Challenges to Achieving Measles Elimination, Georgia, 2013–2018

Controlling measles outbreaks in the country of Georgia and throughout Europe is crucial for achieving the measles elimination goal for the World Health Organization’s European Region. However, large-scale measles outbreaks occurred in Georgia during 2013–2015 and 2017–2018. The epidemiology of these outbreaks indicates widespread circulation and genetic diversity of measles viruses and reveals persistent gaps in population immunity across a wide age range that have not been sufficiently addressed thus far. Historic problems and recent challenges with the immunization program contributed to outbreaks. Addressing population susceptibility across all age groups is needed urgently. However, conducting large-scale mass immunization campaigns under the current health system is not feasible, so more selective response strategies are being implemented. Lessons from the measles outbreaks in Georgia could be useful for other countries that have immunization programs facing challenges related to health-system transitions and the presence of age cohorts with historically low immunization coverage.

defined as persons aged <15 years, adults, as persons aged >15 years. We analyzed the proportion of cases of different age groups by the age group of potential sources. Χ 2 test was used for statistical comparisons. In the analysis of the costs of measles outbreaks, the expenses were converted into USD using exchange rates at the time when they were incurred.
Various indicators are used to assess sensitivity, specificity, timeliness and completeness of measles surveillance (1,2). The main indicators reviewed in this report are included in the Appendix Table 2.

Immunization Coverage Survey: Methods
Below is the summary of the background and methods of the immunization coverage survey conducted in Georgia in 2015-2016 applicable to coverage with measles-mumps-rubella vaccine (MMR). The full report is available online (4). Immunization coverage in Georgia had been high until 1990, but declined in the 1990s, during the immediate period after the regaining of independence and subsequent armed conflicts and economic crisis. Although immunization services have improved in the last decade, challenges remain, as demonstrated by continued occurrence of outbreaks of measles. As of 2015, at the time of planning the survey, national coverage estimates for the first and second doses of measles-containing vaccines (MCV1 and MCV2, respectively) reported by Georgia to WHO (Appendix Table 3) remained largely below the national target of 95% (6).
The accuracy of administrative coverage data was unclear because of difficulties with determining target populations, particularly in the cities where the continuous changes to health care system had greatest impact on primary health care facilities (HCFs). The abolition of geographic catchment areas for HCFs, intense population movement, and existence of uncertain number of children not registered with HCFs resulted in greater difficulties with assessing coverage in large cities than in smaller towns and rural areas. Administrative coverage data have not been validated for over a decade, as no independent nationwide coverage surveys have been conducted in Georgia since a Multiple Indicator Cluster Survey (MICS) in 1999 (8).
Immunization data could not be analyzed for the MICS survey conducted in 2005 because immunization cards for ~85% of households were not stored at home (8).
Because of the lack of independent validation of the coverage data in Georgia and ongoing uncertainty with target populations, we conducted a nationwide immunization coverage survey during 2015-2016 to assess coverage with vaccines included in the routine immunization schedule through 5 years of age.

Survey Population and Vaccine Doses Assessed
Most standard protocols for immunization coverage surveys (MICS, DHS, EPI cluster survey) only include vaccines given during the first year of life and first dose of measles vaccine, but this approach leaves out later doses, such as second dose of measles-containing vaccines, and doses after primary series for diphtheria-tetanus toxoids and polio vaccines. The coverage with vaccine doses recommended after 12 months of age in Georgia has not been independently assessed previously. Therefore, we decided to assess coverage with all vaccines included in the immunization schedule before the age 6 years.
Per NCDC request, and because of greater uncertainties with accuracy of reported coverage data in cities, the survey was designed to allow obtaining separate estimates for three largest cities of Georgia -Tbilisi (2015 population 1,100,000), Batumi (154,000), and Kutaisi (148,000), which together account for 38% of total population of the country (3). Therefore, these three cities and rest of Georgia were surveyed separately and nationwide estimates were obtained by pooling the data from these surveys. The areas currently not under Georgian Government control (South Ossetia and Autonomous Republic of Abkhazia) were excluded because of lack of population data, inaccessibility and security concerns. Since very few families in Georgia keep their children's immunization cards at home (8) and parental recall is not considered a reliable source of a child's immunization history, we obtained information on immunizations from HCFs where children receive immunization services, in accordance with recently revised WHO guidance on conducting immunization coverage surveys (10). were initially sampled but were subsequently found to have moved overseas.

Design and Sample Size
A complex, stratified, multi-stage design was used for the survey (Appendix Table 4).
The country was divided into four survey domains consisting of the three largest cities (Tbilisi, Kutaisi, and Batumi) and the rest of the country. In the three large city domains, simple random sampling (SRS) was used to select children [primary sampling units (PSU)] from each of the three age groups. The fourth domain, consisting of the populations not residing in one of the three largest cities, was divided into seven strata. In the first stratum, which included Rustavi and Poti, participants within each age group were selected by SRS because the sampling frame had no easily identifiable subdivisions to be used as sampling units for cluster survey. Five strata required a two-stage cluster design. In the first stage, settlements (village/town) were selected by probability proportionate to population size (PPS), followed by an SRS of children within each age group. The last stratum, representing the remaining 54 districts of Georgia, required a 3stage cluster design. In the first stage, districts were selected by PPS, followed by selection of settlements (village/town) by PPS, followed by a SRS of children within each of the three age groups. Very small settlements were pooled to create sampling unit with >10 children in it. (1.2%) were found to have moved to other countries, resulting in 8,147 children eligible for the survey. We obtained immunization information for 7,723 of them for an overall enrollment rate of 94.5%, and 424 (5.2%) children could not be found. In all birth cohorts and domains, >90% of eligible participants were enrolled (range, 90.4%-98.0%).

Survey Procedures
The relevant population subsets were extracted from the Civil Registry birth registration database via the Immunization Management Module link. The residence codes were assigned to each administrative unit based on child's address. If actual address was different from the child's legal address, the actual address was used to assign the child to sampling unit, accounting for some population movement and reducing the proportion of children who could not be located.
Participant selection process was performed by survey coordinators. SRS was applied using an online random number generator (www.random.org). The survey field teams were given lists of selected children with their addresses and, if known, HCF indicated in the Immunization Management Module (the list and contact information of HCF is available through the Health Information Management System). For children with known HCFs, the teams visited HCFs to locate the immunization records of children selected for the survey.
If the child's immunization records could not be located at the listed HCF or no HCF was listed, the teams visited the child's residence and, after providing an information sheet about the survey, asked parents/guardians if the child had received at least one vaccination. If the answer was positive, parents/guardians were asked to provide information about HCF where the child receives immunizations. If the immunization card was available at home, the data were obtained on-site. Otherwise, the team visited the HCF indicated by a parent/guardian to obtain immunization records. If the child was unvaccinated per parent/guardian report, this information was noted in the interview form and no further attempts to locate records for this child were undertaken. Children who could not be found were not replaced by selecting another child. The survey field teams were comprised of personnel from NCDC, CDC/GID, CDC South Caucasus Office, FELTP graduates and from local Public Health Centers of survey areas. Before beginning fieldwork, the survey personnel received comprehensive training on the survey objectives, methodology, and procedures for data collection.

Data Management and Analysis
The statistical software Epi Info 7 was used for data entry. Analysis was conducted using SAS v9.4 and R v3.3. Analyses accounted for the complex survey design and sampling weights.
We report Wilson-Score confidence intervals for proportions using survey procedures in SAS The estimates of coverage were compared to the national target of 95% coverage for all doses. The target does not specifically refer to timely coverage, therefore, in the analysis we applied it to overall coverage by the time of the survey. The survey results were also compared to corresponding administrative coverage reported through GEOVAC system. GEOVAC, the existing system for administrative reporting of coverage in Georgia, is based on the data provided by HCFs to NCDC and only reflects children registered with HCFs.

Ethical Issues
The coverage survey protocol was reviewed by Human Subject Research Coordinator, GID/CGH/CDC and Ethical Committee, NCDC, and determined to be an evaluation of public health program rather than human subject research.