Triangulation of measles vaccination data in the United Kingdom of Great Britain and Northern Ireland

Bull World Health Organ 2019;97:754–763 | doi: http://dx.doi.org/10.2471/BLT.18.229138 Research 754 Introduction A considerable amount of immunization coverage and sur- veillance data are available nationally, regionally and glob- ally.1 Often, however, these data could be better used to aid decision-making on national and subnational immunization programmes. Particularly where a disease is close to being eliminated and the remaining few percent of susceptible indi- viduals are being targeted, the need for accurate data increases as vaccine coverage increases.2 Synthesizing data from two or more sources (i.e. data triangulation) is a pragmatic approach to optimizing the use of existing data, thereby improving data quality and gaining insights into the performance of vaccine programmes.1 In this study, we used the example of measles in England to illustrate how the triangulation of routine data sources, namely different sources on coverage of the combined measles, mumps and rubella (MMR) vaccine and measles incidence data, can help evaluate data quality and provide estimates of population immunity, which can be used to inform a national measles elimination strategy. As these data sources are available in most settings and for many diseases, with varying degrees of granularity and quality, our approach should be broadly replicable. Measles is a viral infection transmitted by the respiratory route and one of the most contagious human diseases.3 In 2016, approximately 90 000 deaths were attributable to the disease globally,4 down from more than 550 000 in 2000 thanks to the accelerated roll-out of measles immunization programmes.4 The commitment to eliminate measles (and rubella) is an important part of global efforts to improve health and reduce inequality. To achieve and maintain elimination, the World Health Organization (WHO) recommends that countries attain 95% coverage with two doses of measles-containing vaccine by the time children are 5 years of age.5 Modelling suggests that measles can be eliminated in most contexts if the proportion of children younger than 5 years susceptible to measles is less than 15% and the proportion of susceptible individuals aged 5 years and older is less than 5%.6 In England, before measles vaccine was introduced in 1968, 160 000 to 800 000 cases of measles were notified and around 100 deaths from acute disease were recorded each year.7 After the combined MMR vaccine was introduced in 1988, coverage rapidly reached 90% and disease incidence fell to a very low level. In 1994, a large catch-up programme was undertaken with measles–rubella vaccine and in October 1996, a second MMR vaccine dose was added.7 However, in 1998 a British doctor published a now-discredited study suggesting a link between MMR vaccine and autism.8 The resulting intense media interest had a substantial impact on MMR vaccine cov- erage, which dropped to about 80% in the late 1990s and early 2000s and took many years to recover.9 Since this fall, several catch-up campaigns have been implemented to address gaps in population immunity. Measles cases continued to rise and in 2006, endemic transmission became re-established in the United Kingdom of Great Britain and Northern Ireland. The incidence peaked at 3.2 per 100 000 population in 2012 and decreased to 1.0 per 100 000 in 2016.10 The United Kingdom is committed to measles elimination and has developed a national strategy in line with the European Vaccine Action Plan 2015–2020.11,12 Since 2012, coverage for the first MMR vaccine dose in children aged 24 months has been consistently over 90%.9,10 In 2016 and 2017, coverage for the first dose in children aged 5 years reached 95% for the first time.9 Since 2016, imported cases of measles have led to several outbreaks, with some limited spread in the population, particularly among individuals who missed the MMR vaccine when they a National Infection Service, Public Health England, 61 Colindale Avenue, London NW9 5EQ, England. Correspondence to Michael Edelstein (email: Michael.edelstein@phe.gov.uk). (Submitted: 27 December 2018 – Revised version received: 7 June 2019 – Accepted: 19 June 2019 – Published online: 23 August 2019 ) Triangulation of measles vaccination data in the United Kingdom of Great Britain and Northern Ireland Michael Edelstein,a Joanne White,a Antoaneta Bukasa,a Vanessa Salibaa & Mary Ramsaya Objective To illustrate how data triangulation involving routine data sources can optimize data usage and provide insights into vaccine programme effectiveness by considering measles vaccination and disease incidence data in England. Methods We obtained data on measles, mumps and rubella (MMR) vaccine coverage in birth cohorts from 1985 to 2016 from child health records and adjusted for under-ascertainment and catch-up campaigns. We assumed that the population had no natural immunity and that vaccine effectiveness was 95% for one dose and 99.75% for two doses. Vaccinations done outside the routine schedule and in people who entered England after the age of immunization were identified from primary care records. Measles susceptibility was defined as the percentage of individuals who were not immune despite all vaccination activities. We triangulated measles susceptibility and incidence data. Findings Median susceptibility was 4.6% (range: 1.2–9.2). Among cohorts eligible for two MMR vaccine doses, those born between 1998 and 2004 were most susceptible. Measles incidence was highest in these cohorts. Data from primary care and child health records were comparable for cohorts after 2000, suggesting that little supplementary vaccination took place. For cohorts before 2000, primary care data quality was insufficient for accurately estimating coverage. Conclusion Triangulating routine data on measles vaccination coverage and disease surveillance provided new insights into population immunity and helped identify vulnerable groups, which was useful for prioritizing public health actions to close gaps in immunity. This approach could be applied in any country that routinely records vaccine coverage and disease incidence. 755Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policyMichael Edelstein et al. were younger and in undervaccinated communities.13 Overall, the incidence of measles in England quadrupled between 2017 and 2018.14 Routine coverage and surveillance data in England Vaccine coverage in England is esti- mated by two methods. The first uses data from local Child Health Informa- tion Systems, which provide data to the cover of vaccination evaluated rapidly (COVER) programme.15 The second involves an online platform called ImmForm that automatically extracts immunization data from approximately 95% of primary care facilities (i.e. gener- al practices).16 The COVER programme is designed to collect data on coverage by specific target ages (i.e. by children’s second and fifth birthdays) and because it includes the entire population eligible for vaccination, provides the most ac- curate estimates of coverage at the time of data collection. As coverage is not routinely assessed again, COVER data will not accurately reflect the current status of a given birth cohort many years later. Vaccines given at an older age may not be recorded and neither the numera- tor nor denominator in the coverage calculation will include individuals who arrived in England after their fifth birthday. Nevertheless, COVER data are used for reporting to WHO on the WHO and United Nations Children's Fund’s (UNICEF’s) Joint Reporting Form. By contrast, ImmForm data rep- resent vaccine coverage recorded by general practices at the time of data extraction and include anyone in a specific birth cohort who was registered with the practice at that time (i.e. 2017 to 2018 for our study). Consequently, the data should cover vaccinations given either through routine vaccination programmes, during national catch-up campaigns, opportunistically, or outside England to any individual of any age. The accuracy of ImmForm data depends on the quality and completeness of clini- cal coding at each facility, these charac- teristics are known to have the greatest influence on immunization data quality globally.1 In the United Kingdom, clini- cians are legally required to report sus- pected measles cases to the public health services. In addition, national surveil- lance systems require all suspected cases to be confirmed by laboratories using either an immunoglobulin-M antibody test on a serum or oral fluid specimen or a polymerase chain reaction technique.11 Data triangulation Data triangulation involves the synthesis of two or more data sources with the aim of assisting programme planning and decision-making. The process can identify and address limitations in any single data source or data collection method. In addition, deeper insights can be achieved by examining comple- mentary data and putting them into a broader context. A recent report from a WHO-commissioned expert group on immunization data quality recom- mended that data triangulation should become the default approach to data analysis and use in its Expanded Pro- gramme on Immunization.1 The aim of our study was to use the example of measles elimination in England to demonstrate how data sources that are routinely available as part of the Expanded Programme on Immunization can be synthesized to improve vaccination data quality and to generate new information, for example on measles susceptibility and gaps in immunity, that can help guide decision- making on vaccine policy. Methods In calculating the proportion of the Eng- lish population susceptible to measles, we assumed a vaccine effectiveness of 95% for one MMR vaccine dose and 99.75% for two doses.17–19 In addition, we assumed there was no natural im- munity because the level of circulating disease in the country over the past 30 years was low. To estimate coverage of routine immunization for each birth cohort between the year from April 1985 to March 1986 (i.e. 1985–1986) and the year from April 2015 to March 2016 (i.e. 2015–2016), we used COVER data on the first and second MMR vaccine doses. Data collected at children’s fifth birthdays were generally available for birth cohorts from 1992–1993 until 2012–2013. We used data collected at the second birthday for individuals in birth cohorts after 2012–2013, who were too young during our study period to have had coverage of two doses assessed at 5 years of age, and for individuals in birth cohorts before 1992–1993, who were born before the second dose was included in the vaccination schedule. We applied a 50% coverage underestimate correction factor to COVER estimates. This percentage was based on a study that examined the extent of underes- timation by checking the vaccination status of individuals not recorded as vaccinated in Child Health Information Systems.20 Further, to consider the pos- sibility that the 50% correction factor was too high, we examined the effect of a 25% correction factor in a sensitivity analysis. Several birth cohorts included in the study were eligible for supplemen- tary immunization in national catch-up campaigns (Table 1). Coverage data have been published for the 1994 and 2013 campaigns;21,22 for other campaigns, we used the best estimates from Public Health England (unpublished data). Fig. 1 describes how we determined the level of protection from measles among cohorts that were eligible for participa- tion in catch-up campaigns. To ascertain the level of opportunistic vaccination after the routine vaccination age and outside of catch-up campaigns, we tri- angulated vaccine coverage data from the COVER programme (i.e. routine vaccination data only) with data from ImmForm, which potentially captures any vaccinations given up to the time of data extraction. For each birth cohort, we calculated susceptibility (S) to measles, which was defined as the percentage of individuals in the birth cohort who had not been vaccinated or who were probably not immune despite routine, supplementary or opportunistic vaccination, using the equation: S X Y= − ⋅ + ⋅( ) ( )[ ]100 0 9975 0 95. . (1) Where X is the percentage of birth cohort who received ≥ 2 vaccine doses and Y is the percentage of birth cohort who received 1 dose. Then, using the size of the population in each age band in 2017 obtained from the Office of National Statistics,23 we calculated the number of susceptible individuals in the population and overall population susceptibility. Susceptibility in each birth cohort was compared to the target immunity level required to keep the reproductive number (R0) below one and, therefore, interrupt transmission in the population (R0 is the number of 756 Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policy Michael Edelstein et al. Ta bl e 1. M ea sle s v ac cin at io n co ve ra ge a nd in di vi du al s s us ce pt ib le to m ea sle s, by b irt h co ho rt , E ng la nd , 1 98 5– 20 16 Bi rt h co ho rt a Ap pl ica bl e ca tc h- up ca m pa ig n Ro ut in e va cc in at io n co ve ra ge ,c % Ad ju st ed ro ut in e va cc in at io n co ve ra ge ,d % Pr op or tio n su sc ep tib le to m ea sle s,e % Es tim at ed n o. of in di vi du al s i n co ho rt in 2 01 7 Es tim at ed n o. of su sc ep tib le in di vi du al s Im m un ity le ve l s uffi cie nt to in te rr up t tr an sm iss io n? f Na m e an d da te b Co ve ra ge , % Fi rs t M M R va cc in e do se Se co nd M M R va cc in e do se Fi rs t M M R va cc in e do se Se co nd M M R va cc in e do se 20 15 –2 01 6 N on e N A 91 .2 N A 95 .6 N A 9. 2 67 4 80 7 61 9 47 Ye s 20 14 –2 01 5 N on e N A 91 .6 N A 95 .8 N A 9. 0 67 5 04 5 60 6 87 Ye s 20 13 –2 01 4 N on e N A 91 .4 N A 95 .7 N A 9. 1 68 2 35 6 61 9 92 Ye s 20 12 –2 01 3 N on e N A 94 .9 87 .2 95 .4 93 .6 3. 0 69 9 25 0 20 8 13 Ye s 20 11 –2 01 2 N on e N A 95 .0 87 .6 97 .5 93 .8 2. 9 72 1 70 8 21 0 70 Ye s 20 10 –2 01 1 N on e N A 94 .9 87 .6 97 .4 93 .8 3. 0 70 7 07 5 21 2 12 Ye s 20 09 –2 01 0 N on e N A 94 .6 88 .4 97 .3 94 .2 3. 1 69 4 48 0 21 5 29 Ye s 20 08 –2 00 9 N on e N A 94 .4 88 .4 97 .2 94 .2 3. 2 68 4 37 0 21 9 00 Ye s 20 07 –2 00 8 M M R 20 08 U nk no w n (lo w ) 94 .2 88 .3 97 .1 94 .1 3. 3 68 9 76 9 22 7 62 Ye s 20 06 –2 00 7 M M R 20 08 U nk no w n (lo w ) 93 .5 87 .0 96 .8 93 .5 3. 6 66 7 81 8 24 0 41 Ye s 20 05 –2 00 6 M M R 20 08 U nk no w n (lo w ) 92 .4 84 .6 96 .2 92 .3 4. 2 65 4 36 6 27 4 83 Ye s 20 04 –2 00 5 M M R 20 08 U nk no w n (lo w ) 91 .5 83 .0 95 .7 91 .5 4. 7 62 7 40 7 29 4 88 Ye s 20 03 –2 00 4 M M R 20 08 U nk no w n (lo w ) 89 .9 80 .1 94 .9 90 .1 5. 5 61 6 97 5 33 9 34 N o 20 02 –2 00 3 M M R 20 08 a nd 20 13 U nk no w n (lo w ) an d 10 .8 87 .3 74 .7 93 .6 87 .3 6. 2 59 9 47 2 37 1 67 N o 20 01 –2 00 2 M M R 20 08 a nd 20 13 U nk no w n (lo w ) an d 10 .8 86 .8 73 .2 93 .4 86 .6 6. 4 58 9 60 6 37 7 35 N o 20 00 –2 00 1 M M R 20 08 a nd 20 13 U nk no w n (lo w ) an d 10 .8 86 .0 73 .0 93 .0 86 .5 6. 8 60 5 72 4 41 1 89 N o 19 99 –2 00 0 M M R 20 08 a nd 20 13 U nk no w n (lo w ) an d 10 .8 88 .6 74 .0 94 .3 87 .0 5. 7 62 3 26 2 35 5 26 N o 19 98 –1 99 9 M M R 20 08 a nd 20 13 U nk no w n (lo w ) an d 10 .8 89 .6 74 .6 94 .8 87 .3 5. 2 64 5 13 3 33 5 47 N o 19 97 –1 99 8 M M R 20 08 a nd 20 13 U nk no w n (lo w ) an d 10 .8 90 .5 74 .6 95 .3 87 .3 4. 8 65 7 00 5 31 5 36 Ye s 19 96 –1 99 7 M M R 20 08 a nd 20 13 U nk no w n (lo w ) an d 10 .8 90 .8 74 .0 95 .4 87 .0 4. 7 68 3 22 5 32 1 12 Ye s 19 95 –1 99 6 M M R 20 08 U nk no w n (lo w ) 91 .7 74 .2 95 .9 87 .1 4. 8 68 9 51 1 33 0 97 Ye s 19 94 –1 99 5 M M R 20 08 U nk no w n (lo w ) 92 .6 74 .7 96 .3 87 .3 4. 3 69 6 48 4 29 9 49 Ye s 19 93 –1 99 4 M M R 20 08 U nk no w n (lo w ) 93 .5 76 .4 96 .8 88 .2 3. 9 72 5 25 0 28 2 85 Ye s 19 92 –1 99 3 M M R 20 08 U nk no w n (lo w ) 94 .1 74 .4 97 .0 87 .2 3. 7 73 1 67 1 27 0 72 Ye s 19 91 –1 99 2 M M R2 1 99 6 an d M M R 20 08 60 a nd u nk no w n (lo w ) 92 .4 N A 96 .2 N A 5. 9 75 7 32 8 44 6 82 N o (c on tin ue s. . . ) 757Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policyMichael Edelstein et al. additional cases each disease case gener- ates in a susceptible population; if R0 < 1, transmission will not carry on).6 Table 1 shows whether each cohort achieved this target or not. To further validate our susceptibility estimates, we compared the age-specific incidence of laboratory- confirmed cases of measles by year of diagnosis (restricted to cases with an onset between 2010 and 2018) with age-specific susceptibility. We focused on age-specific incidence and suscep- tibility rather than on annual incidence to identify reported cases that occurred in birth cohorts eligible for vaccination and to help us discover underprotected groups that could be targeted by practi- cal changes to vaccination programmes (for example, by identifying appropriate age groups for a catch-up campaign). All susceptibility calculations were performed using Microsoft Excel (Mi- crosoft Corporation, Redmond, United States of America). All coverage and surveillance data were collected through national routine surveillance systems and the study was conducted using ag- gregated data as part of routine surveil- lance activities. No specific funding or formal ethical approval was required. Results Table 1 shows coverage of the first and second MMR vaccine doses in birth cohorts between 1985–1986 and 2015– 2016. The small differences between COVER and ImmForm estimates for birth cohorts between 2000–2001 and 2012–2013 (Fig. 2; ImmForm data at the children’s fifth birthday were not avail- able after this date) suggest that little op- portunistic vaccination took place after routine immunization and that no large groups of unvaccinated foreign-born children were registered with general practices in England. Since the quality of primary care data (i.e. ImmForm data) was low for birth cohorts before 2000–2001, coverage estimates were uncertain (Fig. 2). Overall, measles susceptibility among people born between 1985 and 2016 was 4.6% (range: 1.2–9.2), which corresponds to 975 920 individuals in these birth cohorts in 2019 (Table 1). Of individuals who were eligible for the second MMR vaccine dose from October 1996 onwards, those born between 1998 and 2004 were in birth cohorts classified as not having a sufficiently high level of immunity to prevent measles transmis-Bi rt h co ho rt a Ap pl ica bl e ca tc h- up ca m pa ig n Ro ut in e va cc in at io n co ve ra ge ,c % Ad ju st ed ro ut in e va cc in at io n co ve ra ge ,d % Pr op or tio n su sc ep tib le to m ea sle s,e % Es tim at ed n o. of in di vi du al s i n co ho rt in 2 01 7 Es tim at ed n o. of su sc ep tib le in di vi du al s Im m un ity le ve l s uffi cie nt to in te rr up t tr an sm iss io n? f Na m e an d da te b Co ve ra ge , % Fi rs t M M R va cc in e do se Se co nd M M R va cc in e do se Fi rs t M M R va cc in e do se Se co nd M M R va cc in e do se 19 90 –1 99 1 M M R2 1 99 6 an d M M R 20 08 60 a nd u nk no w n (lo w ) 92 .7 N A 96 .4 N A 5. 7 77 7 36 8 44 3 10 N o 19 89 –1 99 0 M M R2 1 99 6 an d M M R 20 08 60 a nd u nk no w n (lo w ) 92 .0 N A 96 .0 N A 6. 1 76 6 80 4 46 7 75 N o 19 88 –1 98 9 M ea sle s– ru be lla 19 94 92 89 .8 N A 94 .9 N A 1. 2 76 0 18 3 9 12 2 Ye s 19 87 –1 98 8 M ea sle s– ru be lla 19 94 92 87 .2 N A 93 .6 N A 1. 4 76 9 94 1 10 7 79 Ye s 19 86 –1 98 7 M ea sle s– ru be lla 19 94 92 90 .8 N A 95 .4 N A 1. 2 75 0 90 8 9 01 1 Ye s 19 85 –1 98 6 M ea sle s– ru be lla 19 94 92 77 .9 N A 88 .9 N A 2. 0 75 8 38 5 15 1 68 Ye s To ta l N A N A N D N D N D N D 4. 6 21 3 82 6 86 97 5 92 0 N A M M R: m ea sle s, m um ps a nd ru be lla ; M M R2 : s ec on d do se o f m ea sle s, m um ps a nd ru be lla v ac ci ne ; N A: n ot a pp lic ab le ; N D : n ot d et er m in ed . a Bi rth c oh or ts in cl ud ed in di vi du al s b or n be tw ee n Ap ril in o ne y ea r a nd M ar ch in th e fo llo w in g ye ar . b Ca tc h- up c am pa ig ns w er e ca rri ed o ut in 1 99 8, 2 00 8 an d 20 13 w ith m ea sle s, m um ps a nd ru be lla (M M R) v ac ci ne , in 1 99 4 w ith m ea sle s– ru be lla v ac ci ne a nd in 1 99 6 w ith th e se co nd M M R va cc in e do se (M M R2 ; F ig . 2 ). c Va cc in e co ve ra ge w as d et er m in ed u sin g da ta fr om lo ca l C hi ld H ea lth In fo rm at io n Sy st em s, w hi ch p ro vi de d at a to th e co ve r o f v ac ci na tio n ev al ua te d ra pi dl y (C O VE R) p ro gr am m e. d Ad ju st ed v ac ci ne c ov er ag e w as c al cu la te d us in g a 50 % c ov er ag e un de re st im at e co rre ct io n fa ct or : f or e xa m pl e, if th e es tim at ed p ro po rti on o f t he c oh or t n ot c ov er ed b y va cc in at io n w as 8 % , 4 % (i .e . 5 0% o f 8 % ) w as a dd ed to th e un ad ju st ed co ve ra ge . e Th e su sc ep tib ilit y ca lc ul at io n to ok in to a cc ou nt b ot h ro ut in e an d ca tc h- up v ac ci na tio n, a s d es cr ib ed in th e m et ho ds . f Th e im m un ity le ve l w as re ga rd ed a s s uffi ci en t t o in te rru pt m ea sle s t ra ns m iss io n w he n th e re pr od uc tiv e nu m be r ( R 0 ) w as u nd er 1 , w hi ch c or re sp on de d to a n im m un ity le ve l o ve r 8 5% in c hi ld re n ag ed u nd er 4 y ea rs a nd o ve r 9 5% in th os e ag ed 5 ye ar s a nd o ve r. (. . . co nt in ue d) 758 Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policy Michael Edelstein et al. sion (Table 1). Surveillance data on mea- sles cases confirmed that, among birth cohorts eligible for full vaccination, the incidence was highest in these cohorts (Fig. 3). Likewise, birth cohorts between 1989–1990 and 1991–1992 did not ap- pear to achieve a sufficiently high level of immunity. However, high susceptibility in those cohorts was not matched by high disease incidence (Fig. 3), which indi- cates that coverage of the 1996 catch-up campaign was probably underestimated. The incidence of disease in individuals born in 2010 and 2011 was higher than that in more susceptible individuals born between 2000 and 2003 (Fig. 3). How- ever, most cases in those born in 2010 and 2011 occurred before all children in their birth cohort became eligible for the first MMR vaccine dose (Fig. 3). When a coverage underestimate correction factor of 25% rather than 50% was applied in the sensitivity analysis, susceptibility estimates suggested that no birth cohort between 1989–1990 and 2006–2007 achieved a sufficiently high level of immunity to prevent measles transmission. However, this apparently high susceptibility was not matched by a high observed disease incidence. Con- sequently, the correction factor of 50% was more likely to be correct. Discussion Our application of data triangulation to measles vaccination in England, which considered data on vaccine coverage (in- cluding supplementary immunization) in individual birth cohorts and data on the age-specific incidence of measles, illustrates that the approach provides a pragmatic, simple and useful way of generating and validating disease sus- ceptibility estimates. The concordance between data sources we observed for specific birth cohorts confirmed that data quality was high in those years, such information is helpful for evaluat- ing data-driven targeted vaccination. In contrast, we found that coverage and incidence data were discordant for individuals in birth cohorts between 1989–1990 and 1991–1992, which sug- gested that one of the two sources was inaccurate. Discordant findings can trigger further investigation and lead to improved data quality. The use of triangulation also en- abled us to determine that, despite good coverage overall, there was a high proportion of susceptible individuals among those born between 1998 and 2004 (who were aged between 15 and 22 years in 2019), even after adjusting for coverage under-ascertainment. This finding was consistent with disease in- cidence data, which showed that most cases and outbreaks in recent years oc- curred in this age group.24 In England, where vaccine coverage is assessed using various methods, our synthesis of data from two different vaccine coverage sources led to insights into the vaccina- tion status of the population. This data synthesis enabled us to estimate the magnitude of opportunistic vaccination, as well as evaluate the data quality of primary care vaccination records. For example, we found that data quality was not sufficient to accurately estimate vaccine coverage among adults born abroad. In addition to the limitations in- herent in both COVER and ImmForm data, the study had several other limita- tions. First, coverage data for catch-up campaigns were less accurate than for routine immunization. In particular, data were not collected for the 2008 catch-up campaign (unpublished re- gional evaluations suggest coverage was low). Second, a London-only, catch- up campaign took place in 2004 and achieved a minimum of 24% coverage for individuals born between 1985 and 2004.25 This campaign was not included in our study because it was regional. Consequently, susceptibility in eligible birth cohorts may have been overesti- mated at the national level. Third, cover- age in national catch-up campaigns was assumed to be the same in all areas and eligible birth cohorts. Fourth, although 95% of general practices contributed to ImmForm data, the proportion varied between cohorts. Moreover, only 50% of practices reported data on cohorts born before September 1995. However, as these practices were spread across the country, coverage estimates for Fig. 1. Flowchart for determining level of protection from measles in birth cohorts, England, 1985–2016 0 95 99.75 Level of protection (%) Birth cohort eligible for catch-up campaign Received catch-up dosea Received second catch-up doseb Received second catch-up doseb Received second catch-up doseb Did not receive second catch-up doseb Did not receive second catch-up doseb Did not receive second catch-up doseb Received MMR2 Received catch-up dosea Did not receive catch-up doseaDid not receive MMR2 Did not receive catch-up dosea Did not receive MMRIReceived MMRI MMR1: first dose of measles, mumps and rubella vaccine; MMR2: second dose of measles, mumps and rubella vaccine. a For birth cohorts eligible for one catch-up campaign. b For birth cohorts eligible for more than one catch-up campaign. 759Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policyMichael Edelstein et al. these cohorts were unlikely to have been biased. Fifth, although COVER data included single-antigen measles vaccine for birth cohorts between 1985 and 1987, ImmForm data did not. This discrepancy may explain why estimated coverage in these birth cohorts was lower for ImmForm than COVER data (Fig. 2). Sixth, during the early to mid- 2000s, a small number of parents opted to have their children vaccinated using a private, unlicensed, single measles vaccine.26 This vaccine was not included in either COVER or ImmForm data. A 2007 study involving children born in 2001 and 2002 estimated that use of this vaccine could have increased coverage for all measles-containing vaccines by around 2% in individuals born in the early 2000s.26 When high-quality census data that can be linked to age-specific disease susceptibility estimates are available, the number of susceptible individuals cur- rently in the population can be deduce, thereby enabling the size and timing of future outbreaks to be modelled. Data triangulation can improve the accuracy and precision of coverage estimates, which is vital in areas where coverage is high, and increase confidence in data. In contrast, a recent study that used incomplete and inaccurate infor- mation overestimated the number of susceptible individuals in England by a factor of 1.8.27 Our analytical approach involved only routinely available data sources, which are not exclusive to the United Kingdom or other high-income coun- tries.28 Any country that routinely records disease incidence and vaccine coverage could consider a similar ap- proach for measles and other diseases. However, the value of the information produced will depend on the accuracy and precision of the data available and on knowledge of how data quality varies over time. A comprehensive report on improving immunization data quality and use that was recently presented to WHO’s Strategic Advisory Group of Experts on immunization is available online.1 The report provides strategic guidance on improving data at the national level. In countries where data on individual years are not available, wider age ranges could be considered. Although the resulting insights would be less detailed, the analysis would still be useful for validating coverage and surveillance data and for identifying sus- ceptible age groups. Currently, WHO is planning to publish a framework for, and guidance on, data triangulation to help countries routinely adopt the approach.1 One alternative to using routine data sources is to conduct much costlier and resource-intensive seroprevalence stud- Fig. 2. Measles vaccine coverage, by birth cohort, England, 1985–2016 M ea sle s v ac cin e c ov er ag e ( % ) 100 90 80 70 60 50 40 30 20 10 0 Birth cohort Two doses (COVER programme) One dose (COVER programme) Two doses (ImmForm) One doses (ImmForm) 1985- 1986 1986- 1987 1987- 1988 1988- 1989 1989- 1990 1990- 1991 1991- 1992 1992- 1993 1993- 1994 1994- 1995 1995- 1996 1996- 1997 1997- 1998 1998- 1999 1999- 2000 2000- 2001 2001- 2002 2002- 2003 2003- 2004 2004- 2005 2005- 2006 2006- 2007 2007- 2008 2008- 2009 2009- 2010 2010- 2011 2011- 2012 2012- 2013 2013- 2014 2014- 2015 2015- 2016 First MMR vaccine dose (introduced 1988) Second MMR vaccine dose (introduced 1996) MMR catch-up campaign 2008 MMR catch-up campaign 2013 Single measles vaccination (introduced 1968) MMR catch-up campaign 1988 MR catch-up campaign 1994 MR catch-up campaign 1996 COVER: cover of vaccination evaluated rapidly (programme); MMR: measles, mumps and rubella; MMR2: second dose of measles, mumps and rubella vaccine; MR: measles and rubella. Notes: Estimates from the cover of vaccination evaluated rapidly (COVER) programme up to 1989 include coverage of both the single measles vaccine and the first measles, mumps and rubella (MMR) vaccine. For the COVER programme, coverage data were collected at the children’s second birthday for those in birth cohorts after 2012–2013 and before 1992–1993; otherwise data were collected at their fifth birthday. Birth cohorts included individuals born between April in one year and March in the following year. 760 Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policy Michael Edelstein et al. ies. In England, where the quality of both coverage and incidence data is relatively high, the added value of seroprevalence studies is limited. Current arrangements rely on residual blood samples from hos- pitals, which may not be representative of the general population, particularly for younger age groups. Although national estimates of disease susceptibility can help iden- tify at-risk birth cohorts, they may not reflect inequalities at the local level. In England, vaccine coverage varies by ethnicity, social deprivation and geographical location.29 Consequently, coverage is heterogeneous and the burden of measles and rubella falls disproportionately on specific commu- nities.30,31 Herd immunity extends the benefits of national immunization pro- grammes to unvaccinated individuals, thus intrinsically reducing inequalities, but its impact will depend on local and overall vaccine coverage and population mixing patterns. When a large number of unvaccinated individuals live in close proximity, their community becomes vulnerable to outbreaks. Better vaccine coverage across the whole population should be accompanied by targeted ef- forts to assess the risk, specific needs and characteristics of undervaccinated communities and, thereby, close any gaps in immunity.32 In conclusion, triangulating exist- ing data sources on routine vaccina- tion coverage and vaccine-preventable disease surveillance can generate new insights into a population’s level of im- munity and help prioritize public health actions aimed at closing gaps in immu- nity. Use of this approach in England helped establish that, despite achieving high MMR vaccine coverage, measles susceptibility in particular age cohorts was sufficiently high to sustain disease transmission. Nevertheless, susceptibil- ity in all population subgroups cannot be estimated using routine data sources alone. Ad hoc studies are needed for undervaccinated groups, such as adults born abroad. Triangulating coverage and incidence data, in particular, is a useful way of maximizing the quality of data on vaccine-preventable diseases and should be used more widely. Forthcoming guid- ance from WHO and its partners on the triangulation of data from the Expanded Programme on Immunization will help countries improve data use and quality and, ultimately, help control vaccine- preventable diseases. ■ Competing interests: None declared. صخلم ةيلماشلا ادنلريأو ىمظعلا ايناطيبرل ةدحتلما ةكلملما في ةبصلحا حاقل تانايب ثلثم نمضتي يذلا تانايبلا طيطختل نكمي فيك حيضوت ضرغلا تانايبلا مادختسا ينستح لىإ يدؤي نأ ،ةينيتور تانايب رداصم عضو قيرط نع حاقللا جمانرب ةيلاعف لوح ةبقاث ىؤر يرفوتو .رابتعلاا في اترلجنإب ضرلماب ةباصلإا راركتو ةبصلحا حاقل تانايب فاكنلاو ةبصلحا حاقل ةيطغت لوح تانايب لىع انلصح دقل ةقيرطلا لىإ 1985 نم ديلاولما تاعوممج في (MMR) ةينالملأا ةبصلحاو مدعل ًاقفو اهليدعتب انمقو ،لافطلأا ةحص تلاجس نم ،2016 ميهدل نكي لم ناكسلا نأ انضترفاو .ةعباتلما تلاحمو ققحتلا ،ةدحاو ةعرلج 95% تناك حاقللا اذه ةيلاعف نأو ،ةيعيبط ةعانم ةياعرلا تلاجس للاخ نم فرعتلا مت .ينتعرلج 99.75%و لىعو ،ينيتورلا ططخلما جراخ تتم يتلا تمايعطتلا لىع ةيلولأا فيرعت مت .عينمتلا نس دعب اترلجنإ لىإ اولخد نيذلا صاخشلأا Fig. 3. Measles susceptibility and incidence, by birth cohort, England, 1985–2018 M ea sle s i nc id en ce (p er 1 0 00 0 00 p op ul at io n) 600 500 400 300 200 100 0 M ea sle s su sc ep tib ili ty (% ) 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 1985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018 2010Year of case occurence: Cases ocurring before all children in their birth cohorts became eligible for the first MMR vaccine dose Susceptibility 2011 2012 2013 2014 2015 2016 2017 2018 Calendar year of birth MMR: measles, mumps and rubella. 761Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policyMichael Edelstein et al. 摘要 大不列颠及北爱尔兰联合王国的麻疹疫苗接种数据三角互证 目的 通过考虑英国的麻疹疫苗接种和疾病发病率数据 来说明涉及常规数据源的数据三角互证如何优化数据 使用情况并提供有关疫苗计划效果的见解。 方法 我们从 1985 年至 2016 年的儿童健康记录中获 得了出生队列中麻疹、腮腺炎和风疹 (MMR) 疫苗覆 盖率的相关数据并根据归因不足和弥补活动进行了调 整。我们假设该人群没有天然免疫力，并且一剂疫苗 有效率为 95％，两剂疫苗有效率为 99.75％。在初级 保健记录中确定了在常规计划之外以及在免疫接种年 龄后进入英国的人群中进行的疫苗接种。麻疹易感率 是指不论各种免疫活动，非免疫人口所占的百分比。 我们对麻疹易感率和发病率数据进行了三角互证。 结果 中位数易感率为 4.6％（范围 ：1.2-9.2）。在符合 两种 MMR 疫苗剂量的队列中，1998 年至 2004 年间出 生的人群最易感染。麻疹发病率在这些队列中最高。 2000 年以后，来自初级保健和儿童健康记录的数据在 队列中具有可比性，表明很少进行补充疫苗接种。对 于 2000 年以前的队列，初级保健数据质量不足以准确 估算覆盖率。 结论 麻疹疫苗接种覆盖率和疾病监测的三角互证常规 数据针对人群免疫力提供了新的见解并有助于识别易 感群体，从而有利于优先考虑公共保健措施，以缩小 免疫差距。这种方法可以应用于任何经常记录疫苗覆 盖率和发病率的国家。 Résumé Triangulation des données relatives à la vaccination contre la rougeole au Royaume-Uni de Grande-Bretagne et d'Irlande du Nord Objectif Illustrer comment la triangulation des données incluant des sources de données ordinaires peut optimiser l'utilisation des données et donner une idée de l'efficacité du programme de vaccination en considérant des données relatives à la vaccination contre la rougeole et à l'incidence de la maladie en Angleterre. Méthodes Nous avons obtenu des données sur la couverture vaccinale contre la rougeole, les oreillons et la rubéole dans les cohortes de naissances de 1985 à 2016 à partir des dossiers médicaux des enfants et nous les avons ajustées en fonction de la sous-détermination et des campagnes de rattrapage. Nous sommes partis du principe que la population n'avait pas d'immunité naturelle et que l'efficacité du vaccin était de 95% pour une dose et de 99,75% pour deux doses. Les vaccinations pratiquées en dehors du calendrier habituel et chez des personnes entrées en Angleterre après l'âge de vaccination ont été identifiées à partir des dossiers relatifs aux soins primaires. La sensibilité à la rougeole a été définie comme le pourcentage d'individus qui n'étaient pas immunisés malgré toutes les activités de vaccination. Nous avons triangulé les données relatives à la sensibilité à la rougeole et à l'incidence de la rougeole. Résultats La sensibilité médiane était de 4,6% (étendue: 1,2-9,2). Parmi les cohortes pouvant bénéficier de deux doses de vaccins ROR, celles nées entre 1998 et 2004 étaient les plus sensibles. L'incidence de la rougeole était plus élevée dans ces cohortes. Les données tirées des dossiers relatifs aux soins primaires et des dossiers médicaux des enfants étaient comparables pour les cohortes ultérieures à 2000, ce qui laisse entendre que peu de vaccinations supplémentaires ont été pratiquées. Dans le cas des cohortes antérieures à 2000, la qualité des données sur les soins primaires était insuffisante pour permettre une estimation précise de la couverture. Conclusion La triangulation des données ordinaires sur la couverture vaccinale antirougeoleuse et la surveillance de la maladie a jeté une lumière nouvelle sur l'immunité de la population et aidé à identifier les groupes vulnérables, ce qui a permis d'établir l'ordre de priorité des actions de santé publique destinées à combler les lacunes en matière d'immunité. Cette approche pourrait être appliquée dans tous les pays qui enregistrent systématiquement la couverture vaccinale et l'incidence de la maladie. Резюме Триангуляция данных о вакцинации против кори в Соединенном Королевстве Великобритании и Северной Ирландии Цель Проиллюстрировать, как триангуляция данных с применением стандартных источников данных может оптимизировать использование данных и помогает определять наиболее эффективные способы реализации программы прививок в контексте данных о вакцинации от кори и распространенности заболевания в Англии. لم نيذلا دارفلأل ةيوئلما ةبسنلا انهأ لىع ةبصلحاب ةباصلإا ةيلباق طيطختب انمق .ميعطتلا ةطشنأ لك نم مغرلا لىع يننصمح اونوكي .ابه ةيلعفلا ةباصلإاو ،ةبصلحاب ةباصلإا ةيلباق تانايب لىإ 1.2 :ىدلما) 4.6% وه ةباصلإا ةيلباق طسوتم ناك جئاتنلا حاقل تاعرج لىع لوصحلل ةلهؤلما تاعومجلما ينب نم .(9.2 رثكلأا يه 2004و 1998 يماع ينب ديلاولما تناك ،MMR هذه ينب لىعلأا وه ةبصلحاب ةباصلإا لدعم ناك .ةباصلإل ةضرع ةيلولأا ةياعرلا تلاجس نم ةاقتسلما تانايبلا تناك .تاعومجلما دعب ام تاعومجلم ةبسنلاب ةنراقملل ةلباق لافطلأا ةحص تلاجسو .طيسب لييمكت ميعطت لىع لوصلحا مت هنأ حضوي امم ،2000 ماع تانايب ةدوج تناك ،2000 ماع لبق ام تاعومجملل ةبسنلاب .قيقد لكشب ةيطغتلا ريدقتل ةيفاك يرغ ةيلولأا ةياعرلا دض ميعطتلا ةيطغت لوح ةينيتورلا تانايبلا طيطتخ ىدأ جاتنتسلاا ةعانم لوح ةديدج ةبقاث ىؤر ميدقت لىإ ،ضارملأا ةبقارمو ةبصلحا يتلاو ،رطخلل ةضرعلما تاعومجلما ديدتح في ةدعاسلماو ،ناكسلا تارغث جلاعل ةماعلا ةحصلا تاءارجلإ ةيولولأا حنلم ةديفم تناك ينيتور لكشب موقت ةلود ةيأ في بولسلأا اذه قيبطت نكمي .ةعانلما .ضرلماب ةباصلإا لدعمو حاقللا ةيطغت تانايب ليجستب 762 Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policy Michael Edelstein et al. Методы Авторы рассмотрели данные об охвате вакциной от кори, краснухи и свинки (MMR) младенцев, родившихся в период с 1985 по 2016 год, которые получены из медицинских карточек детей, и скорректировали их с учетом недостаточного выявления и кампаний подчищающей иммунизации. Авторы исходили из предположения, что естественный иммунитет в популяции отсутствует и что эффективность вакцинации составляла 95% после однократной дозы и 99,75% после двух доз. Случаи вакцинации с нарушением обычного графика и вакцинации лиц, попавших в Англию позже установленного возраста иммунизации, идентифицировались по первичным медицинским документам. Подверженность заболеванию корью определялась как процентная доля лиц, которые не приобрели иммунитет, несмотря на вакцинацию. Авторы провели триангуляцию данных по частоте заболевания корью и подверженности этому заболеванию. Результаты Медианная подверженность заболеванию составила 4,6% (диапазон: 1,2–9,2). Среди когорт, которым полагалось получить две дозы вакцины MMR, наиболее подвержены заболеванию были дети, рожденные в период с 1998 по 2004 год. В этих когортах распространенность кори была наивысшей. Данные первичных учреждений здравоохранения и детских медицинских карт для когорт лиц, родившихся после 2000 года, были сопоставимыми, что позволяет предположить, что дополнительная вакцинация почти не проводилась. Для когорт лиц, родившихся до 2000 года, качество данных первичных медицинских документов было недостаточным для точного определения охвата вакцинацией. Вывод Триангуляция стандартных данных об охвате вакцинацией от кори и надзоре за распространенностью заболевания позволила сделать новые выводы об иммунном статусе популяции и помогла выявить подверженные риску группы, что было важно для расстановки приоритетов в общественном здравоохранении с целью принятия мер по ликвидации пробелов в иммунизации. Такой подход применим в любой стране, которая регулярно фиксирует охват населения вакцинацией и частоту возникновения соответствующих заболеваний. Resumen Triangulación de los datos de vacunación contra el sarampión en el Reino Unido de Gran Bretaña e Irlanda del Norte Objetivo Ilustrar cómo la triangulación de datos con fuentes rutinarias de datos puede optimizar el uso de los mismos y proporcionar información sobre la eficacia de los programas de vacunación al considerar la vacunación contra el sarampión y los datos de incidencia de la enfermedad en Inglaterra. Métodos Se obtuvieron datos sobre la cobertura de la vacuna contra el sarampión, las paperas y la rubéola (SPR) en las cohortes de nacimiento de 1985 a 2016 a partir de los registros de salud infantil y se ajustaron para tener en cuenta las campañas de recuperación y de los errores por la falta de observaciones. Se asumió que la población no tenía inmunidad natural y que la efectividad de la vacuna era del 95 % para una dosis y del 99,75 % para dos dosis. Las vacunas realizadas fuera del horario de rutina y en personas que ingresaron a Inglaterra después de la edad de vacunación fueron identificadas en los registros de atención primaria. La susceptibilidad al sarampión se definió como el porcentaje de individuos que no eran inmunes a pesar de todas las actividades de vacunación. Se triangularon los datos de susceptibilidad e incidencia del sarampión. Resultados La mediana de susceptibilidad fue de 4,6 % (rango: 1,2- 9,2). Entre las cohortes elegibles para dos dosis de la vacuna triple viral, las que nacieron entre 1998 y 2004 fueron las más susceptibles. La incidencia de sarampión fue mayor en estas cohortes. Los datos de los registros de atención primaria y de salud infantil fueron comparables para las cohortes después de 2000, lo que sugiere que se realizó muy poca vacunación suplementaria. Para las cohortes anteriores a 2000, la calidad de los datos de atención primaria fue insuficiente para estimar con precisión la cobertura. Conclusión La triangulación de los datos de rutina sobre la cobertura de la vacunación contra el sarampión y la vigilancia de las enfermedades proporcionó nueva información sobre la inmunidad de la población y ayudó a identificar a los grupos vulnerables, lo que resultó útil para establecer prioridades en las medidas de salud pública a fin de subsanar las deficiencias en materia de inmunidad. Este enfoque podría aplicarse en cualquier país que registre de forma rutinaria la cobertura de la vacuna y la incidencia de la enfermedad. References 1. Key backgrounds documents. Meeting of the Strategic Advisory Group of Experts (SAGE) on Immunization, Geneva, Switzerland, 2–4 April 2019. Geneva: World Health Organization; April 2019. Available from: https:// www.who.int/immunization/sage/meetings/2019/april/SAGE_Yellow_ Book_April_2019.pdf?ua=1 [cited 2019 Jun 7]. 2. Brown DW, Burton AH, Feeney G, Gacic-Dobo M. Avoiding the will o′ the wisp: challenges in measuring high levels of immunization coverage with precision. World J Vaccines. 2014 August;4(3):97–9. doi: http://dx.doi. org/10.4236/wjv.2014.43012 3. David L, editor. Control of communicable diseases manual. 19th edition. Washington, DC: American Public Health Association; 2008. 4. Measles. Key facts. Geneva: World Health Organization; 2018. Available from: http://www.who.int/news-room/fact-sheets/detail/measles [cited 2019 Jun 7]. 5. Measles vaccines: WHO position paper – April 2017. Wkly Epidemiol Rec. 2017 04 28;92(17):205–27. PMID: 28459148 6. Funk S, Knapp JK, Lebo E, Reef SE, Dabbagh AJ, Kretsinger K, et al. Combining serological and contact data to derive target immunity levels for achieving and maintaining measles elimination. Cold Spring Harbor: bioRxiv. 201574 [Preprint]. 2019. doi: http://dx.doi.org/10.1101/201574doi: http://dx.doi.org/10.1101/201574 7. Measles: the green book, chapter 21. London: Public Health England; 2013. Available from: https://www.gov.uk/government/publications/measles- the-green-book-chapter-21 [cited 2019 Jun 7]. 8. Wakefield AJ, Murch SH, Anthony A, Linnell J, Casson DM, Malik M, et al. Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children. Lancet. 1998 02 28;351(9103):637–41. Retraction in: Lancet. 2010 Feb 6;375(9713):445. doi: http://dx.doi. org/10.1016/S0140-6736(97)11096-0 PMID: 9500320 9. Childhood vaccination coverage statistics, England, 2016–17. Leeds: NHS Digital; 2017. Available from: https://digital.nhs.uk/data-and-information/ publications/statistical/childhood-vaccination-coverage-statistics/ childhood-vaccination-coverage-statistics-england-2016-17 [cited 2019 Jun 7]. 10. UK measles and rubella elimination indicators and status. London: Public Health England; 2018. Available from: https://www.gov.uk/government/ publications/measles-and-rubella-elimination-uk/uk-measles-and-rubella- elimination [cited 2019 Jun 7]. 11. Measles and rubella elimination UK strategy. London: Public Health England; 2019. Available from: https://www.gov.uk/government/ publications/measles-and-rubella-elimination-uk-strategy [cited 2019 Jun 7]. 763Bull World Health Organ 2019;97:754–763| doi: http://dx.doi.org/10.2471/BLT.18.229138 Research Data triangulation and immunization policyMichael Edelstein et al. 12. European vaccine action plan 2015–2020. Geneva: World Health Organization; 2014. Available from: http://www.euro.who.int/en/health- topics/disease-prevention/vaccines-and-immunization/publications/2014/ european-vaccine-action-plan-20152020-2014 [cited 2019 Jun 7]. 13. Measles outbreaks across England. London: Public Health England; 2018. Available from: https://www.gov.uk/government/news/measles-outbreaks- across-england [cited 2019 Jun 7]. 14. Laboratory-confirmed cases of measles, rubella and mumps, England: October to December 2018. London: Public Health England; 2019. Available from: https://assets.publishing.service.gov.uk/government/uploads/ system/uploads/attachment_data/file/791349/hpr0819_mmr.pdf [cited 2019 Jun 7]. 15. Amirthalingam G, White J, Ramsay M. Measuring childhood vaccine coverage in England: the role of Child Health Information Systems. Euro Surveill. 2012 04 19;17(16):20149. PMID: 22551461 16. Edelstein M, Crocker-Buque T, Tsang C, Eugenio O, Hopson T, Pebody R, et al. Extracting general practice data for timely vaccine coverage estimates: the England experience. Vaccine. 2017 09 12;35(38):5110–4. doi: http://dx.doi. org/10.1016/j.vaccine.2017.08.015 PMID: 28822644 17. Rivest P, Bédard L, Arruda H, Trudeau G, Remis RS. Risk factors for measles and vaccine efficacy during an epidemic in Montreal. Can J Public Health. 1995 Mar-Apr;86(2):86–90. PMID: 7757898 18. King GE, Markowitz LE, Patriarca PA, Dales LG. Clinical efficacy of measles vaccine during the 1990 measles epidemic. Pediatr Infect Dis J. 1991 Dec;10(12):883–8. PMID: 1766702 19. Vitek CR, Aduddell M, Brinton MJ, Hoffman RE, Redd SC. Increased protections during a measles outbreak of children previously vaccinated with a second dose of measles-mumps-rubella vaccine. Pediatr Infect Dis J. 1999 Jul;18(7):620–3. doi: http://dx.doi.org/10.1097/00006454-199907000- 00010 PMID: 10440438 20. Simone B, Balasegaram S, Gobin M, Anderson C, Charlett A, Coole L, et al. Evaluation of the measles, mumps and rubella vaccination catch-up campaign in England in 2013. Vaccine. 2014 Aug 6;32(36):4681–8. doi: http://dx.doi.org/10.1016/j.vaccine.2014.05.077 PMID: 24996125 21. Gay N, Ramsay M, Cohen B, Hesketh L, Morgan-Capner P, Brown D, et al. The epidemiology of measles in England and Wales since the 1994 vaccination campaign. Commun Dis Rep CDR Rev. 1997 Feb 7;7(2):R17–21. PMID: 9046124 22. Evaluation of vaccine uptake during the 2013 MMR catch-up campaign in England. Report for the national measles oversight group on uptake of the measles, mumps and rubella (MMR) vaccine. London: Public Health England; 2014. Available from: https://www.gov.uk/government/ publications/evaluation-of-vaccine-uptake-during-the-2013-mmr-catch- up-campaign-in-england [cited 2019 Jun 7]. 23. Estimates of the population for the UK, England and Wales, Scotland and Northern Ireland. Mid-2017. London: Office for National Statistics; 2019. Available from: https://www.ons.gov.uk/peoplepopulationandcommunity/ populationandmigration/populationestimates/datasets/ populationestimatesforukenglandandwalesscotlandandnorthernireland [cited 2019 Jun 7]. 24. le Polain de Waroux O, Saliba V, Cottrell S, Young N, Perry M, Bukasa A, et al. Summer music and arts festivals as hot spots for measles transmission: experience from England and Wales, June to October 2016. Euro Surveill. 2016 Nov 3;21(44):30390. doi: http://dx.doi.org/10.2807/1560-7917. ES.2016.21.44.30390 PMID: 27881230 25. Hitchen L. London’s catch-up vaccination campaign against measles, mumps, and rubella reaches less than a quarter of children who were unvaccinated. BMJ. 2008 09 24;337:a1797. doi: http://dx.doi.org/10.1136/ bmj.a1797 PMID: 18815174 26. Sonnenberg P, Crowcroft NS, White JM, Ramsay ME. The contribution of single antigen measles, mumps and rubella vaccines to immunity to these infections in England and Wales. Arch Dis Child. 2007 Sep;92(9):786–9. doi: http://dx.doi.org/10.1136/adc.2006.109223 PMID: 17412744 27. Over 20 million children worldwide missed out on measles vaccine annually in past 8 years, creating a pathway to current global outbreaks – UNICEF. New York: United Nations Children's Fund; 2019. Available from: https:// www.unicef.org/press-releases/over-20-million-children-worldwide- missed-out-measles-vaccine-annually-past-8-years [cited 2019 Jun 7]. 28. Segatto C, Samad S, Mengue SS, Rodrigues G, Flannery B, Toscano CM. Historical analysis of birth cohorts not vaccinated against rubella prior to national rubella vaccination campaign, Brazil. J Infect Dis. 2011 Sep 1;204 Suppl 2:S608–15. doi: http://dx.doi.org/10.1093/infdis/jir357 PMID: 21954255 29. Byrne L, Ward C, White JM, Amirthalingam G, Edelstein M. Predictors of coverage of the national maternal pertussis and infant rotavirus vaccination programmes in England. Epidemiol Infect. 2018 01;146(2):197–206. doi: http://dx.doi.org/10.1017/S0950268817002497 PMID: 29239285 30. Stewart-Freedman B, Kovalsky N. An ongoing outbreak of measles linked to the United Kingdom in an ultra-orthodox Jewish community in Israel. Euro Surveill. 2007 09 20;12(9):E070920.1. PMID: 17900430 31. Cohuet S, Bukasa A, Heathcock R, White J, Brown K, Ramsay M, et al. A measles outbreak in the Irish traveller ethnic group after attending a funeral in England, March–June 2007. Epidemiol Infect. 2009 Dec;137(12):1759–65. doi: http://dx.doi.org/10.1017/S0950268809002714 PMID: 19470194 32. Letley L, Rew V, Ahmed R, Habersaat KB, Paterson P, Chantler T, et al. Tailoring immunisation programmes: using behavioural insights to identify barriers and enablers to childhood immunisations in a Jewish community in London, UK. Vaccine. 2018 07 25;36(31):4687–92. doi: http://dx.doi. org/10.1016/j.vaccine.2018.06.028 PMID: 29945834