Review of the Measures

Category 1: Physical Health
Category 2: Mental Health
Category 3: Meeting Basic Daily Needs
Category 4: Getting Around
Category 5: Meeting Household Needs
Category 6: Talking with People
Stabilization Measures (Categories 7-12)
Category 13: Utilization Measures

Category 1: Physical Health

1. Improvement in Dyspnea

Dr. Gerteis explained that all three groups of respondents questioned whether an agency could have an influence on dyspnea. Debra Saliba, M.D., M.P.H., suspected that many physicians are not used to working in multidisciplinary teams. As a clinician who works in such teams, Dr. Saliba knows of several approaches that could make a significant difference to this characteristic.

Dr. Golden pointed out that the standard deviation for this measure is not large. He wondered whether agencies perceive dyspnea as something they cannot affect. Dr. Shaughnessy explained that when the mean is around 50 percent, as it is for this item, the standard deviation is likely to be low. Dr. Shaughnessy cautioned against overly emphasizing the standard deviation because it is, in part, an artifact. Dr. Bernard added that the variability between the 25th and 75th percentiles is at least 16 percent, and some variability between agencies is desirable.

Dr. Fitzgerald requested additional clarification on the standard deviation of the risk adjusted measure compared to the measure without risk adjustment. Dr. Hittle explained that agency-level standard deviation usually, but not always, goes down. Risk factors can be distributed in widely different ways across agencies, so risk-adjusting can increase or decrease agency-level variation.

Ms. Fredland noted that home health agencies are more likely to stabilize than to improve dyspnea. Her agency has undertaken a major initiative to improve dyspnea outcomes but with little success so far. The OASIS question does not separate out patients who are only short of breath at night. Home health agencies can address that.

Pamela Teenier asked about the risk factors for dyspnea. Dr. Bernard pointed out that these do not have much effect, as the adjusted and unadjusted rates are not very different. Dr. Shaughnessy agreed that this was true in the aggregate, but risk factors can make a significant difference for individual agencies.

Dr. Bernard said that the way in which dyspnea reflects pulmonary disease in this country makes it very important. Yet she wondered how much agencies can influence this for the better. With chronic diseases, when an agency lengthens the time it takes for someone to get worse, the agency is doing well. This is difficult to capture with these measures. Dyspnea is a very important indicator because of the prevalence of the problem and the number of cases that home health agencies see. Dr. Bernard said she probably would not select this measure for CMS public reporting.

Ms. Fredland predicted that when the public and others see these measures, they will not understand why agencies cannot improve. For this reason, improvement in dyspnea is not a valid measure to present to the public. Stabilizing dyspnea would be appropriate for public reporting.

Ms. Ketcham pointed out that 36 percent of agencies chose this measure and 56 percent of those demonstrated improvement. Mary Nguyen, R.N., found it difficult to answer this question for patients with minimal or moderate dyspnea at rest. These patients have a chronic condition, so they are not moderate on admission nor minimal on discharge. They are always short of breath.

Dr. Golden said that dyspnea is a common symptom that probably needs more attention. Dr. Saliba said it could be improved in several ways, and should be included in the NHQR. However, Ms. Fredland argued that this measure should not be used for CMS public reporting.

Dr. Saliba wondered whether the 58.8 percent of agencies that succeeded in improving this outcome did so through reclassification. Ms. Crisler replied that the OASIS manual had a chapter on data accuracy, which is critical for any data-driven system.

Ms. Terry said that recoding occurs when, as part of quality improvement, staff members are taught to understand the questions better while the dyspnea measure does not become more amenable to changes. Dr. Golden indicated that in the first year of these measures, some recoding will occur, but this will stabilize over time.

Dr. Bernard believes that the evaluation for OASIS measures is more subjective than for HEDIS measures, but the measures will stabilize over time.

Dr. Fitzgerald suggested that the logical follow-up to actionability is responsiveness, or whether an outcome improves as a result of a change in behavior by the agency. For dyspnea, 58 percent of agencies actually improved the outcome, which indicates that this was an actionable change that produced a responsive reflection in the indicator. But some panel members believed that the changes produced reflected differences in how the data were collected, rather than responses to process changes. Dr. Fitzgerald is seeking indicators that agencies can actually change and used to hold agencies accountable.

Ms. Ketcham suggested that distinguishing between short- and long-term homecare patients made sense for these measures. Agencies can see an improvement in patients discharged from the hospital with shortness of breath, but this is less likely to occur in the long-term care population.

Ms. Clark noted that a regulatory requirement for data accuracy exists, and this is probably one of the most common citations for agencies. Regulators determine data accuracy from clinical record reviews and find that data are not always accurate for every question. A learning curve is involved. Although the OASIS data collection has been ongoing for 3-4 years, agencies and clinical staff are still learning how to use these data. This will be the case for every one of these items.

Dr. Saliba asked whether the agencies that participated in the pilot were taught how to use OASIS. If so, some of the success rate is less likely to reflect reclassification, because so much effort was made to ensure that the classification was correct. Ms. Clark replied that agencies have had ongoing intensive instruction in OASIS data collection and accuracy of data. New staff members have less experience, and others are slow to change their habits.

Ms. Terry said that in large agencies that have significant turnover and several hundred people collecting data, training is ongoing. Also, nurses admit patients, but therapists discharge them. As a result, the coding issue is always present. Ms. Fredland added that teaching nurses to think in the terms used in OASIS is difficult, because these are not necessarily the terms they typically use. Nurses do not think of a lesion as a rash, for example.

2. Improvement in Status of Surgical Wounds
3. Improvement in Number of Surgical Wounds

The panel members considered most of the remaining measures in groups. Dr. Shaughnessy pointed out that the second paper under Tab 7 of the background notebook listed all of the risk factors tested for every model.

Overview of Risk Adjustment and Outcome Measures: Download in PDF format (1.5 MB) PDF Help

Dr. Golden pointed out that these measures are not risk-adjusted, yet the outcome measure properties sheets provide risk-adjusted numbers. Dr. Shaughnessy explained that these figures refer to risk adjustment that has been done to date but was not included in the CMS agency-level reports because it was not ready for publication.

Ms. Terry stated that surgical wounds have been difficult for homecare to address. Agency staff members do not usually look at wounds when they admit a patient, because they usually improve on their own. If a wound is healed at discharge, agencies are supposed to indicate that the patient had no wound and skip the question. This OASIS item has been difficult for agencies to answer correctly. Stabilization is easier, because improvement does not always occur.

Ms. Scott pointed to a difficulty with the number of surgical wounds. Sometimes a large open wound will close partially, becoming five small wounds instead of one large wound. This would be coded as a decline, even though the patient improved clinically. This is an important measure, but the current design of the questions leads to data that are not acceptable. Ms. Fredland added that physicians often remove the bad tissue from wounds, making them appear worse, but this is needed for healing. This question is flawed because of how it is worded. If it were worded differently, it would be appropriate for CMS public reporting.

Dr. McGee pointed out that improvement in the number of wounds was referred to as healing well after an operation for consumers. Dr. Gerteis explained that consumers did not understand number of wounds at all but did understand healing well after an operation.

Dr. Bernard stated that of the agencies that selected improvement in number of surgical wounds, 87.5 percent saw an improvement, and this measure had a reasonable amount of variability. Dr. Fitzgerald was concerned that these indicators are not risk-adjusted. Where risk-adjusted measures are available, it is difficult to defend using non-risk-adjusted measures in CMS public reporting. Dr. Shaughnessy and his colleagues are working on risk adjustment, but this item has caused much difficulty. The researchers have received input about important risk factors, but they are usually restricted to individual patients. It may be that the way that these wounds are treated and cared for is the dominant explanatory factor in outcomes. In addition, great variation exists in wound care among home health agencies. Dr. Golden suggested that nutritional status might be a key factor, but Dr. Shaughnessy did not have those data. Dr. Shaughnessy pointed out that agencies that selected this measure did make an improvement.

Mr. Lindberg requested clarification on the patients for whom the measure was calculated. Dr. Hittle replied that standard exclusions were used. Wound status is not collected on patients whose episodes end with inpatient facility or death. Patients who are nonresponsive at the start of care are excluded from all end-result outcomes, because they confound the risk-adjustment process too much. They are very different from all others whom the home health agency treats. Nonresponsive patients represent a very small percentage of home health patients.

Ms. Terry noted that the questions regarding the status of surgical wounds have been difficult to answer. Moreover, this measure is not risk-adjusted, and Ms. Terry has concerns about such items. If the measures are not risk-adjusted, agencies might do poorly on them, because hospitals are discharging sicker patients. Dr. Bartlett pointed out that 13 items are not risk-adjusted because national data are not currently available to do this adequately. Risk adjustment might be possible later. Dr. Sangl pointed out that the form that panel members would complete has a section for comments that could indicate such concerns. Dr. Thoumaian added that panel members could indicate that they would select the measure for CMS public reporting if it were risk-adjusted.

Mr. Fitzgerald did not believe the wound status measure would be valid for CMS, because it is not risk adjusted. This measure should be in NHQR, which might lead to a research priority in this area.

Dr. Bernard agreed that the measure would be appropriate for AHRQ but not CMS, because consumers found this to be relevant and it is a reason for seeking home healthcare after an acute episode that includes surgery. Dr. Bernard did not believe risk adjustment would make a significant difference at the state or national level, although it might be used for comparing different agencies.

Dr. Murtaugh characterized this measure as actionable and an area in which home healthcare plays an important role. However, questions about how well it is measured have been raised. Consumers regard home health agencies as places where postsurgical wounds are treated.

Ms. Nguyen noted that a component of this and all wound questions is out of the hands of the agency, because agencies are often at the mercy of what the physician does or does not know about wound care and is willing to order. Surgeons are not always willing to try different approaches.

Ms. Cronin suggested adding this measure to a "developmental" list. She was struck by the comments about how the question is asked, and suggested considering how to ask the question in ways that could be publicly reported for consumers. When these data collection points were designed, public consumer reporting was not considered.

Dr. Golden suggested that agencies might refuse patients if this measure is not risk-adjusted, because they would be afraid of the impact of including such patients in their report cards. Ms. Fredland agreed that patients sometimes do not receive care because agencies cannot afford to keep them or because the risk involved is too great. Ms. Scott's agency would not change its admission criteria due to such concerns, but others might.

Ms. Terry argued that this item is important. Changing the wording would make it easier for agencies and produce better data. Agencies are selecting patients based on finances and, when these measures come out, they might refuse to accept wound-care patients because doing so would cause them to lose money.

Dr. Fitzgerald pointed out that the providers trying to improve their care will continue to receive the 41-indicator OBQI set, whether or not data on surgical wound healing are publicly reported. Dr. Shaughnessy suggested that this measure could be risk-adjusted as well as it is possible to do now, which may be preferable to not risk adjusting at all. CMS implemented OBQI only recently, and CMS formally entertains suggestions to improve OASIS items. Dr. Shaughnessy expects that all of this will improve over time.

4. Improvement in Urinary Tract Infection

Ms. Ketcham pointed out that treating urinary tract infections (UTIs) reduces incontinence, which is a key problem for family caregivers. Dr. Shaughnessy noted that the number of patients on whom this measure is based is quite small, 54,000 out of one million, or about 5 percent. Dr. McGee added that the measure has almost no variation in the interquartile range, which, when added to the small size of the base, would seem to eliminate it.

Dr. Gerteis said consumers thought that treating UTIs was more important for how patients are treated medically. Dr. McGee asked about consumer and policymaker perceptions of the ability of home healthcare agencies to have an impact on incontinence. Ms. Scott believed that this creates an opportunity for providers. Discharge planners and physicians recognize that agencies can have an impact on wounds, because they are visible, but they do not know how much agencies can affect activities of daily living (ADLs). Ms. Scott supported including such measures to help move this agenda forward. Dr. Bernard supported the inclusion of incontinence, not UTIs.

5. Improvement in Urinary Incontinence

Mr. Fitzgerald wondered why such a small number of agencies selected this measure. Dr. Shaughnessy explained that when agencies address incontinence, they can make a difference, although most agencies do not believe that they can have an impact.

Ms. Fredland's agency does a great deal of teaching about bladder training and has reduced urinary incontinence to a low level. Nevertheless, it is more difficult to address with Alzheimer's patients and others with dementia. Ms. Fredland favors this measure because of the results seen at her agency and because addressing incontinence makes a significant difference to a patient's life. The incontinent patient is often embarrassed and reluctant to leave the home.

Estimates indicate that about 20 percent of the homecare population is incontinent. Ms. Cronin pointed out that incontinence is not a Nursing Home Compare measure. Dr. Golden noted that the presence of Foley catheters is a Nursing Home Compare measure, but Dr. Shaughnessy explained that it is a process measure. Ms. Cronin supported the inclusion of this measure but was concerned about whether this issue resonates with the public for CMS public reporting. Dr. Gerteis explained that consumers thought that incontinence is a natural degenerative process, and there is not much agencies can do about it. Dr. Bernard pointed out that incontinence must be addressed to help keep homecare patients out of nursing homes. Ms. Teenier added that addressing incontinence can have an impact on overall health.

Ms. Scott suggested adding incontinence to the developmental list, because the only options in the incontinence question are whether the patient is or is not incontinent. Some patients who are coping successfully probably fall somewhere between these extremes.

Dr. Hittle explained that this measure is calculated based on a composite of two different items, the urinary incontinence item and when the incontinence occurs. Those who are removed from the catheter receive credit for improving, as do those who move from being always incontinent to timed voiding.

Dr. Saliba asked whether someone who has a catheter because of an enlarged prostate and is not a candidate for surgery will be considered a treatment failure. Dr. Hittle replied that this measure is risk-adjusted, and the risk adjustment should account for such situations. Dr. Golden pointed out that, given the scale, someone who is incontinent and is put on a catheter would be considered a treatment failure, because he or she would be moving the wrong way on the scale. Dr. Hittle explained that this is based on the value decision that having a catheter is worse. Ms. Teenier agreed that consumers are likely to seek more independence, and having a catheter makes the patient more dependent.

6. Improvement in Bowel Incontinence

Ms. Cronin pointed to the low number of patients affected. According to Dr. Hittle, eligible patients represent approximately 7 percent of the population of Medicare and Medicaid patients.

Dr. McGee noted that urinary and bowel incontinence are correlated, so there is some redundancy and urinary continence appeared to be a good measure for both AHRQ and CMS reporting. Dr. Shaughnessy pointed out that bowel incontinence risk adjusts a little better but this issue might be overridden by sample size and other factors.

Dr. McGee asked for clarification on better risk-adjusting. Dr. Shaughnessy explained that the more factors in the risk model, the more meaningful it is for clinicians and policymakers, because the researchers took into consideration a variety of factors that can also influence the outcome. Another question to consider in selecting measures is how much variation in the outcome risk factors they explain. Bowel incontinence has a slightly but not significantly higher variation explained than that of urinary incontinence. The clinical arguments should outweigh the statistical arguments.

Dr. McGee assumed that the researchers tested all of the risk-adjustment variables available and there would be natural variation in the degree to which various measures would require risk adjustment to be equitable. As a result, having many variables in the risk-adjustment model does not make a measure more credible. This is strictly a statistical interpretation that does not go far enough, said Dr. McGee, unless some measures have certain risk factors that clinicians can identify that have not been available for risk adjustment. From a practical standpoint, applying the risk-adjustment models often makes little difference, and people tend to overestimate the degree to which factors make a difference. Dr. Shaughnessy agreed. Statisticians sometimes set an artificial goal to try to include as many risk factors as possible in the model. It is better to do as well as one can with the known risk factors to determine whether care is making a difference.

Ms. Fredland pointed out that most patients with bowel incontinence have Alzheimer's disease, dementia, or neurological deficits that make their incontinence difficult to address. Although there is much that agencies can do for urinary incontinence, they can do little for bowel incontinence, unless the patient has colitis, in which case a physician must provide treatment.

Dr. Bartlett asked about the possibility of developing composite scores for any of the measures considered so far. Dr. Shaughnessy replied that he and his colleagues have experimented with ways to develop subgroups of these 41 measures. For the physiologic category, they determined whether all of these measures (except one of the wound measures) could be combined to produce a composite measure that provides useful information about the physiological care provided by a home health agency. The researchers developed a composite measure for each of the 13 categories based on the average across all measures. Dr. Golden noted that, in other similar meetings, composite measures have generated mixed responses for technical reasons. It is difficult for the panel members to endorse composite measures now without field testing. Dr. Shaughnessy agreed and explained that the composite measures were developed simply to show that it can be done. These are not ready for release.

Dr. McGee agreed that major reporting challenges with composites exist. These measures require a background explanation and very clear and direct statements to consumers and policymakers about what the home health agency can do to affect the outcome. It is almost impossible to do this with composite measures. Composite measures would not necessarily make it easier for people to understand a report and would yield a global statement likely to be misinterpreted, which is not fair to the agencies. Dr. McGee suggested that the panel focus on the individual measures, which are easier to explain. Dr. Shaughnessy agreed that these composite measures are not appropriate for CMS public reporting. However, home health agency staff have often requested something more general than 41 measures to provide a sense of how they are doing overall before selecting more specific measure for quality improvement. This type of information might be valuable for providers.

Dr. Murtaugh suggested that if consumers see that no agency's results are in the same direction on every indicator, they will not obtain an overall sense of whether a given agency is good or not. Dr. Shaughnessy pointed out that this can happen even with eight measures.

Top of Page

Category 2: Mental Health

7. Improvement in Behavior Problem Frequency
8. Improvement in Cognitive Functioning
9. Improvement in Confusion Frequency
10. Improvement in Anxiety Level

Dr. Shaughnessy noted that these four measures have a low risk adjustment R2. Also, the subpopulation to which they pertain is small. Most of these measures were not risk-adjusted at the national level.

Ms. Terry argued that many of the mental health measures affect a small number of patients and are difficult for home healthcare agencies to address. Mental health problems usually are not the main reason why a patient is receiving home healthcare. Very few services are psychiatric, and these issues may not be amenable to treatment. Most of the measures in this category were not high on Ms. Terry's list for either the NHQR or CMS public reporting.

Dr. Bernard supported the confusion measure, which has physiologic implications. If an agency can stabilize a patient after discharge from the hospital, cognitive function often improves as confusion dissipates. Stabilizing the patient's underlying problems helps reduce confusion. Dr. Saliba distinguished delirium from dementia. Organic dementia, such as Alzheimer's disease, is usually a deteriorating disease. Ms. Nguyen was concerned about all of the mental health measures for agency-to-agency comparisons. Yet the confusion measure might be valuable for the NHQR.

Ms. Clark pointed out that confusion can be related to improper medication administration and many other issues related to general improvement in patient care. Confusion is not just a mental process. Dr. Bernard assumed that the risk model addresses dementia and underlying Alzheimer's disease.

Dr. Saliba pointed out that patients at highest risk for delirium are those with cognitive impairment, who are most likely to be tipped over by an infection or improper treatment. These patients should not be excluded from the measure.

Ms. McCall said that more drug therapy options to treat anxiety are now available. Such treatment appears to be the only measure for which a homecare nurse could suggest to a physician that drug therapy might be appropriate. Ms. Fredland regarded anxiety as the most promising of the measures, because homecare agencies can address it effectively through breathing exercises. Patients often have great anxiety because they do not understand their medications, and agencies can reduce anxiety by teaching about medications. Ms. Fredland said the anxiety measure would probably make both lists, except it is not risk-adjusted. Dr. Shaughnessy said that he and his colleagues are working on this now. The underlying reliability is 0.612, which just makes the cut of 0.6. This measure probably will be improved and will then lend itself better to risk adjustment and quality applications. Dr. Shaughnessy was not certain how much more variance will be explained as a result.

Ms. Teenier explained that when staff members perform assessments on behavior questions, they must sometimes rely on someone else to tell them whether the patient is anxious. Patients themselves often deny having anxiety, and staff cannot determine whether anxiety is present through observation alone. Mr. Fitzgerald pointed out that this is reflected in this measure's low reliability.

Ms. Terry said that anxiety is difficult to identify and address. However, confusion is an appropriate measure for the NHQR. It is risk-adjusted and would be appropriate for CMS public reporting because it can be addressed by home health agencies. Dr. Gerteis noted that consumers and physicians believe that cognitive functioning deficits could be improved by home health agencies.

Ms. Scott was reluctant to make home healthcare agencies responsible for improving cognitive functioning, because it is difficult to assess and often difficult to improve in the timeframe in which home health agencies see patients (typically 20-40 days). Ms. Scott encouraged the NHQR to include measures of outcomes that affect many people and therefore deserve national attention, including issues that would be addressed from the social policy/resource perspective. Dr. McGee suggested that, in that case, it might be important to include at least one measure from this category. He asked whether cognitive functioning is the strongest one. Ms. Clark pointed out that, given the aging of the population, the population to which this applies will increase.

Ms. Scott said that the mental health issue that attracts her the most is depression, but it is not included in the measure set. Dr. Bernard pointed out that the sample size for confusion is larger than for any other mental health measure. Robin Remsburg, Ph.D., R.N., added that all of these items showed 100-percent improvement, which speaks well for being able to address them.

Top of Page

Category 3: Meeting Basic Daily Needs

11. Improvement in Eating
12. Improvement in Upper Body Dressing
13. Improvement in Lower Body Dressing
14. Improvement in Bathing
15. Improvement in Grooming
16. Improvement in Management of Oral Medications

Top of Page

Category 4: Getting Around

17. Improvement in Ambulation/Locomotion
18. Improvement in Toileting
19. Improvement in Transferring

Dr. Gerteis reported that all three groups of respondents thought that grooming and housekeeping were the least important of these measures and did not understand the distinction between upper and lower body dressing. Dr. Shaughnessy noted that this group of measures had better statistical properties than those of the previous categories.

Dr. Fitzgerald pointed out that improvement in eating has not been targeted. Dr. Shaughnessy said that this item also had the lowest sample size. He suggested choosing the most meaningful of the ADLs, which likely would not include improvement in eating.

Dr. Saliba argued that eating problems are not that important or mutable, but individuals who need help with eating typically have another ADL or instrumental ADL (IADL) dependency. Dr. Remsburg said that eating is the last ADL to be affected. Eating is important, but other ADLs such as ambulation and toileting probably affect a larger proportion of homecare patients. Ms. Scott agreed that those two measures are more important.

Ms. Fredland supported the inclusion of ambulation, toileting, transferring, and medication management for both the NHQR and CMS public reporting. The last measure is essential, because if patients do not take their medications correctly, they develop confusion and do not improve. Ambulation also leads to deterioration and must be included. Toileting difficulties lead to living in poor hygiene. Transferring is essential to moving at all. Eating affects too small a number of patients, and by the time a patient reaches the point of an eating problem, homecare agencies can do little to help.

Frances Petrella, R.N., would include bathing, because it is important for the beneficiaries of homecare to be as independent as possible. This area is closely associated with the four measures suggested by Ms. Fredland for both reporting purposes. Ms. Teenier agreed. The elderly population understands the goal of moving toward independence, which is what most agencies strive for.

Dr. Hittle noted that bathing, ambulation, and transfer apply to the largest percentage of the population, although none of these measures applies to a small percentage. The measure that really stands out in terms of its risk adjustment, size of the C-statistic, and R2, is toileting. The risk adjustment for transferring is not as robust but is still adequate. Dr. Shaughnessy found no statistical reason to exclude any of these measures.

Dr. Saliba pointed out that the measures supported by other panel members are not treated with occupational therapy except, perhaps, for transferring. By eliminating grooming, dressing, and other measures, the list excludes much of what occupational therapists address. Ms. Teenier suggested that grooming could be subsumed under dressing, and dressing could be added to the list.

Ms. Terry said that all of these measures appeared promising for both lists, but hesitated to include so many ADLs because they measure many similar kinds of improvements. If the list is limited to 12, it would be inappropriate for half of these to be ADLs. Ms. Terry suggested selecting a smaller number. Medication management is different from the others and should clearly be included. It is one of the major issues that homecare addresses. Toileting and ambulation are very important, and therapy addresses the important goal of getting patients to move about.

Ms. Ketcham thought that ambulation is important, because when it is improved, toileting and transferring also can improve. Consumers do not understand upper body dressing, but it correlates with everything and is probably the best measure.

Ms. Nguyen pointed out that the transferring and toileting measures do not indicate improvement in OASIS if the patient uses an assistive device. Similarly, if a patient is admitted on a rolling walker and is discharged with a quad cane, they have made an enormous improvement, but this is not reflected in the ambulation score. Dr. Hittle said that this issue is being raised with the OASIS Technical Expert Panel.

Dr. Bernard pointed out that measures 16 and 30 both address the management of oral medications. Dr. Hittle explained that measure 16 refers to improvement and measure 30 to stabilization, which means non-worsening or non-decline.

Dr. Hittle explained that one reason why the ambulation measure is calculated as Ms. Nguyen described is that previous reliability studies showed that clinicians had difficulty distinguishing between the 0 and 1 levels when assistive devices were used. A way to address this difficulty may be found in the future. Dr. Shaughnessy suggested that this problem should not preclude including this measure now, because the problem might be addressed in the future.

Dr. McGee said that toileting is important, and urinary incontinence covers one aspect of this issue. She supported the inclusion of upper and lower body dressing. Grooming enhances dignity and is also important, although consumers did not support its inclusion. Dr. McGee wondered whether upper and lower body dressing need to be separated for CMS public reporting and whether grooming could be included. Consumers are likely to regard all of these measures as belonging together. Dr. Shaughnessy noted that these outcome measures are highly related to each other.

Mr. Lindberg suggested that if these items could be grouped, consumers could be presented with a more manageable list of four or five items. Dr. Kosiak explained that AHRQ considered the use of composite measures across the board for hospitals and nursing homes and felt that more research and discussion were required. AHRQ is seeking advice from panel members about which measures to include. It will eventually determine whether measures can be combined. Dr. Thoumaian added that CMS's focus is on specific outcome measures for public reporting.

Mr. Lindberg said that the risk of choosing too many measures is that they will overwhelm other measures that panel members believe are more important. Panel members should simply select all of the measures that they consider appropriate, and CMS and AHRQ can then choose how many of those to use. Dr. Bartlett reminded panel members that they were asked to select their top 12 or so measures.

Mr. Lindberg supported the inclusion of upper body dressing, which is a good determinant of other deficits. Dr. Saliba would combine upper and lower body dressing. She would not add grooming, because it is so highly correlated with upper body dressing that if upper body dressing is included, grooming would probably be included. Dr. Shaughnessy suggested that panel members choose one of the existing measures, because the combined measure has not been field-tested.

Dr. Kelley suggested that if panel members believe that several measures go together, they should select the one that best represents that domain of quality. Panel members were not selecting home health quality measures for perpetuity. AHRQ would establish a process for reviewing its NHQR measures. Panel members should therefore share their opinions on issues that should be considered in the future.

Dr. Saliba retracted her earlier suggestion that upper and lower body be combined, because upper body dressing correlates better with some of the other measures that panel members suggested excluding.

Ms. Clark characterized upper and lower body dressing as two separate entities. Some patients can do one but not the other. Grooming is more closely aligned with bathing, according to the definition used for OASIS.

20. Improvement in Pain Interfering with Activity

Ms. Fredland argued that pain has a major impact on mobility and can be treated by homecare agencies. If this measure were risk-adjusted, she would include it.

Dr. Shaughnessy explained that the R² is approximately 7 percent, which is fairly low by almost any standard. The C-statistic is around 0.65, which also is low. Patients with pain are a very heterogeneous population, because there are so many reasons for pain. Dr. Shaughnessy and his colleagues recently examined stabilization, rather than improvement, of pain in orthopedic patients, and this did much better in terms of risk adjustment. The researchers are trying to redefine the subpopulation to which the measure pertains to identify a more homogenous population that will allow better risk adjustment. The researchers agree about the importance of this measure.

Dr. Fitzgerald pointed out that this measure is in line with nursing home measures. Ms. Nguyen added that it affects just about every other outcome.

Dr. Gerteis noted that consumers, doctors, and discharge planners did not find pain important, because they were not sure how agencies could address it. They believe that home health agencies address conditions that are visible and tangible, unlike pain.

Dr. Saliba viewed pain as important and pervasive and was not sure of the importance of risk adjustment. Knowing the etiology or risk factors helps home health agencies better treat pain. Pain would not be risk-adjusted for nursing homes either, so the pain measure would at least be comparable across sites. Dr. Shaughnessy noted that pain can be modestly risk-adjusted, but the risk adjustment is not as explanatory as it is with other measures, because the etiology of pain varies so much. Dr. Saliba suggested that clinical reasons for the risk adjustment should exist. Risk adjustment is needed for outcomes for which patients are at high risk and when agencies can do little to help. Adjustment will ensure that agencies are not given inappropriately low scores. Yet home health agencies should always address pain, whatever the cause or risk factor. Dr. Shaughnessy noted that many agencies have done this with some success.

Ms. Nguyen suggested including pain for the NHQR because it is still not treated well in the United States. Additional emphasis might improve the situation. Ms. Petrella suspected that pain is not well assessed.

Dr. Murtaugh suggested adding pain to the developmental list, because of the risk-adjustment issue. Experts on pain have suggested using a 0-10 scale for cancer, and he wondered if the same could be done in OASIS. Dr. Bernard noted that Ms. Scott's agency uses a 0-10 scale, but this is not reported, because it is not part of the OASIS database. Dr. Shaughnessy and his colleagues have explored this option.

Top of Page

Category 5: Meeting Household Needs

21. Improvement in Light Meal Preparation
22. Improvement in Laundry
23. Improvement in Shopping
24. Improvement in Housekeeping

Ms. Terry found the IADLs to be troublesome in homecare. Agencies are not really paid to address these, and observing them is almost impossible. For example, agency staff almost never observe a patient doing laundry. Their assessments are based on self-reports. ADLs are a better measure of homecare. Ms. Nguyen agreed and added that the ADLs correlate with the IADLs.

Ms. Fredland pointed out that patients do not always report on IADLs truthfully. If patients are able to do their own shopping, they are not homebound, which would disqualify them from Medicare homecare. Ms. Fredland did not support including the IADL measures.

Ms. Clark explained that an agency's assessment of the IADLs is based on a patient's ability to do the task in question and interventions needed to complete these tasks. These are vital for homecare overall because, although agencies do not address these tasks directly, they can help patients find persons to do these tasks for them. But the IADLs are not suitable for NHQR or CMS public reporting.

Dr. Golden asked for clarification on light meal preparation. Ms. Clark explained that this refers to a patient's ability to heat a meal in the microwave and to open the door to receive the meal. Confusion affects this ability, but this is a very low priority. If the other problems are addressed, light meal preparation will be addressed.

Mr. Lindberg thought these measures would be helpful if they could be used to evaluate how well an agency performs assessments to help an individual obtain the additional assistance they need to resolve these issues. But the current questions do not serve this purpose.

Dr. Shaughnessy noted that past consensus panels agreed that the IADLs were important. However, at that time, the researchers did not have some of the correlations in outcomes between ADLS and IADLs. Dr. Hittle added that the IADLs perform better statistically than do some of the mental health measures. Ms. Crisler suggested that the cognitive component of IADLs should not be overlooked. They require physical and cognitive abilities and are key in the ability to stay at home.

Mr. Lindberg said that housekeeping stands out in its relationship to mental health. But Ms. Teenier pointed out that males and females tend to give very different answers to the housekeeping questions. Also, separating the ability to do housekeeping from whether the patient actually does housekeeping can be very difficult. Ms. Crisler explained that this is why some items have a gender-based responses. Housekeeping involves more muscle mass. If improvement is seen in this area, an agency can tell that the patient is doing better.

Ms. Teenier pointed out that a dirty house might affect lung or wound status, but it has less impact on overall health than some other measures. Dr. Shaughnessy added that this measure also has lower reliability.

Ms. Clark pointed out that housekeeping is a culturally responsive item, because what may be a clean house to one person might not be to someone else. Patients often are unwilling to reduce their messiness, because they do not regard it as a problem. The clinician making the judgment brings a cultural background to the response. Ms. Clark considered the housekeeping measure to be a low priority.

Dr. McGee had mixed reactions to these measures. She wanted to include a measure that addresses cognitive function, and she was not impressed by the confusion measures. For the NHQR, this is an important public policy issue to address. Leaving out the IADLs is of concern because much effort is required to sustain frail people in their homes, and more than nursing care is needed. ADLs are the best measures to include in this set, partly because of the concern with accountability for the IADLs. An agency can assess IADL needs and make referrals, but it is not fair to hold it accountable for the patient's ability to do these things, especially given how subjective these are to assess. The NHQR and CMS public reporting can explain that a reason to include the IADLs is how much impact they have, together with sufficient cognitive functioning, on people's ability to live independently. Ms. Teenier added that little reimbursement for addressing IADLs is available.

Dr. Saliba reported that a team that examined the ability of certain outcomes to predict rehabilitative costs found that IADLs were the most robust predictors of cost. If IADLs are not addressed, it is difficult to treat other problems. It might be appropriate to include one of these measures in the NHQR because of the need for reimbursement.

Dr. Gerteis noted that consumers and others she spoke to had similar concerns. Many believed that patients could obtain sufficient help with laundry, shopping, housekeeping and other such tasks from friends and family. But if a patient cannot prepare even light meals, it is more difficult for others to provide care on a regular basis.

Dr. Bernard pointed to the impact of IADLs on allowing someone to remain in his or her home. Laundry, shopping, and housekeeping are intermittent needs, and as long as the patient has cognitive ability, he or she can negotiate for these. The ability to prepare light meals enables a patient to live independently, so Dr. Bernard might support the inclusion of this measure.

Top of Page

Category 6: Talking with People

25. Improvement in Speech or Language
26. Improvement in Phone Use

Dr. Saliba pointed out that measure 25 covers speech therapy. After a stroke, 35-40 percent of people have difficulty communicating. Stroke accounts for a relatively small proportion of the homecare population, but improvement in speech or language is the only outcome that is addressed by the speech therapist, which makes it worth considering.

Ms. Terry noted that the speech measure is not risk-adjusted and only a small percentage of patients receives speech therapy in the home. Dr. Saliba sees a fair number of patients who cannot communicate with anyone, which could be addressed with a speech board. Ms. Terry noted that by the time a patient needs speech therapy, he or she often is not homebound and can obtain therapy elsewhere.

Dr. Fitzgerald pointed out that this measure applies to 20 percent of the population. Dr. Hittle added that a sufficient number of patients with speech-language impairment are eligible for improvement on this measure. Ms. Fredland suggested that this may be due to difficulty with understanding, which is not addressed by speech therapy. Ms. Crisler pointed out that speech-language pathologists address hearing as well as speaking. Dr. Hittle suggested that when this measure was developed, too much may have been combined into a single item. However, the item's reliability is adequate.

Dr. Golden asked whether someone who does not regain verbal abilities but can point or use assistive devices would be considered to have improved on this measure. Ms. Crisler replied that this patient would not reflect improvement, because this measure addresses oral communication only.

Ms. Teenier addressed actionability. This is not a highly used service for various reasons, such as homebound status, therefore this outcome is not one on which agencies can have a major impact.

Ms. Fredland pointed out that telephone use is related to cognitive functioning. Patients who can use the telephone are alert and oriented, whereas those who cannot are confused. This is not an appropriate measure to include, because a more direct of cognitive functioning is preferable.

Dr. Fitzgerald noted that none of the agencies selected this measure.

Ms. Crisler explained that telephone use falls under safety. For AHRQ's framework, with this elderly homebound population, safety is a very important factor. But Ms. Terry pointed out that it is not a major focus in homecare, although this would be appropriate for a long-term care program. This is a low priority in homecare.

Dr. Bernard suggested that teaching someone how to call 911 or an emergency contact is important if a patient lives alone. Ms. Teenier said that calling 911 and calling a friend are separate skills. The phone use question is really two different questions.

Dr. Kelley noted that AHRQ is emphasizing patient-centeredness, timeliness, and safety. He asked panel members to suggest where additional work is needed on OASIS measures so that they can better address these AHRQ priorities. Dr. Bernard suggested that if these two measures were changed, they might address safety. They do not do so as written.

Top of Page

Stabilization Measures (Categories 7-12)

28. Stabilization in Bathing
29. Stabilization in Grooming
30. Stabilization in Management of Oral Medications
31. Stabilization in Light Meal Preparation
32. Stabilization in Laundry
33. Stabilization in Shopping
34. Stabilization in Housekeeping
35. Stabilization in Cognitive Functioning
36. Stabilization in Anxiety Level
37. Stabilization in Transferring
38. Stabilization in Speech or Language
39. Stabilization in Phone Use

Dr. Fitzgerald believed that these measures are not intuitive, and thus not appropriate for CMS public reporting because they are difficult to understand. For the NHQR, stabilization differs from improvement, and some research/policy measures may be appropriate to consider. CMS public reporting should rely on the improvement measures for these outcomes. Ms. Fredland agreed, except for cognitive functioning, where stabilizing the patient is very important for both CMS public reporting and the NHQR.

Dr. Bernard expressed concern about the limited variability of these indicators. For example, the bathing measure ranges from 89 percent at the 25th percentile to 94 percent at the 75th percentile. The measures that are most appropriate should have some variability, otherwise, comparing agencies is not useful. Presenting the trend on the national level over the years shows how agencies are addressing chronic illness and supportive needs.

Dr. Shaughnessy explained that, for many types of patients, improvement is not the purpose of homecare. Assessing whether a patient does not decline is an appropriate way to gauge success for these types of patients. In addition, because of the way it is defined, there is a statistical idiosyncrasy in that stabilization measures tend to have much higher mean rates, typically in the 80-percent range. The limited variation for these measures is an artifact of the statistics.

Ms. Ketcham suggested that the measures would be more meaningful if patients could be categorized as short- or long-term. For longer-term homecare, the goal is stabilization. That is not the case for shorter-term patients. Dr. Shaughnessy believed that this is possible and perhaps AHRQ should take this into consideration in its reporting, although chronic and acute may be a more appropriate distinction. For patients with chronic conditions who have short episodes, the goal is stabilization.

Dr. McGee disagreed that consumers would not intuitively understand the stabilization measures, because much depends on how they are described. Some research shows that people are more impressed by what can go wrong than what can go well, so the stabilization measures could be more powerful in attracting consumers' attention. Dr. McGee supported distinguishing between acute and chronic patients, which would make the measure easier to explain. The stabilization measures do not always indicate no change. Dr. McGee wondered whether including improvement and stabilization measures for the same outcomes would be appropriate for the NHQR, because this might be confusing to consumers.

Dr. Bernard believed that including both would be helpful. The condition of patients with certain chronic illnesses deteriorates over the years. Home health agencies try to reduce this decline and keep the patient functioning at a certain level for as long as possible. Stratifying the indicators among the two populations would be a good idea.

Ms. Terry suspected that only a small number of chronic patients are cared for by homecare agencies, because most Medicare homecare is for short-term, acute episodes. Homecare was developed as a long-term care program, but this changed as a result of reimbursement. Today, the average patient is in Medicare homecare for approximately 47 days.

Dr. Murtaugh supported separating post-acute from chronic patients, although this would not be easy using existing data. He proposed length of stay as an alternative. Dr. Shaughnessy replied that these distinctions can be approximated reasonably well from the OASIS data based on prior hospital stay, length of stay, and a list of conditions that tend to be chronic rather than acute. Stabilization is relevant to patients with serious cardiac and pulmonary problems, for whom other types of outcomes need to be controlled so that the conditions do not deteriorate. In some situations, a lack of decline is important.

Ms. Fredland observes many chronic patients—elderly residents of her region who worked in heavy manufacturing that led to chronic conditions. Some regions of the country have a higher percentage of chronic patients, although the national level is low.

Ms. Nguyen pointed out that acute patients have shorter lengths of stay, and her agency has chronic patients who are frequent readmissions and can be stabilized in a short period of time. Dr. Shaughnessy explained that these are considered short-stay patients in OASIS. Ms. Nguyen supported separating patients into two categories, especially for consumer reporting.

Dr. Gerteis did not test the stabilization items separately. Consumers did not understand them. The stabilization items were described as staying the same or not getting worse. Consumers had difficulty understanding the concept rather than the wording.

Dr. Golden explained that the OASIS measures are not chronic care measures. They only measure what happened during the timeframe in which the patient was seen by the homecare agency. The stabilization measure is a way to avoid saying that the patient lost function during care, which is a negative measure.

Ms. Terry pointed out that chronic heart failure patients generally require longer term home care, but they often have short acute exacerbations. Medicare and Medicaid allow homecare agencies to keep only a small percentage of chronic patients in their programs for certain specific conditions. Even though these patients have chronic conditions, they are seen for short periods of time, when the agencies attempt to stabilize them. Stabilization includes improvement. For dyspnea, stabilization is a better measure in some ways. But stabilization is a difficult concept to understand.

Dr. Bernard asked whether the denominator for the stabilization measures is the same as for the improvement measures. Dr. Hittle responded that the population for the improvement measures includes anyone who has the ability to improve. This means that none of the patients included in the improvement measures are functioning at the optimal level. If 20 percent of the patients of one agency are in the most independent category and 40 percent of a second agency's patients, then the second agency has much less opportunity to improve the condition of its patients. The measures are risk-adjusted for this. The stabilization measures exclude patients who are at the lowest end of the scale and who therefore cannot deteriorate further.

Dr. Shaughnessy added that for the dyspnea improvement measure, the denominator includes only those who can improve and therefore excludes those at the highest end of the dyspnea scale. To be in the denominator, a patient must be at risk of being in the numerator.

Dr. Hittle pointed out that if patients with no potential for deterioration were included in the stabilization measures, they would all receive a score of 1, meaning that they were stabilized. Similarly, if patients who are doing so well on a given outcome that they have no potential to improve were included in the improvement measures, they would all receive a score of 0, indicating no improvement. This is why these groups of patients were excluded. Dr. Golden added that the stabilization and improvement groups have some overlap in the middle, but the stabilization group excludes the lowest performers, while the improvement group excludes the highest performers.

Dr. Bernard argued against including the stabilization measures, not because they are difficult to understand, but because the population for which stabilization is relevant is not identified. Stabilization simply offers another way of discussing the data, and the improvement measures are probably sufficient for this purpose. Dr. Hittle pointed out that eliminating the stabilization measures would exclude all indicators that patients are deteriorating at a higher rate. Ms. Teenier suggested replacing stabilization with deterioration. Mr. Fitzgerald added that stabilization does not measure the outcomes of interest. We need the inverse of stabilization to identify the patients whose condition deteriorated. But stabilization frames the issue in a positive way.

Dr. Hittle suggested that if different measures for the chronic and acute populations were established, this should not be done by controlling what is in the numerator and the denominator. It should be done by stratifying on some other measure that distinguishes between the two populations. Dr. Shaughnessy said that this could easily be done for AHRQ reports.

Ms. Ketcham noted that if long-term homecare can help patients maintain their levels in bathing, light meal preparation, and other important functions, then these patients can probably remain in the home. Dr. Bernard suggested that this requires defining the episode of care differently, because one admission might not be sufficient to detect whether this is happening for a patient with chronic readmissions. Ms. Ketcham suggested using the length of stay in homecare to describe these patients.

Ms. Clark proposed that these statistics capture the patient who is admitted multiple times with chronic illness, but not the patient who is never discharged. The statistics are based on admission and discharge. Patients who are with agencies for years are not included, except in the case mix through length of stay. Dr. Shaughnessy suggested that these patients could be captured, but whether this is worth doing depends on whether the population in question is sufficiently large. Ms. Ketcham noted that whether this is worth doing should also be based on how many healthcare dollars are spent on this small population.

Ms. Clark pointed out that most patients require acute care and are referred to a home health agency by a hospital discharge planner or physician for a short length of stay. During this period, the patient does not have the opportunity to investigate and choose an agency. But patients receiving service after a recertification period who may not be satisfied with their current agency do have the opportunity to investigate other agencies. For these patients, the CMS public reporting would be very useful.

Dr. Shaughnessy suggested that when agencies care for patients with acute problems, other dimensions of the patients' health need to be addressed to prevent decline. Stabilization measures are therefore needed even for acute patients.

Dr. Remsburg suggested that improvement and stabilization together are meaningful. Even for acute patients, not getting worse is important. These two types of measures provide different types of information about the care an agency is providing. Whether a provider is improving the patients who can be improved and maintaining the rest are important. Dr. Fitzgerald pointed out that providers will continue to receive all 41 indicators.

Dr. McGee felt it was crucial that the NHQR provide a very specific profile of the patients whose data are presented. The report needs to make clear that it does not capture people who are in the system for the long term, who may be small in number but large in cost.

Dr. McGee argued that the real audience for CMS public reporting is intermediaries, such as discharge planners, family members, and physicians, who help steer patients toward certain agencies. Public reporting has value even for those with a sudden episode, because publicizing results has an impact.

The outcomes addressed by the stabilization measures are a subset of the improvement measures and are not the ones that seemed to be most highly favored in the panel members' discussions. Perhaps the NHQR should address the stabilization measures in conjunction with or instead of the improvement measures and explain which group is involved and why looking at these issues in this different way might be helpful. It is difficult for readers to shift gears from persons who are doing well to persons who are doing poorly. This raises the risk that they will misinterpret the data.

Ms. Terry wondered why only a subset of the improvement measures are included in the stabilization measures. For example, dyspnea is not included, although it would be understandable and appears to be more relevant. Many of the IADLs are included. Dr. Shaughnessy responded that these 41 measures emerged from the comments of many people. Most groups said that in spite of the problems with IADLs, they mean a great deal and are not completely captured by the ADLs and cognitive functioning. The acute patients that homecare agencies typically see have chronic dimensions of their health that should not be neglected. Dr. Shaughnessy said that dyspnea might be an appropriate stabilization outcome, but it was necessary to make choices.

Dr. Hittle added that the researchers started out with more stabilization measures, but many did not meet the minimum standard for risk adjustment, which is why they were dropped. Of those that remained, some have a reasonable degree of risk-adjustment, and others are borderline. With stabilization measures at 95 percent, it is statistically difficult to show much impact.

Dr. Saliba suggested that part of the problem with stabilization is that it is clearly meant to stand alone. She was concerned that providers would not receive credit for improvements on the stabilization measures. This argued for excluding improvement from stabilization and presenting both improvement and stabilization. If only stabilization is presented, it must include improvement.

Dr. Kelley noted that a goal is to prevent confusion. The NHQR will not be organized as a chart book. Rather, the measures will be synthesized and written up in a short report, which should prevent some of the confusion that might arise with a chart book format. The report's appendix will describe the numerator, denominator, and other statistical considerations for each measure, which should also help clear up some confusion.

Julie Crocker, M.S.N., R.N., agreed that chronically ill and acute patients represent different populations with very different needs in terms of utilization for CMS public reporting.

Dr. Bernard pointed out that for a postsurgery patient who has other chronic illnesses and needs home health primarily for rehabilitation or wound repair, the agency should ensure that in the course of treatment, these chronic illnesses do not decline. Separating the two populations does not appear feasible. When an agency intervenes with an acute problem, it is important that other underlying diseases not worsen.

Ms. Ketcham considered including in her list of improvement measures upper body dressing, because it correlates with other improvement measures; oral medication management, which is what home health agencies do; ambulation, which is the object of much focus by agencies; and pain measures. For stabilization, she would include bathing, light meal preparation, cognitive functioning, and transferring, which patients need to be able to do so that they can be discharged to the community. All of these measures are suitable for both purposes. Stratifying the samples is desirable, yet there is value in including them as they are.

Dr. McGee found the domain labels for the measures misleading, because they do not necessarily reflect the measures included. Dr. Gerteis explained that the domains were created for this meeting, and have no independent importance. Dr. Hittle added that the measures were categorized into preexisting domains that the IOM committee developed, and he welcomed suggestions for rearranging the categories. Dr. Shaughnessy suggested generalizing the category for the stabilization measures as "sustaining health."

Dr. Saliba found it difficult to label stabilization as more important than improvement for transferring. Some patients lose the ability to transfer during an illness, but improve during homecare. Others do not transfer well before they are hospitalized, and the goal is merely to maintain their current level. The measure does not discriminate between those for whom stabilization is good and those for whom improvement is more appropriate.

For this domain, Ms. Fredland preferred the title, "sustaining health in the home." Bathing is needed to sustain a person in the home. Cognitive functioning is also important; once it declines, patients usually can no longer be maintained safely in their homes.

Ms. Terry found the concept of stabilization complicated and difficult to understand. Cognitive functioning is difficult for homecare and is probably not a main reason for homecare. The fact that this measure is not risk adjusted also makes it problematic. Helping persons with bathing is a major part of homecare. Improvement in this outcome is probably more appropriate than is stabilization.

Dr. Fitzgerald believed that the NHQR should report areas that are difficult to address and challenges facing the home health community. This would focus the attention needed to produce changes. Any indicators that home health professionals regard as not actionable should perhaps be included in the NHQR, while measures that are clearly actionable belong in the CMS public reporting. Dr. Sangl noted that actionability is a common criterion for both purposes. Mr. Fitzgerald wanted to draw attention to the lack of actionability through the NHQR to demonstrate the need for research in these areas.

Dr. Saliba said that this would make the NHQR a white paper that lays out a research agenda. Because this criterion is not used for many other elements in the report, homecare would appear to be performing more poorly than other services. The nursing home measures, for example, will focus on issues for which process measures and outcome measures are available. It is not clear that homecare should take the lead on this.

Dr. Fitzgerald believed that the straightforward issues for which a great deal of variation among agencies exists and which can be improved when agencies focus on them, are suitable for consumer reporting. This challenges agencies to focus on them and makes agencies feel confident that they can improve them. The ones that do not improve much when agencies focus on them are more of a national priority. Ms. Ketcham wondered how all of the data in the report would fit together if Dr. Fitzgerald's suggestions were followed. The report must show how nursing homes, hospitals, and homecare fit together and how to best handle patients across the continuum.

Dr. Kosiak stated that some of these issues will be examined from a broader perspective. The distinctions between the chronically ill and acute populations might come across in other frameworks. Dr. Kelley added that the IOM framework is AHRQ's guide for examining these issues, but it has areas that need to be fleshed out. This is the first report, and changes will be made as the report evolves.

Dr. Murtaugh pointed out that the evidence base for what is actionable will differ across provider settings in the NHQR. In most cases, evidence is available and everyone in the field believes that certain practices are effective, but the science base for that is not strong. Dyspnea represents problems in terms of stabilization. Stabilization rates are probably very high because it does not apply to many patients, although it does apply to the minorities with heart failure or chronic obstructive pulmonary disease. Dr. Murtaugh recently completed a randomized trial that focused on improving symptomatology. The intervention had a positive effect on symptoms, although the investigators did not use outcome measures. This lends some scientific credence to the belief that homecare can improve certain conditions.

Ms. Terry wondered how the NHQR would be distributed to the public. She hoped that it would include explanations with the graphs. For the risk-adjusted measures, the reports should explain expected and observed rates in a simple way. Dr. Kelley explained that the graphs shown to panel members were from Nursing Home Compare, which is an example of what the CMS public reporting is likely to include. The NHQR is a report to Congress and other policymakers and will not necessarily be written for the public. However, AHRQ does plan to develop an NHQR product for the public based on the report. Dr. Kelley and his staff would work closely with experts at AHRQ who can translate technical concepts to a lay audience.

Dr. Bernard cautioned against including indicators for the NHQR that appear to be problematic and have no variability. The NHQR should include indicators that have variability and are amenable to interventions, so that disparities can be identified in different regions and communities. Otherwise, the measures chosen will not be useful.

Ms. Fredland supported the inclusion of cognitive functioning and bathing as stabilization measures. Cognitive functioning is critical to keep patients in their homes.

Ms. McCall asked whether it would be appropriate to hold an agency responsible for cognitive functioning. Ms. Terry replied that cognitive functioning is different from confusion but difficult to distinguish at the first visit from confusion, depression, and other symptoms. Cognitive functioning is never the main focus of a patient's care. Ms. Fredland agreed. But if the report is to address longer term patients, cognitive functioning is important. Ms. McCall agreed that agencies should be held accountable in such cases. But home health agencies need to be comfortable with being held accountable for these measures as legitimate outcome measures for homecare. Ms. Terry argued against including cognitive functioning, because it is not easily amenable to being addressed in the short term and is not a major reason why patients receive homecare. Homecare agencies can make a difference in many other measures in the short term. Dr. Saliba said that few interventions decrease the rate of decline in cognitive functioning.

Ms. Nguyen suggested that including such measures in CMS public reporting might lead agencies to take resources away from outcomes for which very little room for improvement exists and that are the reason why the patient was referred for homecare. This is in light of the short period of time and small number of visits available to make accomplishments that are often major. Ms. Teenier pointed out that the inter-rater reliability is 0.63, which is at the lower level of acceptability. Ms. Clark agreed that very few homecare skills can address this.

Ms. Fredland has often found that cognitive functioning is an issue of depression, which can be treated because it often results from various Vitamin B12 deficiencies. Most homecare agencies do not consider whether they can keep these patients stable. Ms. Ketcham believes that this argues for improvement rather than stabilization. Dr. Saliba suggested that the reversible causes of dementia are very few. Supplementing Vitamin B12 in someone with profound cognitive impairment will not lead to sudden improvement.

Dr. Remsburg said that the stabilization and improvement measures are closely related. If she were to choose bathing, medication management, and light meal preparation for improvement, she would also choose these outcomes for stabilization. She is less certain about transferring, but if she did choose it she would choose both the improvement and stabilization measures.

Top of Page

Category 13: Utilization Measures

27. Discharged to Community
40. Any Emergent Care Provided
41. Acute Care Hospitalization

Ms. Teenier asked which questions were used to derive the scores on these measures. Referring to Tab 8 in the background notebook. Dr. Hittle explained that measure 27 is based on question M00870 (Where is the patient after discharge from your agency?). If the patient is discharged to a hospice or hospital, then the patient is not discharged to the community, whether he or she was formally discharged from the agency or not. Even if the patient is not formally discharged, this is still considered the end of the episode of care, because the patient is in someone else's care.

Acute care hospitalization is based on a response to question M0855 (To which inpatient facility has the patient been admitted?). Question M0890 (If the patient was admitted to an acute care hospital, for what reason was he/she admitted?) covers emergency, urgent, and elective hospitalizations. Of all discharges to hospitals, Dr. Hittle said 85 percent were for emergent or unscheduled care, 6 percent were for urgent care, 6 percent were for elective care, and 3 percent were unknown.

Ms. Teenier asked whether discharge to the community includes everyone who was admitted to a facility in the denominator. Dr. Hittle replied affirmatively. Dr. McGee asked whether the patients with unknown status were excluded and Dr. Hittle replied that they were removed from both the denominator and the numerator. The validation sample has one million cases for hospitalization, but only 995,000 cases are included; the remainder have a response of unknown on question M0870.

Ms. Teenier suggested that these questions ask the same thing, with one positive and the other negative. Dr. Hittle explained that acute care hospitalization excludes rehabilitative nursing home and inpatient hospice, which are in the denominator but not the numerator (which includes hospitals only). Dr. Shaughnessy added that the acute care and community discharge measures are very highly correlated, more than any other two measures. Dr. Hittle noted that they are so highly correlated that they are almost the same.

Ms. Teenier suggested that the nursing home may be the most appropriate environment for a patient to receive care, but discharge to a nursing home may be perceived negatively by the community. If an agency did not recommend that the patient be discharged to a nursing home, the agency would be told on evaluation that it did not take the patient's needs into consideration.

Question M0855 does not include nursing homes or rehabilitation facilities. Ms. Teenier suggested that discharge to a nursing home or rehabilitation facility may be a better measure of accountability for homecare agencies, because it may be best for the patient. Dr. Fitzgerald pointed out that the same can be said for hospitalization. The reasons why a patient needs to be hospitalized may be out of the control of the home health agency, so this should be removed from the numerator.

Ms. Fredland pointed out that question M0855 has "no inpatient facility admission" as an option, which means that the patient remained in the community. This question provides information on exactly where the patient went, whether to a hospice, nursing home, rehabilitation facility, or hospital. Therefore, this question covers the same information as M0870. Ms. Fredland would select question M0855 over M0870, because it covers remaining in the community. Ms. Clark pointed out that this ignores the skip pattern in the questionnaire, because question M0855 is only completed if the answer to question M0830 on emergent care is "no" or "unknown." Dr. Hittle added that question M0855 is not completed if the patient is discharged to any destination other than an inpatient facility.

Dr. McGee suggested that some of the items under reason for hospitalization, question M0895, could be deleted.

Dr. Fitzgerald suggested that if someone with an acute myocardial infarction were hospitalized, that should not be in the numerator as a failure of the home healthcare agency. Dr. Shaughnessy replied that, at present, these patients are not excluded. The actual numbers of patients in some of these groups are so small that this is not very important, although it can be done.

Dr. Fitzgerald argued in favor of making these adjustments. Some of the items in question M0895 should be deleted because they do not indicate a failure on the part of the home health agency. Dr. Shaughnessy pointed out that it is difficult to draw the line between what could and what could not be the agency's doing. Ms. Crisler said that if a patient has been exhibiting a symptom for weeks while under homecare, it is not appropriate to exclude him or her.

Ms. McCall raised the issue of the timing of the hospitalization data collection, asking whether data are collected on patients who are hospitalized 2 days after discharge and whether agencies track this. The OASIS measures do not include such hospitalizations. Dr. Murtaugh pointed out that the agency may not know the reason for the hospitalization. In many cases, an agency staff member visits the home only to learn that the patient has gone to the hospital. There are many unknowns in this category, although this information would be very helpful.

Dr. McGee suggested that it would be much cleaner and more defensible to measure the number of people admitted to the hospital during homecare rather than to adjust by excluding certain reasons. She was inclined to let the case mix adjustment and aggregation of data take care of these issues, because not all of this can be second-guessed. The preface to the display of data would list some reasons why people enter the hospital during homecare and explain that home health agencies are not accountable for all of them.

Ms. Ketcham argued against including the discharge to the community measure, because premature home health discharges cannot be measured. She would include acute care hospitalization and emergent hospitalization, because these can be addressed by home health agencies. Many patients end up in the hospital because of a failure of the clinician to assess early symptoms or provide appropriate treatment. Ms. Ketcham wondered whether a patient going to the hospital for a planned visit, such as for chemotherapy, is risk-adjusted for agencies with many of such patients. Dr. Shaughnessy replied that neoplasms are among the approximately 50 different risk factors included, and many of the factors discussed by panel members are on this list.

Ms. Terry argued that one of the primary purposes of homecare is to prevent rehospitalization. Therefore, it is a good indicator, although risk adjustment is very important. People will regard risk-adjusted rehospitalizations of 26-28 percent as high. This is a very important indicator of agency performance. If the rate is 20 percent, the agency is doing very well.

Seventy percent of patients in Ms. Terry's agency come from the hospital. Many patients who are discharged after cardiac surgery return to the hospital within 3 days, usually because they were discharged prematurely from the hospital and not because of anything the homecare agency did. Patients are being discharged sooner and do not do well in homecare. Other variables are difficult for homecare to control, and this must be understood.

Dr. Hittle explained that although patients who are hospitalized for chemotherapy are transferred rather than discharged, the hospitalization is considered the end of the episode of care because the patient is in someone else's care. One of the challenges of measuring outcomes in homecare is the definition of an episode of care. For example, agencies have different policies regarding whether a patient goes to a nursing home for respite care or to a hospital. Some agencies discharge their patients after they are in the hospital for 1 day, whereas others keep the case open for 2 weeks. The definition developed by the University of Colorado researchers was endorsed by the homecare agencies in the demonstration. When a patient leaves the agency's care for a specified period of time, it is considered the end of the episode of care.

Ms. Terry pointed out that since the advent of PPS, under Medicare, agencies cannot discharge patients until the end of the episode, even if the patients are hospitalized. The OASIS definition is based on the way homecare agencies discharge patients. Dr. Hittle explained that the episode of care is over because the agency has finished with the patient for the time being. A new episode begins when the patient returns. Dr. Shaughnessy added that this is an issue for payment, not for OASIS. Ms. Terry explained that when a patient goes to the hospital, the agency completes a transfer form, which is a short version of the discharge form. Dr. Shaughnessy noted that information on intervening hospitalizations is collected on the shorter form that addresses the hospitalization. Health status data are collected on discharge from the agency in a more formal sense when the patient is not discharged to the hospital.

Dr. Shaughnessy said that all hospitalized patients are excluded from the denominator for the health status outcome measures.

Ms. Crisler explained that the very first three-agency pilot demonstration collected health status data that had to be recreated at the last time point prior to hospitalization. If the home health agency staff did not see the patient prior to hospitalization, the accuracy of these data was highly suspect, which is why this information was dropped. Dr. Hittle added that for practical purposes for providers, the health status items pertain to patients who are not hospitalized or institutionalized on discharge from the agency. Dr. Bernard believed that this information is critical for some of the stabilization measures. The NHQR needs to make clear that this pertains only to the population that remains in the community after completion of the homecare experience. Dr. Shaughnessy noted that the instructions for interpreting the outcome report address this topic, but it also should be covered in the CMS consumer reporting.

Dr. Golden pointed out that the hospital readmissions that do not reflect homecare services are random events that are equally random for all agencies. These readmissions will therefore be adjusted for in the mean. However, if a homecare agency has patients who tried to live at home but ended up in a nursing home, the population is skewed. Some institutions might try harder to avoid sending patients to a nursing home. Ms. Teenier is not sure that the interventions of the agency can make a difference in keeping patients at home, because this is often attributable to the family's efforts. Dr. Golden suggested that an agency might admit more patients because of a hospital's philosophy, that is, a system bias.

Dr. Golden supported including these measures in both reports, even if an agency has many readmissions because of patients discharged too soon from the hospital. Public report cards create tensions. If an agency believes that it received a negative report because of such policies, this would create a dialog to address issues that would not otherwise be addressed.

Ms. Clark supported including acute care hospitalizations. In the survey process, clinical records are reviewed, and, if a patient had emergent care and rehospitalization, the agency's efforts are examined. If the patient had been exhibiting symptoms for several days without any intervention and was then rehospitalized, a direct correlation would be assumed. This is important for the agency's report card.

Dr. Bernard also supported the inclusion of acute care hospitalizations. With these indicators, 0 is never targeted as a goal. Variations—why one is higher and another lower—would be studied. The data do not identify the issue, prompting the agency to look further to see whether the case mix is different. Emergency care is a critical indicator that might be affected by home health more than acute care hospital admissions. Emergent care is not risk-adjusted, but this is not a concern. Dr. Shaughnessy noted that this measure has potential for risk adjustment. Dr. Hittle added that the emergent care measure is based on question M0830.

Dr. Fitzgerald described a major movement to make readmission rates more sensitive to failures in primary discharge. An opportunity exists to make this hospitalization rate more sensitive to home healthcare and more valid as a measure. If myocardial infarctions, chemotherapy, and scheduled surgical procedures were excluded from the numerator and perhaps the denominator, the measure would be more valid. Perhaps the measure could include a buffer period of 2-3 days for readmissions, so that readmissions that occurred within 2 or 3 days after admission to homecare would not be considered a failure of the agency. The goal is to be as precise as possible when giving a negative report, and as generous as possible in excluding factors from the numerator. Any indicator can be expressed in a positive mode. If 25 percent of patients return to the hospital, 75 percent do not, which is how the measure could be expressed for CMS public reporting. But Ms. Clark pointed out that the 2-3 day buffer period might not be appropriate if an agency delayed that long before seeing the patient for the first time.

Dr. Saliba noted that chemotherapy is a prescheduled readmission. Most of the others might be affected by the home health agency. Based on this limited set of administrative data, it will be difficult to distinguish between avoidable and unavoidable hospitalizations, so total hospitalizations must be used. Dr. Shaughnessy commented that risk factors can translate into some of these exceptional categories. Dr. Hittle noted that the first phase will use the measures as currently defined. Refining this measure could be a research priority in the future. Mr. Fitzgerald believed that the precision of this measure could be increased considerably.

Ms. Terry suggested that the timeframe is important when the measure is refined. If a patient is readmitted on the same day he is admitted, this usually indicates a problem. Risk adjustment is important for this measure. If it is not risk-adjusted, it will not be meaningful, because important reasons will not be taken into account.

Dr. Golden said that too few measures to address hand-offs are available in the system. Most in the field would say that someone admitted for 23-hour observation had a hospital and not an emergent visit. Ms. Crisler noted that such patients are seldom formally admitted, which is the criterion. Dr. Bernard's research indicates that this is an insurance issue, not a patient issue. But in Dr. Golden's institution, a patient in observation status is cared for by an inpatient team, not an emergency team. Dr. Bernard would categorize this as an inpatient stay. Ms. Nguyen bases the decision on whether the patient was formally admitted to the hospital. If the patient was not admitted to the hospital, the homecare agency is still responsible for that patient.

Dr. McGee asked whether the populations covered by the acute hospitalization and emergent care measures overlap. Dr. Hittle replied that responses to questions M0830 and M0840 are collected for patients who are transferred to an inpatient facility. Dr. Hittle explained that the second paragraph in Box 2b of Measure 40 behind Tab 13 should say, "Patients who are nonresponsive at start of care are excluded." Ms. McCall argued that patients who are admitted to the emergency room and then admitted to the hospital are billed only for their inpatient stay and should therefore not be included. It would be better for a separate measure to address whether the patient had emergent care only that was not followed by a hospitalization.

Dr. McGee believed that the current measure, which includes emergency care whether or not it leads to hospitalization, is better because this is how people will view this issue. The title of this measure is unlikely to include a qualifier. Observation stays are "pseudo-hospitalizations" that are outside the control and influence of the homecare agency.

Ms. Terry said that acute care hospitalization is cleaner for homecare. Determining emergent care is very difficult. Staff may go into the home two or three times a week, and patients do not always report whether they went to the emergency room. It is difficult to collect accurate data on emergent care, especially when this includes emergency visits to a physician's office. Acute care hospitalization is a cleaner statistic because agencies know when a patient is admitted to the hospital. Ms. Nguyen added that, in many cases, the patient goes to the emergency room through the intervention of the homecare nurse, because critical symptoms were not addressed earlier.

Dr. Bernard would like to disaggregate admissions. Often, the way someone is admitted to the hospital is through the emergency room. There would be some overlap between the acute care hospitalization and the emergent care measures. Acute care seems more consistent for homecare. Dr. Bernard would like to include an emergent care measure, one that reflects true emergency room use. On a policy level, this is very important. It would be good for home health to demonstrate that it is associated with a decline in emergent care. The measure, as currently defined, does not provide this information.

Ms. Fredland agreed that the current emergency care measure is not appropriate because it does not distinguish between admissions. Without the information provided by Questions 840 (emergent care reason) and 895 (reason for hospitalization), there is no way for the public to know how agencies do with a particular diagnosis. An agency can prevent hospitalizations by keeping patients hydrated and preventing nausea, vomiting, and constipation. The public will want know this when selecting an agency. Dr. Shaughnessy pointed out that this is an adverse-event measure for a low-frequency event that cannot be risk-adjusted very well, which is why it is provided descriptively to agencies. He asked whether panel members want such infrequent outcomes to be included in measures for CMS public reporting.

Dr. Saliba argued that it might not be appropriate to admit patients who do not do well to the emergency department before admitting them to the hospital. A good team can often arrange for them to be admitted without waiting in the emergency department. Some agencies use emergency departments inappropriately as a substitute for making a home visit when a patient calls with new problems. Emergent care is disruptive and puts patients at high risk for problems.

Dr. Shaughnessy pointed out that the agency-level variation in both the standard deviation and interquartile range is quite acceptable.

Ms. Teenier explained that the problem with the emergent care question is that it includes more than just emergency room visits, for example, emergency visits to the physician based on an appropriate intervention by the homecare nurse. Ms. Clark pointed out that homecare agencies are often directed to send a patient to the emergency room by the physician.

Ms. Terry suggested that ideally, patients would be seen by teams, but in reality, homecare staffs have difficulty reaching physicians by telephone. They sometimes try for days while a patient's condition deteriorates. Sending older patients to the emergency room is very bad. This measure includes appropriate interventions, such as sending a patient to a physician. Separating these issues would be helpful. Some physicians tell agencies that they will meet the patient at the emergency room.

Dr. Saliba argued that emergency department care is a quality issue in that physicians are not responding to agencies appropriately. This may not be mutable but quality problems are not inevitable. This is relevant in a quality report card.

Ms. Terry contrasted the "cleanliness" of acute care hospitalization with emergency care, which includes emergency visits to the physician that probably represent an appropriate use of resources. Ms. McCall agreed that this measure needs to reflect emergency room care only.

Ms. Ketcham found that physicians underreport emergent visits. Ms. Fredland argued that agencies cannot control emergency room visits if they are recommended by a physician. In some cases, the home health nurses argue that the condition could be treated at home, but the physician disagrees.

Dr. Fitzgerald noted that this speaks to the issue of hand-offs between physicians and home health agencies. When patients who are not ready are handed off from the hospital to the home health agency, the home health agency should not be penalized and these patients should not be included in the statistics. The quality failure would then be lost to the measurement scheme, but it belongs to someone other than the home health agency. Dr. Shaughnessy asked whether patients who go to the emergency room on the basis of a physician request should be removed from the measure. Dr. Fitzgerald was comfortable with the measures as written, saying none of the indicators needed to be changed for CMS or AHRQ purposes. If they were changed, these patients should not be included in the scores of home health agencies. In quality measurement, the patient should not be in the numerator if the homecare agency could not have prevented the patient from being there.

Dr. Shaughnessy said that OBQI demonstration agencies were asked that acute care hospitalization be a target outcome, but the agencies demonstrated considerable resistance because they could not control physician behavior. However, many improved their communications with physicians as a result of this effort. Dr. Shaughnessy and his colleagues have been impressed by the capacity of home health agencies to make a difference in areas in which they initially perceived they could do little.

Ms. Ketcham characterized this as a community practice problem. If an agency benchmarks well with other agencies in this area, this can be conveyed to the public. If benchmarking problems indicate a systems problem in the community, it is time to change the practice.

Top of Page

   Previous Section    Contents     Next Section