Susan Lister,1 Peter B McIntyre,1 Margaret A Burgess,1 Eddie D O'Brien2
Studies reviewedA total of 108 out of 448 studies reporting published and unpublished data on immunisation coverage in Australian children from 1990 to 1998 were eligible for inclusion.
Fifty-one higher quality studies were selected from a range of designs and settings shown in Tables 1-7. These tables summarise the location, design, sample size and age, response rate, strategy and method of validation for each study. Several studies, which indirectly reported coverage data, were included in this review. For example, some evaluated the implementation of immunisation reminder systems19-21 others evaluated the effectiveness of the Parent Held Record (PHR),22 or the accuracy of parental report23 or described an immunisation campaign.24 A preliminary evaluation of the ACIR was also included.25
Excluded studies26-29 were primarily those investigating immunisation procedures and processes rather than coverage. Seventeen letters and editorials14,30-46 and a review document on the role of parents and service providers47 were also excluded as they did not contain original data on immunisation coverage.
Type of publicationThe majority (72%) of the coverage studies included in the review were published in either peer-reviewed journals or State/Territory communicable diseases bulletins. Of the remainder, five were Master of Public Health treatises or PhD theses48-52 and 12 studies were either published as government reports or were included in State government annual reports.7,53-64
Study populationsMost studies were either population-based (52%) or from specific settings such as child-care (9%), schools (29 %) or population sub-groups such as Aborigines or persons of non-English speaking background (6%). Six studies which were based in clinical or related settings were included in the review, five of which are shown in Table 7.23,65-68 Many of the population based studies, including the ABS, used appropriate random sampling methods but the findings may not be generalisable beyond the population from which they were drawn. Thirty school-based studies50,62,69-96 including all measles outbreak investigations were identified, with some overlap between primary and high schools in some studies. Nine of these are presented in Table 5 and five outbreak investigations are outlined in Table 6. Ten childcare studies were included,55,97-108 six of which are presented in Table 4.
Study designCross-sectional study designs accounted for 66% of the total, followed by birth cohorts (28%), of which three were retrospective51,109,110 and two prospective.104,111 All other prospective cohort studies used data from registers in Victoria (Victorian Maternal and Child Health Nurses (VMCHN) database),7 the Australian Capital Territory (ACT) (ACT Central Vaccination Register),6 Queensland8,9 and the Northern Territory (NT).10,11
Top of page
Sample size and response ratesThe sample size in the studies varied widely, from a final sample of 69 two year old children in child-care102 to over 6,700 children in the 1995 ABS immunisation survey,5 shown in Tables 1-5. In many studies, particularly those using birth cohort methods, low response rates reduced power and generalisability. Furthermore, some studies reported significant differences in the characteristics of respondents and non-respondents, further reducing their representativeness.109,110 However, several studies using random sampling achieved response rates of above 90%.5,16,112-114 Two studies used the World Health Organization (WHO) method115 of cluster random sampling, one in a sparsely populated rural region58 and one in metropolitan Melbourne.116
Vaccines studiedThe majority of studies measured all age-appropriate vaccines with a small number measuring individual vaccines only (Table 3). Studies evaluating individual vaccines were primarily focused on recently introduced vaccines such as Hib6,61,117 and hepatitis B118,119 or were part of outbreak investigations for measles72,79,91 rubella,49,120 or pertussis.15,103
Validation of immunisation statusThe majority of studies in the Tables 1-7 validated at least a subset of their data by sighting Parent Held Records (PHRs) or by contacting the vaccination provider. A small number of serological surveys were conducted23,54,68,118,121 while immunisation databases used only data transmitted from providers.
Validation by Parent Held Record (PHR)Most parents use the PHR when it is available and refer to it as their primary source of information. Hall et al found that 80.1% of parents of children 2-4 years used the PHR which was completed correctly in most cases.58 With some exceptions,121,122 most studies in this review investigating the validity of parental report found significant differences between reported and validated levels of vaccination. For example, a study in Northern Sydney found only 60% agreement between parental report of vaccination status and provider records.110
The 1995 ABS immunisation survey used both parental recall and viewing of PHRs (60.6%) for children aged 3 months to 6 years. Forty-seven per cent of children with records available were fully immunised compared with 33.1% of all children in the survey. The ABS study classified children as unimmunised if the parent could not recall the exact number of doses for each vaccine, which is likely to have underestimated true immunisation coverage.5
A study of children attending childcare facilities in Queensland also found that those with a PHR were more likely to be fully immunised at 2 years of age.97 The PHR is less likely to be useful among disadvantaged groups, as illustrated by a study of Aboriginal families which found that there was no documentation of vaccinations for 52% of children.114
Serologic validation of parental recallA serosurvey in Western Sydney medical centres found only 74% of the children in the total sample had protective levels of measles antibody, compared with 84% from parentally reported vaccination, a positive predictive value of 84%.23 A population-based serologic survey in NSW found a non-significant difference in measles immunity between parents with written records (84% immunity) and those using parental report (76% immunity).121
Serological surveysThe small number of serological surveys largely focused on specific diseases, for example measles and rubella,23,49,54,121 including two studies54,121 that were opportunistically added to larger studies, for example the National Survey of Lead in Children. These studies gave valuable data on age-specific seroprevalence but were unable to differentiate between acquired and vaccine induced antibody (Causer et al, 1998; personal communication).
Regional immunisation registersCoverage estimates from population immunisation databases are more likely to be accurate than parental recall, but underestimates coverage if providers do not report all immunisations given. The VMCHN database7 calculates age-specific coverage rates using all births in Victoria as the denominator and all children attending Maternal and Child Health (MCH) clinics as the numerator. The data are likely to be accurate for children up to 1 year of age, as approximately 90% of all children in this age group attend the clinics,7 but as attendance progressively falls and children may be immunised by more than one provider, ascertainment through MCH clinics is less complete over this age.
Other regional databases include the NT immunisation database,10,11 the ACT Central Vaccination Register6 and the Vaccination Information Vaccination Administration System (VIVAS) in Queensland.9,21 The VIVAS system includes a vaccine distribution scheme which increases reporting of immunisation encounters. The VIVAS and ACT databases have been modified to transmit data for the ACIR on a centralised reporting basis since early 1996, but the other separate databases have been discontinued.
Top of page
School entry immunisation certificatesSchool entry immunisation certificates were made compulsory in NSW, Victoria and the ACT during the period covered in this review. A number of studies have assessed the quality of these data and estimated coverage for children entering school. Many schools do not have a completed certificate for all children in kindergarten or Year 181,96 and those certificates which are completed substantially overestimate immunisation.71
Acceptance of non-statutory evidence of immunisation was identified as the major factor leading to overestimation of compliance with the legislation.71,93,95 Two studies evaluating the effectiveness of the new legislation in Victoria and NSW found inconsistencies in the issuing and administration of certificates and evidence of schools. To reduce the impact of over-reporting, some studies excluded incomplete certificates from their analysis75,78 and one study concluded that school certificates should not be used for assessing coverage.83
Childcare immunisation certificatesImmunisation certificates at entry into licensed childcare services have recently been introduced in many Australian states, some on a compulsory basis (NSW, Victoria, ACT). In 1995 in NSW, the Statewide Sentinel Immunisation Surveillance System (SSISS) database for childcare was established, containing a systematic random sample of childcare immunisation records.101 Among 745 children aged 2 years of age, 70% of children were recorded fully immunised (excluding Hib). Prior to the introduction of legislation in the ACT in 1994, 34% of records in a childcare centre were found to be incomplete during an outbreak investigation for pertussis.103
A survey in the NT following the introduction of a voluntary program in 1995, found that centres had documentation for only 66% of children.106 Similarly, a study reviewing documentation of immunisation records in Family Day Care and childcare centres in NSW in 1995 found that many parents in Family Day Care had not submitted their certificates to the provider.102 Overall it seems likely that similar problems to school immunisation certificates (poor quality and low levels of compliance) are prevalent with childcare immunisation certificates.
Measures of immunisation status
Full immunisation, partial immunisation and timelinessThe outcome measure used in the majority of studies was 'fully immunised' and the ages most commonly used in the assessments were 24-35 months and school entry (aged 4-6 years). A summary of outcome measures for age groups of 1 year, 2 years and school entry are shown in Tables 8-10. A small number of studies did report both the coverage and timeliness of the vaccinations, with only 21% of vaccinations given on time in a Northern Sydney study (within 1-2 weeks for the infant doses and 4-6 weeks for the 12 and 18 month doses).110 In childcare, although 66% of 2 year old children were completely immunised, only 24% had been immunised on time.99,100 Several studies measured the level of partial immunisation where one or more doses of vaccine were given but others had been missed. The proportion of partially immunised children ranged from 11% to 33.5%.58,69,99,100,106,110,123
Age of assessmentMany studies showed a drop in the proportion of children immunised after the age of 12 to 18 months. In the ABS 1995 survey, 88.5% of children were fully immunised against D/T at 1 year falling to 63% at 2 years. These data indicate that while many children received the 12 month single dose of MMR, they did not receive the fourth doses of DTP and Hib at the age of 18 months. A childcare coverage survey found that coverage for the primary series of DTP and Hib was over 95%, falling at 18 months by 22%-26%.106 Similarly, coverage of only 65% for the fourth doses of DTP and Hib was found in a regional Queensland database, where coverage for the first three doses had been 81%-84%.8
Vaccines assessedMany studies reviewed reported coverage for separate vaccines, which while providing more information, made comparisons by age more complex. Moreover, while most studies reported the levels of the combined vaccine DTP, a small number reported levels of the combined diphtheria and tetanus (CDT) vaccine, which excludes pertussis, in their definition of 'fully immunised'.8,15
The major change to the Immunisation Schedule during the review period was the introduction of Hib in April 1993, with free vaccine available from July 1993. As a result, several studies5,50,113 undertaken around this time give estimates of full immunisation coverage including and excluding Hib, with large differences between the estimates. For example, a school entry study using PHRs in 1996 found that full immunisation was 47% if Hib was included but that this increased to 74.5% if Hib was excluded.50 Similarly, the ABS survey estimated that full immunisation for children aged 0-6 years was 33.1% with Hib and 52.1% without Hib.5
Immunisation status of 1 year old childrenOnly a small number of studies reported immunisation status for children at approximately 12 months of age (Table 8). There is some variation in the estimates for full immunisation as some include MMR and others exclude these antigens. Estimates from the ABS 1995 survey were 51.4% including Hib and 70.8% excluding Hib. Estimates of full immunisation from other studies were generally higher, for example the NT database reported levels of 75% including Hib.11 Where rates for DTP are reported, however, these are similar to the ABS 1995 estimates. Assessment of immunisation status in a cohort of children in Melbourne suggested that 92%-93% of children aged either 9 or 16 months were fully immunised, much higher than both the 1995 ABS estimate and estimates from the VMCHN database. However, a serosurvey assessing the prevalence of measles immunity in NSW found that 77% of children aged 12-23 months were immune, which is 10% lower than the levels reported in the 1995 ABS survey but consistent with a measles vaccine efficacy of around 90%.121
Immunisation status of 2 year old childrenTable 9 illustrates the range of outcome measures used for children of around 2 years of age in several studies dated from 1994, including the 1995 ABS survey. Most of these studies included 4 doses of DTP and OPV in their definition of 'fully immunised', but at least one of these studies116 included only 3 doses of DTP and OPV in their assessment. Estimates for both full immunisation and DTP showed a very wide range (51%-88% and 58%-93% respectively).
The ABS 1995 rate of full immunisation for 2 year old children is much lower than any other reported estimate for this age group. The other studies that estimated full immunisation, including Hib, ranged from 51.1% - 66.5% compared to 34.3% for the ABS survey. Estimates excluding Hib ranged from 60.3%-87.8%, compared to 51% for the ABS survey. A cross-sectional, population-based, cluster sample survey in Newcastle, NSW,113 reported estimates both with and without Hib of 51.1% and 77% respectively. While the reported immunisation coverage for DTP and OPV in the two Victorian studies were similar,7,116 estimates for the same vaccines differed in the NT, although one of these was a childcare centre based study.10,106
There were also major variations in coverage reported for studies using the same method of data collection. For example, the Victorian immunisation database7 showed that 85.3% of children aged 18 months to 3 years in 1996-97 were immunised for DTP while the urban Darwin database in the NT reported that only 60% of children had been immunised for DTP in 1996.10 The pattern for all other vaccines in these databases was similar with the exception of Hib and it is unclear whether these are caused by technical problems with the database or due to actual differences in coverage between the States.
Top of page
Immunisation status at school entryTable 10 shows estimates from school entry and other school surveys, which as discussed above, show a higher level of full immunisation coverage than for younger children. In WA between 80% and 86.5% of children were fully immunised (excluding Hib).81,125 In NSW, 89% of children were classified as fully immunised at school entry, based on parental reports.69 Overall, the estimates shown in Table 10 are higher than those from the ABS 1995 survey but are likely to be overestimates (see Methods).
Childcare settingsStudies based in childcare centres using provider held documentation, show estimates of full immunisation for 2 year olds ranging from 60.3%-85%.97,99,100,106-108 One study in the NT found that 67% of 2 year old children with childcare immunisation certificates were fully immunised and that this rate was higher than in the general population of 2 year olds in Darwin,10 although estimates in Darwin were lower than those using data from all seven NT databases.11 A prospective cohort study in Perth found higher rates of full immunisation (excluding Hib) than other comparable studies, with levels ranging from 86% at 12 months to 85% at 24 months.104,105 However, these data were based on parental report only and are likely to be overestimates.
Immunisation status in States and TerritoriesState by State coverage results from population-based studies and for children aged 2 years are reported in Table 11 and are compared with the ABS 1995 results. In all regions, with the exception of Tasmania and one study in the NT, the fully immunised coverage estimates from studies other than the ABS are much higher. This appears to be largely due to low proportions of DTP in the ABS data, (48%-66%). Rates for OPV and MMR, however, are largely comparable with the ABS estimates and as expected, Hib rates are lower in the ABS than the other studies. Interestingly, the ABS estimates vary considerably between States, with Tasmania showing full coverage of only 23.9% (including Hib) compared to 42.5% in WA. Estimates without Hib for these two States were 37.3% and 58.3% respectively.
HibTwo studies in the review assessed the pre and post uptake of Hib after its introduction in 1993 and one of these also compared the age-specific incidence of Hib with changes in uptake of Hib.6,61,122 In Sydney, uptake of Hib was estimated at 9% in May 1993 rising to 48% in August 1993 for children under 18 months of age and rising from 31% to 45% in the same period for children aged 19-60 months.61,122 The ACT immunisation register estimated that 68% of 9 month old children and 34% of 2 year old children had received Hib vaccine by March 1995.6
Hepatitis BThree studies assessing the level of coverage of hepatitis B in specific populations, including Aboriginal and non-English speaking groups were involved in the review.68,118,119 Coverage was generally low, with only 54% of Aboriginal children (median age of 24.5 months) in North Queensland immune to hepatitis B.118 In one study however, 81.6% of infants from 'at risk' groups in Victoria received one or more doses of hepatitis B vaccine.119
Aboriginal communitiesSome studies assessing coverage in Aboriginal children from remote communities in the NT and WA suggested a higher than average rate of immunisation.112,125 For example, the NT study estimated that over 97% of 2 year old children had been immunised for all vaccines other than fourth doses of DTP and polio.112 Conversely, studies in urban and less remote rural areas found a much lower level of coverage in Aboriginal children compared to non-Aboriginal children from the same populations.57,114 For example, a study in western NSW estimated that only 60% of Aboriginal children aged between 2 and 4 years of age were fully immunised compared to 84.1% of the non-Aboriginal children.57 Furthermore, a study in the North Coast region of NSW reported ABS estimates which showed the immunisation status for Aboriginal children was only half that of the overall population.114
Top of page
Evaluating Australian coverage studiesThe 51 higher quality studies summarised in Tables 1-7 included 38 published in peer-reviewed journals, the two ABS surveys (1989-90 and 1995)5,16 and 12 government reports, treatises and abstracts. Attempts to evaluate all these studies were difficult for reasons previously outlined in this review due largely to variations in age groups, vaccines studied and definitions of immunisation status, in addition to study design and other criteria used for assessing quality. Generalisability and reliability at a national population level were key determinants of quality and in this review only the ABS surveys fulfilled these criteria.
The studies with highest validity at a State level were databases using provider-held records and calculating coverage from a prospective birth cohort, such as the VMCHN database,7 the ACT vaccination database6 and the NT database.10,11 The NT database was also sensitive enough to give data on specific small populations such as remote Aboriginal communities.112
The second category of coverage studies was cross-sectional studies with appropriate sampling and high response rates that are likely to have high validity for the specific populations included. These included a cluster sample from metropolitan Melbourne,116 a cluster sample from Newcastle, NSW113 and a cross-sectional study of Hib vaccine coverage in Sydney.61,122 A household study in central Sydney investigating coverage of measles and rubella used serology to validate parental report and while this provided optimum validation it was only generalisable to the local area.53 All these surveys had positive features but generalisability of the findings was limited by the differences in vaccine delivery and uptake between regions in Australia.
The third category of coverage surveys was retrospective birth cohorts, which shared the problem of low response rates and selection bias. Response rates in Sydney,110 Western Australia51 and Queensland109 were 49%-58%. It is likely that respondents have higher immunisation coverage than non-respondents and that this study design will overestimate coverage.
The ABS immunisation survey remains the reference standard in this review for generalisability and reliability. This is followed by State immunisation databases and then a small number of coverage surveys. However, all these designs exhibit different biases and comparisons are difficult. A national, prospective birth cohort design in which data is complete is clearly the ideal method, complemented by ad hoc coverage surveys to assess coverage in small populations with special needs such as urban Aboriginal communities.
Comparison of data from the ACIR with the ABSTable 12 compares the proportion of children fully immunised with DTP, OPV and Hib in each State and Territory using the ABS 1995 survey and the ACIR coverage estimates.18 The estimates for DTP and OPV from the ABS are much higher than those for the ACIR for all States/Territories with the exception of Queensland, probably reflecting the more complete data available from the VIVAS reporting system which is linked to the vaccine supply.
It is important to note that ABS estimates for full immunisation at 1 year were derived from a cross-sectional sample at 12-23 months of age whereas the ACIR reports the status of a birth cohort at 12 months of age. This means that children who received the third dose of a vaccine scheduled in the first year of life after 12 months of age are deemed immunised by the ABS but not by the ACIR, which may slightly increase the ABS estimates.
As expected, Hib coverage is lower in the ABS estimates than in those from the ACIR because the ABS survey was conducted soon after the introduction of Hib vaccine, while ACIR data refer to a 1996 birth cohort (Figure 1). The comparison shows how rapidly uptake of Hib vaccine occurred, with an average increase of 17% for all of Australia. The only region with similar Hib vaccine estimates in both the ACIR and the 1995 ABS survey is the NT, where the two dose primary course and widespread publicity may have promoted early Hib uptake.
Figure 1. Immunisation coverage for one year olds: ABS 1995 and ACIR 1998
Overall, ACIR estimates should be viewed as preliminary minimum estimates consistent with the estimates from the ABS survey. Coverage as measured by the ACIR is likely to improve dramatically with the introduction of a range of incentives for parents and providers to immunise and to report to the ACIR.
Top of page
Comparisons with overseas studiesImmunisation coverage estimates overseas in comparable industrialised countries such as the UK and the USA are higher than in Australia. Estimates for five other English speaking countries are shown in Table 13. The vaccination coverage statistics for children at 12 months of age in the UK for three doses of DTP, OPV and Hib are between 91.7% and 92.8%, compared with estimates from the ABS 1995 survey of 83%-86% for DTP and OPV (Figure 2).126
Figure 2. Immunisation coverage at age one year: international comparisons
Coverage for children aged 2 years in the UK is higher than at 12 months, contrasting with the situation in Australia in which a reduction in coverage occurs after 12 months of age. However, the UK immunisation schedule does not include a fourth dose of DTP or Hib at 18 months, in contrast to Australia. In the UK, regional databases submit immunisation data to a centralised register from which quarterly reports are prepared. This system commenced in 1987 and uses the birth cohort method to analyse data, which is similar to the ACIR. It also incorporates a financial incentives scheme for GPs to encourage high immunisation levels and the British national target now exceeds 90% coverage for each antigen.127
In the USA, the National Immunisation Survey (NIS)128 was initiated in 1994 as the main method to estimate coverage for children aged 19-35 months. In this quarterly random telephone survey, response rates are around 67% and providers are also contacted for verification of immunisation status. The 12 month estimate for the period January to December 1997 was 95% for three doses of DTP or CDT, 91% for OPV and 93% for Hib and 91% for a measles-containing vaccine (MCV).128 The equivalent proportions for the 1995 ABS survey are comparable for measles but much lower for DTP (at only 63% for diphtheria /tetanus and 58% for pertussis) and also for OPV (87%). Estimates of full immunisation at around 24 months of age show that 78% of children in the USA were fully immunised in comparison to 51.3% of children in Australia.128
Prior to the implementation of the NIS, coverage was measured by ad hoc surveys and it was found that while most children (87%) were fully immunised at school entry because of legislation requiring compulsory immunisation, the proportions for children aged 2 years old were much lower at 44%.129 Coverage for very young children in the USA has therefore increased substantially since 1994.
The Canadian experience mirrors that of the USA in that immunisation estimates were not collected in a standardised manner until 1994, when a system which collects data by mailed questionnaire for four cohorts of children turning 2 years of age during 1994-96 was implemented.130 Coverage for this period ranged from 85%-87% for four doses of DTP, and was 90% for polio. Coverage for MMR was high at 97% and has remained fairly constant while the lower rates for Hib vaccine reflect its introduction during the period of data collection. Vaccine specific estimates increased by 1% -3% above baseline in the period from 1994 to 1996.130
New Zealand immunisation estimates were obtained by coverage surveys and by health benefit claim data. Immunisation estimates in the 1990s show a progressive increase in coverage from a relatively low level. For example, a cohort study of children born in 1990-1991 found a complete immunisation rate of 75% by 6 months of age131 while a more recent cohort study in Christchurch in 1995 estimated that 93% of children had been fully immunised by 8 months of age.133 Health Benefit data from 1994 are shown in Table 13, with an estimated a range of 79%-87% for most vaccinations for children aged 12-18 months.133 Recent New Zealand coverage data for the same age group indicates that coverage levels of DTP/Hib in 1997 dropped slightly to 87.3% after an overall increase to 90.6% in 1996.134 This may be due to changes in processing claims rather than a real decrease. Coverage for Maori and Pacific Islander children was lower in northern New Zealand than for all other children, reflecting a similar pattern similar to Aboriginal children in Australia.58,114,135
While there is some variation with the immunisation schedules in some overseas countries, the available data suggest that the UK, USA and Canada outperform both Australia and New Zealand particularly for immunisation with pertussis containing vaccines. The differentials are especially large when comparing coverage for the fourth dose of DTP, with estimates in Australia falling well behind those of the other countries in Table 13. The UK appears to have the highest coverage levels of all the countries surveyed, with estimates of between 90%-95% for all antigens shown in this table. While the Australian data are not as recent as all the other estimates in Table 13, they remain the only population-based estimate available for comparison, while the ACIR is still in the developmental stage.
Coverage levels in most other countries in the world are reported on the WHO Internet site.136 These data are obtained from routine national reports without details of methodology and are therefore difficult to compare with those outlined above.
Top of page
DiscussionThe best estimate of immunisation status in Australian children, or reference standard, in 1998 is the 1995 ABS survey. It is the only national coverage study that is generalisable to the whole population, despite concerns about reliance on invalidated parental report for almost half of the responses. However, it is likely that the ACIR will take over this role when it becomes fully established.
The estimate for full immunisation in the ABS study for children aged 2 years (51%), even when Hib was excluded from the analysis, was substantially lower than those in other coverage studies in this review (range 60% to 88%). This appears to be due largely to lower reported levels of DTP rather than the levels of OPV and MMR. With the exception of serological surveys which tend to focus on assessing coverage for single vaccines in specific populations, the ABS was also the best estimate of coverage for specific vaccines and 'milestones'.
When estimates from the ABS 1995 survey were compared with the first 'milestone' data from the ACIR (Table 12), the ABS estimates were higher than the ACIR for most vaccines and some statewide differences did emerge in this comparison. However, any meaningful interpretation is complicated by technical difficulties with the ACIR resulting in underestimates of coverage at present. This may improve with the introduction of financial incentives to GPs based on their performance as recorded by the ACIR, from July 1998. The exception was the difference between the low levels of Hib in the ABS 1995 survey (during the phasing-in of the vaccine) and the higher estimates of the ACIR. This comparison is useful as it shows how quickly the uptake of Hib has occurred in younger Australian children.
This review included both published and unpublished literature, reducing the possibility of publication bias resulting in an overly optimistic view of immunisation coverage. The extent to which the large variation in coverage estimates was attributable to methodological problems in study design and analysis was often impossible to assess, as very few publications provided adequate data. It was difficult to determine whether the net effect of a particular study had been to underestimate or overestimate vaccine coverage.
The higher quality studies were larger and population based. Retrospective cohorts suffered from low response rates and selection bias, leading to an overestimate in coverage. Cross-sectional studies, although largely free of bias, provided only a snapshot at one point in time and could not monitor trends. Most of the immunisation databases cited in the review were derived from prospective birth cohorts, as was the ACIR and the well established COVER database in the UK. This method is clearly the best design if reporting is complete, and provided there has been adequate time for the operation of the database to become established.
Serological surveys, while the most accurate method for validating coverage, have sample sizes that are necessarily limited and therefore differences between population subgroups cannot be examined. At the whole population level, serological surveys have been shown to be very useful in demonstrating trends in susceptibility to vaccine preventable diseases, as in the UK.136 The first population-based serological surveys using opportunistically collected sera are currently underway in Australia sponsored by the National Centre for Disease Control and conducted by the NCIRS. These surveys will be repeated on a two or three-yearly basis.
In contrast, parental recall is the simplest means of estimating immunisation status, but frequently overestimates coverage. The PHR is more accurate than parental recall, but is dependant on the quality of the information added to the record and the book being kept up-to-date. Well maintained provider-held records overcome these problems but must be adequately linked for children who have multiple providers and/or high mobility. Providers are the source of data for the State/Territory immunisation databases in this review, for the ACIR and for coverage in the UK126 and the USA.128
As methodology strongly influences the final estimates, this review supports the need for a nationally consistent methodology, facilitating comparisons between regions. A national, prospective birth cohort design in which data are complete is clearly the ideal method, complemented by ad hoc surveys to assess coverage in small and specific populations, for example urban Aboriginal communities.
Immunisation coverage in developed countries overseas, particularly the UK, is reportedly higher than in Australia (over 90% for all antigens at 24 months). Both the USA and Canada collect national data regularly using national surveys. The USA has recorded increased uptake in infants and very young children since the implementation of their quarterly surveys. The UK has a well established centralised immunisation database which reports coverage data on a quarterly basis. The use of 1995 data from Australia, however, meant that it was not as up to date as the other overseas estimates. A fundamental requirement in comparing such estimates is the need for timely, accurate Australian data which should be achievable through the ACIR.
Failure to develop a regular system of measuring immunisation coverage which is high quality and generalisable at a national population level will result in continued confusion about the true levels and trends of coverage in Australia. It is only by having a sensitive and timely system that interventions to improve and maintain coverage can be assessed and gaps identified. Only when true coverage rates regularly exceed 90% can we hope to achieve and maintain the levels of herd immunity needed to interrupt transmission of vaccine preventable diseases in Australia. We will then be able to eliminate diseases such as measles and take a responsible role in the world-wide eradication efforts planned by the World Health Organization.
This article was published in Communicable Diseases Intelligence Volume 23, No 6, 10 June 1999.