Date sent: Sat, 15 Dec 2001 11:42:46 +0100 Source: The Lancet Volume 358, Number 9297, pp 1946-1953 Date: December 8, 2001 URL: http://www.sciencedirect.com/science/journal/01406736 Predictions and associations of fatigue syndromes and mood disorders that occur after infectious mononucleosis ------------------------------------------------------------------------- Peter D White, Janice M Thomas, Hillar O Kangro, William D A Bruce-Jones, John Amess, Dorothy H Crawford, Shirlyn A Grover, Anthony W Clare Departments of Psychological Medicine (P D White MD, W D A Bruce-Jones MPhil, Prof A W Clare MD), Computer Sciences (J M Thomas MSc), Virology (H O Kangro PhD), and Haematology (J Amess MB), St Bartholomew's and the Royal London School of Medicine and Dentistry, London; Department of Virology, St George's Hospital Medical School, London (S A Grover); and Department of Medical Microbiology, University of Edinburgh, Edinburgh, UK (Prof D H Crawford DSc) Correspondence to: Dr Peter D White, Department of Psychological Medicine, St Bartholomew's and the Royal London School of Medicine and Dentistry, St Bartholomew's Hospital, London EC1A 7BE, UK (e-mail:P.D.White@qmul.ac.uk) Summary Background Certain infections can trigger chronic fatigue syndromes (CFS) in a minority of people infected, but the reason is unknown. We describe some factors that predict or are associated with prolonged fatigue after infectious mononucleosis and contrast these factors with those that predicted mood disorders after the same infection. Methods We prospectively studied a cohort of 250 primary-care patients with infectious mononucleosis or ordinary upper-respiratory-tract infections until 6 months after clinical onset. We sought predictors of both acute and chronic fatigue syndromes and mood disorders from clinical, laboratory, and psychosocial measures. Findings An empirically defined fatigue syndrome 6 months after onset, which excluded comorbid psychiatric disorders, was most reliably predicted by a positive Monospot test at onset (odds ratio 2.1 [95% CI 1.4-3.3]) and lower physical fitness (0.35 [0.15-0.8]). Cervical lymphadenopathy and initial bed rest were associated with, or predicted, a fatigue syndrome up to 2 months after onset. By contrast, mood disorders were predicted by a premorbid psychiatric history (2.3 [1.4-3.9]), an emotional personality score (1.21 [1.11-1.35]), and social adversity (1.7 [1.0-2.9]). Definitions of CFS that included comorbid mood disorders were predicted by a mixture of those factors that predicted either the empirically defined fatigue syndrome or mood disorders. Interpretation The predictors of a prolonged fatigue syndrome after an infection differ with both definition and time, depending particularly on the presence or absence of comorbid mood disorders. The particular infection and its consequent immune reaction may have an early role, but physical deconditioning may also be important. By contrast, mood disorders are predicted by factors that predict mood disorders in general. Introduction Chronic fatigue syndrome (CFS) is now generally regarded as a valid diagnosis, although the disorder is heterogeneous in form or aetiology.1-3 Aetiology can best be divided into predisposing, triggering, and maintaining factors.1,2 Certain infections can trigger both long-term fatigue and CFS. For instance, long-term fatigue, CFS, or both, have been reported after infectious mononucleosis,4-8 viral hepatitis,9 viral meningitis,10 and Q fever.11 Clinically diagnosed viral illnesses in primary care may be a minor risk factor for CFS.12 By contrast, common upper-respiratory-tract infections are not.7,13 Of the people who contract the triggering infections, between 9% and 42% have CFS afterwards.7,8,10,11 Therefore, predisposing and maintaining cofactors must determine this outcome. Predisposing factors in CFS (whether or not it follows an infection) may include personality14 and premorbid mood disorders.10,13 Maintaining factors may include current mood disorders,15 social adversity,16 illness beliefs,17 inactivity,17,18 and deconditioning.19,20 Some of these factors may be important in prolonging fatigue after corroborated infections. In particular, reduced exercise capacity and deconditioning9 and psychological distress10,12 have been noted. Buchwald and colleagues8,21 reported predictions of both "non-recovery" (mainly with fatigue) and "psychological distress" in 150 patients with infectious mononucleosis. They found that both higher temperature and physical incapacity at onset predicted non-recovery at 2 months. Female sex, family support, and life events before onset predicted non-recovery 6 months after onset.8 Psychological distress 6 months after onset was predicted both by life events after onset and by relative inactivity just before onset.21 We have already reported, from a previous study of the participants reported here, that adverse life events and difficulties are associated with or predict both psychiatric and depressive disorders, but not the fatigue syndrome that follows infectious mononucleosis.22 We have also shown that infectious mononucleosis, or the agents that cause it, may be an aetiological factor in CFS.7 We now report other potential causes and their interactions in acute and chronic fatigue syndromes after this disorder, exploring two hypotheses. First, that the fatigue syndrome is associated with immunological or virological evidence of delayed recovery after infectious mononucleosis, as well as premorbid psychiatric disorders, social adversity, and physical deconditioning; second, that mood disorders are associated with a premorbid psychiatric history, personality, and social adversity. Top Methods Participants The detailed methods have been described previously.7,22,23 We wrote twice to 469 patients who presented to a primary-care doctor with suspected infectious mononucleosis or an ordinary upper-respiratory-tract infection. Individuals with infectious mononucleosis were recruited consecutively from patients who had a positive Monospot test for the heterophil antibody, undertaken by the two local haematology departments. Individuals with upper-respiratory-tract infections were selected by local general practitioners to ensure masking of the interviewer; a blood sample was sent for a Monospot test to exclude mononucleosis. Participants were aged between 16 and 65 years. Reasons for exclusion were: pregnancy, breastfeeding, imminent departure from the country for a long period, and high risk of HIV infection but no consent for counselling and testing. Two cohorts were studied: students (from the City University, London, and medical and nursing students from St Bartholomew's Hospital, London) and general-practice patients from the boroughs of Hackney and Islington in London. In our previous papers from this study, we classified the participants as follows: primary Epstein-Barr virus (EBV) infection; glandular fever not caused by EBV; and ordinary upper-respiratory-tract infection.7,22,23 These divisions were primarily based on the results of laboratory investigations, but we also used the physical presentation (glandular fever or an upper-respiratory-tract infection) to classify the latter two groups in the absence of defining laboratory data. For this paper, we reanalysed group membership purely on the basis of laboratory data. We classified participants into two groups: those with infectious mononucleosis and those without. The criteria for infectious mononucleosis were a positive Monospot test and at least 10% atypical lymphocytes; a positive result in the test for IgM antibody to EBV capsid antigen; or both these criteria. A primary EBV infection was confirmed by a positive result for IgM antibody to EBV capsid antigen and seroconversion (or quadrupling) of IgG antibodies to EBV capsid antigen and/or seroconversion of antibodies to EBV nuclear antigen.24 Five patients were excluded because they had no clinical or serological evidence of an infection at onset. 118 participants had infectious mononucleosis, 127 did not. Of the 118 participants with infectious mononucleosis, 108 (92%) had primary EBV infections, and a further four (3%) had a positive Monospot test and atypical lymphocytes, with no other cause confirmed. The infections in the remaining six (5%) individuals in the infectious mononucleosis group were human herpesvirus 6 (three cases) and cytomegalovirus (three cases). In the non-infectious-mononucleosis group (n=127) there were: 11 participants with hepatitis A, ten with influenza B, eight with adenovirus, seven with definite or probable streptococcus, 16 with miscellaneous infections (influenza A, Toxoplasma gondii, rubella, coxsackievirus, Mycoplasma pneumoniae, Coxiella burnetti), 14 with atypical lymphocytes alone (no other cause found), 11 with a positive Monospot test but less than 10% atypical lymphocytes (no other cause found), 12 with neutrophilia (no other cause found), and 38 with no cause identified. One investigator (PDW) interviewed all participants, masked to serological and diagnostic data, 1, 2, and 6 months after the onset of their infectious symptoms. Primary-care records were available and examined 2.5 years after onset of infection in 176 (70%) of 250 participants and 132 (60%) of 219 non-participants. Written informed consent was obtained from all participants. The study was approved as ethically satisfactory by the East London and the City Health Authority research ethics committee. To control for a type II error, a power analysis was undertaken before the project started, based on a mean frequency of significant fatigue or ill-health after infectious mononucleosis, estimated from previous studies. Combination of the most relevant studies produced a mean frequency of 23% for significant fatigue 3 months after onset of infectious mononucleosis.5-6 The most important time to compare frequencies was thought to be 2 months after onset, when physical recovery should have occurred. Therefore, an estimate of 30% was made for the infectious-mononucleosis group. We assumed that the frequency of significant fatigue in the non-infectious-mononucleosis group would be no more than 5%.13 With alpha of 0.05 and a power required of 0.90, a minimum of 54 participants would be required for each group. We studied more cases than this minimum so that we could model associations and predictions. Procedures All participants were interviewed and examined in the same way, with the interviewer masked to both the general practitioner's diagnosis and the results of all investigations. The first interview was carried out as soon as possible (mean 1 month) after clinical onset. The second and third interviews were carried out 2 and 6 months after onset. All blood samples were drawn in the morning. A standard history was taken at the first interview. Questions included time from first symptom to height of infection, number of days in bed, time off work or study, current drugs, and smoking and alcohol consumption. Open questions were asked about any past medical history of illness and treatments. Particular questions were asked about past and current history of atopy (eczema, hay fever, or asthma) and past history of tonsillectomy and infectious mononucleosis. Primary-care records corroborated this information. Weight and height were measured at the first interview to allow the calculation of the body-mass index. The interviewer examined each participant's throat and both anterior and posterior triangles of the neck for signs of tonsillitis, pharyngitis, and lymphadenopathy. The presence of at least one enlarged lymph node was taken as evidence of lymphadenopathy and the presence of both throat redness and swelling as evidence of throat inflammation. All participants completed an assessment of their exercise work capacity and cardiovascular response to exercise at each interview. Because all participants were recovering from infections, we did not use a robust measure of aerobic fitness, validated on people without infections. Instead, we modified a 1 min step test, which was designed to measure fitness in untrained young women.25 The participant's resting pulse rate was measured after at least 1 h sitting down, just before the exercise. Participants were then asked to walk up and down a flight of nine standard stairs (each of height 0.18 m) at a reasonable but not fixed pace, for 1 min. The pulse rate was measured 30 s after completion of exercise; the recording took 20 s. We calculated power output as the number of stairs climbed, multiplied by the height of each stair, and the participant's weight. The cardiovascular response to exercise was calculated as the difference between the resting and postexercise pulse rates (exercise pulse-rate difference), in case the postexercise pulse rate (which would normally be used) was raised partly because of the systemic effects of the infection. A measure of physical fitness was calculated by dividing the number of stairs climbed by the exercise pulse-rate difference. 2 years after a participant's last interview, PDW examined all available primary-care records, using a standard form. From these data, we corroborated the date of onset, past medical and psychiatric history, and the number of consultations for illness in the 12 months before onset of the illness being studied. The same information was sought from non-participants. The socioeconomic class of participants and their fathers was calculated.26 The schedule for affective disorders and schizophrenia27 was given at each interview to establish research diagnostic criteria for current and past psychiatric diagnoses.28 Participants were asked standard questions about psychiatric disorders and treatment in their first-degree relatives. The hospital anxiety and depression scale measures self-reported anxiety and depression in patients who are physically ill.29 Social adversity in the previous 12 months was measured at the last interview, by use of the life events and difficulties schedule.30 This semistructured interview has been used extensively to study the relation between social adversity and both psychiatric and physical illnesses.30 Audiotape recordings of the interviews were rated, with masking of all diagnoses, by one investigator (WDAB-J). Two derived variables were analysed. Severe social adversity ("provoking agents") included severe life events, or major social difficulties, independent of both infection and psychiatric illness, that have previously been found to trigger depressive illness.30 Moderate social adversity ("significant stressors") consisted of all independent life events and difficulties which lasted at least 3 months, with some degree of threat to the participant.22,30 The Eysenck personality questionnaire is a self-reported measure of certain personality traits.31 We report extroversion and emotionality (neuroticism) in this paper. We used two strategies to overcome the possibility that current illness (state) might influence the measurement of personality traits. First, we asked participants to answer the questionnaire with the instruction: "Try to disregard your illness when answering these questions and answer 'yes' or 'no' according to how you feel or behave when you are your usual self".32 This strategy provided scores close to those obtained when participants are well, in a previous study.32 We also wrote to the nearest relative of the participant and asked them to answer the Eysenck personality questionnaire as though they were the participant, a strategy which we found gave reasonable correlations with self-reports.33 We also corroborated the family history of psychiatric disorders, with the same questions as asked of the participants. Blood samples were taken at all three interviews and serum and plasma were prepared. All serum samples were frozen on the day of venesection and stored at -20C. Consecutive samples from each individual were tested in batches, by the same assay. All laboratory investigations were done without knowledge of clinical data. Investigations included total and differential white-blood-cell counts and the Monospot slide test for the heterophil antibody.34 EBV-specific investigations included indirect immunofluorescent IgG antibodies to the capsid antigen and nuclear antigen, and both restricted and diffuse early antigen, against already established EBV cell lines.24 We used an enzyme immunoassay to measure IgM antibody to EBV capsid antigen. Other antibody assays have been described previously.7 A regression assay was carried out with separated mononuclear cells. This test measures the ability of a person's cytotoxic T cells to kill autologous EBV-infected B lymphocytes.35 CD4 and CD8 subsets were measured by a fluorescence-activated cell sorter, the mononuclear white cells having previously been separated on a density gradient. We were able to measure only a limited number of T-cell subsets because of technical difficulties. Other tests included serum gamma -glutamyltranspeptidase, aspartate aminotransferase, and bilirubin concentrations. There are two commonly used consensually agreed definitions of CFS36,37 and one empirically derived definition of an acute fatigue syndrome23 (table 1). Because of the uncertain nosological status of CFS we chose to examine all three definitions. Table 1: Criteria for the three fatigue syndromes ----------------------------------------------------------------------------------- Criterion Empirically defined Oxford criteria CDC CFS37 fatigue syndrome23 CFS36 ----------------------------------------------------------------------------------- Main symptom Abnormal physical Severe physical Severe fatigue, not fatigue27 and mental fatigue relieved by rest Minimum 2 weeks 6 months 6 months duration Other symptoms Two of the following: None imperative; Four required of: hypersomnia, poor sleep disturbance, new headache, concentration, and muscle pain multiple joint retardation, supportive pain, muscle pain, irritability, or unrefreshing sleep, anhedonia impaired memory or concentration, sore throat, tender cervical or axillary lymph glands, postexertion malaise Functional Significant: <75 Substantial Substantial impairment on social assessment scale27 Medical Known physical causes Known physical causes Extensive list of exclusions of fatigue of fatigue exclusions Psychiatric Everything, apart Current: psychosis, Current or past: exclusions from phobias, non- bipolar affective psychosis, bipolar specific somatoform disorder, substance affective disorder, disorder, and misuse, eating melancholic neurasthenia disorder, organic depression, brain disorder substance misuse, eating disorder, organic brain disorder CDC=Centers for Disease Control and Prevention. CFS=chronic fatigue syndrome. ----------------------------------------------------------------------------------- Statistical analysis To check for bias by non-participation and differences between groups, chi 2 statistics were used for categorical data. Mann Whitney or independent t tests were used for interval and continuous variables. We used univariate examinations of relative risks of all potentially predictive and associated independent variables of the different definitions of the fatigue syndrome23,36,37 (table 1) and mood disorders. Mood disorders were defined as all depressive disorders (major, minor, and intermittent) combined with general anxiety and panic disorders.28 We particularly concentrated on hypothesised risks (immune response, previous psychiatric illnesses, personality, social adversity, and deconditioning). Relative risks for continuous and interval variables were calculated by categorising the variables, with a split at the median. The comparison group for individuals with a fatigue syndrome was those who were well. The comparison group for those with mood disorders was those who were well and those who had the empirical fatigue syndrome (table 1). All significantly predictive variables were then entered into exploratory forward stepwise logistic regression models to predict the presence or absence of a fatigue syndrome or a mood disorder. The final most explanatory model was a compromise between strength of association or prediction and loss of power because of missing data. Interactions between relevant variables in the final model were sought and entered into the models, when possible. Models were made for 1, 2, and 6 months. Continuous and interval data were transformed by natural logarithms, when necessary and possible, to provide normally distributed data. Because there were only 16 cases of the empirical fatigue syndrome at 6 months, we added in the 26 cases of "fatigue not otherwise specified"7,23 to reduce type II error. Similarly, we added the 18 cases of "idiopathic chronic fatigue"37 to the 17 cases of CFS according to the definition of the Centers for Disease Control and Prevention (CDC) found at 6 months.7 These two categories defined participants with prolonged unexplained abnormal fatigue, but with insufficient accompanying symptoms or disability to qualify for the full syndrome.23,37 To decrease the amount of missing data in two hypothesised variables, we imputed some missing values for the Eysenck personality and physical fitness scores. For the personality scores we used the relative's ratings when self-ratings were missing (three cases) and vice versa (16 cases). After this procedure, mean scores and SDs varied by no more than 0.2. For the fitness data, we found that median scores varied by no more than 0.04 units across interviews: 1.94 (IQR 1.40-3.00), 1.94 (1.43-2.88), and 1.90 (1.40-2.67) at 1, 2, and 6 months, respectively, with Pearson's correlations of 0.50 for scores at 1 and 2 months (p<0.001) and 0.54 for scores at 2 and 6 months (p<0.001). We therefore transposed data from a participant's nearest available interview. This adjustment occurred for 65, 28, and 13 participants at 1, 2, and 6 months, respectively. Top Results 250 people (59% of those eligible) took part. The commonest reasons for non-participation were no contact (62% of non-participants, in most cases because of inaccurate addresses) and protocol exclusions (20%). Only 37 (8%) of 469 potential participants refused to participate. There were no significant differences between participants and non-participants in terms of the proportion with infectious mononucleosis at onset, preonset or postonset primary-care records of psychiatric or fatigue diagnoses, or the number of surgery attendances before or after onset. Participants were more likely than non-participants to be female (51% vs 40%) and were slightly older (median 22 vs 21 years; p<0.01). Five patients were excluded because they had no clinical or serological evidence of an infection at onset. 135 participants (55%) were from the student cohort; 110 (45%) were from general primary care. Significantly more students than primary-care patients had ordinary upper-respiratory-tract infections. There were no significant differences in the incidence or prevalence of fatigue syndromes or psychiatric disorders between the two cohorts at any time. Therefore, the two cohorts were combined for further analyses, although sample, age, sex, vocabulary IQ, and socioeconomic class were examined in univariate analyses. There were no significant differences between the infectious-mononucleosis and non-infectious-mononucleosis groups in distribution by age, sex, socioeconomic class, or personality. There were no significant differences in preonset social adversity, psychiatric history, or surgery attendance. Participants with infectious mononucleosis were less likely to be from the student sample (p=0.002) and had a slightly lower vocabulary IQ (mean 105 [SD 10] vs 109 [9]; p=0.004). Neither past psychiatric history nor past history of a mood disorder had a significant predictive influence on the empirically defined fatigue syndrome (table 2), however past history was defined and measured. Similarly, neither social adversity nor personality was a significant risk for the empirical fatigue syndrome, with the exception of extroversion, which was a negative risk at 6 months only for those participants with the empirical fatigue syndrome.33 Table 2: Univariate relative risks (95% CIs) for the three differently defined fatigue syndromes ----------------------------------------------------------------------------------- Relative risk (95% CI) of fatigue syndrome Empirical fatigue Oxford CDC syndrome23 CFS36 CFS/ICF37 1 month 2 months 6 months 6 months 6 months ----------------------------------------------------------------------------------- Number with syndrome 87 65 42 38 35 Demographic data Sex (F/M) 1.5 1.2 1.4 1.3 1.4 (1.1-2.0)* (0.8-1.7) (0.8-2.4) (0.7-2.3) (0.7-2.5) Age 0.9 1.1 1.1 0.8 1.2 (0.7-1.2) (0.7-1.6) (0.6-1.8) (0.4-1.4) (0.7-2.3) Sample (GP/student) 1.3 1.6 1.8 3.1 2.5 (1.0-1.8)* (1.1-2.3)* (1.0-3.0)* (1.6-5.9)* (1.3-4.5)* Vocabulary IQ 0.9 0.6 0.5 0.8 0.5 (0.6-1.3) (0.3-1.1) (0.2-1.1) (0.3-1.7) (0.2-1.2) Socioeconomic class 1.5 1.7 1.5 2.6 2.3 (1.0-2.1)* (1.0-2.9) (0.7-3.0) (1.0-6.3)* (0.9-5.8) Father's 1.0 1.2 1.6 1.1 1.8 socioeconomic class (0.8-1.4) (0.8-1.8) (0.9-2.8) (0.6-2.1) (1.0-3.4) Clinical and laboratory data Infectious 2.0 4.3 3.2 2.1 2.5 mononucleosis (1.4-2.9)* (2.3-7.9)* (1.6-6.4)* (1.1-4.0)* (1.3-5.1)* Positive Monospot 2.3 5.1 2.5 2.0 2.2 at onset (1.6-3.3)* (2.7-9.7)* (1.4-4.7)* (1.1-3.7)* (1.1-4.3)* EBV IgM positive at 1.8 2.6 3.2 2.9 2.1 1 month (1.3-2.5)* (1.5-4.5)* (1.6-6.7)* (1.3-6.4)* (0.9-4.8) EBV IgM positive at 1.8 1.7 1.4 1.4 1.5 interview (1.3-2.5)* (1.2-2.6)* (0.6-3.0) (0.6-3.5) (0.6-3.7) Cervical 1.8 4.9 1.4 1.6 1.7 lymphadenopathy (1.2-2.7)* (2.5-9.7)* (0.8-2.5) (0.8-3.0) (0.9-3.3) Atopy 1.0 0.8 1.6 0.6 1.0 (0.8-1.3) (0.6-1.3) (0.9-2.7) (0.3-1.1) (0.5-2.0) AST at 1 month 1.2 2.4 2.3 1.7 1.8 (0.8-1.6) (1.3-4.5)* (1.0-4.8) (0.7-4.1) (0.7-4.8) gamma GT at 1 month 1.2 1.4 1.5 1.4 1.3 (0.8-1.6) (0.8-2.3) (0.7-3.0) (0.6-3.1) (0.5-3.5) Bilirubin at 1 month 0.6 0.7 0.7 0.8 0.9 (0.5-0.9)* (0.4-1.2) (0.3-1.3) (0.4-1.8) (0.3-2.3) Fatigue at onset 1.8 2.0 2.1 1.8 1.5 (1.4-2.5)* (1.3-3.0) (1.0-4.3)* (0.9-3.2) (0.8-2.9) Time in bed at onset 1.6 2.0 1.9 3.2 2.2 (1.2-2.3)* (1.2-3.2)* (0.9-4.0) (1.5-7.1)* (1.1-4.6)* Exercise power 0.8 1.1 0.9 0.8 0.8 (0.6-1.1) (0.7-1.7) (0.4-1.8) (0.4-1.5) (0.4-1.6) Fitness 0.7 0.6 0.4 0.4 0.5 (0.5-0.9)* (0.4-1.0)* (0.2-0.9)* (0.2-0.9)* (0.2-1.0)* GP attendance in 1.1 1.2 1.5 4.1 3.0 yearbefore onset (0.8-1.6) (0.6-2.1) (0.6-3.7) (1.7-10.3)*(1.3-7.0)* Psychosocial data PPD at any time 1.2 1.1 1.6 1.8 1.6 (0.9-1.6) (0.7-1.6) (0.8-3.3) (1.0-3.2) (0.8-3.4) PPD in year before 1.3 1.5 1.1 1.5 1.7 onset (1.0-1.7) (1.0-2.2) (0.6-2.0) (0.8-2.7) (0.9-3.1) PPD in 2 weeks 1.3 1.3 0.8 1.9 1.5 before onset (0.9-2.1) (0.7-2.4) (0.3-2.2) (1.0-3.7) (0.5-4.5)* GP record of PPD 1.4 1.6 1.9 4.0 2.3 (1.0-2.0) (0.9-2.8) (1.0-3.7) (2.2-7.4)* (1.2-4.6)* GP record of any 1.4 1.2 1.8 3.7 2.0 premorbid (1.0-2.1) (0.6-2.3) (0.9-3.6) (2.0-6.8)* (1.0-4.0) psychiatric treatment PMD at any time 1.1 1.2 1.5 2.1 2.2 (0.8-1.5) (0.8-1.8) (0.9-2.6) (1.2-3.8)* (1.2-3.9)* PMD in year before 1.2 1.3 1.1 1.5 1.7 onset (0.8-1.7) (0.8-2.1) (0.6-2.1) (0.8-2.8) (0.9-3.2) PMD in 2 weeks before 1.3 1.5 0.7 2.9 2.8 onset (0.7-2.3) (0.7-3.4) (0.1-4.5) (1.5-5.6)* (1.4-5.8)* GP record of PMD 1.2 1.3 2.1 3.5 2.1 (0.8-2.0) (0.6-2.9) (1.0-4.4)* (2.0-6.3)* (1.0-4.4) HADS depression score NA 1.5 2.1 1.6 2.1 at 1 month (0.9-2.5) (1.0-4.5)* (0.7-3.7) (0.8-5.7) HADS anxiety score at NA 1.2 2.0 3.9 2.5 1 month (0.7-2.0) (1.0-4.2) (1.4-11.0)*(0.9-6.9) Mood disorder at 2 NA NA 1.6 4.4 3.0 months (0.7-3.4) (2.6-7.5)* (1.6-5.5)* Severe social 1.0 1.6 1.6 1.9 2.4 adversity (0.7-1.5) (1.0-2.7) (0.6-4.1) (0.8-4.5) (1.2-4.7)* Moderate social 1.1 1.1 0.9 1.0 1.0 adversity (0.8-1.5) (0.7-1.8) (0.5-1.6) (0.4-2.2) (0.5-2.4) Self-rated 0.8 1.0 0.6 1.0 1.1 extroversion (0.6-1.1) (0.7-1.5) (0.3-1.2) (0.6-1.8) (0.5-2.2) Relative-rated 0.8 0.9 0.8 0.6 0.6 extroversion (0.6-1.1) (0.6-1.3) (0.5-1.4) (0.3-1.1) (0.3-1.2) Self-rated 1.1 1.3 1.1 1.9 1.7 emotionality (0.8-1.5) (0.9-2.0) (0.7-1.9) (1.0-3.5)* (0.8-3.6) Relative-rated 0.9 1.1 1.1 2.3 1.8 emotionality (0.7-1.2) (0.8-1.7) (0.6-1.8) (1.2-4.2)* (0.9-3.3) ICF=idiopathic chronic fatigue; AST=aspartate aminotransferase; gamma GT=gamma-glutamyltranspeptidase; PPD=premorbid psychiatric disorder; PMD=premorbid mood disorder; HADS=hospital anxiety and depression scale. NA=not applicable. GP=general practitioner. *p<0.05. **Empirical fatigue syndrome and fatigue nototherwise specified. ----------------------------------------------------------------------------------- Participants with the empirical fatigue syndrome had higher CD8 lymphocyte counts than those who were well, but only at 1 month and not thereafter. This finding applied to all participants and as a trend in those with infectious mononucleosis: in that group, participants with the fatigue syndrome had a median CD8-lymphocyte count of 1.21x109/L (IQR 0.69-1.83), compared with 0.92x109/L (0.40-1.06) in those who were well (p=0.07; 36 participants in total). There were no significant differences in CD3 and CD4 counts. In contrast to the empirical fatigue syndrome, the two consensually agreed definitions of CFS had predictions that were a mixture of the factors that predicted the empirical fatigue syndrome or mood disorders (tables 2 and 3). In particular, preonset primary-care attendance and previous mood disorders were predictive of these definitions of CFS. In contrast to the empirical fatigue syndrome, the strongest risk for having a mood disorder occurred if a participant had had a premorbid mood or psychiatric disorder, whether measured by interview or primary-care record (table 3). Other significant risks were social adversity and emotional personality (both by self-rating and relative scoring). Table 3: Univariate relative risks of mood disorders ----------------------------------------------------------------------------------- Relative risk (95% CI) of mood disorder 1 month 2 months 6 months ----------------------------------------------------------------------------------- Number with mood 43 39 36 disorder Demographic data Sex (F/M) 0.7 (0.4-1.4) 1.0 (0.6-1.9) 1.3 (0.7-2.6) Age 1.1 (0.7-1.9) 1.5 (0.8-2.7) 1.0 (0.5-2.0) Sample (GP/student) 1.7 (1.0-2.9) 2.2 (1.2-4.0)* 1.1 (0.6-2.2) Vocabulary IQ 0.4 (0.2-1.0) 1.1 (0.5-2.5) 0.8 (0.4-2.1) Socioeconomic class 0.9 (0.4-1.7) 1.3 (0.6-2.7) 1.5 (0.7-3.4) Father's socio- 1.1 (0.6-1.9) 0.8 (0.4-1.6) 1.2 (0.6-2.5) economic class Clinical and laboratory data Infectious 1.3 (0.7-2.2) 0.8 (0.4-1.4) 1.3 (0.7-2.4) mononucleosis Positive Monospot 1.2 (0.7-2.1) 0.8 (0.4-1.3) 1.1 (0.6-2.3) at onset EBV IgM positive at 1.4 (0.8-2.5) 0.9 (0.5-1.9) 1.4 (0.7-2.9) 1 month EBV IgM positive at 1.4 (0.8-2.5) 0.9 (0.4-1.6) 1.4 (0.6-3.2) interview Cervical lymphadeno- 1.0 (0.5-1.9) 1.0 (0.5-2.0) 1.1 (0.5-2.5) pathy Atopy 0.7 (0.4-1.3) 0.6 (0.3-1.1) 0.6 (0.3-1.3) AST at 1 month 1.1 (0.6-2.1) 0.6 (0.3-1.2) 0.8 (0.3-1.9) gamma GT at 1 month 1.6 (0.8-2.9) 1.2 (0.6-2.6) 0.9 (0.4-2.0) Bilirubin at 1 month 1.3 (0.7-2.5) 0.7 (0.4-1.5) 0.3 (0.1-0.8)* Fatigue at onset 1.8 (1.0-3.2)* 2.4 (1.2-4.5)* 0.8 (0.4-1.9) Time in bed at onset 1.4 (0.8-2.4) 1.4 (0.8-2.7) 1.5 (0.7-3.2) Exercise power 1.7 (0.9-3.0) 1.1 (0.6-1.9) 1.1 (0.5-2.3) Fitness 1.0 (0.6-1.7) 0.9 (0.5-1.6) 0.8 (0.4-1.7) GP attendance in 1.6 (0.8-3.1) 1.8 (0.9-3.7) 5.0 (1.5-16.2)* year before onset Psychosocial data PPD at any time 2.7 (1.5-4.9)* 4.3 (2.2-8.7)* 4.7 (1.9-12.0)* PPD in year before 3.3 (1.9-5.6)* 6.5 (3.3-12.5)* 4.5 (2.1-9.4)* onset PPD in 2 weeks 5.3 (3.3-8.4)* 6.5 (3.9-10.8)* 6.1 (3.1-11.8)* before onset GP record of PPD 3.3 (1.9-5.6)* 5.4 (2.9-10.2)* 4.2 (1.9-9.0)* GP record of any 2.6 (1.5-4.6)* 5.0 (2.7-9.3)* 4.3 (2.2-8.4)* premorbid psychiatric treatment PMD at any time 2.2 (1.3-3.8)* 2.8 (1.6-5.1)* 3.3 (1.5-6.9)* PMD in year before 2.3 (1.3-3.8)* 3.0 (1.7-5.2)* 3.5 (1.8-7.0)* onset PMD in 2 weeks 3.8 (2.4-6.3)* 5.5 (3.5-8.9)* 7.9 (4.4-14.4)* before onset GP record of PMD 3.1 (1.7-5.6)* 3.8 (2.2-6.6)* 3.1 (1.4-6.8)* HADS depression NA 2.8 (1.2-6.5)* 1.3 (0.5-3.4) score at 1 month HADS anxiety score NA 4.4 (1.6-12.1)* 3.2 (1.1-9.2)* at 1 month Mood disorder at NA NA 10.4 (5.1-21.4)* 2 months Severe social 1.4 (0.7-2.8) 3.8 (2.1-6.8)* 6.4 (2.7-15.0)* adversity Moderate social 1.7 (0.8-3.7) 3.3 (1.3-8.1)* 2.9 (0.9-9.5) adversity Self-rated 0.6 (0.4-1.1) 0.7 (0.4-1.2) 0.6 (0.3-1.2) extroversion Relative-rated 0.4 (0.2-0.7)* 0.5 (0.3-0.9)* 0.4 (0.2-0.8)* extroversion Self-rated 3.8 (1.9-7.6)* 3.6 (1.8 -7.1)* 6.1 (2.2-17.0)* emotionality Relative-rated 2.4 (1.3-4.4)* 2.4 (1.3-4.6)* 7.6 (2.4-24.5)* emotionality AST=aspartate aminotransferase; gamma GT=gamma-glutamyltranspeptidase. PPD=premorbid psychiatric disorder; PMD=premorbid mood disorder; HADS=hospital anxiety and depression scale; NA=not applicable; GP=general practitioner. *p<0.05. ----------------------------------------------------------------------------------- Variables that neither predicted nor were associated with fatigue syndromes or mood disorders (those not shown in tables 2 or 3) included: insidious onset, total number of physical symptoms at onset, previous tonsillectomy, throat inflammation, taking antibiotics at any time after onset (multiple courses or penicillins versus non-penicillins), any other medication, body-mass index, family psychiatric history, number of friends or confidants, atypical lymphocyte count, IgG antibodies to EBV capsid antigen, antibodies to EBV nuclear antigen, antibody titres against EBV restricted and diffuse early antigen, and the regression assay. Table 4 shows the final models for the empirical fatigue syndrome at 1, 2, and 6 months. Having a positive Monospot test close to onset was predictive for all three times (with this variable being closely associated with having infectious mononucleosis). Enlarged cervical glands were associated at 1 month and 2 months. Physical fitness was negatively associated with the fatigue syndrome at 1 month and a negative predictor for 2 and 6 months, whether actual or imputed fitness data were used. Time spent in bed at onset predicted a fatigue syndrome at 2 months. There was a significant correlation between initial fatigue and bed rest (r=0.49, p<0.001), which probably explains why the latter was not included in the model at 1 month. There were no significant inverse correlations between bed rest and physical fitness at any interview (r<0.15 at all interviews). Although there were interactions found between certain variables at all three times, none entered the final models. At no time did premorbid or postmorbid psychiatric or mood disorders enter a model. Similarly, neither postmorbid depression nor anxiety scores from the hospital anxiety and depression scale entered a model. Table 4: Logistic regression models for the empirically defined fatigue syndrome ----------------------------------------------------------------------------------- Odds ratio (95% CI) of fatigue syndrome 1 month 2 months 6 months ----------------------------------------------------------------------------------- Number in analysis 114 154 179 Monospot positive at 1.80 (1.13-2.87)* 2.47 (1.47-4.14)*** 2.14 (1.37-3.34) onset Fatigue at onset 1.77 (1.14-2.73)** .. .. Enlarged cervical 1.80 (1.12-2.89)* 2.34 (1.44-3.80)*** .. glands at interview Log fitness (****) 0.34 (0.16-0.73)** 0.34 (0.14-0.84* 0.35 (0.150-0.81)* Days in bed at .. 1.13 (1.06-1.21)*** .. onset Extrovert personality .. .. 0.87 (0.79-0.97)*** score *p<0.05, **p<0.001, ***p<0.001 ****Measured at 1 month in models for 1 and 2 months, and at 2 months in model for 6 months. p values for goodness of fit of models: p=0.65 at 1 month, 0.72 at 2 months, and 0.51 at 6 months. The proportions of participants correctly classified at 1 month, 2 months, and 6 months were 71%, 81%, and 83% for all cases and 78%, 64%, and 25% for fatigue syndrome patients. ----------------------------------------------------------------------------------- When only those participants with infectious mononucleosis infection were examined, the explanatory models did not change, apart from only physical fitness being predictive of empirically defined fatigue syndrome at 6 months. The mean duration of illness in those with a fatigue state at 6 months was 18 weeks (SD 7). Table 5 shows the models for CFS at 6 months, as defined by the Oxford or the CDC criteria. Although bed rest and physical fitness were in both CFS models, having infectious mononucleosis and having a positive Monospot test at onset were not. Instead, previous mood disorder and emotional personality (for the Oxford definition) were predictive. There was a positive interaction between emotional personality and bed rest at onset. Both CFS definitions were more likely to occur in the non-student sample. Table 5: Logistic regression models for the Oxford and CDC criteria for CFS ----------------------------------------------------------------------------------- Odds ratio (95% CI) ------------------------------------------------------------- Oxford defined CFS36 CDC-defined CFS CFS and ICF37 ----------------------------------------------------------------------------------- Number in analysis 182 171 Interaction of days ** .. x emotiona personality score Days in bed .. 1.07 (1.02-1.12)** Sample (GP/student) 3.00 (1.63-5.49)*** 1.64 (1.02-2.61)* Mood disorder at 2 2.32 (1.22-4.42)** Premorbid mood 1.82 (1.15-2.89)** disorder Log fitness at 0.35 (0.22-0.90)* 0.40 (0.16-1.00)* months *p<0.05, **p<0.01, ***p<001. p values for goodness of fit of models: p=0.71 for Oxford criteria; 0.70 for CDC criteria. GP=general practitioner. The proportions of participants correctly classified were 87% for Oxford and 84% for CDC for all cases and 34% and 21% respectively for CFS cases. ----------------------------------------------------------------------------------- Since both the Oxford and CDC CFS criteria allow comorbid mood disorders, we further analysed these data to see whether previous mood disorder and emotional personality were predictive of CFS due to the presence of comorbid mood disorders. Ten of 13 participants with both Oxford-defined CFS and a current mood disorder had had a mood disorder 4 months earlier, compared with five of 25 participants without a comorbid mood disorder (odds ratio 5.11 [95% CI 1.68-15.6]). Eight of ten participants with both CFS defined by the CDC criteria and a current mood disorder had had a mood disorder before their infection, compared with 11 of 25 participants without a comorbid mood disorder (odds ratio 5.09 [0.89-29.0]). The mean Eysenck personality questionnaire emotionality score was 14.7 (SD 5.4) in participants with Oxford-defined CFS and a comorbid mood disorder, compared with 9.8 (5.0) in those without (p=0.008). There were insufficient numbers to model infectious mononucleosis participants alone. Of the 35 participants with CFS or idiopathic chronic fatigue by the CDC criteria, 29 (83%) participants also met Oxford CFS criteria, and 24 (69%) also met criteria for the empirically defined fatigue state (fatigue syndrome and fatigue not otherwise specified). Of the 38 participants with Oxford CFS, 29 (76%) participants also met the criteria for CFS or idiopathic chronic fatigue (ICF) by the CDC criteria, and 22 (58%) also met criteria for the empirically defined fatigue state. Table 6 shows the models for mood disorders. The models were reliable over time, with premorbid psychiatric disorder, emotional personality, and social adversity in all the models. No interactions between variables entered the models at 1 month or 2 months. At 6 months, there was a significant interaction between emotional personality and severe social adversity, which replaced social adversity in the model (p=0.01). The interaction was omitted from the final model because it did not improve the model statistics. Infectious mononucleosis never predicted a mood disorder. Table 6: Logistic regression models for mood disorders ----------------------------------------------------------------------------------- Odds ratio (95% CI) of mood disorder 1 month 2 months 6 months ----------------------------------------------------------------------------------- Total number in model 172 196 186 Number with mood 33 34 28 disorder Any psychiatric 2.54 (1.45-4.44) 2.77 (1.70-4.51) 2.33 (1.38-3.92) disorder just before onset Emotionalpersonality 1.25 (1.13-1.38) 1.14 (1.04-1.25) 1.21 (1.11-1.35) score Moderatesocial 2.10 (1.08-4.06)* .. .. adversity Severe social .. 1.96 (1.22-3.15) 1.72 (1.03-2.88) adversity *p<0.05, **p<0.01, ***p<0.001. p values for goodness of fit of models: p=0.97 for 1 month, 0.02 at 2 months, p<0.001 at 6 months. The proportions of participants correctly classified were 85% at 1 month, 89% at 2 months, and 90% at 6 months for all cases and 48%, 50%, and 50% for cases of mood disorder. ----------------------------------------------------------------------------------- Discussion We found that predictions of a fatigue syndrome depended primarily on the particular definition used, and secondly on when participants were examined. The most reliable predictors of the empirically defined syndrome were a positive Monospot at onset and lower physical fitness. Whereas mood disorders were predicted by a premorbid psychiatric history, emotional personality, and social adversity, none of these factors predicted having an empirical fatigue syndrome. The consensually defined chronic fatigue syndromes (Oxford and CDC) were predicted by a mixture of the factors that predicted the empirical fatigue syndrome (lower fitness and bed rest) and mood disorders (previous history and emotional personality). Neither univariate nor logistic regression analyses supported the original hypotheses that a previous psychiatric history or social adversity would predict the empirically defined syndrome. Both mood disorder at 2 months and emotional personality (neuroticism) predicted Oxford-defined CFS. Premorbid mood disorder predicted CDC-defined CFS. These predictions of CFS were related more to having a comorbid mood disorder than to having CFS itself, although the 95% CIs for this confirmation were wide owing to small numbers. The significant effect of the sample in both CFS models may have been related to the higher proportion of infectious mononucleosis in the non-student sample causing more CFS. These data support the difference in nosology and aetiology between both acute and chronic fatigue syndromes (of relatively short duration) and mood disorders. They also suggest that the Oxford36 and CDC37 criteria for CFS should be used with caution or only with stratification by mood disorders in aetiological studies.3 Almost all previous studies have found close associations between CFS and both previous and current psychiatric disorders or social adversity.1,2,10, 13,15,16 Why were our data so different? The three most likely reasons are differences in sampling, definitions, and timing. Few previous studies of CFS have been prospective or have followed a corroborated infection. The one exception also reported primary-care patients, but the participants had upper-respiratory-tract infections at onset, not infectious mononucleosis.13 The one other prospective study of infectious mononucleosis also found no link between premorbid psychiatric disorders and non-recovery.8 All previous definitions of CFS have been derived consensually, rather than empirically, and most have included comorbid mood disorders (table 1).3,36,37 Associations with psychosocial factors would be expected. The third explanation is that we studied participants in the first 6 months of their illness, whereas most previous studies observed patients several years after onset. This explanation would fit with the hypothesis that psychosocial factors are unimportant in a fatigue syndrome of several months duration, but may become more important with time, as suggested by our finding that extroversion was a negative predictor at 6 months, but not beforehand. Having a positive Monospot test at onset was always the first factor entered into the model of the empirical fatigue syndrome. Having a positive result for IgM antibody EBV capsid antigen at 1 month was a marker for both infectious mononucleosis and a positive Monospot, since it became non-significant after one of these two factors were entered into the models. The Monospot test was itself primarily a marker for infectious mononucleosis. However, its presence at onset may have reflected more than this feature, because it was included in the models for infectious-mononucleosis participants alone, and it has previously been shown to predict slower recovery after infectious mononucleosis.38 Cervical lymphadenopathy was part of the model for those participants with infectious mononucleosis, which suggests that this feature was not simply a marker of the original infection. This association has been reported before with infectious mononucleosis.5,6 The CD8-lymphocyte data are preliminary but not unexpected in view of the known association between CD8 cells and acute infectious mononucleosis.39 The importance of the initial heterophil antibody, supported by the association with cervical lymphadenopathy, and the trend towards greater numbers of CD8 lymphocytes close to onset, suggests that an interaction between the initial virus infection and the consequent immune response may have an aetiological role in the acute fatigue syndrome, close to onset. This immune response may be related to release of particular cytokines, which in turn may cause an acute fatigue syndrome, as part of acute sickness behaviour.40 The relation between initial fatigue and bed rest suggests that more severe fatigue at onset may have led to a longer time in bed, or that a common factor, such as either acute sickness behaviour40 or avoidant coping,2 led to one or the other. Neither preonset consulting behaviour nor the number of symptoms at onset predicted a fatigue syndrome, whereas there was an interaction between bed rest and emotional personality in the model for Oxford-defined CFS. Physical fitness was a reliable negative predictor of, or an association with, both the empirical and the consensually defined fatigue syndromes, for both the actual and imputed data. Although some caution should be exercised in interpretation of these findings in view of our necessarily adapting the usual step test, the reliability across time and definitions was remarkable. We could not measure fitness before onset, so we cannot determine the direction of causality between bed rest and fitness. Bed rest may have led to deconditioning,17,18,20 although the lack of a significant inverse correlation between the two variables casts doubt on this idea. The negligible change in physical fitness over the three interviews suggests that fitness may have been a stable trait that independently predisposed some participants to a fatigue syndrome. This explanation would also suggest that the amount of bed rest at onset was an independent predictor of prolonged fatigue (as in the models for CDC and Oxford CFS), possibly by being a marker for consequent inactivity.18 The previous findings that inactivity just before the onset of infectious mononucleosis predicted subsequent symptoms is consistent with this notion,21 as is physical incapacity at onset predicting non-recovery 2 months after infectious mononucleosis.8 Physical inactivity does seem to contribute to fatigue in CFS.17,18 The association between deconditioning and both postinfectious and chronic fatigue syndromes has been found before.9,19,20 Since our measure of deconditioning predicted a fatigue syndrome, and since reversal of deconditioning in CFS with graded exercise therapy is associated with increased exercise capacity,20 we suggest that deconditioning is an important maintaining factor for postinfectious fatigue. Prevention of postinfectious fatigue by an early return to physical activity may be possible. A controlled trial of patients recovering from infectious mononucleosis found that bed rest led to a greater number of fatigued patients when compared with unrestricted movement.4 Although there is some evidence that life events can trigger CFS16 and non-recovery after infectious mononucleosis,8 this study and others suggest that social adversity triggers mood disorders that may in turn cause fatigue, rather than CFS itself.21,23,41 There was no single predictor of a fatigue syndrome after infection. Having infectious mononucleosis and deconditioning were both important. Other aetiological factors changed with time and included immune, behavioural, and personality factors. Since an interactive model of maintaining factors in CFS suggests that certain illness beliefs lead to inactivity, which itself leads to more fatigue,17 future studies should look for interactions between certain illness beliefs (which we did not measure), personality, inactivity, and deconditioning. Although infectious mononucleosis and deconditioning were consistent predictors in all three models, we emphasise that exploratory regression analyses are prone to produce false associations. Our models were weaker explanations at 6 months than nearer onset. We used less stringent criteria for CFS at 6 months. These data should be interpreted with caution until they have been corroborated in other studies. We can be more confident that the findings are representative of primary-care patients in general, because of the similar general practice outcomes at 2.5 years between participants and non-participants.7 A previous psychiatric history was the most consistent predictor of mood disorders along with both the personality trait of emotionality (neuroticism) and social adversity. Although we must be cautious about interpreting these data without replication, particularly in view of the poor fits of the models at 2 and 6 months, the consistency of these predictions at three different times suggests that the findings are reliable. A previous history is an important risk factor for mood disorders after other virus infections.42 Social adversity is a risk factor both for mood disorders in general30 and after infectious mononucleosis.21,22 The link between mood disorders and premorbid emotional personality traits has been noted many times before.43 Even the interaction between emotional personality and social adversity has been reported.43 We did not find that postviral mood disorders were related to physical markers of the original infection. Aetiological predictions for postinfectious mood disorders did not differ from those of mood disorders in general. Contributors Peter White designed the study, carried out all the interviews, collected the general practice data, analysed the data with Janice Thomas, and wrote the paper. John Amess selected all potential participants. John Amess, Dorothy Crawford, Hillar Kangro, and Shirlyn Grover provided laboratory data. Hillar Kangro determined patients' group membership. Bill Bruce-Jones coded and analysed the life events and difficulties schedule data. Anthony Clare supervised and helped to design the study. All investigators commented on and approved the final draft. Acknowledgments We thank all the general practitioners in the City of London and the boroughs of Hackney and Islington for their active support, especially John Coulson, Marion Fleetwood, Michael Glenn, Jenny Hope, and Gill Neaman; Katharine Luke, Russell Blacker, Mike Horton, Peter Sanders, Leszek Borysiewicz, Raymond Heath, Trevor Silverstone, and Harold Stern for their help and advice; and Matthew Hotopf and Anthony Pinching who helpfully commented on an earlier draft. The Mental Health Foundation and the Linbury Trust provided funding. Top References 1 Royal Colleges of Physicians, Psychiatrists, and General Practitioners. Chronic fatigue syndrome: report of a joint working group. London: Royal Colleges of Physicians, Psychiatrists, and General Practitioners: 1996, 1-58. 2 Wessely SC, Hotopf M, Sharpe M. Chronic fatigue and its syndromes. Oxford: Oxford University Press, 1998. 3 Lloyd AR. Chronic fatigue and chronic fatigue syndrome: shifting boundaries and attributions. Am J Med 1998; 105: 7S-10S. 4 Dalrymple W. Infectious mononucleosis: 2, relation of bed rest and activity to prognosis. Postgrad Med 1964; 35: 345-49. 5 Thompson DS, Godleski J, Herman S. Prognosis post infectious mononucleosis . J Am Coll Health Assoc 1969; 17: 453-57. 6 Lambore S, McSherry J, Kraus AS. Acute and chronic symptoms of mononucleosis. J Fam Pract 1991; 33: 33-37. 7 White PD, Thomas JM, Amess J, et al. Incidence, risk and prognosis of acute and chronic fatigue syndromes and psychiatric disorders after glandular fever. Br J Psychiatry 1998; 173: 475-81. 8 Buchwald DS, Rea TD, Katon WJ, Russo JE, Ashley RL. Acute infectious mononucleosis: characteristics of patients who report failure to recover. Am J Med 2000; 109: 531-37. 9 Berelowitz GJ, Burgess AP, Thanabalasingham T, et al. Post-hepatitis syndrome revisited. J Viral Hepat 1995; 2: 133-38. 10 Hotopf M, Noah N, Wessely S. Chronic fatigue and psychiatric morbidity following viral meningitis; a controlled study. J Neurol Neurosurg Psychiatry 1996; 60: 495-503. 11 Ayres JG, Flint N, Smith EG, et al. Postinfection fatigue syndrome following acute Q fever: follow up study of patients involved in the 1989 outbreak in the West Midlands. QJM 1998; 91: 105-23. 12 Cope H, David A, Pelosi A, Mann A. Predictions of chronic "post viral" fatigue. Lancet 1994; 344: 864-68. 13 Wessely S, Chalder T, Hirsch S, et al. Postinfectious fatigue: prospective cohort study in primary care. Lancet 1995; 345: 1333-38. 14 Buckley L, MacHale SM, Cavanagh JTO, Sharpe M, Deary IJ, Lawrie SM. Personality dimensions in chronic fatigue syndrome and depression. J Psychosom Res 1999; 46: 395-400. 15 Clark MR, Katon W, Russo J, et al. Chronic fatigue: risk factors for symptom persistence in a 2.5 year follow-up study. Am J Med 1995; 98: 187-95. 16 Theorell T, Blomkvist V, Lindh G, et al. Critical life events, infections, and symptoms during the year preceding chronic fatigue syndrome (CFS): an examination of CFS patients and subjects with a non-specific life crisis. Psychosom Med 1999; 61: 304-10. 17 Vercoulen JHMM, Swanink CMA, Galama JMD, et al. The persistence of fatigue in chronic fatigue syndrome and multiple sclerosis: development of a model. J Psychosom Res 1998; 45: 507-17. 18 Van der Werf SP, Prins JB, Vercoulen JHMM, van der Meer JWM, Bleijenberg G. Identifying physical activity patterns in chronic fatigue syndrome using actigraphic assessment. J Psychosom Res 2000; 49: 373-79. 19 Riley MS, O'Brien CJ, McCluskey DR, et al. Aerobic work capacity in patients with chronic fatigue syndrome. BMJ 1990; 301: 953-56. 20 Fulcher KY, White PD. Strength and physiological response to exercise in patients with the chronic fatigue syndrome. J Neurol Neurosurg Psychiatry 2000; 69: 302-07. 21 Katon W, Russo J, Ashley RL, Buchwald D. Infectious mononucleosis: psychological symptoms during acute and subacute phases of illness . Gen Hosp Psychiatry 1999; 21: 21-29. 22 Bruce-Jones WDA, White PD, Thomas JM, et al. The effect of social adversity on the fatigue syndrome, psychiatric disorders and physical recovery, following glandular fever. Psychol Med 1994; 24: 651-59. 23 White PD, Grover SA, Kangro HO, et al. The validity and reliability of the fatigue syndrome that follows glandular fever. Psychol Med 1995; 25: 917-24. 24 Henle W, Henle GE, Horwitz CA. Epstein-Barr virus specific diagnostic tests in infectious mononucleosis. Hum Pathol 1974; 5: 551-65. 25 Harvey VP, Scott GD. The validity and reliability of a one-minute step test for women. J Sports Med 1970; 10: 185-92. 26 Office of Populations Censuses and Surveys. Classification of occupations and coding index. London: HM Stationery Office, 1980. 27 Endicott J, Spitzer RL. A diagnostic interview: the schedule for affective disorders and schizophrenia. Arch Gen Psychiatry 1978; 35: 837-44. 28 Spitzer RL, Endicott J, Robins E. Research diagnostic criteria: rationale and reliability. Arch Gen Psychiatry 1978; 35: 773-82. 29 Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983; 67: 361-70. 30 Brown GW, Harris TO. Life events and illness. London: Unwin Hyman Ltd, 1989. 31 Eysenck HJ, Eysenck SBG. Manual of the Eysenck personality questionnaire. London: Hodder and Stoughton, 1975. 32 Kendell RE, Discipio WJ. Eysenck personality inventory scores of patients with depressive illnesses. Br J Psychiatry 1968; 114: 767-70. 33 White PD, Nias DKB. A comparison of self-report and relative ratings of personality. Personality Individual Diff 1994; 16: 801-03. 34 Lee CL, Davidsohn I, Panczysyn O. Horse agglutinins in infectious mononucleosis II: the spot test. Am J Clin Pathol 1968; 49: 12-18. 35 Svedmyr E, Jondal M. Cytotoxic effector cells specific for B cell lines transformed by Epstein-Barr virus are present in patients with infectious mononucleosis. Proc Natl Acad Sci USA 1975; 72: 1622-26. 36 Sharpe MC, Archard LE, Banatvala J, et al. A report: chronic fatigue syndrome . J R Soc Med 1991; 84: 118-21. 37 Fukuda K, Straus SE, Hickie I, et al. The chronic fatigue syndrome: a comprehensive approach to its definition and study. Ann Intern Med 1994; 121: 953-59. 38 Fleisher GR, Collins M, Fager S. Limitations of available tests for diagnosis of infectious mononucleosis. J Clin Microbiol 1983; 17: 619-24. 39 Callan MFC, Tan L, Ogg GS, et al. Direct visualisation of antigen-specific CD8+ T cells during the primary immune response to Epstein-Barr virus in vivo. J Exp Med 1998; 187: 1395-402. 40 Dantzer R, Bluthe RM, Gheusi G et al. Molecular basis of sickness behaviour. Ann N Y Acad Sci 1998; 856: 132-38. 41 Ray C, Jefferies S, Weir WRC. Life events and the course of chronic fatigue syndrome. Br J Med Psychol 1995; 68: 323-31. 42 Perkins DO, Stern RA, Golden RN, Murphy C, Naftolowitz D, Evans DL. Mood disorders in HIV infection: prevalence and risk factors in a nonepicenter of the AIDS epidemic. Am J Psychiatry 1994; 151: 233-36. 43 Van Os J, Park SBG, Jones PB. Neuroticism, life events and mental health: evidence for person-environment correlation. Br J Psychiatry 2001; 178 :(suppl 40) S72-77. -------- (c) 2001 Elsevier Science Direct