Estimating the basic reproduction number of measles in low-and middle-income settings using 172 seroprevalence studies: A modelling approach

Figures
Abstract
The basic reproduction number, R0, is an epidemiological measure to describe the transmissibility of infectious diseases and evaluate the potential effect of interventions. Measles R0 has historically been considered to be between 12–18, with contextual factors contributing to its heterogeneity, but has rarely been estimated using data across multiple countries. Our study aims to estimate measles R0 in low- and middle-income countries using a standardised database of population-based serosurveys. We fitted an age-structured compartmental model of measles transmission dynamics and vaccination (DynaMICE) to the age-specific seroprevalence data extracted from a recent systematic review. Using Markov Chain Monte Carlo, we estimated setting-specific posterior distributions of R0 in 172 studies with unique survey years and locations from 57 countries. Bootstrapped samples of R0 estimates were pooled by study characteristics, including survey period, geography, and overall bias in sampling, measurement, and reporting results. Measles R0 estimates varied substantially across serostudies, ranging from 0.93 (95% credible interval (CrI): 0.70–1.00) to 147 (95% CrI: 76.5–208), with fewer than 13% of studies having median R0 values in the range of 12–18. Pooled R0 estimates showed smaller medians and variation in serostudies conducted after 2000 or including the adult population, while no distinguishable variation was identified across the World Health Organization regions. Our revised estimates demonstrated the wide range of measles R0 across low- and middle-income settings and highlighted the importance of considering the heterogeneity in measles transmissibility when modelling epidemics and planning interventions and vaccination strategies.
Citation: Fu H, Sbarra AN, Russell T, Abbas K, Auzenbergs M, Jit M (2026) Estimating the basic reproduction number of measles in low-and middle-income settings using 172 seroprevalence studies: A modelling approach. PLOS Glob Public Health 6(7): e0006731. https://doi.org/10.1371/journal.pgph.0006731
Editor: Orvalho Augusto, University of Washington, UNITED STATES OF AMERICA
Received: July 4, 2025; Accepted: June 7, 2026; Published: July 17, 2026
Copyright: © 2026 Fu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The input data and computer code for this submitted manuscript are included in a public repository: https://doi.org/10.5281/zenodo.19166369.
Funding: This work was supported by the Gates Foundation and Gavi, the Vaccine Alliance (INV-034281 & previously OPP1157270 / INV-009125 HF, KA, MA, and MJ), via the Vaccine Impact Modelling Consortium (www.vaccineimpact.org). ANS also received funding from the National Institutes of Health (F31AI167535, T32AI165369). KA is supported by the Japan Agency for Medical Research and Development (JP223fa627004). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Measles is a contagious disease prone to epidemics and has resulted in a large health burden worldwide [1,2]. The basic reproduction number (R0) of measles is generally recognised as one of the highest amongst common vaccine-preventable diseases, with suggestions that each primary measles case can generate 12–18 secondary infections when the population is completely susceptible [3]. Due to this characteristic of high transmissibility, measles requires a high level of two-dose vaccine coverage for disease control. However, a systematic review of R0 estimates reveals that measles R0 varied widely from 1.43 to 770 by country settings [4]. This variation in measles transmissibility can affect the estimation of measles burden and vaccine impact in individual settings. However, it was unclear the extent to which this variation stemmed from the diversity in data sources or estimation methods across studies, as opposed to true heterogeneity.
Serosurveys examine the presence of specific antibodies to understand the immunity profile against infectious diseases within a population over a defined period of time. Because presence of these measles-specific antibodies is understood to be a correlate of protection from the clinical disease, seroprevalence data stratified by age, vaccination history, and other selected characteristics can be used to inform susceptibility gaps in the population and understand the heterogeneity of transmissibility across settings. With appropriate sampling and data collection methods, serosurveys provide a less biased snapshot of the population-level immunity profiles in the community, compared to notification data that are subject to under-ascertainment and underreporting issues in the surveillance systems [5].
To address the heterogeneity of measles transmissibility across low- and middle-income countries (LMICs) that contribute to the majority of global measles burden [2], we extracted age-specific seroprevalence data from serostudies included in a recently conducted systematic review [6]. We then inferred the setting-specific R0 by fitting a dynamic model of measles transmission and vaccination to the seroprevalence dataset [7–9]. The comparison of model-based R0 estimates by setting-specific characteristics will enhance the understanding of the key factors that contribute to the heterogeneity of measles transmissibility across settings.
Methods
Serological data
We obtained 221 articles from a 2024 systematic review by Sbarra et al. that summarised the scope and bias in population-level measles serosurveys [6]. Surveys that did not provide information on the age distribution of the sampled population (n = 16) were excluded. We also excluded survey data from non-representative populations, such as immunocompromised patients, pre-term babies, and refugees or border populations (n = 21). Based on the bias assessment in the previous review [6], we further excluded surveys having a critical bias in the validity of the tools used to measure serology (n = 19). We did not exclude serosurveys that were rated lower quality because they had insufficient information when reporting serological results, because a standard reporting style was not widely used, especially in early studies, and because reporting completeness would not necessarily reflect the implementation quality of serosurveys.
From the 165 included articles, a total of 172 unique studies were defined with a unique combination of country, survey year, and first author (Table A in S1 Appendix). We extracted 13 unique studies from 6 articles, including survey data for multiple years or multiple countries. Where only the publication years were available, we approximated the survey years using the average duration between the survey and publication years in studies with complete information (3.7 years). In mother-newborn studies examining both the samples from mother-newborn pairs, we extracted the data only from mothers. Most such studies reported a high correlation in the antibody concentration between mother and newborn. The age-specific seroprevalence estimates were derived from surveys conducted between 1963 and 2019 in 57 LMICs, with the majority found in China, India, Türkiye, and Brazil (Fig A in S1 Appendix). Most of the included studies (n = 132, 76.7%) consisted of seroprevalence estimates across multiple age groups, and only a small proportion did not cover children (n = 21, 12.2%).
Measles transmission and vaccine impact model
Dynamic Measles Immunization Calculation Engine (DynaMICE) is a compartmental model developed to capture measles transmission and assess the health impact of vaccination strategies at the country level. The model consists of four epidemiological states – maternally immune, susceptible, infectious, and recovered (Fig B in S1 Appendix), further stratified by the doses of measles-containing vaccines (MCV) received. DynaMICE is incorporated with fine-scale age structure, including 156 weekly age groups for children aged under 3 years old and 98 yearly age groups for those aged between 3 and 100 years old. In addition, modelling the force of measles infection takes into account annual seasonality via a sine function, pathogen transmissibility via R0, and age-dependent mixing pattern via a contact matrix. We assumed the incubation period for measles to be 14 days and immunity following infection to be lifelong [10]. The proportion of newborns with maternal immunity was directly informed by the proportion of immune adults at child-bearing age (18–35 years old, ‘recovered’ compartments). We derived country-specific demographics from the United Nations World Population Prospects [11]. Country-specific synthetic contact matrices were obtained from Prem et al. [12] and modified to match the age structure of DynaMICE.
In modelling measles vaccination in DynaMICE, we applied country-specific vaccination schedules and coverage reported in the World Health Organization (WHO) Immunization Data portal [1]. Routine MCV1 and MCV2 doses were provided at the country-specific recommended age, and supplementary immunisation activities were implemented as one-off campaigns in the targeted age groups in the middle of the calendar year. We assumed vaccine effectiveness for MCV1 to vary linearly by age [13], two-dose vaccine protection to be capped at 98% [14], and no further protection from three or more MCV doses. Details of model structure, equations, parameters, and key assumptions have been included in the previous publication [8].
Model fitting
We fitted DynaMICE to each unique set of age-specific seroprevalence data per country-time period using Markov Chain Monte Carlo (MCMC) with component-wise Metropolis algorithm and random order for updating parameters in each blocked iteration [15]. Four MCMC chains for each study were conducted with starting parameters randomly drawn from their prior distributions. Three parameters―R0, duration of maternal immunity protection, and vaccine effectiveness for the first MCV dose―were fitted, and they were assumed to be constant over time within the same serostudy. We selected less informative prior parameter distributions to cover a variety of settings for the 172 included serostudies (Table 1). We extracted prior distributions for R0 and vaccine effectiveness from the systematic reviews [4,13] that gathered all the available evidence, and for the duration of maternal immunity, we used information from a longitudinal study with multiple follow-ups [16]. For 69 serosurveys conducted prior to country-specific MCV introduction, we removed the parameter for vaccine effectiveness and fitted the seroprevalence data with a two-parameter model. We constructed the likelihood by assuming the seroprevalence in each age group to be independently binomially distributed. We conducted diagnostic checks for convergence of MCMC chains and goodness-of-fit, with a targeting effective sample size above 200. We obtained 500 posterior samples for each study and calculated the medians and 95% credible intervals (CrI) of fitted parameters.
In our model-based outputs, we added up the non-susceptible populations ― maternally immune, infectious, and recovered sub-populations (compartments) [8] ― to calculate the seroprevalence in the corresponding age groups and calendar years. We included all the non-susceptible states because IgG antibodies, as the detection target in serosurveys, cannot distinguish between maternally acquired, infection-acquired and vaccine-induced immunity. Models were first simulated using the country-specific demographics in 1980 [11] and assumed no implementation of vaccination until they reached equilibrium. From 1980 to the survey year, we applied annual inputs of demographics and MCV coverage estimates [1] (Fig C in S1 Appendix). For serosurveys conducted before 1980, we extracted the model estimates in 1980, considering that seroprevalence was likely similar during earlier years, given low coverage of vaccination and other control measures for measles.
Factors associated with R0
Measles R0 is influenced by demographic, socioeconomic, and environmental factors, which change contact behaviours and transmission risk per contact in the population [4]. Although the DynaMICE model has accounted for country-specific age structure and age-dependent contact patterns, other factors could further contribute to the temporal and geographical variation in measles R0. In addition, the survey quality and available details of seroprevalence data also affect the estimation of R0 through fitting to seroprevalence data. To understand these dependencies, we evaluated the association between R0 estimates and different study features, including survey period, age of the survey population, number of age-specific data points, and potential methodological biases. We used the bias assessment outcomes in Sbarra et al. [6], which evaluated the processes of sample selection, assay protocol, handling of equivocal results, and other information on serology testing methodology presented in the original serostudies under systematic, pre-defined algorithms. Serostudies were categorised into low, moderate, severe, and critical levels of biases in sampling, measurement and reporting serology. Unweighted bootstrapping was used to pool R0 estimates and compare their distributions across the selected study features. For each selected feature, we filtered a subset of serostudies, and from their posterior samples, we randomly drew 10,000 bootstrapped samples with replacement.
We assessed the association of median R0 for each study with average household size [17], population density [11], and Human Development Index (HDI) [18]. HDI is a country-specific and time-dependent summary measure combining the dimensions of health and longevity, education, and living standard. When these data were missing at specific survey years, we performed linear interpolation for household size and for HDI and population density, we used the estimate at the closest year on a country basis.
Estimated measles cases
Using the posterior parameters of R0, duration of maternal immunity, and vaccine effectiveness, we estimated the annual numbers of measles cases during the survey year and the two years before and after. We calculated the mean annual cases over the five years to reduce the influence of year-to-year fluctuations in the time series. We compared the model-based case estimates with the WHO country-specific reported cases [19] to assess their consistency in measuring measles burden and the potential issue of underreporting. We also compared our model-based estimates with measles incidence estimates from the Global Burden of Disease (GBD) Study 2021 [20], as it is a key publicly available resource for health decision makers.
Sensitivity analysis
We investigated the goodness-of-fit of the two-parameter and three-parameter models by calculating the root mean square error (RMSE) between the observed seroprevalence data and median posterior predictions. As the number of age data points is different in each serostudy, the RMSE metric enables comparison across studies by averaging the total error size. To assess how model fit affects the estimation of R0, we ranked the RMSE and conducted a sensitivity analysis on the pooled R0 results after removing 20% of the studies with the highest error size.
Results
Posterior R0 estimates
We conducted MCMC diagnostics and presented parameter trace plots, prior and posterior parameter distributions, pairwise parameter correlations, effective sample sizes, and Gelman-Rubin statistics in Figs D–G in S1 Appendix. We obtained 500 sets of posterior parameters from the converged MCMC chains. The posterior model estimates of age-specific seroprevalence show a good fit using the RMSE measure (Fig H in S1 Appendix), with two-thirds of the studies having an average difference less than 10% between the model results and observed data (S1 Fig). There was a positive correlation between median R0 and vaccine effectiveness estimates (Spearman’s correlation coefficient (R)=0.3007, p-value (p)=0.0020). However, the correlation became weaker after excluding studies with a severe or critical bias in study population selection, serological measurement and result reporting (Fig I in S1 Appendix, R = 0.2558, p = 0.0485). No significant correlation between R0 and duration of maternal immunity was found (R = 0.0122, p = 0.8739).
Fig 1 presents the median and 95% CrIs of the R0 posterior distributions obtained from fitting the DynaMICE model to age-specific seroprevalence data in 172 individual studies. We observed a right skewed distribution of R0 estimates, ranging widely across studies from 0.93 (95% CrI: 0.70–1.00) to 147 (95%CrI: 76.5–208). Only 12.8% (n = 22) of the median estimates were in the commonly cited range of 12–18 [3], while 75.0% (n = 129) of them presented a value below 12 and 12.2% (n = 21) above 18. In 15.2% of the included studies (n = 26), posterior R0 showed a broad range of 95% CrIs, with the difference between lower and upper bounds exceeding 10. These R0 estimates with a broad range tended to result from serosurveys with a smaller sample size, fewer age-specific data points, or conducted in the pre-vaccination era. Geographically, R0 estimates concentrated in a range of less than 15 based on serosurveys conducted in the WHO Eastern Mediterranean Region, while in other regions, a few upper R0 estimates could reach 30 and above. The variation in R0 estimates remained large between studies in the same country, as seen in median R0 estimates dispersing over a range of 20 in China, India, Brazil, and Türkiye (Fig J in S1 Appendix).
The median posterior R0 estimates are presented with 95% credical intervals, ordered from highest to lowest in each WHO region. Arrows indicate that the upper R0 estimates exceed 40. Two studies (JAM-1986 and IND-1983) with the lower bound estimate above 40 are not shown. See the study reference list in Table A in S1 Appendix.
Factors associated with R0
Across different study periods, age groups, and bias levels of serosurveys, pooled R0 estimates mostly presented a right-skewed distribution and often peaked at a value <10 (Fig 2). However, R0 estimates based on the same study feature could still have wide variation, as seen in the multimodal curves of pooled R0 estimates in studies with a larger number of seroprevalence data points and a lower bias in sampling, measuring, and reporting measles serology. Our findings suggested that the survey period and age of populations may be associated with the peaks and shapes of pooled R0 distributions. The median R0 estimates were 8.99 (interquartile range (IQR)=10.4) in the serostudies conducted prior to 1990, 12.8 (IQR = 14.0) in 1991–2000, 4.48 (IQR = 3.62) in 2001–2010, and 5.17 (IQR = 3.73) in 2011–2019, showing a smaller value and narrower variability in the estimates based on more recent studies. A wider variation and a greater value were also found in R0 estimates based on studies involving only children (median = 13.2, IQR = 11.6) compared to those with both children and adults (median = 5.19, IQR = 4.42). We also inspected six serostudies reporting R0 below 1 and found they were conducted in near-elimination settings where MCV1 coverage level tended to be high and close to the overall seropositivity in the population. On the other hand, two serostudies with the highest R0 values (>50) were conducted in the pre-vaccination era, where high seroprevalence in very young children was observed. The factors affecting the overall variation in R0 remained important when we looked at influential factors for variation within countries. However, across different countries, the associated factors often had different directions of effect on R0 (Fig K in S1 Appendix).
Density of 10,000 bootstrapped R0 estimates by each selected feature is presented.
In exploring other country- and year-specific factors in the broader context, we identified a weak negative correlation (R = –0.2787, p = 0.0002) between the posterior median R0 and Human Development Index, suggesting that measles transmissibility decreased with improvement in health and socioeconomics. We also found a weak positive correlation between median R0 and average household size (R = 0.2437, p = 0.0015). Nonetheless, no significant correlation was found with population density (R = –0.0361, p = 0.6385). Based on the univariable log-linear regression analysis on R0, HDI was a significant predictor but with limited explanation of the variance (R-squared = 3.93%, Fig 3).
Regression coefficients and R-squared metrics were derived based on univariable log-linear models. Where available, data were matched to the year each serosurvey was conducted.
Estimated measles cases
WHO data on reported measles cases was not available, especially in the early years when case surveillance and notification systems had not yet been fully operated. Among 155 studies included for comparison (Fig L in S1 Appendix), 60% of our median model-based estimates (n = 93) show greater numbers than the WHO reported cases. Assuming these model-based estimates reflect the actual magnitude of measles incidence, we found that a proportion of 1.27% measles cases were reported to WHO (25th–75th percentiles: 0.56%–2.94%). However, in 36 studies for comparison, the upper bound model estimates reveal < 100 cases, while the WHO data reported > 1000 cases. We also noted that the majority of serostudies where model-estimated measles cases were lower than the WHO-reported cases were conducted in China (n = 23) and having severe bias in measuring and reporting serological results.
The comparison with the GBD 2021 Study was available for 134 serostudies (77.9%) conducted between 1988–2019 (Fig M in S1 Appendix). Our model-based case estimates generally differed from the GBD incidence estimates, with 84 serostudies (62.7%) having uncertainty intervals that did not overlap with those from GBD. Compared to GBD, our model was more likely to present lower case estimates and the scale of difference in median estimates was notable (lower by >80% for all 62 serostudies with non-overlapping interval estimates).
Sensitivity analysis
Fitting the DynaMICE model to the age-specific seroprevalence data, we presented a median RMSE of 6.16% (25th–75th percentiles: 2.40%–11.6%). Those serostudies with a poorer fit were with more age-specific data points for fitting, conducted in earlier years, and less likely to have low or moderate bias in overall serosurvey quality, but no particular trend was seen by WHO regions (Table B in S1 Appendix). We conducted a sensitivity analysis restricting the analysis to the 80% best-fitting studies, with RMSEs below 15.2%. Compared to the main analysis with a full set of studies, pooled R0 estimates of the restricted set generally showed similar point estimates and wide intervals for WHO regions, study periods, age groups, and serological measurement and reporting biases (Fig 4).
Median and 95% credible intervals of pooled R0 estimates from studies with each selected feature are presented. Arrows indicate the upper estimates exceed 50.
Discussion
This analysis systematically estimated measles R0 using published age-specific serological data across LMICs. Based on dynamic model simulation of measles transmission and vaccination, we identified great variability in R0 across 172 study settings. Our model-based measles R0 estimates were mostly below the commonly cited range between 12–18 and demonstrated large heterogeneity even within the same countries or WHO regions. The value of R0 tended to be higher in serosurveys conducted prior to 2000 or including only the children population, and it was also found to be negatively associated with the Human Development Index. Furthermore, the serology-informed R0 estimates we applied to quantify the measles incidence suggested a low reporting of disease burden in the surveillance systems.
The great variability in measles R0 identified in our analysis was also reported in a comprehensive review by Guerra et al. [4]. Both our analysis and Guerra’s review found that most R0 estimates to be outside of the commonly cited range of 12–18, which was derived from a single study taking early data in England and New York [3]. However, Guerra et al. [4] summarised estimates from independent modelling studies based on surveillance and/or serology data sources, and it was therefore not able to separate to what extent the R0 variability could be attributed to the methodological differences. Another multi-country analysis based on the same model structure reported more than five times greater measles transmission rates in the United Kingdom than in Ethiopia, but only two LMICs were included, and the homogenous mixing assumption limited further inspection of the heterogeneity of transmissibility [21]. Instead, our analysis included all the available serological data in LMICs conducted prior to 2019 and utilised the DynaMICE model incorporated with fine-scale age structure and country-specific demographics and contact patterns. With a systematic framework for serosurvey selection and model parameterisation, we improved the quantification of measles R0 heterogeneity and assessment of potential contributing factors.
Advancements in healthcare, education and living standards could partly explain the variability in measles R0, as suggested in previous studies and our regression analysis with HDI [4]. However, we did not find a correlation between R0 and average household size and population density. In DynaMICE, the effect of household structure and crowdedness on age-related mixing across countries may have been partially adjusted through the incorporation of synthetic social contact matrices, which were constructed with contact diary surveys and demography data [8,22]. In such cases, other indicators like clustering of susceptibility or proportion of large-sized households could be more informative to explain the complicated mechanism of household transmission [23,24]. In addition to selecting appropriate indicators, addressing the interactions between them is essential to disentangle the heterogeneity in measles transmissibility. An advanced understanding of factors associated with the R0 heterogeneity will help characterise measles transmission in settings where serodata are unavailable [23] and update disease burden under evolving settings.
We estimated annual numbers of measles cases by leveraging the strength of serological data in capturing disease incidence regardless of symptom occurrence and access to healthcare [25]. The comparison between our serology-based burden estimates and WHO notification data suggested that fewer than 5% of cases were reported in most study settings. Underreporting was also identified in a previous modelling study [26], and the WHO notification data were prone to underestimation due to low propensity to seek care in settings with weak healthcare systems, low case ascertainment, preferential reporting of severe cases, and variable surveillance sensitivity. We also compared our measles case numbers with GBD 2021, the only publicly available incidence estimates at the country level. We found limited alignment in the magnitude of disease burden between the two studies, largely owing to methodological differences―the GBD estimates were derived from a mixed-effects linear regression model with routine and campaign MCV coverage as covariates and fitted to the WHO notification data [27]. As our seroprevalence-based case estimates are subject to several limitations, the comparison with the WHO and GBD estimates of measles cases should be seen as exploratory only.
Our study has several limitations. First, we assumed that each serosurvey was nationally representative and used country-level demography and vaccine coverage inputs for model fitting. However, serosurveys were usually conducted in geographically restricted populations due to time, cost, and personnel constraints, especially in LMICs [28]. Whilst high-spatial-resolution estimates of routine MCV coverage are available for addressing subnational heterogeneity [29], local data on campaign coverage, contact patterns, and population structure are still limited in LMICs. In many serostudies included in our analysis, reported information was insufficient in defining the geographic areas of the catchment population [6]. A mismatch of population representativeness could lead to either overestimation or underestimation of measles R0, as subnational coverage can be greater or lower than national coverage (Fig N in S1 Appendix). The interaction between coverage and other model inputs could further complicate any prediction on the direction of bias in R0 estimation. Nonetheless, our exploratory analysis of a small number of serostudies (n = 44) showed that the pooled R0 estimates did not change substantially when we removed serostudies in settings with >10% difference between national and subnational coverage levels (Fig O in Appendix).
Second, we assumed that R0 and social contact rates were constant over time. Our pooled R0 analysis by study period suggested a potential temporal trend. Assuming a constant R0 could underestimate measles transmissibility in the years before 2000 but overestimate it in the later years. Even though we addressed some of the temporal effects in DynaMICE through yearly inputs of population age structure, social and environmental factors can still contribute to temporal variation in measles transmissibility. For example, factors such as urbanisation, education, infrastructure improvement, and major public health events like COVID-19 may alter measles transmission and contact behaviour in particular settings [4,30]. The temporal trend of R0 is likely to be non-linear, sensitive to occasional events, and jointly driven by multiple time-varying components. Repeated cross-sectional serosurveys in the same population could facilitate the characterisation of the temporal trends in measles transmissibility.
Third, the serostudies included in our analysis used different laboratory methods and antibody levels to define seropositivity, in part likely due to weak evidence for an antibody threshold correlating to protection against measles infection [31]. Although we excluded studies with a critical bias in measuring seropositivity [6], we were not able to consider the sensitivity and specificity of the laboratory methods because the original serostudies gave limited information on whether laboratory protocols were specifically followed, and on sample quality at time of testing. We also did not perform sensitivity analyses across studies that used different thresholds to define seroprotection. Should information on serological test performance be incorporated into model fitting, the updated R0 estimates could change in either direction, depending on the relative values of test sensitivity and specificity. In addition to the performance of serological tools, serosurveys with sufficient age-specific data points and a broad age range (e.g., covering both children and adults) will help address the changing vaccination strategies over time and reduce bias and uncertainty in estimating measles R0.
Fourth, we did not consider the waning of measles IgG antibodies in the population, which could lead to underestimation of R0 in matching seroprevalence in older adults. However, existing observations showed that the decrease in IgG antibody concentration is slow and hence will likely have a small effect on seroprevalence [32]. In LMICs, where measles transmission is mostly endemic, high-level immunity in the population could be sustained with infection. Thus, the effect of immunity waning in adults is less of a concern in most of the serostudies included in this analysis, except for those in the near-elimination settings.
In conclusion, we have comprehensively explored measles R0 across LMICs by fitting to the available age-specific seroprevalence data extracted from a systematic review, accounting for country-specific demographic characteristics, contact patterns, and immunisation programmes with the DynaMICE model. We have illustrated the significant variability in measles R0 estimates across different settings. We demonstrated that the variation in R0 is not purely driven by differences in model structure, but also dependent on serosurvey features and country-specific demographic and developmental characteristics. Further research to improve the quality, granularity, and completeness of data collection for serostudies and other model inputs will enhance the estimation of measles R0 and understanding of its key determinants. Our findings highlight the importance of heterogeneities at the country level and subnational level for planning measles interventions and vaccination strategies in LMICs. Local R0 values are crucial to identify the speed at which epidemics spread and the level of immunity needed to prevent transmission. Setting-specific serological survey data should be used to support epidemiological analyses and decision-making wherever available.
Supporting information
S1 Appendix. Supplementary methods and results.
https://doi.org/10.1371/journal.pgph.0006731.s001
(PDF)
S1 Fig. Observed data and model estimates of age-specific seroprevalence.
Study reference is included in S1 Appendix.
https://doi.org/10.1371/journal.pgph.0006731.s002
(PDF)
References
- 1.
World Health Organization. Immunization data. Accessed 2023 November 20. https://immunizationdata.who.int/listing.html
- 2. Minta AA, Ferrari M, Antoni S, Lambert B, Sayi TS, Hsu CH. Progress toward measles elimination - worldwide, 2000-2023. MMWR Morb Mortal Wkly Rep. 2024;73(45):1036–42. pmid:39541251
- 3. Anderson RM, May RM. Directly transmitted infections diseases: control by vaccination. Science. 1982;215(4536):1053–60. pmid:7063839
- 4. Guerra FM, Bolotin S, Lim G, Heffernan J, Deeks SL, Li Y, et al. The basic reproduction number (R0) of measles: a systematic review. Lancet Infect Dis. 2017;17(12):e420–8. pmid:28757186
- 5. Gibbons CL, Mangen M-JJ, Plass D, Havelaar AH, Brooke RJ, Kramarz P, et al. Measuring underreporting and under-ascertainment in infectious disease datasets: a comparison of methods. BMC Public Health. 2014;14:147. pmid:24517715
- 6. Sbarra AN, Cutts FT, Fu H, Poudyal I, Rhoda DA, Mosser JF, et al. Evaluating scope and bias of population-level measles serosurveys: a systematized review and bias assessment. Vaccines (Basel). 2024;12(6):585. pmid:38932314
- 7. Auzenbergs M, Fu H, Abbas K, Procter SR, Cutts FT, Jit M. Health effects of routine measles vaccination and supplementary immunisation activities in 14 high-burden countries: a Dynamic Measles Immunization Calculation Engine (DynaMICE) modelling study. Lancet Glob Health. 2023;11(8):e1194–204. pmid:37474227
- 8. Fu H, Abbas K, Klepac P, van Zandvoort K, Tanvir H, Portnoy A, et al. Effect of evidence updates on key determinants of measles vaccination impact: a DynaMICE modelling study in ten high-burden countries. BMC Med. 2021;19(1):281. pmid:34784922
- 9. Verguet S, Johri M, Morris SK, Gauvreau CL, Jha P, Jit M. Controlling measles using supplemental immunization activities: a mathematical model to inform optimal policy. Vaccine. 2015;33(10):1291–6. pmid:25541214
- 10.
Strebel PM, Papania MJ, Gastañaduy PA, Goodson JL. Measles vaccines. In: Plotkin SA, Orenstein WA, Offit PA, Edwards KM, editors. Plotkin’s Vaccines. Seventh ed. Elsevier; 2018. 579-618.e21.
- 11.
United Nations Population Division. World population prospects 2024. https://population.un.org/wpp/. Accessed 23 July 2024.
- 12. Prem K, van Zandvoort K, Klepac P, Eggo RM, Davies NG, Centre for the Mathematical Modelling of Infectious Diseases COVID-19 Working Group. Projecting contact matrices in 177 geographical regions: an update and comparison with empirical data for the COVID-19 era. PLoS Comput Biol. 2021;17(7):e1009098.
- 13. Hughes SL, Bolotin S, Khan S, Li Y, Johnson C, Friedman L, et al. The effect of time since measles vaccination and age at first dose on measles vaccine effectiveness - A systematic review. Vaccine. 2020;38(3):460–9. pmid:31732326
- 14. Sudfeld CR, Navar AM, Halsey NA. Effectiveness of measles vaccination and vitamin A treatment. Int J Epidemiol. 2010;39 Suppl 1(Suppl 1):i48-55. pmid:20348126
- 15. Metropolis N, Rosenbluth AW, Rosenbluth MN, Teller AH, Teller E. Equation of State Calculations by Fast Computing Machines. J Chem Physics. 1953;21(6):1087–92.
- 16. Leuridan E, Hens N, Hutse V, Ieven M, Aerts M, Van Damme P. Early waning of maternal measles antibodies in era of measles elimination: longitudinal study. BMJ. 2010;340:c1626. pmid:20483946
- 17.
United Nations Population Division. Household size and composition. 2022. https://www.un.org/development/desa/pd/sites/files/undesa_pd_2022_hh-size-composition.xlsx
- 18.
United Nations Development Programme. Human development index. 2024. https://hdr.undp.org/sites/default/files/2023-24_HDR/HDR23-24_Statistical_Annex_HDI_Table.xlsx
- 19.
World Health Organization. Provisional monthly measles and rubella data. Accessed 2024 June 26. https://www.who.int/teams/immunization-vaccines-and-biologicals/immunization-analysis-and-insights/surveillance/monitoring/provisional-monthly-measles-and-rubella-data
- 20.
Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2021 (GBD 2021) results. Institute for Health Metrics and Evaluation; 2022. https://vizhub.healthdata.org/gbd-results/
- 21. Trentini F, Poletti P, Merler S, Melegaro A. Measles immunity gaps and the progress towards elimination: a multi-country modelling analysis. Lancet Infect Dis. 2017;17(10):1089–97. pmid:28807627
- 22. Prem K, Cook AR, Jit M. Projecting social contact matrices in 152 countries using contact surveys and demographic data. PLoS Comput Biol. 2017;13(9):e1005697. pmid:28898249
- 23. Papadopoulos T, Vynnycky E. Estimates of the basic reproduction number for rubella using seroprevalence data and indicator-based approaches. PLoS Comput Biol. 2022;18(3):e1008858. pmid:35239641
- 24. Kuylen E, Willem L, Broeckhove J, Beutels P, Hens N. Clustering of susceptible individuals within households can drive measles outbreaks: an individual-based model exploration. Sci Rep. 2020;10(1):19645. pmid:33184409
- 25. Haselbeck AH, Im J, Prifti K, Marks F, Holm M, Zellweger RM. Serology as a tool to assess infectious disease landscapes and guide public health policy. Pathogens. 2022;11(7):732. pmid:35889978
- 26. Eilertson KE, Fricks J, Ferrari MJ. Estimation and prediction for a mechanistic model of measles transmission using particle filtering and maximum likelihood estimation. Stat Med. 2019;38(21):4146–58. pmid:31290184
- 27. GBD 2021 Causes of Death Collaborators. Global burden of 288 causes of death and life expectancy decomposition in 204 countries and territories and 811 subnational locations, 1990-2021: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. 2024;403(10440):2100–32. pmid:38582094
- 28. Thompson KM, Odahowski CL. Systematic review of measles and rubella serology studies. Risk Anal. 2016;36(7):1459–86. pmid:26077609
- 29. Local Burden of Disease Vaccine Coverage Collaborators. Mapping routine measles vaccination in low- and middle-income countries. Nat. 2021;589(7842):415–9. pmid:33328634
- 30. Delamater PL, Street EJ, Leslie TF, Yang YT, Jacobsen KH. Complexity of the Basic Reproduction Number (R0). Emerg Infect Dis. 2019;25(1):1–4. pmid:30560777
- 31. Bolotin S, Hughes SL, Gul N, Khan S, Rota PA, Severini A, et al. What is the evidence to support a correlate of protection for measles? A systematic review. J Infect Dis. 2020;221(10):1576–83. pmid:31674648
- 32. Bolotin S, Osman S, Hughes SL, Ariyarajah A, Tricco AC, Khan S, et al. In elimination settings, measles antibodies wane after vaccination but not after infection: a systematic review and meta-analysis. J Infect Dis. 2022;226(7):1127–39. pmid:35417025