Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities

Hello! This is Troy Hillman, Director of Quality and Health Policy at the American Medical Rehabilitation Providers Association (AMRPA). AMRPA is the national trade association representing over 800 inpatient rehab facilities, which includes both freestanding inpatient rehab facilities and rehabilitation units of acute care hospitals. AMRPA appreciates the opportunity to comment on CBE #4630: Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities. As we have stated in previous comments as part of the Fiscal Year 2024 IRF Proposed Rule, AMRPA did not support the adoption of this measure into the IRF QRP, and the concerns that we had about this measure at that time are consistent with the concerns we have today about this measure being considered for endorsement.

First, this measure should be identified as competing with two currently endorsed measures: CBE #2635: Discharge Self-Care Score, as well as CBE: #2636: Discharge Mobility Score. As identified in the PQM Measure Evaluation Rubric, the measure developer must indicate how the new measure is superior to existing measures as part of the importance criteria for endorsement, and we ask that the PQM E&M Committee require an evaluation of competing measures as part of the consideration for endorsement of this measure.

AMPRA also has concerns about the imputation method utilized for calculating assessment data. The development of this measure for use in the IRF QRP suggested that the imputation method was necessary for those instances where functional assessment data was considered “missing.” Except for the dash value, where no information was available for the assessment, all other “activity not attempted” codes are indicative of an affirmative functional assessment value and should not be considered as missing. Instead, each code represents a circumstance where a patient is not capable of performing the activity, suggesting that their performance should be considered at least as the most severe level, or code 01 “dependent.” AMRPA recommends that the PQM E&M Committee evaluate the imputation method, and whether this accurately represents the patient's functional status at the time of assessment.

AMRPA also asks that you consider whether this measure meets the feasibility criteria required for endorsement, where the measure developer considered the people, tools, tasks, and technologies necessary to implement this measure. While the data elements utilized in the measure calculations are already part of the IRF-PAI, IRFs would require a significant amount of investment in technology to implement the imputation methodology required to manage the performance of this measure. Providers that are unable to calculate imputed values on their own have limited ability to identify what the patient's measure scores may be in order to determine what risk adjusted target value is required to contribute towards a positive result.

Finally, AMRPA members are concerned about the suggestion that this is a cross-setting measure and the unintended consequences that may result from such a designation. This measure differs across post-acute care settings as the risk adjustment methodology utilizes setting-specific covariates and setting-specific coefficients. For example, each setting has a model intercept for the expected discharge function score for each patient to meet or exceed, in order to perform well on this measure. For IRFs, that intercept is a little over 34, while for SNFs the model intercept is 30, and for LTACs the model intercept is 14. This suggests that depending on a setting, a patient with the exact same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation, and LTACs having the lowest expectation. To truly be considered a cross-setting measure, AMRPA members believe that the patient expectations should be consistent, regardless of setting. 

Additionally, in suggesting this as a cross-setting measure that is publicly reported, these measures have the unintended consequence of potentially being used to limit patient access to certain settings based upon performance results. While CMS and the measure developers documented concerns about providers denying access to certain patients who may not perform well on this measure, AMRPA questions whether consideration was given for referral sources utilizing this information to direct patients to alternative settings which may not provide the appropriate level of services to produce high quality outcomes. We ask that the committee consider these unintended consequences when evaluating this measure for endorsement. Again, AMPRA appreciates the opportunity to comment on these, and will be submitting additional written comments for consideration of this measure for endorsement. Thank you very much.

On behalf of Carolinas Rehabilitation-Patient Safety Organization (PSO), commonly referred to as EQUADR^SM, we appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  EQUADR^SM is a PSO supporting the performance improvement of inpatient rehabilitation hospitals across the country.

This measure does not meet the requirements for endorsement as the measure is administratively difficult to manage, produces results inconsistent with other endorsed measures, and could produce unintended consequences impacting access to IRF care.  The concerns are detailed below for PQM E&M Committee’s consideration of these reasons when evaluating this measure for endorsement.

First, clinicians are struggling to understand and manage performance on this measure.  While the data elements utilized in the ‎measure are part of the Inpatient Rehabilitation Facility – Patient Assessment Instrument (IRF-PAI), the imputation methodology to manage ‎the performance of this measure requires an infeasible ‎investment in technology and training to implement.  Providers that are unable to calculate imputed values on ‎their own have limited ability to identify what the patients’ measure scores are ‎and what the risk-adjusted target values are to create a positive ‎result.  ‎ The imputation method is unnecessary, as existing endorsed measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients) do not require imputed values and instead utilize values that are representative of the patient’s functional status at admission and discharge.  This measure does not meet the feasibility criteria required for ‎endorsement, as this measure requires impractical investment in the people, tools, tasks, and ‎technologies necessary to implement this measure. 

Second, this measure produces results that are inconsistent with two currently endorsed functional measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients).  The existing endorsed measures utilize all the functional items in the Self-Care and Mobility domains and produce an accurate representation of the functional improvement and discharge functional status of every IRF patient.  The Discharge Function Score measure utilizes 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional assessment of IRF patients.  Utilizing less than half of all functional items included in the IRF-PAI and existing endorsed measures produces values that are inconsistent and are not representative of the functional improvements and discharge functional status the patients that achieve as part of their IRF care. We do not believe that the new measure is superior to existing measures and should not replace the existing measures.  This measure does not meet the importance criteria required for ‎endorsement.

Finally, there are unintended consequences that may result from the endorsement of this measure.  This measure is identified as “Cross-Setting”, yet it differs significantly across post-acute care settings.  While the functional items included in this measure are similar across post-acute care settings, the measure calculations and risk-adjusted patient expectations are very different.  The measure considered for endorsement for Home Health has a different developer and different imputation and risk-adjustment methodology.  Depending on post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation and Long-term acute care hospital (LTCH) patients having the lowest expectation.  To be considered a “cross-setting” measure, the measure calculations and patient expectations should be standardized regardless of setting. Additionally, since these measures are publicly reported on Care Compare, there is the unintended consequence of measure values being used to limit patient access to certain settings based upon reported results.  The characterization of that this measure as cross-setting measures implies that comparison of results can be made between IRFs, SNFs, LTCHs and Home Health providers.  Payers and referral sources already are using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes.  This measure does not meet the use and usability criteria required for ‎endorsement.

As detailed above, this measure does not meet all the criteria necessary for endorsement, and EQUADR^SM therefore asks the PQM E&M Committee to not support the endorsement of this measure.  

On behalf of Gundersen Health System, we appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  

This measure does not meet the requirements for endorsement as the measure.  It is administratively difficult to manage, produces results inconsistent with other endorsed measures, and could produce unintended consequences impacting access to IRF care.  Please consider details below for PQM E&M Committee’s consideration of these reasons when evaluating this measure for endorsement.

Performance on this measure is difficult to understand. The imputation methodology is not able to be replicated nor are we able to pinpoint patient specifics to fully understand how improvement is able to be made on an individual patient level.  This makes it meaningless as a measure to work toward improvement. The Federal Register Final Rule explains the imputation recodes missing functional status data to the most likely value and produces an estimate of scores on items with missing values.  This data is not representative of the patients’ actual function and has the risk of jeopardizing representation of the functional benefit inpatient rehabilitation had on patients served by inpatient rehabilitation. The time, resources and technologies needed to implement and understand this measure would be far higher than the value it would provide to meeting a cross-setting metric.

In addition, this measure utilizes only a fraction of the functional items included in the IRF-PAI.  Therefore, the measure is not representative of the functional improvements and discharge functional status of patients receiving care in an Inpatient Rehabilitation Facility.  This only serves to severely undercut the value of IRF level of care.  It also does not have a cognitive or neuropsychology element that many times can be a primary focus of IRF level of care.  The public reporting of this measure on Care Compare does not represent accurately the benefit of the functional gains our patients make on our Rehab Unit.  The unintended consequences of displaying this information for consumers and other public entities such as insurance companies can lead to barriers to access the most appropriate level of care. Payers and referral sources are already using information to direct patients to alternative settings which might not be in the patients’ best interest. When it is difficult to understand, replicate and doesn’t represent the actual benefits of care provided.  There are varied measure calculations and risk-adjusted patient expectations in the various post-acute care settings.  Depending on the post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations and the IRF setting patients having the highest expectation.  This does not serve to be a cross-setting measure. The risk to limit patient access to care is a big concern and should be considered as a reason not to endorse this measure. 

As I outlined here, this measure does not meet all the criteria necessary for endorsement, and Gundersen Health System therefore asks the PQM E&M Committee to not support the endorsement of this measure. 

Sincerely,

Deborah Head, OTR/L Rehab Program Manager

Gundersen Health System  

On behalf of MossRehab we appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF). 

This measure does not meet the requirements for endorsement as the measure is administratively difficult to manage,  results produced are inconsistent with other endorsed measures, and could produce unintended consequences impacting access to IRF care.  

Clinicians are struggling to understand and manage performance on this measure.  While the data elements utilized in the ‎measure are part of the Inpatient Rehabilitation Facility – Patient Assessment Instrument (IRF-PAI), the imputation methodology to manage ‎the performance of this measure requires an infeasible ‎investment in technology and training to implement.  Providers that are unable to calculate imputed values on ‎their own have limited ability to identify what the patients’ measure scores are ‎and what the risk-adjusted target values are to create a positive ‎result.  ‎ The imputation method is unnecessary, as existing endorsed measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients) do not require imputed values and instead utilize values that are representative of the patient’s functional status at admission and discharge.  This measure does not meet the feasibility criteria required for ‎endorsement, as this measure requires impractical investment in the people, tools, tasks, and ‎technologies necessary to implement this measure. 

This measure produces results that are inconsistent with two currently endorsed functional measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients).  The existing endorsed measures utilize all the functional items in the Self-Care and Mobility domains and produce an accurate representation of the functional improvement and discharge functional status of every IRF patient.  The Discharge Function Score measure utilizes 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional assessment of IRF patients.  Utilizing less than half of all functional items included in the IRF-PAI and existing endorsed measures produces values that are inconsistent and are not representative of the functional improvements and discharge functional status the patients that achieve as part of their IRF care.  Additionally, the public reporting of this measure on Care Compare shows that measure performance is inconsistent with the two other endorsed measures, and some IRFs are reporting that patients who meet expectations for the Discharge Self-Care and Discharge Mobility measures are identified as not meeting expectations for the Discharge Function Score measure.  Because of these inconsistencies, we do not believe that the new measure is superior to existing measures and should not replace the existing measures.  This measure does not meet the importance criteria required for ‎endorsement.

There are potential unintended consequences that may result from the endorsement of this measure.  This measure is identified as “Cross-Setting”, yet it differs significantly across post-acute care settings.  While the functional items included in this measure are similar across post-acute care settings, the measure calculations and risk-adjusted patient expectations are very different.   The measure considered for endorsement for Home Health has a different developer and different imputation and risk-adjustment methodology.  Depending on post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation and Long-term acute care hospital (LTCH) patients having the lowest expectation.  To be considered a “cross-setting” measure, the measure calculations and patient expectations should be standardized regardless of setting. Additionally, since these measures are publicly reported on Care Compare, there is the unintended consequence of measure values being used to limit patient access to certain settings based upon reported results.  The characterization of that this measure as cross-setting measures implies that comparison of results can be made between IRFs, SNFs, LTCHs and Home Health providers.  Payers and referral sources already are using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes.  This measure does not meet the use and usability criteria required for ‎endorsement.

As detailed above, this measure does not meet all the criteria necessary for endorsement, and MossRehab therefore asks the PQM E&M Committee to not support the endorsement of this measure. 

I appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  

This measure does not meet the requirements for endorsement as the measure is administratively difficult to manage, produces results inconsistent with other endorsed measures, and could produce unintended consequences impacting access to IRF care.  The concerns are detailed below for PQM E&M Committee’s consideration of these reasons when evaluating this measure for endorsement.

First, clinicians are struggling to understand and manage performance on this measure.  While the data elements utilized in the ‎measure are part of the Inpatient Rehabilitation Facility – Patient Assessment Instrument (IRF-PAI), the imputation methodology to manage ‎the performance of this measure requires an infeasible ‎investment in technology and training to implement.  Providers that are unable to calculate imputed values on ‎their own have limited ability to identify what the patients’ measure scores are ‎and what the risk-adjusted target values are to create a positive ‎result.  ‎ The imputation method is unnecessary, as existing endorsed measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients) do not require imputed values and instead utilize values that are representative of the patient’s functional status at admission and discharge.  This measure does not meet the feasibility criteria required for ‎endorsement, as this measure requires impractical investment in the people, tools, tasks, and ‎technologies necessary to implement this measure. 

Second, this measure produces results that are inconsistent with two currently endorsed functional measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients).  The existing endorsed measures utilize all the functional items in the Self-Care and Mobility domains and produce an accurate representation of the functional improvement and discharge functional status of every IRF patient.  The Discharge Function Score measure utilizes 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional assessment of IRF patients.  Utilizing less than half of all functional items included in the IRF-PAI and existing endorsed measures produces values that are inconsistent and are not representative of the functional improvements and discharge functional status the patients that achieve as part of their IRF care.  Additionally, the public reporting of this measure on Care Compare shows that measure performance is inconsistent with the two other endorsed measures, and some IRFs are reporting that patients who meet expectations for the Discharge Self-Care and Discharge Mobility measures are identified as not meeting expectations for the Discharge Function Score measure.  Because of these inconsistencies, we do not believe that the new measure is superior to existing measures and should not replace the existing measures.  This measure does not meet the importance criteria required for ‎endorsement.

Finally, there are unintended consequences that may result from the endorsement of this measure.  This measure is identified as “Cross-Setting”, yet it differs significantly across post-acute care settings.  While the functional items included in this measure are similar across post-acute care settings, the measure calculations and risk-adjusted patient expectations are very different.   The measure considered for endorsement for Home Health has a different developer and different imputation and risk-adjustment methodology.  Depending on post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation and Long-term acute care hospital (LTCH) patients having the lowest expectation.  To be considered a “cross-setting” measure, the measure calculations and patient expectations should be standardized regardless of setting. Additionally, since these measures are publicly reported on Care Compare, there is the unintended consequence of measure values being used to limit patient access to certain settings based upon reported results.  The characterization of that this measure as cross-setting measures implies that comparison of results can be made between IRFs, SNFs, LTCHs and Home Health providers.  Payers and referral sources already are using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes.  This measure does not meet the use and usability criteria required for ‎endorsement.

If representing your IRF: 

As detailed above, this measure does not meet all the criteria necessary for endorsement, and [Insert hospital/corporation name] therefore asks the PQM E&M Committee to not support the endorsement of this measure.  

If making an individual statement: 

As detailed above, this measure does not meet all the criteria necessary for endorsement, and therefore ask the PQM E&M Committee to not support the endorsement of this measure.

Kara Simpson, VP Clinical Operations for Ernest Health, representing 30 inpatient rehabilitation hospitals and 7 LTACHs.  On behalf of our organization, we fully agree with all of the points brought forward by Troy and AMRPA with the setting of the Discharge Function Score in QRP across settings.  We do not endorse this for all of the same reasonings.  Thank you.

I appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  

This measure does not meet the requirements for endorsement as the measure is administratively difficult to manage, produces results inconsistent with other endorsed measures, and could produce unintended consequences impacting access to IRF care.  The concerns are detailed below for PQM E&M Committee’s consideration of these reasons when evaluating this measure for endorsement.

First, clinicians are struggling to understand and manage performance on this measure.  While the data elements utilized in the ‎measure are part of the Inpatient Rehabilitation Facility – Patient Assessment Instrument (IRF-PAI), the imputation methodology to manage ‎the performance of this measure requires an infeasible ‎investment in technology and training to implement.  Providers that are unable to calculate imputed values on ‎their own have limited ability to identify what the patients’ measure scores are ‎and what the risk-adjusted target values are to create a positive ‎result.  ‎ The imputation method is unnecessary, as existing endorsed measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients) do not require imputed values and instead utilize values that are representative of the patient’s functional status at admission and discharge.  This measure does not meet the feasibility criteria required for ‎endorsement, as this measure requires impractical investment in the people, tools, tasks, and ‎technologies necessary to implement this measure. 

Second, this measure produces results that are inconsistent with two currently endorsed functional measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients).  The existing endorsed measures utilize all the functional items in the Self-Care and Mobility domains and produce an accurate representation of the functional improvement and discharge functional status of every IRF patient.  The Discharge Function Score measure utilizes 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional assessment of IRF patients.  Utilizing less than half of all functional items included in the IRF-PAI and existing endorsed measures produces values that are inconsistent and are not representative of the functional improvements and discharge functional status the patients that achieve as part of their IRF care.  Additionally, the public reporting of this measure on Care Compare shows that measure performance is inconsistent with the two other endorsed measures, and some IRFs are reporting that patients who meet expectations for the Discharge Self-Care and Discharge Mobility measures are identified as not meeting expectations for the Discharge Function Score measure.  Because of these inconsistencies, we do not believe that the new measure is superior to existing measures and should not replace the existing measures.  This measure does not meet the importance criteria required for ‎endorsement.

Finally, there are unintended consequences that may result from the endorsement of this measure.  This measure is identified as “Cross-Setting”, yet it differs significantly across post-acute care settings.  While the functional items included in this measure are similar across post-acute care settings, the measure calculations and risk-adjusted patient expectations are very different.  The measure considered for endorsement for Home Health has a different developer and different imputation and risk-adjustment methodology.  Depending on post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation and Long-term acute care hospital (LTCH) patients having the lowest expectation.  To be considered a “cross-setting” measure, the measure calculations and patient expectations should be standardized regardless of setting. Additionally, since these measures are publicly reported on Care Compare, there is the unintended consequence of measure values being used to limit patient access to certain settings based upon reported results.  The characterization of that this measure as cross-setting measures implies that comparison of results can be made between IRFs, SNFs, LTCHs and Home Health providers.  Payers and referral sources already are using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes.  This measure does not meet the use and usability criteria required for ‎endorsement.

As detailed above, this measure does not meet all the criteria necessary for endorsement, and therefore ask the PQM E&M Committee to not support the endorsement of this measure.

On behalf of Madonna Rehabilitation Hospital, we appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  

This measure does not meet the requirements for endorsement as the measure is administratively difficult to manage, produces results inconsistent with other endorsed measures, and could produce unintended consequences impacting access to IRF care.  The concerns are detailed below for PQM E&M Committee’s consideration of these reasons when evaluating this measure for endorsement.

First, clinicians are struggling to understand and manage performance on this measure.  While the data elements utilized in the ‎measure are part of the Inpatient Rehabilitation Facility – Patient Assessment Instrument (IRF-PAI), the imputation methodology to manage ‎the performance of this measure requires an infeasible ‎investment in technology and training to implement.  Providers that are unable to calculate imputed values on ‎their own have limited ability to identify what the patients’ measure scores are ‎and what the risk-adjusted target values are to create a positive ‎result.  ‎ The imputation method is unnecessary, as existing endorsed measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients) do not require imputed values and instead utilize values that are representative of the patient’s functional status at admission and discharge.  This measure does not meet the feasibility criteria required for ‎endorsement, as this measure requires impractical investment in the people, tools, tasks, and ‎technologies necessary to implement this measure. 

Second, this measure produces results that are inconsistent with two currently endorsed functional measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients).  The existing endorsed measures utilize all the functional items in the Self-Care and Mobility domains and produce an accurate representation of the functional improvement and discharge functional status of every IRF patient.  The Discharge Function Score measure utilizes 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional assessment of IRF patients.  Utilizing less than half of all functional items included in the IRF-PAI and existing endorsed measures produces values that are inconsistent and are not representative of the functional improvements and discharge functional status the patients that achieve as part of their IRF care.  Additionally, the public reporting of this measure on Care Compare shows that measure performance is inconsistent with the two other endorsed measures, and some IRFs are reporting that patients who meet expectations for the Discharge Self-Care and Discharge Mobility measures are identified as not meeting expectations for the Discharge Function Score measure.  Because of these inconsistencies, we do not believe that the new measure is superior to existing measures and should not replace the existing measures.  This measure does not meet the importance criteria required for ‎endorsement.

Finally, there are unintended consequences that may result from the endorsement of this measure.  This measure is identified as “Cross-Setting”, yet it differs significantly across post-acute care settings.  While the functional items included in this measure are similar across post-acute care settings, the measure calculations and risk-adjusted patient expectations are very different.   The measure considered for endorsement for Home Health has a different developer and different imputation and risk-adjustment methodology.  Depending on post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation and Long-term acute care hospital (LTCH) patients having the lowest expectation.  To be considered a “cross-setting” measure, the measure calculations and patient expectations should be standardized regardless of setting. Additionally, since these measures are publicly reported on Care Compare, there is the unintended consequence of measure values being used to limit patient access to certain settings based upon reported results.  The characterization of that this measure as cross-setting measures implies that comparison of results can be made between IRFs, SNFs, LTCHs and Home Health providers.  Payers and referral sources already are using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes. Madonna Rehabilitation Hospital houses several levels of care, and this measure will make it hard for them to distinguish outcomes between the levels if all the outcomes are being figured differently.  This measure does not meet the use and usability criteria required for ‎endorsement.

As detailed above, this measure does not meet all the criteria necessary for endorsement, and Madonna Rehabilitation Hospital therefore asks the PQM E&M Committee to not support the endorsement of this measure.  

On behalf of Encompass Health, we appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  

We have serious concerns with the proposed Discharge Function measure, both with the recoding of the “NA” IRF-PAI section GG items for use in the measure calculation and the proposed combination of self-care and mobility into a single measure.  We do not support this proposed Discharge Function measure and respectfully urge that it not be finalized but instead be subject to further consideration and refinement prior to being endorsed by the PQM E&M Committee.

Recoding ANA codes by using an algorithm-based imputation model to calculate the Discharge Function measure would supersede the clinical judgement of caregivers when assessing their patients, pursuant to current CMS guidance. It is not at all surprising that patients who are required to benefit from intensive rehabilitative therapy during their stay at a rehabilitation hospital may not be able to attempt certain self-care and mobility functional tasks safely at the time of their admission. 

In addition to our concern that the new Discharge Function measure’s recoding of ANA codes runs counter to IRF caregivers’ clinical judgement, we also have concerns on the use of the bootstrapping sample to develop the Discharge Function measure imputation model and risk adjustment.  Acumen’s report noted “clinical and function characteristics was different between observations with and without NAs.”  A model built solely upon the functional abilities of patients who had no NA assessments should not be imposed on patients who had NA assessments.  

Furthermore, the different codes combined into the model as NA codes: Activity not attempted (codes 07, 09, 10, and 88), a dash (-), and a skip (^) have significantly different CMS guidance and specifications for their use.  Codes 07, 09, 10, 88, or skip (^) do not represent “missing” data as described in the Proposed Rule.  Rather, these codes represent clinical information that the patient was incapable of performing a task for reasons specified by CMS in the IRF-PAI manual. The only “missing” functional status data is represented by a dash (-), where the item was unable to be scored at all. These dashes are typically used for patients who had a short stay or unplanned discharge, and an item could not be assessed (with nonsensical responses to the BIMS assessment, as instructed by CMS, being the exception).  The other activity not attempted codes were assessed by a clinician and coded appropriately, so the data are not “missing.” 

Additionally, we do not support the combination of self-care and mobility items into a single discharge function score, rather they should be assessed and reported independently to provide the most accurate assessment of a patient’s abilities and disabilities.  Furthermore, these measures were approved for use in the IRF setting as separate measures compiled of a different set of GG items (NQF #2635 and #2636). Further, some patients have an imbalance of impairment between upper mobility and lower mobility.  For example, a frontal stroke patient would have more limited function in their lower extremities and be disadvantaged in their Mobility scores, while a central spinal cord syndrome patient would have more limited function in their upper extremities and be disadvantaged in their Self-Care scores.

For a true “cross-setting” functional measure that yields truly comparable results, more must be done to standardize data across PAC settings and the measure set must be able to collect and differentiate patients’ abilities and disabilities in a wide range of functional levels, such as patients who may be more functionally independent in the HH setting and LTCH patients who are more functionally dependent.  The measure should prioritize ensuring that these distinctions in patient characteristics are preserved.

Finally, to our knowledge this is the first time CMS will be implementing a quality measure score with data not actually collected by clinicians but imputed based on the model developed.  CMS and Acumen should release more of the data and methodology so the PQM E&M Committee may verify the model outcomes appropriately. In particular, the report is somewhat unclear how walk versus wheelchair patients for purposes of assessing ANA status were accounted for.  In our analyses, depending on how the model handles dashes and ANA codes for walk and wheel patients, the sample of patients without an NA can range from over 60 percent to over 90 percent.  This wide discrepancy shows the complexity of developing this measure and in verifying its results. 

We are concerned that the outcomes calculated in this measure will be inaccurate and not represent the true functional gain achieved by patients during their course of treatment in a rehabilitation hospital, as the expected discharge function score will be calculated for patients with NA codes based on the function items of patients without NA codes. 

To summarize, we are concerned that the Discharge Function measure has not been appropriately developed or tested, will supersede the clinical judgement of providers in assessing their patients’ functional capabilities, and may not be equitable among IRF patients.  We do not support the PQM E&M committee endorsing this measure and recommend CMS engage with providers to better understand the assessment of patients in a clinical setting before moving any further. 

As detailed above, this measure does not meet all the criteria necessary for endorsement, and Encompass Health, therefore asks the PQM E&M Committee to not support the endorsement of this measure.  

On behalf of Mary Free Bed Rehabilitation Hospital, we appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  

This measure does not meet the requirements for endorsement as the measure could be administratively difficult to manage (feasibility), does not appear to produce new information (importance), and could produce unintended consequences impacting access to IRF care (usability).  Our concerns regarding the PQM E&M Committee’s consideration of endorsement for this measure are detailed below .

First, responsibility for imputation and scoring is unclear.  While the data elements utilized in the ‎measure are part of the Inpatient Rehabilitation Facility – Patient Assessment Instrument (IRF-PAI), the imputation methodology to manage ‎the performance of this measure comes with burden. The report is not clear on whether CMS would incorporate imputation and scoring algorithms into their reporting systems or if providers are responsible for doing so. If this becomes the providers’ responsibility there is room for error in submitting scores if the algorithms are not properly implemented. ‎In addition, evidence to support use of a new imputation method which differs from the method used for the existing Self-Care and Mobility scores (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients) is insufficient. Although the new imputation method demonstrated reductions in bias and error, the report did not provide information on whether the new method would change results relative to current procedure. And, if the new procedure is better, should it not be implemented for all quality measures to ensure consistency in reporting? Thus, we feel this measure does not meet the feasibility criteria required for ‎endorsement, as this measure may place unnecessary burden on providers for scoring and imputation and may perform different from existing imputation methods. 

Second, this measure produces results that are paradoxically both similar  and inconsistent with two currently endorsed functional measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients).  The existing endorsed measures utilize all the functional items in the Self-Care and Mobility domains and produce an accurate representation of the functional improvement and discharge functional status of every IRF patient.  The Discharge Function Score creates a composite measure by utilizing 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional assessment of IRF patients. The report notes high correlations (r=0.85-0.91) between the new measure and the existing Self-Care and Mobility discharge and change scores as one would expect for a measure derived from these existing measures. Of concern is that these correlations are not independent and possibly inflates as the scores based in part items that are in both comparators (3 Self-Care items and 7 mobility items). However, the correlation with Discharge to Community (NQF #3479) was only 0.24, indicating a positive by weak association with this item. Another problem arises from the inclusion of facilities with as few as 20 cases in the model. While results for facilities with large samples might be stable, facilities with small samples could be biased or fluctuate widely from one quarter to another. One solution would be to include patient data from earlier than the past year, although this would increase the duration of time over which patient outcomes are reflected in their quality scores. Despite similarities in scores by correlation, the public reporting of this measure on Care Compare shows that measure performance is inconsistent with the two other endorsed measures, and some IRFs are reporting that patients who meet expectations for the Discharge Self-Care and Discharge Mobility measures are identified as not meeting expectations for the Discharge Function Score measure. This discrepancy could create problems with interpretation of the Self-Care, Mobility, and the Discharge Function Score collectively, erode confidence in the QRP measures, or complicate decisions made based on these scores. At a minimum, we would like to see a comparison of the new and existing measures to determine where and to what extent discrepancies arise. If discrepancies are limited to only facilities with small samples, the new measure may be comparable to existing measures. However, if discrepancies also occur with large facilities, the impact of implementation on providers should be examined and reported. Further, if the objective is to replace the existing measures, why include only a subset of each of the Self-Care and Mobility items. As noted, the full item sets are already measured so why include only a portion of the available information? For these reasons, we do not believe that the new measure is superior to existing measures and thus does not meet the importance criteria required for ‎endorsement.

Finally, there are unintended consequences that may result from the endorsement of this measure.  This measure is identified as “Cross-Setting”, yet it differs significantly across post-acute care settings.  While the functional items included in this measure are similar across post-acute care settings, the measure calculations and risk-adjusted patient expectations are very different.  The measure considered for endorsement for Home Health has a different developer and different imputation and risk-adjustment methodology.  Depending on post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations established by setting-specific models. Although the patient populations of IRF, long-term acute care hospitals (LTCH), skilled nursing facilities (SNFs), and Home Health Associations may overlap, they have differences imposed on them by their respective eligibility criteria and/or timelines for admission. Thus, creating separate expectations based on regression equations that only represent patients from within each setting does not permit comparisons for facilities from different settings.  To be considered a “cross-setting” measure, the measure calculations and patient expectations must be standardized regardless of setting. As these measures are publicly reported on Care Compare, there is potential for misuse, whether unintended or intended, of measure values being used to limit patient access to certain settings based upon reported results.  The characterization of that this measure as cross-setting measures implies that comparison of results can be made between IRFs, SNFs, LTCHs and Home Health providers which the statistical methods do not support.  Payers and referral sources already are using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes.  Thus, we feel this measure does not meet the use and usability criteria required for ‎endorsement.

As detailed above, this measure does not meet all the criteria necessary for endorsement, and Mary Free Bed Rehabilitation Hospital therefore asks the PQM E&M Committee to not support the endorsement of this measure.  

The American Medical Rehabilitation Providers Association (AMRPA) appreciates the opportunity to follow-up with additional detail on the verbal comments provided on November 21 related to CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities. AMRPA is the national trade association representing over 800 inpatient rehabilitation facilities (IRFs), ‎including freestanding IRFs and rehabilitation units of acute care hospitals, which focus on the care and ‎functional recovery of some the most vulnerable Medicare beneficiaries – such as patients with ‎ stroke, brain injury, and spinal cord injury. ‎ 

As noted in our prior verbal comments, AMRPA members believe that CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities is a competing measure for two existing endorsed measures, CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients.  The two existing endorsed measures separate patient function into two separate domains, Self-Care and Mobility, and utilize all the functional status items included on the IRF-PAI.  Using separate measures and all the functional items for quality recognizes that IRF patients have significantly different functional abilities or disabilities requiring the intensive therapies provided only in the IRF setting.   The measure under consideration for endorsement combines these two domains of patient function and utilizes 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional ‎assessment of IRF patients.  Utilizing less than half of all functional items included in the IRF-PAI ‎and existing endorsed measures produces values that are inconsistent and are not ‎representative of the functional improvements and discharge functional status that patients ‎achieve as part of their IRF care‎.  

Additionally, the public reporting of this measure on Care Compare shows that ‎performance on the Discharge Function Score measures is inconsistent with the two other endorsed measures.  For example, in the September 2024 Care Compare refresh, 153 of the over 1100 IRFs with reported values (roughly 14%) appear to have a value for the Discharge Function Score measures that is less than their performance on both existing endorsed measures that utilize all functional items.  Similarly, AMRPA member IRFs are reporting ‎that patients who meet expectations for the Discharge Self-Care and Discharge Mobility ‎measures are identified as not meeting expectations for the Discharge Function Score ‎measure.‎ As identified in the PQM Measure Evaluation Rubric, the measure developer must indicate how the new measure is superior to existing measures as part of the Importance criteria for endorsement.  Based upon the inconsistencies in the measure performance against two existing endorsed measures currently displayed in public reporting, AMRPA does not believe that this measure should be considered superior to the existing endorsed measures and therefore does not meet the Importance criteria necessary for endorsement.

AMRPA also has concerns about the imputation method utilized for calculating assessment data.  As AMRPA stated in comments to the FY 2024 IRF Proposed Rule:

The imputation method should not be used to recode instances where the assessment data is not “missing.” 

CMS and the original measure developer (Acumen) suggest that “missing” functional status data include patients who are coded with the following assessment values:

• Code 07, Patient refused: if the patient refused to complete the activity.
• Code 09, Not applicable: if the patient did not attempt to perform the activity and did not perform this activity prior to the current illness, exacerbation, or injury.
• Code 10, Not attempted due to environmental limitations: if the patient did not attempt this activity due to environmental limitations. Examples include lack of equipment, weather constraints.
• Code 88, Not attempted due to medical condition or safety concerns: if the activity was not attempted due to medical condition or safety concerns.
• Skipped Value, Item was skipped based upon a prior response.
• Dash Value, Indicates “No information.”

Except for the Dash Value (where no information was available for the assessment), all other values are indicative of an affirmative functional assessment value and should not be considered as “missing.” Each code represents a circumstance where the patient is not capable of performing the activity, suggesting that their performance should be considered at least at the most severe level (Code 01 – Dependent, where if the activity was attempted it would require a helper to complete the entire activity for the patient or require two or more helpers to complete the activity). Code 07, Patient Refused rarely occurs and, in these situations, the refusal from the patient to perform the activity suggests that it would require a helper to complete the activity for the patient. Code 09, Not Applicable indicates that the patient did not perform the activity and did not perform the activity prior to the IRF stay, suggesting that it would require a helper to complete the activity for the patient. Code 10, Environmental Limitations indicates that if the activity was attempted without equipment or in a compromised weather situation, the patient would have required a helper to complete the activity for them. Code 88, Medical Condition or Safety Concerns indicates that, if the activity was attempted, it would impact the patient’s medical condition or put the patient in an unsafe situation, unless the activity was completed by a helper. Finally, Skipped Value indicates that the patient was unable to perform a prior activity of a similar nature, meaning that the patient would be similarly unable to perform this activity without a helper. In these scenarios, the data is not “missing” and, instead, the coded value represents a situation where the most dependent level is the only clinically relevant option. Therefore, the statistical imputation model should not be used for these code values and instead should only be used when the Dash Value, representing no information, is coded.

The imputation method will be operationally difficult for clinicians to know the patient values and manage performance. 

Without a technological/software solution or other means to calculate the imputed values in real-time, clinicians will be left to wonder what the patient’s function score will be at admission or discharge and would have to wait for CMS to publish measure data to know whether their patient met or exceeded the expected discharge score. Under these circumstances, where the values are not readily available, performance on this measure becomes guesswork for any patient requiring the use of an imputed value. Quality measures are meant to be actionable and meaningful; however, the imputation method will make it extremely difficult for IRFs to know what is needed to improve performance and provide meaningful results to patients.

AMRPA also asks that you consider whether this measure meets the feasibility criteria required for endorsement. The measure developer must consider the people, tools, tasks, and technologies necessary to implement this measure.  As AMRPA noted in our comments to the FY 2024 IRF Proposed Rule:

Costs associated with managing this measure have not been considered, such as software ‎updates associated with the various measure calculations and training/education ‎for clinicians.

While CMS states that “…this measure adds no additional provider burden since it would be calculated using data from the IRF-PAI that IRFs are already required to collect,” IRFs will still need to educate and train their clinicians on the new measure, incorporate discussion of this measure into their interdisciplinary team meetings, and create a solution that will calculate imputation values and the risk-adjusted expected discharge function score values in order to manage performance. Clinicians will need to understand which items are included in this new discharge function score measure, the implications of the imputation method and the use of the “missing” codes, and how performance on this measure may differ from other quality measures. Clinicians will also need to understand how to review measure results on Provider Preview Reports and other CMS reporting tools. We would estimate that the time necessary to educate/train one clinician on this measure would take at least one hour, and the review of Provider Preview Reports and other CMS quality measure reports to take at least one hour every quarter, or 4 hours annually. Using the adjusted hourly wages provided in the proposed rule, the cost of educating, training, and managing this measure for one clinician would be between $250 (Licensed Practical and Licensed Vocational Nurses) to $450 (Physical Therapists) annually. Across the over 1,100 IRFs, this would represent an annual cost burden of between $275,000 and $495,000. Should additional education and training be provided to other clinicians, this will add another $50-90 per clinician.

Technology-related costs should also be considered, as the imputation method and risk-adjusted expected scores require advanced calculations to be able to monitor patient progress toward their expectation. We will not speculate what the average cost for such software development may be, but instead note that any costs associated with these needs have not been considered as part of the proposed adoption of this measure. 

AMRPA members believe that these costs must be considered, as they would significantly impact the values in the IRF QRP Impact Table, Accounting Statement Table, and conclusion of estimated payment changes per discharge displayed in the proposed rule. We further encourage CMS to consider these costs when assessing all new or modified measures in order to more accurately reflect the burden IRF QRP changes pose for IRFs and other providers.

While the data elements utilized in the measure calculations are already part of the IRF-PAI, IRFs would require a significant investment in technology to implement the imputation methodology required to manage the performance of this measure.  Providers that are unable to calculate imputed values on their own will have limited ability to identify what the patients’ measures scores may be order determine what risk-adjusted target value is required to contribute towards a positive result.  For these reasons, AMRPA believes that this measure does not meet the feasibility criteria required for endorsement.

Finally, AMRPA members are concerned about the suggestion that this is a cross-setting measure and the unintended consequences that may result from this designation.  This measure differs across post-acute care settings as the risk-adjustment methodology utilizes setting specific covariates and setting-specific coefficients.  For example, each setting has a model intercept for the expected discharge function score for each patient.  For IRFs the model intercept is 34.1701, for SNFs the model intercept is 30.0118, and for LTCHs the model intercept is 14.5276.  This suggests that depending on setting a patient with the same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation and LTCH patients having the lowest expectation.  To be considered a cross-setting measure, AMRPA members believe that the patient expectations should be consistent regardless of setting.

Additionally, in suggesting this as a cross-setting measure that is publicly reported, these measures have the unintended consequence of potentially being used to limit patient access to certain settings based upon results.  While CMS and the measure developers documented concerns about providers denying access to certain patients who may not perform well on this measure, AMRPA questions whether consideration was given for referral sources using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes.  We ask that the committee consider these unintended consequences when evaluating this measure for endorsement.

As previously stated, AMRPA appreciates the opportunity to provide these additional detailed comments related to CBE #4630 – Cross-Setting Discharge Function Score ‎for Inpatient Rehabilitation Facilities. We ask that the PQM E&M Advanced Illness and Post-Acute Care Recommendation Committee consider our comments when evaluating this measure for endorsement.  Should the committee require any additional information related to our comments on the measure, we ask that you please reach out to AMRPA Director of Quality and Health Policy, Troy Hillman ([email protected]) or AMRPA President, Kate Beller ([email protected]).

I appreciate the opportunity to comment on CBE #4630 – Cross-Setting Discharge Function Score for Inpatient Rehabilitation Facilities (IRF).  

This measure does not meet the requirements for endorsement as the measure is administratively difficult to manage, produces results inconsistent with other endorsed measures, and could produce unintended consequences impacting access to IRF care.  The concerns are detailed below for PQM E&M Committee’s consideration of these reasons when evaluating this measure for endorsement.

First, clinicians are struggling to understand and manage performance on this measure.  While the data elements utilized in the ‎measure are part of the Inpatient Rehabilitation Facility – Patient Assessment Instrument (IRF-PAI), the imputation methodology to manage ‎the performance of this measure requires an infeasible ‎investment in technology and training to implement.  Providers that are unable to calculate imputed values on ‎their own have limited ability to identify what the patients’ measure scores are ‎and what the risk-adjusted target values are to create a positive ‎result.  ‎ The imputation method is unnecessary, as existing endorsed measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients) do not require imputed values and instead utilize values that are representative of the patient’s functional status at admission and discharge.  This measure does not meet the feasibility criteria required for ‎endorsement, as this measure requires impractical investment in the people, tools, tasks, and ‎technologies necessary to implement this measure. 

Second, this measure produces results that are inconsistent with two currently endorsed functional measures (CBE #2635 Discharge Self-Care Score for Medical Rehabilitation Patients and CBE #2636 Discharge Mobility Score for Medical Rehabilitation Patients).  The existing endorsed measures utilize all the functional items in the Self-Care and Mobility domains and produce an accurate representation of the functional improvement and discharge functional status of every IRF patient.  The Discharge Function Score measure utilizes 3 of the 7 Self-Care items and 7 of the 17 Mobility items included in the functional assessment of IRF patients.  Utilizing less than half of all functional items included in the IRF-PAI and existing endorsed measures produces values that are inconsistent and are not representative of the functional improvements and discharge functional status the patients that achieve as part of their IRF care.  Additionally, the public reporting of this measure on Care Compare shows that measure performance is inconsistent with the two other endorsed measures, and some IRFs are reporting that patients who meet expectations for the Discharge Self-Care and Discharge Mobility measures are identified as not meeting expectations for the Discharge Function Score measure.  Because of these inconsistencies, we do not believe that the new measure is superior to existing measures and should not replace the existing measures.  This measure does not meet the importance criteria required for ‎endorsement.

Finally, there are unintended consequences that may result from the endorsement of this measure.  This measure is identified as “Cross-Setting”, yet it differs significantly across post-acute care settings.  While the functional items included in this measure are similar across post-acute care settings, the measure calculations and risk-adjusted patient expectations are very different.   The measure considered for endorsement for Home Health has a different developer and different imputation and risk-adjustment methodology.  Depending on post-acute care setting, a patient with the same patient characteristics will need to meet or exceed different expectations, with IRF patients having the highest expectation and Long-term acute care hospital (LTCH) patients having the lowest expectation.  To be considered a “cross-setting” measure, the measure calculations and patient expectations should be standardized regardless of setting. Additionally, since these measures are publicly reported on Care Compare, there is the unintended consequence of measure values being used to limit patient access to certain settings based upon reported results.  The characterization of that this measure as cross-setting measures implies that comparison of results can be made between IRFs, SNFs, LTCHs and Home Health providers.  Payers and referral sources already are using this information to direct patients to alternative settings which may not provide the appropriate services to produce high quality outcomes.  This measure does not meet the use and usability criteria required for ‎endorsement.

As detailed above, this measure does not meet all the criteria necessary for endorsement, and therefore I ask the PQM E&M Committee to not support the endorsement of this measure.