Factors Associated with the Uptake of Rotavirus and Pneumococcal Conjugate Vaccines among Children in Armenia: Implications for Future New Vaccine Introductions
1. Introduction
To better plan for immunization programs and further reduce the burdens of rotavirus and pneumococcal diseases in Armenia, our objectives were to (1) quantify uptake of the RV and the PCV among children living in Armenia in 2015–2016 and (2) examine the individual- and community-level factors associated with the non-uptake of each dose of the vaccines. Accomplishing these objectives may identify areas of improvement and foci for interventions.
2. Materials and Methods
This project was a secondary analysis of an existing de-identified data set from the ADHS and was determined by The George Washington University Institutional Review Board to not be human research. The analytic sample for this analysis consisted of all living children aged 0 to 35 months with documented birthdates. The subsamples for each vaccination were composed of children eligible for that particular vaccination and dose based on the recommendations of Armenia’s National Immunization Program schedule and the year the vaccine was introduced.
Vaccination information was collected for all children under three years of age in each household. Documentation of immunization was obtained either from child health cards, which were maintained by local health facilities, or immunization passports, which were kept by the child’s parent or guardian. Data were collected from both sources if available. If neither was available, vaccination history reported by the mother was recorded. If there was no vaccination documentation and the mother could not recall any details, the vaccination was decidedly not administered, as per ADHS documentation.
Uptake, defined as the documented receipt of each of the doses of the RV and the PCV, was assessed by determining the weighted proportion of children who had received each dose. Frequencies, proportions, and 95% confidence intervals (CIs) for the estimates of uptake of each dose of the RV and the PCV were calculated.
Multivariable multilevel logistic regression models were used to examine the individual and contextual factors associated with RV and PCV non-uptake, defined as those without a documented date of vaccination, among children with vaccination cards and those eligible for the particular vaccination. Potential factors, based on a literature review, included individual-level factors such as the child’s sex, age, siblingship, and birth order, as well as the mother’s age, education, employment status, antenatal care, and place of delivery, and the household’s wealth index. Contextual or community-level factors, such as distance to health clinic (categorized as “not a big problem” and “big problem” by the ADHS), place of residence, and region, were also considered. Multicollinearity was assessed by examining the variance inflation factor (VIF) using less than five as criteria to rule out multicollinearity. Birth order and siblingship were found to be multicollinear, resulting in birth order being excluded from subsequent analyses. Wealth index, a composite measure created by the ADHS using principal component analysis, was based on a household’s ownership of certain assets, housing construction materials, and access to water and sanitation. For the purposes of this analysis, wealth index was further grouped into three categories: poorer/poorest as low, middle as middle, and richer/richest as high. Armavir, due to its size and composition as the largest province in Armenia with both rural and urban regions, was the reference for these analyses.
All tests were two-sided and p < 0.05 was used to define statistical significance. Due to the study design of the ADHS, all statistical methods utilized techniques and included weights where possible to account for the complex sampling design. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). No imputation for missing data was performed.
3. Results
For RV2, among the children living in urban areas, older age was again associated with lower odds of non-uptake (6 to 11 months: aOR = 0.30, 95%CI: 0.12–0.75; 12 to 17 months: aOR = 0.17, 95%CI: 0.06–0.53; 18 to 23 months: aOR = 0.14, 95%CI: 0.04–0.47; 24 to 29 months: aOR = 0.22, 95%CI: 0.06–0.78; 30 to 35 months: aOR = 0.24, 95%CI: 0.08–0.72). Additionally, among the children living in urban areas, those whose mothers viewed distance to the health clinic as a big problem had almost three times the odds of non-uptake as those who did not (aOR = 2.94, 95%CI: 1.09–7.89). Among the children living in rural areas, non-uptake was highest in wealthier children, with those in the high wealth index category having almost four times the odds of non-uptake (aOR = 3.76, 95%CI: 1.56–9.05).
4. Discussion
The 2015–2016 ADHS is the first data collection cycle after the RV and the PCV were added into Armenia’s National Immunization Schedule providing an opportunity to examine the initial uptake of both vaccines. The uptake of the RV was found to be high, but the uptake of the PCV was much lower. The initial uptake of the RV was in line with coverage levels of the other childhood vaccinations (tuberculosis (BCG), diphtheria–tetanus–pertussis (DTP), polio (Pol), and measles containing vaccine (MCV)), which ranged from 76.1% for the third dose of DTP to 91.7% for BCG in the 2015–2016 ADHS. PCV uptake, although lower, followed the same pattern as the other vaccines which require a course of three vaccinations (DTP and Pol). This may suggest that the two NVIs examined in this study are on track to have relatively high coverage levels once they become more established components of the National Immunization Schedule. Although more time has passed since the introduction of the RV compared to the PCV, which may have affected uptake, other factors (such as those examined in this study) may also be involved. Administering the RV orally as opposed to intramuscularly may be a reason for the differences in uptake, as this mode of delivery can be more appealing, especially in light of the multiple vaccinations children receive at the recommended ages. The high initial uptake of both RV doses is encouraging for the prospects of increasing PCV uptake. PCV1 and PCV2 are due at the same times as RV1 and RV2, demonstrating that mothers are interacting with the healthcare system around the recommended time. In an effort to increase the uptake of the PCV, healthcare providers can utilize this interaction with mothers and stress the importance of receiving all vaccinations as recommended in an attempt to overcome any potential apprehension among mothers during their visits.
Multilevel logistic regression models also allowed us to examine the effect of clustered sampling on the uptake of NVIs by calculating the ICC. For both doses of the RV, there was evidence of significant variability in the odds of non-uptake due to intra-cluster differences among children living in urban areas. For the PCV series, only PCV2 demonstrated significant variability. The observed decreases in the ICCs for RV1 and RV2 in rural areas and for PCV2 suggest that the included factors explain some of the variability in the non-uptake of these vaccinations between clusters. For both doses of the RV, among children living in urban areas the ICC did not decrease after adding explanatory variables. The inclusion of individual- and community-level factors should have had an impact on the ICCs since they are expected to explain some of the variance in the outcome of the model. The lack of an observed change indicates that factors other than those which were included in this analysis may be responsible for intra-cluster variability. Further studies on the differences between clusters could provide helpful information for future NVIs.
There are some limitations to this analysis that are important to consider. Vaccination data were based solely on the information recorded by trained field workers from cards completed by healthcare professionals, and the accuracy of this transcription is unknown. A vaccine that is not documented may be delayed but eventually received. Additionally, the sample was restricted to only those with documented vaccination information in order to curb recall bias, but this does usually introduce potential bias if having a vaccination card is associated with being vaccinated. Since most children presented health cards to the survey staff, this may not have been the case. There were some vaccinations reported by mothers that were not considered to be received, as there was no documented date, in an effort to have a conservative estimate of the uptake. The frequency of a child having only a maternal report of receipt of a vaccination was low and thus should not have a major impact on the estimates. Therefore, by restricting analyses to children with health cards and considering only those with a documented date of vaccination as received, uptake may be underestimated. Additionally, the 2015–2016 ADHS only allows for a relatively short timeframe when evaluating NVI implementation, and more time between introduction and analysis may be needed to provide a better understanding of uptake. Nonetheless, this study provides a baseline for future studies centered around the uptake of the RV and the PCV and sheds light on the mechanisms behind NVIs. Potential factors were limited to what was collected in the ADHS, and, consequently, other factors not evaluated may be associated with the uptake of the RV and the PCV. Furthermore, due to the small sample sizes in some regions, which limited the ability to evaluate the association of regions with non-uptake, associations observed should be taken as an indication for further studies with larger sample sizes in each region in order to obtain more conclusive results. Lastly, due to the cross-sectional nature of the data, causal inferences cannot be made, and factors that are found to be significant should thus be classified in association with the non-uptake of the RV and the PCV.
5. Conclusions
Using the most recent ADHS data, we were able to examine the uptake and associated factors of the two most recent NVIs among children living in Armenia. Although uptake was high for the RV, there is room for improvement for the PCV series, especially the third dose. Future studies should examine the impact of these vaccinations on the occurrence of related diseases in Armenia and assess the timing of each dose of both vaccines in order to provide more information on the uptake and coverage of these NVIs. A deeper look into other potential factors, such as structural factors, as well as maternal attitudes and knowledge, may offer insights that could be used to strengthen NVI activities among children living in Armenia. A better understanding of the uptake of the RV and the PCV is crucial in order to promote optimal coverage when using these vaccinations and better protect the children, as well as the general population, in Armenia.