We have shown that reported MCV coverage, which is determined using clinic-registered children as the denominator, yields a higher coverage estimate than 5 qualitatively different coverage assessment methods that were obtained during a measles outbreak investigation. Because officially reported coverage estimates in China do use clinic-registered children as the denominator, reported coverage is likely to be higher than actual coverage. This higher-than-actual reported coverage may provide a partial explanation why indigenous circulation of measles has continued in China despite more than 25 years of a 2-dose MCV vaccination policy, the strategy of using measles SIAs, and high reported coverage with MCVs.
Conducting SIAs is useful to address coverage inequities and rapidly close population immunity gaps in targeted age groups. This has been demonstrated both in China and elsewhere in the world [9-12]. However, SIAs should not be considered superior to routine immunization [13], because they provide vaccination in an intermittent manner, allowing for accumulation of susceptible children between campaigns. Experience from previous elimination programs has demonstrated that a 2-dose measles vaccine policy has been highly successful in achieving and maintaining measles elimination status [14-16]. In the United States, there have been three efforts to eliminate measles targeting 1967, 1982, and 1996. Over the years, the U.S. experienced several failures, but systematically incorporated the lessons learned from each failure into subsequent efforts, and finally achieved the goal in 2000 [17,18]. Key lessons learned from the efforts include: (1) elimination requires very high MCV vaccination coverage by age 2 years, (2) a second dose of measles vaccine is needed to achieve satisfactory levels of immunity, and (3) coverage assessment is crucial [17,19].
In this outbreak, the 2010–2012 birth cohorts, which became age-eligible for measles vaccination following the 2010 campaign, had the highest incidence of measles. Fully 80% of the 108 MCV-eligible cases were unvaccinated. This pattern raised our interest in providing alternative estimates of true coverage as a check on the high reported coverage that was calculated using doses administered data with clinic-registered children as the denominator. Each result of the five independent measurement methods revealed that coverage of 2010, 2011, and 2012 birth cohorts was below 85%, which is much lower than the 95% target objective needed to eliminate measles, and much lower than coverage that was officially reported.
Several recently-published and unpublished studies also showed that low coverage of new birth cohorts after the 2010 campaign was a key factor contributing to China’s measles resurgence [8]. These outbreaks were due to failure to provide measles vaccine for children at recommended ages. Although the 2011 national survey showed high coverage for both MCV1 and MCV2 at the national level, low-coverage areas were also identified in the survey [5], indicating that “pocket areas” exist. A recently-published coverage survey also showed that coverage among a migrating population was lower than among local populations [20].
To eliminate measles in China, learning from other elimination programs and from domestic measles outbreaks is critically important. Several strategies are important for China. First, increasing and maintaining high (≥95%) coverage of both MCV1 and MCV2 through routine immunization is the top priority at this stage. The 1996–2009 birth cohorts (in some provinces back to the 1990 birth cohorts) have been targeted by SIAs, and many fewer cases have been reported from these age-group since 2011 [21]. The majority of new, potential susceptibles will come from new born children. Maintaining high coverage through routine immunization can minimize the number of susceptibles, and consequently avoid periodic epidemics. China is undergoing a large urban migration, leading to over 236 million migrant workers in 2012 [22], which complicates the administration of vaccines given in series and given during times of family movement. Conducting outreach to migrant children becomes an additional need for closing immunity gaps.
Second, delivering both MCV1 and MCV2 to 8-month and 18-month old children on time is essential. Several studies indicated that children born to vaccinated mothers lose their maternal antibodies earlier than children whose mothers were immune through infection [23-25]. Seroprevalence data from this study also showed the trend of rapidly decreasing seroprevalence - from 60% in 0–1 months to the lowest level of 18.2% in infants 6–7 months of age. China provides MCV1 at one of the youngest ages globally – 8 months. This young age is designed to protect children as early as possible, prior to risk of getting infection. To avoid future outbreaks, it is crucial to attain high coverage levels by timely vaccination [20], so that herd immunity can be robust enough to protect children too young to vaccinate.
Third, frequent monitoring of coverage is important to identify program areas that need to be strengthened. Outbreaks of measles can also be used to find program weaknesses, but outbreaks are lagging indicators. In contrast, coverage can be a leading indicator to find areas in need of additional effort before an outbreak occurs [17]. The unrealistically high coverage reported is not consistent with China’s current measles epidemiology. Having higher-than-actual reported coverage can undermine confidence in vaccination strategies by giving false sense that the programme cannot improve coverage, and may lead to a lack of understanding why outbreaks are occurring.
Fourth, coverage assessments should strive to include all children in an area. A purpose of coverage measurement is to identify areas at risk of disease outbreaks. It would not have been possible to use reported coverage to predict the L County outbreak because reported coverage ranged from 99.4% to 99.9% for each of the birth cohorts affected by the outbreak. Most likely, the restriction of the denominator to clinic-registered children resulted in missing children in the area who should have been registered in the clinic and vaccinated. Given the large urban migration in China, including all children in an area in coverage assessments will be an important component in an effective measles elimination strategy. When a measles outbreak occurs in a community, an initial step for public health officials can be to use population estimates to determine whether the number of children registered in community clinics is consistent with population estimates. If inconsistent, community based surveys using rigorous methodology may be indicated [26].
Fifth, enhancing outbreak analysis and response activities to close immunity gaps will be needed to eliminate measles. Given the imbalance of socioeconomic development and immunization program capacity, small area SIAs are likely to be needed as a supplement to routine immunization in less developed areas. These SIAs depend on the epidemiology of measles and identification of immunity gaps. Surveillance and outbreak analysis can provide additional information to determine the extent to which missed opportunities to vaccinate occur, and then decide which types of actions should be taken [27].
Limitations
This study has several limitations, primarily the limitations of each of the 5 methods. The administrative method used Statistical Year Book birth records, and therefore would miss in- and out-migration among young children. The house-to-house survey is a convenience sample of neighborhood- and age-matched children, and is not representative of the entire county. Since these children did not get measles despite living near cases, their coverage may have been higher than the county average. The clinic review will miss children who never registered with the clinic, and can include children who moved away from the area. The VE equation method relies on the stability of the VE estimate. The serological survey was a convenience sample of children seen in the 2 hospitals, and previous studies comparing commercially available EIA assay versus the gold standard plaque reduction–neutralization assay have demonstrated that EIA was less sensitive, but is a reliable identifier of measles-seronegative individuals [28]. Although each of the 5 methods used in this study is imperfect and has advantages and disadvantages, all the 5 results come to the same conclusion that actual coverage is lower than reported coverage that is determined with a clinic-based denominator. We feel that a key strength of this study is the convergence of results from 5 qualitatively different methods. We therefore believe that our conclusion is supported by the evidence provided.
Recommendation
We recommend review and evaluation of the methods for estimating officially-reported vaccination coverage levels in China, with a goal of identifying feasible coverage assessment methods that will provide useful information for the immunization program in China. Methods that include children not registered in immunization clinics should be sought.