HPV Vaccination Programs in Low and Middle-Income Countries
HPV Vaccination Programs in Low and Middle-Income Countries
This study provides insight into factors that impact the outcome of HPV vaccine campaigns in lowest-income countries. Among the key findings of this study are: first, high VUR (mean VUR = 88.7%), and VA D3-D1 (mean VA = 90.8%) were found across the 21 programs assessed; second, school-based vaccine delivery models and program management by an NGO each had a positive and statistically significant impact on VUR; third, duration of the vaccination campaign appears also as a predictive factor for VUR and VA, with increased speed of vaccination predicting higher VUR and VA.
Program management by NGOs was significantly associated with VUR. NGOs are typically smaller and face fewer internal bureaucratic hurdles than a Ministry of Health. Consequently, the finding that NGO management of a program could predict better VUR compared with MoH management may be related to the relative speed with which each type of organization can implement and execute a program, rather than reflecting overall capabilities or competencies of either type of institution. It should also be noted that programs implemented by NGOs have benefited from the authorization and the support of the national health authorities of the country.
Our results found also that program duration is a significant predictor of VUR (p = 0.03) and shows an association with VA (p = 0.07). Shorter duration predicted improved vaccination indicators, suggesting that a longer interval between vaccine shipment and initiation of the third vaccine dose reflects difficulty in vaccinating girls against HPV in a timely manner. A lack of momentum within a given program may lead to loss of interest in completing all three doses of vaccine among girls and their parents. Similarly, a delay in completing all three doses of vaccine may reflect logistical difficulties related to vaccine supplies or personnel. While speed of vaccination has not typically been included in standard efficacy metrics for vaccine programs, these data could suggest that it could be as an aggregate indicator of a well-run program. This result should be confirmed in further studies, and we hope that other investigators will include speed of vaccination in their future studies in order to determine if this metric is a reproducible and reliable predictive factor for program performance. It should be noted that specific program duration was not a requirement imposed on Program grantees by the GAP.
Of the three types of vaccine delivery models assessed in this study, the school-based model was found to be the strongest positive predictive factor of higher VUR. School-based programs likely have the strongest impact on VUR; daily attendance of target girls at school allows them to be vaccinated more quickly than might occur at a health clinic that requires the girl to make a special trip. Given that this finding is based on data from a relatively large number of programs, we believe that this result supports the use of school-based models as a way to optimize delivery of HPV vaccines to school-aged girls. School-based delivery models may be more effective at delivering HPV vaccine because girls between the ages of 9 and 13 years, the target population recommended by the WHO for HPV vaccination, are likely to be present at school in the 12 programs.
The association between school-based models and increased VUR is supported by other studies that demonstrate the relevance of such models in delivering HPV vaccine. A high rate of HPV vaccine coverage (93.2% after three doses) was achieved in Rwanda using a school model. One study found that school-based models have also been effective in demonstration programs in Peru, Uganda, Viet Nam, and India (vaccine coverage ranged from 82.6% to 96.1%), That report suggests that schools may be especially well-suited to reaching girls in the targeted HPV vaccine age range. Additional studies also support the effectiveness of school-based HPV vaccine delivery programs in developing countries. Similarly, a study in Brazil found VURs of 87.5%, 86.3%, and 85.0% at D1, D2 and D3 time respectively, in school-based vaccination delivery programs. In that study, no significant differences in VUR were observed between public and private schools or urban and rural schools.
Other studies indicate that school-based models may be more effective than age-based models. With 84% of children in developing world attending primary school in 2006, school-based models should be considered as an effective method for delivering HPV vaccine to target populations. However, as previously reported, school-based programs may face obstacles if vaccination dates occur on non-school days, and clinic and/or door-to-door follow up may be required in order to complete the three-dose vaccine series. Within the scope of GAP, girls moving away during school breaks or between school years were reported as a key factor that increased the number of girls lost to follow-up. In order to address this issue, it is important that the full vaccination course is administered during the academic school year. This again advocates for the need to carefully monitor the duration of the campaign and to ensure that girls receive all three doses within the recommended vaccine administration schedule that runs for a period of six months. We should also note that while an earlier analysis of only 8 programs participating in GAP found that mixed models were more effective than school-based models, the current analysis contains a larger number of programs, includes a higher number of vaccinated girls, and also utilizes more sophisticated statistical analyses.
Community involvement actions appeared to impact the efficacy of the programs, especially the VUR. Programs in which communities were engaged in following-up with girls participating in the vaccination campaign had a higher mean VUR compared with programs that did not engage the community in this activity. This action had no effect on VA, suggesting that enrollment in the vaccine campaign is a bigger hurdle than getting girls who have enrolled to complete the three vaccine doses. There was an interesting trend toward increased VUR with an increase in the number of community involvement actions, suggesting a dose-effect response. However, this trend did not reach statistical significance, which may indicate that the type of activity in which communities are involved may be more important than the overall level of community engagement.
Understanding which aspects of an HPV vaccine campaign are most influenced by community standards, morals, and expectations may help in developing community engagement actions that impact VUR and VA. Community sensitization about the availability and value of vaccinating school-aged girls against HPV may impact vaccine uptake. A study in Brazil found that the method used initially to notify parents about the vaccine had a significant impact on vaccine indicators. That study found that information disseminated by schools was more important than information provided by local media with respect to vaccine uptake.
The inclusion of key messages regarding the safety and efficacy of the vaccine had a positive impact on VUR. This finding again suggests that initial enrollment is a greater barrier to completing the three doses of vaccine than is the need to return for repeated vaccine administrations. Key messages that address safety and efficacy at the launch of a vaccine campaign may help to increase enrollment by easing safety concerns and educating girls and their parents about the potential benefits of protection against the long-term consequences of HPV infection. The relatively high rate of VA D3-D1 (90.8%) found in the 21 programs suggests that most girls who participate in the first dose of the campaign remain in the program and complete the vaccine series. These results are consistent with those of a study conducted in Brazil, which found a 97.2% three-dose adherence rate in a school-based HPV vaccine delivery model. This finding is important for those countries interested in implementing an HPV vaccination program to consider. Given the availability of census data from the Ministry of Education, and the high density of targeted girls within school populations, school-based delivery models appear to be the optimum approach for reaching girls eligible for vaccination against HPV. Significant effort should be made at the start of any HPV vaccine program to identify all eligible girls and determine the most effective way to include them in the vaccine campaign.
For low and middle-income countries, it is a challenge to conduct effective cost analyses around HPV vaccination. Many countries adapt existing cost models, but these have been developed for high-income countries and may not be relevant. Model predictions using six different vaccination models suggest that vaccination can reduce the incidence of HPV infection and cervical cancer in a cost-effective manner. In these models, factors that influence cost effectiveness are discount rate, duration of vaccine protection, vaccine price, and HPV prevalence. In an effort to provide relevant data that could be used in designing HPV vaccine programs for low- and middle-income countries, we gathered cost data for seven of the GAP programs.
The mean cost per FIG and vaccine dose across the seven programs was estimated to be USD 8.80 and USD 2.70, respectively. Due to the small number programs with cost data available for this study, it was not possible to detect a significant difference in cost per FIG or per vaccine dose among the three models of vaccine delivery in this analysis. However, a study conducted in Tanzania found that the cost per FIG was lower for class-based delivery models compared with age-based delivery in a large schools-based delivery programs including a total of 134 primary schools. Another study in Tanzania found that implementation of a nationwide HPV vaccine program was associated with significant non-vaccine costs, such as financing of pre-introduction activities, development of new delivery infrastructure, and the deployment of new human resources or reallocation of existing personnel. A study that was conducted in Peru, Uganda and Viet Nam and included five HPV vaccination projects found that the cost per vaccine was lower when vaccine delivery was integrated into health services compared with school-based and integrated outreach. Given the importance of building and enhancing the infrastructure for delivering health care to pre-adolescents, investments in non-vaccine costs related to HPV vaccination campaigns could be amortized over a broader array of health services that deliver a variety of interventions to the HPV vaccine target population.
While our cost data include staffing costs for the programs assessed, they do not provide more detailed information on other non-vaccine costs categories that could substantively impact the implementation or sustainability of the programs. A review of the available data from low- and middle-income countries found that HPV vaccination is cost-effective and potentially cost saving, especially in settings without organized cervical screening programs.
Our study has several limitations. First, the 21 programs used various sources of census data that may have been potentially inaccurate to calculate their target populations, which could impact VUR results. Our finding of VUR greater than 100% in three programs may indicate an under-estimation of the target population and/or recruitment of girls from outside of the original target area, suggesting that the methodologies used for determining the target population may be suboptimal. Second, because 21 programs were included in the study, a potential lack of statistical power could be discussed, particularly for the analysis between the sensitization methods used in different programs and VUR and VA results. Consequently, the results from a given model may not accurately reflect the potential performance of that model over an entire population and larger vaccination programs. Third, we have defined program duration as the time from the date of shipment of first vaccine dose to the date of delivery of the third vaccine dose. Although this may not be the most precise measurement of program duration, the definition that we used yielded interesting results and warrants further evaluation. Despite these limitations, this study provides important insight into factors that impact VUR and VA in a very large sample of HPV vaccine programs reaching more than 217,000 girls within a broad range of low and middle-income country contexts.
Discussion
This study provides insight into factors that impact the outcome of HPV vaccine campaigns in lowest-income countries. Among the key findings of this study are: first, high VUR (mean VUR = 88.7%), and VA D3-D1 (mean VA = 90.8%) were found across the 21 programs assessed; second, school-based vaccine delivery models and program management by an NGO each had a positive and statistically significant impact on VUR; third, duration of the vaccination campaign appears also as a predictive factor for VUR and VA, with increased speed of vaccination predicting higher VUR and VA.
Program management by NGOs was significantly associated with VUR. NGOs are typically smaller and face fewer internal bureaucratic hurdles than a Ministry of Health. Consequently, the finding that NGO management of a program could predict better VUR compared with MoH management may be related to the relative speed with which each type of organization can implement and execute a program, rather than reflecting overall capabilities or competencies of either type of institution. It should also be noted that programs implemented by NGOs have benefited from the authorization and the support of the national health authorities of the country.
Our results found also that program duration is a significant predictor of VUR (p = 0.03) and shows an association with VA (p = 0.07). Shorter duration predicted improved vaccination indicators, suggesting that a longer interval between vaccine shipment and initiation of the third vaccine dose reflects difficulty in vaccinating girls against HPV in a timely manner. A lack of momentum within a given program may lead to loss of interest in completing all three doses of vaccine among girls and their parents. Similarly, a delay in completing all three doses of vaccine may reflect logistical difficulties related to vaccine supplies or personnel. While speed of vaccination has not typically been included in standard efficacy metrics for vaccine programs, these data could suggest that it could be as an aggregate indicator of a well-run program. This result should be confirmed in further studies, and we hope that other investigators will include speed of vaccination in their future studies in order to determine if this metric is a reproducible and reliable predictive factor for program performance. It should be noted that specific program duration was not a requirement imposed on Program grantees by the GAP.
Of the three types of vaccine delivery models assessed in this study, the school-based model was found to be the strongest positive predictive factor of higher VUR. School-based programs likely have the strongest impact on VUR; daily attendance of target girls at school allows them to be vaccinated more quickly than might occur at a health clinic that requires the girl to make a special trip. Given that this finding is based on data from a relatively large number of programs, we believe that this result supports the use of school-based models as a way to optimize delivery of HPV vaccines to school-aged girls. School-based delivery models may be more effective at delivering HPV vaccine because girls between the ages of 9 and 13 years, the target population recommended by the WHO for HPV vaccination, are likely to be present at school in the 12 programs.
The association between school-based models and increased VUR is supported by other studies that demonstrate the relevance of such models in delivering HPV vaccine. A high rate of HPV vaccine coverage (93.2% after three doses) was achieved in Rwanda using a school model. One study found that school-based models have also been effective in demonstration programs in Peru, Uganda, Viet Nam, and India (vaccine coverage ranged from 82.6% to 96.1%), That report suggests that schools may be especially well-suited to reaching girls in the targeted HPV vaccine age range. Additional studies also support the effectiveness of school-based HPV vaccine delivery programs in developing countries. Similarly, a study in Brazil found VURs of 87.5%, 86.3%, and 85.0% at D1, D2 and D3 time respectively, in school-based vaccination delivery programs. In that study, no significant differences in VUR were observed between public and private schools or urban and rural schools.
Other studies indicate that school-based models may be more effective than age-based models. With 84% of children in developing world attending primary school in 2006, school-based models should be considered as an effective method for delivering HPV vaccine to target populations. However, as previously reported, school-based programs may face obstacles if vaccination dates occur on non-school days, and clinic and/or door-to-door follow up may be required in order to complete the three-dose vaccine series. Within the scope of GAP, girls moving away during school breaks or between school years were reported as a key factor that increased the number of girls lost to follow-up. In order to address this issue, it is important that the full vaccination course is administered during the academic school year. This again advocates for the need to carefully monitor the duration of the campaign and to ensure that girls receive all three doses within the recommended vaccine administration schedule that runs for a period of six months. We should also note that while an earlier analysis of only 8 programs participating in GAP found that mixed models were more effective than school-based models, the current analysis contains a larger number of programs, includes a higher number of vaccinated girls, and also utilizes more sophisticated statistical analyses.
Community involvement actions appeared to impact the efficacy of the programs, especially the VUR. Programs in which communities were engaged in following-up with girls participating in the vaccination campaign had a higher mean VUR compared with programs that did not engage the community in this activity. This action had no effect on VA, suggesting that enrollment in the vaccine campaign is a bigger hurdle than getting girls who have enrolled to complete the three vaccine doses. There was an interesting trend toward increased VUR with an increase in the number of community involvement actions, suggesting a dose-effect response. However, this trend did not reach statistical significance, which may indicate that the type of activity in which communities are involved may be more important than the overall level of community engagement.
Understanding which aspects of an HPV vaccine campaign are most influenced by community standards, morals, and expectations may help in developing community engagement actions that impact VUR and VA. Community sensitization about the availability and value of vaccinating school-aged girls against HPV may impact vaccine uptake. A study in Brazil found that the method used initially to notify parents about the vaccine had a significant impact on vaccine indicators. That study found that information disseminated by schools was more important than information provided by local media with respect to vaccine uptake.
The inclusion of key messages regarding the safety and efficacy of the vaccine had a positive impact on VUR. This finding again suggests that initial enrollment is a greater barrier to completing the three doses of vaccine than is the need to return for repeated vaccine administrations. Key messages that address safety and efficacy at the launch of a vaccine campaign may help to increase enrollment by easing safety concerns and educating girls and their parents about the potential benefits of protection against the long-term consequences of HPV infection. The relatively high rate of VA D3-D1 (90.8%) found in the 21 programs suggests that most girls who participate in the first dose of the campaign remain in the program and complete the vaccine series. These results are consistent with those of a study conducted in Brazil, which found a 97.2% three-dose adherence rate in a school-based HPV vaccine delivery model. This finding is important for those countries interested in implementing an HPV vaccination program to consider. Given the availability of census data from the Ministry of Education, and the high density of targeted girls within school populations, school-based delivery models appear to be the optimum approach for reaching girls eligible for vaccination against HPV. Significant effort should be made at the start of any HPV vaccine program to identify all eligible girls and determine the most effective way to include them in the vaccine campaign.
For low and middle-income countries, it is a challenge to conduct effective cost analyses around HPV vaccination. Many countries adapt existing cost models, but these have been developed for high-income countries and may not be relevant. Model predictions using six different vaccination models suggest that vaccination can reduce the incidence of HPV infection and cervical cancer in a cost-effective manner. In these models, factors that influence cost effectiveness are discount rate, duration of vaccine protection, vaccine price, and HPV prevalence. In an effort to provide relevant data that could be used in designing HPV vaccine programs for low- and middle-income countries, we gathered cost data for seven of the GAP programs.
The mean cost per FIG and vaccine dose across the seven programs was estimated to be USD 8.80 and USD 2.70, respectively. Due to the small number programs with cost data available for this study, it was not possible to detect a significant difference in cost per FIG or per vaccine dose among the three models of vaccine delivery in this analysis. However, a study conducted in Tanzania found that the cost per FIG was lower for class-based delivery models compared with age-based delivery in a large schools-based delivery programs including a total of 134 primary schools. Another study in Tanzania found that implementation of a nationwide HPV vaccine program was associated with significant non-vaccine costs, such as financing of pre-introduction activities, development of new delivery infrastructure, and the deployment of new human resources or reallocation of existing personnel. A study that was conducted in Peru, Uganda and Viet Nam and included five HPV vaccination projects found that the cost per vaccine was lower when vaccine delivery was integrated into health services compared with school-based and integrated outreach. Given the importance of building and enhancing the infrastructure for delivering health care to pre-adolescents, investments in non-vaccine costs related to HPV vaccination campaigns could be amortized over a broader array of health services that deliver a variety of interventions to the HPV vaccine target population.
While our cost data include staffing costs for the programs assessed, they do not provide more detailed information on other non-vaccine costs categories that could substantively impact the implementation or sustainability of the programs. A review of the available data from low- and middle-income countries found that HPV vaccination is cost-effective and potentially cost saving, especially in settings without organized cervical screening programs.
Our study has several limitations. First, the 21 programs used various sources of census data that may have been potentially inaccurate to calculate their target populations, which could impact VUR results. Our finding of VUR greater than 100% in three programs may indicate an under-estimation of the target population and/or recruitment of girls from outside of the original target area, suggesting that the methodologies used for determining the target population may be suboptimal. Second, because 21 programs were included in the study, a potential lack of statistical power could be discussed, particularly for the analysis between the sensitization methods used in different programs and VUR and VA results. Consequently, the results from a given model may not accurately reflect the potential performance of that model over an entire population and larger vaccination programs. Third, we have defined program duration as the time from the date of shipment of first vaccine dose to the date of delivery of the third vaccine dose. Although this may not be the most precise measurement of program duration, the definition that we used yielded interesting results and warrants further evaluation. Despite these limitations, this study provides important insight into factors that impact VUR and VA in a very large sample of HPV vaccine programs reaching more than 217,000 girls within a broad range of low and middle-income country contexts.