Objectives To assess childhood vaccination coverage at first, second and fifth birthdays by ethnicity in London between 2006/2007 and 2010/2011 and identify factors relating to lower coverage.
Design Data concerning receipt of diphtheria-containing vaccines were extracted from child health information systems (CHISs) and sent to the Health Protection Agency.
Setting Nine London Primary Care Trusts (PCTs).
Participants Records for 315 381 children born April 2001–March 2010.
Main outcome measures Receipt of a full primary course of diphtheria-containing vaccines at first and second birthdays, and a primary course and preschool booster at fifth birthday.
Results Consistently good vaccine coverage of the primary course (>88% at first birthday, >89% at second birthday) was achieved across the five largest ethnic groups. Coverage of the preschool booster at fifth birthday was >65% across the five largest ethnic groups. Lowest coverage was observed in smaller ethnic groups. Deprivation was not a strong indicator of coverage overall, and for most ethnic groups there was no relationship between deprivation and coverage. Coverage was significantly lower in children not assigned to a general practitioner practice in the CHIS.
Conclusions Smaller, less well-established ethnic groups within a PCT may require specific targeting to ensure children are fully immunised and to improve record keeping. Unregistered children need particular attention and may be missed by current scheduling processes in London. In order to monitor the impact of the current National Health Service (NHS) reorganisation on inequalities in access to healthcare data on country of birth, in addition to ethnicity, should be available for analysis.
- Comm Child Health
- Data Collection
- Health Service
Statistics from Altmetric.com
What is already known on this topic
The UK has a universal childhood immunisation programme with overall high vaccine coverage rates; these are lower in London where there is also an increasingly ethnically diverse population.
Previous studies have identified differences in vaccination coverage by ethnicity, but the capacity for monitoring this using data routinely collected within London has not been assessed.
Reducing inequalities in childhood immunisation has been identified as a priority by the National Institute for Clinical Excellence.
What this study adds
In general, the largest ethnic groups have good vaccination coverage, but newer, smaller communities within a PCT may need particular attention.
Improvements in record keeping and transfer of information are associated with improvements in reported vaccination coverage.
Children not registered with a general practitioner, or without up-to-date GP practice details in the child health information system, have lower recorded vaccination coverage and are at risk of missing out on key primary care initiatives.
The UK childhood immunisation programme currently includes a 5-in-1 vaccine that protects against diphtheria, tetanus, pertussis, polio and Haemophilus influenzae type b (DTaP/IPV/Hib) offered at 2, 3 and 4 months of age (primary course) and a preschool booster between 3 years 3 months and 5 years of age (dTaP/IPV or DTaP/IPV).1 Diphtheria-containing vaccines have been routine for decades and are uncontroversial, thus providing an indication of primary care access in children.
Child health information systems (CHISs) are managed by child health departments and one of their many functions is to schedule and record the immunisations given to children,2 although this varies locally3; in London, general practitioners (GPs) are often responsible for scheduling. Since 2006, the majority of Primary Care Trusts (PCTs, statutory bodies responsible for ensuring the availability of health services in a geographical area) in London have transferred to RiO CHIS. Vaccination coverage across the UK is currently monitored by the Cover of Vaccination Evaluated Rapidly (COVER) programme.4 Coverage is not uniform throughout the country; in particular coverage in London is lower,5 ,6 but recent increases, correlated with the introduction of National Health Service (NHS) London's Immunisation Improvement Programme, have reduced the gap between London and the rest of England.7
In 2010, 34% of people living in London were estimated to have been born abroad and 56% of children born in London were born to non-UK born mothers.8 ,9 London has an increasingly ethnically diverse population.10 Reducing differences in immunisation uptake among children is a priority identified by the National Institute for Clinical Excellence.11 This study explored the capacity for routine CHIS data in London to monitor vaccination coverage by ethnicity.
Nine London PCTs, who responded to a request at a London Immunisation Network (a group of London immunisation coordinators) meeting and via email, participated (figure 1). Participation was voluntary. The study PCTs represented a range of London PCTs in terms of vaccination coverage (coming from all four quartiles of 2010/2011 London PCT COVER data for diphtheria-containing vaccines at first, second and fifth birthdays).
A script written for the RiO CHIS was used to extract the following fields for each PCT’s responsible population (children registered with a GP in the PCT or unregistered children living within the PCT's geographical boundary): month and year of birth, date of receipt of each diphtheria-containing vaccine (see appendix 1—web only), gender, ethnic group, nationality, postcode and GP practice code. Where the child's record was linked to a maternal record, basic demographic data were extracted from the mother's record.
In total records for 315 381 children born April 2001–March 2010 were extracted. There were 185 534 children born April 2005–March 2010 (first birthday cohort), 180 477 born April 2004–March 2009 (second birthday cohort) and 164 000 born April 2001–March 2006 (fifth birthday cohort).
To maintain anonymity, postcodes were replaced by lower super output area codes12 prior to sending to the Health Protection Agency. Deprivation scores for each area were assigned using the Income Deprivation Affecting Children Index 2010.13 The following deprivation quintiles were created: 0–0.130 (least deprived), 0.131–0.250, 0.251–0.340, 0.341–0.440 and 0.441–0.960 (most deprived).
Vaccination status was computed for each child at each age evaluated; either fully immunised or not fully immunised (including completely unimmunised and partially immunised). Only month and year of birth were extracted, therefore a child born in April 2005 was considered to have received a dose within 1 year (by ‘first birthday’) if the vaccine was received between April 2005 and April 2006 inclusive. Fully immunised status was assigned at first and second birthdays if at least three primary immunisations (vaccines containing high-dose diphtheria; ‘D’) were recorded within 1 and 2 years, respectively. Fully immunised was assigned at fifth birthday if at least four high-dose diphtheria-containing vaccines were recorded, with the fourth (or subsequent) dose received between ages 3 and 5 years. The fourth (or subsequent) dose could also be the low-dose diphtheria dTaP/IPV vaccine. This coding is more conservative than some COVER extractions, which may only require the final vaccine (‘Part 3’ for a primary course or preschool booster within the appropriate age range for fifth birthday assessment) to be recorded for a child to be considered fully immunised.
Individual RiO ethnicity categories were used for larger groupings and combined into broader categories for those with small numbers of children (see appendix 2—web only).
Analyses were conducted in STATA SE/V.12.0 statistical software. Crude coverage proportions with exact 95% CIs were calculated for each variable. Multivariable logistic regression was used to determine whether ethnicity differences were due to other available data on gender, deprivation, PCT and year of birth. We also tested for interactions in the model between ethnicity and PCT and ethnicity and deprivation.
Overall, vaccination coverage was 87% at first birthday, 87% at second birthday and 60% at fifth birthday. An increase in coverage over the study period was seen for all cohorts by year of birth. In general, individual PCT coverage in this study was similar or lower than COVER data for the same period due to more conservative coding; the coverage in this data set also reflects low coverage in London.
Nationality was available for <2.5% of all children. Ethnicity was recorded for 66% (121 657) of children at first birthday, 64% (115 941) at second birthday and 56% (91 877) at fifth birthday. To examine representativeness of recorded ethnicity data, known ethnicity data for children born 2005–2009 in each PCT were compared with Greater London Authority ethnicity projections for children aged 0–4 years in 2009.14 The proportion in black and minority ethnic (BME) groups in the study data set was within 5% of the Greater London Authority projected BME proportion for all PCTs except one, which had a higher proportion of BME children in the data set (63%) than projected (49%). Overall, the study PCTs had a similar proportion of BME children (51%) to the projection for Greater London (53%).
Consistently good coverage of the primary course (>88% at first birthday, >89% at second birthday) was achieved across the five largest ethnic groups. Coverage of the preschool booster at fifth birthday was >65% across the five largest ethnic groups (figure 2). Although some of the smallest ethnic groups had good coverage, the lowest coverage in each cohort was among the smaller ethnic groups and those with unknown ethnicity. Adjusting for gender, deprivation, PCT and year of birth did not substantially change the ethnicity patterns in coverage (model details in appendix 3—web only). There was evidence of interaction between PCT and child ethnicity for all three age cohorts (p<0.001); this was most pronounced for white-Polish populations and related to the size of the white-Polish population within a PCT. Where white-Polish populations were larger (410 and 383 children), coverage at first birthday in this group (90% and 88%, respectively) was closer to the average for the PCT. Two PCTs with smaller white-Polish populations (77 and 140 children; all other PCTs had <50 children in this group) had significantly lower coverage at first birthday in their white-Polish populations (67% and 69%, respectively) than the average for their PCT. Similar interactions relating to the white-Polish population size were observed at second and fifth birthdays.
Gender was recorded for >99.9% of children overall. There was no difference in coverage between males and females at first birthday, although coverage was fractionally higher (<1%) for females at second and fifth birthdays (p<0.01).
Deprivation scores were assigned to 98% (309 552) of records overall. Coverage across quintiles ranged between 86% and 88% at first birthday, between 87% and 88% at second birthday and between 59% and 63% at fifth birthday, lower coverage in general relating to higher deprivation. Interaction between ethnicity and deprivation was significant in each cohort (p<0.001). Across each age cohort, a trend of reducing coverage by increasing deprivation was seen only for white-British and Not known groups. The opposite trend was observed for Indian and white-Other/Mixed/Unspecified at first and second birthdays only. Trends were not seen for other ethnicities.
At the time of data extraction 14 022 (4.4%) children in the data set were not assigned to a GP (ie, did not have a GP practice code recorded because they were unregistered, moving between practices or records were not current). The proportion of children not assigned to a GP in each PCT at the time of data extraction ranged from 1.1% to 7.0%. Vaccination coverage was 52% versus 88% in children without a GP practice code versus those with a GP practice code at first birthday, 55% versus 89% at second birthday and 21% versus 63% at fifth birthday. Significant differences in coverage between children with and without a GP practice code assigned were seen across all PCTs in all three cohorts. Overall, 2.2% of white-British children were not assigned to a GP practice compared with 4.2% of non-white-British and 5.8% of children without ethnicity recorded.
Overall 42% (131 077) of records in the data set were linked to a maternal record. This varied considerably by age and PCT. At first birthday, 112 306 (61%) child records were linked to a maternal record, 87 998 (49%) at second birthday and 32 395 (20%) at fifth birthday. At first birthday, linkage to a maternal record varied across PCTs from 16% to 87%. However, in all PCTs the proportion of children linked to a maternal record improved over time, sometimes dramatically (eg, in one PCT from 2% (children born 2005/2006) to 61% (children born 2009/2010)). Overall, vaccination coverage at first birthday was higher for linked (90.8%) children versus unlinked (81.3%) children (p<0.001) (further subanalyses of maternal data for linked children were not conducted as they were not considered representative). As linkage improved over time for each PCT, the difference in coverage between linked and unlinked children became more pronounced. Of those children in the data set with ethnicity recorded, 51% were linked to a maternal record compared with 27% of children without ethnicity recorded.
The largest ethnic groups in each cohort had good vaccination coverage. Ethnic groups with lowest coverage were generally smaller and those with unknown ethnicity. Interactions between PCT and ethnicity were observed for a minority of ethnic groups, for white-Polish populations (for whom migration to the UK has increased since Poland joined the European Union in 2004) this related to the size of the population in a PCT. Differences in coverage between ethnic groups were not explained by adjustment for gender, deprivation, PCT or year of birth. Deprivation was not a strong indicator of coverage overall, and for most ethnic groups there was no relationship between deprivation and coverage. Data completeness was a key factor in determining the vaccination coverage recorded (as evidenced by the low coverage in children with unknown ethnicity and those not linked to a maternal record). Children not assigned to a GP in the CHIS had lower vaccination coverage than those with a GP practice code recorded. Routinely collected data from the RiO CHIS can be used for basic analysis of vaccination coverage by ethnicity, with adjustment for certain factors.
Strengths and limitations of study
This is the first study to explore the capacity for data routinely collected within the RiO system to provide vaccination coverage data by ethnicity for London and includes smaller ethnic groups such as Somali and white-Polish. However, some ethnicities, for example, Romanian, are not captured individually. In other published analyses, additional factors have been studied such as family size, maternal smoking, maternal education and lone parenthood,15 ,16 but this study was restricted to CHIS fields. Deprivation in this study was assigned based on postcode, relating deprivation to a geographical area.
Although vaccination is unlikely to be recorded incorrectly, no record of vaccination could reflect failure to immunise or failure to record. This is more likely where immunisations were given in other geographical areas and would lead to lower measured coverage in those who move into the PCT, particularly at an older age. This may therefore explain the lower coverage in population groups more likely to have moved since birth,17 including recent migrants to the UK. Failure to record vaccination is also often associated with missing demographic data. It is difficult to disentangle true improvements in coverage from improvements in data quality. Data quality issues relating to information migrated from legacy systems have previously been identified.18 The lower vaccination coverage observed among children without a link to a maternal record may result from data relating to the mother and immunisations either not being added or transferred across to the current PCT records for children born outside the PCT (either in another PCT or abroad). Other information such as a child's ethnicity may also not have been transferred across. Obtaining a vaccination history from children arriving from abroad can be problematic; even if the vaccines received abroad are known, they may not be coded within the CHIS and so may be omitted from the record. Children who are not linked to a maternal record may also be those who are living with relatives, possibly moving frequently, or are looked-after children. More information such as country/area of birth/previous residence would aid understanding.
Comparison with other studies
Improvements in vaccination coverage over time, differences in coverage between PCTs and lower coverage of the preschool booster at fifth birthday compared with coverage of primary vaccines at first and second birthdays were expected.5 Previously, the Millennium Cohort Study identified differences in coverage by ethnicity.16 A study in Manchester found that white infants were least likely to be vaccinated with primary vaccines, and that for white infants (as found here) lower coverage was significantly associated with living in a deprived area.15 For black infants or black British infants and Pakistanis, there was no significant association between deprivation and immunisation.15
Conclusions and policy implications
We have shown that monitoring coverage by ethnicity is possible and could be used to identify groups with low recorded immunisation coverage. Such findings should be explored to determine whether there is a genuine need to improve coverage or a need to improve data quality. London's population is highly mobile making it challenging to maintain accurate health records as children move across PCT boundaries and change GP, particularly in the first year of life when the primary vaccine course is offered. The absence of data could indicate less contact with the health system, both in terms of opportunities for immunisation and maintaining records. Children in London are invited for vaccination by GPs; those not registered with a GP are at serious risk of missing out on immunisations and need particular attention. Registration with a GP can be particularly low among certain migrant populations.19
Although it is encouraging to see data completeness and vaccination coverage improving, the NHS is currently undergoing a major reorganisation and previous experience has shown that reorganisation negatively impacts on the quality of data20; this is therefore likely to continue to present challenges. However, it is also an opportunity to influence the CHIS service specification to ensure that fields such as ethnicity and country of birth are accurately recorded. Directors of Public Health in Local Authorities will have a duty to scrutinise and challenge the NHS for how well and how equitably it provides immunisation services—coverage by ethnicity should be one of the key metrics by which this role is undertaken.
We thank the following local collaborators and data managers within PCTs who provided data for this study: Oladapu Osubu (Oxleas/Bexley), Patricia Stephens and Ian Kirkwood (Camden), Khalida Aziz and David Griffiths (Hounslow), Gladys Xavier and Saleem Yasir (Outer North East London), Elizabeth Bell (Havering) and Fiona White and Chris Lovelace (Sutton and Merton). Thanks also to the London Immunisation Leads Network for the opportunity to present the initial proposal for this study, the Health Protection Agency and North London research and development offices for their assistance with study permissions and to David Freeman for database design. We also thank the following for their helpful comments on the proposal and/or drafts: Gayatri Amirthalingam, Mary Ramsay, Helen Bedford, David Elliman, Mark Johnson and Jane Jones.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
Note: The Health Protection Agency is now part of Public Health England
Contributors Contributors (specific author contributions, please note that contributions of other persons are listed in the acknowledgements). KSW (guarantor) had the idea for the study, designed the study, coordinated the collection of data, conducted the analysis and interpretation of results and drafted and revised the manuscript. JCJvW designed the study, wrote and piloted the data extraction script, provided data from Ealing and critically appraised the manuscript drafts. NA provided statistical guidance with respect to the planning of the study, analysis and interpretation of the study data and critically appraised the manuscript drafts. KG designed the study and critically appraised the manuscript drafts. JCJvW designed the study, interpreted the study results and critically appraised the manuscript drafts.
Competing interests None.
Ethics approval This study involved secondary use of non-identifiable data from health records; under the revised governance arrangements for research ethics committees (RECs) REC review was not required. Management permission for research within each PCT was obtained via research and development offices.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data submitted to the study by individual PCTs are available to those PCTs.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.