Nonpolio enteroviruses (NPEVs) are a common cause of febrile illness, herpangina (HA), and hand, foot and mouth disease (HFMD) in children worldwide. HA is characterized by vesicular exanthems of the oropharynx and soft palate; in HFMD, lesions also occur on the extremities. The diseases are largely self-limited, but severe neurologic and cardiopulmonary complications occasionally occur and during the last 10–20 years have been associated with outbreaks of enterovirus A71 (EV-A71) and coxsackievirus A16 (CV-A16) in East and Southeast Asia.1 The risk of severe complications is small in infected individuals, but in populations with high prevalence of NPEV, they remain a main threat to infants and young children. With the greatest NPEV burden in this region, China reported over 7 million children with HFMD (1.1% severe cases) during 2008–2012, of which 45% were associated with EV-A71, followed by CV-A16 and other NPEVs.2
In this East and Southeast Asia region, an annual outbreak of NPEV starts when the temperature rises in spring (depending on latitudes), and the prevalent subtypes vary each year with complex patterns.3 NPEV surveillance strategies in this region typically involve sampling from children with HFMD and HA for NPEV testing. In Japan, South Korea and Taiwan, sentinel surveillance has been in place since 1990s and was more inclusive, including patients of HFMD, HA and some other illnesses.4–7 However, because low-precision reports generally show that EV-A71 and CV-A16 are more frequently detected in HFMD than HA,8,9 subsequent countries (including China) that introduced surveillance in the 2000s only sample viruses from patients with HFMD.2,10,11
There has been lack of representative reports of the epidemiology and virus characteristics of HA (see Herpangina Outbreak Reports, Supplemental Digital Content 1, http://links.lww.com/INF/D482). Compared with Japan and South Korea, Taiwan is the warmest country in which HA is routinely monitored and so its seasonal epidemiology may be relevant to the most affected region in Asia. Here, we characterize the epidemiology of HA in the Taiwan NPEV surveillance, and compare it with HFMD and other relevant viral syndromes.
MATERIALS AND METHODS
Study Design and Data Collection
The National Health Insurance (NHI) is a universal, single-payer healthcare system that covers more than 99% of population in Taiwan.12 The NHI databases, accessible through the Health and Welfare Data Science Center of the Ministry of Health and Welfare, recorded diagnoses in all outpatient and emergency care visits, and hospital admissions since the late 1990s. Between 2000 and 2015, all diagnoses were coded by physicians according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).
The Taiwan National Virologic Surveillance (NVS) is a routine laboratory surveillance network that tracks the dynamic pattern of circulating respiratory viruses and enteroviruses across the country in a real-time manner. The network harmonizes diagnostic methods for detection and characterization of emerging NPEVs, and enables real-time alerts at the regional as well as national levels.6 Since late 1990s, 2–5 respiratory samples (throat or nasopharyngeal swabs) were collected each week by each sentinel physician from outpatients and inpatients with influenza-like illness and NPEV-specific syndromes (including HFMD and HA) in the community, preferably within 3 days of disease onset.1,13 Each year, approximately 1000–3000 clinical samples collected through NVS were coordinated by Taiwan Centers for Disease Control (TWCDC), and sent to dozens of designated laboratories across the country for standardized viral identification. All viruses were isolated using rhabdomyosarcoma, MRC-5, Vero, A549, LLC-MK2 or HEp-2 cell lines; the isolates were then identified by immunofluorescent assays using 24 antibodies against polioviruses types 1–3 and NPEVs.14 Immunofluorescent assay nontypable isolates, such as enterovirus D68, parechovirus or rhinoviruses, were sent to the TWCDC central laboratory for genotyping,14 but the genotyping information has not been processed for research use, and therefore was not included in this report.
We included personal identification number, date of sampling and date of symptom onset from respiratory samples collected from patients under 20 years of age in the 2002–2015 NVS databases, linked testing results to the NHI and obtained information of all healthcare visits (including ICD-9-CM diagnoses) 0–6 days before the sampling date, as respiratory shedding of enteroviruses usually is limited to a week or less. The linked samples were classified according to ICD-9-CM diagnoses recorded during these visits into six syndromes: HFMD (code 074.3), HA (code 074.0), other enterovirus complications (codes 074, 079.1 or 079.2), other viral exanthems (VE; codes 050–059), respiratory infections including pneumonia and influenza (P&I; codes 480–488) and minor acute respiratory illnesses (ARIs; codes 460–478).1,15 The classifications were hierarchical and mutually exclusive in the exact order listed above. For example, samples recorded with both HFMD and HA were classified as HFMD-related samples, and samples recorded with HA, P&I and ARI were classified as HA-related samples. The number of other enterovirus complications was small (less than 500) and was then excluded from analyses after this hierarchical classification.
The study protocol was reviewed and approved by the Institutional Review Board at National Taiwan University Hospital (201705028RINA), Chang Gung Memorial Hospital (105-4832C) and TWCDC (105205). The research was performed in accordance with the relevant guidelines and regulations. The Institutional Review Boards waived the requirement for obtaining individual informed consent because only anonymized datasets were used in this study.
The diagnosis-specific representation rates of the included viral samples were calculated at monthly basis by dividing the number of samples collected and recorded in the NVS with the number of diagnosis-specific events aggregated from the entire Taiwanese population recorded in the NHI. Representation rates were examined in strata by sex and age (boys and girls versus under 1, 1–2, 3–5, 6–12, >12 years), calendar year, calendar month, 6 regions (5 regions in Western Coast from north to south plus Eastern Coast) and 4 levels of urbanization prespecified for ~400 administrative regions in Taiwan. These were factors found to be associated with the numbers of HFMD and HA diagnosis recorded in the healthcare system.16,17
Among viral samples we calculated the proportion of viral isolation results (all NPEV, nonenteroviruses or negative) by diagnosis and identified nontrivially represented (>5%) NPEV subtypes. We also examined the NPEV positive rates by diagnosis, in strata by sex and age, by calendar year, by calendar month and by quintiles of monthly diagnosis incidence of HFMD and HA. A monthly diagnosis incidence was the estimated number of that diagnosis in every 1000 people in each of the 168 calendar months in 2002–2015, adjusted for sex and age and county-urbanization strata and assuming a Poisson distribution. The estimated diagnostic incidence of the 168 months was then ranked into quintiles. NPEV positive rates were adjusted for sex and age strata using a logistic model. Means, standard errors, medians and interquartile ranges of estimates were obtained and interpolated through bootstrapping with 50 resamplings. Meta-regressions (metareg, STATA 15.1, StataCorp LLC, College Station, TX) were used for post hoc test of differences in positive rates by sampling characteristics.
To estimate the monthly burden of EV-A71 and CV-A16 related to HA versus HFMD, we estimated the distribution of prevalent NPEV subtypes during a calendar year. Based on and following our findings that NPEV positive rates were dependent on the background number of HA events, we also estimated the sex- and age-standardized NPEV positive rates in HA (or in HFMD) according to the background number of HA in each month. The estimated NPEV positive rates and subtype-specific proportions were then applied to observed numbers of HA or HFMD in each of the 168 calendar year-months to calculate the burden of EV-A71 and CV-A16 infections presented with HA versus HFMD.
During 2002–2015, 175,608 samples with a valid personal identifier recorded were collected in the Taiwan NVS from patients under age 20 years, among whom 95% were respiratory samples and 61% were collected from outpatients, resulting in 104,779 outpatient respiratory samples. In 98,110 (94%) of the samples, at least one outpatient visit could be identified 0–6 days before the sampling date after linkage to their healthcare records in the NHI databases. These 98,110 samples were classified by ICD-9-CM diagnoses into 6976 HFMD-related, 15,389 HA-related, 465 other enterovirus-related, 2217 VE-related, 19,350 P&I-related, 45,995 ARI-related and 7718 other samples. Sample representation rates were generally consistent across sex and age, time, region and level of urbanization, except they tended to be disproportionally high in regions, seasons or populations associated with lower follow-up child-months (see Fig. S1, Supplemental Digital Content 2, http://links.lww.com/INF/D483).
The overall positive rates of NPEV were not substantially different in samples related to HFMD (48%) and HA (43%), but were substantially lower in samples related to VE (14%), P&I (4%), and ARI (11%) (Fig. 1). NPEV subtypes identified from samples related to HFMD were predominantly CV-A16 (40%) or EV-A71 (34%). By contrast, other samples were not dominated by any NPEV subtype (≤16% in all subtypes) although CV-A16 and EV-A71 was nontrivially represented in samples related to HA (8% and 5%, respectively), and CV-A16 was nontrivially represented in samples related to VE (8%) and ARI (5%).
During the same period in Taiwan, there were more than 3 times visits of HA (4.0 millions) than HFMD (1.2 millions) in patients under age 20 years. We estimated that there was an EV-A71-associated HA for every 2.2 EV-A71-associated HFMD occurred (10th–90th percentile 1.1–2.8), and a CV-A16 associated HA for every 1.8 CV-A16-associated HFMD occurred (10th–90th percentile 0.55–3.0). Compared with surveillance for HFMD only, the inclusion of HA in Taiwan’s NVS had detected an EV-A71 outbreak a few months earlier in mid-2011 (see Fig. S2, Supplemental Digital Content 3, http://links.lww.com/INF/D484).
We found NPEV positive rates varied substantially by age (highest in age 3–5 years) and by sex: the positive rates in boys were slightly higher than in girls of the same ages in general, regardless of diagnosis (Fig. 2). This variation in positive rates by sex and age was largely consistent with (but not identical to) age- and sex-specific diagnosis incidence of HFMD or HA.
Sex- and age-standardized NPEV positive rates fluctuated moderately by year (P for fluctuation <0.001 by year in all diagnoses) (Fig. 3A). The yearly fluctuations were not related to yearly diagnostic incidence of HFMD or HA (P ≥ 0.3 for HFMD- or HA-related samples and P > 0.04 for other samples). In HFMD-related samples, the yearly fluctuations of NPEV positive rates were not associated with predominant subtypes (P = 0.3 for EV-A71 versus CV-A16), but the positive rates in HA-related samples may be higher in EV-A71 prevalent years than in CV-A16 prevalent years (P = 0.01, difference by 1.16 folds).
When the data of same calendar months across all included years were aggregated, moderate but statistically significant fluctuations of NPEV positive rates were also observed in all diagnoses by calendar month (P for fluctuation < 0.001) (Fig. 3B), but were not significantly associated with HFMD or HA monthly diagnosis incidence (P > 0.05 in all fluctuations). Instead of aggregating calendar months, we further stratified all 168 months in years 2002–2015 into quintiles by monthly diagnosis incidence of HFMD or HA, and found that NPEV positive rates in HA-related, VE-related, P&I-related and ARI-related samples were all strongly dependent on diagnosis incidence of HA (Fig. 4) (P < 10–12 in all trends). For a typical boy at age 3–5 years who has HA, this would confer to a positive rate of 54% (95% confidence interval [CI]: 52%–55%) during a peak season of HA (top 20% diagnostic incidence, ~25 HA in 1000 child-months) and a positive rate of only 35% (95% CI: 32%–38%) during a trough season of HA (bottom 20% diagnostic incidence, ~3.6 HA in 1000 child-months). By contrast, the NPEV positive rate in HFMD-related samples did not depend on diagnosis incidence of HA (P for trend = 0.7) or HFMD (P for trend = 0.6) (Fig. 4). A similar pattern dependent on diagnosis incidence of HFMD was also noted despite a smaller sample size and weaker significance (P for trend = 0.01 for HA-related, 0.09 for VE-related, 0.03 for P&I-related and 0.001 for ARI-related samples).
During the period of 2002–2015, NVS tested over 150,000 samples to monitor circulating NPEV subtypes in Taiwan. The sentinel-based NVS system is by and large similar between Japan, South Korea and Taiwan, but with the unique personal identification numbers recorded across various health systems in Taiwan, we were able to map ~100,000 viral samples individually to the outpatient healthcare system, so that all primary care diagnoses in the week before sampling can be identified. By doing so, we were then able to assess the sampling rates. The ~7000 HFMD-related and ~15,000 HA-related respiratory samples represented 6.0 and 4.1 in every 1000 follow-up months, respectively, equivalent to about one in every 150 HFMD and one in every 250 HA that were known to the healthcare system were sampled. There were oversampling in low-prevalence areas, seasons and populations, but this oversampling was necessarily for meaningful statistical estimates to be made in these data-deprived strata.
This is the first report on HA that provides relevant information at a large scale. In our analysis, the positive rates in HFMD (48%) were higher than in HA (43%) and were more consistent than in HA during either prevalent or nonprevalent seasons. Unlike an early report in Japan where similar numbers of HA and HFMD cases were observed by sampling sentinels,18 the overall number of HA in Taiwan was 3 times greater than the number of HFMD. Although the presence of EV-A71 and CV-A16 subtypes in HA was considerably lower than their presence in HFMD, our estimation suggested still one concurrent HA-associated EV-A71 infection when every 2.2 HFMD-associated EV-A71 infection occurred, and one concurrent HA-associated CV-A16 infection for every 1.8 HFMD-associated CV-A16 infection occurred. This is equivalent to additional ~30% of EV-A71 or CA16 cases if HA cases are accounted for. The comparison of high NPEV positive rates in HA (43% on average) and low NPEV positive rates in respiratory diagnoses (P&I and ARI, ≤11% on average) in all age and sex, year, month and diagnostic density strata, as shown in Figures 2–4, confirmed that these additional cases were largely attributed to HA but not to the background viral shedding.
Most countries sample NPEVs in HFMD cases, targeting EV-A71 and CV-A16 that are associated with outbreaks and severe complications. Nevertheless, our study assessed the benefit of HA monitoring and estimated that Taiwan may have had observed detectable EV-A71 activity in the community a few months earlier in 2011 because of the inclusion of HA.19 Therefore, although sampling of HFMD cases is an appropriate strategy, there might be benefits for selected NPEV subtypes when HA information was additionally obtained. Whether or not a country or a region should include HA as a part of NPEV surveillance, however, depends on local disease burden and severity, circulating virus subtypes and costs.
One major limitation is that we mainly estimated the burden of EV-A71- and CV-A16-associated HA and HFMD cases in Taiwan but not the cost–benefit of HA surveillance. Any additional cost and benefit related to HA surveillance should be assessed against comparator sampling policies and local epidemiology. Other limitations include not having a sufficient size to take into account the variation in the distribution of NPEV subtypes within a year, not being able to assess sampling bias related to sentinel characteristics, not being able to include nonrespiratory samples, not being able to completely distinguish active diseases from chronic shedding of viruses and not being able to identify emerging subtypes (such as coxsackievirus A6)20,21 or subtypes unrelated to HA or HFMD (such as enterovirus D68).14,22–24 Our modest NPEV positive rates in these samples (>40%) were comparable to other surveillance-based reports (typically less than 60%),25,26 but the real NPEV infection rate in HA or HFMD is likely to be higher if sampling and testing were controlled in a more strict, less generalizable setting, or if more sensitive viral detection methods were adopted, such as direct detection of viral sequences. For our community-led purpose, we also did not include any samples from hospital admissions.
Community-based NPEV intelligence in Asia is vibrant, but generally is dominated either by HFMD reports or by reports of sporadic subtype-specific HA outbreaks (see Herpangina Outbreak Reports, Supplemental Digital Content 1, http://links.lww.com/INF/D482). Given all the viral samples and the detectable year-round activity of NPEV in Taiwan across 14 years in the 21st century, this is the first representative and systematic report for viral characteristics of HA and its comparison to HFMD. Our report highlighted that EV-A71 or CV-A16 infections in HA are not exceptional cases, and the sheer amount of these HA cases cannot be considered purely as a result of diagnostic or recording errors. Our report of concurrent HA and HFMD cases also suggested that the presentation of NPEV, either with HA or with HFMD, may not be entirely explained by viral subtypes. In all, our finding confirms that HFMD monitoring is reliable, but there can be measurable additional benefit when HA is also monitored.
The authors are grateful to the Health and Welfare Data Science Center, Ministry of Health and Welfare, Taiwan for their support in data coordination.
1. Liu DP, Wang TA, Huang WT, et al. Disease burden of enterovirus infection in Taiwan: implications for vaccination policy. Vaccine. 2016;34:974–980.
2. Xing W, Liao Q, Viboud C, et al. Hand
, foot, and mouth disease in China, 2008-12: an epidemiological study. Lancet Infect Dis. 2014;14:308–318.
3. Pons-Salort M, Grassly NC. Serotype-specific immunity explains the incidence of diseases caused by human enteroviruses. Science. 2018;361:800–803.
4. Hyeon JY, Hwang S, Kim H, et al. Accuracy of diagnostic methods and surveillance
sensitivity for human enterovirus, South Korea, 1999-2011. Emerg Infect Dis. 2013;19:1268–1275.
5. National Institute of Infectious Diseases. Pathogen surveillance
system in Japan and Infectious Agents Surveillance
Report (IASR). IASR. 2010;31:69–70. Available at: https://idsc.niid.go.jp/iasr/31/361/tpc361.html
. Accessed February 20, 2019.
6. Tseng FC, Huang HC, Chi CY, et al; CDC-Taiwan Virology Reference Laboratories and Sentinel Physician Network. Epidemiological survey of enterovirus infections occurring in Taiwan between 2000 and 2005: analysis of sentinel physician surveillance
data. J Med Virol. 2007;79:1850–1860.
7. Ho M, Chen ER, Hsu KH, et al. An epidemic of enterovirus 71 infection in Taiwan. Taiwan Enterovirus Epidemic Working Group. N Engl J Med. 1999;341:929–935.
8. Yamashita T, Ito M, Taniguchi A, et al. Prevalence of coxsackievirus A5, A6, and A10 in patients with herpangina
in Aichi Prefecture, 2005. Jpn J Infect Dis. 2005;58:390–391.
9. Park K, Lee B, Baek K, et al. Enteroviruses isolated from herpangina
-foot-and-mouth disease in Korean children. Virol J. 2012;9:205.
10. Podin Y, Gias EL, Ong F, et al. Sentinel surveillance
for human enterovirus 71 in Sarawak, Malaysia: lessons from the first 7 years. BMC Public Health. 2006;6:180.
12. Hsing AW, Ioannidis JP. Nationwide population science: lessons from the Taiwan national health insurance research database. JAMA Intern Med. 2015;175:1527–1529.
13. Lin TY, Twu SJ, Ho MS, et al. Enterovirus 71 outbreaks, Taiwan: occurrence and recognition. Emerg Infect Dis. 2003;9:291–293.
14. Huang YP, Lin TL, Lin TH, et al. Molecular and epidemiological study of enterovirus D68 in Taiwan. J Microbiol Immunol Infect. 2017;50:411–417.
15. Yang TO, Huang WT, Chen MH, et al. Diagnostic uncertainty of herpangina
-foot-and-mouth disease and its impact on national enterovirus syndromic monitoring. Epidemiol Infect. 2016;144:1512–1519.
16. Yang TO, Huang WT, Chen MH, et al. Sex differences in common childhood infections in Taiwan. Int J Infect Dis. 2018;75:115–117.
17. Yang TO, Huang WT, Chen MH, et al. Seasonal synchrony in incidences of common infectious diagnoses in early childhood among neighbouring regions. Int J Infect Dis. 2014;28:214–216.
18. Eshima N, Tokumaru O, Hara S, et al. Age-specific sex-related differences in infections: a statistical analysis of national surveillance
data in Japan. PLoS One. 2012;7:e42261.
19. Tsai YC, Lin CH, Huang WT, et al. The epidemic of enterovirus 71 in Taiwan, 2011–2012. Taiwan Epidemiol Bull. 2012;28:79–81.
20. Ogi M, Yano Y, Chikahira M, et al. Characterization of genome sequences and clinical features of coxsackievirus A6 strains collected in Hyogo, Japan in 1999-2013. J Med Virol. 2017;89:1395–1403.
21. Lin TL, Li YS, Huang CW, et al. Rapid and highly sensitive coxsackievirus a indirect immunofluorescence assay typing kit for enterovirus serotyping. J Clin Microbiol. 2008;46:785–788.
22. Harvala H, Jasir A, Penttinen P, et al. Surveillance
and laboratory detection for non-polio enteroviruses in the European Union/European Economic Area, 2016. Euro Surveill. 2017;22:1600807.
23. Abedi GR, Watson JT, Nix WA, et al. Enterovirus and parechovirus surveillance
—United States, 2014-2016. MMWR Morb Mortal Wkly Rep. 2018;67:515–518.
24. Hawkes MT, Vaudry W. Nonpolio enterovirus infection in the neonate and young infant. Paediatr Child Health. 2005;10:383–388.
25. Zhuang ZC, Kou ZQ, Bai YJ, et al. Epidemiological research on hand
, foot, and mouth disease in Mainland China. Viruses. 2015;7:6400–6411.
26. Kim HJ, Hyeon JY, Hwang S, et al. Epidemiology and virologic investigation of human enterovirus 71 infection in the Republic of Korea from 2007 to 2012: a nationwide cross-sectional study. BMC Infect Dis. 2016;16:425.