Private Hospital Data Collection Review Final Report

3.2 Potential areas for streamlining or improving comparability

Private Hospital Data Collection Review - Final Report

Page last updated: 14 June 2012

      3.2.1   Cancer data

      There is no clear opportunity for streamlining the collection of national cancer data from private hospitals, nor for improving comparability in this area.

      3.2.2   Perinatal data

      There is no clear opportunity for streamlining the collection of perinatal cancer data from private hospitals, nor for improving comparability in this area.

      3.2.3   NHCDC

      There is no clear opportunity for streamlining the collection of NHCDC data from private hospitals.

      There may be an opportunity to overcome some comparability issues through preparation of a series of experimental estimates’ reports, in addition to the usual NHCDC reports with their standard caveats. For example, one report may be on the pharmacy cost for private hospitals versus the pharmacy cost for public hospitals. The experimental estimate reports would build on the findings from the Productivity Commission Research Report into Public and Private Hospitals.

      Benefits

      • Increased use of and understanding of cost data and differences between sectors.
      • Additional reports for the public and private sector (i.e. improved feedback loop).

      Enablers

      • Data is already collected and standardised.
      • Building on work undertaken by Productivity Commission.
      • Some stakeholders (state governments, media) are more interested in comparisons between sectors.

      Barriers

      • Additional work for the Department to create reports based on cost data.
      • Achieving agreement on estimating cost areas (e.g. user cost of capital).
      • Inappropriate use of experimental estimates.
      • Some stakeholders (private hospitals) are more interested in comparisons within rather than between sectors.

      Recommendation:
      Undertake increased analysis of NHCDC information and develop a series of “experimental estimate” reports.

      3.2.4   Safety and quality data

      There is significant uncertainty over the mechanisms that will operate for collation of safety and quality data at a national level. This Review finds that the expectations are that nationally mandated safety and quality indicators are expected to be able to be derived from currently collected data. However, the procedures and protocols for successfully doing this are yet to be proven, given the difficulties encountered in testing to date.

      3.2.5   PHEC

      The current processes and systems surrounding PHEC offer an opportunity for streamlining that would result in small savings in effort at the private hospital level and some reduction in effort at the State and Territory level. This opportunity would require the transfer of responsibility for PHEC from the ABS to the AIHW. Any benefits gained would require no additional investment or transfer effort for private hospitals themselves.

      Figure 6 illustrates the current process for collecting the PHEC data, with the ABS operating the collection and States and Territories providing summary activity data on behalf of around 95% of all private hospitals.

      Figure 7 illustrates the data flows under the proposed operating model for PHEC, with AIHW operating the collection. One key change in the flows is that summary activity data for all hospitals with data in the APC would be prepared in-house by the AIHW, with only those hospitals who do not submit data to the APC having to prepare and submit their own summary activity data. States and Territories would not need to summarise nor provide any data on behalf of private hospitals. Based on the private hospital coverage of APC (see Section 3.1.6), around 98% of private hospitals would have their data summarised by AIHW, with approximately 10 hospitals extracting and providing their own summary activity data.

      The other key change in data flows is of course that all private hospitals would provide summary financial, workforce and other operating data to the AIHW instead of to the ABS.

      Top of page

      Figure 6 Overview of current PHEC data flows.
      Figure 6 shows an overview of current PHEC data flows


      Figure 7 Overview of proposed new PHEC data flows.
      Figure 7 shows an overview of proposed new PHEC data flows.

      Top of page

      Benefits

      The potential benefits from moving to this proposed model are:

      • fewer private hospitals who need to extract and submit their own summary activity data;
      • Australian Hospital Statistics reports could report public hospital and private hospital data for the same financial year;
      • improved data for inclusion into Australian Hospital Statistics reports and their users;
      • reduced effort on behalf of States and Territories; and
      • greater flexibility in terms of access to de-identified PHEC data.

      Enablers

      There are several factors that would make it easier to move to the proposed model with AIHW assuming responsibility for the PHEC collection and reporting role:

      • The AIHW has the legislative protections and track record to give confidence to private hospitals and their representatives that their data will remain secure and unable to be used in a potentially identifying fashion without individual hospitals’ consent.
      • The ABS is likely to be amenable to a negotiated change in responsibility, subject to reasonable conditions. These conditions would include: confidence that any change would maintain the existing high participation rate in PHEC, by private hospitals, thereby maintaining the integrity of the PHEC time series; continuing access to the PHEC data that ABS requires for its non health related statistics, such as National Accounts reporting; and a transitional process that maximises the return on ABS investment in the review and re-engineering of the PHEC processes, forms and data items.
      • The AIHW has the experience, systems and other infrastructure to be able to successfully take on the PHEC responsibility and deliver it successfully.
      • The existing time frame for APC reporting would mean that PHEC reporting would be able to happen at least in line with current ABS time lines. By way of a comparison, each (financial) year’s Australian Hospital Statistics report is produced around 10 months after the relevant year’s end, while the corresponding Private Hospitals, Australia report is published around 11 months after the year’s end.

      Barriers

      There are significant barriers to this approach being practicable. These include:

      • The APC does not contain unique hospital identifiers for private hospitals in some jurisdictions (see Section 3.1.6). Such identifiers need to be attached to APC episode records to allow data for each individual hospital to be summarised correctly. Without such identifiers, this proposed model will not work. The jurisdictions that currently do not provide hospital identifiers are likely to resist the necessary change to make this approach work. This Review found that, even if private hospitals provided explicit consent for their APC data to be identified in this way, the jurisdictions would be unlikely to comply with this consent. This view existed even in those jurisdictions where admitted episode data are required as a condition of licence. In this case, the jurisdictions argue that they are responsible for the quality of the data and so will be seen as vicariously responsible for the quality of the summary results generated by AIHW but beyond the jurisdiction’s control.

        This Review considered ways to circumvent this barrier. Possible ways to achieve this might be to effectively replace the APC at the national level with an improved PHDB. Achieving this would require a long term strategy to modify the PHDB to collect the core APC items, increase resources devoted to its operation and invest those resources in improving the coverage and quality of the data. This would need to succeed to the point where the revamped PHDB was adequate to generate the necessary summary activity data for most private hospitals. This option, if sufficiently successful, would have the additional benefit of eliminating the need for States and Territories to submit APC data to the AIHW for private hospitals.

        Top of page This Review found that this option would most likely be more difficult to deliver than the preferred option of achieving collaboration with the States and Territories to provide hospital identifiers to the APC. Nevertheless, it remains as a long term alternative strategy.


      • The AIHW does not have the same statutory power to compel private hospitals to provide PHEC data as the ABS does through the Census and Statistics Act 1905. This creates a risk of lowering the PHEC response rates for those sections of the PHEC relating to other than summary activity.


      • AIHW timelines to release the Australian Hospital Statistics reports are generally getting shorter and private hospitals may have shorter amount of time to respond to PHEC.

      Recommendation:
      That the responsibility for the Private Health Establishments Collection be transferred to the Australian Institute of Health and Welfare.

      A transitional plan for delivering this recommendation is included in Chapter 4.


      Option 1   Update the format of HCP and PHDB data interchange files

      Currently each of the HCP and PHDB require data to be submitted using a collection specific, fixed field text file (see Section 3.1.6). This format makes the process of modifying file preparation software (by private hospitals) and file import software (insurers for HCP, the Department for PHDB) more difficult than it could be, when a change is required.

      There are more flexible data interchange formats available, which are widely used and would simplify the processes of making changes to individual fields (names, data type), adding new fields and removing existing fields. The simplest such format to implement would be a comma separated values (CSV) text file. This is a standard format111 that is widely implemented and which most data management products are able to import or export without the need for software modification. The CSV file would consist of a header record that contains the names of the fields for the data and subsequent records would each contain one episode’s data.

      111 Internet Assigned Numbers Authority

      Adopting this format for either HCP or PHDB would require further modification to record layouts beyond simply declaring a preferred CSV format. Both collections require a header record, which contains summary information and has a different format than the episode records that follow it in the file. The CSV file format does not allow for such an arrangement. However, it should be simple for file import and export procedures to be modified to output the header record’s fields as additional fields on each episode record rather than as a separate record at the beginning of the CSV file.

      The more flexible file format also makes it practicable for a hospital to develop a single data extraction module for fields common to both HCP and PHDB. In effect, the private hospital extracts 3 files – one containing records with fields unique to HCP, one containing records with fields unique to PHDB and one containing records with fields common to both. The files would need to have a common field that allows them to be linked (an episode identifier). The PHDB file then would be created by linking the 2 relevant files and the HCP file analogously prepared.

      This would mean that private hospitals could reduce the effort required to modify PHDB and HCP data extraction software when future changes to fields common to both are made. Choosing to adopt this approach should be a decision left to private hospitals to make.

      Top of page

      Benefits

      • The order of fields within the CSV file’s records is not important, provided all fields are present for each record, the fields are in the same order throughout the file and the first record contains the names of the fields. This means that a private hospital can choose the most convenient order when outputting fields to the CSV file.
      • Adding new fields to the collection becomes simpler. A new field can be added with less risk that the output file’s format will be corrupted through record layout errors.
      • When a new field is added to the collection, it will not “break” the import procedures for recipients of the data if that field is erroneously specified or omitted altogether. This means that other fields in the file can be tested for validity independently of errors with the new field.
      • When an existing field is removed from the collection, that variable simply needs to be omitted when exporting the file. No special effort needs to be made to adjust the variable ordering structure of the output file, other than excluding the field concerned.
      • These last two benefits make further evolution of the collections more practicable.

      Enablers

      • CSV is widely implemented and used as a data interchange format and most data management software incorporates standard import and export capability for CSV files. This means that private hospitals’ and insurers’ software suppliers do not need to write and maintain modules for writing or reading the CSV files.
      • The changes required to export and import procedures should be relatively easy to implement successfully, requiring minimal effort.

      Barrier

      • This requires every private hospital to change their data export procedure for each collection, every insurer to change their data import procedure (for HCP), every insurer to change their export procedure (for HCP) and the Department to change its import procedure for each collection.

      While there is an initial cost for all parties in moving to this modus operandi, the benefits of making the change will accrue over time. Every time a change is required to the fields included in a collection, there will be a saving in the effort required to modify, test and implement data extraction and import routines. As such, there is value in pursuing a change to a more flexible format such as CSV, regardless of other changes being adopted or not.

      There are also other format options available, that would provide similar flexibility, some of which allow greater functionality. For example, Extensible Markup Language (XML) is a standard file format specification for machine readable files112. It provides a level of functionality that allows files to be read by a wide range of applications, not just data management software. Adoption of an XML format would open up possibilities for browser based manipulation of data files, among other possibilities.

      Top of page

      112 Extensible Markup Language (XML) 1.0 (Fifth Edition). World Wide Web Consortium (W3C).

      However, there would be a significant cost to implementing and maintaining XML format specifications, in terms of developing the file specification, modifying software to produce the XML file and testing the software’s performance. It would require far more effort and time to implement such a solution than the benefits warrant. For this reason, the simpler, CSV option is preferred.

      Recommendation:
      That CSV file format with field names in the first record be adopted as the standard file format for PHDB and HCP files. This recommendation covers all HCP files, including HCP1, HCP2 and AN-SNAP.


      Option 2   Modify the APC file formats at State and Territory level to match PHDB file formats for common fields

      The detailed comparison between the PHDB and the APC showed that there are many commonalities between the two collections. The APC is the minimum set of data elements agreed for mandatory collection and reporting at a national level about care provided to admitted patients in Australian hospitals. Jurisdictional health authorities specify what private hospitals must submit monthly in terms of patient care and they then supply this information to the AIHW in the APC format annually (see Section 2.1.3).

      The PHDB has clinical, demographic, benefit and charge data for private hospital patient episodes nationally. In total there are 76 data items in each submission. The PHDB collects a large amount of patient and episode level benefit and charge information which is not within scope of the APC (see Section 3.1.6). Summary comparisons between the data specifications for PHDB and APC NMDS are shown in Table 10.

      Table 10 Comparison of APC NMDS and PHDB.

      Comparison APC NMDS vs. PHDB
      Identical 18 items. Activity when injured, Additional diagnosis, Admission date, Care type, Date of birth, Inter-hospital contracted patient, Mental health legal status, Number of days of hospital-in-the-home care, Number of qualified days for newborns, Place of occurrence of external cause of injury (ICD-10-AM), Principal diagnosis, Procedure, Separation date, Sex, Total leave days, Total psychiatric care days, Urgency of admission, Weight in grams (measured).
      Mappable 15 items. Admitted patient election status, Area of usual residence, Australian State/Territory identifier (establishment), Diagnosis related group, Establishment number, Establishment sector, External cause, Hospital insurance status, Intended length of hospital stay, Major diagnostic category, Mode of admission, Mode of separation, Person identifier, Region code.
      In APC not PHDB 4 items. Country of Birth, Indigenous Status, Source of referral to public psychiatric hospital, condition onset flag.
      In PHDB not APC 43 items. Predominantly information about particular types of care information e.g. Coronary care unit charges, coronary care unit days etc.

      Top of page

      At present, each State and Territory specifies its own range of variables and file format for the admitted episode data that it collects from private hospitals. No two jurisdictions adopt the same file format nor do the file formats co-incide with PHDB or HCP format specifications, in terms of record layouts.

      The consequences of this are that private hospitals currently have to maintain 3 separate data extraction procedures for admitted episode data across these 3 collections. Moreover, when a field common to all 3 collections is changed then the private hospital needs to change all 3 data extraction procedures separately. In the case of large private hospital operator operating across multiple jurisdictions but with central data extraction procedures, the effort is considerably larger113.

      113 Note that although we refer to 3 collections there are actually 8 separately operated jurisdictional admitted episode collections.

      As discussed above in relation to the CSV file format option for HCP and PHDB, this effort could be reduced to a single change requirement if HCP, PHDB and jurisdictional admitted episode collections shared a common file specification format, at least for that subset of fields common to all 3 collections.

      However, the information that jurisdiction’s legislation dictate must be collected from private hospitals is often greater than the APC NMDS. This may not be a barrier though as many of the additional “state items” like coronary care unit days are already in the PHDB.

      Nonetheless, there is value in exploring whether a change to allow common file format specification for selected data fields can be achieved. To do this would require one or more jurisdictions to agree to participate in a pilot process to test the approach. Similarly, one or more private hospitals would need to agree to participate in such a pilot test.

      Benefits

      • Potentially reduced effort required for private hospitals to prepare data extracts for HCP, PHDB and jurisdictional admitted episode collections.
      • Reduced effort for private hospitals to modify data extraction procedures when a change occurs to a field common to 2 or more of the 3 collections.
      • In jurisdictions where private hospitals experience difficulty meeting jurisdictional admitted episode data submission requirements, compliance with those requirements is likely to become easier and data quality to improve. This would have a consequential benefit for APC data quality at the national level, particularly as common fields include a number of APC NMDS agreed items.

      Enablers

      • If the above option of adopting CSV formats for PHDB and HCP is implemented, then it makes the process of delivering the common file specification substantially easier. It would simplify any changes to data extraction and import procedures for private hospitals and jurisdictions, respectively.
      Top of page

      Barriers

      • Agreement is needed for jurisdictions to modify their data collection formats and data import procedures for private hospitals.
      • It may be that, for this approach to deliver worthwhile benefits, jurisdictions might have to agree to lose detail for some fields where they use supersets of APC NMDS agreed code sets.
      • If the above CSV file format option is not successfully implemented, then an agreed HCP, PHDB and jurisdictional common data items’ format will be required, requiring much more effort to implement.
      • A significant issue here is with the AR-DRG and ICD versions. While PHDB and HCP have a field for AR-DRG version, jurisdictions will generally only accept the current version as specified in the APC NMDS. A similar issue can arise with ICD codes for diagnoses. Private hospitals are bound to provide insurers with diagnostic and procedural information to insurers according to the AR-DRG versions, ICD versions or MBS codes specific in their contracts. For some insurers, these differ from currently mandated versions in the APC NMDS. This creates a disconnection between relevant code sets in HCP, PHDB and APC for some fields that are nominally common to all 3 collections.
      • Protocols would be needed for delivering agreement between a larger range of stakeholders for future changes to items within PHDB, HCP and APC NMDS. Currently, APC NMDS and related data development is driven primarily by public sector needs. There is representation of private hospitals in this process but this has only marginal influence on outcomes. The above issue with AR-DRG and ICD versions is a symptom of this situation, resolution of which will require meaningful consideration of implications for insurers and private hospitals when deciding how and when to implement changes to AR-DRG and ICD versions.

      This will require finding a jurisdiction that would be willing to pilot such a process. A suitable jurisdiction may be Victoria as they have a well established data collection (Victorian Admitted Episodes Dataset) and expressed interest in such an option during this Review. NSW may also be interested in exploring this option, as they have just commenced a process of replacing their Inpatient Statistics Collection system for private health facilities.

      The pilot would also involve selected private hospitals rather than a wholesale change for all private hospitals. A suitable model might be to seek the involvement of a corporate private hospital provider willing to participate. A corporate provider would have greater incentive to be involved, consistent with the greater benefits they could realise, and better infrastructure available to support its involvement.

      Such a pilot study would need to operate under a governance and oversight arrangement involving the Department, the participating jurisdiction and the private hospital sector. NHISSC would also have an interest in the outcomes of the pilot.

      If the pilot is successful then it would create pressure to roll the approach out to other jurisdictions. This would arise as the benefits to private hospitals would be significant and the potential benefits to jurisdictional admitted episode collections and consequently to the APC also may be significant.

      Even though PHDB and state based collections are run for each financial year this pilot preparatory work could commence immediately.

      Recommendation:
      That jurisdictions and private hospitals be approached to undertake a pilot test of a process for effecting a common file format for those data fields common to PHDB, HCP and the APC NMDS.

      Recommendation:
      That, subject to the above pilot succeeding, the successful model for common specification of common fields be rolled out to all private hospitals and all jurisdictions.


      Top of page

      Option 3   Modify governance arrangements for data development and implementation affecting private hospitals and insurers

      There have been issues with disconnection between data development and implementation of APC NMDS items, and operating constraints for the private hospital sector (see Section 3.1.7). In addition, this Review is recommending changes that require greater alignment of data sets and data collection processes for at least HCP, PHDB and the APC.

      These recommendations will require active participation from private hospitals, health insurers, States and Territories, the Department, AIHW and ABS, to deliver successfully. Importantly, the level of involvement from these agencies will need to be maintained in the future, to minimise the risk that future changes to one or more of these collections re-introduce duplication, inefficiency or other problems.

      At present, the HCP Working Group (HCPWG) is responsible for oversight of the HCP and comprises representation from insurers and Commonwealth Departments (DVA and the Department). The PHDB is maintained by the Department, with input from the Private Hospitals Working Group (PHWG) which comprises representation from private hospitals, AIHW, ABS and the Department. The APC NMDS is overseen by the National Health Information Standards and Statistics Committee (NHISSC), which comprises representation from States and Territories, the Commonwealth (the Department, DVA, Medicare Australia114, ABS, AIHW), the National Health Chief Information Officers’ Forum (NHCIOF), the National e-Health Transition Authority (NeHTA) and APHA. The New Zealand Ministry of Health is accorded observer status115.

      114 From 1 July 2011, Medicare Australia will cease to exist and this role is expected to filled by a representative from the Commonwealth Department of Human Services.
      115 National Health Information Standards and Statistics Committee Business Rules Terms of Reference and Business Rules, February 2009. NHISSC.

      Structures and protocols to improve communication among these 3 groups should be put in place. In addition, NHISSC should consider establishing a formal procedure for assessing impact of proposed changes to APC NMDS data items and metadata on the private hospital sector. That protocol should actively engage the HCPWG and PHWG and should extend beyond implementation requirements, feasibility and timetables to also include business impacts.

      Such procedures could be established under the existing NHISSC business rules, which allow NHISSC to “appoint working groups where necessary”. Such working groups can include membership from outside of NHISSC, such as sector representatives and members of HCPWG and PHWG.

      NHISSC also allows for informal observers to attend its meetings and, as noted above, the New Zealand Ministry of Health has been accorded formal observer status. It would be appropriate for the health insurance industry to seek the same observer status with NHISSC, to provide it with a means to contribute to discussions as needed and maintain timely awareness of national health data developments.

      This Review finds also that regular meetings of HCPWG and PHWG, to consider joint data issues, should be scheduled. These meetings would provide opportunities to identify common concerns or opportunities with respect to the HCP, PHDB and APC nexus. They would also provide opportunities to discuss APC related issues and capture the full range of industry issues requiring consideration at NHISSC meetings.

      Benefits

      • Improved communication among key stakeholder groups in the management of national data development and implementation directly affecting the private hospital sector.
      • Reduced risks of problems arising through inadequate consideration of implications of proposed changes for the business of private hospitals and insurers.
      • Greater ownership of changes to HCP, PHDB and APC NMDS metadata and processes HCP, PHDB and APC NMDS metadata and processes among private sector stakeholders.
      Top of page

      Enablers

      • The structures necessary to effect this option are in place and the proposed changes can be adopted within existing business rules.
      • Representatives of private hospitals and insurers are keen for greater input into national data development and implementation activities. There is recognition that national performance monitoring processes will make use of the outputs from these activities and that it is important that the private sector’s view is adequately represented in decision making processes.

      Barriers

      • Engagement between insurers and private hospitals can be complicated by the fact that they often are in adversarial positions when negotiating contracts. This makes collaborative decision making more difficult.


      Recommendation:
      That the health insurance industry formally seek membership of NHISSC as an observer. This could be either as a permanent (observer) member or on an occasional basis, when issues specific to health insurers are to be considered.

      Recommendation:
      That PHWG and HCPWG meet at least once per year to discuss data related issues. This meeting should take place in the December quarter, to allow sufficient time for issues requiring implementation in the following financial year to be identified and considered.

      Recommendation:
      That NHISSC be asked to develop a formal protocol for assessing business and related impacts of proposed changes to APC NMDS data items and associated metadata on the private hospital and health insurance sector.


      Top of page

      Option 4   Central collection portal for PHDB and HCP data submission

      The similarity between the PHDB and HCP has led to suggestions that a central portal be established. Such a portal would receive a single file from a private hospital of all their separations for a given period. The private hospital would have access rights only to upload its own data.

      This file would then be stored in a database accessible to the Department, DVA and other insurers through the same portal. The insurers would have access privileges allowing them to download only those records with their insurer identifier on them and only the HCP specific data from those records.

      Insurers then would add their HCP specific information to the records.

      The Department would have access rights allowing it to download only PHDB specific and HCP specific fields.

      Figure 8 illustrates how such a portal would operate.

      There are some prerequisites for such a portal to operate. First, a suitable entity to operate the portal would need to be identified or created and the necessary organisational infrastructure put into place. This would include ensuring adequate security protocols to protect confidentiality and privacy of the identified patient information that the portal would hold.

      Second, PHDB and HCP file specifications would need to be replaced with a single file specification that would apply to all episode records. private hospitals’ data extraction and submission procedures would need to be modified to accommodate the new format. Suitable governance and monitoring arrangements would need to be established as part of this step.

      Third, the portal itself would need to be built and tested, before it could be implemented. Part of the functionality of the portal would need to be a data validation engine, which would automatically review a file of submitted data and generate an error report for the submitting hospital.

      Top of page Figure 8 Central collection portal for HCP and PHDB.
      Figure 8 shows the central collection portal for HCP and PHDB.

      Benefits

      • PHDB and HCP data would be generated and transmitted in a single file, for each private hospital. It would eliminate the need for separate files to be prepared for each insurer (HCP data) and for the Department (PHDB).
      • HCP data would be downloaded in a single file, for each insurer. It would eliminate the need to obtain separate files from each private hospital.
      • PHDB and HCP data would be able to be downloaded in a single file by the Department. It would eliminate the need to obtain separate files from each private hospital (PHDB) and from each insurer (HCP).

      Enabler

      • The Private Health Insurance Act 2007 provides the legislative framework to require private hospitals and insurers to submit their data via such a portal.

      Barriers

      • While the Private Health Insurance Act 2007 provides a legislative framework to require compliance, it does not provide a strong basis for enforcing compliance.
      • There is no obvious entity that could fill the role of portal manager. No one agency or organisation appears sufficiently acceptable to all stakeholders, in terms of managing the privacy and confidentiality risks.
      • There would be significant effort required for private hospitals and insurers to modify their data management procedures to comply with such a portal’s business processes.

      Having considered the benefits, barriers and enablers, this Review finds that this option is unlikely to be achievable at this point in time. However, as PHDB, HCP and APC processes continue to evolve in the future, an opportunity may arise to reconsider whether a solution along these lines – entire or partial – will have become feasible.


      Top of page

      Option 5   Using ECLIPSE for hospital to insurer HCP reporting

      Consideration should be given to using ECLIPSE for transmission of HCP data by private hospitals to insurers. Currently, there is an outdated HCP specification within ECLIPSE (see Section 3.1.6. However, this part of the ECLIPSE record is not used by health insurers nor is it populated by private hospitals.

      If this specification within the ECLIPSE record was updated to match the current HCP requirement, then private hospitals could populate these fields when preparing their ECLIPSE records and transmit their HCP data to insurers via the ECLIPSE record. This would integrate HCP data transmission into another business process, eliminating the need for a separate extraction and transmission process for HCP data.

      Adopting an approach such as this would require establishment of an initial arrangement between the Department and Medicare Australia116. There would also need to be an ongoing arrangement for maintaining the currency of the HCP specification as and when future changes will be made. Medicare Australia is amenable to such arrangements, on the basis that they are consistent with the policy directions it receives from the Australian Government and they are able to recover the costs of initial and subsequent changes.

      116 More correctly with the Commonwealth Department of Human Services, after 1 July 2011.

      At present, ECLIPSE is not used for all private hospital admitted episodes leading for which health insurers receive a claim (see Section 3.1.6). This means that transmission of HCP data ECLIPSE cannot be implemented for all private hospitals and all health insurers at once.

      Benefits

      • Significant reduction in effort for private hospitals and insurers through elimination of an entire business process.

      Enablers

      • ECLIPSE has substantial market penetration, being used for around 19% of all private hospital admitted episode claims with health insurers. That market penetration will continue to grow as insurers increasingly engage private hospitals to change to ECLIPSE as the preferred method for claims’ lodgement.
      • The outdated HCP specification within the current ECLIPSE record requires only small modifications to be brought up to date.
      • Partial implementation of this approach is possible for a private hospital and a health insurer, and even partial implementation of this solution offers benefits to private hospitals and insurers. For example, if a private hospital uses ECLIPSE to lodge claims with half of the insurers with which it deals, it can eliminate half of the HCP data preparation and transmission burden by switching HCP data transmission to ECLIPSE for those insurers. Moreover, it can progressively reduce the burden further as it implements ECLIPSE with other insurers. A similar observation applies from the perspective of the health insurer receiving HCP data via ECLIPSE from some, but not all, private hospitals.
      • For private hospitals who start using ECLIPSE in the future, implementing HCP transmission into ECLIPSE will be part of the overall ECLIPSE implementation and should require little additional effort or cost. Similarly for health insurers who move to receiving HCP data via ECLIPSE.
      Top of page

      Barriers

      • There is an implementation cost for a private hospital to move to HCP transmission via ECLIPSE, and for a health insurer to move to HCP receipt via ECLIPSE. This should effectively be a once only cost, as once the software module is developed to populate the HCP fields in ECLIPSE, it can be re-used each time a new insurer adopts ECLIPSE for claims’ lodgement. A similar observation applies to a health insurer receiving HCP data via ECLIPSE from some, but not all, hospitals.
      • Medicare Australia will seek payment for costs associated with updating the HCP specification within ECLIPSE. Similarly, it will seek payment for future modifications to the specification. The Department should seek an arrangement whereby it pays only for work done and not for any “licence fee” type arrangement. In this context, it must be noted that the availability of up to date HCP transmission capability adds value to the ECLIPSE product.
      • There is an unresolved legal question about the marketing model Medicare Australia uses for ECLIPSE (see Section 3.1.6). This question has not been tested in a court of law and so the risk remains that Medicare Australia may have to change its marketing model in the future. For example, ECLIPSE is currently marketed freely to users but could be forced to operate under a fee-based model. This would affect market penetration and potentially cause some users to cease using the product.
      • There is a possibility that competing products may enter the marketplace and displace ECLIPSE. In this case, distributors of the competing products may or may not choose to include HCP data transmission as a product feature.

      The above barriers suggest that it would be unwise to move to a model where ECLIPSE is the mandated medium for transmitting HCP data between private hospitals and insurers, in the short to medium term. However, there remains value in creating an environment in which private hospitals and insurers have a choice of using ECLIPSE for transmitting HCP data, particularly as the benefits of dong so can be large. For example, a corporate provider managing dozens of hospitals could achieve substantial savings.

      The landscape with respect to market penetration and the risks identified above should be reviewed in the medium term to assess whether further changes in this regard are warranted.

      Recommendation:
      That the ECLIPSE record specification be updated to permit transmission of HCP data according to the current HCP specification.

      Recommendation:
      That the ECLIPSE record specification thereafter be maintained to ensure its capability to transmit HCP data remains current.


      Option 6   Pilot project to link claims’ data and inpatient data for private episodes in public hospitals

      To obtain this missing information such as AR-DRG, diagnosis and procedure codes it may be possible to link the claim form and the data held by state health departments. Possible fields that could be used to link the two datasets include:
      Top of page

      • Patient Unit Record Number/Episode Number/Hospital Record Number;
      • Sex;
      • Date of Birth;
      • Hospital;
      • Admission Date; and
      • Separation Date.

      This project could take the form of a pilot. It would require obtaining involvement from one insurer and one jurisdictional health department. The first stage would involve the Department meeting with an insurer and a jurisdiction health department to articulate the project. The Department would specify the reason for undertaking the linking, the benefits of undertaking the linking and the reporting and data distribution processes. A prior requirement would be to confirm the willingness of the AIHW to participate in such a pilot as an independent analyst and that there are no legislative nor other restrictions to prevent its participation.

      The next step would involve the insurer extracting the agreed linking fields from their system. They would then supply this data to the jurisdiction health department. The jurisdiction health department would then attempt to link the insurer data to episodes in their system.

      The jurisdictional health department would then report back on the results to stakeholders. The report would identify:

      • the successful number of records linked and how many false matches were obtained;
      • how long this data linking exercise takes;
      • how long it would take if it were to involve linking data from multiple insurers; and
      • how often this linking process could be undertaken (that is, monthly, quarterly or annually).

      Information such as DRG and Principal Diagnosis could be supplied back to the insurer(s) by the jurisdictional health department in an agreed format. This would enable the insurer to load the data into their systems and undertake detailed comparisons (DRG level) between the public and private sector in terms of costs. It is likely this would be attractive to insurers but the validity of this view would be tested as part of the pilot.

      The preferred plan for convergence focuses on obtaining HCP information from public hospitals on privately insured patient stays. Apart from charge information state and territory health departments already collect the same information that is specified in HCP from public hospitals.

      Top of page

      Table 11 Comparison of APC NMDS and HCP

      Comparison APC NMDS vs. HCP
      Identical 18 items. Activity when injured, Additional diagnosis, Admission date, Care type, Date of birth, Inter-hospital contracted patient, Mental health legal status, Number of days of hospital-in-the-home care, Number of qualified days for newborns, Place of occurrence of external cause of injury (ICD-10- AM), Principal diagnosis, Procedure, Separation date, Sex, Total leave days, Total psychiatric care days, Urgency of admission.
      Mappable 15 items. Area of usual residence, Australian State/Territory. identifier (establishment), Condition onset flag, Diagnosis related group, Establishment number, Establishment sector, External cause, Hospital insurance status, Intended length of hospital stay, Major diagnostic category, Mode of admission, Mode of separation, Person identifier, Region code.
      In APC not HCP 4 items. Country of Birth, Indigenous Status, Funding source for hospital patient, Source of referral to public psychiatric hospital.
      In HCP not APC 43 items. Predominantly information about particular types of care information e.g. Coronary care unit charges, coronary care unit days etc.

      If the States and Territories make available the information about episodes of care where privately insured patients elect to be treated as private patients in a public hospital, this will greatly improve comparability and transparency across the hospital system. It will enable more detailed comparative information to be made available about cost of stays by private patients whether they be in public or private hospitals.

      The availability of such comparative information will contribute to creating a more competitive environment across both the public and private sectors, holding out the prospect, over time, of a more efficient hospital system as whole that will be better able to contain costs.

      In a climate of increasing health costs, mostly due to an ageing population, it is essential to view the Australian hospital system as a whole and not as a disparate set of public and private hospitals that until now, have largely operated in isolation. If hospital services can be subject to greater competition, where jurisdictions may be able to outsource some procedures to private hospitals, if safety and quality is not compromised, but the procedure is more cost effective, downward pressure on cost can be increased through greater competition.

      Experience with State and Territory health authorities indicates they are open to developing a more competitive environment, but limitations in the ability to compare costs across the private and public hospital sectors have restricted this environment from developing. Allowing wider access to diagnosis and procedure information about private hospital patient stays in public hospitals would support greater efficiency across the whole Australian hospital system.

      Top of page

      It is primarily the identical and mappable items for which insurers are interested in receiving information. More specifically the fields of:

      • principal (and additional) diagnosis;
      • procedure codes; and
      • DRG.

      If insurers have this information they can then compare stays between the public and private sectors. It also enables insurers to better help their members by potentially developing programs to assist their members in managing their conditions outside of the hospital setting.

      The best approach to obtaining this information is for the Department to establish a project with a particular state or territory health department to undertake a data linkage exercise. This was identified as the best approach in the workshop. Outlined below are the prerequisites required to delivering the plan and the barriers, enablers and timeframes for undertaking this.

      The prerequisites to delivering this option are identifying an insurer that would participate in this pilot process; identifying a jurisdiction to participate; and establishing governance and management structures for the project, across the Department, HCPWG and jurisdiction.

      The choice of health department is a decision for the department. We suggest initial contact be made with the Victorian Department of Health to discuss this project as they have the most established admitted episode data collection (Victorian Admitted Episode Dataset – VAED). They also have a dedicated data linkage unit.

      The choice of insurer is also a decision for the Department and we suggest contacting either Medibank Private or the Australian Health Services Alliance. In discussions with these stakeholders throughout Work Stream 1 both demonstrated a willingness to participate in a process to obtain information on privately insured stays in public hospitals. This was reinforced at the Work Stream 2 workshop.

      Top of page

      Benefits

      • Improved data for insurers.
      • Insurers can compare public and private hospital stays.
      • Collaboration between stakeholders.
      • Increased use of private hospital data.

      Enablers

      • Data is already collected by jurisdiction health departments – this makes both the pilot and ongoing linkage easier.
      • Some jurisdictional health departments have expressed a willingness to be involved in such a project – again making both the pilot and ongoing linkage easier.

      Barriers

      • Additional work for insurers through having to supply data.
      • Additional work for jurisdictional health departments.
      • Obtaining jurisdiction health department consent to participate in this project.
      • Rolling out the pilot to all insurers would take time.
      • Cost of data linkage – this is a barrier to be overcome for both the initial pilot and any ongoing linkage process.

      Recommendation Investigate the feasibility of conducting a data linking exercise between jurisdiction health department and insurers for improved HCP information.