Cost Per Case-Mix Weighted Activity For Complex Continuing Care in Ontario

J O I N T P O L I C Y A N D P L A N N I N G C O M M I T T E E Cost Per Case-Mix Weighted Activity For Complex Continuing Care in Ontario Cost per Resource Utilization Groups (RUG-III) -weighted Patient Days (Using 1997/98 Year End MIS Trial Balance & CIHI OCCPS/MDS 2.0 Data) TECHNICAL PAPER FROM THE JPPC COMPLEX CONTINUING CARE FUNDING WORK GROUP Prepared for the JPPC by Gary F. Teare, Ph.D. Providence Centre, Scarborough JPPC Reference Document RD #8-12, September 1999 AN ONTARIO MINISTRY OF HEALTH AND ONTARIO HOSPITAL ASSOCIATION PARTNERSHIP

All inquiries and questions pertaining to the methodology described in this document should be sent by e-mail at RUGS@jppc.org or by facsimile to Howard Baker, Technical Planning Consultant, JPPC Secretariat, at (416) 934-0711. Any concerns pertaining to the MIS TRAIL BALANCE DATA used in the calculation should be directed to your Finance and Information Consultant at the Ministry of Health. Any concerns pertaining to the OCCPS/MDS 2.0 DATA used in the calculation should be directed to your Continuing Care Client Support contact at the Canadian Institute for Health Information.

September 1999 To: Hospital Chief Executive Officers/Presidents Ministry of Health Ontario Hospital Association The Ontario Joint Policy and Planning Committee (JPPC) Complex Continuing Care Funding Working Group (CCCFWG), a sub-committee of the JPPC Hospital Funding Committee, firmly believes that to more appropriately serve Ontario s residents, we as an industry must strive to accurately measure what it is we do for our patients. For the first time, the JPPC is distributing comparative reports that indicate provincial and facility-specific 1997/98 costs per RUG-III-weighted patient day to Ontario hospitals with Ministry of Health designated chronic beds. These comparative reports, included with this technical document and its companion summary document (JPPC Reference Document #8-11), enhance our understanding in this regard. The purpose behind releasing this information at this time is two-fold. First, it is hoped that facilities will use this information to identify data quality and reporting issues and to use the insights gained from this information to make any necessary final changes to their data from 1998/99 and beyond, for both cost and patient activity. Second, the CCCFWG would appreciate feedback from the field regarding the work completed thus far. This feedback will be used during Phase Two of its work to use 1998/99 data to develop a funding model for implementation for April 2000. The CCFWG is also planning technical briefing sessions for the fall 1999 to assist facilities in interpreting these comparative reports as well as understanding the methodology used in their construction. It is for these reasons that we strongly urge facilities to use this information for internal purposes only. This data is not meant to indicate relative efficiencies or suggest proposed future funding allocations. The CCCFWG s concerns regarding data quality remain a prominent issue to be addressed through its future work plans. One way to do so is to ensure that the data upon which evidenced-based decisions are made is reliable and accurate. The JPPC anticipates that fiscal year 2000-01 will represent a re-vitalization of the interest in and focus upon data quality and measurement. The JPPC committees are orienting their work plans along this theme and intend to strengthen their role as leaders in this regard. We look forward to the industry s continued support and commitment of our efforts. Sincerely, Greg Fougère Greg Fougère, Chair, JPPC Complex Continuing Care Funding Working Group

Table of Contents 1.0 PATIENT ASSESSMENT, CASE-MIX CLASSIFICAT ION, AND RELATIVE RESOURCE WEIGHTING... 1 1.1 PATIENT ASSESSMENT INSTRUMENT: THE RESIDENT ASSESSMENT INSTRUMENT MINIMUM DATA SET (RAI-MDS)...1 1.1.1 Reliability and Validity of the MDS...1 1.2 THE RESOURCE UTILIZATION GROUPS (RUG-III) CASE-MIX CLASSIFICATION SYSTEM FOR CONTINUING CARE... 2 1.3 RELATIVE RESOURCE WEIGHTING: CASE-MIX INDICES... 2 1.4 RUG-III CASE-MIX INDICES FOR ONTARIO...3 2.0 CALCULATION OF PATIENT VOLUMES FOR ONTAR IO CHRONIC CARE BEDS...4 2.1 DATA QUALITY: ASSESSING THE ONTARIO MDS DATA...4 3.0 CALCULATION OF RUG-III CASE-MIX INDEX WEIGHTED PATIENT DAYS (RUG-WEIGHTED PATIENT DAYS)... 5 3.1 RUG-III CLASSIFICATION FOR EACH MDS ASSESSMENT ASSOCIATED WITH AN EPISODE OF CARE...6 3.1.1 RUG items missing from the Quarterly assessment...6 3.2 CALCULATION OF THE PATIENT DAYS ASSOCIATED WITH THE RUG-III CLASSIFICATION FOR EACH MDS ASSESSMENT WITHIN AN EPISODE....6 3.2.1 Episodes for which no MDS assessment data were available...7 3.2.2 An Example of How Patient Days were Calculated and Associated with RUG Groups... 8 3.3 WEIGHTING OF PATIENT DAYS... 10 3.3.1 Patient days associated with a RUG group... 10 3.4 CALCULATION OF FACILITY TOTAL RUG-WEIGHTED PATIENT DAYS... 10 3.4.1 Facility total RWPD... 10 3.4.2 Calculation of a facility-level case-mix summary score: the Facility Case-mix Index (FCMI)... 10 4.0 CALCULATION OF COST PER RWPD... 11 4.1 LINKAGE OF TOTAL RUG-WEIGHTED PATIENT DAYS TO FACILITY COST DATA FROM THE MANAGEMENT INFORMATION SYSTEMS DATA... 11 4.2 REFERENCE DATA AND CALCULATIONS USED TO CHECK THE VALIDITY OF THE PATIENT VOLUMES (DAYS) CALCULATED USING THE OCCPS/MDS DATA... 11 4.2.1 Calculation of Patient Days based on Admission and Discharge Data in the OCCPS... 12 4.2.2 Agreement Between the Number of Patient Days Calculated in the RUGWPD Process Versus Facility Total Patient Days from Other Sources... 12 REFERENCES CITED... 14 GLOSSARY OF TERMS... 16 DATA VALIDATION/RECONCILIATION GUIDELINE... 18 APPENDICES... 23

1.0 Patient Assessment, Case-mix Classification, and Relative Resource Weighting 1.1 Patient Assessment Instrument: The Resident Assessment Instrument Minimum Data Set (RAI-MDS) The Resident Assessment Instrument (RAI) was originally developed in the United States as a response to evidence of major quality problems within the nursing home sector (Zimmerman, 1997). The RAI was developed to standardize the patient assessment process and assist care planning by health care providers. The RAI consists of a Minimum Data Set (MDS) of elements with descriptions of common definitions, and Resident Assessment Protocols (RAPs) used to assist more detailed reviews of problems identified in 18 functional areas. Examples of domains addressed in the assessment include cognitive and communication patterns, physical functioning, health conditions, and special treatments and procedures. The MDS is not intended to replace or substitute the full clinical record. Instead, it is intended to provide a summary of the patient s functional abilities and physical, social, and psychological needs to form the basis for the development and periodic updating of the care plan. The first version of the RAI-MDS was introduced in 1990. A commitment to on-going research meant that improvements were made to the instrument culminating in the release of version 2.0 in 1995. The Ontario Ministry of Health mandated the use of the RAI-MDS 2.0 for all patients in designated chronic care beds, starting in 1996. The Health Care Financing Administration (HCFA) in the U.S. has indicated its intent to sponsor periodic reviews of the RAI-MDS to determine whether revisions are needed to reflect changes in clinical practice and advances in assessment technologies (Morris et al., 1997). 1.1.1 Reliability and Validity of the MDS Extensive testing of the reliability and validity of the MDS 2.0 and its predecessors was conducted in their development. Studies in the U.S. of inter-rater reliability of the MDS 1.0 and 2.0 have been reported (Hawes et al, 1995; Morris et al., 1997). The reliability of the MDS in international studies involving seven countries has also been reported (Sgadari et al., 1997). Inter-rater reliability refers to the degree to which the findings of an assessment are the same, when measured independently by two or more observers, given that the actual patient characteristics have not changed. These studies found strong inter-rater reliability of the MDS, particularly in the area of functional domains. While these studies demonstrated that the MDS is reliable in research studies, it is also important that the assessments conducted in day-to-day practice be reliable. Comparison of MDS data from research projects, using research nurses to collect the data, to MDS data collected as part of everyday practice showed nearly identical reliability estimates and similar levels of association between certain variables or subscales (Phillips and Morris, 1998). The validity of MDS data gathered in research and in routine use has also been tested and found to be good in several studies, including studies of criterion, concurrent and predictive validity of MDS subscales and items (Hartmaier et al, 1995; Gambassi et al, 1998; Mor et al 1995; Phillips and Morris, 1998; Blaum et al, 1997; Resnick et al, 1996; Abicht-Swensen and Debner, 1999). An important application of MDS data has been the development of a case-mix classification system known as Resource Utilization Groups (RUG-III). Effective July 1998, the U.S. 1997 Balanced Budget Act implemented the Prospective Payment System (PPS), which reimburses nursing homes under Medicare, based on MDS data Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 1

and the RUG-III algorithm. The Ontario Ministry of Health has announced that funding of chronic care beds in Ontario will be based upon the RUG-III classification system, starting April 1, 2000. 1.2 The Resource Utilization Groups (RUG-III) Case-mix Classification System for Continuing Care Development of the RUG case-mix classification system dates back to the early 1980's, with the work of Fries and Cooney (1985). It was further refined as RUG-II in New York State (Schneider DP, Fries BE, Foley WJ, et al., 1988). Since approximately 75 to 85 % of the cost of continuing care is due to staff costs, the RUG system was designed to classify patients into groups with similar staff resource use. Both the RUG-II and subsequently RUG-III were developed through an extensive clinical consultation process, statistical analysis of measured per-diem resource use, and with careful consideration of the positive and negative incentives that may be introduced when a funding system is based on a case-mix classification system. RUG-III was updated in 1997 in preparation for its use in the prospective payment system (PPS) for the U.S. Medicare nursing home benefit. This PPS was implemented, for all U.S. nursing homes, by the Health Care Financing Administration (HCFA) of the U.S. federal government, on July 1, 1998. RUG-III uses data from MDS assessments to classify continuing care patients into 44 different groups, within each of which patients have homogenous resource use, but among which average resource use is statistically different. The original RUG-III was developed using the measured per-diem resource use from over 7,000 patients in skilled nursing facilities in 6 States (Fries, BE, Schneider DP, Foley WJ, et al., 1994). The RUG-III "1997 update" was based on further resource use measurement in two staff time measurement (STM) studies of nearly 4,000 patients in 10 States. These latter STM studies intentionally included a disproportionate number of patients in heavier care, medically complex or high rehabilitation units of skilled nursing facilities in order to better estimate resource use in these rarer, but costly types of patients. Based on the patient's clinical conditions and certain services received, the RUG-III first classifies patients into one of seven major categories which are ordered hierarchically from most to least resource intensive: Special Rehabilitation, Extensive Services, Special Care, Clinically Complex, Impaired Cognition, Behaviour Problems and Reduced Physical Functions. Subsequently, these categories are further split into 44 groups through a series of splits based on the patient's activities of daily living (ADL) function, other clinical characteristics, or services received. 1.3 Relative Resource Weighting: Case-Mix Indices The case-mix index (CMI) for a group in the RUG classification system is the relative weight of resource use in that group compared to some base resource use level. The base used in the RUG-III system is generally chosen to be the average resource use in the patient population. Thus the case-mix index for a group represents the relative cost of caring for the average patient within that group, compared to the average patient in the population. Although the RUG-III classification system was developed based on measurement of resource use (staff time) for care of residents in U.S. skilled nursing facilities and nursing homes, international studies have demonstrated its applicability to continuing care populations across jurisdictions (Carpenter et al. 1997, Ikegami et al, 1994, Ljunggren et al, 1992). These studies found that, while actual minutes of care per day may vary across jurisdictions, the relative resource use among RUG groups was very similar, regardless of jurisdiction. Most recently, a 1998 study in Alberta, involving Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 2

approximately 2,000 institutional continuing care patients, found that the average daily minutes of nursing care for RUG groups closely matched those found in the U.S. studies (Beaulne et al., 1999). 1.4 RUG-III Case-mix Indices for Ontario By definition, a RUG-III case-mix index is the ratio of the average wage-weighted staff time consumed for care of patients in a particular RUG group to the average wageweighted staff time of all patients in the relevant population. The average amount of staff time (nursing and therapies) associated with care of patients in each of the 44 RUG-III groups in the 1997 RUG-III update is available from the U.S. Health Care Financing Administration s World Wide Web site (http://www.hcfa.gov/medicare/snfpps.htm). Case-mix indices for the U.S. are also posted on that web site, however they are inappropriate for use in Ontario, since the relative wages of nursing and therapy staff are different in Ontario than in the U.S. Ontario wage-weights were calculated as the ratios of licensed nursing, therapy and therapy assistant staff wages to nursing aide wages. The wage figures used for deriving these Ontario wage weights were taken from results of fiscal year 1997/98 Ontario Hospital Association salary surveys (Appendix A). The daily minutes measured for each staff type in the combined 1995 and 1997 HCFA staff time measurement studies were wage-weighted by multiplication by the corresponding staff-type Ontario wage weight. The wage-weighted times for all staff types were summed within RUG-III groups to obtain the total per-diem wage-weighted minutes of care (WWMPD) for each RUG group. It is desirable to define the case-mix value of 1.00 as the Provincial average wageweighted minutes of care per day (WWMPD pavg ), and thus to scale the case-mix indices relative to the average. To calculate the WWMPD pavg : 1. the total wage-weighted minutes of care were calculated by multiplying the Provincial total patient days associated with each RUG group 1 by the WWMPD for the group 2. these were then summed across all RUG groups to give the Provincial total wageweighted minutes per day (WWMPD ptot ), which were subsequently 3. divided by the Provincial total patient days, to yield the WWMPDpavg. The value of the CMI for each RUG group was then calculated as the ratio of the groupspecific WWMPD to the WWMPD pavg (that is, the ratio of the RUG-group WWMPD to the average WWMPD for patients in chronic care in Ontario, in fiscal year 1997/98). 1 See later section "Calculation of the patient days associated with the RUG-III classification for each MDS assessment within an episode" for method of linking patient days to particular RUG groups. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 3

2.0 Calculation of Patient Volumes for Ontario Chronic Care Beds Case-mix weighted chronic care patient volumes were calculated for Ontario hospitals with designated chronic care beds. Patient volumes were expressed in units called Resource Utilization Group weighted patient days (RWPD). This section of the report describes the derivation of those patient volume units. 2.1 Data Quality: Assessing the Ontario MDS data The Ontario MDS data, which are held in the Canadian Institute for Health Information's (CIHI) Ontario Chronic Care Patient System (OCCPS) data bases were checked in several ways to ensure that they would be of sufficient quality for deriving valid case-mix adjusted patient volumes. The internal consistency of the RUG-III Activities of Daily Living (ADL) Index and the convergent validity of that index and the MDS Cognitive Performance Scale (CPS) were reported for the fiscal year 1997/98 OCCPS data in a recently released CIHI report (CIHI, 1999). Both of those MDS subscales are central to the RUG-III classification algorithm. Whether categorized by fiscal quarter, geographic region, or facility size, the internal consistency (Chronbach's alpha) RUG-III ADL Index exceeded 0.90. A threshold of 0.80 for that statistic is typically used as an indicator of good reliability for a scale with parallel items. The correlation of the RUG-III ADL Index and the CPS was examined in the FY 97/98 OCCPS MDS data by fiscal quarter, facility size and region. In all cases the correlation coefficient had a value of approximately 0.5. These results are entirely consistent with clinical expectation and with results obtained in a similar study in the U.S. (Phillips and Morris, 1998). Prior to the analyses for calculating RUG-weighted patient days, the OCCPS data were subjected to extensive data logic checks and a limited number of other accuracy audits. A complete report on the results of the data logic checks is in preparation for the Joint Policy and Planning Committee's Complex Continuing Care Funding Work Group. Briefly, these analyses found that for the vast majority of records the data elements were within the prescribed valid value ranges and items with specific logical relationships conformed to the expected logic. For example, patients recorded as comatose were not also recorded to be engaged in activities. The very small number of exceptions to such logical relationships was not deemed to be a material threat to the validity of the RUG classification and patient volume calculations. Unlike the inpatient data base used in acute care settings, (CIHI's Discharge Abstract Database), the nature of a complex continuing care patient's admission, and the comprehensiveness of the MDS assessment, require that the data be collected and maintained within a relational data base. Moreover, since the MDS patient data are longitudinal (collected at least quarterly) and a patient may also have several care episodes, it is necessary to employ stringent data quality procedures and unique patient identifiers in order to appropriately track patient data within and among admissions to complex continuing care. Duplicate records were known to exist for some admissions in the OCCPS data base. In most cases, the duplicates are present in the data base due to an error in a data submission involving one or more of the seven variables used by the Canadian Institute for Health Information to uniquely identify a particular patient for a particular admission. The use of multiple variables to define admissions provides a safety net to prevent attributing MDS data to the wrong patient. However, it also can result in erroneous duplicate records for some admissions if a facility submits another admission record for the same patient with a different value in one or more of the six admission identifier items other than Date of Entry (MDS item Ab1). This error was found on 143 occasions Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 4

(among 34,044 records) in the OCCPS admission/discharge file used for this analysis. If ignored, these duplicate records would result in inflated patient days estimates for these facilities. The magnitude of the error would vary by the size of the facility affected. There were also duplicate records found among the MDS 2.0 assessment records. The reasons for these errors and some methods to reduce their introduction into the OCCPS data base is the topic of a forthcoming report in preparation for the JPPC. A series of queries were used to identify duplicate records in the OCCPS files and to pick out the records most likely to be in error. In most cases the erroneous record was not difficult to identify. Most often the incorrect record would be somewhat incomplete, or would have a different value of one of the unique identifier variables (for example, sex) than any of the other records for that patient. In the relatively infrequent situation wherein it was difficult to identify the incorrect record, the one with the latest time stamp (indicating it had been received later at CIHI) was selected as the correct record. The erroneous duplicate was then omitted from further analyses. Earlier work with the OCCPS data base indicated that missing discharge dates could create a problem in trying to calculate patient volumes (days) accurately for several facilities. In order to minimize this problem, early in May 1999, CIHI distributed a special "data audit" report to all Ontario facilities with chronic care beds. This report highlighted admissions with no discharge date in the OCCPS data base between July 1996 and May, 1999. Facilities were asked to confirm whether these patients were still residing in their facilities, and to fax back the dates of discharge of those who had been discharged. Discharge dates before April 2, 1998 were gleaned from the responses returned by facilities and the copy of the OCCPS database used for these analyses (for FY 97/98) was updated to include these discharge dates. Over 95% of facilities responded to this special audit request. 3.0 Calculation of RUG-III Case-mix Index Weighted Patient Days (RUG-weighted patient days) In the funding system being developed for Complex Continuing Care (CCC) in Ontario, the basic units of patient volume are days of patient care (patient days). Funding equity is introduced by weighting the days of care for patients by the resource intensity of their care relative to that of the average CCC patient in Ontario. The weights used, called case-mix indices (CMIs), are based on wage-weighted staff time. Here we have labeled these RUG-III case-mix index weighted patient days as RUG-weighted patient days (RWPD). Calculation of RUG-weighted patient days involved four distinct steps that are listed here and then discussed in sequence in this section: 1. RUG-III classification for each MDS assessment associated with an episode of care. 2. Calculation of the patient days associated with the RUG-III classification for each MDS assessment within an episode. 3. Weighting of patient days 4. Calculation of the facility total RWPD and a facility summary score, the Facility Case- Mix Index (FCMI). Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 5

3.1 RUG-III classification for each MDS assessment associated with an episode of care. The structured logic (algorithm) for the RUG-III grouping system is available on the United States Health Care Financing Administration s (HCFA) World Wide Web site (http://www.hcfa.gov/medicare/hsqb/mds20/mdssoftw.htm). After careful consideration, the JPPC CCCFWG selected version 5.11 as the official version of the RUG-III grouper to be used in developing the funding methodology for Complex Continuing Care in Ontario. The RUG grouper can be run in a "hierarchical" or an "index maximizing" mode. The "index maximizing" approach to classification ensures that the patient is classified into the group with the highest case-mix index for which they qualify. For these analyses, the RUG algorithm was run using the "index maximizing" approach to classify patients into a RUG-III group for each MDS assessment they had undergone. 3.1.1 RUG items missing from the Quarterly assessment Data from 105 items of the RAI-MDS are used to classify patients into the 44 RUG-III groups. Unfortunately, in the design of the OCCPS, three items required for RUG classification were omitted from the (reduced) Ontario-specific Quarterly assessment form. These items are (1) percent of calories by tube or parenteral route (item K6a); (2) amount of fluid by tube or parenteral route (item K6b); (3) diagnosis of diabetes mellitus (item I1a). These items do however, appear on the Full assessment form, which is used for the admission assessment (within 14 days of admission), for annual re-assessments, and for assessment after significant clinical change has occurred. While a RUG classification is still possible in the absence of data for these items, it is possible, in relatively rare instances, that patients would be misclassified into a lower RUG group, simply due to the missing information. Assuming that the answers to these three items would be stable over the course of up to a year, we carried forward their values from the most recent previous Full assessment to subsequent Quarterly assessments for each episode of care during FY 97/98. Thus, the potential downward misclassification of patients at Quarterly assessments due to missing data in the three missing RUG items was avoided. 3.2 Calculation of the patient days associated with the RUG-III classification for each MDS assessment within an episode. For each admission to a chronic care bed in Ontario since July 1, 1996 (and for patients occupying a bed at that time), the OCCPS/MDS database contains, at minimum, the admission and discharge (if any) date(s) along with patient identifiers, demographics, source of admission and the discharge status. 2 All patients are to be assessed, using the Full MDS 2.0 clinical assessment no later than the fourteenth day after their admission. The Ontario Ministry of Health has not required Full MDS assessment to be completed for patients staying less than 14 days. For all episodes with MDS assessment data, the number of patient days to be associated with the RUG classification for each MDS assessment was calculated as follows. 1. The days from the admission date to the day before the reference date of first MDS assessment were associated with the RUG classification resulting from that MDS assessment. 2 Although, by law, submissions of data to the OCCPS were required for all patients admitted to or present in chronic care beds since July 1996, it is known that for a relatively small, but unquantified number of cases, no data have been submitted. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 6

2. The days from an MDS assessment reference date until discharge or the day before the reference date of the next subsequent MDS assessment were associated with the RUG classification of the first MDS assessment. Only patient days during FY 97/98 were counted. For patient stays that carried over from the previous, or into the next fiscal year, the patient days within FY 97/98 and their RUG group assignment were calculated as follows. 1. In the event that a patient was carried over into FY 97/98 from the previous fiscal year, the RUG classification for their final assessment in quarter four of FY 96/97 was assigned to their patient days from April 1, 1997 until discharge or the day before the reference date of their first MDS assessment in FY 97/98, whichever came first. 2. In the event that a patient was carried over from FY 97/98 into the next fiscal year, if they had a MDS assessment in quarter four of FY 97/98 the RUG classification for that assessment was assigned to their patient days from the assessment reference date through March 31, 1998. 3. In the event that a patient was admitted in the last days of FY 97/98 but was not assessed until early in the first quarter of FY 98/99, the RUG classification for that assessment was applied to the patient days from the date of entry through March 31, 1998. Occasionally, there were admissions with missing discharge data. For example, there were patients in the OCCPS data set with no discharge date for a particular admission, but who had a subsequent re-admission to the same facility. In other cases, there were patients who did not have a discharge date associated with an admission, however their MDS assessments did not continue throughout the fiscal year. In these cases, calculations of patient days were conducted as indicated in steps 1 to 3, except that the "discharge" date for the admission was taken as either: a) the day before a subsequent admission of the same patient in the same facility, or b) the first day of the fiscal quarter following that in which the last MDS assessment was completed. The OCCPS data base also contains admission and discharge information about episodes of care for which no MDS assessment data are available. Most such episodes are short stays of 13 days or less, for which MDS assessment is not mandatory. For episodes without MDS assessment data, the patient days during FY 97/98 were calculated as the difference between the admission date and the discharge date. April 1, 1997 was taken as the "admission" date for such patients present in a chronic care bed at the start of FY97/98. April 1, 1998 was set as the "discharge" date for such episodes if there was no discharge date in the data base. 3.2.1 Episodes for which no MDS assessment data were available Some of the episodes without MDS assessment resulted from readmission, for a relatively short stay, of a patient who had recently been discharged from chronic care. In cases where this readmission to chronic care was expected 3 and the patient s clinical condition had not changed significantly, then, the patient s next MDS assessment is not required until 90 days after the last MDS of the preceding episode (rather than by day 14 after admission). However, in this kind of circumstance if the patient was discharged 3 That is, the Reason for Assessment (item AA8) on the discharge tracking form submitted for the episode prior to the readmission was was discharged - return anticipated (code = 7). Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 7

from the second episode earlier than the due date for their MDS assessment, there would be no data for RUG classification of the patient days during the second episode. Therefore, patient days for an episode of care, for which no MDS data were available, were associated with the patient s last RUG classification in their most recent previous episode if the patient was discharged from the second episode earlier than 90 days after the date of the MDS assessment in the prior episode. Apart from the scenario just described, patient days during episodes of care for which no MDS data were available could not associated with a RUG group. 3.2.2 An Example of How Patient Days were Calculated and Associated with RUG Groups Figure 1 is a schematic diagram to illustrate how patient days were calculated for RUGweighting. In the illustrated scenario, the patient was first admitted to chronic care on January 13, 1997, and assessed with the MDS on January 27 (before FY 97/98 began). A second (the Quarterly) MDS assessment for this admission was done on April 28, 1997 (early in FY 97/98) and the patient was discharged on May 25. For this admission, the 27 patient days from April 1 through April 27, 1997 were assigned the RUG group from the classification based on the January 27 th MDS assessment. The 27 patient days from April 28 th through May 24 th were assigned the RUG group from the classification based on the April 28 th MDS assessment. The patient was discharged on May 25 th, with return expected. The second admission of the patient was an expected re-entry that occurred within 90 days of the most recent MDS assessment and the patient's clinical condition was not significantly changed in the interim, so a Quarterly MDS assessment was completed in line with the assessment schedule initiated in the first admission. The 55 patient days from July 1 through August 24 th were assigned the RUG group from the MDS assessment on July 28 th. In the week of August 25 th, the patient experienced a significant clinical deterioration and a Full MDS assessment for a significant change was performed on August 25 th. The patient was discharged to acute care sometime after August 25 th, and return of the patient was not expected. Unfortunately, a discharge tracking form was accidentally not submitted to CIHI for this discharge. The patient was re-admitted from acute care on October 1 and a new MDS assessment cycle started by completion of a Full admission assessment on October 15. Since the date of discharge for the patient's previous admission was unknown, the 37 patient days counted from the last assessment on August 25 th through September 30 th, the day before the current admission, were assigned the RUG group from the August 25 th assessment. The 106 patient days from October 1, 1997 through January 14, 1998 were assigned the RUG group from the October 15 assessment. The patient was assessed (Quarterly) on January 15 and the 26 days through February 9 th were assigned to the RUG group for that assessment. The patient was discharged, return not expected, on February 10. The patient was re-admitted March 12, 1998 and assessed with the MDS on April 3 (after the close of FY 97/98). The 20 patient days from March 12 th through March 31 st (the last day of FY 97/98) were assigned to the RUG group for the assessment completed on April 3, 1998. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 8

Figure 1: Schematic diagram to illustrate calculation of patient days component of RUG-weighted patient days. 1A M M D Jan Jan Apr May 13 27 28 25 2A M M d Jul Jul Aug 1 28 25 3A M M D Oct Oct Jan Feb 1 15 15 10 4A M D Mar Apr 21 3 Jan Apr Apr June 1 1 1 30 1997 1997 1998 1998 The figure shows the admission dates (A), MDS 2.0 assessment reference dates (M) and discharge dates (D) in the OCCPS data base for a single, hypothetical patient in chronic care during fiscal year 1997/98. This series of admissions for a patient has been fabricated to illustrate several of the scenarios encountered in calculating patient days from the OCCPS data. A lower-case "d" indicates a discharge that occurred for a particular admission of this patient, but for which no discharge date is recorded in the OCCPS data base. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 9

3.3 Weighting of Patient Days 3.3.1 Patient days associated with a RUG group After the days of care for each episode were assigned to RUG groups (where possible), as described above, all days assigned to each RUG group were summed within that group for each facility. The RUG-group-specific total patient days were then multiplied by the CMI for that RUG group to obtain patient days weighted by relative resource intensity. This multiplication of days by CMI was repeated for each of the 44 RUG groups in order to calculate the distribution of RUG-weighted patient days (RWPD) by RUG group for each facility (to be distributed separately). 3.3.2 Patient days not associated with a RUG group Patient days from episodes with length of stay (LOS) less than 14 days, with no MDS assessment data and which could not be associated with the RUG group of a previous episode, could not be RUG-weighted (Appendix B, Column C). No RUG group could be ascertained for these patient days. Currently there is little evidence available to indicate how these patient days should be weighted. For the purposes of this analysis, such days for each facility were weighted by the average patient day weight calculated for that facility based on patient days that were associated with a RUG classification (Appendix B, Column G). Further work is required to investigate the clinical characteristics of these short-stay patients in order to determine the fairest way to weight these patient days in a funding formula. Such patient days comprised 0.9 % of all Ontario chronic care patient days accounted for in the OCCPS for FY 97/98. Patient days from episodes with LOS of 14 days or more that were not associated with a RUG group (Appendix B, Column D) were assigned the CMI of the lowest RUG group (CMI = 0.4573) by default, since, by Ministry of Health mandate, a MDS assessment should have been submitted for such episodes. Such patient days comprised 0.7% of all Ontario chronic care patient days accounted for in the OCCPS for FY 97/98. 3.4 Calculation of Facility Total RUG-weighted Patient Days 3.4.1 Facility total RWPD The RUG group-specific RWPD were summed together with the non-rug-groupassociated weighted patient days to obtain the total RUG-weighted patient days (Total RWPD) for each facility. (Appendix B, Column H) 3.4.2 Calculation of a facility-level case-mix summary score: the Facility Case-mix Index (FCMI) In order to facilitate case-mix comparisons between facilities, a facility-level index, the Facility Case-mix Index (FCMI), was calculated for each hospital. The FCMI is calculated by dividing the facility s total RUG-weighted patient days by their total (unweighted) patient days during a particular time period (in this case, FY 97/98). The FCMI can be interpreted as the average CMI for the facility. It represents the average resource intensity of patient days in that hospital compared to the average resource intensity of patient days for the Province, which, by definition, have a patient day weight (i.e. CMI) of 1.0. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 10

The FCMI was calculated as the ratio: Total RWPD for the facility Sum of all patient days for the facility where the numerator, the total RUG-weighted patient days (Appendix B, column H) was calculated as the sum of: (a) the sum of patient days weighted by their associated RUG group CMI (Appendix B, column F,) and (b) patient days not associated with a RUG group (Appendix B, column C), weighted by a facility average CMI based solely on patient days that were associated with a RUG group (Appendix B, column G), and (c) patient days not associated with a RUG group (Appendix B, column D), weighted by the default CMI of 0.457. 4.0 Calculation of Cost per RWPD 4.1 Linkage of Total RUG-weighted Patient Days to Facility Cost Data from the Management Information Systems Data Based on financial data supplied by all hospitals to the Ontario Ministry of Health in compliance with Management Information Systems (MIS) requirements, the Ontario Cost Distribution Methodology (OCDM) is used to allocate costs to Direct Care costs (Appendix B, column J) and Overhead costs (Appendix B, column K). The sum of Direct Care costs and Overhead costs yields Total costs (Appendix B, column M). Hospitals with different levels of care (e.g. chronic care, rehabilitation, acute care) must appropriately allocate costs to the different levels of care when submitting their MIS data to the Ministry of Health. Direct Care costs and Total costs allocated to chronic care were derived for each hospital with chronic care beds 4. These cost data were divided by total patient days calculated from the OCCPS/MDS data to yield facility average Direct Care and Total cost per diem (Appendix B, columns N and O, respectively). The data were also divided by total RUGWPD to yield facility average Direct Care and Total cost per RUG-weighted day (Appendix B,.columns P and R, respectively). 4.2 Reference Data and Calculations Used to Check the Validity of the Patient Volumes (Days) Calculated Using the OCCPS/MDS Data Two sources of facility total patient days for FY 97/98, based on data sources external to the MDS, and a further method of calculating patient days using OCCPS data were employed to check the validity of the number of patient days calculated based on the MDS data. The Ministry of Health supplied a report of total chronic care patient days for Ontario hospitals, based on their submissions of midnight census information to the Ministry(Appendix C, column C). The Ministry also provided the total chronic care patient days reported by hospitals in their MIS submissions (Appendix C, column D). Total patient days can also be calculated based solely on admission and discharge information in the OCCPS database. 4 Appendix D has been provided to facilitate auditing of cost allocations within Direct Care. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 11

4.2.1 Calculation of Patient Days based on Admission and Discharge Data in the OCCPS (Appendix C, column E) Records from the Admission/Discharge file were filtered to include only those where: the patient was admitted prior to April 1, 1998 AND the patient was not discharged prior to April 2, 1997 AND the patient was discharged prior to April 2, 1998 OR was not discharged Using these records, chronic care patient days were calculated for all patients in chronic care during fiscal year 1997/98 (FY 97/98) as described below. Accumulation of patient days for each patient admission began at 00:00 o'clock on April 1, 1997, for admissions prior to April 1, 1997 at 00:00 o'clock on the date of entry to the chronic care bed for admissions admitted on any day between April 1, 1997 and March 31, 1998, inclusive Accumulation of patient days associated with each admission ended at 23:59 o'clock March 31, 1998, for patients not discharged on or before March 31, 1998 at 23:59 o'clock of the day before the date of separation, for patients discharged between April 1, 1997 and March 31, 1998, inclusive. Patient days were summed within facilities and across facilities to obtain facility-specific and province-wide total chronic care patient days, respectively. 4.2.2 Agreement Between the Number of Patient Days Calculated in the RUGWPD Process Versus Facility Total Patient Days from Other Sources In general, for most facilities there is substantial agreement among the patient day totals calculated as part of the RWPD calculations and the total patient days from the midnight census, MIS data and MDS admission/discharge information only. Disagreements among these values may occur for several reasons. Total patient days calculated for RWPD may be greater than or (most often) less than those calculated strictly from OCCPS admission/discharge data due to the impact of missing discharge date information in the OCCPS. Wherever a discharge date is missing, the total patient days calculated from OCCPS admission/discharge data only will include all days up to the end of the fiscal year in question, even though the patient may have been discharged some time before the end of the fiscal year, or even before it began. In the RWPD calculations, patient days stop accumulating in these cases on the day before a subsequent re-admission, or at the end of the fiscal quarter in which the last MDS assessment took place if no re-admission occurred. Errors may also have occurred in facilities' counting and reporting of patient days for MIS or in census information. None of these additional measures of patient days is truly a "gold standard" reference. Discrepancies indicate that investigations must be made to determine the correct values and cleanse the data. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 12

Separations from chronic care beds can also be calculated from both the midnight census data submitted to the Ministry of Health and from the OCCPS. Separations based on these two data sources also show substantial agreement and are reported in Appendix C. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 13

References Cited Abicht-Swensen LM, Debner LK. 1999. The Minimum Data Set 2.0: A functional assessment to predict mortality in nursing home residents. American Journal of Hospice and Palliative Care 16(3): 527-532. Beaulne P, Hollander M, Lyle M, Schalm C, Wanke M, Warren S. 1999. Continuing Care Outcomes Project: Classification system development final report to Alberta Health. Pinnell Beaulne Associates Ltd. Edmonton, AB. Blaum CS, O'Neill EF, Clements KM, Fries BE, Fiatarone MA. 1997. Validity of the minimum data set for assessing nutritional status in nursing home residents. American Journal of Clinical Nutrition 66(4): 787-794. Canadian Institute for Health Information (CIHI). Prepared for CIHI by Nenadovic M, Gilbart E, Hallman K, Teare G, Hirdes JP. 1999. Provincial Mini-Status Report: The Quality of Caring: Chronic Care in Ontario April 1, 1997 to March 31, 1998. Carpenter GI, Ikegami N, Ljunggren G, Carrillo E, Fries B. 1997. RUG-III and resource allocation: comparing the relationship of direct care time with patient characteristics in five countries. Age and Ageing 26-S2: 61-66. Fries BE, Cooney LM. 1985. Resource utilization groups: A patient classification system for long-term care. Medical Care 23:110 Fries BE, Schneider DP, Foley WJ, Gavazzi M, Burke R, Cornelius E. 1994. Refining a case-mix measure for nursing homes: Resource utilization groups (RUG-III). Medical Care 32(7): 668-685. Gambassi G, Landi F, Peng L, Brostrup-Jensen C, Calore K, Hiris J, Lipsitz L, Mor V, Bernabei R. 1998. Validity of diagnostic and drug data in standardized nursing home resident assessments. Medical Care 36(2):167-179. Hartmaier SL, Sloane PD, guess HA Koch GG, Mitchell CM, Phillips CD. 1995. Validation of the Minimum Data Set Cognitive Performance Scale: agreement with the Mini-Mental State Examination. Journal of Gerontology A (Biological and Medical Sciences) 50(2): M128-M133. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 14

Hawes C, Morris JN, Phillips CD, Mor V, Fries BE. 1995. Reliability estimates for the Minimum Data Set for nursing home resident assessment and care screening (MDS). The Gerontologist 35(2): 172-178. Ikegami N, Fries BE, Takagi Y, Ikeda S, Ibe T. 1994. Applying RUG-III in Japanese longterm care facilities. Gerontologist 34:628-639. Ljunggren G, Fries BE, Winblad U. 1992. International validation and reliability testing of a patient classification system for long-term care. European Journal of Gerontology 1: 372-383. Mor V, Branco K, Fleishman J, Hawes C, Phillips C, Morris J, Fries B. 1995. The structure of social engagement among nursing home residents. Journal of Gerontology Psychological Sciences 50B(1): P1-P8. Morris JN, Nonemaker S., Murphy K, Hawes C., Fries BE, Mor V, Phillips C. 1997. A commitment to change: Revision of HCFA's RAI. Journal of the American Geriatrics Society 45: 1011-1016. Phillips CD, Morris JN. 1998. The potential for using adminsitrative and clinical data to analyze outcomes for the cognitively impaired: An assessment of the Minimum Data Set for nursing homes. Alzheimer Disease and Associated Disorders 11: 162-167. Resnick NM, Brandeis GH, Baumann MM, Morris JN. 1996. Evaluating a national assessment strategy for urinary incontinence in nursing home residents: reliability of the Minimum Data Set and Validity of the Resident Assessment Protocol. Neurology and Urodynamics 15: 583-598. Schneider DP, Fries BE, Foley WJ, Desmond M, Gormley W. 1988. Case mix for nursing home payment: Resource utilization groups, version II. Health Care Financing Review. 1988 Annual Supplement: 39-52. Sgadari A, Morris JN, Fries BE, Ljunnggren G, Jónsson PV, DuPaquier J-N, Schroll M. 1997. Efforts to establish the reliability of the Resident Assessment Instrument. Age and Ageing 26-S2: 27-30. Zimmerman DR. 1997. The power of information: Using resident assessment data to assure and improve the quality of nursing home care. Generations 21(4): 52-56. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 15

Glossary Of Terms Term Definition Episode (of care) The inpatient period betweeen an admission (whether initial or reentry) and a discharge. Minimum Data Set (MDS 2.0) The standardized, multidimensional patient assessment developed in the U.S. for inpatient continuing care. In 1996, the Ontario Ministry of Health mandated that patients in chronic care beds be assessed, using the MDS 2.0, at least quarterly. Ontario Chronic Care Patient System (OCCPS) The name of the database system, administered by the Canadian Institute for Health Information (CIHI), which contains the MDS 2.0 data collected by Ontario Hospitals with chronic care beds. Resource Utilization Groups-III (RUG-III) The third version of a case-mix classification system for continuing care inpatient settings. This version of the RUG system was first published in 1994 and was updated by the U.S. Health Care Financing Administration in 1997. There are 7 major hierarchical categories, containing a total of 44 classification groups ( RUG groups ). 105 MDS data elements are used in the RUG-III grouper algorithm. Case-Mix Index (CMI) The CMIs are the weights used in RUG-III to reflect the relative resource intensity of different patient groups. A CMI is ratio of the average wage-weighted staff (nursing and therapies) time consumed for care of patients in a particular RUG group to the average wage-weighted staff time of all patients in the relevant population. For example, patients in a RUG group with a CMI of 1.5 would need care that, on average, is 50 % costlier than the average chronic care patient in Ontario. The average patient has, by definition, a CMI of 1.0. Cost Per Case-Mix Weighted Activity For Complex Continuing Care In Ontario (RD #8-12) 16