More than one-third and one-quarter of the four million neonatal deaths each year worldwide are estimated to be due to severe infection and sepsis, respectively [1
]. The majority of these deaths are in developing countries, where home birth continues to be a norm and, unprotected against the environment in poor rural households, newborns are likely to be more vulnerable compared to older infants and adults [3
A wide range of morbidity and mortality has been shown to be influenced by the external environment, including changes in temperature, humidity, and rainfall throughout seasons. Previous studies have reported seasonal variation in the incidence of birth weight [4
], preterm births [9
], hypothermia [3
], viral and upper respiratory infections [3
], pneumonia [22
], typhoid [23
], skin infections [3
], eye infections [3
], encephalitis [22
], meningitis [24
], umbilical infections [3
], malaria [25
], cholera [28
], and non-cholera diarrhea [22
]. Focused on older age groups, the commonality of these hospital-based studies, where environmental factors are regulated to a certain extent, leave limited data on the seasonal variation of morbidities in newborns in the community. To our knowledge, a study in rural Gadchiroli district of India is the only study that specifically examined the association of newborn morbidities with specific environmental factors in a community setting [3
], and none has reported seasonal variation of newborn sepsis.
Countries in South Asia have adopted national strategies to strengthen management of serious neonatal illness, including serious infection, at the community level [42
]. These strategies, typically involving community-based surveillance for serious illness and subsequent management, are resource intensive [44
]. Understanding environmental factors associated with serious newborn illness will provide programmatically useful information to develop strategies for health education, surveillance, and preparedness. This study aimed to assess the association of serious newborn illness in the community with three environmental factors: temperature, rainfall, and humidity. We attempt to establish a programmatically relevant seasonality pattern, based on the levels of environmental factors and discuss potential causal and contributing pathways that explain the epidemiology presented.
provides the monthly averages of environmental factors for 24 months (2004–2005) in Sylhet and 36 months (2004–2006) in Mirzapur. Figure 2
plots these monthly values over the entire surveillance periods along with incidence of very severe disease (VSD) per 100 neonates. Generally, in both sites, May through October was the period when all indices were at their highest levels.
VSD incidences corresponded with temperature and humidity curves in both sites, but not with rainfall (Figure 2
). Figure 3
shows the unadjusted proportions of newborns with VSD by quintiles of each environmental variable. In Sylhet, clear linear associations were observed between VSD proportion and each environmental factor, whereas in Mirzapur the association was obvious only for temperature. Adjusted for covariates, however, only temperature was significantly associated with incidence of VSD in both sites; odds ratios (ORs) were 1.14 (95% CI: 1.08 to 1.21) in Sylhet and 1.06 (95% CI: 1.04 to 1.07) in Mirzapur (Table 2
—Model 1). For HHI, ORs were 1.06 (95% CI: 1.04 to 1.08) in Sylhet and 1.03 (95% CI: 1.01 to 1.04) in Mirzapur (Table 2
—Model 2). In both Sylhet and Mirzapur, newborns born in high-risk months had significantly higher odds of having VSD (ORs: 1.72 (95% CI: 1.32 to 2.23) in Sylhet and 1.62 (95% CI: 1.33 to 1.96) in Mirzapur) (Table 2
—Model 3). These months also corresponded to the highest levels of HHI in Sylhet (Table 1
confirms the higher incidence of VSD during the higher-risk, higher-temperature months. We further examined the individual danger signs by higher and lower risk months. Respiratory rates >70 breaths per minute and fever >101 °F were significantly higher in both sites during the high-risk months as compared to the rest of the year.
To our knowledge, this is one of the very few studies to show variations in very severe disease (VSD) in rural communities in the context of environmental factors. We explored seasonality of VSD by specific measures of environmental factors and found that, unadjusted, temperature, rainfall, humidity, and HHI levels were correlated with incidence of VSD in newborns delivered in rural homes. When adjusted for other background factors, temperature alone and HHI, largely driven by temperature, were found to be associated with incidence of VSD; however, the size of the effect was relatively small. Four consecutive months of the year in Sylhet and six in Mirzapur, with mean temperature levels above 28 °C and higher HHI, had significantly higher odds of incidence of VSD compared to the rest of the year, although the contribution of other confounding factors that were not controlled for in the analysis is unknown. Specific signs of severe disease, i.e., high respiratory rate and fever, were more common in these high-risk months.
As one of the very few studies examining the relationship between environmental factors and VSD in a community context, the high coverage of active surveillance for illness and daily measures of environmental factors over the entire surveillance period of three years provides a finer temporal resolution for correlating changes in environmental factors with incidence of VSD. However, this may mean the statistical associations found between temperature and humidity with odds of severe neonatal illness may have little practical relevance. Application of our findings to programmatic settings should be made with caution, since VSD associated with high case fatality occurs throughout the year and thus requires careful vigilance by the families and the health systems for every newborn. However, seasonal variation in illness was strong, and may be, in part, modulated by temperature and humidity, but likely involves other factors unaccounted for in this analysis.
Our findings should be interpreted considering the limitations that we used household building materials as proxy for a comprehensive wealth index and could not control for birth weight in the regression model. By study design, birth weight was collected in only one site and the time for measurement varied due to logistical constraints of reaching newborns at the earliest possible time. However, we controlled for gestational age.
Few studies have examined seasonal variations of sepsis occurrence; most examined sepsis in adolescents and adults [50
] and only one study was conducted in a developing country [50
]. Contrary to our findings in neonates in a tropical climate, one study conducted in England [53
] and two studies conducted in the United States [51
] in temperate climates showed increases in sepsis cases in adults in the winter months, which Danai et al
. attributed to increases in sepsis incidence occurring from an initial respiratory source [51
]. Other studies have shown that seasonality of disease and environmental fluctuations differ by latitude [35
], although this has not been examined for sepsis specifically. The SEARCH trial in Gadchiroli, with closest similarity with our settings showed no seasonal variation of suspected neonatal sepsis, but found higher incidences of hypothermia, upper respiratory symptoms, umbilical and bacterial skin infections in winter, and of unexplained fever in summer [3
]. Bang et al
. suggested that lack of protection from the effects of the environment played a major role in the seasonal patterns observed [3
Association of serious neonatal illness, and specifically serious neonatal infections, with environmental factors in a rural developing country setting can potentially be explained under two pathways, often sequential or overlapping: (1) pathogenic (bacterial or viral) and (2) behavioral. In both of our sites, VSD was found to be associated with increased temperature levels. In warmer and more humid months, it is likely that a variety of pathogens proliferate in the environment, as demonstrated for Klebsiella pneuminiae
, a common source of community-acquired infections [56
]. An increase in growth may lead to an increase in virulence or inoculums [57
], as well as colonization. The proposition that bacterial growth increases with increasing temperature and humidity is supported by numerous other studies [29
]. However, there are also controversies around the influence of temperature and humidity on bacterial pathogens. Laboratory studies show considerable differences on the effect of temperature or humidity on bacteria strains [62
], indicating that many factors outside of environmental causes are involved in disease transmission. Behavioral patterns are seasonal and could have contributed to contamination and infection transmission. Lack of ventilation in rooms during the summer, which is very common in rural Bangladesh, as well as an increase in the number of individuals indoors during the monsoon season, might have resulted in overcrowding and the transmission of pathogens [3
]. During the summer, neonates might have more contact with others and use less clothing, decreasing protection from the environment and increasing the spread of transmission [68