Machine learning predictive modelling for identification of predictors of acute respiratory infection and diarrhoea in Uganda's rural and urban settings

Despite the widely known preventive interventions, the dyad of acute respiratory infections (ARI) and diarrhoea remain among the top global causes of mortality in under- 5 years. Studies on child morbidity have enormously applied "traditional" statistical techniques that have limitations in handling high dimension data, which leads to the exclusion of some variables. Machine Learning (ML) models appear to perform better on high dimension data (dataset with the number of features p (usually correlated) larger than the number of observations N). Using Uganda's 2006-2016 DHS pooled data on children aged 6-59 months, I applied ML techniques to identify rural-urban differentials in the predictors of child's diarrhoea and ARI. I also used ML to identify other omitted variables in the current child morbidity frameworks. The predictors were grouped into four categories: child characteristics, maternal characteristics, household characteristics and immunisation. I used 90% of the datasets as a training sets (dataset used to fit (train) a prediction model), which were tested or validated (dataset (pseudo new) used for evaluating the performance of the model on a new dataset) on 10% and 30% datasets. The measure of prediction was based on a 10-fold cross-validation (resampling technique). The gradient-boosted machine (ML technique) was the best-selected model for the identification of the predictors of ARI (Accuracy: 100% -rural and 100%-urban) and diarrhoea (Accuracy: 70%-rural and 100%-urban). These factors relate to the household's structure and composition, which is characterised by poor hygiene and sanitation and poor household environments that make children more suspectable of developing these diseases; maternal socio-economic factors such as education, occupation, and fertility (birth order); individual risk factors such as child age, birth weight and nutritional status; and protective interventions (immunisation). The study findings confirm the notion that ARI and diarrhoea risk factors overlap. The results highlight the need for a holistic approach with multisectoral emphasis in addressing the occurrence of ARI and diarrhoea among children. In particular, the results provide an insight into the importance of implementing interventions that are responsive to the unique structure and composition of the household. Finally, alongside traditional models, machine learning could be applied in generating research hypotheses and providing insight into the selection of key variables that should be considered in the model.

Spatial Co-Morbidity of Childhood Acute Respiratory Infection, Diarrhoea and Stunting in Nigeria

In low- and middle-income countries, children aged below 5 years frequently suffer from disease co-occurrence. This study assessed whether the co-occurrence of acute respiratory infection (ARI), diarrhoea and stunting observed at the child level could also be reflected ecologically. We considered disease data on 69,579 children (0–59 months) from the 2008, 2013, and 2018 Nigeria Demographic and Health Surveys using a hierarchical Bayesian spatial shared component model to separate the state-specific risk of each disease into an underlying disease-overall spatial pattern, common to the three diseases and a disease-specific spatial pattern. We found that ARI and stunting were more concentrated in the north-eastern and southern parts of the country, while diarrhoea was much higher in the northern parts. The disease-general spatial component was greater in the north-eastern and southern parts of the country. Identifying and reducing common risk factors to the three conditions could result in improved child health, particularly in the northeast and south of Nigeria.

Women's autonomy in healthcare decision-making and healthcare seeking behaviour for childhood illness in Ghana: Analysis of data from the 2014 Ghana Demographic and Health Survey

INTRODUCTION

The capacity of women to decide on their healthcare plays a key role in their health. In this study, we examined the association between women's healthcare decision-making capacity and their healthcare seeking behaviour for childhood illnesses in Ghana.

MATERIALS AND METHODS

We used data from the 2014 Ghana Demographic and Health Survey. A total sample of 2,900 women with children less than 5 years was used for the analysis. Data were processed and analysed using STATA version 14.0. Chi-square test of independence and binary logistic regression were carried out to generate the results. Statistical significance was pegged at 95% confidence intervals (CIs). We relied on the 'Strengthening the Reporting of Observational Studies in Epidemiology' (STROBE) statement in writing the manuscript.

RESULTS

Out of the 2,900 women, approximately 25.7% could take healthcare decisions alone and 89.7% sought healthcare for childhood illnesses. Women who decided alone on personal healthcare had 30% reduced odds of seeking healthcare for childhood illnesses compared to those who did not decide alone [AOR = 0.70, CI = 0.51-0.97]. With age, women aged 45-49 had 69% reduced odds of seeking healthcare for childhood illnesses compared to those aged 25-29 [AOR = 0.31, CI = 0.14-0.70]. Women from the Northern and Upper West regions had 72% [AOR: 0.28, CI: 0.11-0.70] and 77% [AOR: 0.23, CI: 0.09-0.58] reduced odds of seeking healthcare for childhood illnesses respectively, compared to those from the Western region.

CONCLUSION

Ghanaian women with autonomy in healthcare decision-making, those who were older and those from the Northern and Upper West regions were less likely to seek healthcare for childhood illness. To reduce childhood mortalities and morbidities in Ghana, we recommend educating women such as those who take healthcare decisions alone, older women and women from deprived regions like the Northern and Upper West regions on the need to seek healthcare for childhood illnesses.

Examining the spatial variations of co-morbidity among young children in Ethiopia

Background

Addressing the issues of childhood comorbidity remains a crucial global public health issue due to its consequences in child wellbeing. This study aims to account for nonlinear, spatial effect and to evaluate spatial variation in childhood co-morbidity at cluster level while controlling for important risk factors.

Methods

Using the 2016 Ethiopia DHS data, a multinomial logistic model was assessed by linear, nonlinear and random effects. The study also employed a spatial analysis tool which is Getis-Ord to identify hotspot areas of child comorbidity at the cluster level. The model with fixed, nonlinear and spatial effects identified as the best model to identify risk factors related to the coexistence of childhood illnesses.

Results

The results indicated that statistically significant high hotspots of comorbidity were found in Tigray and Oromia whereas low hotspots were found in Harari and Somali regions. Children between 10 and 15 months old were at high risk of co-morbidity in Ethiopia. Besides, our findings revealed that being male children, not-breastfed children, from households lack of toilet facility, children from households who use spring water, children born first, children from working mother, anemic children and children from uneducated mother are at high risk of multiple illnesses.

Conclusions

Comorbidity in childhood is not random in the country, with high hotspots of comorbidity in the regions of Tigray and Oromia. The results show a critical upshot for a combined morbidity control method for decreasing children’s illnesses and death. The maps remain novel to design appropriate healthcare interventions at regional as well as cluster level. Regions with high hotspots of child comorbidity should be considered for health healthcare interventions.

Household relationships and healthcare seeking behaviour for common childhood illnesses in sub-Saharan Africa: a cross-national mixed effects analysis

Background

Intra-household dynamics play crucial roles in utilisation of healthcare services for children. We investigated the influence of household relationships on healthcare seeking behaviour for common childhood illnesses in four sub-Sahara African regions.

Methods

Data on 247,061 under-five children were extracted from recent Demographic and Health Surveys conducted between 2012 and 2016 in 25 countries. Data were combined and analysed per sub-region. Dependent variables (DVs) were uptake of health facility care for diarrhea and Acute Respiratory Tract Infection (ARI) symptoms. The main independent variable (IV) was household relationship which was represented by maternal marital profile (marital status, family type and number of marriages) and maternal relationship to household head. Mixed effects logit models were fitted to assess independent relationship between the IVs and DVs with adjustment for relevant demographic and socio-economic characteristics at 5% significance level.

Results

The percentage of children who received care for diarrhea and ARI symptoms from health facilities across sub-regions was: Western Africa (WA) 42.4, 44.1%; Central Africa (CA) 32.6, 33.9%; Eastern Africa (EA) 41.5, 48.7% and Southern Africa (SA) 58.9, 62.7%. Maternal marital profile was not associated with healthcare seeking behaviour for diarrhea and ARI symptoms in any of the sub-regions. Children whose mothers were daughter/daughter-in-law to household head were significantly less likely to be taken to health facility for diarrhea treatment in Eastern Africa (AOR = 0.81, CI: 0.51–0.95). Having a mother who is the head of household was significantly associated with higher odds of facility care for ARI symptoms for children from Western (AOR = 1.20, CI: 1.02–1.43) and Southern Africa (AOR = 1.49, CI: 1.20–1.85).

Conclusion

The type of relationship between mother of under-fives and head of households affect health seeking behaviour for treatment of diarrhea and ARI symptoms in Eastern, Western and Southern Africa. Countries in these regions need to adapt best practices for promoting healthcare utilisation for children such that household relationship does not constitute barriers.

Electronic supplementary material

The online version of this article (10.1186/s12913-019-4142-x) contains supplementary material, which is available to authorized users.

Environmental Risks Associated with Symptoms of Acute Respiratory Infection among Preschool Children in North-Western and South-Southern Nigeria Communities

The exposure-disease-stress model places young children in their physical and social contexts and considers the extent and intensity of associational links to symptoms of acute respiratory infection (ARI), taking in to account a range of biological, social, and environment components. This study uses the 2013 Nigeria Demographic and Health Survey to assess the individual and environmental risks present in the North-Western and South-Southern Nigerian communities and examines their associations with ARI symptoms. The descriptive findings show that the prevalence of ARI symptoms is significantly higher among preschool children in the North-Western province (5.7%) than in the South-Southern province (1.4%) (p < 0.001). In addition to regional differences, multilevel logistic models further indicate that the increased likelihood of a child suffering from ARI symptoms is significantly associated with the dry season (aOR 1.42; 95% CI: 1.02–1.97) and household poverty (aOR 1.42; 95% CI: 1.01–1.99), even after adjusting for the cooking fuel used and various other characteristics of the children, households, and communities. These findings underscore the importance of taking into account environmental risks when addressing specific regional variations in ARI symptoms, because these determinants differ between communities in Nigeria. As it is imperative to achieve minimum levels of child health, in order to improve economic development across regions, future health policies aiming to promote child health will benefit from taking a region-specific perspective into consideration.

Factors contributing to regional inequalities in acute respiratory infections symptoms among under-five children in Nigeria: a decomposition analysis

BACKGROUND

Acute respiratory infections (ARI) are major causes of morbidity and mortality in many low-income countries. Although factors associated with ARI symptoms in children under 5 years of age have been identified; however, variation in their prevalence resulting from regional-specific proximate determinants has received little attention. Therefore, we aim to investigate the specific regional determinants of overall and wealth-related inequality in children having ARI in Nigeria over a decade.

METHODS

We analyzed trends in development of ARI symptoms among children under 5 years of age in Nigeria using nationally representative cross sectional surveys carried out in 2003, 2008 and 2013. Overall- and household wealth index based- inequality in the distribution of prevalence of ARI symptoms were estimated by region using Gini index and Concentration Index, respectively. Multivariate logistic regressions for complex survey and decomposition analysis for both indexes were used to calculate percentual contribution.

RESULTS

We found a decreasing trend in development of ARI symptoms over the decade between regions. Children in South Western region had reduced likelihood of developing the symptoms. Concentration index (CI) for the prevalence of ARI symptoms over the years and across regions had negative values (all p < 0.05). Gini index (GI) varies from 0.21 in North East to 0.62 in South Western region. Furthermore, the mapping showed that the extent at which both inequalities contribute to ARI symptoms prevalence in each region is different. The four major sources of wealth-related inequalities were poor households, no maternal education, biomass cooking, and rural area. The major contributors to overall inequalities were having a child aged 6 to 23 months, having no maternal education, having no vaccination card, and having a high birth order/short birth interval.

CONCLUSIONS

Although ARI prevalence decreased over the decade, it has remained unequally distributed between regions and over the time. The sources of those inequalities are context sensitive. Thus, in future health promotion initiatives, it is imperative to account for regional variations in the distribution of ARI.

Socio-demographic and environmental determinants of infectious disease morbidity in children under 5 years in Ghana

BACKGROUND

Globally, diarrhoea and acute respiratory infections (ARIs) have been identified as major threats to child survival. In Ghana, the two conditions are among the top three causes of morbidity and mortality among children under 5 years. An in-depth analysis of the factors associated with these two diseases is warranted, because of their high degree of fatality and also it provides a basis for intervention planning.

OBJECTIVES

To investigate socio-demographic and environmental factors associated with infectious disease morbidity in children under 5 years old in Ghana.

DESIGN

Population-based cross-sectional survey. The study sample comprised 2,790 children aged 0-59 months, drawn from the Ghana Demographic and Health Surveys. The mothers reported whether their children under age 5 had been ill with a cough accompanied by short, rapid breathing (ARI), or diarrhoea with the presence of blood or mucus in the stool, in the 2 weeks preceding the survey.

RESULTS

Children in the 6-11, 12-23, and 24-59 months age groups had, respectively, 3.48 (95% CI=2.23, 5.44), 4.57 (95% CI=3.03, 6.90), and 1.93 (95% CI=1.30, 2.87) increased odds of getting diarrhoea infection compared to those in the youngest age category (0-5). Similarly, children in the 6-11, 12-23, and 24-59 months age brackets were, respectively, 2.64 (95% CI=1.76, 3.97), 2.63 (95% CI=1.81, 3.83), and 1.83 (95% CI=1.29, 2.59) times more likely to have cough compared to children in 0-5 months age brackets. Children who were not breastfeeding had higher odds of childhood diarrhoea (OR=1.33, 95% CI=1.03, 1.73) compared to those who were breastfeeding. Compared to children who were living in households without co-wives, children who were living in households with co-wives had 1.74 increased odds of diarrhoea (95% CI=1.33, 2.27). A unit increase in maternal opinion regarding wife beating was associated with 14% reduced odds of diarrhoea (OR=0.86, 95% CI=0.80, 0.91), while a unit change in the women's attitude towards sex index was associated with 14% reduced odds of childhood cough (OR=0.86, 95% CI=0.77, 0.97).

CONCLUSIONS

Our results show that breastfeeding, polygamous marriage, and maternal decision-making autonomy are significant predictors of child morbidity. Therefore, implementing effective educational programmes that aim at promoting breastfeeding, empowering women, and discouraging polygamous marriages could help save many children from infectious disease morbidity in Ghana.

Environmental factors and childhood Fever in areas of the Ouagadougou Health and Demographic Surveillance System, Burkina Faso

Using data on 825 under-5 children from the Ouagadougou Health and Demographic Surveillance System collected in 2010, this article examines the effects of aspects of the immediate environment on childhood fever. Logit regression models were estimated to assess the effects of the quality of the local environment on the probability that a child is reported to have had a fever in the two weeks preceding the survey, after controlling for various demographic and socioeconomic variables. While the estimated impact of some environmental factors persisted in the full models, the effects of variables such as access to water and type of household waste management decreased in the presence of demographic, socioeconomic and neighbourhood factors. The management of waste water was found to significantly affect the occurrence of childhood fever. Overall, the results of the study call for more efforts to promote access to tap water to households at prices that are affordable for the local population, where the threats to child health appears to be greatest.

Determinants of childhood diarrhea among underfive children in Benishangul Gumuz Regional State, North West Ethiopia

BACKGROUND

Diarrhea is second only to pneumonia as the cause of child mortality worldwide. Developing countries particularly in Sub Saharan Africa including Ethiopia have a high burden of this disease. Studies showed that different factors were associated with the occurrence of childhood diarrhea. Therefore, this study was aimed to identify determinant factors of diarrhea in underfive children in Benishangul Gumuz Regional State, western Ethiopia.

METHOD

Demographic and Health Survey (DHS) data of 2011 was used for this study. The data was extracted from the National DHS data using data extraction tools. A total of 925 under five children were selected. The logistic regression model was employed to examine the determinants of childhood diarrhoea. Both bivariate and multivariate data analysis was performed using SPSS version 16.0.

RESULT

The results of this study indicated that low level of maternal education [AOR = 1.81, 95% CI (1.12,2.76)], absence of toilet facility [AOR = 3.5, 95% CI (2.4, 5.2)], improper child stool disposal methods [AOR = 2.05, 95%CI (1.36, 3.10)], having more than two under five children [AOR = 1.73, 95% CI (1.03, 2.93)], higher birth order [AOR = 6.1, 95% CI (3.1,12.2)] and the age of children [AOR = 1.9, 95% CI (1.2, 3.6)] were found to be the risk factors for childhood diarrhea after adjusting for other variables. When toilet facility was stratified by maternal education, it showed that children of mothers who had no education were the most vulnerable in the absence of toilet facilities [OR = 9.16, 95% CI (5.79, 14.48)].

CONCLUSION

Under poor environmental conditions, mothers with primary education and above protected their children against diarrhea better than mothers with no education. Thus, implementing effective educational programs that emphasize environmental health and sanitation practices and encouraging female school enrolment would reduce childhood diarrheal morbidity in the region.

Correlates of diarrhoea among children below the age of 5 years in Sudan

BACKGROUND

The Millennium Development Goals recognise child health and survival as an important socio-development issue.

OBJECTIVES

To determine the correlates of diarrhoea among children aged below 5 years in north Sudan.

METHODS

We conducted secondary data analysis of the Sudan Multiple Cluster Indicators Survey II.

RESULTS

Altogether, 23,295 children were included in the survey. Half (50.0%) of the children were males, and 22.5% of them were of age less than one year. Boys were 3% (p=0.044) more likely to have diarrhoea compared to girls. Compared with the oldest age group (48-59 months), children less than 6 months of age and those aged 36-47 months had 25% and 18% lower prevalence of diarrhoea, respectively, while children aged 6-24 months and those aged 24-35 months had 1.5 fold and 1.17 fold higher prevalence of diarrhoea. Children in urban areas were 6% more likely to have diarrhoea. Children from households with 1 or 2 people per room were 8% less likely to have diarrhoea compared to children from households with more than 3 people per room.

CONCLUSIONS

Diarrhoea was associated with child's age, gender, and social status. Our findings provide a useful baseline for interventions and comparisons with future studies.

Spatial variation of salt intake in Britain and association with socioeconomic status

OBJECTIVES

To evaluate spatial effects of variation and social determinants of salt intake in Britain.

DESIGN

Cross-sectional survey.

SETTING

Great Britain.

PARTICIPANTS

2105 white male and female participants, aged 19-64 years, from the British National Diet and Nutrition Survey 2000-2001.

PRIMARY OUTCOMES

Participants' sodium intake measured both with a 7-day dietary record and a 24-h urine collection. By accounting for important linear and non-linear risk factors and spatial effects, the geographical difference and spatial patterns of both dietary sodium intake and 24-h urinary sodium were investigated using Bayesian geo-additive models via Markov Chain Monte Carlo simulations.

RESULTS

A significant north-south pattern of sodium intake was found from posterior probability maps after controlling for important sociodemographic factors. Participants living in Scotland had a significantly higher dietary sodium intake and 24-h urinary sodium levels. Significantly higher sodium intake was also found in people with the lowest educational attainment (dietary sodium: coeff. 0.157 (90% credible intervals 0.003, 0.319), urinary sodium: 0.149 (0.024, 0.281)) and in manual occupations (urinary sodium: 0.083 (0.004, 0.160)). These coefficients indicate approximately a 5%, 9% and 4% difference in average sodium intake between socioeconomic groups.

CONCLUSIONS

People living in Scotland had higher salt intake than those in England and Wales. Measures of low socioeconomic position were associated with higher levels of sodium intake, after allowing for geographic location.

An audit of ear, nose and throat diseases in a tertiary health institution in South-western Nigeria

Introduction

This study is aimed at determining the pattern of ear, nose and throat diseases and their relationship with socio-demographic factors with auditing intent in a tertiary hospital in South-western Nigeria.

Methods

Medical records of patients managed at the Department of Otorhinolaryngology, University College Hospital, Ibadan, Nigeria from 2006 to 2010 were reviewed for all essential clinical data.

Results

There were 2641 (52.8%) males and 2360 (47.2%) females. Two thousand and fifty (41%) patients had age ≤15years old. Sixty three percent of the patients were Christians, 37% were Muslims and less than 1% had other religions. There were more patients in lower occupational classes than those in the upper classes. The average number of patients with ear, nose and throat diseases managed per month was eighty three. Patients with ear diseases were 3136 (62.7%), the nose diseases were 1153 (23.0%), the throat diseases were 479 (9.6%) and head/neck diseases were 233 (4.7%).

Conclusion

This study showed that otitis media, obstructive adenoid, foreign bodies in the ear and throat infections were the common ear, nose, throat disorders seen in patients aged ≤15years whereas, hearing loss, rhinosinusitis and tumors were the common disorders of ear, nose and throat seen in patients aged 16 years and above. Although these disorders are not yet considered to be of public health importance, they contribute significantly to the existing burden of health problems in our environment. Therefore, there is a need for improved public awareness on ear, nose and throat diseases.

Home gardening is associated with Filipino preschool children's dietary diversity

Dietary diversification through home gardening is a sustainable strategy that can address multiple micronutrient deficiencies. This cross-sectional study aimed to determine the association between home gardening and the dietary diversity of preschool-aged children. Households with children aged 2 to 5 years (n=200) were surveyed from Baras and Angono in the province of Rizal, Philippines in January 2008. Food security was determined based on the US Department of Agriculture Food Security Questionnaire. Dietary diversity score was based on the number of unique food groups consumed during the past 24 hours. The Student t test was performed to compare means between groups (households with gardens vs households without gardens) whereas proportions between groups were compared using Pearson's χ(2) analyses. Multiple linear regression was performed to model the adjusted regression coefficients for the quantitative outcome variables by exposure variable. Around 52.5% of children were from households with a fruit and/or vegetable garden. Children from households with gardens had higher dietary diversity scores whether using the all-inclusive dietary diversity score (6.12 vs 5.62; P=0.040) or applying a 10-g minimum intake for each food group (5.89 vs 5.37; P=0.044) compared with children who lived in homes without a garden. Children from households with gardens were significantly more likely to eat vegetables more frequently (χ(2)=9.06; P=0.029). The presence or absence of a garden was not significantly associated with food security. Having a home garden was positively associated with the child's diet diversity and with frequency of vegetable consumption. Households without gardens may benefit from interventions promoting gardens as a means to improve diet quality.

Orphanhood and vulnerability: a conduit to poor child health outcomes in Rwanda

The HIV epidemic in sub-Saharan Africa has caused many children to become orphaned and vulnerable. Recent studies show that orphaned and vulnerable children (OVC) lack the basic necessities for survival and development. These children are particularly at high risk of poor health and poverty. Although the poor health outcomes of these children are well documented, the complexities of the factors that mediate their health outcomes have not been systematically studied. The aim of this paper is to examine how the complex relationships between and within the proximate and socio-economic determinants mediate the poor health outcomes of children through their OVC status. The analyses considered graphical chain modelling of morbidity data from a sample of 3745 children aged below five years from the 2005 Rwandan Demographic and Health Survey. The results show that OVC status influences the risk of childhood morbidity both directly and indirectly and also as a conduit through which other significant proximate factors and socio-economic factors operate.

Diarrhoea, acute respiratory infection, and fever among children in the Democratic Republic of Congo

Several years of war have created a humanitarian crisis in the Democratic Republic of Congo (DRC) with extensive disruption of civil society, the economy and provision of basic services including health care. Health policy and planning in the DRC are constrained by a lack of reliable and accessible population data. Thus there is currently a need for primary research to guide programme and policy development for reconstruction and to measure attainment of the Millennium Development Goals (MDGs). This study uses the 2001 Multiple Indicators Cluster Survey to disentangle children's health inequalities by mapping the impact of geographical distribution of childhood morbidity stemming from diarrhoea, acute respiratory infection, and fever. We observe a low prevalence of childhood diarrhoea, acute respiratory infection and fever in the western provinces (Kinshasa, Bas-Congo and Bandundu), and a relatively higher prevalence in the south-eastern provinces (Sud-Kivu and Katanga). However, each disease has a distinct geographical pattern of variation. Among covariate factors, child age had a significant association with disease prevalence. The risk of the three ailments increased in the first 8-10 months after birth, with a gradual improvement thereafter. The effects of socioeconomic factors vary according to the disease. Accounting for the effects of the geographical location, our analysis was able to explain a significant share of the pronounced residual geographical effects. Using large scale household survey data, we have produced for the first time spatial residual maps in the DRC and in so doing we have undertaken a comprehensive analysis of geographical variation at province level of childhood diarrhoea, acute respiratory infection, and fever prevalence. Understanding these complex relationships through disease prevalence maps can facilitate design of targeted intervention programs for reconstruction and achievement of the MDGs.