We received laboratory notifications of 855 potential participants with IgM positive results over a five year study period. We were able to contact 430 of these through their family doctor, and 253 (59%) of them agreed to detailed longitudinal assessments. The demographic and illness characteristics of these 253 participants (summarised in ) were not significantly different from an additional 177 who agreed to be followed by self report only—age 34.3 versus 37.0 years, difference = -2.7 years (95% confidence interval -6.08 to 0.67) years; sex (per cent male) 57% versus 56%, difference = 2% (-10% to 13%). We found a non-significant trend towards higher baseline symptom scores and worsened disability parameters in the self report cohort compared with the main cohort. These groups of patients did not differ from those who declined enrolment (data not shown).
Demographic characteristics of the cohort (n=253). Values are numbers (percentages) unless stated otherwise
The demographic features of the main cohort were consistent with the expected patterns of exposure to these pathogens; Epstein-Barr virus infection was most common in adolescents and young adults, Q fever was most common in men (largely because of the nature of the occupational exposure, such as meat working or shearing), and Ross River virus was most common in participants with outdoor activities that increase the likelihood of mosquito bites.23-25
In all three infection groups, approximately 25% of the original serological diagnoses were not confirmed by our more stringent criteria applied longitudinally. This is consistent with the recognised limitations of diagnoses made on the basis of detection of IgM antibodies in a single serum sample.
Factor analysis of symptoms
We derived six symptom domains. Two factors seemed to capture the physical and mental distress of acute illness, and we labelled them “acute sickness” (including items such as “headaches” and “fevers”) and “irritability” (including “feeling irritable or cranky” and “rapidly changing moods”). We also identified four other factors reminiscent of classic descriptions of post-infective or chronic fatigue states. These included: a “fatigue” factor (with items such as “prolonged tiredness after activity” and “feeling tired after rest or relaxation”), a “musculoskeletal pain” factor (with items such as “pains in your arms or legs” and “joint pain”), a “mood disturbance” factor (with items such as “feeling nervous or tense” and “feeling unhappy or depressed”), and a “neurocognitive disturbance” factor (featuring “poor memory” and “poor concentration”). Scores on the fatigue factor showed the strongest and most consistent correlations with functional impairment: “days out of role in the past month” (baseline r = 0.22, P < 0.01; three months r = 0.37, P < 0.01; six months r = 0.25, P < 0.01).
Incidence of post-infective fatigue syndrome
The case rate for provisional post-infective fatigue syndrome was 35% (87/250) at six weeks, 27% (67/250) at three months, 12% (29/250) at six months, and 9% (22/250) at 12 months. No difference in these case rates existed between the initial infective agents ().
Fig 1 Survival curves of post-infective fatigue syndrome by infective agent after onset of acute infection. Kaplan-Meier analyses of proportion of participants within each infective subcohort who remained as cases. Test of equality across groups: χ (more ...)
The medical, psychiatric, and laboratory assessments of the 29 provisional cases of post-infective fatigue syndrome at six months led to exclusion of one participant on medical grounds and none on psychiatric grounds. The 28 cases of chronic fatigue syndrome, termed here confirmed post-infective fatigue syndrome, included 14 men and 14 women with a mean age of 37 (range 17-63) years, including five participants with confirmed Epstein-Barr virus infection, three with Q fever, 13 with Ross River virus, and eight with unconfirmed infection. The 28 cases did not differ in age or sex when compared with either all participants with serological confirmation—age 36.0 versus 32.2 years, difference = 3.8 (-3.0 to 10.6) years; sex (per cent male) 55% versus 58%, difference = -3.4% (-19% to 28%)—or all enrolled participants—age 36.0 versus 34.1 years, mean difference = 1.9 (-8.7 to 4.9) years; sex (per cent male) 55% versus 58%, difference = -3% (-20% to 26%).
The rates of premorbid psychiatric diagnoses in the confirmed cases of post-infective fatigue syndrome and the matched (recovered) control participants, determined by formal psychiatric assessment of both groups at six months, were comparable—21% versus 17%, difference = 5% (-18% to 27%)—as were the rates of intercurrent psychiatric disorders—21% versus 10%, difference = 11% (-10% to 33%). Similarly, the rates of psychiatric disorder between cases and all remaining participants, detected by the structured interview at baseline, did not differ—premorbid psychiatric disorder 23% versus 14%, difference = 9% (-23% to 13%); intercurrent psychiatric disorder 23% versus 10%, difference = 13% (-8% to 28%). Interestingly, the case rates of provisional post-infective fatigue syndrome in the self report cohort were significantly higher at six and 12 months (35% and 32%) than in the main cohort. Higher rates of disability were also reported in the self report cohort.
Characteristics of post-infective fatigue syndrome
If the same pathophysiology underpinned all the clinical aspects of the acute infective illness and the post-infective fatigue state, we would predict that the individual symptom factors that we had derived empirically would resolve in a uniform manner across the time points assessed. In fact, we found substantial variation, particularly early in the course of the illness. In the group of 28 confirmed cases of post-infective fatigue syndrome, the median score on the acute sickness factor rapidly dropped to zero, whereas the median scores for fatigue, musculoskeletal pain, and neurocognitive disturbance remained high (). When we compared the kinetics of resolution of the symptom factors for the group as a whole, again the acute sickness and irritability factors showed the greatest initial speed of resolution. By contrast, the fatigue and neurocognitive disturbance factors showed significant reductions only late in the course of the illness (). These differences were most significant in the period between baseline and three months, when planned contrasts showed that the key construct of fatigue differed from all other factors (all P < 0.05), with the exception of neurocognitive disturbance. When we compared the gradients between three and six months, significant differences no longer existed, suggesting that the symptom domains had become more uniform and stable over time.
Fig 2 Pattern of change in individual symptom factors in participants with and without post-infective fatigue syndrome. Median (horizontal bar) and 25th/75th centiles (box extremities) of normalised factor scores for six symptom domains in confirmed cases of (more ...)
Fig 3 Differential rates of resolution of individual symptom factors after acute infection. Scores on each of six symptom factors for each participant (n=229) over 12 months in three divided time periods, calculated from factor analysis. Mean symptom scores (more ...)
Importantly, these final symptom patterns were also highly stereotyped, regardless of the original infective trigger. Planned contrasts of the patterns of resolution of the six symptom factors by infective subcohorts revealed that only musculoskeletal pain showed significant differences in prevalence and natural history in the early post-infective period (baseline to three months: Ross River virus v Epstein-Barr virus, P < 0.001; Ross River virus v not confirmed, P < 0.01; Ross River virus v Q fever, P < 0.01). The central symptom domains of post-infective fatigue syndrome did not differ between the infection groups at later time points.
Risk factors for acute sickness
Demographic characteristics did not generally predict the scores on the six symptom factors recorded at baseline (). We saw an association between older age and the fatigue score during the acute illness. Serologically confirmed Ross River virus infection was associated with the severity of the musculoskeletal pain factor, consistent with the propensity of this infection to cause arthralgia. Higher neuroticism and external locus of control scores were associated with more severe mood disturbance.
Demographic, psychological, and infective risk factors predicting symptom severity in acute illness (n=229). Values are standardised β coefficients from regression analysis
Risk factors for post-infective fatigue syndrome
The predictors of post-infective fatigue syndrome over the 12 months after acute infection were largely limited to the factor scores that reflect severity of acute illness (). Importantly, premorbid and intercurrent psychiatric disorder did not show predictive power for post-infective fatigue syndrome at any time point.
Risk factors for post-infective fatigue syndrome (n=229). Values are standardised β coefficients from regression analysis