Predictive Modelling to Determine Risk of Attrition from ART Services Due to Loss-to-Follow-Up Among Clients on Antiretroviral Therapy in Zimbabwe

In the vast landscape of modern medicine, few achievements are as profound as turning a once-fatal diagnosis into a manageable chronic condition. For people living with HIV, daily medication known as antiretroviral therapy, or ART, acts as a powerful shield, keeping the virus suppressed and allowing the immune system to function normally. However, the success of this medical miracle depends entirely on a simple, human factor: consistency. Patients must take their medication every day and return to clinics regularly to collect their supplies and monitor their health. When a patient stops coming to the clinic without a formal goodbye, they are said to be lost to follow-up. This gap in care is dangerous; without the medication, the virus rebounds, the risk of drug resistance rises, and the threat of death returns. In many parts of the world, keeping patients engaged in this long-term care journey is the single biggest hurdle to controlling the epidemic.

In the rural districts of Zimbabwe's Matabeleland North province, researchers set out to understand exactly who is most likely to fall through the cracks of this care system and why. They turned to a massive collection of digital health records, looking back over nearly a decade of patient history to find patterns that could predict who might stop coming to the clinic. By analyzing the records of nearly 3,800 adults and adolescents who started treatment between 2010 and 2018, the team built a model to identify the specific moments and characteristics that signal a high risk of dropping out. Their goal was not just to count the missing patients, but to map the terrain of vulnerability so that health workers could intervene before a person was lost.

The study revealed a stark reality: losing patients to follow-up is a persistent challenge, with more than a third of the people in the study eventually disappearing from the clinic records. Over the course of the study, which tracked patients for an average of six years, the team calculated that for every 1,000 months of observation, nearly nine patients would fail to return for their medication. While this rate might seem small in isolation, the cumulative effect over years and across thousands of people represents a significant loss of life and a major barrier to ending the epidemic. The data showed that the risk of dropping out is not evenly distributed; it is heavily concentrated in specific groups and specific times.

The most striking finding was that age plays a critical role in retention. Younger patients, specifically adolescents between the ages of 15 and 19, faced a significantly higher risk of leaving care compared to older adults. The analysis showed that adults were substantially less likely to drop out than their teenage counterparts. This suggests that the transition from childhood to adulthood, with its unique social pressures and developmental challenges, creates a fragile period where staying on treatment becomes difficult. Furthermore, the length of time a person had been on treatment mattered immensely. The first few years of therapy were the most dangerous period for attrition. Patients who managed to stay in care for five years or longer became remarkably stable, with their risk of dropping out dropping to a level so low it was almost negligible compared to those in their early years of treatment.

Other factors painted a more complex picture. While one might assume that being married or having a higher level of education would protect a patient from dropping out, the data did not support this as a clear rule. In fact, the study found that being married or having a tertiary education did not significantly lower the risk of loss to follow-up in this rural setting. This hints that the barriers to care are deep and structural, perhaps related to the fear of stigma or the difficulty of traveling to clinics, which affect people regardless of their social standing or education. The researchers also noted that the risk of dropping out was highest during the early post-initiation phase, driven by the initial side effects of medication and the psychological adjustment to a lifelong treatment regimen.

To make sense of these patterns, the researchers used a statistical approach that tracks the time until an event happens, in this case, the event of a patient missing an appointment. They confirmed that their model was robust and reliable, ruling out the idea that the results were due to random chance. The study did not find that gender or the initial strength of a patient's immune system were strong predictors of who would leave care; instead, the timeline of treatment and the age of the patient were the dominant forces. The researchers acknowledged that their data came from electronic records, which means some patients might have moved to other clinics without telling anyone, a phenomenon known as a silent transfer. This possibility suggests that the true number of people lost to follow-up might be even higher than what the records show.

The implications of these findings point toward a need for targeted support rather than a one-size-fits-all approach. The study suggests that health programs should focus intensely on the first two years of treatment, a window of high vulnerability where patients are most likely to default. It also highlights the urgent need for specialized support for adolescents, who face unique hurdles in staying engaged with care. The authors propose that using these predictive insights to flag high-risk individuals before they miss an appointment could allow clinics to reach out with peer support or flexible services. By understanding that the risk is highest for the young and the newly treated, health systems can direct their limited resources to where they are needed most, ensuring that the life-saving promise of antiretroviral therapy is kept for everyone, not just those who can navigate the system without help.