Estimates of the national burden of respiratory syncytial virus in Kenyan children aged under 5 years, 2010–2018

To estimate the rates of outpatient burden and non-medically attended ARI for the base region (base rates, from where rates were extrapolated to the other regions), we used data previously collected by the Kenya Medical Research Institute – Center for Global Health Research (KEMRI–CGHR) with support from the US Centers of Disease Control and Prevention (CDC) in Siaya County, Nyanza region, western Kenya. Siaya County has high mortality among children aged < 5 years, as well as high malaria and HIV prevalence [24, 25]. Data collected from 2010 to 2018 among children aged < 5 years visiting St. Elizabeth Lwak Mission Hospital (LMH), which is the referral clinic for the Asembo Population-Based Infectious Disease Surveillance (PBIDS) platform, were used to estimate the Nyanza region ARI base rates [26]. Asembo PBIDS is located within the Siaya Health and Demographic Surveillance System (HDSS) [27], a comprehensive surveillance program for evaluating disease burden/indicators and impact of public health interventions. The PBIDS covers the population within a 5-km radius from LMH; here, the residents were asked about children’s history of ARI in the last 14 days from the date of interview and health-seeking patterns [26, 28]. ARI was defined as an acute respiratory infection (onset within 10 days from interview date) with cough, difficulty breathing, sore throat, or runny nose.

For the estimation of the base rates of hospitalized SARI and severe ARI that was not hospitalized (non-hospitalized SARI), data collected through pediatric surveillance of RSV at the Kilifi County Referral Hospital (CRH) from 2010 to 2018 were used. This surveillance platform is a component of HDSS run by KEMRI–Center for Geographic Medical Research Coast (CGMR–C) funded by the Welcome Trust, UK. Kilifi County has low per capita income, malaria is endemic, and < 5 year and infant mortality ratios are high at 41 and 28.2 per 1000 live births, respectively [29, 30]. We defined SARI as hospitalization with an acute respiratory infection (i.e., cough or difficulty breathing with an onset within the past 10 days from interview date). The location of the base sites, sentinel facilities, regions, and counties are presented in Fig. 1.

The steps used to estimate RSV-associated outpatient ARI are summarized in Fig. 2. We estimated the age-specific annual rates of outpatient ARI for the Asembo PBIDS by dividing the age-specific number of ARI cases by the age-specific population of PBIDS participants. To account for potential variation in health-seeking behavior among patients attending LMH vs. other facilities in the study area, this rate was adjusted by applying the proportion of those with ARI seeking care at any facility to the LMH-specific rates (Additional file 1: Eq. 1). Given that the PBIDS participants are offered free treatment for ARI at LMH, the adjusted rates may not be generalizable to the entire Nyanza region (“base region”). To estimate the outpatient ARI in the base region, we therefore multiplied the rates by the Kenya Demographic and Health Survey (KDHS) health-seeking estimate for the entire base region divided by the health-seeking estimate from the PBIDS (Additional file 1: Eq. 2). Rates of outpatient ARI in the other 7 regions were then calculated by first estimating region-specific adjustment factors based on the relative risk of known risk factors for ARI (Additional file 1: Eq. 3) obtained from Tazinya et al. [20]. The adjustment factors also incorporated the region-specific prevalence of these risk factors, as reported from the demographic and health survey conducted in 2014, and the region-specific health-seeking behavior for ARI as summarized in Additional file 2: Table S1a [31, 32]. We then applied the region-specific adjustment factors to the ARI base rate to estimate the region-specific outpatient ARI rates (Additional file 1: Eq. 4). Similar approaches have previously been used to extrapolate disease burden rates from one base region to the national level [21, 22].

Data on RSV testing at LMH, from nasopharyngeal (NP) and/or oropharyngeal (OP) swabs collected from outpatients who presented with ARI, were available up to 2014. We aggregated this over the years stratified by age categories to estimate age-specific proportions of RSV positive ARI between 2015 and 2018. These proportions were applied to the ARI rates to estimate the age-specific RSV-associated outpatient ARI rate in the base region (Additional file 1: Eq. 6). Data on RSV testing were not available for any other outpatient facility; we therefore generated an ARI to SARI RSV positivity ratio from the base region (i.e., age-specific percent RSV-positive ARI from LMH divided by age-specific percent RSV-positive SARI from LMH and Siaya CRH), which was applied to the proportion of RSV-positive SARI estimated for the other regions to obtain the proportion of RSV-positive ARI for each region assuming that this ratio would be relatively similar across regions. We applied the regional-age-specific proportion RSV positive to the regional rates of outpatient ARI to obtain the regional age-specific rates of RSV-associated outpatient ARI (Additional file 1: Eq. 6).

We used 2018 data from a Health Utilization Survey (HUS) conducted in Siaya County—Nyanza region, Nakuru County—Rift Valley region, Kakamega County—Western region, and Marsabit County—Eastern region, to estimate the proportion of ARI cases seeking care https://​bmcpublichealth.​biomedcentral.​com/​articles/​10.​1186/​s12889-023-15252-3. For regions where the HUS was not conducted, we extrapolated the proportion of ARI cases seeking care from Siaya County by multiplying the Siaya proportion of ARI cases seeking care by region-specific proportion of individuals seeking care for ARI (as defined in KDHS—cough and difficulty breathing > 2 days) divided by the proportion of individuals seeking care for ARI (as defined in KDHS) in Nyanza. To estimate the regional-age-specific rates of non-medically attended ARI, we divided the regional-age-specific rates of outpatient ARI by the region-specific proportion of ARI patients who sought care as estimated from the HUS and KDHS ARI [31, 32] then subtracted the rate of outpatient ARI (Fig. 2 and Additional file 1: Eq. 5). Assuming similar RSV positivity as outpatient RSV, we estimated the regional age-specific rates of non-medically attended ARI associated with RSV by applying the regional age-specific proportion positive for RSV to the regional rates of non-medically attended ARI (Additional file 1: Eq. 7).

Age-specific annual base rates (Coastal region) of hospitalized SARI were estimated by dividing age-specific number of hospitalized SARI cases from Kilifi HDSS residents at the Kilifi CRH by the age-specific population of Kilifi HDSS participants (Fig. 2 and Additional file 3: Eq. 1). This rate was adjusted to include residents hospitalized elsewhere (~ 20%) using admissions data for inpatient hospitals within the Kilifi HDSS catchment area available from the District Health Information Software 2 (DHIS2) (https://​dhis2.​org/​). Regional rates of hospitalized SARI in the other 7 regions relied on a region-specific adjustment factor (Additional file 3: Eq. 2) based on the relative risk and regional prevalence of risk factors for SARI. The region-specific adjustment factor was then applied to the SARI base rate to estimate the region-specific hospitalized SARI rates (Additional file 3: Eq. 3).

To estimate the proportion of SARI cases positive for RSV, we divided the number of RSV-positive cases by SARI cases. NP and/or OP samples obtained from patients hospitalized with SARI at Kilifi and Siaya CRHs were systematically tested for RSV and therefore represented the Coastal and Nyanza regions respectively. Samples were only intermittently tested for RSV in the other regions; we therefore used data from Siaya CRH together with Kakamega CRH for the Western region. We used data from Kenyatta National hospital (NH) and Nakuru and Nyeri CRHs to estimate the proportions for Rift Valley, Central, and Nairobi regions (Fig. 1). For Eastern and Northeastern regions, we used the national average percent RSV positive from the SARI cases in the absence of region-specific data. Regional age-specific rates of hospitalized SARI associated with RSV were then estimated by applying the age-specific proportion RSV-positive cases among the hospitalized SARI patients to the regional age-specific rates of hospitalized SARI, (Additional file 3: Eq. 6).

Using the HUS data described above (unpublished data), we first estimated the proportion of reported pneumonia cases (SARI cases) that resulted in hospitalization—defined as hospitalized cases of a reported episode of respiratory illness with cough and difficulty breathing for > 2 days or a clinician’s diagnosis of pneumonia within the past 12 months [33]. For regions where the HUS was not conducted, we extrapolated the proportion of hospitalized SARI cases from Siaya County by multiplying the Siaya proportion of hospitalized SARI by region-specific proportion of individuals seeking care for ARI divided by the proportion of individuals seeking care for ARI in Nyanza (Fig. 2 and Additional file 3: Eq. 5). Non-hospitalized SARI cases were those who reported an episode of pneumonia as defined above with a clinician’s recommendation for hospitalization or presentation with any one of the danger signs (inability to breastfeed or drink, persistent vomiting, convulsions or seizures, loss of consciousness), yet they were not hospitalized. The rate of non-hospitalized SARI was estimated by dividing the regional-age-specific rates of hospitalized SARI by the region-specific proportion of hospitalized SARI then subtracting the rate of hospitalized SARI (Additional file 3: Eq. 4). Assuming similar proportion of RSV positivity for non-hospitalized SARI as for hospitalized SARI, regional-age-specific rates of non-hospitalized RSV-associated SARI were estimated by applying the age-specific proportion of RSV positivity among the hospitalized SARI cases to the regional age-specific rates of non-hospitalized SARI (Additional file 3: Eq. 7).

RSV-associated in-hospital deaths were estimated by first calculating the age-specific RSV-associated case fatality rate (CFR) using data from Kilifi and Siaya and then applying the age-specific CFRs to regional-age-specific hospitalized SARI rates associated with RSV infection (Fig. 2) to get the RSV-associated mortality rate. We then obtained the number of in hospital deaths by applying the age-specific RSV-associated mortality rate to the age-specific population. Given that data on out-of-hospital deaths are missing for Kenya, we first derived multipliers from in and out of hospital RSV-associated deaths by age groups (0–5, 6–11, 12–59, and 0–59 months) from LMIC estimates by Li et al. [5]. We then applied those age-specific multipliers to the age-specific in-hospital deaths to obtain the number of out-of-hospital deaths. The age-specific in-hospital CFRs and derived age-specific out- and in-hospital multipliers are summarized in Additional file 2: Table S1c.

The county-specific annual numbers of outpatient and non-medically attended ARI and their respective RSV-associated cases (Additional file 1: Eq. 8–11) and the hospitalized and non-hospitalized SARI and RSV-associated SARI cases (Additional file 3: Eq. 8–11) were estimated by applying the region-specific rates to the population of each county in the region. The national, regional, and county population data were obtained from the 2019 national census data [34] and extrapolated for the other years using the World Bank annual percentage growth rate for Kenya [35]. Since census data were only available in broader age categories, we applied population proportions from Siaya and Kilifi HDSS to the census data to obtain fine age bands used in this study. County borders are displayed in Fig. 1.

To estimate adjustment factors to apply to base rates, we estimated the prevalence of each risk factor by running 1000 iterations while allowing it to vary within a binomial distribution. The 50th, 2.5th, and 97.5th percentile values were reported as the estimate and 95% CIs, respectively. ARI rates are reported per 1000 population, whereas SARI and death rates are reported per 100,000 population. Age groups included monthly delineated categories up to 11 months, 12–14 months, 15–17 months, 18–20 months, 21–23 months, 24–35 months, 36–47 months, and 48–59 months. Analyses were performed using Stata version 16 (StataCorp, College Station, Texas) and R version 3.6.3 (Vienna, Austria). Maps of counties, regions, and facilities were assembled using QGIS version 3.10.13 (https://​qgis.​org).

Using data collected from 2010 through 2018, we estimated the national annual rate of outpatient and non-medically attended ARI among Kenyan children aged < 5 years as 1725 (95% CI 1577–1899) and 1896 (95% CI 1733–2088) cases per 1000 children respectively (Table 1). The national rates of hospitalized and non-hospitalized SARI were estimated as 2501 (95% CI 2182–2869) and 7717 (95% CI 6734–8855) cases per 100,000 children respectively (Table 2). The distribution by year and regions are provided in Additional file 2: Tables S2, S4, S6, and S8.

Among Kenyan children aged < 5 years, the average annual rate of RSV-associated outpatient ARI was 206 (95% CI 186–229) cases per 1000 children (Table 1). The highest rate of RSV-associated outpatient ARI was observed in children aged 4 months (410, 95% CI 343–489) followed by 5 months of age (390, 95% CI 322–463), whereas the lowest rate was observed among children aged 48–59 months (60, 95% CI 39–86). Coast (298, 95% CI 254–345) and Eastern regions (297, 95% CI 262–336) had the highest, whereas Nairobi had the lowest rates of outpatient RSV-associated ARI (29, 95% CI 16–45) cases per 1000 children (Fig. 3 and Additional file 2: Table S3). When considering the disease burden over time, the rate of RSV-associated outpatient ARI was lowest in 2015 – 137 (95% CI, 124–153) (Additional file 2: Table S3). The national annual rate of RSV-associated non-medically attended ARI was 226 (204–252) cases per 1000 children (Table 1).

The national annual rate of RSV-associated hospitalized SARI among children < 5 years of age was 349 (95% CI 303–404) per 100,000 children (Table 2). The highest rate was observed in 1-month-old infants at 3420 (95% CI 2911–4012) and the lowest in children aged 48–59 months at 13 (95% CI 7–21). The rate of RSV-associated hospitalized SARI steadily decreased with increase in age with the highest rates observed in younger age groups (< 6 months old and even higher among children aged < 3 months) and lowest in older children (Fig. 4). Eastern and Nairobi regions had the highest 400 (95% CI 338–474) and the lowest 138 (95% CI 84–214) annual rate of RSV-associated hospitalized SARI, respectively (Fig. 4 and Additional file 2: Table S7). The rates were highest in 2010—547 (95% CI 475–640)—and lowest in 2017—168 (95% CI 146–193) (Additional file 2: Table S7). The national annual rate of RSV-associated non-hospitalized SARI was estimated as 1077 (95% CI 934–1247) per 100,000 population (Table 2).

The annual rate of RSV-associated in-hospital deaths per 100,000 children aged < 5 years was estimated at 9 (95% CI 7–13) in Kenya (Table 3). RSV-associated deaths declined with increasing age: the rate of RSV-associated in hospital deaths among infants aged < 6 months was 44 (95% CI 25–71) (Table 3).

Among Kenyan children aged < 5 years, the total annual number of RSV-associated outpatient ARI was estimated to be between 1,114,748 and 1,372,458, while the number of RSV-associated non-medically attended ARIs was estimated to be between 1,222,626 and 1,510,303 cases (Additional file 2: Table S10). The counties with most outpatient ARI cases were Kilifi, Meru, and Nakuru, while Samburu, Kirinyaga, and Lamu had the lowest, in that order. For non-medically attended ARI, Mandera, Wajir, and Kakamega were the leading counties, while Samburu, Kirinyaga, and Lamu had the least cases in that order.

The annual number of RSV-associated hospitalized SARI cases was estimated to be between 18,160 and 24,213, whereas the number of RSV-associated non-hospitalized SARI cases was 55,977 to 74,736 (Additional file 2: Table S10). Nakuru, Kilifi, and Nairobi had the highest number, while Lamu, Kirinyaga, and Taita–Taveta had the least number of RSV-associated hospitalized SARI in that order. We also estimated that the highest number of RSV-associated non-hospitalized SARI were highest in Nakuru, Mandera, and Kilifi and lowest in Lamu, Kirinyaga, and Vihiga counties.

Further, we estimate the annual number of RSV-associated deaths to be between 1914 and 3553 cases with in-hospital and out-of-hospital deaths ranging from 420 to 779 and from 1495 to 2774 cases, respectively (Additional file 2: Table S10). Nakuru County had the highest number of RSV-associated deaths, whereas Lamu County had the lowest number of RSV-associated deaths.