Exchanging MDS Data Across Sectors of the Health Care System: Clinical Considerations

By John P. Hirdes, Ph.D., Margaret Riddell, RN, BA, Nancy Curtin Telegdi, RN, MA

Acknowledgements

This project was supported by a financial contribution from the Health Transition Fund, Health Canada. The views expressed herein do not necessarily represent the official policy of federal, provincial or territorial governments. Additional support was provided to various aspects of this research by the Ontario Ministry of Health and Long-term Care, Ontario Hospital Association, the Ontario Joint Policy and Planning Committee, Canadian Institutes for Health Research, Providence Centre Foundation, and interRAI.

In July 1996, the Ontario Ministry of Health mandated the use of the MDS 2.0 assessment instrument for all chronic units and hospitals on a quarterly basis. Interest in this instrument grew rapidly as other provinces recognized the potential benefits of the MDS 2.0 as a data source for case mix based funding, care planning, outcome measurement and quality improvement. For example, Saskatchewan and the Yukon have recently mandated the implementation of the instrument for all long-term care facilities in those jurisdictions.

As interRAI began to develop new MDS instruments for home care, mental health, postacute care (rehabilitation), acute care and palliative care, an integrated health information system linking the major sectors of the health care system emerged as a clear possibility (Hirdes et al., 1999). These new MDS instruments were designed to address the specific needs, strengths and preferences of distinct populations in health care. At the same time, they share a common clinical emphasis on function rather than diagnosis alone, common terminology, common data collection methods, and a common core set of items.

The advantages of having a compatible approach to assessment across the health care system led Nova Scotia, Manitoba and British Columbia to initiate implementation studies of multiple MDS instruments. With similar measures, one can compare how the needs of similar individuals are responded to in different sectors of the health care system. Moreover, these instruments can allow health professionals from one sector to communicate with their counterparts from another sector through a common clinical language.

For persons receiving services, poor communication, repetitive questions and interruptions in care have been the source of considerable frustration as they move through different parts of the health care system. The MDS series of instruments can go a long way to addressing these problems, provided that these assessment data are used effectively and in a clinically appropriate manner.

It may seem like a relatively simple proposition to “zap” data from one care provider to another before the patient even arrives at the doorstep of the second organization. However, the possibility of this data exchange raises numerous new questions that must be addressed in the implementation process. For example, how do considerations of privacy affect the exchange of data between health care agencies, even if it is done in the name of an individual’s care? What are the technical considerations for dealing with differences in coding between MDS instruments that were the results of new research on item scaling, design requirements for a particular population or chance variations that occur with any complex development process? How can we reduce assessment burden by streamlining data collection between sectors?

The term auto-population refers to the use of the data elements from a previously completed assessment to automatically complete data elements in a subsequent assessment. This has considerable appeal to persons interested in reducing the time it takes to complete assessments, particularly in the case of transitions between agencies using compatible MDS instruments (e.g., home care and nursing homes). If the computer can automatically fill out the item, why should the resident have to be assessed in that domain area a second time? Simply put, the resident may have changed. The development of clinical guidelines related to data exchanges between MDS assessments has been an important focus of the RAI-Health Informatics Project (RAI-HIP).

RAI-Health Informatics Project (RAI-HIP)

The RAI-Health Informatics Project (RAI-HIP) was a two-year study funded by Health Canada’s Health Transition Fund, which aimed to examine the use of the MDS series of assessment instruments for community, hospital and institutional settings. The RAIHIP effort was comprised of four main sub-studies: a) Integrated Health Information Systems (IHIS) - a longitudinal study of persons receiving home care, long-term care, in-patient mental health or acute care in 6 Ontario cities. These individuals were assessed as part of normal clinical practice using RAI instruments designed for the sector in which they were receiving care. They were followed longitudinally until their discharge from that sector of health care; b) Home Care Quality Indicators (HCQI) – a cross-sectional study of Community Care Access Centre (CCAC) clients in 10 Ontario cities done in collaboration with parallel studies in the US and Japan. Data from the study were used to develop and validate HCQIs measuring the process and outcomes of home care; c) RAI-Mental Health (RAI-MH) Pilot Implementation – a study done in collaboration with the Ontario Joint Policy and Planning Committee (JPPC) to support voluntary implementation of the RAI-MH in 34 psychiatric units and hospitals across Ontario. This study was done in conjunction with the JPPC’s staff time measurement study (STM) with the aim of developing a case-mix classification system for in-patient psychiatry; and d) RUG-III/RAI-Post Acute Care (RAI-PAC) study – this research involved a comprehensive, critical review of the literature on the use of Resource Utilization Groups (RUG-III) for funding long-term care and complex continuing care. Also, the RAI-PAC was piloted tested in rehabilitation hospitals and complex continuing care settings to evaluate its potential use in those settings.

Clinical Guidelines for Data Exchanges

One deliverable of the IHIS sub-study was the development of guidelines for clinicians on how to use data from a previous MDS assessment when completing intake assessments. The preliminary recommendations of the RAI-HIP team in this regard are provided below. The transition from home care to a long-term care facility was used as the context for these recommendations; however, most of the principles underlying these specifications would also pertain to other transitions.

In deciding whether a particular data element might be a candidate for auto-population there are four key factors to consider:

To create guidelines for the potential auto-population of items from MDS HC and MDS 2.0, items from both instruments were first organized to identify those that were a direct match or could be “cross-walked” by recoding one or both into the same scale for responses. These pairs of matched items were then categorized into the following groups:

Only the first two categories of items are recommended for auto-population between instruments over time. For the remaining categories, prior assessments can be used as sources of information that clinicians should evaluate as they complete the new assessment.

Even though some items in a given section may be auto-populated from an earlier assessment, no section should be regarded as complete until the clinician has fully considered all remaining items. For example, important new diagnoses might be missed if the assessor fails to consider diseases or medical conditions not previously identified.

Table 1 lists items that can be considered highly unlikely to change from the last assessment to current one. The project team was therefore confident in identifying these as candidates for auto-population if the previous assessment was completed within the past year.

Table 2 shows items that could likely be auto-populated provided that the assessor reviews and confirms the values of these items prior to signing off the assessment. These items could be auto-populated with relative ease from the MDS-HC to MDS 2.0, but the coding differs somewhat between assessments. The MDS-HC requires specification of whether a disease is subject to active monitoring by a home care professional, whereas the MDS 2.0 focuses only on the presence or absence of conditions that affect the resident’s status or treatments received.

The balance of the MDS items should not be considered candidates for auto- population; however, previous MDS assessments could be regarded as information sources that assessors should consult in determining the individual’s current status. Information from prior assessments may be particularly useful for tracking outcomes over time and can inform the clinician of whether the characteristics s/he now observes are new or long-standing conditions. This is helpful, for example, in detecting delirium.

Table 3 lists items for which a substantial portion of individuals can be expected to have remained relatively stable within 90 days of the current assessment reference date. However, due to the possibility that change could occur within 90 days, assessors should review this carefully with attention to the time lines specified in each item.

Table 4 shows items that may be consistent within a 30 day time frame, but whose status may change over a longer time frame. It should also be noted that the absence of some these conditions may be less stable over time than their presence. For example, once a feeding tube is used it may continue to be used.

Table 5 provides examples of items that may change rapidly making them unsuitable candidates for auto-population. These items should generally be fully reassessed, because one cannot be certain that the person’s status has remained stable.

To Auto-populate or Not to Auto-populate

The primary conclusion of our initial consideration of data exchanges between sectors using the MDS instruments is that there are relatively few circumstances in which autopopulation is a reasonable clinical strategy. For the overwhelming majority of assessment items, one cannot a priori rule out the possibility that a meaningful clinical change has occurred. Of course, further research needs to be done to obtain precise normative data on the typical levels of stability in these items. However, such research is unlikely to provide a great deal of new evidence in favour of auto-population.

It should also be recognized that the transition event (e.g., nursing home admission) might itself be sufficiently stressful to trigger changes even in items that tend to be relatively stable in most circumstances. Therefore, the utmost of caution should be used when making assumptions that the person’s traits are unchanged.

One might ask what are the potential pitfalls of auto-population? There may be many. From the perspective of the resident, new needs and new strengths may go undetected. Care plans may no longer be appropriate if the individual has changed for the better or the worse. The failure to prevent a small decline may well be the start of a cascade of events leading to catastrophic decline. From the perspective of the organization, the failure to detect declines in function may have resource implications because these changes are not reflected in the organization’s case-mix. On the other hand, the failure to detect improvements in function means that the organization’s quality of care measured by such outcomes would be under-estimated. For policymakers and program evaluators, auto-population can reduce sensitivity to change making it more difficult to identify cost-effective programs and services.

The promise of an integrated health information system based on the MDS instruments is not realized through a reduced reliance on health professionals by the automatic completion of assessments. Rather, improvements in the continuity, outcomes and quality of care can best be achieved when historical assessment information is organized in a way that provides health professionals with a complete view of the person’s needs, strengths and preferences over time.