Levels, trends and socio-demographic determinants of infant and under-five mortalities in and around slum areas of Dhaka city, Bangladesh

Analyzing and forecasting under-5 mortality trends in Bangladesh using machine learning techniques

BACKGROUND

Under-5 mortality remains a critical social indicator of a country's development and economic sustainability, particularly in developing nations like Bangladesh. This study employs machine learning models, including Linear Regression, Ridge Regression, Lasso Regression, Bayesian Ridge, Decision Tree, Gradient Boosting, XGBoost, and CatBoost, to forecast future trends in under-5 mortality. By leveraging these models, the study aims to provide actionable insights for policymakers and health professionals to address persistent challenges.

METHODS

Data from the 1993-94 to 2017-18 Bangladesh Demographic and Health Survey (BDHS) was analyzed using advanced machine learning algorithms. Key metrics, including Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), R-squared, and Mean Absolute Percentage Error (MAPE), were employed to evaluate model performance. Additionally, k-fold cross-validation was conducted to ensure robust model evaluation.

RESULTS

This study confirms a significant decline in under-5 mortality in Bangladesh over the study period, with machine learning models providing accurate predictions of future trends. Among the models, Linear Regression emerged as the most accurate, achieving the lowest MAE (4.05), RMSE (4.56), and MAPE (6.64%), along with the highest R-squared value (0.98). Projections indicate further reductions in under-5 mortality to 29.87 per 1,000 live births by 2030 and 26.21 by 2035.

CONCLUSIONS

From 1994 to 2018, under-5 mortality in Bangladesh decreased by 76.72%. While the Linear Regression model demonstrated exceptional accuracy in forecasting trends, long-term predictions should be interpreted cautiously due to inherent uncertainties in socio-economic conditions. The forecasted rates fall short of the Sustainable Development Goal (SDG) target of 25 deaths per 1,000 live births by 2030, underscoring the need for intensified interventions in healthcare access and maternal health to achieve this target.

On the interpretability of the SVM model for predicting infant mortality in Bangladesh

BACKGROUND

Although machine learning (ML) models are well-liked for their outperformance in prediction, greatly avoided due to the lack of intuition and explanation of their predictions. Interpretable ML is, therefore, an emerging research field that combines the performance and interpretability of ML models to create comprehensive solutions for complex decision-making analysis. Conversely, infant mortality is a global public health concern affecting health, social well-being, socio-economic development, and healthcare services. The study employs advanced interpretable ML techniques to anticipate and understand the factors affecting infant mortality in Bangladesh, overcoming the shortcomings of the conventional logistic regression (LR) model.

METHODS

By utilizing the global surrogate model and local individual conditional expectation (ICE) interpretability technique, the interpretable support vector machine (SVM) has been used in this study to reveal significant characteristics of infant mortality using data from the Bangladesh Demographic and Health Survey (BDHS) 2017-18. To investigate intricate decision-making analysis of infant mortality, we adapted SVM and LR techniques with the hyperparameter tuning parameters. These models' performances were initially assessed using the receiver operating characteristics (ROC) curve, run-time, and confusion matrix parameters with 100 permutations. Afterward, the SVM model's model-agnostic explanation and the LR model's interpretation were compared to enhance advanced comprehension for further insights.

RESULTS

The results of the 100 permutations demonstrated that the LR model (Average: accuracy = 0.9105, precision = NaN, sensitivity = 0, specificity = 1, F1-score = 0, area under the ROC curve (AUC) = 0.6780, run-time = 0.0832) outperformed the SVM model (Average: accuracy = 0.8470, precision = 0.1062, sensitivity = 0.0949, specificity = 0.9209, F1-score = 0.1000, AUC = 0.5632, run-time = 0.0254) in predicting infant mortality, but the LR model had a slower run-time and it was unable to predict any positive cases. The interpretation of LR analysis revealed that infant mortality rates decrease when mothers give birth after over two years, with higher educational attainment, overweight or obese mothers, working mothers, and families with polluted cooking fuel having lower rates. The local ICE interpretability technique, which depicts individual influences on the average likelihood of dying within the first birthday, explores the interpretable SVM model that mothers with normal BMIs, giving birth within two years, using less polluted cooking fuel, working mothers, and having male infant were more likely to experience infant death. The interpretable SVM model based on the global surrogate model also reveals that working mothers who used polluted cooking fuel at home and working women who used less polluted cooking fuel but had a longer period between pregnancies than two years would have higher infant death rates. Even among non-working mothers who used polluted cooking fuel and gave birth within two years of the preceding one, infant death rates were higher.

CONCLUSIONS

The interpretable SVM model reveals global interpretations help clinicians understand the entire conditional distribution, while local interpretations focus on specific instances, providing different insights into model behavior. Interpretable ML models aid policymakers, stakeholders, and families in understanding and preventing infant deaths by improving policy-making strategies and establishing effective family counseling services.

Caesarean delivery and neonatal mortality: evidence from selected slums in and around Dhaka city, Bangladesh- A prospective cohort study

BACKGROUND

This study examined the neonatal mortality for newborn of women who delivered by caesarean section or vaginally using a prospective cohort.

METHODS

A total of 6,989 live births registered from 2016 to 2018, were followed for neonatal survival from the selected slums of Dhaka (North and South) and Gazipur city corporations, where icddr,b maintained the Health and Demographic Surveillance System (HDSS). Neonatal mortality was compared by maternal and newborn characteristics and mode of delivery using z-test. Logistic regression model performed for neonatal mortality by mode of delivery controlling selected covariates and reported adjusted odd ratios (aOR) with 95% confidence interval (CI).

RESULTS

Out of 6,989 live births registered, 27.7% were caesarean and the rest were vaginal delivery; of these births, 265 neonatal deaths occurred during the follow-up. The neonatal mortality rate was 2.7 times higher (46 vs. 17 per 1,000 births) for vaginal than caesarean delivered. Until 3rd day of life, the mortality rate was very high for both vaginal and caesarean delivered newborn; however, the rate was 24.8 for vaginal and 6.3 per 1,000 live births for caesarean delivered on the 1st day of life. After adjusting the covariates, the odds of neonatal mortality were higher for vaginal than caesarean delivered (aOR: 2.63; 95% CI: 1.82, 3.85). Additionally, the odds were higher for adolescent than elderly adult mother (aOR: 1.60; 95% CI: 1.03, 2.48), for multiple than singleton birth (aOR: 5.40; 95% CI: 2.82, 10.33), for very/moderate (aOR: 5.13; 95% CI: 3.68, 7.15), and late preterm birth (aOR: 1.48; 95% CI: 1.05, 2.08) than term birth; while the odds were lower for girl than boy (aOR: 0.74; 95% CI: 0.58, 0.96), and for 5th wealth quintile than 1st quintile (aOR: 0.59, 95% CI: 0.38, 0.91).

CONCLUSION

Our study found that caesarean delivered babies had significantly lower neonatal mortality than vaginal delivered. Therefore, a comprehensive delivery and postnatal care for vaginal births needed a special attention for the slum mothers to ensure the reduction of neonatal mortality.

Preterm birth and neonatal mortality in selected slums in and around Dhaka City of Bangladesh: A cohort study

BACKGROUND

Although under-five mortality has declined appreciably in Bangladesh over the last few decades, neonatal mortality still remains high. The objective of the study is to assess the level and determinants of preterm birth and the contribution of preterm birth to neonatal mortality.

METHODS

Data for this study came from selected slums in and around Dhaka city, where; since 2015, icddr,b has been maintaining the Health and Demographic Surveillance System (HDSS). The HDSS data were collected by female Field Workers by visiting each household every three months; however, during the visit, data on the Last Menstrual Period (LMP) were also collected by asking each eligible woman to ascertain the date of conception. Gestational age was estimated in complete weeks by subtracting LMP from the date of the pregnancy outcome. In this study, 6,989 livebirths were recorded by HDSS during 2016-2018, and these births were followed for neonatal survival; both bivariate and multivariate analyses were performed.

RESULTS

Out of total births, 21.7% were born preterm (before 37 weeks of gestation), and sub-categories were: 2.19% for very preterm (28 to 31 weeks), 3.81% for moderate preterm (32 to 33 weeks), and 15.71% for late preterm (34 to 36 weeks). The study revealed that preterm babies contributed to 39.6% of neonatal deaths; however, the probability of death was very high on the 1st day of birth (0.124 for very preterm, 0.048 for moderate preterm, 0.024 for late preterm, and 0.013 for term birth), and continued until the 3rd day. In the regression analysis, compared to the term neonates, the odds of neonatal mortality were 8.66 (CI: 5.63, 13.32, p<0.01), 4.13 (CI: 2.69, 6.34, p<0.01) and 1.48 (CI: 1.05, 2.08, p<0.05) respectively for very, moderate, and late preterm birth categories. The population attributable fraction for neonatal mortality was 23%, and sub-categories were 14% for very preterm, 10% for moderate preterm, and 6% for late preterm.

CONCLUSIONS

Although urban slums are in proximity to many health facilities, a substantial proportion of preterm births contribute to neonatal deaths. So, pregnant women should be targeted, to ensure timely care during pregnancy, delivery, and post-partum periods to improve the survival of new-borns in general and preterm birth in particular.

Factors affecting practices of recently delivered women on maternal and neonatal health care in selected rural areas of Bangladesh

Bangladesh has made laudable progress in maternal and child health (MCH). Maternal and child mortalities have reduced substantially accompanied by stellar rise in immunization and contraceptive prevalence rate (CPR). However, such success is distributed unevenly throughout and the country is among one of the top ten countries with highest number of neonatal and under-five children mortalities. Rural Bangladesh is home to more than half of the country's total population. Yet, disparity in access to healthcare services and information are overt in these areas. Utilization of maternal health services (MHS) is low whereas maternal and child mortalities are high in the rural areas. Thus, this cluster randomized cross sectional study was conducted with the aim to observe the practices that rural women followed in regards to maternal and child health and factors that affected these practices. Primary data was collected from 550 respondents using a structured questionnaire within the time period September-October 2019. All our participants were recently delivered women (RDW), defined in our study as women of reproductive age (15-49 years) who had delivered a child recently, i.e. 12 months prior (September 2018 - August 2019) the data collection. We conducted logistic regression and multivariate analysis to analyze data. Results from this study depict that while 96.3% of RDW opted for ANC visits and 99.1% fed colostrum to their newborn, fewer have had institutional deliveries and the number of RDW who had PNC was only 64.7%. Education was found to be the most prominent factor that affected practices employed by RDW. The more educated a respondent was, the greater the chance was of her engaging in appropriate maternal and child health practices. The RDW preferred and visited private facilities the most to obtain healthcare services with private medical doctors being one of the prime sources of healthcare information for the respondents. On the contrary, monthly expenditure exerted no statistically significant impact on the aforementioned practices. Thus, results of our study imply that interventions enhancing education and health knowledge of women and engaging private sector be designed for improving maternal and neonatal health care in rural areas of Bangladesh.

Adult mortality trends in Matlab, Bangladesh: an analysis of cause-specific risks

OBJECTIVE

With socioeconomic development, improvement in preventing and curing infectious diseases, and increased exposure to non-communicable diseases (NCDs) risk factors (eg, overweight/obesity, sedentary lifestyle), the majority of adult deaths in Bangladesh in recent years are due to NCDs. This study examines trends in cause-specific mortality risks using data from the Matlab Health and Demographic Surveillance System (HDSS).

DESIGN, SETTINGS AND PARTICIPANTS

We conducted a follow-up study from 2003 to 2017 using data from Matlab HDSS, which covers a rural population of 0.24 million (in 2018) in Chandpur, Bangladesh. HDSS assessed the causes of all deaths using verbal autopsy and classified the causes using the 10th revision of the International Statistical Classification of Diseases. We examined 19 327 deaths involving 2 279 237 person-years.

METHODS

We calculated annual cause-specific mortality rates and estimated adjusted proportional HRs using a Cox proportional hazards model.

RESULTS

All-cause mortality risk declined over the study period among people aged 15 and older, but the risk from stroke increased, and from heart disease and cancers remained unchanged. These causes were more common among middle-aged and older people and thus bore the most burden. Mortality from causes other than NCDs-namely, infectious and respiratory diseases, injuries, endocrine disorders and others-declined yet still constituted over 30% of all deaths. Thus, the overall mortality decline was associated with the decline of causes other than NCDs. Mortality risk sharply increased with age. Men had higher mortality than women from heart disease, cancers and other causes, but not from stroke. Lower household wealth quintile people have higher mortality than higher household wealth quintile people, non-Muslims than Muslims.

CONCLUSION

Deaths from stroke, heart disease and cancers were either on the rise or remained unchanged, but other causes declined continuously from 2003 to 2017. Immediate strengthening of the preventive and curative healthcare systems for NCDs management is a burning need.

Community and individual level determinants of infant mortality in rural Ethiopia using data from 2016 Ethiopian demographic and health survey

The infant mortality rate remains unacceptably high in sub-Saharan African countries. Ethiopia has one of the highest rates of infant death. This study aimed to identify individual-and community-level factors associated with infant death in the rural part of Ethiopia. The data for the study was obtained from the 2016 Ethiopian Demographic and Health Survey. A total of 8667 newborn children were included in the analysis. The multilevel logistic regression model was considered to identify the individual and community-level factors associated with new born mortality. The random effect model found that 87.68% of the variation in infant mortality was accounted for by individual and community level variables. Multiple births (AOR = 4.35; 95%CI: 2.18, 8.69), small birth size (AOR = 1.29; 95%CI: 1.10, 1.52), unvaccinated infants (AOR = 2.03; 95%CI: 1.75, 2.37), unprotected source of water (AOR = 1.40; 95%CI: 1.09, 1.80), and non-latrine facilities (AOR = 1.62; 95%CI: 1.20) were associated with a higher risk of infant mortality. While delivery in a health facility (AOR = 0.25; 95%CI: 0.19, 0.32), maternal age 35–49 years (AOR = 0.65; 95%CI: 0.49, 0.86), mothers receiving four or more TT injections during pregnancy (AOR = 0.043, 95% CI: 0.026, 0.071), and current breast feeders (AOR = 0.33; 95% CI: 0.26, 0.42) were associated with a lower risk of infant mortality. Furthermore, Infant mortality rates were also higher in Afar, Amhara, Oromia, Somalia, and Harari than in Tigray. Infant mortality in rural Ethiopia is higher than the national average. The government and other concerned bodies should mainly focus on multiple births, unimproved breastfeeding culture, and the spacing between the orders of birth to reduce infant mortality. Furthermore, community-based outreach activities and public health interventions focused on improving the latrine facility and source of drinking water as well as the importance of health facility delivery and received TT injections during the pregnancy.