Health And Demographic Surveillance Site Research Articles

Text mining of verbal autopsy narratives to extract mortality causes and most prevalent diseases using natural language processing

Verbal autopsy (VA) narratives play a crucial role in understanding and documenting the causes of mortality, especially in regions lacking robust medical infrastructure. In this study, we propose a comprehensive approach to extract mortality causes and identify prevalent diseases from VA narratives utilizing advanced text mining techniques, so as to better understand the underlying health issues leading to mortality. Our methodology integrates n-gram-based language processing, Latent Dirichlet Allocation (LDA), and BERTopic, offering a multi-faceted analysis to enhance the accuracy and depth of information extraction. This is a retrospective study that uses secondary data analysis. We used data from the Agincourt Health and Demographic Surveillance Site (HDSS), which had 16338 observations collected between 1993 and 2015. Our text mining steps entailed data acquisition, pre-processing, feature extraction, topic segmentation, and discovered knowledge. The results suggest that the HDSS population may have died from mortality causes such as vomiting, chest/stomach pain, fever, coughing, loss of weight, low energy, headache. Additionally, we discovered that the most prevalent diseases entailed human immunodeficiency virus (HIV), tuberculosis (TB), diarrhoea, cancer, neurological disorders, malaria, diabetes, high blood pressure, chronic ailments (kidney, heart, lung, liver), maternal and accident related deaths. This study is relevant in that it avails valuable insights regarding mortality causes and most prevalent diseases using novel text mining approaches. These results can be integrated in the diagnosis pipeline for ease of human annotation and interpretation. As such, this will help with effective informed intervention programmes that can improve primary health care systems and chronic based delivery, thus increasing life expectancy.

Levels, trends and inequalities in mortality among 5-19-year-olds in Tanzania: Magu Health and Demographic Surveillance Study (1995-2022).

For the past two decades, health priorities in Tanzania have focussed on children under-five, leaving behind the older children and adolescents (5-19 years). Understanding mortality patterns beyond 5 years is important in bridging a healthy gap between childhood to adulthood. We aimed to estimate mortality levels, trends, and inequalities among 5-19-year-olds using population data from the Magu Health and Demographic Surveillance Site (HDSS) in Tanzania and further compare the population level estimates with global estimates. Using data from the Magu HDSS from 1995 to 2022, from Kaplan Meir survival probabilities, we computed annual mortality probabilities for ages 5-9, 10-14 and 15-19 and determined the average annual rate of change in mortality by fitting the variance weighted least square regression on annual mortality probabilities. We compared 5-19 trends with younger children aged 1-4 years. We further disaggregated mortality by sex, area of residence and wealth tertiles, and we computed age-stratified risk ratios with respective 95% confidence intervals (CIs) using Cox proportional hazard model to determine inequalities. We further compared population-level estimates in all-cause mortality with global estimates from the United Nations Inter-agency Group for Child Mortality Estimation and the Global Burden of Disease study by computing the relative differences to the estimates. Mortality declined steadily among the three age groups from 1995 to 2022, whereby the average annual rate of decline increased with age (2.2%, 2.7%, and 2.9% for 5-9-, 10-14-, and 15-19-year-old age groups, respectively). The pace of this decline was lower than that of younger children aged 1-4 years (4.8% decline). We observed significant mortality inequalities with boys, those residing in rural areas, and those from poorest wealth tertiles lagging behind. While Magu estimates were close to global estimates for the 5-9-year-old age group, we observed divergent results for adolescents (10-19 years), with Magu estimates lying between the global estimates. The pace of mortality decline was lower for the 5-19-year-old age group compared to younger children, with observable inequalities by socio-demographic characteristics. Determining the burden of disease across different strata is important in the development of evidence-based targeted interventions to address the mortality burden and inequalities in this age group, as it is an important transition period to adulthood.

Underweight associated with water, sanitation, and hygiene among women of reproductive age in Arba Minch Health and Demographic Surveillance Site, Southern Ethiopia.

Ethiopia is one of the sub-Saharan African countries where underweight among women of reproductive age is high, and it is predisposing to low birth weight, preterm birth, and reduced resistance to infections. Poor water, sanitation, and hygiene (WASH) and drinking water polluted with disease-causing microorganisms lead to undernutrition. Therefore, this study aimed to assess the association between WASH and underweight among women of reproductive age at Arba Minch Health and Demographic Surveillance Site (HDSS), Southern Ethiopia. A community-based cross-sectional study was conducted at Arba Minch HDSS, South Ethiopia, from April to May 2022. Women of reproductive age were included in the study. Data were collected using a pre-tested questionnaire and checklist. Multivariate analysis was conducted to assess the association separately for WASH indicators with underweight status of the women by adjusting for sociodemographic characteristics. Adjusted Odds Ratio (AOR) with a 95% confidence interval was used to assess the association. The prevalence of underweight and overweight/obesity among women of reproductive age was 7.82% (95% CI: 5.60, 10.81) and 12.32% (95% CI: 9.50, 15.83), respectively. Latrine hygiene and use were significantly associated with being underweight. The odds of being underweight among those who use clean latrine were 0.43 (95% CI: 0.20, 0.92) compared to those who use unclean latrine or do not have toilet facilities. The utilization of clean latrine facilities is negatively associated with being underweight among women of reproductive age. From WASH components, latrine utilization and keeping the latrine clean to use need to get focused on reducing the risk of being underweight among women of reproductive age.

Family network and household composition: a longitudinal dataset derived from the Karonga Health and Demographic Surveillance System, in rural Malawi.

Proximity to family, household composition, and structure are often studied as outcomes and as explanatory factors in a wide range of scientific disciplines. Here, we describe a large longitudinal dataset (currently including data from over 70,000 individuals from 2004 to 2017), including data on household structure, proximity to kin, population density, and other socio-demographic factors derived from data from the Karonga Health and Demographic Surveillance Site (HDSS) in Northern Malawi. We present how the dataset is generated, list some examples of how it can be used, and provide information on the limitations that affect the types of analyses that can be carried out.

Early life ambient air pollution, household fuel use, and under-5 mortality in Ghana

IntroductionEnvironmental exposures, such as ambient air pollution and household fuel use affect health and under-5 mortality (U5M) but there is a paucity of data in the Global South. This study examined early-life exposure to ambient particulate matter with a diameter of 2.5 µm or less (PM2.5), alongside household characteristics (including self-reported household fuel use), and their relationship with U5M in the Navrongo Health and Demographic Surveillance Site (HDSS) in northern Ghana. MethodsWe employed Satellite-based spatiotemporal models to estimate the annual average PM2.5 concentrations with the Navrongo HDSS area (1998 to 2016). Early-life exposure levels were determined by pollution estimates at birth year. Socio-demographic and household data, including cooking fuel, were gathered during routine surveillance. Cox proportional hazards models were applied to assess the link between early-life PM2.5 exposure and U5M, accounting for child, maternal, and household factors. FindingsWe retrospectively studied 48,352 children born between 2007 and 2017, with 1872 recorded deaths, primarily due to malaria, sepsis, and acute respiratory infection. Mean early-life PM2.5 was 39.3 µg/m3, and no significant association with U5M was observed. However, Children from households using “unclean” cooking fuels (wood, charcoal, dung, and agricultural waste) faced a 73 % higher risk of death compared to those using clean fuels (adjusted HR = 1.73; 95 % CI: 1.29, 2.33). Being born female or to mothers aged 20–34 years were linked to increased survival probabilities. InterpretationThe use of “unclean” cooking fuel in the Navrongo HDSS was associated with under-5 mortality, highlighting the need to improve indoor air quality by introducing cleaner fuels.

Understanding the health system utilisation and reasons for avoidable mortality after fatal injury within a Three-Delays framework in Karonga, Northern Malawi: a retrospective analysis of verbal autopsy data

ObjectivesTo use verbal autopsy (VA) data to understand health system utilisation and the potential avoidability associated with fatal injury. Then to categorise any evident barriers driving avoidable delays to care...

Family network and household composition: a longitudinal dataset derived from the Karonga Health and Demographic Surveillance System, in rural Malawi

Proximity to family, household composition, and structure are often studied as outcomes and as explanatory factors in a wide range of scientific disciplines. Here, we describe a large longitudinal dataset (currently including data from over 70,000 individuals from 2004 to 2017), including data on household structure, proximity to kin, population density, and other socio-demographic factors derived from data from the Karonga Health and Demographic Surveillance Site (HDSS) in Northern Malawi. We present how the dataset is generated, list some examples of how it can be used, and provide information on the limitations that affect the types of analyses that can be carried out.

Developing Climate Change and Health Impact Monitoring with eHealth at the South East Asia Community Observatory and Health and Demographic Surveillance Site, Malaysia (CHIMES).

Malaysia is projected to experience an increase in heat, rainfall, rainfall variability, dry spells, thunderstorms, and high winds due to climate change. This may lead to a rise in heat-related mortality, reduced nutritional security, and potential migration due to uninhabitable land. Currently, there is limited data regarding the health implications of climate change on the Malaysian populace, which hinders informed decision-making and interventions. This study aims to assess the feasibility and reliability of using sensor-based devices to enhance climate change and health research within the SEACO health and demographic surveillance site (HDSS) in Malaysia. We will particularly focus on the effects of climate-sensitive diseases, emphasizing lung conditions like chronic obstructive pulmonary disease (COPD) and asthma. In our mixed-methods approach, 120 participants (>18 years) from the SEACO HDSS in Segamat, Malaysia, will be engaged over three cycles, each lasting 3 weeks. Participants will use wearables to monitor heart rate, activity, and sleep. Indoor sensors will measure temperature in indoor living spaces, while 3D-printed weather stations will track indoor temperature and humidity. In each cycle, a minimum of 10 participants at high risk for COPD or asthma will be identified. Through interviews and questionnaires, we will evaluate the devices' reliability, the prevalence of climate-sensitive lung diseases, and their correlation with environmental factors, like heat and humidity. We anticipate that the sensor-based measurements will offer a comprehensive understanding of the interplay between climate-sensitive diseases and weather variables. The data is expected to reveal correlations between health impacts and weather exposures like heat. Participant feedback will offer perspectives on the usability and feasibility of these digital tools. Our study within the SEACO HDSS in Malaysia will evaluate the potential of sensor-based digital technologies in monitoring the interplay between climate change and health, particularly for climate-sensitive diseases like COPD and asthma. The data generated will likely provide details on health profiles in relation to weather exposures. Feedback will indicate the acceptability of these tools for broader health surveillance. As climate change continues to impact global health, evaluating the potential of such digital technologies is crucial to understand its potential to inform policy and intervention strategies in vulnerable regions.

Comparing full immunisation status of children (0–23 months) between slums of Kampala City and the rural setting of Iganga District in Uganda: a cross-sectional study

BackgroundImmunisation remains the most cost-effective public health intervention in preventing morbidity and mortality due to Vaccine-Preventable Diseases (VPDs). The study aims to compare the differences in immunisation coverage amongst children aged 0 to 23 months living in slums of Kampala city and Iganga as rural districts in Uganda.MethodsThis study utilises data from a cross-sectional survey done in 2019 in the slums of Kampala City and the rural district of Iganga within the Health and Demographic Surveillance Site (HDSS). It included 1016 children aged 0–23 months and their parents. A logistic regression model was used to analyse the relationship between multiple independent variables and the binary dependent variables (fully immunised) using Stata statistical software. The measures of association were odds ratios reported with a corresponding 95% confidence interval.ResultsOut of the 1016 participants, 544 participants live in the rural area and 472 participants in the slums. Slums had 48.9% (n = 231) of fully immunised children whilst rural areas had 43.20% (n = 235). The multivariate analysis showed that children living in slums are more likely to be fully immunised as compared to their counterparts in rural areas (Odds ratio:1.456; p = 0.033; CI:1.030–2.058). Immunisation coverage for BCG (98.9%), Polio 0 (88.2%), Penta1 (92.7%), and Pneumo1 (89.8%) were high in both settlements. However, the dropout rate for subsequent vaccines was high 17%, 20% and 41% for Penta, pneumococcal and rota vaccines respectively. There was poor uptake of the new vaccines with slums having 73.4% and 47.9% coverage for pneumococcal and rota vaccines respectively and rural areas had 72.1% and 7.5% for pneumococcal and rota vaccines respectively.ConclusionThe low full immunisation status in this study was attributed to the child’s residence and the occupation of the parents. Lack of education and poor access to messages on immunisation (inadequate access to mass media) are other contributing factors. Educational messages on the importance of immunisation targeting these underserved populations will improve full immunisation coverage.

Widening the lens of population-based health research to climate change impacts and adaptation: the climate change and health evaluation and response system (CHEERS).

Climate change significantly impacts health in low-and middle-income countries (LMICs), exacerbating vulnerabilities. Comprehensive data for evidence-based research and decision-making is crucial but scarce. Health and Demographic Surveillance Sites (HDSSs) in Africa and Asia provide a robust infrastructure with longitudinal population cohort data, yet they lack climate-health specific data. Acquiring this information is essential for understanding the burden of climate-sensitive diseases on populations and guiding targeted policies and interventions in LMICs to enhance mitigation and adaptation capacities. The objective of this research is to develop and implement the Change and Health Evaluation and Response System (CHEERS) as a methodological framework, designed to facilitate the generation and ongoing monitoring of climate change and health-related data within existing Health and Demographic Surveillance Sites (HDSSs) and comparable research infrastructures. CHEERS uses a multi-tiered approach to assess health and environmental exposures at the individual, household, and community levels, utilizing digital tools such as wearable devices, indoor temperature and humidity measurements, remotely sensed satellite data, and 3D-printed weather stations. The CHEERS framework utilizes a graph database to efficiently manage and analyze diverse data types, leveraging graph algorithms to understand the complex interplay between health and environmental exposures. The Nouna CHEERS site, established in 2022, has yielded significant preliminary findings. By using remotely-sensed data, the site has been able to predict crop yield at a household level in Nouna and explore the relationships between yield, socioeconomic factors, and health outcomes. The feasibility and acceptability of wearable technology have been confirmed in rural Burkina Faso for obtaining individual-level data, despite the presence of technical challenges. The use of wearables to study the impact of extreme weather on health has shown significant effects of heat exposure on sleep and daily activity, highlighting the urgent need for interventions to mitigate adverse health consequences. Implementing the CHEERS in research infrastructures can advance climate change and health research, as large and longitudinal datasets have been scarce for LMICs. This data can inform health priorities, guide resource allocation to address climate change and health exposures, and protect vulnerable communities in LMICs from these exposures.

Performance evaluation of machine learning and Computer Coded Verbal Autopsy (CCVA) algorithms for cause of death determination: A comparative analysis of data from rural South Africa.

Computer Coded Verbal Autopsy (CCVA) algorithms are commonly used to determine the cause of death (CoD) from questionnaire responses extracted from verbal autopsies (VAs). However, they can only operate on structured data and cannot effectively harness information from unstructured VA narratives. Machine Learning (ML) algorithms have also been applied successfully in determining the CoD from VA narratives, allowing the use of auxiliary information that CCVA algorithms cannot directly utilize. However, most ML-based studies only use responses from the structured questionnaire, and the results lack generalisability and comparability across studies. We present a comparative performance evaluation of ML methods and CCVA algorithms on South African VA narratives data, using data from Agincourt Health and Demographic Surveillance Site (HDSS) with physicians' classifications as the gold standard. The data were collected from 1993 to 2015 and have 16,338 cases. The random forest and extreme gradient boosting classifiers outperformed the other classifiers on the combined dataset, attaining accuracy of 96% respectively, with significant statistical differences in algorithmic performance (p < 0.0001). All our models attained Area Under Receiver Operating Characteristics (AUROC) of greater than 0.884. The InterVA CCVA attained 83% Cause Specific Mortality Fraction accuracy and an Overall Chance-Corrected Concordance of 0.36. We demonstrate that ML models could accurately determine the cause of death from VA narratives. Additionally, through mortality trends and pattern analysis, we discovered that in the first decade of the civil registration system in South Africa, the average life expectancy was approximately 50 years. However, in the second decade, life expectancy significantly dropped, and the population was dying at a much younger average age of 40 years, mostly from the leading HIV related causes. Interestingly, in the third decade, we see a gradual improvement in life expectancy, possibly attributed to effective health intervention programmes. Through a structure and semantic analysis of narratives where experts disagree, we also demonstrate the most frequent terms of traditional healer consultations and visits. The comparative approach also makes this study a baseline that can be used for future research enforcing generalization and comparability. Future study will entail exploring deep learning models for CoD classification.

Aflatoxin exposure among children of age 12–59 Months in Butajira District, South-Central Ethiopia: a community based cross-sectional study

BackgroundThe continued provision of safe food, free of aflatoxin remains a huge challenge in developing countries. Despite several favourable climatic conditions that facilitate aflatoxin contamination in Ethiopia, there is little information showing aflatoxin exposure in children. Therefore, this study assessed aflatoxin exposure among young children in Butajira district, South-Central Ethiopia.MethodsCommunity based cross-sectional study stratified by agro-ecology was employed in Health and Demographic Surveillance Site (HDSS) of Butajira. The study included 332 children aged 12–59 months and were selected by simple random sampling technique using the HDSS registration number as a sampling frame. We collected data on dietary practice and aflatoxin exposure. Aflatoxin M1 concentration in urine was measured by Enzyme-Linked Immunosorbent assay (ELISA). The data analysis was carried out using STATA.ResultsDetectable urinary Aflatoxin M1 was found in 62.4% (95% CI: 56.9 – 67.5%) of the children at a level ranging from 0.15 to 0.4 ng/ml. Children living in lowland agro-ecological zone had [AOR = 2.11 (95% CI; 1.15, 3.88] odds of being exposed to aflatoxin as compared to children living in highland agro-ecological zone. Children at lower socio-economic status [AOR = 0.27 (95% CI; 0.14, 0.50] and medium socio-economic status [AOR = 0.47 (95% CI; 0.25, 0.87] had 73% and 53% lower odds of being exposed to aflatoxin as compared to children in the higher socio-economic status, respectively.ConclusionsAflatoxin exposure among young children was very high in South-Central Ethiopia. This high aflatoxin exposure might emphasize the need for aflatoxin exposure mitigation strategies in Ethiopia. Especially, raising awareness of the community towards aflatoxin exposure is very crucial. In addition, further research is required to assess long-term aflatoxin exposure and its association with child growth and development.

Prevalence of COVID-19 Vaccine Hesitancy in a Rural Setting: A Case Study of DIMAMO Health and Demographic Surveillance Site, Limpopo Province of South Africa

Background: The primary purpose of introducing the COVID-19 vaccine was to fight the pandemic. However, the vaccine was not well received worldwide. This challenge has threatened the effective implementation and roll-out of COVID-19 immunization campaigns. The challenge of vaccine hesitancy was reported to be more prevalent in rural areas due to various factors such as cultural beliefs, misinformation, poverty, lack of education, and distrust of vaccines. Yet there seems to be a scarcity of studies determining the prevalence of vaccine hesitancy in deep rural areas of Limpopo. Purpose: The study aimed to explore the prevalence of COVID-19 hesitancy among the rural black population in South Africa. Study Methods: A longitudinal quantitative study was conducted with data from the DIMAMO Health and Demographic Surveillance Site (HDSS) database for 2020 to 2022. A non-probability total sampling technique was used to select the respondents. Trained fieldworkers collected data using an electronic data capture questionnaire. A comparison of categorical variables was performed using Chi-Square in SPSS version 26 and the statistical significance was set at p &lt; 0.05, with a 95% confidence interval to analyze the data. Results: The limited availability of vaccination sites in Limpopo Province, South Africa, was associated with a reduced certainty that the vaccine would be accepted, as reported in the current study. The prevalence of visiting traditional healers was significantly higher in non-vaccinated than vaccinated participants among the rural black population, indicating a different cultural belief among the rural black population that existed before and during the COVID-19 pandemic. Conclusion: The present study findings show diverse factors of concern associated with vaccination hesitancy for COVID-19 among rural black people. Lack of education, gender, not being diagnosed with COVID-19, not being a Christian, visiting traditional healers, vaccine mistrust, unknown side effects, and a lack of confidence in the vaccine itself. Conspiracy theories were factors that impacted vaccine acceptance among black people living in rural areas. The prevalence of visiting traditional healers was significantly higher in non-vaccinated than vaccinated participants. Therefore, the present study findings emphasize the need to collectively integrate and utilize the traditional healers in the South African healthcare system.

Breast Cancer Screening in Semi-Rural Malaysia: Utilisation and Barriers.

Breast cancer (BC) is the commonest cancer in Malaysia. Delayed diagnosis is a significant cause of BC mortality in the country. Early diagnosis and screening are vital strategies in mortality reduction. This study assessed the level of utilisation and barriers for breast self-examination (BSE), clinical breast examination (CBE) and mammogram in a semi-rural population in Malaysia and compared these across the different ethnic groups. This cross-sectional study was conducted among women aged 40 years and above, embedded within a health and demographic surveillance site (HDSS) in Segamat, Malaysia. Trained data collectors collected data on screening and barriers during home visits. Study participants (n = 250) were aged 59.4 ± 10.9 years and represented Malaysia’s three major ethnic groups. Practice of regular BSE, CBE uptake (ever) and mammogram (ever) was 23.2%, 36% and 22.4%, respectively. Regular BSE practice was highest in the Malay ethnic group and least among the Chinese. Regular CBE was very low in all ethnic groups (<5%). Mammogram uptake was highest among Chinese (34.4%), followed by Indians (30.4%) and Malays (16.6%). After adjusting for other socio-demographic variables, Malay ethnicity was positively associated with regular BSE (adjusted OR = 5.26, 95% CI 2.05, 13.50) and negatively associated with having had a mammogram (adjusted OR = 0.3, 95% CI 0.15, 0.57). Lower education was negatively associated (adjusted OR = 0.36, 95% CI 0.17, 0.74) with mammogram attendance (ever). Emotional and financial barriers were the most reported types of barriers, specifically, fear of diagnosis (74.8%), cost of diagnosis (69.6%) and fear of losing a breast (66.4%). Malay women more commonly reported most barriers compared to other ethnic groups. Screening uptake was low among semi-rural women in Malaysia. Implementing culturally appropriate interventions that consider ethnic differences is crucial to empowering women to engage in BC screening initiatives in these communities.

The impact of equity factors on receipt of timely appropriate care for children with suspected malaria in eastern Uganda

IntroductionMalaria accounts for more than one-tenth of sub-Saharan Africa’s 2.8 million annual childhood deaths, and remains a leading cause of post-neonatal child mortality in Uganda. Despite increased community-based treatment in Uganda, children continue to die because services fail to reach those most at risk. This study explores the influence of two key equity factors, socioeconomic position and rurality, on whether children with fever in eastern Uganda receive timely access to appropriate treatment for suspected malaria.MethodsThis was a cross-sectional study in which data were collected from 1094 caregivers of children aged 6–59 months on: illness and care-seeking during the previous two weeks, treatment received, and treatment dosing schedule. Additional data on rurality and household socioeconomic position were extracted from the Iganga-Mayuge Health and Demographic Surveillance Site (HDSS) database. A child was considered to have received prompt and appropriate care for symptoms of malaria if they received the recommended drug in the recommended dosing schedule on the day of symptom onset or the next day. Unadjusted and adjusted logistic regression models were developed to explore associations of the two equity factors with the outcome. The STROBE checklist for observational studies guided reporting.ResultsSeventy-four percent of children had symptoms of illness in the preceding two weeks, of which fever was the most common. Children from rural households were statistically more likely to receive prompt and appropriate treatment with artemisinin-combination therapy than their semi-urban counterparts (OR 2.32, CI 1.17–4.59, p = 0.016). This association remained significant following application of an adjusted regression model that included the age of the child, caregiver relationship, and household wealth index (OR 2.4, p = 0.036). Wealth index in its own right did not exert a significant effect for children with reported fever (OR for wealthiest quintile = 1.02, CI 0.48–2.15, p = 0.958).ConclusionsThe findings from this study help to identify the role and importance of two key equity determinants on care seeking and treatment receipt for fever in children. Whilst results should be interpreted within the limitations of data and context, further studies have the potential to assist policy makers to target inequitable social and spatial variations in health outcomes as a key strategy in ending preventable child morbidity and mortality.

The Impact of Rainfall Variability on Diets and Undernutrition of Young Children in Rural Burkina Faso.

Background: Climate change and consequent increases in rainfall variability may have negative consequences for the food production of subsistence farmers in West Africa with adverse impacts on nutrition and health. We explored the pathway from rainfall through diet up to child undernutrition for rural Burkina Faso.Methods: The study used data of a dynamic cohort with 1,439 children aged 7–60 months from the Nouna Health and Demographic Surveillance Site (HDSS) for 2017 to 2019. We assessed data on diets, height, weight, household characteristics, and daily precipitation (from 1981 to 2019). Principal component analysis was used to identify distinct child dietary patterns (Dietary Pattern Scores, DPS). These were related to 15 rainfall indicators by area to obtain a precipitation variability score (PVS) through reduced rank regression (RRR). Associations between the PVS and anthropometric measures, height-for-age (HAZ), and weight-for-height (WHZ), were examined using multi-level regression analysis.Results: Stunting (HAZ < −2) and wasting (WHZ < −2) were seen in 24 and 6% of the children. Three main dietary patterns were identified (market-based, vegetable-based, and legume-based diets) and showed mixed evidence for associations with child undernutrition. The RRR-derived PVS explained 14% of the total variance in these DPS. The PVS was characterized by more consecutive dry days during the rainy season, higher cumulative rainfall in July and more extremely wet days. A 1-point increase in the PVS was associated with a reduction of 0.029 (95% CI: −0.06, 0.00, p < 0.05) in HAZ in the unadjusted, and an increase by 0.032 (95% CI: 0.01, 0.06, p < 0.05) in WHZ in the fully adjusted model.Conclusion: Rainfall variability was associated with dietary patterns in young children of a rural population of Burkina Faso. Increased rainfall variability was associated with an increase in chronic undernutrition, but not in acute undernutrition among young children.

One-week recall period gives a more accurate estimate of exclusive breastfeeding practice than 24-h recall among infants younger than six months of age

BackgroundThe World Health Organization recommends a 24-h recall period to estimate breastfeeding practice of mothers of infants aged younger than six-months. Though 24-h recall was preferred for its low recall bias and for practical reasons, it can overestimate exclusive breastfeeding practice (EBF). Validating this indicator will help account for the deviation from the true estimate. This prospective cohort study measured accuracy of the 24-h recall method and validates a week recall as an alternative approach for use in a small sample population.MethodThe study was conducted from March to April 2018 involving 408 mother-infant pairs living in Butajira Health and Demographic Surveillance Site (HDSS), Southern Ethiopia. Participants were prospectively followed for 14 consecutive days; where their breastfeeding practice in the past 24 h was measured daily. Exclusive breastfeeding prevalence estimate obtained using the 24-h recall method and recall periods spanning a varying number of days (short period recalls) was compared against the cumulative of the responses from a prospectively measured repeated 24-h recalls over the course of 14 days. McNemar statistics was used to assess statistical significance of the difference in the EBF prevalence estimates of the single 24-h recall and the reference standard. Sensitivity, specificity, positive predictive value and negative predictive values were calculated to determine the level of accuracy. Receiver Operating Characteristics curve was used to measure the difference in performance between the two methods.ResultThe highest prevalence (71.4%) of exclusive breastfeeding practice was estimated using the single 24-h recall method whereas the lowest breastfeeding practice (47.1%) was obtained from a cumulative of 14 repeated 24-h recalls. A week recall (a recall over 7 days’ period), resulted in the smallest discrepancy in estimate (7.1%) as compared to cumulative estimate of 14 repeated 24-h recalls. Comparing against our reference standard, a week recall had 96.7% sensitivity and 83.5% specificity in estimating exclusive breastfeeding practice.ConclusionsUsing single 24-h recall method overestimated exclusive breastfeeding prevalence. However, a week recall gave an estimate close to the estimate from the standard method. A week recall has a potential to balance the tradeoff between the accuracy of EBF estimates and the resource implication of using multiple prospective measurements that have a proven superior accuracy.

Spatial distribution of under immunization among children 12\u201323\u2009months old in Butajira HDSS, southern Ethiopia

BackgroundImmunization is essential to prevent between 2 and 3 million deaths globally each year and it is widely accepted that it is one of the most cost-effective health interventions. Despite all its advantages, immunization in Ethiopia is still far from the target set by the United Nations Sustainable Development Goals to achieve universal immunization by all countries in 2030. The 2016 Ethiopian Demographic and Health Survey (EDHS) reported an overall full immunization rate of only 38.3%. The objective of this study was to evaluate the spatial distribution of under immunization in 12 to 23 months old children and further identify the determinants of under immunization clustering in the Butajira Health and Demographic Surveillance Site (HDSS).MethodsWe conducted a community based sectional survey from March to April, 2016 in Butajira HDSS. We collected data on immunization status from a total of 482 children between the age of 12 to 23 months. We randomly selected household and interviewed mothers and /or observed vaccination cards when available to collect data on child’s immunization status. We also collected the geographic location of all villages within the ten Kebeles using a Handheld Global Positioning System (GPS) (Garmin GPSMAP®). We analyzed the spatial distribution of under immunization and clustering using the SatScan® software which employs a purely spatial Bernoulli’s model. We also ran a logistic regression model to help evaluate the causes of clustering.ResultsWe found that only 22.4% [95% CI: 18.9, 26.4%] of children were fully immunized. This study identified one significant cluster of under immunization among children 12–23 months of age within the Butajira HDSS (relative risk (RR) = 1.24,P < 0·01). We found that children residing in this cluster had more than 1.24 times risk of under immunization compared with children residing outside of the identified cluster. We found significant differences with regard to Maternal Tetanus Toxoid immunization status and place of delivery between cases found within a spatial cluster and cases found outside the cluster. For example, the odds of home delivery is more than two times [AOR 2.21: 95%CI; 1.06, 4.63] among children within an identified spatial cluster than the odds among children found outside the identified cluster.ConclusionsUnder immunization of 12–23 months old children and under immunization with specific vaccines such as Polio, BCG, DPT (1–3) and Measles clustered geographically. Spatial studies could be effective in identifying geographic areas of under immunization for targeted intervention like in this study to gear health education to the specific locality.

Tobacco use and associated factors among adults reside in Arba Minch health and demographic surveillance site, southern Ethiopia: a cross-sectional study

BackgroundTobacco use is one of the world-leading preventable killers. There was a varied prevalence of tobacco use and cigarette smoking across different areas. The aim of the study was to assess the prevalence and factors associated with current tobacco use among adults residing in Arba Minch health and demographic surveillance site (HDSS).MethodsA community-based cross-sectional study was conducted among adults residing in Arba Minch HDSS in 2017. The estimated sample size was 3368 individuals which were selected by simple random sampling techniques using Arba Minch HDSS dataset. Data collection tools were obtained from the WHO STEPwise. Current use of tobacco, which defined as the current use of smoked and/or smokeless tobacco, was considered as the dependent variable. A binary logistic regression model was used to identify candidate variables for the multivariable logistic regression model. An adjusted odds ratio (AOR) at a p-value of less than 0.05 was used to determine a statistically significant association between independent and dependent variables.ResultThe prevalence of tobacco use among adults was 20.2% (95% CI: 18.9–21.6%). The current use of smoked and smokeless tobacco were 17.1% (95%CI: 15.8–18.4%) and 9.7% (95%CI: 8.8–10.8%), respectively. The current use of tobacco was significantly associated with sex (female [AOR 0.54; 95%CI: 0.42–0.68] compared to men), age group (35–44 [AOR 1.57; 95%CI: 1.14–2.17], 45–54 [AOR 1.99; 95%CI: 1.45–2.74], and 55–64 [AOR 3.26; 95%CI: 2.37–4.48] years old compared to 25–35 years old), physical activity (moderate physical activity level [AOR 0.65; 95%CI: 0.44–0.96] compared with low) and residency (highland [AOR 4.39; 95% CI: 3.21–6.01] compared with at lowlander). Also, heavy alcohol consumption (AOR 3.97; 95% CI: 3.07–5.12), and Khat chewing (AOR 3.07(95%CI: 1.64–5.77) were also associated with the use of tobacco among the study participants.ConclusionNearly one in five adults used tobacco currently in the study area, which is more than the national reports. Interventions for the reduction of tobacco use need to give due attention to men, older adults, uneducated, poor, and highlanders.

Prevalence of heavy episodic drinking and associated factors among adults residing in Arba Minch health and demographic surveillance site: a cross sectional study

BackgroundAlcohol consumption is associated with different types of illnesses; particularly heavy episodic drinking is one of the risk factors for the disease burden of alcohol intake. The aim of the study was to assess the prevalence of heavy episodic drinking and associated factors in Arba Minch Health and Demographic Surveillance Site (HDSS).MethodsA community-based cross-sectional study was conducted in 2017 among adult residents of Arba Minch HDSS. Using Arba Minch HDSS database, 3368 individuals were selected by simple random sampling techniques. From WHO STEPS instruments, step one was applied for this study. Variables with a p-value of less than 0.10 for bivariate analysis entered into a multivariable logistic regression model to outline the independent predictors of the heavy episodic drinking. To assess the presence of an association between dependent and independent variables, a p-value of less than 0.05 was considered.ResultsThe prevalence of heavy episodic drinking was 13.7% (95% CI: 12.6–14.9). The study has shown that heavy episodic drinking was significantly associated with occupation (daily laborer [AOR = 0.49; 95% C.I: 0.29–0.85] and housewives [AOR = 0.63; 95% C.I: 0.45–0.88] compared with farmers), wealth index (2nd quintiles [AOR =0.55; 95% C.I: 0.41–0.74) and 3rd quintiles [AOR = 0.66; 95% C.I: 0.46–0.93] compared with 1st quintiles), and climatic zone (midland [AOR = 1.80;95% CI: 1.11–2.93), highland [AOR = 1.95;95% CI: 1.19–3.18] compared with lowland). In addition, tobacco use [AOR = 4.28;95% CI: 3.38–5.43], and khat use [AOR = 4.75; 95% CI: 2.66–8.50) were also associated with heavy episodic drinking among the study participants.ConclusionsMore than one in ten adults reported heavy episodic drinking in the study area. Intervention programs that aim to prevent heavy episodic drinking should be designed appropriately for individuals from lower wealth status, and for highlander.