The protocol was reviewed and approved by the Ethical Review Boards of the Kenya Medical Research Institute (KEMRI) (SSC#932) and the Institutional Review Board of the US Centers for Disease Control and Prevention (CDC) (IRB# 4566).

The Centers for Disease Control and Prevention-Kenya (CDC-K) and KEMRI have been collaboratively conducting population-based infectious disease surveillance (PBIDS) for pneumonia, diarrhea, fever and jaundice since late 2005 in 2 regions in Kenya: Kibera, a large, informal urban settlement in Nairobi, and Lwak, a rural area in western Kenya. The study regions and surveillance methods have been described previously [21], [22]. Kibera has a population density of ∼70,000 persons/km². Homes are single-level, built with mud, wood, and metal sheeting, chaotically scattered along unpaved roads and open sewers. A majority of men work as security guards, servants, casual laborers, or small business merchants within the city [21]. The altitude of Nairobi is approximately 1,600 meters, minimizing the likelihood for substantial transmission of malaria. Lwak has a population density of ∼325 persons/km². Multi-family compounds are spread throughout the area, often with great distance between compounds. Residents are predominantly farmers and fishermen. Malaria is endemic. Approximately 16% of febrile patients in Kibera have malaria parasitemia (CDC-Kenya, unpublished data), whereas >50% of ILI cases in Lwak are associated with malaria-positive blood smears [23]. Temperatures are different in the two sites; Lwak is hotter and has more rainfall compared to Nairobi. The daily mean temperature in Nyanza, the province where Lwak is located, ranges from a high of 30.8°C in February to a low of 27.7°C in July. Annual rainfall is approximately 1,400 mm/year, peaking in April–May and November–December. In contrast, mean daily maximum temperature in Nairobi ranges from a high of 25.6°C in February and March to a low of 20.6°C in July. Annual rainfall in Nairobi averages approximately 1000 mm/year with peak rainfall in April–May and November–December [24].

Approximately 28,000 and 25,000 persons participated in PBIDS in Kibera and Lwak, respectively. Community interviewers visited participating households regularly to inquire about illnesses among household members. For the household morbidity surveillance, community interviewer visits were conducted biweekly in Kibera until September 2009, when visits were increased to weekly across half of the study site. Beginning February 1, 2010, community interviewer household visits were conducted weekly across the entire study site. In Lwak, household visits were conducted biweekly until they were increased to weekly on January 4, 2010. During household visits, community interviewers asked participants if they had experienced cough, fever, diarrhea, and other symptoms since the previous visit. If the resident reported currently or previously having any symptom since the previous visit, the community interviewer collected detailed information about symptom onset and duration, measures temperature, 1-minute respiratory rate, and, among children, evaluated lower chest wall indrawing and stridor.

Each surveillance site had 1 field clinic that provided free medical care to all study residents. In Kibera, residents could attend Tabitha Clinic, an outpatient facility owned by Carolina for Kibera (Chapel Hill, NC) and staffed and equipped by KEMRI-CDC. In Lwak, residents could attend St. Elizabeth Lwak Mission Hospital, which had inpatient and outpatient facilities. All participants lived within 1 and 5 kilometers from the clinics in Kibera and Lwak, respectively. Sick residents who visited the clinic were questioned regarding symptoms of present illness. Additionally, information about vital signs, physical exam, diagnosis, treatment, and outcome were collected along with specimens from patients meeting certain clinical criteria.

During the study period, nasopharyngeal and oropharyngeal swabs were collected at field clinics from patients with ILI or acute lower respiratory illness (ALRI). ILI was defined as an axillary temperature ≥38°C and cough or sore throat [25]. The ALRI case definition differed by patient age. ALRI for persons aged <5 years was defined, according to the World Health Organization's Integrated Management of Childhood Illness guidelines for severe pneumonia [26], as cough or difficulty breathing, and 1 of the following symptoms: inability to drink or breastfeed, convulsions, lower chest indrawing, loss of consciousness, lethargy, vomiting, stridor, or blood oxygen saturation <90%. In persons aged ≥5 years, ALRI was defined as an axillary temperature ≥38°C or blood oxygen saturation <90%, and either cough, difficulty breathing, or chest pain. All specimens were processed at the KEMRI-CDC International Emerging Infections Program Laboratory in Nairobi. Specimens were tested for influenza A and B by real-time reverse transcriptase-polymerase chain reaction (RT-PCR); those testing positive for influenza A were further subtyped by RT-PCR. The RT-PCR testing used the CDC pH1N1 testing protocol [27]. Specifics of the laboratory testing were described previously [28].

A pH1N1 case was defined as any laboratory-confirmed pH1N1 infection in an ILI or ALRI patient evaluated at the field clinic during August 1, 2009–February 5, 2010 in Kibera or August 1, 2009–January 20, 2010 in Lwak. The study periods were determined on the basis of available data at the time of analyses. Only patients who had a household visit ≤20 days after their symptom onset date were included in the study. Date of symptom onset was calculated by using data from the clinic questionnaire. The most common reason that patients did not have an interview in the 20 days after symptom onset was that the patient or household proxy was not at home during the community interviewer visit. In households with ≥1 laboratory-confirmed pH1N1 illness, the person with the earliest symptom onset date was considered the index patient. A case household was defined as a household with a laboratory-confirmed pH1N1 patient; a case-household person was a household contact of the pH1N1 patient.

To estimate the underlying illness rate in the community, we randomly selected from the PBIDS database 4 comparison households per case household, matched by the number of children aged <5 years. Comparison households did not have laboratory-confirmed pH1N1 cases, and had to have had a household interview ≤10 days before or after the matched patient symptom onset date.

We calculated the number of ILI cases among case-household and comparison-household persons by using data from household visits. For case-household persons, we considered laboratory-confirmed pH1N1 infections or episodes of ILI that occurred ≤8 days after the index patient's symptom onset date as a secondary cases. The cutoff of 8 days was determined on the basis of average duration of pH1N1 shedding in Kibera as determined in a recent study [28]. For comparison household members, a case was defined as an episode of ILI that occurred ≤8 days before the household morbidity surveillance interview date.

We compared the sex distribution of the case and comparison households by using the Cochran-Mantel-Haenszel (CMH) test, adjusted for the matching factor. We compared the age and family size of the 2 groups by using the Wilcoxon signed-rank test for matched samples. We calculated the age-adjusted risk ratio (aRR) and corresponding 95% confidence interval (CI) of secondary ILIs in index households in Kibera and Lwak by using log-binomial regression [29]. We calculated the risk ratio (RR) and its corresponding 95% CI of secondary ILIs among case households and cases of ILI among comparison households. We assessed the role of age in pH1N1 transmission with respect to susceptibility (e.g., getting infected) and infectiousness (e.g., ability to infect others). For susceptibility, we compared the proportion of secondary ILIs among household members aged <5 years with those aged ≥5 years by calculating RRs and corresponding 95% CIs. To determine the age group that was most infectious, we categorized pH1N1 patients into 3 age groups (<5, 5–15, and ≥15 years) and compared the proportion of secondary ILIs associated with the index pH1N1 patient among each age group by using chi-square tests.

We identified 170 laboratory-confirmed pH1N1 infections in Kibera and 83 in Lwak. As presented in Figure 1, among the 170 laboratory-confirmed Kibera cases, 55 cases were excluded from the study because of missing household identification information (6); missing clinic data (6); because no household interview occurred ≤20 days after symptom onset (32); or because they were not the first laboratory-confirmed case in the household (11). Among the 83 laboratory-confirmed Lwak cases, 17 were excluded because of missing household identification information (8); because no household interview occurred ≤20 days of symptom onset (2); or because they were not the initial laboratory-confirmed case in the household (7). Therefore, we included 115 (Kibera) and 66 (Lwak) patients, in our analysis.

In Kibera, the median (range) age among patients was 7.1 (0.6–41.3) years, and 59 (51%) were female (Table 1). In Lwak, the median (range) age was 9.7 (0.2–55.9) years, and 35 (53%) were female (Table 2). The first laboratory-confirmed pH1N1 infections were confirmed on August 3 and September 29, 2009, in Kibera and Lwak, respectively (Figure 2). The number of laboratory-confirmed cases in Kibera was highest in early November and decreased thereafter. Lwak’s initial peak was during late November, approximately a week later than that in Kibera, and increased again at the end of December.

In Kibera, 311 (53%) and 1175 (52%) of the case household and comparison household members were female, respectively. The median (range) age was 16.6 (0.1–67.5) and 15.8 (0.04–70.2) years among case and comparison households, respectively (Table 1). In Lwak, 176 (50%) and 674 (52%) of the case and comparison household members were female, respectively. The median (range) age was 15.5 (0.1–87.5) and 16.6 (0–90.4) years, among case and comparison households, respectively (Table 2).

In Lwak, 8.2% of the case household members had secondary ILI, whereas 3.3% of the comparison household members had ILI (RR, 2.6; 95% CI, 1.6–4.3). In Kibera, 4.6% of the case household members had secondary ILI, whereas 2.7% of the comparison household members had ILI (RR, 1.8; 95% CI, 1.1–2.8) (Table 3). The proportion of secondary ILI cases among case households was significantly lower in Kibera compared with Lwak, (aRR, 0.5; 95% CI, 0.3–0.9). The mean (standard deviation) serial interval was 4.0 (1.6) and 3.2 (3.0) in Kibera and Lwak, respectively.

The proportion of secondary cases of ILI among case households was greater among children aged <5 years than among persons aged ≥5 years in both sites (Table 4). In Lwak, the difference was statistically significant; children aged <5 years were 3.8-fold more likely to have had a secondary case of ILI than persons aged ≥5 years (22.5% versus 5.9%; RR 95% CI, 1.9–7.5). In Kibera, children aged <5 years were 1.4-fold more likely to have had a secondary case of ILI compared with persons aged ≥5 years, but the difference was not statistically significant (6.1% versus 4.3%; RR 95% CI, 0.6–3.4).

In both sites, the proportion of secondary cases of ILI differed according to the index pH1N1 patient age group, but the difference was not statistically significant in either site (P  = .33 in Kibera, and P  = .06 in Lwak) (Table 4). In Kibera, the proportion of secondary ILI was highest when the index patient was aged ≥15 years (6.7%, 8/120), second-highest when the index patient was aged <5 years (5.0%, 11/219), and lowest when the index patient was aged 5–15 years (3.3%, 8/245). In Lwak, the proportion of secondary ILI was highest when the index patient was aged 5–15 years (11.2%, 23/206), second-highest when the index patient was aged ≥15 years (4.8%, 4/84), and lowest when the index patient was aged <5 years (3.2%, 2/62).