• View in gallery

    Monthly distribution of diarrhea incidence (cases per month), mean daily temperature (°C), aridity index (minimum daily temperature/humidity), and rainfall (millimeters per month) from January to December (averaged over the study period for all districts).

  • View in gallery

    Spatial distribution of population-adjusted average monthly diarrhea (top left), average annual temperature (top right), average annual aridity index (bottom right), and average annual rainfall levels (bottom left) in 2014. Both diarrhea and rainfall were log transformed for better contrast with color scales.

  • View in gallery

    Comparison of fitted vs. observed diarrhea incidence rate in three provinces (Herat, Kabul, and Balkh). Maps illustrate the average for the period from January–September 2016, whereas graphs represent the correlation of fitted vs. observed data points for the same period.

  • 1.

    McMichael AJ, Woodruff RE, Hales S, 2006. Climate change and human health: present and future risks. Lancet 367: 859869.

  • 2.

    Patz JA, Epstein PR, Burke TA, Balbus JM, 1996. Global climate change and emerging infectious diseases. JAMA 257: 217223.

  • 3.

    Patz JA, Campbell-Lendrum D, Holloway T, Foley JA, 2005. Impact of regional climate change on human health. Nature 438: 310317.

  • 4.

    Serdeczny O, Adams S, Baarsch F, Coumou D, Robinson A, Hare W, Schaeffer M, Perrette M, Reinhardt J, 2017. Climate change impacts in sub-Saharan Africa: from physical changes to their social repercussions. Reg Environ Change 17: 15851600.

    • Search Google Scholar
    • Export Citation
  • 5.

    Jepsen MR, Simonsen J, Ethelberg S ,2009. Spatio-temporal cluster analysis of the incidence of Campylobacter cases and patients with general diarrhea in a Danish county, 1995–2004. Int J Health Geogr 8: 11.

    • Search Google Scholar
    • Export Citation
  • 6.

    Carlton EJ, Woster AP, DeWitt P, Goldstein RS, Levy K, 2015. A systematic review and meta-analysis of ambient temperature and diarrhoeal diseases. Int J Epidemiol 45: 117130.

    • Search Google Scholar
    • Export Citation
  • 7.

    Herrador BRG, De Blasio BF, MacDonald E, Nichols G, Sudre B, Vold L, Semenza JC, Nygård K, 2015. Analytical studies assessing the association between extreme precipitation or temperature and drinking water-related waterborne infections: a review. Environ Health 14: 29.

    • Search Google Scholar
    • Export Citation
  • 8.

    Carlton EJ, Eisenberg JN, Goldstick J, Cevallos W, Trostle J, Levy K, 2013. Heavy rainfall events and diarrhea incidence: the role of social and environmental factors. Am J Epidemiol 179: 344352.

    • Search Google Scholar
    • Export Citation
  • 9.

    Troeger C et al. 2018. Global disability-adjusted life-year estimates of long-term health burden and undernutrition attributable to diarrhoeal diseases in children younger than 5 years. Lancet Glob Health 6: e255e269.

    • Search Google Scholar
    • Export Citation
  • 10.

    Bartsch SM, Lopman BA, Ozawa S, Hall AJ, Lee BY, 2016. Global economic burden of norovirus gastroenteritis. PLoS One 11: e0151219.

  • 11.

    Lelieveld J et al. 2012. Climate change and impacts in the eastern Mediterranean and the Middle East. Clim Change 114: 667687.

  • 12.

    Lloyd SJ, Kovats RS, Armstrong BG, 2007. Global diarrhoea morbidity, weather and climate. Clim Res 34: 119127.

  • 13.

    Khalil I et al. 2016. Burden of diarrhea in the eastern Mediterranean region, 1990–2013: findings from the global burden of disease study 2013. Am J Trop Med Hyg 95: 13191329.

    • Search Google Scholar
    • Export Citation
  • 14.

    Troeger C et al. 2018. Estimates of the global, regional, and national morbidity, mortality, and aetiologies of diarrhoea in 195 countries: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Infect Dis 18: 12111228.

    • Search Google Scholar
    • Export Citation
  • 15.

    Anwar MY, Burnham G, 2016. Trends in infectious disease incidence among children in Afghanistan at a time of public health services expansion. EMHJ 22: 778785.

    • Search Google Scholar
    • Export Citation
  • 16.

    Wallace MR, Hale BR, Utz GC, Olson PE, Earhart KC, Thornton SA, Hyams KC, 2002. Endemic infectious diseases of Afghanistan. Clin Infect Dis 34(Suppl 5): S171S207.

    • Search Google Scholar
    • Export Citation
  • 17.

    World Health Organization, 2017. World Health Statistics 2017: Monitoring Health for the SDGs. Geneva, Switzerland: WHO. Available at: https://www.who.int/gho/publications/world_health_statistics/2017/en/. Accessed July 30, 2017.

    • Search Google Scholar
    • Export Citation
  • 18.

    Raleigh C, Jordan L, Salehyan I, 2008. Assessing the impact of climate change on migration and conflict. Paper Commissioned by the World Bank Group for the Social Dimensions of Climate Change Workshop, Washington, DC: The Social Development Department, The World Bank Group, 5–6. Available at: https://www.researchgate.net/profile/Clionadh_Raleigh/publication/255519298_Assessing_the_Impact_of_Climate_Change_on_Migration_and_Conflict/links/58c6a15392851c0ccbff63fb/Assessing-the-Impact-of-Climate-Change-on-Migration-and-Conflict.pdf.

  • 19.

    Mendelsohn R, Dinar A, Williams L, 2006. The distributional impact of climate change on rich and poor countries. Environ Dev Econ 11: 159178.

    • Search Google Scholar
    • Export Citation
  • 20.

    Aich V, Akhundzadah N, Knuerr A, Khoshbeen A, Hattermann F, Paeth H, Scanlon A, Paton E, 2017. Climate change in Afghanistan deduced from reanalysis and coordinated regional climate downscaling experiment (Cordex)—south Asia simulations. Climate 5: 38.

    • Search Google Scholar
    • Export Citation
  • 21.

    Wagner AL, Mubarak MY, Johnson LE, Porth JM, Yousif JE, Boulton ML, 2017. Trends of vaccine-preventable diseases in Afghanistan from the disease early warning system, 2009–2015. PLoS One 12: e0178677.

    • Search Google Scholar
    • Export Citation
  • 22.

    Lowe R, Bailey TC, Stephenson DB, Graham RJ, Coelho CA, Carvalho MS, Barcellos C, 2011. Spatio-temporal modelling of climate-sensitive disease risk: towards an early warning system for dengue in Brazil. Comput Geosci 37: 371381.

    • Search Google Scholar
    • Export Citation
  • 23.

    Alexander KA, Carzolio M, Goodin D, Vance E, 2013. Climate change is likely to worsen the public health threat of diarrheal disease in Botswana. Int J Environ Res Public Health 10: 12021230.

    • Search Google Scholar
    • Export Citation
  • 24.

    Starr JM, Campbell A, Renshaw E, Poxton IR, Gibson GJ, 2009. Spatio-temporal stochastic modelling of Clostridium difficile. J Hosp Infect 71: 4956.

    • Search Google Scholar
    • Export Citation
  • 25.

    Khalidi NA, 1989. Demographic profile of Afghanistan. Canberra, ACT, Australia. International Population Dynamics Program, Dept. of Demography, Research School of Social Sciences, the Australian National University. Available at: https://www.academia.edu/11860590/Demographic_Profile_of_Afghanistan_1979.

    • Search Google Scholar
    • Export Citation
  • 26.

    Fonseca PA, Hacon SD, Reis VL, Costa D, Brown IF, 2016. Using satellite data to study the relationship between rainfall and diarrheal diseases in a southwestern Amazon basin. Cien Saude Colet 21: 731742.

    • Search Google Scholar
    • Export Citation
  • 27.

    Konno T, Suzuki H, Katsushima N, Imai A, Tazawa F, Kutsuzawa T, Kitaoka S, Sakamoto M, Yazaki N, Ishida N, 1983. Influence of temperature and relative humidity on human rotavirus infection in Japan. J Infect Dis 147: 125128.

    • Search Google Scholar
    • Export Citation
  • 28.

    Xu Z, Liu Y, Ma Z, Toloo GS, Hu W, Tong S, 2014. Assessment of the temperature effect on childhood diarrhea using satellite imagery. Sci Rep 4: 5389.

  • 29.

    Prasetyo D, Ermaya Y, Martiza I, Yati S, 2015. Correlation between climate variations and rotavirus diarrhea in under-five children in Bandung. Asian Pac J Trop Dis 5: 908911.

    • Search Google Scholar
    • Export Citation
  • 30.

    Curtis SL, Hossain M, 1998. The effect of aridity zone on child nutritional status. Calverton, MD: Macro International. Available at: https://www.popline.org/node/530081.

  • 31.

    Colston JM et al. 2018. Evaluating meteorological data from weather stations, and from satellites and global models for a multi-site epidemiological study. Environ Res 165: 91109.

    • Search Google Scholar
    • Export Citation
  • 32.

    Shi L, Liu P, Kloog I, Lee M, Kosheleva A, Schwartz J, 2016. Estimating daily air temperature across the southeastern United States using high-resolution satellite data: a statistical modeling study. Environ Res 146: 5158.

    • Search Google Scholar
    • Export Citation
  • 33.

    dos Reis J, Rennó C, Lopes E, 2017. Validation of satellite rainfall products over a mountainous watershed in a humid subtropical climate region of Brazil. Remote Sens 9: 1240.

    • Search Google Scholar
    • Export Citation
  • 34.

    Kolstad EW, Johansson KA, 2011. Uncertainties associated with quantifying climate change impacts on human health: a case study for diarrhea. Environ Health Perspect 119: 299305.

    • Search Google Scholar
    • Export Citation
  • 35.

    Wang Y, Gu J, 2015. A hybrid prediction model applied to diarrhea time series. Fuzzy Systems and Knowledge Discovery FSKD). 2015 12th International Conference on Fuzzy Systems and Knowledge Discovery (FSKD), Zhangjiajie, China, 1096–1102. Available at: https://ieeexplore.ieee.org/abstract/document/7382095.

    • Search Google Scholar
    • Export Citation
  • 36.

    Liu Y, Wang X, Pang C, Yuan Z, Li H, Xue F, 2015. Spatio-temporal analysis of the relationship between climate and hand, foot, and mouth disease in Shandong province, China, 2008–2012. BMC Infect Dis 15: 146.

    • Search Google Scholar
    • Export Citation
  • 37.

    Raghavan RK, Hanlon CA, Goodin DG, Davis R, Moore M, Moore S, Anderson GA, 2016. Bayesian spatiotemporal pattern and eco-climatological drivers of striped skunk rabies in the North Central Plains. PLoS Negl Trop Dis 10: e0004632.

    • Search Google Scholar
    • Export Citation
  • 38.

    Chib S, Winkelmann R, 2001. Markov chain Monte Carlo analysis of correlated count data. J Bus Econ Stat 19:428435.

  • 39.

    Lee D, Rushworth A, Napier G, 2016. CARBayesST Version 2.2: An R Package for Spatio-temporal Areal Unit Modelling with Conditional Autoregressive Priors. Technical Report. University of Glasgow. Available at: https://cran.r-project.org/web/packages/CARBayesST/vignettes/CARBayesST.pdf.

    • Search Google Scholar
    • Export Citation
  • 40.

    Leroux BG, Lei X, Breslow N, 2000. Estimation of disease rates in small areas: a new mixed model for spatial dependence. Statistical Models in Epidemiology, the Environment, and Clinical Trials. New York, NY: Springer, 179191.

    • Search Google Scholar
    • Export Citation
  • 41.

    Rushworth A, Lee D, Mitchell R, 2014. A spatio-temporal model for estimating the long-term effects of air pollution on respiratory hospital admissions in Greater London. Spat Spatiotemporal Epidemiol 10: 2938.

    • Search Google Scholar
    • Export Citation
  • 42.

    Congdon P, 2005. Bayesian Models for Categorical Data. Hoboken, NJ: John Wiley & Sons. Available at: https://onlinelibrary.wiley.com/doi/book/10.1002/0470092394.

    • Search Google Scholar
    • Export Citation
  • 43.

    Geweke J, 1991. Evaluating the Accuracy of Sampling-Based Approaches to the Calculation of Posterior Moments. Minneapolis, MN: Federal Reserve Bank of Minneapolis, V196.

    • Search Google Scholar
    • Export Citation
  • 44.

    Checkley W, Epstein LD, Gilman RH, Figueroa D, Cama RI, Patz JA, Black RE, 2000. Effects of EI Niño and ambient temperature on hospital admissions for diarrhoeal diseases in Peruvian children. Lancet 355:442450.

    • Search Google Scholar
    • Export Citation
  • 45.

    Zhou X, Zhou Y, Chen R, Ma W, Deng H, Kan H, 2013. High temperature as a risk factor for infectious diarrhea in Shanghai, China. J Epidemiol 23: 418423.

    • Search Google Scholar
    • Export Citation
  • 46.

    Azage M, Kumie A, Worku A, Bagtzoglou AC, Anagnostou E, 2017. Effect of climatic variability on childhood diarrhea and its high risk periods in northwestern parts of Ethiopia. PLoS One 12: e0186933.

    • Search Google Scholar
    • Export Citation
  • 47.

    Metcalf CJE, Walter KS, Wesolowski A, Buckee CO, Shevliakova E, Tatem AJ, Boos WR, Weinberger DM, Pitzer VE, 2017. Identifying climate drivers of infectious disease dynamics: recent advances and challenges ahead. Proc Biol Sci. 284: 20170901.

    • Search Google Scholar
    • Export Citation
  • 48.

    Bentham G, Langford IH, 2001. Environmental temperatures and the incidence of food poisoning in England and Wales. Int J Biometeorol 45: 2226.

    • Search Google Scholar
    • Export Citation
  • 49.

    Black RE, Lanata CF, 1995. Epidemiology of diarrheal diseases in developing Countries. Infections of the Gastrointestinal Tract. New York, NY: Raven Press, 1336.

    • Search Google Scholar
    • Export Citation
  • 50.

    Ansari SA, Springthorpe VS, Sattar SA, 1991. Survival and vehicular spread of human rotaviruses: possible relation to seasonality of outbreaks. Rev Infect Dis 13: 448461.

    • Search Google Scholar
    • Export Citation
  • 51.

    McEldowney S, Fletcher M, 1988. The effect of temperature and relative humidity on the survival of bacteria attached to dry solid surfaces. Lett Appl Microbiol 7: 8386.

    • Search Google Scholar
    • Export Citation
  • 52.

    Yue RPH, Lee HF, 2018. Pre-industrial plague transmission is mediated by the synergistic effect of temperature and aridity index. BMC Infect Dis 18: 134.

    • Search Google Scholar
    • Export Citation
  • 53.

    Drayna P, McLellan SL, Simpson P, Li SH, Gorelick MH, 2010. Association between rainfall and pediatric emergency department visits for acute gastrointestinal illness. Environ Health Perspect 118: 14391443.

    • Search Google Scholar
    • Export Citation
  • 54.

    Qureshi AS, 2002. Water resources management in Afghanistan: the issues and options. IWMI 49.

  • 55.

    Bhavnani D, Goldstick JE, Cevallos W, Trueba G, Eisenberg JN, 2014. Impact of rainfall on diarrheal disease risk associated with unimproved water and sanitation. Am J Trop Med Hyg 90: 705711.

    • Search Google Scholar
    • Export Citation
  • 56.

    Bandyopadhyay S, Kanji S, Wang L, 2012. The impact of rainfall and temperature variation on diarrheal prevalence in sub-Saharan Africa. Appl Geogr 33: 6372.

    • Search Google Scholar
    • Export Citation
  • 57.

    Singh RB, Hales S, De Wet N, Raj R, Hearnden M, Weinstein P, 2001. The influence of climate variation and change on diarrheal disease in the Pacific Islands. Environ Health Perspectives 109:155159.

    • Search Google Scholar
    • Export Citation
  • 58.

    Pitzer VE, Viboud C, Lopman BA, Patel MM, Parashar UD, Grenfell BT, 2011. Influence of birth rates and transmission rates on the global seasonality of rotavirus incidence. J R Soc Interface 8:15841593.

    • Search Google Scholar
    • Export Citation
  • 59.

    Anwari P, Debellut F, Pecenka C, Parwiz SM, Clark A, Groman D, Safi N, 2017. Potential impact and cost-effectiveness of rotavirus vaccination in Afghanistan. Vaccine 36: 77697774.

    • Search Google Scholar
    • Export Citation
  • 60.

    Anwari P, Safi N, Payne DC, Jennings MC, Rasikh S, Waciqi AS, Parwiz SM, 2018. Rotavirus is the leading cause of hospitalizations for severe acute gastroenteritis among Afghan children< 5 years old. Vaccine 36: 77657768.

    • Search Google Scholar
    • Export Citation
  • 61.

    Levy K, Hubbard AE, Eisenberg JN, 2008. Seasonality of rotavirus disease in the tropics: a systematic review and meta-analysis. Int J Epidemiol 38: 14871496.

    • Search Google Scholar
    • Export Citation
  • 62.

    Patel MM, Pitzer V, Alonso WJ, Vera D, Lopman B, Tate J, Viboud C, Parashar UD, 2013. Global seasonality of rotavirus disease. Pediatr Infect Dis J 32: e134e137.

    • Search Google Scholar
    • Export Citation
  • 63.

    Frost A, Wilkinson M, Boyle P, Patel P, Sullivan R, 2016. An assessment of the barriers to accessing the Basic Package of Health Services (BPHS) in Afghanistan: was the BPHS a success? Global Health 12: 71.

    • Search Google Scholar
    • Export Citation
  • 64.

    Mokdad A et al. 2018. Burden of diarrhea in the eastern Mediterranean region, 1990–2015: findings from the Global Burden of Disease 2015 study. Int J Public Health 63: 109121.

    • Search Google Scholar
    • Export Citation
  • 65.

    Lioubimtseva E, Cole R, 2006. Uncertainties of climate change in arid environments of Central Asia. Rev Fish Sci 14: 2949.

  • 66.

    Wei K, Wang L, 2013. Reexamination of the aridity conditions in arid Northwestern China for the last decade. J Clim 26: 95949602.

  • 67.

    Patz JA, Grabow ML, Limaye VS, 2014. When it rains, it pours: future climate extremes and health. Ann Glob Health 80: 332344.

Past two years Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 753 275 32
PDF Downloads 374 137 18
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

 

 

Diarrhea Patterns and Climate: A Spatiotemporal Bayesian Hierarchical Analysis of Diarrheal Disease in Afghanistan

View More View Less
  • 1 Department of Epidemiology, University of Louisville School of Public Health and Information Sciences, Louisville, Kentucky;
  • | 2 Department of Biostatistics, Yale School of Public Health, New Haven, Connecticut;
  • | 3 Department of Epidemiology of Microbial Diseases, Yale School of Public Health, New Haven, Connecticut

Subject to a high burden of diarrheal diseases, Afghanistan is also susceptible to climate change. This study investigated the spatiotemporal distribution of diarrheal disease in the country and how associated it is with climate variables. Using monthly aggregated new cases of acute diarrhea reported between 2010 and 2016 and monthly averaged climate data at the district level, we fitted a hierarchical Bayesian spatiotemporal statistical model. We found aridity and mean daily temperature were positively associated with diarrhea incidence; every 1°C increase in mean daily temperature and 0.01-unit change in the aridity index were associated with a 0.70% (CI: 0.67%, 0.73%) increase and a 4.79% (CI: 4.30%, 5.26%) increase in the risk of diarrhea, respectively. Average annual temperature, on the other hand, was negatively associated, with a 3.7% (CI: 3.74%, 3.68) decrease in risk for every degree Celsius increase in annual average temperature. Temporally, most districts exhibited similar seasonal trends, with incidence peaking in summer, except for the eastern region where differences in climate patterns and population density may be associated with high rates of diarrhea throughout the year. The results from this study highlight the significant role of climate in shaping diarrheal patterns in Afghanistan, allowing policymakers to account for potential impacts of climate change in their public health assessments.

INTRODUCTION

The role of climate in shaping the dynamics of human diseases is under increased scrutiny in recent years,1 as a number of studies illustrate the potential for global2 and regional3 impacts of changes in climate on population health. The impact of climate on the distribution of diseases, infectious diseases in particular, is a vivid demonstration of how changing climatic factors could have implications for the population well-being.4 Acute diarrhea is one of the conditions that has been shown to exhibit a spatial and temporal distribution linked with climatic variables,5 but there is conflicting evidence regarding associations with temperature and precipitation. Whereas all-cause and bacterial diarrhea incidence tends to be associated with warmer temperatures, viral causes of diarrhea tend to exhibit negative associations with temperature,6 and both positive and negative associations between diarrhea and precipitation have been observed.7,8 Understanding dynamics of this interaction is important because diarrhea is a leading cause of morbidity and mortality, with more than 55 million years of life lost among children < 5 years of age attributed to diarrhea in 2016 alone.9 Hence, alteration to this pattern could pose a significant challenge to public health globally, as well as adding economic costs on top of tens of billions of dollars presently spent annually.10

For susceptible regions like the eastern Mediterranean, where changes in climate are predicted to be more severe,11,12 the urgency to understand diarrheal patterns is critical as these places already experience a disproportionately higher burden of disease13; the latest report estimates that the diarrhea-related death rate was significantly higher in parts of the eastern Mediterranean than the global average, particularly among vulnerable groups.14 Afghanistan is typical of countries situated in elevated risk regions with a less developed economy. Diarrhea is the second most prevalent disease (behind only respiratory infections) in a country that is home to numerous endemic infections,16 most with high mortality rates.17 Diarrhea is estimated to cause ∼20% of morbidities reported among children < 5 years of age and 11% of morbidities in all age-groups attending health facilities in Afghanistan in 2013.15

Given the high rate of diarrheal disease and susceptibility of the region to climate change,18,19 where average temperature in the period between 2006 and 2099 is predicted to increase from 1.7°C to as high as 6.4°C (depending on global carbon emission projections) and precipitation is predicted to decrease 20, it is of particular interest to understand the distribution and spatiotemporal dynamics of diarrheal disease in Afghanistan and how they are associated with climate variables. One way to quantify these associations is to statistically model the distribution of acute diarrhea across space and time, adjusted for climatic factors. Although past studies had noted seasonal patterns of diarrheal diseases peaking in the summer in the country,21 no studies to our knowledge have explicitly examined the association between climate variables and diarrhea incidence at the subnational level in Afghanistan. Such climate-sensitive models can be integrated into disease early warning systems22 and provide a platform for analyzing future trends.23

In this work, we use a hierarchical statistical modeling approach, fitted in the Bayesian setting,24 to investigate the spatiotemporal distribution of acute diarrhea in Afghanistan and to quantify the role that climatic variables play in shaping the pattern of diarrhea across the country. Such a model can be used to measure whether and how any change in climate could alter the distribution of acute diarrhea in the country, allowing for policymakers to account for potential impacts of climate change in their public health assessments, and provides a framework that could be extended to similar settings.

METHODS

Diarrhea data.

Monthly data on the number of diarrhea cases were provided by the Ministry of Public Health in Afghanistan, based on reports from April 2010 through September 2016. April is the start of solar Hijri Shamsi calendar year presently in use in the country. The data include reports from more than 2000 health facilities, covering practically all parts of the country’s territory. All cases were included, regardless of whether sourced from public or private health facilities. Incident diarrhea cases diagnosed at designated facilities are reported at monthly intervals to a national database under supervision of health management information system department of Ministry of Public Health; recurrent (i.e., occurring within 1 week of first diagnosis for the same individual) and chronic cases (persisting over 4 weeks from the first diagnosis) were excluded in the reporting process. In our study, diarrheal cases were aggregated to the district level to ensure each spatial unit includes at least one municipality with a sizeable number of residents; thus, every data point in the study represents the total number of cases per month for a selected district. Presently, more than 400 districts in Afghanistan are named by various sources, but data were analyzed using the 2005 administrative classification (n = 398), which effectively covers all parts of the country. Even though no national census has been conducted since 1979,25 updated annual district data from the Afghanistan Central Statistics Office (cso.org.af) were used as the closest available approximation of district population sizes. Monthly population size for a given district was estimated from the annual data using linear interpolation. This process was performed for each district separately.

Climatic data.

Existing literature highlights the role of rainfall,26 temperature,27 change in maximum or minimum temperature,28 humidity,29 and aridity30 as potential indicators for diarrhea patterns. Therefore, in the absence of land-based weather data, several satellite-based measurements of meteorological variables were used as potential predictors of diarrhea cases in a district across time. Data were extracted from the publicly available Earth Observing System Data and Information System that tallies and integrates meteorological and environmental measurements provided through orbital satellites operated primarily by the National Ocean and Atmospheric Administration of the United States, including rainfall (millimeters/month) [sourced from Tropical Rainfall Measuring Mission satellite (TRMM)], monthly averaged daily mean temperature (at 2 m above the earth’s surface), monthly averaged daily maximum temperature, and monthly averaged daily minimum temperature (originating in The Modern-Era Retrospective analysis for Research and Applications, Version 2 [MERRA-2] database). Although a degree of caution has been advised when considering satellite-based meteorological observation (mostly because of underestimation),31 both MERRA-based near-surface temperature measurements32 and TRMM-based monthly average precipitation readings33 were found to be highly correlated with land-based observations. We extracted the lowest available spatial resolution of 0.5 × 0.625 degrees (∼56 × 58 km) for all variables except rainfall, which had smaller resolution data available for the region (0.25 × 0.25 degree or ∼28 × 23 km). Direct measurement of aridity was unavailable. Instead, a composite index23 was synthesized by taking the ratio of daily minimum temperature over specific humidity, which provides a rough measure of dryness less dependent on temperature; the result was then scaled by multiplying by 106 to obtain a more interpretable fractional unit (0.01).

Because impacts of climate factors may vary as topography changes, two additional variables were constructed to account for terrain and landscape variability: Diurnal temperature variation (daily maximum temperature minus daily minimum temperature) and Average Annual temperature (average of mean daily temperature in a selected year). Both variables can be considered as surrogates for terrain, where diurnal temperature variation is expected to be greatest in dry and flat settings, whereas average annual temperature should be lower for mountainous regions. Diarrhea and climate data are provided as additional files in the Supplemental Material.

Statistical methods.

We initially analyzed the baseline distribution of diarrhea as well as climate variables to explore their trends, seasonality, peaks, and geographic characteristics. To avoid multicollinearity between the predictor variables, we calculated the Pearson correlation coefficient between each pair of predictor variables to identify highly correlated variables and removed covariates found to be highly correlated (r > 0.9) with the main predictors (mean daily temperature, rainfall, and aridity). We then used the Poisson regression to examine the univariate associations between climatic predictors (mean daily temperature, aridity, rainfall, average annual temperature, and diurnal temperature variation) with diarrhea incidence; district-level population (log-scale) was included as an offset term. We also included time since the start of the study (in months) as a predictor to test for any (linear) trends in diarrhea incidence.

Spatiotemporal statistical approach.

Failing to account for spatial and temporal correlations when statistically modeling space–time data may lead to inaccurate uncertainty quantification for the estimated parameters and, therefore, incorrect statistical inference for the associations of interest. In our setting, it is possible that there may be more similarities in diarrheal risk for neighboring districts than those further apart and at months closer in time even after adjusting for the climatic predictors; therefore, it is important to investigate this possibility and to account for the remaining spatiotemporal correlation. A variety of statistical tools have been developed to capture the effect of climatic factors on diarrhea patterns,8,34,35 but we chose to use a Bayesian hierarchical approach.24 The use of Bayesian hierarchical disease mapping techniques, particularly in conjunction with climatic factors, is on the rise in recent years.36,37 The increasing availability of software packages that can efficiently fit hierarchical models has made it possible to practically select this as the preferred platform to assess and quantify the role that climate variables play in shaping the geographic and temporal distribution of diarrhea across the country.

Model selection.

We used the following statistical framework for count data first proposed by Chib and Winkelmann38 as the basis for accounting for spatiotemporal correlation in the model:
YtdPoisson(μtd);ln(μtd)=XtdTβ+ln(populationtd)+ψtd, 
where Ytd is the number of observed diarrhea cases during the time point t in the district d, μtd is the expected number of diarrhea cases during the time point t in the district d, XtdTβ is the vector of covariates multiplied by their respective regression coefficients, populationtd is the population during the time point t in the district d, and ψtd captures the latent spatiotemporal correlation (structured random effects). As diarrhea cases are aggregated at the district level, the random effects need to be defined similarly.

In our initial analysis, we observed heterogeneous clusters of districts that exhibited consistently different risk levels relative to their adjacent geographical regions. The observation necessitated a flexible structure for the spatiotemporal random effects, whereby districts would be similar (auto-correlated) only if they are in the same cluster (risk-level group), but different if otherwise. In a recently developed R package (R Core Team, Vienna, Austria), CARBayesST (Lee & Rushworth, Glasgow, UK),39 a platform for efficiently fitting a range of statistical models, has been developed for investigating space–time patterns in areal-level data. This includes a configuration that allows for detection of clusters that have different levels of risk compared with their spatial and temporal neighbors,40 accommodating both spatial surface smoothing and a local error process. This model, first proposed by Lee and Lawson,30 is similar to the method proposed by Rushworth et al.,41 where a set of spatially and temporally auto-correlated latent effects is assumed; but here, rather than the spatiotemporal latent structure being captured by only a single term (ψtd) for all regions, a second component (λCtd) is added, which accounts for the local (clustering) effect (C). This relaxation allows adjacent space–time data points (e.g., Dtd,Dij) to be auto-correlated only if they are in the same cluster (λCtd=λCij), but exhibits a stepwise change if estimated to be in different clusters.

To identify the correct value of the cluster parameter (i.e., number of dissimilar clusters), a random subset of data for 2011 were selected, different values were assigned to the cluster parameter, and the corresponding models were fit to the data (Supplemental Appendix 2B). Deviance information criterion (DIC) and log marginal predictive likelihood (LMPL)42 were used to compare models with different numbers of clusters, where over- or under-assigned numbers were penalized by an increase in DIC or a decrease in LMPL scores. A similar approach was also used to assess which climate variables contribute to improved model fit. We chose another random subset of data from 2014 to 15 and fit models with different combinations of predictors to compare them versus the null; the chosen model was then fitted to the entire dataset. Results were obtained based on 100,000 Markov chain Monte Carlo (MCMC) iterations, thinned (×25) to reduce the autocorrelation among posterior samples, and the first 50,000 samples discarded as burn-in. We used the Geweke diagnostic43 to assess the convergence of the MCMC sampler. For computational stability, predictor variables were scaled to have a mean of zero and variance of one before model fitting, and back-transformed to real values after results were obtained. To assess models’ validity, a subset of fitted values from January to September 2016 were extracted and compared with the observed incidence rate in the three most populous provinces (Herat, Kabul, and Balkh) that represent different geographic regions in the country.

Microsoft Excel (Office 2013, Microsoft Corp., Redmond, WA) was used for initial data cleansing. Statistical analyses were conducted using the R statistical program (V 3.4; R Core Development Team, Vienna, Austria) and the CARBayesST package.39

RESULTS

Initial analysis revealed a seasonal trend in diarrhea cases, with peak incidence occurring in summer and reduced incidence during winter months (Figure 1). Temporally, the peak in diarrhea cases coincided with a rise in air temperature and rainfall (Figure 1).

Figure 1.
Figure 1.

Monthly distribution of diarrhea incidence (cases per month), mean daily temperature (°C), aridity index (minimum daily temperature/humidity), and rainfall (millimeters per month) from January to December (averaged over the study period for all districts).

Citation: The American Journal of Tropical Medicine and Hygiene 101, 3; 10.4269/ajtmh.18-0735

Average daily temperature across all districts was 12.2°C, and rainfall was almost 40 mm/month, which suggests a temperate climate. Temperature-wise, the northeastern and central regions, which mostly correspond to mountain ranges forming the western end of the Himalayas including Hindu Kush and Pamir, are coldest (Figure 2); temperature stays below the national average even during the summer (Supplemental Appendix 1A). The southeast plains are the warmest part of the country (Figure 2).

Figure 2.
Figure 2.

Spatial distribution of population-adjusted average monthly diarrhea (top left), average annual temperature (top right), average annual aridity index (bottom right), and average annual rainfall levels (bottom left) in 2014. Both diarrhea and rainfall were log transformed for better contrast with color scales.

Citation: The American Journal of Tropical Medicine and Hygiene 101, 3; 10.4269/ajtmh.18-0735

Although the overall temporal trend of the aridity index is inversely associated with temperature, the spatial distribution of aridity across different seasons (Supplemental Appendix 1B) illustrated that aridity is negatively correlated with the rise in temperature in some eastern and central regions, and positively correlated with temperature in the rest of the country (Supplemental Appendices 1A and 1B). Because the east and northeast receive the greatest share of rainfall (Figure 2), the inverse trend of aridity could be the result of the waning monsoon currents (east) and deposition of precipitation in the shape of snow during the winter (northeast). This could also explain the larger difference in the aridity index between seasons in eastern Afghanistan (comparing winter versus summer) (Supplemental Appendix 1B), where less rain-fed districts experience smaller fluctuations throughout the year. Because the eastern and central regions form a substantial proportion of reporting units despite their small geographic dimensions, the overall trend of aridity is perhaps biased toward them.

A high degree of spatial disparity in diarrhea incidence was observed throughout the country (Figure 2). North-central and eastern districts reported the highest incidence, and scarcely populated areas in the south reported lower rates of diarrhea on average in the population-adjusted map, but no obvious spatial correlations were observed with temperature, rainfall, or aridity variations. On average, 709 cases of diarrhea were reported from each district per month during the study period, but the number of diarrhea cases per district per month was highly variable (Table 1).

Table 1

Monthly distribution of unadjusted diarrhea cases and climatic variables per district (April 2010–September 2016)

VariableUnit/monthMeanStandard deviationMinimumMaximum
DiarrheaNo.7091004025,067
Minimum daily temperatureCentigrade5.269.46−27.4232.65
Maximum daily temperatureCentigrade19.4311.08−16.8245.48
Mean daily temperatureCentigrade12.2210.46−21.9436.68
RainMillimeter39.146.60686.6
AridityRatio of minimum daily temperature/specific humidity (0.01)0.0730.0310.0170.348
Diurnal temperature variationCentigrade14.272.474.9840.36
Average annual temperatureCentigrade/year18.696.06−1.7432.02

For the average annual temperature, the yearly distribution across all districts is reported.

Both maximum and minimum daily temperatures were highly correlated with the mean daily temperature (Pearson’s correlation > 0.90) and thereby dropped from proceeding analysis to avoid multicollinearity (Supplemental Appendix 2A). Univariate Poisson analysis of the relationship between diarrhea incidence and climate variables suggested positive associations with temperature (P < 0.001), diurnal temperature variation (P < 0.001), and annual average temperature (P < 0.001), but negative associations with aridity (P < 0.001) and rainfall (P < 0.001) (Supplemental Appendix 2B). Results also indicate a very small but statistically significant positive temporal trend in diarrhea incidence. Moran’s test for spatial dependence, using the residuals from a Poisson model, suggested a statistically significant correlation (value = 0.22721, observed rank = 10,001, P-value < 0.005) that indicates a bias in univariate Poisson analysis, necessitating the consideration of spatial structure.

Model fit.

Spatial patterns (Figure 2) suggested heterogeneous, multiple cluster focal points of diarrhea. We found that allowing for 13 distinct clusters provided the best fit to the subset of data from 2011 (Supplemental Appendix 2C) and used this value for the cluster parameter in subsequent models that were constructed using the entire range of data.

We developed several models using this approach and used DIC, LMPL, and root mean square error (RMSE) criteria to select the best fit combination. Results (Table 2) illustrate that model 3, which does not include rain and diurnal temperature variation, has the best fit (DIC = 53,128.82, LMPL = −23,183.82, RMSE = 1.17) and indicate that perhaps rainfall and diurnal temperature variation are not major predictors for diarrhea at the available spatiotemporal scale; all other factors were found to be significant predictors.

Table 2

Comparison of model fit vs. the full model (model 1) using a subset of data (2014–2015)

ModelPredictor(s)DICp.d.LMPLRMSE
1Time + mean daily temperature + rain + aridity index + average annual temperature + diurnal temperature59,076.234,316.58−25,847.241.58
2Time + mean daily temperature + rain + aridity index + average annual temperature56,597.633,193.19−24,939.731.44
3Time + mean daily temperature + aridity index + average annual temperature53,128.822,820.02−23,183.821.17
4Time + mean daily temperature + rain + average annual temperature54,035.582,934.83−24,025.241.31
5Time + mean daily temperature + rain61,241.923,091.51−27,370.071.95
6Time, mean daily temperature + aridity index + average annual temperature + diurnal temperature60,232.444055.11−26,576.001.67

DIC = deviance information criterion; LMPL = log marginal predictive likelihood. Model 3, which excluded rain and diurnal temperature variation, showed the best fit, with lowest DIC, highest score for p.d. and LMPL, and lowest RMSE.

Aridity index and mean daily temperature were both significantly and positively associated with diarrhea incidence; every 1°C increase in the mean daily temperature was associated with a 0.70% increase in relative risk for diarrhea, and a 0.01-unit change in the aridity index was associated with a 4.79% increase in the risk of diarrhea (Table 3). Average annual temperature, on the other hand, was negatively correlated with diarrhea incidence, with a 3.7% decrease in risk for every degree Celsius increase in annual average temperature (Table 3); this implies that even though seasonally varying variables such as daily mean temperature and aridity are positive predictors for diarrhea, regions with a colder overall climate have higher baseline-level risk, after adjusting for other predictors. A statistically significant, negative temporal trend in diarrhea incidence was also observed.

Table 3

Spatiotemporal Hierarchical Bayesian Model for diarrhea incidence risk per district, using the localized structure

Variable (unit)Change in risk, median (%)95% CI
Time (month)−0.08(−0.09, −0.08)
Aridity (0.01)4.79(4.30, 5.26)
Mean daily temperature (C0)0.70(0.67, 0.73)
Average annual temperature (C0)−3.71(−3.74, −3.68)

Reported median values are for percent relative risk increase or decrease.

A subset of fitted values from January to September 2016 were extracted and compared with the observed incidence rate in the three most populous provinces (Herat, Kabul, and Balkh) that represent different geographic regions the country (western, northern, and central regions, respectively), for internal validation. The model demonstrated a good fit to the spatial and temporal distribution of diarrhea in the three provinces (Figure 3).

Figure 3.
Figure 3.

Comparison of fitted vs. observed diarrhea incidence rate in three provinces (Herat, Kabul, and Balkh). Maps illustrate the average for the period from January–September 2016, whereas graphs represent the correlation of fitted vs. observed data points for the same period.

Citation: The American Journal of Tropical Medicine and Hygiene 101, 3; 10.4269/ajtmh.18-0735

DISCUSSION

Our results highlight the role of climate in shaping the spatiotemporal distribution of diarrheal disease in Afghanistan. Temporally, the trend illustrates a small but statistically significant drop in rates of diarrhea over time. There is an annual seasonal pattern starting in early summer, peaking around July–August, and dropping after October. This period corresponds with the rise in daily mean temperature and drop in relative humidity and rainfall levels. Spatially, districts forming scarcely populated desert areas in the south reported fewer cases, whereas the eastern region near the Pakistani border (which experiences a fringe monsoon climate) and districts in central and northern Afghanistan (two sides of the Hindu Kush mountain range) reported most cases on average. The high variability in reported diarrhea incidence even for districts close to each other reflects perhaps the heterogeneous map of clustered transmission groups.

The positive association we identified between temperature and diarrhea incidence reflects similar results reported by comparative studies27,44 and likely reflects increased growth and survival of bacterial pathogens at warmer temperatures.45 Aridity index also exhibited a positive association with diarrhea incidence. Indeed, comparing maps of climate variables and diarrhea incidences from four different seasons (Supplemental Appendix 1, Figures A–C), we observed that the rise in diarrhea reports is closely correlated with an increase in temperature and aridity index during the summer months in central, northern, and western parts of the country, consistent with findings from other settings with a dry climate.46

Although both average annual temperature and diurnal temperature were used to measure the interaction between topography and climate variables, only the former has been found to play a role in the spatiotemporal distribution of diarrheal diseases in other settings. The relationship we identified illustrates a negative association, whereby districts with higher average annual temperature—a surrogate for warmer, flatter settings—reported fewer cases after adjusting for other climate variables.

Despite significant associations with climatic variables, deciphering the process(es) underlying the link is challenging, given the multiplicity of pathways by which climate factors can shape infectious disease transmission,47 including diarrheal infections. In the absence of information about the specific etiologic agents of diarrhea in our dataset, we can only suggest hypotheses about possible driving mechanisms. Traditionally, elevated diarrhea rates in warmer seasons are suggested to be mediated through factors such as enhanced pathogen survival in the environment,45 elevated exposure to contaminated food,48 increased water intake in hotter conditions,49 or more frequent contact among individuals. Similarly, drier weather is also associated with increased survivability of viral and bacterial pathogens, including those that cause diarrheal diseases.50,51 Concurrence of aridity and hotter temperature was found to be strongly associated with the elevated rate of diarrheal incidences in settings similar to Afghanistan23 and has been suggested to facilitate communicability of other infections, such as plague.52

Although in comparative studies, rainfall was found to be an important predictor of diarrhea patterns,8,53 likely because of the increased contamination of drinking water, it was not found to be a major predictor in our study. This could be the result of the combined effect of geography, seasonality, and reporting factors. Overall, the country is located in a region where rainfall levels are typically low54 and perhaps short of necessary thresholds5557 to facilitate the transmission of diarrheal infections among the population. Furthermore, rainy seasons generally correspond to winters in major parts of the country, when precipitation does not necessarily equate to immediate increased water flow on the ground. Districts with higher altitudes receive the larger proportion of precipitation, where it typically deposits in the form of snow. Coincidently, it is the same period of the year when population activities wind down, roads close, and communications get interrupted, particularly in rural districts. As the socioeconomic activities pick up again in late spring, rain stops in almost all parts of the country, with the exception of eastern districts that have separate climate conditions. However, because this study used monthly average rainfall measures, the possibility of an association between acute rainfall events and diarrhea at finer time intervals (e.g., weekly periods) cannot be ruled out.

An important finding was that eastern districts demonstrate distinct diarrhea patterns and higher overall incidence (Supplemetal Appendix 1C). This could be the result of differences in the timing of climate trends. Although the rest of the country observes a drop in humidity and increased aridity during the summer, eastern districts receive increasing rainfall with the onset of the monsoon in Southeast Asia. The trend reverses during the winter months, which in combination with the start of snowfall in nearby northeast and central districts exacerbates aridity in the eastern region in winter rather than summer (Supplemenatl Appendix 1B). Thus, the risk level for diarrhea is sustained through a “double-punch mechanism”: rise of temperature in summer and increased aridity in winter seasons (Supplemental Appendix 1C), consistent with predictions in our model.

In combination with this monsoon climate, density of risk groups might play a role in the East as well. Population is highly concentrated in the eastern districts and is much lower than average in southern districts. In the context of high birth rates, the densely populated districts are expected to have a bigger pool of susceptible individuals (infants and children); in such settings, it has been illustrated that sustained transmission of viral causes of diarrhea (e.g., rotavirus) could be maintained year-round.58 Whereas the dominant proportion of acute diarrhea cases among the general population of Afghanistan is caused by bacteria, including cholera, enterotoxigenic Escherichia coli, Shigella, Salmonella, and Campylobacter,16 which are often associated with warmer temperatures, among children (particularly those under the age of five), viral pathogens such as Rotavirus and Norovirus are the predominant causes of diarrhea.16,59 Rotavirus, in particular, was found to cause 52% of acute gastroenteritis hospitalizations among children < 5 years old in Kabul and Herat.60 Although rotavirus typically peaks in the cooler, drier months in temperate regions and some tropical countries,61,62 it was found year-round in Afghanistan.42 In localities where no large-scale vaccination has been initiated, socioeconomic development has been found to be stronger predictor of rotavirus seasonality than climate factors.62 In this study, we did not have data for age categories (e.g., younger than versus older than 5 years) nor individual pathogens; therefore, it was not possible to discern whether the high incidence of diarrhea in the eastern region is associated with elevated risk among children in densely populated districts or can be attributed to differences in climate (or a combination of both). This illustrates that diarrhea is a multifaceted problem in which many variables need to be considered for a comprehensive understanding of observed patterns.

A notable limitation of our analysis is that proposed models are based on monthly incidence data and climate patterns, and therefore less likely to capture impacts of acute weather events such as heavy rainfalls or heat waves on diarrhea. Furthermore, the probability of underreporting in a number of remote localities due to inaccessibility of medical services63 could potentially lead to bias in the observed spatial pattern, particularly for less populated areas. Also, data provided for this study come mostly for outpatient departments of designated health facilities, where most cases are diagnosed based on clinical symptoms alone. However, given the high overall incidence64 and relative ease of diarrhea diagnose (which is straightforward and nonspecific), the bias in reporting, and therefore spatial patterns, should be relatively low.

Although our analysis highlights the role of climatic factors, other variables may have even larger effects in structuring diarrhea incidence in Afghanistan, including access to clean water, education level (particularly among women), population concentration (urban versus rural settings), birth rate, and population movements. Although our model implicitly controls for the spatial clustering of such variables, it could possibly be improved by explicitly incorporating those aspects should more data be obtained in future studies.

Our study has highlighted the prominent role climate plays in determining the geographic and temporal pattern of diarrhea in Afghanistan. Although many factors may contribute to higher rates of diarrhea, including environmental, societal, and economic elements, the role of climate is important to consider, as climate change is likely to worsen future trends of diarrheal diseases in the country.1,12,34 Given the projected 1.7–6.4°C increase in mean temperature in the coming decades,20 there could be a corresponding > 1% to 4.5% increase in monthly diarrhea rates on average based on extrapolation of results from this study. However, this is without consideration of the concurrent effect of aridity (which is likely to also increase with climate change65,66), and does not take into account the indirect effects that climate change may have on diarrheal incidence (e.g., through population movements or changes in diet67) or the impact of future interventions. Results from this study illustrate the potential of a climate-based spatiotemporal model to capture diarrheal diseases patterns in Afghanistan, which could be integrated into an early warning system, providing a platform for public health planners and policy-makers to assess future trends and evaluate the potential value of interventions.

Supplemental Appendix

Acknowledgments:

We sincerely thank Afghanistan’s Ministry of Public Health and Dr. Sayed Yaqoob Azimi, MD-MPH, Information Management Specialist at UNICEF Afghanistan Country Office, for providing data and insightful information that were used for our analysis.

REFERENCES

  • 1.

    McMichael AJ, Woodruff RE, Hales S, 2006. Climate change and human health: present and future risks. Lancet 367: 859869.

  • 2.

    Patz JA, Epstein PR, Burke TA, Balbus JM, 1996. Global climate change and emerging infectious diseases. JAMA 257: 217223.

  • 3.

    Patz JA, Campbell-Lendrum D, Holloway T, Foley JA, 2005. Impact of regional climate change on human health. Nature 438: 310317.

  • 4.

    Serdeczny O, Adams S, Baarsch F, Coumou D, Robinson A, Hare W, Schaeffer M, Perrette M, Reinhardt J, 2017. Climate change impacts in sub-Saharan Africa: from physical changes to their social repercussions. Reg Environ Change 17: 15851600.

    • Search Google Scholar
    • Export Citation
  • 5.

    Jepsen MR, Simonsen J, Ethelberg S ,2009. Spatio-temporal cluster analysis of the incidence of Campylobacter cases and patients with general diarrhea in a Danish county, 1995–2004. Int J Health Geogr 8: 11.

    • Search Google Scholar
    • Export Citation
  • 6.

    Carlton EJ, Woster AP, DeWitt P, Goldstein RS, Levy K, 2015. A systematic review and meta-analysis of ambient temperature and diarrhoeal diseases. Int J Epidemiol 45: 117130.

    • Search Google Scholar
    • Export Citation
  • 7.

    Herrador BRG, De Blasio BF, MacDonald E, Nichols G, Sudre B, Vold L, Semenza JC, Nygård K, 2015. Analytical studies assessing the association between extreme precipitation or temperature and drinking water-related waterborne infections: a review. Environ Health 14: 29.

    • Search Google Scholar
    • Export Citation
  • 8.

    Carlton EJ, Eisenberg JN, Goldstick J, Cevallos W, Trostle J, Levy K, 2013. Heavy rainfall events and diarrhea incidence: the role of social and environmental factors. Am J Epidemiol 179: 344352.

    • Search Google Scholar
    • Export Citation
  • 9.

    Troeger C et al. 2018. Global disability-adjusted life-year estimates of long-term health burden and undernutrition attributable to diarrhoeal diseases in children younger than 5 years. Lancet Glob Health 6: e255e269.

    • Search Google Scholar
    • Export Citation
  • 10.

    Bartsch SM, Lopman BA, Ozawa S, Hall AJ, Lee BY, 2016. Global economic burden of norovirus gastroenteritis. PLoS One 11: e0151219.

  • 11.

    Lelieveld J et al. 2012. Climate change and impacts in the eastern Mediterranean and the Middle East. Clim Change 114: 667687.

  • 12.

    Lloyd SJ, Kovats RS, Armstrong BG, 2007. Global diarrhoea morbidity, weather and climate. Clim Res 34: 119127.

  • 13.

    Khalil I et al. 2016. Burden of diarrhea in the eastern Mediterranean region, 1990–2013: findings from the global burden of disease study 2013. Am J Trop Med Hyg 95: 13191329.

    • Search Google Scholar
    • Export Citation
  • 14.

    Troeger C et al. 2018. Estimates of the global, regional, and national morbidity, mortality, and aetiologies of diarrhoea in 195 countries: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Infect Dis 18: 12111228.

    • Search Google Scholar
    • Export Citation
  • 15.

    Anwar MY, Burnham G, 2016. Trends in infectious disease incidence among children in Afghanistan at a time of public health services expansion. EMHJ 22: 778785.

    • Search Google Scholar
    • Export Citation
  • 16.

    Wallace MR, Hale BR, Utz GC, Olson PE, Earhart KC, Thornton SA, Hyams KC, 2002. Endemic infectious diseases of Afghanistan. Clin Infect Dis 34(Suppl 5): S171S207.

    • Search Google Scholar
    • Export Citation
  • 17.

    World Health Organization, 2017. World Health Statistics 2017: Monitoring Health for the SDGs. Geneva, Switzerland: WHO. Available at: https://www.who.int/gho/publications/world_health_statistics/2017/en/. Accessed July 30, 2017.

    • Search Google Scholar
    • Export Citation
  • 18.

    Raleigh C, Jordan L, Salehyan I, 2008. Assessing the impact of climate change on migration and conflict. Paper Commissioned by the World Bank Group for the Social Dimensions of Climate Change Workshop, Washington, DC: The Social Development Department, The World Bank Group, 5–6. Available at: https://www.researchgate.net/profile/Clionadh_Raleigh/publication/255519298_Assessing_the_Impact_of_Climate_Change_on_Migration_and_Conflict/links/58c6a15392851c0ccbff63fb/Assessing-the-Impact-of-Climate-Change-on-Migration-and-Conflict.pdf.

  • 19.

    Mendelsohn R, Dinar A, Williams L, 2006. The distributional impact of climate change on rich and poor countries. Environ Dev Econ 11: 159178.

    • Search Google Scholar
    • Export Citation
  • 20.

    Aich V, Akhundzadah N, Knuerr A, Khoshbeen A, Hattermann F, Paeth H, Scanlon A, Paton E, 2017. Climate change in Afghanistan deduced from reanalysis and coordinated regional climate downscaling experiment (Cordex)—south Asia simulations. Climate 5: 38.

    • Search Google Scholar
    • Export Citation
  • 21.

    Wagner AL, Mubarak MY, Johnson LE, Porth JM, Yousif JE, Boulton ML, 2017. Trends of vaccine-preventable diseases in Afghanistan from the disease early warning system, 2009–2015. PLoS One 12: e0178677.

    • Search Google Scholar
    • Export Citation
  • 22.

    Lowe R, Bailey TC, Stephenson DB, Graham RJ, Coelho CA, Carvalho MS, Barcellos C, 2011. Spatio-temporal modelling of climate-sensitive disease risk: towards an early warning system for dengue in Brazil. Comput Geosci 37: 371381.

    • Search Google Scholar
    • Export Citation
  • 23.

    Alexander KA, Carzolio M, Goodin D, Vance E, 2013. Climate change is likely to worsen the public health threat of diarrheal disease in Botswana. Int J Environ Res Public Health 10: 12021230.

    • Search Google Scholar
    • Export Citation
  • 24.

    Starr JM, Campbell A, Renshaw E, Poxton IR, Gibson GJ, 2009. Spatio-temporal stochastic modelling of Clostridium difficile. J Hosp Infect 71: 4956.

    • Search Google Scholar
    • Export Citation
  • 25.

    Khalidi NA, 1989. Demographic profile of Afghanistan. Canberra, ACT, Australia. International Population Dynamics Program, Dept. of Demography, Research School of Social Sciences, the Australian National University. Available at: https://www.academia.edu/11860590/Demographic_Profile_of_Afghanistan_1979.

    • Search Google Scholar
    • Export Citation
  • 26.

    Fonseca PA, Hacon SD, Reis VL, Costa D, Brown IF, 2016. Using satellite data to study the relationship between rainfall and diarrheal diseases in a southwestern Amazon basin. Cien Saude Colet 21: 731742.

    • Search Google Scholar
    • Export Citation
  • 27.

    Konno T, Suzuki H, Katsushima N, Imai A, Tazawa F, Kutsuzawa T, Kitaoka S, Sakamoto M, Yazaki N, Ishida N, 1983. Influence of temperature and relative humidity on human rotavirus infection in Japan. J Infect Dis 147: 125128.

    • Search Google Scholar
    • Export Citation
  • 28.

    Xu Z, Liu Y, Ma Z, Toloo GS, Hu W, Tong S, 2014. Assessment of the temperature effect on childhood diarrhea using satellite imagery. Sci Rep 4: 5389.

  • 29.

    Prasetyo D, Ermaya Y, Martiza I, Yati S, 2015. Correlation between climate variations and rotavirus diarrhea in under-five children in Bandung. Asian Pac J Trop Dis 5: 908911.

    • Search Google Scholar
    • Export Citation
  • 30.

    Curtis SL, Hossain M, 1998. The effect of aridity zone on child nutritional status. Calverton, MD: Macro International. Available at: https://www.popline.org/node/530081.

  • 31.

    Colston JM et al. 2018. Evaluating meteorological data from weather stations, and from satellites and global models for a multi-site epidemiological study. Environ Res 165: 91109.

    • Search Google Scholar
    • Export Citation
  • 32.

    Shi L, Liu P, Kloog I, Lee M, Kosheleva A, Schwartz J, 2016. Estimating daily air temperature across the southeastern United States using high-resolution satellite data: a statistical modeling study. Environ Res 146: 5158.

    • Search Google Scholar
    • Export Citation
  • 33.

    dos Reis J, Rennó C, Lopes E, 2017. Validation of satellite rainfall products over a mountainous watershed in a humid subtropical climate region of Brazil. Remote Sens 9: 1240.

    • Search Google Scholar
    • Export Citation
  • 34.

    Kolstad EW, Johansson KA, 2011. Uncertainties associated with quantifying climate change impacts on human health: a case study for diarrhea. Environ Health Perspect 119: 299305.

    • Search Google Scholar
    • Export Citation
  • 35.

    Wang Y, Gu J, 2015. A hybrid prediction model applied to diarrhea time series. Fuzzy Systems and Knowledge Discovery FSKD). 2015 12th International Conference on Fuzzy Systems and Knowledge Discovery (FSKD), Zhangjiajie, China, 1096–1102. Available at: https://ieeexplore.ieee.org/abstract/document/7382095.

    • Search Google Scholar
    • Export Citation
  • 36.

    Liu Y, Wang X, Pang C, Yuan Z, Li H, Xue F, 2015. Spatio-temporal analysis of the relationship between climate and hand, foot, and mouth disease in Shandong province, China, 2008–2012. BMC Infect Dis 15: 146.

    • Search Google Scholar
    • Export Citation
  • 37.

    Raghavan RK, Hanlon CA, Goodin DG, Davis R, Moore M, Moore S, Anderson GA, 2016. Bayesian spatiotemporal pattern and eco-climatological drivers of striped skunk rabies in the North Central Plains. PLoS Negl Trop Dis 10: e0004632.

    • Search Google Scholar
    • Export Citation
  • 38.

    Chib S, Winkelmann R, 2001. Markov chain Monte Carlo analysis of correlated count data. J Bus Econ Stat 19:428435.

  • 39.

    Lee D, Rushworth A, Napier G, 2016. CARBayesST Version 2.2: An R Package for Spatio-temporal Areal Unit Modelling with Conditional Autoregressive Priors. Technical Report. University of Glasgow. Available at: https://cran.r-project.org/web/packages/CARBayesST/vignettes/CARBayesST.pdf.

    • Search Google Scholar
    • Export Citation
  • 40.

    Leroux BG, Lei X, Breslow N, 2000. Estimation of disease rates in small areas: a new mixed model for spatial dependence. Statistical Models in Epidemiology, the Environment, and Clinical Trials. New York, NY: Springer, 179191.

    • Search Google Scholar
    • Export Citation
  • 41.

    Rushworth A, Lee D, Mitchell R, 2014. A spatio-temporal model for estimating the long-term effects of air pollution on respiratory hospital admissions in Greater London. Spat Spatiotemporal Epidemiol 10: 2938.

    • Search Google Scholar
    • Export Citation
  • 42.

    Congdon P, 2005. Bayesian Models for Categorical Data. Hoboken, NJ: John Wiley & Sons. Available at: https://onlinelibrary.wiley.com/doi/book/10.1002/0470092394.

    • Search Google Scholar
    • Export Citation
  • 43.

    Geweke J, 1991. Evaluating the Accuracy of Sampling-Based Approaches to the Calculation of Posterior Moments. Minneapolis, MN: Federal Reserve Bank of Minneapolis, V196.

    • Search Google Scholar
    • Export Citation
  • 44.

    Checkley W, Epstein LD, Gilman RH, Figueroa D, Cama RI, Patz JA, Black RE, 2000. Effects of EI Niño and ambient temperature on hospital admissions for diarrhoeal diseases in Peruvian children. Lancet 355:442450.

    • Search Google Scholar
    • Export Citation
  • 45.

    Zhou X, Zhou Y, Chen R, Ma W, Deng H, Kan H, 2013. High temperature as a risk factor for infectious diarrhea in Shanghai, China. J Epidemiol 23: 418423.

    • Search Google Scholar
    • Export Citation
  • 46.

    Azage M, Kumie A, Worku A, Bagtzoglou AC, Anagnostou E, 2017. Effect of climatic variability on childhood diarrhea and its high risk periods in northwestern parts of Ethiopia. PLoS One 12: e0186933.

    • Search Google Scholar
    • Export Citation
  • 47.

    Metcalf CJE, Walter KS, Wesolowski A, Buckee CO, Shevliakova E, Tatem AJ, Boos WR, Weinberger DM, Pitzer VE, 2017. Identifying climate drivers of infectious disease dynamics: recent advances and challenges ahead. Proc Biol Sci. 284: 20170901.

    • Search Google Scholar
    • Export Citation
  • 48.

    Bentham G, Langford IH, 2001. Environmental temperatures and the incidence of food poisoning in England and Wales. Int J Biometeorol 45: 2226.

    • Search Google Scholar
    • Export Citation
  • 49.

    Black RE, Lanata CF, 1995. Epidemiology of diarrheal diseases in developing Countries. Infections of the Gastrointestinal Tract. New York, NY: Raven Press, 1336.

    • Search Google Scholar
    • Export Citation
  • 50.

    Ansari SA, Springthorpe VS, Sattar SA, 1991. Survival and vehicular spread of human rotaviruses: possible relation to seasonality of outbreaks. Rev Infect Dis 13: 448461.

    • Search Google Scholar
    • Export Citation
  • 51.

    McEldowney S, Fletcher M, 1988. The effect of temperature and relative humidity on the survival of bacteria attached to dry solid surfaces. Lett Appl Microbiol 7: 8386.

    • Search Google Scholar
    • Export Citation
  • 52.

    Yue RPH, Lee HF, 2018. Pre-industrial plague transmission is mediated by the synergistic effect of temperature and aridity index. BMC Infect Dis 18: 134.

    • Search Google Scholar
    • Export Citation
  • 53.

    Drayna P, McLellan SL, Simpson P, Li SH, Gorelick MH, 2010. Association between rainfall and pediatric emergency department visits for acute gastrointestinal illness. Environ Health Perspect 118: 14391443.

    • Search Google Scholar
    • Export Citation
  • 54.

    Qureshi AS, 2002. Water resources management in Afghanistan: the issues and options. IWMI 49.

  • 55.

    Bhavnani D, Goldstick JE, Cevallos W, Trueba G, Eisenberg JN, 2014. Impact of rainfall on diarrheal disease risk associated with unimproved water and sanitation. Am J Trop Med Hyg 90: 705711.

    • Search Google Scholar
    • Export Citation
  • 56.

    Bandyopadhyay S, Kanji S, Wang L, 2012. The impact of rainfall and temperature variation on diarrheal prevalence in sub-Saharan Africa. Appl Geogr 33: 6372.

    • Search Google Scholar
    • Export Citation
  • 57.

    Singh RB, Hales S, De Wet N, Raj R, Hearnden M, Weinstein P, 2001. The influence of climate variation and change on diarrheal disease in the Pacific Islands. Environ Health Perspectives 109:155159.

    • Search Google Scholar
    • Export Citation
  • 58.

    Pitzer VE, Viboud C, Lopman BA, Patel MM, Parashar UD, Grenfell BT, 2011. Influence of birth rates and transmission rates on the global seasonality of rotavirus incidence. J R Soc Interface 8:15841593.

    • Search Google Scholar
    • Export Citation
  • 59.

    Anwari P, Debellut F, Pecenka C, Parwiz SM, Clark A, Groman D, Safi N, 2017. Potential impact and cost-effectiveness of rotavirus vaccination in Afghanistan. Vaccine 36: 77697774.

    • Search Google Scholar
    • Export Citation
  • 60.

    Anwari P, Safi N, Payne DC, Jennings MC, Rasikh S, Waciqi AS, Parwiz SM, 2018. Rotavirus is the leading cause of hospitalizations for severe acute gastroenteritis among Afghan children< 5 years old. Vaccine 36: 77657768.

    • Search Google Scholar
    • Export Citation
  • 61.

    Levy K, Hubbard AE, Eisenberg JN, 2008. Seasonality of rotavirus disease in the tropics: a systematic review and meta-analysis. Int J Epidemiol 38: 14871496.

    • Search Google Scholar
    • Export Citation
  • 62.

    Patel MM, Pitzer V, Alonso WJ, Vera D, Lopman B, Tate J, Viboud C, Parashar UD, 2013. Global seasonality of rotavirus disease. Pediatr Infect Dis J 32: e134e137.

    • Search Google Scholar
    • Export Citation
  • 63.

    Frost A, Wilkinson M, Boyle P, Patel P, Sullivan R, 2016. An assessment of the barriers to accessing the Basic Package of Health Services (BPHS) in Afghanistan: was the BPHS a success? Global Health 12: 71.

    • Search Google Scholar
    • Export Citation
  • 64.

    Mokdad A et al. 2018. Burden of diarrhea in the eastern Mediterranean region, 1990–2015: findings from the Global Burden of Disease 2015 study. Int J Public Health 63: 109121.

    • Search Google Scholar
    • Export Citation
  • 65.

    Lioubimtseva E, Cole R, 2006. Uncertainties of climate change in arid environments of Central Asia. Rev Fish Sci 14: 2949.

  • 66.

    Wei K, Wang L, 2013. Reexamination of the aridity conditions in arid Northwestern China for the last decade. J Clim 26: 95949602.

  • 67.

    Patz JA, Grabow ML, Limaye VS, 2014. When it rains, it pours: future climate extremes and health. Ann Glob Health 80: 332344.

Author Notes

Address correspondence to Mohammad Y. Anwar, Department of Epidemiology, University of Louisville School of Public Health and Information Sciences, 485 E. Gray St., Louisville, KY 40202. E-mail: m0anwa02@louisville.edu

Financial support: J. L. W. was supported by CTSA grant numbers UL1 TR001863 and KL2 TR001862 from the National Center for Advancing Translational Science (NCATS), components of the National Institutes of Health (NIH), and NIH roadmap for Medical Research. V. E. P. was supported by the National Institutes of Health/National Institute of Allergy and Infectious Diseases (NIH/NIAID) grant R01AI112970. The contents of this manuscript are solely the responsibility of the authors and do not necessarily represent the official view of NIH.

Disclosure: This study used aggregate de-identified data and was therefore exempt from ethics committee review.

Authors’ addresses: Mohammad Y. Anwar, Department of Epidemiology, University of Louisville School of Public Health and Information Sciences, Louisville, KA, E-mail: m0anwa02@louisville.edu. Joshua L. Warren, Department of Biostatistics, Yale School of Public Health, New Haven, CT, E-mail: joshua.warren@yale.edu. Virginia E. Pitzer, Department of Epidemiology of Microbial Diseases, Yale School of Public Health, New Haven, CT, E-mail: virginia.pitzer@yale.edu.

Save