Bulletin of the World Health Organization

Estimated incidence of influenza-virus-associated severe pneumonia in children in El Salvador, 2008–2010

Wilfrido Clara a, Julio Armero b, David Rodriguez b, Celina de Lozano b, Luis Bonilla c, Percy Minaya d, Rafael Chacón e, Jorge Jara c, Natalia Blanco c, Marc-Alain Widdowson f, Joseph Bresee f, Xiyan Xu f, Alexander Klimov f, Eduardo Azziz-Baumgartner f & Nivaldo Linares-Perez f

a. Centers for Disease Control and Prevention, Central American Region (CDC-CAR), Embassy of the United States in San Salvador El Salvador, Final Blvd Santa Elena, Antiguo Cuscatlan, La Libertad, El Salvador.
b. Ministry of Health, San Salvador, El Salvador.
c. Universidad del Valle, Guatemala City, Guatemala.
d. Training of Epidemiology and Public Health Intervention Network, Atlanta, United States of America (USA).
e. Council of Ministers of Health of Central America and the Dominican Republic, San Salvador, El Salvador.
f. Influenza Division, Centers for Disease Control and Prevention, Atlanta, USA.

Correspondence to Wilfrido Clara (e-mail: aclara@gt.cdc.gov).

(Submitted: 24 October 2011 – Revised version received: 05 July 2012 – Accepted: 10 July 2012 – Published online: 06 August 2012.)

Bulletin of the World Health Organization 2012;90:756-763. doi: 10.2471/BLT.11.098202


Influenza is a vaccine-preventable disease that annually affects 5–10% of the population worldwide.1 This burden is well documented in upper-income, temperate countries, where influenza surveillance has been conducted for years, and is being increasingly understood in middle-income countries, where surveillance has substantially improved during recent years. In the United States of America, for example, an annual average of 228 635 hospitalizations for pneumonia and influenza occurred during 1979–2001 and 23 207 deaths were associated with influenza viruses during 1976–2007.2,3 In addition, the United States spent an average of 10.4 billion United States dollars annually on the treatment of influenza.4 Similarly, in high-income and middle-income countries such as Singapore and Thailand, influenza has substantial disease and economic burdens.5,6

Less is known about the burden of influenza in low-income countries; however, investments in epidemiology and laboratory capacity as a result of efforts to comply with the 2005 International Health Regulations and prepare for influenza pandemics have yielded data about the circulation and burden of influenza in low-income tropical countries. For example, studies now suggest that the rate of influenza-virus-associated severe acute lower respiratory tract infection is approximately 2 cases per 1000 child–years among children aged < 5 years in low-income countries.7 National influenza surveillance data from Bangladesh suggest that approximately 67 000 persons are hospitalized annually as a result of influenza virus infection and that the rate of influenza-virus-associated severe acute lower respiratory infection is 1 case per 1000 child–years among children aged < 5 years.810 Similar findings have been reported in Guatemala, Kenya, Nicaragua, the Philippines, Thailand and Viet Nam.7,1114

Although data on the influenza burden are useful to guide investments in influenza prevention and control, these data are not available in many tropical countries in Latin America. In 2004 El Salvador, the smallest and most densely populated country in Central America (293 inhabitants/km2),15 introduced influenza vaccination among children 623 months old, adults > 60 years old and persons with certain pre-existing medical conditions. In 2007 influenza sentinel-hospital surveillance was begun in El Salvador to better guide influenza prevention and control efforts.16 Although vaccine coverage among children and adults aged > 60 years reached 8595% during 2010,17,18 few data were available about the burden of influenza in El Salvador. These data may help the El Salvador Ministry of Health to better assess the value of influenza vaccination among people vulnerable to severe influenza,19 adjust vaccination policies and determine the number of influenza vaccine doses to purchase. The primary objective of this study was to estimate the population-based incidence of severe pneumonia associated with influenza virus among children aged < 5 years in El Salvador during 2008–2010 on the basis of findings from sentinel surveillance and a health-care utilization survey. We supplemented these findings with a comparison of the antigenic characteristics of influenza virus strains isolated from case patients with vaccine formulations for the northern and southern hemispheres during the study period.


Case identification

The San Juan de Dios Hospital in Santa Ana, the largest public hospital in the western region of El Salvador, was the first sentinel surveillance site for severe acute respiratory infection established in the country. For this study, a case patient with severe pneumonia was defined as a patient aged < 5 years hospitalized with cough or difficulty breathing and at least one danger sign (i.e. chest in-drawing, stridor while calm, convulsions, inability to drink, lethargy, unconsciousness or intractable vomiting).16,20 Hospital physicians identified all patients with severe pneumonia and collected demographic and clinical data. Nasal and oropharyngeal swab specimens were collected from a convenience sample of 5 or 6 consecutive patients per week. This site was supported by the Influenza Program for the Central American Region of the United States Centers for Disease Control and Prevention (CDC). The hospital regularly provides 35–48% of all respiratory samples tested for influenza virus in the El Salvador national influenza surveillance system.

Sample processing

Respiratory samples were transported in virus-transport medium (i.e. tryptase-phosphate broth) to the hospital’s laboratory and tested by indirect immunofluorescence for influenza A and B viruses; respiratory syncytial virus; parainfluenza virus types 1, 2 and 3; and adenovirus. All influenza-virus-positive samples and 10% of influenza-virus-negative samples were sent to the National Influenza Centre at the El Salvador Ministry of Health (San Salvador) for quality control analysis by immunofluorescence testing. Starting in 2009 approximately 25% of samples that underwent quality control analysis were further tested using real-time reverse-transcription polymerase chain reaction (qRT-PCR).21

A convenience sample of influenza-virus-positive specimens was sent to the CDC (Atlanta, United States) every six months for antigenic characterization. There, influenza viruses were isolated in Madin-Darby canine kidney cells and underwent haemagglutination inhibition testing using post-infection ferret antisera.22 Antigenic characteristics of the virus isolates were reported to the National Influenza Centre by the CDC’s Influenza Division each year. We classified strains as dominant if they accounted for at least 70% of influenza virus isolates during a particular year and as co-dominant if they accounted for at least 40% but less than 70% (< 70%) of isolates.23 To determine whether the influenza vaccine formulation used in El Salvador during the study period matched the circulating influenza virus strains, we compared these strains with those used to prepare influenza vaccine formulations for the northern hemisphere (the formulation used in El Salvador until 2010) and the southern hemisphere.

Health-care utilization survey

To estimate the denominator for our incidence calculation, we performed a health-care utilization survey to determine the size of the sentinel hospital’s catchment population. During November and December 2009, we used multistage cluster sampling to survey 1663 households in urban and rural locations in five counties of the Santa Ana Department, where 80% of case patients who had severe pneumonia and were admitted to the sentinel hospital lived during 2008 (i.e. the hospital catchment area). For both urban and rural areas, we selected a random sample of census tracts or cantons and then a random sample of households within each of these. Investigators obtained informed consent and surveyed household members to determine who had developed symptoms of and sought care for sudden-onset fever, cough or sore throat (i.e. influenza-like illness) during the preceding month. In addition, the investigators determined where ill individuals had sought care (e.g. at the sentinel hospital or another facility) and whether they were admitted to hospital for severe pneumonia during the preceding year and, if so, the hospital to which they were admitted. We assumed that the pattern of health-care services use by age group was similar every year.

Statistical analyses

We compared case patients who were tested with those who were not tested for respiratory viruses with respect to age, sex and severity of illness by use of χ2 and tests (SPSS, version 17.0; SPSS Statistics, New York, United States of America). To estimate the annual incidence of influenza-virus-associated pneumonia, we first used the monthly proportion of case patients who tested positive for influenza virus at San Juan de Dios Hospital to impute the monthly number of case patients who would have tested positive for influenza virus if all had been tested. We then adjusted this estimate by use of census data,24 the proportion of days each year that samples were collected, and the proportion of those surveyed in the catchment area who, in the previous year, developed sudden-onset fever, cough or sore throat and were subsequently admitted to the sentinel site and not to another hospital (Appendix A, available at: http://xurl.es/pvcvz).


Case patient characteristics

During 2008–2010, physicians at the sentinel hospital identified 2554 case patients aged < 5 years with severe pneumonia: 495 cases (19%) occurred in 2008, 1367 (54%) occurred in 2009 and 692 (27%) occurred in 2010. Samples from 608 (24%; 153 [31%] in 2008, 217 [16%] in 2009 and 238 [34%] in 2010) were collected and tested for respiratory viruses.

Table 1 summarizes select demographic characteristics among tested and untested case patients. During 2008–2009, the median age and prevalence of male sex among tested and untested case patients were similar. However, during 2010, the median age among tested case patients was significantly less than that among untested case patients (6 versus 11 months; P < 0.001). The percentage of tested case patients who died was significantly greater than the percentage of untested case patients who died, both in 2009 (9% versus 0%; P < 0.001) and in 2010 (4% versus 0%; P < 0.001).

Of the 608 case patients with severe pneumonia who underwent testing, 113 (19%) tested positive for at least one of the following respiratory viruses: influenza virus, parainfluenza virus, adenovirus and respiratory syncytial virus. Influenza virus was detected in 37 case patients (6%; 95% confidence interval, CI: 4–8) who provided samples for testing. The median age of influenza-virus-positive case patients was 10 months (interquartile range: 4–12 months). The frequency of influenza virus detection among tested case patients aged < 2 years was higher than that among older tested case patients during 2008 and during 2009 but not during 2010 (Table 2).

The proportion of case patients who tested positive for influenza virus was highest each year from June to September (Table 2). We also observed an increase in influenza-virus-positive cases at the end of December 2009, during the 2009 influenza pandemic, when influenza A virus subtype H1N1pdm (i.e. the pandemic strain) was detected in 17 of 21 patients (81%) with severe pneumonia.

Survey findings

A total of 7683 household members (mean number of members per household: 4.6) were detected during the health-care utilization survey. Participants reported that 464 household members (6%; 95% CI: 4–7) developed influenza-like illness one month before the interview. Of these people, 233 (50%; 95% CI: 45–55) sought medical care. During the year before the interview, 22 household members (0.3%; 95% CI: 0.2–0.4) were hospitalized with a history of influenza-like illness. Among them, 5 of 8 (63%) aged < 5 years were hospitalized at the sentinel hospital.

Overall and annual incidence

Table 3 summarizes the data used to estimate the incidence of influenza-virus-associated severe pneumonia among children aged < 5 years in El Salvador during 2008–2010. The incidence during the study period was estimated to be 3.2 cases per 1000 person–years (95% CI: 2.8–3.7). The greatest annual incidence occurred in 2009, with 7.6 cases per 1000 person–years (95% CI: 6.5–8.9); 6.7 cases per 1000 person–years were attributable to influenza A virus subtype H1N1pdm. Incidence estimates were significantly lower during the other study years: 1.5 cases per 1000 person–years (95% CI: 1.0–2.0) for 2008 and 0.6 cases per 1000 person–years (95% CI: 0.3–1.0) for 2010.

Matching between vaccines and recovered isolates

During 2008–2010, El Salvador used influenza vaccine formulations manufactured for use in the northern hemisphere but haemagglutination inhibition testing showed that the influenza virus strains identified in the study matched antigens from both the northern and southern hemisphere formulations (Table 4). The northern hemisphere formulation seemed to be a better match than the southern hemisphere formulation in 2008 because it included the dominant circulating strain (A/Brisbane/59/2007 [H1N1]-like virus). On the other hand, both formulations were considered best matches to circulating strains in 2010 because both contained the same strains.


This study, which is, to our knowledge, the first to estimate the incidence of influenza-virus-associated severe pneumonia among children aged < 5 years in El Salvador, suggests that young children are frequently hospitalized as a result of influenza. If the 2008 incidence of influenza-virus-associated severe pneumonia among children aged < 5 years hospitalized in San Juan de Dios Hospital (1.5 cases per 1000 person–years; 95% CI: 1.0–2.0) was similar throughout El Salvador, we estimate that from 600 to 1100 influenza-virus-associated hospitalizations occurred among children in this age group during 2008 [(1.0–2.0 cases per 1000 person–years) × (555 893 individuals aged < 5 years)]. Almost all cases of influenza-virus-associated severe pneumonia identified during 2009 were attributed to influenza A virus subtype H1N1pdm. Although influenza due to pandemic and seasonal strains had similar clinical manifestations, younger children had higher attack rates during the pandemic year than during 2008 and 2010. This may be explained by the lack of pre-existing immunity against the pandemic influenza strain among children.26,27 The incidence of influenza-virus-associated severe pneumonia requiring hospitalization among children in this study was similar to those reported in other low-income and lower-middle-income countries, including Bangladesh, Guatemala, Kenya and the Philippines, but lower than those reported in Nicaragua (3 cases per 1000 person–years), Thailand (5 cases per 1000 person–years) and Viet Nam (9 cases per 1000 person–years).714

The incidence of influenza-virus-associated severe pneumonia requiring hospitalization is important because influenza is a vaccine-preventable infection and because respiratory illnesses are among the leading causes of death among children in lower-middle-income countries such as El Salvador.28 During 2008–2010, Ministry of Health hospitals in El Salvador annually admitted approximately 14 000 children aged < 5 years with severe pneumonia. In addition, in El Salvador severe pneumonia causes 14% of deaths among children aged < 5 years.17 Infection with influenza virus also predisposes children to infections with other common pathogens associated with severe illness (e.g. pneumococcal and staphylococcal pneumonia).29 Although our study did not explore the association between influenza and bacterial pneumonia, in 2010 El Salvador initiated a bacterial pneumonia surveillance system and introduced pneumococcal conjugate vaccine.30,31

During the study period, El Salvador used the northern hemisphere influenza vaccine formulation. Our data suggest that both northern and southern hemisphere vaccine formulations matched influenza viruses circulating in El Salvador during 2008 and 2010, with the exception of the pandemic strain. In a similar analysis of unpublished 2003–2010 data, the Ministry of Health concluded that the southern vaccine formulation most often matched influenza viruses identified among Salvadorian patients. As of 2011 the Ministry of Health uses the southern hemisphere vaccine formulation to target children aged < 5 years, elderly people, health-care personnel, pregnant women and people with pre-existing medical conditions during influenza vaccination campaigns starting in March rather than in January, when they were formerly begun. Despite high reported vaccine coverage during 2010, it is unclear how many eligible children receive a second dose of influenza vaccine. It may be useful for the Ministry of Health to verify that coverage is high among children and other target groups and to explore strategies to sustain high vaccination coverage. In addition, vaccination coverage among pregnant women and other persons with pre-existing medical conditions remains low, at approximately 33%. Knowledge of rates of influenza-associated hospitalization and other burden of disease data might improve communication of the potential risk of influenza and help mobilize persons and resources in El Salvador for successful influenza vaccine campaigns.

Our findings might also help public health officials in El Salvador to better determine the potential value of empirically treating severe pneumonia in children aged < 5 years with oseltamivir during influenza epidemics. In addition, data on the rates of influenza-associated hospitalization have the potential to help El Salvador explore the potential value of targeted non-pharmaceutical interventions such as hand washing campaigns among groups at high risk of hospitalization as a result of influenza virus infection.

Although there is still little information about influenza seasonality in tropical countries, unpublished data from 2003–2011 suggest that El Salvador typically has influenza epidemics during June to September (M. Melendez et al., e-mail correspondence, 5 July 2012), concurrent with El Salvador’s rainy season, as often occurs in tropical countries in Asia, Africa and South America.3235 Improved understanding of the magnitude and timing of annual influenza activity can help health authorities improve the timing of influenza vaccination campaigns and recommendations to commence empirical antiviral treatment of people with suspected influenza virus infection. In addition, continued surveillance can help determine the best vaccine formulation for El Salvador.

We believe it important for countries like El Salvador to explore sustainable ways to support national laboratories with supplies and reagents for qRT-PCR testing for respiratory viruses; to train clinical and surveillance personnel to identify, register, code and notify all severe pneumonia cases at the sentinel sites; and to use standard case definitions along with standardized operating procedures to avoid selection bias and misclassification. Influenza surveillance can be expensive for low-middle income countries. For example, starting in 2007 the CDC’s Central American Regional Office provided El Salvador with approximately US$ 350 000 in technical and economic assistance to strengthen influenza surveillance. Sustainability should be addressed with long-term strategic planning to promote the rational use of limited resources.

This study has several limitations. First, the collection of samples from patients was not systematic. As a result, 62% of specimens were from children with very severe pneumonia, which might have resulted in selection bias. Second, qRT-PCR was unavailable at El Salvador’s National Influenza Centre before 2009 and was performed on approximately 25% of samples obtained during 2009–2010. Therefore, almost all of the respiratory samples in our study underwent immunofluorescence analysis for influenza virus and respiratory syncytial virus, which is less sensitive than qRT-PCR for detecting respiratory viruses.36 In addition, 16% of specimens were collected ≥ 5 days after symptom onset, when immunofluorescence and qRT-PCR testing are both less likely to detect respiratory viruses. Finally, because under-reporting might have arisen, we should have evaluated the sentinel site’s adherence to the severe pneumonia case definition, to better understand the magnitude of bias. These limitations may have yielded conservative estimates of the incidence of influenza-virus-associated severe pneumonia among children who required hospitalization.

Annually, El Salvador has a considerable incidence of influenza-virus-associated pneumonia among children aged < 5 years, for whom lower respiratory tract infections are a leading cause of death.17 An integrated approach to preventing and mitigating childhood pneumonia in El Salvador could include a programme to control and prevent influenza. Our findings suggest that studies to verify whether influenza vaccine coverage is high among children and other target groups are warranted and that vaccination strategies that have been proven to achieve high coverage should be sustained. The incidence of influenza-virus-associated severe pneumonia in El Salvador may help renew efforts to increase seasonal influenza vaccination coverage among target groups and to expand vaccination coverage among other vulnerable groups (e.g. pregnant women).


We thank Elizabeth de Cuellar and Iván Aparicio (San Juan de Dios Hospital); Guadalupe de Guzmán (El Salvador National Influenza Center); Patricia Alberto and Ana Beatriz Sánchez (El Salvador Ministry of Health); Nancy Cox, Carolyn Bridges, Ann Moen and, Tomas Rodriguez, Erin Burns and Mary M McCauley (CDC Influenza Division); Otavio Oliva (Pan American Health Organization); and International Centre for Diarrhoeal Disease Research, Bangladesh. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the CDC or the institutions affiliated with the authors.

Competing interests:

None declared.