https://orcid.org/0000-0002-9742-8091
Figures
Abstract
Zero-dose children remain highly vulnerable to vaccine-preventable diseases and can sustain transmission even in highly vaccinated populations. The WHO Immunization Agenda 2030 has prioritised reaching out to these children. We assessed the spatial distribution of zero-dose children together with the associated risk factors in a provincial capital in the Democratic Republic of Congo. A cross sectional survey was conducted in the city of Kikwit between September 28 and October 14, 2022. Data were collected both at household and health area level. QGIS and SATscan were used to describe and identify hotspots among zero-dose children, and a mixed effect logistic regression model was used to identify risk factors. Overall, 1,863 children aged 12–23 months were enrolled. Kikwit city had a 16.3% zero-dose prevalence, with significant variation between and within health zones. Two hotspots were identified through geospatial analysis, each spanning multiple health areas. Multilevel analysis revealed significant clustering at health area level and found six associated risk factors. These include the absence of home visits by community health workers (aOR = 1.90), living more than a kilometre from a health centre (aOR = 1.95), the mother’s lack of tetanus vaccination (aOR = 3.16), and inability to name a vaccine-preventable disease (aOR = 3.20). However, secondary (aOR = 0.56) or tertiary (aOR = 0.21) education of mothers/guardians and belonging to Bunda (aOR = 0.36) or Mbala (aOR = 0.52) ethnicity reduced the risk of zero-dose. We observed a high prevalence of zero-dose children with a heterogeneous spatial distribution of epidemiological importance. Due to sub-zonal diversity, a health zone approach to reduce zero-dose immunization appears very limited. Zero-dose prevalence was related to the community health workers’ home visit, to the distance of residence to a health centre and to household-level factors. Geospatial results could help in targeting priority health areas and communities for vaccination.
Citation: Mutwadi AM, Madinga JN, Vanlerberghe V, Mbala PK, van der Sande MAB (2024) Geospatial and multilevel clustering of zero-dose children in Kikwit, Democratic Republic of the Congo in 2022. PLOS Glob Public Health 4(2): e0002617. https://doi.org/10.1371/journal.pgph.0002617
Editor: Sheetal Prakash Silal, University of Cape Town, SOUTH AFRICA
Received: August 11, 2023; Accepted: February 3, 2024; Published: February 29, 2024
Copyright: © 2024 Mutwadi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The scripts used for conducting the analysis, the study protocol, and data dictionaries are publicly available on GitHub at the following URL: https://github.com/armandmutwadi. The pseudonymized data that supports the findings of this study is retained at the Institute of Tropical Medicine in Antwerp. Please note that the database contains geolocation data pertaining to a vulnerable population group. This makes it unsuitable for public sharing, even in an anonymized format. Therefore, due to confidentiality and ethical considerations, access to the database will only be granted to external parties after a formal request has been made and subsequently authorized. Interested parties may obtain access to this data upon approval of a motivated and written request submitted to ITM (ITMresearchdataaccess@itg.be).
Funding: This work was partially funded by the Belgian Directorate for Development Cooperation and Humanitarian Aid (DGD) through the multiyear program (2022–2026) for the institutional collaboration between the Institut National de Recherche Biomedicale (INRB, recipient of this funding), DRC, and the Institute of Tropical Medicine (ITM) of Antwerp, Belgium. No personal grants were involved. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Zero-dose (ZD) children are those who have not received a single dose of Diphtheria Tetanus and Pertussis (DTP) vaccine as part of the routine vaccination schedule by the age of 12 months. There were an estimated 12.4 million ZD children in 2020, the majority living in sub-Saharan Africa or other conflict-affected areas [1, 2]. Pre-Covid-19 pandemic investigations indicated that nearly 50% of vaccine preventable deaths occurred among ZD children [3]. The World Health Organisation Immunization Agenda 2030 (IA2030) identified reaching out to ZD children as a strategic priority and has set a target of a 25% reduction in the number of ZD children by 2025 and 50% by 2030 [1, 4].
Global vaccination coverage has significantly increased since the WHO Expanded Programme on Immunization (EPI) in 1974 as a result of routine immunization (RI) and additional targeted campaigns. However, over the past ten years, coverage levels have decreased or stagnated in a number of countries, including the Democratic Republic of the Congo (DRC). Surveys between 2010–2019 in 92 countries recorded prevalences of ZD in children aged 12 to 23 months from 5.2% in upper-middle income countries to 11.1% in low-income countries, for a total pooled prevalence of 7.7% [5].
According to routine data from the DRC national EPI, the coverage of DTP1 vaccine was consistently above 90% in the last 3 years, and even surpassed 100% in 2022 [6]. However, the routine gathering of immunisation data poses challenges and has limitations in precisely assessing vaccination coverage, thereby undermining its dependability for decision-making in low- and middle-income countries (LMIC) [7]. Population-based vaccination surveys are commonly employed to gather coverage rates that accurately represent the actual situation. Based on household vaccination surveys conducted in 2020 and 2021 in DRC, there was a minor decline in DTP1 coverage rates, decreasing from 83.0% to 80.9% correspondingly, and a rise in the prevalence of ZD children from 17.0% to 19.1% [8, 9]. These rates are above the previous estimate for low-income countries. In 2021, it has also been described that there were wide variations between DRC provinces and between health zones. For example, the Mai-Ndombe province had a ZD prevalence of 41.8% while its nearest neighbour, the Kwilu had 10.3% [10].
A health zone (HZ) in DRC is a vast surface with 100,000 to 200,000 inhabitants having a secondary level hospital, and is subdivided in health areas, covering about 10.000 inhabitants having one health centre and various community-based health posts and sites. The health centre is a first-level health care facility with a minimum package of activities, including immunization service, which is provided at the centre and during outreach activities. A health centre serves a population within a maximum radius of action of 8 to 15 kilometres [11]. It is not routinely followed up if vaccination coverage is heterogeneous across health areas. This information could guide catch-up vaccination efforts to reduce ZD prevalence. The major factor contributing to disparities between vaccination coverage and susceptibility to vaccine-preventable diseases (VPDs) is the poor management of routine immunisation (RI) and vaccination campaigns [12, 13]. Spatial data are needed to identify geographically vulnerable populations beyond survey data which are identifying individual and household characteristic -based risk groups. Globally, geospatial analysis are increasingly used to create zonal-level health and demographic maps [14–17].
The aim of this study was to investigate the geospatial distribution and prevalence of ZD children by conducting a survey among children aged 12 to 23 months in all health areas within Kikwit city, DRC. The national EPI is interested in this specific age range due to its relevance because children in this age group should have received all routine vaccines according to the national childhood vaccination schedule for 2022, and for comparability purposes, as numerous studies and systematic reviews have focused on this age group [18–20]. We also set out to assess risk factors related to ZD status among these children at both household and health area levels taking into account the variation between health areas.
Methods
Study design and setting
A cross sectional survey was conducted at the household and health area level between September 28th and October 14th, 2022, in the city of Kikwit, located in the Province of Kwilu at the western part of DRC. The city has two HZ, the North and the South HZ (Fig 1).
Sampling
Household level.
The survey targeted children aged 12 to 23 months at the moment of survey. Respondents were heads of household, and the mother or guardian of infant. Fathers were included in the study to provide the characteristics of household; in case of unavailability of father, this information was collected from the mother. Information on vaccine uptake was taken from the immunization card and if the card was not available, we relied on the mother’s or guardian’s recall. The study was powered to estimate a 50% ZD with 3% precision, taking into account a design effect of 2, and 10% non-response. Two stage cluster sampling was done. All the health areas within each HZ were included in the study. The North HZ has 25 health areas, and South 24. For the two zones (49 health areas = 49 clusters), the total sample size needed was 1764 children (900 children for the North and 864 for the South HZ) which means thirty-six children per cluster (health area, secondary sampling unit) was calculated.
In practical terms, we proceeded as follows:
- ■ First, in each cluster, 6 avenues in the urban setting (or 6 villages for the rural places) were selected by simple random sampling based on the list of avenues/villages provided.
- ■ Then, in each selected avenue/village, 6 households were selected based on a systematic random sampling technique.
Health area level.
In each health area, the main health centre was visited to collect data on the health system, vaccination activities and demographic characteristics of its population. Hence, 49 health centres were included in this study. The respondent at health centre level was the head nurse or his/her deputy. Data were collected with an interviewer-based questionnaire.
Data collection
The data was collected using “Kobocollect” on mobile devices [21]. All data, including GPS coordinates, were transmitted each day to a secure virtual server after verification for accuracy by the field supervisor.
Outcomes
ZD child was defined as a child who had not received a single dose of Diphtheria Tetanus and Pertussis vaccine (DTP vaccine). According to the immunization schedule in DRC (Table 1), the first DTP should be administered at 6 weeks after birth, the second one at 10 weeks and the third, and last one at 14 weeks [22]. Furthermore, data were collected for possible risk factors as reported elsewhere (such as age, sex, migrant status, marital status, ethnicity, religion, mother’s profession, mother’s education, antenatal care attendance, previous maternal vaccination such as tetanus vaccine during pregnancy (maternal recall), name of at least one VPD, father’s profession, community health workers’ home visit, distance between household and the health centre) and at health area, on the availability of a refrigerator in the health facility, vaccine stockout during the last 12 months at health facility level, and accessibility of the health facility [23–32].
Statistical analyses
Completed questionnaire databases were transferred to IBM SPSS statistics version-27 and R studio 2023.03.0 Build 386 software for analysis.
Categorical variables were summarized as frequencies and proportions, continuous variables as mean with standard deviation (SD) or median with interquartile range, depending on the normality of distribution. The distance between household and the main health centre of the health area was computed using geographical coordinates with QGIS. The ZD prevalence was assessed separately by source of information (vaccination card and mother recall), and with both sources pooled.
The association between dependent and independent variables was determined using crude and adjusted odds ratios (OR) with 95% confidence intervals (95% CI). We have applied a Bonferroni correction, considering a p-value of 0.05 divided by 6 (number of comparisons at the final model) as statistically significant. We used two-level mixed-effect logistic regression modelling to simultaneously investigate health area and household level factors associated with the ZD children, allowing between health area variation.
Three multilevel logistic regression models were fitted. Model 1 was an empty model that contained only health area-specific random effects to model between health area variation with ZD children. Model 2 included the health area-specific random effects and the respondent characteristics, Model 3 included health area-specific random effects, respondent characteristics and the health area level characteristics.
From the bivariate regression, all variables with a p-value <0.25 were included in the multivariate analysis [33]. The final model was selected throughout a backward stepwise approach, based on the smallest Akaike information criterion (AIC). Multicollinearity among independent variables was checked by estimating the variance inflation factor. The “ANC visit” variable was removed from the multivariate analysis because it was constant to “mother receiving tetanus vaccine during the last pregnancy” variable. Intra-Class Correlation (ICC) and proportional change in variance (PCV) were used to quantify the magnitude of the effect of health area itself on ZD children [34]. We evaluated the accuracy of the final model using the receiver operating characteristic (ROC) by measuring the area under the curve (AUC).
Spatial analysis
The household’s geographic locations were subjected to a spatial point patterns analysis [35]. All enrolled households were connected to latitude and longitude data. The population density data was downloaded from GRID3 DRC Gridded Population Estimates [36]. An exploratory analysis using the Ripley’s K function [37]was performed to determine whether ZD children tended to occur close to other ZD children [38]. The maximum and minimum values for the x and y coordinates were used to construct a grid containing the coordinates of ZD children [39]. The K function was then transformed into an L function which is easier to interpret visually [37]. The analysis was done in R Studio version 2023.03.0 built 386 for Windows using the Spatstat package.
Spatial clusters (hotspots) were identified using the spatial scan statistic in the SaTScan software [40]. A circular window shape was used with a maximum spatial cluster size of 50% of the population at risk. A likelihood ratio test is applied to each window, which is considered as a potential cluster, to test the null hypothesis of absolute spatial randomness against the alternative hypothesis that there is an elevated risk within the window as compared to outside the window. The inference was done by using Monte Carlo simulations where replications of the data set under the null hypothesis were generated. A Bernoulli distribution was used as the probability model and a cluster was significant if it had a maximum likelihood ratio higher than the maximum likelihood ratio from the most likely cluster generated in the random data set [41]. The level of significance was set at p = 0.05. All statistically significant clusters were mapped on to the study area using QGIS version 3.30.3.
Missing data
Five children out of 1,868 enrolled (0.3%) did not have DTP and covariate information and were excluded from analysis.
Ethical approval
Data collectors ensured that participants understood the purpose of the study and formal written informed consent was obtained from the parent/guardian at household level and from the head/deputy nurse at health facility level prior to participation in the study. The study protocol was approved by both the Institute of Tropical Medicine, Antwerp IRB board (IRB Approval letter FRM-1257v3.0) and the Kinshasa School of Public Health Ethics Committee (Approval N° ESP/CE/129/2022).
Results
Characteristics of households, mothers/guardians and children
Overall, 1,863 children aged 12 to 23 months were included in the analysis. There were 210 (11.3%) children in a migrant or returnee household; 33.1% had vaccination cards; and almost 67% of mothers/guardians declared to have not received visit of CHWs the three last months (Table 2). Nearly 16% of mothers/guardians were not able to quote at least one VPD (Table 3). The median distance between households and the main health centre inside a health area was 467 meters, with a minimum of 2.81 meters and a maximum of 4693.3 meters.
Health areas characteristics
92% of health areas in the North Kikwit HZ reported stockout during the last 12 months while in the South HZ, 54.2% of health areas were affected by a stockout (p = 0.007). In both North and South HZ some parts were inaccessible due to ravines (p = 0.65). 70% of the health areas had refrigerators in the South and 64% in the North (p = 0.61). (Fig 2).
Prevalence of zero-dose children
Overall, 616/1863 (33.1%) of households presented vaccination cards. For the remainder 1247 households, DTP vaccination status was based on maternal recall. Based on the vaccination card (n = 616), the prevalence of ZD was 3.5% (13/371) in the North HZ and 4.9% (12/245) in the South and based on maternal recall (n = 1247), this prevalence was 17.9% (105/585) in the North and 26.1% (173/662) in the South.
Overall, 303/1863 children (16.3%) were considered as ZD considering both vaccination card and mother recall, less in the North 12.3% (118/956) than in the South HZ 20.4% (185/907) (p <0.001). Significant variation in prevalence was found between health areas (p<0.001 for both HZ). In the North HZ, three health areas had a ZD prevalence equal to 0, and the highest prevalence was 33.3%. In the South HZ, one health area had a ZD prevalence equal to 0, while three health areas had a prevalence > 40%. (Fig 3).
Geospatial distribution of ZD children
Geographical point pattern analysis revealed a spatial clustering with two Hotspots of ZD children in Kikwit city. Hotspot 1 was restricted in the South HZ and had a relative risk of 2.48 of being zero-dose. Hotspot 2 covered six health areas in the South HZ and four in the North HZ and had a relative risk of 1.77 (Table 4) (Fig 4)
Children living in these hotspots have a high risk of being ZD, whereby we observed that this risk was higher in hotspot 1 than in hotspot 2 (prevalence difference p-value 0.036). The expected number of ZD cases was based on the overall prevalence (16.3%).
There is an overlap between of Hotspots between ZD children and the lack of CHWs’ home visit. Health areas with a high prevalence of ZD children were located more on the outskirts of the city and those with a low prevalence were located more in the middle and west of the city. The CHWs home visit rate was less than 25 per 100 households in 4/5 health areas in the hotspot 1. Four health areas out of five (80%) lacked a refrigerator in the hotspot 1, and four out of twelve (33.3%) in the hotspot 2. Ten health areas (83.6%) in the hotspot 2 had stockout during the previous twelve months, while one (20%) in the hotspot 1.
Risk factors of ZD children
In multivariate analysis, six variables were associated with ZD prevalence. The absence of visits from CHWs (aOR = 1.90), household located more than 1 kilometre from a health centrer (aOR = 1.95), not receiving maternal tetanus vaccination during the previous pregnancy (aOR = 3.16), and mothers/guardians unable to name at least one VPD (aOR = 3.20) significantly increased the risk of ZD. On the other hand, children of the Bunda and Mbala ethnical groups (respectively, aOR = 0.36 and 0.52) and those from mother/guardians with secondary (aOR = 0.56) or university (aOR = 0.21) level of education were at lower risk for ZD (Table 5). Vaccine stockout, lack of a refrigerator and inaccessibility of some part of the health area were not significantly associated with ZD prevalence.
The ROC showed an area under the curve (AUC), of 83.9% (95%CI 81.5% - 86.2%), indicating the model was adequate in differentiating those not having received a single dose of DTP vaccine from those receiving it. Details regarding health area-level variance of the multilevel logistic models predicting zero-dose children and the random effect results are included in the next section.
Random effect results
After considering both household and health area-level covariates (Model 3), health area-level variability was reduced from 49.3% to 29.2% (Table 6). The model also showed that 40.7% of health area-level variance on zero-dose was explained by the combined factors at both the household and health area levels.
Although model 1 has the smallest AIC, this model included only the health area random effect, and no fixed effect was considered.
Discussion
We found a zero-dose children prevalence of 16.3% in Kikwit city. This finding is higher than in earlier studies of LMICs and the province of Kwilu, to which the city belongs [5, 10] but less than the prevalence throughout the country of 19.1% [10, 25]. However, we found substantial differences in the prevalence between and within the two HZ of the Kikwit city. The South HZ was more affected by ZD children then the North HZ.
Global health experts have recently focused on pinpointing and targeting VPD-risk populations geographically. This helps identify underserved communities and facilitates more precise interventions and resource allocation [42]. Specific risk factors related to individuals/ households might be useful in identifying at-risk groups. However, it can be challenging to consider these characteristics when they are spread out over the geographical area being studied, particularly during vaccination campaigns. Our study demonstrates and sheds light on the existence of ‘geographical’ pocket and communities of ZD children within HZs in the city of Kikwit. These results provide evidence to decision-makers to develop targeted strategies where more effort is needed. We have identified two major hotspots for ZD children in the city, which were mainly located in the South HZ, with one overlapping some health areas in the North HZ. For a child located in this geographical area, the risk of not being vaccinated against DTP is twice as high compared to other parts of the city. As DTP is administered together with Haemophilus influenzae type B and Hepatitis B in the pentavalent vaccine in DRC, the risk of not being vaccinated against these pathogens is similar as ZD. Well beyond the timing of these vaccines, this child also has a high risk of not receiving measles and yellow fever vaccine, even while these are routinely given at a later age [22].
The Immunization Agenda 2030 aims to significantly reduce (50%) the number of ZD children by 2030 by achieving high and equitable coverage levels [1, 4]. The process of achieving this goal can be made easier with the use of geospatial and multilevel analysis.
In this study, within a single HZ, the distribution of ZD children was very heterogeneous, varying significantly between health areas (sub-zonal level). Thus, having information on the prevalence of ZD at the HZ level only is of limited value to target interventions, as there can be substantial heterogeneity between health areas. For instance, the zero-dose prevalence in the South HZ was 20.4%; while at the sub-zonal level, there is one health area with a ZD prevalence equal to zero and another health area with a ZD prevalence of more than 40%, so clearly the two communities do not have the same priority for intervention. To achieve equitable coverage levels, it is necessary that information on the level of coverage is detailed down to the sub-zonal level (health area-level).
The geospatial analysis also revealed that health areas located on the outskirts of the city were more affected by ZD children than those located in the middle of the city, and the middle of the city has a higher population density than the outskirts, which is in line with another study demonstrating that ZD children are more likely residing in missed communities that are difficult to access, or that receive less attention from political and health decision-makers [1]. The geospatial analysis could help not only in prioritizing missed communities but also in monitoring the resource allocation. 80% of health areas lacked a refrigerator and 83.3.6% had stockout during the previous twelve months, in the first hotspot. The geospatial approach has the potential of enhancing access to not just immunization activities, but also other primary health services missed by these communities.
We observed a substantial clustering of ZD prevalence within Kikwit city using multilevel analysis, with an Intra-class correlation of 29.2% (> conventional threshold of 5%) [43]. This suggests that health area-level factors account for 29.2% of the variance of the prevalence of ZD children in the city (after controlling for household-level factors), which is in line with previous studies in this field [29, 44], even though we did not find a significant association with vaccine stockout, absence of a refrigerator or accessibility problems in the health area. However, this study revealed a significant association between ZD prevalence and the CHW home visit and the distance between households and health centre which can be considered proxies of the general health system functioning. This highlights the need for interventions that target not only household or individual vaccination related factors, but also factors at community/health area level, such as community mobilization by CHWs and bringing the vaccination services closer to the households. The absence of CHWs’ home visits nearly doubled the risk of being zero-dose. CHWs cannot replace qualified medical professionals, but they are vital to the community and can increase service uptake. CHW programmes in deprived regions provide promotive, preventive care and patient referral services, helping to improve population health and advance universal health coverage [26, 27, 45]. However, it is unlikely that CHWs will perform this task on a volunteer basis given the increasing time demands on them. The possibility of paying them for their time and supporting their transportation expenses should be taken into account when defining nation-wide health policies.
Households located more than one kilometre from the health centre were associated with twice the risk of having ZD children compared with those located less than one kilometre away. According to the operating standards for a health centre in the DRC, the health centre serves a population within a maximum radius of action of 8 to 15 km [11] This result highlights the need to have several locations offering the immunization service within a health area and emphasises the importance of outreach in the delivery of vaccination to children.
Our findings showed that children of mothers/guardians with at least secondary education were less likely to be zero-dose than those with no schooling or primary education. Mothers/guardians who couldn’t name a single VPD had a threefold risk of having a ZD child compared to those who named at least one VPD. Several studies have linked maternal education to immunization rates [25, 28, 32, 46, 47]. Education of mothers positively impacts children’s health by providing good information and influencing beliefs, perceptions, and practices that unintentionally harm children’s health. Education brings knowledge, eliminates ignorance, and increases understanding of diseases. The fact that some mothers couldn’t name one VPD showed that the immunization information was inadequate for potential stakeholders in the area.
Mothers who did not receive a tetanus vaccine during pregnancy were more likely to have a ZD child than those who did. It exhibits similarities to the occurrences observed in India [48]. ANC informs women about vaccination [28–31], but agreeing to be vaccinated is a separate concept. During ANC sessions, health education is provided to encourage women to get vaccinated. CHWs play a crucial role in providing health education within the community. They possess the ability to effectively tackle cultural obstacles (such as home birth) that may hinder the utilisation of healthcare services through health education. The impact of promoting tetanus vaccination for pregnant women appears to go beyond the mother’s health and may have an impact on her willingness to prioritize vaccination of the child.
Our study also found that households belonging to the Bunda and Mbala ethnicities were about half as likely to have ZD children as Pende ethnicity in the same city. DRC is a country of great cultural diversity, with over 450 ethnic groups, and previous studies in Africa settings have demonstrated the influence of ethnicity on behaviour towards vaccination [49, 50]. It will be necessary to deepen the knowledge on this, in order to understand through a qualitative study why in these ethnic groups, children were less likely to be ZD compared to the Pende ethnic group.
The strength of this study lies in the collection of household data and not basing our results on routinely collected surveillance data that is known to have weaknesses [7], and the sampling method, covering all 49 health areas of the two HZ of Kikwit city in two weeks’ time. Having information from the health centres besides the households enabled us to understand the situation beyond household-level considerations, by taking into account other factors in the health system. The limitations of this study could be that DTP vaccination data were obtained from vaccination cards, which may contain errors or be difficult to read or interpret. We observed a few transcription errors (7.3%) on the vaccination cards, the date (day, month, and year) on which a specific vaccine is administered must be indicated on the card, but some vaccination cards indicate "day and month" or "month and year" or simply a tick beside the administered vaccine. These errors could be confusing for interpretation when making calculations. If the card was not available, vaccination data were collected from the mother/guardian’ recall. Several research in the same field rely on maternal recall as a method to gather data on childhood immunisation. This is because vaccination cards are less accessible at household level and vaccination records at health facilities level are inadequately maintained [8, 9, 25, 51, 52]. Nevertheless, the maternal recall of childhood immunization may be biased.
At health area level, we collected data on the availability of refrigerator, the accessibility of some avenues/villages, and the vaccine stockout during the last 12 months, but could have included other factors, such as the supervision of vaccination sites, the number of vaccination sessions during a given period (week, month), the motivation or incentive of health professionals, etc. Cluster analysis is a statistical technique and assumes no underlying knowledge of the situation or how population may behave. In other words, it is just clustering the data around a series of central points–which way it may or may not make sense, that is why we combined this technique with a multilevel regression approach not only to identify the existence of clusters but also risk factors, which made results more understandable. And this study was cross-sectional and hence, cannot define trend over time.
This study provided us with the prevalence of ZD children at a fine-scale (health area) level to enable targeted interventions to reach such children. This data can also be used in future research to assess the performance of other techniques of estimating ZD children at fine-scale prevalence [19, 53]
Conclusion
Our findings demonstrated a high prevalence of ZD children with a substantial heterogeneity in the city and identified two hotspots within the South HZ, with one of the hotspots overlapping with health areas in North HZ. Due to sub-zonal diversity, a health zone approach to reduce ZD immunization appears limited. ZD prevalence was related to the CHWs’ home visit, to the distance of residence to a health centre and to household-level factors. Geospatial results could help in targeting priority health areas and communities for vaccination with the support of CHWs.
Supporting information
S1 Data. A QGIS-compatible geospatial file containing the results of geospatial analysis, which may be accessed using QGIS software.
https://doi.org/10.1371/journal.pgph.0002617.s001
(QGS)
Acknowledgments
The authors would like to express their warm thanks to the children, parents/guardians, and healthcare workers who agreed to participate in this study. A particular thanks to Joel Kanyama, Serge Mukawu and Théo Emboni who contributed in important aspects to the current work.
References
- 1. GAVI. Zero-dose children and missed communities. 2023 [cited 12 Mar 2023]. Available: https://www.gavi.org/our-alliance/strategy/phase-5-2021-2025/equity-goal/zero-dose-children-missed-communities#fn5
- 2. Utazi CE, Aheto JMK, Wigley A, Tejedor-Garavito N, Bonnie A, Nnanatu CC, et al. Mapping the distribution of zero-dose children to assess the performance of vaccine delivery strategies and their relationships with measles incidence in Nigeria. Vaccine. 2023;41: 170–181. pmid:36414476
- 3. Taneja G, Datta E, Sapru M, Johri M, Singh K, Jandu HS, et al. An Equity Analysis of Zero-Dose Children in India Using the National Family Health Survey Data: Status, Challenges, and Next Steps. Cureus. 2023;15. pmid:36851944
- 4. WHO. Immunization Agenda 2030. 2021; 1–24. Available: https://www.who.int/immunization/ia2030_Draft_One_English.pdf?ua=1
- 5. Cata-Preta BO, Santos TM, Mengistu T, Hogan DR, Barros AJD, Victora CG. Zero-dose children and the immunisation cascade: Understanding immunisation pathways in low and middle-income countries. Vaccine. 2021;39: 4564–4570. pmid:33744046
- 6. Système National d’Information Sanitaire en RDC. Taux de couverture vaccinale en DTC1, système national d’information sanitaire, RDC. 30 Nov 2023 [cited 1 Dec 2023]. Available: https://snisrdc.com/dhis-web-data-visualizer/index.html#/
- 7. Dolan SB, Carnahan E, Shearer JC, Beylerian EN, Thompson J, Gilbert SS, et al. Redefining vaccination coverage and timeliness measures using electronic immunization registry data in low- and middle-income countries. Vaccine. 2019;37: 1859–1867. pmid:30808566
- 8. Paul Samson Dikasa Lusamba. Rapport d’enquête de couverture vaccinale chez les enfants de 6–23 mois en RDC (2021). Kinshasa; 2021. pp. 139–140. Available: https://espkinshasa.net/enquete-de-couverture-vaccinale-chez-les-enfants-de-6-23-mois-en-republique-democratique-du-congo/
- 9. Lusamba Paul Samson Dikasa, Nyandwe Jean, Mafuta Eric. Rapport d’Enquête de couverture vaccinale chez les enfants âgés de 6 mois à 23 mois en République Démocratique du Congo, 2020. Kinshasa; 2020. pp. 65–67.
- 10. Lusamba Paul Samson Dikasa. Rapport d’enquête de couverture vaccinale chez les enfants de 6 à 23 mois en République Démocratique du Congo, 2021. Kinshasa; 2021. pp. 76–77. Available: https://espkinshasa.net/enquete-de-couverture-vaccinale-chez-les-enfants-de-6-23-mois-en-republique-democratique-du-congo/
- 11. Ministère de la santé publique, Secrétariat Général. Recueil des normes de création, d’organisation et de fonctionnement des structures de la zone de sante en République Démocratique Du Congo. Kinshasa; 2019. Available: http://www.scogo.cd/_iserv/dlfiles/dl.php?ddl=normes-zs-actualisees-juillet-2012x.pdf
- 12. WHO. Limitations related to globally reported immunization subnational data. 2023. Available: https://www.who.int/teams/immunization-vaccines-and-biologicals/immunization-analysis-
- 13.
Cutts FT, Danovaro-Holliday MC, Rhoda DA. Challenges in measuring supplemental immunization activity coverage among measles zero-dose children. Vaccine. Elsevier Ltd; 2021. pp. 1359–1363. https://doi.org/10.1016/j.vaccine.2020.11.050 pmid:33551302
- 14. Utazi CE, Thorley J, Alegana VA, Ferrari MJ, Takahashi S, Metcalf CJE, et al. Mapping vaccination coverage to explore the effects of delivery mechanisms and inform vaccination strategies. Nature Communications 2019 10:1. 2019;10: 1–10. pmid:30967543
- 15. Takahashi S, Metcalf CJE, Ferrari MJ, Tatem AJ, Lessler J. The geography of measles vaccination in the African Great Lakes region. Nat Commun. 2017;8. pmid:28541287
- 16. Utazi CE, Thorley J, Alegana VA, Ferrari MJ, Takahashi S, Metcalf CJE, et al. High resolution age-structured mapping of childhood vaccination coverage in low and middle income countries. Vaccine. 2018;36: 1583–1591. pmid:29454519
- 17. Mosser JF, Gagne-Maynard W, Rao PC, Osgood-Zimmerman A, Fullman N, Graetz N, et al. Mapping diphtheria-pertussis-tetanus vaccine coverage in Africa, 2000–2016: a spatial and temporal modelling study. The Lancet. 2019;393: 1843–1855. pmid:30961907
- 18. Cata-Preta BO, Santos TM, Mengistu T, Hogan DR, Barros AJD, Victora CG. Zero-dose children and the immunisation cascade: Understanding immunisation pathways in low and middle-income countries. Vaccine. 2021;39: 4564–4570. pmid:33744046
- 19. Biswas A, Tucker J, Bauhoff S. Performance of predictive algorithms in estimating the risk of being a zero-dose child in India, Mali and Nigeria. BMJ Glob Health. 2023;8. pmid:37821114
- 20. Wendt A, Santos TM, Cata-Preta BO, Arroyave L, Hogan DR, Mengistu T, et al. Exposure of Zero-Dose Children to Multiple Deprivation: Analyses of Data from 80 Low- and Middle-Income Countries. Vaccines (Basel). 2022;10. pmid:36146646
- 21. Nampa IW, Mudita IW, Riwu Kaho NPLB, Widinugraheni S, Lasarus Natonis R. The KoBoCollect for Research Data Collection and Management (An experience in Researching the Socio-Economic Impact of Blood Disease in Banana). SOCA: Jurnal Sosial, Ekonomi Pertanian. 2020;14: 545.
- 22. UNICEF. Calendrier vaccinal pour les enfants de 0 à 15 mois en RDC | UNICEF. [cited 17 Mar 2023]. Available: https://www.unicef.org/drcongo/calendrier-vaccinal-enfants-rdc
- 23. Ashbaugh HR, Hoff NA, Doshi RH, Alfonso VH, Gadoth A, Mukadi P, et al. Predictors of measles vaccination coverage among children 6–59 months of age in the Democratic Republic of the Congo. Vaccine. 2018;36: 587–593. pmid:29248265
- 24. Nchinjoh SC, Saidu Y, Agbor VN, Mbanga CM, Jude Muteh N, Njoh AA, et al. Factors Associated with Zero-Dose Childhood Vaccination Status in a Remote Fishing Community in Cameroon: A Cross-Sectional Analytical Study. Vaccines (Basel). 2022;10. pmid:36560465
- 25. Ishoso DK, Danovaro-Holliday MC, Cikomola AM-W, Lungayo CL, Mukendi J-C, Mwamba D, et al. “Zero Dose” Children in the Democratic Republic of the Congo: How Many and Who Are They? Vaccines (Basel). 2023;11: 900. pmid:37243004
- 26.
Zulu JM, Perry HB. Community health workers at the dawn of a new era. Health Research Policy and Systems. BioMed Central Ltd; 2021. https://doi.org/10.1186/s12961-021-00761-7 pmid:34641904
- 27. Hartzler AL, Tuzzio L, Hsu C, Wagner EH. Roles and functions of community health workers in primary care. Ann Fam Med. 2018;16: 240–245. pmid:29760028
- 28. Legesse E, Dechasa W. An assessment of child immunization coverage and its determinants in Sinana District, Southeast Ethiopia. BMC Pediatr. 2015;15. pmid:25886255
- 29. Adedokun ST, Uthman OA, Adekanmbi VT, Wiysonge CS. Incomplete childhood immunization in Nigeria: A multilevel analysis of individual and contextual factors. BMC Public Health. 2017;17. pmid:28270125
- 30. Babalola S, Lawan U. Factors predicting BCG immunization status in northern Nigeria: A behavioral-ecological perspective. Journal of Child Health Care. 2009;13: 46–62. pmid:19240190
- 31. Koumaré AK, Traore D, Haidara F, Sissoko F, Traoré I, Dramé S, et al. Evaluation of immunization coverage within the Expanded Program on Immunization in Kita Circle, Mali: a cross-sectional survey. BMC Int Health Hum Rights. 2009;9. pmid:19828057
- 32. Mekonnen ZA, Mekonnen ZA, Gelaye KA, Were MC, Tilahun B. Timely completion of vaccination and its determinants among children in northwest, Ethiopia: A multilevel analysis. BMC Public Health. 2020;20. pmid:32527248
- 33. Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in logistic regression. Source Code Biol Med. 2008;3. pmid:19087314
- 34. Austin PC. Intermediate and advanced topics in multilevel logistic regression analysis. 2017. pmid:28543517
- 35. Ripley BD. Spatial point pattern analysis in ecology. Developments in numerical ecology. 1987; 407–429.
- 36.
GRID3. GRID3 DRC Gridded Population Estimates, Version 2.0 (Kinshasa, Kongo-Central, Kwango, Kwilu, and Mai-Ndombe provinces) | GRID3 Data Hub. 17 Feb 2021 [cited 7 Jul 2023]. Available: https://data.grid3.org/maps/GRID3::grid3-drc-gridded-population-estimates-version-2-0-kinshasa-kongo-central-kwango-kwilu-and-mai-ndombe-provinces/about
- 37. Ripley B. D. (1977). Modeling Spatial Patterns. Journal of the Royal Statistical Society. Series B (Methodological), 39, 172–212.—References—Scientific Research Publishing. [cited 3 Apr 2023]. Available: https://www.scirp.org/(S(czeh2tfqyw2orz553k1w0r45))/reference/ReferencesPapers.aspx?ReferenceID=1314678
- 38. Dixon Philip M. Ripley’s K function. Encyclopedia of Env. El-Shaarawi AH, Piegorsch WW, editors. Encyclopedia of Environmetrics. Chichester; 2002. Available: https://www3.nd.edu/~mhaenggi/ee87021/Dixon-K-Function.pdf
- 39. Ripley BD. Tests of “Randomness” for Spatial Point Patterns. Journal of the Royal Statistical Society: Series B (Methodological). 1979;41: 368–374.
- 40. Kulldorff M. A spatial scan statistic. 2007;26: 1481–1496.
- 41. Kulldorff M, Mostashari F, Duczmal L, Katherine Yih W, Kleinman K, Platt R. Multivariate scan statistics for disease surveillance. Stat Med. 2007;26: 1824–1833. pmid:17216592
- 42. Utazi CE, Thorley J, Alegana VA, Ferrari MJ, Takahashi S, Metcalf CJE, et al. Mapping vaccination coverage to explore the effects of delivery mechanisms and inform vaccination strategies. Nat Commun. 2019;10: 1–10. pmid:30967543
- 43.
Heck RH, Thomas S, Tabata L. Multilevel Modeling of Categorical Outcomes Using IBM SPSS. Routledge; 2013. https://doi.org/10.4324/9780203808986
- 44. Ditekemena JD, Nkamba DM, Mutwadi A, Mavoko HM, Nelson J, Fodjo S, et al. COVID-19 Vaccine Acceptance in the Democratic Republic of Congo: A Cross-Sectional Survey. 2021; 1–11. pmid:33672938
- 45. WHO. Community Health Worker programmes in the WHO African region: Evidence and Options-Policy Brief. Ed. 2017. 2017. Available: http://apps.who.int/bookorders.
- 46. Tadesse H, Deribew A, Woldie M. Predictors of defaulting from completion of child immunization in south Ethiopia, May 2008 A case control study. BMC Public Health. 2009;9. pmid:19463164
- 47. Johri M, Rajpal S, Subramanian S V. Progress in reaching unvaccinated (zero-dose) children in India, 1992–2016: a multilevel, geospatial analysis of repeated cross-sectional surveys. Lancet Glob Health. 2021;9: e1697–e1706. pmid:34798029
- 48. Sissoko D, Trottier H, Malvy D, Johri M. The influence of compositional and contextual factors on non-receipt of basic vaccines among children of 12-23-month old in India: A multilevel analysis. PLoS One. 2014;9. pmid:25211356
- 49. Masters NB, Wagner AL, Carlson BF, Muuo SW, Mutua MK, Boulton ML. Childhood vaccination in Kenya: socioeconomic determinants and disparities among the Somali ethnic community. Int J Public Health. 2019;64: 313–322. pmid:30535788
- 50. Mohamud AN, Feleke A, Worku W, Kifle M, Sharma HR. Immunization coverage of 12–23 months old children and associated factors in Jigjiga District, Somali National Regional State, Ethiopia. BMC Public Health. 2014;14. pmid:25146502
- 51. Cutts FT, Izurieta HS, Rhoda DA. Measuring Coverage in MNCH: Design, Implementation, and Interpretation Challenges Associated with Tracking Vaccination Coverage Using Household Surveys. PLoS Med. 2013;10. pmid:23667334
- 52. Mahamud A, Burton A, Hassan M, Ahmed JA, Wagacha JB, Spiegel P, et al. Risk factors for measles mortality among hospitalized Somali refugees displaced by famine, Kenya, 2011. Clinical Infectious Diseases. 2013;57: 160–166. pmid:23821730
- 53. Saidu Y, Agbor VN, Di Mattei P, Nchinjoh SC, Edwidge NN, Njoh AA, et al. Development of a composite scoring system to rank communities at high risk of zero-dose children in Cameroon: A geospatial analysis. J Glob Health. 2023;13. pmid:37971948