Trends, Distribution and Variations in Neonatal Sepsis Rates in Tanzania Health Facilities based on DHIS2 Data from 2021 to 2025

Technical Summary: Trends, Distribution, and Variations in Neonatal Sepsis Rates in Tanzania (2021–2025)

Problem Statement
Neonatal sepsis remains a primary driver of neonatal mortality in Tanzania, contributing to over 20% of the 24 per 1,000 live births neonatal mortality rate. While the Tanzania Ministry of Health (MoH) and stakeholders integrated the neonatal sepsis rate indicator into the District Health Information System 2 (DHIS2) to facilitate national, regional, and facility-level decision-making, there has been a scarcity of comprehensive analyses assessing the temporal trends and geographical variations of this indicator. Previous studies in Tanzania largely relied on primary data from specific facilities, limiting the availability of national and regional trend data. This study addresses the gap by utilizing routine DHIS2 data to evaluate the trends, distribution, and regional variations of neonatal sepsis rates across Tanzania's health facilities from 2021 to 2025.

Methodology
This descriptive cross-sectional study analyzed routine aggregated data extracted from the DHIS2 database.

• Data Source: The study utilized data from 138 health facilities (119 public, 19 private) across mainland Tanzania that reported neonatal sepsis data between January 2021 and December 2025. Ten facilities were excluded due to incomplete data.
• Definitions: Neonatal sepsis was defined according to WHO and Tanzania National Standard Treatment Guidelines (2020) as a clinical syndrome in infants aged 0–28 days with systemic signs of infection (e.g., poor feeding, convulsions, respiratory distress, fever/hypothermia). The rate was calculated as cases per 1,000 live births.
• Data Management: Data were cleaned in Microsoft Excel (v20) to ensure completeness and consistency. Statistical analysis was performed using R (v4.5.0).
• Analytical Approach: Descriptive statistics were used to calculate frequencies and proportions. Trends were visualized via line graphs. Variations in trends across subgroups (facility level, ownership, sex, location, zones, and regions) were assessed using Ordinary Least Squares (OLS) linear regression. Significance was determined at a p-value ≤ 0.05 with 95% Confidence Intervals (CI). Geographical distributions were mapped using QGIS (v3.6.3).

Key Results

• Overall Trend: The study observed a slight overall decline in the national neonatal sepsis rate, decreasing from 12.2 per 1,000 live births in 2021 to 10.9 per 1,000 live births in 2025. However, the OLS trend analysis indicated this reduction was not statistically significant (p = 0.101).
• Facility Level Variations:
  • Secondary Facilities: Reported the highest rates throughout the period, though they showed a marked reduction from 64.1 (2021) to 41.9 (2025) per 1,000 live births. This trend was not statistically significant (p = 0.566).
  • Primary Facilities: Reported a slight increase from 11.7 (2021) to 12.4 (2025) per 1,000 live births.
  • Tertiary Facilities: Showed an overall increase from 7.9 (2021) to 9.8 (2025), with a sharp spike in 2024.
• Ownership and Location:
  • Ownership: Private facilities showed a pronounced decrease (18.4 to 14.3 per 1,000), while public facilities showed a marginal decrease (10.9 to 10.5 per 1,000). Neither was statistically significant.
  • Location: Urban areas reported higher rates than rural areas but experienced a decline (21.5 to 14.5 per 1,000). Conversely, rural areas showed a gradual increase (8.2 to 9.7 per 1,000).
• Sex Disaggregation: Male neonates consistently had higher sepsis rates than females, though both groups showed slight, non-significant declines.
• Geographical Distribution:
  • Zones: The Coastal Zone reported the highest pooled rate (17.8 per 1,000), while the South West Highlands Zone reported the lowest (8.7 per 1,000).
  • Regions: The five regions with the highest rates were Lindi (21.3), Manyara (20.5), Iringa (19.9), Mtwara (19.5), and Morogoro (18.7) per 1,000 live births. Only Geita (4.9) and Rukwa (4.5) reported rates below 5 per 1,000.
• Statistical Significance: Across all subgroups (facility level, ownership, sex, zones, and location), the observed variations in trends were substantial but not statistically significant (all p-values > 0.05).

Key Contributions

• National-Level Analysis: This study provides one of the first national-level assessments of neonatal sepsis trends in Tanzania using routine DHIS2 data over a five-year period, moving beyond facility-specific primary data studies.
• Granular Disaggregation: It offers a detailed breakdown of sepsis rates by facility level (primary, secondary, tertiary), ownership (public, private), sex, and geographical zones/regions, highlighting specific high-burden areas (e.g., Coastal Zone, Lindi Region) and high-performing areas (e.g., Geita, Rukwa).
• Data Utilization: The study demonstrates the utility of DHIS2 for monitoring neonatal health indicators and identifying geographical disparities that may inform resource allocation.

Significance and Claims
The authors claim that while the overall national trend shows a slight decline, the lack of statistical significance suggests that the improvements are not yet robust or consistent across the health system. The study highlights substantial variability in sepsis rates across different subgroups, indicating uneven health system performance and service availability.

The significance of the findings lies in identifying specific high-burden zones (Coastal) and regions (Lindi, Manyara, etc.) where targeted interventions are needed. The authors suggest that the higher rates in secondary facilities and urban areas may be attributed to the referral of complex cases and the concentration of tertiary hospitals, while regional variations may be influenced by cultural practices and adherence to Infection Prevention and Control (IPC) measures.

Limitations and Recommendations
The study acknowledges limitations, including the use of aggregated data which prevents causal inference regarding individual risk factors, potential variations in data quality and diagnostic criteria across facilities, and the reliance on clinical diagnosis where laboratory confirmation may be lacking.

Based on these findings, the authors recommend:

1. Strengthening IPC interventions, particularly in high-burden zones and secondary facilities, through supportive supervision and mentorship.
2. Enhancing the quality and reliability of routine neonatal sepsis data through better verification mechanisms and capacity building for data management.
3. Prioritizing high-quality neonatal care and IPC in urban areas with high birth volumes to ensure early detection and management.