When this measure comes back for maintenance, the committee would like to see:
- Implementation data (to include patients 18 years and older) that examines whether the measure is associated with improved nutritional status or related clinical endpoint
This composite measure assesses the percentage of hospitalizations for adults aged 18 years and older at the start of the inpatient encounter during the measurement period with a length of stay equal to or greater than 24 hours who received optimal malnutrition care during the current inpatient hospitalization where care performed was appropriate to the patient's level of malnutrition risk and severity. Malnutrition care best practices recommend that for each hospitalization, adult inpatients are screened for malnutrition risk, assessed to confirm findings of malnutrition risk or concern raised through a hospital dietitian referral order, and, if identified with a "moderate" or "severe" malnutrition status in the current performed malnutrition assessment, receive a current "moderate" or "severe" malnutrition diagnosis and have a current nutrition care plan performed. A version of this measure, assessing performance only for adults aged 65 years and older, is currently endorsed and active in the IQR program; this submission describes a substantive change in the measure, as the population is changed to all adults aged 18 and older.
Measure Specs
General Information
Malnutrition is a leading cause of United States (U.S.) morbidity and mortality. Evidence suggests that 20% to 50% of all patients are malnourished or at risk of malnutrition at the time of hospital admission, with up to 31% of these malnourished patients and 38% of well-nourished patients experiencing nutritional decline during their hospital stays. Insufficiency of available nutrients needed to promote healing and rehabilitation may lead to an increased risk of medical complications, including depression of the immune system, impaired wound healing, muscle wasting, and increased mortality. Malnutrition and weight loss can also contribute to sarcopenia, or a loss of skeletal muscle mass and function, which also impedes an individual’s recovery, mobility, ability to perform daily activities, and independence.
The presence of a malnutrition diagnosis is unique in that it can have complex physiological causes, as well as be multifactorial, with environmental, economic, and psychological origins being possible also. This makes identifying and treating malnutrition an effective step to improve health equity in acute care. There is an inherent connection between malnutrition, food insecurity, and health equity. Food insecurity is present in households concerned about food running out, dietary quality and variety, and quantity of food consumed. Screening for malnutrition can be of significance in identifying and addressing health inequities when malnutrition is caused by food insecurity.
Though malnutrition can be present on admission, it can also develop throughout a hospital course despite a baseline of adequate nutrition status. Hospitalized patients are vulnerable to nutritional decline for many reasons, including dietary restrictions in preparation for medical testing and treatments, as well as poor appetites, nutritional intolerance, and gastrointestinal problems resulting from existing medical conditions, hospitalization-related stress and anxiety, side effects from medications, and other medical, behavioral, and cultural reasons. Insufficient intake causes further decline in the nutrition status of patients who are malnourished at the time of hospital admission. Hospitalized malnourished patients also have a greater risk of complications, such as development of hospital-acquired infections, functional decline, and in-hospital death. A patient’s nutrition status is also considered a key factor in “post-hospital syndrome,” a period of increased susceptibility to poor outcomes immediately following hospitalization.5
The Global Malnutrition Composite Score (GMCS) electronic clinical quality measure (eCQM) uses the evidence- and consensus-based nutrition care workflow that incorporates both clinical risk factors and patient preferences to evaluate hospital performance into four steps that occur exclusively in the hospital setting. These include the malnutrition risk screening performed by a nurse, RD/RDN, or any other appropriate professional; nutrition assessment performed by an RD/RDN; malnutrition diagnosis documented by a physician or other qualified healthcare professional; and documentation of a nutrition care plan of malnutrition interventions that is developed by an RD/RDN. A version of this measure evaluating performance in adults aged 65 years and older is currently endorsed and active in the CMS IQR program. This submission represents a substantive change, as the measure population will now include all adults aged 18 years and older.
Measured entities will document all data elements directly in the Electronic Health Record. Data will be extracted from the Electronic Health Record by utilizing the value set assigned to the data elements. The report will be patient level.
Numerator
This is a continuous variable measure.
"Measure Observation 1" = "Encounters with Malnutrition Risk Screening and Identified Result"; "Measure Observation 1" identifies hospital encounters where a "Malnutrition Risk Screening" was performed with a current identified "Malnutrition Screening Finding of Not At Risk Result" or current "Malnutrition Screening Finding of At Risk Result" OR a "Hospital Dietitian Referral" was ordered.
"Measure Observation 2" = "Encounter with Nutrition Assessment and Identified Status"; "Measure Observation 2" identifies hospital encounters where a "Nutrition Assessment" was performed with a current identified "Nutrition Assessment Status Finding of Well Nourished or Not Malnourished or Mildly Malnourished", "Nutrition Assessment Status Finding of Moderately Malnourished", or "Nutrition Assessment Status Finding of Severely Malnourished".
"Measure Observation 3" = "Encounters with Malnutrition Diagnosis"; "Measure Observation 3" identifies hospital encounters where a current "Malnutrition Diagnosis" was documented.
“Measure Observation 4” = “Encounters with Nutrition Care Plan”; "Measure Observation 4" identifies hospital encounters where a current "Nutrition Care Plan" was performed.
"Population 5 Measure Observation TotalMalnutritionComponentsScore" equals the sum of ("Measure Observation 1" plus "Measure Observation 2" plus "Measure Observation 3" plus "Measure Observation 4")
"Population 6 Measure Observation TotalMalnutritionCompositeScore as Percentage" = 100 * ("TotalMalnutritionComponentsScore" divided by "TotalMalnutritionCompositeScore Eligible Denominators").
-For each hospitalization, Population Criteria 6 represents the sum of performed Measure Observations 1, 2, 3, and 4 divided by the number of clinically eligible occurrences.
All Measure Observations
This includes all the needed data elements to identify a qualifying encounter
During inpatient encounter and/or associated emergency department and/or observation encounter(s)
- Ethnicity
- Payer
- Race
- Administrative Sex
- Encounter Type
- Inpatient Admission Time
- Inpatient Discharge Time
- Date of Birth
- Ethnicity: Extensional CDCREC, OID 2.16.840.1.114222.4.11.837
- Payer: Intensional SOP, OID 2.16.840.1.114222.4.11.3591
- Emergency Department Visit: Extensional LOINC, OID 2.16.840.1.113883.3.117.1.7.1.292
- Race: Extensional CDCREC
- Observation Services: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1111.143
- Encounter Inpatient: Extensional SNOMED, OID 2.16.840.1.113883.3.666.5.307
- ONC Administrative Sex: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1
Malnutrition Risk Screening/ Hospital Dietitian Referral
Malnutrition risk screening performed by nursing, RDN, or appropriate professional -OR- Hospital Dietitian Referral Ordered
- Documented Malnutrition Risk Screening
- Documented Malnutrition Risk Screening Time Stamp
- Documented Malnutrition Risk Screening Result
- Documented Hospital Dietitian Referral
- Documented Hospital Dietitian Referral Time Stamp
- Malnutrition Risk Screening: Extensional LOINC, OID 2.16.840.1.113762.1.4.1095.92
- Malnutrition Screening Finding of Not at Risk Result: Extensional LOINC, OID 2.16.840.1.113762.1.4.1095.34
- Malnutrition Screening Finding of at Risk Result: Grouping, OID 2.16.840.1.113762.1.4.1095.89
- Malnutrition Screening Finding of at Risk Result: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1095.38
- Malnutrition Screening Finding of at Risk Result: Extensional LOINC, OID 2.16.840.1.113762.1.4.1095.94
- Hospital Dietitian Referral: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1095.91
Nutrition Assessment
Nutrition Assessment performed by RD/RDN in patients screened and identified with malnutrition risk -OR- a Hospital Dietitian Referral
- Documented Nutrition Assessment
- Documented Nutrition Assessment Time Stamp
- Documented Nutrition Assessment Result
- Nutrition Assessment: Extensional LOINC, OID 2.16.840.1.113762.1.4.1095.21
- Nutrition Assessment Status Finding of Well Nourished or Not Malnourished or Mildly Malnourished: Grouping (SNOMED, LOINC), OID 2.16.840.1.113762.1.4.1095.96
- Nutrition Assessment Status Finding of Well Nourished or Not Malnourished or Mildly Malnourished: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1095.48
- Nutrition Assessment Status Finding of Well Nourished or Not Malnourished or Mildly Malnourished: Extensional LOINC, OID 2.16.840.1.113762.1.4.1095.95
- Nutrition Assessment Status Finding of Moderately Malnourished: Grouping (SNOMED, LOINC), OID 2.16.840.1.113762.1.4.1095.47
- Nutrition Assessment Status Finding of Moderately Malnourished: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1095.44
- Nutrition Assessment Status Finding of Moderately Malnourished: Extensional LOINC, OID 2.16.840.1.113762.1.4.1095.98
- Nutrition Assessment Status Finding of Severely Malnourished: Grouping (SNOMED, LOINC), OID 2.16.840.1.113762.1.4.1095.43
- Nutrition Assessment Status Finding of Severely Malnourished: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1095.42
- Nutrition Assessment Status Finding of Severely Malnourished: Extensional LOINC, OID 2.16.840.1.113762.1.4.1095.97
Malnutrition Diagnosis
Malnutrition diagnosis documented by a physician or eligible provider in patients with Moderate or Severe Malnutrition result from current Nutrition Assessment
- Documented Malnutrition Diagnosis
- Documented Malnutrition Diagnosis Time Stamp
- Malnutrition Diagnosis: Grouping (SNOMED, ICD10), OID 2.16.840.1.113762.1.4.1095.55
- Malnutrition Diagnosis: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1095.53
- Malnutrition Diagnosis: Extensional ICD-10, OID 2.16.840.1.113762.1.4.1095.54
Nutrition Care Plan
Nutrition care plan performed by RD/RDN in patients with Moderate or Severe Malnutrition result from current Nutrition Assessment
- Documented Nutrition Care Plan
- Documented Nutrition Care Plan Time Stamp
- Nutrition Care Plan: Extensional SNOMED, OID 2.16.840.1.113762.1.4.1095.93
Denominator
TotalMalnutritionCompositeScore Eligible Occurrences" is 4 except in the following instances:
-If a "Malnutrition Risk Screening" was performed and a "Malnutrition Screening Finding of Not At Risk Result" was identified AND "Hospital Dietitian Referral" was not ordered, then the "TotalMalnutritionCompositeScore Eligible Occurrences" is 1.
-If a "Malnutrition Risk Screening" was performed OR a "Hospital Dietitian Referral" was ordered AND a "Nutrition Status Finding of Well Nourished or Not Malnourished or Mildly Malnourished" was identified OR a Nutrition Assessment was not completed , then the “TotalMalnutritionCompositeScore Eligible Occurrences” are 2.
-For the reporting facility, the Population Criteria 6 averages the performance of each "TotalMalnutritionCompositeScore as Percentage" across all eligible hospitalizations during the measurement period.
For any qualifying encounter with a patient at least 18 years of age or older with an inpatient status length of stay of at least 24 hours, the eligible occurrences (mathematical denominator) equals:
- 1 when there is a Not At Risk Result from Malnutrition Screening AND no Hospital Dietitian Referral order is present
- 2 when a Malnutrition Screening results in an At Risk result OR there is a Hospital Dietitian Referral AND the Nutrition Assessment results in a Well Nourished or Not Malnourished or Mildly Malnourished result OR there is no documented Nutrition Assessment.
In all other scenarios, the eligible occurrence is 4.
No specific codes or value sets are associated with the eligible occurrences (mathematical denominator).
Exclusions
None
None
Measure Calculation
See attached diagram.
No measure stratification.
No minimum sample size is required to determine measure performance.
Point of Contact
N/A
United States
Tamaire Ojeda
Commission on Dietetic Registration
120 S. Riverside Plaza
Suite 2190
Chicago, IL 60606
United States
Importance
Evidence
Addressing malnutrition improves health outcomes and quality of life and decreases complications, hospital readmissions and length of stay, as well as care delivery costs[1]. Clinical consensus recommendations underscore the benefits of early malnutrition identification and systematic nutrition care interventions; coupled with interdisciplinary collaboration, these are critical to remediating malnutrition across the care continuum. Nutrition care best practices also include the engagement of patients and their families in development and implementation of nutrition care plans during hospitalization and upon discharge to enhance recovery and improved outcomes. Studies have demonstrated that implementation of comprehensive nutrition care pathways from inpatient admission through discharge improved identification of patients at high risk of malnutrition, and decreased time to nutrition consult, length of hospital stay, and 30-day readmission rate. Evidence also demonstrates the use of malnutrition quality measures assists health systems to identify malnutrition quality of care performance gaps[2] and improve outcomes when implemented in conjunction with comprehensive quality improvement efforts.[3]
The components of this measure are supported by clinical guidance that recommends the following: (1) malnutrition screening for patients admitted into the acute inpatient care setting; (2) nutrition assessment for patients identified at-risk of malnutrition or with a hospital dietitian referral order to form the basis for appropriate nutrition interventions; (3) appropriate recognition, diagnosis, and documentation of the nutrition status of a patient in order to (4) address their condition with an appropriate plan of care and communicate patient needs to other care providers.4 These components were originally proposed as four individual measures, though ultimately combined into one composite score based on feedback from both the National Quality Forum (NQF) and the Centers for Medicare & Medicaid Services (CMS).
The process for risk identification, assessment, diagnosis, and treatment of malnutrition necessitates a multi-disciplinary care team that begins with the identification of an initial risk population for a more thorough physical assessment by registered dietitian (RD) or registered dietitian nutritionists (RDN). The RDN in turn provides the necessary treatment recommendations to address nutritional status and the clinical indicators that inform a medical diagnosis of malnutrition completed by a physician. The four component measures individually will only provide a fraction of the necessary information on quality of care for patients at-risk of malnutrition. For example, knowing which patients have been assessed out of those who were initially identified as at-risk, but not knowing if the appropriate proportion of patients were screened upon admission, would be an insufficient assessment of quality of care.
Implementation of this measure supports timely malnutrition risk screening and hand off to the RD/RDN for appropriate nutritional assessment for patients at-risk of malnutrition during the current hospitalization. For patients identified with a moderate or severe malnutrition status from the nutrition assessment, best practice also recommends a medical diagnosis by a physician or other qualified healthcare professionals and the execution of the nutrition care plan by an RD/RDN. Evidence demonstrates that implementing a standardized protocol for screening, assessment, diagnosis, and care planning results in better identification of malnourished patients and subsequent improvements in rates of nutrition intervention for the malnourished.[4] Outcomes modeling, and those reported in other studies, also demonstrate the benefits to patient outcomes, including reduced risk of 30-day readmissions, length of hospital stay, and complications, as well as improved quality of life after hospitalization.
[1] Sriram K, Sulo S, VanDerBosch G, et al. A comprehensive nutrition-focused quality improvement program reduces 30-day readmissions and length of stay in hospitalized patients. JPEN J Parenter Enteral Nutr. 2017; 41(3): 384-391.
[2] Wills‐Gallagher J, Kerr KW, Macintosh B, Valladares AF, Kilgore KM, Sulo S. Implementation of malnutrition quality improvement reveals opportunities for better nutrition care delivery for hospitalized patients. Journal of Parenteral and Enteral Nutrition. 2022;46(1):243-248. doi:10.1002/jpen.2086
[3] Valladares AF, Kilgore KM, Partridge J, Sulo S, Kerr KW, McCauley S. How a Malnutrition Quality Improvement Initiative Furthers Malnutrition Measurement and Care: Results From a Hospital Learning Collaborative. Journal of Parenteral and Enteral Nutrition. 2021;45(2):366-371. doi:10.1002/jpen.1833
[4] Mueller C, Compher C & Druyan ME and the American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.) Board of Directors. Nutrition Screening, Assessment, and Intervention in Adults. Journal of Parenteral and Enteral Nutrition. 2011; 35 (1): 16-24. A.S.P.E.N. Clinical Guidelines (wiley.com)
Measure Impact
The provision of nutrition care in alignment with a patient’s malnutrition risk or malnutrition severity has been shown to improve a variety of outcomes of interest. Early hospital-based malnutrition identification and documentation allows care teams to address a patient’s condition with an appropriate plan of care and communicate patient needs to other care providers. Identifying and addressing malnutrition early in the episode of care is associated with reduced lengths of stay, 30-day readmission rates, hospital-acquired conditions, and overall healthcare costs[1],[2],[3]. A randomized controlled trial of 652 hospitalized, malnourished older adults aged 65 years and older evaluated the use of a high-protein oral nutritional supplements for its impact on patient outcomes reporting significant reductions in 90-day mortality[4]. Nutrition support for patients identified with risk for malnutrition or malnutrition improves clinical outcomes[5].Nutrition assessments conducted for at-risk patients identified by malnutrition screening using a validated screening tool was associated with key patient outcomes including less weight loss, reduced length of stay, improved muscle function, better nutritional intake, and fewer readmissions. Additionally, a study of 733 patients from more than a dozen hospitals identified that the completion of a validated assessment for patients who were hospitalized was able to detect predictors of outcomes for malnutrition, such as hospital length of stay, and readmission within 30 days after discharge[6].
[1] Lew CCH, Yandell R, Fraser RJL, Chua AP, Chong MFF, Miller M. Association Between Malnutrition and Clinical Outcomes in the Intensive Care Unit: A Systematic Review. Journal of Parenteral and Enteral Nutrition. 2017;41(5):744-758. doi:10.1177/0148607115625638
[2] Meehan A, Loose C, Bell J, Partridge J, Nelson J, Goates S. Health System Quality Improvement. J Nurs Care Qual. 2016;31(3):217-223. doi:10.1097/NCQ.0000000000000177
[3] Fry DE. Patient Characteristics and the Occurrence of Never Events. Archives of Surgery. 2010;145(2):148. doi:10.1001/archsurg.2009.277
[4] Deutz NE, Matheson EM, Matarese LE, et al. Readmission and mortality in malnourished, older, hospitalized adults treated with a specialized oral nutritional supplement: A randomized clinical trial. Clinical Nutrition. 2016;35(1):18-26. doi:10.1016/j.clnu.2015.12.010
[5] Mueller C, Compher C, Ellen DM. A.S.P.E.N. Clinical Guidelines. Journal of Parenteral and Enteral Nutrition. 2011;35(1):16-24. doi:10.1177/0148607110389335
[6] Jeejeebhoy KN, Keller H, Gramlich L, et al. Nutritional assessment: comparison of clinical assessment and objective variables for the prediction of length of hospital stay and readmission. Am J Clin Nutr. 2015;101(5):956-965. doi:10.3945/ajcn.114.098665
The current measure in place to evaluate the quality of malnutrition care includes only patients aged 65 years and older. There is evidence supporting the presence of malnutrition in adults of all ages.[1] Expanding this measure to include all adults aged 18 and older will better capture quality of malnutrition care for all hospitalized adults, as malnutrition has a significant impact on care and outcomes throughout the life cycle. Additionally, this expansion will help address the known major gap between the presence of malnutrition and its diagnosis and treatment, and major impact of food insecurity identification to help improve health equity.
The voice of patients, families, and caregivers is essential to the provision of high-quality care. To ensure that malnutrition care included this unique perspective, a National Dialogue was convened in 2018 among multi-stakeholder representatives from providers, social workers, payers, professional societies, patient and caregiver advocacy groups, and community-based service providers. Participants voiced concern that nutrition status is rarely considered when evaluating overall health status, and that the term “malnutrition” often has a negative connotation among patients and caregivers because of implied fault. However, there was a consensus among the entirety of the group that evaluation of nutrition status and the provision of high-quality nutrition care should be integrated into protocols, pathways, and models throughout the care spectrum.
Performance Gap
Testing activities involving statistical analysis used EHR data from 28 facilities, comprised of academic medical centers, critical access hospitals, and short-term acute care facilities. The sample included a mix of rural and urban facilities, and two EHR systems (Epic [N=9] and Cerner [N=19]).
| Overall | Minimum | Decile_1 | Decile_2 | Decile_3 | Decile_4 | Decile_5 | Decile_6 | Decile_7 | Decile_8 | Decile_9 | Decile_10 | Maximum | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mean Performance Score | 90.3 | 83.7 | 84.0 | 86.1 | 87.9 | 88.8 | 89.5 | 90.5 | 91.8 | 92.6 | 94.2 | 98.0 | 98.2 |
| N of Entities | 28 | 1 | 3 | 3 | 3 | 2 | 3 | 3 | 2 | 3 | 3 | 3 | 1 |
| N of Persons / Encounters / Episodes | 145,846 | 310 | 4945 | 8644 | 10,299 | 5724 | 11,824 | 33,921 | 19,748 | 13,118 | 9877 | 27,746 | 6304 |
Equity
Equity
The importance of nutrition for health was marked when nutritional care was raised to the level of a human right, in close relationship to two well-recognized fundamental rights: the right to food and the right to health. Defined as the state in which everyone has a fair and just opportunity to attain their highest level of health, it is critical to include malnutrition care as a measure of improvement in health equity. Malnutrition is unique because it not only has deeply complex physiological causes, but also a multifactorial environmental, economic, and psychosocial origin. Furthermore, the COVID-19 pandemic reinforced that SDOH, including access to nutritious food, have a major impact on people’s health, well-being, and quality of life and that SDOHs are intrinsically linked to health equity. As such, an individual’s health can be influenced not only by genetics and family circumstances, but also by the environment, policies, and community they live in.
There is an inherent connection between malnutrition, food insecurity, and health equity. Food insecurity can be counted as the most relevant to cause or affect malnutrition. Food insecurity is present in households concerned about food running out, diet quality and variety, and quantity of food consumed. Screening for malnutrition can be of useful in identifying and addressing health inequities when malnutrition is caused by food insecurity. Addressing malnutrition through the implementation of quality measures that include a nutrition care plan provided by an RDN can help reduce disparities in accessing healthy food and health care. Malnutrition can be directly or indirectly affected by social determinants of health, making it a key diagnosis to improve health inequities. In addition, of all the healthcare settings, acute care houses all the possible resources and community contacts to support a patient with a malnutrition diagnosis.
The main goal of the GMCS is to measure performance related to identification and treatment of malnutrition in the acute care setting. Identifying malnutrition helps flag those who are food insecure and, conversely, identifying food insecurity may suggest the presence of risk of malnutrition. In addition, the GMCS includes an individualized nutrition care plan tailored to address any social determinant of health directly related to the malnutrition diagnosis, hence improving health equity for the patient. More importantly, because it is an interdisciplinary measure, GMCS includes many professional resources, potentially improving care while ensuring a more comprehensive discharge plan.
Feasibility
Feasibility
For the assessment of feasibility, we collected data via a feasibility scorecard, which was filled out by three hospital systems. Our goal was to determine the ease of collecting the data needed to report the measure(s) and to determine what changes, if any, would be needed in the hospital’s workflow or EHR to support implementation. We developed feasibility scorecards using the CBE’s standard scorecard template, customizing each scorecard to include all the data elements needed to calculate each measure. The scorecard includes questions about current and future capabilities related to four categories: workflow, data availability, data accuracy, and data standards. The scorecard rates each data element as 0 or 1. For data elements that received a score of 0 on any of the four feasibility criteria, we asked clinicians to provide a narrative description of what is required to achieve future feasibility.
Three hospital systems, two using Epic and one using Meditech for their EHR systems completed feasibility scorecards ahead of the workflow assessments to (1) show the availability of each data element required to calculate the measure score and (2) identify feasibility concerns at the sites. The scorecard assessed the ease of collecting each data element and the extent to which sites anticipate being able to collect more challenging data elements in the future.
Overall, the completed feasibility scorecards confirmed the feasibility of implementing the GMCS measure in clinical settings. Two hospital systems (using Epic) rated all data elements needed to compute the GMCS measure as feasible across all four categories. One hospital system (using Meditech) rated all data elements as feasible for workflow but reported feasibility issues (a score of 0) on some GMCS-specific data elements (e.g., Intervention Performed: Nutrition Care Plan, Assessment Performed: Nutrition Assessment Status) for the data availability, accuracy, and standards categories. This hospital system indicated that all data elements are supported by their workflow and would be feasible in the EHR system, but because the GMCS measure is not currently implemented at their sites, they do not have the data elements’ value sets coded and available. However, they did not anticipate any feasibility concerns for future implementation.
Given these findings supporting the feasibility of the GMCS with the current specifications, we have not made any adjustments to the final measure in response to the feasibility assessment.
Proprietary Information
Scientific Acceptability
Testing Data
Testing activities involving statistical analysis used EHR data from 28 facilities, comprised of academic medical centers, critical access hospitals, and short-term acute care facilities. The sample included a mix of rural and urban facilities, and two EHR systems (Epic [N=9] and Cerner [N=19]).
In addition, to evaluate the importance, usability and use, and face validity of the measure, we developed a web survey with a battery of questions assessing 1) clinicians’ views on whether the measure is evidence-based and important to making significant gains in healthcare quality (importance), and 2) whether potential audiences are using or could use performance results for both accountability and performance improvement to achieve the goal of high-quality, efficient healthcare for individuals or populations (usability and use). The survey was distributed to nutrition and dietetics practitioners affiliated with the Academy of Nutrition and Dietetics. Responses were collected between 3/18/24 and 3/31/24, and we received a total of N = 48 survey responses.
None
Our analysis is based on data from 145,846 eligible encounters. Eligible encounters were defined as encounters of patients who are 18 years or older and with a hospital length of stay of 24 hours or longer with any admission diagnosis between 1/1/2022 and 12/31/2022.
The sample comprises patient encounters with a mean age of 60.4 years (standard deviation = 19.1). 52.9% of the sample fell within the 18-64 age group, while 47.1% were aged 65 and older. Just over a half (54.3%) of the sample were female, and 45.7% are male. Ethnicity-wise, 99.2% of the encounters were for the non-Hispanic patients, with 0.6% being Hispanic and 0.2% missing data. The racial composition shows that 70.2% of the encounters were White, 22.5% were Black or African American, and smaller percentages belonged to other racial categories, including American Indian or Native Alaskan, Asian, Native Hawaiian or Pacific Islander, and others. Additionally, 1.2% of the sample are marked as UTD (untested or unknown), with 0.1% missing race data.
The sample displays both similarities and differences in comparison to national estimates. The mean age of individuals in our sample is 60.4 years, which is higher than the national average for inpatient admissions (49.9 years). The higher mean age in our sample may be higher than the national average because GMCS was initially developed, piloted, and validated for screening malnutrition risk in adults aged 65 years and older. Additionally, the patient populations of the facilities in our sample were skewed towards older patients. A slightly larger proportion of individuals in our sample were aged 65 and older (47.1% vs. a national estimate of 40%), while a somewhat smaller proportion fell within the 18-64 age group (52.9% vs the national estimate of 60%). Additionally, our sample has a slightly higher percentage of females (54.3%) compared to the national estimate of 49.5%, and a slightly lower percentage of males (45.7%) compared to the national estimate of 50.5%.
A higher percentage of individuals in our sample are non-Hispanic (99.2%) compared to the national estimate of 80.9%, while a lower percentage are Hispanic (0.6%) compared to the national estimate of 19.1%. When examining racial composition, our sample has a higher percentage of White individuals (70.2%) compared to the national estimate of 75.5%, and a higher percentage of Black or African American individuals (22.5%) compared to the national estimate of 13.6%. Conversely, our sample has lower percentages of individuals from other racial categories such as American Indian or Native Alaskan, Asian, and Native Hawaiian or Pacific Islander, compared to national estimates.
Our analysis is based on data from 145,846 eligible encounters. Eligible encounters were defined as encounters of patients who are 18 years or older and with a hospital length of stay of 24 hours or longer with any admission diagnosis between 1/1/2022 and 12/31/2022.
The sample comprises patient encounters with a mean age of 60.4 years (standard deviation = 19.1). 52.9% of the sample fell within the 18-64 age group, while 47.1% were aged 65 and older. Just over a half (54.3%) of the sample were female, and 45.7% are male. Ethnicity-wise, 99.2% of the encounters were for the non-Hispanic patients, with 0.6% being Hispanic and 0.2% missing data. The racial composition shows that 70.2% of the encounters were White, 22.5% were Black or African American, and smaller percentages belonged to other racial categories, including American Indian or Native Alaskan, Asian, Native Hawaiian or Pacific Islander, and others. Additionally, 1.2% of the sample are marked as UTD (untested or unknown), with 0.1% missing race data.
The sample displays both similarities and differences in comparison to national estimates. The mean age of individuals in our sample is 60.4 years, which is higher than the national average for inpatient admissions (49.9 years). The higher mean age in our sample may be higher than the national average because GMCS was initially developed, piloted, and validated for screening malnutrition risk in adults aged 65 years and older. Additionally, the populations of the samples facilities was skewed towards older patients. A slightly larger proportion of individuals in our sample were aged 65 and older (47.1% vs. a national estimate of 40%), while a somewhat smaller proportion fell within the 18-64 age group (52.9% vs the national estimate of 60%). Additionally, our sample has a slightly higher percentage of females (54.3%) compared to the national estimate of 49.5%, and a slightly lower percentage of males (45.7%) compared to the national estimate of 50.5%.
A higher percentage of individuals in our sample are non-Hispanic (99.2%) compared to the national estimate of 80.9%, while a lower percentage are Hispanic (0.6%) compared to the national estimate of 19.1%. When examining racial composition, our sample has a higher percentage of White individuals (70.2%) compared to the national estimate of 75.5%, and a higher percentage of Black or African American individuals (22.5%) compared to the national estimate of 13.6%. Conversely, our sample has lower percentages of individuals from other racial categories such as American Indian or Native Alaskan, Asian, and Native Hawaiian or Pacific Islander, compared to national estimates.
Reliability
Signal-to-noise reliability
Using signal-to-noise reliability, we tested the extent to which a facility’s quality of malnutrition care can be distinguished from that of other facilities using the GMCS. In other words, signal-to-noise reliability tests the precision of measure scores. To compute signal-to-noise reliability, we estimated the proportion of observed variability in the GMCS that is due to differences between facilities in the completeness of malnutrition care (signal variance), as opposed to variability due to differences in care within facilities (noise variance).
For each facility, we computed the facility-specific noise variance as the sample variance of the patient scores on Measure Observation 6 divided by the number of encounters in the facility minus 1. For the signal variance, we used the iterative empirical Bayes method to estimate a single value for all facilities. We then calculated a reliability coefficient for each facility as the ratio of signal variance to the sum of the signal variance and noise variance for that facility. A reliability of 1 indicates perfect reliability, where all variation in the GMCS measure scores reflects between-facility differences rather than within-facility differences.
Test-retest reliability
Using test-retest reliability, we assessed the stability of the GMCS across random samples of encounters. In other words, we tested the extent to which the GMCS measure scores are affected by sampling variability in the encounters used to compute the scores.
To do so, we drew 1,000 bootstrap samples (i.e., sampling with replacement) of encounters stratified by facility, where we kept the original number of encounters within each facility. The randomly sampled sets of encounters from a given facility are assumed to reflect an independent set of re-measurement of the GMCS scores for each facility. Adequate reliability is assumed if the GMCS measure scores calculated from the random datasets for the same IPF are similar.
Within each bootstrap sample, we computed the GMCS measure score for each facility and grouped the 1,000 samples into 500 pairs. We then calculated Spearman’s correlation (rho) and the intraclass correlation coefficient (ICC) between those measure scores in each pair of samples to assess how stable the facilities’ measure scores remains as they get computed on a different, randomly sampled set of encounters. Spearman’s correlation quantifies the strength of the rank-order association between the measure scores in each pair, where a value of 1 indicates perfect positive association. The ICC quantifies the strength of absolute agreement between the facility scores in each pair, where a value of 1 indicates perfect reliability. Following the calculations, we examined the distribution of the resulting 500 Spearman’s correlations and ICCs. Adequate reliability is assumed if the GMCS measure rates calculated from the random datasets for the same facility are similar. Note that unlike signal-to-noise, test-retest reliability does not provide a separate reliability coefficient per facility.
Signal-to-noise reliability
The summary statistics reveal high reliability across the board with coefficients averaging at 0.96 and ranging from 0.69 to 1.00. The median signal-to-noise reliability for the GMCS measure is sufficiently high (above the CBE threshold of 0.6) for all facilities (0.99). Using the CBE threshold of 0.60, all facilities in our sample would be considered to have a sufficiently high reliability.
Test-retest reliability
Spearman’s correlation averaged at 0.97 and ranged from 0.91 to 0.99, and the ICC averaged at 0.96 and ranged from 0.83 to 0.99. Given that a Spearman’s correlation of above 0.90 is often considered very strong, and an ICC of above 0.75 is considered good reliability, these results provide evidence of high reliability for the GMCS.
The results of the reliability analysis indicate that the GMCS is a highly reliable measure, both with respect to signal-to-noise and test-retest reliability. The high reliability coefficients obtained in the signal-to-noise analysis suggest that the GMCS can be reliably used to distinguish between facilities in terms of their completeness of malnutrition care. The high reliability coefficients obtained in the test-retest analysis support that the GMCS is highly stable in the face of random sampling variability in the encounters used to compute the measure scores.
| | Overall | Minimum | Decile_1 | Decile_2 | Decile_3 | Decile_4 | Decile_5 | Decile_6 | Decile_7 | Decile_8 | Decile_9 | Decile_10 | Maximum |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Reliability | 0.96 | 0.78 | 0.82 | 0.93 | 0.95 | 0.97 | 0.98 | 0.99 | 1.00 | 0.99 | 0.97 | 1.00 | 1.00 |
| Mean Performance Score | 90.3 | 83.7 | 84.0 | 86.1 | 87.9 | 88.8 | 89.5 | 90.5 | 91.8 | 92.6 | 94.2 | 98.0 | 98.2 |
| N of Entities | 28 | 1 | 3 | 3 | 3 | 2 | 3 | 3 | 2 | 3 | 3 | 3 | 1 |
| N of Persons / Encounters / Episodes | 145,846 | 310 | 4945 | 8644 | 10,299 | 5724 | 11,824 | 33,921 | 19,748 | 13,118 | 9877 | 27,746 | 6304 |
Validity
Empirical validity testing
We found significant differences in the malnutrition screening result across all four patient and encounter characteristics. The direction of the differences was as expected according to prior research, where encounters of older patients had more “at-risk” results than younger patients, longer stays more than shorter stays, readmission encounters more than non-readmissions, and encounters of non-Hispanic black patients more than non-Hispanic white patients.
The φ (phi) effect size, is a measure of association used to determine the strength and direction of the relationship between two categorical variables. It is similar to the Pearson correlation coefficient for continuous variables but is specifically designed for binary (yes/no) data. The φ coefficient ranges from -1 to 1, where -1 indicates a perfect negative association, 1 indicates a perfect positive association, and 0 indicates no association between the variables. For φ effect size coefficient, the values of 0.10, 0.30 and 0.50 indicate, respectively, small, medium and large effects.
The Cohen’s h effect size, is a measure of the difference in proportions between two groups. It is used for comparing the proportions of two groups when the data are binary (e.g., success/failure, yes/no). Cohen's h provides a standardized measure of effect size, allowing for comparisons across different studies or contexts. For Cohen’s h effect size coefficient, the values of 0.20, 0.50 and 0.80 indicate, respectively, small, medium and large effects.
Unexpectedly, we found non-significant differences for the age and LOS factors, meaning that facilities’ GMCS measure scores did not differ meaningfully across encounters of older and younger patients and across longer and shorter stays. On the other hand, as hypothesized, we found significant differences in facilities’ GMCS measure scores across the readmission status and race/ethnicity factors. The direction of the differences was as expected, where the GMCS computed among readmission encounters were lower than that of non-readmission encounters, and the GMCS computed among encounters of non-Hispanic black patients were lower than that of non-Hispanic white patients.
Cohen's d is a measure of effect size commonly used in the context of comparing the means of two groups. For Cohen’s d effect size coefficient, the values of 0.20, 0.50 and 0.80 indicate, respectively, small, medium and large effects. The Wilcoxon signed-rank test is a non-parametric statistical test used to compare two related samples or paired data, which we used in place of the paired sample t-test in cases where the normality assumption was not met. The Wilcoxon test can be used to calculate an effect size denoted as r, which represents the extent to which the measure score in one group differ from the scores in another group. To interpret the r effect size, we used the same cutoffs we used for interpreting Cohen’s d.
Data element validity
The results demonstrate no variation in the validity or the data elements by sites, with all sites reporting 100% agreement on all tested data elements.
Face validity
The responses to the web survey revealed overall support among clinicians for the face validity of the GMCS measure. Most clinicians agreed or strongly agreed that the GMCS measure score is an accurate reflection of malnutrition care quality (84.8%; Question a), and most clinicians agreed or strongly agreed that it can be used to distinguish between good and poor quality of malnutrition care (73.9%; Question b). No clinicians strongly disagreed with either of these two statements. Of the clinicians who disagreed with these statements, a commonly expressed concern was that the GMCS measure only captures the completion or documentation of the care components, rather the quality or effectiveness of the provided assessments and interventions.
Finally, we received several recommendations to strengthen the face validity of the GMCS measure. These recommendations included the addition of monitoring and evaluation of care to assess care quality and impacts, to define specific aspects needed in a nutrition care plan, to add a discharge component to the measure, and to promote dietitians’ (who are the experts in nutrition) active participation and lead in implementing the measure.
We found significant differences in the malnutrition screening result across all four patient and encounter characteristics. The direction of the differences was as expected according to prior research, where encounters of older patients had more “at-risk” results than younger patients, longer stays more than shorter stays, readmission encounters more than non-readmissions, and encounters of non-Hispanic black patients more than non-Hispanic white patients.
Unexpectedly, we found non-significant differences for the age and LOS factors, meaning that facilities’ GMCS measure scores did not differ meaningfully across encounters of older and younger patients and across longer and shorter stays. On the other hand, as hypothesized, we found significant differences in facilities’ GMCS measure scores across the readmission status and race/ethnicity factors. The direction of the differences was as expected, where the GMCS computed among readmission encounters were lower than that of non-readmission encounters, and the GMCS computed among encounters of non-Hispanic black patients were lower than that of non-Hispanic white patients.
The responses to the web survey revealed overall support among clinicians for the face validity of the GMCS measure. Most clinicians agreed or strongly agreed that the GMCS measure score is an accurate reflection of malnutrition care quality (84.8%; Question a), and most clinicians agreed or strongly agreed that it can be used to distinguish between good and poor quality of malnutrition care (73.9%; Question b). No clinicians strongly disagreed with either of these two statements. Of the clinicians who disagreed with these statements, a commonly expressed concern was that the GMCS measure only captures the completion or documentation of the care components, rather the quality or effectiveness of the provided assessments and interventions.
Finally, we received several recommendations to strengthen the face validity of the GMCS measure. These recommendations included the addition of monitoring and evaluation of care to assess care quality and impacts, to define specific aspects needed in a nutrition care plan, to add a discharge component to the measure, and to promote dieticians’ (who are the experts in nutrition) active participation and lead in implementing the measure.
The above results collectively provide evidence of high validity for the GMCS measure.
First, the two sets of statistical hypotheses tested in the empirical analyses provide evidence of validity for the GMCS based on its relationship with readmission status and race/ethnicity. In particular, the results indicate that as suggested in the literature, readmission encounters and encounters of non-Hispanic black patients are at higher risks of malnutrition compared to non-readmission encounters and encounters of non-Hispanic white patients, respectively. Further, facilities had significantly lower performance on the GMCS when computed among these higher-risk encounters compared to their lower-risk counterparts. These statistical findings together suggest that the facility scores relate to readmission status and race/ethnicity as expected, providing a form of validity evidence for the GMCS. We did not find evidence of validity for the GMCS with regards to its hypothesized relationship with age and LOS. This may be in part due to the relatively small sample size (N = 28) used in the paired t-tests in the second set of hypotheses. Though this is unlikely to impact overall scoring, future research may benefit from testing data from a larger number of facilities.
Second, the data elements used to construct the GMCS measure were validated by showing perfect agreement with the source EHR data, providing evidence that the data elements are accurate reflections of the malnutrition care provided to patient encounters.
Finally, the results from the clinician web survey suggest that there is overall agreement among clinicians regarding the face validity of the GMCS measure as being an accurate reflection of malnutrition care quality.
Risk Adjustment
Use & Usability
Use
Applies to Eligible Hospitals and Eligible Critical Access Hospitals participating in the IQR program.
Usability
The Global Malnutrition Composite Score for a facility demonstrates how many of the clinically eligible components of evidence-based malnutrition care were documented for the qualifying population. The goal for the final score is for its value to be closer to 100%. Because it is a composite score, facilities can evaluate which component is not at 100% completion, and thus, directly address the gap(s) in service to address malnutrition care. In addition, the GMCS follows the evidence- and consensus-based Nutrition Care Process (NCP), a quality improvement process designed to standardize terminology and improve consistency in nutrition care to improve outcomes. The four steps of the NCP, Nutrition Assessment and Reassessment, Nutrition Diagnosis, Nutrition Intervention, and Nutrition Monitoring and Evaluation, are captured in the measure observations of the GMCS. Though not a formal part of the NCP, nutrition screening is a process by which individuals enter the NCP after identification as at risk for malnutrition. Consistency in these steps allows for improved and individualized care to treat patients based on their level of malnutrition and/or malnutrition risk based on both clinical risk factors and patient preferences.
The process for risk identification, assessment, diagnosis, and treatment of malnutrition necessitates an interprofessional care team that begins with the identification of an initial risk population for a more thorough nutrition-focused physical examination (NFPE) by the RDN. The RDN in turn provides the necessary treatment recommendations to address nutritional status and the clinical indicators that inform a medical diagnosis of malnutrition completed by a physician or other eligible provider. The four components, when measured only individually, provide merely a fraction of the necessary information on quality of care for patients at-risk of malnutrition. For example, knowing which patients have been assessed out of those who were initially identified as at-risk, but not knowing if the appropriate proportion of patients were screened upon admission, would be an insufficient assessment of quality of care. The GMCS’ inclusion of both individual components and a composite score allows organizations to evaluate data and drive change based on their individual performance.
The GMCS supports timely malnutrition risk screening and hand off to the RDN for appropriate nutrition assessment for patients at-risk for malnutrition. For patients identified with a moderate or severe malnutrition from the nutrition assessment, best practice also recommends a medical diagnosis by a physician or other qualified healthcare professional and the execution of the nutrition care plan by an RDN. Evidence demonstrates that implementing a standardized protocol for screening, assessment, diagnosis, and care planning results in better identification of malnourished patients and subsequent improvements in rates of nutrition intervention for the malnourished.
Nutrition Screening
Nutrition screening is defined as the process of identifying individuals who may have a nutrition diagnosis and therefore may benefit from nutrition assessment and interventions by an RDN. All hospitalized patients should be screened for risk of malnutrition and/or nutrition-related problems. Nutrition screening tools are intended to be simple and require little to no training to execute, thereby allowing for completion by a variety of health professionals, particularly in the acute care setting.
As a workflow element of the NCP, nutrition screening plays an integral role in clinical workflows as the entry point to the NCP to ensure patients are properly channeled to RDNs for nutrition assessments and associated interventions. Likewise, referrals to and/or care pathways including the RDN or nutrition, serve as a parallel route, allowing clinicians to use their judgement in identifying patients that may benefit from RDN assessment and interventions. Because this is a requirement by The Joint Commission, this workflow should already be present in acute care facilities. Though optimization work may be needed, little effort should be required from implementors.
If there is a need to improve the score in this observation, measure entities may benefit from identifying barriers to completing the screening. A process improvement project can be done to identify and address the facility specific barrier(s). As a reminder, in the absence of nutrition screening, a hospital dietitian referral and/or care pathway decision point can prompt the RDN to complete a nutrition assessment, identify the nutrition status, and develop a patient-specific nutrition care plan of malnutrition treatments and interventions. Creating a standardized set of criteria for initiating a dietitian referral may improve appropriate referral rates while minimizing referrals that do not truly require RDN assessment and/or intervention.
Nutrition Assessment
The second step in optimal nutrition care for patients identified as at risk for malnutrition is performing a nutrition assessment. Nutrition assessment is a systematic approach for collecting, classifying, and synthesizing essential data to describe nutritional status. Using structured nutrition care terminology supports clear and consistent communication of nutrition care indicators representing the unique contribution of the RDN, relevant to supporting a nutrition status problem(s), and understandable to the interdisciplinary team. In the case of validated malnutrition assessment tools, structured nutrition concepts are useful because the indicators are defined and determined to be accurate to support malnutrition identification. RDNs provide a critical analysis of the nutrition findings and compare them against suitable reference standards with a distillation of the most relevant data to support the existence of a nutrition problem and its etiology, also known as the cause and/or contributing factors, of a nutrition problem.
If the scores for this observation were to be low, measure entities are encouraged to involve the facility RDN, and any other affected staff, in reviewing the current process of alerting the RDN when there is a referral or an At Risk nutrition screening result. Once the current process is defined, and gaps are identified, staff can develop a process that addresses the needs with the available resources.
Nutrition Diagnosis
Documentation of a malnutrition diagnosis in the electronic health record (EHR) may vary widely based on hospital policies, RDN and physician/eligible clinician practices, and use of the NCP. Diagnosis and documentation of malnutrition to address a patient’s condition with an appropriate plan of care and communication of patient needs to other care providers is essential to the provision of high-quality care. The nutrition diagnosis problem, found in the Nutrition Assessment, using structured NCP terminology in a PES (problem, etiology, and sign/symptom) statement is a concise communication to the physician and/or eligible clinician of the finding, its cause, and the supporting evidence, thus offering a focused summary of the nutrition status and eliminating the need for providers to review and digest the RDN nutrition assessment analysis.
If this observation has a low score, measure entities will need to address identify the common reasons for the gap between RDN nutrition diagnosis during the Nutrition Assessment, and the physician or eligible clinician’s malnutrition diagnosis. Most of the time, educating the physician or eligible clinician in the process of the RDN nutrition diagnosis, and how the PES statement supports the malnutrition diagnosis, is a step towards improving this score. Resources including the EHR system can be leveraged to support improved communication between RDN and physician or eligible clinician and close the gap in care.
Nutrition Care Plan
Optimal nutrition care best practice is comprised of appropriate recognition, diagnosis, and documentation of the nutrition status of a patient to address their condition with an appropriate plan of care addressing the cause of the problem and communicating patient needs to other care providers. Nutrition interventions that address the malnutrition diagnosis in hospitalized patients are key to support patient outcomes.
Appropriate and timely identification of patients eligible for a nutrition assessment, along with subsequent physician or eligible clinician notification of nutrition problem statements and recommended interventions including markers for monitoring the effectiveness of the intervention, are critical in diagnosing malnutrition and providing the necessary follow-up care.
It is important to note that as an established process, the RDN usually includes the Nutrition Care Plan as part of the Nutrition Assessment done in that observation step. Addressing low scores in this step is expected to be mostly through the analysis of steps in place to ensure the Nutrition Care Plan is developed by the RDN, once the malnutrition diagnosis is identified, and that this is actually coded/mapped to the correct data set in the reporting process.
Since initial publication, a total of 21 JIRA tickets have been submitted for the 65+ version of this measure. Additionally, the measure has received numerous expert reviews throughout two separate Annual Update cycles. Several main themes emerged from these tickets, namely:
- Calculation errors in specific scenarios (negative malnutrition risk screening followed by completion of other components; negative nutrition assessment followed by completion of other components)
- Overlapping codes in nutrition screening and assessment value sets
- Confusion over recommended screening tools based on selected value set codes
- Unclear wording in header description of measure observations and aggregation
- Measure calculation logic that more fairly captures performance of measure observations
Based on issues identified in JIRA tickets, several changes were made to measure logic, and corresponding header language, namely:
- Update to the measure logic to eliminate calculation errors
- Adjusting age parameter to start during encounter rather than measurement period
- Removal of redundant logic in Measure Observation definitions
- Inclusion of malnutrition diagnoses from previous encounters to better align with clinical practice
- Inclusion of completed Measure Observations during adjacent Emergency Department and/or Observation encounters associated with the eligible Inpatient Encounter
- Updates to names and content of nearly every measure value set to reduce confusion, eliminate overlap, and better capture clinical practice
- Updates to eligible occurrences (mathematical denominator) to more fairly measure performance
Because no formal measure reporting has yet taken place, no concrete data is available to determine the impact of updates to measure logic and header on performance scores. However, the recent rate of JIRA ticket submissions related to the 65+ population version of measure has consistently declined, potentially indicating that issues and questions have been satisfactorily addressed. Likewise, all feedback received during the two Annual Update cycles, including input from a multidisciplinary technical expert panel and patient representatives, has been satisfactorily addressed.
Over the course of measure testing, update, and review, it was clear that in certain circumstances, the logic led to erroneously high performance scores. As a result, a Known Issue, EKI-21, was published to provide guidance to implementers.
Because no formal measure reporting has yet taken place, the QMS team is unaware of other impacts, including those that may directly affect patients. Anecdotally, implementers frequently communicate to QMS that the measure has increased the visibility and perceived importance of malnutrition care in their facilities, but no data is available to substantiate these statements.
Comments
Staff Preliminary Assessment
CBE #3592e Staff Assessment
Importance
Strengths:
- The developer cites a few studies as supporting evidence that addressing malnutrition via early identification and nutrition interventions improves health outcomes (e.g., reduced hospital readmissions rates and length of stay).
- The developer assessed performance scores using data from 28 facilities, including a mix of rural and urban and across two electronic health record systems (Epic and Cerner). Performance scores ranged from 84% to 98% across the 28 facilities, with an overall mean of 90.3%. The developer also conducted a topped-out analysis and found that despite a relatively narrow performance gap, there is room for quality improvement even among the test sample of high-performing hospitals, as 16 (57.1%) of facilities performed either worse than or not significantly different from the average facility in the data.
- The developer explains a National Dialogue was convened in 2018 among multi-stakeholder representatives (providers, payers, patients, and caregiver advocacy groups, etc.). There was consensus from the group that evaluation of nutrition status and provision of high-quality nutrition care should be integrated into protocols, pathways, and models throughout the care spectrum.
- The developer notes that a version of this measure evaluating performance in adults aged 65 years and older is currently endorsed and active in the CMS Hospital Inpatient Quality Reporting program. The submitted measure represents a substantive change, as the measure population will now include all adults aged 18 years and older.
Limitations:
- The developer cites 2011 clinical practice guidelines from American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.) however the grading and level of evidence appear to be on the moderate to lower end of the grading scale (levels C or E).
- The logic model shows the clinical workflow for malnutrition care but does not clearly show the relationship between health care processes and the desired health outcome.
- Beyond the 2018 National Dialogue, the developer does not provide evidence of more recent discussions with patients/caregivers about measure meaningfulness.
Rationale:
- There is a business case for the measure with some supporting evidence. More recent evidence showing the importance of the measure to patients should be considered.
Feasibility Acceptance
Strengths:
- The developer conducted a feasibility assessment of this eCQM in three hospital systems representing two electronic health record vendors, Epic and Meditech. The developer provided the required eCQM Feasibility Scorecard. The Epic hospitals system rated all data elements as feasible.
The developer notes that this is not a proprietary measure, indicating that the measure can be used without substantial burden.
Limitations:
- The developer does not indicate in their feasibility assessment whether all required data elements used to calculate this measure are routinely generated and used during care delivery. The developer explained that the Meditech hospital system reported feasibility issues in data availability, accuracy, and standards for several measure-specific data elements due to data element value sets not being coded. However, the Meditech hospital system did not anticipate any feasibility concerns once the measure was implemented in their system.
Rationale:
- The developer conducted a feasibility assessment across three hospital systems representing two electronic health record vendors, Epic and Meditech. The developer identified some issues with data element availability, accuracy, and standards at one of the three hospital systems, but explained these issues would be resolved with measure implementation. The developer does not indicate in their feasibility assessment whether all required data elements used to calculate this measure are routinely generated and used during care delivery.
Scientific Acceptability
Strengths:
- The measure is clear and well defined.
- Signal-to-noise reliability for all facilities is above the threshold of 0.6 ranging from 0.69 to 1 with an average of 0.96.
- Test-retest reliability (Spearman Rank and ICC) met the threshold of 0.
Limitations:
- Low number of entities in reliability calculations.
Rationale:
- The measure is well defined. Reliability was assessed at both the patient and entity level. Reliability statistics are above the established thresholds.
Strengths:
- The developer conducted accountable entity-level empirical validity using data from 28 hospitals. As hypothesized and in alignment with the literature, the developer found significant differences in measure scores across readmission status and race/ethnicity factors. Unexpectedly, the developer did not find significant differences in measure scores across age or length of stay, but explained this may be due to the relatively small sample size of facilities. The developer conducted data-element level validity testing in a total sample of 180 encounters across six facilities and found 100% agreement on all tested data elements. Finally, the developer collected face validity feedback from clinicians using a web-based survey. 73.9% of clinicians agreed or strongly agreed that the measure can be used to distinguish good from poor quality of care.
Limitations:
- Within the face validity survey, some clinicians noted a commonly expressed concern that the measure only captures completion/documentation of the care components, rather than the quality/effectiveness of the assessment or intervention.
Rationale:
- The developer assessed measure validity using accountable entity-level empirical validity, data-element level validity, and face validity. The interpretation of the empirical results supports an inference of validity.
Equity
Strengths:
- The developer described how the measure contributes to efforts to address inequities in healthcare. The measure includes an individualized nutrition care plan tailored to address any social determinant of health directly related to the malnutrition diagnosis, hence improving health equity for the patient.
- Within the validity section of the measure submission, the developer conducted empirical testing to assess whether encounters' malnutrition screening results differ significantly in non-Hispanic black patients vs. non-Hispanic white patients. The developer found that measure scores of non-Hispanic black patients were lower than that of non-Hispanic white patients.
Limitations:
- For future submissions, the developer may consider whether the measure can identify statistical differences in performance across other racial categories and/or economical status.
Rationale:
- The developer evaluated disparities in performance by race/ethnicity. The developer assessed how to measure contributes to efforts to address inequities in health care.
Use and Usability
Strengths:
- The developer indicated the measure is currently in use in the Hospital Inpatient Quality Reporting program.
The developer explained that because the measure is a composite score of four components, facilities can evaluate which component is not at 100% completion, and thus, directly address the gap(s) in service to address malnutrition care. The developer describes how measure entities can improve performance on each component if the observation has a low score. For example, for measure observation #3 (nutrition diagnosis) educating the eligible clinician in the process of the RDN nutrition diagnosis, and how the PES statement supports the malnutrition diagnosis. - The developer described that feedback on measure performance and implementation is collected via JIRA. The developer described that prior measure feedback resulted in several changes to the measure logic and corresponding header language.
Limitations:
- The developer notes that data from measure reporting is not yet available, making it challenging to accurately assess year-over-year performance.
Rationale:
- The measure is currently used in the Hospital IQR program. The developer describes actions measures entities can take to improve performance on each component if the observation has a low score. The developer did not report any findings on the progress on improvement as data from measure reporting are not yet available.
Committee Independent Review
Nearly Topped Out
Importance
Disagree with staff assessment. Although the performance gap shown is relatively modest, there remains a substantial proportion of facilities that could substantially improve their performance (i.e. deciles 4 and below are less than 89%). I do not believe additional discussions with patient groups about the meaningfulness of the measure are necessary.
Feasibility Acceptance
Disagree with staff assessment. The feasibility issue identified seems to be idiosyncratic to the single site assessed, and not a widespread issue with specifications.
Scientific Acceptability
Agree with staff assessment. Testing supports the reliability of the numerator and denominator.
Agree with staff assessment. Testing supports the validity of the measure result.
Equity
Agree with staff assessment. I would have preferred to see disparities identified in the performance statistics, but the logic supporting a health equity value to this measure is sound.
Use and Usability
Disagree with staff assessment. From my reading of the criteria for maintenance of endorsement, progress on improvement/change over time isn't necessary to show, although I agree it would be useful in this context given that it is an IQR measure.
Summary
A useful measure that addresses an important area, but change over time scores are not available and are an important consideration as the Committee assesses whether continued endorsement and implementation of the measure presents a substantial opportunity for improvement, or whether performance is "stuck" and the measure may need to be retired. Will need this examination on next maintenance review.
do not support
Importance
agree with staff assessment
Feasibility Acceptance
agree with staff assessment
Scientific Acceptability
agree with staff assessment
Hospitals widely vary in the ability to get nutrition assessment/consult on weekends. This measure would measure the ability of the hospital to provider services on weekends which may be discriminatory against smaller or more rural hospital systems. That the face validity surveys did not comment on this may be a reflection on the stakeholders who were invited to participate.
Equity
Hospitals widely vary in the ability to get nutrition assessment/consult on weekends. Patients are not randomly distributed to hospitals with disadvantaged groups being more likely to go to hospitals with fewer resources. Unclear whether this measure would distinguish between hospital quality or the nutritional status of the patients. Further research needed.
Use and Usability
agree with staff assessment
Summary
Overall this measure is very complex and thus difficult to interpret the results. I have concerns that it assumes that all hospitals have equal availability of nutrition services. Increasing the measure to 72 hours of admission would potentially address the weekend issue.
Support with minor importance issues addressed
Importance
Agree with staff assessment. Additional information incorporated into logic model and citations supporting improved outcomes would be beneficial. In addition, including guideline grades and strength of recommendations if applicable would be helpful in form.
Feasibility Acceptance
Disagree with staff assessment. The feasibility scorecard workflow evaluation indicates that no facilities identified workflow issues and which is defined as the extent to which the data elements necessary to calculate the measure are captured during the standard process of care. The explanation for the Meditech site seems reasonable.
Scientific Acceptability
Agree with staff assessment for accountable entity level testing. Sample appears adequate and results indicate high reliability.
Agree with staff assessment.
Equity
Agree with staff assessment. Would be useful to present stratified results in the equity section.
Use and Usability
Disagree with staff assessment. Since data are not yet available from IQR, it appears the measure meets the evaluation criteria for maintenance measures in the E&M guidebook. In addition, feedback received appears to have been addressed in revisions to specifications.
Summary
Minor issues with regarding documenting evidence in importance section. Overall appears the measure is feasible, reliable, valid and usable. Data from program implementation will be important for the next evaluation given that performance rates are relatively high from the limited sample.
Support this measure
Importance
It is an important measure with follow up for patients that are facing malnutrition. It may allow for interventions to be put in place that can help sustain the patient upon discharge and possibly prevent future hospitalizations
Feasibility Acceptance
Seemed to be acceptable to those reporting the measure.
Scientific Acceptability
There is a direct linkage between malnutrition and health.
The need to identify and address malnutrition were presented.
Equity
Need to identify and address the disparity --- and see if interventions can be put in place whether related to SDOH or other issues.
Use and Usability
In the locations identified, it has been recorded. For any patient hospitalized that has a malnutrition condition, it should be recorded as addressing the malnutrition could result in improved health.
Summary
Important to record for the health and well-being of persons that are malnutritioned.
Support...
Importance
Agree with staff assessment. Need more information about the quality of evidence supporting each of the measure components.
Feasibility Acceptance
Do not agree with staff assessment. The measure is feasible for testing sites that use Epic. For the sites that use Meditech, the developer stated that the workflow for the measure is supported.
Scientific Acceptability
Agree with staff assessment. The testing methods are unusual, but assuming they are sound, the results indicate reliability.
Agree with staff assessment. However, I am concerned with the results of the analyses of the GMCS measure with age and LOS and would like more discussion about why the results did not align with the stated hypotheses.
Equity
Agree with staff assessment.
Use and Usability
Agree with staff assessment.
Summary
... but only if more evidence supporting each of the measure components is provided (including its quality and consistency).
CBE #3592e: Global Malnutrition Composite Score
Importance
Agree with Staff Preliminary Assessment
Feasibility Acceptance
Agree with Staff Preliminary Assessment
Scientific Acceptability
Agree with Staff Preliminary Assessment
Agree with Staff Preliminary Assessment
Equity
Agree with Staff Preliminary Assessment
Use and Usability
Agree with Staff Preliminary Assessment
Summary
Complicated. Overall this measure is very complex which makes it difficult to interpret the results - which we do not have yet. This measure is just getting started in the IQR program and we do not yet have good foundation data. We need more time to evaluate performance. I also wonder if it assumes that all hospitals have equal availability of nutrition services. Small hospitals may be challenged by weekend admissions and increasing the requirement to within 72 hours of admission would potentially close this gap.
Important measure that needs…
Importance
This is an important measure addressing an issue that receives far too little attention. My concern is that malnutrition may be more of a problem after admission than at admission, but there is no criterion for checking malnutrition status, or the reasons for it, after admission. It is good that the age limit has been lowered to include all adults, as this cn happen at any age.
Feasibility Acceptance
Agree with staff assessment
Scientific Acceptability
Agree with staff assessment
Agree with staff assessment
Equity
Agree with staff assessment
Use and Usability
Agree with staff assessment
Summary
Important measure that needs a little work to optimize it .
support
Importance
likely a net benefit to measurement as well as a performance gap
Feasibility Acceptance
while one can anticipate feasibility issues in some settings, the information provided suggests that reasonable mechanisms for data capture are possible
Scientific Acceptability
agree with staff assessment
agree with staff assessment
Equity
agree with staff assessement
Use and Usability
At the moment, Met seems like the best category based on the evaluation rubric. When data from measure reporting becomes available, usability should be reconsidered.
Summary
No specific issues exist to preclude endorsement at this time.
Public Comments
Public Comment Shared During May 29 Listening Session
Hi! I'd like to say I'm very happy that the developer has changed the age to 18 and above, because before this did not incorporate individuals who perhaps have digestive diseases or eating type of disorders. And there's a number of individuals for other reasons that, perhaps you have to check their malnutrition, individuals that are in an advanced state of some type of physical degenerative state. So, I'm very glad that they've [the developer] changed this, and I think we'll be able to incorporate this across many areas. And I just wanted to put my support in there for that. Thank you.