Background
Rotavirus is the most common cause of vaccine-preventable severe acute gastroenteritis (AGE) among infants and young children worldwide [
1,
2]. In 2013, RVAGE was estimated to cause 215,000 global deaths among children < 5 years of whom 2% (~ 4000) were from Kenya [
3] alone. In Kenya, RVAGE accounts for 19% (~ 9000) of annual hospitalizations among children < 5 years [
4]. Two RV vaccines Rotarix® (GlaxoSmithKline), and RotaTeq® (Merck & Co.), are approved and recommended by the World Health Organization (WHO) for global use [
5]. Efficacy and effectiveness studies of these vaccines have shown significant reduction in AGE and RVAGE associated hospitalizations and deaths among children < 5 years in both clinical trials and in settings where they have been incorporated into the national immunization programs [
1,
6‐
10]. Consistent with data from Mexico [
6] and Brazil [
7,
11], African countries that were early introducers of RV vaccines including Malawi [
10], Ghana [
12], and Rwanda [
13], have shown remarkable declines in childhood morbidity and mortality associated with AGE and RVAGE. Furthermore, the cost benefit of these vaccines has equally been demonstrated [
4,
7,
14] .RV vaccine (Rotarix®) was introduced into the Kenya national immunization program in July 2014. Recent population-based data on pre-vaccine disease rates are not available in Kenya. However, such data are needed to evaluate the impact of vaccination program and may help county and national level governments, regional and global decision makers with evidence needed to support investment in these vaccines.
We examined baseline rates of AGE and RVAGE specific hospitalizations and deaths among children < 5 years in rural western Kenya from January 1, 2010 to December 31, 2013 before RV vaccine introduction in Kenya.
Discussion
This study documents comprehensive, age-stratified population-based hospitalization and mortality rates associated with AGE and RVAGE before introduction of RV vaccines among Kenyan children < 5 years in a rural community whose demographic and healthcare seeking characteristics are well described [
16,
19]. Unlike other WHO rotavirus surveillance study sites in Africa, our hospital-based surveillance site for rotavirus is unique for a few reasons. First, it is supported by an ongoing HDSS which monitors population denominators and conducts verbal autopsy [
16]. Second, our surveillance hospital is the only regional public referral hospital for the local HDSS and the only in-patient facility within the HDSS, making our surveillance data representative of the population as shown from our current data and as previously observed [
15]. Furthermore, the advantages of a population-based incidence rate are two-fold. First, they provide an opportunity to estimate number of people affected by a disease. Second, they can help to project the number of illness episodes that can be prevented with effectively known interventions such as vaccines [
25].
Our 4-year study’s most important findings are that before RV vaccine introduction in Kenya; approximately 90 and 60% of RVAGE hospitalized children were aged < 2 years and < 1 year, respectively, and that hospitalizations and mortality associated with AGE and RVAGE were highest among infants. Furthermore our data suggests that children bearing the greatest burden of morbidity and mortality associated with AGE and RVAGE were infants aged 6–11 months. This finding is similar to observations from neighboring Sudan where pre-RV vaccine data indicates that 91 and 61% of rotavirus hospitalizations occurred before 2 years and 1 year respectively [
26]. Furthermore, our finding is consistent with observations from the first 2 years of the current study [
15], a study conducted at the coastal region of Kenya [
27], Global Enteric Multicenter Study (GEMS) [
2] and other studies conducted in Europe [
28] before introduction of RV vaccines. Our observation that 21% of hospitalizations among children < 5 years in the HDSS were due to AGE is similar to 23% reported previously from mid-term analysis of our current study [
15], 22% reported from Kilifi HDSS in coastal region of Kenya [
27], 21% from neighboring Mwanza region in Tanzania [
29], and 21% from Ethiopia [
30]. In addition, our finding that 26% of hospitalized AGE case patients were infected with RV remains similar to the rate of 27% reported from mid-term analysis of our current surveillance data [
15] and to 29% from Kilifi HDSS at the Kenyan coast [
27]. These observations suggest that AGE and RVAGE burden in our setting is comparable to those from other settings in Kenya and neighboring countries before RV vaccine introduction. Our observation that rates of hospitalization due AGE and RVAGE declined over the study period before vaccine introduction may be associated with unknown non-RV vaccine intervention factors. However, the proportion of all deaths that were associated with AGE and RVAGE did not follow the same pattern. Thus, these observed trends are difficult to explain, though in part may reflect the effects of other interventions. Although widespread distribution and use of zinc and ORS as part of devolved government development efforts in Kenya has been described [
31] and may be a contributing factor to the observed decline in diarrhea burden in this setting, such argument remains speculative and prompts further investigation. This trend however is consistent with other observations from a recent community-based survey conducted in this setting [
19], and is not dissimilar to the global trend of diarrhoea and rotavirus disease [
3,
32].
Our data show that rotavirus was more commonly detected among infants. Moreover, RVAGE presented with more severe episodes than non-RVAGE as characterized by severe dehydration, vomiting and low grade fever —an observation similar to other previous studies [
30,
33‐
35]. Rotavirus is the most common cause of severe dehydrating diarrhoea and is the leading pathogen associated with moderate-to-severe diarrhoea (MSD) [
35], as further reaffirmed by GEMS — the largest diarrhoea etiology case-control study ever conducted in countries representing the highest disease burden regions located in Africa and Asia [
2]. Severe dehydration caused by diarrhoea in children is a major cause of preventable morbidity and mortality in Kenya [
31]. As commonly observed in our setting and consistent with the caretakers healthcare seeking trends in Kenya [
19,
31], delay in seeking care for childhood diarrhoea and reducing amount of fluid and food intake during childhood AGE illness can lead to severe disease. Our current study found that case-fatality among RVAGE was not significantly different from non-RVAGE cases, suggesting that rotavirus may not be associated with mortality in hospital based studies as shown from other studies [
33,
36]. This finding supports the assumption that seeking care for RVAGE from a health care facility enables access to appropriate rehydration, which would then reduce the risk of death from the disease.
Understanding seasonality of rotavirus can help formulate hypothesis for assessing potential factors influencing transmission and guide policy makers in deciding on appropriate interventions and approaches that can work in local settings for improving case management during peak seasons [
37]. For example, in settings such as USA, rotavirus seasons have been observed to be delayed, shortened, and diminished [
4,
38] after vaccine introduction. In our current analysis, rotavirus detection peaked in months which are locally known to be usually warm and dry. Our current findings are consistent with recent observations from Kenya [
15,
33], and remains similar to findings from other studies conducted in Burkina Faso [
37], Peru and Bangladesh [
39] before rotavirus vaccine introduction in those settings. Although there is no unifying rotavirus seasonality pattern globally [
40], it’s spread by the faecal-oral route remains agreeable [
35], and even airborne or droplet transmission has been postulated [
41]. The later attribute potentially makes the virus transmission route also to resemble that of other non-enteric respiratory infectious diseases such as measles [
42].These observations suggests that a drop in humidity and rainfall combined with dry soil could potentially increase additional chance for transmission through aerial contaminated faecal materials since survival of rotavirus may still be favored in such conditions as described elsewhere [
41,
43].
Treating RVAGE is expensive. In Kenya, it has been estimated that rotavirus disease cost the national healthcare system $10.8 million each year, and that a 2-dose rotavirus vaccine (RVV) series can avert ~ 2500 deaths, ~ 6000 hospitalizations and ~ 860,000 clinic visits with a cost saving of $2.1 million annually [
4].RV vaccines have been shown to be effective in reducing the hospitalizations and death due to diarrhoea in children and the protective effect potentially lasts through 2nd year of life [
1,
44].While the benefit of these vaccines has been documented in other African countries where they were introduced ahead of Kenya, such as in South Africa [
45], Rwanda [
3], Ghana [
12],and Togo [
46], population-level benefits of RVV are yet to be demonstrated from Kenya.
A possible limitation of our current study is that many rotavirus-associated fatalities are likely associated with delay in healthcare seeking [
5]. Furthermore, VA relies on signs, symptoms and circumstances prior to death to assign cause of death which is subject to misclassification error, and therefore the method as applied in our current study may lead to over or under-estimation of mortality [
36]. Our methodology for estimating diarrhoea deaths attributable to rotavirus was based on the following 3 assumptions: (i) that in the absence of treatment, the hospitalized severe cases would not have survived; (ii) the treatment effect on survival of severe diarrhoea is equal for rotavirus and non-rotavirus diarrhoea; and that (iii) the rotavirus attributable fraction of severe diarrhoea observed in the sentinel hospital are generalizable to the source population within each age stratum as already described elsewhere [
36]. Maintaining caution when interpreting these estimates is important since we recognize that such assumptions may affect the validity and generalizability of the estimates to the general population. However, since there are currently no reliable data for the direct measurement of the proportion of diarrhoea deaths that are attributable to rotavirus [
22,
36] especially in the high disease burden regions located mostly in low-and middle income countries such as in our setting [
3], we believe our methodology remains more reasonable, robust and applicable as recommended by WHO [
22,
36]. Moreover, our current hospital surveillance data suggests agreeable representation of the source population consistent with previous observations [
15].
Acknowledgements
This study includes data generated by the KEMRI and CDC joint operated HDSS, which is a member of the International Network for the Demographic Evaluation of Populations and their Health (INDEPTH). We acknowledge the contributions of and thanks to; WHO country office in Kenya and WHO-AFRO office based in Brazzaville; thanks to Ministry of Health staff for supporting the data collection and processing; the entire KEMRI and CDC study team at SCRH. Special thanks to Dr. Daniel R. Feikin and, Dr. Kayla F. Laserson (CDC), Dr. Amek Nyagwara (KEMRI-CGHR), Charles Mwitherero (WHO), Caroline Maina (MoH, Kenya), Alice Ngereso (WHO), Prof. Kirsi Lumbe Sandat, Tiina Kangasala and Catarina. Stahle-Nieminen (University of Tampere), Linet Aluoch Sewe, Collins Okello, Pamela Kanga, Peter Jaron and Ken Ruttoh (KEMRI-CGHR) for supporting the study operations. We are grateful to the caretakers in the Karemo, Asembo and Gem communities who participated in this work. This manuscript is published with the approval of the Director, KEMRI.