Improving Services for HIV-Exposed Infants in Zambia and Cameroon Using a Quality Improvement Collaborative Approach

To bridge the gap between what is known and what is done, quality improvement collaboratives (QICs) enable health programs to rapidly address quality challenges at scale. Two QICs in Cameroon and Zambia improved coverage of early infant HIV testing and initiating antiretroviral therapy in HIV-exposed infants. The QIC approach empowers health care workers to design solutions tailored for their specific settings.

Introduction: Early infant diagnosis (EID) and rapid antiretroviral therapy (ART) initiation are lifesaving interventions for HIV-infected infants. In Cameroon and Zambia, EID coverage for HIV-exposed infants (HEIs) is suboptimal and the time to ART initiation for infants infected with HIV often exceeds national standards despite numerous policy and training initiatives. In each country, MOH led project design and site selection. MOH and ICAP provided quality improvement training and monthly supportive supervision, which enabled HF teams to conduct root cause analyses, design and implement contextually appropriate interventions, conduct rapid tests of change, analyze monthly progress, and convene at quarterly learning sessions to compare performance and share best practices. Results: In Cameroon, EID testing coverage improved from 57% (113/197 HEIs tested) during the 5-month baseline period to 80% (165/207) in the 5-month endline period. In Zambia, EID testing coverage improved from 77% (4,773/6,197) during the 12-month baseline period to 89% (2,144/2,420) during the 3-month endline period. In a comparison of the same baseline and endline periods, the return of positive test results to caregivers improved from 18% (36/196 caregivers notified) to 86% (182/211) in Cameroon and from 44% (94/214) to 79% (44/ 56) in Zambia. ART initiation improved from 44% (94/214 HIVinfected infants) to 80% (45/56) in Zambia; the numbers of HIVinfected infants in Cameroon were too small to detect meaningful differences. Conclusions: QICs improved coverage of timely EID and ART initiation in both countries. In addition to building quality improvement capacity and improving outcomes, the QICs resulted in a "change package" of successful initiatives that were disseminated within each country.

INTRODUCTION
S ince the release of the Joint United Nations Programme on HIV/AIDS (UNAIDS) Fast Track declaration in 2015, the global community has worked to achieve HIV epidemic control by 2030 by ensuring that at least 95% of people living with HIV are aware of their status, 95% of those aware of their status are linked to antiretroviral therapy (ART), and 95% of those on ART have achieved viral suppression. 1,2 Although many countries have made remarkable progress toward reaching these goals for adults, the same cannot be said for infants and children. In 2018, for example, the global community achieved only 59% of pediatric ART coverage targets. 3 High-quality national programs are essential to prevent early mortality due to pediatric HIV, which peaks at 3-4 months 4 and approaches 50% by 2 years of age. 5 Unfortunately, the health systems required to deliver HIV testing and treatment services to infants and children are lacking in many settings, and for two-thirds of HIV-infected children in Africa, Asia, and the Americas, HIV is only diagnosed when the children have advanced immunodeficiency, leading to high rates of preventable morbidity and mortality. 6 The interventions needed to prevent motherto-child transmission (PMTCT) of HIV and swiftly identify HIV-infected infants and link them to ART can be conceptualized as a cascade of services ( Figure 1). Health systems must consistently and correctly identify and engage HIV-infected pregnant women, provide ART for those not already on treatment, and deliver a package of services to their HIV-exposed infants (HEIs). These early infant diagnosis (EID) interventions include maternal counseling, HIV testing before 8 weeks of age, rapid return of results to parents/caretakers and treating clinicians, and prompt ART initiation for HIV-infected infants.
Cameroon and Zambia are both facing generalized HIV epidemics, with adult HIV prevalence of 3.1% and 11.5%, respectively. 7,8 Although Zambia's PMTCT and EID coverage are substantially higher than those in Cameroon (Table 1), neither country has optimized their PMTCT-EID cascade, and substantial gaps remain in HIV services for infants born to HIV-infected women. 9 Both Cameroon and Zambia have national PMTCT and EID policies, strategies, guidelines, training curricula, and systems for supportive supervision, supply chain management, and program monitoring and evaluation. Despite these national efforts, consistent implementation of EID services is lacking. 10,11 While variation exists within the quality and/or effectiveness of the implementation of these health system inputs, bridging this ongoing "know-do gap" has become critically important for improving patient outcomes.
The know-do gap between established standards of care (what we know) and the ability of health systems to produce improved outcomes (what we do) has become an area of focus for country HIV programs and international donors. 12 The use of quality improvement (QI) methodologies has been successful in closing challenging know-do gaps such as those seen in Cameroon and Zambia. [13][14][15] In particular, the QI collaborative (QIC) approach has shown great promise in improving health programs in low-resource settings. [16][17][18][19][20][21] To bridge the EID know-do gap in Cameroon and Zambia, ICAP partnered with the Ministry of Health (MOH) in each country, donors, and implementing partners to design and implement the QIC projects to improve 3 key steps in the cascade: EID coverage, timely return of HIV test results, and rapid ART initiation for infants found to be HIV infected.

QIC Methodology
ICAP and its MOH partners used a wellestablished QIC approach based on the Institute for Healthcare Improvement (IHI) known as the Breakthrough Series, which supports multiple health facilities (HFs) to address the same quality challenge at the same time to achieve rapid, measurable, and sustained improvements. 18,21 The QIC approach begins with convening key MOH stakeholders to identify the specific health care quality challenge and kick-start the design of the QIC (Figure 2). The partners collaborate to select project HF sites and develop shared QIC aims (targets), indicators, and a measurement strategy. Baseline assessments inform the development of appropriate and specific QIC aim statements and provide data with which to monitor and assess QIC progress. 21,22 Multidisciplinary QI teams are established at each participating HF with various cadres of HF staff; after baseline training and orientation, each QI team is supported to identify contextually appropriate interventions and perform rapid iterative tests of change using the Model for Improvement and its plan-do-study-act (PDSA) cycles. 13 The PDSA approach helps teams test The QIC approach has shown great promise in improving health programs in lowresource settings.
The PDSA approach helps teams assess changes and possible improvements; the results guide whether to implement, modify, or abandon the proposed intervention.
changes and see whether they yield improvements; the results are then analyzed to decide whether to implement, modify, or abandon the proposed intervention. If the intervention does not achieve the desired results, it is modified or replaced and the PDSA cycle is repeated. 23 HFs then come together for quarterly meetings, in which they compare progress and share interventions and innovations. 22 In addition to building QI capacity and improving targeted outcomes, QICs often develop a "change package" of tools, strategies, and best practices that can be shared across teams, scaled up, and widely disseminated. This approach has been shown to produce large-scale improvements in both high-and lowresource settings. 19,24,25 In both Cameroon and Zambia, ICAP supported the national MOHs to design the QICs in collaboration with a panel of expert stakeholders including MOH leaders, the Health Resources and Services Administration (HRSA), the Centers for Disease Control and Prevention (CDC), and local implementing partners. MOHs had final approval of site selection, indicators, and other key project elements. Because of this collaborative MOH-led design process, the 2 QICs were slightly different in their design and focus ( Table 2).  Abbreviations: ART, antiretroviral therapy; EID, early infant diagnosis; PMTCT, prevention of mother-to-child transmission of HIV.

Baseline Assessment
Retrospective aggregate monthly data were abstracted from HF registers by HF staff using a standardized paper-based tool. HF and ICAP staff then entered these baseline data into a standalone District Health Information System (DHIS2) 28 database and conducted descriptive analyses. In Cameroon, the baseline assessment included 5 months of data (October 2015 to February 2016). In Zambia, the baseline assessment included 12 months of data (January to December 2016).

Staffing, Training, and Project Launch
Each participating HF assembled a multidisciplinary QIC team, including maternal-child health

FIGURE 2. Quality Improvement Collaborative Approach Used to Improve Early Infant Diagnosis and Antiretroviral Therapy Initiation and Health Facilities in Cameroon and Zambia a
Abbreviations: QI, quality improvement; QIC, quality improvement collaborative. a Adapted from the Institute for HealthCare Improvement Breakthrough Series.
Each HF assembled a multidisciplinary QIC team, including nurses, medical officers, laboratory technicians, lay counselors, and/ or HF managers.
Improving Services for HIV-Exposed Infants in Zambia and Cameroon www.ghspjournal.org Global Health: Science and Practice 2021 | Volume 9 | Number 2 and ART nurses, medical officers, laboratory technicians, lay counselors, and/or HF managers. Each team attended an initial 1-week workshop, at which ICAP provided training on QI tools and methods and QIC indicators, as well as refresher training on national EID and infant ART guidelines. In Zambia, materials from the national QI curriculum were used throughout the 1-week workshop. In both countries, Cameroonian and Zambian national QI leaders served as expert trainers and facilitators. Using their baseline data, HF teams used process maps, fishbone diagrams, and driver diagrams to identify root causes associated with low EID coverage, loss to follow-up, and low rates of ART initiation for HIV-infected infants. Teams then selected and prioritized interventions ("change ideas") tailored to their HF and designed to produce rapid improvement. Institutionalizing QI and building local QI capacity at the central and facility levels are critical to achieving and maintaining epidemic control. Evidence shows that country MOH QI capacity building requires action-based learning and ongoing skill building of QI systematic approaches to produce self-sustaining and scalable change. 29,30 Effective QI training requires experiential learning whereby both staff and leaders within QI teams can directly apply knowledge within the context of ongoing QI projects using the currently available resources. In both countries, the initial QI workshops were designed to provide QI teams composed of MOH central, district, and facility staff with the opportunity to apply QI theory to everyday work directly at each HF through the use of common practical approaches and tools. 27,28

Supportive Supervision and Quarterly Learning Sessions
Each month, in collaboration with local partners and MOH, ICAP provided on-site supportive supervision, QI coaching, and mentoring on data collection and analysis to the QIC teams at each HF. The supportive supervision visits were led through each MOH district health team per their routine HIV program supervision schedule. The visits provided MOH QI mentors with the opportunity to help HF teams identify successful change ideas and link progress to real-time data in DHIS2. The use of PDSA cycle implementation is the heart of QI and is well embedded into the QIC approach. 29 The PDSA framework is grounded in continuous learning and guides thoughtful teambased actions. The tool is intended to help health care workers (HCWs) implement and practically assess if a desired change is leading to improvement in real time. 31 The process typically involves conducting several different "tests of change" Abbreviations: ART, antiretroviral therapy; EID, early infant diagnosis; HEI, HIV-exposed infant.
Improving Services for HIV-Exposed Infants in Zambia and Cameroon www.ghspjournal.org Global Health: Science and Practice 2021 | Volume 9 | Number 2 followed by systematically documenting, analyzing, adapting, retesting, and re-evaluating the iterative PDSA cycles using the PDSA worksheets as a tool. 32 Although this essential QI methodology appears straightforward and clear, recent literature suggests that real-world application is more complex, with wide variation in how changes are tested and challenges that cause teams to implement changes with lack of rigor and consistency. 33,34 General parameters and group consensus for the identification of successful change interventions included a demonstration of HCW compliance with implementation, staff satisfaction with the intervention, improvements in monthly data, and predictions about sustainability. Quarterly follow-up learning session workshops were convened for each project. QIC teams reported their progress using the shared indicators and described their PDSA cycle results. These peer-to-peer meetings provided HF staff with the opportunity to share lessons learned, best practices, failed ideas, and successful tools. These meetings also enabled each QIC team to benchmark their progress against other HFs via friendly competition and to communicate with senior leaders regarding their experiences while advocating for above-site, district level interventions. Each team was also provided with time to plan for the next quarter's activities. During the final learning session, the highest-performing teams were recognized with awards.

Data Collection and Analysis
Each month, HF QI teams collected aggregate anonymized data, shared the data with ICAP using standardized paper forms, and plotted their data on annotated run charts. ICAP staff entered the data into a dedicated DHIS2 database that was systematically reviewed every month for data quality. If errors were identified, HFs were contacted to obtain the correct information. Microsoft Excel 2012 was used to generate monthly descriptive statistics and graphs showing progress toward targets for each HF as well as the performance of the collaborative as a whole. QIC indicator performance was assessed for each HF during the implementation period and the range, mean, and median across HFs was calculated. In addition to descriptive statistics, project baseline performance was compared with performance during the final endline period (3-5 months) of the intervention period using the chi-squared test of independence. Run charts were constructed by the QIC HF teams, who entered the data every month. Run chart rules of analysis were utilized to monitor processes, measure performance to the aim, and measure the impact of change interventions. 35,36 Dissemination of Successful Change Ideas After 15 months of implementation, each project convened a final stakeholders' meeting in which experiences, results, and the package of successful changes and best practices (Table 3) were shared with MOH, regional-and district-level health leaders, implementing partners, and representatives of the U.S. President's Emergency Plan for AIDS Relief (PEPFAR) agencies. During this final meeting, higher-performing HFs were invited to present their project results and change interventions, and MOHs and their partners developed strategies for sustainability, scale-up, and spread of improvements to other parts of their respective country. Funding for each project was completed at the end of the implementation of each QIC. The MOHs in Zambia and Cameroon will lead the scale-up and institutionalization of the QICs in their respective countries.  May 2018) in Zambia. Root cause analyses revealed that typical barriers to implementation of the EID cascade included staff knowledge deficits, unclear roles and responsibilities, process breakdowns, and systems bottlenecks. In response, the QIC teams tested interventions related to test result management, improved staff and client education, staffing modifications, workflow process modifications, commodity management, documentation, and data quality improvements ( Table 3).

Ethical Review
The PDSA method provided HCWs with a simple algorithm for implementing, testing, and Although the essential QI methodology appears straightforward and clear, the real-world application is more complex.
The PDSA method provided HCWs with a simple algorithm for implementing, testing, and adapting improvement interventions.
Improving Services for HIV-Exposed Infants in Zambia and Cameroon www.ghspjournal.org Global Health: Science and Practice 2021 | Volume 9 | Number 2 adapting improvement interventions. The multidisciplinary HF teams worked together to develop and test contextually specific interventions tailored for their sites and communities. For example, a few QIC teams in Cameroon identified the common practice of pediatric caregivers providing incorrect contact information to nurses, which hindered necessary follow-up. In response, these teams developed enhanced patient education, including one-on-one counseling and focus groups, to build trust between clients and staff and to explain the importance of HIV testing at recommended intervals, as well as the rationale behind requesting contact information. As observed in other settings, QIC leadership teams found that while PDSA is simple in theory, application in real life can be somewhat complicated and requires QI teams to thoughtfully dissect their data while drawing shared conclusions about the intervention and its subsequent "success" or "failure" with progress to the QIC aim. The time required for HF teams to become more independent with PDSA cycle management ranged from 6 to 9 months. During site support visits, MOH leadership and QIC teams initially observed common challenges with QI team PDSA cycle implementation including inadequate planning of the "who, what, where, when, and how" of the change; poor documentation; implementation on too large of a scale; failure to secure sitelevel leadership buy-in; poor data quality; and poor communication between team members. Over time, HF teams mastered the skills needed to independently identify, implement, and test change ideas. Table 3 outlines the most successful change interventions identified through group consensus and QI team professional expertise using data and subjective assessment of each intervention.
The QIC approach itself and the facility-level changes led to an improvement in performance in all 3 steps in the EID cascade: (1) early HIV testing for infants under 8 weeks of age, (2) timely return of EID results to caregivers, and (3) rapid ART initiation for infants found to be HIV infected (Tables 4 and 5, Figure 3).
In Cameroon, EID testing for HEIs under 8 weeks of age improved from an aggregate performance of 57% (113 tested of 197 eligible for testing) during the 5-month baseline period to 80% (165 tested of 207 eligible for testing) during the 5-month endline period (P<.01). In Zambia, EID testing improved from an aggregate performance of 77% (4,773 infants under 8 weeks of age tested of 6,197 total infants tested) during the 12-month baseline period to 89% (2,144 infants under 8 weeks tested of 2,420 total infants tested) during the 3-month endline period (P<.01). In a comparison of the same baseline and endline periods, the return of positive test results to caregivers improved from 18% (

DISCUSSION
Despite the availability of relevant policies, guidelines, and training initiatives aimed at improving the HEI cascade in Cameroon and Zambia, quality challenges had persisted for years. We found that the use of QIC methodology enabled providers to bridge the know-do gap and was associated with substantial improvements in HEI testing coverage, return of results to caretakers, and swift linkage of HIV-infected infants to treatment in both countries.
The use of clear targets, defined processes, collaborative problem-solving, and ongoing performance evaluation were critical enablers of the successful QI collaboratives. The social dynamics of the QIC approach enhanced the sense of shared purpose and community among HCWs, fostered teamwork and friendly competition, and built leadership support while creating an internal enabling environment at the facility level, characteristics of QI projects that have been noted elsewhere. [16][17][18] Quarterly learning sessions, monthly data collection, and QI mentoring encouraged the rapid and sustained improvements and facilitated diffusion of innovation. These fundamental activities provided site-level teams with consistent and supervised opportunities to identify and address ongoing challenges to program implementation while continuously measuring progress toward the aim.
Strengths of the project included MOH leadership; the number of HFs; the magnitude and consistency of improvements in these critically important and challenging service delivery domains; the similarity of the findings in 2 countries with different HIV epidemics and EID responses; and the development of resources, Abbreviations: ART, antiretroviral therapy; DBS, dried blood sample; EID, early infant diagnosis; HEI, HIV-exposed infant; PCR, polymerase chain reaction.
The use of clear targets, defined processes, collaborative problem-solving, and ongoing performance evaluation were critical enablers of the successful QI collaboratives.
Improving Services for HIV-Exposed Infants in Zambia and Cameroon www.ghspjournal.org methods, and tools that can be used at additional HFs.
As highlighted by Kruk et al. 37 and the Lancet Global Health Commission on High Quality Health Systems, more global deaths are due to poor-quality care than insufficient access to health services, and high-quality health systems could prevent more than 8 million deaths a year in low-and middle-income countries. Investing in quality management-including the development of quality standards, measurement of quality indicators, and implementation of QI methodologies -is increasingly a priority of both MOHs and global health donors. For example, PEPFAR highlights the importance of QI methods in its guidance, funds QICs in multiple partner countries, and supported a multi-year QI capacity-building course for health ministry partners across sub-Saharan Africa. 30 In 2018, the World Health Organization, the World Bank, and the OECD collaborated on a call to action for quality health services, calling for the development of national health care quality policies and strategies inclusive of improvement methods and interventions. 38 Multiple MOHs have incorporated QI methodsincluding QICs-into national policies and guidelines, ensuring that these activities take place irrespective of donor involvement. In Mozambique, HF teams implement QI projects with support from MOH, and quarterly provincial supervision occurs routinely with or without donor support (Isabel Pereira, MD, CDC Mozambique, personal communication, 2019). In Tanzania, MOH HIV program leaders routinely perform QI-focused supportive supervision and mentoring activities independent of donor involvement. 39 Experience shows that quality management, including the use of QI methods and tools, is a high-value and sustainable approach to health systems strengthening. Not all quality challenges require QICs, however, and identifying when a QIC is the optimal intervention is a priority for MOHs and other implementers. As above, important criteria include a high-priority quality challenge shared by multiple sites; a refractory quality challenge that has not improved following simpler interventions, such as training, quality assurance, and/or single-site QI activities; and an enabling environment including available HF staff and strong leadership support.

Limitations
As with most QIC projects and time-series analyses, inferring causality between the intervention and the results in the current study is limited by the absence of a control or comparison group, and generalizability is limited by the nonrandom selection of HFs. 40 However, expert consensus suggests that randomized studies of QI project effectiveness are likely to be an inappropriate study method and that statistical process control FIGURE 3. Improvements in Early Testing, Timely Return of Test Results, and Antiretroviral Initiation From Baseline to Endline After Implementing a Quality Improvement Collaborative Approach, Cameroon and Zambia Improving Services for HIV-Exposed Infants in Zambia and Cameroon www.ghspjournal.org Global Health: Science and Practice 2021 | Volume 9 | Number 2 methods such as the use of run charts are the preferred approach to determining project success. 41,42 Additional limitations of our analysis include the fact that the number of HEIs identified each month was generally quite small, making the use of percentages less informative than it would be with larger samples. It is also possible that some of the changes identified during the intervention were the result of improved data quality rather than improvements in service delivery.
QICs are a relatively resource-intensive intervention, requiring substantial time and effort on the part of HF teams and their district-level mentors. As noted, each project described in this article included stakeholder engagement, training, quarterly in-person learning sessions, and hundreds of supportive supervision visits to HFs over the 12-month project lifespan. This methodology is clearly not appropriate for all quality challenges, but in the context of a high-priority quality shortfall with a substantial know-do gap where other interventions have not succeeded, it is a critically important addition to the health systems toolkit. In the case of EID, for example, the relative cost of a time-limited QIC pales in comparison to the cost of HIV testing, prevention, and treatment services, as well as the cost of low-quality care.

CONCLUSIONS
Despite robust evidence, supportive policies, national guidelines, and widespread training initiatives, the provision of effective testing and treatment services to HEIs has lagged in countries around the world, with dire consequences for the infants of HIV-infected women. The use of QIC methodology can effectively bridge this know-do gap by empowering HCWs to design solutions tailored to their specific settings. The well-established approach used in these projects is resource-intensive, and additional exploration may be warranted to determine if less intensive approaches can be as effective.
Funding: The Cameroon and Zambia projects were supported by the U.S. President's Emergency Plan for AIDS Relief through the U.S. Health Resources and Services Administration Quality Improvement Capacity for Impact Project Award under the terms of Cooperative Agreement no. 1U1NHA285550100. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the funding agency.
Author contributions: GD wrote the manuscript in collaboration with all authors and led the project, conceptual framework, and overall analysis of data. All authors discussed and collaboratively agreed upon the design, results, analysis, and planning of the manuscript. TA, MM, and PS significantly contributed to the writing and reviewing of the manuscript. JNB, MP, TPB, RB, MAB, ME, LH, SJ, ZFK, PK, RL, MMM, FT, LW, and MR provided critical revision and final approval of the version to be published. MP designed the concept of the project from Zambia and led project implementation. TA led project concept development in Cameroon, QI activities with the collaborating sites, and final change intervention analysis. JPA led the project data collection in Cameroon and significantly contributed to the analysis with interpretation for writing and reviewing the manuscript. JNB designed the concept of the project from the Zambia Ministry of Health. TPB designed the concept of the project from the Cameroon Ministry of Health. RB designed the concept of the project from the headquarter level. MAB designed the concept of the project from Zambia. ME designed the concept of the project from Cameroon. LH designed the concept of the project from Zambia. SJ designed the concept of the project from the headquarter level. ZFK designed the concept of the project from the Cameroon Ministry of Health. PK designed the concept of the project from Zambia. SCK led the project data collection in Zambia and led the analysis and significantly contributed to the interpretation for writing and reviewing the manuscript. RL designed the concept of the project from the Zambia Ministry of Health. CMM led the design of the monitoring and evaluation strategy from the headquarter level, data collection tools, information system development, data analysis, and interpretation. MMM designed the concept of the project from Cameroon. MM led the design of the monitoring and evaluation strategy from Zambia, data collection tools, information system development, and data analysis. PS led project data analysis and interpretation from the regional level. FT and LW designed the concept of the project from the headquarter level. MR conceived the original project concept and design, overall project leadership, and supervision.