Food and Chemical Toxicology 40 (2002) 12571261 www.elsevier.

com/locate/foodchemtox

Caeine: behavioral eects of withdrawal and related issues

P.B. Dewsa,*, C.P. OBrienb, J. Bergmanc
a

New England Regional Primate Research Center, Harvard Medical School, One Pine Hill Drive, Southborough, MA 01772-9102, USA b Department of Psychiatry, University of Pennsylvania/VA Med. Center, 3900 Chestnut St, Philadelphia, PA 19104-6178, USA c McLean Hospital, Harvard Medical School, 115 Mill Street, Belmont, MA 02178-9106, USA Accepted 28 August 2001

Abstract Acquired tolerance to some behavioral eects of caeine in humans is widely assumed to occur but is poorly documented and appears, at most, to be of low magnitude. Withdrawal from regular consumption of caeine has been reported to result in a variety of symptoms, including: irritability, sleepiness, dysphoria, delerium, nausea, vomiting, rhinorrhea, nervousness, restlessness, anxiety, muscle tension, muscle pains and ushed face. Some of these same symptoms have been reported following excess intake of caeine. The prevalence of symptoms reported on withdrawal in dierent studies also covers a wide range from 11% or less to 100%. It is suggested that the evidence leads to the conclusion that non pharmacological factors related to knowledge and expectation are the prime determinants of symptoms and their reported prevalence on withdrawal of caeine after regular consumption. # 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Acquired tolerance; Adaptive changes; Alcohol; Binge mode; Cocaine; Double-blind; Heroin; Locomotor activity; Non-pharmacological factors; Observer rating; Prevalence of symptoms; Prospective clinical type study; Questionnaires; Retrospective survey study; Tolerance; Withdrawal symptoms

1. Introduction Tolerance, withdrawal symptoms and dependence will be discussed. They are related phenomena of wide occurrence in pharmacology. Tolerance is the reduced eect of an agent that often results from regular administration of the agent over a period of time, a few days or a week or two, depending on the agent, so that larger doses are needed to produce the original eect. Withdrawal symptoms are symptoms that can occur when such regular administration is discontinued. Withdrawal symptoms often occur when tolerance has developed, but the quantitative relationship between degree of tolerance and consistency of occurrence and intensity of withdrawal symptoms is usually not strong and, again, depends on the agent. There is no universally accepted denition of dependence, but a widely accepted denition is that dependence on an agent is present during regular administration of an agent if discontinuance precipitates withdrawal symptoms (OBrien, 1995).

* Corresponding author. Tel.: +1-508-624-8144; fax: +1-508-6248197. E-mail address: peter_dews@hms.harvard.edu (P.B. Dews).

Dependence is thus dened in terms of withdrawal symptoms. Reasons for the poor relationship between tolerance and withdrawal are clear if tolerance is recognized as an adaptive phenomenon. Adaptive changes are usual when physiological systems are subjected to continuing external inuence. As an example, consider the adaptive changes that take place when a subject moves from sea level to an altitude of 5000 m and stays there. The partial pressure of oxygen at 5000 m is about 80 mmHg, down from 150 mmHg at sea level. A partial pressure of 80 mmHg is too low for full oxygenation of blood as it passes through the lungs. Adaptive changes occur. For example, respiration increases, cardiac output increases and the concentration of red cells in the blood is increased by cellular proliferation in bone marrow. The rst two changes can take place quite quickly, but the last is slower. When the subject returns to sea level the changes in respiration and circulation can regress rather quickly, although if the stay at 5000 m has been prolonged and the heart has hypertrophied that takes some time to regress. The return of the concentration of red cells to normal takes many days. During that time, the subject has blood with a higher viscosity than normal, making circulation more dicult: withdrawal symptoms

0278-6915/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0278-6915(02)00095-9

1258

P.B. Dews et al. / Food and Chemical Toxicology 40 (2002) 12571261

from being withdrawn from high altitude if you will. The homology to adaptation to regularly administered chemical agents is obvious. The regular administration of the agent produces changes in the physiological systems aected by the agent. Adaptive changes take place to reduce the changes caused by the agent. When administration of the agent is discontinued the countervailing adaptations are unopposed and the result can be symptoms. Agents modify, directly and indirectly, more than one physiological system, and the adaptive changes in the dierent systems will dier in kind and eectiveness and in the rate at which they disappear when administration of the agent is discontinued. As with altitude, the degree of tolerance seen for the dierent systems can vary, and the frequency, nature and intensity of withdrawal symptoms related to the dierent physiological systems can likewise vary. It is thus unreasonable to expect to be able to make generally applicable quantitative statements about tolerance to an agent. Even when attention is focused on a single eect of an agent, it is logistically dicult to generate a pharmacologically satisfactory quantitative account of tolerance. In classical studies of drug antagonisms (and tolerance is a special case of antagonism, self-antagonism) quantitative assessments often could be made on isolated tissues at the rate of one determination every 2 or 3 min. To obtain doseeect curves for tolerance, rst tolerance must be developed by exposure to agent, a matter commonly of 1 week or more, and then the determination of changes in eects of acute doses to assess tolerance can be made usually only at the rate of one per day, at best. The completion of the matrix of dierent levels of exposures and doseeect curves at each level thus is a formidable undertaking, certainly for a drug with the pharmacokinetics of caeine: and does not appear to have been done even in outline. For completeness, all the determinations should be repeated for dierent initial periods of developed tolerance: say, a week, a month, a year, 3 years. . .. While the demonstration that tolerance can develop to a particular eect of an agent is relatively easy and has been done innumerable times, the demonstration lacks a quantitative context. It should be noted that although the phenomenon of tolerance has fascinated many pharmacologists and others, and much ne work has been done, the pressure of the clinic for quantitative information has been lacking. Clinicians are aware of the existence of tolerance, but dosage of regularly taken therapeutically important drugs is determined by assessing the desired eects and adjusting dosage to optimize. Whether the adjustment of dosage is made necessary by development of tolerance or by other changes in the patient may be of importance in indicating changes in progress of the disease, a quantitative account of tolerance to the agent would not be much help in diagnosis.

2. Tolerance to caeine That tolerance to caeine can develop in some species and under some regimes is beyond doubt. As an example, Finn and Holtzman (1986) exposed rats to more than 50 mg/kg per day of caeine, obtained from the drinking water. After a week, they started giving the rats a series of doses of caeine from 3 to 100 mg/kg assessing the eects on locomotor activity. While in rats that had not been drinking caeine, 3 mg/kg caused about a 50% increase in locomotor activity, no dose caused an increase of more than about 25% in the caffeine drinking rats. Thus the tolerance to caeine was great and was not surmountable by increasing the dose. Intake of 50 mg/kg per day is about half a log unit above what the heaviest human users obtain from dietary sources. Tolerance to some behavioral eects of caeine in human users in the range that people consume caeine is widely assumed to occur but has been surprisingly little documented. It was shown many years ago that 150 mg caeine taken 3040 min before retiring delayed sleep in non-coee drinkers but not in coee drinkers, but the dierence between the groups was small (Colten et al., 1968). The dierence could be related to factors determining whether an individual chose to be a coee drinker and not to tolerance to caeine due to the regular consumption of coee. In any case, there seem to have been no subsequent studies that challenge the conclusion of the authors of the study that tolerance to [behavioral eects of] caeine in man appears to be of low magnitude. For a more recent account of the subject, see Fredholm et al. (1999).

3. Withdrawal symptoms A review of caeine dependence studies by Griths and Woodson (1988) listed a wide variety of withdrawal symptoms that had been reported when consumers abruptly discontinued caeine consumption, such as headache, irritability, sleepiness, lethargy, mental confusion, insomnia, psychomotor impairment, hand or limb tremor, weakness, dysphoria, delirium, nausea, vomiting, rhinorrhea, nervousness, restlessness, anxiety, muscle tension, muscle pains and ushed face. It is interesting that some of these same symptoms have also been reported in the opposite situation, that of excess consumption of caeine, namely mental confusion, insomnia, tremor, nausea, nervousness, restlessness and ushed face (DSM IV). The symptoms are reported by the subject, and often little or nothing can be conrmed by physical examination by a physician. Not only are reported withdrawal symptoms variegated but the frequency of occurrence reported by subjects in dierent studies also varies widely. There are two general ways of investigating withdrawal phenomena.

P.B. Dews et al. / Food and Chemical Toxicology 40 (2002) 12571261

1259

One way is to survey a sample of individuals who regularly drink caeinated beverages and ask them whether they have ever omitted their beverage and, if so, whether they had symptoms and, if so, what they were. This is a retrospective survey study. The second way is to select a sample of regular drinkers and switch them to a caeine-free diet, usually striving to make the switch double-blind; that is, such that neither subject nor rater know when the switch is made. This is a prospective clinical type study. Both approaches have been used to assess caeine withdrawal. The following paragraphs discuss representative studies and do not constitute an exhaustive review. A large survey of daily caeine consumers was conducted by telephone by Harris Laboratories of Lincoln, NE (Dews et al., 1999). Subjects were asked do you consume caeinated beverages (e.g. coee, tea, colas) on a regular daily basis?: some 6839 of 11,211 subjects queried answered armatively. These subjects were further asked: Have you had problems or symptoms on stopping caeine in the past? Some 752 (11%) self reported withdrawal symptoms. In a questionnaire study by Greden et al. (1978) on 83 hospitalized subjects, 18 were considered high consumers of caeine ( > 750 mg/day): two (11%) reported they got a headache if they omitted their morning coee. In a later study, also on hospitalized subjects, and directed specically at withdrawal headache, 205 were questioned (Greden et al., 1980). Forty-two (20%) checked headache on abstention from coee (53 did not know). Another questionnaire study was conducted on the wives of 183 graduate students by Goldstein and Kaizer (1969). One item on questionnaire was: How do you think you would feel if you did not have any coee at all in the morning? While only 11 (6%) responded yes to Get a headache, 70 (38%) responded armatively to Feel half awake A questionnaire was completed by 135 psychiatric patients in a study by Winstead (1976). Of 34 patients estimated to be consuming more than 500 mg/day caeine, average of 6.7 cups coee per day, 10 (23%) reported somatic symptoms on withdrawal, but 26 (76%) reported anxiety. In this series of survey studies, the prevalence of symptoms on abstinence from caeine ranged from 11 to 76%. The range of reported incidence of symptoms in prospective clinical trials is even greater than in the survey studies. In a study on medical students, Goldstein (1964) deprived subjects of caeine-containing beverages from lunchtime for the rest of the day. Then, 30 min before retiring, they took a capsule containing either 150 mg caeine or lactose. Next morning, 18 h after their last caeine for lactose subjects and 8 h after for caeine subjects, they were asked if they had a headache. Among regular consumers of 2 or more cups of coee per day, 14 of 67 (21%) reported headache

after lactose the night before but only four out of 74 (5%) after caeine the night before. The survey study previously mentioned on 11,211 subjects was followed by a controlled experiment, also conducted by Harris, on some 55 of the subjects who had reported symptoms on withdrawal from caeine. When decaeinated coee was substituted for caeinated coee, blindly, only ve of 18 (28%) actually reported symptoms on withdrawal in spite of the fact that all 18 had reported that they had had withdrawal symptoms in the past. The self-reported withdrawal symptoms included sleepy, drowsy, no energy, shaky, nervous, insomnia, irritable, stressed, nausea, diarrhea, stomach upset and headache. van Dusseldorp and Katan (1990) report that 38 of 45 (84%) subjects in a double-blind withdrawal study did not realize when their coee was switched to decaeinated, but 19 (42%) reported more headache while only ve (11%) reported less when on decaeinated coee. Out of 99 subjects self-diagnosed as caeinedependent 11 were selected and completed a 2-day study, being given caeine on one day and not the other, blindly (Strain et al., 1994). Nine of the 11 (82%) gave descriptions of symptoms that were diagnosed as due to withdrawal. Of the 11 subjects, seven had coee and four had soft drinks as their usual source of caffeine. The two subjects who had no symptoms on withdrawal were coee drinkers, despite the fact that coee is a richer source of caeine than soft drinks. For a study of experimental headache, Dreisbach and Pfeier (1943) recruited 22 subjects, including ve who had had typical periodic migraine headaches for at least 2 years. They were asked to discontinue caeine beverages and were given capsules containing caeine or lactose. They took the same number of capsules daily but the proportion containing caeine was progressively increased until after 6 or 7 days they were taking around 780 mg/day caeine. When the caeine capsules were replaced by lactose only capsules, 14 (82%) reported headache. Other symptoms reported were nausea, vomiting, rhynorrhea, mental depression, drowsiness, yawning and disinclination to work. Some of the headache, nausea and vomiting could have been due to migraine. All seven subjects in a study by Griths et al. (1986) reported symptoms on withdrawal (100%), as did all 62 subjects in a study by Silverman et al. (1992). In this study, 32 (52%) reported headache compared with 2% under baseline conditions. In a non-blind study by Naismith et al. (1970) all 20 subjects (100%) reported withdrawal symptoms. In studies of post-operative headache in subjects withdrawn from caeine as part of pre- operative fasting, of 233 patients 22% of coee drinkers reported symptoms but only 7% of non-caeine drinkers (Weber et al., 1993). Finally, in a couple of studies that involved regular dosing of caeine and then abrupt withdrawal, but where the object of the study was physiological

1260

P.B. Dews et al. / Food and Chemical Toxicology 40 (2002) 12571261

measures, for example, cardiovascular, rather than reported symptoms both Robertson et al. (1981) in 18 subjects and Ammon et al. (1983) in 10 subjects reported no complaints on withdrawal. Thus the incidence of withdrawal symptoms in the clinical studies ranged from 22% (0% in studies not on symptoms of withdrawal) to 100% in a study when subjects knew when they were withdrawn. Studies have been made to obtain more objective assessment of performance decline in caeine withdrawal, for example on psychomotor tests (Phillips-Bute and Lane, 1997; Lane and Phillips-Bute, 1998). In general, the declines have been surprisingly small and nonsignicant, even at times when rating scales showed the changes noted above. One test that has shown signicant declines is a vigilance task, in which the proportion of targets detected declined and reaction time increased (Lane and Philips-Bute, 1998). Unfortunately, the results are presented as a composite of the 30 subjects so it is not possible to see how many subjects were measurably aected or how large the eect was in individual subjects. There are many dierences between these dierent studies. The subjects ranged from hospitalized psychiatric patients to healthy people going about their daily business. Reports on symptoms were obtained by questionnaires completed by subject, by observer rating and by telephone. The focus of the studies ranged from withdrawal symptoms in general to particular interest in headache. The subjects in some studies were recruited without regard to reported history of withdrawal symptoms; other studies recruited subjects who reported a history of withdrawal symptoms, even dependence. All these factors must contribute to the variability of symptoms and their prevalence. Even more striking, however, is the high prevalence reported, 100%, in the non-blind study compared to the low prevalence (0%) in studies not directed towards withdrawal. Less extreme, but still striking, are the dierences between prevalence reported from studies conducted in laboratories identied as concerned with caeine withdrawal problems in the literature and by the institutional press oce, the studies of Griths et al. (1986) (100%), Silverman et al. (1992), (100%) and Strain et al. (1994) (82%), and the report from the Harris study (Dews et al., 1999) (28%), within the range of the noise level. Harris Laboratory had not been linked to caeine withdrawal studies and so subjects had no reason to expect that such a study was being conducted. Workers agree that coee with caeine has discernible eects and therefore can be distinguished from the non-caeinated state, so that double-blind cannot be maintained in knowledgeable subjects with some focus on withdrawal. The most parsimonious explanation of much of the variability is that caeine withdrawal causes a subtle syndrome, close to background noise with symptoms

and prevalence primarily determined by non-pharmacological factors. Such a conclusion is by no means novel, see for example, Smith et al. (1993). Consider the following example. In the review of caffeine dependence by Griths and Woodson (1988), only one of eight reports published prior to 1943 mentioned headache. Following Dreisbach and Pfeiers 1943 description of severe headache on withdrawal, some 10 of 16 studies reported headaches on withdrawal. Once headache had been highlighted as a consequence of withdrawal, reports of headache on withdrawal became more frequent. There are anecdotal accounts of severe caeine withdrawal symptoms with prostration. Until a reasonable number of individuals have been documented as exhibiting severe caeine withdrawal symptoms under controlled conditions, the evidence will remain anecdotal. Severe reactions were reported to ingestion of aspartame, including epileptiform convulsions, blindness and incapacitating nervousness. In the light of the vast experience of aspartame since, the anecdotal reports no longer command credence. The demographics of reports of caeine withdrawal symptoms are interesting. In the Harris survey of 11,211 subjects (Dews et al., 1999) 5.5% of women, but only 0.9% of men complained of symptoms upon withdrawal that interfered with their normal daily activities (although, as was subsequently found, as many as 2/3 of even this population may have failed to report symptoms in a clinical trial). Such a high ratio of women to men is not generally seen for complaints not related to physical sex characteristics, nor is it seen for withdrawal of other agents such as heroin or phenytoin, which are prone to cause withdrawal symptoms. It has been seen, however, in complaints of noxebo eects after consumption of behaviorally inert substances such as aspartame. (Noxebo is a neologism coined to designate the opposite of placebo. A placebo pleases, a noxebo indisposes, both without pharmacogical basis). Of the 4000 complaints of symptoms after consumption of aspartame received by the FDA between 1985 and 1993, 76% were made by women. The reports of caeine withdrawal symptoms continuing for days, weeks or even months raise the possibility that some individuals may miss the enhancement of performance they customarily receive at appropriate times from consumption of caeine (Smith et al., 1993). They report their lower (non-caeine) level of performance at times as continuance of withdrawal symptoms. Obviously, such a state of plain normalcy scarcely qualies as a withdrawal syndrome. The foregoing discussion is not intended to make the case that caeine withdrawal symptoms do not occur but rather to put them in the context of symptoms and complaints by people in the course of their everyday lives.

P.B. Dews et al. / Food and Chemical Toxicology 40 (2002) 12571261

1261

4. Caeine dependence According to the denition of dependence given in the Introduction, the acceptance that there can be caeine withdrawal symptoms signies that there can be dependence on caeine. In contemporary consideration of maintenance of abuse of drugs, withdrawal symptoms are not accorded the importance they once were. Formerly it was believed that the forestalling of withdrawal symptoms was a prime mover in sustaining abuse of drugs. There are drugs of abuse, such as heroin, that produce a clear withdrawal syndrome, which can be avoided by regular use. In contrast, many cocaine or alcohol abusers use their drug compulsively, but intermittently in a binge mode. That is, intensive abuse over a day or few days followed by voluntary abstention. They typically experience only minimal withdrawal symptoms consisting of tiredness when they abstain from cocaine, or insomnia and nervousness when the alcohol binge is over. Consideration of the abuse of such drugs leads to the conclusion that a pattern of abuse can be just as strongly maintained when abstinence symptoms are inconsequential, as with cocaine, or relatively rare, as with alcohol. Forestalling delerium tremens is not believed to be a prime factor in maintaining the pattern of drinking of most alcoholics. Conversely, there are many therapeutic agents such as antihypertensive medications that produce tolerance and, if stopped abruptly, lead to withdrawal symptoms, but are not abused. So, dependence, as dened, and abuse should not be associated, and discussing caeine in terms of drugs of abuse trivializes the dangers of such drugs as cocaine.

References
Ammon, H.P.T., Bieck, P.R., Mandalaz, D., Verspohl, E.J., 1983. Adaptation of blood pressure to continuous heavy coee drinking in young volunteers. A double-blind crossover study. British Journal of Clinical Pharmacology 15, 701706. Colten, T., Gosselin, R.E., Smith, R.P., 1968. The tolerance of coee drinkers to caeine. Clinical Pharmacology and Therapeutics 9, 3139. Dews, P.B., Curtis, G.L., Hanford, K.J., OBrien, C.P., 1999. The frequency of caeine withdrawal in a population-based survey and in a controlled, blinded experiment. Journal of Clinical Pharmacology 39, 12211232. Dreisbach, R.H., Pfeier, C., 1943. Caeine withdrawal headache. Journal of Laboratory and Clinical Medicine 28, 12121219. Finn, I.B., Holtzmann, S.G., 1986. Tolerance to caeine-induced stimulation of locomotor activity in rats. Journal of Pharmacology and Experimental Therapeutics 238, 542546.

Fredholm, B.B., Battig, K., Holmen, J., Nehlig, A., Zvartau, E.E., 1999. Actions of caeine in the brain with special reference to factors that contribute to its widespread use. Pharmacological Reviews 51, 83133. Goldstein, A., 1964. Wakefulness caused by caeine. Naunyn Schmiedebergs Archives fur Experimentalle Pathologie und Pharmakologie 248, 269278. Goldstein, A., Kaizer, S., 1969. Psychotropic eects of caeine in man. III. A questionnaire survey of coee drinking and its eects in a group of housewives. Clinical Pharmacology and Therapeutics 10, 477488. Greden, J.F., Fontaine, P., Lubetsky, M., Chamberlin, K., 1978. Anxiety and depression associated with caeinism among psychiatric inpatients. American Journal of Psychiatry 135, 963966. Greden, J.F., Victor, B.S., Fontaine, P., Lubetsky, M., 1980. Caeinewithdrawal headache: A clinical prole. Psychosomatics 21, 411 413, 417418. Griths, R.R., Bigelow, G.E., Liebson, I.A., 1986. Human coee drinking: reinforcing and physical dependence producing eects of caeine. Journal of Pharmacology and Experimental Therapeutics 239, 416425. Griths, R.R., Woodson, P.P., 1988. Caeine physical dependence: a review of human and laboratory animal studies. Psychopharmacology 94, 437451. Lane, J.D., Phillips-Bute, B.G., 1998. Caeine deprivation aects vigilance performance and mood. Physiology and Behavior 65, 171175. Naismith, D.J., Akinyanju, P.A., Szanto, S., Yudkin, J., 1970. The eect in volunteers of coee and decaeinated coee on blood glucose, insulin, plasma lipids and some factors involved in blood clotting. Nutrition and Metabolism 12, 144151. OBrien, C.P., 1995. Drug addiction and drug abuse. In: Hardman, J.G., Limbird, L.E. (Eds.), Goodman and Gilmans The Pharmacological Basis of Therapeutics, ninth ed.. Macmillan, New York, pp. 557577. Phillips-Bute, B.G., Lane, J.D., 1997. Caeine withdrawal symptoms following brief caeine deprivation. Physiology and Behavior 63, 3539. Robertson, D., Wade, D., Workman, R., Woosley, R.L., Oates, J.A., 1981. Tolerance to the humoral and hemodynamic eects of caeine in man. Journal of Clinical Investigation 67, 11111117. Silverman, K., Evans, S.M., Strain, E.C., Griths, R.R., 1992. Withdrawal syndrome after the double-blind cessation of caeine consumption. New England Journal of Medicine 327, 11091114. Smith, A.P., Brockman, P., Flynn, R., Maben, A., Thomas, M., 1993. Investigation of the eects of coee on alertness and performance during the day and night. Neuropsychobiology 27, 217223. Strain, E.C., Mumford, G.K., Silverman, K., Griths, R.R., 1994. Caeine dependence syndrome: Evidence from case histories and experimental evaluations. Journal of the American Medical Association 272, 10431048. van Dusseldorp, M., Katan, M.B., 1990. Headache caused by caeine withdrawal among moderate coee drinkers switched from ordinary to decaeinated coee: a 12 week double blind trial. British Medical Journal 300, 15581559. Weber, J.G., Ereth, M.H., Danielson, D.R., 1993. Perioperative ingestion of caeine and postoperative headache. Mayo Clinic Proceedings 68, 842845. Winstead, D.K., 1976. Coee consumption among psychiatric inpatients. American Journal of Psychiatry 133, 14471450.