Using participatory action research to improve immunization utilization in areas with pockets of unimmunized children in Nigeria

Background: In 2005, Nigeria adopted the Reaching Every Ward strategy to improve vaccination coverage for children 0–23 months of age. By 2015, Ogun state had full coverage (100%) in 12 of its 20 local government areas, but eight had pockets of unimmunized children, with the highest burden (37%) in Remo North. A participatory action research (PAR) approach was used to facilitate implementation of local solutions to contextual barriers to immunization in Remo North. This article assesses and seeks to explain the outcomes of the PAR implemented in Remo North to understand whether and possibly how it improved immunization utilization. Methods: The PAR intervention took place from 2016 to 2017. It involved two (4-month) cycles of dialogue and action between community members, frontline health workers and local government officials in two wards of Remo North, facilitated by the research team. The PAR was assessed using a pre/post-intervention-only design with mixed methods. These included household surveys of caregivers of 215 and 213 children, respectively, 25 semi-structured interviews with stakeholders involved in immunization service delivery and 16 focus group discussions with community members. Data were analysed using the Strategic Advisory Group of Experts (SAGE) vaccine hesitancy framework. Results: Collaboration among the three stakeholder groups enabled the development and implementation of solutions to identified problems related to access to and use of immunization services. At endline, assessment by card for children older than 9 months revealed a significant increase in those fully immunized, from 60.7% at baseline to 90.9% (p <.05). A significantly greater number of caregivers visited fixed government health facilities for routine immunization at endline (83.2%) than at baseline (54.2%) (p <.05). The reasons reported by caregivers for improved utilization of routine immunization services were increased community mobilization activities and improved responsiveness of the health workers. Spillover effects into maternal health services enhanced the use of immunization services by caregivers. Spontaneous scale-up of actions occurred across Remo North due to the involvement of local government officials. Conclusion: The PAR approach achieved contextual solutions to problems identified by communities. Collection and integration of evidence into discussions/dialogues with stakeholders can lead to change. Leveraging existing structures and resources enhanced effectiveness. In this section we describe the study design, sampling and recruitment of the respondents, and the conceptual framework used in data analysis. We used a pre-test/post-intervention only approach to evaluate the outcomes of the PAR. Baseline (situational analysis) in May 2016 and endline assessments (in April 2017) were carried out using mixed methods comprising a household survey (HHS), secondary data analysis of the NHMIS, focus group discussions (FGD) and semi-structured interviews (SSI). We used concurrent mixed methods designs at baseline and endline—the quantitative and qualitative data were collected in parallel, within the same time frame. Integration was carried out during data analysis and interpretation of results. The qualitative interviews were used to explain the results of the survey and to gain more insight into contextual factors. Quantitative methods included a survey at the household level targeting caregivers responsible for the vaccination of at least one under-five child, and secondary analysis of NHMIS data to track immunization coverage. Qualitative methods included FGDs with community men and women—used to explore the uptake of the intervention by the communities and changes in knowledge, attitudes and utilization of immunization. This was triangulated with SSIs of key stakeholders. The SSIs helped us explore system challenges and to understand whether there was a match (or mismatch) between community views and the views of other stakeholders. Baseline and endline data collection was carried out by a team of two quantitative and two qualitative researchers, 14 enumerators and eight qualitative research assistants. Remo North was purposively selected for this study because of the burden of unimmunized children. Two focal wards were selected. These were Ipara and Ilara, with high and low immunization coverage, respectively, according to the 2015 NHMIS data. We wanted to determine the range of facilitators and barriers to immunization, and whether there were differences among the sites which could explain the outcomes. In terms of characteristics, Ilara is essentially a remote and rural farm settlement, while Ipara is semirural, having more commercial activity and a more organized structure with numbered streets. Enumeration of households for the survey was conducted by officials of the National Population Commission. This exercise identified houses with children under 5 years of age, who were the focus of enquiry. The HHS sampling was conducted using the World Health Organization (WHO) modified two-stage cluster sampling method [22]. Using probability proportional to size techniques, we identified the clusters for the study. Thirty clusters were selected across the two wards—12 in Ilara and 18 in Ipara. To identify households, in each cluster an arbitrary but central starting point was identified. Consecutive houses along this path were visited to identify households eligible for inclusion. One under-five child was selected from households in seven consecutive homes. Where more than one eligible child was present in a household, one was selected using a table of randomly generated numbers. All eligible children were selected in the seventh household of each cluster, as required by this method. The respondents in this study were caregivers of under-five children in the selected wards. Individuals were eligible if they were currently domiciled in the ward. Information was obtained primarily from the mother/primary caregiver. Interviews took between 25 and 40 minutes to complete. Most interviews were conducted in Yoruba. The study collected data from 210 adults relating to 215 children at baseline and from 210 adults relating to 213 children at endline. These were different sampled populations. Primary qualitative data were also collected at baseline and endline using topic guides. Sixteen FGDs (8 per ward) were carried out in each period. Respondents were community members (young women/men and older women/men), and usually 6–8 in a group. Adults who were caregivers or involved in the immunization decision-making relating to a child were included in the FGDs. Research assistants recruited participants with the help of community mobilizers. A total of 25 key informants consisting of frontline health workers, policy-makers, local government implementers, religious and traditional leaders, and WDC and SMC members were recruited for SSIs at baseline and endline using purposive sampling. Additionally, SSIs were conducted with the 24 JAC members at endline. These were carried out to determine whether the PAR approach worked in the context and with the planned implementation structures and processes. The SSIs and FGDs lasted about 60 and 90 minutes, respectively, and were audio-recorded with the respondents’ consent. We used the theoretical framework for vaccine hesitancy by the WHO Strategic Advisory Group of Experts (SAGE) [23] to understand the PAR outcomes and explore the differences between baseline and endline. The model mapped the determinants of vaccine hesitancy in 13 countries and describes attitudes towards vaccination as a continuum ranging from complete acceptance to total refusal. It differentiates between contextual, individual, group and vaccine/vaccination-specific factors that influence immunization acceptance and utilization. We regrouped our study outcomes according to the themes in the hesitancy framework, and only the outcomes that emerged from the study were included in the adapted framework. For instance, we did not include the design of vaccine programme delivery (see Fig. 2). We explored whether immunization utilization had changed and the main drivers of change. Conceptual framework Primary quantitative data were analysed using SPSS [Statistical Package for the Social Sciences] version 21 software. The primary study outcome was immunization completeness. This was assessed as three doses of DPT/pentavalent vaccine as well as measles and yellow fever recorded as administered in an immunization card. To assess the association between covariate factors and immunization coverage, a univariate analysis was carried out for each factor and immunization coverage. All statistically significant factors/variables from the univariate analysis were included in a multivariate logistic regression model. Crude odds ratios were determined for each variable. Primary qualitative data were analysed using NVivo 11. The primary outcomes assessed were changes in access to and utilization of immunization services. An inductive approach and open thematic coding were used. Transcripts were read by two qualitative researchers, coded and common themes and sub-themes identified according to the research objectives. A third qualitative researcher coded a few transcripts in order to ratify the codes and themes/sub-themes identified. The respondent characteristics are described first. Then the outcomes of the PAR are presented according to the conceptual framework. Findings are compared and contrasted between different groups and between Ilara and Ipara wards where possible. HH survey These results profile, at endline, 213 children and their caregivers (210) studied across 210 households in the study area. Half of the sampled children were older than 2 years of age and almost 51% were male. Similarly, at baseline (a different sampled population), 215 children and their 210 caregivers were studied across 210 households in the study area. Most of the caregivers were Yoruba (89% at baseline and 83% at endline)—this reflected the general population ratio between indigenous people and migrants. Table ​Table22 details a comparison of gender, religious affiliation and socioeconomic characteristics including educational level of the respondents in both wards—in the baseline and endline surveys—and shows that the samples are comparable. Respondent characteristics—household survey Values in the same row not sharing the same subscript are significantly different at p < 0.05 1Tests are adjusted for all pairwise comparisons (within a row) using the Bonferroni correction FGDs The characteristics of the FGD participants at endline were similar to those at baseline. The participants were separated into different groups based on age and gender—the young men and women groups consisted of participants who were aged 18 to 39 years. The older women and men groups consisted of participants aged 40 to 65, with an outlier of 73. There were more Christians than Muslims. All the participants but one (a widow) were married. SSI At endline, a total of 24 PAR participants (12 in each ward) were interviewed in Ilara and Ipara. Sixteen out of these 24 JAC members interviewed were female. Interviews were also carried out with 25 key informants—policy-makers, local government officials, health workers and key community stakeholders involved in immunization service delivery. Six health workers were interviewed at endline compared with four at baseline, due to deployments that had taken place during the past year. Two WDC members and two religious leaders (an imam and pastor) were interviewed in each ward. The foremost traditional leaders in the wards (Kabiyesi in Ilara and Baale in Ipara) were interviewed as well. The outcomes of the PAR are presented by addressing the main question first—did immunization utilization change? Then we examine how and why, using the conceptual framework. We present the quantitative findings first, then the qualitative. Where there are differences between Ilara and Ipara, we highlight these. The primary study outcome was immunization completeness—assessed as three doses of DPT/pentavalent vaccine as well as measles and yellow fever recorded as administered on an immunization card. According to the routine vaccination schedule in Nigeria, the final antigens (measles and yellow fever vaccines) are administered at 9 months of age. The analysis of immunization completeness in the HHS encompassed all children between 9 and 59 months who should have plausibly achieved this outcome. Only 56 children (32.6%) over 9 months of age (n = 172) at baseline had immunization cards available for inspection. Availability of cards for assessment improved at endline (and was statistically significant) to 88 (52.4%) of 168 children over 9 months. At endline, assessment by card for children older than 9 months revealed a statistically significant increase from baseline (60.7%) (50% in Ilara, 67.7% in Ipara), to 90.9% (90.6% in Ilara, 91.1% in Ipara) (p < 0.05) of children having received all vaccinations (refer to Table ​Table33 for details). However, when immunization status was assessed by card and recall, 146 (84.9%) of the 172 children over 9 months were assessed as fully immunized at baseline. At endline, similar figures—albeit a bit lower, but not statistically significant—of complete immunization were found, namely 136 (81.0%) of the 168 children older than 9 months. Immunization utilization in the wards, household survey Values in the same row not sharing the same subscript are significantly different at p < 0.05 1Tests are adjusted for all pairwise comparisons within a row using the Bonferroni correction The consensus in the FGDs was that immunization utilization by caregivers in Ipara and Ilara for their children had improved in the past year. Most policy-makers and local government officials commented that the coverage data from Remo North now showed fewer red and yellow indicators, indicating that the number of unimmunized children was declining and immunization-seeking behaviour had increased. The NHMIS categorizes access and utilization of immunization with numbers and colour codes ranging from 1 (deep green) for good access/utilization to 4 (red) for poor access/utilization. Ilara has moved from category 4 to 2 now on routine immunization (RI). It’s very encouraging. —LGA official 2– PAR participant, Ilara Caregivers in the HHS were asked about their most recent immunization visit (see Additional file 1). Significantly more caregivers visited fixed government health facilities for immunization services at endline (83.2%) than at baseline (54.2%) (p < 0.05). Also, a significantly higher proportion of caregivers in Ipara (88.7%) accessed routine immunization at fixed government facilities than in Ilara (75%) (p < 0.05) at endline. Interestingly, there was higher utilization of mobile or outreach services at baseline (34.8%) than at endline (10.6%). In terms of indirect costs, significantly more caregivers in the HHS were of the opinion that services were much cheaper (38.1%) at endline than at baseline (16.2%). Intensified efforts on community mobilization with the JAC/WDC members were highlighted during the FGDs, and healthcare professionals were described as more motivated in carrying out community mobilization. This appeared to have encouraged greater facility use. According to the young women in both wards, health workers had become more responsive to the communities’ immunization needs during the past year—immunization was carried out on time, and greater availability of health workers and vaccines was noted. According to an LGA official: …In the past, community members always complained of the attitude of the health workers—that they were too harsh and not accommodating. It’s not like that now. —LGA official 1– PAR participant, Ipara Some of the respondents in the FGDs also attributed improvements in immunization utilization to improved relationships between the community members and health workers which had resulted from the dialogues and action. Improved availability of vaccines for routine immunization was frequently mentioned in the discussion groups and by a majority of the SSI respondents in both wards; however, the perception of indirect costs of immunization remained the same at endline. Young women in Ilara and Ipara frequently reported that they still contributed 100 naira (approximately US$ 0.30) towards the transportation cost of the vaccines from the LGA headquarters to the wards. There was evidence of knowledge and awareness of immunization and its value. In the HHS, at baseline, similar to endline, the majority (95.7% and 96.1%) of caregivers stated that immunization prevents diseases, with polio and measles being the vaccine-preventable diseases that they were most aware of. Health facilities were the predominant sources of information on child health (91%), similar to baseline (91%). Provision of information on immunization was reported to be the most important function of the WDC by a little over 40% of respondents. The survey findings were supported by the findings in the FGDs. A notable difference at endline was that the young women groups in both wards spoke more knowledgeably about immunization and contributed more to the discussions than at baseline. Young men and women groups in both wards also reported adverse events following immunization (AEFI) as an important reason why some people refused to take immunization for their children. The SMC was adjudged to be the most active in immunization via mobilization of the communities. At endline, there were more frequent reports from the young women in both wards about passing on information about immunization to their neighbours. Leaders of the non-indigenous groups were reported to provide information on immunization to their groups in both wards. This information usually related to the dates and times of immunization, the value of immunization and information on AEFI. The content of the information was provided by the health workers, and the language barrier was overcome by the use of these mediators. SSI respondents frequently reported improved attitudes of caregivers towards immunization, and this was also the general view in the FGDs. An important reason given for this improved attitude in both wards was reduced fear regarding AEFI. Participants reported that this was due to intensified health talks on AEFI given to mothers during facility visits and outreaches in the past year. JAC members in Ilara also stated that home visits by health workers, especially in the course of tracking defaulters, provided opportunities for the husbands to be educated on AEFI. Nevertheless, some young women in both wards still commented that AEFI was distressing and discouraging: Respondent (R)1: Going for immunization doesn’t take anything. It’s just the issues that arise after. Like the sleepless nights. Not being able to sleep till morning (because of children crying from the pain at the immunization sites or fever). R3: Truly, immunization is good for children. The only issue is that the arm injected gets swollen and is filled with pus. Why is that? —Young women, Ilara Several key achievements relating to the overall health services in the past year were reported by PAR participants and FGD groups in Ilara. They included renovation of the health facility and reinstitution of antenatal care and delivery services. These were credited as the main reason for improved utilization of health and immunization services by caregivers. In Ipara, the provision of water supply in the health facility and delivery services for women in their first pregnancies were major achievements reported by the participants. Changes in the health services reported by SSI respondents and during the FGD are summarized in Table ​Table44. Perceived changes in health services Formerly, we only have one staff available here, and after two weeks, we won’t see the staff again and the facility will be locked, but now thing aren’t like that anymore. Whenever you come it is either you meet one or two or three persons on duty —Young woman, Ilara One woman in my house fell sick around 1.00 am; they took her to the health centre and they attended to her. If it was before, as at last year, it was not like that; but we thank God for the relationship between the joint action committee and the community, it brings about good results. —Older man, Ilara R: Health workers are not enough. Where there are supposed to be three people doing a job, we find only one person I: Do you always meet the ones available on ground? R: Yes, but they are not enough —Older women, Ipara The issue of water is already solved. We have been able to connect water to the health facility, and with that, health workers are happier and mothers are happier to know that they would not have to bring buckets of water to the facility —Chairman JAC, Ipara Multivariate logistic regression was performed for children over the age of 9 months to identify factors associated with completion of immunization based on assessment of cards and recall. Statistically significant factors were location (ward) and caregiver education (see Table ​Table5).5). The likelihood of complete immunization for children older than 9 months in Ipara was 2.72 (CI 1.45–5.11, p = 0.002) compared with children in Ilara. Caregivers with a higher level of education were 5.09 times (CI 1.32–19.62, p = 0.018) as likely to fully immunize their children as uneducated caregivers. This trend continued: caregivers with secondary and primary education were respectively four times (CI 1.66–9.64, p = 0.002) and 2.93 times (CI 1.19–7.24, p = 0.02) as likely as uneducated caregivers to fully immunize their children. Multivariate logistic regression results on determinants of complete immunization, household survey *p < 0.05 (indicates statistical significance at a 5% level of significance) The SMC and WDC are historical influences in terms of their collaboration with the immunization sector. The JAC continued with this collaboration in the implementation of the JAPS. Caregiver knowledge regarding who the SMC members were increased significantly at endline, from 20.5% to 48.1%, and their knowledge of WDC members also increased significantly, from 27.1% to 52.4%. Similar figures were reported across wards. Multivariate logistic regression showed that children over 9 months of age were 3.68 times as likely to be fully immunized when caregivers had knowledge of SMC members (CI 1.44–9.46). The involvement of the local government implementers resulted in a spillover of the strategies used in the JAPs into other wards in Remo North. Monitoring data showed that functional WDCs increased from 100 pre-PAR to 164 post-PAR. Correspondingly, immunization coverage in Remo North increased from 66% in 2016 to 86% in 2017, and the proportion of unimmunized children dropped from 30% to 9% (see Fig. 3). While the trend in immunization uptake cannot be attributed solely to the PAR, it appears that the intervention was an important contributor to the acceleration of the upward trend from 2016 to 2017 in Remo North. Trend of immunization uptake in Remo North