To respond to the need for a tool that consolidates and presents data in an accessible format, the publicly accessible, online dashboard for VL was developed in 2017 through the collaboration among the International Training and Education Center for Health (I-TECH), the Côte d’Ivoire Ministry of Health (MOH), the United States President’s Emergency Plan for AIDS Relief (PEPFAR), the United States Centers for Disease Control and Prevention (CDC), The United States Agency for International Development, the Clinton Health Access Initiative, and the Kenyan MOH [24,25]. I-TECH obtained the software source code in the HTML for the Kenyan VL dashboard and adapted it for the Ivorian context to visualize monthly aggregate VL data from the open-source enterprise-level laboratory information system (OpenELIS; Digital Initiatives Group at the University of Washington [26]), the electronic laboratory information system in Côte d’Ivoire [27,28]. The visualization tool was Highcharts, and the data sources were monthly OpenELIS datasets at the laboratory level. The use case was to provide analyzed, critical information for monitoring program performance and progress toward achieving the goals for viral load testing and suppression and EID. The target users were the CDC and the PEPFAR implementing partners, policy makers at different levels of the MOH in Côte d’Ivoire, and clinical and laboratory service providers.

Data visualizations produced by the VL dashboard included bar charts and pie charts for the number of tests, turnaround time, and viral suppression with disaggregation by gender, age group, and reasons for seeking care. All charts except for the national average turnaround time could be disaggregated by month, region, district, clinical laboratory, or PEPFAR implementing partners. Users could access the disaggregated visualizations through the navigation tabs at the top of the dashboard webpage. A dedicated tab showed charts about viral suppression by gender, age, reason for the test request, and antiretroviral regimen. All data visualizations, source datasets, and a dashboard user guide were globally accessible and available for download.

In 2020 the EID dashboard was created based on the VL dashboard to display monthly aggregate EID data from OpenELIS [29]. The types of visualizations and functions are similar to those described above for the VL dashboard. More details about the development process, information visualized, and types of visualizations of the VL and EID dashboards were described elsewhere [24].

We conducted a qualitative study including 2 focus group discussions (FGDs) and 12 semistructured in-depth interviews (IDIs). Specifically, the FGDs were intended to gather general perspectives that the participants converged or diverged on about data-driven decision-making and the VL and EID dashboards. The IDIs were intended to probe for greater depth of response with themes that arose in the FGDs. Data collection took place from September 2021 to April 2022 through in-person or online meetings.

The study population consisted of existing and potential users of the VL and EID dashboards. Existing users were staff members of national and regional clinical laboratories, the CDC office in Côte d’Ivoire, and PEPFAR implementing partners. Potential users were from MOH divisions and clinical laboratories that reportedly had not used the dashboards, but the I-TECH project team and existing users thought these entities would find the dashboards useful.

To enroll existing users, we purposefully sampled individuals who were already using the dashboards based on previous meetings to participate in the first FGD and subsequent IDIs. We then used snowball sampling to ask the FGD participants to introduce us to other existing users from additional organizations to participate in the IDIs.

To enroll potential users, we first purposefully sampled a staff member of the Programme National de Lutte contre le Sida (PNLS; National AIDS Control Program) who had shared with the study team that they had not used the dashboards. The study team also identified individuals who had not used the dashboards and were heads or deputy heads of the regional clinical laboratories in Côte d’Ivoire. We purposefully chose PNLS because, as the national technical group created by the MOH to guide the national HIV/AIDS response, PNLS knew about the dashboards but reported that they did not routinely use them. Regional laboratories might also be interested in the dashboards because they were referral sites that processed all VL and EID samples referred from other laboratories in the region. We invited identified potential users to participate in the second FGD and subsequent IDIs. Recruitment stopped once we confirmed that all targeted types of organizations were represented and that thematic saturation had been reached.

The conceptual framework for the use of health data in decision-making (referred to as the “data use framework” hereafter) informed the interview guides for the 2 FGDs [4,30]. The data use framework builds on the Performance of Routine Information System Management (PRISM) framework created by the MEASURE Evaluation group [4,30,31]. The framework highlights key activities that most directly influence data demand and use and subsequently the building blocks in the health system and ultimately health outcomes. We chose to focus on the activities that would influence data demand and use to see if and how they were reflected in the participants’ responses.

The relevant constructs of the Consolidated Framework for Implementation Research (CFIR) informed the interview guides for the IDIs (Table 1) [32,33]. The CFIR is a meta-theoretical framework designed to explain the factors that affect translation of interventions into practice and their sustainment in practice across different contexts [32,33].

The interview guides were initially developed in English and later translated into French. The FGDs and IDIs were recorded, and comprehensive notes were taken. On average, the FGDs lasted 59 minutes, and the IDIs lasted 32 minutes. The recordings were transcribed verbatim in French and translated into English for analysis.

A combination of deductive and inductive approaches was used to analyze the data. Codes were created using the activities that influence data demand and use from the data use framework, the CFIR constructs, as well as those derived from the data. All transcripts were coded by one primary coder (YH). YH first chose one FGD transcript and one IDI transcript to code with the initial codebook that consisted of the deductive codes. After adding inductive codes to the codebook, YH revised the coding in the first two transcripts and continued coding the rest of the transcripts. Any subsequent changes in the codebook or coding approach were applied to transcripts that had been coded previously. We summarized the coded information into themes that corresponded to the research questions. The two frameworks served as the basis for the themes, with additional themes derived from the data that did not fit the constructs of the two frameworks. Two additional team members (YRK and PHA) reviewed the coded data and thematic summaries to validate the analysis and ensure consistency in interpretation. The coding and thematic analysis were conducted in ATLAS.ti (8 Windows; ATLAS.ti Scientific Software Development GmbH).

This study was determined to be nonhuman participants research by the University of Washington Institutional Review Board and approved by Côte d’Ivoire Comité National d’Ethique des Sciences de la Vie et de la Santé (CNESVS, Ivorian Institutional Review Board; reference number 006‐21/MSHP/CNESVS-km). All participants were unpaid volunteers and provided written informed consent. To ensure data confidentiality, the data collected were stored in a project-specific, encrypted instance of Google Drive managed by the secure server environment of the University of Washington that is compliant with the US Health Insurance Portability and Accountability Act (HIPAA) guidelines.

We conducted one FGD and 7 IDIs with existing users and another FGD and 5 IDIs with potential users. There were 26 participants across the FGDs and IDIs; 11 (42.3%) were female; 17 organizations were represented (Table 2). The participants from the CDC and PNLS were program officers; the participants from the PEPFAR implementing partners were clinical program leads or strategic information team members; and the participants from national or regional referral laboratories were heads or deputy heads of the laboratories.

This study highlights the integral role of data-driven decision-making in HIV program implementation and laboratory operations in Côte d’Ivoire. The findings reveal that data are widely used across organizational levels to monitor performance, address challenges, and inform decisions, but application varies between referral laboratories and PEPFAR implementing partners. While some organizations focus on operational monitoring, others integrate data into clinical decision-making, reflecting differences in roles and priorities. While existing users valued the dashboards for tracking performance and benchmarking, potential users faced barriers related to access, awareness, and training. Concerns about the need for additional indicators suggest opportunities for improvement. Strengthening capacity-building and data integration could enhance the impact of the dashboards on decision-making. Articles that describe dashboards displaying HIV-related data are not uncommon [34-40]. This study not only describes the features of the VL and EID dashboards in Côte d’Ivoire but also explores contextual factors, dashboard complexity, hardware availability, and internet connectivity that influenced the dashboard user experience.

The participants not only recognized the importance of data but also reported concrete behaviors of using data in decision-making. The recognition and behaviors were shared by both leaders and other staff members in an organization. This data culture, especially among leadership, seemed to be more supportive than in some LMICs reported in other studies [41,42]. The finding in our study could reflect a sample of participants who highly valued data, since more than half of the participants were existing dashboard users and had monitoring and evaluation as one of their key responsibilities.

We observed a virtuous cycle starting from a supportive data use culture at PEPFAR partner organizations and clinical laboratories in Côte d’Ivoire, which made people willing to use the VL and EID dashboards in their work. MOH supervisors can use the dashboards to provide regular feedback and comparative results to health facilities and laboratories, showcasing to both the data producers and users how the submitted data were used and enhancing the value of data collection and use [43-45]. The perceived value of data ultimately fed back into a supportive data use culture.

The participants at clinical laboratories shared that they submitted data to PEPFAR implementing partners monthly in addition to the routine data collection and use process established by the MOH. This indicates parallel data systems due to donors’ data demands, which may have been burdensome and led service providers to view themselves only as data producers [4,46]. Having access to useful tools such as the VL and EID dashboards that visualize the data that the laboratories submitted might motivate them to assume the role of data users [47].

All participants reported that their organizations had processes to validate data or ensure data quality, which gave them confidence in the quality of the data they collected, reported, and used. Key data validation or quality assurance activities include using the data validation report in OpenELIS, the electronic laboratory information system, and conducting routine data quality assessments. Our prior research supports this impression by showing that OpenELIS, the electronic laboratory information system feeding data into the dashboards, improved data timeliness and completeness significantly [48]. However, since poor data quality remains a barrier to data-driven decision-making in LMICs [41], more targeted research is necessary to understand the correlation between data quality, users’ confidence in data quality, and actual data use behaviors.

The findings about how the study participants viewed the dashboards conformed with what other studies reported, but the perceived facilitators and barriers for using the dashboards were an addition to the evidence. Ease of use was one of the most assessed dashboard characteristics [21], and the study participants found the VL and EID dashboards simple and straightforward. Simple innovations have a higher likelihood of effectiveness due to their ability to enhance user satisfaction and shorten the learning curve [49-51]. For dashboards that intend to simplify the data synthesis and analysis process for users, low complexity is an important determinant of effectiveness.

The market pressure from peer organizations may have motivated existing users to use the dashboards. This phenomenon was especially present among the organizations that worked directly with PEPFAR. It would be interesting to explore among the potential users who work indirectly with PEPFAR if the same phenomenon would apply when some potential users start using the dashboards while others do not. Late adopters often face significant pressure to implement innovations, especially when competitors or colleagues have already adopted them; when others in the field are using an innovation, individuals and organizations may feel compelled to follow suit [52]. In order for market pressure to exist, potential users first must know about the innovation. Marketing the dashboards through training and awareness activities as tools for building an organization’s credibility could be necessary before market pressure can build.

The existing users, all based in Abidjan or other urban areas, were concerned that potential users in other regions of the country might not have the necessary equipment or Internet connection to access the online dashboards and that they might need training to learn to use the dashboards. It is important to ensure the precondition of necessary equipment is met for achieving the outcome of increased user pool in the future [53]. Providing training or a support hotline might also be useful strategies to engage potential users [54].

The FGDs and IDIs were carried out by I-TECH personnel who were involved in the development and implementation process of the dashboards, potentially leading to social desirability bias in reporting barriers or issues. However, since we explained to the participants that the study was formative rather than evaluative, the social desirability bias might not have manifested to a great extent. Other limitations include potential difficulties in recalling or misunderstanding the questions. To partially address these limitations, we provided explanations of the interview questions and encouragement to offer honest critiques and suggestions for improvement.

In addition, the study sample was purposively selected and not intended to be statistically representative of all existing or potential users of the dashboards. Since this was a qualitative study, we did not collect objective observations or produce quantitative estimates of the phenomena discussed by the informants. Thus, we were unable to assess how common particular experiences or opinions may have been. Future mixed methods research could help corroborate the qualitative insights reported by triangulating them with quantitative data on dashboard use patterns and implementation outcomes.

Finally, since a single study team member coded all transcripts, there is a possibility of subjective bias or limited interpretive diversity in the coding process. This risk was mitigated by independent review and validation of the thematic analysis by 2 additional team members.

This study demonstrated that decision makers and program staff in Côte d’Ivoire’s HIV response all valued data-driven decision-making. They recognized the utility of the HIV VL and EID dashboards in monitoring performance and guiding decisions. While PEPFAR partners valued laboratory, clinical, and pharmacy data and used them in client- or program-level monitoring, laboratories demanded data on laboratory operational processes to improve workflow and quality control. These insights contribute to the broader understanding of effective data use and dashboard design in Côte d’Ivoire.