Human Papillomavirus Vaccine Delivery in Mozambique: Identification of Implementation Performance Drivers Using the Consolidated Framework for Implementation Research (CFIR)

Implementation research is concerned with the uptake of health interventions in different settings and the identification of context-specific implementation success barriers and facilitators. This study provides implementation findings from Mozambique's human papillomavirus (HPV) 2-year vaccine pilots in an effort to inform and, hopefully, expedite, national scale-up. Mozambique was chosen because the country conducted demonstration projects in 3 different districts with extremely different economic resources and sociocultural practices.

With the aim of standardising terminologies and approaches for identifying and documenting implementation determinants, the Consolidated Framework for Implementation Research (CFIR) was published in 2009. The CFIR is organised into 5 domains (intervention characteristics, outer setting, inner setting, characteristics of individuals and process) that are subdivided into 39 constructs. To date, to the knowledge of the authors of this paper, the CFIR has not been used to study implementation of vaccination programmes in low- and middle-income countries (LMICs), despite several studies employing it to explore HPV vaccine delivery in the United States (US). Thus, "The goal of this paper is to show the utility of CFIR in identifying and documenting implementation barriers and facilitators for the scale-up of interventions in LMIC health systems, such as that of Mozambique."

A post-implementation interpretive evaluation was carried out in 5 steps:

1. Defining the innovation: The innovation of interest was the vaccine delivery model rather than the vaccine itself. The country chose to explore a periodic, school-based vaccine delivery model, on the premise that most girls aged 9-14 years would be found in schools.
2. Defining the unit of analysis, site inclusion, and performance criteria: There were 3 district pilot sites, and the researchers decided to collect data from all of them. They classified the sites as high, intermediate, or low performing based on HPV vaccination coverage achieved during demonstration project implementation.
3. Initial CFIR construct selection and data collection: The researchers developed a semi structured interview guide to capture information on the selected CFIR constructs and used it to conduct 40 key informant interviews (KIIs) at the Ministry of Health (MOH) central level and all 3 demonstration districts. KIs included district and health facility immunisation staff, Ministry of Education managers, and teachers, central-level informants from MOH, research institutes, and immunisation programme partners.
4. Final CFIR construct selection: The final constructs for analysis were selected through the coding process.
5. CFIR construct valence and strength rating: The researchers compared valence and strength ratings of CFIR constructs, across diverse implementation sites, so as to explain drivers and barriers to implementation success.

Eighteen constructs emerged from KIs' responses as implementation influencers. The study revealed 8 significant drivers of implementation success or failure that either distinguished or influenced implementation performance. They are: adaptability, complexity, financial resources, organisational culture and workers' beliefs about the innovation, training, intervention recipients' perceptions, engaging the right opinion leaders, and decentralisation of planning processes. Details about some of the strongly distinguishing constructs related to communication include:

• With regard to a construct that describes the extent to which vaccine recipients' needs are accurately known and prioritised, as well as the barriers to reaching them: The low- and intermediate-performing districts' health worker respondents reported a failure of the delivery model design to address some elements that subsequently compromised their ability to reach all eligible girls in their districts. For instance, while health workers were cognisant that these girls were in the community, not knowing their precise location became a barrier to reaching them.
• Available resources: This could be seen in social mobilisation activities, which clearly demonstrated effects of disparity in resource availability. The high-performing site included an extra component, beyond regular methods implemented in all demonstration districts. Regular methods included both television and radio spots and health worker and community leader discussions, while in the higher-resourced district, engagement of community volunteer activists, who conducted door to door educational visits, was an additional strategy. The activists received remuneration for their work; thus, a similar strategy could not be adapted in the less resourced districts where funds were scarce or unavailable.
• Organisational incentives and rewards: The lack of timely financial support in the low-performing district resulted in negative repercussions for HPV implementation and acceptance. Disgruntled community leaders lost trust in the local health directorate, who they believed had received money from the central MOH level to pay for community leader support but were refusing to disburse it. Consequently, they not only failed to complete tasks they had agreed on (pre-vaccination registration of girls) but went further and used their important positions as opinion leaders to discourage parents from allowing their daughters to be vaccinated, further diminishing implementation effectiveness.
• Access to knowledge and information: Teachers in the high-performing site received training to execute various activities during preparation and implementation phases of HPV vaccination. They were trained in pre-registration of eligible girls, education for vaccine-eligible girls and their parents about the upcoming HPV campaign, location of appropriate vaccination venues in schools, queuing and management of girls during vaccination, and registration of vaccination in registries and on girls' cards. In addition, teachers were trained in monitoring vaccinated girls for adverse effects following immunisation. Given all the activities that were missed where teachers were not trained, this construct strongly distinguished performance.
• Learning climate: The high-performing district (Gavi-funded) included an evaluation component while the intermediate and low-performing districts did not.
• Opinion leaders: Organisers in the low-performing district inadvertently excluded an important opinion leader group: mosque imams. The high- and intermediate-performing districts did not experience similar problems despite excluding religious leaders from social mobilisation activities, because of the predominance of Christianity.
• Innovation participants: Engagement of innovation participants was compromised by community beliefs in the low-performing district. According to KIs from this district, people did not understand why the vaccine was being offered to only girls instead of all children, as is done with other vaccines familiar to the population.

Organisational culture, which in this study refers to the culture and practices of the Mozambican MOH workers, was a non-distinguishing factor. Nevertheless, it strongly influenced implementation in a positive manner. Health workers at all levels of the Mozambican health system abide by a culture of resilience, whereby making do under challenging circumstances is the norm, including sacrificing when a matter is deemed to be of national priority or interest. Health worker respondents from all sites stated that they went out of their way to ensure the demonstration project was implemented within planned timelines, and they viewed their selection as pioneer HPV vaccination district as an honour.

Another non-distinguishing construct that nonetheless mattered was champions. The HPV vaccine demonstration project implementation in all the sites was positively influenced by the engagement of Mozambique's first lady, who identified cervical cancer as one of her legacy campaigns. Her involvement drove the inclusion of the 2 non-Gavi-funded districts in the demonstration project and facilitated the successful realisation of the added demonstration projects despite limited funding. In addition, there was high public visibility of the vaccine's introduction due to wide media coverage of the launch ceremony that she officiated, a factor that contributed to increased demand generation in all sites.

In short, these findings could provide guidance for HPV vaccination programmes' stakeholders on practices that can be replicated and those that should be avoided during scaling up of HPV vaccine delivery in the country and other LMICs with similar health and socioeconomic settings. For example, the study shows that social mobilisation needs to be informed by evidence-based research while still capturing and valuing recipients' perceptions, beliefs, and attitudes. In addition, the identification of the most influential opinion leaders is key to implementation success. Community members informed health workers in the highly Muslim populated demonstration district that they had not taken their girls for the HPV vaccination because messages had only been relayed via radio but not by their imams in the mosques. Interviewees deemed the period of social mobilisation prior to and throughout implementation to be insufficient.

The researchers conclude by explaining why they found the CFIR to be a useful and practical tool for researching health system implementation success determinants. Finally, they recommend that the MOH garner local and international support to develop strategies that take into account the implementation barriers outlined in this study when planning for countrywide expansion.