One co-author (JMS) is currently employed by Sanofi Pasteur. All other co-authors have declared that no competing interests exist. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.
Conceived and designed the experiments: HCB MC ST P. Sawatwong SJO JMS P. Srisaengchai P. Sawanpanyalert SAM PA. Performed the experiments: MC ST PP SN P. Sawatwong DD SC. Analyzed the data: HCB DD SJO P. Srisaengchai LFP SAM. Contributed reagents/materials/analysis tools: HCB MC SJO P. Srisaengchai LFP P. Sawanpanyalert PA. Wrote the paper: HCB MC ST SJO JMS SAM. Expert guidance and critical review: P. Sawanpanyalert PA.
Data on the burden of the 2009 influenza pandemic in Asia are limited. Influenza A(H1N1)pdm09 was first reported in Thailand in May 2009. We assessed incidence and epidemiology of influenza-associated hospitalizations during 2009–2010.
We conducted active, population-based surveillance for hospitalized cases of acute lower respiratory infection (ALRI) in all 20 hospitals in two rural provinces. ALRI patients were sampled 1∶2 for participation in an etiology study in which nasopharyngeal swabs were collected for influenza virus testing by PCR.
Of 7,207 patients tested, 902 (12.5%) were influenza-positive, including 190 (7.8%) of 2,436 children aged <5 years; 86% were influenza A virus (46% A(H1N1)pdm09, 30% H3N2, 6.5% H1N1, 3.5% not subtyped) and 13% were influenza B virus. Cases of influenza A(H1N1)pdm09 first peaked in August 2009 when 17% of tested patients were positive. Subsequent peaks during 2009 and 2010 represented a mix of influenza A(H1N1)pdm09, H3N2, and influenza B viruses. The estimated annual incidence of hospitalized influenza cases was 136 per 100,000, highest in ages <5 years (477 per 100,000) and >75 years (407 per 100,000). The incidence of influenza A(H1N1)pdm09 was 62 per 100,000 (214 per 100,000 in children <5 years). Eleven influenza-infected patients required mechanical ventilation, and four patients died, all adults with influenza A(H1N1)pdm09 (1) or H3N2 (3).
Influenza-associated hospitalization rates in Thailand during 2009–10 were substantial and exceeded rates described in western countries. Influenza A(H1N1)pdm09 predominated, but H3N2 also caused notable morbidity. Expanded influenza vaccination coverage could have considerable public health impact, especially in young children.
Increasing evidence supports influenza viruses as an important cause of disease in tropical and subtropical regions, particularly in young children and older adults
The first cases of influenza A(H1N1)pdm09 infection were detected in the United States in April 2009. The ensuing pandemic resulted in heightened influenza surveillance worldwide. However, few surveillance systems included established catchment populations, meaning that population-based estimates of influenza A(H1N1)pdm09 incidence are limited, especially in tropical and resource limited settings. Robust incidence data are needed to estimate the burden of the 2009 pandemic and are important to inform policy discussions about prevention and control strategies for influenza in general.
The first cases of influenza A(H1N1)pdm09 were reported in Thailand in May 2009. Although influenza A(H1N1)pdm09 quickly became the predominant circulating influenza subtype, circulation of seasonal influenza viruses continued. We used ongoing active, population-based surveillance in two Thailand provinces
We have conducted active, population-based surveillance for community-acquired pneumonia in all twenty hospitals in two Thailand provinces since 2002–2003
A case of acute lower respiratory infection (ALRI) was defined as evidence of both active infection (at least one of: reported fever, reported chills, measured temperature >38.2 or <35°C, or an abnormal white blood cell count or differential) and lower respiratory tract disease (at least one of: abnormal breath sounds, documented tachypnea, cough, sputum production, or dyspnea) in a hospitalized patient. Every other ALRI case-patient was systematically sampled for possible participation in an etiology study. Sampling was done by ward (e.g., male, female, pediatric) and chronologically based on admission time. Chest radiographs, if performed, were digitized and interpreted by a panel of radiologists using standard criteria as previously described
All adult participants and guardians of children aged <18 years provided written consent for study enrollment. A CDC Institutional Review Board and the Ethical Review Committee of the Thailand Ministry of Public Health approved this study.
Nasopharyngeal swab specimens were collected from enrolled patients using a polyester swab (Puritan®, Guilford, ME, USA). In Nakhon Phanom, nasopharyngeal specimens were collected using flocked swabs (FLOQSwabs™, Copan, Murrieta, CA, USA) starting in July 2010. Specimens were collected at the time of enrollment and were stored at 4–8°C for <24 hours before being frozen at −70°C. Specimens were transported weekly on dry ice to the Thailand National Institute of Health. All specimens were tested for influenza A and B viruses by real-time reverse transcription polymerase chain reaction (rRT-PCR)
As part of the response to the 2009 influenza pandemic, additional information on underlying conditions, influenza vaccination status, and antiviral treatment was collected using a standard questionnaire on a non-systematic sample of study patients from October 2009 through August 2010.
The study period included January 2009 through December 2010. Clinical and demographic characteristics were compared among patients infected with influenza A virus subtypes and those infected with influenza B virus. Dichotomous variables were compared by chi-square or Fisher’s exact test and continuous variables by t-test or Kruskal-Wallis. The crude age-specific incidence of influenza-associated ALRI hospitalizations was calculated for the 2-year study period by dividing the number of rRT-PCR-confirmed influenza cases by the combined population of the two provinces
During the study period, 28,076 patients were hospitalized with ALRI (
Of the 7,207 enrolled patients, 902 (12.5%) tested positive for an influenza virus, including 190 (7.8%) of 2,436 children aged <5 years. Compared to influenza-negative patients, influenza-positive patients were less likely to be age <5 years (21% vs. 36%) and less likely to be ≥65 years (14% vs. 23%) (p<0.01). After adjusting for age, influenza-positive patients were more likely than influenza-negative patients to have documented fever >38.0°C (56% vs. 37%, p<0.01), but less likely to die (0.44% vs. 1.7%, p = 0.01), require intubation (1.2% vs. 3.4%, p<0.01), need oxygen therapy (21% vs. 34%, p<0.01), or to have chest radiographs consistent with pneumonia (37% vs. 58%, p<0.01). Nasopharyngeal specimens were collected from influenza-positive patients a mean of 0.86 days after admission compared to 1.04 days for influenza-negative patients (p<0.01).
Among 902 influenza cases, 779 (86%) were influenza A, 120 (13%) were influenza B, and three were positive for both influenza A (one each of influenza A(H1N1)pdm09, H1N1, and H3N2) and B. Among influenza A cases, 417 (54%) were influenza A(H1N1)pdm09, 270 (35%) were H3N2, one was A(H1N1)pdm09 and H3N2, 59 (7.6%) were H1N1, and 32 (4.1%) were not subtyped.
Patients infected with influenza A(H1N1)pdm09 virus were similar in age to those with H1N1 and influenza B but younger than those with H3N2 (
A(H1N1)pdm09 (N = 417) | A(H1N1) (N = 59) | A(H3N2) (N = 270) | A, unsubtyped (N = 32) | B (N = 120) | p-value | |
Age, years, median (range) | 16 (0–92) | 14 (0–82) | 45 (0–95) | 15.5 (0–80) | 15 (0–86) | <0.01 |
Age group, years | ||||||
Age <5 | 86 (20.6) |
8 (13.6) | 60 (22.2) | 5 (15.6) | 31 (25.8) | <0.01 |
5–17 | 136 (32.6) | 24 (40.7) | 30 (11.1) | 14 (43.8) | 39 (32.5) | |
18–49 | 117 (28.1) | 11 (18.6) | 56 (20.7) | 4 (12.5) | 25 (20.8) | |
50–64 | 45 (10.8) | 8 (13.6) | 59 (21.8) | 5 (15.6) | 7 (5.8) | |
≥65 | 33 (7.9) | 8 (13.6) | 65 (24.1) | 4 (12.5) | 18 (15.0) | |
Temperature >38°C | 243 (58.4) | 36 (61.0) | 153 (56.7) | 16 (50.0) | 57 (47.5) | 0.17 |
Fever by report | 409 (98.1) | 55 (93.2) | 266 (98.5) | 31 (96.9) | 118 (98.3) | 0.07 |
Days in hospital, median | 3 | 3 | 3 | 3 | 3 | 0.71 |
Death | 1 (0.2) | 0 | 3 (1.1) | 0 | 0 | 0.29 |
Intubation | 7 (1.7) | 1 (1.7) | 2 (0.8%) | 0 | 1 (1.3) | 0.70 |
Oxygen therapy | 79 (19.0) | 14 (23.7) | 71 (26.4) | 5 (15.6) | 20 (16.7) | 0.07 |
Co-infections | ||||||
RSV | 30 (7.2) |
0 | 29 (10.9) | 1 (3.1) | 8 (6.7) | 0.03 |
Adenovirus | 2 (0.48) |
1 (1.7) | 0 | 0 | 0 | 0.20 |
4 (2.1) of 188 |
0 of 27 | 8 (4.7) of 170 |
0 of 13 | 1 (2.1)of 47 | 0.38 |
|
Blood culture positive of those with culture | 1 of 121 |
0 of 23 | 0 of 92 | 0 of 9 | 0 of 35 | |
CXR with pneumonia (of those with CXR read) | 68 (33.7) of 202 | 9 (31.0) of 29 | 48 (32.2) of 149 | 5 (33.3) of 15 | 20 (35.1) 57 | 0.97 |
Pearson chi-square test for homogeneity, excluding influenza A cases without subtype.
Kruskal-Wallis test, excluding influenza A cases without subtype.
Among those tested for
One A(H1N1)pdm09-infected patient was also positive for adenovirus and RSV.
One patient with
One patient positive for both influenza A(H1N1)pdm09 and A(H3N2) tested positive for
CXR, chest x-ray.
Co-infection with RSV occurred in 30 (7.2%) patients with influenza A(H1N1)pdm09, 29 (11%) with H3N2, and 8 (6.7%) with influenza B virus (7.6% of all influenza-infected patients were RSV-positive vs. 9.3% of influenza-negative patients). Of 68 influenza-RSV co-infections, 56% occurred among children aged <5 years with little variation by influenza subtype. No patients co-infected with influenza and RSV died or required mechanical ventilation, and mean hospital stay did not differ significantly between influenza-RSV-co-infected and influenza-only-infected patients (3.9 vs. 3.6 days, p = 0.35). Among 445 influenza-infected patients aged >17 years tested for pneumococcus by urine antigen assay, 12 (2.7%) were positive compared to 140 (4.4%) of influenza-negative patients. One (25%) of four fatalities in influenza-infected patients was co-infected with pneumococcus compared to 11 (2.5%) of 440 non-fatalities (Fisher’s exact p = 0.10). Patients co-infected with pneumococcus were more likely to require oxygen therapy than those who tested negative for pneumococcus (58% vs. 28%, p = 0.04) but were not more likely to need mechanical ventilation; co-infected patients had longer hospital stays (mean 5.0 vs. 3.8 days, p = 0.23), though not significantly so. Three (0.34%) influenza-infected patients tested positive for adenovirus compared to 145 (2.3%) influenza-negative patients. Blood cultures were performed in 282 (31%) of influenza-infected patients, and only one was positive for a potential pathogen (
Influenza peaks occurred twice in each study year, with smaller peaks occurring in January – March and larger peaks occurring during July – September (
The crude incidence rate of influenza-associated ALRI hospitalizations during 2009–2010 was 35 per 100,000 person-years; after adjusting for sampling and enrollment, the overall rate was 136 per 100,000 person-years (95% CI = 127–144 per 100,000). Adjusted rates were highest in children aged <5 years at 477 per 100,000 person-years (95% CI = 414–547 per 100,000) and adults aged ≥75 years at 407 per 100,000 person-years (95% CI = 312–519 per 100,000) (
To account for the fact that non-enrolled ALRI patients were more likely to have severe disease (mechanical ventilation or death) than enrolled patients and that a lower proportion of severely ill patients tested positive for influenza viruses, we re-calculated the adjusted incidence using influenza-positive percentages specific to severe vs. non-severe disease. The adjusted incidence accounting for severity was 135 per 100,000 person-years, almost identical to the incidence not accounting for severity (136 per 100,000 person-years).
From October 2009 through August 2010, supplemental data were collected for 475 (14%) study patients, including 31% (82/261) of patients who tested positive for an influenza virus compared to 13% (393/3126) of those who tested negative (p<0.01). Among these 475 patients, 443 (93%) received antiviral treatment (all oseltamivir); 50% received treatment within 2 days of symptom onset and 97% within 2 days of admission. Influenza-negative patients were as likely to receive therapy as those who were influenza-positive (92% vs. 98%, p = 0.09). Influenza vaccination was documented for 32 (7.2%) patients, but only one patient was vaccinated after the monovalent influenza A(H1N1)pdm09 vaccine was available. Co-morbid conditions were identified among 23% of influenza-positive versus 38% of influenza-negative patients (p = 0.01); asthma was most common (16% of influenza-positive).
We documented the substantial incidence and the seasonality of influenza in rural Thailand for a 2-year period that included the 2009 pandemic using ongoing active, population-based surveillance for hospitalized cases of influenza-associated ALRI. Influenza incidence was largely attributable to influenza A(H1N1)pdm09 virus but with important contributions from influenza A(H3N2), and from influenza B. Our results add to the limited but growing literature on the high burden of influenza in tropical settings
The incidence of hospitalized influenza cases overall was highest in children aged <5 years and adults 75 years and older (>400 per 100,000 person-years in both groups), which is typical of rates documented previously in Thailand
Few data on influenza incidence during the pandemic are available from tropical regions. A preliminary report from Guatemala found influenza A(H1N1)pdm09 hospitalization rates of 7.1–8.8 per 100,000 persons during 8 months when circulation was peaking
The co-circulation of seasonal and pandemic influenza viruses during the 2-year period allowed us to compare characteristics of patients infected with different subtypes during the same season. Patients infected with H3N2 virus were significantly older than those infected with influenza A(H1N1)pdm09, H1N1, or influenza B, although the proportion of cases aged <5 years was not significantly different. We found no difference in clinical severity in patients infected with influenza A(H1N1)pdm09 and other influenza viruses. A comparison of patients with pandemic and other influenza viruses in Guatemala found similar characteristics, although patients in the Guatemala study had more severe disease overall
Our findings likely underestimate the rate of severe influenza cases and deaths by 2–3 times, due to challenges enrolling patients with severe illness. ALRI surveillance in these provinces identified 32 deaths among 9,778 children during 2009–10, but only six were enrolled in our study and tested for influenza viruses. An early report of the first 272 hospitalized influenza A(H1N1)pdm09 cases in the U.S. found that 25% required intensive care, 15% needed mechanical ventilation, and 7% died
The low case fatality proportion in our sites should be considered in light of the availability and use of oseltamivir in Thailand, where national guidelines during the pandemic recommended treatment of influenza-infected patients with complicated illness (e.g., pneumonia, altered level of consciousness), failure to improve within 48 hours, and high-risk conditions (e.g., pregnancy, chronic illnesses). Data on a subset of our patients showed that >90% of those hospitalized with ALRI received oseltamivir, 50% within two days of symptom onset. Oseltamivir access was likely improved by domestic production of a generic formulation and by prepositioning the drug at government hospitals. Further, hospital access in Thailand is very good, allowing patients to be treated before illness progression. Our lower case fatality proportion may also have been related to the low prevalence of co-morbid conditions (23%) compared to other studies
Similar to trends seen in the U.S. and other parts of the world, Thailand had a rapid disappearance of the H1N1 virus and relatively lower infection rates from influenza A(H1N1)pdm09 virus in older adults compared to previous influenza seasons. However, in contrast to the U.S. and most temperate regions where influenza A(H1N1)pdm09 accounted for >99% of influenza infections
Few studies have systematically assessed the prevalence of other pathogens among influenza-infected patients by means other than blood culture
Strengths of our analysis include systematic sampling from active surveillance for ALRI, lack of reliance on clinical reporting or testing, and a known catchment area. However, the findings must be interpreted in light of several limitations. We lacked data on some of the known or potential risk factors for severe influenza infection, including pregnancy and obesity. Our incidence estimates relied on statistical adjustments that assumed enrolled and non-enrolled patients had identical prevalences of influenza positivity. Patients admitted on nights and weekends were under-represented in our sample and may have differed from patients admitted on weekdays in terms of demographics, disease severity, and influenza prevalence. Finally, data on co-infections with bacteria other than pneumococcus relied on blood culture as the single diagnostic modality, and <30% of patients had cultures, 25% of whom were pre-treated with antibiotics.
Using active, population-based surveillance, we found a high incidence of hospitalized influenza ALRI infections in Thailand, exceeding estimates from most other nations and confirming influenza’s substantial contribution to severe respiratory illness, including pneumonia, in Thailand and likely in Southeast Asia. The high disease burden in young children is particularly noteworthy and underscores the role that influenza vaccination should play in strategies to prevent childhood pneumonia globally
We thank Peera Areerat and Asadang Ruayajin of the Provincial Health Offices in Nakhon Phanom and Sa Kaeo, respectively, for their collaboration and advocacy for this work. We recognize Sununta Henchaichon, Charatdao Bunthi, and Jens Levy of the U.S. Centers for Disease Control and Prevention, International Emerging Infections Program, Nonthaburi, Thailand, for their support of this research. Alexander Klimov and Stephen Lindstrom of the US CDC, Atlanta, Georgia are also acknowledged for their contributions to the development of laboratory capacity that supported this study. We thank Ladarat Patinawin from the Bureau of Epidemiology at the Thailand Ministry of Public Health for assistance with national data on influenza in Thailand.