Acute and Chronic Sleep Deprivation in Residents – Cognition and Stress Biomarkers

Abstract

Background

Insufficient sleep affects circadian hormonal profiles and inflammatory markers, and may modulate attention, executive functioning and decision-making. Medical professionals and specifically resident physicians, who are involved in long-term nightshift schedules during their post-graduate training, are prone to acute and chronic sleep deprivation and disruption, putting them at risk for making medical errors.

Aim

Evaluate the impact of chronic and acute-on-chronic sleep deprivation and disruption among residents on selected physiological and cognitive measures.

Methods

Thirty-three medical and surgical residents were evaluated twice - at baseline and after a 26-hour shift. Eighteen young attending physicians who did not engage in nightshift schedules served as controls and were evaluated once. Measures included morning cortisol and high-sensitivity C-reactive protein (hs-CRP), computerized tests of attention and behavior, the Behavior Rating Inventory of Executive Function, a risk-taking questionnaire and the Pittsburgh Sleep Quality Index.

Results

Residents, but not attendings, reported chronic sleep disruption and deprivation. Residents at baseline exhibited reduced morning cortisol levels and elevated hs-CRP levels, compared to attendings. Residents at baseline had impaired global executive function compared to attendings. A nightshift with acute sleep deprivation further reduced residents’ executive function. Residents at baseline and after a nightshift demonstrated increased impulsivity and slower processing time than attendings. Residents and attendings did not differ in risk-taking tendencies which were assessed in a separate cohort.

Conclusions

In a real-life setting, resident physicians exhibit increased low-grade systemic inflammation (hs-CRP) and impaired HPA-axis function. Their chronic sleep curtailment is associated with greater impulsivity, slower cognitive processing, and impaired executive function. Future research is warranted to understand how improving working schedule by increasing sleep duration may minimize the short-term and potential long-term risks to physicians in training.

Introduction

Sleep is a pillar of human health¹. Insufficient sleep has deleterious effects on health^2,3, increasing the risk of cardiovascular disease, premature aging and obesity⁴. Insufficient sleep is also associated with poor mental and cognitive functioning^5,6. Even one night of sleep deprivation, defined as having less than 5 hours of uninterrupted sleep, leads to a detectable decline in cognitive performance⁷. Sleep disorders have been recently declared a public-health epidemic⁸, with about 25% of the US adult population suffering from chronic sleep-deprivation⁹. Sleep deprivation is particularly prevalent among shift-workers, including medical professionals¹⁰.

While sleep deprivation and disruption impairs cognitive functions such as attention, executive performance, and risk-taking behaviors, the biological underpinnings of these effects are not fully understood¹¹. It was found that sleep deprivation and disruption affects the circadian hormonal and the immune systems¹². For example, short sleep duration and sleep disruption increases C-reactive protein (CRP) levels^13,14, and impaired sleep quality changes cortisol secretion patterns¹⁵. Relatedly, sleep deprivation increases systemic inflammation, which in turn disrupts the blood-brain barrier¹⁶ (for a more extensive review, see ref.¹⁷). Furthermore, it has been suggested that hormonal and inflammatory pathways can influence cognitive function¹⁸. For example, increased secretion of cortisol impairs some executive functions¹⁹ and higher levels of CRP are associated with impaired cognition²⁰.

The first goal of the current study is to provide a comprehensive cognitive and biochemical evaluation of the effects of sleep disruption in the same study. Second, our study aims to assess the differential effects of chronic and acute-on-chronic sleep deprivation on cognitive impairments, and their biological underpinnings. Notably, most research to date has investigated the effect of acute sleep deprivation, i.e. a single night of insufficient sleep. Little information exists regarding the cognitive effects associated with chronic sleep deprivation, i.e. repetitive bouts of insufficient night sleep over prolonged periods of times²¹. Acute total sleep deprivation cannot be extrapolated to predict the effects of chronic sleep deprivation since they have common but also different neuro-behavioral consequences²².

We chose to focus our investigation on resident physicians. During their postgraduate training, residents are involved in nightshift schedules, which subject them to acute sleep deprivation and disruption, as well as to stress and fatigue²³. Since their residency lasts several years, their sleep disruption is also chronic^24,25. This provides a unique opportunity to test both chronic and acute-on-chronic sleep deprivation. Furthermore, cognition impairment among physicians is particularly worrisome, as such impairment may cause medical errors that might harm patients^26,27. Some studies documented the overall impact of nightshift schedules on residents’ cognitive and performance skills^28–35. For example, implementation of schedules with protected sleep period while on call resulted in an increase in overnight sleep duration and improved alertness the morning after the nightshift³⁰. Other studies have focused on the assessment of residents before and after a night with little sleep. For example, the effect of a nightshift on residents’ cognitive performance is comparable to the effect of ingesting alcohol at levels close to the legal limit for driving³⁶. The current study goes beyond previous research on residents’ sleep in (1) testing the hormonal and inflammatory pathways that may connect sleep and cognition, a (2) testing both chronic and acute-on-chronic effects of sleep deprivation and disruption on several cognitive and behavioral measures.

Our study evaluated three groups of physicians: (a) Young attending physicians who do not participate in a nightshift schedule and have no evidence of acute or chronic sleep deprivation; (b) Resident physicians at baseline who participate in a nightshift schedule, and thus suffer by definition from chronic sleep deprivation and disruption. This group has slept well for three consecutive nights and thus did not suffer from acute sleep deprivation; (c) Resident physicians post-call - same residents as in (b) evaluated a second time in the morning after a nightshift. This design allowed us to test the isolated effect of chronic sleep disruption, by comparing residents at baseline to attendings. It further allowed us to examine the additive effect of acute sleep deprivation superimposed on chronic sleep deprivation and disruption by comparing residents to themselves after a nightshift. Since the vast majority of residents suffer from chronic sleep disruption due to shift-work, we did not examine the isolated effect of acute sleep deprivation in a non-sleep deprived population. We expected to find impairments in residents at baseline as compared to attendings, due to chronic sleep deprivation and disruption (even though both groups had slept the night before the assessment). We predicted that acute sleep deprivation would further deteriorate residents’ cognitive and biological impairments and exacerbate the differences between resident and attending physicians.

Methods:

Study population

Fifty-one physicians (16 females) from various medical and surgical specialties (supplemental Table 1) were recruited (age: 32.5±4.9 years, range: 25–46 years), 33 residents and 18 attending physicians (Table 1). To minimize age differences between the groups, we recruited young attendings (age 46 and below), although the residents recruited were still younger on average than attendings. This difference was adjusted for in later analyses. All participants were recruited from the medical staff personnel of the Hadassah Medical Center, Jerusalem, Israel. The experimenter personally called physicians at the hospital and invited them to take part in the study, in return for a breakfast voucher. We screened for stimulant medication use and excluded those physicians who were using stimulants for attention-deficit disorder (n=2). Caffeine consumption was highly prevalent in all participants and was not accounted for in the study. All residents performed between 4 to 8 26-hour nightshifts per month during a residency period lasting 4–6 years (median time in residency – 2 years, interquartile range – 2 years). Thus all residents, by definition, suffered from chronic sleep deprivation and disruption. To ensure acute sleep deprivation in the resident group post-call, only those residents who slept less than 5 hours during a nightshift (lasting 26 hours) were included. Furthermore, baseline residents’ assessment was only performed after residents were off nightshifts for the preceding three nights. All attendings were recruited from specialties that do not involve any nightshifts. Attendings were recruited only if they stated that they have slept more than 5 hours for the preceding three nights. All subjects provided written informed consent, and the study was approved by the Helsinki Committee (IRB) of Hadassah-Hebrew University Medical Center Jerusalem, Israel (protocol # - 0065-16 HMO).

Table 1.

Demographic and Sleep Characteristics in Resident and Attending Physicians

Baseline Characteristics	Resident Physicians (n=33)	Attending Physicians (n=18)	p value
Age (years)	30.5±3.7	36.3±4.8	<0.001
Sex (Male/Female)	21/12	14/4	0.34
BMI (kg/m2)	25.1±3.8	25.4±2.9	0.76
Married (%)	63.6%	86.6%	0.03
Hours slept the night before testing (hours)	6.1±0.9	6.5±0.9	0.05
Average sleep per day in past month (hours)	6.2±0.6	6.9±0.7	0.001
Self-reported sleep quality (PSQI)
Sleep quality	0.8±0.7	0.6±0.7	0.32
Sleep latency	0.8±0.9	0.4±0.5	0.17
Sleep duration	1±0.7	0.5±0.6	0.01
Sleep efficiency	0.1±0.4	0±0	0.09
Sleep disturbances	1±0.5	0.9±0.4	0.67
Sleep medication use	0.06±0.2	0±0	0.30
Daytime dysfunction	1±0.9	0.8±0.6	0.41
Global PSQI	5.1±2.5	3.4±1.9	0.02

Data are presented as Mean ± Standard Deviation

Study Design:

All participants completed a demographic questionnaire and the Pittsburgh Sleep Quality Index (PSQI) followed by the MOXO-CPT task, the BRIEF-A questionnaire, and blood draw between 08:00-10:00 AM. Residents completed the assessments on two different occasions, at the same time of the day, after a nightshift and after three nights of normal sleep. To avoid an order effect, residents were randomly assigned to perform the post-call assessment first or second. Risk taking was measured in a separate cohort (30 residents at baseline, 34 residents after nightshift and 28 attendings) (Figure 1, supplementary Table 2).

Figure 1:

Study Design

Sleep Assessment:

The self-rated items of the PSQI assess sleep quality and disturbances over a 1-month time interval. It generates seven component scores³⁷ (range of subscale scores, 0–3), namely, sleep quality; sleep latency; sleep duration; habitual sleep efficiency; sleep disturbance; use of sleeping medication; and daytime dysfunction. The sum of these seven component scores yields a global score of subjective sleep quality (range, 0–21); higher scores represent poorer subjective sleep quality, and a score above 5 is considered indicative of poor sleep³⁸.

Attention Assessment:

The MOXO-Continuous Performance Test (MOXO-CPT) is an eight-stage standardized computerized test designed to assess attention performance and screen for ADHD related symptoms. The total duration of the test is 18 minutes^39,40. Typical CPT tasks require the participant to sustain attention over a continuous stream of stimuli and respond to a pre-specified target⁴¹. Performance indices include scores in four domains: attention, timing, impulsivity, and hyperactivity, whereby higher scores (number of correct/incorrect attempts) indicate a higher degree of impairment in the specific domain.

Executive Function Assessment:

The Behavior Rating Inventory of Executive Function-Adult Version (BRIEF-A) is a 75-item self-report questionnaire exploring everyday behaviors, in which executive functions are implicated. It is designed to capture higher-order cognitive processes required to properly engage in real-world, goal-directed behaviors, and has been validated in different clinical settings⁴². The BRIEF-A consists of nine subscales (Inhibit, Shift, Emotional Control, Self-Monitor, Initiate, Working Memory, Plan/Organize, Task Monitor, and Organization of Materials), two index scales (Behavioral Regulation Index (BRI) and Metacognition Index (MTI)), and a general composite score, the Global Executive Composite (GEC). The BRI is composed of the Inhibit, Shift, Emotional Control, and Self-Monitor scales and measures the ability to appropriately regulate behavioral and emotional responses. The MTI is composed of the Initiate, Working Memory, Plan/Organize, Task Monitor, and Organization of Materials scales and measures the ability to actively organize information and solve problems as they are encountered during completion of complex tasks. Higher scores reflect lower executive function, with T scores above 65 considered clinically significant.

Risk-Taking Assessment:

We employed a previously validated questionnaire to examine risk-taking tendency⁴³. We used two parallel sets of questions to increase reliability. The first set consisted of 10 questions in each of which participants had to choose between a lottery and a sure gain. The lottery always offered a 50% chance to gain 100 Shekels (NIS). The sure option offered a sure gain; the size of the gain ranged from 5 NIS in the first question to 95 in the 10^th question, in 5 NIS increments. For example, participants had to indicate whether they preferred a 50% chance to gain 100 NIS or a sure gain of 60 NIS. The second set of questions consisted of 11 items with different amounts. The lottery always offered a 50% chance to gain 15 NIS. The sure option offered gains that ranged from 1 NIS in the first question to 11 NIS in the 11^th question, in increments of 1 NIS. The number of questions in which participants chose the lottery option was added, yielding a global risk-taking score, ranging from 0 (no risk-taking) to 21 (maximum risk-taking).

Biochemical analyses of inflammatory and hormonal markers:

Venous peripheral blood was drawn from all participants at 08:00-10:00 AM following the cognitive assessments. Serum was sent immediately for analysis at a central laboratory for high-sensitivity C-reactive protein levels (hs-CRP, expressed as mg per dL) and free cortisol levels (expressed as nmol per L).

Statistical analyses:

Continuous variables were compared using paired t-tests between residents at baseline and residents post-call, and unpaired t-tests between residents at baseline and attendings. Dichotomous variables were compared with χ² tests or Fisher’s exact test. Correlations between variables were assessed by Pearson’s correlation coefficient. hs-CRP and cortisol levels were naturally log-transformed before analysis. To control for age differences between the study groups, we followed up the correlations with linear regression in which the age variable was introduced. Statistical significance was assumed at a p-value < 0.05. All data were analyzed using SPSS version 25 (IBM, NY, USA).

Results

Sleep disruption

In the PSQI, residents reported shorter average sleep duration during the preceding month as compared to attendings (6.2±0.6 vs. 6.9±0.7 hours/night, p=0.001). Residents also reported poorer sleep quality than attendings (PSQI scores: 5.1±2.5 vs. : 3.4±1.9, respectively, p=0.02). Residents slept 3±1.2 (range: 1–5) hours on the nightshift before the study, and of those a mean of 2.4±1.4 hours of continuous uninterrupted sleep (Table 1).

Attention, executive function, and risk-taking

Significant differences emerged when comparing MOXO-CPT scores in residents at baseline to attendings. Residents exhibited increased impulsivity and slower processing speed, whereas attention and hyperactivity did not differ. In a regression analysis where age was also introduced as an independent variable, those differences were not statistically significant. None of the four attention domain scores differed between residents at baseline and residents post-call (Table 2).

Table 2.

Cognitive functioning in Resident and Attending Physicians

The MOXO-Continuous Performance Test Scores
	Attending Physicians	p value¥	Resident Physicians Baseline	Resident Physicians post-call	p value*
Attention	33.3±0.3	0.37	33.3±0.8	33.3±0.6	0.89
Timing	24.7±4.6	0.05	27.2±3.2	26.6±3	0.35
Impulsivity	0.7±0.5	0.03	1.1±0.76	1.2±1.1	0.29
Hyperactivity	0.4±0.4	0.09	0.8±1.2	0.7±06	0.42
Behavior Rating Inventory of Executive Function (T scores)
Behavioral Regulation	44.7±7.5	0.05	49.2±9	52.6±9.3	0.18
Metacognition	42.3±5.9	0.002	48.5±8.8	51.3±10.4	0.05
Global Executive Composite	42.7±6.5	0.008	48.7±8.5	53±10.5	0.02

Data are presented as Mean ± Standard Deviation

^*Within-Subject comparison before and after night shift

^¥Between-subject comparison of baseline resident physicians’ vs attending physicians’ parameters. MOXO-CPT scores indicate number of correct/incorrect attempts – higher scores indicating poorer performance. Lower T scores on BRIEF-A indicate better executive function.

The BRI and MTI sub-scores of the BRIEF-A, and accordingly the GEC score, were all higher in the residents at baseline as compared to attendings, indicating impaired executive function. Residents post-call had even higher scores than residents at baseline (Table 2, Figure 2). In a regression analysis where age and/or type of specialty (medical/surgical) were introduced, the pattern of differences between the GEC scores of attendings and residents did not change.

Figure 2:

BRIEF-A scores in Attending Physicians and in Resident Physicians both before or after a nightshift

Risk-taking tendencies did not differ between the three groups (Supplemental Table 2).

Biomarkers of inflammation and hormonal function

Average hs-CRP levels in residents at baseline were marginally higher than those of the attendings (0.28±0.24 vs. 0.16±0.14 mg/dL, p=0.067). hs-CRP levels did not significantly differ between residents at baseline and residents post-call (p=0.43). We further conducted a comparison of hs-CRP levels of each group to the previously established threshold of 0.3mg/dL, which has been associated with increased cardiovascular risk and mortality⁴⁴. hs-CRP levels among residents at baseline were in the vicinity of this risk-associated cut-off value (t(32)=−0.391, p=0.698) while hs-CRP levels among attendings were significantly lower (t(16)=−3.95, p=0.001). Additionally, 16/33 (48.5%) of the residents at baseline had hs-CRP levels of 0.3mg/dL or higher, compared with only 4/17 (23.5%) of the attendings (p=0.129) (Figure 3). In a regression analysis where age was introduced as an independent variable, we obtained the same pattern of results.

Figure 3:

hs-CRP and Cortisol levels in Attending Physicians and in Resident Physicians both before or after a nightshift

Significant differences emerged when comparing morning serum cortisol levels scores in residents at baseline to attendings (275±95 vs. 358±115 nmol/L, p<0.001). Additionally, cortisol levels in residents at baseline were lower than those of attendings post-call (396±127 nmol/L; p<0.001) (Figure 3). In a regression analysis where age was introduced as an independent variable, the same pattern of results emerged.

Associations between sleep disruption, cognition and biomarkers

In a correlation analysis, three out of seven sleep components measured by the PSQI were associated with greater impairment of executive function – lower sleep efficiency, reported sleep disturbances and daytime dysfunction. Furthermore, shorter sleep duration was associated with reduced attention. Finally, daytime dysfunction was associated with higher hs-CRP levels and reported sleep disturbances with increased cortisol. See Table 3 for the full list of correlations of sleep parameters.

Table 3.

Correlations between chronic sleep disturbances and cognitive measures and serum biomarkers

	CRP	CORTISOL	BEHAVIORAL REGULATION	METACOGNITION	GLOBAL EXECUTIVE COMPOSITE	ATTENTION	TIMING	IMPULSIVITY	HYPERACTIVITY
SLEEP QUALITY	0.212	−0.115	−0.059	0.108	0.045	0.077	−0.156	0.094	0.097
SLEEP LATENCY	−0.119	0.129	−0.206	−0.147	−0.175	−0.136	0.110	−0.284	−0.125
SLEEP DURATION	−0.116	−0.174	0.064	0.112	0.108	−0.304*	0.162	0.136	−0.132
SLEEP EFFICIENCY	0.137	0.027	0.378**	0.209	0.310*	−0.046	−0.079	0.030	0.068
SLEEP DISTURBANCES	0.166	0.337*	0.409**	0.171	0.301*	−0.227	−0.222	0.165	0.168
SLEEP MEDICATONS	0.297*	0.045	0.039	−0.042	−0.013	0.059	−0.083	−0.172	−0.016
DAYTIME DYSFUNCTION	0.321*	0.051	0.339*	0.365**	0.389**	0.166	0.029	0.192	0.070
GLOBAL PSQI	0.174	0.052	0.198	0.217	0.239	−0.110	−0.018	0.069	0.020

Data are presented as r correlation coefficient

^*Correlation is significant at the 0.05 level (2-tailed).

^**Correlation is significant at the 0.01 level (2-tailed).

Discussion

The current study explored the effects of chronic and acute-on-chronic sleep deprivation and disruption, associated with residency training, on cognitive performance, systemic inflammation and cortisol levels. Participants were residents before and after a nightshift and young attending physicians. We found that chronic sleep disruption and deprivation, caused by the overall schedules of residency training, was associated with poorer executive functioning and increased impulsivity, and was accompanied by lower morning cortisol levels and a trend towards increased inflammation. Acute partial sleep disruption, following a nightshift, was associated with further executive function impairment in residents and with increased cortisol secretion. Impulsivity and inflammation were not affected beyond the impairment levels associated with chronic sleep deprivation and disruption.

Attention, executive function and risk taking

Cognitive performance and behavior were altered in our cohort of chronically sleep-deprived resident physicians in a dose-response manner. Residents had higher scores than the attendings, indicating a greater executive function impairment. Executive functioning is viewed as the central control system underpinning complex behaviors, such as attention, planning and goal-setting, inhibitory abilities and cognitive flexibility^45,46. Sleep deprivation studies have usually reported executive functioning domain-specific associations, such as cognitive control in task-goal switching⁴⁷, motor response inhibition⁴⁸, reaction time and inhibitory control⁴⁹, and divergent thinking⁵⁰. We tried to overcome these limitations by using the BRIEF-A, a validated questionnaire designed to capture a wide range of higher-order cognitive processes required to properly engage in goal-directed behaviors (although not validated previously on physicians). Our study further expands the current base of knowledge by assessing executive functioning under natural settings and real-life conditions, providing greater ecological validity.

Chronic sleep deprivation was associated with impulsive behavior in residents in our study, which was not further exacerbated by acute sleep deprivation, as measured by the MOXO-CPT computerized tool. Our results are supported by some prior evidence: self-regulation impairments, including loss of the ability to manage cognition and emotion, that are manifest as difficulties in controlling impulsive behavior and maintaining adequate attention are affected by prolonged lack of sleep^51,52. Similar to our results, another study found no significant increase in impulsive behavior following acute sleep deprivation⁵³. Impulsivity, as seen here in the context of relatively young, inexperienced physicians, making critical decisions, often under stress, may result in compromised medical care. Indeed, some studies have shown that extended 24-hour shifts increase the rate of medical errors⁵⁴. At the same time, other studies found that extending physicians’ sleep duration did not improve patient-centered outcomes^55–58.

Risk-taking scores were similar across the three groups in our study. Several studies suggested an increase in physicians’ risk-taking behavior after acute sleep deprivation^59,60. However, studies investigating risk taking in other settings failed to reach a consistent conclusion^61,62.

Biomarkers of inflammation and hormonal function

A trend towards an increased systemic inflammation, as reflected by elevated hs-CRP, was detected in residents compared to attendings. This finding concurs with a long list of studies associating inflammation with total night sleep deprivation, partial night sleep deprivation, or chronic restriction of sleep duration¹². From existing literature, it appears that persistent disturbances of sleep are necessary for inflammatory signaling to be translated into an increase in systemic markers of inflammation⁶³. Accordingly, we did not observe increases in hs-CRP levels following a single night of partial sleep deprivation during the nightshift. Our findings are also in line with a large meta-analysis⁶⁴ where chronic sleep disruption, but not shorter sleep duration, was linked to increased systemic inflammation. Notwithstanding, the elevated hs-CRP levels in the residents group, which were in the higher cardiovascular risk range, deserve attention, and potentially suggest that the working schedules of residents may be fraught with an enhanced risk contributing to both cardiovascular and all-cause mortality⁴⁴. Physicians are clearly exposed to chronic stress and have been reported at increased risk for burnout and depression ^{34,35,65–67}, all of which are also linked to increased inflammation and cardiovascular disease⁶⁸. Thus, an improved understanding of the major determinants of the adverse consequences of residents’ schedules is critical to enable personalized tailoring of such training and risk mitigation.

Morning cortisol levels were reduced in the residents group with chronic sleep deprivation (baseline), and rose to normal levels, similar to those of the attendings, following acute sleep deprivation (post call). Since blood samples were drawn between 08:00 and 10:00 AM, coinciding with the physiological peak in blood cortisol, the reduced cortisol concentrations could be related to two major processes: (a) the HPA pathway has reconfigured its responsiveness in the context of chronic stress. The sustained requirements to increase cortisol after partial sleep disruption during a nightshift led to the dampening of the baseline cortisol output. This explanation is supported by recent evidence showing that two nights of 4 hours in bed were associated with a reduction in ACTH response to CRH injection, suggestive of a decreased pituitary sensitivity to CRH in a state of sleep debt⁶⁹. (b) the normal timing of morning cortisol rise is shifted to a different time of the day. Acute insufficient sleep results in elevation of evening cortisol levels^70,71, which has been hypothesized to reflect altered HPA axis recovery from the circadian-driven morning stimulation^71–73. As we tested the cortisol levels only once, this study does not permit any inferences as to which of the two possibilities is at play. Still, our findings are important since the homeostatic regulation of cortisol is crucial to a myriad of physiological processes, including innate immunity, which could explain the elevated hs-CRP levels⁷⁴ in residents at baseline. Moreover, low cortisol has been suggested as a key factor in the development of stress-related disease and burnout^15,75, and our findings are in line with the evidence that HPA axis activity decreases with longer duration of chronic stress⁷⁶. While lower morning cortisol in residents at baseline has not been documented before, prior studies documented dysregulation of the HPA axis in residents after a nightshift^77–79.

Strengths and limitations

The strength of our study lies in the comprehensive cognitive and biochemical assessment of the chronic effects of interrupted sleep over time in a real-life environment. Although night shiftwork is associated with shorter sleep duration⁸⁰, many of the detrimental effects attributable to shiftwork seem to originate from the associated chronic circadian misalignment⁸¹. Our study reflects the combined effects of circadian disruption associated with iterative changes in day and night work schedules, as well as the sleep disruption involved in the specific work schedule during clinical residency training.

The observational (rather than experimental) nature of the study encompasses some methodological limitations that should be acknowledged. Apart for disrupting sleep, residency introduces also differences in work load and stress^35,82,83. Resident physicians may differ from attendings in age, lifestyle, specialty, income, and professional experience (note that we recruited the youngest, least experienced attendings, and controlled for age in our analyses). While we find that sleep was associated with the different outcomes, it is still possible that other factors also contributed to our observed effects. However, attendings were the closest possible control, as practically all residents in our medical system participate in nightshifts and medical students are less suitable to serve as a control group.

Cognitive functioning was measured in our study through validated, well-established behavioral tools^84,85. Sleep was assessed with the PSQI, a widely used questionnaire which has been applied to a large variety of settings, including shift work, with strong reliability and both internal and external validity³⁸. Still, the PSQI is based on subjective reporting, which may have led to underestimation of the true degree of sleep deprivation⁸⁶. Even if underestimation occurred, it occurred in all groups, and cannot account for the observed differences. Additionally, in contrast to laboratory conditions, we could not monitor all of our cognitive and biological outcome measures throughout the day to assess for circadian changes and for windows of a specific vulnerability.

The current study was cross-sectional, and did not investigate the potential longitudinal effects of the above-mentioned cognitive perturbations. Future research may unravel long-lasting effects of chronic sleep disruption on health and cognitive domains. Future research is also needed to test whether letting residents sleep more (e.g., by limiting the on-call period) may solve some of the problems documented in this study. We note that adding sleep may not be sufficient for improving patient care, due to residents’ overall fatigue and burnout^23,87. Such changes in training schedules, while improving sleep, might cause deterioration in education opportunities and interruption in continuity of care^56,57 and result in more medical errors⁵⁸.

Conclusions

In conclusion, our findings suggest that chronic sleep deprivation and disruption caused by postgraduate residency training impose adverse cognitive effects of impaired executive function and increased impulsivity. These are also associated with maladaptive cortisol levels and a trend towards increased low-grade systemic inflammation. Some of the effects of chronic sleep deprivation are further aggravated by the acute sleep restriction of night calls. Further research is needed to examine whether incorporating biological and functional considerations into working schedules, may minimize the short-term and potential long-term risks of physician training.

Abbreviation List

BRIEF-A

Behavior Rating Inventory of Executive Function-Adult Version

BRI

Behavioral Regulation Index

GEC

Global Executive Composite

hs-CRP

high-sensitivity C-Reactive Protein

MOXO-CPT

The MOXO-Continuous Performance Test

MTI

Metacognition Index

PSQI

Pittsburgh Sleep Quality Index