Loss to follow-up (LTFU) from rifampicin-resistant tuberculosis care (RR-TB) is challenging current TB control efforts [
1]. This incomplete treatment can lead to further resistance to available anti-TB drugs and the transmission of drug-resistant TB (DR-TB) to contacts [
2]. Of those who are LTFU, the majority die within just a few years of leaving treatment, if not reengaged in care [
3]. In South Africa (SA), recent studies indicate that between 6 and 16% of people undergoing RR-TB treatment are LTFU [
4,
5], meaning that they have missed greater than two consecutive months of treatment [
6]. Given the estimated 21,000 incident cases of drug-resistant TB that occur annually in South Africa [
7], addressing LTFU is a public health imperative.
Despite the dire implications of LTFU, effective interventions to reduce it are lacking. Known risk factors for LTFU in SA include male sex, younger age, HIV status, alcohol use, substance use, and site of TB disease, while protective factors include stable housing and steady employment [
8]. Unfortunately, most of these risk factors are either non-modifiable or the distal results of complex societal problems that are challenging for the TB control program to address immediately. Identifying new, modifiable, and actionable risk factors is imperative to decreasing LTFU and improving RR-TB treatment outcomes.
One potential risk factor for LTFU that has yet to be thoroughly investigated is distance traveled to receive RR-TB treatment. Within KwaZulu-Natal (KZN) and Eastern Cape (EC) provinces, RR-TB treatment is initiated and monitored at centralized and decentralized facilities that provide access to the specialized medications and expertise needed to treat RR-TB [
9,
10], requiring some patients to travel long distances to initiate and/or receive RR-TB care [
11]. Although RR-TB treatment decentralization has been ongoing since 2011 [
9], participants in more recent qualitative studies still express that transportation time and cost are barriers to care retention [
12]. Travel distance is an obvious common factor underlying these findings; however, no prior studies have analyzed its impact on RR-TB care retention in SA. Accordingly, this analysis aims to evaluate the relationship between travel distance to RR-TB treatment on LTFU across 10 RR-TB treatment sites in KZN and EC provinces in SA.
Discussion
This analysis examined the relationship between distance traveled to RR-TB treatment and LTFU in the KZN and EC, South Africa. South Africa has made great strides in decentralizing RR-TB, transitioning to an ambulatory care model for RR-TB and increasing the number of RR-TB treatment initiation sites from 17 to 658 [
9]; however, KZN and EC still have the least decentralized RR-TB services in SA [
9]. This analysis demonstrates that one-quarter of patients in these provinces must travel at least 60 km to receive RR-TB care, a substantial burden when you consider the road quality and transportation access issues in SA, as well as the cost and time associated with these trips.
This analysis demonstrated that these long travel distances (> 60 km) adversely impact care retention, increasing the odds of LTFU from RR-TB care by 91%. Despite lack of RR-TB research on this topic, these findings are congruent with older research demonstrating poor outcomes (i.e., death) for patients traveling long distances (> 60 km) to receive DS-TB treatment at district hospitals, which was conducted when DS-TB diagnosis and intensive phase treatment still took place at more centralized sites [
21]. Research conducted within other African countries also links long travel distance to decreased TB treatment success [
23,
24], increased death during TB treatment [
25], and TB treatment delay [
26]; however, the findings on LTFU are mixed. Despite the conflicting findings about distance and LTFU in Africa, the preponderance of the global research demonstrates a relationship between increased distance to TB treatment and decreased patient engagement [
27‐
30]. Further, this data is supported by numerous qualitative studies in which people with TB indicated that increased distance from care and travel costs negatively impacted engagement, adherence, and care retention [
12,
31‐
33].
Our analysis reinforces the current, but limited, data demonstrating that increased travel distance to treatment is linked to poor TB outcomes, regardless of drug-resistance pattern. One potential way to intervene on this relationship within South African RR-TB treatment facilities is more frequent down referral to and care coordination with primary healthcare facilities (PHCs). PHCs are available within 5 km of 90% of South Africans, allowing people with RR-TB to receive treatment closer to home [
34]. Based on current decentralization policies, monitoring of RR-TB treatment is within the purview of PHCs [
9]. Newer, short-course regimens, like BPaL [
35], offer hope for greater integration within the PHC clinics. Efforts focused on building the capacity of the healthcare workers at PHCs, particularly nurses [
36,
37], to effectively monitor RR-TB treatment would allow SA to capitalize on existing infrastructure to bring RR-TB closer to patients and improve treatment outcomes.
Beyond increasing geographic access, other covariates provide insight into patient engagement in care. The specific site where someone received treatment impacted their odds of LTFU. Discussions with the parent study staff revealed additional site level characteristics that may have impacted LTFU. Most notably, differences in treatment site accessibility via bus or taxi routes were hypothesized to have impacted LTFU, an important topic for future patient engagement research.
The transition away from injectable regimens has led to a notable reduction in the odds of LTFU, consistent with the initial studies reporting the outcomes of all-oral RR-TB regimens in SA [
5]. All-oral regimens eliminated the need to travel to health facilities daily for aminoglycoside injections, likely decreasing the impact of travel distance on care outcomes. Most individualized regimens were the result of side effects from or contraindications to injectables which may explain why individualized regimens also reduced the odds of LTFU.
Among the nonmodifiable risk factors, we continue to see that younger males are at increased risk for LTFU, making them potential targets for interventions that promote retention [
8]. The relationship between HIV and LTFU is less clear cut. Although our analysis showed HIV increased risk for LTFU, past research on this relationship is mixed [
8,
38]. In populations with high rates of RR-TB/HIV coinfection, increased regimen complexity, additional side effect burden [
39], and the cumulative impact of HIV and TB stigma may contribute to poorer retention of people living with HIV in RR-TB care [
40]. Finally, this study links completion of secondary school with decreased odds of LTFU. Such a relationship between education level and LTFU has been seen in other geographic areas [
22,
41], but not in recent studies within South Africa [
42,
43]. Although there is limited ability to modify the education level of adults in treatment for RR-TB, there is potential to provide disease- and treatment-specific education. Such educational interventions may decrease LTFU, as better TB-specific knowledge has been shown to decrease LTFU in other areas of the globe [
44].
One major limitation of this study is varying ward sizes. Because ward size is based on population, wards that are less densely populated are geographically larger. As a result, ward centroids are a less accurate proxy for residential locations in larger wards. In future studies, we recommend the collection of geolocations for residences. This data is important for more accurate distance-to-treatment analyses, and may also be useful in tracking patients who are lost from care, which is why it was initially recommended in the first
Policy Framework on Decentralisation and Deinstitutionalized Management for South Africa [
45]. Additionally, we were unable to capture transit type, taxi/bus routes, transit time, transit cost, and migration/movement throughout treatment, which may more accurately represent the experience of traveling to and from RR-TB treatment than simple distance. Finally, there were not enough sites to use a multilevel modeling approach, but treatment site was controlled for in the analysis. Despite these limitations, this is still one of the first analyses to characterize distance traveled to receive RR-TB treatment in SA and link it to LTFU, thus supporting the importance of continued decentralization efforts.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.