Use of Routinely Collected National Data Sets for Reporting on Induced Abortion in Australia examines the utility of the available routinely collected national data sources for enumerating induced abortion in Australia. It outlines a methodology for estimating the number of induced abortions in Australia using data from Medicare and from the AIHW National Hospital Morbidity Database.
This report will be of particular interest to health planners and researchers and those interested in the utility of national data collections relating to health service provision.
This report examines the utility of the available routinely collected national data sources for enumerating induced abortion in Australia. It outlines a methodology for estimating the number of induced abortions in Australia using the Medicare data and the National Hospital Morbidity Database (NHMD) data. Both data sets were used because neither has complete ascertainment of induced abortion.
The Medicare data includes information on services provided to patients other than those admitted to hospital, and to private patients admitted to hospital, for which Medicare claims have been presented and processed. The NHMD data includes information on almost all hospitalisations in Australia. Private patients treated as admitted patients in hospitals are included in both data sets.
Induced abortion may be defined as the termination of pregnancy through medical or surgical intervention (WHO 2005; FIGO 1999). The number of induced abortions in Australia was estimated using data from the NHMD for admitted patients in all states and territories, and Medicare data for out-of-hospital services for those states and territories in which abortion services are provided in non-hospital facilities as well as in hospitals. The alternative method of using Medicare data for all claims for induced abortion services whether in-hospital or out-of-hospital, and data for public patients treated in hospitals from the NHMD was examined, but it was found that this method could not be used because induced abortion could not be specifically identified in the Medicare data for private patients admitted to hospital.
The methodology developed for this report will be used by the AIHW to regularly report on the estimated number of induced abortions in Australia.
Criteria for extracting data on induced abortion from the NHMD were developed. They are females with:
Both a diagnosis code and an abortion-related procedure code are required because correct coding of induced abortion requires both codes to be assigned and neither the procedure codes nor the diagnosis codes are specific for induced abortions. The procedure codes represent procedures that are undertaken for induced abortion and for other reasons (such as following spontaneous abortion). The diagnosis codes may be assigned when a patient is admitted for an abortion procedure but, for some reason, the procedure is not carried out. The presence of both an abortion-related procedure code and an abortion-related diagnosis code effectively provide two pieces of information that indicate an induced abortion has occurred, rather than only one, less specific piece of information.
These criteria may over-estimate the number of induced abortions because separations where it was not specified that the Medical abortion was complete are included. Under-enumeration may result from the exclusion of a relatively small number of separations with a diagnosis of O05 Other abortion or O06 Unspecified abortion and the possible non-use of codes O04.5–O04.9 Medical abortion, complete or unspecified for cases with gestation of more than 20 weeks.
These criteria were validated using data from the abortion notifications data collections in South Australia and Western Australia and found to be satisfactory for enumeration of induced abortion in the NHMD. Overall, the discrepancies between the NHMD data for South Australia extracted using these criteria and the data reported to the abortion notification data collection in South Australia were relatively small (0.2% more induced abortions reported to the NHMD than the notifications in 2002 and 0.3% fewer in 2003). For Western Australia, the discrepancies were larger compared to those for South Australia, with 10.2% more induced abortions carried out in hospitals reported to the notifications data collection than were reported to the NHMD in 2002 and 5.4% more in 2003.
The classification of induced abortion in the Medicare data was examined. A number of MBS-item numbers were considered to be either intended to be used for induced abortion or theoretically related to induced abortion (see Chapter 2). However, none of them are specific for induced abortion because they could be applicable to other types of pregnancy with abortive outcomes. There is no diagnostic information available in the Medicare data, so no indication for the procedure is available.
To achieve complete coverage for induced abortion in Australia, both the NHMD and the Medicare data sets are needed, because neither has complete coverage of induced abortion. It was proposed that an estimate of induced abortion in Australia could be determined, either by adding non-hospital services in the Medicare data to separations in the NHMD, or by adding public patient separations in the NHMD to hospital and non-hospital services in the Medicare data.
The latter method cannot be used because, as described above, induced abortion cannot be specifically identified in the Medicare data. Therefore the number of induced abortions would likely be over-estimated and the degree of this over-estimation cannot be determined using the available data.
The former method could be used for the estimate if the following assumptions are made:
If these assumptions are correct, then the number of induced abortions in Australia could be estimated by adding the number of non-hospital services for MBS-item 35643 Evacuation of the contents of the gravid uterus by curettage or suction curettage in the Medicare data to the number of separations with induced abortion in the NHMD data.
Hospitals included in the Medicare data differ from those in the NHMD data. In the Medicare data, some hospitals, although licensed by the relevant state or territory health authority, are not declared by the Commonwealth for Medicare and private health insurance purposes. Therefore, services provided in these hospitals would be classified as non-hospital services in the Medicare data provided for this report, and would be included twice (because they would also be included in the NHMD which includes data from hospitals licensed by the state and territory health authorities), thus potentially over-estimating the number of induced abortions. In the NHMD, the coverage of private hospitals is incomplete for some jurisdictions, so the number of induced abortions may be under-estimated.
The hospitals included in the Medicare data and the NHMD data were different among the states and territories, and legislation affecting where abortion services are provided also varies among the states and territories. Therefore, the states and territories were considered separately when developing the methodology for estimating the number of induced abortions in Australia. An estimate of the number of induced abortions in Australia for 2003 was determined using:
For Queensland the number of induced abortions would likely be accurate.
For South Australia, the number of induced abortions would likely be accurate. As noted above the number of induced abortions in the NHMD was slightly higher than the number of notifications of induced abortion to the South Australian Abortion Statistics Collection (SAASC) in 2002 (0.2%) and slightly lower in 2003 (0.3%).
For the Northern Territory, the number of induced abortions may be under-estimated because the coverage of private free-standing day hospitals in the Northern Territory is incomplete in the NHMD.
Under state and territory legislation, induced abortion is undertaken in both hospitals and non-hospital facilities in these jurisdictions.
For New South Wales, the number of induced abortions would likely be over-estimated because some facilities which provide abortion services may be regarded as hospitals in one data set, but as non-hospitals in the other. The number of induced abortions may also be under-estimated because there were fewer separations with induced abortion reported to the NHMD than there were in-hospital services for MBS-item 35643 Evacuation of the contents of the gravid uterus by curettage or suction curettage in the Medicare data.
For Victoria, the number of induced abortions may be under-estimated because the coverage of private hospitals in Victoria is incomplete in the NHMD.
For Tasmania, there is no information available that indicates that the method would not be accurate (i.e. there is no evidence of gaps or overlaps in coverage).
For the Australian Capital Territory, the number of induced abortions may be under-estimated because the coverage of private free-standing day hospitals in the Australian Capital Territory is incomplete in the NHMD.
This method was used because of possible differences in the definition of hospitals in the Medicare data and the NHMD, evidenced by the considerable discrepancy between private patient separations in the NHMD (2,702 separations) and the number of in-hospital services in the Medicare data (834 services). Adding the number of separations with induced abortion from the NHMD to the number of non-hospital services for MBS-item 35643 Evacuation of the contents of the gravid uterus by curettage or suction curettage from the Medicare data would result in possible marked double counting of services provided in these facilities.
The estimate was then adjusted to account for the estimated 13.1% of private patients who receive induced abortion services but who do not claim a Medicare benefit ( Nickson et al. 2004 ). This was applied to the non-hospital Medicare services for MBS-item 35643 Evacuation of the contents of the gravid uterus by curettage or suction curettage for New South Wales, Victoria, Tasmania and the Australian Capital Territory.
These data are presented in more detail in Chapter 3
Induced abortion was defined in the NHMD as separations with a diagnosis of O04.5–O04.9 Medical abortion, complete or unspecified and an abortion-related procedure. Separations with a diagnosis of O06.5–O06.9 Unspecified abortion, complete or unspecified from a private free-standing day hospital facility(ies) in Victoria were also included (see Chapter 2). These data are presented in more detail in Chapter 4.
These data are for services provided in 2003 for MBS-item 35643 Evacuation of the contents of the gravid uterus by curettage or suction curettage for which Medicare claims were presented and processed. These data are presented in more detail in Chapter 5.
Identification of induced abortions at or after 20 weeks gestation is limited in the NHMD and the Medicare data. Other national routinely collected data sets and state-based data collections were therefore assessed as sources of data on these induced abortions.
Induced abortion at or after 20 weeks gestation can be identified in the NHMD as separations with a diagnosis of O04.5–O04.9 Medical abortion, complete or unspecified and an induced abortion-related procedure, which also have duration of pregnancy recorded as 20 completed weeks (ICD-10-AM diagnosis codes O09.3–O09.5). The number of induced abortions at or after 20 weeks gestation is likely under-estimated in the NHMD because some induced abortions for known or suspected fetal abnormality or damage at or after 20 weeks gestation may not be identifiable in the NHMD. This is because it is not clear from the coding instructions (ACS 1511 Termination of pregnancy) that a diagnosis of O04 Medical abortion is required in these cases.
MBS-item 16525 Management of second trimester labour may be applicable for some induced abortion services at or after 20 weeks gestation. However, this MBS-item is not specific for induced abortion. Also, this MBS-item would be used for induced abortions carried out in the 14th–19th weeks, but not for those in the third trimester of pregnancy.
Induced abortions occurring at 20 weeks gestation or more are in scope for the National Perinatal Data Collection. However, induced abortions cannot be identified separately from stillbirths and live births.
Induced abortions occurring at or after 20 weeks gestation are included in the ABS Perinatal Mortality Data. However induced abortions cannot generally be separately identified in this data set.
Induced abortions at or after 20 weeks gestation are included in various state-based data collections, including from perinatal mortality committees, abortion notification collections in Western Australia and South Australia, and congenital anomalies data collections. The extent to which induced abortions can be identified varies among the collections and among the states and territories.
Data development work that could enhance routine reporting of induced abortion in national data sets is presented in Chapter 8. Involvement of government, service providers, relevant medical colleges and professional bodies, and information experts would be important for the development of any enhanced data collection arrangements. The data development work includes: