- Association Between Malnutrition and Clinical Outcomes
- Patients requiring intensive care unit (ICU) care and diagnosed with malnutrition (using Subjective Global Assessment [SGA]) may have higher hospital mortality compared to well-nourished patients requiring ICU care.
- Patients requiring ICU care and diagnosed with malnutrition (using SGA) are likely to experience prolonged hospital length of stay compared to well-nourished patients requiring ICU care.
- Patients requiring ICU care and diagnosed with malnutrition (using Mini Nutritional Assessment [MNA]) may experience more hospital acquired complications compared to well-nourished patients requiring ICU care.
- Patients hospitalized due to traumatic injury and screened at risk of malnutrition (using Nutritional Risk Screening [NRS]-2002) may experience more hospital acquired conditions compared to well-nourished patients.
- Patients hospitalized with heart failure and diagnosed with malnutrition (using several different measurement tools) may have higher mortality compared to well-nourished patients with heart failure.
- Patients hospitalized with cancer and diagnosed with malnutrition (using SGA) may experience prolonged hospital length of stay compared to well-nourished patients.
- Patients hospitalized with cirrhosis awaiting transplantation and diagnosed with malnutrition (using SGA) may have higher pre-transplant mortality compared to well-nourished patients.
- Effectiveness of Screening on Clinical Outcomes
- No studies met inclusion criteria to address effectiveness of screening or diagnostic assessment on clinical outcomes, primarily because studies lacked an appropriate control group.
- This evidence gap underscores the need for future research that addresses the effectiveness of various measurement tools for malnutrition on clinical outcomes. Such research is vital to standardize malnutrition assessment and further understand its downstream implications on patient-relevant outcomes.
- Effectiveness of Hospital-Initiated Interventions for Malnutrition
- Hospital-initiated malnutrition interventions (i.e., specialized nutrition care, protein/calorie supplementation) likely decrease mortality compared to usual care.
- Hospital-initiated malnutrition interventions may improve quality of life compared to usual care.
- No difference was observed between hospital-initiated malnutrition interventions and usual care for length of stay, readmission rates, and hospital acquired conditions compared to usual care.
- Evidence was insufficient to address the effect of hospital-initiated malnutrition interventions on activities of daily living and discharge disposition compared to usual care.
Objectives. To review the association between malnutrition and clinical outcomes among hospitalized patients, evaluate effectiveness of measurement tools for malnutrition on clinical outcomes, and assess effectiveness of hospital-initiated interventions for patients diagnosed with malnutrition.
Data sources. We searched electronic databases (Embase®, MEDLINE®, PubMed®, and the Cochrane Library) from January 1, 2000, to June 3, 2021. We hand-searched reference lists of relevant studies and searched for unpublished studies in ClinicalTrials.gov.
Review methods. Using predefined criteria and dual review, we selected (1) existing systematic reviews (SRs) to assess the association between malnutrition and clinical outcomes, (2) randomized and non-randomized studies to evaluate the effectiveness of malnutrition tools on clinical outcomes, and (3) randomized controlled trials (RCTs) to assess effectiveness of hospital-initiated treatments for malnutrition. Clinical outcomes of interest included mortality, length of stay, 30-day readmission, quality of life, functional status, activities of daily living, hospital acquired conditions, wound healing, and discharge disposition. When appropriate, we conducted meta-analysis to quantitatively summarize study findings; otherwise, data were narratively synthesized. When available, we used pooled estimates from existing SRs to determine the association between malnutrition and clinical outcomes, and assessed the strength of evidence.
Results. Six existing SRs (including 43 unique studies) provided evidence on the association between malnutrition and clinical outcomes. Low to moderate strength of evidence (SOE) showed an association between malnutrition and increased hospital mortality and prolonged hospital length of stay. This association was observed across patients hospitalized for an acute medical event requiring intensive care unit care, heart failure, and cirrhosis. Literature searches found no studies that met inclusion criteria and assessed effectiveness of measurement tools. The primary reason studies did not meet inclusion criteria is because they lacked an appropriate control group. Moderate SOE from 11 RCTs found that hospital-initiated malnutrition interventions likely reduce mortality compared with usual care among hospitalized patients diagnosed with malnutrition. Low SOE indicated that hospital-initiated malnutrition interventions may also improve quality of life compared to usual care.
Conclusions. Evidence shows an association between malnutrition and increased mortality and prolonged length of hospital stay among hospitalized patients identified as malnourished. However, the strength of this association varied depending on patient population and tool used to identify malnutrition. Evidence indicates malnutrition-focused hospital-initiated interventions likely reduce mortality and may improve quality of life compared to usual care among patients diagnosed with malnutrition. Research is needed to assess the clinical utility of measurement tools for malnutrition.
The table below summarizes our findings for Key Question 1 of our review (concerning the association between malnutrition and clinical outcomes). Our searches identified 6 SRs (including 43 relevant unique studies) evaluating the association between malnutrition and clinical outcomes among hospitalized patients.
We sought to address three Key Questions:
- Key Question 1. What is the association between malnutrition and clinical outcomes among hospitalized patients?
- How do outcomes vary depending on measures or tools used to detect malnutrition?
- Are patient-related risk factors, such as increased age or certain pre-existing health conditions, associated with poorer clinical outcomes?
- Key Question 2: What is the effectiveness of screening or diagnostic assessment for malnutrition among hospitalized adults?
- In studies that report on clinical outcomes, what is the accuracy of screening or diagnostic tools for malnutrition?
- In studies that report on clinical outcomes, what is the effectiveness of screening or diagnostic tools on measures of nutrition (nutritional stores)?
- What is the impact of the use of screening or diagnostic tools on clinical outcomes?
- Key Question 3: Among patients diagnosed with malnutrition, what is the effectiveness of hospital-initiated interventions used to treat malnutrition on clinical outcomes?
Findings in Relation to What Is Known
Our review underscores many known limitations of research on nutritional measurement tools: these problems include varied definitions of malnutrition in the literature, lack of validated tools, and lack of an accepted reference standard.
As anticipated, we found malnutrition studies have employed a wide range of definitions for malnutrition. For example, one prior SR of nutrition-focused interventions for hospitalized patients by Feinberg et al. (2017) included 244 RCTs. However, over 40% of these trials (105 of 244) did not use a commonly available measurement tool to confirm the diagnosis of malnutrition, instead relying on presence of severe disease, weight loss, BMI (or other biomarkers), or clinical opinion to define a malnourished population. Thus, we focused on more recent literature, our SR captured studies published from 2000 onwards to try to capture literature better aligned with current recommendations for screening and diagnosis established by the GLIM taskforce (2019). However, we found that relatively few studies have used criteria aligned with the GLIM criteria to define malnourished patients. Since the GLIM consensus criteria were established relatively recently, it is likely that not enough time has passed for studies to employ these criteria in trials. The few studies identified in our searches that focused on validation of GLIM criteria were excluded for KQ2 as they did not meet study design criteria for our report (i.e., they were uncontrolled trials).
Even when malnourished patients are identified using validated tools, inconsistent agreement and reliability may be problematic, particularly given the large number of different instruments currently in use. Skipper et al. (2012) performed an SR to assess the validity of available screening tools used to identify patients at-risk of malnutrition. Authors found that available tools only achieved moderate- rather than high-level validity, agreement, and reliability, and that there were large variations in these measures for all tools. Authors attributed the large range in validity and reliability to researchers using different reference standards to validate tools and suggested that use of a single reference standard would narrow the ranges of reliability and validity. Differences in validity of measurement tools likely contributed to variations observed in the SOE for the association of malnutrition with poor clinical outcomes. Use of unvalidated instruments as reference standards was also a common reason for excluding studies from KQ 2.
Regarding hospital-initiated interventions, our findings aligned with prior reviews for the important outcome of mortality, despite large differences in how malnutrition was defined. Two previous systematic reviews, one by Feinberg et al. (2017) and the other by Gomes et al. (2019), examined the efficacy and safety of various nutrition-focused interventions used to treat hospitalized patients. These reviews included studies published prior to 2000 and included trials of patients designated malnourished based on biometrics (e.g., BMI, weight loss, serum albumin level), severity of disease (e.g., any intensive care unit admission) or clinical judgment. Like our review, these reviews also found that nutrition-focused interventions decreased mortality.
Implications for Clinical Practice, Education, Research, or Health Policy
Our SR was intended to inform the deliberations of a congressional panel charged with developing quality measures for malnutrition-related hospital readmissions. Such measures would potentially support assigning accountability for the assessment and treatment of malnutrition in hospitalized adults, with an emphasis on the needs of older frail adults. Although our review confirmed that malnutrition is associated with poor outcomes and that specific interventions may be beneficial, it also highlights many challenges of drawing conclusions from this evidence base, starting with fundamental questions regarding how malnutrition should be defined and measured.
Variations in how malnutrition is defined and measured pose a challenge for hospitals seeking to standardize processes for screening, further assessing, and documenting diagnosis of malnutrition. As previously mentioned in the Background section of this review, national survey data indicate that only 38% of hospital professionals report using a recognized tool to screen for malnutrition, and only 23% report using one for diagnostic assessment in those who screen positive. Another challenge to standardizing these processes is ensuring that the definition or criteria for malnutrition reflected within available measurement tools are consistent with current documentation and coding requirements for malnutrition. For instance, the International Classification of Diseases, 10th edition (ICD-10) coding system has different codes for varying levels of severity of malnutrition that are not always aligned with how malnutrition is defined using GLIM or ASPEN criteria.
Further research such as a multi-arm trial that randomizes patients to different measurement tools would allow researchers and practitioners to further understand the clinical utility of each tool, including downstream potential benefits and harms. Similarly, studies that acknowledge the notable overlap in variables utilized amongst these measurement tools may help identify which variables have the greatest sensitivity and specificity, impact on clinical outcomes, and lead to development of a comprehensive tool. For example, some tools may benefit from removing outdated variables, such as BMI. Such research could support the complex and evolving task of disentangling disease (i.e., severity of illness) and nutritional status.
As for malnutrition interventions, no studies were identified specific to the effect of parenteral or enteral nutrition in patients diagnosed with malnutrition, representing a significant evidence gap in malnourished hospitalized patients. Further research in this area is essential to determine which malnourished patient populations benefit from specific types of interventions. Key guidance for hospitals is needed on how to standardize processes for screening, diagnosing and documenting malnutrition in order to inform development of quality measures and to improve patient outcomes.
This review does not provide cost information.
As noted, in consultation with our TEP and to align with current consensus recommendations for screening and diagnosis, our review was limited to studies that confirmed a diagnosis of malnutrition using commonly available measurement tools. Thus, our findings regarding clinical outcomes and interventions are applicable to other patients with malnutrition determined in this way. However, conversely, it remains unclear to what extent these findings are generalizable across malnourished populations not identified in this way.
Using these criteria to define malnutrition also indirectly led to exclusion of studies assessing more invasive nutritional interventions for hospitalized patients such as parenteral nutrition or enteral nutrition. Studies assessing these interventions initiated treatment based on severity of illness, clinical judgement, or surrogate markers of malnutrition, such as blood serum markers and other biometrics. Thus, our findings only extend to two types of hospital-initiated malnutrition interventions, specialized nutrition care (i.e., consultation with a nutrition specialist and individualized goals) and oral protein/calorie supplementation. Also, while included studies enrolled older patients and patients with a range of underlying clinical conditions, we did not have enough studies to conduct subgroup analysis to determine if the effects of treatment differed depending on patient characteristics.
Limitations and Suggestions for Future Research
A key challenge for assessing studies of malnutrition is determining how malnutrition should be defined. Although many studies have defined malnutrition using biomarkers (e.g. BMI, weight loss, serum albumin levels) experts have expressed concern that these measures are not reliable indices of malnutrition by themselves. For instance, serum albumin levels often fluctuate in response to physiological stress and other factors unrelated to a patient's nutritional status. Similarly, metrics such as BMI fail to account for variations related to gender, age, race, or body type. Other studies have used severity of disease (e.g. any intensive care admission) as a proxy for or criterion to intervene on malnutrition often without formal diagnostic assessment. Therefore, there is wide variability in how malnutrition has been identified and studied.
The wide range of definitions (and measurement tools) has created challenges for clinical practice and malnutrition research. To address this problem, in 2019, the GLIM taskforce recommended a two-step approach to identifying malnutrition that involves 1) screening for malnutrition using a valid tool, and 2) performing a formal diagnostic assessment for those who screen positive.18 A formal diagnosis of malnutrition according to GLIM recommendations requires patients to have at least one etiologic factor (reduced food intake, hypercatabolic burden of disease) and one phenotypical factor (non-volitional weight loss, low body mass index (BMI), low skeletal muscle mass). However, GLIM recommendations have yet to be clinically validated or widely applied in clinical practice or research settings.
Finally, defining malnutrition requires the validation of measurement tools against a gold standard. However, through discussions with our TEP, we recognized that there currently is no universally agreed upon gold standard for malnutrition assessment and measurement. For the purposes of this report, we selected, with input from our TEP and SMEs, imaging modalities to quantify and evalute body composition (i.e., muscle and adipose tissues) as the gold standard and SGA as a semi-gold standard for classifying malnutrition. However, use of imaging specifically to assess malnutrition is infrequent and has important limitations, including cost, radiation exposure, and need for serial studies. In addition, at present, none of the current assessment tools (malnutrition measurement tools, the GLIM framework, or imaging modalities) capture micronutrient deficiencies. Development of an accepted gold standard for defining malnutrition is key to supporting future clinical care and research.
Effectiveness of Measurement Tools
This review highlights several important knowledge gaps in the current literature that future research needs to address. One is the relatively small number of studies that used available measurement tools to identify malnutrition. As noted, many studies identified malnutrition based only on biometrical measures, such as serum albumin levels, BMI, and weight, despite consensus that albumin and BMI should not be used to define malnutrition in practice or research. Thus, future studies assessing the impact of malnutrition on outcomes or evaluating malnutrition-focused interventions should use known tools to establish malnutrition status. Future studies would also benefit from stratifying patients by age, gender, and frailty.
Second is the absence of studies meeting inclusion criteria that address key question 2. The absence of studies addressing the clinical utility (effectiveness) of measurement tools for nutrition screening and diagnostic assessment (KQ2) does not necessarily imply that these tools are ineffective. However, it reveals the need for appropriately designed studies to better understand the downstream consequences of nutrition screening, including subsequent diagnostic assessment, management, and clinical outcomes is extremely important given that hospitals are mandated to provide nutrition screening for all hospitalized patients within 24 hours of admission.
One way to indirectly address this is to determine if one measureâ€”SGA, imaging modalities, or the new GLIM criteriaâ€”better captures clinically important malnutrition. To assess which measure is more effective, one could envision a multi-arm clinical trial that compares multiple tools and techniques. For example, a study could screen hospitalized patients as mandated by the Joint Commission, and further assess at-risk patients with each of these diagnostic assessment tools; results of one diagnostic assessment tool could then be used to randomize patients (i.e., those diagnosed with malnutrition) to nutritional interventions. This would provide better understanding of the clinical course for patients who test negative by various diagnostic assessments and provide insights on potential harms of using specific tools. Furthermore, given significant overlap in the variables utilized in the tools, future research could also support identification of which variables have the greatest impact on sensitivity and specificity in prospective clinical studies.
Finally, the studies addressing the efficacy of malnutrition-focused interventions were limited to studies of specialized nutrition care (consultation with a dietitian to set goals for protein and caloric intake) or increased protein/calorie supplementation. These studies had several shortcomings, including high risk of bias and poor reporting of treatment-related adverse events. Most studies were rated as high risk of bias or had some concerns with outcome-level risks of bias. In most cases, studies failed to report adequate randomization, had unknown allocation concealment, and were unclear if all pre-specified outcomes were reported. Some studies also had high dropout rates (>20%) and did not blind outcome assessors. These limitations, along with inconsistencies in the findings for some outcomes and lack of precision for others, downgraded the overall strength of the evidence to low or insufficient for most outcomes. These studies either did not report on harms of treatment or reported on harms in a manner that did not allow us to synthesize the data. Thus, our review does not capture harms associated with the assessed treatments. Specifically, future studies should randomize patients diagnosed with malnutrition (i.e., using a diagnostic assessment tool) to different interventions, such as parenteral and enteral nutrition, conduct subgroup analyses to assess the benefits of nutritional intervention in subpopulations, and document harms associated with treatment.
Uhl S, Siddique SM, McKeever L, Bloschichak A, D’Anci K, Leas B, Mull NK, Tsou AY. Malnutrition in Hospitalized Adults: A Systematic Review. Comparative Effectiveness Review No. 249. (Prepared by the ECRI–Penn Medicine Evidence-based Practice Center under Contract No. 75Q80120D00002.) AHRQ Publication No. 21(22)-EHC035. Rockville, MD: Agency for Healthcare Research and Quality; October 2021. DOI: 10.23970/AHRQEPCCER249. Posted final reports are located on the Effective Health Care Program search page.