Tobacco Documents Online

I Ps~ chopharmacology (5988) 94 : 437-451 I f eviews Caffeine physical dependence : 9 review of human and laboratory animal studies Roland R. Griffithsl•2 and Phillip P. Woodson' Department of Psychiatry and Behavioral Sciences and 2 Department of Neuroscience, e Johns Hopkins University School of Medicine, 720 Rutland Avenue, Baltimore, MD 21205, USA Ibstract. Although caffeine is the most widely used beha- orally active drug in the world, caffeine physical depen- dence has been poorly characterized in laboratory animals nd only moderately well characterized in humans. In hu- ans, a review of 37 clinical reports and experimental stu- ies dating back to 1833 shows that headache and fatigue are the most frequent withdrawal symptoms, with a wide ariety of other signs and symptoms occurring at lower equency (e.g. anxiety, impaired psychomotor perfor- ance, nausea/vomiting and craving). When caffeine with- drawal occurs, severity can vary from mild to extreme (i.e. capacitating). The withdrawal syndrome has an onset at -24 h, peak at 20-48 h, and duration of about I week. he pharmacological specificity of caffeine withdrawal has been established. The proportion of heavy caffeine users ~ho will experience withdrawal'symptoms has been esti- ated from experimental studies to range from 25% to 00%. Withdrawal symptoms have been documented after relatively short-term exposure to high doses of caffeine (i.e. I-Is days of > 600 mg/day). Although animal and human udies suggest that physical dependence may potentiate the reinforcing effects of caffeine, human studies also demon- rate that a history of substantial caffeine intake is not necessary condition for caffeine to function as a rein- tircer. The similarities and differences between caffeine and classic drugs of abuse are discussed. y words: Caffeine - Caffeinism - Coffee - Tea - Physical pendence - Withdrawal - Reinforcer - Drug self-adminis; ation - Subjective effects - Drug dependence- Drug abuse - Humans - Animals I Introduction affeine is the most widely used behaviorally active drug the world (Gilbert 1984), with 82-92% of adults in North America regularly consuming caffeine (Gilbert 1976a; Gra- Om 1978). Presently, worldwide per capita caffeine con- mption has been estimated to be 70 mg per day which the equivalent of a large cup of instant coffee or a small cup of ground coffee for every man, woman and child (Gil- rt 1984; Barone and Roberts 1984). In the United States If d Canada, daily per capita caffeine consumption has been timated to be 211 and 238 mg, respectively. These figures are,, about half those estimated for the United Kingdom r4 mg) and Sweden (425 mg), which are particularly ~olfprint reyucs•ts to: R.R. Grifliths Psychopharmacology 2' Springer-Verlag 1988 heavy tea and coffee consuming countries, respectively (Gil- bert 1984). After oral administration to humans, caffeine is rapidly and completely absorbed, reaching maximal plasma levels at about 30 min (Blanchard and Sawers 1983). The princi- pal pharmacological actions of acute caffeine administra- tion are to stimulate the central nervous system, act on the kidney to produce diuresis, stimulate cardiac muscle, decrease peripheral vascular resistance while increasing cerebrovascular resistance, increase gastric and other secre- tions, and relax smooth muscle, most notably bronchial muscle (Rall 1985; Battig 1985). With repeated or chronic administration, tolerance occurs to various behavioral and physiological actions of caffeine (Hirsh 1984; Finn and Holtzman 1986). Caffeine is principally eliminated by me- tabolism in the liver and has an average plasma half-life of about 3-6 h in humans (Kalow 1985). The chronic use of a tolerance-inducing drug which has a moderate to rapid elimination rate makes that compound a good candidate for producing physical dependence as manifested by biochemical, physiological or behavioral dis- ruptions occurring upon termination of drug administra- tion. Although clinically significant caffeine physical depen- dence has been periodically described in medical reports dating back at least over the last century and a half, caffeine physical dependence is not widely recognized by the lay population or by health-care professionals. For example, the most recent version of the influential diagnostic manual of the American Psychiatric Association (DSM-III-R) does not acknowledge the existence of caffeine physical depen- dence (American Psychiatric Association 1987). The purpose of this paper is to review and evaluate the current scientific understanding of the physical depen- dence producing effects of caffeine. Sections II and III re- view studies documenting caffeine physical dependence in laboratory animals and humans, respectively. The subse- quent section (Section IV) reviews human studies suggesting that physical dependence may be an important determinant of the reinforcing effects of caffeine. Section V discusses pharmacological and physiological mechanisms of caffeine physical dependence, while the final section (Section VI) considers the controversial question of whether caffeine can meaningfully be considered to be a drug of abuse. Physical dependence in laboratory animals Although methods have been well established for evaluating the physical dependence potential of various drug classes in laboratory animals (Martin 1977; Brady and Lukas I

r I I I I I I I I I I I I I I 438 1984). surprisingly few studies have been conducted with caffeine. Six reports have been published using rats, most of which document substantial behavioral disruptions fol- lowing cessation of chronic caffeine dosing (Boyd et al. 1965: Vitiello and Woods 1977; Carney 1982: Holtzman 1983: Finn and Holtzman 1986; Holtzman and Finn 1987). The most reliable caffeine withdrawal effect has been decreased locomotor activity. Boyd and colleagues (1965) reported that termination of drug treatment (100 days of 190 mg/kg/day caffeine intragastrically for 5 days a week) was followed by a decrease in locomotor activity to half that before withdrawal and half that in nondrugged control animals. The decreased locomotor activity lasted I week, and was accompanied by a slight but significant decrease in colonic temperature and an increase in urinary protein and glucose. Holtzman showed that decreased locomotor activity during caffeine withdrawal is dose dependent. Groups of rats which drank an average of 19 or 36 mg/kg/ day caffeine for 6 weeks showed no change in locomotor activity, whereas animals consuming 67 mg/kg/day showed decreased locomotor activity following substitution of water for caffeine (Finn and Holtzman 1986). This decrease lasted 2 days and was maximal on the 1st day of withdrawal (about 50% of pre-withdrawal levels). In a similar experi- ment by Holtzman (1983), decreased locomotor activity was shown following substitution of water for caffeine (11 weeks of exposure to caffeine; approximately 160 mg/ kg/day during the last 7 weeks). This decrease lasted 4 days and was maximal on the 2nd day of withdrawal (about 80% of prewithdrawal levels). Disruption of operant schedule-controlled behavior during caffeine withdrawal has been less reliably demon- strated than the decreased locomotor activity. Carney (1982) showed a 50% decrease in food-maintained lever pressing behavior on days when rats were injected intraperi- toneally with saline rather than a standard daily dose of 32 mg/kg caffeine. Another study (Holtzman and Finn 1987), however, failed to show disruption of lever pressing in rats during caffeine withdrawal (50 and 90 mg/kg/day for 1 month). The only other preclinical study to provide information about caffeine withdrawal was one which used a taste-aver- sion paradigm to provide evidence for both the aversive properties of caffeine in naive rats and the aversive proper- ties of absence of caffeine in rats repeatedly exposed to caffeine (Vitiello and Woods 1977). In this study injections of caffeine to naive rats produced a dose-related avoidance of a novel flavor associated with caffeine. However, rats which had previously received injections of caffeine on each of 12 days (approximately 1.5-12 mg/kg/day) showed a dose-related avoidance of a novel flavor associated with the absence of caffeine. In contrast to the numerous reports and studies of caf- feine withdrawal in humans to be reviewed in Section III, the relatively few reports investigating the effects of caffeine withdrawal in laboratory animals is striking, particularly given the established utility of such animal models for char- acterizing physical dependence and pharmacological mech- anisms of action of various drug classes (Martin 1977; Brady and Lukas 1984). The only reliable behavioral effect of caffeine withdrawal to have been clearly documented in laboratory animals is decreased locomotor activity in rats. Parametric studies are needed to define the behavioral and species generality of these caffeine withdrawal effects. Given the relatively subtle nature of the observable behav- ioral effects of caffeine withdrawal in humans, it should be anticipated that characterization of analogous effects in laboratory aninfats may prove to be an experimental challenge. Physical dependence in humans As in research with laboratory animals, experimental meth- ods have been established for evaluating the physical depen. dence potential of drugs in humans (Martin 1977; Bra& and Lukas 1984; Petursson and Lader 1984). Caffeine phys_ ical dependence, as revealed by a withdrawal syndrome fol- lowing cessation of chronic caffeine dosing, has been clearly and repeatedly documented. Tables I and 2 summarize 37 reports, including case reports, clinical observations, ex- perimental studies and survey studies, which provide infor- mation about the signs, symptoms and time course of the caffeine withdrawal syndrome. In assembling the tables, 25 published reports which may have some relevance to caffeine dependence were pur- posely excluded. Twelve reports of abrupt and/or gradual withdrawal from caffeine after chronic, high dose caffeine consumption were excluded from the tables because it wa, unclear whether caffeine withdrawal signs or symptoms were explicitly looked for or documented (Roch 1914; Ross 1971; Greden 1974, case # 2; Molde 1975; De Freitas and Schwartz 1979; Foxx and Rubinoff 1979; Hyner 1979; Ber- nard et al. 1981 ; Young et al. 1982; Khoury and Maltbie 1984; James et al. 1985, 1987). Eleven reports describing various withdrawal signs and symptoms occurring upon abstinence after long-term use of products containing caf- feine in combination with other pharmacologically active compounds (e.g. aspirin, phenacetin, antipyrine, oxyco- done) were excluded from the tables because the results cannot be attributed solely to caffeine (Schilling 1928; Ides- trom 1960; Miller 1960; De Busscher and Varenne 1966: Gault et al. 1968; Kielholz 1970; Murray 1973; Burns 1977: Gardos 1977; Babington and Monson 1982; Granella et al. 1987). Two reports were also excluded which described withdrawal signs and symptoms after long-term use of com- bination products containing theophylline, an analog of caffeine (Laux 1979; Horowitz et al. 1982). Although a re- port by Vojtechovsky and 9afratova (1972) was cited in a prominent review paper as providing a possible example of caffeine withdrawal headache (Gilbert 1976 a), this report was excluded from the tables because it is not clear that the study involved a meaningful duration of caffeine absti- nence. Signs and symptoms of caffeine withdrawal. Tables I and 2 show that headache is the most frequently reported with- drawal symptom (19 reports). Possibly related, two addi- tional case reports described "fullness" in head and pres- sure in head or facial flushing, but no headache as part of the withdrawal syndrome (Cobbs 1982; Wilkin 19861 Caffeine withdrawal headache has been characterized a, being gradual in development (Dreisbach and Pfeiffer 194?: Greden et al. 1980; Roller 1981), diffuse (Dreisbach and Pfeiffer 1943; Greden 1974; Greden et al. 1980), throbbing (Dreisbach and Pfeiffer 1943; Greden 1974; Greden et al. 1980), severe (Bridge 1893; Dreisbach and Pfeiffer 1943: Naismith et al. 1970; Greden 1974; Greden et al. 1980. Weil and Rosen 1983 p 183; Rainey 1985), and phenomeno- 2C463994'79

439 fable 1. Summary of case reports and clinical observations relevant to caffeine withdrawal in humans Reference Udon(1833) # uelliot (1885 a) I t endel (1889) I ndge (1893) Clinical observations; patients who were heavy habitual coffee users abstained from coffee Clinical observations; patients who were habitu- al coffee users abstained from coffee t illes de Ia Tourette N=1 Case report; patient was a heavy habitual coffee nd Gasne (1895) [Case * 2] user; description of effects of overnight absti- nence I transky (1932) I tagner(1939) Subjects Type of report and history of caffeine use N=1 Case report; description of effects of abstaining from morning tea N=2 Case reports; two patients (3-12 cups of coffee/ [Cases # 3 day) with signs and symptoms of caffeinism, ab- & 6] stained from coffee N=1 Case report; patient was a heavy habitual coffee [Case # 5] user: description of effects of overnight absti- nence N=1 Case report; patient who had been eating up to and over 5 handfuls of roasted coffee beans and consuming several cups of coffee daily for 2 yrs abstained from consuming coffee beans N=1 Case report (first-person account); heavy user of caffeinated beverages; description of symp- toms of intermittent abstinence N=1 Case report: patient who consumed coffee pre- ~ranklin et al. (1948) N=36 ~ I I ~ Shorofsky and tamm (1977) pared from 250 to 375 g of ground coffee daily abstained from coffee when hospitalized with a variety of medical problems Clinical observations; during a semi-starvation experiment with male subjects, large amounts of coffee or tea (limited to a maximum of 9 cups/ day) were consumed; subjects were occasionally exposed to 3-day periods of greatly reduced fluid intake In Der Beeck (1961) N=1 Case report; patient was a heavy coffee drinker (coffee prepared from at least 100 g of ground coffee each day): description of symptoms of coffee abstinence N=1 Case report: patient consumed an estimated 1.5-1.8 g caffeine/day in coffee; coffee intake was restricted to one cup/day N=1 Case report; patient who was a heavy coffee [Case # 1] drinker (10-12 cups/day over last three weeks) abstained from coffee N=1 Case report: patient consumed caffeine-contain- [Case * 3] ing beverages and analgesics (estimated 1.5 g caffeine/day): description of symptoms appar- ently correlated with reduced caffeine intake Clinical observation; heavy coffee users; de- scription of symptoms during short-term (24-h) religious fasts Withdrawal signs and symptoms Always experienced mental confusion which was relieved by tea Concurrent medical problems and insufficient de- tail make these case reports difficult to interpret; upon complete or partial abstinence from coffee, both patients experienced insomnia which was apparently suppressed by coffee consumption Psychomotor impairment due to marked limb tremor occurred after overnight abstinence and was apparently suppressed by coffee consump- tion Weakness and dysphoria occur during the first weeks of coffee cessation; protracted withdrawal signs or symptoms may occur for several months In susceptible individuals, severe incapacitating headaches described as occurring during first day or two after abrupt coffee abstinence; constipa- tion described as occurring occasionally Marked hand tremor after overnight abstinence was partially suppressed after first cup of coffee in the morning Patient became tired and sleepy Abstinence from caffeinated beverages associated with a mild irritability/uneasiness combined with tiredness, sleepiness (yawning), and impairment in work- and thought-related activities Delirium described as a possible withdrawal sign Although not explicitly evaluated, a few subjects complained of headache and increased lassitude during periods of restricted caffeine and fluid in- take Feelings of apathy, lethargy, listlessness during coffee abstinence were suppressed by coffee con- sumption None Fatigue was reported for I wk after coffee absti- nence Severe headache; relief of headache obtained with analgesics containing caffeine but not with caffeine-free analgesics Withdrawal headache described as a common problem I

I I I I I I I I I I A I I I I I I I 440 Table 1. (continucd) Reference - Subjects Grcdcn et al. (1980) N= 1 Gibson (1981) N= 1 Cobbs (1982) N=1 Weil and Rosen N=2 (1983 p 183) Rainey (1985) N=1 Wilkin (1986), N=1 Type of report and history of caffeine use Case report; patient who was a heavy caffeine consumer (greater than 500 mg caffeine/day over 20 years) sharply reduced or eliminated caf- feine intake of intermittent coffee abstinence Case report: depressed patient was a heavy cof- fee drinker (10-15 cups/day): description of ef- fects of abrupt caffeine abstinence upon admis- sion to a metabolic ward Case report: diabetic patient consumed up to 30 cups of coffeeiday; description of symptoms Case report (first-person account); two heavy habitual coffee users switched to decaffeinated coffee Case report; patient consumed an estimated 834 mg caffeine/day in caffeine-containing bev- erages and analgesic tablets; treatment consisted of abstinence from all methylxanthines Case report; patient consumed 8-12 cups of cof- fee/day excepting weekends when 1 cup/day was consumed logically distinct from migraine headache (Dreisbach and Pfeiffer 1943). Although headache is the most frequently reported withdrawal symptom, it should be noted that sev- eral clinical reports have concluded that acute caffeine (Marburg 1899; Hollingworth 1912; Dreisbach and Pfeiffer 1943) or coffee (Guelliot 1887; Schulte 1950; Harrie 1970) administration can induce headache in some individuals. In addition to headache, a constellation of symptoms frequently reported during caffeine withdrawal is character- ized by fatigue (i.e. mental depression, let-down, fatigue, weakness, lethargy, lassitude, apathy, listlessness, tiredness, sleepiness, drowsiness, yawning, disinclination to work, lazy, and decreased activeness and alertness). Tables I and 2 show that 15 separate reports described such symptoms. A third possible dimension of the caffeine withdrawal syndrome is anxiousness (i.e. anxious, nervous, jittery, shaky, muscle tension, restless, and insomnia). Although this dimension has been described in eight reports in Ta- bles 1 and 2, the supporting evidence is not as compelling as that for headache and fatigue (Goldstein et al. 1969; Griffiths et al. 1986 a). A wide variety of other signs and symptoms have been reported to occur during caffeine withdrawal, but at a rela- tively low frequency. These signs and symptoms, with the number of reports from the tables indicated in parentheses, Withdrawal signs and symptoms Headache began 18-20 h after abstinence, peak- ing 3 h after onset and lasting at least 36 h unless caffeine consumed: accompanied by rhinorrhea, fatigue, yawning: patient reported being able to smell coffee, even when none was present; head- ache was suppressed by coffee consumption On 2nd and 3rd day of caffeine abstinence urinary MHPG increased along with anxiety and headache Periods of coffee abstinence associated with "fullness" and pressure in head (onset 3-5 h), "let-down " (onset 4-6 h), lethargy, fine motor impairment, confusion: other possible symptoms included muscle stiffness, blurred vision, slowed speech, diaphoresis, and anxiety attacks; with- drawal symptoms apparently suppressed by cof- fee consumption Severe headache (both individuals) and tiredness (one individual) occurred on 2nd day after switching to decaffeinated coffee; symp- toms were suppressed by caffeinated coffee con- sumption Headache, irritability, nervousness, vomiting, in- somnia, restlessness, and lethargy were reported over a 3-day period. Weekend syndrome of facial reddening accompa- nied by "blood shot" eyes and sensations of "fullness" and warmth; symptom onset late Sa- turday morning and most intense on Saturday evening include impaired psychomotor preformance and/or marked limb tremor (5), irritability/uneasiness (4), rhinorrhea (3), nausea/vomiting (2), confusion (2), diaphoresis (2), muscle pains/stiffness (2), inability to work effectively (2), de- creased contentedness (1), dysphoria (1), blurred vision (1). slowed speech (1), constipation (1), scleral injection (1), fa- cial warmth (1), delirium (1), olfactory hallucination (1). decreased cigarette smoking (1), increased cerebral blood flow (1), increased urinary MHPG (1), decreased lympho- cyte fl-adrenoceptor sensitivity (1), lowered serum calcium (1), and elevated serum phosphorus (1). "Craving" for cof- fee has also been described, although not empirically docu- mented, as a coffee withdrawal symptom (Goldstein et al 1974; Rippere 1984). For the most part, these miscellaneou, signs and symptoms have not been rigorously evaluated in experimental studies and thus their reliability as part of the caffeine withdrawal syndrome remains to be estab- lished. Ser;eritv. When signs or symptoms of caffeine withdrawal occur, the severity can vary from mild to extreme. At its worst, caffeine withdrawal has been repeatedly documented to be incompatible with normal functioning and sometime- totally incapacitating (Kingdon 1833; Bridge 1893; Dreis- back and Pfeiffer 1943 ; Goldstein and Kaizer 1969; Greden 264639r;)481

r 441 t,bte 2. Summary of experimental and survey studies relevant to caffeine withdrawal in humans Reference Subjects orst et al. N= 7 adult men 1934) k reisbach and N = 22 mostly men ; feiffer (1943) graduate and medical students I i rGoldstein (1964) N=approximate- I I ly 109 [Experiments c and d] mostly men ; medical students Goldstein and N=183 wives of grad- I Kaizer (1969) uate students ' I Goldstein et al. N= 56 wives of gradu- ~(1969) ate students I Naismith et al. N=20 male and fe- l(1970) male staff from a de- partment of nutrition I~ instead (1976) N= 135 mostly adult men; inpatients on an acute psychiatric ward I (1978) adult psychiatric inpa- tients lGreden et al. N=205 hospitalized (1980) I Greden et al. N=83 mostly men; patients Design Withdrawal not blind; after receiving caf- feinated coffee (estimated 3-4 mg/kg caf- feine) once daily for 1-8 wks, coffee dosing was abruptly terminated Single-blind; caffeine administered in cap- sules in increasing doses over 6-7 days to 600-750 mg, day; placebo capsules substi- tuted for caffeine capsules Double-blind; subjects abstained from caf- feine after lunch and received 150 mg caf- feine or placebo in decaffeinated coffee at bedtime over 4 successive nights with both treatments being given twice Questionnaire survey (no experimental ma- nipulation); consequences of omission of morning coffee Double-blind; subjects abstained from caf- feine after dinner and received placebo, 150 mg or 300 mg caffeine (free base) in de- caffeinated coffee the following morning (about 9 A.M.); this procedure occurred re- peatedly over a 9-day period with each treatment being given 3 times Not blind; after a 10-day baseline period (estimated dietary caffeine intake of 560 mg/day) subjects switched abruptly to decaffeinated coffee Questionnaric survey (no experimental ma- nipulation) of occurrence of coffee with- drawal symptoms: 25% of the group were defined as heavy coffee users (i.e. estimated to consume at least 500 mg/day on at least 2 study days) Questionnaire survey (no experimental ma- nipulation) of occurrence of headache on omission of morning caffeine Questionnaire survey (no experimental ma- nipulation) of occurrence of headache upon stopping routine caffeine consumption Withdrawal signs and symptoms Results of this non-blind study are difficult to interpret; some suggestive evidence for impaired psychomotor performance during 1st wk of with- drawal Lethargy in morning, cerebral fullness at noon, and headache in early afternoon, reaching peak intensity 3-6 h later; nausea, vomiting, rhinorr- hea, lowered serum calcium and elevated serum phosphorus accompanied headache in some sub- jects: other withdrawal symptoms included men- tal depression, drowsiness, yawning, and disin- clination to work: withdrawal headache was sup- pressed by caffeine; 82% of the subjects reported definite or severe withdrawal headache while the remaining subjects reported very slight or no headache In subjects with histories of heavy coffee use (5 or more cups/day), morning headache occurred sig- nificantly more frequently after placebo (25% of trials) than after caffeine (3% of trials); among moderate coffee users (2-4 cups/day) the fre- quency of headache was nonsignificantly higher after placebo (12% of trials) than after caffeine (7% of trials); in light coffee users (0 or I cups/ day) the frequency of headache was equally low ,.after placebo and caffeine Moderate (3-4 cups/day) and especially heavy coffee users (5-10 cups/day) reported irritability, inability to work effectively, nervousness, rest- lessness, lethargy, headache; light coffee users (1-2 cups/day) did not report these symptoms; percentages of light, moderate and heavy coffee users reporting headache were 0, 9, and 8%, re- spectively Compared to subjects who were not regular cof- fee drinkers, heavy coffee users (5 or more cups/ day) reported being less alert, active and content, and more sleepy, irritable, and jittery/nervous/ shaky after caffeine abstinence; caffeine generally produced a dose-related suppression of with- drawal symptoms (including headache) in the heavy users All subjects reported lassitude and severe head- ache within 12 h of caffeine abstinence: symp- toms disappeared after a further 36 h Anxiety withdrawal symptoms were reported more frequently by the heavy user group (26%) than by other patients (5%) 11 % reported headache; no significant differ- ences between light, moderate or heavy caffeine users 20% of total sample (28% of the 152 who ans- wered the question) reported caffeine withdrawal headache; those reporting headache had higher mean caffeine intake (616 mg/day) than those not reporting headache (395 mg/day)

442 I I I I I I I I I I I I I I I Table 2. (continued) Reference _ Subjects Design White et al. N=36 college stu- Double-blind: subjects abstained from caf- (1980) dents feine for at least 3 hrs: muscle tension was measured with electromyogram before and 30 min after either 300 mg caffeinc citrate (N= 19) or placebo (N= 17); anxiety and reaction time assessed after drug or placebo administration Mackenzie et al. N = 7 adult men Not blind: subjects abstained from caffeine (1981) after at least 3 months of daily coffee drink- ing with estimated caffeine intake of 400 mg/day Robertson et al. N= 18 adult men and Double-blind; caffeinated barley-based (1981) women beverage served 3 times/day (750 mg caf- feine/day) for 7 days followed by substitu- tion of a placebo barley-based beverage Roller (1981) N = 1 adult man A heavy coffee drinker (900-1100 mg caf- feine/day) abstained from caffeine for a 72-h period ; theophylline was given at 24 h into this 72 h period; at the end of 72 h either caffeinated coffee (approx 115 mg caffeine) or decaffeinated coffee was given in a blinded fashion: this protocol was rep- licated on 9 occasions Victor et al. N= 124 male and fe- Questionnaire survey (no experimental ma- (1981) male medical inpa- nipulation) of occurrence of caffeine with- tients drawal headache Ammon et al. N=10 adult male stu- Double-blind; approximately half the sub- (1983) dents jects were switched to decaffeinated coffee after 504 mg/day caffeine in coffee for 4 wks Edelstein et al. N<430 residents of a Not bli.nd: subjects consumed 3 or more (1983) psychiatric hospital caffeinated beverages/day; decaffeinated beverages were substituted for caffeinated beverages Carter (1984) N=32 Double-blind; regular coffee drinkers (38 cups/week for 19 years): approximately half of the subjects were switched to decaf- feinated coffee after five days of caffeinated coffee Mathew and N= 16 adult men and Not blind; heavy caffeine users (estimated Wilson (1985) women 986 mg/day) and light caffeine users (esti- mated 126 mg/day) abstained from caffeine for 24 h Griffiths et al. N=7 adult men; most Double-blind; heavy caffeine users were (1986a) with histories suggest- switched to decaffeinated coffee after con- ing drug or alcohol suming caffeinated coffee for an average of abuse 10 consecutive days (mean of 1.25 g/day caffeine base during last 5 days) et al. 1980; Cobbs 1982; Weil and Rosen 1983 p 183; Rainey 1985; Griffiths et al. 1986a). For example, in one early report of caffeine withdrawal, Bridge (1893) noted that in susceptible individuals, " Headache often occurs dur- Wiibdrawal signs and symptoms Before caffcine or placebo, high caffeine con- sumers (376 mg over previous 24 h) showed more muscle tension than did low consumers (87 mg over previous 24 h): among subjects receiving placebo, anxiety was positively correlated with level of prior caffeine consumption On day 3 of caffeine abstinence a decrease in lym- phocyte beta-adrenoceptor sensitivity occurred in all subjects: by day 7 some sensitivity was re- gained in 5 of 7 subjects Despite nearly complete tolerance to caffeine's effects on several humoral and hemodynamic variables, substitution of placebo did not result in any detectable effects on these measures Withdrawal symptoms started after about 6 hrs with headache; shortly thereafter came lassitude, then rhinorrhea and leg muscle pains, followed by diaphoresis; after 16 h of abstinence came gen- eral muscle pains (flu-like symptoms); these symptoms gradually increased to a maximum in- tensity at a later time and were suppressed by caffeinated coffee consumption 24% of all subjects reported withdrawal head- ache: differences between low, moderate, and high caffeine consumers were apparently not sig- nificant Although tolerance developed to caffeine's effects on blood pressure, blood pressure was not af- fected by substitution of decaffeinated coffee Headache occurred during first 1-2 wks of decaf- feinated beverages No withdrawal symptoms occurred during actual period of decaffeinated coffee, although symp- toms did appear during periods that subjects be- lieved they were on decaffeinated coffee A bilateral increase in cerebral blood flow oc- curred in several frontal regions of the heavy caf- feine user group only; although a"few" subjects reported withdrawal symptoms, no significant in- crease in headache occurred Orderly syndrome developed with onset latency of 19 h, peaking on the ist or 2nd day, then de- creasing over the next 5-6 days; subjects reported more headache and being more sleepy, more lazy. less alert and less active; other withdrawal effects included changes in staff ratings of subject mood and behavior, decreased cigarette smoking, and trend in psychomotor performance impairment; 100% of subjects reported withdrawal headache ing the first day or two in so severe a degree as to compel the individual to keep his bed..." An early experimental study by Dreisbach and Pfeiffer (1943) documented that. "headache as extreme in severity as the subjects had ever G4~~:~4~3 2

l perienced was produced by the sudden withdrawal of caf- me." This extreme headache occurred in 55% of 38 trials on 22 individuals comprising a relatively unselected subject pulation. 7i me course. The caffeine withdrawal syndrome follows an orderly time course. Onset generally has been reported to c ur 12-24 h after terminating caffeine intake (Dreisbach d Pfeiffer 1943; Goldstein 1964; Goldstein and Kaizer 969; Goldstein et al. 1969; Naismith et al. 1970; Greden et al. 1980; Mathew and Wilson 1985; Griffiths et al. I 86a; Wilkin 1986), although two studies have described set as early as 3 or 6 h (Roller 1981 ; Cobbs 1982). Fatigue has been described as preceding headache or cerebral full- ess in some reports (Dreisbach and Pfeiffer 1943) but not others (Roller 1981; Cobbs 1982). Intensity of caffeine tithdrawal has generally been described as peaking at 20-48 h of abstinence (Dreisbach and Pfeiffer 1943; Greden al. 1980; Griffiths et al. 1986a; Wilkin 1986). The dura- n of caffeine withdrawal has most often been described Tbe about I week (Horst et al. 1934; Greden 1974; Mack- enzie et al. 1981; Griffiths et al. 1986a), although substan- troal differences across individual subjects have been noted riffiths et al. 1986a) and one clinician has suggested that rotracted withdrawal signs or symptoms may occur for several months after terminating caffeine intake (Mendel ~889). ~harm acological specificity. The pharmacological specificity of the physical dependence to caffeine has been established I y a number of different observations. First, the severity f withdrawal appears to be an increasing function of the caffeine maintenance dose. Such dose dependence has been shown repeatedly in studies comparing groups of subjects at tdiffer in self-selected histories of caffeine consumption oldstein 1964; Goldstein and Kaizer 1969; Goldstein et al. 1969; Naismith et al. 1970; Winstead 1976; Greden i I t al. 1980; White et al. 1980; Mathew and Wilson 1985), though an absence of dose dependence has been occasion- ly described (Greden et al. 1978; Victor et al. 1981). It should also be noted, however, that dose dependence has ot yet been demonstrated in a prospective experimental udy. A second type of observation suggesting the pharma- logical specificity of caffeine physical dependence is that withdrawal can be produced by caffeine administered in psules (Dreisbach and Pfeiffer 1943) as well as by caffeine dministered in beverages (cf. Tables I and 2). Third, caf- ine ingested in capsules or tablets (Dreisbach and Pfeiffer 1943; Greden 1974), or in beverages (Kingdon 1833; Guel- ~t 1885a; Gilles de la Tourette and Gasne 1895; In Der eck 1961; Goldstein et al. 1969; Greden et al. 1980; oller 1981; Cobbs 1982; Weil and Rosen 1983 p 183) can suppress symptoms of caffeine withdrawal induced by caf- tne abstinence. Fourth, the magnitude of suppression by affeine of symptoms induced by caffeine abstinence is an ncreasing function of caffeine dose (Goldstein et al. 1969). Fifth, caffeine is more effective in suppressing withdrawal eadache than oxygen inhalation or administration of ace- ylsalicylic acid, benzedrine sulfate or amyl nitrite (Dreis- ach and Pfeiffer 1943). Iroportion of population at risk and individual di%lerences. rom the available data, it is difficult to estimate what proportion of heavy caffeine consumers will experience 443 symptoms after caffeine abstinence. There are four pub- lished experimental studies which prQvide information about the proportion of heavy caffeine using subjects (esti- mated >_ 500 mg caffeine/day) experiencing caffeine with- drawal headache. In two studies, 100% of subjects reported headache (Griffiths et al. 1986a; Naismith et al. 1970). A considerably lower figure of 25% can be estimated from a study by Goldstein (1964) based on the total number of withdrawal trials conducted. This relatively low rate of withdrawal may be due to a relatively shorter period of withdrawal (approximately 18 h) which terminated with breakfast after a night of sleep. All three of these studies involved potentially biased subject groups that were specifi- cally selected because of the subjects' high use of caffeine. The fourth study (Dreisbach and Pfeiffer 1943), in contrast, involved a relatively unbiased population of students with varied histories of caffeine use. During caffeine abstinence after a period of experimentally administered caffeine in capsules, 82% of the subjects experienced headache, with no clear influence of baseline rate of caffeine use (i.e. all three subjects who were previously caffeine abstainers expe- rienced headache during withdrawal of experimentally ad- ministered caffeine). The percentage of heavy caffeine users experiencing caf- feine withdrawal headache in these experimental studies is high relative to that reported by heavy caffeine users in retrospective questionnaire surveys: 8% (Goldstein and Kaizer 1969), 11 % (Greden et al. 1978), and <_ 10% (Win- stead 1976). Th3 discrepancy might be attributable to the possibility that heavy users infrequently omit daily caffeine and have little experience with withdrawal headache (Gre- den et al. 1978). It should also be noted that, although caffeine with- drawal symptoms have been frequently reported in experi- mental studies, such symptoms are not an invariant conse- quence of termination of high dose consumption. There are instances in which individuals with histories of heavy caffeine use apparently experienced no symptoms upon abrupt termination or restriction of caffeine (Reimann 1967; Carter 1984). Also, as described previously, 12 re- ports of abrupt and/or gradual withdrawal from caffeine after chronic, high dose caffeine consumption were ex- cluded from the tables because it was unclear whether caf- feine withdrawal signs or symptoms were explicitly looked for or documented. It is possible that some or all of these reports represent instances in which withdrawal signs and symptoms were totally absent. It is important to recognize, however, that absence of symptoms does not necessarily indicate an absence of physical dependence. For example, Mathew and Wilson (1985) showed that caffeine abstinence produced reliable increases in cerebral blood flow but only inconsistent reports of withdrawal symptoms. A more thorough understanding of individual differ- ences will be important for determining what proportion of the general population is a risk for experiencing signifi- cant caffeine withdrawal effects. Animal and human studies have clearly documented substantial differences between in- dividual subjects in the behavioral effects of caffeine (e.g. Goldstein et al. 1965b; Logan et al. 1986; Seale et al. 1986), including the reinforcing effects (Griffiths and Woodson 1987). Consistent with these observations, human caffeine withdrawal studies have documented substantial differences across subjects with respect to incidence (Dreisbach and Pfeiffer 1943) or continuance (Griffiths et al. 1986a) of I

r ~ ~ ~ 1 ~ 1 i ~ t ~ 1 i headache after abrupt caffeine withdrawal. The possibility should be explored that heavy coffee users are inherently more susceptible to the reinforcing andior physical depen- dence producing effects of caffeine. Further research will also be necessary to establish the reliability of and mecha- nism(s) for these individual differences in caffeine reinforce- ment and physical dependence. Minimum dosing parameters for caJJeine withdrawal. Al- though there is relatively little information available on the Pre- 1 Withdrawal (calfemated coltee) Subjective Effects of Caffeine Withdrawal 20, 1 5" 05, . 00 30.1 2.51 1 51 2 3 4 5 6 7 8 9 10 Consecutive Days of Withdrawal (decatteinated coffee) minimum dosing parameters necessary for the expression of caffeine physical dependence, some evidence suggests that withdrawal phenomena may be detectable after rela- tively short-term exp©surc to high caffeine doses or after long-term exposure to relatively low doses. Two studies documented caffeine withdrawal headache upon abrupt ter- mination of high doses of caffeine (terminal doses >_ 600 mg/day) administered for 6-15 days. One of these studies was conducted with three subjects who were nor- mally caffeine abstainers (Dreisbach and Pfeiffer 1943) 2.01 1.51 ~ 1.0 0.5 0.0 T- r Pre- 1 W it hdrawal (calteinated cdfee) 2.0't 1.5' m _ ~ 1.0' Q 0.51 10-1 _ _ 0.0 T r Pre- 1 Withdrawal (cafteinated coffee) 2 3 4 5 6 7 8 9 Consecutive Days of Withdrawal (decaffeinated coffee) 10 T r Pre- 1 Withdrawal (caffeinated coffee) 2 3 4 5 6 7 8 9 Consecutive Days of Withdrawal (decaffeinated coffee) 10 N 2 3 4 5 6 7 8 9 10 Consecutive Days of Withdrawal (decalteinated coffee) 2.0-1 ~ F~~+ ~ r ~ m 10 ~ ~ S 05 ~ ~ 0 0J 0 T Withdrawal (calfeinated coffee) Pre- 1 2 3 4 5 6 7 8 9 10 Consecutive Days of Withdrawal (decafteinated coffee) Fig. 1. Caffeine withdrawal: effects of substituting decaffeinated coffee for caffeinated coffee on subject-rated adjective clusters in seven subjects. The decaffeinated phase was preceded by a mean of 10 successive days of drinking only caffeinated coffee (100 mg caffeine per cup). v-Axes: 12:30 P.M. ratings on five adjective clusters. x-Axes: consecutive days; pre-withdrawal data show mean daily results from the 5 days which preceded substitution of decaffeinated coffee. Each data point with brackets indicates mean ± I SEM for seven subjects (N=7). Filled data points indicate which decaffeinated coffee days were significantly different (P<0.05) from the 5-day pre- withdrawal period. (Figure adapted from Griffiths et al. 1986a) ~

I hile the other study involved three subjects with histories f heavy caffeine use who were caffeine abstinent for 13-17 days (Griffiths et al. 1986a). The latter study also ggested that 11-15 days of high dose caffeine exposure ~iay not be sufficient for producing the maximal degree of caffeine physical dependence as reflected in frequency of headache. A suggestion that withdrawal can occur after ffng-term exposure to relatively lower doses of caffeine has een provided in clinical descriptions (Schlesinger 1931) and in two studies which showed nonsignificant elevations in withdrawal symptoms after chronic, self-selected exposure two to four cups of coffee/day (Goldstein 1964; Gold- 11ein and Kaizer 1969). Given the large size of the popula- tion at risk, it will be important for future research to deter- ~ ine whether a clinically significant low dose caffeine with- rawal syndrome can be reliably detected. Therapeutic detoxification from high doses of caffeine. In I esponse to the toxic manifestations of caffeinism and asso- iated health risk concerns, therapeutic detoxification from affeine has been and is presently often suggested as a logi- cal therapeutic strategy. Clinicians have made widely vary- l ng recommendations about the best procedures for ac- omplishing caffeine detoxification, including: (1) abrupt bstinence (Roch 1916); (2) gradual dose tapering (Guelliot 1885b; Bridge 1893; Greden 1981; Khoury and Maltbie ~984) sometimes in the context of structured behavior mod- fication programs (Foxx and Rubinoff 1979; Bernard et al. 1981; James et al. 1985); and (3) pharmacological replace- ment with caffeine-containing medication (Rugh 1896; ~transky 1932; Dreisbach and Pfeiffer 1943; Greden 1981). number of clinicians have also recommended the use of various other medications to provide symptomatic relief from the discomfort of caffeine withdrawal (Cole 1833; tuelliot 1885b, 1887; Bridge 1893; Dreisbach and Pfeiffer 43; Greden 1981; Khoury and Maltbie 1984). The diver- sity of treatment strategies for caffeine detoxification prob- ably reflects widely varying opinions and knowledge about the severity and frequency of a caffeine withdrawal syn- drome. Given the documented wide individual differences in severity and duration of caffeine withdrawal (Dreisbach Iand Pfeiffer 1943; Griffiths et al. 1986a), perhaps the wisest approach to caffeine detoxification is to deal with each case individually. Because of ease of implementation, abrupt ces- sation in a supportive therapeutic context can be attempted nitially. If significant withdrawal symptoms develop during abrupt abstinence, the more involved procedures of caffeine treplacement therapy, supplemental medication (e.g. acetyl- salicylic acid), or structured behavioral dose tapering pro- ~ grams should be used (cf Greden 1981 for thoughtful sug- gestions about practical implementation of caffeine detoxi- fication procedures in the context of medical practice). With continued progress in the precise characterization of the caffeine withdrawal syndrome, future research should focus Ion the development of empirically based treatment regi- mens of optimal efficacy. I An illustrative example of caf'feine withdrawal. A recent characterization of behavioral and subjective aspects of the caffeine withdrawal syndrome in humans was provided in an experiment in which seven subjects with histories of ~ heavy coffee drinking were switched abruptly, under dou- ble-blind conditions, from caffeinated coffee to decaffein- ated coffee for 10 or more days (Griffiiths et al, 1986a). 445 Objective Effects of Caffeine• Withdrawal 13, 0 S a~ r m "' 11 ~ ct; a L 9 C' 7 51 'i T f Pre• 1 Withdrawal (cafleinated coHee) 100, 95 1 90 1 ~ " 851 ~ CL T 1- Pre• 1 2 3 4 S 6 7 8 9 10 Consecutive Days of Withdrawal (decafteinated coffee) 2 3 4 5 6 7 8 9 10 Consecutive Days of Withdrawal (decaffeinated coffee) Withdrawal (caffeinated coffee) 80J 20 T Pre. 1 2 3 4 5 6 7 8 9 10 Withdrawal (cafleinated coffee) Consecutive Days of Withdrawal (decaffeinated coffee) Fig. 2. Caffeine withdrawal: effects of substituting decaffeinated coffee for caffeinated coffee on objective measures of subject be- havior. r-Axes: composite score from mean of 12:30 P.M. and 8:30 P.M. staff ratings of seven adjective clusters, psychomotor performance on a circular lights task, and number of cigarettes smoked. Other details are similar to those in legend of Fig. 1. (Fig- ure adapted from Griffiths et al. 1986a) Mean caffeine intake over the 5 days preceding the with- drawal phase was 1.25 g/day (range across subjects, 0.86-1.63 g/day). The withdrawal phase was the first occa- sion during their experimental participation (which aver- aged 19 days) that subjects were exposed to decaffeinated coffee for more than a 24 h period. As will be described in more detail in the following section, substitution of decaf- feinated coffee did not significantly affect number of cups of coffee consumed but was associated with transient de- I

I I I I I I I I I I I I I I I I 446 creases in subject ratings of coffee liking. The results of the experiment also showed that substitution of decafTein- ated coffee ,.,produced an orderly withdrawal syndrome which peaked on day I or 2 after substitution and then gradually subsided. Caffeine withdrawal produced signifi- cant increases in the Fatigue scale and significant decreases in the Vigor scale on the Profile of Mood States question- naire, a standardized mood state inventory (McNair et al. 1971). Figure 1 shows significant effects of caffeine with- drawal on five subject-rated adjective clusters: 1) sleepy, tired, drowsy, half-awake; 2) lazy, sluggish; 3) alert, atten- tive, observant, able to concentrate; 4) active, stimulated, energetic; 5) headache. These were rated by the subjects on a 4-point scale (0=definitely does not apply: and 3=very strongly applies) on the basis of how they were feeling at the present time. In this study the most sensitive and reliable subject-rated withdrawal symptom was head- ache which occurred in all seven subjects and, for the group, remained significantly elevated over prewithdrawal levels through the 3rd day after substitution of decaffeinated cof- fee (Fig. 1). Objective behavioral measures of caffeine withdrawal had the same time course as the subjective effects (Figs. 1 and 2). The top panel of Fig. 2 shows that changes in sub- ject behavior were prominent enough to be detectable in ratings by observers who were blind to drug condition. For this measure, observers used a 4-point scale to rate the subject on seven adjective clusters (alert, content, active, sleepy, talkative, lazy, and irritable) on the basis of observ- ing the subject over a 2-h period. The figure also shows that caffeine withdrawal was associated with a significant decrease in the number of cigarettes smoked as well as a trend toward'disruption of psychomotor performance. Physical dependence as a determinant of caffeine reinforcing effects in humans A large number of experimental studies have evaluated var- ious caffeine-induced subjective effects that might plausibly be related to the reinforcing properties of caffeine. This literature shows that, in contrast to amphetamine which generally produced elevations in ratings indicating "eu- phoria" and "well-being," caffeine did not consistently produce such effects (cf Weiss and Laties 1962; Chait and Griffiths 1983; Battig 1985). A number of studies, in fact, showed that caffeine produced "dysphoric" changes in mood such as increases in anxiety and nervousness (e.g. Goldstein et al. 1965a; Greden 1974; Rapoport et al. 1981; Chait and Griffiths 1983 ; Charney et al. 1984). Some of the clearest initial experimental evidence for positive mood changes produced by caffeine came from a survey and a clinical pharmacology study which showed that after overnight caffeine abstinence, heavy coffee users (_ 5 cups per day) reported pleasant and desirable effects of coffee drinking and caffeine administration in contrast to coffee abstainers who reported unpleasant and undesir- able effects (Goldstein and Kaizer 1969; Goldstein et al. 1969). The self-selected nature of the subject populations made it unclear whether the observed difference between heavy users and coffee abstainers was related to physical dependence or, alternatively, reflected some other pre-ex- isting difference between these groups. The reinforcing properties of caffeine in humans have been investigated more directly by adapting procedures w Caffeinated , -*-- Decaffeinated -4- 3.0 ~ c>-e w I h T l/1 11l T I I/Y y 1 ln C) Y (~ I~ ~ 1 ~ /I 1 1 Mr ~ J U_ ,I 1 1~ V CM` 2.0 1,0 0 , , , 7-7 r r 5 10 i r T___r 15 , ~ ~ CONSECUTIVE DAYS Fig. 3. Effects of substituting decaffeinated coffee for caffeinated coffee on subject rated coffee liking in seven subjects. The decaf. feinated phase was preceded by a mean of 10 successive days of drinking only caffeinated coffee (100 mg caffeine per cup), y-Axes: 8:30 P.M. ratings of coffee liking. x-Axes: consecutive days. Each data point with brackets indicates mean ± I SEM for seven sub- jects (N = 7). Filled data points indicate which decaffeinated coffee days were significantly different (P<0.05) from the 5-day period preceding substitution of decaffeinated coffee. (Figure adapted from Pharmacology Biochemistry & Behavior, R.R. Griffiths and P.P. Woodson, Reinforcing properties of caffeine: Studies in hu- mans and laboratory animals, in press, Pergamon Journals, Ltd) originally developed in the animal drug self-administration laboratory (Griffiths et al. 1980). To date, five such human caffeine self-administration studies have been reported (Kozlowski 1976; Podboy and Malloy 1977; Griffiths et al. 1986a, b; Griffiths and Woodson 1987). These reports, which have been reviewed recently (Griffiths and Woodson 1987), show that under appropriate conditions caffeine can serve as a reinforcer. The results of one of these reports extend the observa- tions by Goldstein (Goldstein and Kaizer 1969; Goldstein et al. 1969), discussed above, by providing the first experi- mental demonstration of the potentiation of the behavioral reinforcing effects of caffeine by physical dependence (Grif- fiths et al. 1986a). This report involved a series of experi- ments investigating the self-administration and reinforcing effects of caffeine in coffee in subjects with histories of heavy coffee drinking who resided on a research ward. One experiment, described in more detail in Section III, involved substitution of decaffeinated for caffeinated coffee under double-blind conditions. Although substitution of decaf- feinated coffee did not significantly affect number of cups of coffee consumed, Fig. 3 shows that coffee liking de- creased significantly on the first 2 days after substitution and subsequently progressively increased to pre-substitu- tion levels. This transient decrease in liking was probabiN related to the caffeine withdrawal syndrome which was as- sessed concurrently on other subjective and objective mea- sures and showed a similar time course (cf Section III and Figs. 1 and 2). To the extent that liking might predict rein- forcing effects, these data provided suggestive evidence that. in subjects physically dependent on caffeine, decaffeinated coffee may be aversive relative to caffeinated coffee. The implication that physical dependence may poten- tiate the relative reinforcing effects of caffeinated (100 mg per cup) versus decaffeinated coffee was investigated more directly in a subsequent experiment which utilized a double-