Caffeine Enhances Sustained Attention Among Adolescents

Abstract

Despite the growing interest in caffeine use and its effects among adolescents, and a large literature on caffeine and attention among adults, there is a lack of experimental work examining the impact of caffeine on sustained attention among adolescents. We evaluated the acute effects of caffeine (vs. placebo) during a long (33-min) classic vigilance task among 31 adolescents (aged 12–17; 15 female; median caffeine use = 28 mg/day). We predicted a dose-dependent effect of caffeine, which would attenuate declines in target detection over time (i.e., a vigilance decrement). In each of 3 visits, participants completed an identical pairs continuous performance task beginning ~25 min after consumption of noncaloric flavored water containing placebo, 1 mg/kg, or 3 mg/kg caffeine (order counterbalanced). Percent hits for low probability targets across 12 100-trial blocks was the primary outcome measure. As predicted, the linear decline in hits across trial blocks was attenuated by caffeine (Caffeine vs. Placebo × Block Linear, p = .01), with significant improvements in Blocks 9–12 (ps < .03). Compared to 1 mg/kg, 3 mg/kg caffeine resulted in earlier improvement in target detection (Drug Dose × Block Quadratic, p = .001). This study demonstrated that caffeine acutely and dose-dependently improves sustained attention among adolescents. These results were likely due to the attention-enhancing effect of caffeine, rather than withdrawal reversal, as our sample was characterized by light to moderate caffeine use. This study provides the foundation for further work on the impact of chronic caffeine consumption on cognitive function during adolescence.

Caffeine is the most widely consumed psychostimulant in the world (Nehlig, 1999). Developmentally, low levels of caffeine use in early childhood (2- to 5-year-olds; 25 mg/day) escalate in early adolescence (e.g., 68 mg/day in 12- to 16-year-olds; Branum, Rossen, & Schoendorf, 2014). Over the past decade, the percentage of adolescents using caffeine appears to be stable, but adolescents are consuming drinks with higher caffeine content, such as coffee and energy drinks as opposed to soda (Ahluwalia, Herrick, Moshfegh, & Rybak, 2014; Branum et al., 2014; J. L. Temple, 2019). Given that energy drinks have greater caffeine content, adolescents are at increased risk of negative effects of caffeine, such as anxiety, jitteriness, and at larger doses, even cardiac arrhythmias and tachycardia (Bernstein et al., 1998; Cannon, Cooke, & McCarthy, 2001; J. L. Temple, 2019).

In addition, there is a growing concern that chronic caffeine use by adolescents may worsen the sleep problems that are already common during this developmental period (e.g., Owens, Adolescent Sleep Working Group, & Committee on Adolescence, 2014). Indeed, adolescents report using caffeine to maintain focus and reduce fatigue during tasks that require sustained attention (e.g., classes at school) or to compensate for insufficient or inadequate sleep (Owens et al., 2014; Owens & Weiss, 2017). For example, adolescents report “getting through the day” as a primary motive for caffeine consumption (Bryant Ludden & Wolfson, 2010), and teenagers who consume more than 50 mg/day report using caffeine to stay awake and for its stimulant effects (Ludden, O’Brien, & Pasch, 2017; J. L. Temple, Dewey, & Briatico, 2010; see also Visram, Cheetham, Riby, Crossley, & Lake, 2016).

The pharmacology of caffeine is broadly consistent with these subjective reports. Specifically, caffeine acts as an antagonist to adenosine receptors, which plays a role in wakefulness (Olson, Thornton, Adam, & Lieberman, 2010; Ribeiro & Sebastião, 2010). Based on the pharmacologic effects of caffeine and subjective reports of users, it is widely believed that caffeine improves or maintains attention over long periods of time. However, despite the growing interest in caffeine use during adolescence (e.g., Ahluwalia et al., 2014; Branum et al., 2014; Owens et al., 2014; J. L. Temple, 2019), experimental evidence examining the impact of caffeine on sustained attention in teenagers is lacking.

Even in the extensive literature on caffeine and performance in adults, very few adequately assess the construct of sustained attention. Conceptually, sustained attention, or vigilance (we use these terms interchangeably), is the ability to maintain alertness and respond to infrequent target stimuli over a long period of time (e.g., Huang-Pollock, Karalunas, Tam, & Moore, 2012; Mackworth, 1948). In classic vigilance studies, which often involve performing a monotonous, slow-paced task, target detection decreases as time-on-task increases, creating a “vigilance decrement” (Davies & Parasuraman, 1982; Mackworth, 1948; Rosvold, Mirsky, Sarason, Bransome, & Beck, 1956). Manipulations that maintain vigilance, such as reinforcement or stimulant medication (e.g., Bubnik, Hawk, Pelham, Waxmonsky, & Rosch, 2015), act not by improving performance at the outset of the task but by reducing the vigilance decrement over time.

In contrast, studies in the adult caffeine and attention literature typically employ very brief tasks (10 min; e.g., Baer, 1987; Dodd, Kennedy, Riby, & Haskell-Ramsay, 2015; Elkins et al., 1981; Haskell, Kennedy, Milne, Wesnes, & Scholey, 2008), often with a high rate of stimulus presentation that may emphasize processing speed over vigilance (e.g., 100 stimuli/minute; e.g., Dodd et al., 2015; Smit & Rogers, 2000; Smith, Christopher, & Sutherland, 2013). Perhaps most importantly, most studies to date have ignored changes in performance as a function of time-on-task (e.g., Childs & de Wit, 2006; Clubley, Bye, Henson, Peck, & Riddington, 1979; Conners, 1979; Hewlett & Smith, 2006; Judelson et al., 2013; Smit & Rogers, 2000), a hallmark feature of vigilance. Finally, even in the small number of studies that demonstrate attenuation of the vigilance decrement by caffeine over time (J. G. Temple et al., 2000; see also Frewer & Lader, 1991; cf. null effects by Loke & Meliska, 1984), the interpretation is complicated by difficulty distinguishing between true beneficial effects of caffeine and withdrawal reversal. Among habitual caffeine users, any improvement in vigilance performance may reflect eliminating the effects of withdrawal symptoms, rather than a true benefit of caffeine (James & Rogers, 2005). Abrupt abstinence from caffeine use can lead to withdrawal symptoms (Silverman & Griffiths, 1992) even when consumption is less than 100 mg per day (Evans & Griffiths, 1999). Consistent with this perspective, Smit and Rogers (2000) found that caffeine improved attention performance among people with habitually high, but not low, levels of caffeine consumption after an overnight abstinent period. However, naïve caffeine-using populations have been used to address withdrawal reversal (e.g., Childs & de Wit, 2006) as they are unlikely to exhibit withdrawal symptoms. In the present study, we address the issue by examining the acute effects of caffeine on youth who have not yet become habitual users. Caffeine use typically begins in childhood and adolescence (Ahluwalia & Herrick, 2015), and youth do not exhibit withdrawal symptoms to the extent that adults do (Bernstein et al., 1998; Heatherley, Hancock, & Rogers, 2006; see also Graczyk et al., 2018).

The present study sought to fill critical gaps in our knowledge about the effects of caffeine on vigilance among low caffeine using adolescents. We focus on this sample both because of the growing interest in caffeine use and its effects during adolescence, the period of caffeine initiation and escalation, and because the focus on light caffeine users confers methodological advantages, namely mitigation of the concern over withdrawal reversal. We employed a classic vigilance paradigm, in which participants must respond only to infrequent target stimuli, and we examined changes in performance across a 33-min period. Adolescents (aged 12–17) completed a double-blind, placebo-controlled evaluation of acute caffeine (placebo, 1 mg/kg caffeine, and 3 mg/kg caffeine) across three laboratory visits. For a 60-kg adolescent, 1 mg/kg is comparable to a 12-oz soda, while 3 mg/kg is more comparable to a large strong coffee or an energy drink. The use of multiple dosages is important as the effects of caffeine on cognitive processes may be dose-dependent (e.g., Childs & de Wit, 2006). We predicted that caffeine would attenuate the linear decline in target detection across trial blocks (i.e., the vigilance decrement) when compared to the placebo condition. We also hypothesized that 3 mg/kg caffeine would further attenuate the vigilance decrement compared to 1 mg/kg.

Method

Participants

Thirty-four 12- to 17-year-olds were recruited from a parent study examining taste perception, salivary protein composition, and caffeine consumption. Participants in the parent study were recruited from the local community via flyers in public schools and from past studies where participants provided permission to be contacted about future research. Participants in the parent study were self-reported nonsmokers and were willing to abstain from caffeine for 24 hr before appointments. Exclusion criteria included metabolic or endocrine disorders (which would impact caffeine metabolism), pregnancy, history of olfactory disorders, use of medications known to affect caffeine metabolism (e.g., psychostimulants), and daily caffeinated beverage intake. Upon completion of the parent study, participants were recruited for the present study by phone and e-mail. Of the 34 adolescents who participated, data for three participants were incomplete due to software malfunctions. Table 1 presents demographics and other participant characteristics.

Table 1.

Sample Characteristics

Characteristic	M (SD)	Mdn/n (%)	Range
Age	15.12 (1.59)	14.63	12–17
Weight (kg)	64.03 (16.72)	60.15	33.85–94.15
Height (cm)	165.19 (10.39)	166.08	147.62–184.62
Average daily caffeine consumption (mg/day)	70.80 (79.37)	28.42	7.81–321.3
ESS-CHAD	7.97 (3.88)	8.00	1–15
Gender
Male		18 (53)
Female		16 (47)
Race
Asian		2 (6)
Black		5 (15)
White		24 (70)
American Indian/Native Alaskan		1 (3)
Other		2 (6)
Ethnicity
Hispanic or Latino		3 (9)
Non-Hispanic or non-Latino		30 (91)
Parental education
High school		2 (6)
Some college		6 (18)
Completed college		9 (28)
Graduate school		16 (48)
Household income
<$9,999		1 (3)
$10,000–49,999		6 (18)
$50,000–69,999		5 (15)
$70,000–89,999		7 (21)
$90,000–109,999		5 (15)
$110,000–139,999		5 (15)
>$140,000		4 (11)

Note. n = 34. ESS-CHAD = Epworth Sleepiness Scale for Children and Adolescents.

Design and Procedure

All procedures were approved via the University at Buffalo Institutional Review Board for Protocol HRP-503 and the Teen Sports Drink Study. The study was a within-subjects, double-blind, placebo-controlled evaluation in which caffeine dose (0 [placebo], 1, and 3 mg/kg) varied across visits. Participants completed three visits (intervisit interval range = 4–10 days). During the parent study, participants sampled and rated seven novel noncaloric flavored water drinks (made by combining flavored water additives; e.g., Mio, Kool-Aid, Crystal Light). The fourth ranked beverage flavor was selected for use during study visits. Caffeine and placebo solutions were prepared and coded by an experimenter not involved in data collection. Food-grade anhydrous caffeine (Sigma) was added to deionized water heated to 140°F and stirred for 25 min. Caffeine and placebo solutions were then aliquoted into 14-mL vials, labeled A, B, or C, and frozen. On the day of the visit, the appropriate vial was thawed at room temperature and added to the assigned beverage based on body weight. Drug order was counterbalanced across participants within gender. Participants were asked to not eat or drink anything except water for 2 hr before the appointment.

During Visit 1, adolescents provided informed assent while parents provided informed consent. Next, height and weight were assessed, and participants completed a brief reward preference survey (for an operant task completed after the attention task in each visit; Tonkin & Hawk, 2019). At each visit, the participant was given 5 min to drink 500 mL of the flavored water containing caffeine or placebo. During a 25-min absorption period, participants completed several questionnaires, including the Caffeine Use Questionnaire, adapted from Miller (2008) and used to determine daily caffeine usage, and the Epworth Sleepiness Scale for Children and Adolescents (ESS-CHAD; Johns, 2015), which was used to determine levels of daytime sleepiness (see Janssen, Phillipson, O’Connor, & Johns, 2017). Several filler measures were also included. After the questionnaires, the participant was introduced to and practiced an identical pairs continuous performance task (IP-CPT; Cornblatt, Risch, Faris, Friedman, & Erlenmeyer-Kimling, 1988, described next). Testing started an average of 24 min (SD = 2 min) after caffeine consumption and concluded 33 min later. Thus, the attention task occurred well within the active window of caffeine as peak plasma concentrations occur between 30 and 120 min after consumption and caffeine has a half-life of 5 hr (Bruce, Scott, Lader, & Marks, 1986; Mandel, 2002; White et al., 2016; Willson, 2018). After the IP-CPT, pilot data were collected on a 12-min operant task (Tonkin & Hawk, 2019). All visits started between 11:00 a.m. and 7:00 p.m. as adolescents generally consume caffeine later in the day (see Bryant Ludden & Wolfson, 2010). At the conclusion of Visit 3, participants were remunerated $50.

Identical Pairs Continuous Performance Task (IP-CPT)

Going back to the classic work of Rosvold et al. (1956), the CPT is designed to require sustained attention to a relatively monotonous stream of stimuli in order to detect relatively rare targets. To avoid ceiling effects that occur with simpler versions (e.g., Halperin, Wolf, Greenblatt, & Young, 1991; Hawk et al., 2018), participants completed an identical pairs CPT adapted from prior work (e.g., Halperin, Trampush, Miller, Marks, & Newcorn, 2008; Rhodes & Hawk, 2016; see also Kato et al., 2018). Participants were asked to attend to a series of four-digit numbers on a computer monitor, pressing the keyboard space bar only when the stimulus was identical to the stimulus immediately preceding it (percent targets = 10%).

Participants initially completed a brief practice (28 trials) at a slow pace (600-ms stimulus duration; 2,000-ms response window). All participants achieved the standard of correctly detecting at least two of the three targets with one or fewer false alarms to nontargets before proceeding. The actual task consisted of 1,200 trials; each consisted of a 150-ms stimulus presentation and a 1,500-ms response window (~36 stimuli/minute). The IP-CPT was programmed using E-Prime 2.0 software (Psychology Software Tools, Pittsburgh, PA).

Data Reduction and Analysis

As in prior work (e.g., Kato et al., 2018; Rhodes & Hawk, 2016; see reviews by Huang-Pollock et al., 2012), correct target detection, as indicated by percent hits ([number of targets identified/number of targets presented] × 100), was the primary outcome measure. To examine the vigilance decrement as a function of time-on-task percent hits was calculated for each of the twelve 100-trial blocks. Percent false alarms (responses to nontargets) was examined across trial blocks to ensure that changes in target detection were not simply a function of the overall response rate.

Percent hits and false alarms were analyzed in separate 3 Drug × 12 Trial Block analyses of variance. To evaluate caffeine effects, we focused on two planned orthogonal single-df contrasts: drug (average of 1 and 3 mg/kg) versus placebo and dose (1 vs. 3 mg/kg). To evaluate the effects of time, linear and quadratic trends across trial blocks were analyzed. Gender was evaluated as a moderator as previous literature suggests sex differences on the effects of caffeine among adolescents (Graczyk et al., 2018; J. L. Temple, Ziegler, Martin, & de Wit, 2015; J. L. Temple, Ziegler, Graczyk, & Crandall, 2017b). Additional moderators were considered in supplemental exploratory models: drug order, age, and daytime sleepiness (age and daytime sleepiness were mean centered).

Results

Percent Hits

Figure 1a presents mean percent hits (targets detected) for all Caffeine × Block conditions. Consistent with a vigilance decrement, percent hits declined linearly across trial blocks (block linear: F(1, 30) = 64.153, p < .001, d = 1.42; block quadratic: F(1, 30) = .898, p = .351, d = 0.19). Percent hits were higher during the caffeine visits compared to the placebo visit (drug vs. placebo: F(1, 30) = 10.251, p = .003, d = 0.58) and tended to be higher during 3 mg/kg versus 1 mg/kg caffeine (F = 3.439, p = .07, d = 0.33).

Figure 1.

Mean percent hits (Panel a) and false alarms (Panel b) for all Drug × Trial Block Conditions.

As predicted, the effects of caffeine and block interacted. Compared to placebo, caffeine attenuated the linear decline in percent hits across trial blocks (Caffeine vs. Placebo × Block Linear, F(1, 30) = 6.765, p =.014, d = 0.47), with significant differences in Blocks 9–12 (ps < .03). Compared to 1 mg/kg of caffeine, 3 mg/kg resulted in improved target detection earlier in the task (Drug Dose × Block Quadratic, F(1, 30) = 13.292, p = .001, d = 0.70), with higher percent hits in Blocks 4, 5, 7, and 8 (ps < .05). Data regarding hit reaction time can be found in the online supplemental materials as it is often reported as a primary outcome measure.

Percent False Alarms

Figure 1b presents mean percent false alarms for all Caffeine × Block conditions. As can be seen, false alarm rates were generally low (all means <4%). On average, false alarms decreased modestly and linearly across trial blocks (block linear: F(1, 30) = 4.785, p = .04, d = 1.30; block quadratic: F(1, 30) = .328, p = .57, d = 0.46). Although false alarms were generally comparable between caffeine and placebo conditions (caffeine vs. placebo: F(1, 30) = .477, p = .495, d = 0.12), false alarms were somewhat higher for placebo compared to active caffeine in Block 1 (p = .08), resulting in a statistically significant Caffeine versus Placebo × Block Quadratic interaction, F(1, 30) = 6.466, p = .016, d = 0.40. Percent false alarms tended to be higher during 1 mg/kg caffeine condition compared to 3 mg/kg caffeine, F(1, 30) = 4.068, p = .05, d = 0.36, a pattern that did not significantly differ across trial blocks (Drug Dose × Block Linear and Quadratic, Fs < 1, d = 0.12).

Moderators

Gender, drug order, and age all failed to significantly moderate the effects of caffeine (drug, dose, and interactions with block, all ps > .05). However, daytime sleepiness (as indicated by the ESS-CHAD) moderated the effect of caffeine on target detection (percent hits drug vs. placebo: F(1, 29) = 6.459, p = .017). Higher daytime sleepiness was associated with a larger improvement in hit rate during caffeine visits compared to the placebo visit.

Discussion

The present study provided the first test of acute caffeine effects on sustained attention among adolescents using a paradigm well-suited for assessing lapses in vigilance. We addressed many of the limitations of prior work by employing a classic vigilance task requiring the detection of infrequent, relatively difficult-to-detect targets over a long period of time (~30 min). Consistent with our hypotheses, acute caffeine attenuated the vigilance decrement observed under placebo conditions, and the beneficial effect of caffeine was dose-dependent (i.e., greater for 3 mg/kg vs. 1 mg/kg caffeine).

The pattern of findings is consistent with our hypothesis that vigilance—or sustained attention—is improved by acute caffeine consumption. As can be seen in Figure 1a, performance during the early minutes of the task was comparable across drug conditions. As time-on-task increased, rates of target detection declined. This was more pronounced in the placebo condition compared to the caffeine conditions. Given that the effects of caffeine on attention are not evident until approximately 10 min into the task, null findings found in prior work may be explained by the use of relatively brief attention tasks (e.g., Baer, 1987; Dodd et al., 2015; Owen, Parnell, De Bruin, & Rycroft, 2008; tasks lasting 5 to 10 min).

It also seems unlikely that the present findings could be explained by withdrawal reversal, a major concern in work with chronic caffeine users. For example, J. G. Temple et al. (2000; Study 2) found that the vigilance decrement observed in a placebo group was attenuated in a group given 1.1 mg/kg caffeine. However, all participants (mean age = 21) were asked to abstain from caffeine use for 12 hr prior to participation, raising the possibility that the group differences in vigilance were caused by a withdrawal-induced performance decline over time in the placebo group, rather than a true beneficial effect of caffeine on sustained attention. In contrast, participants in the present study consumed a median consumption of only 28 mg of caffeine per day. Because the majority of participants in the present study typically consumed less caffeine than is usually found in a single cup of coffee, brewed tea, or most caffeinated soft drinks (Ahluwalia & Herrick, 2015), it is highly unlikely that withdrawal reversal could account for the observed dose-dependent effect of caffeine on the vigilance decrement.

Our supplemental analyses lend support to the hypothesis that caffeine effects on vigilance were greater among those with more daytime sleepiness. Although subjective sleepiness was only measured at Visit 1, these data are broadly consistent with studies in adults that demonstrate caffeine attenuates cognitive decline associated with sleep deprivation (e.g., M. Beaumont et al., 2001; Kamimori et al., 2015; Patat et al., 2000; Ramakrishnan et al., 2014). Interestingly, chronic sleep deprivation is common among adolescents, at least partly due to the combination of delayed circadian rhythms and school requirements for early awakening (Owens et al., 2014; J. L. Temple, 2009). It is plausible that the effects of caffeine on attention could contribute to the escalation of caffeine use to offset daytime sleepiness. However, the literature on the role of chronic caffeine use in higher doses is much more complicated as chronic consumption may lead to tolerance and the loss of any cognitive benefits from caffeine (see R. Beaumont et al., 2017). Moreover, chronic caffeine use among youth may worsen sleep problems and reduce the quality of sleep (Aepli, Kurth, Tesler, Jenni, & Huber, 2015; Clark & Landolt, 2017), creating a feedforward cycle of escalating caffeine use and sleep disruption. One way to get higher doses of caffeine is to use high-dose energy drinks, which are growing in popularity among adolescents (Clark & Landolt, 2017; J. L. Temple, 2019). Energy drinks may be related to possible sleep disruption and other health concerns (Ehlers, Marakis, Lampen, & Hirsch-Ernst, 2019; J. L. Temple, Ziegler, Graczyk, & Crandall, 2017a; Zucconi et al., 2013). Ultimately, a better understanding of the complex interplay of caffeine, sleep, and cognitive performance will require longitudinal studies that utilize rigorous measurement in each domain, including indices of sustained attention like the one employed in the present study.

It is important to note that we did not assess increases in subjective arousal or attention but rather focused on evaluating the objective impact of caffeine on vigilance. That said, we recognize that either or both subjective and objective effects could contribute to the escalation of caffeine use in youth. The study also would have been strengthened by biological confirmation of the caffeine manipulation. In addition, the present sample was one of convenience, which may limit the generalizability of the findings.

In sum, the present study is novel in demonstrating that caffeine acutely improves vigilance among adolescents. That is, caffeine had no effect early in the task when target detection was high for all conditions; however, the degree to which attention waned overtime was attenuated by caffeine in a dose-dependent manner. A strength of this study was the use of light caffeine users, which suggests caffeine improves sustained attention, rather than reversing the effects of caffeine withdrawal. However, it will be critical to extend this research to examine the impact of chronic caffeine use, and its interaction with other factors such as sleep, to fully understand the role of caffeine in cognitive function during adolescence.

Public Health Significance.

The present laboratory study is the first to demonstrate a dose-dependent effect of caffeine on sustained attention among adolescents. Because participants were relatively light users of caffeine, the results suggest this is a true beneficial effect, rather than simply reversal of caffeine withdrawal.