The methods used in this report are adapted from and align with the methods previously applied in the series of reports on Vaccine preventable diseases and vaccination coverage in Aboriginal and Torres Strait Islander people, Australia.13,132
General issues regarding data on vaccine preventable diseases
Three sources of routinely collected data were used for this report. Notification data were obtained from the National Notifiable Diseases Surveillance System (NNDSS) and hospitalisation data from the Australian Institute of Health and Welfare (AIHW) National Hospital Morbidity Database. Mortality data were provided by the Australian Bureau of Statistics (ABS).
Comparisons between the notification, hospitalisation and death data should be made with caution since these datasets differ in their purposes of collection, reporting mechanisms, accuracy, timeliness and period of reporting.
In this report, in order to provide the most recent information available, and to accommodate the varied reporting formats, data of different time periods (representing the available data for the most recent years since the previous report in the series) have been selected from each dataset for review. As there are no unique identifying codes to link records for the same individuals across these datasets, and due to differences in defining a case and in the completeness and the accuracy of the data in each dataset, it is not possible to interpret deaths and hospitalisations as subsets of notifications.
For some diseases, there are no specific ICD codes that correspond to the particular disease condition of interest. This will limit the validity of comparisons between notification data and hospitalisation or death data (e.g. for invasive pneumococcal disease). The algorithms used to select surrogate ICD codes to match the hospitalisation case definitions for invasive pneumococcal disease are explained in the disease chapter. Hospitalisation data for invasive Haemophilus influenzae type b disease has not been presented because no type-specific ICD code exists.
Indigenous status identification in the datasets used in this report
In this report, the term ‘Indigenous status’ refers to a field in a dataset in which information is recorded on whether or not a person has been identified as Aboriginal and/or Torres Strait Islander. It is reported as either one of two categories: an individual who was identified in a record as an Aboriginal and/or Torres Strait Islander person is categorised as ‘Indigenous’; an individual whose Indigenous status was recorded as non-Indigenous or was not stated or inadequately described is categorised as ‘Other’.
The quality of Aboriginal and Torres Strait Islander health statistics depends on the accuracy of Indigenous population estimates and completeness and reporting accuracy in the collection of Indigenous status information for the disease of interest. Considerable work has been done in recent years by agencies such as the ABS, AIHW and state and territory governments on assessing and improving the quality of Aboriginal and Torres Strait Islander statistics in national and state and territory administrative data collections.133-136 More work is needed to improve the quality of the data, as there are large variations in quality between data collections (particularly for notifications), and within the same data collections, there are variations between jurisdictions and over time.
The NNDSS was established in its current form in 1991, and includes de-identified information about cases of notifiable vaccine preventable diseases (VPDs) notified to state and territory authorities under their respective public health legislation. Prior to 2004, state and territory notification criteria were based on the National Health and Medical Research Council (NHMRC) surveillance case definitions,137 with various modifications applied in different jurisdictions. Since 2004, all jurisdictions have applied the new national case definitions for notifiable diseases endorsed by the Communicable Diseases Network Australia.138 The case definitions for notifications for each of the included VPDs are described in Appendix B.
The data collected by the NNDSS are continually updated by jurisdictions. There could be minor variations between NNDSS data in this report and annual reports of the NNDSS (Australia’s Notifiable Disease Status reports)16 and other reports that include national notifiable diseases data, since different data versions were used for analysis. In this report, disease notifications primarily consist of cases with a date of diagnosis between 1 January 2007 and 31 December 2010 (4 years), as at August 2011. Historical notification trend data included in this report have been updated from previous reports in this series.13,132 Previous reports on data prior to 2005 analysed notifications by date of onset as collected from the clinical history, where available, or the specimen collection date for laboratory-reported cases. As of mid-2005, a date of diagnosis field was generated for all NNDSS records. For each notification record, a date of diagnosis is derived from the date of onset, or, where that is not supplied, the earliest date recorded among the following fields: date of specimen, date of notification, or date when the notification was received (the only mandatory date field).16
The variables extracted for analysis of each disease were: date of diagnosis, Indigenous status, age at diagnosis, and the state or territory from which the notification was received. Data for specific serotypes/serogroups of the causal organisms have been presented for invasive pneumococcal disease and meningococcal disease. Following an assessment of the completeness of reporting in the Indigenous status data field (see below), all jurisdictions were included for reporting of VPD notifications in this report for the period 2007 to 2010. This contrasts with previous reports in this series where notification data from only selected jurisdictions were analysed and reported.
Notification data are presented for invasive Hib disease, hepatitis A, acute hepatitis B, measles, meningococcal disease and pneumococcal disease in their respective chapters, and rare diseases (diphtheria, tetanus, poliomyelitis and rubella) in a combined chapter. Summary data are presented in Appendix C. No notification data are presented for influenza, mumps or pertussis, due to the low level of completeness of the Indigenous status field across most jurisdictions. Data on rotavirus and HPV disease notifications have not been included in this report as these diseases are not nationally notifiable. Notification data for varicella-zoster infections (including chickenpox and herpes zoster) have not been included in this report because national notifications are not available prior to vaccine funding, and due to the low level of completeness of Indigenous status in these data.
Indigenous status identification in notification data
The proportion of notifications that lack identification of Indigenous status was examined by jurisdiction, year and disease. An acceptable level of completeness of Indigenous status identification in the records was defined as at least 60% for a substantial majority of the diseases analysed. This level of completeness was achieved for all jurisdictions during this reporting period of 2007–2010. After establishing that notification incidence estimates were not dominated by any one of the jurisdictions (data not shown), estimates are presented for all jurisdictions combined.
Tasmania and the Australian Capital Territory were excluded from the disease trend data and graphs covering the period 2000–2010, due to varying but generally low levels of Indigenous status identification over the whole of this period.
Other issues to be noted when interpreting notification data
A major limitation of the notification data is that they represent only a proportion of all the cases occurring in the community, due to under-reporting. This proportion may vary between diseases, over time, and across jurisdictions.16 An infectious disease that is diagnosed by a laboratory test is more likely to be notified than if it is diagnosed only on clinical grounds. Data accuracy may also vary among jurisdictions due to the use of different case definitions for surveillance (prior to adoption of the national case definitions) and varying reporting requirements and mechanisms by medical practitioners, hospitals and laboratories.
The AIHW National Hospital Morbidity Database has received administrative, demographic and clinical information about patients admitted to public and private hospitals in Australia since 1993. Almost all hospitalisation episodes in public and private hospitals are captured.139 Data are received by financial year of hospital ‘separation’ (the process by which an admitted patient completes an episode of care by being discharged, dying, transferring to another hospital, or changing type of care).139 The four most recent (financial) years for which data were available (2005/2006, 2006/2007, 2007/2008, 2009/2010) are included in this report. Available data for analysis were final, and future revisions are not anticipated. Trend data for pneumococcal disease and ‘influenza and pneumonia’ by age group have been analysed and presented by calendar years rather than financial years to facilitate comparison with notification data, taking into account the seasonality of these conditions.
Data from jurisdictions in which Indigenous status identification of hospitalisation records exceeded 80% within the reporting periods are included for reporting of hospital rates and rate trends by Indigenous status, in accordance with AIHW recommendations136,139 (see ‘Indigenous status identification in hospitalisation data’ below). For hospitalisation records where data on the jurisdiction of residence are missing, the jurisdiction in which the hospitalisation occurred is used to replace the jurisdiction of residence datum.
Data were extracted based on the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD-10-AM). Most states and territories began using ICD-10-AM in 1998/1999, and since 1999/2000, all jurisdictions use the new classification. The codes used to select the specific condition(s) for reporting of each of the included VPDs are described in Appendix B and/or in the respective disease chapter. Eligible separations included those with the code of interest listed in the principal diagnosis (the diagnosis chiefly responsible for the episode of hospital admission) or in any other additional diagnosis fields (i.e. conditions or complaints either coexisting with the principal diagnosis or arising during the episode of care).139 For hepatitis B, only hospitalisation records with acute hepatitis B as the principal diagnosis were included, consistent with previous practice in this report series and the series of national surveillance reports on VPDs.13,24,132,140-143
The variables extracted for analysis of each disease were: age at admission, state or territory of residence, Indigenous status, year of separation, and separation (discharge) diagnoses (principal and additional diagnoses – up to 31 diagnoses prior to 2003/2004, and up to 50 diagnoses since 2003/2004, were recorded for each hospital separation).
Hospitalisation data are presented for hepatitis A, acute hepatitis B, influenza, influenza and pneumonia, measles, meningococcal disease, mumps, pertussis, pneumococcal disease, rotavirus and varicella-zoster infections (chickenpox and herpes zoster, separately) in their respective chapters, and rare diseases (diphtheria, tetanus, poliomyelitis, and rubella) in a combined chapter. Summary data are presented in Appendix D. No hospitalisation data are presented for invasive Hib disease, as there is no type-specific code for invasive Hib disease within the ICD-10 classification system.
HPV hospitalisations are not included in this report, as the conditions measured by hospitalisation are of limited value in monitoring HPV-related disease. The most appropriate indicators are Pap test abnormalities and genital warts. The relevant datasets for these are outside the scope of this report and they have been reported on elsewhere.144,145 However, data on Aboriginal and Torres Strait Islander people has not been published.
Indigenous status identification in hospitalisation data
The previous (second) report in this series included aggregated hospital separation data from five jurisdictions: New South Wales, the Northern Territory, Queensland, South Australia and Western Australia.13 In this report, data from six Australian jurisdictions (all except Tasmania and the Australian Capital Territory) were included for reporting of aggregated hospital separations and rates by Indigenous status for the 4-year period July 2005 to June 2010, as recommended by the AIHW.136 (About 96% of the Aboriginal and Torres Strait Islander population resides in these six jurisdictions.) Subsequently, it has been recommended that data from 2011–2012 onwards may include all jurisdictions.146 For trend reporting for the period 1999/2000–2009/2010, data were included for only four jurisdictions (the Northern Territory, Queensland, South Australia and Western Australia), where Indigenous status data quality has been demonstrated as satisfactory over the whole period. (About 60% of the Aboriginal and Torres Strait Islander population resides in these four jurisdictions.) This data inclusion/exclusion selection is consistent with AIHW recommendations regarding acceptable data quality for analysis based on Indigenous identification.136,139 Data from these selected jurisdictions are not necessarily representative of Aboriginal and Torres Strait Islander people living in other jurisdictions.139
Jurisdictional differences in data quality, including the degree of Indigenous under-identification, should be considered when interpreting the results. The analysis of hospitalisation rates over time should also be interpreted with caution, as hospitalisation rates for Aboriginal and Torres Strait Islander patients may be affected to a varying degree by improved identification over the period being analysed.133,136
Other issues to be noted when interpreting hospitalisation data
Hospitalisations generally represent the more severe end of the morbidity spectrum of a disease, and the extent to which ICD-coded hospitalisation data can reflect the burden of the disease of interest varies between diseases.
There are also limitations associated with the use of ICD codes to identify cases. Errors that cause the ICD code to differ from the true disease include both random and systematic measurement errors. These errors may occur either along the patient pathway (e.g. level of details documented in medical records, clinicians’ experience) or along the paper trail (e.g. transcribing errors, coder errors such as miss-specification, unbundling (assigning codes for all the separate parts of a diagnosis rather than the overall diagnosis) and upcoding (using reimbursement values to determine the order of coding)).147 It is difficult to gauge the relative importance of hospitalisations where the coded disease of interest was not the principal diagnosis but was recorded as an additional diagnosis for that hospitalisation episode.
In the National Hospital Morbidity Database, there is one record for each hospital admission/separation episode. This means that there are separate records for each readmission or inter-hospital transfer. This is unlikely to have a major impact on the numbers reported for most diseases reviewed in this report, as they are mostly acute illnesses, but the implications of the potential but unquantified impact of this limitation need to be considered for each VPD individually.
Hospitalisation data may also be affected by variations in admission practices over time, between public and private sectors, and across various states and territories.139 Variation in availability and access to hospitals across different geographic regions should also be noted when comparing and interpreting hospitalisation data for different population groups.
The registration of deaths is the responsibility of the eight individual state and territory Registrars of Births, Deaths and Marriages. As part of the registration process, information on the cause of death is supplied by the medical practitioner certifying the death, or by a coroner. The information is provided by individual Registrars to the ABS for coding and compilation into aggregate statistics. In addition, the ABS supplements this data with information from the National Coroners Information System.148
Since 1997, the International Classification of Diseases, 10th Revision (ICD-10) has been used to identify the cause of death. The problems associated with the accuracy of ICD coding used for hospital separations, discussed above, may also be relevant for mortality data. The codes used to select the specific condition(s) for reporting of each of the included VPDs are described in Appendix B.
Unit file records of registered deaths were not available for analysis for this report. Unpublished mortality data on selected VPDs, by Indigenous status, aggregated for the period 2006–2010 were obtained directly from the ABS. In this current data release, 2006, 2007 and 2008 data are final, but 2009 and 2010 data are subject to further revisions.148
Due to incomplete identification of Indigenous status in the records, data on deaths in Aboriginal and Torres Strait Islander people are included from five jurisdictions only (New South Wales, the Northern Territory, Queensland, South Australia and Western Australia), as per recommendations of the ABS.148 In November 2010, the Queensland Registrar of Births, Deaths and Marriages advised the ABS of an initiative which resulted in the registration (in 2010) of some 374 previously unregistered deaths that occurred between 1992 and 2006 or at an unknown time. Of these, around three-quarters (284) were deaths of Aboriginal and Torres Strait Islander Australians.148 After consulting with the ABS, it was decided that, for the purposes of this report, these death records would be excluded from the dataset from which deaths attributed to VPDs were derived.
This report includes data on death records where the disease of interest was documented as the ‘underlying cause of death’ (i.e. the single disease that ‘initiated the train of morbid events leading directly to death’),148 and separately where a disease was one of the multiple causes of death (i.e. ‘either the underlying cause, the immediate cause, or any intervening causes, and those conditions which contributed to death but were not related to the disease or condition causing death’).148 In this report, deaths where the disease was one of the multiple causes, but not the underlying cause, are referred to as a ‘contributing’ cause.
As per ABS requirements to protect the confidentiality of individuals,148 some of the exact counts of <5 deaths in aggregate cannot be reported, but instead are reported as a range. Additional counts have to be reported in ranges to prevent possible back calculations. For diseases where the total number (or subtotal by age group) of recorded deaths from selected jurisdictions is small, only the total counts or range for all ages by Indigenous status are reported. Where possible, for selected diseases, death counts are reported by three age groups (<5 years, 5–49 years and ≥50 years) by Indigenous status.
Mortality data are reported and analysed by the year in which the death was registered rather than by the year in which the death occurred. This avoids problems associated with incomplete data for the latest available year. In recent years, less than 5% of deaths in a particular calendar year are registered in the subsequent year,149 the bulk of which are deaths that occurred in December of that calendar year.
In Australia, information on the cause of death is reported routinely for every death on a standard Medical Certificate of Cause of Death completed by a medical practitioner or a coroner. The person completing the certificate must nominate the underlying (principal) cause of death and any associated conditions.149 The accuracy in ascertaining the cause of death may vary according to the experience of the practitioner, the complexity of the disease process, the circumstances of the death, and whether post-mortem autopsy was performed. Studies comparing clinical and autopsy diagnoses have found that infectious diseases were not uncommonly a missed or discordant diagnosis, although vaccine preventable diseases were not specifically identified.150,151 In the case of pertussis and tetanus, studies have documented that deaths due to these diseases, which can be otherwise identified through disease surveillance systems and hospitalisation records, sometimes go unrecorded on death certificates.152,153
Calculations and statistical methods
Calculation of rates
All rates were calculated using the mid-year estimated resident populations released by the ABS as the population denominator. Rates are presented as annual rates or average annual rates per 100,000 total population, or population by Indigenous status and by age groups as appropriate.
Age-standardised rates by Indigenous status, for all age groups in aggregate, are also reported. The direct standardisation method is used to calculate rates for all age groups combined, using the ABS 2006 population estimates as the standardising population. Interpretation and comparison of the standardised rates for each disease between ‘Indigenous’ and ‘Other’ Australians should take into account the limitations of age-standardised rates in representing the overall disease burden, including the issues arising from small number of events, age distribution of events, misclassification, population size and distribution, and under-identification of Indigenous status. Accordingly, rates were not standardised when case numbers were less than 20.154
Rate ratios for Indigenous versus non-Indigenous Australians were calculated for most of the reported diseases, including age-specific rate ratios, where appropriate.
STATA version12 was used for statistical analysis and to calculate 95% confidence intervals for rates. The method of Draper was used for confidence intervals for age-standardised rates.155
It is also important to note that high disease rates may be observed even with small absolute numbers of cases in jurisdictions with small populations (e.g. the Australian Capital Territory, the Northern Territory, Tasmania), and a small change in the numbers may result in a relatively large change in rates.
The statistical significance of time trends was assessed based on whether or not confidence intervals for individual years overlapped.
Population denominators for calculation of rates
For notification data, all rates were calculated using the mid-year estimated resident populations for the corresponding calendar year for the respective age group and/or jurisdiction as the population denominator.
For hospitalisation data, the mid-year population estimate for the first half of the financial year was used as the denominator; for example, the 2009 mid-year population estimate is used to calculate rates for 2009/2010. (This is consistent with previous reports in this series.)
Estimates from the B series of ABS experimental estimates and projections, Aboriginal and Torres Strait Islander Australians, 1991 to 2021 (based on the 2006 Census)156 were used as the population denominators for calculations of rates for Indigenous people. This differs from AIHW recommendations to use the most recent Census with a year ending in one (1), i.e. 2001.154 Therefore, if age-standardised rates are produced by other organisations that cover the same time periods and groups of ICD codes as this report, there may be some inconsistencies between the two. However, it was thought more appropriate to use the 2006 Census as those population estimates are closer to the periods analysed in this report.
The population denominators for calculation of rates for ‘other’ (presumed non-Indigenous) people were derived by subtracting the corresponding estimates for Aboriginal and Torres Strait Islander population of the relevant jurisdictions and age groups from the estimates of the total ABS-estimated resident population as at June of the corresponding year (based on latest estimates as at November 2011).
This publication is available as a downloadable document.