Combining serological and contact data to derive target immunity levels for achieving and maintaining measles elimination

Background Vaccination has reduced the global incidence of measles to the lowest rates in history. However, local interruption of measles virus transmission requires sustained high levels of population immunity that can be challenging to achieve and maintain. The herd immunity threshold for measles is typically stipulated at 90–95%. This figure does not easily translate into age-specific immunity levels required to interrupt transmission. Previous estimates of such levels were based on speculative contact patterns based on historical data from high-income countries. The aim of this study was to determine age-specific immunity levels that would ensure elimination of measles when taking into account empirically observed contact patterns. Methods We combined estimated immunity levels from serological data in 17 countries with studies of age-specific mixing patterns to derive contact-adjusted immunity levels. We then compared these to case data from the 10 years following the seroprevalence studies to establish a contact-adjusted immunity threshold for elimination. We lastly combined a range of hypothetical immunity profiles with contact data from a wide range of socioeconomic and demographic settings to determine whether they would be sufficient for elimination. Results We found that contact-adjusted immunity levels were able to predict whether countries would experience outbreaks in the decade following the serological studies in about 70% of countries. The corresponding threshold level of contact-adjusted immunity was found to be 93%, corresponding to an average basic reproduction number of approximately 14. Testing different scenarios of immunity with this threshold level using contact studies from around the world, we found that 95% immunity would have to be achieved by the age of five and maintained across older age groups to guarantee elimination. This reflects a greater level of immunity required in 5–9-year-olds than established previously. Conclusions The immunity levels we found necessary for measles elimination are higher than previous guidance. The importance of achieving high immunity levels in 5–9-year-olds presents both a challenge and an opportunity. While such high levels can be difficult to achieve, school entry provides an opportunity to ensure sufficient vaccination coverage. Combined with observations of contact patterns, further national and sub-national serological studies could serve to highlight key gaps in immunity that need to be filled in order to achieve national and regional measles elimination. Electronic supplementary material The online version of this article (10.1186/s12916-019-1413-7) contains supplementary material, which is available to authorized users.

Measles, a highly contagious immunising infection, could be a future target for 2 eradication. [1,2] Since the introduction of vaccination in the late 1960s, mortality and 3 morbidity from measles has declined drastically. [3] Nevertheless, outbreaks continue to 4 occur, and achieving regional elimination, or interruption of transmission, has been 5 challenging. [4] 6 Control of measles is achieved through vaccination in early childhood, and the 7 vaccine is part of routine immunisation schedules worldwide. Typically, vaccination 8 targets are established for the level of coverage required to achieve "herd immunity", or 9 the level of population immunity necessary to prevent outbreaks occurring. [5] For 10 measles, this level has been found to be in the range of 90-95%. [6] Strictly speaking, 11 however, any target based on vaccination coverage only affects the current and future 12 birth cohorts. With sufficient susceptibility in older age groups, outbreaks can occur 13 even at high levels of vaccination coverage. To assess the ability of a country or region 14 to achieve and maintain elimination, that is the sustained absence of endemic 15 transmission, immunity levels must therefore be considered across all age groups. These 16 levels are affected by historical and current routine vaccination coverage, but also by 17 vaccination campaigns and past outbreaks that conferred natural immunity. 18 For this reason, in the late 1990s, the World Health Organization (WHO) European 19 Region (EURO) derived age-specific target immunity profiles, or the levels of immunity 20 necessary in different age groups to achieve elimination. [7] These profiles are widely 21 applied within and occasionally outside Europe. Based on a basic reproduction 22 number (or number of secondary cases produced by a typical infective in a totally 23 susceptible population) of 11, it was recommended to ensure that at least 85% of 1-4 24 year olds, 90% of 5-9 year olds and 95% of 10 year olds and older possess immunity 25 against measles.
[8] These immunity targets are distinct from recommendations on 26 vaccination coverage levels. Gaps in immunity can exist despite high routine coverage if 27 coverage targets were not met in the past, or because of population migration. 28 Immunity targets reflect the effect of susceptibility in all age groups and highlight the 29 potential need for campaigns to close any gaps in immunity. 30 The aforementioned target immunity levels derived in the late 1990s were based on 31 assumed age-specific contact patterns based on pre-vaccination measles epidemiology in 32 England and Wales. Since then, much work has gone into better quantifying the 33 amount of transmission-relevant contact occurring between different age groups. Diary-based studies have been conducted across Europe [9,10], as well as in 35 Vietnam [11] China [12], Uganda [13], Zimbabwe [14] and elsewhere. While other 36 methods for measuring social contact patterns exist [15][16][17], contact data from diary 37 studies have become the de facto standard for studying age-specific infectious disease 38 dynamics. Mathematical models of transmission based on these observed patterns have 39 consistently outperformed those based on homogeneous mixing. [18][19][20] 40 Here, we aimed to evaluate current guidelines on target immunity levels for measles 41 taking into account contact patterns observed in diary studies. To this end, we 42 combined the observed age-specific social mixing patterns with observed or hypothesised 43 immunity levels to calculate a contact-adjusted immunity, akin to the mean level of 44 immunity across the population but taking into account that some age groups have 45 more contact with each other than others. We validated this method by testing the 46 extent to which contact-adjusted immunity levels based on serological studies conducted 47 around in the late 1990s / early 2000s could have been used to predict the case load in 48 the following decade. We then calculated hypothetical contact-adjusted immunity levels 49 if previously recommended immunity levels were achieved in a range of settings where 50 contact studies have been undertaken, and assessed whether these levels would have 51 been sufficient for achieving and maintaining elimination. Lastly, we compared these 52 results to alternative scenarios of greater or lower immunity than the recommendation 53 to test whether an alternative recommendation would be more justified once mixing 54 patterns were taken into account.

57
We considered the annual number of measles cases reported by each country to WHO. 58 We used serological studies conducted in 17 countries of the WHO EURO as part of the 59 ESEN2 project to determine immunity levels at the times of the studies [25]. Equivocal 60 samples were interpreted as positive as in the original study, but we also tested 61 scenarios where they were removed from the sample or interpreted as negative. We took 62 into account uncertainty by drawing from the individual samples using a 63 bootstrap (n = 100), and using the re-sampled immunity levels with re-sampled contact 64 matrices to estimate contact-adjusted immunity. Since contact studies were not available for all countries in ESEN2, contact studies from representative countries were 66 used where necessary (for mediterranean countries: Italy; for Eastern European 67 countries: Poland; for Sweden: Finland; for Ireland: Great Britain). 68 We used diary studies available on the Zenodo Social Contact Data 69 Repository (https://zenodo.org/communities/social_contact_data), to 70 determine contact matrices for 17 countries and the Hong Kong Special Administrative 71 Region of China [26][27][28][29][30], and a further study conducted in five countries of South East 72 Asia.

73
Contact-adjusted immunity 74 To calculate the effective reproduction number R, we used an age-structured SIR-type 75 model [21,22]. In this model, the force of infection λ i experienced by age group i is the 76 sum of the forces of infection exerted on those in age group i by those in the same and 77 all other age groups: where λ ij is the force of infection exerted by age group j on age group i, β ij is the 79 infection rate, or the rate at which individuals in age group i contact individuals out of 80 a total number N j in age group j and become infected if these are infectious, and I j is 81 the number of infectious people in age group j. This formulation of the force of 82 infection assumes that the rate of infection between two random individuals depends on 83 their ages only, and that the probability of a given contacted member of age group j to 84 be with someone infectious depends on population-level prevalence of infection only.

85
The infection rate β ij can be further split, where p Inf is the probability that a contact between an susceptible and infectious person 87 leads to infection, here assumed age-independent, and φ ij is the number of contacts an 88 individual of age group j makes with those of age group i per unit time.

89
The basic reproduction number R 0 is defined as the spectral radius (or largest 90 eigenvalue) of the next-generation matrix (NGM) K [23] In the age-structured SIR-type model, the elements of the next-generation matrix K are 92 where q is a scale factor that, in the simplest case, is the probability of infection upon where, again, ρ denotes the spectral radius and K is a matrix with elements In classical mathematical epidemiology in a well-mixed population, the relationship 98 between the basic reproduction number R 0 and the effective reproduction number R is 99 where r is the proportion of the population that is immune. We interpret as contact-adjusted immunity, that is the equivalent of population immunity once 101 age-specific contact patterns are taken into account.

102
Homogeneous mixing approximation 103 An assumption of homogeneous mixing is equivalent to assuming that φ ij = δn j , that is 104 the rate of contact of group i being with group j depends only on an overall level of 105 contact δ and the proportion n j = N j /N of the population that are in group j, 106 N = N j being the overall population size. This, in turn, means that the infection 107 rate is β ij = δp inf n j and the force of infection (Eq. 1) is independent of age group: with q = p Inf D Inf . This matrix has rank 1 (as all rows are equal) and its only non-zero 113 eigenvalue is given by the trace: If the proportion immune of those in age group i is r i , the elements of K are and where r is the proportion of the population that is immune, recovering the expression 117 used above to define contact-adjusted immunity.

118
Contact matrices 119 We established contact matrices from diary studies conducted in a range of different 120 settings using a bootstrap, randomly sampling P individuals with replacement from the 121 P participants of a contact survey. We then determined a weighted average d ij of the 122 number of contacts in different age groups j made by participants of each age group i, 123 giving weekday contacts 5/2 times the weight of weekend contacts. We further obtained 124 symmetric matrices, i.e. ones fulfilling c ij n i = c ji n j by rescaling This gave the elements of the contact matrix φ ij = c ij /T , scaled by the time period T 126 over which contacts were measured (usually 24 hours).

127
Differences in contact patterns (due to factors such as cultural difference, schooling, 128 population density or demography) could be expected to underlie differences in the 129 value of the basic reproduction number R 0 between countries [24]. It is unclear, though, 130 whether such differences could be measurable in diary studies, or whether it is masked 131 by differences in study design and data collection. Because of this, we tested two models 132 of contact: one where the basic reproduction number R 0 is independent of the measured 133 contact matrix, and one where it scales with the mean number of contacts (weighted by 134 the population in each age group), such that contact-adjusted immunity becomes where c is a contact scale factor, calculated as the spectral radius of a given contact 136 matrix divided by the mean of the spectral radii across countries.

137
Predicting outbreaks from seroprevalence data 138 We established population-level immunity levels from seroprevalence data and estimated 139 contact-adjusted immunity by combining them with observed contact patterns. We used 140 a threshold for the derived contact-adjusted immunity levels to classify countries as following the seroprevalence study. We also calculated the Brier score, a so-called proper 145 forecasting score, to determine the predictive ability of the seroprevalence studies when 146 using the probability of being above the established threshold as a probability for 147 experiencing outbreaks. The Brier score is defined as where N is the number of observations (here: N = 17, the number of countries in 149 which seroprevalence was established in the ESEN2 study), f t is the probability of Overall, the 17 countries that took part in the ESEN2 study reported 59,494 measles 159 cases to WHO in the 10 years following the serological studies. The number of cases 160 experienced by individual countries varied widely ( Fig. 1  Contact-adjusted immunity levels estimated based on the serological profiles (taking 165 equivocal samples to be positive) were better correlated with the case load than plain 166 immunity levels (i.e., population-averaged immunity not taking into account age-specific 167 contact patterns). Comparing the immunity levels with the maximum number of cases 168 per million in the 10-year period yielded negative correlation of 0.42 (Spearman's rank 169 correlation, 90% credible interval (CI) 0.27-0.53, p = 0.12) for contact-adjusted 170 immunity and 0.23 (90% CI 0.17-0.29, p = 0.37) for plain immunity. Notable outliers in 171 the correlation between immunity levels and case load were Latvia (contact-adjusted 172 immunity 67%, plain 85%, 16 cases over 10 years), Lithuania (contact-adjusted 173 immunity 86%, plain 94%, 58 cases) in one direction, and Spain (contact-adjusted 174 immunity 95%, plain 99%, > 3000 cases) and Israel (contact-adjusted immunity 94%,  To test the predictive ability of estimated seroprevalence levels in combination with 181 age-specific mixing we split the countries into those that experienced large outbreaks in 182 the 10 years following the serological studies and those that did not. We set the  Fig. 1). We then tested different threshold immunity 185 levels (ranging from 80% to 99%, in increments of 1%) and classified countries as being 186 at risk of outbreaks or not based on whether their estimated immunity levels fell below 187 the threshold or not.   With alternative scenarios, the reproduction numbers changed (Fig. 3B-E). Raising 207 immunity in under-5-year olds by 5% to 90% would increase adjusted immunity levels   Contact-adjusted immunity in different theoretical scenarios, with age-specific mixing as measured in diary studies. Each column represents one of the scenarios of age-specific immunity (top), with differences between the settings given by their different mixing patterns. Scenarios from left to right: A) Current target levels. B) 5% higher immunity in under 5 year olds. C) 5% higher immunity in 5-9 year olds. D) 5% lower immunity in 10-14 year olds. E) 5% higher immunity in 5-9 year olds and 5% lower immunity in 15-19 year olds.
immunity in 5-to-9-year olds by 5% to 95% would sharply increase contact-adjusted 211 immunity. In this scenario, all countries would have 5% or less probability of being at 212 risk of outbreaks, with 16 out of 18 at less than 1% risk (exceptions: Hong Kong 5%, 213 Netherlands 3%).

214
In scenarios immunity in 5-to-9 year olds was raised but a gap in immunity was 215 introduced in older generations contact-adjusted immunity dropped below the threshold 216 level of 93% in some settings. A scenario of reduced immunity in 10-to-14-year olds by 217 5% to 90% while retaining higher immunity in younger age groups resulted in elevated 218 risks of outbreaks in 13 out of 18 countries. A scenario of reduced immunity in 14-to-19 219 year olds by 5% to 90% while retaining higher immunity in younger age groups resulted 220 in elevated risks of outbreaks in 11 out of 18 countries.

222
Taking into account age-specific mixing patterns and applying these to immunity levels 223 observed across Europe, we were better able to predict outbreaks than by considering immunity alone. This improvement was not large enough to be considered certain or 225 statistically significant given the relatively small sample size. Yet, combined with 226 previous evidence that observed age-specific mixing improve the accuracy of 227 mathematical models, we believe that there is a strong case for taking these into 228 account when interpreting the results of serological studies [18][19][20].

229
A threshold of 93% contact-adjusted immunity was found to be the best predictor of 230 outbreaks in the subsequent decade, with approximately two-thirds of countries 231 correctly assessed to either be facing large outbreaks or not. In the absence of any more 232 detailed information on setting-specific basic reproduction numbers, however, such a 233 threshold will only ever be an approximation. On the other hand, setting-specific 234 parameters are difficult to establish, are subject to method-specific biases and can span 235 a wide range of values [24,35].

236
In principle, country-specific reproduction numbers would, to some degree, depend 237 on the frequency and types of contact within the population and should therefore be 238 amenable to measurement in contact studies such as the ones used here. However, 239 scaling estimated susceptibility levels with the relative number of contacts reported in 240 each study gave results almost identical to the simpler version not using such scaling.

241
These results indicate that differences in survey methodology mask any such difference 242 in contacts that would be reflected in the value of R 0 . We argue that aiming to achieve 243 93% or greater contact-adjusted immunity in a population is a pragmatic choice that 244 can be informed by measurable quantities, that is age-specific immunity levels and 245 mixing patterns.

246
Current guidelines on target immunity levels are based on estimates derived almost 247 20 years ago, and were based on assumed mixing patterns matched to pre-vaccination 248 data from England and Wales. We have used transmission models in combination with 249 recently observed age-specific contact patterns from a variety of European and some 250 non-European settings to assess whether these guidelines are sufficient for achieving 251 measles elimination. We investigated a range of settings with different demographic 252 profiles and cultural contexts: from high-income settings characterised by low birth 253 rates and an ageing population (e.g., Germany or the United Kingdom) to having 254 more (Vietnam) or less (Taiwan) recently undergone the demographic transition to low 255 birth rates, or characterised by a high birth rate and young population (Uganda).

256
With observed mixing patterns, several settings were found to be at risk of 257 outbreaks even if they achieved previously recommended target immunity levels, 258 including ones with very different demographic profiles. Achieving 95% immunity in 259 5-to-9 year olds, on the other hand, would reduce transmission sufficiently to achieve 260 elimination in all except the most extreme scenarios.

261
The importance of immunity levels in 5-to-9 year olds presents both a challenge and 262 an opportunity: Levels as high as 95% in this age group can only be maintained Where there were less stringent vaccination requirements at school entry, more case of 276 measles were observed. [40] Analyses of pre-elimination measles outbreaks in the US 277 indicated that transmission occurred among highly vaccinated school-aged populations, 278 suggesting that higher population immunity levels are needed among school-aged 279 children compared to preschool-aged children. [41] It has been proposed that minimum 280 coverage levels as low as 80% at the second birthday of children may be sufficient to 281 prevent transmission among preschool-aged children in the United States if population 282 immunity is at least 93% among over-5 year olds. [42] 283 While our results stress the role of 5-to-9 year olds, they also highlight the 284 importance of not having gaps in immunity in older age groups. This is particularly 285 important close to elimination as a lower force of infection pushes cases into older age 286 groups. [43] Given the higher rate of complications of measles when experienced at older 287 age, ensuring immunity among adults will be important not only for interrupting 288 transmission, but also to prevent serious episodes of disease. [44] 289 Our study has several limitations. The delineation of countries into having 290 experienced outbreaks or not is somewhat arbitrary, if in agreement with a milestone 291 towards measles eradication established by the World Health Assembly [45]. Depending 292 on the local situation with respect to measles elimination, a country may decide to 293 apply less or more stringent immunity thresholds. Moreover, population immunity 294 represents past levels of vaccine coverage or natural infection which may not be 295 reflective of the future. For example, immunity may be high just after a major outbreak 296 but such outbreaks could occur again if coverage is sub-optimal. An important caveat is 297 therefore that seeing immunity sufficient to interrupt transmission does not guarantee 298 that elimination is maintained if current levels of coverage are insufficient.

299
Lastly, the contact-adjusted immunity levels we estimated from serological studies 300 did not always correctly predict where outbreaks could be expected. On the one hand, 301 Latvia and Lithuania did not experience large numbers of cases in spite of low levels of 302 contact-adjusted immunity. These two are among the smallest in our group of countries 303 for which we had serological data available and may be at lower risk of imported cases. 304 Still, they would have been expected to have seen more cases given the results of the 305 serological studies in 2003 and 2004, respectively. Latvia in particular reported 306 immunity levels as low as 76% among all age groups and 62% in 5-to-9 year olds in 307 2003, but only reported 16 cases of measles in the 10 years 2004-13. To our knowledge, 308 there were no supplementary immunisation activities that could explain the absence of 309 outbreaks. It would be of value to determine whether these countries are now at high 310 risk of large outbreaks in spite of having previously interrupted transmission, or whether 311 there were issues with the serological tests conducted.

312
Israel and Spain, on the other hand, experienced large numbers in spite of high levels 313 of contact-adjusted immunity. Two potential causes for this discrepancy suggest 314 themselves: First, drops in vaccination coverage as well as vaccination campaigns may 315 have changed the risk of outbreaks during the 10 years following the serological studies. 316 Second, serology based on residual and population-based samples may not always be 317 representative of relevant immunity levels. In Spain, a disproportionate number of cases 318 occurred in young adults [46], but there was nothing in the serological data to suggest 319 that this might be expected. Moreover, if those lacking immunity are preferentially in 320 contact with each other because they cluster socially or geographically, outbreaks could 321 occur in these groups, and population-level serology might not provide a good estimate 322 of realised immunity levels in outbreak settings. In Israel, outbreaks occurred in 323 orthodox religious communities with very low vaccination coverage. [47] 324 Further sub-national serological and epidemiological studies, particularly in 325 low-income countries at high risk of measles outbreaks, could generate key insights on 326 the relationship between immunity levels, heterogeneity of susceptibility and outbreak 327 risk. [49,50] At the same time, further studies of contact patterns across settings, 328 combined with models of such patterns where no data have been collected, will make it 329 possible to expand our results to other countries and regions. [51] Combined with 330 observations of contact patterns, these would serve to highlight key gaps in immunity 331 that need to be filled in order to achieve national and regional elimination and, 332 ultimately, global eradication of measles. 333