Browsing by Author "Odur, Benard"

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    A Multilevel Decomposition of Time Variation in the Risks of Infant Mortality in Rural Uganda: UDHS 1995–2016
    (East African Journal of Health and Science, 2023) Odur, Benard; Nansubuga, Elizabeth; Wamala, Robert; Atuhaire, Leonard
    The study assessed the contribution of maternal, child, paternal, household, proximate, and community-level factors to infant mortality risk time variation in rural Uganda between 1995 and 2016. Five rounds of Uganda Demographic and Health Survey data sets were used, and a multilevel mixed-effect logistic regression model was applied to decompose the contribution of different factors to time variation in the risks of infant mortality. All live births that were made five years before the surveys of 1995, 2001, 2006, 2011, and 2016 were considered, with infants who did not survive beyond one year treated as the outcome variable analysis, excluding those who were born less than 12 months before the survey. The fixed part of the model helped us detect the significant variables in determining infant mortality, and yet the random part of the model helped us quantify the amount of time variation in the risks of infant mortality explained by the selected variables. The child-level determinants of infant mortality were sex, birth order, and weight. Among the maternal factors, the study revealed that marital status, access to ANC, use of contraceptives, maternal education level, and preceding birth interval were consistent deterrents of infant mortality, while household size, sanitation, and wealth index remained critical. While controlling for other factors in the rural areas, time variation in the risks of infant mortality was dependent on community factors (such as region, community hygiene, and prenatal care utilization rate), proximate factors (such as access to prenatal care, contraceptives use, place of delivery, and the number of ANC visits), maternal factors (such as marital status, educational level, age, parity, preceding birth interval, desire for pregnancy, and breastfeeding), and endowment. It was observed that the changes in the risks of infant mortality over the period were explained by community (30.7%), proximate (22.7%), maternal (41.0%), and endowment (37.9%). Child-level factors explained 28.2%, and paternal-level education level explained only 30.1%. Remarkably, household-level factors captured 32.3% of the changes in infant mortality. A higher proportion of the explained variation in the risk of infant mortality across communities (PCV) was captured by child, paternal, maternal endowment, and household factors. Interventions to accelerate the reduction in infant mortality should target birth spacing to at least two years, girl child education to at least o level, joint household decision-making in having children, avoiding teenage pregnancies, postnatal care utilization, enforcing at least four ANC visits during pregnancy, improving household sanitation, and increasing access to safe water at household-levels
  • Thumbnail Image
    Item
    Parametric Versus Non-Parametric Models for Predicting Infant Mortality within Communities in Uganda using the 2016 Uganda Demographic and Health Survey Data
    (East African Journal of Health and Science, 2023) Odur, Benard; Nansubuga, Elizabeth; Odwee, Jonathan; Atuhaire, Leonard
    Machine learning techniques have been infrequently used to identify community-based infant mortality risks. Achieving SDG 3 Targets 3.2 and 3.3 could be expedited by early detection of at-risk infants within communities. This study aimed to devise a community-centric algorithm for predicting infant mortality. We analysed UDHS 2016 data containing birth records for 22,635 children born within the five years preceding the survey, excluding those born within a year of the interview date. Twelve machine learning models were evaluated for their predictive capabilities using the area under the receiver operating characteristic curve (AUC ROC) in Python. Data subsets were divided into training and testing sets in a 2:1 ratio. Among the evaluated models, CatBoost showed superior performance with an AUC ROC of 74.9%. The five most influential variables for the CatBoost model were postnatal care utilisation, paternal age, household size, preceding birth interval, and maternal age. While the algorithm’s best performance was achieved using 28 variables, it still exhibited robust predictive power when limited to the top 8 or 10 variables. Hence, CatBoost stands out as an effective tool for identifying community-based infant mortality risks
  • Thumbnail Image
    Item
    Prediction equations for body composition of children and adolescents aged 8e19 years in Uganda using deuterium dilution as the reference technique
    (Clinical Nutrition ESPEN, 2018) Ndagire, Catherine T.; Muyonga, John H.; Odur, Benard; Nakimbugwe, Dorothy
    Background and aims: Body composition is important as a marker of both current and future health status. Bioelectrical impedance analysis (BIA) is a simple and accurate method for estimating body composition in field, clinical and research settings, if standard protocol procedures are followed. However, BIA requires population-specific equations since applicability of existing equations to diverse populations has been questioned. This study aimed to derive predictive equations for Total Body Water (TBW), Fat Free Mass (FFM) and Fat Mass (FM) determinations with BIA and anthropometric measurements in a population of children and adolescents aged 8e19 years in Uganda. Methods: A cross-sectional study was conducted among 203 children and adolescents aged 8e19 years attending schools in Kampala district (also referred to as Kampala city since the city is conterminous with the district), Uganda through a two-stage cluster sample design. Deuterium dilution method (DDM) was used as the reference measure while BIA and anthropometric measures were used to create the new body composition prediction equations through multivariate regression. Results: The new prediction equations explained 88%, 87% and 71% of the variance in TBW, FFM % and of FM respectively with no statistical shrinkage upon cross-validation. The linear regression models proposed in this study were well adjusted with respect to TBW, FFM and FM. Log of TBW obtained by DDM ¼ 0.0129 Impedance index þ 0.0055 Age þ 0.0049Waist Circumference þ 0.1219Ht2 þ 2.0388. Log of FFM obtained by DDM ¼ 0.0197 FFM obtained by BIA e 0.0181 sex code e 0.00055 Impedance þ 3.1761. Log of FM obtained by DDM ¼ 0.0634 FM obtained by BIA e 0.1881 sex code þ 0.0252 Weight þ 0.5273.