Introduction

Childhood diarrhoea remains a major concern in developing Bangladesh [1]. There is a strong association between childhood diarrhoeal disease and several climate factors. For example, an increasing number of diarrhoea cases due to Vibrio cholera has been observed with both high and low rainfall [2], while rising temperature as well as high and low rainfall both predict increasing non-cholera cases [3]. These might be closely related to raising of river levels [3] resulting in flooding that effects the most vulnerable groups in poor socio-economic areas and where water-satiation practices are are suboptimal [4]. Flooding is also associated with rapid land coverage (loss of arable land, habitat destruction and the decline in natural vegetation cover) in the Dhaka megacity [5–7]. Moreover, overcrowding, poor sewer systems and waste disposal with water stagnantion, increasing surface temperature and environmental pollution together [7] and concurrently facilitates other infections including typhoid and dengue fever in Dhaka [8–10].

Epidemiological studies suggested a strong association between childhood malnutrition and increased risk of infectious diarrhoea [11]. Among all causes of childhood diarrhoea, rotavirus is one of the most significant attributable pathogens [1] and cellular attachment with healthy cells in the brush border of the intestine is fundamental in the pathophysiology of the rotavirus infection [12–14]. As a result and, compared to undernourished children, rotavirus infection disproportionately affects well-nourished children [15,16].

Global warming with climate change effects ecosystems and might also predict emergence and re-emergence of various infections [17–19] [20]. Data from the long established diarrhoeal disease surveillance system (DDSS) [21] of the Dhaka Hospital at International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b) observed an association of an increased incidence of rotavirus diarrhoea with high temperature, low humidity and high river-levels [22] with seasonal variability [23, 24]. The facility also experienced a sustained rising trend of rotavirus infection over the last two decades [25] despite a significant improvement of water and sanitation practices in the capital city and its neighbouring catchment areas [24]. Similar to other countries, the profile of childhood malnutrition in Bangladesh has shifted over time with a more recent decrease in under nutrition and increase in over nutrition [24]. Thus, there might be an individual or combined contribution of changing climate factors and decrease in childhood under nutrition to the rising trend of rotavirus infection among children less than 5 years old.

The DDSS has maintained an around the clock diarrhoeal disease surveillance system for the last three decades [15,21,24]. This surveillance system prospectively collects information on diarrhoeal patients and the aetiology, including rotavirus. We have linked this data with different meteorological indicators such as temperature, rainfall, sea level pressure and humidity over the last two decades (1993 to 2012). This enables determination of the long term relationship of the changing pattern of childhood malnutrition, meteorological indicators, and other socio-demographic and sanitation factors with rotavirus diarrhoea in children less than 5 years of age in urban Dhaka, Bangladesh.

Materials and methods

Results

The yearly estimated number of under 5 years children with diarrhoea admitted to the hospital in 1993 was 5,206 and that increased to 6,892 in 2012. Similarly, the proportion of rotavirus cases during this period gradually increased from 20% in 1993 to 43% in 2012 (Chi-square for trend: p = 0.010) (Fig 1). An abrupt increase in rotavirus infection was observed in 1999 and that remained for the rest of the period.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Moving average of yearly estimated to total patients and under-5 years children, proportion of rotavirus, underweight, stunting, wasting, mean temperature, rainfall, sea level pressure and humidity (1993–2012).

https://doi.org/10.1371/journal.pone.0179418.g001

In contrast, the proportion of children who were underweight, stunted or wasted decreased from 1993 to 2012 from 58% to 29% (Chi squared for trend: p<0.001), 53% to 21% (p<0.001) and 32% to 22% (p = <0.001) respectively (Fig 1). During this time, the mean temperature increased from 25.76°C to 26.61°C (p = 0.07) while the mean rainfall, sea level pressure and humidity decreased from 234.92 mm to 111.75 mm (p = 0.5), 1008.30 mm of hg to 1006.61 mm of hg (p = 0.02) and 76.63% to 70.26% (p<0.001) respectively.

In the unadjusted ARIMA model (Table 1), there is significant association between monthly proportion of rotavirus infection and underweight. A decrease in one unit of the monthly proportion of underweight coincided with a 0.26 increase in the monthly proportion of rotavirus infection [(95% CI: -0.43, -0.094) p = 0.002]. The association remained when adjusted for monthly mean temperature, rainfall, sea level pressure and humidity (Table 1; model 1). When adjusted for years stratum, the effect of monthly change in proportion of underweight decreased to -0.24 (95% CI: -0.42, -0.06) (Table 1, model 2). Moreover, in the final model (Table 1, model 3) adjusted for other socio-demographic factors, it accounted for a-0.19 (95% CI:-0.376, -0.003). A similar association was also observed for wasting in which one unit decrease in monthly proportion of wasting resulted in an increase of the monthly proportion of rotavirus infection by -0.27 (95% CI: -0.455, -0.075) after adjusting for all the co-variates including meteorological indicators. However, there was no association with stunting (full model for underweight, stunting and wasting are describe in S3 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Association between the monthly proportion of rotavirus infection and underweight, stunting, wasting for under-5 children in urban Dhaka, Bangladesh.

Adjusted and unadjusted seasonal ARIMA models were fitted to the proportion of children with rotavirus infection. Covariates are outlined in the model definitions below.

https://doi.org/10.1371/journal.pone.0179418.t001

When interactions were modelled, the association between monthly rotavirus infection and underweight remained significant with a variation in effect size as expected and effect size ranged between -0.14 to -0.26 with an additional variability which ranged between 0.7% to 1.52% in rotavirus infection with unadjusted model by adding different interaction terms (see S4 Table). For wasting, the effect size ranged between -0.17 and -0.36 with a variability of 0.90% to 1.14% with unadjusted model (S6 Table). Details of interaction modeling for stunting are provided in S5 Table.

Discussion

Findings of the present study indicate a striking inverse relationship of wasting (acute malnutrition) and underweight (mixed acute and chronic malnutrition) with childhood rotavirus infection after controlling for different climate, socio-demographic and sanitation practices among under-5 years children in Dhaka. However, while a similar association with stunting (chronic malnutrition) was also observed it was not statistically significant. This nutrition interaction in which rotavirus disproportionately affects better nourished children, has been previously described [35]. Notably, this is in contrast to childhood malnutrition as an established risk factor for diarrhoeal disease due to most other enteric pathogens [36].

It is evident that rotavirus infection is associated with the variability of climate factors [22] with strong seasonal increases especially during cooler and dries months [23,32,33]. A one degree centigrade increase of temperature above a threshold (29 degrees centigrade) was associated with a 40% increase in rotavirus diarrhoea and there was a linear inverse relationship between the number of cases of rotavirus diarrhoea with relative humidity [22]. To our knowledge, this is the only study that estimated the effect of changing childhood malnutrition and its association with rotavirus infection after controlling for the ambient temperature, rainfall, sea level pressure and humidity (Model 1). These analyses indicate that not only climatic factors but also improved nutrition or over nutrition might have a significant role in the pathogenesis of rotavirus infection [12–14]. However, other man-made influences including unplanned urbanization with rapid land coverage and reduced green spaces in the Dhaka megacity and improper sewerage might be additional contributors to the causal path [5–7,37,38]. After adjustment for host (age and sex) and demographic (residence and under 5 children in the household) characteristics and sanitation practices, the significant association indicates a strong relationship between rotavirus infection and acute and chronic malnutrition.

Rotavirus, a major pathogens responsible for acute watery diarrhoea, perhaps requires healthy intestinal epithelium for attachment and pathogenesis [14]. In animal models, malnutrition results in a decreased number of cells and impaired epithelial proliferation in the small intestinal mucosa [12] that inhibit cellular attachment by the virus [13]. In addition to natural infection, it is plausible that a predilection of rotavirus for healthy, nutritionally intact intestinal epithelium is a factor in the observed immune response to Rotarix oral vaccine in which meant seroconversion rates to Rotarix were 86%, 75%, and 63% in high, middle, and low income countries respectively [39].

However, immune-compromised, severely malnourished children [11] are more prone to have infection rotavirus [16]. A reduction of acute malnutrition and, increases in the proportion of well-nourished children, as well as the recent trend towards an increasing prevalence of childhood obesity among under-5 children at DSSS [24], might positively contributed for an acute infection [15]. On the other hand, genetic variation might predict chronic malnutrition such as stunting [40], which may not be interlinked with acute infection [41].

In the present study, we observed that in the early 1990’s (1993) the proportion of all the malnutrition subtypes (underweight, wasting and stunting) were relatively high among the under5 yeras children and the proportion of rotavirus was low as was mean ambient temperature. However, from 1993 onwards, the proportion of under 5 years children with malnutrition gradually decreased and rotavirus diarrhoea increased. During 1999–2001, there was a sudden peak of rotavirus diarrhoea associated with a decreased mean level of all of the examined climate factors, in addition to reduced malnutrition rates which might have further affected the association. It is perhaps relevant that from 2002 onwards, the proportion of children under 5 years with malnutrition, mean relative humidity, rainfall and sea level pressure all decreased steadily whereas mean ambient temperature gradually increased. We employed a series of interaction models between all three components of malnutrition and climate factors and year stratum (Model 4–13; S4, S5 and S6 Tables) and estimated the variability of associations. The association largely remained significant with some degree of variability in the effect size.

Strengths and limitations

Considering methodologic strengths of the current study are the inclusion of long-term observations with unbiased systematic samples and the quality assessment of nutrition status by trained personnel, and quality laboratory performance with same method of detection of rotavirus over the study period. The adjusting for possible confounding socio-economic factors and performing seasonal ARIMA models with a series of likely interactions are the other strengths. However, we did not consider other potentially important factors like vegetation index, per-capita land, river water level, or land surface temperature which might influence the causal path in transition of rotavirus disease [8,9]. Hospital based data might not adequately represent the population at large and might have a bias towards patients with less severe rotavirus illness and who seek care less often at hospital facilities. Finally, patients enrolled in the DDSS were self-referred and there is no subject-specific information on of their conditions, including spatial variability.

Conclusion

Findings of the present study indicate an inverse association between acute childhood malnutrition and rotavirus diarrhoea among a population of children less than 5 years of age in urban Dhaka, Bangladesh after adjusting for important climate, socio-demographic and sanitation factors. An insignificant inverse association; however, was observed for chronic malnutrition. The relationship between recent nutritional transition and rotavirus infection in the mega city might be additionally influence by other man-made factors. As a result, an alternative intervention other than man made factors may be warranted such as mass vaccination may be needed to prevention and control of childhood rotavirus infection.

Supporting information

Acknowledgments

Hospital surveillance was funded by icddr,b and the Government of the People's Republic of Bangladesh through IHP-HNPRP. icddr,b acknowledges with gratitude the commitment of the Government of the People's Republic of Bangladesh to the icddr,b’s research efforts. icddr,b also gratefully acknowledges the following donors which provided unrestricted support to icddr,b since 1993 from when the data is used for this study: Australian Agency for International Development (AusAID), Government of the People's Republic of Bangladesh, Canadian International Development Agency (CIDA), Embassy of the Kingdom of the Netherlands (EKN), Swedish International Development Cooperation Agency (Sida), Swiss Agency for Development and Cooperation (SDC), and Department for International Development, UK (DFID).