Increased risk of chronic fatigue syndrome
following herpes
zoster: a population-based study
S.-Y. Tsai & T.-Y. Yang & H.-J. Chen & C.-S. Chen &
W.-M. Lin & W.-C. Shen & C.-N. Kuo & C.-H. Kao
Introduction
Chronic fatigue is associated with several viral infections, as indirectly evidenced by the existence of viral particles in patients with persistent chronic fatigue [1–4]. Chronic fatigue
syndrome (CFS) is a specific disorder accompanied by unexplainable major symptoms, such as persistent fatigue [5].
Postviral fatigue remains a possible etiology for the unexplainable symptoms of chronic fatigue, such as herpes zoster
(HZ) infection. A previous study reported that HZ infection of the peripheral ganglia might be associated with some cases of CFS, with the most likely cause of infection being a neurotropic herpes virus, particularly varicella zoster virus (VZV)
[3, 6–8]. The symptoms of CFS might be produced by infection of the peripheral ganglia, with fatigue caused by postviral infection of the autonomic ganglia as one of the possible etiologic factors of CFS [9]. VZV is known to reactivate frequently in the peripheral ganglia of previously healthy adults and cause sudden, debilitating illness, indicating that it is a likely cause of CFS [10]. Another study showed monozygotic twins discordant among CFS patients, indicating that
HZ infection elevated the odds ratio, exhibiting an approximately 2-fold risk [11, 12]. Based on the aforementioned
assumptions and evidence, we propose that HZ infection might be associated with the development of CFS. The diagnostic criteria of CFS are common, based on the 1994 Fukuda definition, which requires a person to have
severe persistent fatigue for at least 6 months, which is necessary to rule out the well-known causes of fatigue [5]. Furthermore,
four or more symptoms in addition to fatigue
(which involves clinically evaluated, unexplained, persistent, or relapsing chronic fatigue), are present, such as unusual postexertion fatigue, impaired memory or concentration, unrefreshing sleep, headache, muscle pain, joint pain, sore throat, and tender cervical nodes [13].
However, the diagnostic tools for treating CFS are still inefficient and nonspecific. According to the definition of CFS, the first step in excluding CFS is to rule out cancer and other common chronic disorders [14]. Based on the exclusion criteria, a diagnostic workflow must be obtained to preclude common systematic disorders such as multiple sclerosis [15–17], human immunodeficiency virus (HIV) infection
[18, 19], or systemic lupus erythematosus [20]. Recent studies have documented that fatigue might be associated with
postviral infection [21, 22]. We demonstrated that the risk of CFS might also be associated with HZ infection, which we propose based on the hypothesis of association between postviral neuropathy and CFS.
We used a population-based prospective study of the National Health Insurance Research Database (NHIRD) of Taiwan to survey the hazard ratios (HRs) and cumulative incidence rates of CFS for different risk levels between patients with and without HZ infection.
Materials and methods Study design
We conducted a population-based prospective cohort study based on the original claims data of one million beneficiaries randomly sampled from the NHIRD. All personal identifications were encrypted by the National Health Research Institutes before the data were released. We also obtained the approval of the Institutional Review Board of China Medical University Hospital (CMU-REC-101-012). Previous studies have presented related information on the NHIRD [13, 14], demonstrating diagnostic accuracy and validity [15, 16]. Study population
We used the International Classification of Disease, Ninth Revision (ICD-9), for obtaining diagnostic codes and identifying
HZ infection. We identified 9,205 patients with newly diagnosed HZ infection (ICD-9-CM, code 053) and 36,820 patients without HZ infection from the registry of ambulatory and patient claims data between 2005 and 2007; the date of diagnosis of HZ infection was considered the index date. We excluded patients who had CFS (ICD-9-CM, code 780.71), multiple sclerosis (ICD-9-CM, code 340), HIV infection (ICD-9, code 042), or systemic lupus erythematosus (ICD-9, code 710.0) at baseline, or missing information on sex or age. For each patient with HZ infection, four study patients were selected randomly by frequency matching according to sex, age (per 5 years), and index year. Both cohorts were followed up until they were diagnosed with cancer-related fatigue (CRF), or until the patients were censored because of loss to follow-up, withdrawal from the national health insurance system, or it reached the end of 2010.
Comorbidity variables
The following diseases were considered potential confounders in the association between HZ infection and CFS. The baseline comorbidity history of each patient included cancer (ICD-9-CM, codes 140–208), rheumatoid arthritis (ICD-(ICD-9-CM, code 714), hyperthyroidism (ICD-9-CM, code 242), diabetes (ICD-9-CM, code 250), renal disease (ICD-9-CM, codes 582– 583.7, 585, 586, and 588), chronic hepatitis (ICD-9-CM, codes 571, 572.2, 572.3, 572.8, 573.1–573.3, 573.8, and 573.9), and depression (ICD-9-CM, codes 296.2–296.9). Statistical analysis
We compared the distributions of age, sex, and comorbidities between the case and control groups by using the Chi-square test. The incidence rates of CFS were calculated according to
the follow-up time until the end of 2010 or the date of diagnosis ofHZ infection, death, or loss to follow-up.We used
the Kaplan–Meier (K-M) estimation method to construct cumulative incidence curves of CRF for the HZ and non-HZ
cohorts, and a log-rank test to determine whether these K-M curves were statistically different. Cox proportional hazards regression analysis was performed to measure the effects of HZ infection on the risk of CFS. HRs and 95 % confidence
intervals (CIs) were calculated in the model.We performed all statistical analyses using the SAS statistical package (version 9.2 forWindows; SAS Institute Inc., Cary, NC, USA).We set the statistical significance at α=0.05 and depicted the survival curves using R statistical software (version 2.14.1 for
Windows). Results
In the claims data of 2005–2007, 9,205 patients with HZ infection met the eligibility criteria, with 36,820 patients forming the non-HZ group (Table 1). The two groups were similar in sex and age distributions, with an approximatemean age of 56 years. Cancer, hyperthyroidism, diabetes, renal disease, chronic hepatitis, and depression were more prevalent in the HZ group than in the non-HZ group at baseline.
In Table 2, the incidence rates of CFS were higher in the HZ cohort (4.56 per 1,000 person-years) than in the non-HZ cohort (3.44 per 1,000 person-years). Adjusted HRs (aHRs) showed that the HZ cohort had an elevated, 1.29-fold risk of CRF compared with the non-HZ cohort when adjusted for age, sex, and comorbidities.When stratified by sex (crude HR for CFS), women showed a significantly higher risk of CFS (1.39, 95 % CI = 1.11–1.75) in the comparison between the HZ and non-HZ cohorts. However, the aHR of female patients showed a significant, 1.36-fold risk, after adjustment for age, sex, and comorbidities. The stratified age group of 40–64 years showed an elevated crude HR of developing CFS, indicating an approximately 1.65-fold risk, whereas the aHRs showed a 1.58-fold risk, despite exhibiting a significantly higher risk of CFS. Among individuals without cancer, rheumatoid arthritis, diabetes, renal disease, chronic hepatitis, or depression, patients with HZ infection weremore likely to develop CFS than
non-HZ patients [aHR: 1.27 (95 % CI = 1.06–1.51) for those without cancer, 1.30 (95 % CI = 1.09–1.54) for those without rheumatoid arthritis, 1.30 (95 % CI = 1.09–1.54) for those without hyperthyroidism, 1.25 (95 % CI = 1.04–1.52) for those without diabetes, 1.28 (95 % CI = 1.07–1.53) for those without renal disease, 1.36 (95 % CI = 1.08–1.71) for those without chronic hepatitis, and 1.31 (95 % CI = 1.10–1.56) for
those without depression, respectively]. As shown in Table 3, in patients receiving antiviral treatment, the crude HR reduced to 0.85 (95 % CI = 0.53–1.35) with nonsignificance; even the aHRs reduced to 0.80 (95 % CI = 0.50–1.28). However, in patients with antiviral treatment, there was a reduced, 15 % risk of CFS compared with patients without antiviral treatment. The cumulative incidence rate in the HZ cohort indicated a persistent, significantly higher risk of CFS than in the
non-HZ cohort; these rates were ascending during the followup years, as shown in Fig. 1 (log-rank test, p=0.001).
Discussion
CFS was clearly described in the 1994 Fukuda definition, and the diagnostic flowchart indicated that CFS diagnosis rules out other causes of CFS [5]. Furthermore, CFS is similar to other disorders, such as fibromyalgia, but it still lacks efficient
treatment [23, 24]. For similar disorders, medication for treatment is available as a benefit of the National Health Insurance
(NHI) system of Taiwan. However, we excluded no other disorders similar to HZ in the non-HZ cohort because doing so would have led to underestimating the risk of CRF between the HZ and non-HZ cohorts. A previous study documented that CFS might be associated with herpes virus infection. However, we found that the HZ cohort that had no existing CFS before baseline had an elevated risk of CFS when comparing the claims data of the HZ and non-HZ cohorts in the
NHI system. VZVinfects a wide variety of cells in the central and peripheral nervous systems, explaining the diversity of
VZV-associated clinical disorders, such as CFS [25]. There were several limitations of this HZ-related CSF
study. First, the HZ-related CFS was heterogeneous regarding the non-CFS cohort composition: we did not restrict healthy people as the comparison group. We might have
underestimated the risk of CFS despite adjusting for CFSand HZ-related comorbidities, such as rheumatoid arthritis [26], hyperthyroidism [27], diabetes [28], renal disease [29], chronic hepatitis [30], and depression [31]. Second, a recent study showed that CFS might overlap CRF in clinical diagnosis [32, 33], even when accompanied by a similar mechanism
such as immunological response [34, 35]. However,
CRF had no ICD-9 code; hence, we could not clearly identify this symptom in the NHIRD. We attempted to adjust all cancers to stratify CRF. Among the cancer-free patients, the risk of developing CSF between the HZ and non-HZ cohorts showed statistical significance. Third, we confirmed that HZ infection is a possible risk factor for developing CFS [9, 36] when we adjusted for cancer and other comorbidities. Moreover, the results of antiviral treatment reflected that the etiology of HZ-induced CFS was unique against viral treatment [9]. However, it is difficult to clarify misclassification and
competition between disorder diagnoses in clinics as the possible reasons for the hypothesis of association between CFS
and HZ infection. We considered that HZ-related CFS might have existed between the HZ and non-HZ cohorts, and did not account for any other comorbidity. In addition, Taiwan
launched the NHI system in 1995, operated by a single buyer, the government. All insurance claims should be scrutinized by medical reimbursement specialists and peer review. The diagnoses of CRF and HZ infection were based on the ICD-9 code
determined by clinical physicians. Therefore, the diagnoses
and codes for CRF and HZ infection should be accurate and reliable. Furthermore, CFS did not have any diagnostic tool
and specific treatment for the application of insurance benefits, so we considered that the hidden bias of insurance benefits would be less.
We determined that the HZ cohort without comorbidity for each patient had a significantly higher risk of CSF than that of the non-HZ cohort; these results might indicate the existence of HZ-related CFS, which reduced the occurrence of cancer and other fatigue-associated morbidities. However, to provide evidence of HZ-induced CFS, we must still clarify the etiology of CFS and HZ infection by using a more detailed
approach that enhances the accurate diagnosis of CFS. Conclusion
The results showed that herpes zoster (HZ) infection might be associated with the risk of chronic fatigue syndrome (CFS), and even HZ infection without any
comorbidity exhibited an elevated risk of CFS. Patients with HZ infection required further prevention of fatigue
by receiving antiviral treatment. For improving the prevention of CFS in clinics, we recommend adopting a
strategy for patients with HZ infection that involves HZ vaccines and antiviral treatment [9]. Our study highlights the need for a better understanding of the association between HZ infection and the risk of CFS. These findings of this population cohort study provide pivotal evidence of postviral fatigue among patients with HZ infection. It addresses an important issue concerning public health and can be adopted in other countries.
