Predictors of loss to follow-up among patients receiving antiretroviral therapy in Njombe Region, Tanzania, 2017-2021

This retrospective cohort study of 37,642 patients in Njombe, Tanzania, reveals that 35.6% were lost to follow-up between 2017 and 2021, with significant predictors including poor pharmacy refill adherence, being single or divorced, and residing in the Wangingombe district, while the 2021 cohort showed improved retention.

The Gist

Imagine the fight against HIV as a long-distance relay race. The runners are the patients, the baton is their medication (antiretroviral therapy, or ART), and the finish line is a healthy, virus-free life. For the race to be won, every runner must stay in the game and keep passing the baton from one hand to the next without dropping it.

This study is like a race official looking back at the last five years (2017–2021) of a specific, very tough section of the track in Njombe, Tanzania. They wanted to find out: Why do some runners drop out of the race before they cross the finish line?

Here is the breakdown of what they found, using simple analogies:

1. The Big Picture: A High Dropout Rate

In this region, which has a very high number of people living with HIV, the researchers looked at nearly 38,000 runners.

• The Problem: About 36% of them stopped running (dropped out) before the study ended. In the world of HIV treatment, dropping out is dangerous because it lets the virus wake up, become stronger, and potentially spread to others.
• The "Ghost" Runners: The study found that the official daily count of dropouts was much lower than the reality. It's like a coach only counting people who missed one day, while the study counted anyone who missed six months. The real number of people who vanished from care was much higher than the daily reports showed.

2. The "Storm" of 2020

The study noticed a massive spike in dropouts in 2020.

• The Analogy: Imagine a sudden, violent storm (the COVID-19 pandemic) hit the race. Roads were blocked, buses stopped running, and people were afraid to leave their homes.
• The Result: In 2020, the dropout rate jumped to nearly half of the runners. However, by 2021, the race got better. Why? The organizers (health officials) changed the rules. Instead of forcing runners to visit the track every month, they started giving them a three-month supply of water (medication) at once. This "Multi-Month Dispensing" meant runners didn't have to travel as often, and fewer people got lost.

3. Who Was Most Likely to Drop Out? (The Predictors)

The researchers looked for patterns to see who was most likely to quit. Here is what they found:

• The "Young and Restless" (Age 19–28):

  • The Metaphor: Think of this group as teenagers who just got their driver's license. They are trying to drive the car alone for the first time.
  • The Finding: Young adults were twice as likely to drop out compared to older people. When kids are young, their parents drive them to the clinic and make sure they take their medicine. But when they turn 19, they have to drive themselves. Without that "parental safety net," many young adults get lost or forget the route.

• The "Solo Runners" (Marital Status):

  • The Metaphor: Running a marathon is hard. It's much easier if you have a running buddy holding your hand.
  • The Finding: People who were single or divorced were more likely to drop out than those who were married. Married people often have a partner reminding them to take their pills and taking them to the clinic. Single people often lack that daily support system.

• The "Terrain" Matters (District of Residence):

  • The Metaphor: Imagine the race track has different sections. Some are flat and paved (easy to run), while others are muddy and full of potholes (hard to run).
  • The Finding: Where you lived mattered a lot.
    • Wanging'ombe: This was the "muddy section." Runners here were twice as likely to drop out compared to the main district.
    • Makete: This was the "smooth paved section." Runners here stayed in the race the best.
    • Why? It's likely due to how far the clinic is, how much it costs to get there, or how busy the clinic staff are.

• What Didn't Matter:

  • Surprisingly, gender (being a man or a woman) didn't change the odds of dropping out. In many other places, men drop out more often, but in this specific region, men and women faced the same challenges.

4. The Solution: Tailored Coaching

The study concludes that you can't use a "one-size-fits-all" strategy to fix this. You can't just shout "Keep running!" to everyone. You need specific coaches for specific problems:

• For the Young Adults: They need a "buddy system" or peer groups to help them navigate the transition to adult care.
• For the Single Runners: They need extra emotional and social support so they don't feel alone.
• For the Muddy Districts (like Wanging'ombe): The health system needs to bring the water station closer to the runners. This might mean community clinics or mobile teams so people don't have to travel far.
• For Everyone: Keep using the "three-month supply" rule. It works!

The Bottom Line

The race to beat HIV in Njombe is still going, but too many runners are dropping out. The study shows that young people, single people, and people living in specific difficult districts need extra help to stay in the race. By fixing these specific holes in the track, Tanzania can help more people cross the finish line and stay healthy.

Technical Summary

1. Problem Statement

Loss to follow-up (LTFU) is a critical barrier to the success of Antiretroviral Therapy (ART) programs, particularly in high-prevalence regions like Njombe, Tanzania. LTFU leads to viral rebound, drug resistance, increased mortality, and undermines national goals (e.g., UNAIDS 95-95-95 targets). While previous studies have identified general predictors of attrition (e.g., male gender, low CD4 counts), there is a lack of granular, district-level evidence regarding the specific demographic and clinical drivers of disengagement in Tanzania's highest HIV-prevalence region. Furthermore, the impact of the COVID-19 pandemic on retention patterns in this specific context required investigation.

2. Methodology

• Study Design: Retrospective facility-based cohort study.
• Setting: Njombe Region, Tanzania (comprising four districts: Njombe, Wanging'ombe, Makete, and Ludewa).
• Data Source: National Care and Treatment Clinic (CTC2) database managed by the National AIDS and Sexually Transmitted Infections Control Programme (NASHCOP).
• Study Period: January 1, 2017, to December 31, 2021.
• Population: 39,566 People Living with HIV (PLHIV) initiated on ART within the region.
  • Exclusion criteria: Patients with missing demographic/ART initiation data or those transferred out of the region.
• Outcome Definition (LTFU): Defined conservatively as absence from care for ≥180 days after the last scheduled visit without documented death or transfer. (Note: This is stricter than the national standard of 90 days to ensure long-term disengagement accuracy).
• Variables:
  • Independent: Age, gender, marital status, district of residence, year of ART initiation, pharmacy refill adherence, and viral load status.
  • Dependent: LTFU status (Binary: Yes/No).
• Statistical Analysis:
  • Descriptive statistics for demographics.
  • Bivariate analysis using Chi-square and t-tests.
  • Multivariable binary logistic regression to identify independent predictors (Adjusted Odds Ratios - AOR).
  • Software: Microsoft Excel, Power BI, and IBM SPSS Statistics v26.

3. Key Results

A. Prevalence and Trends

• Overall LTFU: 13,446 patients (34.0%) were lost to follow-up over the five-year period.
• Temporal Trends:
  • LTFU peaked in 2019 (47.3%) and 2020 (46.3%), coinciding with the onset of the COVID-19 pandemic.
  • A dramatic improvement occurred in 2021, with LTFU dropping to 12.5%. This is attributed to the scale-up of Multi-Month Dispensing (MMD) and Differentiated Service Delivery (DSD) models.
• Geographic Disparities:
  • Wanging'ombe: Highest LTFU rate (47.8%).
  • Njombe: 33.2%.
  • Ludewa: 32.1%.
  • Makete: Lowest LTFU rate (23.7%).

B. Predictors of Loss to Follow-up (Multivariable Analysis)

The logistic regression model identified several significant independent predictors:

Predictor Category	Significant Finding	Adjusted Odds Ratio (AOR)	Interpretation
District of Residence	Wanging'ombe vs. Njombe	2.09 (95% CI: 1.96–2.23)	>2x higher risk of LTFU.
	Makete vs. Njombe	0.71 (95% CI: 0.66–0.76)	Significantly lower risk (protective).
Age Group	Young Adults (19–28) vs. Elderly (≥70)	2.07 (95% CI: 1.68–2.55)	Highest vulnerability; >2x risk.
Marital Status	Single vs. Married	1.25 (95% CI: 1.18–1.33)	Higher risk of disengagement.
	Divorced vs. Married	1.44 (95% CI: 1.28–1.62)	Highest risk among marital categories.
Year of Initiation	2021 vs. 2017	0.25 (95% CI: 0.22–0.29)	75% reduction in risk (protective).
Gender	Male vs. Female	Not Significant (p = 0.873)	No independent association found.

Note: Contrary to some previous literature, gender was not a significant predictor in this cohort. However, lower pharmacy refill adherence and specific marital statuses were strongly associated with attrition.

4. Key Contributions

1. Granular Geographic Insight: The study provides the first district-level breakdown of LTFU in Njombe, highlighting that Wanging'ombe requires urgent, targeted intervention compared to the better-performing Makete district.
2. Demographic Refinement: It challenges the generalization that "males" are universally at higher risk, identifying young adults (19–28 years) and unmarried individuals as the primary high-risk groups in this specific context.
3. Pandemic Impact Analysis: It quantifies the disruption caused by the pandemic (2019–2020 peaks) and documents the efficacy of programmatic adaptations (MMD/DSD) implemented in 2021 that successfully reversed the trend.
4. Methodological Rigor: By using a 180-day LTFU definition and a large, census-level dataset (N=39,566), the study offers a robust estimate of long-term retention rather than just point-prevalence.

5. Significance and Recommendations

• Policy Shift: The findings argue against "one-size-fits-all" interventions. Strategies must be district-specific (e.g., intensified tracing in Wanging'ombe) and demographically targeted (e.g., peer support for young adults).
• Programmatic Optimization: The success in 2021 validates the expansion of Differentiated Service Delivery (DSD) models, such as Multi-Month Dispensing, to reduce clinic visit frequency and transport costs.
• Social Support: The strong association between single/divorced status and LTFU suggests a need for enhanced psychosocial support and community-based care models for patients lacking spousal support.
• Data Utilization: The study underscores the value of longitudinal cohort analysis over routine dashboard metrics for identifying long-term retention trends and evaluating the true impact of interventions.

Conclusion: While LTFU remains a critical challenge in Njombe, the study identifies actionable levers for improvement. Targeting young adults, unmarried patients, and specific high-burden districts with tailored retention strategies is essential for sustaining the ART program and achieving epidemic control.