The Subcommittee on Populations of the National Committee on Vital and Health Statistics was convened on Friday, January 22 in the Hubert H. Humphrey Building in Washington, D.C. The meeting was open to the public. Present:
The Subcommittee on Populations held an all-day meeting on January 22, to educate themselves about the data issues relating to what is commonly called "post-acute care." (The Subcommittee is looking for a more appropriate term.) The presenters and program:
Overview of post-acute care
Post-acute care: data, findings and sources
Outcome research and overview of quality of care issues
Pediatric post-acute care and quality issues
Data issues
The following questions, posed by Dr. Kane in his overview, were to some extent addressed by all of the day's speakers:
One of Dr. Kane's emphases was the lack of a basis for systematic decision-making in discharge planning. This relates to another big issue in post-acute care: accountability. He stressed the amount of improvement that might be accomplished by focusing on post-hospital care and holding hospitals accountable for good discharge decisions. He concluded by outlining several questions about quality, which other speakers also addressed:
Dr. Kane advocated doing something to create a consistent measure, without trying to "capture each nuance of each respondent." This position received support from other presenters.
Dr. Liu outlined the trends in the growth of post-acute care and the distribution among the different types of providers. He, along with his fellow presenters, stressed the need for a coherent, integrated system in terms of payment, terminology, and data. He also identified several pressing research needs, such as comparisons of the different providers and outcomes research. He focused on better assessment instruments as a major priority for improving information in this area.
Dr. Kramer is starting a study to look at the quality indicators needed to measure post-acute care outcomes, how to collect those indicators, and what is available. He reviewed several key issues in post-acute care that affect quality assessment, and their implications in terms of information needs. He called particular attention to the need for self-report among the quality assessment tools.
Mr. Webb is President and CEO of Village Care of New York, a community-based not-for-profit that provides clinical services for people with AIDS. His experience bears out the analyses of the previous speakers regarding the lack of rationality in data and discharge planning. Village Care is building a database using the format and intent of the MDS and OASIS, and facing the challenge of the lack of standardized quality indicators in the MDS.
Dr. Ireys reported that for children with special needs, the field has finally come together around a non-categorical functional approach to definition, including behavioral, developmental, emotional or physical conditions as well as conditions that lead to dependency on compensatory mechanisms. He then discussed data sources and data gaps regarding this population, and shared some basic data and characteristics, e.g: A chart book funded by the Maternal and Child Health Bureau (availability projected for early 2000) with data from the NHIS child disability supplement. Discussion with the Subcommittee focused on risk adjustment, the potential for alliances with advocates for adults with special needs, and the challenges of coding and developing appropriate functional measures for children.
Ms. Nonemaker's of HCFA Office of Clinical Standards and Quality has the lead in developing clinical measures and information systems on post-acute care and long-term care. She described the current work on the MDS and OASIS and other systems as: short-term steps toward the long-range goal of building standardized, beneficiary-centered data elements to help consistently measure a beneficiary's clinical characteristics and care needs. These measures would be used in a regulatory quality monitoring system to rate and compare providers on their performance in key areas, and to inform consumers on their choices. HCFA is moving toward building a core data set or common assessment instrument to be used across all post acute care settings. She predicted that the process will take 10 to 20 years. She also described the MDS data structure and its potential role in quality assurance.
Dr. Zimmerman discussed the use of the MDS for quality assurance in the nursing home setting. He called attention to the fallacy of using quality indicators to prematurely reach conclusions and terminate providers, rather than using them to guide ongoing improvement efforts. He also cautioned against "over-adjusting" in risk adjustment and thereby disregarding important causal factors related to quality of care. He stressed the importance of timing of events, a point underscored by Dr. Iezzoni.
Integrative Health Services is a for-profit health care company with long-term care facilities, home care locations, hospices, and long-term acute care hospitals around the country, and a single data system. Dr. Kilgore added his voice to the call for uniform measures across sites of care.
Dr. Hargraves is a sociologist who talks with patients about what happens to them during care. He stressed the importance of talking to patients in order to find out what is important to them. His other major point is the importance of having standardized survey instruments to permit following people and making comparisons over time.
Dr. Iezzoni began by noting the self-educational purpose of this meeting and the Subcommittee's desire to find a more appropriate term than "post-acute" for care across the continuum of settings, the focus of this NCVHS Subcommittee on Populations project. She asked those present to introduce themselves and then welcomed the first speaker.
Dr. Kane said he would offer some concepts to think about. First, "nobody invented post-acute care"; it simply emerged, and then rose to prominence after the imposition of changes in the hospital payment system that led to shorter hospital stays. Through this process, institutions other than hospitals have come to occupy the center of the health policy stage.
For simplicity, post-acute care can be thought of as post-hospital care; but in terms of the services provided, hospitalization has become often superfluous or unnecessary. This leads to the oxymoron, "direct admission post-acute care." An arbitrary definition may be inevitable. The term "subacute care" doesn't work, either. Many of the terms used are a function of who pays, while the services look the same, regardless of what terms are used. Dr. Kane argued that care triggered by a hospitalization, or post-hospital care, really is or should be the focus. He further indicated that discharge planning is "a travesty in this country," with no organization or consistency. Those making the decisions are uninformed about what is the right decision.
Post-acute care comprises three major vendor types: home health care (i.e., Medicare certified vendor); nursing homes (Medicare and non-Medicare); and rehabilitation (both in- and out-patient). Medicare regulations influence the service delivery of all three, "not necessarily for the better." And "the hidden hero of post-acute care is the family," the chief source of informal care.
Defining the timeframe, or "episode" of care is difficult to define. This is compounded by the difficulty in defining where care starts. Statistically, this timeframe may be at approximately three or four months, but Medicare home health care is becoming more long-term care oriented. The other two forms of post-acute care have shorter tenures.
Dr. Kane called six issues to the Subcommittee's attention, and then briefly discussed each one:
Dr. Iezzoni asked whether his comments would apply simply to Medicare or also to the private sector. He said that the private sector has generally followed Medicare models and not been creative about alternatives; thus the comments do have implications for privately financed care.
(1.) Attempts to model who gets care have been unable to account for much of the variation because the behavior is not systematic. Findings do show that the factors influencing who gets this care include diagnosis; functional status (a strong role, involving "careful case selection or cherry-picking"); cognitive status; and living situation. In planning, no systematic prognosis is taking place; decision makers have varying amounts of information about the patient's clinical status, and this does not seem to govern the decision. Rather, "the first train leaving the station is the best one to be on, regardless of where it is going." There is an attraction to getting people into care that is "easier to package."
(2.) Little information exists about what kind of difference post-acute care makes, and it is a difficult thing to study scientifically. Analysis of rehabilitation shows it to be the most effective; but it is expensive.
(3.) There is a big variation in what things cost. Rehabilitation is the most expensive in terms of both direct and indirect costs; i.e., rehabilitation does not seem to save money "down the road," at least with its current case mix. This raises the question of whether the goal is optimum or efficient care.
Another issue is accountability: people receive many kinds of care in their "post-acute care careers," and it is not clear who should be held accountable for results. Dr. Kane's analyses always argue that "the first decision is the critical decision," on which all others are contingent. For example: in looking at data, it is important to know whether the discharge is from hospital or home health care.
(4.) Given that "we have created something that we never meant to do," in response to economic incentives, one solution is to develop some form of point-of-service capitation that makes it possible to bundle services.
Some "great minds" are contemplating what the right bundling would be. One model involves bundling post-acute care services, with a program to allocate patients to the most appropriate service and to be responsible for their care. The second model addresses the lack of a locus of responsibility by placing it at the feet of the hospital, by bundling the payment for hospitalization and post-hospital care. This makes sense clinically, said Dr. Kane, encouraging different behavior by hospital discharge planners. However, the competence of hospitals to do post-acute care may be questionable and seems confined to "high-tech institutional care."
Indeed, hospitals do not need to be in charge of this kind of system; a new corporation could be created that would contract with the hospital and the post-acute care system and manage the latter. Some models are being developed to do that. Realizing potential economic benefits from this approach needs a large-scale integration that goes beyond the current "boutique operations."
Dr. Kane then turned to the information needs with regard to post-acute care. He identified the following:
He defined this as information on prognosis, relationship of selected characteristics to outcomes; the service capacity of post-acute care vendors; what is a unit of service; who is able to care for whom (getting at the phenomenon of "downward delegation"); the specific content of post-acute care; etc.
Finally, Dr. Kane raised some questions about quality:
Asked which of the information and data questions are the highest priority, Dr. Kane pointed to "what makes a difference?" A reasonable database would allow the use of epidemiological methods. Given the current lack of data for developing policy, one option is to turn the problem over to managed care institutions and then try to calculate a fair pay rate. This is the way "most of medical care" is being handled right now, and "not very well." The second approach is to try and understand the system and manage it intelligently, an approach that will require much more data.
Dr. Starfield pointed out that if this domain is not limited to post-hospital care, then holding hospitals responsible for it all makes little sense. She also commented on the difficulties that stem from lumping together facilities-oriented and services-oriented data, and the possibility of "splitting the pie" in another way to come up with a preferable data strategy.
Dr. Kane agreed that ideally, there would be no designations about who gives what care; but that is a long way off. He agreed about data collection, and added that even hospital data are poor sources of information about "what goes on there." The goal is to approximate the ability to track every individual from birth. Still, the current basis for the nation's health statistics is likely to continue (although alternative models are being developed, for example, in Minnesota). He asserted that a key aspect of the problem is how the end of hospitalization --or "launch coordination"-- is handled. Thus, attention should be focused there. He granted that those who don't come through the hospital will be lost to that system; but it could be adapted to accommodate them. The alternative is "a very amorphous system with no social structures to it;" thus difficult to build a data system from.
Dr. Starfield returned to Dr. Kane's point about the defects in hospital data and said she favored starting anew.
Ms. Coltin asked whether data exist on what percent of admissions to various settings are direct, whether the clinical conditions differ, and any changes over time. Dr. Kane commented on the complexity of these issues and noted some of the available data sources.
Don Muse from the audience asked if any research had correlated the type of post-acute care with referring provider characteristics. Dr. Kane has studied that question in terms of selection bias in self-referral by hospitals, finding none. No data from disinterested research studies suggest any patterns of this kind.
Dr. Liu predicted that it will take 25 years or more to answer all the questions raised by Dr. Kane. For his talk, he referred to tables supplied to the Subcommittee, based mostly on claims data.
Medicare spending on skilled nursing facilities (SNFs) has risen dramatically since 1990 (about four-fold), although length of stay has remained fairly constant. The change is an increase in utilization, with more people using the SNF benefit. Similarly, the home health benefit increased more than four-fold. Visits per person also increased greatly during that period.
These trends led to a call for prospective payment systems, mandated in the 1997 Balanced Budget Act as "the immediate kneejerk response." Related post-acute care issues, however, were not addressed, including many enumerated by Dr. Kane.
Medicare spending on in-patient care and SNF and home health care since 1974 has accounted for two-thirds of Medicare spending growth. There appears to be no simple offset between in-patient care and post-acute care. This raises research questions about the potential for substitution and about how much of the expanded growth in post-acute care is due to inappropriate use.
Some one-quarter of all hospital discharges under Medicare result in the use of a post-acute care provider. There are more discharges to SNFs and home health than to rehab and long-term care hospitals, of which fewer exist. About 18 percent move from one post-acute care provider to one of a different type within an episode.
Dr. Liu noted the need for a more rational post-acute care system, with consistent payment across Medicare providers. They are now reimbursed in different ways and have different eligibility policies and requirements.
He posed the question: "How broken is the post-acute care system now?" He showed some research looking at the patient overlap between SNF, home health, and rehab in terms of 32 DRGs, looking at the value of DRGs as indicators. The research does not clarify the degree of overlap in patient types across post-acute care settings. Much more information is needed, especially to compare post-acute care providers. There also is little information on outcomes of patient care. Dr. Liu noted that his department has initiated research on these questions.
Asked about race/ethnicity and gender data on the patients, Dr. Liu said some basic information on these factors does exist through the Medicare claims.
Dr. Iezzoni noted the need for a measure of function in order to distinguish among patients; this is one of the data gaps the Subcommittee expects to focus on.
Dr. Newacheck asked if Dr. Liu felt data gaps can be filled by modifying existing data systems and tools, or whether new ones are needed. Dr. Liu responded that new surveys with adequate "Ns" are desirable, but not likely, because of their cost. He is not sure whether additional data can be gotten through claims. Better assessment instruments may be the way to go for information on disability. Asked how he would use additional resources if they existed, he referred again to the assessment instruments. He also supported continued data collection on the MCBS.
Dr. Kane commented that the MCBS is a wonderful tool, but not a source of information on post-hospital care, for which real-time events are needed. He called for building enriched data collection devices into an ongoing system.
Dr. Mor pointed out that in data used to compare different sites, the intervening differential risk of mortality as a function of the place would affect the interpretation of transition rates.
Dr. Kramer is starting a study funded by ASPE to look at the quality indicators needed to measure post-acute care outcomes, how to collect those indicators, and what is available.
He reviewed several key issues in post-acute care:
Dr. Kramer then turned to the implications of the above for information needs, highlighting these issues:
Dr. Kramer then described his current research project, looking at quality indicators in terms of four conditions. They are developing measures related to some 120 indicators.
Dr. Iezzoni asked him if he thought HCFA's data systems were amenable to the comprehensive quality measurement strategy he is developing. Dr. Kramer answered, "No." He further indicated that some elements are present in the existing instruments, but others are not. For example, the absence of self-report is "a real problem." He expressed hope, however, that things would come together. Another dilemma stems from having multiple sources. His goal is to develop something akin to an SF-36 instrument for post-acute care, including an identification of "self report". He indicated that his project would end with a specific recommendation for instrumentation. The first phase ends in October.
Dr. Starfield noted the Committee's work on standards and its interest in the Healthy People 2010 goals regarding equity and eliminating disparities. She asked whether the panelists regard the issue of disparities as part of their work. Dr. Kramer said his group has looked at these issues, however, the samples are too small to differentiate among ethnic groups in terms of outcomes. Dr. Kane pointed to the need for sampling strategies to yield sufficient numbers in target groups. These variables are included in their analyses but race has not emerged as significantly affecting outcomes in post-acute care; however, this may be a function of the study design.
Mr. Webb is President and CEO of Village Care of New York, a community-based not-for-profit that provides clinical services for people with AIDS. He said that his experience bears out the analyses of the previous speakers: "There is no system," and the hospital data on individuals being discharged are not usable. Moreover, there is great competition among institutions for post-acute care clients. Managed care has disrupted any kind of system development. A significant percentage of the patients in his system are having astonishing turnarounds thanks to new drug research. This has shifted the relative proportions of services devoted to end of life care, nursing home and home health care, and so on.
Village Care is building a database using the format and intent of the MDS and OASIS. There are no standardized quality indicators for the MDS, which is problematic. Dr. Webb described the challenges of using their own data to try and manage quality in a competitive environment. One of their goals is to reduce rehospitalizations, and another is to improve functional status and independence; and they are trying to use MDS to manage these outcomes although it is very weak in the mental health area.
The general idea is to "manage across time, place and discipline," which they are finding very difficult, with few models available.
Village Care has "patched together a system of care" enabling them to vary the intensity of support and provide a variety of services in a variety of settings. All this is done in an environment in which "hospitals still dominate the landscape" and "drive the definition of health care." Village Care is working with hospitals and educating primary care practitioners to promote quality of care in a community setting.
Mr. Webb commended the Subcommittee for addressing "one of the more important issues that Medicare needs to face," with the goal of systems and data that are driven by patient needs and care costs.
Asked about data availability, Mr. Webb said he has no confidence that any data are available relating structural characteristics to clinical outcomes. Dr. Kane observed that methodologically, the problem is the use of large systematic databases. The issues that need to be addressed, such as the three-hour rule and nurse delegation, cannot be easily pursued through extant data sets; more innovative programs are needed. Dr. Kramer observed that the biggest dilemma is the way staffing data are collected across facilities, making it difficult to disaggregate to the individual level.
Dr. Mor asked Mr. Webb and Dr. Kramer about their success in trying to work with more than one dataset or setting. Mr. Webb said the data item typically does not mean the same thing in two different settings. Dr. Kramer agreed, adding that the greatest dilemma comes in choosing which data items are best for measuring quality. Dr. Mor said he was trying to get at the demand context, which differs from one setting to another. Dr. Kane commented on the number of factors that affect measurement; he advocated doing something to create a consistent measure, without trying, at least initially, to "capture each nuance of each respondent."
The meeting resumed after lunch with a comment from Gene Clark of Beverly Enterprises on the reason for the dramatic rise in the cost of nursing homes in 1990. He attributed it to the implementation of the OBRA statute that required providers to meet the highest practicable level of functioning for residents.
After noting that NCVHS member Dr. Paul Newacheck is a leading expert on the epidemiology of children who need long-term care, Dr. Ireys began by discussing how that population is defined -- an important issue in the field. One approach has been based on diagnoses or ICD-9 codes; the drawbacks are that it is difficult to encompass some 200 chronic serious health conditions, and that this approach counts conditions, not children. Another one, which focuses on children with activity limitations, has been used frequently, producing trend data. Its drawback is that it is a narrow focus that does not encompass children with many serious health needs. Recently, the field has come together around a non-categorical functional approach to definition, including behavioral, developmental, emotional or physical conditions as well as conditions that lead to dependency on compensatory mechanisms (e.g., insulin pumps).
A major source of epidemiologic data is the 1994-95 Childhood Disability Supplement of the NHIS. Although it is "rapidly aging," due to the changes in the health care system since then, it provides essential information on this population for the first time ever. Other data sources are primarily local area and program-based surveys. There are still many data gaps.
Regarding population characteristics, 15 to 18 percent of all children have a special health need or a chronic condition. Six to seven percent have an activity limitation, and two or three percent have functionally severe conditions. He indicated that a chart book funded by the Maternal and Child Health Bureau will be ready in early 2000, with data from the NHIS child disability supplement. In contrast with adults, where some 10 conditions account for most long-term care needs, children are affected by more than 200 chronic health conditions, most of which are rare (except asthma).
Other characteristics of this population:
Dr. Ireys highlighted several data needs in this area:
In conclusion, he stressed the need for better data in order to reduce the risk for poor outcomes; to create a more coherent and coordinated system than the one that has grown up largely in response to advocacy efforts; and to give children in this population a better chance at happiness and quality of life.
Dr. Newacheck asked about prospective payment for children in managed care, the status of risk adjustment techniques, and the need for more data to do capitation appropriately. Dr. Ireys said Hopkins is trying to evaluate risk adjustment methods in terms of their ability to predict costs for children. The results are expected soon. Also, NACHRI has developed a risk adjustment system that will be ready within a year. Thus, understanding is beginning to emerge about how to do risk adjustment for children; however, getting managed care organizations to adopt these procedures may be a long way off.
Dr. Starfield commented on the commonalities in data issues between children and adults with special needs and the potential for forming alliances. She noted the heterogeneity and amount of comorbidity among adults as well as children, and the common focus on the continuum of care. Dr. Ireys agreed that the similarities are important; one common issue is the challenge of gathering data from multiple service systems and sites. A special issue with children is the importance of getting data from the school system, where many services are provided.
Dr. Iezzoni asked how well ICD-9-CM reflects the diagnoses of children, and whether an ADL-type scale for a child would be different from that for an adult. On the second question, Dr. Ireys noted that the age of a child relates to very different expectations, so functional measures in a data set would have to be different as a function of development. The NHIS Disability Supplement represents the state of the art, "and even that has problems." Regarding ICD-9-CM, he said the codes do not wholly capture this population; providers have to "haggle" with Medicaid or MCOs when no code exists or the diagnostic process is still underway.
Recalling the earlier reference to the insulin pump, Dr. Iezzoni noted the importance of including assistive devices and technologies in the thinking about people with disabilities. Good ways are needed to capture information on the ways these devices affect people's ability to function.
Dr. Handler called attention to the work of Dr. George Brenneman of the American Indian Health Project at Hopkins. He is developing longitudinal data on American Indian children with disabilities.
This HCFA Office has had the lead in developing clinical measures and information systems on post-acute care and long-term care. Ms. Nonemaker asserted that there is less of a "vast wasteland" in post-acute care data than people may think. She underscored Dr. Kane's recommendations that the field move ahead and not be paralyzed by trying to create perfect measures and systems. Her Office is guided by many of the themes outlined in this meeting, such as using common data elements and using a patient-centered approach. They are working to build standardized, beneficiary-centered data elements to help consistently measure a beneficiary's clinical characteristics and care needs. These measures would then be used in a regulatory quality monitoring system to rate and compare providers on their performance in key areas, and to inform consumers on their choices.
Ms. Nonemaker expressed agreement with other presenters' comments about the great variation among providers and settings and the need to be able to compare across settings. Largely as a result of the Balanced Budget Act, HCFA is establishing provider-specific payment systems as a short-term step toward their long-term goals. This is part of a 10- to 20-year plan for an integrative post-acute care system, which aims to cap costs without simply shifting to another setting. They are moving toward building a core data set or common assessment instrument to be used across settings. It will be necessary to collect data for several years to understand how such a system would be configured and the rates derived.
The SNF prospective payment system began in July 1998 and will be phased in over the next three years. Progress is being made on the home care front. HCFA is getting ready to implement the OASIS data set. On the rehab front, they expect to use an MDS-based approach although no final decisions have been made on this. The objective is to have the same measures of clinical characteristics and care needs in both MDS and OASIS, to serve as the foundation for future payment and quality monitoring systems.
Ms. Nonemaker discussed MDS. It is designed to have information come through states to HCFA for all residents in Medicare- and Medicaid-certified facilities. This will create longitudinal records for individuals. The information will be used for payment, quality, and research. The information infrastructure has a standardized hardware platform in each state, which will be able to generate state-level quality indicators. The same infrastructure could collect information from various provider settings. The regulations for MDS transmission went into effect June 22. HCFA is now building a national repository.
Ms. Nonemaker said HCFA's instrument development study makes some assumptions about substitutability of care. Measurement systems need to understand the individuals going into particular types of care. She described a study of 1200 patients in rehab hospitals, long-term-care hospitals and nursing homes (SNFs), using the same kinds of measures. This study found similarities are between the patients in the SNFs and rehab hospitals. This is the first time this comparison has been made, and HCFA is just beginning to study the data. She indicated that they hope to build measures for the understudied area of palliative and terminal care, tailored to patient characteristics rather than settings.
In regard to quality measurement, HCFA has begun a project to develop and test indicators for quality of life in nursing homes. In another project, it will develop and validate measures and indicators for quality and appropriateness of services in post-acute care and long-term care. Dr. Mor is a principal investigator on that project, which will result in recommendations to HCFA about measures that are ready for testing. A major task is to develop markers for the post-acute population.
A related project will look at integrating quality indicators into the long-term care survey process. Ms. Nonemaker predicted a much more objective and standardized use of data in the regulatory process within five years. Information will also be distributed to providers for their own QA and QI activities, and to consumers. HCFA's own value-based purchasing is another potential application.
These quality indicators and markers are to be applied based on characteristics of the patient and population, not the care setting. They will be developed from data obtained from a variety of data sources. Ms. Nonemaker said she agreed with others about the importance of self-reported data. Dr. Mor's contract includes work on how to get the information, ensure that it is reliable, and operationalize it in to the Medicare program.
Dr. Newacheck asked whether any of the work described above applies to the roughly 100,000 children in institutions. Ms. Nonemaker said HCFA requires facilities to do the MDS, and there are questions about its appropriateness for children. The agency is now starting an analysis of the under-65 nursing home population that will look separately at residents under age 18. The goal is to develop a pediatric version of the MDS, if one seems needed. She and Dr. Newacheck noted that there are no data on the number of children in these facilities. HCFA will be working on these issues.
Ms. Coltin discussed with Ms. Nonemaker issues related to the time periods for assessing post-acute care patients.
Ms. Nonemaker mentioned that the uniform needs assessment instrument is seen as the vehicle for achieving more commonality between OASIS and MDS.
Dr. Iezzoni commented on the issue of reliability and credibility with respect to all these data, and on the many "opportunities for subjectivity" in the assessment instruments. She asked how HCFA would make sure the data are accurate, reliable and valid. Ms. Nonemaker said elements that prove not to be reliable will be improved or discarded. MDS data have had "good, if not very good" levels of reliability. HCFA is starting to develop a system to look at the accuracy of MDS data. Dr. Iezzoni observed that the standard way of testing reliability will not work in this context. Ms. Nonemaker mentioned other ways to check reliability, and stressed that "this is not a done deal."
Dr. Zimmerman gave more detail on developments with using the MDS for quality assurance in the nursing home setting. He observed that any changes needed in the model for other settings need to happen quickly. He described the instrument, which has 24 quality indicators covering 11 domains of care. The major concern about them is the limited extent to which they cover quality of life. There are outcome and process indicators and some that combine the two.
Turning to the use of these indicators for internal quality improvement and external quality assurance, Dr. Zimmerman said they are used to generate facility reports and quality indicator profiles comparing facilities with others in a peer group. There also is a series of resident reports. He showed examples of a facility report and a resident report. His group has developed and is testing software that by mid-1999 will make identical data available via the Internet to all survey agencies and facilities in the U.S. Currently, some providers have this information and others do not.
The issues involved in using this information for quality purposes are similar in various post-acute care settings. One of the largest relates to differences in the way the indicators and the data are to be used -- for immediate decisions or for subsequent review and quality improvement. This goes to the heart of the reliability and validity issues raised by Dr. Iezzoni. Some managed care organizations are prematurely rushing to judgment about providers on the basis of one set of indicators, rather than to using them to target next steps for improvement.
Methods and policy issues include the question of how to do risk adjustment and how to make appropriate comparisons. Dr. Zimmerman cautioned against over-adjusting for risk and adjusting out the very factors that bear on quality of care -- e.g., if someone is at higher risk because of previous care issues or a failure to intervene in time. Another issue concerns attribution: whether the condition developed in the institution, and even if it did, whether it may be due to poor care. The appropriateness of standard and threshold levels is also important; another issue relates to "the rush to aggregation" and the relationship between quality measures. Any attempt to rate across dimensions has to be done very carefully.
Dr. Zimmerman speculated that the measures in long-term care have limited applicability to post-acute care. Assessments must take into account preexisting chronic impairments. Dr. Zimmerman discussed his use of the term "post-acute" (meaning, after a period of intensive care, in or out of the hospital) and agreed that the term is used in a host of ways, some of which involve no prior acute or hospital care.
Dr. Iezzoni underscored Dr. Zimmerman's point that knowing the timing of events is critical and must be part of whatever data set is developed. Dr. Zimmerman commented on the difficulty of capturing this information. Dr. Iezzoni observed that the electronic medical record may be the only solution.
At Dr. Iezzoni's request, Dr. Zimmerman explained that he does not like the term "quality of life" because it is typically treated as distinct from quality of care, rather than as an overarching dimension that encompasses a person's clinical, medical, and functional status as well as social and other factors.
Ms. Coltin commented on the positive reasons, in the short run, for prompt action by MCOs to narrow their networks in response to quality indicators. In the long run, they will want to develop consistent partnerships for quality improvement. Dr. Zimmerman "hesitantly agreed" with this interpretation but cautioned against leaping to conclusions using data with limited predictive validity. Ms. Coltin pointed out that the field now has several sources of information on which to make sound decisions.
IHS is a for-profit health care company with long-term care facilities, home care locations, hospices, and long-term acute care hospitals around the country, and a single data system. He showed a patient-centered data model that includes clinical indicators, demographics, risk factors, treatment plan, regulatory backdrop, health status, and social environment, among other factors. The company is shifting to an MDS-based system for measuring health status. Dr. Kilgore offered recommendations that essentially support having uniform measures and consistency across sites.
Dr. Hargraves is a sociologist who talks with patients about what happens to them during care. With little time left for his presentation, he stressed the importance of talking to patients in order to find out what is important to them. His other major point is the importance of having standardized survey instruments to permit following people and making comparisons over time. In talking with patients, he and colleagues have gotten their permission to link to administrative data and discharge data.
He noted that subgroup analysis is important -- e.g., only heart attack or hip replacement patients.
In response to a question from Dr. Mor, Dr. Hargraves said recall is a key factor in patient surveys. He described some of their approaches. For example, they do not ask people very specific questions they can't answer, such as regarding time-related events. They found the consistency of seeing the same provider to be very important, and something people did remember. To another question, he said that patient satisfaction surveys tend to be "too fuzzy" because "everybody is satisfied." He and his colleagues elicit details about events and look for the criteria patients use to judge quality. Dr. Hargraves then commented on how they handle patients who are unable to respond. One tip is to be sure to use the same proxy for re-interviews.
In response to a question from Dr. Liu, Dr. Kilgore said IHS has not done a dynamic analysis of patient flow and optimal care levels as patients go through the system.
Dr. Iezzoni then thanked the presenters and adjourned the meeting, saying the Subcommittee would discuss its findings on February 2.
I hereby certify that, to the best of my knowledge, the foregoing summary of minutes is accurate and complete.
_______________________________________________
Chair Date