All patients 18 years of age or older were asked for verbal consent for screening and, if they met the case definition, written, informed consent to participate in the surveillance study. Relatives of adult patients who were unconscious or unable to provide consent on enrollment were asked to provide written, informed consent for their relative to participate, and this consent was renewed directly with the patient on regaining consciousness. Parents or guardians of children <18 years old were asked for verbal consent to screen their child to determine eligibility, after which written, informed consent was requested from the parents or guardians and written, informed assent from children aged 7 to 17 years old. The protocol received approval from the institutional review boards of the Universidad del Valle de Guatemala (UVG; Guatemala City, Guatemala) and Centers for Disease Control and Prevention (CDC; Atlanta, GA) and approval from the Guatemalan Ministry of Public Health and Social Welfare (MSPAS; Guatemala City, Guatemala). All positive pH1N1 results were reported immediately to the MSPAS, who informed the local public health authorities and patients in each site. The policy of the MSPAS was to provide oseltamivir treatment free of charge to all patients confirmed with pH1N1, although few patients received timely antiviral treatment due to limited stocks and delays in reporting results.

Guatemala, with a population over 14 million, has a gross national income per capita of $2680 and is considered a middle-income country by the World Bank (http://data.worldbank.org/indicator/NY.GNP.PCAP.CD, accessed on 1 September 2010). Seasonal influenza vaccination is part of the national immunization schedule for persons ≥50 years old and healthcare workers, but coverage among these groups is low despite the vaccine being provided free of charge when available.

A description of the surveillance system for hospitalized pneumonia and ambulatory ILI in Guatemala has been presented previously [22]. Briefly, we conducted prospective surveillance for hospitalized pneumonia and ILI in the Departments of Santa Rosa (total population: 319,963) and Quetzaltenango (total population: 705,301). In Santa Rosa, surveillance for hospitalized pneumonia began in November 2007 and was conducted at the only hospital in the department, the National Hospital of Cuilapa. This hospital is a 176-bed regional referral hospital with a four-bed pediatric intensive care unit (ICU) and a four-bed adult ICU. In Quetzaltenango, surveillance for hospitalized pneumonia began in February 2009 and was conducted at the Western Regional Hospital, one of two general-purpose public hospitals in the department. The Western Regional Hospital is a larger facility with 425 beds, including 22 pediatric and six adult ICU beds. Both hospitals provide free care and serve mostly low- and mid-income populations.

Surveillance for ILI in public ambulatory clinics began in Santa Rosa in November 2007 in one health center (staffed by at least one physician) and was then expanded to five additional health posts (staffed by nurses) in June 2009 in response to the pH1N1 pandemic. Surveillance for ILI in ambulatory clinics in Quetzaltenango began in July 2009 in three health centers and one health post. Health care is also provided free of charge at these ambulatory clinics.

Prior to the pandemic, surveillance was limited to residents of the catchment area of each facility, but this geographic restriction was lifted in May 2009 to assist with monitoring the pandemic.

A case of pneumonia was defined as a patient admitted to the hospital with at least one sign of acute infection and at least one respiratory sign or symptom from the respective columns in Table 1. A case of ILI was defined according to PAHO/CDC guidelines as a patient presenting to an ambulatory health clinic with a measured temperature >38°C and either cough or sore throat [23]. Suspect cases were identified prospectively by study nurses through review of ward registers or patient chief complaints to find any patients with respiratory-related illnesses. In addition, because this surveillance system also collects information on gastrointestinal, neurological and febrile illnesses, patients admitted for, or presenting with, complaints related to these syndromes were also screened to determine whether they met the case definition for pneumonia (hospital) or ILI (ambulatory clinic).

In the hospital, clinical and epidemiologic data were collected from both patient interviews and chart reviews. The highest temperature (axillary) documented in the first 24 hours after admission was recorded. A standard review of chest radiographs was undertaken by a study radiologist, and a standard pulmonary physical exam was performed by a study physician. Oxygen saturation was measured by study nurses with a pulse oximeter. Patients were asked whether they had been diagnosed with any of the following chronic conditions: asthma, other lung disease, diabetes, cancer, chronic cardiovascular disease (hypertension or heart disease), liver disease, kidney disease or any immunocompromised condition (including HIV/AIDS).

In the ambulatory clinics, axillary temperature was measured by study nurses and all other clinical and epidemiologic data were collected through patient interview. In both the hospital and ambulatory clinics, care sought for the current illness episode prior to hospital admission or presentation to the ambulatory clinic was reported by the patient or caregiver, along with any medicines taken prior to admission or presentation. Diarrhea was defined as three or more liquid or loose stools in a 24-hour period during the last seven days. Trained study nurses took nasopharyngeal swabs (NP) from all eligible and consenting patients with pneumonia and ILI, whereas oropharyngeal (OP) swabs were also taken from pneumonia patients; NP and OP swabs were put into one tube with viral transport media and stored at 4°C until they could be processed and sent to the laboratory at the UVG.

All patients with hospitalized pneumonia were asked to return three to six weeks after discharge for a follow-up visit. Patients who did not return by six weeks were followed with a phone call to determine their vital status.

A laboratory-confirmed case of influenza A was defined as a case of pneumonia or ILI with influenza A virus infection as determined by real-time reverse transcription polymerase-chain reaction (rRT-PCR). Subtyping for influenza A to differentiate viruses as seasonal H1N1 and H3N2 and pH1N1 was conducted using a standardized CDC protocol [24]. We tested for adenovirus [25] and other respiratory viruses (human metapneumovirus, human parainfluenza viruses 1 to 3, influenza B and respiratory syncytial virus [RSV]) using CDC rRT-PCR protocols (D. Erdman, pers. comm.). A rapid antigen test (BinaxNow™, Inverness Medical Professional Diagnostics, Princeton, NJ) to detect infection with Streptococcus pneumoniae was used in urine samples from all persons >10 years of age.

Data were collected using preprogrammed, hand-held personal digital assistants. Only data from 2008 and 2009 were included in this report. SAS v. 9.1 (Cary, NC) was used for analysis. The case fatality proportion (CFP) was calculated as the number of deaths among the cases of influenza A that occurred during hospitalization or within six weeks of discharge divided by the total number of cases of influenza A among hospitalized pneumonia patients. For non-normally distributed continuous variables, the Mann-Whitney U-test was used when two influenza subtypes were compared, and the Kruskal-Wallis analysis of ranks was used when three influenza subtypes were compared. Pearson's chi-square statistic or Fisher's exact test were used to test for differences in categorical variables between patients infected with different influenza A virus subtypes and a P value of <0.05 was considered statistically significant. Variables with missing data for >10% of the patients were analyzed to ensure no difference in terms of age, sex or influenza A subtype between those with and without data.

The median age of hospitalized pneumonia patients with influenza A was 3 years, and was lower for seasonal H1N1 (2 years) than for pH1N1 (4 years), but the difference was not statistically significant (P = 0.48) (Table 2). The proportions of hospitalized pneumonia patients aged <1 year old with seasonal H1N1 (15/38; 39%) and pH1N1 (28/76; 37%) were similar (P = 0.42) (Figure 4a). Young adults 15 to 29 years old made up a larger proportion of hospitalized pneumonia patients with pH1N1 (10/76; 13%) than seasonal H1N1 (1/38; 3%) but the difference was not statistically significant (P = 0.1). The proportion of hospitalized pneumonia patients 60 years or older with seasonal H1N1 (3/38; 8%) and pH1N1 (5/76; 7%) were similar (P = 1.0). Overall, there was no significant difference in the age distribution of hospitalized pneumonia patients between seasonal H1N1 and pH1N1 (P = 0.21).

The age distribution of the ILI patients with influenza A differed from that of hospitalized pneumonia patients (Figure 4b) (P<0.001), with the median age of ILI patients (8 years) greater than that for hospitalized pneumonia patients (3 years). Among ILI patients, there was no statistical difference between the age distribution of pH1N1 and seasonal H1N1 (P = 0.22) or pH1N1 and H3N2 (P = 0.25) (Table 3). The proportion of ILI patients with pH1N1 that was <1 year old (20/162, 12%) was not significantly different from the proportion of ILI patients with either seasonal H1N1 (10/51, 20%; P = 0.25) or H3N2 (5/21, 24%; P = 0.17). Similarly, the proportion of ILI patients with pH1N1 that were ≥60 years old (3/162, 2%) was not significantly different from seasonal H1N1 (0/51, 0%; P = 1.0) or H3N2 (0/21, 0%; P = 1.0). The group most affected by ILI due to influenza A was school-age children from five to 14 years old with 108 (44%) cases of influenza A; there was no statistically significant difference in the proportion of pH1N1 cases aged 5 to 14 years old (80/162, 49%) in comparison to seasonal H1N1 (17/51, 33%; P = 0.05) or H3N2 (6/21; 29%; P = 0.10).

Children <5 years old admitted with pneumonia and found to have influenza A were more likely to have sought care previously (42/67, 63%) than persons ≥5 years old (29/61, 48%), but the difference was not statistically significant (P = 0.09). There was no difference in the proportion of children <5 years old and persons ≥5 years old with respect to having taken any medication, including antipyretics and antivirals, within the 72 hours prior to admission (data not shown). Children <5 years old admitted with pneumonia were more likely to present with wheezing (55/72, 76%) than persons ≥5 years old (29/64, 45%; P = 0.0002), but were less likely to present with tachypnea (persons <5 years old: 28/71, 39%; persons ≥5 years old: 48/65, 74%; P<0.0001). Headaches, muscle aches and shivering were significantly more common among persons ≥5 years old admitted with pneumonia compared to children <5 years old (data not shown). There was no significant difference by age in the proportion of hospitalized pneumonia patients with cough, difficulty breathing, shortness of breath, sputum production, sore throat, fever, rhinorrea, pleuritic chest pain, oxygen saturation <90% or diarrhea. Children <5 years old were more likely to be coinfected with another respiratory virus (27/73, 37%) than were persons ≥5 years old (5/65, 8%; P<0.001). There were no significant differences in severe outcomes (admission to the ICU, mechanical ventilation or death) between children <5 years old and persons ≥5 years old (data not shown).

Among ILI patients with laboratory-confirmed influenza A, children <5 years old (12/56, 21%) were no more likely to have sought care elsewhere prior to presentation at the ambulatory clinic than persons ≥5 years old (33/139, 24%; P = 0.73). Children <5 years old were more likely to present with cough (65/65, 100%) than persons ≥5 years old (139/148, 94%; p = 0.04). Diarrhea was also more commonly reported from children <5 years old (10/65, 15.4%) than persons ≥5 years old (8/148, 5%; p = 0.02). Patients ≥5 years old were significantly more likely to report muscle aches and shivering than patients <5 years old (data not shown). There was no difference in the proportion of children <5 years old and persons ≥5 years old in terms of viral coinfection (data not shown). Among ILI patients, there were no significant differences in the presence of cough, sore throat, rhinorrhea, sputum production, pleuritic chest pain, difficult breathing or shortness of breath by age (data not shown).

About 40% of the influenza A cases were females, and there was no significant difference in sex between pH1N1 and seasonal H1N1 (P = 0.68; Table 2). Treatment-seeking behavior and treatment received prior to hospital admission for pneumonia was similar between patients with seasonal H1N1 and pH1N1. The interval between symptom onset and hospital admission was similar for both subtypes, between four and five days.

Clinical presentation of hospitalized pneumonia patients with influenza A was similar between patients infected with seasonal H1N1 and pH1N1; cough, difficulty breathing and shortness of breath were the most common symptoms reported by patients with both subtypes (Table 2). However, a measured temperature ≥38°C within the first 24 hours of admission was less frequent among hospitalized pneumonia patients with pH1N1 (62%) than among those with seasonal H1N1 (84%; P = 0.01). We did not evaluate the frequency of vomiting or nausea among these patients, but there was no difference in the proportion of patients reporting diarrhea by subtype (P = 0.25).

As reported previously [22], in this population, the proportion of hospitalized pneumonia patients with influenza A reporting a chronic or underlying medical problem was very low, less than 20%, and there was no significant difference between patients with seasonal H1N1 and pH1N1 (P = 0.93). Asthma was the most common condition reported. There were no women aged 14 to 44 years old hospitalized with seasonal H1N1; more than a quarter (29%) of the female patients with pH1N1 in this age range was pregnant, but this amounted to only two women.

Almost a quarter (22%) of hospitalized pneumonia patients with influenza A were admitted to the ICU (Table 4). This proportion was higher for patients infected with pH1N1 (28%) than for seasonal H1N1 (18%) but the difference was not statistically significant (P = 0.28). We examined the proportion of hospitalized pneumonia patients with influenza A admitted to the ICU before the pandemic began in May 2009 (7/34, 21%) and after (23/104, 22%), but there was no statistical difference (P = 0.85). The proportion of patients with respiratory distress who required mechanical ventilation was more than three times higher among the hospitalized pneumonia patients with pH1N1 (11%) compared with those with seasonal H1N1 (3%), but the difference was not statistically significant (P = 0.14).

The CFP was three times higher for hospitalized pneumonia patients with pH1N1 (15%) compared to seasonal H1N1 (5%), but this difference was not statistically significant (P = 0.21) (Table 4). The two deaths associated with seasonal H1N1 both occurred in patients between 50 and 59 years of age during their hospitalization. Out of the 11 deaths among patients with pH1N1, nine occurred in the hospital and two within one day of discharge. Four of the patients with pH1N1 who died were <1 year old, two were between one and 15 years old, two were between 20 and 49 years old, and three were between 50 and 59 years old. There were no deaths of patients with pH1N1 ≥60 years old.

Coinfection with another respiratory virus was similar between hospitalized pneumonia patients with seasonal H1N1 and pH1N1 (P = 0.52) (Table 2). However, there were different respiratory viruses associated with coinfection by influenza A subtype: adenovirus was the most common viral coinfection among the hospitalized pneumonia patients with seasonal H1N1, with 13% of the patients coinfected. In contrast, RSV was the most common viral coinfection among hospitalized pneumonia patients with pH1N1, coinfecting 14% of these patients. The significant difference in the proportion of patients with pH1N1 compared with seasonal H1N1 coinfected with RSV was due to a much greater incidence of RSV in 2009 than in 2008, with the peak number of cases coinciding with the 2009 influenza pandemic (Figure 3).

Among the 39 hospitalized pneumonia patients <5 years old with pH1N1, there was no difference in the proportion admitted to the ICU between those patients coinfected with RSV (3/10, 30%) than those not coinfected with RSV (9/29, 31%; P = 1.0). A greater proportion (2/10, 20%) of the patients coinfected with RSV than the patients not coinfected with RSV (2/29, 7%) required mechanical ventilation, but the difference was not statistically significant (P = 0.27). Similarly, the CFP among patients coinfected with RSV (3/10, 30%) was higher than that among those not RSV coinfected (2/29, 7%), but the difference was not statistically significant (P = 0.1). Of the four deaths associated with pH1N1 among this age group, three (75%) patients were coinfected with RSV.

There was no difference in the proportion of children and adults ≥10 years old who were coinfected with S. pneumoniae by subtype (Table 2).

Less than a quarter (23%) of the ILI patients with influenza A sought care outside the home before presenting at the health center or health post, and this percentage did not differ significantly by influenza A subtype (P = 0.61) (Table 3). The median number of days from onset of symptoms to presentation at an ambulatory clinic for patients with influenza A was two days and was similar across all influenza subtypes (P = 0.11).

Clinical signs and symptoms of patients with ILI were similar across influenza A subtypes (Table 3), apart from difficulty breathing and pleuritic chest pain, which were significantly less likely to be reported by ILI patients with pH1N1 than those with seasonal H1N1 or influenza A (H3N2). We found a higher proportion of patients with diarrhea among those with seasonal H1N1 (14%) than either pH1N1 (7%) or H3N2 (5%), although this difference was not statistically significant (P = 0.24).

The proportion of ILI patients with influenza A who were coinfected with another respiratory virus was higher among ILI patients with seasonal H1N1 (14%) and H3N2 (14%) than pH1N1 (9%), but the difference was not statistically significant (P = 0.57). Whereas the proportion of ILI patients with pH1N1 who were coinfected with RSV (7%) was higher than the proportion of seasonal H1N1 (0%), the proportion of ILI patients with H3N2 coinfected with RSV was similar (10%), and there was no overall difference by influenza A subtype (P = 0.12).