Table of contents

Introduction Communication DRG Classification Review Cost Weights Development Benchmarking and Reporting Casemix Analysis Clinical Documentation Manual Audit P4P Quality Incentive Pilot Programme Casemix Project Structural Change International Collaboration Appendices

2 3 6 8 10 12 14 17 21 23 26

4
1

Casemix Beyond Resource Allocation

Introduction
Casemix information has been developed since 2008 to support the direction of Hospital Authority (HA) on Pay for Performance (P4P). It is geared towards promoting productivity and quality improvement through a fair and transparent mechanism that reflects performance of hospitals. Casemix is a patient classification system under which patient care activities are allocated into a set of scientifically defined categories identified by a specific set of codes and descriptions. In many developed countries, Casemix has become a common tool for measuring the activities of a hospital or health system. Since its implementation in HA, awareness among clinical colleagues was raised regarding its importance for both accuracy and completeness of clinical documentation in quality care; as well as the cost implication at patient activity level. Our stakeholders also envisaged that Casemix information can be made more meaningful from a clinical perspective and that HA hospitals can learn and improve through sharing of best practices based on Casemix information. Casemix development is an evolving process that merits long term investment from a corporate perspective. The experience gained and lessons learnt from the past few years have laid down a solid foundation for clinical data validation, refinement of cost data at patient episode level, as well as the future use of Casemix information beyond its initial focus of resource allocation. The locally adapted Casemix system in HA, Diagnosis Related Groups (DRGs), aims to provide stakeholders with updated and comprehensive information to enhance their understanding of the efficiency of hospital services delivery and the facilitation of improvement in clinical care with a view to promote best practices in HA.

4
2

Casemix Beyond Resource Allocation

Communication
Communication and staff engagement is vital to the development of Casemix system. Under the governance structure of Casemix project, leaders and representatives from the Co-ordinating Committee/ Central Committee (COCs/ CCs) as well as cluster and hospital management have been actively involved in identifying and addressing areas of concern in the developmental and implementation process. Besides, multiple channels have been established to enhance communication with our stakeholders across the board.

Channels of communication to engage stakeholders

Cluster Review Meetings Cluster Review Meeting (CRM) is a half yearly forum for the exchange of experience and information between clusters and the corporate Casemix Office. Its focus is cluster Casemix data quality and their performance. This platform allows the Casemix Office to update hospitals with its key developments while stakeholders from clusters may raise any underlying issues. In 2011, an additional session named Casemix Roundtable was introduced along with CRMs. This further provided clusters with a unique opportunity to further develop understanding and use of Casemix information for stakeholders in specific areas. Clusters involvement in and feedback from CRMs has been constructive. A total of 101 participants attended the latest round of CRMs held in August and September 2011.

4
3

Casemix Beyond Resource Allocation

Number of CRM Participants CCEs HCEs Frontline (COS/CON/AC/SMO/MO/RS) HIRM Officers Finance and Administrative Staff Total Participants 6 16 52 16 11 101

The Casemix website The Casemix website (http://fin.home/casemix) is regularly updated with key information available to staff for review or comment. An email address (casemix@ha.org.hk) has been set up for staff to make enquiries and/or to request for further information when needed.

4
4

The Casemix newsletters In addition, the publication of Casemix newsletters started in July 2011. This served as a vehicle to disseminate information to clusters, connect with parties of common interest, and raise staff awareness of the application of Casemix in performance monitoring and benchmarking beyond resource allocation.

Casemix Beyond Resource Allocation

Casemix Reporting System (Data-Mart) With the wonderful help of the IT team, a new data-mart system called Casemix Reporting System (CRS) is underway. It is a common interface built for facilitating cluster users in data exploration and extraction for multiple purposes including modeling, analyses, cost weights development, classification review and reporting. CRS may store a wide range of data for a minimum of five years including clinical, pharmacy, services and Casemix DRGs for acute inpatient data.

Our commitment Casemix Office places great emphasis on communicating with corporate wide stakeholders to ensure Casemix information benefits HA hospitals/ clusters in performance monitoring, services planning and resource allocation.

4
Casemix Beyond Resource Allocation 5

DRG Classification Review

Background
Since the development of Casemix in HA, the International Refined Diagnosis Related Groups (IR-DRG) classification system was adopted for its relatively high compliance with the existing diseases classification system used in the Clinical Management System (CMS) of HA. The IR-DRG was an off-the-shelf product where grouping logics were pre-defined. This classification system was applied to all acute inpatient care episodes (except acute inpatient psychiatric service) in HA hospitals. In 2009, a preliminary overview of the classification system was conducted through consultation with the Casemix Clinical Committee (CCC) and Health Informatics team. As it was originally created in the United States, there was a need to increase the representativeness of IR-DRG for patient classification in HA. However, a more comprehensive review of the IR-DRG should be undertaken when the completeness of clinical documentation in the CMS was further enhanced, in particular the input of principal diagnosis into individual patient record. Thanks to the effort of frontline colleagues, the cases of missing principal diagnosis dropped from 23,935 to 897 from 2008/09 to 2010/11.

4
6

Continuous DRG Class ification Review

The vast improvement in clinical documentation evident in 2010/11 made it feasible to conduct a more comprehensive approach in reviewing the classification system. A plan to continuously review, refine and make improvements to the classification was endorsed by the Casemix Development Committee (CDC) with a view to appropriately reflect patients complexities and related treatment costs locally. This on-going annual reviewing process of the DRG classification system with alternating year focuses on Complications or Co-morbidities (CC) levels and DRG Families.
Casemix Beyond Resource Allocation

2nd Comprehensive Review Focus: Grouping of DRG Families

2011 1st Comprehensive Review Focus: Grouping of CC levels 2009 Preliminary Review Focus: Grouping of DRG Families

1st Comprehensive DRG Classification Review

This comprehensive review is being conducted in 2011/12. The Severity Level as per secondary diagnosis (SDx) is the first component of the DRG classification system identified for review. Severity Level reflects the level of complexity of diseases when presented as complications or co-morbidities. SDx carrying significant gaps between their pre-defined resource consumption levels in the off-the-shelf IR-DRG and their actual resource consumption in HA are to be selected for review. An approach of cost analysis coupled with clinical consensus is adopted in identifying these SDx. Advice and consensus will be sought from the CCC, CDC and Co-ordinating Committee representatives on the new HA Severity Level.

4
7

Casemix Beyond Resource Allocation

Cost Weights Development

Background
Cost weights are one of key components in the development of Casemix information. They reflect the RELATIVE resources used to treat patients based on the severity and complexity of their conditions. To better reflect HAs local situation, a set of local cost weights was built based on the acute inpatient (IP) activities and costing data of HA. The clinical data required were largely available through routinely captured information from the Clinical Management System (CMS) of HA while the costing data were extracted from the annual specialty costing results submitted by seven clusters.

4
8

The Journey of Development

The first set of HA cost weights (Version 1) was developed in 2008 based on the Specialty Costing per day and length of stay data of the respective DRGs with some appropriate adjustment upon consultations with clinicians. However, the drawback was the averaging effect of the higher surgical cost per patient with operations and those without in the same surgical ward. In 2009, the HA cost weights (Version 2) were refined to more accurately reflect the level of complexity and resource utilisation based on the Specialty Costing data. Actual utilisation data for operation, pharmacy, radiology and pathology at aggregated level were used to allocate cost to DRGs instead of by length of stay. The HA cost weights (Version 3) were further enhanced in 2010 by allocating cost to individual episode level, with additional refinements to reflect surgeons time cost in operating theatre. The enhancement more appropriately reflected cost for short stay day cases and reduced fluctuation in cost weights due to extreme outlier cases.
Casemix Beyond Resource Allocation

Riding on Version 3, the HA cost weights (Version 4) were refined in 2011 by allocating cost of medical devices to DRGs, mainly for Cardiology and Orthopaedic services to reflect the corresponding consumption of medical devices. While the current Version 4 is demonstratively superior to earlier versions, it still has a number of significant limitations. These limitations include the availability of data and the need to relate financial data to clinical care processes at granular level. HA is continuously reviewing and developing its costing methodology to progressively improve the appropriateness of cost allocation to patients.

4
Casemix Beyond Resource Allocation 9

Benchmarking and Reporting

Casemix Reports to stakeholders To provide stakeholders with updated and comprehensive information, Casemix Office distributes reports to all clusters on a monthly, quarterly and bi-annually basis. These regular analyses allow clusters to benchmark their performance in productivity, complexity, coding completeness and allocative as well as technical efficiency. Casemix information facilitates clusters in planning and monitoring their services as well as identifying areas for improvement. Furthermore, Casemix Office started to issue a bi-annual performance report to cluster managements in 2011. This additional report provides a high level summary of clusters performance. Graphs in the report describe how individual cluster tracks its Casemix performance indicators against the HA overall performance. These indicators include: Weighted Episodes (WE) Casemix Index (CMI) Complexity levels Length of Stay (LOS) and Average Length of Stay (ALOS)

4
10

Relative Stay Index (RSI) and related analyses (e.g. Bed days changes by DRG Families)

In addition, a watch list mechanism has been put in place for screening and monitoring sudden and/ or adverse changes in the average length of stay of DRGs and service activities. The development of Casemix benchmarking and reporting is an evolving process. Regular reviews of the performance indicators will be conducted to better reflect the performance at hospital, cluster and overall HA levels.
Casemix Beyond Resource Allocation

4
Casemix Beyond Resource Allocation 11

Casemix Analysis
Casemix is a useful tool for health care management by providing related resource consumption profile with specific treatment patterns. It is a way of defining a hospitals output or product by identifying clinically similar or homogeneous groups of patient that utilise similar bundles of treatment, tests and services with respect to resource use. Based on Diagnosis Related Groups (DRGs), Casemix information with demography, health status and health care professional opinion, may be used in strategic planning, priority setting for resource allocation and service development at corporate-wide, cluster or hospital levels. The potential of Casemix information in improving the existing hospital planning, budgeting and operation processes may be demonstrated by a case study.

4
12

Case Study
Based on 2010/11 DRG costing data, the study describes and compares volume activity, workload, resource utilisation and respective Casemix/ complexity in HA by four age groups which are categorised as under age 1, age 1-17, age 18-64 and age 65+.
Age <1
35%

Age1-17

Age18-64

Age 65+

40%

50%

47%

48% 8% 8%

47% 6% 6%

38% 5% 7%

42% 5% 6%

Headcount

No. of Episode

LOS[1]

WE[2]

[1] Length of Stay [2] Weighted Episode measurement of resource used adjusted by Casemix

Casemix Beyond Resource Allocation

Age <1 Cost per patient ($) Casemix Index (CMI) Episode per patient LOS per patient 18,000 0.906

Age 1-17 18,000 21,000 0.805 0.884 1.90 5.30

Age 18-64 22,000 0.892

Age 65+ 33,000

Age 65+ higher than Age <65 by 57%

1.174 33% 2.34 9.73 23% 84%

General observations
Higher acute inpatient service utilisation and resource used in adult and aged patients for 2010/11 Higher in cost (57%) for elderly than non-elderly patient is mainly due to sicker or more complex conditions in terms of Casemix index, greater number of admissions and longer length of stay

Common Ambulatory DRGs

Age <1 022360 AP Cataract Procedures 053130 AP Echocardiography 063130 AP Colonoscopy 113150 AP Dialysis 159170 AM Neonate,birthwt > 2499g X X X X X X X Age 1-17 Age 18-64 Age 65+ X

4
13

Common Inpatient (multi-day) DRGs

Age <1 04416 IM Simple pneumonia & whooping cough 044171 IM COPD 054122 IM Heart Failure w/CC 14613 IM Vaginal delivery 15817 IM Neonate, birthwt >2499g without major procedure X X Age 1-17 X Age 18-64 Age 65+ X X X

Casemix Beyond Resource Allocation

Clinical Documentation Manual Audit

Background
In 2009, a first baseline manual audit was conducted for 2008/09 data. The objective was to assess the appropriateness of reported diagnoses and/ or procedures of various specialties across seven clusters in HA. The results indicated that vast improvement was required especially in the area of secondary diagnoses reporting. The results also suggested that further education and training in clinical documentation guidelines was important from clinicians perspectives. In the first year of implementation of Pay for Performance, significant improvement in clinical documentation was witnessed. However, some variations in the extent of reported clinical conditions among the clusters were observed.

4
14

Clinical Documentation Manual Audit

During 2010/11, twenty-two grouping standards were developed through consultation with Clusters and COCs. These grouping standards aimed to align the variable practices of clinicians reporting of secondary diagnoses by refining the criteria used in the off-theshelf IR-DRG behind the screen, so as to minimize any unnecessary and inconsistent grouping of cases and procedures into the different Severity Levels among hospitals and clusters. These standards were applied only to the DRG grouping process and did not alter or affect the front-end medical records input by the clinicians. In order to validate these standards and to assess the current documentation practices, a second and major manual audit was then conducted.

Casemix Beyond Resource Allocation

Methodology
Approximately 10,000 patient records were audited between January and March 2011. A stratified and randomized sample was selected with approximately 30 records per standard against each major hospital. Each record was audited by staff from another cluster or HAHO using a predefined methodology.

Audit Results
The following observations were made when validating the grouping standards: Compliance rates were fairly similar and consistent among seven clusters Some standard criteria (e.g. acute renal failure) could not help to distinguish more severe patients

4
15

Some standard criteria (e.g. secondary thrombocytopenia) were invalid

The following observations were made in the audit against definitions of secondary diagnoses: Compliance rates were fairly similar and consistent among seven clusters Inappropriate reporting was not evident

Casemix Beyond Resource Allocation

Continuous Clinical Documentation Data Review Mechanism

Having a data review programme in place is essential to regularly and continuously assess the appropriateness of clinical documentation data for DRG grouping process. Inconsistency in reporting of diagnoses and procedures will lead to wrong grouping of DRGs and misinterpretation of complexity and resource implication of cases. The Computerised Edit Check was developed to co-exist with the manual audit to ensure data quality is being monitored. It is a set of indicators which identifies secondary diagnoses that do not meet the criteria of DRG Grouping Standards. International Classification of Diseases - 9 - Clinical Modification (ICD-9-CM) codes that are identified will then be removed for the grouping process.

4
16 Casemix Beyond Resource Allocation

P4P Quality Incentive Pilot Programme

Background
As quality improvement is one of the three components of performance in the Pay for Performance (P4P) model, a P4P Quality Incentive Pilot Programme was developed by introducing financial incentives to strengthen quality enhancement initiatives in 2010/11.
HAs Pay for Performance Model

Performance (P4P) = G + Q + STW

Growth in Targeted Areas of Need Quaity & Incentive Pool

PERFORMANCE
Service Improvement/ Technology/ Workforce

Objectives
The P4P Quality Incentive Pilot Programme aimed to: (a) Provide reward/ recognition for quality service

4
17

(b) Provide incentive to clusters to improve quality especially through system enhancement (c) Drive quality performance (d) Be outcome and patient-centred

(e) Reflect the total quality performance of a cluster by a comprehensive list of Quality Performance Indicators (QPIs)

Casemix Beyond Resource Allocation

Components of the Programme

A set of Quality Performance Indicators (QPIs) were selected and refined from the Key Performance Indicator (KPI) framework to measure quality under four strategically prioritised areas: (i) access; (ii) safety; (iii) appropriateness of care; and (iv) specific disease management/ integrated care. Through consultation with clinical groups and along the corporate directions, a stretched target was set for each individual QPI.
Strategic Priority Areas Quality Performance Indicators Waiting time SOPD - routine category 1. Medicine 2. Surgery 3. Psychiatry 4. Orthopaedics Cancer treatment waiting time 5. Breast cancer 6. Colorectal cancer 7. MRSA bacteremia for acute episodes Safety 8. Casemix-adjusted unplanned readmission rate 9. Fracture hip surgery (pre-op LOS) Performance Target New case booking for routine cases 75th percentile at 52 weeks

4
18

Access

90% of patients < 55 days from diagnosis to first definitive treatment < 0.1258 MRSA bacteremia in acute beds per 1,000 acute patient days HAs best performance 70% of fracture hip surgery with pre-op LOS 2 days 35% of DM patients treated in GOPD and SOPD with HbA1c of <7% 65% with BP < 140/90 mmHg

Specific disease 10. DM HbA1c control in management/ each cluster (combine integrated SOPC and GOPC) care 11. Hypertension - BP control for GOPC patients

Casemix Beyond Resource Allocation

A sum of HK$50 million was earmarked for the programme. Financial incentive was provided to clusters via two instalments during the year dovetailing with the half year and full year KPI performance reports. Quality was either measured against a stretched standard (a measure of achievement) or against individual clusters self performance in the previous year (a measure of improvement). Clusters were rewarded for performing close to the target or for significant improvement.

Results
Generally speaking, improvements were observed among majority of the indicators. More clusters demonstrated improvement instead of deterioration in performance in 9 out of 11 indicators. All clusters showed improvement in 3 indicators, namely MRSA bacteremia rate, diabetic control and blood pressure control for chronic diseases management.

4
19

Casemix Beyond Resource Allocation

4
Way Forward
The Quality Incentive Programme had been instrumental in fostering a culture of quality improvement amongst clusters. Financial incentive is important in promoting a change in behaviour, but it is not the only means to motivate change. An in-depth and qualitative evaluation is underway to evaluate the impact and effectiveness of the programme in achieving its objectives. The evaluation may assist in the refinement of the programme to strengthen its effect in improving the quality of patient care.
20 Casemix Beyond Resource Allocation

Casemix Project Structural Change

In 2008, the Project Steering Committee (PSC) chaired by the Chief Executive of HA, was set up to steer and provide direction on the development of a Casemix system in HA. Subsequently, various committees were established to solicit clinical and technical support to the development of the system and reported to the PSC.

4
Casemix Beyond Resource Allocation 21

After two years of implementation of the Casemix system, its development had evolved into a consolidation phase. Therefore a transformation of the Casemix governance structure was deemed appropriate. The new project structure was endorsed by the PSC and Directors meeting, with 28 September 2011 as the effective date. It will continue its role in future Casemix developments in supporting HAs performance monitoring and quality assurance, the Government funding review and the development of DRG Package Charging mechanism for Government healthcare financing reform.

4
22 Casemix Beyond Resource Allocation

International Collaboration
This year, members of the Casemix Office travelled overseas to present to researchers, analysts, Casemix experts, government officials and other stakeholders the development of Casemix in HA. Active sharing of information and exchange of experience with international experts definitely added value to the betterment of our locally adapted HA Casemix system.

International Establishments
Last year, HA was awarded the Best Casemix Innovation Prize at the 26th Patient Classification Systems International (PCSI) Conference. This year, amongst the abstracts submitted from all participating countries, all 5 abstracts from HA were accepted for oral presentations at the 27th PCSI Conference, held in Montral, Canada. Three of the abstracts, based on a blinded review by at least 3 members of the PCSI Scientific Committee, received the highest scores and were selected as inclusion in the supplement to be published in the BMC Health Services Research (http://www.biomedcentral.com/14726963/11?issue=S1&page=1).

4
23

Casemix Beyond Resource Allocation

4
24

Participation at International Conferences

1st International Health Management Congress by the Australasian College of Health Service Management (ACHSM) in Rotorua, New Zealand 28th International Society for Quality in Health Care (ISQua) Conference in Hong Kong

27th Patient Classification Systems International (PCSI) Conference in Montral, Canada 37th World Hospital Congress by the International Hospital Federation in Dubai, United Arab Emirates

2nd Johns Hopkins Asia Pacific Casemix Conference in Hong Kong

Casemix Beyond Resource Allocation

Presentations at International Conferences

Titles Clinical Documentation Manual Audit Evaluating quality of care in Hong Kong through identification of Potentially Preventable Readmissions within the current unplanned readmissions indicator framework Pay-for-Performance Quality Incentive Programme - One-Year Pilot Programme Perception towards Casemix System by Clinicians after the First Year of Implementation in Hong Kong: A Survey Real-time Monitoring of Patient Outcome - VLAD Authors K. Fan, M. Wu, L. Lau, S. Gillett, K. H. Lee, D. Yeung K.H. Lee, M. Berlinguet, L. Wang, K. Mitchell

Y.S. Choi, K.H. Lee, S. Gillett, G. Lam G. Lam, S.H. Lee, D. Yeung D. Yeung, K. Wong, S. Gillett, K.H. Lee

Abstracts are attached as appendices and are also available on the Casemix website (http://fin.home/casemix).

4
25

Casemix Beyond Resource Allocation

Appendix I
Clinical Documentation Manual Audit
K. Fan, M. Wu, L. Lau, S. Gillett, K.H. Lee, D. Yeung Introduction Preliminary audit studies in HA (Hospital Authority) hospitals have shown that diagnosis and procedure data reported in the electronic records were very accurate. However, appropriateness is as important as accuracy in clinical documentation. With the introduction of internal resource allocation based on Casemix Pay for Performance (P4P), the relevance of clinical documentation became apparent. During the first year under the new P4P, there was significant improvement in clinical documentation, but large variations were observed between clusters in the extent to which specific clinical conditions were reported. In addition, clinicians were puzzled by the perverse incentives to report diagnoses and procedures entirely for financial reasons. As a consequence, HA introduced the concept of grouping standards, that is, a series of agreed upon rules that would describe when specific International Classification of Diseases codes carry significant resource implications. During 2010-11, 22 grouping standards were developed through consultation with clusters and representatives of clinical specialties. In order to validate these standards, and to assess the accuracy and appropriateness of current documentation practices, a second and major manual audit was conducted. Methods This manual audit of approximately 10,000 patient records was undertaken in January and March 2011. A stratified, randomized sample of records was extracted, with approximately 30 records applied against each of the major hospitals. Each hospitals records were audited using a predefined methodology by staff from other clusters or Hospital Authority Head Office. Results This paper describes the manual audit and examines the implications of its results for appropriate clinical documentation. Conclusions Auditing is an important tool in ascertaining the accuracy and appropriateness of clinical documentation practices, as well as in validating existing grouping standards.

4
26

Casemix Beyond Resource Allocation

Appendix II
Evaluating quality of care in Hong Kong through identification of Potentially Preventable Readmissions within the current unplanned readmission indicator framework
K.H. Lee, M. Berlinguet, L. Wang, K. Mitchell Introduction Potentially preventable readmissions (defined as readmissions that could likely have been prevented by better care during the initial admission, improved discharge planning, or improved post-discharge care) can provide important hospital quality information and is compared overall and at hospital cluster level with the currently used indicator in Hong Kong: the unplanned readmission indicator. Methods We used all 2009/10 acute care admissions, 590,247 stays from all 25 major acute hospitals of Hong Kong, collapsed in 7 hospital clusters. We used the Potentially Preventable Readmissions (PPR) from 3M Inc. as the method to designate hospital readmissions within 30 days of an initial admission (IA) as preventable or not. The PPR method necessitates the All-Patient Refined Diagnosis Related Groups (APR DRGs), a grouping classification from the same developers of the International Refined DRGs (IRDRG), currently used to group and measure case-mix in Hong Kong. Hence, we examined all possible combinations of categories of IAs and readmissions. The readmission is defined as potentially preventable if there is a plausible clinical relationship between the patients APR DRG during the IA and the APR DRG during the readmission. (HCF Review 2008; 30;75-91). The US Florida state publishes state-wide adjusted PPR rates for public hospitals and we use its 2008 norm to compare with HAs PPR rates.(www.floridahealthfinder. gov/compareCare/SelectProcedureCondition.aspx). Some limitation in the comparison with the Florida norm was warranted given the relatively high number of ungroupable stays in the APRDRG classification , 43,657 stays or 7.4% of all cases. This may be due to post-coding mapping using valid International Classification of Diseases (ICD-9) codes from different years, this was being acceptable by the IRDRG grouper but not accepted by the APRDRG grouper. The Hong Kong Hospital Authority (HA) used unplanned re-admissions for acute care as an important performance indicator for several years. However in 2010, HA introducedv a new definition of unplanned readmissions to more closely link performance to specific hospitals. Under this new definition an unplanned readmission is documented when there is an admission via Accident & Emergency (A&E) department to the same specialty in any HA hospital within 28 days of discharge from the index episode. Results Using the current HA indicator in 2009/10, the overall crude unplanned readmission rate was 10.4% with a range between 8.2% and 11.4% by hospital cluster. The observed PPR rate was 8.74% (34 352 PPR/ 392 958 at risk stays (once logical clinical exclusions are applied)) and between 6.9% and 10.4% by hospital cluster. The indirect standardized rate (using Florida norm) was 1.52 (cluster range: 1.24 to 1.73). In ranking, the best performing cluster in PPR came third in URR; the least performing were the same; four clusters changed 2 ranks or more. Conclusions We observed a significant reduction in percentage of the clinically defined potentially preventable readmissions (PPR) in relation to the broader, but less specific, unplanned readmission rate (URR). Quality indicator measuring readmission rate is important. The current measurement of unplanned readmissions in Hong Kong may benefit from refinements to identify clinically related potentially preventable readmissions.

4
27

Casemix Beyond Resource Allocation

Appendix III
Pay-for-Performance Quality Incentive Program One-Year Pilot Program
Y.S. Choi, K.H. Lee, S. Gillett, G. Lam Introduction The Hospital Authority (HA) is a statutory body that was established under the Hospital Authority Ordinance. It has managed public hospitals in Hong Kong since 1991. Hospitals in Hong Kong are divided into seven geographically based clusters. HA has designed a Pay-for-Performance (P4P) model which includes incentives to promote productivity and quality. In the second year of this models implementation, financial incentives have been introduced to strengthen its focus on quality indicators. Methods A set of 11 Quality Performance Indicators (QPI) was selected and developed from a framework of existing Key Performance Indicators (KPI) that were agreed upon by the HA Board of Hospitals and their senior executives. There were two systems of performance measurement: 1. Cluster hospitals whose achievement was close to target. 2. Cluster hospitals that showed improvement over their prior years performance level.

4
28

Performance targets to be achieved by clusters were set for each QPI. With dual measurement, an innovative method for measuring and rewarding quality performance was developed. Results There were improvements in all except for two indicators in the program, and all clusters showed improvement in three indicators. The HA overall result achieved preset targets in five indicators. The reward received by individual clusters from this program ranged from 63% to 88% of their total maximum potential quality reward. Conclusions This paper gives an overview of HAs P4P Quality Incentive Program. The results after a oneyear pilot were mixed; however, there was more improvement than deterioration in performance measurement among all QPIs. The program has been successful in fostering a culture among clusters to continuously strive for quality, and HA will continue to assess the impact of the program. The program will then be refined and broadened as more data and feedback are gathered.

Casemix Beyond Resource Allocation

Appendix IV
Perceptions of the Casemix System by Clinicians after the First Year of Implementation in Hong Kong: a Survey
G. Lam, S.H. Lee, D. Yeung Introduction The Hong Kong Hospital Authority (HA) introduced a Pay-for-Performance (P4P) resource- allocation policy using a Casemix system in late 2008. Clinicians played a vital role in its implementation, especially with regard to the accuracy of clinical data. The purpose of this study was to: 1. Assess the short-term impact of Casemix-based funding as perceived by clinicians on clinical practice and quality of patient care after one year of implementation. 2. Examine any association between the characteristics of the clinicians (rank and specialty) and their perceived impact of the Casemix system on clinical service. 3. Identify the barriers encountered by clinicians on the effective implementation of this new policy. Methods A pilot quantitative study was done in March 2010 on a large, public general hospital in Hong Kong. All clinicians working in the hospital were recruited. A self-administered questionnaire, developed using recommendations from available literature, was used. Three aspects were looked at: the background characteristics of the clinicians, their perceptions about the impact of the Casemix system, and the clinicians knowledge of the Casemix system. Five-point Likert scale, true-false, and open-ended questions were used. Analyses were performed to examine the relationship between the variables using Pearsons chi-square test or Fishers exact test, where appropriate. Results 1. 520 questionnaires were sent out and the response rate was 17.3%. 2. More than 2/3 of the respondents did not perceive any change in their clinical practice or the quality of care, efficiency of work, or fairness of resource allocation. 3. More than 2/3 of the respondents agreed that there was improvement in clinical documentation, but at the expense of their time. 4. 60% of the respondents did not agree that the system induce gaming. 5. Participants knowledge of the Casemix system was generally poor, particularly among junior clinicians who, in addition, had a lower participation rate in the Casemix promulgation session conducted by the HA Casemix Office. The junior clinicians also expressed anxiety about having to carry out clinical documentation without being given clear guidance. Conclusions After the first-year implementation of the P4P/Casemix policy in HA, clinicians did not perceive any negative impact on patient service, and they agreed that there was improvement in clinical documentation. The perceived lack of both knowledge and access to knowledge among the junior clinicians needs to be addressed. Lastly, a post-implementation survey was found to be useful in providing evidence to facilitate the formulation of communication strategies in the implementation of a corporate Casemix system.

4
29

Casemix Beyond Resource Allocation

Appendix V
Real-time Monitoring of Patient Outcome VLAD
D. Yeung, K. Wong, S. Gillett, K.H. Lee Introduction Comparisons of hospital-specific performance are typically presented using cross-sectional analyses; however, they can be inadequate in quality care monitoring. This may be due to: 1) averaging over time can hide important variations in the clinical experience of individual patient; 2) even if a problem was identified, the traditional approach might not give front line doctors any help in identifying the patients involved. To overcome these issues, we needed to monitor performance in terms of the sequences of good and bad outcomes over time, as well as to target specific patients for clinical review. This can be applied in real time if required. Traditionally, this has been done by using cumulative sum (CUSUM) charts. However, these charts are difficult to understand from a clinicians point of view. Recently, there is increasing interest in an alternative approach - Variable Life-Adjusted Displays (VLAD) with statistical control limits. These charts are much easier to interpret clinically and use a real-time monitoring manner. In this paper, we explored the concept of VLAD as applied to measuring the outcome (inhospital mortality) of in-patients with Acute Myocardial Infarction (AMI) in Hong Kong Hospital Authority hospitals.

4
30

Methods We used 4000 acute in-patients with AMI as principal diagnosis (500 consecutive AMI cases per 8 major hospitals), starting from April 2009, to investigate the VLAD approach. The VLAD chart with control limits showing cumulative differences between expected and observed mortality of patients. The expected mortality calculated from risk model was based on age, sex, risk of mortality (3M International Refined-DRGs (IRDRGs)), 5 DRGs* (common procedures undertake) for prediction of the in-hospital mortality of AMI. Control limits were set to allow one false positive in every 1,200 patients (or one every two years for major hospitals), based upon a Markov Chain Monte Carlo Simulation. Remarks * DRG04120: IP NON-COMPLEX RESPIRATORY SYSTEM PROCEDURES; DRG05115: IP CARDIAC CATHETERIZATION; DRG05106: IP OTHER CARDIOTHORACIC PROCEDURES; DRG04102: IP LONG TERM MECHANICAL VENTILATION WITHOUT TRACHEOSTOMY; DRG05140: IP PERCUTANEOUS CARDIOVASCULAR PROCEDURES Results For these 8 major hospitals, half of them did not flag a signal, i.e. VLAD within control limits. Three of them hit the lower control limit, i.e. these three hospitals had sufficiently cumulatively more deaths than expected; while one of them hit the upper control limit. Conclusions The VLAD charts can provide a useful visual tool for real-time monitoring of patient outcomes. By identifying the changes in pattern of outcomes over time we can flag specific groups of patients for clinical review. This could potentially provide earlier identification of factors impacting on the quality of patient care and leading to improved patient outcomes.

Casemix Beyond Resource Allocation

Appendix VI
Casemix Project Governance

Chairman: Ms Nancy TSE, HOF D(F)

Casemix Executive Group

Convener/ Co-Chairman: Dr L C CHONG, KEC CSC(Surg) / TKOH COS(SURG) Co-chairman: Dr S H LEE, QEH CON(PAED)

Clinical Committee

Development Committee
Chairman: Dr Hong FUNG, NTEC CCE/PWH HCE

Subcommittees Grouping Standards Working Group

- Clinical Documentation - Costing
Chairman: Mr Ken FAN, HOF CIM(Casemix Office)

Chairman: Dr Joseph LUI, KEC CCE / UCH HCE

Chairman: Dr C K CHAN, QMH Deputy HCE I / QMH CONS(MED) Co-chairman: Dr C K HO, TMH Deputy HCE / COS(OPH)

4
31

- Benchmarking/Reporting

- Other (Complexity, Outliers and Linked Episodes)

For details of memberships and terms of reference, please visit the Casemix website (http://fin.home/casemix).

Casemix Beyond Resource Allocation

Appendix VII
Acknowledgements
Project Steering Committee (PSC) Chairman Dr P Y LEUNG Secretary Dr Cissy CHOI Members Dr W L CHEUNG Dr L C CHONG Dr Hong FUNG Dr S F HUNG Dr S H LEE HOCS D(CS) Convener/Co-Chairman of CCC and KEC CSC(Surg) / TKOH COS(SURG) Chairman of CDC and NTEC CCE/PWH HCE KWC SD(Mental Health) & KCH HCE Co-Chairman of CCC and QEH CON(PAED) HOQ&S D(Q&S) HOS&P D(S&P) HOF D(F) HOF CM(FP) HOF Casemix Clinical Coordinator CE

4
32

Dr H W LIU Dr S V LO Ms Nancy TSE Dr Deacons YEUNG

Last but not least, special thanks to former members of various Casemix committees. Their invaluable contribution shaped and strengthened the development of Casemix. Dr Gordon AU, QMH Hon CONS(ONC) Dr Andrew HO, TMH Asso Cons(M&G) Dr H F HO, KCC CSC(Medical) / QEH COS(A&E) Dr S K LAM, KWH CON(O&G) Dr C C LAU, HKEC CCE / PYN HCE / WCH HCE Dr Emmy LAU, PYNMED AC Dr C S LI, QEH DCOS(MED) & CON(MED) Prof. Raymond LIANG, HKU Prof(MED) Dr K H MAK, KWH CON(ORT) Dr Fei Chau PANG, HOCS CM(MG) Dr T C PUN, QMH CONS(O&G)

Published by the Casemix Office, Finance Division, Hospital Authority Email: Casemix@ha.org.hk Copyright. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form without prior permission of the copyright owner.

Casemix Beyond Resource Allocation