Abstract
Underfive years old acute watery diarrhea (U5AWD) accounts for most diarrheal diseases’ burden, but little is known about the adjusted effect of meteorological and socioeconomic determinants. A dataset containing the seasonal numbers of U5AWD cases at the district level of Iran is collected through MOHME. Accordingly, the districtlevel standardized incidence ratio and Moran’s I values are calculated to detect the significant clusters of U5AWD over sixteen seasons from 2014 to 2018. Additionally, the author tested twelve Bayesian hierarchical models in order to determine which one was the most accurate at forecasting seasonal number of incidents. Iran features a number of U5AWD hotspots, particularly in the southeast. An extended spatiotemporal model with seasonally varying coefficients and space–time interaction outperformed other models, and so became the paper’s proposal in modeling U5AWD. Temperature demonstrated a global positive connection with seasonal U5AWD in districts (IRR: 1.0497; 95% CrI: 1.0254–1.0748), owing to its varying effects during the winter ((IRR: 1.0877; 95% CrI: 1.0408–1.1375) and fall (IRR: 1.0866; 95% CrI: 1.0405–1.1357) seasons. Also, elevation (IRR: 0.9997; 95% CrI: 0.9996–0.9998), piped drinking water (IRR: 0.9948; 95% CrI: 0.9933–0.9964), public sewerage network (IRR: 0.9965; 95% CrI: 0.9938–0.9992), years of schooling (IRR: 0.9649; 95% CrI: 0.944–0.9862), infrastructuretohousehold size ratio (IRR: 0.9903; 95% CrI: 0.986–0.9946), wealth index (IRR: 0.9502; 95% CrI: 0.9231–0.9781), and urbanization (IRR: 0.9919; 95% CrI: 0.9893–0.9944) of districts were negatively associated with seasonal U5AWD incidence. Strategically, developing geoinformation alarm systems based on meteorological data might help predict U5AWD highrisk areas. The study also anticipates increased rates of U5AWD in districts with poor sanitation and socioeconomic level. Therefore, governments should take appropriate preventative actions in these sectors.
This is a preview of subscription content, access via your institution.
Data availability
Meteorological dataset analyzed during the current study is available in the IRIMO repository, https://data.irimo.ir/login/login.aspx. Elevation data can be obtained from https://www.calcmaps.commapelevation/. The raw data of different stages of Iran National Population and Housing Census and the Households Income and Expenditure Survey (HIES) is available at https://www.amar.org.ir/english. The other datasets generated and/or analyzed during the current study are not publicly available due to MOHME policies but are available from the corresponding author on reasonable request.
Code availability
All the software codes regarding model specifications are available here: https://www.rinla.org/home.
Abbreviations
 WHO:

World Health Organization
 IHME:

Institute for Health Metrics and Evaluation
 U5AWD:

Underfive acute watery diarrhea
 GBD:

Global Burden of Disease
 SDG 3:

Sustainable Development Goal 3
 BHM:

Bayesian hierarchical modeling
 MCMC:

Markov chain Monte Carlo
 INLA:

Integrated nested Laplace approximation
 MOHME:

Ministry of Health and Medical Education
 IHSR:

Infrastructuretohousehold size ratio
 YOS:

Years of schooling
 IRIMO:

Iran Meteorological Organization
 HIES:

Households Income and Expenditure Survey
 SCI:

Statistical Center of Iran
 PCA:

Principal component analysis
 DHS:

Demographic and Health Surveys
 SIRs:

Standardized incidence ratios
 LISAs:

Local indicators of spatial association
 IR:

Incidence rate
 iid:

Independent and identically distributed
 ICAR:

Intrinsic conditional autoregressive
 BYM:

BesagYorkMollié
 RW1:

Firstorder random walk
 DIC:

Deviance information criterion
 \({P}_{DIC}\) :

Effective number of parameters for DIC
 WAIC:

Watanabe–Akaike Information Criterion
 \({P}_{WAIC}\) :

Effective number of parameters for WAIC
 LS:

Logarithmic score
 CrI:

Credible interval
 MCrIW:

Mean credible interval width
 MAE:

Mean absolute error
 SD:

Standard deviation
 CI:

Confidence interval
 IQR:

Interquartile range
 HH:

HighHigh
 IRR:

Incidence rate ratio
References
Abuzerr S, Nasseri S, Yunesian M, Hadi M, Zinszer K, Mahvi AH, Nabizadeh R, Abu Mustafa A, Mohammed SH (2020) Water, sanitation, and hygiene risk factors of acute diarrhea among children under five years in the Gaza Strip. J Water Sanit Hyg Dev 10(1):111–123
Adane M, Mengistie B, Medhin G, Kloos H, Mulat W (2017) Piped water supply interruptions and acute diarrhea among underfive children in Addis Ababa slums, Ethiopia: a matched casecontrol study. PLoS ONE 12(7):e0181516. https://doi.org/10.1371/journal.pone.0181516
Agegnehu MD, Zeleke LB, Goshu YA, Ortibo YL, Mehretie Adinew Y (2019) Diarrhea prevention practice and associated factors among caregivers of underfive children in Enemay District, Northwest Ethiopia. J Environ Public Health 2019:5490716. https://doi.org/10.1155/2019/5490716
Agustina R, Sari TP, Satroamidjojo S, BoveeOudenhoven IM, Feskens EJ, Kok FJ (2013) Association of foodhygiene practices and diarrhea prevalence among Indonesian young children from low socioeconomic urban areas. BMC Public Health 13(1):1–12
Alebel A, Tesema C, Temesgen B, Gebrie A, Petrucka P, Kibret GD (2018) Prevalence and determinants of diarrhea among underfive children in Ethiopia: A systematic review and metaanalysis. PLoS ONE 13(6):e0199684. https://doi.org/10.1371/journal.pone.0199684
Alemayehu B, Ayele BT, Kloos H, Ambelu A (2020a) Individual and communitylevel risk factors in underfive children diarrhea among agroecological zones in southwestern Ethiopia. Int J Hyg Environ Health 224:113447. https://doi.org/10.1016/j.ijheh.2019.113447
Alemayehu B, Ayele BT, Melak F, Ambelu A (2020) Exploring the association between childhood diarrhea and meteorological factors in Southwestern Ethiopia. Sci Total Environ 741:140189
Anselin L (1995) Local indicators of spatial association—LISA. Geogr Anal 27(2):93–115
Anwar MY, Warren JL, Pitzer VE (2019) Diarrhea patterns and climate: a spatiotemporal Bayesian hierarchical analysis of diarrheal disease in Afghanistan. Am J Trop Med Hyg 101(3):525–533
Ashkenazi S, Schwartz E (2020) Traveler’s diarrhea in children: new insights and existing gaps. Travel Med Infect Dis 34:101503. https://doi.org/10.1016/j.tmaid.2019.101503
Azage M, Kumie A, Worku A, C. Bagtzoglou A, Anagnostou E (2017) Effect of climatic variability on childhood diarrhea and its high risk periods in northwestern parts of Ethiopia. PLoS One 12(10):e0186933
Berde AS, Yalçın SS, Özcebe H, Üner S, KaradağCaman Ö (2018) Determinants of childhood diarrhea among underfive year old children in Nigeria: a populationbased study using the 2013 demographic and health survey data. Turk J Pediatr 60(4):353–360. https://doi.org/10.24953/turkjped.2018.04.001
Besag J, York J, Mollié A (1991) Bayesian image restoration, with two applications in spatial statistics. Ann Inst Stat Math 43(1):1–20
Bivand R, Altman M, Anselin L, Assunção R, Berke O, Bernat A, Blanchet G (2015) Package ‘spdep’. The Comprehensive R Archive Network
Blangiardo M, Cameletti M (2015) Spatial and spatiotemporal Bayesian models with RINLA. John Wiley & Sons, Hoboken
Blangiardo M, Boulieri A, Diggle P, Piel FB, Shaddick G, Elliott P (2020) Advances in spatiotemporal models for noncommunicable disease surveillance. Int J Epidemiol 49:I26–I37. https://doi.org/10.1093/ije/dyz181
Budhathoki SS, Bhattachan M, Yadav AK, Upadhyaya P, Pokharel PK (2016) Ecosocial and behavioural determinants of diarrhoea in underfive children of Nepal: a framework analysis of the existing literature. Tropical Medicine and Health 44(1):1–7. https://doi.org/10.1186/s4118201600069
Carlton EJ, Eisenberg JNS, Goldstick J, Cevallos W, Trostle J, Levy K (2014) Heavy rainfall events and diarrhea incidence: the role of social and environmental factors. Am J Epidemiol 179(3):344–352. https://doi.org/10.1093/aje/kwt279
Chou WC, Wu JL, Wang YC, Huang H, Sung FC, Chuang CY (2010) Modeling the impact of climate variability on diarrheaassociated diseases in Taiwan (1996–2007). Sci Total Environ 409(1):43–51
De Albuquerque C, Roaf V (2012) On the right track: good practices in realising the rights to water and sanitation.
Delahoy MJ, Cárcamo C, Huerta A, Lavado W, Escajadillo Y, Ordoñez L, Vasquez V, Lopman B, Clasen T, Gonzales GF (2021) Meteorological factors and childhood diarrhea in Peru, 2005–2015: a time series analysis of historic associations, with implications for climate change. Environ Health 20(1):1–10
Diouf K, Tabatabai P, Rudolph J, Marx M (2014) Diarrhoea prevalence in children under five years of age in rural Burundi: an assessment of social and behavioural factors at the household level. Global Health Action 7(1):24895. https://doi.org/10.3402/gha.v7.24895
D’Souza RM, Hall G, Becker NG (2008) Climatic factors associated with hospitalizations for rotavirus diarrhoea in children under 5 years of age. Epidemiol Infect 136(1):56–64. https://doi.org/10.1017/S0950268807008229
Dunn G, Johnson GD, Balk DL, Sembajwe G (2020) Spatially varying relationships between risk factors and child diarrhea in West Africa, 2008–2013. Math Popul Stud 27(1):8–33. https://doi.org/10.1080/08898480.2019.1592638
Eesteghamati A, Gouya M, Keshtkar A, Najafi L, Zali MR, Sanaei M, Yaghini F, El Mohamady H, Patel M, Klena JD, Teleb N (2009) Sentinel hospitalbased surveillance of Rotavirus diarrhea in Iran. J Infect Dis 200(SUPPL. 1):S244–S247. https://doi.org/10.1086/605050
Elliott P, Savitz DA (2008) Design issues in smallarea studies of environment and health. Environ Health Perspect 116(8):1098–1104. https://doi.org/10.1289/ehp.10817
Feleke H, Medhin G, Kloos H, Gangathulasi J, Asrat D (2018) Householdstored drinking water quality among households of underfive children with and without acute diarrhea in towns of Wegera District, in North Gondar, Northwest Ethiopia. Environ Monit Assess 190(11):1–12. https://doi.org/10.1007/s1066101870334
Gelman A, Hwang J, Vehtari A (2014) Understanding predictive information criteria for Bayesian models. Stat Comput 24(6):997–1016. https://doi.org/10.1007/s1122201394162
GómezRubio V (2020) Bayesian inference with INLA. CRC Press
Hashizume M, Armstrong B, Hajat S, Wagatsuma Y, Faruque ASG, Hayashi T, Sack DA (2007) Association between climate variability and hospital visits for noncholera diarrhoea in Bangladesh: effects and vulnerable groups. Int J Epidemiol 36(5):1030–1037. https://doi.org/10.1093/ije/dym148
IHME (2020) Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2019 (GBD 2019) Results. Institute for Health Metrics and Evaluation Seattle. https://vizhub.healthdata.org/gbdcompare/. Accessed 14 July 2021
KnorrHeld L (2000) Bayesian modelling of inseparable spacetime variation in disease risk. Stat Med 19(17–18):2555–2567. https://doi.org/10.1002/10970258(20000915/30)19:17/18%3c2555::aidsim587%3e3.0.co;2%23
KumiKyereme A, AmoAdjei J (2016) Household wealth, residential status and the incidence of diarrhoea among children underfive years in Ghana. J Epidemiol Glob Health 6(3):131–140. https://doi.org/10.1016/j.jegh.2015.05.001
Li R, Lai Y, Feng C, Dev R, Wang Y, Hao Y (2020) Diarrhea in under five yearold children in nepal: a spatiotemporal analysis based on demographic and health survey data. Int J Environ Res Public Health 17(6):2140. https://doi.org/10.3390/ijerph17062140
Mallick R, Mandal S, Chouhan P (2020) Impact of sanitation and clean drinking water on the prevalence of diarrhea among the underfive children in India. Child Youth Serv Rev 118:105748. https://doi.org/10.1016/j.childyouth.2020.105478
MartínezBello DA, LópezQuílez A, TorresPrieto A (2017) Bayesian dynamic modeling of time series of dengue disease case counts. PLoS Negl Trop Dis 11(7):e0005696. https://doi.org/10.1371/journal.pntd.0005696
Masinaei M, Eshrati B, Yaseri M (2020) Spatial and spatiotemporal patterns of typhoid fever and investigation of their relationship with potential risk factors in Iran, 2012–2017. Int J Hyg Environ Health 224:113432. https://doi.org/10.1016/j.ijheh.2019.113432
McMichael AJ, CampbellLendrum DH, Corvalán CF, Ebi KL, Githeko A, Scheraga JD, Woodward A (2003) Climate change and human health: risks and responses. World Health Organization
Mertens A, Balakrishnan K, Ramaswamy P, Rajkumar P, Ramaprabha P, Durairaj N, Hubbard AE, Khush R, Colford JM Jr, Arnold BF (2019) Associations between high temperature, heavy rainfall, and diarrhea among young children in rural Tamil Nadu, India: A Prospective Cohort Study. Environ Health Perspect 127(4):47004
Miettinen OS (1974) Proportion of disease caused or prevented by a given exposure, trait or intervention. Am J Epidemiol 99(5):325–332. https://doi.org/10.1093/oxfordjournals.aje.a121617
Misriyanto E, Sitorus RJ (2020) Analysis of environmental factors with chronic diarrhea in toddlers in Jambi City in 2019. Int J Sci Soc 2(4):300–310
Moors E, Singh T, Siderius C, Balakrishnan S, Mishra A (2013) Climate change and waterborne diarrhoea in northern India: impacts and adaptation strategies. Sci Total Environ 468–469:S139–S151. https://doi.org/10.1016/j.scitotenv.2013.07.021
Moraga P (2017) SpatialEpiApp: A Shiny web application for the analysis of spatial and spatiotemporal disease data. Spat SpatioTemp Epidemiol 23:47–57. https://doi.org/10.1016/j.sste.2017.08.001
Mukabutera A, Thomson D, Murray M, Basinga P, Nyirazinyoye L, Atwood S, Savage KP, Ngirimana A, HedtGauthier BL (2016) Rainfall variation and child health: effect of rainfall on diarrhea among under 5 children in Rwanda, 2010. BMC Public Health 16(1):1–9. https://doi.org/10.1186/s1288901634359
Mulatya DM, Ochieng C (2020) Disease burden and risk factors of diarrhoea in children under five years: evidence from Kenya’s demographic health survey 2014. Int J Infect Dis 93:359–366. https://doi.org/10.1016/j.ijid.2020.02.003
Musengimana G, Mukinda FK, Machekano R, Mahomed H (2016) Temperature variability and occurrence of diarrhoea in children under fiveyearsold in Cape Town metropolitan subdistricts. Int J Environ Res Public Health 13(9):859. https://doi.org/10.3390/ijerph13090859
Negesse Y, Taddese AA, Negesse A, Ayele TA (2021) Trends and determinants of diarrhea among underfive children in Ethiopia: crosssectional study: multivariate decomposition and multilevel analysis based on Bayesian approach evidenced by EDHS 2000–2016 data. BMC Public Health 21(1):1–16. https://doi.org/10.1186/s12889021101913
Omona S, Malinga GM, Opoke R, Openy G, Opiro R (2020) Prevalence of diarrhoea and associated risk factors among children under five years old in Pader District, northern Uganda. BMC Infect Dis 20(1):1–9. https://doi.org/10.1186/s1287902047700
Online K (2014) Report from the Department of Economics about the Results of Implementing the First Phase of Targeting: The People Have Become Poor, the Class Gaps Have Increased. Khabar Online. http://www.khabaronline.ir/detail/363531/Economy/1379. Accessed 14 July 2021 2014
ParedesParedes M, Okhuysen PC, Flores J, Mohamed JA, Padda RS, GonzalezEstrada A, Haley CA, Carlin LG, Nair P, DuPont HL (2011) Seasonality of diarrheagenic Escherichia coli pathotypes in the US students acquiring diarrhea in Mexico. J Travel Med 18(2):121–125
Pettit LI (1990) The conditional predictive ordinate for the normal distribution. J Roy Stat Soc: Ser B (methodol) 52(1):175–184
Piel FB, Fecht D, Hodgson S, Blangiardo M, Toledano M, Hansell AL, Elliott P (2021) Smallarea methods for investigation of environment and health. Int J Epidemiol 49(2):686–699. https://doi.org/10.1093/IJE/DYAA006
Prasetyo D, Ermaya Y, Martiza I, Yati S (2015) Correlation between climate variations and rotavirus diarrhea in underfive children in Bandung. Asian Pacif J Trop Dis 5(11):908–911. https://doi.org/10.1016/S22221808(15)609550
Robert CP, Elvira V, Tawn N, Wu C (2018) Accelerating MCMC algorithms. Wiley Interdisc Rev: Comput Stat 10(5):e1435. https://doi.org/10.1002/wics.1435
Rue H, Martino S, Chopin N (2009) Approximate Bayesian inference for latent Gaussian models by using integrated nested Laplace approximations. J R Stat Soc Ser B Stat Methodol 71(2):319–392. https://doi.org/10.1111/j.14679868.2008.00700.x
Singh RBK, 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 Perspect 109(2):155–159. https://doi.org/10.1289/ehp.01109155
Spiegelhalter DJ, Best NG, Carlin BP, Van Der Linde A (2002) Bayesian measures of model complexity and fit. J R Stat Soc Ser B Stat Methodol 64(4):583–616. https://doi.org/10.1111/14679868.00353
Tate JE, Burton AH, BoschiPinto C, Parashar UD, Agocs M, Serhan F, De Oliveira L, Mwenda JM, Mihigo R, Ranjan Wijesinghe P, Abeysinghe N, Fox K, Paladin F (2016) Global, regional, and national estimates of rotavirus mortality in children <5 years of age, 2000–2013. Clin Infect Dis 62:S96–S105. https://doi.org/10.1093/cid/civ1013
Thiam S, Cissé G, Stensgaard AS, NiangDiène A, Utzinger J, Vounatsou P (2019) Bayesian conditional autoregressive models to assess spatial patterns of diarrhoea risk among children under the age of 5 years in mbour, senegal. Geospat Health 14(2):321–328. https://doi.org/10.4081/gh.2019.823
UNICEF (2016) Diarrhoea remains a leading killer of young children, despite the availability of a simple treatment solution. UNICEF data: monitoring the situation of children and women
Vaghefi SA, Keykhai M, Jahanbakhshi F, Sheikholeslami J, Ahmadi A, Yang H, Abbaspour KC (2019) The future of extreme climate in Iran. Sci Rep 9(1):1–11
Wang Y, Li J, Gu J, Zhou Z, Wang Z (2015) Artificial neural networks for infectious diarrhea prediction using meteorological factors in Shanghai (China). Appl Soft Comput 35:280–290
Wang X, Yue Y, Faraway JJ (2018) Bayesian regression modeling with INLA. Chapman and Hall/CRC, London
Wang H, Di B, Zhang T, Lu Y, Chen C, Wang D, Li T, Zhang Z, Yang Z (2019) Association of meteorological factors with infectious diarrhea incidence in Guangzhou, southern China: a timeseries study (2006–2017). Sci Total Environ 672:7–15
Wasihun AG, Dejene TA, Teferi M, Marugán J, Negash L, Yemane D, McGuigan KG (2018) Risk factors for diarrhoea and malnutrition among children under the age of 5 years in the Tigray Region of Northern Ethiopia. PLoS ONE 13(11):e0207743. https://doi.org/10.1371/journal.pone.0207743
Watanabe S (2010) Asymptotic equivalence of Bayes cross validation and widely applicable information criterion in singular learning theory. J Mach Learn Res 11:3571–3594
WHO (2017) WHO Fact Sheet. World Health Organization. https://www.who.int/newsroom/factsheets/detail/diarrhoealdisease. Accessed 14 July 2021
Xu Z, Huang C, Turner LR, Su H, Qiao Z, Tong S (2013) Is diurnal temperature range a risk factor for childhood diarrhea? PLoS One 8(5):e64713
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
Zhang Y, Bambrick H, Mengersen K, Tong S, Hu W (2021) Using internetbased query and climate data to predict climatesensitive infectious disease risks: a systematic review of epidemiological evidence. Int J Biometeorol 65(12):2203–2214. https://doi.org/10.1007/s00484021021554
Acknowledgements
The author gratefully acknowledges Dr. Håvard Rue, Dr. Finn Lindgren, and Dr. Elias T. Krainski for their valuable suggestions and technical supports in Rinla discussion group (https://groups.google.com/g/rinladiscussiongroup). The author also appreciates Dr. Babak Eshrati for extracting the disease dataset from MOHME’s communicable diseases surveillance system.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Ethics approval
The Iranian Ministry of Health and Medical Education (MOHME) waives the need for approval in analyses where only aggregated surveillance data are provided.
Data protection, confidentiality and privacy
Not applicable.
Consent to participate
Not applicable.
Consent for publication
Not applicable.
Competing interests
The author declares no competing interests.
Appendix
Appendix
This appendix provides an overview of the 12 models used throughout the article. It is worth noting that all models use the same notations to condense explanations and facilitate comparisons across models:
Model 1: The ordinary multivariate regression model (nonspatial and nontemporal) as:
where \({\beta }_{0}\) is the intercept; \({X}_{k,it}\) is the value of \(k\) th covariate in district \(i\), season \(t\); and \({\beta }_{k}\) is the estimated coefficient of \(k\) th covariate, showing its linear fixed effect through space and time.
Model 2: The spatial ecological regression model as:
where \({u}_{i}\) and \({v}_{i}\) are the spatial unstructured and structured random effect in district \(i\), respectively; \({\mathcal{N}}_{i}\) is the number of districts sharing boundaries with \(i\) th district; and \({\sigma }_{i}^{2}\) is the variance between \(i\) th district’s effect and neighbors. In every model, we selected an independent and identically distributed (iid) Gaussian distribution as diffuse prior for \({u}_{i}\) and the intrinsic conditional autoregressive (ICAR) structure as the prior distribution of \({v}_{i}\) (Eqs. 9–10). \({u}_{i}+{v}_{i}\) togetherly construct a prominent convolution model known as Besag York Mollié (BYM) (Besag et al. 1991).
Model 3: The time series regression model as:
where \({\gamma }_{t}\) and \({\phi }_{t}\) are the temporal unstructured and structured random effect in year \(t\), respectively. In every model, we selected an iid Gaussian distribution as noninformative prior for \({\gamma }_{t}\) (Eq. 12). We utilized the seasonal latent model available in RINLA as the prior of \({\phi }_{t}\), assuming \(\sum_{t=1}^{4}{\phi }_{t}=0\). We formulated the density for seasonal prior in Eq. 13.
Model 4: A seasonally varying coefficient model containing nonstationary random effects for meteorological variables,
which can be stated more straightforwardly as follows:
where \({X}_{1,it}\), \({X}_{2,it}\), and \({X}_{3,it}\) represent rainfall, temperature, and wind speed of district \(i\) in season \(t\); \(\{{X}_{4,it},\dots ,{X}_{11,it}\}\) are the remaining covariates that are just linear predictors having fixed effects; \(\{{\xi }_{1,t},{\xi }_{2,t},{\xi }_{3,t}\}\) are the variablespecific seasonally varying structured random effects for which we specified cyclic firstorder random walk (RW1) priors (Eqs. 16–17).
Model 5: A spatiotemporal model (combining model 2 and model 3),
Model 6: The spatial ecological regression with seasonally varying coefficients for meteorological variables (combining model 2 and model 4),
Model 7: The time series regression model with seasonally varying coefficients for meteorological variables (combining model 3 and model 4),
Model 8: The spatiotemporal model with seasonally varying coefficients for meteorological variables (combining models 2–4),
Model 9: Model 8 + space–time interaction between spatially unstructured (\(u\)) and temporally unstructured (\(\gamma\)) random effects (referred to as interaction type I (KnorrHeld 2000)),
where \({\delta }_{it}\) is the mentioned interaction term in district \(i\), season \(t\); \({I}_{u}\) and \({I}_{\gamma }\) are 429 × 429 and 16 × 16 identity matrices, respectively and \(\otimes\) is the Kronecker product operator used to build a block matrix out of \({I}_{u}\) and \({I}_{\gamma }\).
Models 9–12 have the same formulation for \({\mu }_{it}\) and only the distribution of \({\delta }_{it}\) differs.
Model 10: Model 8 + space–time interaction between spatially unstructured (\(u\)) and temporally structured (\(\phi\)) random effects (interaction type II) as:
where \({R}_{\phi }\) is a 16 × 16 matrix with RW1 structure.
Model 11: Model 8 + space–time interaction between spatially structured (\(v\)) and temporally unstructured (\(\gamma\)) random effects (interaction type III),
where \({R}_{v}\) is a 429 × 429 matrix with ICAR proximity structure which we specified the prior in Eq. 10.
Model 12: Model 8 + space–time interaction between spatially structured (\(v\)) and temporally structured (\(\phi\)) random effects (interaction type IV),
Finally, it should be noted that all betweencovariate interaction effects have been entered into the models above using a forward approach. However, they did not show any subtle effects, and followingly we decided not to include them in main analyses.
We briefly introduce eight widely used model fit metrics that were utilized to compare the models in the study:
Deviance information criterion
Deviance information criterion (DIC) (Spiegelhalter et al. 2002) is a predictive accuracy measure consisting two components given by,
where \(y\) and \(\theta\) are the data and likelihood parameters vector, respectively; \(D\) is the expected deviance of the model; \(\overline{D }\) is the posterior mean of deviance and \({P}_{D}\) is the effective numbers of parameters. \({P}_{D}\) is defined via:
where \(D(\overline{\theta })\) is the evaluated deviance for the parameters posterior mean.
DIC is the most used measure in evaluating Bayesian models. In summary, lower values of \(\mathrm{DIC}\) indicate better performance as the first component (\(\overline{D }\)) indicates goodness of model fit and, simultaneously, the effective number of parameters (\({P}_{D}\)) is the penalty component for model complexity.
Watanabe–Akaike information criterion
Watanabe–Akaike information criterion (WAIC, also known as Widely Applicable Information Criterion) is developed by Watanabe (Watanabe 2010) to compare Bayesian models’ predictive accuracy given by,
where the first component implies log pointwise posterior predictive density and \({P}_{W}\) is the effective numbers of parameters. \({P}_{W}\) is defined via:
The overall structure and interpretation of WAIC is similar to DIC. Based on recent BHM studies, WAIC offers a number of benefits over DIC, since it employs a more genuine Bayesian approach by making the use of the whole posterior distribution (instead of posterior mean) in calculating the outofsample expectation. Also, it is preferable in case of singular models with nonidentifiable parameterization. On the other side, WAIC is less often utilized in practice due to the extra computational work required and potential challenges in dealing with structured data (Gelman et al. 2014).
Crossvalidated logarithmic score
A Bayesian leaveoneoutcrossvalidation measure is conditional predictive ordinate (CPO) (Pettit 1990) which is calculated for every observation as follows:
where \({y}_{it}\) is the observation in district \(i\), season \(t\), and \({y}_{it}\) is the vector of observations after removing \({y}_{it}\). Crossvalidated logarithmic score (LS) is defined as logscore of CPO values via:
where \(I\) and \(T\) are the number of districts and seasons, respectively. Lower values of \(\mathrm{LS}\) show higher predictive ability.
95% credible interval coverage
A summary statistic used for model selection is 95% credible interval coverage which is given by,
where \({\widehat{y}}_{97.5,i}\) and \({\widehat{y}}_{2.5,i}\) are the 97.5% and 2.5% quantiles of the posterior samples estimated in district \(i\), season \(t\), respectively and \(I(.)\) represents the indicator function. Closer values of coverage to 100 imply better fit.
Mean credible interval width (MCrIW)
Mean credible interval width (MCrIW) is another evaluation statistic defined as:
Higher values of \(\mathrm{MCrIW}\) demonstrate bigger uncertainty in the fitted values of the model.
Mean absolute error (MAE)
One of the most common metrics used to evaluate model fitness in the literature is mean absolute error (MAE). It is defined as:
Lower values of \(\mathrm{MAE}\) indicate better model fit. It should be noted that observing extremely large \(\mathrm{MAE}\) values is not necessarily a good clue and may declare overfitting issues.
Table 5
Table 6
Rights and permissions
About this article
Cite this article
Masinaei, M. Estimating the seasonally varying effect of meteorological factors on the districtlevel incidence of acute watery diarrhea among underfive children of Iran, 2014–2018: a Bayesian hierarchical spatiotemporal model. Int J Biometeorol 66, 1125–1144 (2022). https://doi.org/10.1007/s00484022022639
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00484022022639