatry. 1993;50:31-35.
107. search; 1985. 29. Kendler KS, Neale MC, MacLean CJ, Heath
8. Griffiths RR, Woodson PP. Reinforcing effects 19. Beck AT, Ward CH, Mendelson M, Mock J, AC, Eaves LJ, Kessler RC. Smoking and major
of caffeine in humans. J Pharmacol Exp Ther. 1988; Erbaugh J. An inventory for measuring depres- depression: a causal analysis. Arch Gen Psychiatry.
246:21-29. sion. Arch Gen Psychiatry. 1961;4:561-571. 1993;50:36-43.
9. Evans SM, Griffiths RR. Dose-related caffeine 20. Beck AT, Steer RA. Beck Depression Inven- 30. Greden JF, Walters A. Caffeine. In: Lowinson
discrimination in normal volunteers: individual dif- tory Manual. New York, NY: Psychological Corp; JH, Ruiz P, Millman RB, Langrod JG, eds. Sub-
ferences in subjective effects and self-reported cues. 1987. stance Abuse: A Comprehensive Textbook. Balti-
Behav Pharmacol. 1991;2:345-356. 21. McNair DM, Lorr M, Droppleman LF. EdITS more, Md: Williams & Wilkins; 1992:357-370.
10. Chait LD. Factors influencing the subjective Manual for the Profile of Mood States. San Diego,

Downloaded From: http://jama.jamanetwork.com/ by a New York University User on 06/05/2015