Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Caffeine physical dependence and withdrawal in humans has been documented in numerous case reports and experimental studies (cf. Griffiths and Woodson, 1988a; Griffiths and Mumford, 1995). The most common withdrawal symptoms include increases in headache, drowsiness, and work difficulty (including impaired concentration) and decreases in feelings of contentment and sociability (e.g., Griffiths et al., 1990a; Silverman et al., 1992; Hughes et al., 1993; Strain et al., 1994; cf. Griffiths and Mumford, 1995). Although severity of caffeine withdrawal can vary within and across individuals, it can produce clinically significant functional behavioral impairment (Silverman et al., 1992; Strain et al., 1994; Streufert et al., 1995) and has been implicated in a weekend migraine headache syndrome (Couturier et al., 1992) and postoperative headache (Galletly et al., 1989; Fennelly et al., 1991; Weber et al., 1993). There is convincing evidence that caffeine withdrawal symptoms enhance the reinforcing effects of caffeine and thus are integrally related to the maintenance of long-term patterns of caffeine self-administration (e.g., Griffiths et al., 1986a; Hughes et al., 1993; Evans et al., 1994; Richardson et al., 1995; Rogers et al., 1995; Schuh and Griffiths, 1997; Garrett and Griffiths, 1998).

Although caffeine withdrawal has been well documented, prospective parametric studies of dosing conditions for inducing physical dependence are almost nonexistent. The present set of experiments was undertaken to examine the effects on caffeine withdrawal of manipulating several major caffeine dosing parameters (within-day dosing frequency, maintenance dose, magnitude of dose reduction, and duration of dosing).

Although previous studies have demonstrated caffeine withdrawal occurs under widely varying within-day dosing conditions [e.g., daily dose divided into either 2 (Garrett and Griffiths, 1998) 3 (Evans and Griffiths, 1992), or 10 (Griffiths et al., 1990a) equal portions over the day], the effect of varying within-day dosing pattern has not been studied systematically, and, to our knowledge, no study has examined whether caffeine withdrawal occurs after once-a-day dosing. Thus, the first experiment was designed to explore the role of within-day caffeine dosing frequency by examining caffeine withdrawal in participants who were maintained on 300 mg of caffeine each day either as a single dose in the morning or as three 100-mg doses spaced 4 to 6 h apart over the day.

To our knowledge, there have been no prospective experimental studies of the effects of caffeine maintenance dose on caffeine withdrawal. Although previous studies have shown that the incidence of caffeine withdrawal increased as a function of daily self-reported caffeine dose (e.g., Goldstein and Kaizer, 1969; Goldstein et al., 1969; Galletly et al., 1989; Fennelly et al., 1991; Weber et al., 1993), these studies provide only suggestive evidence that caffeine withdrawal is related to the amount of caffeine consumed because caffeine dose was not manipulated directly, but was self-selected by the individuals themselves. The second experiment examined the role of daily caffeine dose on caffeine withdrawal by maintaining subjects on different doses of caffeine (100, 300, and 600 mg/day) before withdrawal.

Another dosing parameter that has remained unstudied is the magnitude of the caffeine dose required to prevent (i.e., suppress) caffeine withdrawal in individuals already physically dependent on caffeine. Therefore, the third experiment was designed to determine the extent to which lower doses of caffeine would suppress or reduce caffeine withdrawal symptoms in individuals chronically maintained on a constant dose of caffeine. This was assessed by maintaining individuals on 300 mg/day caffeine throughout the study and periodically substituting a range of lower caffeine doses (200, 100, 50, 25, and 0 mg/day).

Finally, no previous study has systematically explored the effect of duration of daily caffeine exposure on subsequent caffeine withdrawal. Although it is well established that long-term daily exposure (e.g., for a period of months) to moderate or high doses of caffeine can result in subsequent caffeine withdrawal, two studies suggested that caffeine withdrawal can occur after relatively short-term (i.e., 6-15 days) exposure to caffeine (Dreisbach and Pfeiffer, 1943; Griffiths et al., 1986a). The fourth experiment was designed to determine directly the minimum duration of chronic caffeine exposure necessary for the expression of caffeine withdrawal. This was assessed by maintaining individuals on placebo throughout the study and intermittently switching them to caffeine (300 mg/day) for periods of 1, 3, 7, or 14 days, followed by a return to placebo for 1 week.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects

Volunteers were recruited from the Francis Scott Key Medical Center campus via posters and flyers. All participants were medically healthy based on medical history, a complete physical examination, electrocardiogram, urine toxicology for abused drugs, and, for female participants, a urine pregnancy test. In addition, no one had any current psychiatric disorders including psychoactive substance abuse or dependence (other than nicotine) based on a psychiatric interview using DSM-III-R criteria (American Psychiatric Association, 1987). All participants had at least a high school education to ensure an adequate understanding of the consent form, questionnaires, and protocol requirements. Individuals with any medical problems contraindicating caffeine consumption, as well as individuals who did not report current daily caffeine consumption, were excluded.

These studies were approved by the Institutional Review Boards of the Francis Scott Key Medical Center and Johns Hopkins University School of Medicine. Before research participation, experimental procedures were discussed with participants and they signed a consent form that outlined the procedures to be used and listed possible side effects of the drugs. Participants were told that the purpose of the study was to evaluate the effects of compounds found in coffee, tea, chocolate, and soda on mood and behavior. To maintain the double-blind and minimize expectancy effects related to caffeine, participants were told that the compounds they might receive included chlorogenic acids, diterpenes, caffeine, tannin, theophylline, and inactive placebo and that the drugs and doses could vary from day to day. Participants were paid for their participation. Before beginning the experimental phase, all participants completed a detailed dietary survey from which daily caffeine intake was calculated for 7 days immediately before starting the study.

Table 1 shows the demographics for participants across the four experiments, which were conducted sequentially over a 4-year period. Different individuals participated in each of the four experiments: 15, 17, 19, and 25 participants completed experiments 1, 2, 3, and 4, respectively. The majority of participants were women, and the average age was between 28 and 30 years across the experiments. The average prestudy caffeine consumption ranged from 241 mg/day in experiment 1 to 294 mg/day in experiment 3 and was similar to estimated daily average caffeine consumption of 280 mg per adult consumer in the United States (Barone and Roberts, 1996).

Dietary Restrictions

In all experiments, participants were instructed to abstain completely from caffeinated and decaffeinated coffee, caffeinated and decaffeinated tea, all chocolate-containing products, caffeinated soft drinks, and various over-the-counter medications for the full duration of the study (ranging between 54 days in experiment 3 to 73 days in experiment 2). Participants were given a detailed list of allowed and restricted beverages and over-the-counter medications. Participants were instructed to report any use of prescription or nonprescription medications, except those previously approved by the investigator (e.g., oral contraceptives). Participants were permitted alcohol in moderation after the evening capsule until bedtime, and cigarette smoking was not restricted. To verify compliance with the dietary restrictions, saliva samples were collected on an unannounced quasi-random basis and analyzed for caffeine. Participants were told that failure to comply with the dietary restrictions could result in discontinuation from the study and/or loss of study payments. Overall, data from seven participants (8.4% of those who completed an experiment) were excluded from the data analyses because of noncompliance (Table 1).

Daily Experimental Procedures

In all experiments, participants reported to the research unit in the morning (Monday through Friday) between 7:00 and 10:00 AM. As described below, they filled out a caffeine withdrawal questionnaire and the Profile of Mood States (POMS) questionnaire each morning (before the first capsule of the day). They then ingested a single capsule under the supervision of a research assistant and nurse and left the research unit. Participants returned to the research unit 4 to 6 h later at midday (12:00 PM) to fill out the two questionnaires and ingest another single capsule. Before leaving the unit at midday, participants were given two more sets of questionnaires to be filled out at 4:00 and 8:00 PM (in experiment 1, participants also were given another single capsule in a blister pack for 4:00 PM). Participants were instructed to fill out the third set of questionnaires and take the third capsule (experiment 1 only) approximately 4 to 6 h after the midday capsule (4:00 PM); the last set of questionnaires was to be filled out at approximately 8:00 PM. Participants recorded the times at which they filled out the forms and ingested capsules. The next morning (with the exception of weekends and holidays) participants returned the 4:00 and 8:00 PM forms. When participants returned to the research unit on Fridays for their midday capsule, they were provided with capsules and questionnaires for the remainder of Friday (4:00 and 8:00 PM) and the weekend. On Monday mornings, participants returned the empty blister packs and the questionnaires. On occasion, when participants knew they would be unable to return to the research unit on a weekday (e.g., doctor's appointment, sick child, funeral, holiday, etc.), they were provided their capsules and questionnaires to take with them. Each participant was issued a digital watch (Casio Data Bank), which permitted setting of daily, multiple alarms to remind the participant when to take capsules, complete questionnaires, and provide saliva samples.

Subjective-Effects Questionnaires

In all four experiments, participants completed a caffeine withdrawal questionnaire four times each day [approximately 8:00 AM (before the first capsule of the day), 12:00 PM, 4:00 PM, and 8:00 PM]. The 26 items were selected based on previous research (Griffiths et al., 1990a,b) as representing dimensions that are sensitive to caffeine withdrawal. In experiment 1, the caffeine withdrawal questionnaire had an additional seven items, but these were not used in the data analysis. Participants rated each of the 26 items on a 4-point scale from "not at all" (0) to "very much" (3), based on how they felt at the present time. The items included: 1) irritable/cross/grumpy, 2) alert/attentive/observant, 3) lightheaded/dizzy, 4) upset stomach, 5) well being, 6) blurred vision, 7) desire to socialize/talkativeness, 8) anxious/nervous, 9) urge to do task-/work-related activities, 10) drowsy/sleepy, 11) ability to concentrate, 12) cerebral fullness, 13) muscle pain or stiffness, 14) yawning, 15) energy/active, 16) runny nose, 17) jittery/shaky, 18) depressed, 19) lethargy/fatigue/tired/sluggish, 20) muzzy/foggy/not clear-headed, 21) content/satisfied, 22) headache, 23) flu-like feelings, 24) sweating, 25) self-confidence, and 26) limb tremor.

In all four experiments, participants also completed the 65-item POMS questionnaire four times daily [approximately 8:00 AM (before the first capsule of the day), 12:00 PM, 4:00 PM, and 8:00 PM]. The POMS questionnaire is an adjective-rating questionnaire generally considered to be a standardized mood-state inventory (McNair et al., 1971) and has been shown previously to be sensitive to the effects of caffeine and caffeine withdrawal (Griffiths et al., 1990a,b; Evans and Griffiths, 1992; Silverman et al., 1992). Participants rated each of the 65 items on a 5-point scale from "not at all" (0) to "extremely" (4), based on how they felt at the present time. Seven empirically derived scales were scored for the 65-item version of the POMS: Tension-Anxiety, Depression-Dejection, Anger-Hostility, Vigor, Fatigue, Confusion-Bewilderment, and Friendly. One additional factor, Total Mood Disturbance, was calculated as a composite factor.

Capsule Preparation and Dosing

For all experiments, caffeine and placebo capsules (size 0 gelatin) were prepared from combinations of caffeine-anhydrous (United States Pharmacopeia) and powdered lactose. The color of the capsules varied across participants, but the capsule color for a given participant remained the same throughout the study. Placebo and caffeine capsules were identical in appearance. In experiment 1 participants received one capsule three times each day and in experiments 2 through 4 they received one capsule two times each day. Capsules were ingested with approximately 150 ml of water in the presence of a nurse or research assistant, with the exception of the 4:00 PM (experiment 1) and weekend capsules. Participants, nurses, and research assistants were blind to drug conditions.

Saliva Samples

On an unannounced quasi-random schedule, participants provided a 5-ml saliva sample in the morning before taking the first capsule of the day and again at 4:00 PM by expectorating directly into a test tube. On weekends and holidays participants were provided saliva tubes and labels in the event that samples were required; participants opened an envelope in the morning (before filling out the morning forms) that contained a piece of paper with "please give saliva samples today" or "no saliva samples today." Participants were instructed to freeze or refrigerate saliva samples over the weekend and return them on Monday morning. All saliva samples were frozen and stored. Caffeine concentrations in selected samples were analyzed (Labstat Incorporated, Kitchener, Ontario, Canada) using gas chromatography with nitrogen-phosphorous detection and capillary column (modified from Jacob et al., 1981) with 5-methylcotinine as the internal standard for caffeine. If saliva samples indicated noncompliance with the dietary restrictions, all data from the participant were excluded from the analyses. The criterion for compliance was that, for a given individual, the caffeine concentration from a saliva sample after 2 days of placebo administration was 25% or less than the caffeine concentration from a saliva sample taken on the morning after receiving 300 mg of caffeine the previous day. There was no evidence of noncompliance in experiments 1 and 2; data from three participants in experiment 3 and from four participants in experiment 4 were excluded because of noncompliance.

Individual Experiments

All four experiments used a within-subject, double-blind, crossover design with the order of dosing conditions in each experiment counterbalanced across participants.

Experiment 1: Interval of Caffeine Dosing. Throughout the 63-day study, 15 participants ingested three capsules daily spaced 4 to 6 h apart (e.g., 8:00 AM, 12:00 PM, and 4:00 PM). Participants were maintained on 300 mg/day caffeine in capsules either as a single 300-mg dose in the morning and two placebo capsules in the afternoon or as 100-mg capsules three times a day. On the first day of the experiment, all participants received 100 mg of caffeine three times. When participants were exposed to the single 300-mg dose in the morning, the morning dose of caffeine was increased by 50 mg/capsule (with a corresponding decrease in the caffeine dose of the other two daily capsules) over a period of 4 days. To assess caffeine withdrawal, participants were administered caffeine for 5 to 9 consecutive days followed by the substitution of placebo capsules for 2 consecutive days. Participants then were returned to caffeine maintenance for 5 to 9 consecutive days. This sequence was repeated two more times, such that participants were exposed to three 2-day placebo substitutions at each dosing interval. At this time, participants were switched to the other dosing interval (i.e., either 300 mg in the morning or 100 mg three times a day) and were maintained on the second dosing interval for 5 to 9 consecutive days before a 2-day placebo substitution. As with the first dosing interval, participants were exposed to three 2-day placebo substitutions.

Experiment 2: Caffeine Maintenance Dose. Throughout the 73-day study, 17 participants ingested two capsules daily spaced 4 to 6 h apart (e.g., 8:00 AM and 12:00 PM). Participants were maintained on either 100, 300, or 600 mg/day of caffeine given in two equally divided doses each day. On the first day of the experiment all participants received 50 mg of caffeine two times (100 mg/day). For those participants assigned to one of the higher maintenance doses first, the dose of caffeine was increased by 50 mg/capsule (100 mg/day) to the scheduled maintenance dose. The scheduled maintenance dose was maintained for at least 7 consecutive days before placebo was substituted for 2 consecutive days. Participants then were returned to that caffeine maintenance dose for at least 7 consecutive days. Then, participants were exposed to another 2-day placebo substitution. This sequence was repeated for the other two caffeine maintenance doses such that each participant was exposed to two 2-day placebo substitutions at each maintenance dose. Between dose conditions, the dose of caffeine was gradually increased or decreased by 100 mg/day (50 mg/capsule) until the new maintenance dose was reached.

Experiment 3: Suppression of Caffeine Withdrawal. Throughout the 54-day study, 19 participants ingested two capsules daily spaced 4 to 6 h apart (e.g., 8:00 AM and 12:00 PM). Initially, participants were maintained on 300 mg/day of caffeine given in two equally divided doses each day. Approximately every 7 days, a lower dose of caffeine (300, 200, 100, 50, and 25 mg/day) or placebo was substituted for 2 consecutive days. Within a given day, the substitution dose was divided equally between the two capsules. Participants then were returned to the caffeine maintenance dose of 300 mg/day for at least 7 consecutive days before participants were exposed to another 2-day dose substitution. This sequence was repeated such that each participant was exposed to all six dose substitutions.

Experiment 4: Duration of Caffeine Exposure. Throughout the 60-day study, 25 participants ingested two capsules daily spaced 4 to 6 h apart (e.g., 8:00 AM and 12:00 PM). For the first 7 days, all participants were maintained on placebo in capsules given twice a day. After this placebo "washout" week, participants were administered caffeine (300 mg/day given in two equally divided doses) for either 1, 3, 7, or 14 consecutive days. Each duration of caffeine exposure was followed by 7 consecutive days of placebo, and each participant was exposed to all caffeine durations.

Data Analysis

Cluster Analysis. The 26 items of the caffeine withdrawal questionnaire were subjected to a hierarchical cluster analysis. This was done to reduce the number of measures and facilitate the analyses because the symptoms and severity of caffeine withdrawal for any single dimension can vary both within and across individuals (see Griffiths et al., 1990a; Griffiths and Mumford, 1995). All participants across the four experiments were exposed to a condition in which they were maintained on 300 mg/day caffeine followed by a 2-day (or more) placebo substitution. Thus, a total of 76 individuals were included in the cluster analysis. For experiments 1 through 3, the peak ratings over the 2 caffeine days immediately preceding placebo substitution were compared with the peak ratings over the 2 placebo days for each item separately, using the first exposure to a 300-mg/day caffeine and 2-day placebo substitution. For experiment 4, the peak ratings over the last 2 days of the 14 consecutive 300-mg/day caffeine condition were compared with the peak ratings over the first 2 days of placebo after the 14-day caffeine condition. Peak effect over 2 days was used because maximal withdrawal symptoms occur within 24 to 48 h after the cessation of caffeine (e.g., Dreisbach and Pfeiffer, 1943; Griffiths et al., 1990a; Evans and Griffiths, 1992). Peak effect for each item was based on the expected direction during caffeine withdrawal. The maximum value was calculated for items 1, 3, 4, 6, 8, 10, 12, 13, 14, 16, 17, 18, 19, 20, 22, 23, 24, and 26. The minimum value was calculated for items 2, 5, 7, 9, 11, 15, 21, and 25.

Experiment 1: Interval of Caffeine Dosing. A three-factor repeated-measures ANOVA was conducted for each of the five clusters from the caffeine withdrawal questionnaire, headache, and each of the eight subscales of the POMS questionnaire, using the peak ratings (i.e., maximum or minimum values as defined above) over the 2 caffeine days immediately preceding each 2-day placebo substitution and the peak ratings over each 2-day placebo substitution. The three factors were Drug (caffeine versus placebo), Number of daily doses (once a day versus three times a day), and Replication (three replications). Planned comparisons (two-tailed tests) were conducted to examine differences between caffeine and placebo by collapsing data across the number of daily doses and replications.

Experiment 2: Caffeine Maintenance Dose. A three-factor repeated-measures ANOVA was conducted for each of the measures as described above for experiment 1. The three factors were Drug (caffeine versus placebo), Maintenance dose (100 versus 300 versus 600 mg/day), and Replication (two replications). Planned comparisons (two-tailed tests) were conducted to 1) examine differences between caffeine and placebo for each caffeine maintenance dose and 2) examine differences between the placebo substitution at the 600-mg dose and the placebo substitution at the 100-mg dose.

Experiment 3: Suppression of Caffeine Withdrawal. A two-factor repeated-measures ANOVA was conducted for each of the measures using the peak ratings over the 2 caffeine maintenance days immediately preceding each 2-day dose substitution and the peak ratings over each 2-day dose substitution. The two factors were Drug (caffeine maintenance versus dose substitution) and Dose Substitution (300 mg versus 200 mg versus 100 mg versus 50 mg versus 25 mg versus placebo). Planned comparisons (two-tailed tests) were conducted to 1) examine differences between caffeine maintenance and dose substitution for each of the six dose substitutions and 2) examine differences between the placebo dose substitution and each of the other five dose substitutions.

Experiment 4: Duration of Caffeine Exposure. A one-factor repeated-measures ANOVA was conducted for each of the measures. There were five drug conditions in the ANOVA: peak ratings over the last 2 days of the 14 consecutive 300-mg/day caffeine condition and peak ratings over the first 2 days of placebo immediately after each caffeine duration (1, 3, 7, and 14 days). Planned comparisons (two-tailed tests) were conducted to 1) examine differences between caffeine and placebo after each duration of caffeine and 2) examine differences between the placebo days.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Caffeine Withdrawal Across All Participants. Caffeine withdrawal for all 76 participants across the 4 experiments was determined using their initial exposure to placebo after maintenance on 300 mg/day caffeine. Table 2 shows that substitution of placebo for caffeine significantly increased Headache, Headache/Poor Mood, Tiredness, Fatigue (POMS), Confusion-Bewilderment (POMS) and Total Mood Disturbance (POMS) and decreased Activity/Alertness, Vigor (POMS), and Friendly (POMS).

Experiment 1: Interval of Caffeine Dosing. The ANOVA showed no significant differences between the two dosing intervals with regard to caffeine withdrawal symptoms. As shown in Table 2, the planned comparisons between caffeine and placebo, collapsed across the two dosing intervals and the three replications, revealed significant changes on eight scales. Figure 1 shows that compared with days on which participants were maintained on caffeine, substitution of placebo was associated with significant increases in Headache (p < .001), Headache/Poor Mood (p = 0.017), Tiredness (p = 0.003), Flu-Like Symptoms (p = 0.001), Fatigue (POMS; p = 0.036), and decreases in Vigor (POMS; p = 0.025). Activity/Alertness (p = 0.018) and Friendly (POMS; p = 0.043) also showed decreases. Taken together, these results indicate that 300 mg/day caffeine results in a reliable caffeine withdrawal syndrome when placebo is substituted and that caffeine withdrawal occurs whether 300 mg of caffeine is ingested as a single dose in the morning or as three divided doses over the day.

View larger version (44K): [in this window] [in a new window]   Fig. 1.   Effects of substituting placebo for caffeine on selected measures, collapsed across the caffeine-dosing interval (i.e., 100 mg three times/day or 300 mg once/day). Bars show mean peak ratings or scores for the last 2 caffeine days and the corresponding 2-day placebo substitutions; brackets show +1 S.E.M. (n = 15). For each of the six measures, difference between caffeine and placebo was significant (p < .05). Range of possible ratings for Headache and scores for Headache/Poor Mood, Tiredness, and Flu-Like Symptoms is 0 to 4; the range of possible scores for Fatigue (POMS) is 0 to 28 and, for Vigor (POMS), is 0 to 32.

Experiment 2: Caffeine Maintenance Dose. Based on the ANOVA, significant Drug × Dose interactions were obtained on 9 of the 14 scales [Headache, Headache/Poor Mood, Activity/Alertness, Tiredness, Vigor (POMS), Fatigue (POMS), Confusion-Bewilderment (POMS), Friendly (POMS), Total Mood Disturbance (POMS); data not shown]. Consistent with the ANOVA, the planned comparisons shown in Table 2 indicate that relative to data from days on which participants received caffeine, substitution of placebo was associated with significant changes in 5 scales when maintained on 100 mg/day, 6 scales when maintained on 300 mg/day, and 10 of the 14 scales when maintained on 600 mg/day caffeine. In addition, when the placebo substitution after 600 mg/day caffeine was compared with the placebo substitution after 100 mg/day caffeine, there were significant differences in the magnitude of effect, with the highest dose showing greater caffeine withdrawal. Placebo substituted after 600 mg caffeine was significantly different than placebo substituted after 100 mg caffeine for Headache (p < .001), Headache/Poor Mood (p < .001), Depression-Dejection (POMS; p = 0.016), Friendly (POMS; p = 0.007), and Total Mood Disturbance (POMS; p < .001).

View larger version (58K): [in this window] [in a new window]   Fig. 2.   Effects of substituting placebo for different maintenance doses of caffeine on selected measures. Bars show mean peak ratings or scores for the last 2 caffeine days at each caffeine maintenance dose and the corresponding 2-day placebo substitutions; brackets show +1 S.E.M. (n = 17). *, significant difference between caffeine and placebo for a given maintenance dose. , placebo substitution at the 600-mg caffeine dose was significantly different from placebo substitution at the 100-mg caffeine dose (p < .05). Range of possible ratings or scores for all measures is 0 to 4.

Experiment 3: Suppression of Caffeine Withdrawal. The ANOVA showed no overall effects of Drug, but did show significant Dose Substitution effects for Activity/Alertness (p = 0.029), Tiredness (p = 0.002), Fatigue (POMS; p = 0.009), and Total Mood Disturbance (POMS; p = 0.039) and Drug × Dose Substitution interactions for Headache (p = 0.009) and Vigor (POMS; p = 0.026). The planned comparisons between each 2-day dose substitution and the corresponding 2-day period of 300-mg/day caffeine maintenance showed that substitution of lower caffeine doses or no caffeine (placebo) resulted in an increased number of withdrawal symptoms. Table 2 shows that relative to when participants received 300 mg/day caffeine, substitution of either the maintenance dose (300 mg/day caffeine) or 200 mg/day caffeine was not associated with any significant changes in mood. In contrast, substitution of 100, 50, 25, or 0 mg/day caffeine was associated with significant changes in two, one, three, and seven scales, respectively.

View larger version (73K): [in this window] [in a new window]   Fig. 3.   Effects of substituting various doses of caffeine or placebo for caffeine (300 mg/day) on selected measures. Bars show mean peak ratings or scores for the last 2 caffeine days preceding each 2-day dose substitution and the corresponding 2-day placebo substitutions; brackets show +1 S.E.M. (n = 19). *, significant difference between the caffeine maintenance condition and the corresponding dose substitution. a and b indicate comparisons between placebo dose substitution and other dose substitutions; within the same graph, any dose-substitution bar labeled a is significantly different from placebo dose substitution b at p < .05. Range of possible ratings or scores for all measures is 0 to 4.

Experiment 4: Duration of Caffeine Exposure. Based on the ANOVA, significant Drug effects were obtained on 5 of the 14 scales [Headache, Headache/Poor Mood, Physical Symptoms, Tiredness, Fatigue (POMS); data not shown]. Table 2 shows that compared with when participants were maintained on caffeine (300 mg/day), substitution of placebo was associated with significant changes on no scales after a single day of caffeine exposure, and on four or more scales after 3 or more days of caffeine exposure. These findings indicate that caffeine withdrawal can develop when placebo is substituted for caffeine after a brief period of exposure to caffeine, i.e., as little as 3 days.

View larger version (75K): [in this window] [in a new window]   Fig. 4.   Effects of substituting placebo after various durations of caffeine exposure on selected measures. Open columns show mean peak rating or score for the last 2 caffeine days (300 mg/day) of the 14-consecutive-day caffeine condition. Hatched columns show mean peak ratings or scores for the first 2 days of placebo substitution after each caffeine duration, either 1, 3, 7, or 14 days of caffeine. Brackets show +1 S.E.M. (n = 25). *, significant difference between each placebo substitution and caffeine. Letters a, b, and c indicate comparisons among placebo substitutions; within the same graph, any two placebo columns designated with the same letter are not significantly different from each other at p < .05. Range of possible ratings or scores for all measures is 0 to 4.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The present study provides the most comprehensive parametric experimental analysis to date of the dosing conditions necessary for the development of caffeine physical dependence as expressed by caffeine withdrawal symptoms upon abrupt cessation of caffeine administration. A particular strength of the present study is that each of the four parametric experiments was conducted using the same general procedure in a different group of moderate daily caffeine consumers.

Caffeine Withdrawal Symptoms. The present study confirms and extends previous findings regarding the symptoms associated with the cessation of caffeine consumption. The major symptom clusters that were affected significantly in each of the four experiments (cf. Table 2) were increases in Headache, Headache/Poor Mood (including irritability, anxiety, feelings of depression), Tiredness (drowsy, yawning, lethargy), Fatigue (POMS), and Total Mood Disturbance (POMS) and decreases in Activity/Alertness (alertness, well being, desire to socialize, urge to do task-related activities, concentration, energy, satisfaction, self-confidence) and Vigor (POMS). These symptom clusters overlap with the four most prominent symptom clusters summarized in a recent comprehensive review by Griffiths and Mumford (1995) of the caffeine withdrawal literature: headache, drowsiness, work difficulty (decreased motivation for work/tasks, impaired concentration), and decreased well being/contentment (including decreased self-confidence, increased irritability). The present results also provide weak support for two less prominent symptom clusters cited in the Griffiths and Mumford (1995) review: decreased friendliness and flu-like symptoms. As shown in Table 2, decreased Friendly (POMS) and Flu-Like Symptoms were intermittently sensitive to the caffeine withdrawal manipulations across the four experiments.

Frequency of Caffeine Dosing. In experiment 1 (caffeine dosing interval), there was no clear evidence that the range or magnitude of caffeine withdrawal symptoms differed when 300 mg caffeine was consumed as a single dose in morning versus as 100 mg at three time points across the day. These findings, in combination with the existing literature, are intriguing given that caffeine withdrawal has been well documented after caffeine consumption in a number of studies that have varied widely with respect to dose and dosing frequency across the day. For example, many studies allowed participants to self-administer caffeine, ranging from 25 to 400 mg/dose, anywhere from four times a day to ad libitum consumption (e.g., Griffiths et al., 1986a,b; Hughes et al., 1991; Oliveto et al., 1992; Evans et al., 1994). Similarly, other studies using experimenter-determined caffeine intake documented caffeine withdrawal when individuals were consuming caffeine twice a day (Silverman et al., 1992; Schuh and Griffiths, 1997) or up to 10 times per day (Griffiths et al., 1990a). Taken together with the present findings, it appears that daily consumption of caffeine, irrespective of the pattern of intake across the day, can result in physical dependence and subsequent withdrawal upon the termination of caffeine intake.

Caffeine Maintenance Dose. The results of experiment 2 (dose effects) provide the first prospective experimental demonstration that both the range and magnitude of caffeine withdrawal symptoms are related to the caffeine maintenance dose. The number of symptom/cluster scales that were altered significantly after the substitution of placebo increased as a function of the daily caffeine maintenance dose (5, 6, and 10 scales at 100, 300, and 600 mg/day caffeine, respectively). In addition, the magnitude of withdrawal effect was greater at 600 mg/day caffeine than 100 mg/day caffeine on several measures including Headache, Headache/Poor Mood, Depression-Dejection (POMS), Friendly (POMS), and Total Mood Disturbance (POMS). These results are consistent with previous studies that reported that the incidence of caffeine withdrawal increased as a function of daily self-reported caffeine dose (e.g., Goldstein and Kaizer, 1969; Goldstein et al., 1969; Galletly et al., 1989; Fennelly et al., 1991; Weber et al., 1993).

Suppression of Caffeine Withdrawal. The results of experiment 3 (suppression of caffeine withdrawal) indicate that when individuals are maintained on 300 mg/day caffeine, a substantial reduction in caffeine consumption, or complete elimination, is necessary for the manifestation of the full, classic withdrawal syndrome. Only when the dose was reduced to 100 mg/day (one-third the maintenance dose) was there any evidence of caffeine withdrawal. In fact, the withdrawal observed tended to be relatively mild and variable even when a mere 25 mg of caffeine was substituted for the maintenance dose of 300 mg, although ratings of Headache tended to increase as the dose of caffeine decreased (cf. Fig. 4). Based on this experiment, the most sensitive caffeine withdrawal symptoms to emerge after a reduction in caffeine dose were Tiredness and Fatigue (POMS). This is consistent with observations from other studies of caffeine withdrawal suggesting that the onset of symptoms such as fatigue and sleepiness occurs before the onset of headache (Dreisbach and Pfeiffer, 1943; Goldstein et al., 1969; Bruce et al., 1991).

Duration of Caffeine Exposure. Experiment 4 (duration of caffeine exposure) showed that relatively short-term exposure (as few as 3 consecutive days) to intermediate doses of caffeine (300 mg/day) is sufficient to produce withdrawal symptoms when caffeine dosing is terminated. As shown in Table 2 and Fig. 4, the withdrawal tended to be somewhat greater after 7 days of caffeine exposure, with no further increases after 14 days of caffeine exposure. To our knowledge, only two previous studies have provided information suggesting that caffeine withdrawal can occur after relatively short durations of caffeine exposure (Dreisbach and Pfeiffer, 1943; Griffiths et al., 1986a). Both studies showed caffeine withdrawal headache upon termination of high caffeine doses (terminal doses 600 mg/day) administered for 6 to 15 days. Thus, the present study documents caffeine withdrawal after lower doses and shorter exposure durations than previously shown.

Limitations. Although the present series of experiments represents the most complete analysis to date of caffeine physical dependence in humans, there were some uncontrolled factors that could have affected the data. First, urine drug screens were not performed on individuals during these experiments. However, an interview, a questionnaire, and a urine drug screen before the study indicated that none of these individuals used illicit drugs. Second, menstrual cycle and related symptoms were not tracked in female participants. Thus, it is unknown how any possible extraneous drug use and any menstrual cycle phase changes could have affected the results. However, given the robust and orderly parametric findings across the four experiments (which were each of about 2 months duration) and the fact that the experiments used double-blind, counter-balanced designs, it seems unlikely that these uncontrolled factors would have altered the major conclusions of these experiments.