Sensorimotor and executive function slowing in anesthesiology residents after overnight shifts

Journal of Clinical Anesthesia 40 (2017) 110–116

Contents lists available at ScienceDirect

Journal of Clinical Anesthesia

Original Contribution

Sensorimotor and executive function slowing in anesthesiology residents after overnight shifts George W. Williams, MD a,b,⁎, Bairavi Shankar, BA c,d, Eliana M. Klier, PhD d, Alice Z. Chuang, PhD e, Salma El Marjiya-Villarreal, MD a, Omonele O. Nwokolo, MD a, Aanchal Sharma, MD a, Anne B. Sereno, PhD d a

Dept. of Anesthesiology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX 77030, United States Dept. of Neurosurgery, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX 77030, United States Dept. of Neuroscience, Rice University, Houston, TX 77005, United States d Dept. of Neurobiology and Anatomy, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX 77030, United States e Dept. of Ophthalmology and Visual Science, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX 77030, United States b c

a r t i c l e

i n f o

Article history: Received 9 January 2017 Received in revised form 31 March 2017 Accepted 4 April 2017 Keywords: Cognitive Sensorimotor Resident Anesthesiology Shift work Sleep deprivation

a b s t r a c t Study objective: Medical residents working overnight call shifts experience sleep deprivation and circadian clock disruption. This leads to deficits in sensorimotor function and increases in workplace accidents. Using quick tablet-based tasks, we investigate whether measureable executive function differences exist following a single overnight call versus routine shift, and whether factors like stress, rest and caffeine affect these measures. Design: A prospective, observational, longitudinal, comparison study was conducted. Setting: An academic tertiary hospital's main operating room suite staffed by attending anesthesiologists, anesthesiology residents, anesthesiologist assistants and nurse anesthetists. Patients: Subjects were 30 anesthesiology residents working daytime shifts and 30 peers working overnight call shifts from the University of Texas Health Science Center at Houston. Interventions: Before and after their respective work shifts, residents completed the Stanford Sleepiness Scale (SSS) and the ProPoint and AntiPoint tablet-based tasks. These latter tasks are designed to measure sensorimotor and executive functions, respectively. Measurements: The SSS is a self-reported measure of sleepiness. Response times (RTs) are measured in the pointing tasks. Main results: Call residents exhibited increased RTs across their shifts (post–pre) on both ProPoint (p = 0.002) and AntiPoint (p b 0.002) tasks, when compared to Routine residents. Increased stress was associated with decreases in AntiPoint RT for Routine (p = 0.007), but with greater increases in sleepiness for Call residents (p b 0.001). Further, whether or not a Call resident consumed caffeine habitually was associated with ProPoint RT changes; with Call residents who habitually drink caffeine having a greater Pre-Post difference (i.e., more slowing, p b 0.001) in ProPoint RT. Conclusions: These results indicate that (1) overnight Call residents demonstrate both sensorimotor and cognitive slowing compared to routine daytime shift residents, (2) sensorimotor slowing is greater in overnight Call residents who drink caffeine habitually, and (3) increased stress during a shift reduces (improves) cognitive RTs during routine daytime but not overnight call shifts. © 2017 Elsevier Inc. All rights reserved.

1. Introduction Shift workers adapting to changes from typical (i.e., 9:00–5:00) work hours contend with issues such as sleep deprivation [1], memory loss [2], cognitive dysfunction [3], and decreased motor performance [4, 5]. Specifically, it is important to investigate the effects of an overnight ⁎ Corresponding author at: Departments of Anesthesiology and Neurosurgery, McGovern Medical School, The University of Texas Health Science Center at Houston, 6431 Fannin Street, Houston, TX 77030, United States. E-mail address: [email protected] (G.W. Williams).

http://dx.doi.org/10.1016/j.jclinane.2017.04.002 0952-8180/© 2017 Elsevier Inc. All rights reserved.

shift in physicians, such as anesthesiologists, because disturbances in circadian rhythms can lead to deficits such as increased distractibility, deviations from typical problem-solving strategies, inability to deal with unexpected events and increased work-related injuries such as needle sticks [6–8]. Sleep deprivation and restriction also result in deterioration of sensorimotor performance [4,5,9–13]. Overnight call shifts are common among anesthesiology residents and practicing attending physician anesthesiologists, and these deficits can lead to medical errors, cause improper care of critically ill patients, or increase the chance of complications [14,15]. This study builds on this previous research by evaluating cognitive and sensorimotor performances of resident

G.W. Williams et al. / Journal of Clinical Anesthesia 40 (2017) 110–116

anesthesiologists with a portable, quick, cost-effective tablet-based tool before and after a single overnight call shift and comparing the findings to similar measurements taken in a routine, daytime shift. In contrast to the negative effects of sleep deprivation mentioned above, previous work has shown that multiple variables can positively affect cognitive function. One study suggests that a mid-day nap can improve cognitive function and alertness among residents during a routine shift [16]. Several studies have demonstrated that caffeine intake can improve cognitive performance and vigilance up to 64 h under extended sleep deprivation conditions [17–19]. Our study will examine whether increased sleep the night before a shift, stress level during shift, and caffeine consumption habits relate to sensorimotor or cognitive changes found across shift. This study employed the Stanford Sleepiness Scale (SSS), a standard subjective scale of sleepiness, as well as two novel, touch-based tablet applications, ProPoint and AntiPoint tasks [20–21]. The tablet tasks allow for rapid data collection using affordable, readily available and portable equipment. The ProPoint task measures sensorimotor function whereas the AntiPoint task measures cognitive function. If sensorimotor or cognitive slowing occurs during an overnight, call shift as compared to a routine, daytime shift, there should be a significant increase in response time (RT) in both the ProPoint and AntiPoint tasks. Additionally, previous research on caffeine consumption, stress, and sleep the night before extended periods of sleep deprivation suggests that residents who sleep more prior to their shift, are more stressed during their shift, or consume caffeine during their shift should display improved cognitive functioning. If true, these effects would be observed by decreases in RT on the AntiPoint task. We hypothesize that sensorimotor and cognitive slowing will occur following overnight, call shifts to a greater degree than the slowing seen following routine, daytime shifts. 2. Materials and methods 2.1. Ethics statement This study was approved by the Institutional Review Board at the University of Texas Health Science Center at Houston (UTHealth) and is in accordance with the ethical principles of the Declaration of Helsinki. Written informed consent for participation in this study was first obtained from all anesthesiology residents who participated in this study. This manuscript adheres to the applicable Equator guidelines. 2.2. Subjects Subjects were recruited and enrolled until target enrollment was met from the Department of Anesthesiology at UTHealth and were residents aged between 25 and 55 years that were rotating through Memorial Hermann Hospital's main operating suite. Residents were excluded if they were rotating at alternate sites or away rotations or were currently using sedatives, hypnotics, or mood elevator medications, or diagnosed with acute stress disorder, adjustment disorder, depression, or sleep related disorders. Thirty eligible subjects were recruited for the Routine group (a daytime shift) and 30 for the Call group (an overnight shift). One subject from the Routine group and two subjects from the Call group were excluded from analysis because they had incomplete pointing task data sets. 2.3. Testing instruments 2.3.1. Pointing tasks The participants performed two tasks on an iPad2 with a video frame refresh rate of 60 Hz. As illustrated in Fig. 1, the stimulus display for both tasks had a circle in the middle of the display with a diameter of 1.4 cm. There were four surrounding square target boxes with a side length of 0.8 cm with centers positioned 4.0 cm away from the center circle.

111

Fig. 1. Tablet tasks. (A) ProPoint task. Central fixation point (circle) and four possible target locations (squares) are indicated. The target location is shown by the filled square and the arrow indicates the direction of correct response direction. (B) AntiPoint task. Same as in (A), but note that arrow indicating direction of correct response direction is now opposite to the target's location (figure adapted from 20).

After touching the center circle, one of the four target boxes would illuminate after a fixed 280 ms delay. In the ProPoint task, the subject tapped the illuminated square as quickly as possible, thus testing sensorimotor control, using a stimulus-driven response. In the AntiPoint task, however, the subject was instructed to tap the square that lay opposite to the one that was illuminated as quickly as possible, thereby testing a more cognitive response that required suppression of an initial motor command in one direction and production of a new one in the opposite direction. Each participant performed each task until obtaining 48 correct trials (each participant completed on average 48.9 trials). For each trial, the response time (RT) and whether the response was correct or not were recorded. RT was defined as the time from onset of the target square to the time the finger touched the correct target. A response was counted as an error if the distance between the correct response location and the iPad-calculated touched location was N 3.3° (1.9 cm) (for additional details see [20–23]). The RT and error data for each subject were recorded on the iPad, and then transferred to a computer for off-line analysis. 2.3.2. Stanford sleepiness scale (SSS) The SSS is a fast, self-reported measure of sleepiness. Participants recorded their level of perceived sleepiness on a scale from 1 to 7, then X, with 1 defined as “feeling active, vital, alert, or wide awake”, 7 defined as “no longer fighting sleep, sleep onset soon; having dream-like thoughts”, and X defined as “asleep.” Subjects completed the SSS before and after their shift. The data were then transferred to a computer for analysis. 2.4. Procedure The Routine group was tested before and after a routine daytime shift, which started at 07:00 and ended at 17:00. The Call group was tested before and after an overnight shift, which started at 15:00 and ended at 07:00 the next morning. There were no consecutive call shifts and no residents were included in both groups. Before the shift, participants first completed a questionnaire, including questions such as age, sex, handedness, and smoking, and hours slept the night before their shift, followed by the SSS and pointing tasks. Typical activity during a call shift involves directly caring for a patient in the operating room for 80–90% of the time of the shift. After the shift, participants completed another questionnaire regarding their shift, such as the number of calls taken, stress level, whether they were habitual caffeine drinkers or not, and the time since last caffeine consumption, followed by the SSS and pointing tasks. Stress level was measured using a numerical rating scale, from 1 to 10, familiar to participants and analogous to the numerical rating scale used for pain, which has been shown to have high reliability [23]. For both pointing tasks, each participant was given 8 practice trials before being allowed to complete the tasks. In order to control for any order bias, half of each subject group started with the ProPoint task

112

G.W. Williams et al. / Journal of Clinical Anesthesia 40 (2017) 110–116

followed by the AntiPoint task, and the other half received the opposite order.

2.5. Statistical analysis 2.5.1. Sample size calculation The estimated standard deviation for RT was approximately 20 ms for ProPoint and 40 ms for AntiPoint based on a previous study [20]. A sample size of 28 per group is sufficient to detect 15 ms and 30 ms differences between groups in RT for ProPoint and AntiPoint tasks at a 5% significance level and 80% power. To account for 10% or less possible subject attrition, 30 participants for each group were recruited.

2.5.2. Errors and trimming Given that there were very few errors (average 1.9%; see Supplementary Table 1), error data were not further analyzed. The remaining correct RT data was further analyzed with a MATLAB script for trimming outliers. The trimming procedure was: (1) exclude any RT that was b150 ms or N800 ms; (2) calculate the mean and standard deviation of remaining trials for each participant, in each testing session, and in each task. Any trial with a RT that was N2.75 standard deviations from the condition mean was removed iteratively (see Supplementary Table 2). Together, these trimming procedures removed 5.1% and 5.3% of the trials for ProPoint and AntiPoint tasks, respectively.

2.5.3. Demographic measures and their group analyses Continuous variables were summarized by mean (± standard error) and group comparisons were made using two-sample t-tests. Discrete variables were summarized by frequency and group comparisons were made using Fisher exact tests. All p-values presented are for two-tailed tests. See Supplemental Table 3 for a summary of descriptive variables and findings. There were 3 missing data points and 6 not applicable responses on the questionnaires that could not be scored (noted in Table 1 and Supplementary Table 3). The analyses for these three demographic measures were performed with the remaining data.

2.5.4. Primary analysis To examine differences in RT, we performed a mixed effects model analysis separately for ProPoint and AntiPoint tasks, where the fixed effects were Group (Call, Routine) and Session (Pre, Post) and the random effect was the participant with an autoregressive order 1 covariance structure. For significant effects or interactions, we followed with mixed effects model planned post-hoc comparisons: (1) between Groups during Pre-session, (2) between Groups during Post-session, (3) between Session change for the Routine group, (4) between Session change for the Call group, and (5) change across Sessions between Groups. A similar mixed effects model analysis and planned post-hoc comparisons were performed for SSS, with the same fixed effects (Group and Session) but the random effect was the participant with compound symmetric covariance structure.

Table 1 Summary of demographics and subject characteristics for each group.

Age (SD) Gender (female, %) Handedness (left, %) Smoker (yes, %) Caffeine consumption (yes, %) Hours slept previous night (SD)a Stress level during shift (1−10) (SD) a

Missing 2 data points.

Routine (n = 28)

Call (n = 29)

P

30.13 (3.09) 12 (41%) 4 (14%) 0 (0%) 19 (66%) 5.83 (2.09) 3.62 (1.74)

29.82 (2.64) 13 (46%) 4 (14%) 0 (0%) 22 (79%) 7.34 (2.65) 4.11 (2.37)

0.68 0.79 1.00 – 0.38 0.02 0.38

2.5.5. Risk factor analyses All primary analyses above were re-run by adding habitual caffeine consumption, hours slept previous night, skipped meal, hours since last caffeine, number of calls, and stress during the shift as fixed effects. A backward elimination procedure was performed to remove the factors that had no significant effect on the outcomes. For all statistical tests, a p-value b0.05 was considered as statistically significant and bolded in the tables. 3. Results 3.1. Demographic and subject characteristics The Routine group (n = 29) had a mean age of 30.1 years (± 3.1, range from 27 to 42) and was 41% female, and the Call group (n = 28) had a mean age of 29.8 (± 2.6, range from 26 to 38) and was 46% female (see Table 1). There were no significant differences between Groups in age, gender, handedness, smoking, habitual caffeine consumption, or reported stress level (see Table 1). There were significant differences in the hours slept the previous night between groups. 3.2. Primary analysis 3.2.1. Group effect 3.2.1.1. Pre-shift. In the ProPoint task (see Fig. 2A and Table 2), there was no difference in RT between groups Pre-shift (480 ms for Routine and 475 ms for Call; t(55) = 0.91, p = 0.37). There was also no difference in RT between groups in the AntiPoint task (597 ms for Routine and 595 ms for Call; t(55) = 0.24, p = 0.81) and in SSS between groups (1.8 for Routine and 1.5 for Call; t(58) = 0.99, p = 0.32) tested before shift (see Table 2). 3.2.1.2. Post-shift. In the ProPoint task (see Fig. 2B and Table 2), there were significant differences in RT between groups tested after shift (483 ms for Routine and 502 ms for Call; t(55) = 3.59, p = 0.0007). Similarly, there were differences in RT between groups in the AntiPoint task (578 ms for Routine and 606 ms for Call; t(55) = 4.15, p b 0.001) and in SSS between groups (2.4 for Routine and 3.8 for Call; t(58) = 4.64, p b 0.001) tested after shift (see Table 2). 3.2.2. Session effect 3.2.2.1. Routine group. In the Routine group (Fig. 3A and Table 2), there was no difference in RT found across session in the ProPoint task (3 ms slower, t(55) = 0.70, p = 0.49). However, there was a significant difference in RT found across session in the AntiPoint task (19 ms faster, t(55) = 2.90, p = 0.005) and a significant difference in SSS across session (difference = +0.6, t(58) = 2.39, p = 0.020). Note that the significant difference in RT for the AntiPoint task was in the opposite direction than what one would have expected (i.e., the residents had faster RTs after a Routine shift). 3.2.2.2. Call group. In the Call group (Fig. 3B and Table 2), there was a significant difference in RT across session in the ProPoint task (27 ms slower, t(55) = 5.35, p b 0.001) and significant difference in SSS (difference = +2.30, t(58) = 9.17, p b 0.001). However, there was no difference in RT found before and after shift in AntiPoint task (11 ms slower, t(55) = 1.63, p = 0.11). 3.2.3. Comparing session effects between groups There was a significant difference in Session effect (pre and post shift) between groups for both tasks and SSS (Fig. 3C and Table 2). The Call group showed greater slowing in the ProPoint task (23 ms, t(55) = 3.32, p = 0.002) and in AntiPoint task (29 ms, t(55) = 3.19,

G.W. Williams et al. / Journal of Clinical Anesthesia 40 (2017) 110–116

113

caffeine consumption, and number of calls during shift, for Routine and Call groups, respectively, did not significantly affect any of the outcome variables. Although the hours slept the previous night were significantly different between groups (Routine = 5.83, Call = 7.34, p = 0.02), it did not affect the pointing tasks and SSS. Two risk factors were found to significantly affect at least one outcome variable. 3.3.1. Caffeine Habitual caffeine consumption showed significant slowing of ProPoint RT on Session effect (t(54) = 4.31, p b 0.001). Residents in the Call group who drank caffeine habitually showed a 65 ms slower Pre-Post difference in ProPoint RT compared to those who did not (p b 0.001). The caffeine effect in ProPoint RT was not observed in the Routine group (p = 0.45). Similarly, the caffeine effect was not significant in both the AntiPoint task (p = 0.14) and the SSS (p = 0.15).

Fig. 2. Group effects. ProPoint and AntiPoint latencies as well as SSS scores are shown for Routine (light gray) and Call (dark gray) groups for both pre-shift (A, unstriped) and postshift (B, striped) measures. Error bars shown are SEs obtained from the mixed effect model. * indicates significant difference at p b 0.05; ns indicates no significant difference at p b 0.05.

p = 0.002) than the Routine group. Note that although Post-shift AntiPoint RT was not significantly slower than Pre-shift in the Call group, the difference in RT for the Call group (11 ms slowing) showed significantly greater slowing than the Routine group, who were actually 19 ms faster after shift. The Call group also showed greater sleepiness on the SSS (1.7, t(58) = 4.79, p b 0.001).

3.3. Risk factor analysis Of all the risk factor studied (see Supplementary Table 3), the number of hours slept the previous night, skipped meals, hours since last

3.3.2. Stress level during shift The effect of stress level during shift was not significant for the ProPoint task (p = 0.46), but was significant for the AntiPoint task (p = 0.008), with a 6 ms decrease in RT per unit increase in reported stress (on a 1–10 scale) during shift. This effect was greater in the Routine group (10 ms decrease/unit, p = 0.007) than in the Call group (3 ms decrease/unit, p = 0.18). The stress level during shift also significantly affected Session SSS (p b 0.001), with a 0.3 decrease in SSS per unit increase in reported stress during shift. In the Routine group, post-Shift SSS increased by 0.18 units for every unit increase in reported stress during shift but was not significant (p = 0.20). However, in the Call group, SSS increased by 0.41 unit for every unit increase in reported stress during shift (p b 0.001). 4. Discussion The aim of this study was to determine if there was measureable and significant slowing in cognitive performance after a single, overnight call shift when compared with a routine daytime shift among anesthesiology residents. The results indicated that there was significant sensorimotor slowing as well as cognitive slowing in residents after a single

Table 2 Estimated unadjusted group/session effects from mixed effect models (SD of average RT from each participant; SE obtained from mixed effect model; bold indicating significant P values, P b 0.05). Task

Session

Routine (SD, SE)

Call (SD, SE)

Group Difference (SD, SE)

Pa

ProPoint

Pre (SD, SE) Post (SD, SE) Session difference (SD, SE) Pb Pre (SD, SE) Post (SD, SE) Session difference (SD, SE) Pb Pre (SD, SE) Post (SD, SE) Session difference (SD, SE) Pb

479.79 (48.12, 3.62) 483.22 (48.77, 3.63) 3.43 (46.92, 4.91) 0.49 597.08 (84.61, 4.70) 578.41 (58.70, 4.69) −18.67 (52.31, 6.44) 0.005 1.80 (1.10, 0.21) 2.40 (1.32, 0.21) 0.60 (1.30, 0.25) 0.020

475.08 (53.91, 3.68) 501.78 (46.21, 3.68) 26.70 (52.38, 4.99) b0.001 595.47 (73.62, 4.81) 606.22 (51.35, 4.80) 10.76 (60.32, 6.59) 0.108 1.50 (0.64, 0.21) 3.80 (1.47, 0.21) 2.30 (1.46, 0.25) b0.001

−4.71 (51.08, 5.16) 18.55 (47.56, 5.17) 23.26 (49.68, 7.00)

0.37

−1.61 (79.40, 6.73) 27.81 (55.21, 6.71) 29.42 (56.38, 9.21)

0.81

−0.30 (0.91, 0.30) 1.40 (1.40, 0.30) 1.70 (1.38, 0.35)

0.32

AntiPoint

SSS

a b

P value obtained from the mixed effect model with group as an independent variable and participant as a random effect variable for before and after shift. P value obtained from a mixed effect model with session as an independent variable and participant as a random effect variable for each group.

b0.001 0.002

b0.001 0.002

b0.001 b0.001

114

G.W. Williams et al. / Journal of Clinical Anesthesia 40 (2017) 110–116

task thus requires inhibition of the reflexive, stimulus-directed response as well as generation of a rule-based voluntary response; both are aspects of cognitive control requiring intact frontal lobe function [34– 36]. The minimalist design and demands of the tasks are key to their success and help to provide a sensitive measure of sensorimotor and executive function that is less affected by confounding variables. For example, typical cognitive tests tend to be more complex and involve additional cognitive processes (such as reading or color identification) that can vary across individuals, increasing variability and decreasing sensitivity in the measurement of these cognitive processes (e.g., slowing due to English as a second language, dyslexia, or color blindness). Indeed, previous work has demonstrated that eye tracking tasks are more sensitive at detecting differences in executive function than standard neuropsychological testing [37–39], and also more sensitive at tracking treatment related cognitive changes [37,40–41].

4.2. Time of day

Fig. 3. Session effects. ProPoint and AntiPoint latencies as well as SSS scores are shown for pre-shift (unstriped) and post-shift (striped) groups for Routine (A, light gray) and Call (B, dark gray) groups. Differences between post-shift and pre-shift measures are also indicated (C). Error bars shown are SEs obtained from the mixed effect model. * indicates significant difference at p b 0.05; ns indicates no significant difference at p b 0.05.

night call compared to a routine day shift. Interestingly, the amount of cognitive slowing between groups (see Fig. 3C middle, Anti, panel) was about 30 ms and commensurate with the amount of cognitive slowing following sub-concussive blows (headers) in a previous study of girls' high school soccer players [20]. Our findings also indicate that the sensorimotor slowing in the Call residents was further exacerbated in those who were habitual caffeine consumers. Stress during shift affected the measured cognitive slowing occurring across shift with increased stress significantly reducing the cognitive slowing in the Routine but not Call group. Interestingly, stress affected the subjective measure of sleepiness by significantly increasing reported sleepiness in Call residents, but not Routine residents. Thus, for Routine residents, stress did not increase reported sleepiness and improved cognitive performance; whereas, for Call residents stress increased reported sleepiness and did not ameliorate cognitive slowing. 4.1. Benefits of tablet tasks Unlike previous studies that use subjective scales of sleepiness [24] or cumbersome measures such as driving simulators [25], which are unlikely to be adopted in medical or other settings, our ProPoint and AntiPoint tasks are objective, sensitive, affordable, and operate on ubiquitously available tablets. These tasks are designed to be completed in ~2 min and require no technical skill to complete, making them particularly useful and applicable in clinical practice. ProPoint and AntiPoint tasks capitalize on methods, paradigms, and analyses well-established for arm movements as well as eye movements (prosaccades and antisaccades) to detect sensorimotor and cognitive changes [20,26– 33]. The ProPoint task measures sensorimotor function using handpointing movements toward a visual target, whereas the AntiPoint task measures cognitive function using hand-pointing movements in the opposite direction (i.e., away) from a visual target. The AntiPoint

A possible confounding variable is the time of day, which varied by Group. The Routine shift began in the morning and the Call shift in the late afternoon, so it is possible that time of day at initial testing could affect RT measurements. Therefore, we compared groups before shifts (see Pre-Shift Results) to see if there were any group differences at this time point in order to determine if this is a confounding variable. Our results showed that there were no differences in performance before shifts in either the ProPoint or AntiPoint tasks, despite the two groups being tested at different times of the day. Although previous work has suggested that sleep inertia follows a circadian rhythm and is significantly higher during the night than during the day [42], this work examined people within an experimental context that was not similar to shift work. Our findings suggest that our resident shift workers do not show the same pattern. In fact, studies have shown that shift workers at least partially adapt their circadian rhythms and show normal performance independent of their shift time [43].

4.3. Sleep Research suggests that saccadic eye movements can measure nighttime sleepiness [44] and that sleep deprivation can impair cognitive functioning and can impair higher-level cognitive functioning even after alertness and vigilance return [45]. Routine shift participants did not sleep during their shifts, but Call participants may have slept during shifts. Further, Call participants also had a greater number of hours slept the previous night. These additional hours of sleep could lower ProPoint and AntiPoint RTs. Thus, it is possible that this extra sleep could improve RTs in the Call group, especially cognitive RTs. If so, this would suggest that the cognitive deficits we report might be greater than we measured here and might be masked by the benefits due to sleep during shift in the Call group. This could be particularly important in practices that have 24 h call shifts, instead of the 16 h call shifts utilized in our institution. The Accreditation Council for Graduate Medical Education (ACGME) currently requires napping for PGY-1 resident physicians if a shift exceeds 16 h of continuous clinical activity [46]. This study potentially validates this policy in the anesthesiology resident population when sensorimotor slowing is considered. Clinical activity during the shift was dependent on surgical case volume demands and not easily controlled. Our study was an observational study and sought to provide an accurate assessment of real-world practice. Observational designs can be chosen for a number of reasons including lack of influence (e.g., structuring sleep schedule for night call anesthesiology residents) as well as ethical concerns. Given our findings, further research may explore whether and to what extent sleep during shift ameliorated the reported sensorimotor and cognitive slowing in residents working an overnight shift.

G.W. Williams et al. / Journal of Clinical Anesthesia 40 (2017) 110–116

4.4. Caffeine Previous research regarding caffeine consumption has shown a dose-response relationship between caffeine intake and improved cognitive performance [47]. However, in regular caffeine consumers, all doses of caffeine (including those smaller than what is normally contained in caffeinated beverages) result in cognitive improvements, and this dose-response relationship flattens with increasing dose [48]. In contrast, in our study we found a greater slowing in those residents consuming caffeine habitually. One possible explanation is that previous studies controlled the amount of caffeine intake and tested participants within a short period of time following caffeine consumption. In the present study we did not control caffeine intake volume and estimating caffeine dose in an observational design was fraught with problems (e.g., controlling for the amount of caffeine found in different beverages and estimates of the size and amounts consumed). The average time since last caffeine was 11.8 h. Thus it is possible the residents would show a caffeine benefit at a short time scale, but that benefit dissipates with time. Further, our data suggests over a long enough time period, there may be a period of withdrawal or “crash”, where at least sensorimotor performance appears to slow. This surmised effect is apparently evident at the end of a call shift. Previous work suggests that caffeine provides increased alertness but limited benefits in verbal memory or motor learning [49]. Our findings regarding the detrimental effect of caffeine consumption suggests that caffeine, while commonly used by practicing physicians, may not be efficacious in improving concentration across longer intervals. 4.5. Stanford sleepiness scale The Stanford Sleepiness Scale (SSS) is widely used as a metric for measuring sleep deprivation. Studies have suggested that the SSS is sensitive to deficits in alertness caused by partial sleep deprivation, but not sensitive enough to be used as a metric for chronic sleep deprivation [50]. Given Routine shift residents did not work for 14 h between shifts and Call shift residents did not work for at least 24 h between shifts, the SSS is an applicable metric for this study. Finally, it is interesting that there were some differences between the objective and subjective measures with respect to the risk factor analyses. Namely, for objective measures, habitual caffeine drinkers showed greater sensorimotor slowing only during Call shifts and stress improved cognitive performance only during Routine shifts. On the other hand, for SSS, a subjective measure, stress resulted in greater reported sleepiness only in the Call residents. Thus, although the self-reported SSS measure may have been more accurate with respect to categorizing whether a resident completed a Routine or Call shift, the objective measures were able to identify risk factors that influenced sensorimotor and cognitive performance. Further, the findings demonstrate that SSS self-reported, perceptual variables associated with fatigue do not necessarily correlate with the more objective, biologically-based performance measures of the tablet tasks. 4.6. Study limitations Call residents may have slept during their shifts, thus future studies need to explore the impact of the number of sleep/nap sessions, the duration of each session, and the total amount of sleep obtained during call shifts and their potential impact on task performance. Additionally, the differentiation between caffeine drinkers was qualitative. We did not quantify the exact amount of caffeine consumed. Future studies could control the number, type, and timing of caffeinated drinks residents consume Note that this study was an initial attempt to determine if our pointing tasks are sensitive enough to detect performance changes during call shifts. It was not clear at the outset that the tasks could even detect differences following a single overnight shift, as previous studies

115

have only been able to detect changes after multiple overnight shifts [25]. Future studies can examine the clinical implications of these changes in latency and quantify reaction time slowing with delay in significant clinical procedures (e.g., time to detect a new lethal arrhythmia, rapid hemodynamic change or change in ventilation). 5. Conclusions Physician anesthesiologists and physicians in other specialties who are required to work overnight call shifts experience sleep deprivation and disruption of their circadian clocks that can lead to deficits in function and increases in workplace accidents. Using a quick, objective, tablet-based measure, we report significant sensorimotor and cognitive slowing after a single, overnight, call shift compared to a routine, daytime shift. This tablet-based tool may be easily used in clinical practice just as it has been validated in an emergency room traumatic brain injury population [21]. Additionally, sensorimotor slowing in the Call residents was further exacerbated in habitual caffeine consumers, and stress was found to improve cognitive performance in Routine shift residents but did not reduce the cognitive slowing in the Call residents. Together, these findings suggest there are measureable sensorimotor and executive function differences following a single call shift. Although these differences are small, they may still be clinically significant in urgent care settings such as the operating room. Further work is needed to determine at what threshold these changes in latency might significantly impair performance. Finally, it is important to note that although there was response slowing in Call residents, self-reported error rates remained extremely low, indicating correct decisions were still reached. In the context of medical decisions, this implies that correct decisions can still be made regarding patient care, but that these decisions take longer to make. Disclosures This work was supported by Mission Connect, a program of TIRR Foundation [grant number 014-119]; and the National Institutes of Health [grant number P30-EY010608]. Acknowledgements We would like to thank Stuart Red (PhD, Department of Neurobiology and Anatomy, McGovern Medical School, UTHealth, Houston, TX, USA) for his efforts in design and data collection and Michelle Won (undergraduate student, Department of Neuroscience, Rice University and Department of Neurobiology and Anatomy, McGovern Medical School, UTHealth, Houston, TX, USA) for feedback on earlier manuscript drafts. We would also like to thank Dr. Brian Crenshaw (Medical Doctor in Private Practice, Anesthesiology, Thibodaux Regional Medical Center, Thibodaux, LA, USA), Mr. Tyrone Burnett (Research Assistant, Department of Anesthesiology, McGovern Medical School, UTHealth, Houston, TX, USA) and Mr. Tariq Syed (Masters of Science, Research Coordinator, Department of Anesthesiology, McGovern Medical School, UTHealth, Houston, TX, USA) for their assistance in data collection and organization. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jclinane.2017.04.002. References [1] Akerstedt T, Arnetz BB, Anderzen I. Physicians during and following night call duty: 41 hour ambulatory recording of sleep. Electroencephalogr Clin Neurophysiol 1990; 76:193–6. [2] Killgore WD, Killgore DB, Day LM, Li C, Kamimori GH, Balkin TJ. The effects of 53 hours of sleep deprivation on moral judgment. Sleep 2007;30:345–52. [3] Wertz AT, Ronda JM, Czeisler CA, Wright Jr KP. Effects of sleep inertia on cognition. JAMA 2006;295:163–4.

116

G.W. Williams et al. / Journal of Clinical Anesthesia 40 (2017) 110–116

[4] Dinges DF, Pack F, Williams K, et al. Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4–5 hours per night. Sleep 1997;20:267. [5] Jewett ME, Dijk DJ, Kronauer RE, Dinges DF. Dose-response relationship between sleep duration and human psychomotor vigilance and subjective alertness. Sleep 1999;22:171–9. [6] Bonnet MH. Sleep deprivation. In: Kryger MH, Roth T, Dement W, editors. Principles and Practice of Sleep Medicine. 3rd ed. Philadelphia, PA: WB Saunders; 2000. p. 53–71. [7] Harrison Y, Horne JA. The impact of sleep deprivation on decision making: a review. J Exp Psychol Appl 2000;6:236–49. [8] Ayas NT, Barger LK, Cade BE, et al. Extended work duration and the risk of self-reported percutaneous injuries in interns. JAMA 2006;296:1055–62. [9] Belenky G, Wesensten NJ, Thorne DR, et al. Patterns of performance degradation and restoration during sleep restriction and subsequent recovery: a sleep dose-response study. J Sleep Res 2003;12:1–12. [10] Deaconson TF, O'Hair DP, Levy MF, Lee MB, Schueneman AL, Codon RE. Sleep deprivation and resident performance. JAMA 1988;260:1721–5. [11] Rollinson DC, Rathlev NK, Moss M, et al. Effects of consecutive night shifts on neuropsychological performance of interns in the ER. Ann Emerg Med 2003;41:400–6. [12] Sleep, Alertness, and Fatigue in Residency (SAFER) education Module, American Academy of Sleep Medicine 2002. [13] Owens JA, Arnedt J, Crouch M, Stahl J. Sleep loss and fatigue in pediatric residents: self-report of sleep patterns and impact on performance. Sleep 2004;27:A149. [14] Gaba DM, Howard SK. Patient safety: fatigue among clinicians and the safety of patients. N Engl J Med 2002;347:1249–55. [15] Gaba DM, Howard SK, Jump B. Production pressure in the work environment: California anesthesiologists' attitudes and experiences. Anesthesiology 1994;81: 488–500. [16] Amin MM, Graber M, Ahmad K, et al. The effects of a mid-day nap on the neurocognitive performance of first-year medical residents: a controlled interventional pilot study. Acad Med 2012;87:1428–33. [17] Wright Jr KP, Badia P, Myers BL, Plenzler SC. Combination of bright light and caffeine as a countermeasure for impaired alertness and performance during extended sleep deprivation. J Sleep Res 1997;6:26–35. [18] Beaumont M, Batejat D, Pierard C, et al. Slow release caffeine and prolonged (64-h) continuous wakefulness: effects on vigilance and cognitive performance. J Sleep Res 2001;10:265–76. [19] Bonnet MH, Gomez S, Wirth O, Arand DL. The use of caffeine versus prophylactic naps in sustained performance. Sleep 1995;18:97–104. [20] Zhang MR, Red SD, Lin AH, Patel SS, Sereno AB. Evidence of cognitive dysfunction after soccer playing with ball heading using a novel tablet-based approach. PLoS One 2013;8:e57364. [21] Fischer TD, Red SD, Chuang AZ, et al. Detection of subtle cognitive changes after mTBI using a novel tablet-based task. J Neurotrauma 2016;33:1237–46. [22] Sereno AB, Patel SS, Shrestha Y, Red SD, inventors. U.S. Patent Application Publication No. US 2014/0249447. Alexandria, VA: U.S. Patent and Trademark Office. 2014. [23] Ferraz MB, Quaresma MR, Aquino LR, Atra E, Tugwell P, Goldsmith CH. Reliability of pain scales in the assessment of literate and illiterate patients with rheumatoid arthritis. J Rheumatol 1990;17:1022–4. [24] Woodrow SI, Park J, Murray BJ, et al. Difference in the perceived impact of sleep deprivation among surgical and non-surgical residents. Med Educ 2008;42:459–67. [25] Huffmyer JL, Moncrief M, Tashjian JA, et al. Driving performance of residents after six consecutive overnight work shifts. Anesthesiology 2016;124:1396–403. [26] Munoz DP, Everling S. Look away: the antisaccade task and the voluntary control of eye movement. Nat Rev Neurosci 2004;5:218–28. [27] Hutton SB. Cognitive control of saccadic eye movements. Brain Cogn 2008;68: 327–40. [28] Ariani G, Wurm MF, Lingnau A. Decoding internally and externally driven movement plans. J Neurosci 2015;35:14160–71.

[29] Buneo CA, Andersen RA. The posterior parietal cortex: sensorimotor interface for the planning and online control of visually guided movements. Neuropsychologia 2006; 44:2594–606. [30] Scott SH. The role of primary motor cortex in goal-directed movements: insights from neurophysiological studies on non-human primates. Curr Opin Neurobiol 2003;13:671–7. [31] Jeter CB, Patel SS, Morris JS, Chuang AZ, Butler IJ, Sereno AB. Oculomotor executive function abnormalities with increased tic severity in Tourette syndrome. J Child Psychol Psychiatry 2015;56:193–202. [32] Hakvoort Schwerdtfeger RM, Alahyane N, Brien DC, Coe BC, Stroman PW, Munoz DP. Preparatory neural networks are impaired in adults with attention-deficit/hyperactivity disorder during the antisaccade task. Neuroimage Clin 2012;2:63–78. [33] Hood AJ, Amador SC, Cain AE, et al. Levodopa slows prosaccades and improves antisaccades: an eye movement study in Parkinson's disease. J Neurol Neurosurg Psychiatry 2007;78:565–70. [34] Nakayama Y, Yamagata T, Hoshi E. Rostrocaudal functional gradient among the predorsal premotor cortex, dorsal premotor cortex and primary motor cortex in goaldirected motor behaviour. Eur J Neurosci 2016;43:1569–89. [35] Funahashi S, Chafee MV, Goldman-Rakic PS. Prefrontal neuronal activity in rhesus monkeys performing a delayed anti-saccade task. Nature 1993;365:753–6. [36] DeSouza JF, Menon RS, Everling S. Preparatory set associated with pro-saccades and anti-saccades in humans investigated with event-related FMRI. J Neurophysiol 2003; 89:1016–23. [37] Hill SK, Reilly JL, Harris MSH, Khine T, Sweeney JA. Oculomotor and neuropsychological effects of antipsychotic treatment for schizophrenia. Schizophr Bull 2008;34: 494–506. [38] Babin SL, Hood AJ, Wassef AA, Williams NG, Patel SS, Sereno AB. Effects of haloperidolon cognition in schizophrenia patients depend on baseline performance: a saccadic eye movement study. Prog Neuro-Psychopharmacol Biol Psychiatry 2011; 35:1753–64. [39] Denheyer M, Kiss ZH, Haffenden AM. Behavioral effects of subthalamic deep brain stimulation in Parkinson's disease. Neuropsychologia 2009;47:3203–9. [40] Broerse A, Crawford TJ, den Boer JA. Parsing cognition in schizophrenia using saccadic eye movements: a selective overview. Neuropsychologia 2001;39:742–56. [41] Gooding DC, Mohapatra L, Shea HB. Temporal stability of saccadic task performance in schizophrenia and bipolar patients. Psychol Med 2004;34:921–32. [42] Scheer FA, Shea TJ, Hilton MF, Shea SA. An endogenous circadian rhythm in sleep inertia results in greatest cognitive impairment upon awakening during the biological night. J Biol Rhythm 2008;23:353–61. [43] Waage S, Harris A, Pallesen S, Saksvik IB, Moen BE, Bjorvatn B. Subjective and objective sleepiness among oil rig workers during three different shift schedules. Sleep Med 2012;13:64–72. [44] Porcu S, Ferrara M, Urbani L, Bellatrecci A, Casagrande M. Smooth pursuit and saccadic eye movements as possible indicators of nighttime sleepiness. Physiol Behav 1998;65:437–43. [45] Killgore WD. Effects of sleep deprivation on cognition. Prog Brain Res 2010;185: 105–29. [46] Iglehart JK. The ACGME's final duty-hour standards — special PGY-1 limits and strategic napping. N Engl J Med 363:1589–1591. [47] Jarvis MJ. Does caffeine intake enhance absolute levels of cognitive performance? Psychopharmacology 1993;110:45–52. [48] Smit HJ, Rogers PJ. Effects of low doses of caffeine on cognitive performance, mood and thirst in low and higher caffeine consumers. Psychopharmacology 2000;152: 167–73. [49] Mednick SC, Cai DJ, Kanady J, Drummond SP. Comparing the benefits of caffeine, naps and placebo on verbal, motor and perceptual memory. Behav Brain Res 2008;193:79–86. [50] Herscovitch J, Broughton R. Sensitivity of the Stanford sleepiness scale to the effects of cumulative partial sleep deprivation and recovery oversleeping. Sleep 1981;4: 83–91.