U.S. Department of Health and Human Services

Medicare Post-Acute Care: Quality Measurement Final Report

Appendices B-K

APPENDIX B. Members of First Expert Panel

APPENDIX C. Sample Rating Form (Back & Neck Condition)

Physical Function Outcomes

Mental Health Outcomes

Quality of Life Outcomes

Utilization Outcomes

Physiology Outcomes

Satisfaction Outcomes

Process

APPENDIX D. Notes From First Expert Panel Meeting

Clinical Panel Meeting Notes for Congestive Heart Failure

Clinical Panel Meeting Notes for Pneumonia

Clinical Panel Meeting Notes for Stroke

Clinical Panel Meeting Notes for Back & Neck

APPENDIX E. Final Ratings and Dropped Quality Indicators

CHF -- Final Ratings

Pneumonia -- Final Ratings

Stroke -- Final Ratings

Back & Neck -- Final Ratings

Discussion of Dropped Quality Indicators

APPENDIX F. Final Quality Indicator Ratings Across All Conditions

APPENDIX G. Barthel Conversion Codes

Barthel Index Replicated Using the MDS 2.0 (10/94n)

Barthel Index Replicated Using the MDS PAC (Draft 8)

Barthel Index Replicated Using the OASIS (Draft B1)

APPENDIX H. Members of Second Expert Panel

APPENDIX I. Notes from Specialist Conference Calls

CHF Conference Call -- Martha Radford, M.D.

Pneumonia Conference Call -- Thomas Marrie, M.D.

Back and Neck Conference Call -- Richard Deyo, M.D.

APPENDIX J. Issues for Second Expert Panel Discussion

APPENDIX K. Project Timeline

APPENDIX B. Members of First Expert Panel

[Top of Page]

APPENDIX C. Sample Rating Form
(Back & Neck Condition)

Table of Contents

Physical Function Outcomes

Mental Health Outcomes

Quality of Life Outcomes

Utilization Outcomes

Physiology Outcomes

Satisfaction Outcomes

Process

Notes

Outcome indicators refer to change (e.g., decline, improvement, recovery) in the item listed.

Indicator Rating Scale:
     0 = negligible value
     1 = definite value
     2 = extremely important for measuring quality of care

Measure Rating:
     R = recommend
     NR = not recommend
     Blank = not familiar

[Top of Page]

APPENDIX D. Notes From First Expert Panel Meeting

Table of Contents

Clinical Panel Meeting Notes for Congestive Heart Failure

Clinical Panel Meeting Notes for Pneumonia

Clinical Panel Meeting Notes for Stroke

Clinical Panel Meeting Notes for Back & Neck

CLINICAL PANEL MEETING NOTES FOR CONGESTIVE HEART FAILURE

December 9, 1998

Indicators discussed:

• Physiology - Mortality
• Quality of life - Well-being
• Physical function - Exercise tolerance
• Utilization - Skilled nursing facility utilization
• Process - Medication compliance
• Mental health - Anxiety
• Additional indicators suggested by panelists

Measures were discussed for:

• Depression
• Patient satisfaction

A. Indicators

1. Physiology

Mortality
As an indicator, it is measurable and believed to be sensitive to quality of care processes (e.g., ACEI use). ACEI use alone is not a good indicator because its utility in diastolic heart failure, common in the elderly, is unclear. Death may not be an undesirable outcome in all circumstances, CHF may be a condition for which profound disability is rare such that patients might welcome death (compared with more disabling conditions or those causing greater suffering). However, assessment of patient goals, advance directives, will be necessary.

2. Quality of Life

Well-being
As an entity in itself, it’s difficult to measure. Validated measures of health related quality of life may be better. Unclear that a single question about subjective well-being is valid or responsive while health related quality of life measures have already been developed.

The indicator “well-being” was re-defined as “disease-specific quality of life” versus general health related quality of life.

3. Physical Function

Exercise Tolerance
As in indicator itself, it crosses over with many other physical function indicators. Since it other physical function measures really assess what people are able to do in their daily lives, exercise tolerance seems a narrow proxy for more important measures. Ambulation, for example, is probably a more meaningful indicator.

4. Utilization

Skilled Nursing Facility Utilization
This may be more a measure of severity of illness as well as lack of social supports, etc. rather than quality of care. However, number of admissions to SNF may be an important indicator, suggesting inadequate care in another setting (e.g., home health). For general utilization measures, hospitalization days was considered a very important indicator of quality in CHF -- and since some CHF acute care occurs in SNFs alone, a combined acute care utilization measure using total number of inpatient days may be useful. ER visits is unlikely to be a good utilization indicator, adding little to hospitalizations since the vast majority of patients visiting an ER with CHF are admitted. All in all, though, it is difficult to know if SNF admission is a good or bad outcome in CHF, so by itself it is not a very good quality indicator. If it is used as a measure, point of origin prior to admission would be important (e.g., home health to SNF).

A new indicator of utilization, total combined inpatient days in hospital and SNF, was added. The measures for SNF days were moved to the indicator list to represent individual quality indicators.

5. Process

Evaluation of Medication Compliance
Several panel members supported this indicator because medication compliance is known to directly affect outcomes like re-hospitalizations and long-term mortality. Several panel members felt it was important in the spectrum of patient education and self-efficacy. Some felt it was too process oriented and that functional outcomes were good enough -- patients who feel well and are functioning well is what matters, not whether they are taking their medications. All agreed that actual patient compliance is difficult to measure, but that measuring a PAC provider’s assessment of patient compliance may be possible, as well as measuring the provider’s provision of medication education. As part of this discussion, members felt that assessment of patients’ knowledge following provider education may be important. The cardiologist on the panel noted that his research group has developed a tool for assessing patient knowledge in CHF.

The indicator was redefined as evaluation of medication compliance: patient education and intervention to assist compliance. A new indicator, patient knowledge, was added.

6. Mental Health

Anxiety
Ambiguous indicator. Formal anxiety disorder, by DSM criteria, is probably rare and not an important indicator in this population. Less severe anxiety may be more common -- but the proper assessment of it and, more importantly, management of it are unknown. Such an indicator might suggest that more individuals should be treated for it, and that’s not at all clear. There is the Hamilton Anxiety scale, though it’s utility in this setting would be uncertain. Some members noted that good medical care of CHF and management of functional disabilities should lead to decreased patient anxiety.

B. Additional Indicators Suggested by Panelists

1. Quality of Life

Role Resumption
Many panelists commented that role resumption, with respect to social interactions, previous responsibilities e.g. as a caregiver, were very important to patients and may not be reflected by health related quality of life of physical function indicators. The Canadian Occupational Performance Measure was suggested as a possible tool to measure role resumption.

2. Process

Assessment of LV Function
Clinically important in CHF. An important care process in some settings, particularly outpatient management of CHF by primary care MDs. Unclear importance in PAC settings -- like vaccinations, PAC providers should assure that such assessment has occurred at some point and if not, provide assessment themselves.

Non-use of Calcium Channel Blockers
Easily measured. Inappropriate to use CaCh blockers in CHF across many studies yet still widely used.

Hypertension Control
Was not discussed by the panel but Michael Rich commented on the important of this as a quality indicator.

C. Measures

Depression
Widely considered a very important indicator by the panel, both the assessment for depression and appropriate treatment when identified -- although whether CHF outcomes improve with treatment of depression is unknown. Measures which can be used by non-mental health professionals are very appealing e.g. the GDS, also specifically developed for an elderly population. A mental health specialist must administer the HAM-D. The CES-D was developed for a younger, healthier population. The BECK counts physical symptoms, is longer than the GDS but would not be unreasonable. The DSM criteria may be too insensitive for depressive symptoms, which are also important in this population. The SKID, from the Diagnostic Interview Schedule is also a possibility and can be administered by telephone for potential follow-up after PAC or for home health patients.

Patient Satisfaction
Panel was unsure about how to measure it. The cardiologist was aware of two patient satisfaction questionnaires that have been used in CHF that we can consider -- though most panel members felt satisfaction was an area in which a general, non-condition specific measure would be appealing. Satisfaction may be difficult to measure because it’s difficult to assign the period/episode during which satisfaction is assessed. Does the episode include the hospitalization? Would different tools be necessary in different settings? A recent JAMA article assessing satisfaction following rehabilitation may be pertinent.

Patient satisfaction, as an indicator, was redefined as “satisfaction with care.”

CLINICAL PANEL MEETING NOTES FOR PNEUMONIA

December 10, 1998

The following indicators were discussed:

• Utilization - Physician visits for pneumonia
• Utilization- Home health care use
• Process - Immunization status
• Quality of life - Days in bed
• Mental health - Delirium
• Physiology - Duration of fever
• Physiology - Duration of fever and chest pain
• Physiology - Pleural exudate, abcess/empyema, pleural effusion
• Additional indicators suggested by panelists

Measures were discussed for:

• Health-related quality of life
• Symptoms

A. Indicators

1. Utilization

Physician Visits for Pneumonia
In and of itself, physician visits may not be a quality measure. Variation in physician visits may be due to practice patterns and differing facility requirements, rather than reflecting real differences in quality. It is also difficult to determine from a record review whether the physician visit was for pneumonia or for some other condition. Assuming that some physician contact is desirable, one may be able to define certain minimum standards of care that would vary by setting. Although there is currently no hard data to support any specific figures, one might propose the following:

We refined the definition of the indicator to be a process measure instead of a utilization measure.

Nursing Home Acquired Pneumonia
If we are measuring quality of care for all pneumonia patients, we can't exclude patients who develop pneumonia while being treated in a SNF. Since the diagnosis and treatment of NH-acquired pneumonia (NHAP) is different from community-acquired pneumonia (CAP), we will need to develop two sets of quality indicators for the two groups. For example, blood cultures, time to antibiotics, and oxygenation are important quality indicators for long-term NH patients, but not for more transient SNF or home health care patients. The same issue will have to be considered for patients who develop CHF or who suffer a stroke while in a SNF. Since an increasing number of patients with CAP are being treated in a SNF without a prior acute admission, we will actually need three sets of quality indicators: (1) for CAP patients with an acute hospitalization and SNF admission, (2) for CAP patients treated in a SNF without hospitalization, (3) for NHAP patients with an acute transfer and return. There may be some difficulty determining exactly when the pneumonia began. The standard measure is the detection of an acute pulmonary radiographic infiltrate, with an incubation period of 10 days during which symptoms (fever, etc.) may occur.

Home Health Care Use
Home health care use is usually seen as a resource outcome, and not a quality measure. For patients for whom home health care is appropriate, then receiving home health is an indicator of quality. If a patient is not a good candidate for home health care, then not receiving home health is an indicator of quality. If we assume that the distinction between the two types of patients is definable, then we can use the item as a quality measure.

We refined the definition of the indicator to be more of a process marker of "appropriate home health care use".

2. Process

Immunization Status
All agreed that pneumonia patients should receive a flu shot and a pneumo-vax. The discussion revolved around who should be responsible for providing it, and how to handle patients who had already received them prior to their pneumonia.

We refined the definition of the indicator to be a measure of "immunization status evaluation", where status is assessed and shots are given when appropriate.

3. Quality of Life

Days in Bed
Using retrospective recall during a patient interview ("In the last x days, how many days did you spend most of the day in bed?") may be difficult if the time period is too long. Experience has shown that the answers tend to cluster at 0-1-2, 7, 14, 21, etc. which results in the item being collapsed into categories. Days of limited activity may be a better measure because it includes days in bed. Perhaps a better measure would be to ask "When did you return to your usual household activities?" The COPM measure of role resumption gets at this, as well as an assessment of the patient's satisfaction with their level of ability. It has been validated in older individuals and takes about 30 minutes to administer.

We added a new quality of life indicator for return to usual activity, with a measure "days until".

4. Mental Health

Delirium
As a quality indicator, delirium itself may not be an appropriate measure. One could imagine a process indicator that rewards facilities for at least checking for delirium, though since the MDS requires an assessment there will be 100% compliance (even if the accuracy of the MDS assessment is in question). The MDS delirium item is considered inaccurate, though the MDS cognitive performance scale has been found to correlate well with the Folstein MMSE. Although an estimated 35% of older patients present with delirium upon acute admission, there are a number of issues with trying to measure it. It may be preferable to move away from delirium specifically and look at a more general cognitive assessment process measure. Since cognitive assessment is required, no one will admit that one has not been performed, and so we could consider testing the patients ourselves and comparing our results to the facility's results (i.e., if we find a cognitive problem but the facility says they did a cognitive assessment and found no problem, then there is a quality issue).

We added a new process indicator for cognitive assessment, and refined the definition of the delirium indicator to reflect change in delirium over time.

5. Physiology

Duration of Symptoms
Since symptoms will have differing times until resolutions and differing severities, each will have to be asked separately, and we will ask duration separately from severity. The key symptoms are cough, SOB, fatigue, and possibly sputum production. Chest pain is seen only in about 50% of pneumonia patients, and the prevalence and severity of chest pain decreases with increasing age. Again, it may be hard to tie symptoms and symptom severity to quality of care, or even to changes in treatment. For now, let's assume that being symptom-free is a preferable outcome regardless of whether or not the treatment itself alleviated the symptoms. An alternative measure could be "symptom-free at time x".

B. Additional Indicators Suggested by Panelists

1. Process

Time Until Begin Antibiotics
Some patients with advance directives may have specified that they do not want antibiotics. We will need to account for this when using this measure. The same will be true for mortality and hospitalization measures for patients with advance directives.

Advance Directive Discussion with Doctor
We want the indicator "advance directive discussion" to mean more than a simple box checked off at admission, and more than a simple DNR note.

We refined the definition of the indicator such that it must require a doctor's note in the chart that advance directives were discussed with patient.

Appropriate Antibiotic Use
This may be difficult to measure, particularly if the antibiotics for NHAP are different than those for CAP. There are published guidelines for these, however, and would not require determining etiology of infection.

Blood Cultures
JAMA article from last November showed that receiving a blood culture in 24 hrs prior to antibiotics was among 4 quality indicators that improved mortality at 30 days. Blood cultures aren't usually done in the SNF/NH setting because they are expensive (equivalent to one day of care), but if SNFs are substituting for hospital care, then they have to be held to the same standard. This indicator does not apply to home health care, however.

2. Physiology

Time to Reach Stability
For this indicator we really mean "stable at discharge" as a way to find problems with patients discharged too early to a non-institutional setting. We could define stable using JAMA article (Holm and others) criteria: systolic BP >100, pulse rate <100, respiratory rate <24, oxygenation >90%, return to baseline mental status. The criteria could be modified to better reflect an aged population. The probability of a patient being unstable at discharge is low because most pneumonia patients getting post-acute care are there because they are deconditioned. It may be more appropriate to measure "deconditioned at discharged".

We refined the definition of the indicator to reflect stability at discharge from SNF.

C. Measures

Health-Related Quality of Life
The SF-36 while common, does not apply to institutionalized patients very well (questions regarding vacuuming, golf), and some patients are offended by it. There is a shorter version, the SF-12. The SIP is a very good instrument, but is very long (136 questions). There is also a shorter version (68 questions). The QWB is 50 questions, self-report, and takes 12 minutes to administer. Other measures are the Life Satisfaction Index, and the Re-Integration with Normal Living Scale. The big problem with all these measures will be their applicability to a frail population.

Symptoms
Generally, these take the form "Do you have x - Yes/No", and if yes then apply a 5-point Likert scale for severity. There is usually a good distribution of responses on the scale. For patients who are cognitively impaired, a proxy report for yes/no is still OK but we might have to develop specific criteria to assess severity (e.g., SOB at rest, SOB while walking, SOB while going up stairs).

CLINICAL PANEL MEETING NOTES FOR STROKE

December 10, 1998

The following indicators were discussed:

• Quality of Life - Self perceived ability to participate in social/recreational activities
• Utilization - Level of care provided
• Physical Function - Gait and balance
• Mental Health - Patient motivation
• Process - PT/OT/ST evaluation
• Process - Stroke prevention
• Process - Driving evaluation
• Physical Function - Communication

Measures were discussed for:

• Self perceived ability to participate in social/recreational activities
• Level of care provided
• Gait and balance
• Communication

A. Indicators

1. Quality of Life

Self Perceived Ability to Participate in Social/Recreational Activities
Please also refer to the closely related discussion of role resumption under congestive heart failure. In general, the panel as a whole placed a moderate-to-high level of importance on this indicator, yet struggled initially to understand how it was similar to role resumption.

The panel identified the importance of further dissecting quality of life into social function, recreation/leisure and role resumption.

With regard to the latter, the main emphasis was on household tasks and caregiving roles rather than productive roles (i.e. employment), although it was recognized that many older adults continue to maintain a role in the workforce. Further, several members pointed out that there may be considerable overlap between household tasks and IADL indicators (such that household tasks should be grouped with function).

Self care was divided into ADLs and IADL/Household. Caregiver stress was added as a separate indicator. Self-perceived ability to participate in social/recreational activities was subsequently dropped as an indicator as its main construct was incorporated in other indicators.

There was strong support for patient-identified quality of life goals (and using such instruments as the Canadian Occupational Performance Measure). Yet for patients with stroke in particular, aphasia or neglect may influence self-perception and ability to formulate personalized goals. Additional measurement points concerned the limitations of using a proxy to measure personalized goals.

2. Utilization

Level of Care Provided
The discussion began with an overall recognition that the ability to determine that the level of care needed matched the level of care received is very important. However, it was also widely recognized that such algorithms do not exist. Practically, some delivery systems arbitrarily use a two-week time interval to evaluate response to therapy. The panel conveyed concern over the use of such time windows, particularly when the variation in individual recovery in stroke is often wide. Cost containment seems to pervade the decisions determining level of care. The panel seemed comfortable recommending that for any level of care, therapy should continue as long as the patient demonstrates improvement.

Nursing home residence was added as an indicator.

3. Physical Function

Gait and Balance
This discussion began attempting to disentangle Gait and Balance from mobility measures encompassed in the more global function and self-care measures (i.e., is this a separate construct from ambulation/mobility?). Some degree of redundancy was identified. In addition, several of the panel members pointed out that the ability to perform safe transfers was the most important factor determining return to home. Gait speed has also been correlated with safety.

As the conversation moved to focus more on mobility, there was some interest in dividing mobility into bed mobility, transfer mobility (sit to stand) and gait. Additional distinctions were made between home mobility and community ability (longer distances, use of bus or escalator).

The panel supported separate indicators: (1) Gait and Balance, (2) Mobility to include bed transfers and walking, and (3) Mobility to include community mobility.

4. Mental Health

Patient Motivation
This indicator was rated fairly highly as many of the panel members recognized its central role. Additional related constructs were discussed, including sense of self-efficacy, work ethic, and sense of optimism. Next there was interest on the perspective from which this indicator would be assessed -- patient, nurse or therapist. The role cognitive function, patient expectation and family support were identified as key factors that may influence motivation. In the particular case of stroke, neglect syndromes may also influence motivation and the deficit may not be perceived.

Concern was raised, however as to whether this would fit the criteria of a quality indicator. Namely, can external quality agencies hold institutions accountable for patient motivation? Patient rights would dictate that they have the right to refuse therapy.

The discussion shifted towards a related and more operational question: What should the evaluation of a patient not participating in therapy consist of? Initially, this might involve starting by talking with the patient to determine their reason(s) (and documenting the reasons). During this conversation, depression and neglect could be assessed. The second stage of the evaluation would consist of physician evaluation of the patient, not only to further elucidate the first stage but also to evaluate whether there were medical or psychiatric contributing factors. The third stage would involve convening a multidisciplinary team to formulate a customized plan that may or may not necessitate the involvement of a mental health professional.

5. Process

PT/OT/ST Evaluation
There was strong support for the addition of separate indicators for: (1) Physical Therapy Evaluation; (2) Occupational Therapy Evaluation; (3) Speech Therapy Evaluation.

In addition, the indicator for integrated, multidisciplinary care was changed to interdisciplinary conference.

Stroke Prevention
The Anticoagulation process measure was modified to encompass stroke prevention more broadly.

Driving Evaluation
The discussion for adding a process measure on driving evaluation was brief. This was categorized as a higher level of function -- community ability -- and would be targeted towards those who were ready for reintegration into the community (and who were driving prior to their stroke). There was support for including driving evaluation as a component of community mobility (see discussion above). It was recognized that this process measure would apply to a minority of patients (estimated at 10-15%).

6. Physical Function

Communication
The discussion of communication assessment quickly proceeded to measurement assessment as there was little disagreement as to the importance of the indicator and the time remaining was short.

B. Measures

Self Perceived Ability to Participate in Social/Recreational Activities
With regard to measures, the Nottingham Health Profile was not rated highly initially, yet this instrument has been used extensively in Britain and has been compared directly with the SF-36 in the setting of Stroke. It contains 45 questions and takes approximately 10 minutes to complete. There was interest in learning more about this instrument as it contains a number of the domains listed above. One identified limitation was that it places a large emphasis on distress, moreso than SIP or SF-36. The SIP has also been used in the setting of stroke and has been modified as such.

Level of Care Provided
With regard to measurement, one suggestion was made to measure days until long-stay nursing home placement. Alternatively, another measure might be number of days in an inpatient setting (hospital, SNF, rehabilitation facility) (where fewer is better). The measure would need to be designed to capture individual patients who were “bouncing back” or experiencing repeat admissions to SNF.

Yet another suggested measure was the number of days until the person returns home (which would include total number of days in rehab, and SNF). Such a measure might be problematic unless individuals who resided in a nursing home prior to their stroke could be readily accounted for.

A similar measure might be designed looking at total number of Medicare days (SNF, rehabilitation) or total number of inpatient days (Medicare days plus non-Medicare days to capture the long stay nursing home group).

Gait and Balance
In terms of measurement, there was a brief discussion of the use of self-report versus performance measures. Several members voiced that the Berg balance test was too lengthy, requiring 20 minutes whereas the Tinetti measure could be done in under 10 minutes. Apparently the Rivermead instrument has not been shown to be predictive of falls.

Communication
The CADL (Communication Abilities of Daily Living) has been validated on both left and right sided stroke. Administration time varies from 30-60 minutes. A short form has not been designed but it may be possible to abstract 1-2 subtests.

The Functional Communication Profile is an older measure that requires advanced training for scoring.

The ASHA Functional Communication Measure is new and validity and reliability studies have been conducted. It looks more closely at daily tasks and breaks down the communicative skills (e.g., going to a movie theater). At present there is only data for patients with aphasia and there is ongoing work on right hemisphere communication. Administration time is approximately 30 minutes.

The Communicative Effectiveness Index (CEI) is a self-report instrument used only in aphasia and has not been used in right hemisphere communication.

The Edinburgh Functional Communication Profile is quick but was not familiar to the panel.

The FIM does include two relevant measures -- comprehension (includes both hearing and understanding) and expression.

The RIC-FAS (Rehabilitation Institute of Chicago) has expanded the FIM communication measure to include reading and writing; speech production and community function. The measure has been used by other centers. Reliability and Validity are only known for the first version -- the measure is now in its 5th version.

CLINICAL PANEL MEETING NOTES FOR BACK & NECK

December 9, 1998

The following indicators were discussed:

• Physiology - Pain, pain intensity, symptom severity
• Physiology - Numbness
• Quality of Life - Social function
• Quality of Life - Days of limited activity
• Process - Neurological checks
• Mental Health - Depression
• Mental Health - Somatization
• Satisfaction - Patient satisfaction

Measures were discussed for:

• Pain
• Physical function

A. Indicators

1. Physiology

Pain, Pain Intensity, Symptom Severity
Symptom severity and pain refer to the same construct; therefore, it is necessary to combine the two into one indicator, called pain. In addition to pain intensity, we should also consider the impact of pain on life functions/activities.

We refined the definition of the indicator such that symptom severity and pain intensity would be replaced by two new indicators: (1) back and neck pain intensity and frequency and (2) leg and arm pain intensity and frequency.

Numbness
Numbness and tingling can be as significant as pain in evaluating symptoms of patients with back and neck conditions such as spinal stenosis. When asked about symptoms, patients often will respond that they are not experiencing pain, but rather a numbness, tingling, or "discomfort." However, numbness is most likely not as sensitive to differences in quality of care (as is pain).

2. Quality of Life

Social Function
Social function seems to overlap a great deal with role resumption (see CHF discussion). However, social function and role resumption do not necessarily refer to the same construct. For example, social function might include such activities as participating in a weekly card game, but this activity would not be considered a "role." A valuable quality indicator with respect to social function would be "return to previous social function" or "perception of self returning to prior situation." However, this may be a difficult indicator to measure because change in social function may or may not be the result of the care that is provided.

Days of Limited Activity
Days of limited activity is a very common, well-regarded quality indicator for PAC. It has high face validity and is frequently cited in the literature. However, there are several issues with using this indicator as a measure of PAC quality. It is most commonly measured with the following item, "During how many of the past 30 days was your activity limited due to your illness?" with responses being 0, 14, 28. Because this measure is not continuous, it is less informative. Furthermore, the term "limited" can be difficult to define. One person's experience of a "limited activity day" may not be the same as another's. Yet another issue of concern is that the days of limited activity might be due to physician's order, rather than the patient's actual status/ability. Measuring limited activity days does have the advantage of picking up clinically meaningful change over time, but generally does not provide any additional information over and above function or social function.

A new indicator, role resumption, was added to the rating sheet. (See discussion of indicators for CHF.)

3. Process

Neurological Checks
Neurological checks in a post-acute setting do not necessarily reflect quality of post-acute care; this is something that should be done in the acute setting. However, ongoing sensory/motor checks are more indicative of quality in the PAC setting. These checks are generally done during the process of other evaluations, such as physical therapy.

A new indicator, physical therapy evaluation, replaced physical therapy use. A new indicator, occupational therapy evaluation, was added to the rating sheet.

4. Mental Health

Depression
It is generally believed that a measure of depression must exist for a condition such as back and neck. However, the incidence of major depression in patients with chronic back pain is much lower than in patients with CHF. Of more consequence than major depression is less severe depression or a general feeling of sadness. This level of depression may be more sensitive to changes in quality of PAC for back and neck patients.

Somatization
For back and neck patients, somatization is of special importance. This is a disorder which can significantly affect (exaggerate) health care resource utilization. However, recognizing and diagnosing these patients can be especially difficult.

5. Satisfaction

Patient Satisfaction
Patient satisfaction is an indicator that is frequently used and well liked. A differentiation must be made between whether we examine patient satisfaction with care or patient satisfaction with where they are currently (or current status).

We refined the definition of the indicator such that patient satisfaction would refer to patient satisfaction with care, rather than patient satisfaction with current status.

6. Utilization

A new indicator, total inpatient days, was added to the rating sheet.

B. Measures

Pain
SF-36 -- Has only two items that measure pain severity. Visual Analogue Scale (VAS) -- Good, but non-specific.

Physical Function
Recommended measures of physical function in back and neck patients are: Oswestry, FIM, SF-36, and self-reported walking distance.

[Top of Page]

APPENDIX E. Final Ratings and Dropped Quality Indicators

Table of Contents

CHF -- Final Ratings

Pneumonia -- Final Ratings

Stroke -- Final Ratings

Back & Neck -- Final Ratings

Discussion of Dropped Quality Indicators

DISCUSSION OF DROPPED QUALITY INDICATORS

As noted in Chapter III, several quality indicators were rated highly by the clinical panel (selected by four or more panel members) but subsequently dropped for various reasons. Following is a list of the indicators that were dropped, including the rationale for dropping them.

A. Global

• Cognition (Mental Health): Although cognition was rated highly by the panels for both pneumonia and stroke, it is more appropriately regarded as a risk adjuster than a measure of post-acute care quality. In all four of our instruments, the Pfieffer Short Portable Mental Status Questionnaire is included for risk adjustment rather than a quality indicator.

• Days of limited activity (Quality of Life): Days of limited activity, rated highly by both the pneumonia and back and neck panels, is encompassed within the broader "role resumption" and "social function" quality of life indicators. Further review of this measure also revealed that it may not be a highly sensitive or accurate measure of limitation because of the tendency of respondents to round the number of days up or down to a figure approximating weeks.

• Recreation/Leisure (Quality of Life): Recreation/Leisure is encompassed within the broader "social/role function" quality of life indicator, measured using the Resumption of Normal Activities questions.

• Home health care use (Utilization): The interpretation of home health care utilization, as a measure of the quality of care, is uncertain. In some cases, high utilization rates may reflect good quality of care with very close follow-up of patients, while in other circumstances it may reflect persistence of patient illness and thus poor quality of care. As such, we did not believe that it clearly and accurately reflects the quality of post-acute care, despite being highly ranked by our panel.

• Skilled nursing facility admission (Utilization): Similar to home health use, SNF admission could be a reflection of appropriate or inadequate post-acute care. Admission to SNF alone does not indicate quality of post-acute care.

• Rehabilitation facility admission (Utilization): See discussion of skilled nursing facility and home health care admission above.

• Caregiver satisfaction (Satisfaction): Caregiver satisfaction was rated highly by the panels for all four conditions. Due to the added data collection burden associated with a caregiver assessment, we dropped this indicator. Recognizing the importance of satisfaction overall, however, we assess patient satisfaction in the patient report components of the instruments (and therefore proxy respondents when a proxy is necessary who are arguably the most burdened caregivers).

• Patient education (Process): Documentation of patient education is unlikely to be reliably documented in medical records. Additionally, the components of adequate or good patient education are extremely difficult to define to develop a meaningful measure. We therefore dropped this as a quality indicator.

B. CHF

• Medication review (Process): Similar to patient education, documentation of medication review will most likely not be consistently available in the medical record. The panel ranked both medication review and assessment of patient compliance with medication very highly. Because of similarities between these two measures, and the implicit review of medications that would occur in an assessment of medication compliance, we left the latter indicator in as a process measure. However, we are uncertain to what degree such information will be documented in the medical record.

C. Back & Neck

• Days in bed (Quality of Life): Similar to days of limited activity, days in bed is encompassed within the broader "role resumption" and "social function" quality of life indicators.

• Total combined PAC (Utilization): This indicator is similar to another indicator, total inpatient days, which is included in the final indicator list.

• Repeat spinal surgery (Utilization): For some patients, repeat spinal surgery may be an appropriate and necessary treatment; while for others, repeat surgery might indicate poor quality of post-acute care. Furthermore, the need for additional surgery is more likely to reflect quality issues related to earlier surgery than quality problems in the post-acute care received.

D. Pneumonia

• Time to reach stability (Physiology): Stability is more sensitive to hospital care than to post-acute care; the majority of patients are stable before transfer to a post-acute facility. This measure was chosen by the panel following a discussion about how Medicare HMOs may bypass acute hospitalization and treat pneumonia patients directly in subacute care units. Such subacute care should therefore scrutinize processes and outcomes of care that are more traditionally hospital-based. However, other measures were chosen for this same purpose (e.g., obtaining blood cultures, appropriate antibiotic use) and should suffice. Such subacute care that bypasses acute hospitalization is likely to affect only a small minority of all pneumonia patients receiving post-acute care.

• Cognition evaluation (Process): While evaluation of cognition in general and delirium in particular were viewed as important for post-acute care of persons with dementia, this indicator was dropped due to concerns that it could not be uniformly assessed from providers' entries to medical records.

E. Stroke

• Gait and balance (Physical Function): No validated self-report measures specifically designed to assess gait and balance exist.

• Level of care provided (Utilization): As with home health care utilization, SNF and rehabilitation hospital utilization, the panel was concerned that transfers to higher levels of care may be a reflection of post-acute care quality. Again, this is an ambiguous measure because transfers to higher levels of care may be necessary and reflect good care in some circumstances while representing poor care in others. Given this ambiguity, this indicator was also dropped.

• Medicare-covered inpatient days (Utilization): This indicator is similar to another indicator, total inpatient days, which is included in the final quality indicator list.

• Motivation evaluation (Process): Motivation, though selected by four of the stroke panel members, is not measurable using a standardized instrument and therefore post-acute care providers cannot be expected to reliably measure and record it.

[Top of Page]

APPENDIX F. Final Quality Indicator Ratings Across All Conditions

[Top of Page]

APPENDIX G. Barthel Conversion Codes

Table of Contents

Barthel Index Replicated Using the MDS 2.0 (10/94n)

Barthel Index Replicated Using the MDS PAC (Draft 8)

Barthel Index Replicated Using the OASIS (Draft B1)

NOTES

ADL Self-Performance (#1 Feeding, #2 Transfer, #3 Grooming, #4 Toileting, #6 Walking, #7 Stairs, #8 Dressing)

0 - independent
1 - supervision
2 - limited assistance
3 - extensive assistance
4 - total dependence
8 - activity did not occur

Bathing Self-Performance (#5 Bathing)

0 - independent
1 - supervision oversight help only
2 - physical help limited to transfer only
3 - physical help in part of bathing activity
4 - total dependence
8 - activity did not occur

Continence Self-Performance (#9 Bowel Continence, #10 Bladder Continence)

0 - continent
1 - usually continent
2 - occasionally incontinent
3 - frequently incontinent
4 - incontinent

NOTES

3 Day ADL Self-Performance (#1 Feeding, #2 Transfer, #3 Grooming, #4 Toileting, #5 Bathing, #6 Walking, #7 Stairs, #8 Dressing)

0 - independent
1 - set up help only
2 - supervision
3 - limited assistance
4 - extensive assistance, with all episodes involving a 1 person physical assist
5 - extensive assistance with at least one episode involving a 2+ person physical assist
6 - total dependence
8 - activity did not occur

Capacity to Perform Instrumental Activities of Daily Living (#7 Stairs)

0 - independent
1 - set up help only
2 - supervision
3 - limited assistance
4 - moderate assistance
5 - maximum assistance
6 - total dependence

Bowel Continence (#9 Bowel Continence)

0 - continent
1 - continent with ostomy (NA)
2 - usually continent
3 - occasionally incontinent
4 - frequently incontinent
5 - incontinent
8 - did not occur

Bladder Continence (#10 Bladder Continence)

0 - continent
1 - continent with catheter (NA)
2 - usually continent
3 - occasionally incontinent
4 - frequently incontinent
5 - incontinent
8 - did not occur

[Top of Page]

APPENDIX H. Members of Second Expert Panel

[Top of Page]

APPENDIX I. Notes from Specialist Conference Calls

Table of Contents

CHF Conference Call -- Martha Radford, M.D.

Pneumonia Conference Call -- Thomas Marrie, M.D.

Back and Neck Conference Call -- Richard Deyo, M.D.

CHF CONFERENCE CALL
Martha Radford, M.D.

• In addition to the disease-specific measures already in the instrument, Dr. Radford suggested measures related to beta blockers, digoxin, and anticoagulation. In a follow-up correspondence she also queried whether a summary measure of "reasonable medical management" could be constructed using a list of commonly-used classes of drugs in heart failure (ACEI, ARB, diuretics, beta-blockers, anticoagulation, digoxin), allow various combinations of these drugs to constitute reasonable medical management.

• Dr Radford suggested including measures related to education and patient empowerment because of the documented association rehospitalization rates. However, it is not clear how to go about measuring this construct. She recommended an article by Carol Ashton for further information on this topic. She noted some unpublished work at Yale suggesting that telephone education programs (without individual patient monitoring or case management) may decrease hospitalizations. Because both Dr. Radford and the expert panel convened in December 1998 supported some kind of measure of patient education/empowerment, a more thorough review of published literature will be conducted to determine the association of education with important CHF outcomes, as well as exploration of the content of such educational programs and how they might be documented by a post-acute care provider.

• With regard to a summary performance measure for CHF, Dr. Radford stated that this is a "reasonable goal," and that the work we are currently doing may ultimately lead to such a measure. She noted, however, that with the current state of knowledge, it would be difficult to sum the different contents areas of quality into a single score.

• Regarding quality of life measures, Dr. Radford suggested that we consult Dr. John Spertus from the University of Missouri, whose previous work includes the Seattle Angina Questionnaire. Dr. Spertus just published a paper documenting the performance of a CHF specific quality of life instrument (Journal of the American College of Cardiology, April 2000).

• Also suggested assessing whether the patient's diet was addressed, and whether the patient is on a low-salt diet.

• Pointed out that post-acute care should be prevention-oriented. We should measure not only if the patient returned to baseline function, but also if bad outcomes were averted (e.g., re-hospitalization, emergency room visits, worsening of condition, progressive decline). Our instrument may be weak in some of these areas except rehospitalization. For example, our instrument can measure decline in basic ADL function, but not subtle declines in higher level functions which are more likely to occur over short follow-up intervals in heart failure.

• Recommended including more assessment items related to higher order function (such as visiting with grandchildren).

• Recommended a 30-day episode of care for CHF because variation in major outcomes like rehospitalization will be captured in a 30 day period. Too long a period will run into subsequent post-acute episodes in a significant proportion of subjects.

• For case mix adjustment, she recommended consulting work performed in this area by Michael Rich or Harlan Krumholz. In addition, she suggested accounting for renal function, prior hospitalization for any condition, terminal illness other than heart failure, and documentation of dementia. Severity of illness could be determined by counting medications.

• For our study, she recommended including only patients with a primary hospital diagnosis of CHF. She noted that the accuracy of the diagnosis is extremely high when it is noted as the primary discharge diagnosis. 

PNEUMONIA CONFERENCE CALL
Thomas Marrie, M.D.

• Suggested we monitor hospitalizations during the post-acute stay as well as after discharge.

• Remarked that for pneumonia, resolution of symptoms is the primary measurement of success. Blood culture may not be a relevant measure for this population since most patients would be well past the point at which blood culture data is useful.

• Suggested including a measure related to assessment and evaluation of adverse effects of antibiotics (e.g., gastrointestinal symptoms which are very common with macrolides), as well as whether the provider addressed the need for smoking cessation program. Also suggested we add a measure of advance directives.

• For the hospital chart review, he recommended including measures related to timely administration of antibiotics, admission blood cultures, and maintenance of function in the acute care facility. These would be used primarily as risk adjusters, allowing for example, poor outcomes to be attributed to poor hospital care instead of post-acute care.

• Dr. Marrie was not aware of a summary performance measure for pneumonia.

• Recommended a 30-day episode of care for pneumonia, but noted at 30 days more than half are still symptomatic, though usually symptoms have significantly improved. Dr. Marrie also noted that 20% of pneumonia subjects are symptomatic for some time period before contracting pneumonia. Many may have chronic symptoms due to other diseases (e.g., COPD) that are always present and are not related to the pneumonia itself. Thus baseline measures of symptoms need to be very clear so that they actually reflect the patient's status in a stable, pre-hospital condition.

• Suggested stratifying patients into categories based on a summary measure of severity of risk factors. Michael Fine's scoring mechanism for mortality and LOS from the PORT may or may not be applicable to post-acute setting, but the technique is worth considering for risk adjustment. Whether all the items in the Fine score are readily available from the hospital chart needs to be determined.

• For case mix adjustment, recommended accounting for use of ventilator, ICU, critical illness, hospital complications, and neuropathy.

• When measuring re-hospitalization rates, be sure to separate hospitalizations for pneumonia from hospitalizations for other conditions.

• For our study, he suggested including patients with a primary or secondary diagnosis of pneumonia, but recommended verifying the diagnosis with x-ray report.

BACK AND NECK CONFERENCE CALL
Richard Deyo, M.D.

• Should ultimately distinguish between medical and surgical back and neck conditions; there might be important differences in length of stay or costs for the different types of conditions. Also, change in symptoms over time may be fundamentally different since surgical patients might be expected to actually improve, rather than just return to baseline, whereas the goal for medical patients might be returning to baseline.

• Suggested rephrasing questions as "Because of your back pain or leg pain (sciatica)…" or "Because of your back problem…" in reference to the pain questions since elderly patients are likely to experience pain from a number of different sources (e.g., DJD, headache).

• For disease-specific measures, he recommended consulting the Rowland and Morris Disability Questionnaire or the Oswestry Questionnaire, which also includes some higher function assessments such as IADLs.

• With respect to responsiveness for detecting clinically important change, he indicated that pain measures are quite responsive to change, but functional measures tend to be less responsive. Overall, he felt that the measures we chose will be responsive to change.

• For back and neck conditions, we will want to assess pain on admission to post-acute care, but not premorbid pain. For pain, the baseline measure will be at admission to post-acute care, since pain and functional impairment may following surgery. Baseline pain would therefore be expected to be responsive to the acute hospital stay (surgery), while the pain following a surgical procedure would be sensitive to the quality of post-acute care.

• Recommended a 1-month or 3-month episode of care for back and neck conditions. He did not think the episode should be any longer than 3-months but was uncertain about whether a 1 or 3 month episode was preferable.

• For case mix adjustment, he recommended capturing whether the patient had a spinal fusion; complication rates and post-acute care placement are much higher for patients with spinal fusion. He also suggested controlling for the number of levels on the spine operated on, comorbidity, wound infections, DVT, new neurologic deficits, other organ complications, perioperative myocardial infarction, and new bowel or bladder dysfunction. Will want to stratify on neck vs lumbar, and on elective surgery vs traumatic injury.

• With respect to a summary performance measure, Dr. Deyo indicated that this has not been done due to the ambiguity about the ideal process of care. He recommended assessing quality through multiple outcome measures.

• For our study, he suggested including only patients with a primary hospital discharge diagnosis of lumbar spinal stenosis. This is likely to be a common diagnosis in an elderly population and would allow a more reasonable comparison of the quality of care in different post-acute settings by assembling a more homogeneous study population.

• With respect to the pain questions, he recommended using just two measures: "How bothersome is the pain?" and "How frequent is the pain?" He noted the disconnect between severity or frequency ratings of pain and how bothered people really are by pain. He noted that asking about how bothersome symptoms are really gets at the impact on the person's life and thus their function while severity and frequency alone do not.

• He also recommended a 10-point (not 20) visual analog scale when asking the patient to rate how much the pain bothers him/her.

[Top of Page]

APPENDIX J. Issues for Second Expert Panel Discussion

9:30 - 10:30 Discussion Session

1. The first expert panel rated a long list of both global and disease specific quality indicators. In the final rank list (Appendix F of this report), the majority of the indicators were global and these indicators served as the basis for the quality measures included in the instruments. Please comment on the balance between the global and disease-specific measures in the patient survey and chart review instruments for the respective condition.

2. Do you think that the measures chosen reflect the quality constructs recommended by the initial expert panel (please see Appendix F)?

3. Are there measures relevant to the post-acute care of older adults that were not included in these quality measurement instruments?

10:45 - 12:15 Discussion Session

1. In your opinion, can the current quality measures be feasibly administered in the respective post-acute care settings?

2. Do you anticipate that these measures will be responsive for detecting clinically important change?

3. Do you foresee problems with floor or ceiling limitations? For example, subjects who cannot perform a given self-care function at baseline cannot recover that function; therefore creating a floor effect?

4. The purpose of this study will be to compare quality of care across settings. Do you have recommendations for sampling comparable populations in different post-acute settings?

5. With respect to selection criteria for study subjects, which post-acute care patients would be most appropriate for quality of care comparison (i.e., not all patients have the condition of interest listed as their primary diagnosis; what techniques could be used to identify a sample of comparable patients across the three different post-acute settings so as to make quality comparisons across settings?). How could potential subjects be identified in the absence of information derived from the MDS/MDS-PAC/OASIS?

1:15 - 2:15 Discussion Session

1. What defines a discrete episode of care for this condition in post-acute care? In particular, what defines the end of a post-acute episode (e.g., rehospitalization? a certain length of time such as 30 days? discharge from the post-acute care provider?). This definition will determine when the follow-up measures are administered.

2. Ultimately these measures will be used for quality comparisons between sites of post-acute care. What variables would you recommend for case mix adjustment?

3. Given the importance of assessing quality, do you feel that the current instruments impose excessive response burden on older patients?

2:30 - 3:30 Discussion Session

1. What criteria would you suggest for determining the need for a proxy respondent? Do you have any experience with particular mental status instruments in post-acute care?

2. Do you feel that the measures in the chart review instruments can be reasonably abstracted from the post-acute care and hospital charts?

3. Do you feel that the quality measures accurately capture key elements for comparing quality for which post-acute care providers can be held accountable?

3:45 - 5:00 Discussion Session

1. Do you believe it is possible to construct a summary performance measure using the measures in these instruments? If so, how would such a measure be constructed and the various domains weighted?

[Top of Page]

APPENDIX K. Project Timeline

[Top of Page]