The Effect of Infective Exacerbations on Sleep and Neurobehavioral Function in Cystic Fibrosis

Information is limited about the effect of an infective exacerbation on sleep and neurobehavioral function in adults with cystic fibrosis (CF). Sleep fragmentation and hypoxemia, which can impair cognitive function in patients with sleep apnea/hypopnea syndrome (1-3), may worsen in patients with CF during pulmonary exacerbations. Hypoxemia impairs cognitive function in humans at altitude and in patients with severe chronic obstructive pulmonary disease (4). Even a single night of sleep fragmentation makes young, healthy subjects sleepier during the day, impairs their subjective assessment of mood, and decreases mental flexibility and sustained attention (5).

A previous report of neurobehavioral function in adults with CF focused on patients with clinically stable severe lung disease and demonstrated objective neurocognitive impairment and increased daytime sleepiness in these patients when compared with healthy control subjects (6). We hypothesized that patients with CF would be more vulnerable to disrupted sleep architecture, worsening sleep hypoxemia, and impaired cognitive performance during a pulmonary exacerbation than during a period of clinical stability, and would improve with treatment. Because all patients could potentially be susceptible to exacerbation-related sleep and cognitive effects, we studied adults with a wide range of pulmonary function. Some of the results of this study have been previously reported in the form of abstracts (7, 8).

METHODS

Subjects and Setting

Subjects older than 16 years were recruited from the Adult Cystic Fibrosis Clinic at Royal Prince Alfred Hospital. Informed consent was obtained, and the study was approved by the Central Sydney Ethics Review Committee.

Cases were studied at the beginning and end of a pulmonary exacerbation. Other patients with CF, volunteering to act as "control subjects," underwent the same testing procedures when in a clinically stable state.

Protocol

Cases answered a validated CF quality-of-life questionnaire (9), underwent polysomnography while breathing room air with evening and early-morning arterial blood gases, also on room air, and had neurobehavioral testing, before and after inpatient treatment of an exacerbation. Further details can be found in the online supplement. Control subjects underwent the same testing procedures on two occasions separated by a period of 10 to 14 days. Neurobehavioral testing was performed either in the afternoon before or on the morning after polysomnography. Twelve cases had neurobehavioral testing performed after spending their first night in hospital. Subjects were tested at the same time of day on each occasion to control for circadian variation in vigilance and alertness. Similar numbers in each group performed the test bout in the afternoon and morning. All subjects were instructed to abstain from alcohol and caffeine-containing drinks for 12 hours before testing.

Sleep studies and arterial blood gases were performed and scored as previously described (10, 11). Total sleep time (TST) minimum average oxyhemoglobin saturation (SpO^sub 2^) was calculated by averaging the lowest saturation measurement in each 30-second period of sleep.

Intelligence was assessed using the Shipley Institute of Living Scale (12). Spirometry. resting daytime SpO^sub 2^, and weight were measured before each test bout. Spirometry was performed according to American Thoracic Society guidelines (13).

Neurobehavioral Function

Neurobehavioral function was evaluated using the Neurobehavioral Assessment Battery (University of Pennsylvania) (14) and the AusEd driving simulator (Woolcock Institute for Medical Research, Sydney, Australia) (15). Further details can be found in the online supplement.

Statistical Analysis

Physiologic data are presented as mean ± SD. The Shapiro-Wilk test was used to test for normality. Levene's test for homogeneity of variance was performed. The paired t test or the Wilcoxon signed rank test was used to compare scores obtained at the first (pre) and second (post) study visits within each group. The Student's t test or Mann-Whitney U test for continuous variables and ?^sup 2^ or Fisher's exact test for categoric variables were used to analyze differences in baseline characteristics between the two groups. Pearson or Spearman correlations were performed. The sample size was calculated a priori to detect a change in arousal frequency of 5/hour during sleep and/or a 200-ms reduction in reaction speed in the driving simulation task for patients with an exacerbation, with a power of 0.80 and a two-sided a of 0.05.

Analysis of covariance was performed to assess the effect of an exacerbation on neurobehavioral and driving simulator performance using PROC GLM in SAS, version 8.2 (SAS Institute, Cary, NC). Age and sex were included as covariates in all neurobehavioral analyses.

RESULTS

Baseline Parameters

When the patients with CF with exacerbations (cases) were in a clinically stable state, the two groups did not differ significantly in terms of age, lung function, education or intelligence levels, body mass indices, and resting daytime Sp^sub O^sub 2^^ (Table 1).

The cases improved their FEV^sub 1^ percent-predicted (45 ± 20 pretreatment vs. 56 ± 24 posttreatment, p < 0.001), FVC percent-predicted (64 ± 21 pretreatment vs. 80 ± 26 posttreatment, p < 0.001), and body mass index (19.7 ± 2.2 pretreatment vs. 20.1 ± 2.1 posttreatmenl, p = 0.01) with treatment. There were no significant changes in the control group between study visits. When admitted, cases were more tachypnoeic than control subjects, both while awake and during stage 2 sleep (cases, 26 ± 7, vs. control subjects, 20 ± 5 bpm; p = 0.004). They also spent a greater proportion of sleep time tachycardie (cases, 10 ± 21% of sleep time with heart rate > 100, vs. control subjects, 1 ± 1% of sleep time with heart rate > 100 bpm; p = 0.03). After treatment, the cases' respiratory rates and proportion of sleep time spent tachycardic did not differ from those of control subjects.

The CF Questionnaire

When admitted with an exacerbation, cases had more respiratory symptoms, a worse perception of their health state, and reduced physical functioning and vitality than control subjects. After inpatient treatment, they reported a reduced treatment burden compared with control subjects. The cases demonstrated significant improvements with treatment in 8 of 12 dimensions. No dimension of the CF questionnaire changed significantly among control subjects between study visits (see Table E1 in the online supplement).

Sleep Architecture

The cases spent more time awake after sleep onset than control subjects (p = 0.02) when unwell. They also had reduced percentage of REM sleep (p = 0.03) and spent fewer minutes in REM sleep (p = 0.04). Cases and control subjects had similar numbers of REM sleep bouts, but the cases' REM bouts were shorter. After treatment, cases had a reduced arousal index (p = 0.001) compared with control subjects and a greater amount of slow-wave sleep (p = 0.02; Table 2).

After treatment of their exacerbation, cases experienced improvements in sleep efficiency, wake after sleep onset, arousal index, time spent in REM sleep, and time spent in slow-wave sleep (Table 2). Control subjects experienced an improvement in sleep efficiency and time in slow-wave sleep on the second study night.

The respiratory disturbance index among both cases and control subjects, on both study nights, comprised mainly hypopneas, with few obstructive apneas. Two control patients who were overweight were found to have sleep apnea/hypopnea syndrome. One control patient had arousals secondary to periodic limb movements. These unexpected findings contributed to the relatively high mean respiratory disturbance indices and arousal indices seen in the control subjects (Table 2).

Gas Exchange

Compared with control subjects, when admitted with an exacerbation, cases had lower minimum nocturnal Sp^sub O^sub 2^^ lower TST minimum average Sp^sub O^sub 2^^, and spent a greater proportion of TST and REM sleep time with an Sp^sub O^sub 2^^ less than 90%. At the completion of treatment, there were no significant differences in these gas exchange variables between cases and control subjects (Table 3).

The treatment of a pulmonary exacerbation resulted in improvements in TST minimum average Sp^sub O^sub 2^^ and percentage of TST spent with an Sp^sub O^sub 2^^ less than 90%. The evening Pa^sub O^sub 2^^ also improved significantly (Table 3). There was no change in morning Pa^sub CO^sub 2^^ or pH. In control subjects, the percentage of TST spent with an Sp^sub O^sub 2^^ less than 90% improved on the second study night and evening Pa^sub O^sub 2^^ worsened.

Neurobehavioral Assessment Battery

Subjective measures. Among cases, treatment of an exacerbation was associated with improvement of all six subscale components of the profile of mood states (POMS) and in total mood disturbance. There were no changes in the scores of control patients (Table E2).

The treatment of an exacerbation resulted in significant improvements in subjective sleepiness, Visual Analog Scale scores and the "general activation" and "deactivation sleep" components of the activation-deactivation checklist (Table 4).

Objective measures. Cases had slower throughput (number of correct items/minute) than control subjects in the serial addition and subtraction task (SAST) when admitted with an exacerbation (cases, 16 ± 4, vs. control subjects, 17 ± 3; F[1, 36] = 5.15; p = 0.03) and a slower response time on the digit symbol substitution task (F[1, 36] = 11.91, p = 0.001), which persisted after treatment (F[1, 36] = 8.48, p = 0.006). With treatment, cases significantly increased the number of correct responses in the SAST (pretreatment, 41 ± 8, vs. posttreatment, 43 ± 8; p = 0.03) and decreased their response time on the psychomotor vigilance task (pretreatment, 320 ± 76, vs. posttreatment, 300 ± 71 milliseconds; p = 0.03). In the SAST, both groups showed similar improvements in throughput on repeat testing (Table E3).

With two of the neurocognitive tasks, sex significantly modified the effect of an exacerbation. Men with exacerbations had a reduced number of correct items per minute in the SAST compared with women with exacerbations (pretreatment: men, 12 ± 2, vs. women, 17 ± 4; p = 0.003; posttreatment: men, 14 ± 3, vs. women, 18 ± 4). Women had more lapses (no response for > 500 milliseconds) in the psychomotor vigilance task with an exacerbation compared with men (pretreatment: women, 6.5 ± 7.0, vs. men, 0.5 ± 0.8; p = 0.03; posttreatment: women, 3.2 ± 5.8, vs. men, 0.3 ± 0.5).

Driving Simulator Performance

Driving simulator data were obtained from 17 cases and 19 control subjects. Analyses were adjusted for age and sex, and there was no effect modification. Among cases, treatment of an exacerbation improved mean brake response time to the appearance of trucks on the road (pretreatment, 1,237 ± 206, vs. posttreatment, 1,119 ± 201 milliseconds; p = 0.01). The mean braking response time did not differ significantly between cases and control subjects either when the cases were unwell or after treatment.

There were no differences between cases and control subjects in steering deviation, speed deviation, or number of crashes.

Correlations

Among cases, associations between FEV^sub 1^ percent-predicted and the three main exacerbation-related effects on sleep (amount of REM sleep, sleep efficiency, hypoxemia) and cognitive function (SAST throughput, digit symbol substitution task response time, psychomotor vigilance task response time) were examined. There was no correlation between FEV^sub 1^ percent-predicted and sleep efficiency or percentage of REM sleep among cases either on admission or at discharge. However, when cases presented with an exacerbation, there was a strong correlation between percent-predicted FEV^sub 1^ and TST minimum average Sp^sub O^sub 2^^ (r^sup 2^ = 0.58, p < 0.001), which persisted after treatment. After adjusting for age and intelligence quotient, there was no association between lung function and performance of cognitive tasks.

DISCUSSION

We have demonstrated that neurobehavioral function and sleep architecture and gas exchange are adversely affected by infective exacerbations and improve with treatment. Disruptions to sleep architecture, including reduced REM sleep, occurred with exacerbations and improved with treatment but were not related to lung function impairment. The extent to which nocturnal hypoxemia worsened with exacerbations was strongly correlated with lung function and, although hypoxemia significantly improved with treatment, it did not reach levels of control subjects by the completion of inpatient treatment. Exacerbations affected performance in tasks assessing reaction speed, concentration, and concrete reasoning. The effect of exacerbations on some tests of cognitive performance differed depending on sex. Our results raise the possibility of a link between the effects of exacerbations on sleep and neurobehavioral function.

A pulmonary exacerbation adversely affected the speed and accuracy with which subjects performed the SAST and the speed with which they performed the digit symbol substitution task. After treatment of their exacerbations, the cases increased their number of correct responses in the SAST and had better response times and fewer lapses in the psychomotor vigilance task. Similarly with the driving task, the braking response times of cases and control subjects were not significantly different, yet the cases were able to significantly improve their response times after treatment, whereas reaction speed of the control subjects remained stable. This suggests that patients with exacerbations, although not noticeably impaired compared with their peers, may not be fulfilling their own cognitive potential.

Sex differences in strategy and performance of computerized neurobehavioral tests are well recognized (16), but there is little information about the differential effects of illness on performance. Two distinct types of solving strategies for neuropsychologic tests have been identified: one, favored by men, aimed for high speed, using a right-hemisphere-type global strategy, and one, favored by women, aimed for high accuracy, using a left-hemisphere-type strategy (17). Previous work has shown that men usually perform faster than women on reaction time tests (16). The fact that our female patients with exacerbations had particularly slow reaction speeds and the male patients with exacerbations displayed reduced accuracy in simple arithmetic would be consistent with the hypothesis that illness increases the vulnerability of patients of each sex to deterioration in tasks in which they are naturally prone to weakness.

Pulmonary exacerbations were associated with less efficient sleep. The time awake after sleep onset and sleep efficiency of the cases improved with treatment and approached values seen in stable control subjects. Treatment improved the cases' arousal index, suggesting that their sleep was more fragmented than usual when they were admitted with an exacerbation. The patients with the worst lung function experienced more severe nocturnal hypoxemia with exacerbations. In some patients, their hypoxemia did not correct to levels approximating those seen in control patients within the 10- to 14-day treatment period. Similar delayed recoveries in other parameters after treatment of an exacerbation have been previously reported (18).

The significant reduction in REM sleep that occurred with exacerbations was not related to lung function impairment. Treatment of the exacerbation increased amounts of both REM and slow-wave sleep, with quantities of REM sleep returning to levels seen in control subjects. Increased coughing may have contributed to the reduction in REM sleep, because protracted coughing in patients with CF has previously been shown to halt progression to REM sleep (19). Somnogenic acute-phase cytokines, interleukin 1ß, interleukin 6, and tumor necrosis factor a, are elevated during exacerbations in patients with CF (18, 20), and increase non-REM sleep intensity at the expense of REM sleep (21). Thus, the REM sleep reduction would also be consistent with an acute response to an infectious stimulus. The increase in slow-wave sleep may have represented recovery from sleep deprivation associated with hospitalization for exacerbations (22). Arousal threshold is increased in slow-wave sleep (23). This may have contributed to the reduced arousal index seen in the cases on their second study night.

Although small numbers precluded direct correlation of neurobehavioral function scores with objective measures of sleep fragmentation or hypoxia, our findings are consistent with a possible link between the sleep and cognitive effects of exacerbations. The subjective components of the Neurobehavioral Assessment Battery relating to feelings of sleepiness, exhaustion, and activation improved markedly after treatment. The SAST is a test of concentration and concrete reasoning, whereas the digit symbol substitution task tests speed, accuracy and concentration. They are sensitive to sleep deprivation, fragmentation, and hypoxic injury (24). The psychomotor vigilance task and driving task are vigilance tasks, designed to measure sustained, focused attention. They assess reaction time, which is sensitive to partial or complete sleep deprivation (25). The absolute amount of REM sleep has also been correlated with intellectual functioning (26) and declines markedly in the case of organic brain dysfunctions of the elderly (27). In addition, both sleep fragmentation and hypoxemia are well-recognized contributors to impaired cognitive function and increased daytime sleepiness in patients with sleep apnea/hypopnea syndrome (1). The possibility of a similar link in patients with CF warrants further study.

Strengths of our study included the presence of a control group of patients with stable CF and our attempt to control for noncognitive influences on test results. Results from both spirometry and the CF questionnaire suggest that patients with CF with exacerbations were successfully differentiated from those without, which is not always straightforward, given the difficulty of accurately defining an exacerbation (28-31). The control group allowed first-night effects on sleep and improvements in cognitive tests because of familiarity or practice to be differentiated from true effects related to the exacerbation and its treatment. Medications, nutritional supplements, and intensity of physiotherapy also have the potential to influence sleep and would have varied slightly between the two groups. Within groups, these factors were similar on both study nights and are therefore unlikely to have influenced differences demonstrated between Study Nights 1 and 2. Multiple between-group comparisons without adjustment for multiple analyses can be justified in an exploratory study such as this one (32). However, the possibility of type I error must be considered, and confirmatory studies doing multiple comparisons would require the use of a more conservative p value (32).

Some neurocognitive and driving simulator variables were not normally distributed. In these cases, Levene's test for homogeneity of variance was performed and was nonsignificant. Given equal variances and equal sample sizes, analysis of covariance is fairly robust, even without fulfillment of the assumption of normality. Limitations to neurobehavioral testing using a battery of cognitive tests include the following: the adequacy of the theory that the tests used represent valid neurocognitive constructs, the potential for test-retest and ceiling-floor effects, and subject motivation (2).

The crucial question arising from this study concerns the clinical relevance of these quantitatively small performance decrements to patients' day-to-day lives. In subjects with sleep apnea/hypopnea syndrome, deficits in cognitive tests of executive function consistently translate into complaints of inattention, impulsivity, and emotional lability in children and declining occupational, interpersonal, or cognitive functioning in adults (2). In children with CF, long-term nocturnal oxygen therapy improves school and work attendance (33), highlighting the possibility that cognitive impairment may be amenable to treatment. We did not examine the impact of exacerbations on activities relevant to young adults with CF, including school and examination performance, work productivity and safety, and on-road driving ability. This is an important direction for future study.

We have demonstrated that exacerbations of lung disease in young adults with CF adversely affect sleep and neurobehavioral performance, regardless of underlying disease severity. These results could have important implications for performance in daily life, because patients often delay admission to hospital to fulfill study or work commitments.

Conflict of Interest Statement: C.J.D. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; D.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; K.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; R.R.C.'s department has received funding from ResMed for clinical trials in treatment of respiratory failure and he has received a travel grant of $1,500 from Respironics; P.T.P.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.

Acknowledgment: The authors thank Dr. Kai Sing Lo for statistical advice, Professor David Dinges for providing the Neurobehavioral Assessment Battery software, and David Joffe, Ben Constable, and Heather Engleman for provision of the AusEd software.