Shifting Prevalence and Risk Factors of Non-Communicable Diseases in Bangladesh: A Comparative Multilevel Analysis of Nationally Representative BDHS Data (2017-2022)

This study analyzes nationally representative BDHS data from 2017-18 and 2022 to reveal a post-pandemic divergence in Bangladesh's non-communicable disease landscape, characterized by a dramatic doubling of diabetes prevalence driven by urban and community-level factors, alongside a decline in hypertension that may reflect survival bias and a weakening of the bidirectional relationship between the two conditions.

The Gist

🏥 The Big Picture: A Health "Plot Twist" in Bangladesh

Imagine Bangladesh's public health system as a busy highway. For years, doctors and officials have been worried about two main traffic jams: High Blood Pressure (Hypertension) and High Blood Sugar (Diabetes).

This study is like a traffic camera that took two photos of the highway: one in 2017 (before the pandemic) and one in 2022 (after the pandemic). The researchers wanted to see how the traffic had changed.

The Shocking Result: The traffic patterns flipped completely.

• The "Silent Killer" (High Blood Pressure): The traffic jam actually got lighter. Fewer people had high blood pressure in 2022 compared to 2017.
• The "Sugar Surge" (Diabetes): The traffic jam got massive. The number of people with diabetes didn't just grow; it more than doubled in just four years.

It's as if one lane of the highway suddenly cleared out, while the other lane became so clogged with cars that it stopped moving entirely.

---

🔍 How They Did the Detective Work

The researchers didn't just guess; they looked at two huge, nationally representative surveys (like taking a census of the whole country).

• The Data: They used official government data from the BDHS (Bangladesh Demographic and Health Survey).
• The Method: Instead of just counting heads, they used a special mathematical tool called "Mixed-Effects Modeling."
  • Analogy: Imagine trying to understand why some students get better grades. A simple count might say "Rich kids get better grades." But a "Mixed-Effects" model is like a smart teacher who realizes that the school the kid attends matters just as much as their wallet. It separates the individual factors (like age or diet) from the community factors (like the neighborhood or local healthcare).

---

🚦 The Key Findings: What Changed?

1. The Diabetes Explosion (The "Sugar Storm")

• The Stat: Diabetes prevalence jumped from 23% to 49%.
• Why? The pandemic acted like a "turbo button" for diabetes.
  • The Virus Effect: The virus itself might have damaged the body's "sugar factories" (pancreas).
  • The Lifestyle Effect: Lockdowns meant people moved less, ate more processed food, and were under more stress.
  • The Urban Penalty: Before, living in the city wasn't a huge risk factor. Now, living in a city (like Dhaka) makes you 62% more likely to get diabetes. The city environment has become a "sugar trap."

2. The Blood Pressure Drop (The "Survival Bias" Mystery)

• The Stat: High blood pressure dropped from 22% to 15%.
• The Mystery: Usually, when people get sicker, blood pressure goes up. Why did it go down?
• The Theory: The researchers suspect "Survival Bias."
  • Analogy: Imagine a group of hikers. If the weakest hikers (those with severe heart and blood pressure issues) fall ill and don't make it to the top of the mountain during a storm (the pandemic), the group remaining at the top looks "healthier" on average. The sickest people were lost to the pandemic, leaving a population that statistically looks like they have less blood pressure, even though the risk factors are still there.

3. The Broken Link Between the Two

• Before: High blood pressure and diabetes were best friends. If you had one, you almost certainly had the other.
• After: They started drifting apart. Having high blood pressure still increases your risk of diabetes, but having diabetes no longer seems to predict high blood pressure as strongly. The pandemic seems to have broken their usual partnership.

---

🎒 Who is at Risk Now?

The study found that the "risk profile" has shifted:

• Age: Being over 40 is still the biggest risk factor for both. It's like being an older car; the engine just wears out faster.
• Wealth:
  • Diabetes: It used to be a "rich person's disease." Now, it's spreading to all wealth groups. The gap between rich and poor is closing, meaning diabetes is becoming a "democratized" problem.
  • Blood Pressure: It is still more common among the wealthy, likely due to richer diets and sedentary lifestyles.
• Education: Having a higher education is still a "shield" against high blood pressure. Educated people seem better at managing their health.
• Location: Cities are now dangerous for your blood sugar. The "Urban Penalty" is real. If you live in a city in 2022, your risk of diabetes is significantly higher than if you live in a village.

---

💡 The Takeaway: What Should We Do?

The researchers conclude that we can't use the same old map to navigate this new territory.

• Old Strategy: "Treat everyone the same way."
• New Strategy: "Target the specific neighborhoods."

Because the risk is now so heavily tied to where you live (especially cities) and community factors, health officials need to stop looking at the whole country as one big block. They need to build "health shields" specifically for urban areas, improve how doctors manage diabetes, and realize that the pandemic didn't just pause life—it permanently changed the rules of the game.

In short: The pandemic didn't just pause the clock; it rewound the tape for blood pressure but fast-forwarded it for diabetes. Bangladesh needs a new playbook to handle this sudden, massive surge in sugar-related health issues.

Technical Summary

1. Problem Statement and Research Gap

Non-communicable diseases (NCDs), specifically Diabetes Mellitus (DM) and Hypertension (HTN), pose a significant public health burden in Bangladesh. While pre-pandemic baselines were established using the 2017–18 Bangladesh Demographic and Health Survey (BDHS), there is a critical knowledge gap regarding how the COVID-19 pandemic altered the prevalence and determinants of these conditions.

• The Gap: Existing literature lacks a comparative analysis of NCD trajectories before and after the pandemic in a low- and middle-income country (LMIC) context.
• The Context: The pandemic introduced unique stressors, including healthcare system interruptions, lockdown-induced sedentary behavior, psychological stress, and potential biological impacts of SARS-CoV-2 infection (e.g., inflammation, beta-cell damage), which may have accelerated NCD onset or altered survival patterns.

2. Methodology

The study employed a comparative, cross-sectional design utilizing secondary data from two nationally representative surveys: BDHS 2017–18 and BDHS 2022.

• Data Source & Sampling:
  • Sample: Pooled weighted data from 12,091 participants (2017–18) and 13,594 participants (2022).
  • Design: Two-stage stratified cluster sampling (Primary Sampling Units based on Enumeration Areas).
  • Biomarkers: Unlike studies relying on self-reports, this analysis used objective biomarker data:
    • Diabetes: Fasting Blood Glucose (FBG) > 6 mmol/L.
    • Hypertension: Mean Systolic BP > 140 mmHg or Diastolic BP > 90 mmHg.
• Statistical Approach:
  • Mixed-Effect Logistic Regression (GLMM): To account for the hierarchical structure of the survey data (individuals nested within clusters/PSUs), the authors used Generalized Linear Mixed Models with random intercepts.
  • Model Selection: Validated via Akaike Information Criterion (AIC) and Likelihood Ratio Tests (LRT), confirming the mixed-effects model provided a superior fit over fixed-effects models.
  • Intra-class Correlation Coefficient (ICC): Calculated to quantify the proportion of variance attributable to community-level heterogeneity.
  • Software: Stata (v17.0) and R (v4.5.1).

3. Key Contributions

• First Post-Pandemic Comparative Analysis: This is the first study to directly compare BDHS 2017–18 and 2022 to assess the "shock" effect of the pandemic on NCD epidemiology in Bangladesh.
• Multilevel Modeling: By utilizing mixed-effects models, the study moves beyond individual-level risk factors to quantify the growing influence of community-level determinants (contextual factors) on disease prevalence.
• Objective Biomarker Reliance: The use of measured biomarkers rather than self-reported diagnoses significantly reduces misclassification bias, providing a more accurate picture of the true disease burden.
• Divergent Trajectories: The study uniquely identifies a divergence in NCD trends, where diabetes surged while hypertension declined, challenging the assumption of uniform NCD growth.

4. Key Results

A. Prevalence Shifts

• Diabetes Mellitus (DM): Prevalence more than doubled, rising from 23% in 2017–18 to 49% in 2022.
• Hypertension (HTN): Prevalence declined from 22% in 2017–18 to 15% in 2022.
  • Interpretation: The authors suggest the decline in HTN may reflect survival bias, where individuals with severe comorbidities (HTN + CVD + COVID-19) experienced higher mortality rates during the pandemic, statistically lowering the prevalence in the surviving 2022 population.

B. Risk Factor Evolution (Multivariate Analysis)

• Age: Remained the strongest predictor for both conditions. The odds of HTN for those >40 years increased from AOR 3.57 (2017) to 4.03 (2022).
• Urbanization (New Risk Factor for DM):
  • In 2017–18, urban residence was not a significant predictor of diabetes.
  • In 2022, urban residence emerged as a significant risk factor (AOR = 1.62; 95% CI: 1.34–1.96), indicating a shift toward urban-specific metabolic risks.
• Socioeconomic Status (SES):
  • Diabetes: The wealth gradient narrowed. While the richest quintile still had higher odds, the magnitude of the disparity decreased (AOR dropped from 2.05 to 1.39), suggesting a "democratization" of diabetes risk affecting lower wealth groups.
  • Hypertension: Higher education remained a protective factor (AOR 0.64 for higher education vs. no education), while wealth remained a positive risk factor.
• BMI:
  • Overweight status became a significant risk factor for diabetes in 2022 (AOR 1.22), whereas it was non-significant in 2017–18.
  • Conversely, overweight status lost its statistical significance as a predictor for hypertension in the 2022 model.
• Bidirectional Relationship: The strong bidirectional link between DM and HTN observed in 2017–18 weakened. In 2022, HTN predicted DM (AOR 1.17), but DM no longer significantly predicted HTN.

C. Community-Level Variability (ICC)

• The ICC for Diabetes rose dramatically from 0.111 (2017) to 0.281 (2022).
• This indicates that 28.1% of the variance in diabetes prevalence in 2022 is attributable to unobserved community-level factors (e.g., local environment, healthcare access, neighborhood economics), a significant increase from the pre-pandemic period.

5. Significance and Implications

• Policy Shift Required: The findings suggest that uniform national strategies are insufficient. The rising impact of community-level factors and urban residence necessitates geographically targeted interventions, particularly in rapidly urbanizing areas.
• Integrated Screening: The weakening of the bidirectional link between DM and HTN implies that screening programs must be decoupled; relying on one condition to flag the other may become less effective.
• SDG 3.4 Alignment: The rapid escalation of diabetes threatens Sustainable Development Goal Target 3.4 (reducing premature NCD mortality). The study highlights the need for early detection and continuity of care to mitigate the long-term burden.
• Survival Bias Warning: The decline in hypertension prevalence serves as a cautionary note for epidemiologists; apparent declines in chronic disease prevalence during crises may mask underlying mortality risks among the most vulnerable.

Conclusion

The post-pandemic landscape of NCDs in Bangladesh is characterized by a rapid acceleration of diabetes driven by urbanization, community-level environmental factors, and biological pandemic effects, contrasted with a stabilization (or apparent decline) of hypertension likely influenced by survival bias. The study underscores the critical need for multilevel public health strategies that address both individual behavioral risks and the emerging structural vulnerabilities of urban and community environments.