Fatiguing illnesses, in particular Chronic Fatigue Syndrome (CFS), are difficult conditions to accurately diagnose and quantify. The Oxford [1] and Centre for Disease Control (CDC) [2] criteria for CFS define a person suffering from CFS as one who has experienced persistent, debilitating mental and physical fatigue for six months or more, where rest is not restorative and the fatigue state is not due to ongoing exertion. There may also be several co-existing symptoms present at any time, including those of a cognitive or neuropsychiatric nature, and once these are established, cognitive, behavioral, emotional, physiological and social factors are thought to work together to perpetuate it [3]. The clinical features of the illness include reports by patients of impaired cognition and research groups have investigated whether an association between mental fatigue and cognitive functioning occurs [4, 5]. Evidence to suggest the existence of impairments in focused attention, speed of processing and performance accuracy in CFS are supported by data which indicate that as sleep deprived, healthy individuals become more fatigued, cognitive deficits become more pronounced [6].

The deficits associated with physical fatigue experienced by CFS sufferers have also been well documented over the past ten years. Decrements in performance on both simple and choice reaction time measures, for example, were indicted by LaManca et al. in 1998 [7] and studies have since revealed impairments in verbal fluency, memory, motor speed, sustained attention and speed of cognitive processing [8-12]. Others have highlighted deficits in verbal and non-verbal memory tasks occurring in this illness [13, 14].

Others have, however, questioned whether cognitive impairment does actually occur [15, 16], especially in the light of apparent discrepancies between objective measures of impairment and the subjective measures of cognitive failures reported by sufferers [17]. There is evidence to suggest that patient-rated perceived levels of impairment appear much higher than their performance on objective measures. Wearden and Appleby [18], although acknowledging this, indicated that the lack of conclusive data may more likely be explained by the fact that the tasks chosen to assess cognition were not of sufficient sensitivity to measure the impairments in CFS. They further proposed that the performance impairments reported were only evident in tests of a more complicated nature, such as tasks requiring the completion of two or more elements simultaneously (the dual-task paradigm). A recommendation made in light of this paper suggested that future research should employ measures that reflect the nature of the specific cognitive complaints patients themselves report.

Methodological considerations aside, factors such as sample demographics, intelligence, anxiety and depression are known to affect cognitive performance. The link between reaction time, age and intelligence, for example, has long since been established [19]. Similarly, mood disorders such as depression and anxiety are associated with memory and additional deficits [20-22]. In light of this, Michiels and Cluydts [23] expressed the concern that it may be the demographic nature of the patient sample and the presence of comorbid psychopathology that were responsible for any cognitive deficits reported and not the illness itself. To illustrate this several instances were highlighted where comparison groups had not been sufficiently matched to the CFS sample and screening for co-morbid anxiety and/or depression were not rigorously applied.

To further complicate this particular area of research, the heterogeneous nature of CFS is well documented [24] dictating that any study investigating cognitive impairment in this illness would require large numbers of patients. Indeed, our previous studies had been conducted on relatively small patient numbers [25] and, although findings were encouraging, it was acknowledged that larger patient numbers were required for a more accurate profile of impairment to emerge. In addition, the analyses conducted would need to be extended to include, not only the potentially confounding factors that co-exist in the illness (such as age, gender and comorbid depression), but also mechanisms (such as the impact of stress, daily hassles and life events) which appear to exacerbate and perpetuate the condition.

In view of the on-going debate as to the existence and exact nature of the cognitive deficits associated with CFS the current study addressed each of the conflicting issues raised by: (a) recruiting a large sample of Chronic Fatigue Syndrome sufferers from a specialized out-patient clinic where strict adherence to the CDC criteria was maintained [2]; (b) comparing the patient group to an age-, gender- and educationally-matched healthy control group; (c) administering a wide range of objective and subjective measures developed from previous studies [8, 25]; and (d) identifying possible confounding factors which could account for any deficits observed and incorporate them into the analysis model.

Ethical approval was granted by the local research ethics committee and informed consent was obtained from all who participated. Data were coded to protect the anonymity of both patient and control groups.

A battery of computerized performance tests were used to assess mood, memory, attention and psychomotor function [8]. Responses were registered using a response box connected to a timer card allowing measurement of reaction times to the nearest millisecond.

Two separate questionnaire booklets were administered in total; one was completed before visiting the clinic or research centre (including demographic data, illness history and symptom severity, health-related behaviours, psychopathology, psychosocial factors and measures of cognition), and one at the time of testing (administered before the completion of the cognitive function tasks and including measures assessing the level of state anxiety, depression, negative and positive affect and fatigue-related symptoms within the previous week).

As factors such as demographics, symptoms, anxiety, depression, mood and psychosocial factors have been shown to be associated with impaired cognition in previous studies, a series of univariate analyses were conducted on each performance task using the individual variables listed in Tables 1-3 as covariates. From these analyses only two variables produced an independent effect which removed the overall group difference in cognition between the CFS and controls; total symptoms and fatigue scores. Table 5 describes the factors exerting an independent effect for each performance task whilst not affecting the differences between the CFS and control groups.

When all the independent factors were co-varied in one analysis, several measures had a significant independent effect (shown in bold) but again the difference between CFS patients and controls only became non-significant when fatigue and symptom severity were co-varied.

The Cognitive Failures Questionnaire (CFQ) was used in this study as a subjective measure of cognitive impairment. To investigate the suggestion that patients with CFS over-estimate the levels of cognitive deficits experienced, a median split was performed on the CFQ scores for each of the two groups and, as a result, four sub-groups were created: (a) CFS patients with low CFQ scores (n=107), (b) controls with low CFQ scores (n=106), (c) CFS patients with high CFQ scores (n=186) and, (d) controls with high CFQ scores (n=16). These were then used as grouping factors in analyses of variance for each of the performance measures with the independent factors highlighted in Table 5 as covariates. These results are shown in Table 6.

There is an overall group effect of CFQ on cognition and, therefore, an association between the subjective CFQ measure and level of cognitive deficit in CFS.

To further investigate the relationship between subjective and objective measures, the cognitive failures questionnaire scores for the patient and control groups were split into their respective quartiles and used as grouping factors.

These analyses reveal clear differences between the CFQ quartiles for the CFS group for both the mean reaction time task (F(3, 285)=3.32, p<0.02) and the Stroop task (F(3, 287)=4.78, p<0.003), but not for the free recall or vigilance tasks. For the controls, however, there were no associations between CFQ scores and deficits for any of the performance measures.

The current study set out to bring some clarity to the issues arising from the on-going debate surrounding the exact nature, and indeed existence, of the cognitive deficits associated Chronic Fatigue Syndrome (CFS). This was achieved by addressing specific comments made by previous reviewers of the literature [17, 18] as well incorporating into the design the views of those who found no evidence of such impairment [7, 15, 16].

To begin with we recruited 307 participants, a much larger number of patients than previous studies, from a single specialised outpatient clinic. To address the issue of accuracy of diagnosis, the Centre for Disease Control (CDC) case definition criteria [2] was used to diagnose all patients invited to take part in the study, which excludes patients exhibiting major psychiatric disorders. It was clear that our sample conformed to the expected demographic profile for Chronic Fatigue Syndrome (CFS): that is, a predominance of middle-aged married females.

When examining the patient sample in more detail we found that the majority of the group (68%) had received a preliminary diagnosis of CFS from their GP before attending the clinic and their mean illness length was five years. Although the majority of patients suggested a physical cause as a possible trigger for their illness (such as influenza), several patients acknowledged a non-physical cause (for example, a stress-related episode).

Half the sample of patients experienced fluctuations in symptom severity although no regular daily pattern emerged. The resultant effect of such variability is that patients experience high levels of uncertainty. Data from a symptom check list provided evidence of the myriad symptoms experienced by CFS sufferers. The study showed that and the majority of patients (41%) reported their condition as ‘bad with some recovery’ on the 5-item current state of health scale at the time of testing. Increased rest and sleep were highlighted by patients as the most effective approaches in their attempt to minimise the negative effects of the illness whilst, on the other hand, exercise, walking, shopping, mental effort and stress were most commonly linked with an exacerbation of symptom severity.

To address criticisms aimed at the inclusion of inappropriate control groups in previously reported studies, we recruited 126 healthy volunteers from a cross-section of the general population who were matched to the patient sample in terms of age, gender, educational standard and socioeconomic group. In addition, the controls completed a similar screening process to the patient group.

Concerns voiced by Wearden and Appleby [18] regarding the suitability and sensitivity of the objective measures of cognitive performance used in previous research were answered by administering tasks which, not only reflected the type of cognitive impairments patients themselves report, but the tests used had been utilised in previous studies of the illness [8, 25] and in studies of fatigue in healthy individuals [6].

In a series of group comparisons we found several differences between the CFS patients and control group. Firstly, and not surprisingly considering the severity of the illness, the controls were significantly more likely to be in employment than the CFS group at the time of testing. Secondly, despite the two groups being matched in terms of educational status, the CFS patients scored significantly lower on the pre-morbid intelligence measure than the controls. There are two possible explanations for this: (a) as intelligence shows a positive correlation with most cognitive measures [39], the converse is also true, namely that deficits in cognitive functioning will be accompanied by lower intelligence scores; (b) patients with depression have been shown to report lower pre-morbid intelligence scores using the NART score [40].

A further area of interest is the role of sleep abnormalities in this illness [8]. Our data suggests that the average number of hours slept per night did not differ between the patient and control groups. However, the two groups were shown to differ in ratings of the extent to which the individual felt rested as a result of sleep – namely sleep quality. Abnormalities in sleep patterns had been shown to affect mood and performance both in CFS patients and healthy controls [8, 27] and were an important factor to consider when examining these data further. In addition, problems with sleep quality are often observed in abnormal psychopathology [41, 42].

The presence of co-morbid mood disorder is also indicated in the patient group. Although these findings contradict previous findings (e.g. Marshall et al. 1996 [14]), it is believed that this study provides a more accurate representation of the illness due to the fact that a larger cohort was assessed in comparison with previous studies. Indeed laboratory testing further highlight the differences between the mood states of the two groups. The CFS patients record significantly lower levels alertness and hedonic tone and have higher anxiety levels than the control group. It is of note that the state anxiety scores of the two participant groups do not differ at the time of testing, indicating that the patient group was not unduly distressed by the test session itself. Trait anxiety scores, however, indicate that patient group exhibit higher general anxiety levels than the controls.

Simple comparisons of performance task data seem to confirm our previous findings of a slowing of motor speed and impairments in immediate recall in CFS. The reaction time task also provides data indicating that there were significant differences between the reaction times of the CFS and control groups when comparing reaction times at minute intervals. In addition, a greater rate of slowing over the three minutes was seen in the CFS group. This ‘time-on-task effect’ was not replicated in the other three minute task (the vigilance task) even though group differences in the reaction times for each minute on the task remain. There is a suggestion here that although both sustained attention and motor speed are affected in CFS, time-on-task effects differ. A possible explanation may be that performance on such tasks are affected differently by fatigue and that these differences may reflect differences in underlying neurotransmitter mechanisms involved in processing the two tasks (Christopher et al. 2005 for a discussion [43]). However, this theory would have to be tested in further wide-ranging studies.

Returning to the vigilance task, the detection rate of repeated digits for each minute was greater in the control group than the patient group and both groups’ detection rate decreased over time. In contrast to our previous findings, where a gradual decline in accuracy over time was reported for both groups [25], the current study indicates that there was no measurable decline in accuracy for the patient group between minutes 1 and 2. A trade-off between reaction time and accuracy (for minutes 1 and 2) is, however, observed in healthy controls and this warrants further investigation.

The final performance task, the Stroop interference task [37], provided evidence to suggest that distraction by irrelevant stimuli is more pronounced in the patient group. These data are contrary to findings reported by Metzger and Denny [44] who found no evidence to suggest that CFS patients perform differently to controls on this task. The inconsistency between their study and the results presented here might be due to the relatively small number of patients tested (40 compare to 307) and the fact that a modified version of the Stroop task was used to assess cognitive function. By modifying the task, some test sensitivity may have been lost and this has been suggested as a possible reason for the lack of conclusive data in the past.

To investigate these data further, a series of additional analyses were conducted, this time identifying a range of factors significantly associated with poorer performance and including these as covariates in the analyses. These included demographic data, pre-morbid intelligence, illness severity, anxiety, depression, psychosocial factors and sleep data. Only two variables, total symptoms and the fatigue sub-scale of the fatigue-related symptoms questionnaire, were identified as factors which act to remove the overall group effect on performance. Although one might expect the level of somatic symptoms, cognitive difficulties and emotional distress to be major factors of influence in this particular illness, these two factors essentially define the two groups. However, several factors were identified as having an independent effect on performance without removing the overall group effect. These factors were, therefore, used as covariates in further analyses to ensure that they could act in cumulative way and affect performance. There was no evidence to suggest that co-morbid anxiety and/or depression play a role in the level of cognitive deficit in this illness.

In order to consider the ability of the patient group to subjectively assess their level of cognitive impairment accurately, data from the Cognitive Failures Questionnaire (CFQ) were used [32]. Previous studies showed that the subjective ratings of cognitive deficits by CFS patients were much higher than the level of deficits identified on objective measures of performance. Patients recording high scores on the Cognitive Failures Questionnaire performed significantly worse than those recording low scores for the two reaction time tasks. These data indicate that patients are more able to accurately subjectively rate their cognitive performance than previous reports have indicated [18].

It was also important to investigate whether illness severity negatively affects cognitive functioning, that is are the more severe cases of CFS more functionally impaired. Illness severity as weighted by splitting the patients according to their fatigue-related symptoms and total symptom scores. These scores were then summed and compared to each performance task allowing us to look at performance across a range of severity scores. The controls performed better on each of the tasks than the patient groups even those rated with low illness severity. When considering the CFS group alone, an association between illness severity and poorer cognitive performance is seen in the psychomotor tasks but not the accuracy or recall tasks. These outcomes are partly in agreement with previous studies [17, 18] but also raises further questions as to the exact nature of the cognitive impairment associated with CFS.

Recruiting patients via a specialised out-patients raises issues about generalisability and this is acknowledged as a limitation of the study. However, our data indicating a measurable deficit in motor speed are in keeping with findings from a recent population-based study [45].

The current study provides compelling evidence, not only for the existence of cognitive deficits in Chronic Fatigue Syndrome, but for the nature of these same impairments. Of interest here is that these objectively measured deficits remain even when factors associated with performance impairments are taken into consideration. Improvements or recovery from the illness should, therefore, be accompanied by improvements in cognition. The possibility of measuring recovery in these terms is indicated here: indeed, recent research provides evidence that some of the performance measures used in this study may be valid indicators of recovery [46]. Further studies are, however, required to investigate the possible mechanisms underlying our findings. One area of interest may be the influence of specific neurotransmitters such as acetylcholine, noradrenalin and dopamine in cognitive functioning. Another is the role of insulin in modulating acetylcholine and noradrenalin which in turn influences cognitive functioning. Similarly, dopamine has also been shown to influence motor functioning, verbal fluency, episodic memory and executive functioning [47, 48]. In addition, several groups have identified neuro-anatomical abnormalities in patients with CFS using neuroimaging techniques. These include reduced regional cerebral blood flow, anatomical abnormalities in cortical and subcortical regions and reduced glucose metabolism. The impact of such abnormalities are likely to impact on cognitive performance and may go some way in offering an explanation for some of the cognitive impairments observed in the current study.

There remains a large gap in the current body of scientific knowledge as to the nature of these abnormalities. One area of interest is ‘cause and effect’: that is, do the abnormalities observed cause CFS or does the decrease in physical activity resulting from the illness produce the abnormality. For example, there have been reports of a link between physical activity and neurogenesis [49]. In addition, recent studies have also suggested that neurogenesis is involved in the action of antidepressants [50]. These findings may explain the increased recovery rates detected in CFS patients prescribed antidepressant medication in our study [51].

Finally, there is an emerging hypothesis based on the model of ‘altered self’ which refers to the immune response to infection [52]. Infectious diseases are often accompanied or followed by periods of fatigue, disturbed sleep and an inability to concentrate. The theory is widened to a brain function model where an extended concept of altered self means that the patient does not return to the non-disease state due to the brains failure to recognise that state [52].