Introduction

Morbidity indicates an unhealthy condition caused by a disease or disorder, which can be mental or physical. The common types of morbidity in children are preventable diseases such as diarrhea, acute respiratory infection (ARI), fever, cough, and breathing problems1. Childhood morbidity is associated with the poor nutritional status of children, which hampers physical growth and decreases the probability of a child’s survival2. Access to secure sanitation, pure drinking water, and hygiene control can decrease the chance of a child’s morbidity and mortality, which leads to a healthier life3. Furthermore, economic development improves the quality of life, which can contribute to the decline of children’s morbidity in any country4. Additionally, mothers can prevent children’s illness and death by utilizing adequate antenatal care, early postnatal care, and skilled birth attendants during delivery5.

There is a notable rate of child mortality in both developed and developing countries due to childhood illnesses. Globally, 5.2 million deaths under five children occurred in 2019, mostly due to preventable and treatable diseases6. Pneumonia and diarrhea account for 1.4 million child deaths in the world each year, and most of these deaths occur in lower- and middle-income countries (LMICs)7. In addition, ARI is another predominant cause of mortality8,9, accounting for approximately 1.3 million child deaths around the globe each year10. The most typical symptom, fever, is seen as a burden of childhood illness. In LMICs, fever is primarily caused by viral infections11,12,13 and is liable to increase pediatric consultations, which has a significant effect on economy14. About two-thirds of child deaths in developing countries are caused by illnesses such as pneumonia, malaria, diarrhea, measles, and malnutrition15.

Due to increased rates of pneumonia and diarrhea, low use of maternal healthcare services, and malnutrition, Bangladesh has one of the highest rates of child mortality in South Asia, ranking fourth in terms of neonatal mortality16. In this country 45 deaths in 1000 live births occur among under five children, where preventable diseases are more likely to cause these deaths.15 Despite significant improvements in different areas, including life expectancy, literacy rate, and maternal and child health, child morbidity is an alarming issue in Bangladesh17. Infectious diseases such as diarrhea, ARI, and fever are more prevalent morbidities among children. In Bangladesh, around 33% and around 5% of children under the age of five experience fever and diarrhea, respectively18, and around one-half million child deaths occur each year because of only diarrhea19. Most ARI-related deaths occur in LMICS due to low healthcare services19, and ARI accounts for 25% of child deaths in Bangladesh each year20.

Most of the studies conducted in LMICs like Bangladesh have focused on identifying the determinants of child morbidity using a logistic regression model18,21,22,23. When data are hierarchical, residual variances are associated with observations. In this situation, logistic regression does not give precise results. In such cases, multilevel logistic regression is utilized24. Previous studies have reported additional sociodemographic determinants of child morbidity; however, the present analysis focused on covariates consistently available and commonly used in BDHS-based morbidity model. Besides, mothers’ age at marriage and father’s occupation are associated with child morbidity and mortality25,26,27. The true effect size might have been underestimated or overestimated without considering these factors. Furthermore, spatial analysis is essential to identify locations where child morbidity is high or low. As a result, policymakers will be able to develop appropriate policies and programs in these areas to reduce the prevalence of child morbidity. Therefore, this study aimed to identify the spatial distribution and determinants of child morbidity among under five children in Bangladesh.

Results

Sample characteristics of the study

Table 1 presents the sociodemographic characteristics of the respondents. The sample was mainly children aged 24–59 months (58.6%), male (51.3%), not wasted (88.1%), and currently breastfed (57.4%); most mothers had secondary education (52.2%), and wealth quintiles were approximately evenly distributed. Most children live in rural areas (68.1%), with the largest shares from Chattogram (17.3%), Dhaka (14.4%), and Sylhet (12.9%). Community-level media exposure was nearly evenly split, with 50.3% in low-exposure and 49.7% in high-exposure areas.

Table 1 Descriptive statistics of morbidity among under-five children by selected background covariates: BDHS 2022.
Full size table

Table 1 shows that overall child morbidity was 33.6%. This table also indicates significant variation in morbidity by child age, mother’s education and household wealth. Morbidities were highest among children aged 12–23 months (42.6%) and lowest among those who were 24–59 months of age (30.8%) (p < 0.001). Children from the richest households had the lowest morbidity (29.5%) (p = 0.031). Regarding maternal education, the lowest morbidity observed among children of mothers with higher education (28.2%) compared to those with no education (29.3%) (p < 0.001). Figure 1 shows that fever was the most common morbidity among under-five children (30.99%), followed by diarrhea (4.83%) and ARI (1.43%).

Fig. 1
Fig. 1
Full size image

Prevalence of childhood morbidities (ARI, fever and diarrhea) among under-five children in Bangladesh.

Spatial distribution of child morbidity

Figure 2 revealed that the spatial autocorrelation of child morbidity was statistically significant (Moran’s I = 0.035, z-score = 2.237, p < 0.05). The hot spot areas and cold spot areas were identified using Getis-Ord Gi* statistic (see Fig. 3). The hot spot areas of child morbidity were found in Rangpur, Khulna, Barisal, and Chattogram division. On the contrary, the cold spot areas were Dhaka, Sylhet, Chattogram, and Mymensingh division. Chattogram division was in both hot spot and cold spot area of child morbidity.

Fig. 2
Fig. 2
Full size image

Spatial autocorrelation of morbidity among under-five children in Bangladesh.

Fig. 3
Fig. 3
Full size image

Hot and cold spot areas of child morbidity in Bangladesh. The map was created by the authors in ArcGIS Desktop (ArcMap) version 10.5 (Esri, Redlands, CA, USA; https://desktop.arcgis.com/). Hotspot and cold-spot clusters were generated using the Getis–Ord Gi* statistic. Administrative boundary shapefiles were obtained from the DHS Program Spatial Data Repository (https://spatialdata.dhsprogram.com/boundaries/) and used for cartographic display.

Factors associated with child morbidity

The findings of the multilevel logistic regression model are shown in Table 2. After adjusting individual-, household- and community-level factors in Model IV, the analysis displayed that children’s age, currently breastfeeding, mother’s education, wealth index, and place of residence were significantly associated with child morbidity. Children aged 24–59 months had a 29% (aOR = 0.71, 95% CI: 0.57, 0.88) lower odds of experiencing morbidity compared to children aged 0–11 months. The odds of developing morbidity were 18% (aOR = 0.82, 95% CI: 0.68, 0.99) lower among children who were currently breastfed than children who were not currently breastfed. Children whose mother completed secondary education had 1.42 (aOR = 1.42, 95% CI: 1.03, 1.95) times more likely to suffer from morbidity than children whose mother had no education. The odds of getting morbidity were 26% (aOR = 0.74, 95% CI: 0.56, 0.97) lower among children belonging to the richest wealth index compared to children from the poorest wealth index. Similarly, children residing in rural areas had 22% (aOR = 0.78, 95% CI: 0.65, 0.95) lower likelihood of developing morbidity than children residing in urban areas. When examining interaction effects, no statistically significant interactions were observed between wealth index and place of residence, or between maternal education and current breastfeeding status.

Table 2 Multilevel logistic regression analysis of individual-, household- and community-level factors related to child morbidity.
Full size table

Random effect results

From Table 2, the variation in the prevalence of child morbidity was found in Model 0 (σ2 = 0.29, 95% CI: 0.18, 0.45). In Model 0, the intra-class correlation coefficient (ICC) explained that differences across clusters accounted for approximately 8.031% of total variability. The proportional change in variance (PCV) for the full model (Model IV) showed that individual-, household-, and community-level factors accounted for around 4.956% of the variability. The log-likelihood value was the lowest in Model IV (log-likelihood = -2509.67) and the Bayesian Information Criterion (BIC) was compared with other models. A substantial improvement in BIC was observed in Model IV (BIC = 5226.72); therefore, this model was selected as the best model to identify the determinants of child morbidity in Bangladesh.

Discussion

Utilizing the nationally representative survey data, our findings revealed that approximately one in three children were suffering from morbidity in Bangladesh. Our spatial analysis found a distinct and significant geographical variation in child morbidity across Bangladesh, indicating the uneven distribution of child health outcomes. We identified distinct hot spot areas of consistently high child morbidity in the divisions of Rangpur, Khulna, Barisal, and Chattogram. However, cold spot areas with a smaller share of child morbidity were identified in the divisions of Dhaka, Sylhet, Chattogram, and Mymensingh. Although fever is a non-specific symptom, its inclusion in line with DHS and MICS morbidity measurement protocols makes it a more generalizable measure to compare to other national and international under-five health studies.

The mixed result that the Chattogram Division is both a hot and cold spot reflects substantial intra-divisional heterogeneity in the risk of child morbidity. This may indicate an urban–rural polarization within the division, with large cities such as Chattogram City having densely populated areas, better access to health care and higher disease reporting compared to remote coastal and hilly subdistricts with low access to health care, poor sanitation and different environmental exposures. Environmental and climatic factors, for example recurrent flooding, high humidity and cyclone-prone costal belt could contribute to higher disease burden in certain clusters while relatively developed peri-urban areas might show lower morbidity on account of better infrastructure and health consciousness. Similar spatial studies in India, Ghana and Malawi have also shown some similar mixed hot–cold pattern of clustering in the same administrative units explained by their internal differences on population density, socio-economic status, infrastructure and use of health care28,29,30. These results emphasize that combined divisional averages may mask important inequalities at local levels, and the importance of sub-divisional spatial analysis and locally targeted public health interventions in geographically diverse settings such as Chattogram.

The distinct geographical variation according to child morbidity is likely to be the result of a mixture of socioeconomic, environmental, and infrastructural conditions29,30. For instance, Rangpur division has been history-declared as one of the most impoverished zones in Bangladesh, coupled with a high percentage of food insecurity and a poor record of access to improved water and sanitation facilities. These are some well-known contributory factors towards communicable diseases like diarrhea and ARI. In the same vein, both Khulna and Barisal divisions29, situated in a vulnerable coastal belt, are easily stricken by natural calamities, such as cyclones, tidal surges, and river erosion. Such environmental disasters affect the entire region, leading to wide-ranging effects of water contamination (including saline intrusion), displacement from healthcare and other basic infrastructures, thus raising the incidence of child morbidity.

Unlike the flood-prone regions of the country, the Dhaka division, being the economic and administrative center of the country, generally boasts relatively better urban infrastructure, better access to advanced healthcare facilities, higher levels of maternal education, and richer households. All these conditions put together make this division a cold spot. Although Sylhet and Mymensingh divisions have also emerged as cold spots, further research is needed to pinpoint the exact protective factors at play in these areas. These could include specific local initiatives, community practices, or environmental conditions which may mitigate the risk of common childhood illnesses.

While not many direct comparisons could be done with other studies from Bangladesh that have adopted the same spatial methodology, findings are parallel with what has been generally observed in epidemiological patterning that usually creates disparities in health outcomes based on different economic developments and vulnerability due to environment across regions. The spatial clustering makes out the clear bulwark for moving from national averages into what areas health challenges are most concentrated, and here public health resource allocation efforts can be targeted and most efficiently used.

Multilevel analysis showed that children’s age was significantly associated with morbidity. Children aged 24 months and older were less likely to have morbidity compared to children aged 0–11 months. This finding was in line with another study conducted in Ethiopia and Uganda31,32. When very young children aged 12–23 begin to crawl and explore the surrounding environment, they are more likely to touch contaminated objects32. As a result, they are easily infected by viruses or bacteria. Conversely, children’s immune systems strengthen with age, reducing the odds of morbidity1. Furthermore, inadequate nutrition in young children aged 0–12 months can lead to morbidity. This is because mothers with low socioeconomic levels, poor sanitary facilities, and contaminated water may not be capable of adequately breastfeeding their babies33.

Although wasting increases the likelihood of child morbidity however, the relationship was not found to be significantly associated with our desired outcome. However, wasting, as a manifestation of acute malnutrition, has important implications for susceptibility and severity of common childhood illnesses. Malnourished children have fewer immune defenses, impaired mucosal integrity and lower potential for recovery from infections. This vulnerability not only increases likelihood of diarrhea and ARI but might also predispose children to recurrent or prolonged febrile episodes. There is global and regional evidence that wasting dramatically increases vulnerability to infection morbidity and mortality among children under five years old. For example, children who are moderately or severely wasted have between three- and nine-times higher risk of having diarrhea and dying or developing an acute respiratory infection compared to well-nourished kids34,35,36. Even though we have examined wasting as an individual level determinant of morbidity, interaction with specific morbidity components like fever, diarrhea and ARI need further exploration following a longitudinal or causal framework. Clearly, nutrition-sensitive and specific interventions need strengthening to break the vicious circle between infections and undernutrition.

Currently breastfeeding has a significant impact on child morbidity. Non-breastfed children were more likely to develop morbidity compared to children who were not currently breastfed. A similar finding was found in another investigation37. In addition, breastfeeding among children is responsible for the increased risk of diarrhea morbidity and mortality38. Breastfeeding can play a crucial role to improve child health and survival, providing energy, antibodies, and nutrients that help children protect against common childhood illnesses, economic, and environmental advantages to society39.

Mother’s education is typically considered to be protective for child health40,41, but in our full model the likelihood of morbidity was greater for children of mothers with secondary education compared to those whose mother had no education, although primary and higher levels were not statistically significant. This unexpected trend may have been due to residual confounding and other context-specific factors that were not completely controlled for in the covariates used. Studies have suggested that crowding is more common for families in urban or peri-urban settings where children might have a higher likelihood of encountering particular forms of mixing and child care, as well as environmental exposures42; partially consistent with this interpretation, morbidity was associated with elevated levels in urban areas in our model (rural residence had lower odds: aOR = 0.78). A second possible explanation is reporting/recognition bias: DHS morbidity indicators are based on caregiver-reported symptoms and mothers with some schooling may be more likely to recognize, recall, and report symptoms (particularly fever) or have healthcare contact which increases detection and reporting. Specifically, our composite morbidity measure is mainly influenced by the proportion reporting fever (30.99%), with diarrhea (4.83%) and ARI (1.43%) being far less frequent; thus, any differences in fever awareness or recall are more likely to have an outsized effect on the composite measure of morbidity. At the descriptive level, offspring of mothers with highest maternal education showed the lowest prevalence of morbidity, indicating heterogeneity across education groups and indicating that the estimate for secondary education may be due to unmeasured contextual or reporting factors, rather than a causal relation. We tested for effect modification by fitting interaction terms (wealth×place of residence; maternal education×current breastfeeding status) to the multilevel model; none reached statistical significance (all 95% confidence intervals included 1), so here we present results from the main-effects model. However, stratified descriptive patterns (i.e., urban vs. rural) and future studies that address richer covariate measurements (for example healthcare access, environmental exposures, or care-seeking behaviors) might elucidate whether this association is due to contextual factors or differential symptom recognition/reporting rather than a causal effect.

Wealth index was an influential determinant of child morbidity. Children from the poorest wealth index were more likely to suffer from morbidity than children from the richest wealth index. This finding was in line with another cross-sectional study of Bangladesh43. Due to poverty, they cannot provide sufficient food and live in poor environments, which adversely affects nutritional status44,45. There is a substantial association between child morbidity and poor nutrition status resulting from food insecurity46.

The present analysis reveals that rural children had reduced odds of getting morbid compared to urban children, contrary to reports from several studies done in Bangladesh and other LMICs where rurality seems to have been associated with higher illness burdens47,48. There are possibly several explanations why this urban–rural paradox occurs. First, DHS morbidity variables are based on symptoms past two-week time of the caregiver’s recall, and this could lead to differential recognition and reporting by place of residence49. Caregivers residing in rural settings may have reduced visibility of or difficulty recalling mild episodes (particularly when care is not sought through formal facilities), whereas urban caregivers generally will have higher levels health awareness and more opportunities for symptom detection due to contact with health staff47. Second, urban living in Bangladesh is diverse, including a majority population that lives in informal settlements where crowded living conditions and risks to infectious disease due to lack of water and sanitation are significantly increased17. Environmental exposures (including air pollution, crowding and childcare mixing) may further enhance risk in a urban environment50. Lastly, differences in healthcare seeking may also play a role: urban caregivers are more likely to seek care, which may increase likelihood of detecting illness and consequently reporting it compared to rural caregivers who may use home remedies or informal providers. Comparable “urban penalty” findings have been described elsewhere in other DHS reports when better availability and reporting in urban settings resulted to a greater actual morbidity despite better infrastructure51. In general, these results indicate that the rural advantage is likely due to a combination of environmental, behavioral and reporting causes rather than an inherent protective effect of rural residence.

There were several strengths and limitations of this study. This research was conducted using the most recent national representative survey. This is the first study in which we explored the hot spots and cold spots of child morbidity in Bangladesh. An advanced statistical model was utilized to assess risk factors for morbidity among under five children, and a wide range of explanatory variables were identified using a literature review. In addition, the explanatory variables were divided into individual-level determinants, household-level determinants, and community-level determinants using a socio-ecological model.

This study should be considered in the light of some limitations. Some important variables, such as ethnicity, hand washing habit, and anemia status, could not be included in this study due to the unavailability of such variables in BDHS 2022. Because morbidity indicators in DHS are based on maternal recall of symptoms, differential reporting by maternal education level may occur and could influence observed associations. Furthermore, the DHS cluster GPS coordinates are given a random displacement to protect confidentiality of respondents (up to 2 km in urban and 5 km in rural areas, with 1% of rural clusters displaced up to 10 km) and compelled to remain within survey regions/administrative boundaries; thus these differences are unlikely to substantially modify our divisions-level spatial patterns but could result on minor uncertainty for finer scale local hotspot estimation. Moreover, since the study is cross-sectional, it is not possible to explore the cause-and-effect relationship between the outcome variable and the explanatory variables.

To effectively tackle child morbidity in Bangladesh, policies must be precise and tailored, drawing on our spatial and determinant findings. We recommend that public health resources and programs be directed toward the subject “hot spot” divisions: Rangpur, Khulna, Barisal, and Chattogram. These areas must receive concentrated efforts that require and address the high burden of child morbidity. Interventions vary in Rangpur, for example, it would require poverty alleviation, food security and WASH infrastructure improvement; in Khulna and Barisal, policy and program integration of climate resilient health services source secured water (salinity issues), and established community-based disaster preparedness regarding child health as vulnerable during the natural calamity. For many divisions such as Chattogram, with both hot and cold spots, more granular, sub-divisional assessments will be necessary to pinpoint localized vulnerable areas for targeted responses. Beyond geography, however, policies should improve and strengthen foundational child health practices through intensified national breastfeeding promotion campaigns and clinical provision of high-quality breastfeeding support by health care providers. It is also critical to answer socioeconomic inequities as a health imperative-expanding and strengthening social safety net programs and poverty reduction initiatives that directly benefit poorest households. We must also make early childhood health a top priority for development with structured health packages designed for infants and toddlers (0–23 months) who are most vulnerable. Finally, policies must advocate for data-driven and multi-sectoral approaches, harmonizing continuous spatial monitoring with national health surveillance, and fostering cooperation among health, education, social welfare, environment, and disaster management sectors. More research also needs to be done to understand the nuanced relation of that secondary maternal education found under this study with child morbidity. In doing so, these strategic recommendations will aid Bangladesh in very significant steps to be made in reducing child morbidity and improving health outcomes among the under five populaces.

Methods

Study design and sample size

The data were taken from the most recent Bangladesh Demographic and Health Survey (BDHS) 202249. The survey was jointly conducted by the National Institute of Population Research and Training (NIPORT) and ICF. The United States Agency for International Development (USAID) in Bangladesh provided financial support to collect data. Two-stage stratified sampling was applied in this survey. In the first stage, a total of 675 enumeration areas (EAs) were selected randomly. After that, 45 households per EA were chosen using systematic sampling in the second stage. After excluding one area due to security concerns, 674 EAs were selected to collect information. 30,330 households were picked for this survey. Ultimately, 30,078 ever-married women were selected for the interview, with a response rate of 99%. The survey methodology was published elsewhere49. A total of 4,005 mothers with children under five years were taken as a study sample.

Outcome variable

In the current study, the outcome variable was child morbidity. This variable was defined by whether child had fever, diarrhea or ARI during last two weeks of the survey. These indicators align with the standard child health modules of the Demographic and Health Surveys (DHS) and UNICEF Multiple Indicator Cluster Surveys (MICS), where caregiver-reported fever, diarrhea, and ARI symptoms are asked for as morbidity measures for under-fives to allow for comparability across countries and over time52,53.

Research in Bangladesh18,21 and other resource poor settings22,23 has consistently identified these three conditions as the leading causes of childhood morbidity. While fever is a symptom rather than a disease it was included alongside diarrhea and ARI for several reasons: (i) in limited-resource settings like Bangladesh, fever is a common indicator of acute infection, and it prompts care-seeking; (ii) it represents a wider spectrum of undifferentiated infectious diseases not otherwise categorized as diarrhea or ARI; and (iii) inclusion follows DHS- and MICS-related surveillance practices increasing the comparability of acute morbidity measures among children.

We additionally present the specific prevalence of fever, diarrhea, and ARI on its own, to maintain interpretability and specificity. Children who reported at least one of these symptoms were coded 1 (present), otherwise they were coded 0.

Explanatory variables

The explanatory variables were selected based on previous research1,17,18,22,31 showing a significant association with child morbidity and the availability of data in the BDHS. Individual-level determinants included children’s age in months (0–11, 12–23, 24–59), sex of children (male, female), wasting (no, yes) and currently breastfeeding (no, yes) were the individual-level factors. Household-level factors were mother’s education (no, primary, secondary, and higher), mother’s currently working status (no, yes), sex of household head (male, female), and wealth index (poorest, poorer, middle, richer, richest). Community-level determinants contained place of residence (urban, rural) and division (Barisal, Chattogram, Dhaka, Khulna, Mymensingh, Rajshahi, Rangpur, Sylhet). The community-level media was categorized as low if the proportion of women who had access to mass media was 0–50% and high if the proportion was 51%–100%. A previous study used this methodology54. We also evaluated other BDHS covariates (e.g., father’s occupation and religion) during stepwise building of the model. Father’s occupation was dropped because of multicollinearity, primarily with the household wealth index (and related SES variables such as maternal education). Religion was excluded for limited diversity in the sample and because it did not enhance model fit, being removed for parsimony.

Statistical analysis

Descriptive statistics were utilized to explain the background characteristics of the respondents. Global spatial autocorrelation of childhood morbidity was tested using Moran’s I 55 that checks whether the spatial pattern of the outcome across a study area is clustered, dispersed or random. A significant positive Moran’s I expresses overall (global) spatial clustering of either high/low values while a negative Moran’s I indicates dispersion; close to zero indicates random distribution. Moran’s I only give a single summary measure for the entire world, while local statistics can tell you where there is clustering.

Cluster analysis was performed using the Getis–Ord Gi* statistic56, which generates a z-score and corresponding p-value for each administrative region in relation to the local sum of itself with its adjacent areas. Hot spots referred to statistically significant positive Gi* z-score (e.g., p < 0.05) indicating spatial clusters of high values of childhood morbidity (high–high clustering). Cold spots were defined as Gi* z-scores that were significant (e.g., p < 0.05) and negative which showed statistically significant spatial clustering of low values (low–low clustering). Areas where the local score was non-significant (Gi* = 0) were described as showing no clustering.

Spatial relationships for both global Moran’s I and local Getis–Ord Gi statistics were delineated by a spatial weights matrix considering the geographic proximity among administrative units. We used a distance-based weights matrix, where neighboring unites that fell within a certain distance threshold was given non-zero weights and thus, the weights were row-standardized such that for each unit the sum of its weights added up to one. The weights specification was identical for the global as well as the local spatial statistics to guarantee transparency and reenactment.

Multilevel logistic regression analysis was used to explore the risk factors associated with child morbidity. Since the nature of DHS data is hierarchical, multilevel logistic model was applied. Individuals are contained in households, and households are contained in clusters24. Five models were employed in this study. Model 0 was a null model where no explanatory variables were considered. Model I adjusted individual-level determinants, Model II included household-level factors and Model III contained community-level determinants. Lastly, we run model IV adjusted individual-, household- and community-level determinants. Adjusted Odds ratio (aOR) with 95% confidence interval was used to report the results. The intra-class correlation (ICC), proportional change in variance (PCV), Akaike information criterion (AIC) and Bayesian information criterion (BIC) were reported to fit the best model. STATA 13 57 was used for all statistical analyses. Maps were produced in ArcGIS Desktop (ArcMap) v10.5 by overlaying Gi* results on DHS administrative boundary shapefiles and exporting the final layout for publication.58.

Ethical consideration

This study utilized secondary data obtained from the DHS. The ICF Institutional Review Board (IRB) approved the DHS survey and informed consent was acquired from all participants. The National Research Ethics Committee of Bangladesh and ICF Macro provided ethical approval. Therefore, ethical approval is not required to carry out this research. All the study procedures were conducted in accordance with the principles of the Declaration of Helsinki as revised in 2013.

Conclusions

Our study unveils that childhood morbidity among the under-five in Bangladesh is very high (33.58%) and, critically, there are huge geographical disparities among the affected areas. We have identified hot spots with high morbidity in the divisions of Rangpur, Khulna, Barisal, and Chattogram, in contrast to cold spots in the divisions of Dhaka, Sylhet, Chattogram, and Mymensingh. The mixed status carries complex internal variations within Chattogram. That spatial pattern is meaningful, indicating where the most concentrated burden of child illness is likely to be associated with factors such as poverty, environmental vulnerabilities, and variations in infrastructure. Other considerations outside the geographic spectrum include older child age, current breastfeeding practices, and wealth of the household. It was surprising to see that mothers’ secondary education has a positive association with morbidity; its more extensive impact still needs to be studied. In general, it calls for urgent targeted spatially informed public health interventions to improve child health in Bangladesh.