Measuring vaccination coverage better will help achieve disease control

Abstract

Timely and accurate measurement of vaccination coverage is required to evaluate the success of vaccine programmes as well as identifying susceptible groups in order to better control disease. Estimating coverage requires knowledge of how many people are eligible for vaccination, and how many have received the vaccine. Achieving this presents a number of challenges in both high and low income settings. Investing in systems that record vaccination coverage better, as an integral part of vaccine strategies, will be a step towards better control of vaccine-preventable diseases.

Communicable disease control, Disease eradication, Immunisation, Information systems, Vaccination

Introduction

Immunisation is one of the most effective public health interventions available. Vaccines will avert over 23 million deaths between 2011 and 2020¹ and are key to achieving the health component of the sustainable development goals (SDGs). Vaccination has led to the eradication of smallpox, for which routine immunisation is no longer required. Polio is the next disease targeted, and since the launch of the global polio eradication initiative in 1988, cases have decreased from 350 000 to 74 reported cases in 2015.¹ When global eradication is not readily achievable, immunisation can eliminate indigenous transmission or contain the disease to a point at which it no longer constitutes a significant public health problem. These three approaches require achieving and sustaining high vaccination coverage through mass vaccination, and targeted and routine programmes. A decrease in vaccination coverage can lead to disease resurgence: in the former Soviet Union for example, the number of diphtheria cases increased from 800 in 1989, before the Soviet Union collapsed, to over 50 000 in 1994. Over 4000 persons died of the disease between 1990 and 1997.²

Measuring vaccine coverage accurately, in a timely manner and at a local level, contributes to disease control by identifying areas and groups that remain at risk, as to inform local or national vaccination strategies. Demonstrating high vaccination coverage is a recurrent performance indicator in vaccine-preventable disease control strategies. For example, WHO's global measles and rubella elimination strategy 2012–2020 aims to achieve and maintain 95% coverage at national and district level.³

Vaccination coverage can be defined as the proportion of individuals eligible for a vaccine who have received it. Estimating vaccination coverage therefore requires two elements: first, the ability to identify and enumerate the number of eligible individuals (the denominator); second, the ability to record and retrieve information about who among eligible individuals has received the vaccine (the numerator). Both numerator and denominator issues can affect vaccination coverage estimates.

Denominator issues

In high income countries, births, deaths and migration are generally well recorded and accurate, and age-specific population estimates exist. In England for example, a census is conducted every 10 years. In addition, electronic patient registers are available, allowing quasi-real time identification of individuals attending a specific primary care facility.⁴

In many low and middle income countries (LMICs), recent, local population estimates are rarely available.⁵ For example, the last population censuses conducted in Angola and the Democratic Republic of Congo were 1970 and 1984, respectively.⁵ The usefulness and reliability of administrative denominator data for identifying target populations and planning immunisation activities is therefore limited. It is not unusual to see coverage estimates exceed 100% because the denominator estimate is out of date, because individuals get vaccinated out of area (for example, in a secondary care facility across an administrative border) and are therefore included in the numerator but not the denominator, or because the area sees large scale seasonal migration. The decision to use administrative denominator data to estimate coverage requires local intelligence on the accuracy of population estimates, but in many circumstances, data quality precludes its use for producing meaningful coverage estimates. Because administrative denominator estimates can be unreliable in LMICs, household coverage surveys are a common source of vaccination coverage data, and can also be used to validate administrative coverage estimates. Survey data allow estimation of vaccination coverage when the size of the target population is unknown.⁶ Advantages of household surveys include a standardised methodology⁷ and an available body of evidence describing good practice.⁸ Surveys are, however, time and resource intensive, requiring high numbers of individuals to be trained for each survey. Surveys are also highly prone to several forms of bias,⁹ although WHO is revising its survey sampling guidance to minimise it.⁷ In addition, national household surveys with enough statistical power to estimate coverage locally are prohibitively expensive. Surveys are generally powered to estimate national or subnational coverage. This limits their precision and their usefulness for providing information to strengthen local system performance.^6,7 Surveys are therefore a complement rather than a substitute to high quality administrative coverage data.⁹

The lack of quality denominator data in low income settings has prompted the experimental use of alternative methods such as satellite imagery, geopositioning and mobile phone call records to estimate population size and migration at the local level.⁵ Serial satellite imagery analysing night time lights has been particularly useful to correct population estimates and by extension coverage following vaccination campaigns. This is particularly useful in LMICs experiencing large seasonal population migration,¹⁰ where population estimation using traditional methods is especially inaccurate. In addition, satellite-based population estimation can be done remotely.¹⁰ This nascent technology currently helps validating survey or administrative population estimates but requires further field testing and validation to be used routinely. Its use can be limited in certain circumstances such as in heavily forested areas or in persistent cloud cover (Diallo MS, personal communication).

Numerator issues

Although most healthcare facilities across the world record vaccine administration, as individuals migrate or receive vaccines in different settings, keeping track of their vaccine history requires either reliable retention of vaccination records by the family or a centralised data management system. A national vaccine register that allows the identification of unimmunised individuals or cohorts in real time is the gold standard. The Australian Immunisation Register, for example, captures all vaccines given to eligible individuals in Australia throughout their life.¹¹ Such systems, however, require a complex and expensive IT infrastructure, generally not available in LMICs or in many high income countries. In France for example, vaccination coverage is estimated from two-yearly school surveys, which do not provide timely data or accurate estimates below the national level.¹² Coverage data in older age groups, a crucial element of information in disease elimination strategies, are not available.¹² In some instances, the number of doses distributed to facilities rather than number of vaccinated individuals is used as a numerator. This does not take into account vaccine wastage, which can be significant if multi-dose vaccine vials are used. In countries where surveys are used to estimate coverage, the numerator is estimated either by looking at immunisation records in the home, asking the child's caretaker or both.⁶ This approach has several limitations: it is prone to bias, a large proportion of children in the least developed countries do not have an immunisation record in the home, information is rarely available for older children or adults, and information is only available on the previous birth cohorts, limiting their use for timely programme interventions.^6,9

Consequences of inaccurate vaccination coverage estimates

There are health and resources consequences, when coverage estimates are inaccurate. Overestimating vaccination coverage can divert attention and resources from immunisation programmes. In Burkina Faso in 2009, despite a national administrative coverage exceeding 95%, a large measles outbreak killed 367 and affected over 54 000 individuals, mostly unvaccinated children.¹³ Conversely, in London in 2013, coverage among 10–16 year olds was estimated to be 61%;¹⁴ however, no large outbreaks occurred, suggesting the low coverage was at least partly caused by inaccurate data. Indeed a study suggested that up to 60% of London children aged 10–16 with no MMR vaccine record were in fact vaccinated.¹⁴ Many who were targeted in a MMR catch-up campaign the same year were therefore already immune. Vaccinating previously vaccinated individuals multiple times—a situation encountered in high and low income settings—is both a waste of resources and an unnecessary medical intervention: both can be mitigated by better coverage data.

Conclusions

Controlling vaccine-preventable diseases requires that eligible populations are identified and vaccinated. Without knowing how many individuals remain at risk, and who and where they are, eradicating or eliminating disease will remain a challenge. Measuring coverage using traditional methods require tools and methods not available in many low income settings. However technological advances such as satellite imagery, geopositioning and mobile phone call records for example are relatively cheap and can help estimate population size and migration at the local level. Further research in the application of these technologies to demography and investment in systems that will enable more accurate recording of vaccination coverage as part of vaccine strategies at the local, national and global level will help achieve better coverage and progress on the path of disease control.

Author's contributions: ME has undertaken all the duties of authorship and is guarantor of the paper.

Funding: None.

Competing interests: None declared.

Ethical approval: Not required.