Data quality statement
National Non‑admitted Patient Emergency Department Care Database
The AIHW has collected and reported on the data in this report under the auspices of the Australian Health Ministers’ Advisory Council, through the National Health Information Agreement.
The data supplied by state and territory health authorities were used by the AIHW to assemble the National Non‑admitted Patient Emergency Department Care Database (NNAPEDCD). The data cover waiting times and other characteristics of presentations to public hospital ED.
The NNAPEDCD provides information on the care provided (including waiting times for care) for non‑admitted patients registered for care in public hospital EDs that have:
- purposely designed and equipped area with designated assessment, treatment, and resuscitation areas
- the ability to provide resuscitation, stabilisation, and initial management of all emergencies
- availability of medical staff in the hospital 24 hours a day
- designated ED nursing staff 24 hours per day 7 days per week, and a designated ED nursing unit manager.
Patients who were dead on arrival are in scope if an ED clinician certified the death of the patient. Patients who leave the ED after being registered/triaged to receive care and then advised of alternative treatment options are also in scope.
The scope includes only physical presentations to ED. Advice provided by telephone or video conferencing is not in scope, although it is recognised that advice received by telehealth may form part of the care provided to patients physically receiving care in the ED. Also excluded from the scope of the NMDS is care provided to patients in general practitioner co‑located units or urgent care centres.
Since 2003–04, data for the NNAPEDCD have been reported annually. The most recent reference period for this data set includes records for Non‑admitted patient ED service episodes between 1 July 2017 and 30 June 2018.
Since 2015–16, jurisdictions were able to provide data for the NNAPEDCD using the NAPEDC NMDS or the NAPEDC NBEDS/DSS. Episodes are included in the NAPEDC NMDS, but excluded for the NAPEDC NBEDS/DSS, where:
- only a clerical service is provided to people supporting a pre‑arranged admission
- people are awaiting transit to another facility, and receive no clinical care.
Summary of key data quality issues
Overall, the quality of the data in the NNAPEDCD is sufficient to be published in this report. However, the following limitations of the data should be taken into consideration when data are interpreted.
States and territories are primarily responsible for the quality of the data they provide. However, the AIHW undertakes extensive validations on receipt of data. Potential errors are queried with jurisdictions, and corrections and resubmissions may be made in response to these edit queries. The AIHW does not adjust data to account for possible data errors or missing values, except where stated.
Comparisons between states and territories and reporting years should be made with reference to the accompanying notes in the chapters and in this appendix. The AIHW takes active steps to improve the consistency of these data over time.
- The NNAPEDCD may not include emergency presentations to hospitals that have EDs that are not in scope for the NAPEDC NMDS/NBEDS.
- The following jurisdictions have provided data to the NNAPEDCD using the NAPEDC NBEDS specification: Queensland (from 2015–16); Victoria and Western Australia (from 2016–17). All other states and territories used the NAPEDC NMDS. The data provided using the NAPEDC NBEDS may not be entirely comparable with data provided using the NAPEDC NMDS
- For 2017–18, waiting times information could not be calculated for about 20,000 emergency presentations (for which waiting times are applicable).
- Changes in definitions for Episode end status in the NMDS and NBEDS between
2015–16 and 2016–17 may affect the comparability of the 2017–18 Episode end status data with that for reporting periods prior to 2016–17.
- For 2015–16, Australian Capital Territory ED care data were not available at the time of publication. Therefore, comparisons over time should be interpreted with caution. The 2015–16 data for Australian Capital Territory may be available for inclusion in future reports.
- Although there are national standards for data on non‑admitted patient ED services, the way those services are defined and counted varies across states and territories, and over time.
- The quality of the data reported for Indigenous status has not been formally assessed; therefore, caution should be used when interpreting these data.
- For Western Australia, the date and time of commencement of clinical care was missing for about 23,000 ED presentations for a Public acute group B hospital in 2016–17, and for about 43,000 in 2015–16. As a result, the 2015–16 and 2016–17 waiting times data for Western Australia (and particularly for Public acute group B hospitals) should be interpreted with caution.
Missing or invalid data
In some cases, the data provided may include missing values (for example, the date/time of physical departure was not recorded), or invalid values (for example, if the time of physical departure was recorded as occurring before the time of presentation).
Because of missing or invalid values (such as time of presentation, or time of start of clinical care), valid waiting time could not be calculated for about 20,000 records with a type of visit of Emergency presentation—this excludes records with an episode end status of Did not wait to be attended to by a health care professional (298,000 records), Dead on arrival (251 records), or Registered, advised of another health care service, and left the emergency department without being attended by a health care professional (about 51,000 records). These records were not used in the derivation of waiting time statistics.
Further, because of missing or invalid values (such as time of start of clinical care, or time of episode end), duration of clinical care could not be calculated for about 55,000 records—this excludes records with an episode end status of Did not wait to be attended to by a health care professional, Dead on arrival, or Registered, advised of another health care service, and left without being attended by a health care professional.
The length of ED stay could not be calculated for about 5,000 records due to missing, or invalid values (such as for time of presentation, or time of physical departure).
How has the scope of the collection changed?
From 2013–14 onwards, the scope of the NAPEDC NMDS (and for the NAPEDC NBEDS/DSS in 2015–16 and 2016–17) has been patients registered for care in public hospital EDs as described in ‘National Non admitted Patient Emergency Department Care Database’.
For 2012–13 and earlier years, the scope of the NAPEDC NMDS was: public hospitals that were classified to peer groups A and B, for the purpose of reporting in Australian hospital statistics for the previous financial year period (using the AIHW’s previous peer group classification). As a result, any comparisons of time series data should take into consideration changes in the scope of the collection from 2013–14 onwards.
How has data coverage changed over time?
Because the scope of the NAPEDC NMDS is restricted to formal EDs, the number of ED presentations reported to the NNAPEDCD does not include all emergency or urgent care provided by public hospitals.
Between 2003–04 and 2013–14, the data coverage of the NNAPEDCD was estimated by comparing the number of ED presentations reported to the NNAPEDCD with the number of non‑admitted patient emergency occasions of service reported to the National Public Hospital Establishments Database (NPHED). The NPHED estimate was considered to be a more complete count of emergency care services, because it included emergency care data for all public hospitals, regardless of whether they had a formal ED, or other arrangements for providing emergency care. This provided an estimate but not an exact measure of the coverage.
For 2014–15, an approximate estimate of coverage was calculated based on emergency occasions of service that were reported to the NPHED in 2013–14. Using this approach, national coverage of the NNAPEDCD was estimated at about 88% in 2014–15. Estimated coverage by remoteness area of the hospital (using the same approach) varied among remoteness areas, ranging from 100% in Major Cities to 18% in Very remote areas (AIHW 2015).
However, emergency occasions of service were not reported to the NPHED from 2014–15 onwards, which meant it was no longer possible to calculate the proportion of all emergency occasions of service that were reported to the NNAPEDCD.
Estimates of coverage from 2015–16 onwards have not been calculated.
Variation in reporting
Variation in hospitals reporting
Between 2013–14 and 2017–18, the number of hospitals that reported ED presentations to the NNAPEDCD was relatively stable for most states and territories and included the major public hospitals in all states and territories.
A summary of the key changes in hospital reporting between 2013–14 and 2017–18 is provided below:
- In New South Wales, Byron Central Hospital commenced providing ED care in 2015–16, replacing care previously provided by Mullumbimby Hospital and Byron Bay Hospital.
- In Queensland, the Sunshine Coast University Hospital opened in March 2017, but this did not constitute a change in coverage, as the ED services were previously provided by a number of smaller hospitals in the region, which reported data for the NNAPEDCD.
- Data for the Royal Children’s Hospital and the Mater Children’s Hospital were included in reporting from 2012–13 to 2014–15. During 2014–15, they were replaced by the Lady Cilento Children’s Hospital. All 3 hospitals reported ED care data in 2014–15.
- In Western Australia:
- Perth’s Children’s Hospital opened in June 2018 and Princess Margaret Hospital closed. The data for these hospitals are reported separately.
- the St John of God Midland Public Hospital opened, and the Swan District Hospital closed in November 2015. Both hospitals were reported in 2015–16.
- in 2014–15, Busselton Health Campus began reporting ED care data, after the Busselton hospital was redeveloped to include a larger ED. This constituted a change in coverage as the activity was previously not reported for the NNAPEDCD.
- also in 2014–15, the Fremantle Hospital’s ED was replaced by the Fiona Stanley Hospital ED. Both hospitals were reported for 2014–15.
Change in coverage due to the opening or closing of hospitals should be taken into account when interpreting changes over time. There was no change in the coverage of the NNAPEDCD between 2016–17 and 2017–18 (assessed by comparing the hospitals included for the two years).
In 2017–18, 286 public hospital EDs reported ED presentations. These included most major public hospitals—classified as Principal referral and women’s and children’s hospitals, Public acute group A hospitals, and Public acute group B hospitals—as well as some smaller hospitals located in regional and remote areas.
Possible variation in triage categorisation
Of the 8.0 million presentations reported to the NNAPEDCD for 2017–18, 98.2% were Emergency presentations, and 1.5% were Return visit, planned. The remaining types of visit accounted for about 0.3%. The proportion of presentations by triage category varied by state or territory.
New South Wales had the highest proportion of emergency presentations that were Non‑urgent (10.3%), and South Australia had the highest proportions of presentations that were Resuscitation (1.4%). Queensland had the highest proportions of presentations that were Emergency (15.8%) and Urgent (46.0%). This may reflect different triage categorisation, differing mixes of patients, or both.
Variation in the proportion of patients admitted to the hospital by triage category may indicate variation in the way EDs triage patients. Although triage category is not a measure of the need for admission to hospital, the proportion of presentations in each category that had an episode end status of Admitted to this hospital can be used to indicate the comparability of the triage categorisation.
The proportion of patients who were subsequently admitted does not include patients referred to another hospital for admission. For example, nationally, about 31% of Emergency presentations had an episode end status of Admitted to this hospital. Victoria, Queensland and the Northern Territory had the highest proportion (36%), and New South Wales had the lowest (26%). For Resuscitation patients, about 75% had an episode end status of Admitted to this hospital nationally, with the proportion ranging from 69% in New South Wales to 87% in the Northern Territory.
Variation in reporting diagnosis information
For the 2017–18 NAPEDC NMDS/NBEDS, diagnosis information was not reported using a uniform classification. The classifications that were reported were:
- Systematized Nomenclature of Medicine—Clinical Terms—Australian version, Emergency Department Reference Set (SNOMED CT‑AU EDRS)
- International Classification of Diseases, 9th Revision, Clinical Modification (ICD‑9‑CM), 2nd edition
- International Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD‑10‑AM) 6th edition, 7th edition, 8th edition, 9th edition, or 10th edition.
Most states and territories reported patients’ diagnoses using a single type of classification. The majority of records (68%) were reported using various editions of ICD‑10‑AM.A principal diagnosis was not reported for about 283,000 records. In addition, about 8,000 records had an ICD-9-CM or a SNOMED CT‑AU diagnosis that did not map to a valid ICD‑10‑AM diagnosis using the methodology outlined Appendix B.
Quality of Indigenous status data
The quality of the data reported for Indigenous status in EDs has not been formally assessed. In addition, the scope of the NNAPEDCD may not include some emergency services provided in areas where the proportion of Indigenous people (compared with other Australians) is higher than average. Therefore, the information on Indigenous status presented in this report should be used with caution.
All states and territories consider the Indigenous status data used in this report to be of a quality appropriate for publication. Indigenous status was not reported for fewer than 1% of ED presentations in 2017–18.
Other factors affecting interpretation of the NNAPEDCD data
This section presents other information about the quality of the data provided for the NNAPEDCD, and factors that may affect the interpretation of the information presented in this report.
Remoteness area of usual residence
In 2017–18, jurisdictions provided area of usual residence using either the 2016 Statistical Area Level 2 or Statistical Area Level 1.
The AIHW mapped the Statistical Area Level 2 area of usual residence information for each presentation to remoteness area categories based on the ABS Australian Statistical Geography Standard Remoteness Structure for 2016. This mapping was done on a probabilistic basis. About 1.5% of records could not be mapped to a remoteness of area of usual residence.
The remoteness area information for 2017–18 are based on the ABS’s ASGS 2016 classification, whereas the remoteness area information reported for 2013–14 to 2016–17 were based on the ABS’s ASGS 2011 classification. Therefore, the remoteness (and socioeconomic status) data presented for 2017–18 are not comparable with similar information presented in earlier reports.
Type of visit
For 2017–18, Victoria, Queensland, and Western Australia provided data for the NNAPEDCD using the NAPEDC NBEDS specifications, for which Patient in transit is not a valid type of visit category. Under the NAPEDC NBEDS specification, patients in transit are included as Emergency presentation.
Episode end status
For the purposes of reporting, the NAPEDC NBEDS episode end status category Transferred for admitted patient care in this hospital (either short‑stay unit, hospital‑in‑the‑home, or non‑emergency department hospital ward) was mapped to the NAPEDC NMDS episode end status category Admitted to this hospital (either short‑stay unit, hospital‑in‑the‑home, or non‑emergency department hospital ward).
For the NAPEDC NMDS, patients who are admitted to the hospital, and subsequently die before leaving the ED are included in the NAPEDC NMDS Episode end status category of Admitted to this hospital (either short‑stay unit, hospital‑in‑the‑home, or non‑emergency department hospital ward).
Victoria, Queensland and Western Australia provided 2017–18 data for the NNAPEDCD using the NAPEDC NBEDS specifications, for which patients who died or otherwise left the ED are not included in the NAPEDC NBEDS category of Transferred for admitted patient care in this hospital (either short‑stay unit, hospital‑in‑the‑home, or non‑emergency department hospital ward). As a result, Victoria, Queensland, and Western Australia data may not be entirely comparable with data provided for other states and territories.
Caution should be used when interpreting changes over time for episode end status because:
- Between 2015–16 and 2016–17, a change in the episode end status data element resulted in a new category—Registered, advised of another health care service, and left the emergency department without being attended by a health care professional (METeOR identifier 616654). As a result, the 2016–17 and 2017–18 data presented for episode end status are not comparable with previous years.
- Between 2014–15 and 2015–16, a change in practice in the certification of death in Victoria resulted in a decrease in the number of presentations with an episode end status of Dead on arrival—from 1,313 in 2015–16 to 111 in 2016–17.
There is a difference between the number of presentations with a type of visit of Dead on arrival (3,402; Table 4.1) and the number of presentations with an episode end status of Dead on arrival (3,510; Table 4.10). All presentations with a type of visit of Dead on arrival had an episode end status of Dead on arrival. However, some presentations with an episode end status of Dead on arrival did not have a type of visit of Dead on arrival. The majority of these presentations were in New South Wales (2,846 with a type of visit of Dead on arrival and 2,938 with an episode end status of Dead on arrival).
Quality of waiting times and length of stay data
For 2017–18, about 20,000 records that should have been included in the calculation of waiting times statistics were excluded, as they did not have a valid commencement of clinical care time recorded.
The criteria used to determine the proportion of Resuscitation patients seen on time varies between jurisdictions, therefore, the proportions of Resuscitation patients seen on time presented in this report may differ from those reported by individual jurisdictions.
Emergency department length of stay
For about 5,000 records, the ED length of stay could not be calculated, as the date and time of physical departure were missing. These records were distributed across multiple hospitals, mainly from New South Wales. Of those, about 800 had an episode end status of Did not wait to be attended by a health care professional, Dead on arrival, or Registered, advised of another health care service, and left the emergency department without being attended by a health care professional
Emergency department duration of clinical care
For about 67,000 records, the duration of clinical care could not be calculated. For about 32,000 of those it was because the date and time of episode end were missing. For about 23,000 it was because the date and time of commencement of clinical care was missing. About 73% of these records had an episode end status of Did not wait to be attended to by a health care professional, indicating that the patient had not received care.
For 2017–18, a duration of clinical care was reported for about 155,000 records with an episode end status of Did not wait to be attended to by a health care professional, Dead on arrival, or Registered, advised of another health care service, and left the emergency department without being attended by a health care professional—for which a time of episode end is not applicable. These records were distributed across multiple hospitals, mainly from Queensland.
NHA Performance indicators
In 2016–17, there was a change in the definition of the indicator Waiting times for emergency department care—proportion seen on time to exclude records for which the episode end status was Registered, advised of another health care service, and left the emergency department without being attended by a health care professional. Therefore, the 2016–17 and 2017–18 data presented for proportion seen on time are not comparable with previous years.
For 2017–18, this resulted in about 49,200 records being excluded from the calculation of this indicator that, in previous years, may have been included. About 48,000 of these records were in New South Wales.
AIHW 2015. Emergency department care 2014–15: Australian hospital statistics. Health services series no. 72. Cat. no. HSE 182. Canberra: AIHW.