EC TWINNING PROJECT Development of National Coding Standards within the Czech DRG System CZ2005/IB/SO/03 Introduction and progress of ARDRGs and CASEMIX in Ireland 1
Introduction The purpose of the presentation is to discuss the introduction in 2005 of ARDRGs in Ireland To Discuss the situation in Ireland prior to this. To examine the method by which the introduction was made To view the Issues in the implementation But first some background Background HIPE (Hospital In-Patient Enquiry) Collects Irish patient activity data on contract to the Irish Department of Health and Children and the Health Service Executive (HSE). Collects information from over 60 hospitals Over one million records every year Administers the National Perinatal Reporting System (NPRS) on behalf of the Department of Health and Children 2
Background (Contd.) Cases are coded in ICD-10-AM This classification was introduced in 2005 Previous to this ICD-9-CM was used Cases are coded by more than 200 specially trained coders in the 60 hospitals Cases are entered using the Windows HIPE (W-HIPE) data entry system Casemix in Ireland CASEMIX in Ireland HIPE Unit, ESRI Coding Policy Collection of Inpatient data Collection of Daycase data. Windows HIPE data entry and reporting software Casemix Unit, HSE Collection of Cost data Preparation of Cost weights CASEMIX Policy Creation of CASEMIX model 3
What is CASEMIX? CASEMIX links clinical and demographic data on HIPE to the cost of treating those patients Activity data Cost data Casemix In Ireland CASEMIX ing for inpatients began in Ireland 1994-1995 The Casemix Project involves 37 hospitals All large Acute Hospitals in Ireland Maternity /Paediatric Hospitals The Casemix model is processed annually by the Casemix Unit in the HSE. Elective daycases are processed separately Outpatients and Casualty patients are not part of the Casemix project. 4
History of CASEMIX in Ireland Year Change 1991 Work started in 15 hospitals 1994 21 hospitals in Casemix (inpatients) 1995 23 hospitals 1996 Inclusion of Daycases 1997 28 hospitals 1998 30 hospitals 1999 Work started in Maternities 2000 32 Hospitals 2002 Work started in Paediatrics 2003 3 Dublin Maternities included 2004 2 Paediatric hospitals included (37 hospitals) Outcome of Casemix Casemix Adjustment 2005 (1,350,000) (850,000) (350,000) 150,000 650,000 1,150,000 5
Grouping in Ireland In Ireland in the 1990s we followed the American system HCFA 9, 12 and 16 Normally update every four years. American system chosen as they were the best at the time. Some support from America. Single DRG vendor. Number of Groups Different grouping schemes have different numbers of DRGs G r o up e r N u m be r of Gr o up s H C F A D R G V e r 9 4 9 0 H C F A D R G V e r 1 2 4 9 4 H C F A D R G V e r 1 6 5 1 1 A R D R G V e r s io n 5.0 6 6 5 A R D R G V e r s io n 5.1 6 6 5 6
Why are we doing this? Why are we doing this? CASEMIX/DRGs is designed to solve the following problems Funding based on historical allocations is unsatisfactory. Inability to compare hospitals due to difference in CASEMIX workload. Deal with the fact that all patients are unique. Deal with the my patients are sicker argument. 7
Effect of Casemix/DRGs 600 500 20 18 16 P atients (thous ands ) 400 300 200 100 0 1980 1981 1982 Inpatients Day Cases Beds 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Source: Acute Hospital Bed Capacity: A National Review. Department of Health and Children, Dublin. The Stationery Office, Dublin 2001. 14 12 10 8 6 4 2 0 B eds (thous ands ) 2005 Data Collection Time Line 2006 Deadline for 2005 coded information Deadline for 2005 Cost information Final Deadline for 2005 coded information Casemix Model Prepared Allocations made 8
HIPE software A central part of the HIPE project is the software used to enter the information. The software is used To collect the data. To validate the data at point of entry. To store the data in a database. To export the data on a monthly basis. To prepare reports on the data. HIPE Software Suite Windows HIPE Data Entry System Windows HIPE Reporter Shared Information System Windows HIPE Batch Coder HIPE Audit Toolkit ICD-10-AM ebook 9
The HIPE database Demographic Date of Birth Sex Marital status Area of residence Clinical Principal diagnosis Secondary diagnoses (19) Principal procedure Secondary procedures (19) Administrative Chart number Admission and discharge dates Type and source of admission Discharge destination and status GMS status Public/private status Specialty Consultant DRGs Before the Change The situation in Ireland prior to the change HCFA 16 (511 groups). Maryland Cost Weights. Single Relative Value for each group. Equivalencing is used to deal with unusually short and long cases. Some special cases dealt with outside the model The last years data processed using this method is 2002. 10
Grouper Selection In 2003, the Department of Health and Children in Ireland commissioned a study on alternative grouping schemes. The main factors guiding the study were The grouper scheme must be adequately supported. The Grouper Scheme must be both recognised and used internationally. Why Change? There were a number of reasons which prompted the change HCFA system was outdated. Costing issues -> Cost weights Grouping system was too rigid. Some minor problems with the Grouper. Appropriateness of the HCFA system in Ireland. Questions on the future direction of HCFA. 11
Grouper Selection A series of groupers were evaluated. HCFA 16 CMS 20.0 ARDRG V4.0, V4.1, V4.2 & V5.0 IRDRG 1.2 APDRG 18.0 These grouper were chosen as they have documentation in English. The source data used was the Irish national HIPE data. 3 years of data was examined ICD-9-CM Grouper Selection An important issue which arose was that Grouper versions are nearly always tied to a specific coding classification ARDRG groupers use ICD10AM HCFA/CMS groupers use ICD9CM etc. The original data has to be Mapped so that it could be used. 12
Grouper Selection A number of statistical measures were used to compare the groupers Reduction in Variance (R Squared) Aim for a high value / Close to 1.0 Measure of how much of the activity is explained by the DRGs Coefficient of Variation Measure of Homogeneity within DRGs Grouper Selection The results Existing grouper (HCFA 16) was poor The CMS 20.0 did not do well The ARDRG and IRDRG performed well. ARDRG was chosen over IRDRG Better training and support More widely used internationally 13
Grouper Selection More details can be found: Measuring Hospital Case Mix: Evaluation of Alternative Approaches for the Irish Hospital System Chris Aisbett, Miriam Wiley, Brian McCarthy, Aisling Mulligan Working Paper No. 192 The ARDRGs were first used in Ireland in Winter 2004 All stakeholders were kept informed via the annual Casemix conference. There were challenges in both activity and cost data. 14
Support Extensive support from Australia and particularly from the state of Victoria. Note: Victoria is similar Ireland. (population, Hospitals etc). Consultants were brought over to Ireland to advise on ARDRGs Data was sent to Australia to be examined. Activity data The initial data used was 2003 data. Big Challenges: Irish discharge status codes were mapped to Australian separation codes. Data is coded in ICD9CM but needed to be in ICD10AM Solution was to engage a consultant to prepare a diagnosis/procedure map so that data coded using ICD9CM can be changed to ICD10AM 15
Activity data This Map was a logical map. It was designed to place patients into the best ARDRGs by changing the diagnosis/procedures. The map could only be for casemix grouping reasons. The map is not useful for coding comparisons. Activity data The map was also used for 2004 data (with some minor changes) Tables showing the HCFA and ARDRG comparisons were prepared for each hospital each year. Comments/Feedback were sought. The ICD10AM coding classification was introduced in 2005 and so, by then, the map is no longer needed. 16
Cost data The hospital cost collection was changed for ARDRGs Original Cost categories (called buckets) were based on the Maryland weights Different cost categories were used based on the Victorian model. Hospitals needed to submit their information differently. Cost weights were adopted from Victoria Some changes were necessary. Weights to be localised for Irish use. Casemix model The casemix model was revamped to deal with different classes of cases. Same day inpatients One day inpatients Normal inpatients (inlier) Non-normal inpatients (outliers) paid at daily rates. This means that there are now five relative values for different types of cases. These changes removed the equivalencing from the model 17
Software Windows HIPE Software was updated New features to use pre-assigned ARDRG groups for cases in 2003 and 2004. New reports for ARDRG/casemix modelling Open tender for provision of grouping software Implementation of grouper interface. Collection of Admission weight Collection of Dialysis cases. Tender for groupers The tender process for groupers in the Irish system took place in 2005 The tender process identified 18 requirements the software must meet. Examples include It must be a Windows program Easy to use Fast Correct version (ARDRG 5.1) Certified 18
Tender for groupers The software provider of the grouper must be able to Update the groups if necessary Provide guaranteed levels of support Implement the grouper in batch mode Tender for groupers Four companies applied for the tender Each submission was evaluated and scored on a number of categories. The winning submission supplied A grouper for unrestricted national use Technical support in the area of ARDRG and Casemix Flexible arrangements for updates. Good value 19
Issues which Arose In general, everyone was very happy with the introduction of ARDRGs There was a steep learning curve for Accountants/Hospital managers/doctors and Medical Record/Coding managers. Some issues arose from the use of Mapping for 2003 and 2004. There was potential that bad habits might be picked up from Australia. Some problems with longitudinal studies. ARDRG Updates The current plan is that the ARDRG system will be updated when the ICD10AM classification system is updated. This will be typically be every four years. Some minor changes have been made to the grouper in the last few years Additional diagnosis codes were added to the coding scheme for Avian Flu and SARS 20
Strengths /Benefits Australian Support The Australian health professionals have been very supportive of the work in Ireland. Support on all aspects of the process Knowledgeable about the complete model. They know how to change the model in Australian and know the consequences of the change. Clear academic process used to form and maintain the ARDRGs. Bigger population than Ireland -> extremely likely to encounter problems before. Strengths /Benefits Modern Grouping System The ARDRG system is comparable with other systems in the world. The system is updateable Grouper is updated on a two yearly cycle. Maintainance is via Clinical/Financial groups in Australia This will ensure that any new medical issues will be reflected in the grouper. 21
Strengths /Benefits Modern Grouping System Improved contract conditions with the grouper implementation mean We can get a better understanding of how the grouper is working. We can change the grouper if needed We will be better able to understand the impact of changes. And the future The new adoption will allow the Irish casemix system to move forward Regular updates. Irish Updates on a four yearly cycle Option to update more frequently Aim to upgrade to V6.? Or V7.? in 2009 22
The future A/E (casualty) programme Adoption or development of programs for capturing and incorporating A/E cases in to the casemix model Out Patient Similarly the adoption or development of programs for capturing and incorporating out-patient cases in to the casemix model. Cost weights program The preparation of Irish version of the cost weights which is completely based in Ireland cost and activity. 23