Digital Health and Data Utilisation for Improved Primary Health Services Delivery: Multi-Site Perspectives from Quality Improvement Teams in Council Hospitals in Tanzania

This study of 12 council hospitals in Tanzania reveals that while fragmented and unreliable digital health systems currently limit their effectiveness, enhancing the compatibility, usability, and reliability of these tools significantly improves access to quality data and drives better primary health service delivery.

The Gist

🏥 The Big Picture: A Hospital's Digital Toolbox

Imagine a hospital as a busy, chaotic kitchen. To run a great kitchen, you need fresh ingredients (data), a good recipe (planning), and a team that knows how to cook (staff).

In Tanzania, hospital managers are trying to upgrade from using paper recipes and chalkboards to using smart tablets and high-tech ovens (Digital Health Systems). The goal is to use these tablets to track ingredients, schedule chefs, and plan menus so everyone gets fed better.

This study asked the kitchen staff (Quality Improvement Teams in 12 hospitals): "Do these new tablets actually help you cook better, or are they just broken, confusing gadgets that get in the way?"

---

🔍 What They Found: The "Broken Tablet" Problem

The researchers surveyed 203 hospital staff members. Here is what they discovered, broken down into simple concepts:

1. The "Glitchy GPS" (Reliability & User-Friendliness)

Imagine trying to use a GPS app that crashes every time you turn a corner, or one that speaks a language you don't understand.

• The Reality: The staff felt the digital systems were unreliable (like a GPS that freezes) and hard to use (like a remote control with too many buttons).
• The Stat: About 65% of staff said the systems were unreliable for things like scheduling staff, and nearly 50% said they weren't user-friendly.
• The Result: Instead of helping, the technology often felt like a burden. Staff spent more time fighting the computer than helping patients.

2. The "Mismatched Puzzle Pieces" (Compatibility)

Imagine trying to fit a square peg into a round hole.

• The Reality: The digital systems often didn't "talk" to each other. One system might track medicine, while another tracks staff, but they don't share information.
• The Stat: Staff were slightly more positive about this (about 40% agreed it helped with ordering medicine), but overall, the systems felt fragmented. It's like having a map of the city, but the traffic lights are on a different map entirely.

3. The "Useless Tool" (Usefulness)

If you give a chef a fancy robot knife that cuts slower than a regular knife, they won't use it.

• The Reality: Many staff didn't see how the digital tools actually helped them do their jobs better.
• The Stat: Over 60% felt the systems weren't useful for planning or budgeting.
• The Cause: The study found the main culprit wasn't the technology itself, but lack of training. It's like giving someone a Ferrari but never teaching them how to drive it.

---

💡 The Good News: When It Does Work

Despite the frustration, the study found a silver lining. When the digital tools were good, they made a huge difference.

• The "Magic Key" Analogy: The study found that if a system is Useful (solves a real problem) and provides Access to Quality Data (gives you the right ingredients), it acts like a magic key.
• The Chain Reaction:
  1. Good Tools → Lead to Better Data.
  2. Better Data → Leads to Smarter Decisions (like knowing exactly how many doctors you need or how many medicines to order).
  3. Smarter Decisions → Lead to Better Patient Care.

The math showed that Access to Quality Data is the most powerful factor. If you have good data, you can make decisions that directly improve how the hospital runs.

---

🚧 Why Is This Happening? (The Root Causes)

The authors explain that the problem isn't just "bad computers." It's a mix of three things:

1. Too Many Different Systems: Hospitals are using a patchwork of different apps that don't connect. It's like trying to build a house with bricks from three different countries that don't fit together.
2. Training Gap: Staff are expected to use complex software without proper instruction. It's like asking someone to perform surgery after reading a manual once.
3. Trust Issues: Because the systems crash or give wrong numbers, staff stop trusting them and go back to paper, which defeats the purpose.

---

🏁 The Conclusion: How to Fix the Kitchen

The study concludes that Tanzania's current digital health efforts are a bit like giving a broken car to a driver and expecting them to win a race. The car (technology) has potential, but it's currently too glitchy and the driver (staff) isn't trained.

The Recommendations:

1. Connect the Dots: Stop using scattered apps. Build one big, unified system (a "Data Warehouse") where all information flows smoothly.
2. Teach the Drivers: Invest heavily in continuous training. Don't just install the software; teach people how to use it effectively.
3. Fix the Engine: Make the systems reliable and easy to use so staff actually want to use them.

The Bottom Line:
Digital health is a powerful engine for improving healthcare, but right now, it's sputtering. If the government fixes the engine (systems) and teaches the drivers (staff) how to drive, these hospitals could run much smoother, saving more lives and using resources much better.

Technical Summary

1. Problem Statement

Despite significant investments in digital health systems (such as DHIS-2) and data generation in Tanzania and Sub-Saharan Africa, there is a critical gap between data availability and data utilization for planning and decision-making.

• The Core Issue: Health data is often underutilized by policymakers and managers, leading to compromised efficiency, accessibility, and quality in primary health service delivery.
• Specific Challenges: Existing digital health initiatives in Tanzanian council hospitals suffer from fragmented, non-interoperable platforms, poor user-friendliness, low reliability, and limited perceived usefulness. These factors hinder the transition from raw data collection to actionable, evidence-based decision-making (e.g., ordering commodities, staff allocation, and budget planning).
• Research Gap: While previous studies have highlighted digital health factors in low-and-middle-income countries (LMICs), there is scant empirical evidence on how specific digital health determinants (compatibility, usability, reliability, usefulness) directly influence data use and subsequent service delivery improvements in Tanzanian council hospitals.

2. Methodology

The study employed a rigorous quantitative approach to evaluate the structural relationships between digital health factors and health service delivery.

• Study Design: Cross-sectional study conducted from March 2023 to March 2024.
• Setting: 12 Council Hospitals across 6 regions in Tanzania (Mbeya, Ruvuma, Kilimanjaro, Mwanza, Mara, and Songwe). Hospitals were purposively selected to include both high-performing (≥3 stars) and low-performing (<3 stars) facilities based on the national star rating system.
• Sampling:
  • Population: Members of Hospital Quality Improvement Teams (QIT).
  • Sample Size: 203 respondents out of a target of 218 (93.1% response rate), calculated using the Yamane formula with a 95% confidence level.
  • Selection: Multi-stage sampling involving stratification by region and performance rating, followed by simple random sampling of hospitals and respondents.
• Data Collection: Structured, self-administered questionnaires using a Likert scale to measure perceptions of digital health system characteristics (compatibility, user-friendliness, reliability, usefulness) and their impact on data use.
• Theoretical Framework: The study utilized the Unified Theory of Acceptance and Use of Technology (UTAUT) framework, adapted to include four moderating variables: Compatibility, Reliability, User-friendliness, and Usefulness.
• Statistical Analysis:
  • Descriptive Statistics: Frequencies, proportions, and means.
  • Structural Equation Modeling (SEM): Partial Least Squares (PLS-SEM) using SmartPLS 3 with bootstrapping techniques to test hypotheses and path coefficients.
  • Validation: Assessed via Cronbach's alpha, Composite Reliability, Average Variance Extracted (AVE), and indicator loadings.

3. Key Contributions

• Empirical Evidence in Tanzania: Provides one of the few multi-site, quantitative studies specifically analyzing the determinants of digital health data utilization within Tanzanian council hospitals.
• Model Validation: Successfully validates a structural model linking digital health system attributes to Access to Quality Data (AQD), Data Use for Planning and Decision-Making (DUPDM), and Improvement of Primary Health Service Delivery (IPHSD).
• Identification of Critical Barriers: Pinpoints that while data exists, the quality of the digital ecosystem (reliability, usability) is the primary bottleneck preventing effective decision-making.
• Hypothesis Testing: Rigorously tests 11 hypotheses, distinguishing between factors that have direct impacts versus those that operate through mediating variables (like access to quality data).

4. Key Results

A. Descriptive Findings (Perceptions)

• Overall Sentiment: 50.2% of respondents disagreed that digital systems effectively support data use (Mean = 2.74 ± 0.87).
• Compatibility: Moderate agreement (40.4% agreed) regarding support for ordering commodities; lower agreement for staff allocation (43.8%) and planning (38.4%).
• User-Friendliness: High dissatisfaction; 47.8% disagreed that systems are user-friendly for planning/budgeting.
• Reliability: Significant negative perception; up to 65.5% disagreed on reliability for staff allocation.
• Usefulness: Low perceived usefulness, with up to 63.5% disagreeing that systems are useful for staff allocation.

B. Structural Model Findings (PLS-SEM)

The analysis confirmed the following significant pathways (Hypotheses Supported):

1. Digital Health Factors $\rightarrow$ Access to Quality Data (AQD):
   • Usefulness (H4): Significant positive influence ($\beta=0.369, p<0.001$).
   • User-friendliness (H2): Significant positive influence ($\beta=0.540, p<0.001$).
   • Compatibility (H1): Significant positive influence ($\beta=0.145, p=0.002$).
   • Reliability (H3): Surprisingly, a significant negative influence was found ($\beta=-0.259, p<0.001$), suggesting that current reliability issues are actively hindering access to quality data.
2. Access to Quality Data $\rightarrow$ Data Use (DUPDM):
   • Strongest predictor in the model ($\beta=0.729, p<0.001$).
3. Data Use $\rightarrow$ Service Delivery Improvement (IPHSD):
   • Significant positive influence ($\beta=0.593, p<0.001$).
4. Direct Effects:
   • Usefulness had a direct positive influence on Data Use ($\beta=0.199, p<0.001$).
   • Direct paths from Digital Health Factors to Service Delivery (H5, H6, H7) and from Access to Quality Data directly to Service Delivery (H10) were rejected (not significant), indicating that these factors work indirectly through data access and usage.

C. Model Fit

• Explanatory Power ($R^2$): DUPDM had high explanatory power (0.780), while AQD (0.501) and IPHSD (0.380) showed moderate power.
• Predictive Accuracy ($Q^2$): All constructs showed predictive relevance ($Q^2 > 0$).

5. Significance and Recommendations

Significance:
The study demonstrates that simply having digital health systems is insufficient. The quality of the user experience (usability, reliability, and perceived usefulness) is the critical determinant that enables access to quality data, which in turn drives data utilization and ultimately improves health service delivery. The findings challenge the assumption that data availability automatically leads to better decision-making; without functional, reliable, and useful systems, data remains a burden rather than an asset.

Recommendations:

1. System Integration: Move away from fragmented, non-interoperable platforms. Authorities should strengthen and integrate existing systems into a unified data warehouse to eliminate redundancy and ensure seamless data exchange.
2. Capacity Building: Implement continuous, comprehensive digital health training for all relevant cadres (clinical and non-clinical) to address the gap between system design and user competency.
3. System Optimization: Focus on improving the reliability and user-friendliness of current tools (e.g., GoTHOMIS, DHIS-2) to build trust among health workers.
4. Policy Shift: Institutionalize a culture of data-informed decision-making by ensuring that digital tools are not just for reporting, but are actively used for planning, resource allocation, and commodity management.

Limitations:
The study was limited to Council Hospitals and Quality Improvement Teams, excluding other facility levels (dispensaries, regional hospitals) and management teams, which may limit the generalizability of the findings to the broader Tanzanian health system.