Sentinel surveillance was conducted at ten general hospitals located in the largest cities of ten provinces in China. In each hospital, surveillance was conducted within the pediatrics ward, the respiratory medicine ward and the intensive care unit. The number of beds in these wards combined was greater than 80 in each of the ten hospitals.

All patients admitted to surveillance wards at the 10 hospitals were screened by nurses and physicians for SARI. A patient >5 years old was defined as having SARI if, upon or during admission, they presented with an acute onset of elevated temperature (axillary temperature ≥38°C) and cough or sore throat, AND tachypnea (respiratory rate ≥ 25/min) or dyspnea (difficulty breathing). A patient ≤5 years old was defined as having SARI if, upon or during admission, they presented with acute onset of cough or dyspnea, AND at least one of the following six signs or symptoms: a) tachypnea (respiratory rate >60/min for ages <2 months, respiratory rate >50/min for ages 2 to <12 months, and respiratory rate >40/min for ages 1 to ≤5 years); b) inability to drink or breastfeed; c) vomiting; d) convulsions; e) lethargy or unconsciousness; f) chest in-drawing OR stridor in a calm child [5]. Per China’s Ministry of Health national influenza surveillance guidelines implemented in October 2009, collection and analysis of data for this project were considered official public health surveillance activities and institutional review board approval was not required [6].

Upon enrollment, physicians registered the SARI case-patient’s name, gender, and date of illness onset. During the hospital admission, physicians completed a standardized case report form (CRF) that recorded demographic information, chronic medical conditions, signs and symptoms of current illness, and laboratory and radiographic data. At hospital discharge, physicians updated the form with information on clinical course during hospitalization, including treatment received, complications and outcomes.

Nurses collected nasopharyngeal (NP) swabs from SARI case-patients within 24 hours of enrollment following standardized procedures. Swabs were immediately placed in viral transport medium (VTM) and stored at 4°C at the local hospital. These specimens were transferred to the closest influenza network laboratory (provincial or prefecture Center for Disease Control and Prevention (CDC) laboratories) within 48 hours of collection. The local influenza network laboratories stored the specimens in VTM at −70°C until they conducted testing by real-time reverse transcription PCR (rRT-PCR) following the standard protocol [7,8].

Based on influenza activity, climate and geography, we divided China into two distinct regions, north and south, for the analysis and interpretation of influenza surveillance data. The Qinling Mountain – Huaihe River serves as the dividing line between north and south China [9-12]. In our sentinel surveillance system, 4 hospitals are located in north China, while 6 are located in south China [Figure 1].

Hospital infection control and local CDC staff entered epidemiologic, clinical, and laboratory data into an electronic database that was transmitted weekly to the national China CDC. We defined a patient with laboratory-confirmed influenza as any SARI case-patient with an NP swab that tested positive for influenza virus by rRT-PCR. We analyzed data that were collected from February 1, 2011- October 27, 2013 with SPSS (v13.0, SPSS, Chicago, IL, USA). Descriptive statistics included frequency analysis for categorical variables such as gender, age group, underlying chronic medical conditions, clinical characteristics and outcomes. We calculated medians and interquartile ranges for continuous variables such as age and length of clinical course. Differences in demographic and clinical characteristics and outcomes were assessed among SARI patients with and without laboratory-confirmed influenza using the chi-squared test or the Fisher’s exact test for nominal variables. We calculated age-adjusted mortality rates using standard techniques [13], with the 2010 China population census serving as the reference population [14]. We used the Wilcoxon rank sum test for ordinal variables. Differences were considered statistically significant at P < 0.05.

Ten hospitals included in the sentinel SARI surveillance system were also ILI sentinel hospitals within the national influenza surveillance system. We compared influenza activity among SARI and ILI patients using surveillance data from the outpatient departments of the ten hospitals in the SARI surveillance system from February 2011 – October 2013.

The Chinese national influenza-like illness (ILI) surveillance system currently includes 409 network laboratories and 554 sentinel hospitals. Surveillance is conducted in sentinel hospital emergency rooms and internal medicine and pediatric outpatient departments. Each surveillance department registers the number of ILI patients seen each day to calculate the proportion of all medical visits that are due to ILI [8]. NP swabs are collected from a sample of ILI patients, and are sent to the nearest network laboratory for influenza testing with rRT-PCR and additional virologic analyses.

Of the 5,268 SARI patients with completed CRFs, 794 (14.2%) were children <1 year old, 1386 (26.3%) were children 1 to <5 years old, and 1050 (19.9%) were adults ≥65 years old [Table 1]. SARI patients had a median age of 12.0 years (interquartile range [IQR], 2.0 - 58.0 years), and 3091 (59%) were male.Table 1

Characteristics of hospitalized severe acute respiratory infection (SARI) patients and laboratory-confirmed influenza patients in 10 sentinel surveillance hospitals in 10 provinces, China, February 1, 2011 to October 27, 2013

Just under 85% of SARI case-patients reported a temperature of ≥38°C after illness onset, with a median peak temperature of 39.0°C (IQR, 38.5 - 39.5°C). After fever, the most common clinical symptom or sign was cough (80%). Among SARI patients, 1,189 (23%) were clinically diagnosed with pneumonia during their hospital admission, and 2,868 (54%) had at least one abnormal finding on chest X-ray. The median duration of hospitalization was 8 days (IQR, 6–12.5), and 61 (1.2%) patients were admitted to an intensive care unit. Ninety-eight (1.9%) patients died during their hospitalization. Among SARI case-patients, 1,407 (27%) had at least one chronic medical condition, such as cardiovascular disease, chronic obstructive pulmonary disease, diabetes mellitus, chronic bronchitis, cancer/tumor, and asthma.

Of 299 SARI patients with laboratory-confirmed influenza, 101 (34%) were <5 years old, 145 (48%) were <15 years old, and 81 (27%) were ≥65 years old [Table 1]. The distribution of cases by age category was different among SARI cases with and without confirmed influenza [Figure 2]. The proportion of SARI cases with confirmed influenza was higher among those aged > = 65 years (27% vs. 19%, p = 0.001) and those 5–9 years old (12.4% vs 7.5%, p = 0.002) and lower among those <1 year old (8.7% vs. 14.9, p = 0.003) and those aged 15–49 years (12.7% vs. 18.9%, p = 0.007). Influenza types/subtypes detected included A(H1N1)pdm09 influenza (pH1N1) (100, 33%), A(H3N2) (80, 27%), A (subtype not identified) (4, 1%), and influenza B (115, 38%). The SARI patients with confirmed influenza had a significantly higher median age than those without confirmed influenza (20.0 years vs. 11.0, p = 0.003), and more often had at least one chronic medical condition (37% vs. 28%, p < 0.001; age adjusted percentage 28% vs. 25%, p < 0.001) including cardiovascular disease (22% vs. 16%, p < 0.001), chronic bronchitis (7%vs. 4%, p < 0.05), and cancer/tumor (4% vs. 3%, p < 0.05) [Table 1]. Although asthma was uncommon in both groups, it was less common among SARI patients with confirmed influenza (1% vs. 2%, p < 0.05). Compared with SARI patients without confirmed influenza, those with confirmed influenza more frequently had fever (95% vs. 89%, p < 0.001), sore throat (34% vs. 33%, p < 0.05) and difficulty breathing (38% vs. 30%, p < 0.05). SARI patients with confirmed influenza were no more likely to have clinical pneumonia than SARI patients without confirmed influenza (22.1% vs. 24.0%, p = 0.925). Mortality was higher among influenza positive SARI patients than influenza negative patients (4.0% vs. 1.8%, p = 0.013), but there was no significant difference after age adjustment (2.7% vs. 1.5%, p = 0.137). Receipt of seasonal influenza vaccination in the prior year was low among both SARI patients with and without confirmed influenza (2.7% vs. 1.1%, p = 0.041).Figure 2

Proportion of severe acute respiratory infection (SARI) patients with and without confirmed influenza by age group, China, Febuary 1, 2011 to October 27, 2013.

In the south, data from the 33 month surveillance period revealed a winter-seasonal peak in SARI patients [Figure 3, Panel B]. There was also an increase in influenza activity during the winter (December to March) and spring months (March to May). In addition, there was an increase in influenza activity during the summer months (July to August) [Figure 3, Panel B]. In contrast to the north, in the 2011–2012 and 2012–2013 influenza seasons in the south, influenza activity peaked during months when the numbers of hospitalized SARI cases were not at their highest. However, in the 2013–2014 season, influenza activity and SARI cases peaked at the same time. The predominant types/subtypes of influenza virus in the south based on isolates from SARI patients were pH1N1 during 2011 spring months (accounting for 35% of isolates collected March-May), B during 2011–2012 winter months (accounting for 71% of isolates collected December-February), A(H3N2) during 2012 spring months (accounting for 64% of isolates collected March-May), A(H3N2) and B during 2012 summer months (accounting for 63% and 37% of isolates collected July-August), and pH1N1 during the 2012–2013 winter months, 2013 spring months and 2013 summer months (accounting for 53% of isolates collected December-February; 83% of isolates collected March-May and 63% of isolates collected July-August) [Figure 4, Panel B]. Similar to the north, the predominant types/subtypes of circulating influenza virus in the south for both the winter and spring months of every influenza season were the same.

In the 4 hospitals in the north, the influenza activity among ILI patients peaked in the winter-spring each year, similar to influenza activity among SARI patients [Figure 5]. Also, as seen with SARI patients, the predominant types/subtypes of virus among ILI patients varied from season to season [Figure 4]. During the study period, the types/subtypes of influenza virus among ILI patients in the north were predominantly pH1N1 and B during 2011 spring months (accounting for 14% and 78% of isolates collected March-May), B during 2011–2012 winter months (accounting for 93% of isolates collected December-February), B and A(H3N2) during 2012 spring months (accounting for 73% and 31% of isolates collected March-May), A(H3N2) and pH1N1 during 2012–2013 winter months (accounting for 54% and 45% of isolates collected December-February), and pH1N1 during 2013 spring months (accounting for 64% of isolates collected March-May).Figure 5

Percent of influenza positive by month of illness onset among hospitalized severe acute respiratory infection (SARI) patients and influenza like illness (ILI) outpatients, China, Febuary 1, 2011 to October 27, 2013. Panel A. Four Northern Hospitals. Panel B. Six Southern Hospitals. *positive rate (%): (1) ILI flu positive rate = percent of influenza positive specimens among all tested specimens for influenza like illness patients. (2) SARI flu positive rate=percent of influenza positive specimens among all tested specimens for severe acute respiratory infection patients.