Adolescents and young adults (AYA), those aged 15 to 39 years, represent a growing share of people living with HIV worldwide. Zambia has one of the highest prevalence rates of HIV among AYA in sub-Saharan Africa; as of 2023, there are approximately 90,000 AYA between the ages of 10 and 19 years living with HIV, with only half of that population receiving antiretroviral therapy (ART) [1]. The AYA population also accounts for half of all new recorded infections in the country [1]. Compared to their peers who are living without HIV, AYA living with HIV are at an increased risk of developing cancer [2-4]. Cancers most frequently diagnosed among AYA living with HIV in Zambia are Kaposi sarcoma (KS), non-Hodgkin lymphoma (NHL), and cervical cancer (CC) [5-9]. The incidence of these cancers increases with a lower cluster of differentiation 4 cell count (CD4); therefore, adherence to ART is paramount [10]. Unfortunately, AYA living with HIV are among the most vulnerable groups with poor adherence to ART, and there is no large-scale cancer control program to facilitate early diagnosis and completion of guideline-recommended treatments in Zambia [11,12].

A cancer program focused on AYA living with HIV and tailored to the Zambian limited resource setting is needed to reduce premature mortality. Data on malignancies within this group are sparse, but late-stage presentation, unplanned breaks from cancer treatment, and abandonment of treatment have been reported among young patients with cancer in resource-limited settings [13-15]. These factors result in high morbidity and premature mortality among AYA living with HIV and diagnosed with cancer. Late-stage presentation and lack of treatment adherence are key drivers of premature cancer mortality among AYA living with HIV in Zambia.

AYA living with HIV are a vulnerable population who face psychosocial issues, stigma, and comorbidities that affect their health care-seeking behavior [16]. In Zambia, young patients living with both HIV and cancer have unmet needs related to information about health risks, coping strategies for medication side effects, positive peer and family support, and activation of personal agency [17-20]. Furthermore, during the transition from pediatric to adult HIV clinics, many of these individuals are lost to care [21-25]. Health care providers who specialize in HIV care and oncology face barriers, such as time constraints, with regard to treating cancer in AYA living with HIV, which inhibits their ability to deliver optimal care [26,27]. There are also no linkages for collaborative management of cancer in AYA living with HIV by primary care providers specializing in HIV care and cancer specialists. At the health system level, some key barriers include weak referral systems and a lack of standardized protocols to implement AYA-friendly care [28-32].

The Zambian guidelines recommend CC screening for all female AYA living with HIV who are sexually active, regardless of age [33]. However, screening among this population is low, despite the availability of free health care services. According to recent studies [18,34], <40% of AYA living with HIV have undergone screening for CC. There are no screening tests available for KS or NHL, but increasing awareness of the risk factors and symptoms, along with physical examinations, could reduce late presentation [13,29]. In fact, strategies to facilitate early detection of KS have shown promise in other African countries. A recent study in Uganda [35] used education messaging (comic strip and video) to increase KS awareness and knowledge. Furthermore, researchers in Zimbabwe developed a physical examination checklist to identify symptoms of KS [36]. The Cancer Diseases Hospital, which is the main referral center for treatments in Zambia, has adopted the National Comprehensive Cancer Network’s (NCCN’s) Harmonized Guidelines for sub-Saharan Africa for AYA oncology to optimize treatment. The NCCN guidelines will be most effective when paired in tandem with adequate treatment adherence support for AYA living with HIV.

AYA-focused programs have been successful in other settings, especially in high-income countries. These programs use a variety of models to promote collaboration across multidisciplinary teams and integrate medical and psychosocial care for AYA [37-39]. In sub-Saharan African countries such as Zambia, multiple barriers need to be addressed to facilitate early diagnosis in the primary care setting and to ensure receipt of full treatment course in cancer centers [40].

We aim to develop and test an integrated model of cancer control for AYA living with HIV that can be delivered as an embedded component in existing HIV treatment programs in primary care facilities and linked with specialized treatment in cancer centers.

This study was approved by the institutional review boards at the Population Council (Protocol 1019) and the ERES (Excellence in Research Ethics and Science) Converge (Ref. No. 2022-Dec-003) and has been registered on ClinicalTrials.gov (NCT06004011). For the AYA living with HIV aged <18 years, we will obtain both written parental consent and assent from the minors. The AYA living with HIV aged ≥18 years will be asked to provide written informed consent. Trained data collectors will administer and document informed consent. Data will be collected via a secure data platform, and no identifiable data will be collected in the baseline and annual surveys. The participants will not receive monetary compensation for participation in study education sessions and other activities. The participants will receive ZK 100 (US $4.23) for completing the data collection at baseline and at annual touchpoints. This study also has a data safety monitoring board that provides ongoing evaluation of the study’s progress. This board has the authority to review any adverse events and provide guidance for any protocol modifications or termination of the study. The board comprises 5 members from nongovernmental organizations and the Zambian Ministry of Health. Any protocol changes, or other modifications, with be shared with the institutional review boards and the data safety monitoring board.

In the HIV treatment facilities, we will generate electronic data files for the study cohort, and in other settings, such as referral facilities and the Cancer Diseases Hospital, the study team will abstract data from electronic databases or paper files. We will also conduct surveys to collect data from the AYA living with HIV and health care providers. The PCs will document their interaction with the AYA living HIV in the PC tracking tool, and the health care provider participation in the ARROW Provider Network will be captured in the study database. These details will be used to determine the “dose” of the ARROW program. All data will be collected by trained data collectors, who will undergo continual monitoring, and study coordinators, who will confirm fidelity to protocol specifications by random in-person review. Refer to Table 2 for the end point definitions and the biological, self-report, and clinical records data collected at baseline and at 12-, 24-, and 36-month follow-ups.

We will use an intent-to-treat approach to test the hypotheses (related to the dual primary and secondary end points) that AYA living HIV in the ARROW program, compared with those receiving one-time education, will have higher uptake of early diagnostic services, increased adherence to treatment, and improved outcomes.

For the facilities assigned to the intervention group in the early detection cohort, we will examine the effects at the individual level across the 2 randomization groups. Given the need to consider the influence of the cluster randomization, we will use generalized estimating equation models to estimate the effect across the trial arms. If the Hausman assumption of correlation between the random and fixed effects is violated, then we will include fixed effects representing cluster identification [74]. For the cancer treatment cohort, we will generate summary statistics for the end points and test statistical significance. We will also conduct logistic regressions to identify key factors that predict adherence.

For both trials, we will adjust for baseline covariates, including sociodemographic factors, should the initial descriptive analyses suggest differences in the distribution of these factors across the study arms. We will also explore the use of propensity scores to control for systematic differences between the AYA living with HIV in the 2 trial arms. In further analyses, we will compare differences in mediating outcomes across the randomized groups and examine the potential moderating roles of key individual behavioral and social factors hypothesized to influence adherence along the cancer care continuum (eg, social support, stigma, and self-efficacy), as shown in Figure 1. We will test the “dose” of the ARROW program received as a covariate in these models. In additional analysis, we will explore differences in the end points by age groups and type of cancer, as appropriate. We will also assess whether knowledge, self-efficacy, social support, and stigma differ by age groups and among AYA living HIV, both with and without cancer.

For measures where we have baseline and 12-, 24-, and 36-month data (eg, self-efficacy and stigma), we will perform difference-in-difference analysis, which will allow for comparisons over time and across the study arms. To understand potential variation across the HIV treatment facilities, we will conduct comparisons to analyze differences in mediating outcomes for the AYA living with HIV and the study end points. Furthermore, to understand ARROW program’s impact on health care providers, we will assess dose of the intervention (network session attendance) and change in knowledge, confidence in ability to care for AYA living with HIV, and peer collaboration.

For assessing the ARROW program, we will collect measures (as outlined in Table 3) from those receiving and implementing ARROW: 250 selected AYA living with HIV with follow-up data collection at the HIV treatment clinics and 250 AYA living with HIV at the Cancer Diseases Hospital, 36 HIV care providers, 12 oncology staff from the Cancer Diseases Hospital, 10 Health Care Collaborative members, and all the PCs. We will collect similar measures focused on the education campaign from those receiving the brochures in the one-time education campaign arm: 250 AYA living with HIV in the early detection cohort, 250 AYA living with HIV in the cancer treatment cohort, and 36 HIV care providers.

Our measurement will include the 4-item questions from the Acceptability of Intervention Measure, the Feasibility of Intervention Measure, and the Intervention Appropriateness Measure [75]. Fidelity will be assessed using data collected in the PC tracking tool and Network meeting. The fidelity measures described in Table 3 will be compiled annually, and summary statistics (eg, proportion of sessions in which all content was delivered) will be generated to assess compliance with study protocols. Sustainability will be assessed using subscales of the Sustainment Measurement System Scale [76]. We will review the scores and fidelity measures and compare scores available for both the ARROW and one-time education arms to assess implementation success and draw lessons to improve and optimize the ARROW program.

We plan to conduct interviews and focus groups to identify the underlying moderating factors associated with the successful implementation of strategies, along with facilitators and barriers to inform future scale-up activities. At 12 months, trained data collectors will interview PCs and health care providers as described for the quantitative measures. We will interview this group to seek feedback on their level of preparedness to implement the strategies, their assessment of the quality of the delivery, the level of responsiveness of the AYA living with HIV, whether they felt there were facilitators to deliver the strategies and potential barriers, and suggestions to address the barriers. We will also interview the AYA living with HIV to seek feedback on the ARROW program and one-time education campaign.

At 36 months, at the end of the trials, we will conduct focus groups with PCs, AYA living with HIV, HIV care providers, and oncology care providers. These focus groups (7-10 participants) will explore topics related to the sustainability of the ARROW program and the one-time education campaign, drawing on the findings from the results of the Sustainment Measurement System Scale survey. We will critically discuss interpretations of the data until they reach a consensus on the dominant themes and meanings. The findings will offer insights to improve the ARROW program and understand perceptions of the one-time education campaign.

We will assess the incremental cost-effectiveness of the ARROW program compared with the one-time education campaign. As shown in Table 4, we will generate the cost of start-up activities to understand the resources required to plan future implementation, the activity-based cost of strategies during the intervention trials, and the estimated cost during scale-up. We will use a previously validated instrument, the Cost Assessment Tool, to collect resource use and cost information [77].

Our main goal is to estimate the implementation cost from the program perspective. We will estimate labor hours by prospectively tracking time spent by each project staff member, including PCs, on a predefined set of activities (individuals will report their time monthly) and use hourly wages to calculate cost. We will capture time spent by health care providers, administrators, and policy makers participating in the ARROW Network and Collaborative using from the meetings. We will also document the expenditures on nonlabor resources to produce education materials, hosting group education sessions, the ARROW Provider Network meetings, and the cost of providing travel support. Using activity-based costing, we will allocate separate costs for supporting early diagnosis services and treatment adherence.

On the basis of these cost estimates and the primary end points from the trial, we will conduct incremental cost-effectiveness analysis by generating the cost per AYA living with HIV undergoing early detection services (physical examinations for KS and NHL and screening for CC) and the cost per AYA living with HIV and cancer completing recommended treatment modalities. Using standard economics methodology, we will explore economies of scale related to the implementation costs that can be achieved during scale-up [77,78]. We will also assess the key drivers of the cost of the ARROW program.

Using the trial data on estimated shift in stage at diagnosis and treatment adherence rates, along with supplemental information on prevalence of KS, NHL, and CC and mortality by stage at diagnosis, we will estimate the projected difference in mortality between ARROW and one-time education for AYA living HIV in Zambia. For CC screening, we will also estimate the incremental change in mortality using a validated microsimulation model, as screening not only will identify CC at an early stage but also can prevent CC by treating precancerous lesions [79,80].

We will derive the cost per life year saved (for all the 3 cancers combined), and our estimates will include both direct and indirect costs (patient time loss). The latter represents the opportunity cost of engaging in health services. We will assess whether the derived costs per life year saved are cost-effective or considered acceptable for implementation based on cost-effectiveness profiles of other strategies or interventions already implemented [81]. The estimated cost and effectiveness will be used to assess return on investment based on resources required to implement the ARROW program and the economic burden that will be reduced by avoiding premature mortality and treatments for adverse events related to KS, NHL, and CC for AYA living with HIV.

Furthermore, we will conduct policy simulations to assess the impact of scaling up ARROW and perform a sensitivity analysis, varying the range of effectiveness and cost estimates, and generate potential best- and worst-case projections. We will create tornado and spider diagrams to display this uncertainty graphically to policy makers. This cost-effectiveness analysis will be complemented by a budget analysis, which will identify the financial outlays that will be required annually to implement the ARROW program at scale.

The ARROW program is an intervention designed to operate at 3 levels: the individual level, the health care provider level, and the health system level. At the individual level, we hypothesize that, compared to those receiving the one-time educational campaign, participants in the ARROW program will have increased uptake of early detection services and adherence to treatment, and this will, subsequently, improve health outcomes. At the health care provider level, we believe that the ARROW Provider Network will create an environment to link HIV pediatric and adult health care providers with their counterparts at the Cancer Diseases Hospital, which will facilitate peer-to-peer learning. Improved linkages between HIV and oncology care, as well as between pediatric and adult health care providers, will benefit both the health care providers and the AYA living with HIV who move through this health care landscape. At the health system level, the ARROW Health Care Collaborative will bring together health system administrators and Zambian Ministry of Health policy makers and foster an environment for collaborative problem solving. We believe that creating a communal space for health system level barriers will result in the successful implementation of solutions to address obstacles to care that AYA living with HIV experience.

First, although cluster randomization reduces contamination across study arms, it increases the risk that the AYA living with HIV in each arm may differ at baseline. The study sampling frame, based on the HIV registers, will allow us to assess baseline differences before facility randomization and adjust our sampling process with propensity score weights. Second, data could be missing because of nonresponse and study attrition. The PCs will maintain regular contact with the AYA living with HIV, and we will use rigorous field data collection practices, including training data collectors, developing protocols, and monitoring fidelity continually. Furthermore, we will address any missing data by including demographic covariates that will serve as proxies for dropout and conducting sensitivity analyses. Third, we anticipate health care provider turnover at the HIV treatment facilities and will conduct refresher training sessions as needed throughout the trial period.

The model tested in Zambia can serve as a blueprint for other sub-Saharan African countries to ensure that AYA living with HIV receive optimal services across the cancer care continuum. We will host consultation and policy forums to share findings and seek feedback on the generalizability of the ARROW program to other sub-Saharan settings. The ARROW program, if shown to be successful, will provide a framework for implementing integrated HIV and cancer services to improve outcomes and reduce mortality among AYA living with HIV.