Physiological and psychological impacts on male medical interns during on-call duty

Abstract

This study aimed to explore the physiological and psychological impacts on male medical interns during on-call duty (OCD). A prospective study of 13 medical interns was carried out using repeated testing during an on-call internal medicine and a duty-free course. There are 10 OCDs per month and each duty is composed of 33.5 consecutive work hours followed by 2 duty-free days for a time period of 3 months. Spectral analysis of R–R intervals was performed to evaluate the participants' heart rate variability. Among the three main spectral components distinguished, a high-frequency (HF) component of 0.15–0.4 Hz and a low-frequency component (LF) of 0.04–0.15 Hz were analyzed. The ratio of LF and HF (LF/HF) was also calculated. Reduced LF/HF, which suggests reduced cardiac sympathetic modulation, was found during duty night work. Increased HF was identified during the first sleep after OCD, which corresponded to the sleepiness scales. The OCD causes temporary emotional influence by the findings of higher anxiety and depression scores. In addition, inattention and impulsiveness, as represented by the continuous performance test score and lower nocturnal sympathetic modulation, were found, which indicated the differences between the internal medicine course and duty-free course. Our study suggests that 10 stressful on-call duties monthly result in reduced sympathetic modulation during duty night work, high anxiety and depression state, and a persistent inattention.

Keywords::

• On-call duty
• medical interns
• heart rate variability
• autonomic nervous system
• sleepiness

Introduction

Within hospitals, among all trainees, interns typically work the greatest number of hours per week (Baldwin and Daugherty 2004). Moreover, on-call duty (OCD) among medical interns is not only characterized by sleep deprivation but also by mental stress. Populations with different degrees of stress and sleep deprivation are also subjected to various types of fatigue-associated negative effects (Takeyama et al. 2005). Therefore, extended shifts and long working weeks among interns may make them especially prone to fatigue-induced errors.

The autonomic nervous system (ANS) is the primary regulator of heart rate. Rhythmic fluctuations in the frequency of impulse conduction along the vagus nerves, modulated by the rate and depth of breathing, result in substantial variations in R–R intervals (RRs), which is known as respiratory sinus arrhythmia (Hirsch and Bishop 1981). RRs are also affected by mental or physical activity, which reduces the average frequency of the impulses conducted along the vagus nerves and increase the activity of the sympathetic nervous system. The application of heart rate variability (HRV) analysis has recently gained popularity as a means of quantifying ANS functioning noninvasively (Malliani et al. 1991; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology 1996; Kuo et al. 2005). Spectral analysis of HRV by Fourier transformation has been categorized into high-frequency (HF), low-frequency (LF), and very LF powers. HF is considered to represent vagal control of the heart rate. The ratio LF and HF (LF/HF) is considered by some investigators to mirror the sympathovagal balance or to reflect sympathetic modulations (Malliani et al. 1991; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology 1996; Kuo et al. 2005).

The brain triggers stress responses that are commensurate with the nature of the stimuli during OCD. Physical stressors, such as sleep deprivation and overtime work, require an immediate systemic reaction that is triggered by these reflexive mechanisms. The brain also responds to psychogenic stressors based on prior experience or innate programming. These responses require processing in the forebrain and can occur in anticipation of or in reaction to stressful events (Herman et al. 2003). The ANS responds to stressor exposure, which can be broadly defined as an actual or anticipated disruption of homeostasis or an anticipated threat to the well-being of an individual, through its sympathetic and parasympathetic arms (Ulrich-Lai and Herman 2009). The consequence of stress can be classified into cognitive, emotional, and somatic responses (Wulff et al. 2010). Therefore, we propose that OCD will impact on autonomic activity, assessment of which will provide an immediate measure of stress, and that the stress-related responses ought to be significant among interns.

A limited number of previous studies have described autonomic modulation among shift workers and surgeons during their work. The surgeons presented with sympathetic hyperactivity during surgery, but night shift workers demonstrated a great reduction in sympathetic activity (Furlan et al. 2000). Considering the properties of work, this study is designed to test the hypothesis that on-call interns have a reduction in sympathetic activity as well as poor performance during night shift work in a similar manner to other shift workers.

The specific aims of this study were (1) to analyze the autonomic functional changes using frequency domain analysis of the HRV among interns and (2) to identify any correlations between subjective evaluations, physiological responses, and overall performance in the same population.

Materials and methods

Participants

We recruited 13 medical interns who were trained at Chang Gung Memorial Hospital for 1 year. They were volunteer recruits from the seventh grade of a medical college student population. All participants were healthy male, aged 25.3 ± 1.9 (mean ± SD) years. They did not suffer from hypnotic drug abuse or excess alcohol, caffeine, or nicotine consumption during the entire period of study. None of them had any medical condition known to involve sleep or the ANS, such as a psychiatric or cardiovascular disease. A 20-min PowerPoint presentation was given to all potential participants to outline the aims of the study. They were aware of this study's intention to obtain objective and subjective evaluations of their performance, assessments of their autonomic functioning, and measurements of their sleepiness and emotional state. The on-call day comprised routine work from 7:30 to 17:00, followed by OCD for 15 h. Thus, the on-call phase lasted for 3 consecutive days: the pre-call day, the on-call day, and the post-call day. The assessments took place during the third month of the internal medicine course. After 3 months, the tests were repeated during the third month of a duty-free course, which acted as a self-control group; such courses included nuclear medicine, pathology, or radiology. The duty-free courses have some specialty, which contained less loading for interns compared with medical intern course because the first line medical staffs are residents not interns. Participants were given a detailed description of the study, and individual informed consent was obtained in written form. The study was conducted from October 2007 to February 2008. The study protocol was approved by the Ethics Committee of Chang Gung Memorial Hospital.

Heart rate variability

Participants underwent electrocardiographic (ECG) recording by means of a miniature physiological signal recorder (TD1, Taiwan Telemedicine Device Company, Taiwan; Kuo and Yang 2009). The small size (5.2 × 3.1 × 1.2 cm) and low weight (11 g) of the recorder produced minimal interference with work or stress on the participants. Figure 1 outlines the protocol. Each intern was monitored by the two-lead digital ambulatory ECG recorder with an accelerometer, which was attached at 21:00 before their sleep during the first night (pre-call). The device continuously recorded for 44 h until removal at 17:00 during the post-call evening. On the following 2 days, the investigation lasted from 21:00 to 17:00 of the next day, which was identified as the post-call day and the second post-call day. After 3 months, the objective evaluation of the ANS was repeated during 1 day of one of their duty-free courses.

Figure 1.  Study design. Participants were aware of this study's intention to obtain objective and subjective evaluations of their performance (CPT), autonomic functioning (HRV), sleepiness (ESS; SSS), and emotional state (HADS). The on-call day comprised routine work from 7:30 to 17:00, followed by OCD for 15 h. The on-call phase lasted for 3 days, one cycle after another; these were the pre-call day, the on-call day, and post-call day. Measurements were obtained during the third month of the internal medicine course. After 3 months, the tests were repeated during the third month of a duty-free course and this was used as a self-control group.

The power spectral density was calculated by computerized fast Fourier transformation to evaluate the RR tachograms by frequency-domain measures. The RR signals to be analyzed were truncated into successive 64 s (4096 points) time segments (windows or epochs) with 50% overlapping. A Hamming window was applied to each time segment to attenuate the leakage effect (Kuo and Chan 1993). We also handle the artifacts by the RR rejection procedure. A temporary mean and SD of all RRs were first calculated for standard reference. Each RR was then validated: if the standard score, i.e. (value-mean)/SD, of an R–R value exceeded three, it was considered erroneous or nonstationary and was rejected (Kuo et al. 1999). The average percentile of R–R rejection according to this procedure was 1%.

Our algorithm then estimated the power density of the spectral components based on fast Fourier transformation every 5 min. The resulting power spectrum was corrected for attenuation resulting from sampling and the application of the Hamming window. Among the three main spectral components distinguished, a HF component of 0.15–0.4 Hz and a LF of 0.04–0.15 Hz were analyzed. The LF/HF was also calculated. All the parameters were logarithmically transformed to correct for the skewness of distribution (Kuo et al. 1999).

Sleep data and accelerometry

We obtained sleep data using a weekly sleep log together with an accelerometer that was attached to the recorder. Acceleration values were stored in the flash memory for each axis, namely X (mediolateral), Y (vertical), and Z (anteroposterior). Each axis had a sampling frequency of 125 Hz and could detect accelerations ranging from − 3 to 3 G.

A vectorial magnitude was calculated as . The quantified magnitude of physical activity was estimated by calculating the root mean square of the vectorial magnitude for each time period (epoch). The accelerometer was used to measure physical activity to control possible effects of physical activity on ANS (Brage et al. 2004). During the calculation of physical activity from three-dimensional acceleration, the signals were mixed together to obtain a vector, , and the vector variation of physical activity was used to quantify physical activity.

Sleepiness evaluation

As a measure of overall sleepiness, interns completed the Epworth Sleepiness Scale (ESS; Johns 1991) during their internal medicine course and then again during their duty-free course. Moreover, to get a sense of the acute changes in sleepiness, interns reported their subjective fatigue using the Stanford Sleepiness Scale (SSS), a validated seven-point scale with corresponding hierarchical statements to describe alertness. On this scale, which has been used in previous studies of medical residents, a score of 1 indicates “feeling active and vital, alert, wide awake” and a score of 7 indicates “almost in reverie, sleep onset soon, losing struggle to remain awake” (Hoddes et al. 1973). During the on-call phase, there were five different time points recorded: noon of the pre-call day, noon of the on-call day, night of the on-call day, morning of the post-call day, and noon of the post-call day. During the control phase, the scale was completed at noon on 2 different days. The SSS data of the not OCD time point and pre-call noon time point were the average of day 1 and day 4 in internal medicine course and the average of day 1 and day 2 in duty-free course, respectively. The ESS and SSS were both used in previous study to assess medical resident's sleepiness (Saxena and George 2005).

Hospital anxiety and depression scale

The Hospital anxiety and depression scale (HADS) comprises seven items in the anxiety subscale (HADS-anxiety) and seven items in the depression subscale (HADS-depression); it does not include any somatic symptoms (Zigmond and Snaith 1983). Each item scores from 0 to 3; therefore, the total score ranges from 0 to 21 for the anxiety and depression subscales. A score above 11 is considered clinically significant for each subscale. HADS was widely used in general population outside the patients in the hospital (Caplan 1994; Fujino et al. 2006; Chan et al. 2010). HADS was also applied to investigate physicians' anxiety and depression (Klaghofer et al. 2010). During the on-call phase, the scale was completed at night during the pre-call, on-call, and post-call days. During the control phase, the scale was completed at night on 2 different days. The Cronbach's αs for the anxiety subscale are 0.923 on the duty-free day, 0.814 on the pre-call day, 0.799 on the duty day, and 0.814 on the post-call day. The Cronbach's αs for the depression subscale are 0.899 on the duty-free day, 0.841 on the pre-call day, 0.797 on the duty day, and 0.841 on the post-call day.

Continuous performance test

The continuous performance test (CPT) is a psychological test for humans and it primarily measures attention. The task presents a time series of visual stimuli to the participant. The participant is instructed to press the space bar in response to a particular single stimulus out of the available set (X task; Hsieh et al. 2005). In this study, participant responses were recorded automatically in a computer using the software Conners' CPT II for Windows. During the on-call phase, the test was performed at noon during the pre-call, on-call, and post-call days. During the control phase, the test was performed at noon on 2 different days. There are six measures: omissions, commissions, hit reaction time (Hit RT), preservations, Hit RT block change, and Hit RT inter-stimulus interval (ISI) change. Omission errors result from a failure to respond to the target. Commission errors are made when a response is given to a non-target. Hit RT is the average speed of the correct responses for the entire test. Preservations are counted when any reaction time is < 100 ms. Hit RT block change measures changes in the reaction time across the duration of the test. Hit RT ISI change examines changes in average reaction time at the different ISIs.

Statistical analysis

Heart rate continuously changes within a person, and repeated measurements were made during heart rate monitoring; we could not treat such measurements as fully independent of the participant. In other words, there were auto-correlations for every heart rate measurement within the same participant, and such measurements were seen as a “random effect” within the same participant. All of the major determinants of heart rate were included in the statistical model in a multiple regression analysis, and they were treated as “fixed effects.” On the basis of this, a linear mixed regression model, a sort of multiple linear regression model, which was powerful to deal with repeated measurements for individuals and contained both fixed and random effects, was used for the statistical analysis of HRV. We analyzed all of the continuous data by a linear mixed regression model, and no data were rejected. We stratified this dataset into sleep or waking groups rather than treat it as a binary explanatory variable. We set four explanatory variables: three of which were binary and one of which was continuous. Quantified activity, as measured by the same recorder, TD1, was set as the continuous explanatory variable. We corrected all HRV indices, RR, HF, LF, and LF/HF, by the value of ln(activity) in the linear mixed-effect model (Tables I and II). In other words, we excluded the confounding factor of position and activity. The first binary explanatory variable representing the long-term effect of the duty course was the internal medicine course. The remaining binary explanatory variables that implied the short-term influence of the duty night during internal medicine were duty night work, post-call, and pre-call, respectively. The long-term effects mean the effects of 3-month internal medicine course compared with duty-free course. In contrast, the short-term effects mean the effects of a single OCD event.

Table I.  Regression coefficients and SD resulting from the construction of a linear mixed-effect model of the HRV among all participants during the awake period.

	Dependent variable
Independent variable	RR (ms)	HF [ln(ms^2)]	LF [ln(ms^2)]	LF/HF [ln(ratio)]
Intercept	487.3(15.2)	3.53(0.153)	6.03(0.119)	2.51(0.086)
Internal medicine course vs. duty-free course	54.2(13.7)*	0.224(0.140)	0.126(0.109)	0.080(0.079)
Duty night work vs. on-call daytime work	− 4.09(14.7)	0.253(0.153)	0.046 (0.120)	− 0.230(0.086)*
Post-call vs. on-call daytime work	− 20.0 (16.1)	0.054(0.165)	− 0.012(0.129)	− 0.092(0.093)
Pre-call vs. on-call daytime work	− 24.4(14.9)	− 0.285(0.156)	− 0.253(0.123)*	0.028(0.087)
ln (activity)	− 55.7(2.15)**	− 0.346(0.024)**	− 0.149(0.020)**	0.201(0.013)**

Note: RR, R–R interval from electrocardiogram; HF, high-frequency; LF, low-frequency components from fast Fourier transformation. *P < 0.05; **P < 0.001.

Table II.  Regression coefficients and SD resulting from the construction of a linear mixed-effect model of the HRV among all participants during the sleep period.

	Dependent variable
Independent variable	RR (ms)	HF [ln(ms^2)]	LF [ln(ms^2)]	LF/HF [ln(ratio)]
Intercept	548.4(39.9)	4.69(0.327)	9.23(0.383)	4.77(0.318)
Internal medicine course vs. duty-free course	22.4(18.4)	− 0.127(0.144)	− 0.234(0.102)*	− 0.332(0.117)*
on-call vs. pre-call	− 33.6(18.7)	0.249(0.151)	− 0.009(0.120)	− 0.133(0.130)
Post-call vs. pre-call	8.36(18.7)	0.555(0.149)*	0.192(0.113)	− 0.048(0.125)
ln (activity)	− 72.1(6.25)**	− 0.249(0.052)**	0.386(0.065)**	0.658(0.053)**

Note: RR, R–R interval from electrocardiogram; HF, high-frequency; LF, low-frequency components from fast Fourier transformation. *P < 0.05; **P < 0.001.

As a result of repeated measurements taken from each participant, a covariance structure was specified for the error term in the mixed-effect model. A reasonable covariance structure for these serial correlated measurements was an autoregressive order one model. The procedure PROC MIXED in SAS 9.1 was used to fit this model. A P-value < 0.05 was considered significant.

All other data were analyzed between control, pre-call, on-call, and post-call days, using non-parametric comparisons, namely Wilcoxon's signed rank test, and Friedman's test followed by Dunn's multiple comparison test due to the small sample size. A value of P < 0.05 with a two-tailed interpretation was considered statistically significant.

Results

As can be seen in Table I, using a linear mixed-effect model and adjusting for the other variable, we found a significant positive correlation between activity and LF/HF, whereas HF and RR were negatively correlated with activity. Following the dynamic changes of HRV over consecutive days, heart rate was found to increase during internal medicine course. Analyzing the waking period, we found a significantly greater decrease in LF/HF when the interns were during night shift than during their day shift (Table I). In addition, we compared HRV during the night shift working with the interns' duty-free nights to exclude any confounding caused by circadian fluctuation. The linear mixed-effect model was applied again. We set the binary explanatory variable “internal medicine course” to represent duty night work or the awake period during the duty-free night. A significant lower LF/HF was still found to be present (Estimate ± SD, − 0.334 ± 0.085 P = 0.004).

Our results indicated that different processes in terms of cardiac autonomic modulation during sleep period seemed to be responding to the long-term and short-term effects of OCD, i.e. 10 on-call duties per month and each duty was composed of 33.5 consecutive work hours followed by 2 duty-free days for a time period of 3 months. LF/HF was reduced during the whole internal medicine course compared with that during the duty-free course. On the other hand, an increase in sleep vagal activity (HF) indicated sensitivity to the short-term impact of OCD (Table II). Compared with all other days, the participants slept significantly less during the on-call night (Table III). Participants during their OCD course had significantly higher scores for ESS than during their duty-free course (P = 0.009; Figure 2A). At the same time, during the night of the on-call day, during the morning of the post-call day, and at noon on the post-call day, the participants had significantly higher scores for SSS compared with control (P < 0.05, 0.05, and 0.01, respectively). In addition, the participants had significantly higher scores for SSS at noon of the post-call day compared with the pre-call day (P < 0.05; Figure 2B).

Table III.  Comparison of the sleep time and CPT results across the different time points.

	NOC	PreO	OCD	PostO
Sleep time (h) (n = 13)	6.12 ± 0.917	6.67 ± 1.18	5.04 ± 0.776*	6.31 ± 1.69
CPT (n = 8)
Omissions (times)	1.00 ± 1.25	7.00 ± 17.8	1.38 ± 1.51	0.625 ± 1.06
Commissions (times)	12.9 ± 7.50	26.1 ± 6.51**	21.3 ± 11.0	22.8 ± 5.75
Hit RT (ms)	313.0 ± 29.8	275.3 ± 19.9**	308.4 ± 74.6	295.2 ± 30.6
Preservation (times)	0.125 ± 0.232	0.625 ± 0.744	1.50 ± 2.73	2.75 ± 5.01
Hit RT block change	0.008 ± 0.014	0.004 ± 0.019	0.005 ± 0.016	0.008 ± 0.019
Hit RT ISI change	0.041 ± 0.022	0.056 ± 0.019	0.043 ± 0.024	0.046 ± 0.014

Note: NOC, non-on-call, PreO, pre-call, OCD, on-call duty, PostO, post-call; Hit RT, hit reaction time; Block, change across duration of test; ISI, inter-stimulus interval. Results are expressed as mean ± SD. *P < 0.05 compared with NOC; **P < 0.01 compared with NOC by Dunn's multiple comparison test.

Figure 2.  ESS and SSS scores for internal medicine and duty-free course. (A) ESS scores for the NOC and OCD phases. (B) SSS scores for control noon (NOC-D), pre-call noon (PreO-D), on-call noon (OCD-D), on-call night (OCD-N), post-call morning (PostO-M), and post-call noon (PostO-D). The results are expressed as Mean ± SD. ††P < 0.01 compared with NOC by Wilcoxon signed rank test, n = 9. *P < 0.05 compared with NOC-D; **P < 0.01 compared with NOC-D; #P < 0.05 compared with PreO-D by Dunn's multiple comparison test, n = 9. The SSS data for NOC duty and pre-call noon were the average of day 1 and day 4 during the internal medicine course and the average of day 1 and day 2 during duty-free course, respectively.

Participants during their on-call and post-call days had significantly higher scores than the control period for the HADS and its subscales, with the exception of the HADS-depression scores during the post-call day (Figure 3, Total, P < 0.001 and 0.05, respectively; Anxiety, P < 0.001 and 0.05, respectively; Depression, P < 0.001). In addition, participants had significantly higher scores during their on-call day compared with their pre-call day, for both total HADS and both subscales (Total, P < 0.05; Anxiety, P < 0.05; Depression, P < 0.05; Figure 3).

Figure 3.  HADS scores and subscale scores for internal medicine and duty-free course. (A) HADS scores, (B) HADS-anxiety [HADS (A)] subscale scores, (C) HADS-depression [HADS (D)] subscale scores for NOC, pre-call (PreO), on-call (OCD), and post-call (PostO) duty. Results are expressed as Mean ± SD. *P < 0.05 compared with NOC; ***P < 0.001 compared with NOC; #P < 0.05 compared with PreO; ##P < 0.01 compared with PreO by Dunn's multiple comparison test, n = 11.

For the CPT, significant differences were only observed in commissions and Hit RT (Table III). The participants during the pre-call day had a higher rate of commissions and a lower rate of Hit RT compared with their non-on-call (NOC) day (Figure 4, Table III).

Figure 4.  Commissions and Hit RT of CPT for internal medicine and duty-free course. (A) Commissions, which were made when a response is given to a nontarget, as measured by the CPT, (B) Hit RT, the average speed of the correct responses for the entire test, as measured by the CPT for NOC, pre-call (PreO), on-call (OCD), and post-call (PostO) duty. The results are expressed as Mean ± SD. **P < 0.01 compared with NOC by Dunn's multiple comparison test; n = 8.

Discussion

In this study, LF/HF, which represented cardiac sympathetic modulation, was found to be lower during OCD than during day-time working and off-call nights for a time period of 3 months. The HRV indices across the three different shifts also showed that nighttime work was associated with reduced cardiac sympathetic modulation compared with the morning and evening working periods. Furthermore, there was a significant decrease in the sympathetic activity during OCD compared with off-duty night. Our study suggests that overnight work and the circadian rhythm synergistically reduce sympathetic activity. It is conceivable that a reduction in sympathetic tone during night work might be accompanied by decreased alertness and a deterioration in working performance; a similar effect has been found to prompt an excess of traffic accidents (Michail et al. 2008) and errors in the performing of a task (Bjerner et al. 1955; Akerstedt 1988; Mitler et al. 1988; Smith et al. 1994).

During the sleep period, increased HF, which represents vagal modulation, was found to occur during the participants' OCD time and this contrasted with a decreased sympathetic index that persisted throughout the internal medicine course. The vagal activity has been related to sleepiness or sleep propensity (Delamont et al. 1998), and it has been demonstrated in previous study that parasympathetic activity predominates before falling asleep (Kuo et al. 2008). The significantly higher ESS score during the internal medicine course and the increased SSS score at midnight during OCD also indicated that the interns were suffering from sleepiness.

It is interesting that our study differs from the results obtained when normal humans undergoing acute sleep deprivation, which has been reported to result in increased sympathetic and decreased parasympathetic cardiovascular modulation (Zhong et al. 2005). The sleep time is on average 5.04 h during on-call and medical interns at Chang Gung Memorial Hospital work 86.7 h weekly, which involves 33.5 consecutive work hours and 10 on-call duties per month during the OCD period. Thus, interns may have adapted to partial sleep deprivation by the time they reach their last month of the internal medicine course. The SSS score increased at midnight for OCD rather than during the morning post-call, which suggests that the sleepiness may be attributable to a heavy workload rather than to sleep deprivation. However, the significant and high ESS scores, which on average are >10, rather suggest that the sleepiness is somewhat similar to sleep disorder involving excessive daytime sleepiness, such as narcolepsy and obstructive sleep apnea (Dauvilliers 2006).

General adaptation syndrome is a model of stress, and uses three stages: alarm, resistance, and exhaustion (Selye 1975). The internal medicine training course appears to have produced a chronic sleep debt and the interns appear to have entered a stage of exhaustion that is not recoverable from during the intervals between on-call duties and thus produces rebounding slow wave sleep. This contrasts with autonomic modulation after acute sleep deprivation, which manifests a response similar to the sympathetic response during the alarm and resistance stages. The latter is similar to surgeons and their first assistants, who have been shown to have sympathetic hyperactivity during operations. Work and life style differences between physicians and surgeons may explain this phenomenon because surgeons have to keep a high level of attention throughout operations, while physicians attend to their patients' complaints in a fragmentary fashion throughout the day (Demirtas et al. 2004).

Patients with depressive and anxiety disorders may be at increased risk of cardiovascular disease as a result of decreased HRV (Gorman and Sloan 2000). The general HRV, especially HF, of interns is not decreased. The interns' anxiety and depression, as measured during their on-call day, showed no relationship to the HRV index. At the core of any psychiatric disorder is an abnormality in neurotransmitter signaling (Wulff et al. 2010). We suggest that this anxiety and depression might be an emotional response to both systemic and psychogenic stresses rather than being associated with depressive and anxiety disorders that meet the Diagnostic and Statistical Manual of Mental Disorders criteria.

Comparing the on-call course with the duty-free course, the interns had a higher commission rate and a lower Hit RT for the CPT, which means greater inattention and increased impulsivity. Interestingly, the significant difference appeared at noon of the pre-call day. Compared with the SSS, although subjectively they did not feel sleepy, objectively their performance was still poorer. There was no apparent difference across the different days during the on-call course, which implies that there may have been a chronic change and that the difference was not caused by a single OCD event. Chronic stress can recruit pathways that are distinct from those involved in acute responses (Ulrich-Lai and Herman 2009). A lower nocturnal sympathetic activity may occur due to repeated chronic stressor exposure. In contrast, temporal anxiety and depression may be diminished by repeated stress exposure. The present findings are in agreement with residents' motor performance that was measured in a previous study, which showed no difference in reaction time post-call versus all other periods for internal medicine. However, internal medicine residents were found to have significantly more major and minor reaction time lapses than their controls (Saxena and George 2005).

Overall, this study differs in design from previous studies in three ways. First and most importantly, to our knowledge this is the only comprehensive study using both subjective and objective evaluations of on-call interns. Second, a duty-free course was used as a self-control group for the participants who were undergoing the internal medicine training course, which had an OCD every 3 days. Finally, the third and last month of both courses were chosen for this study, which ought to minimize any disturbances due to the interns' unfamiliarity with their activity. Such disturbances might play an important role in any autonomic modulation. On the other hand, the 3-month non-duty course provided enough rest to allow recovery from any stress and sleep restriction that had been caused by the internal medicine course. Both the above factors helped facilitate the evaluation of the long-term effects of OCD.

There are several limitations in this study. First, the participants are all male and the sample size is relatively small. Most medical interns are male and the available female interns were relatively unwilling to participate in the study. Furthermore, the HRV is affected by the menstruation cycle and by gender differences (Kuo et al. 1999, 2010; Liu et al. 2003). Considering the increased variability and different patterns of HRV that would be present if female interns were included together with the very small sample size, we specifically did not recruit female participants into this study. Second, we measured sleep time using an objective estimate (actigraphy) as well as by a subjective method (sleep-log diary). These were the two available practical means of data collection during the interns' clinical practice, rather than the golden standard of polysomnography. However, the discrepancy between the sleep-log diary and actigraphy is somewhat troublesome. Medical residents have been found to consistently underestimate or overestimate their actual sleep time compared with controls during a previous study (Saxena and George 2005). On the other hand, our analysis demonstrates that when performing HRV during waking, it is important to specify the activity state. Nonetheless, quantifiable activity was found to correlate with LF/HF positively, but with HF and RR negatively (Table I).

Finally, in order not to bother interns' clinical practice during their stressful OCD, we did not investigate CPT during the duty night so the decreased LF/HF during night time shift could not be shown to correspond to their performance. In addition, this study demonstrated that inattention and impulsiveness in CPT may occur due to repeated chronic stressor exposure. HRV is a useful tool for revealing adverse effects on the cardiovascular system in an occupational setting (Togo and Takahashi 2009). Although shift work is associated with an increased rate of cardiovascular diseases (Knutsson and Boggild 2000; Fujino et al. 2006), a long-term follow-up investigation should be conducted to examine whether an increased rate of cardiovascular diseases will develop, and whether anxiety and depressive disorders will occur.

Conclusions

Our study suggests that current guidelines, which allow 10 on-call duties monthly with an extremely heavy workload in hours, 33.5 consecutive work hours followed by 2 duty-free days for a time period of 3 months, result in a number of different processes that affect cardiac autonomic modulation during the interns' sleep period. Nocturnal vagal activity is elevated after their duty and decreased sympathetic activity is present throughout the internal medicine course. Reduced sympathetic modulation during duty night work, fluctuations in anxiety and depression, and long-term inattention and impulsiveness were found to be the major responses to the stress of OCD.

Acknowledgements

This study was supported by a grant (YM-97A-C-P506) from the Ministry of Education, Aim for the Top University Plan and a grant (NSC-95-2314-B-010-087) from the National Science Council (Taiwan). The authors gratefully acknowledge the cooperation and friendship of participants and classmates of two of the authors' (Yu-Hsuan Lin and Yen-Cheng Ho). We also thank Dr Yeh Yao-Hsien for his excellent technical support and Ms Ying-Hua Huang for her administrative work on manuscript production.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.