An Evaluation of Voluntary Varicella Vaccination Coverage in Zhejiang Province, East China

Background: In 2014 a 2-doses varicella vaccine (VarV) schedule was recommended by the Zhejiang Provincial Center for Disease Control and Prevention. We aimed to assess the coverage of the 1st dose of VarV (VarV1) and the 2nd dose of VarV (VarV2) among children aged 2–6 years through the Zhejiang Provincial Immunization Information System (ZJIIS) and to explore the determinants associated with the VarV coverage. Methods: Children aged 2–6 years (born from 1 January 2009 to 31 December 2013) registered in ZJIIS were enrolled. Anonymized individual records of target children were extracted from the ZJIIS database on 1 January 2016, including their VarV and (measles-containing vaccine) MCV vaccination information. The VarV1 and VarV2 coverage rates were evaluated for each birth cohorts. The coverage of VarV also was estimated among strata defined by cities, gender and immigration status. We also evaluated the difference in coverage between VarV and MCV. Results: A total of 3,028,222 children aged 2–6 years were enrolled. The coverage of VarV1 ranged from 84.8% to 87.9% in the 2009–2013 birth cohorts, while the coverage of VarV2 increased from 31.8% for the 2009 birth cohort to 48.7% for the 2011 birth cohort. Higher coverage rates for both VarV1 and VarV2 were observed among resident children in relevant birth cohorts. The coverage rates of VarV1 and VarV2 were lower than those for the 1st and 2nd dose of MCV, which were above 95%. The proportion of children who were vaccinated with VarV1 at the recommended age increased from 34.6% for the 2009 birth cohort to 75.2% for the 2013 birth cohort, while the proportion of children who were vaccinated with VarV2 at the recommended age increased from 19.7% for the 2009 birth cohort to 48.7% for the 2011 birth cohort. Conclusions: Our study showed a rapid increasing VarV2 coverage of children, indicating a growing acceptance of the 2-doses VarV schedule among children’s caregivers and physicians after the new recommendation released. We highlighted the necessity for a 2-doses VarV vaccination school-entry requirement to achieve the high coverage of >90% and to eliminate disparities in coverage among sub-populations. We also recommended continuous monitoring of the VarV coverage via ZJIIS over time.


Introduction
Since 1998, varicella vaccine (VarV) has been licensed in Zhejiang Province, which is a developed province with a large population of 80 million located in East China. The 1-dose of VarV schedule for routine use in children aged ě12 months was recommended by Zhejiang Provincial Center for Disease Control and Prevention (ZJCDC) since the approval of VarV. VarV is a category II (parent-pay) vaccine in China and the vaccination is voluntary. Since the schedule of 1-dose of VarV was recommended, VarV coverage increased rapidly and a substantial decrease in the incidence of varicella disease, varicella-related morbidity and mortality, and health care costs had been observed. Outbreaks of varicella, however, continued to occur in kindergartens or primary schools among children with high coverage of 1-dose VarV in Zhejiang Province. Some studies reported that approximately 15% of VarV recipients did not achieve protective levels of antibody and VarV induced immunity waned with time [1]. Besides, a large randomized trial indicated that the 2-doses VarV schedule was significantly more effective than the 1-dose VarV schedule [2]. In 2014, ZJCDC updated the recommendation regarding the use of VarV for varicella prevention. The latest recommendation includes a 3-pronged approach: (1) a routine 2-doses VarV schedule, with the 1st dose of VarV (VarV 1 ) administrated at 12-15 months of age and the 2nd dose of VarV (VarV 2 ) administrated at 3-4 years of age; (2) a VarV 2 catch-up vaccination for children and adolescents aged ď12 years who received VarV 1 previously and without a history of varicella disease; (3) all persons aged ě13 years without any evidence of immunity should be given 2 doses of VarV. According to the new recommendation, a minimum interval of 28 days between two doses of VarV is required.
At present, it is important to understand and continuously monitor the coverage of VarV among eligible age groups to guide future policy and intervention design. However, VarV coverage data of eligible children is not available as until now all the category II vaccines have not been included in the provincial level vaccination surveys. Therefore, little is known about the coverage of VarV by demographic characteristics such as gender or migration status. In fact, understanding of the coverage of VarV by these characteristics can provide valuable information for developing interventions tailored to subgroups with lower coverage.
The main objectives of this study included: (1) evaluating the coverage of VarV and exploring its determinants for children aged 2-6 years, and assessing the proportion of children who were vaccinated at the recommended age; (2) comparing the coverage of VarV with the coverage of measles containing vaccine (MCV), which had been included in the national immunization program (NIP).

Data Source
Data on VarV coverage were obtained from Zhejiang Provincial Immunization Information System (ZJIIS). ZJIIS, also known as the immunization registries, is a computerized, population-based system containing demographic and vaccination data for children aged <7 years living in Zhejiang Province since 2004 [3]. All vaccination clinics in Zhejiang Province must participate in ZJIIS. ZJIIS includes a client software deployed at each vaccination clinic and a database deployed in ZJCDC. The ZJIIS database consolidates data from different vaccination clinics through the Internet in real time and provides a tool for supporting effective vaccination strategies at the provider and program management level. Once any immunization clinic in Zhejiang Province is visited for the first time, children will be registered in ZJIIS with a unique identifier. Historical immunization information for migrant children is also included. Demographic information (such as name, date of birth, gender, address, phone number, immigration status) and vaccination records are collected by ZJIIS and these information will be updated in a day if there is any change.
Children aged 2-6 years (born from 1 January 2009 to 31 December 2013) and registered in ZJIIS were enrolled in this study. Appropriately anonymized individual records of target children were extracted from the ZJIIS database on 1 January 2016, including the VarV and MCV vaccination information. Children who were designated in ZJIIS as permanently inactive (i.e., deceased) or "moved or gone elsewhere" were excluded.

Definitions
The VarV 1 coverage was defined as the proportion of children who received VarV 1 , while the VarV 2 coverage was defined as the proportion of children with VarV 1 who received VarV 2 . Valid doses of VarV were defined as VarV 1 administered no earlier than 4 days prior to the first birthday, VarV 2 administered at least 28 days after VarV 1 , and either dose administered on the same day as or ě4 weeks after any other live attenuated vaccine. All the invalid doses were considered as "unvaccinated" and were excluded from analysis.
The data were analyzed by birth cohort to evaluate the coverage of VarV at population or geography (by city) level, and for comparing the coverage of VarV and MCV. 12-month wide cohort was used for analysis among different cities, gender or immigration status. These cohorts were children born between 1 January

Analysis Strategy
The coverage rates of VarV 1 and VarV 2 of the relevant birth cohorts were calculated separately, as well as the proportion of children who were vaccinated at the recommended age 12-15 months for the VarV 1 and 3-4 years for the VarV 2 . The VarV coverage was estimated among strata defined by cities, gender and immigration status. We evaluated separately the coverage of VarV 1 and VarV 2 in parallel to the coverage of the 1st and the 2nd dose MCV, which were scheduled at 8 months of age and 18 months of age, respectively. A χ 2 test for trends in different birth cohorts was performed to evaluate whether there was a significant increase in coverage of VarV over time. We adopted the χ 2 test to examine whether the coverage of VarV was significantly different across subgroup strata defined by gender or immigration status. We compared the coverage of MCV and VarV and evaluated the difference with the χ 2 test. We performed all analysis with SPSS version 13.0 (SPSS Inc., Chicago, IL, USA) and at a significance level of 0.05.

Ethical Considerations
This study was approved by the Ethical Review Board of ZJCDC (No. T-043-R). All the data were anonymous when we exported them from ZJIIS and kept confidential without individual identifiers.  (Table 1).

Overall Coverage of VarV
A rapid increase of the VarV 2 coverage was observed after the new VarV schedule was recommended in 2014. The VarV 2 coverage increased significantly from 31.8% for the 2009 birth cohort to 48.7% for the 2011 birth cohort. Across the three birth cohorts, a significant increase in the VarV 2 coverage over time was observed at both the provincial and city level (p < 0.01). There were significant disparities in VarV 2 coverage across the 10 cities for the three birth cohorts. The coverage rates of VarV 2 of the 2009 birth cohort in WZ, SX, JH, ZS were lower than the average rate of Zhejiang Province. The VarV 2 coverage rates of the 2010 birth cohort in WZ, SX, JH, QZ, ZS were lower than the average rate of Zhejiang Province. The VarV 1 coverage rates of the 2011 birth cohort in WZ, JH, QZ, ZS, and Taizhou (TZ) were lower than the average rate of Zhejiang Province (Table 2).

VarV Coverage across Subgroups
The VarV 1 coverage was 87.6% for males and 87.2% for females in the 2009-2013 birth cohorts, and the VarV 2 coverage was 38.3% for males and 37.3% for females in in the 2009-2011 birth cohorts. There was no significant difference in the VarV coverage between strata defined by gender in the 2009-2013 birth cohorts for the 1st dose, as for the 2nd dose ( Table 3).

Coverage Difference between VarV and MCV
The VarV 1 coverage increased from 84.8% for the 2009 birth cohort to 87.9% for the 2013 birth cohort, and it approached the coverage of the 1st dose of MCV (96.3%-97.8%). The VarV 2 coverage of in the 2009-2011 birth cohorts was relatively consistent and it was significant lower than the 2nd dose coverage of MCV (95.5%-96.2%) ( Table 5).

Discussion
Since the introduction of the 1-dose schedule of VarV, the VarV1 coverage increased significantly from the 2009 birth cohort to the 2013 birth cohort, approaching the goal of >90% for China's NIP vaccines, but it was still lower than the VarV1 coverage in the USA [4], where the VarV1 coverage was 96.2% (95%CI: 96.2%-96.3%) for children aged ≥4 years in 2015. The high VarV1 coverage in the USA is mainly attributed to the 1-dose varicella vaccination school-entry requirement in place since 1999, which was changed to a 2-doses varicella vaccination requirement in 2006 [5].

Discussion
Since the introduction of the 1-dose schedule of VarV, the VarV 1 coverage increased significantly from the 2009 birth cohort to the 2013 birth cohort, approaching the goal of >90% for China's NIP vaccines, but it was still lower than the VarV1 coverage in the USA [4], where the VarV 1 coverage was 96.2% (95%CI: 96.2%-96.3%) for children aged ě4 years in 2015. The high VarV 1 coverage in the USA is mainly attributed to the 1-dose varicella vaccination school-entry requirement in place since 1999, which was changed to a 2-doses varicella vaccination requirement in 2006 [5].
Although the population-based incidence of varicella was not available due to the fact that until now varicella is not a notifiable disease in China, the national public health emergencies reporting system shows a steady decrease in the outbreaks of varicella in children aged ď6 years in recent years.
Achieving a high 2-doses VarV coverage is critical for preventing varicella as the incremental vaccine effectiveness of 2-doses versus 1-dose of VarV was estimated as 63.6% [6]. Furthermore, even among the children who received VarV 1 before, receipt the VarV 2 also is important as the antibody level induced by VarV 1 may wane with time since administration of VarV, which increases the risk of breakthrough varicella [7]. In this study, the VarV 2 coverage was increasing rapidly. The results indicated the new recommendation of VarV vaccination was being quickly and widely accepted by physicians and caregivers and the VarV 2 coverage for the 2011 birth cohort was much higher than that for the older children. These findings also demonstrated the relative success of the introduction of the 2-doses VarV schedule. However, the VarV 2 coverage was still lower than that of VarV 1 , which potentially places unvaccinated children at risk for breakthrough varicella. Breakthrough varicella had been demonstrated as an impact factor in several prolonged outbreaks of varicella in high population intensity settings like schools or kindergartens, because breakthrough varicella is highly infectious and difficult to recognize [8].
Regional disparities in coverage were observed for both VarV 1 and VarV 2 . Children from HZ, NB, TZ, LS had a significantly higher VarV 1 coverage than that of children from other cities. Aside for the reason that VarV was a category II vaccine, this finding might be associated with the school-entry VarV 1 vaccination requirement in these four cities, in addition to the school-entry requirement only for NIP vaccine in Zhejiang Province. We also found that HZ, NB, JX, LS had a higher VarV 2 coverage than other cities. The VarV 2 catch-up campaigns targeting the children aged 4-6 years with VarV 1 which were conducted in 2015 in these four cities might explain the higher coverage of VarV 2 .
Children's age remained a determinant of the VarV 2 coverage, with a higher coverage seen for younger children in this study. The lower VarV 2 coverage among older children might represent the failure to be vaccinated through a catch-up schedule. Another plausible explanation was that some of the older children with VarV 1 might have had breakthrough varicella, but we could not verify this hypothesis from the available data. For children aged ě3 years with only VarV 1 and no history of varicella disease, the visits for the 4th oral polio vaccine at 4 years of age or the diphtheria and tetanus combined vaccine at 6 years of age may present an ideal time for VarV 2 catch-up. Besides, a primary school-entry VarV 2 requirement could be an efficient strategy to improve the coverage.
We did not find any difference in VarV 1 and VarV 2 coverage by children's gender, which was consistent with our previous study [9]. However, we found meaningful disparities in VarV 1 and VarV 2 coverage by children's immigration status. Lower coverage was observed in migrant children and it placed the migrant children at risk of varicella disease and of facilitating the transmission to other susceptible populations, as had been indicated for measles [10]. This result indicated that the utilization of vaccination services in migrant children was poorer than that in resident children, which could be associated with the lower demands for category II vaccines of the migrant caregivers and "service gap" that resulted from current allocation of public health resources. For migrant families, relative lower income and poor awareness of the importance of vaccination with category II vaccines are the main reasons for the lower demand for category II vaccines [9]. It means that a good income frees the household from the struggle of finding work to survive and would provide more spare time for caregivers to seek for public health service like immunization [11]. Besides, many of the migrant caregivers have a lower education level than residents, which results in a poorer understanding of the importance of vaccination of category II vaccines [11]. Furthermore, the public health resources, especially the vaccination nurses, are allocated according to the resident population with no consideration for the migrant population, which results in a "service gap". That means the vaccination service capacity is sometimes inadequate in areas where the volume of migrant children is large or dense. As we know, people believe that medical workers are an authoritative source of information and vaccination nurses can also make more precise recommendations on a case-by-case basis to children' caregivers when necessary. Unfortunately, most vaccination nurses working in vaccination clinics with a large scale migrant people do not have time to explain the importance of vaccinations to every caregivers. Therefore, the potential advantages of improving coverage posed through vaccination nurses' recommendations are not efficiently utilized in these areas. School-entry vaccination requirement, including regulation of 2-doses VarV, had succeeded in improving the coverage, irrespective of children's demographic characteristics such as the immigration status [12]. Therefore, the adoption and enforcement of a 2-doses VarV mandate in Zhejiang Province would not only be efficient in increasing the VarV 2 coverage, but also address the coverage disparity between migrants and residents.
The coverage rates of VarV were significantly lower than those of MCV for both the 1st and the 2nd dose across all the birth cohorts. These findings were consistent with our previous survey conducted in Yiwu in 2014 [9] and demonstrates the importance of vaccines that are free of charge when setting for high coverage goals [13]. A higher VarV coverage could be expected if VarV were included in China's NIP as a category I vaccine. The proportion rates of children who received vaccination at the recommended age of VarV 1 and VarV 2 were increased rapidly over time. The main reason for this was the wide application of ZJIIS in the routine vaccination service in Zhejiang Province. ZJIIS allows the generation of alerts that reminded providers that a vaccination is due or overdue. ZJIIS makes the remind/recall strategy more feasible and efficient, and ensures up-to-date vaccination. Additionally, the parents also had the access to their children's vaccination records and information on the recommendation of vaccines via application programs derived from ZJIIS.
There are three limitations in our study. First, our analysis relied on the vaccination records of children registered in ZJIIS and the children who were not registered might have a lower coverage for all vaccines. As a result, we could have overestimated the coverage. Second, the ZJIIS data provides limited demographic information about the children and their families, so we could not explore other determinants of the coverage of VarV that were mentioned in the previous studies, such as caregivers' education level, or socio-economic status. Third, we could not correlate the coverage of VarV with varicella incidence since varicella was not presently a notifiable disease in China.

Conclusions
Our study showed a rapid increase in the VarV 2 coverage, indicating a growing acceptance of the 2-doses VarV schedule by children's caregivers and physicians after the new recommendation was released. Further decline in the incidence of varicella disease can be expected as a higher 2-doses VarV vaccination coverage is achieved. Although our findings were encouraging, the coverage of VarV is still below the goal set for the NIP vaccine, placing school-age children at high risk of varicella or outbreaks. The findings of our study highlighted the necessity of school-entry VarV vaccination requirements to achieve high coverage (>90%) and to eliminate disparities in coverage among subgroups. We also recommended that the coverage of VarV be continuously monitored via ZJIIS to identify any challenges to the current VarV vaccination program.