Geographic variation in loss to follow-up from HIV care in Tanzania and its association with pharmacy refill adherence in routine programme data

This study of 52,828 people living with HIV across Tanzania's 26 mainland regions reveals significant geographic variation in loss to follow-up and identifies poor pharmacy refill adherence as the strongest independent predictor of disengagement from care, highlighting the value of integrating spatial and refill-based monitoring for targeted retention strategies.

The Gist

Imagine Tanzania's HIV treatment program as a massive, nationwide marathon. Millions of people have started running (started taking their medication), and the goal is for everyone to finish the race strong and healthy. But in any long race, some runners get tired, lose their way, or simply stop running before the finish line. In medical terms, this is called being "Lost to Follow-Up" (LTFU).

This paper is like a race official's report that looks at the entire marathon across all 26 regions of mainland Tanzania to answer three big questions:

1. How many runners are dropping out?
2. Where are the dropouts happening the most?
3. Can we spot the runners who are about to quit before they actually leave the track?

Here is the breakdown of their findings using simple analogies:

1. The Big Picture: The Marathon Has Gaps

The researchers looked at data from over 52,000 runners (people living with HIV) between 2017 and 2021.

• The Problem: About 1 in 4 runners (26.6%) stopped showing up to the checkpoint (the clinic) for more than 6 months.
• The Variation: Just like a marathon has different terrains, the dropout rates varied wildly depending on the region.
  • The "Hilly" Regions (High Dropout): In busy, urban areas like Dar es Salaam and some rural areas like Njombe and Geita, the dropout rate was high (over 32%). It was like a steep, difficult part of the race where many people gave up.
  • The "Flat" Regions (Low Dropout): In places like Mwanza and Iringa, the dropout rate was much lower (around 19-20%). The path was smoother, and more people stayed on track.

2. The Crystal Ball: The "Refill" Clue

The most exciting part of this study is how they tried to predict who was about to quit.

Imagine you have a refill card for your medication. Every time you go to the pharmacy to get your next month's pills, you "stamp" the card.

• The "Good" Runner: Someone who stamps their card on time (getting at least 85% of their pills) is like a runner who is well-fed and hydrated. They are very likely to keep running.
• The "At-Risk" Runner: Someone who misses their stamp (getting less than 85% of their pills) is like a runner who is running out of water. They haven't stopped running yet, but they are in danger of collapsing soon.

The Finding: The study found that if a person missed their pharmacy refills, they were three times more likely to eventually stop going to the clinic entirely.

• The Metaphor: Think of pharmacy refills as the "canary in the coal mine." If the canary stops singing (misses a refill), you know the air is bad (the patient is disengaging) before the miners (the doctors) even realize the patient has left the mine.

3. The Map: Zooming In

The researchers didn't just look at the whole country; they zoomed in like a Google Maps view.

• They found that even within a single region, some specific districts were "danger zones" where people were dropping out, while neighboring districts were doing great.
• Why this matters: Instead of trying to fix the whole country with one big solution, health officials can now look at the map and say, "Hey, District X is having a problem. Let's send extra help there specifically."

4. Who is Most at Risk?

The study also looked at who was most likely to drop out:

• Men: Men were more likely to stop running than women.
• Young People: Teenagers and young adults (19–28) were more likely to drop out than older adults or young children.
• New vs. Old Runners: People who started the race a long time ago were more likely to have dropped out by now, simply because they've had more time to get tired. Newer runners seem to be sticking with it better, likely because the race rules have improved (better support systems).

The Bottom Line

This paper tells us that missing a pharmacy refill is a huge red flag.

Instead of waiting for a patient to disappear for six months before we know they are gone, we can use the pharmacy data as an early warning system. If someone misses their refill, the health system can call them, check in on them, and help them get back on track before they are truly lost.

By combining this "refill radar" with a geographic map, Tanzania can target its resources exactly where they are needed most, ensuring more people finish the marathon healthy and strong.

Technical Summary

1. Problem Statement

Sustained retention in HIV care is critical for achieving viral suppression and controlling the epidemic. However, Loss to Follow-up (LTFU) remains a persistent challenge in sub-Saharan Africa, with reported disengagement rates ranging from 20% to 40% within the first few years of Antiretroviral Therapy (ART) initiation.

• The Gap: While previous studies have documented LTFU in specific Tanzanian regions (e.g., Manyara, Njombe), there is a lack of nationally representative analyses using routine monitoring data to understand the magnitude, spatial distribution, and predictors of LTFU across all 26 mainland regions.
• The Need: There is a need for objective, scalable indicators (beyond self-reported adherence) to identify individuals at risk of disengaging from care early, allowing for targeted retention strategies.

2. Methodology

The study employed a retrospective cohort analysis using routine programmatic data from Tanzania's National Care and Treatment Clinic (CTC-2) database.

• Data Source & Sampling:
  • Population: Individuals living with HIV (PLHIV) receiving ART between January 2017 and December 2021.
  • Sampling Strategy: A simple random sample of ~60,000 individuals was drawn from the national database (containing >2 million records) to ensure computational feasibility while maintaining national representativeness.
  • Final Cohort: 52,828 PLHIV from all 26 mainland regions with valid longitudinal data.
• Definitions:
  • Loss to Follow-up (LTFU): Defined operationally as no recorded clinic visit for ≥180 days after the last scheduled appointment. This definition relies on visit-date gaps rather than facility-reported status fields to ensure standardization.
  • Pharmacy Refill Adherence: Calculated longitudinally as the percentage of days covered by medication possession. Categorized as Good (≥85%) or Poor (<85%).
• Statistical Analysis:
  • Descriptive & Spatial: Geospatial mapping (R) was used to visualize regional and district-level heterogeneity.
  • Regression Modeling: Univariate and multivariable logistic regression models were used to identify predictors of LTFU. The final model adjusted for demographic variables (age, gender, marital status) and geographic region.
• Ethical Considerations: The study used de-identified secondary data; ethical approval was granted by the Muhimbili University of Health and Allied Sciences (MUHAS) IRB with a waiver for individual informed consent.

3. Key Contributions

• National Scale Analysis: This is one of the first studies to provide a comprehensive, nationally representative assessment of LTFU across all 26 mainland regions of Tanzania using routine data.
• Validation of Refill Adherence: It empirically validates pharmacy refill adherence as a robust, objective predictor of disengagement from care within a national routine dataset, offering a scalable alternative to self-reporting.
• Granular Spatial Insights: The study moves beyond regional aggregates to reveal significant district-level heterogeneity, identifying localized clusters of high attrition that are masked in national averages.
• Differentiated Drivers: It highlights that LTFU drivers vary by context (e.g., mobility in dispersed regions vs. demographic factors in mature cohorts), supporting the need for differentiated service delivery models.

4. Key Results

• Overall LTFU Magnitude:
  • 26.6% of the cohort (14,072 individuals) was classified as LTFU during the follow-up period.
  • 73.4% remained in care.
• Geographic Variation:
  • Highest LTFU: Dar es Salaam (33.2%), Njombe (32.9%), Kigoma (32.8%), and Geita (32.7%).
  • Lowest LTFU: Mwanza (19.1%), Iringa (20.3%), and Mara (20.5%).
  • District-Level Heterogeneity: Significant variation existed within regions. For example, in Dar es Salaam, LTFU ranged from 26.9% (Temeke) to 37.4% (Kigamboni).
• Predictors of LTFU:
  • Pharmacy Refill Adherence: The strongest independent predictor. Good adherence (≥85%) was associated with a 66% reduction in the odds of LTFU compared to poor adherence (aOR 0.34; 95% CI 0.32–0.35).
  • Demographics:
    • Gender: Males had higher odds of LTFU than females (aOR 1.22).
    • Age: Adolescents (11–18), young adults (19–28), and the elderly (≥70) had significantly higher odds of LTFU compared to children (0–10).
    • Marital Status: Married individuals had lower odds of LTFU compared to single individuals.
  • Cohort Effect: Individuals entering care in earlier years (2017) had higher LTFU rates than those entering more recently, suggesting programmatic improvements over time.

5. Significance and Implications

• Early Warning System: Pharmacy refill data serves as a practical, low-cost, and objective early warning system. Patients missing refills can be identified and traced before they reach the 180-day LTFU threshold.
• Targeted Interventions: The identification of specific districts with elevated attrition (e.g., Kigamboni in Dar es Salaam) allows health programs to move from broad regional strategies to hyper-localized interventions.
• Policy Recommendation: The authors advocate for integrating refill-based monitoring with spatial analysis into routine HIV program management. This approach supports the design of differentiated service delivery (DSD) models that address specific geographic and demographic barriers to retention.
• Limitations: The study relies on routine data, which may suffer from undocumented transfers (misclassified as LTFU) and lacks socioeconomic variables. However, the large sample size and consistency with prior regional studies support the robustness of the findings.

Conclusion:
The study confirms that LTFU is a major, geographically variable challenge in Tanzania. It establishes that pharmacy refill adherence is a powerful, independent predictor of retention, suggesting that routine program data can be leveraged to proactively identify and support at-risk populations, thereby improving the sustainability of ART delivery.