Outcomes and other characteristics of the nutritional screening and assessment tools validation processes in hospitalized adults: a systematic review Authors: Giovanna Guimarães Lopes Bergamasco, Bruna Rodrigues Lima, Kadine Andrade de Araújo, Nicolly Baião Martins, Silvia Maria Fraga Piovacari, Mayumi Shima, Drielle Schweiger Freitas Bottairi, Samara Cristina Mascarenhas Ferreira, Zelita Melo Fortes de Cerqueira, Adriano José Pereira Published: 2026 ABSTRACT ABSTRACTBackground:Hospital malnutrition has been studied for decades; however, its prevalence remains high, and research in this area is still relevant. Nutritional screening and assessment tools are routinely used in hospital settings.Objective:We aimed to describe and discuss the general characteristics of studies that used nutritional screening and assessment instruments in hospitalized adult populations, with a focus on clinical outcomes.Methods:We conducted a systematic review without meta-analysis. Eligible studies were original prospective or retrospective studies published in Portuguese or English, with no inception date, conducted in hospitalized adult populations, and reporting clinical outcomes. Information sources included PubMed, LILACS, Web of Science, Embase, and Scopus. The search covered articles published up to July 2022. Risk of bias was assessed for all included studies using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. Results presented in the tables were transcribed from the main findings of each evaluated study, and no new statistical analyses were performed.Results:Seventy-seven studies were included, encompassing 20 tools evaluated in hospital settings. The Mini Nutritional Assessment was the most extensively tested in relation to clinical outcomes. Among studies conducted in Brazilian populations, the nutritional assessment tool Subjective Global Assessment and the screening tool Nutritional Risk Screening 2002 were the most frequently studied. Among full-text articles assessed for tool evaluation in hospitals, 77% were excluded because clinical outcomes were not reported.Discussion:The evidence has some limitations. First, the exclusive focus on hospitalized populations may limit generalizability. Second, validation studies that did not evaluate clinical outcomes were not described in detail. Third, the use of "clinical outcomes" as a search term may have led to an underestimation of validation studies focusing solely on diagnostic or screening performance. Overall, nutritional screening and assessment tools commonly used in daily practice have been validated many years ago, largely based on subjective clinical assessments, with relatively few studies reporting clinical outcomes. In addition, most studies were single-center and conducted predominantly in non–Latin American populations. Future research should prioritize multicenter designs, improve population representativeness, and incorporate clinically relevant outcomes.Prospero database registration:ID CRD42022347507. INTRODUCTION Hospital malnutrition has been studied for decades; however, its prevalence remains high, highlighting the continued importance of research in this field.(1,2) Hospital malnutrition is associated with poor prognosis, a high risk of complications, functional impairment, increased morbidity and mortality, prolonged hospital stays, and high readmission rates.(3) In addition, malnutrition imposes a substantial economic burden on healthcare systems.(4,5) Nutritional screening is an essential first step in identifying patients at nutritional risk and should be performed at hospital admission by nutritionists or other healthcare professionals.(6-9) Screening should be quick and straightforward and can be completed by any trained healthcare professional within 24–48 hours of admission.(6-9) Patients identified as being at nutritional risk should then undergo a comprehensive nutritional assessment; those not at risk may be rescreened within 7 days. Despite their clinical importance, many nutritional screening tools currently in use were validated approximately two decades ago.(10-12) Nutritional assessment is a systematic and comprehensive process that integrates nutritional and clinical information and provides the basis for planning nutrition interventions.(13,14) In hospital practice, validated tools are commonly used for assessment, although they are often regarded as semi–gold standards.(13,14) Key limitations include the lack of a single standard tool and heterogeneity in assessed parameters, validation methods, and reported outcomes, which may contribute to misclassification of malnutrition.(14,15)Notably, some widely used nutritional tools were originally validated in retrospective studies and/or relied primarily on subjective parameters or laboratory tests.(12,16) OBJECTIVE We aim to describe and discuss the general characteristics of studies that validated nutritional screening and assessment instruments in hospitalized adult populations, with a focus on clinical outcomes. METHODS This systematic review was conducted without meta-analysis and was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.(17) Eligibility criteria We included original prospective or retrospective studies published in Portuguese or English, with no restriction on inception date, conducted in hospitalized adult human populations, and reporting clinical outcomes. We excluded reviews, editorials, and case reports. Information sources We searched PubMed, LILACS, Web of Science, Embase, and Scopus for articles published up to July 2022. Search strategy The main search strategy used in this review (initiated with a PubMed/MEDLINE search) is described below: (((("malnutrition"[MeSH Terms]) AND ((("mass screening"[MeSH Terms]) OR ("nutrition assessment"[MeSH Terms])))) AND (("physical functional performance"[MeSH Terms]) OR ("treatment outcome"[MeSH Terms]))) OR ((((((((("mass screening"[Title/Abstract]) OR ("nutrition assessment"[Title/Abstract])) OR ("nutritional risk screening"[Title/Abstract])) OR ("mini nutritional assessment short form"[Title/Abstract])) OR ("sga"[Title/Abstract])) OR ("subjective global assessment"[Title/Abstract])) OR ("mini nutritional assessment"[Title/Abstract])) AND ((("malnutrition"[Title/Abstract]) OR ("malnourish"[Title/Abstract])) OR ("malnourished"[Title/Abstract]))) AND (("physical functional performance"[Title/Abstract]) OR ("functionality"[Title/Abstract])))) OR ((((((((("mass screening"[Text Word]) OR ("nutrition assessment"[Text Word])) OR ("nutritional risk screening"[Text Word])) OR ("mini nutritional assessment short form"[Text Word])) OR ("sga"[Text Word])) OR ("subjective global assessment"[Text Word])) OR ("mini nutritional assessment"[Text Word])) AND ((("malnutrition"[Text Word]) OR ("malnourish"[Text Word])) OR ("malnourished"[Text Word]))) AND (("physical functional performance"[Text Word]) OR ("functionality"[Text Word]))). Selection process Three independent researchers systematically identified studies, with support from a librarian from the Instituto de Ensino e Pesquisa Albert Einstein. Study selection was managed using Rayyan (Cambridge, MA, USA; 2023).(18) Titles and abstracts were screened against the inclusion criteria. Disagreements were resolved through reassessment and discussion in a formal consensus meeting. Data collection process and data items For studies included in the full-text analysis, we extracted the following information: country, study design, population, place of hospitalization, diagnosis, tool(s) evaluated, comparators, number of centers, inclusion criteria, sample size, outcomes/metrics, and main results. Data were transcribed directly from the included articles to minimize interpretation bias. Study design (prospective vs retrospective) was not always explicitly reported; in such cases, we classified design based on the described methodology for the purposes of tabulation. When variables were not clearly reported, they were recorded as "not specified." Study risk bias, reporting bias, and certainty assessment Risk of bias was assessed for all included studies using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. A calibration meeting was held before assessment to standardize interpretation of each domain. Five authors conducted the assessments independently, and two authors subsequently reviewed all ratings in full, including one reviewer who had not participated in the initial assessments. Discrepancies were resolved by consensus. All authors reviewed the final report after manuscript completion. Effect measures Because the data were not suitable for meta-analysis, only descriptive analyses were performed. Synthesis methods We verified eligibility by confirming study design, publication language, and population characteristics (human adults hospitalized in a hospital setting) based on the Methods sections of the included articles. The results presented in the tables were transcribed from the main findings of each evaluated study, and no new statistical analyses were performed. References were managed using Mendeley (Mendeley Ltd., United Kingdom; version 2.80.1; 2022). RESULTS Study selection A total of 2,140 articles were found by the search strategy. After peer analysis and evaluation, 77 studies were included in the final analysis. Figure 1 provides the flowchart of article selection and the reasons for exclusions. PRISMA flow diagram of study selection Study characteristics Among the articles eligible for the evaluation of nutritional tools, 77% were excluded from the analysis because clinical outcomes were not considered in any of the analyses. The results below detail the 77 studies included because they used clinical outcomes in testing nutritional tools. The included studies involved populations from 31 countries. Twenty-one studies (28%) had a prospective observational design, and 19 (25%) were prospective cohort studies. Overall, 77% were conducted in single-center settings, while 18% were multicenter studies. In 30 studies, only populations aged >60 years were evaluated. Most studies included clinical and surgical patients (90%). Regarding outcomes, 51 studies (66%) assessed mortality, and 34 assessed length of hospital stay (44%). Functional outcomes, such as frailty and functional capacity, were evaluated in only six studies (8%). Overall, 20 tools were tested in the hospital setting: Mini Nutritional Assessment (MNA), Mini Nutritional Assessment Short Form (MNA-SF), Nutritional Risk Screening 2002 (NRS 2002), Malnutrition Universal Screening Tool (MUST), Malnutrition Screening Tool (MST), Modified Nutrition Risk in Critically Ill (m-NUTRIC), Nutrition Risk in Critically Ill (NUTRIC), Subjective Global Assessment (SGA), Patient-Generated Subjective Global Assessment (PG-SGA), Geriatric Nutritional Risk Index (GNRI), Malnutrition-related Complications Score (MCRS), Automated Nutrition Score (ANS), Prognostic Nutritional Index (PNI), Controlling Nutritional Status (CONUT), Onodera's Prognostic Nutritional Index (O-PNI), Short Nutritional Assessment Questionnaire (SNAQ), Royal Free Hospital-Nutritional Prioritizing Tool (HFR-TNP), 3-Minute Nutrition Screening (3-MinNS), European Society for Clinical Nutrition and Metabolism Diagnostic Criteria for Malnutrition (ESPEN-DCM), and Academy of Nutrition and Dietetics-European Society for Clinical Nutrition and Metabolism Criteria (AND-ESPEN). Studies evaluating the Global Leadership Initiative on Malnutrition (GLIM) criteria in combination with tools were also included. Risk of bias in studies: Risk of bias is summarized in figure 2 and presented by article in figure 3. Overall, approximately 30% of studies were rated as low risk of bias, 45% as some concerns, and 25% as high risk. Summary of risk of bias for each domain and overall risk in the included studies assessed using QUADAS-2 Risk of bias summary for each included study assessed using QUADAS-2 Results of individual studies: The comprehensive analysis of the retrieved articles is presented in table 1. Table 2 provides a summary of the tools discussed, along with their respective outcomes. Table 1Characteristics and main results of the included studiesArticleCountryStudy designPopulationPopulation: place of hospitalizationPopulation; DiagnosisToolComparatorCenterInclusion criteriaSample sizeOutcomeMain resultsGioulbasanis et al.(20) 2011GreeceRetrospective, observational studyMetastatic lung cancer prior to commencement of systemic therapyClinicalOncologicalMNA≥5% weight loss (independent variable)Single-center>18 years171Prognostic prediction: time to disease progression and overall survivalThe MNA better predicts all evaluated outcomes. Malnourished individuals identified by MNA exhibit a reduced average survival of 2.07 monthsMajari et al.(21) 2021IranProspective cohort studyCritically ill patientsICUNeurological; gastrointestinal/ hepatic; cardiac; malignant diseasesMUSTM-NUTRIC and NRS 2002Multicenter>18 years440Length of hospital stay; duration of mechanical ventilation; 28-day mortalityNo association between M-NUTRIC or NRS scores and evaluated outcomesHelminen et al.(22) 2017FinlandProspective cohort studyElderly individuals with hip fractureSurgicalHip fractureMNA-SFMNA completeSingle-center≥65 years594Mobility at 1 and 4 months; quality of life at 1 year; mortality at 1 and 4 months and at 1 year after fractureBoth tools predicted mortality at all timepoints and predicted mobility and quality of life at 4 monthsGoldfarb et al.(23) 2018Canada, USA, and FranceProspective cohort studyElderly individuals undergoing cardiovascular surgerySurgicalCardiovascular diseaseMNA-SFN/AMulticenter≥70 years1,1581-year mortality; 30-day mortality and morbidity; frailtyPatients classified as malnourished by MNA had a threefold higher risk of 1-year mortality. A moderate association was observed between MNA score and frailty (SPPB)Chu et al.(24) 2017ChinaProspective observational studyElderly individuals undergoing elective orthopedic surgerySurgicalOrthopedicMNA-SFN/ASingle-center≥60 years312Mortality at 12 Months Hospital Readmission, Emergency Department Visits, and FunctionalityLower MNA-SF scores associated with increased emergency department visits at 6 months and higher mortality at 12 months. No association was observed with hospital readmission or functional decline. Each 1-point increase in MNA score was associated with a 27% decrease in mortality riskHersberger et al.(25) 2020SwitzerlandRetrospective observational studyMedical inpatient unitsClinicalMixed hospital populationNRS 2002n/aMulticenterAdults with NRS ≥3 points and expected LOS >4 days2,028Mortality at 30 to 180 days ICU admission, hospital readmission and complications at 30 to 180 daysNRS is an independent predictor of mortality and incidence of adverse events at 180 days. NRS 2002 is associated with an ↑ average length of hospital staysHolst et al.(26) 2013Denmark and SwedenProspective cohort studyHospitalized elderly individualsClinical/SurgicalGastrointestinal and neurological diseases; orthopedicMNA, MUST, and NRS 2002Anthropometry and Cognitive Test (SPMSQ)Multicenter>65 years23312-month mortality; anthropometric assessmentNo significant association between the three tools and mortalityDent et al.(27) 2014AustraliaCross-sectional studyHospitalized elderly individuals in the geriatric wardClinicalUninformedGNRI, MNA e MNA-SFIncorporation of calf circumference or BMI into the toolsSingle-centerElderly1726-month mortality; readmission/transfer to higher level of careFull MNA and GNRI predicted adverse outcomes at 6 months better than the MNA-SF versionsGattermann al.(28) 2019BrazilProspective observational studyCritically ill surgical patientsICUGastrointestinal and cardiovascular surgeriesSGAClinical assessmentSingle-center≥18 years76In-hospital mortality; Hospital LOS; ICU LOSNo significant correlation was observed with mortality. Patients classified as malnourished by SGA had a 2.5-fold higher risk of longer hospitalizationMahmoudinezhad et al.(29) 2021IranProspective cross-sectional studyStroke and LOS >48 h.ClinicalNeurologicSGAN/AMulticenter≥65 years349Functional capacityA significant association was observed between ≥3 levels of disability and SGAda Silva et al.(30) 2018BrazilProspective observational studyPatients admitted to the emergency departmentClinicalMixed hospital populationESPEN-DCMSGA, MUST, NRS 2002, SNAQ, and MSTSingle-center≥18 years750LOS; incidence of hospital-acquired infection; in-hospital mortalityNRS 2002 showed a better correlation with mortality. ESPEN-DCM had low diagnostic value but performed better when combined with NRS for prognostic predictionHuang et al.(31) 2021ChinaProspective cohort studyCancer patients undergoing radical gastrectomySurgicalOncologicalGLIM (screened with NRS 2002)Isolated criteria vs combination (1 phenotypic + 2 etiological factors)Single-center≥65 years597Overall survival; disease-free survivalThe combined GLIM criteria outperformed isolated factors and was significantly associated with overall and disease-free survivalColtman et al.(32) 2015USAProspective observational studyPatients admitted to medical, surgical, and neurology intensive care unitsICUClinical and surgical diagnosesRegional Institutional Nutritional Screening ToolSGA and NUTRICSingle-center≥18 years294LOS ICU mortalityNUTRIC, alone or in combination, improved the identification of mortality risk. The tools did not predict risk in patients with a long LOS, suggesting limited suitability for this populationHonda et al.(33) 2016JapanProspective cohort studyPatients with acute heart failureClinicalCardiovascular diseaseGNRIN/ASingle-center≥65 years490Mortality (cardiovascular and all-cause)GNRI was an independent predictor of adverse events in patients with acute heart failureSlee et al.(34) 2016UKRetrospective cohort studyFrail elderly individuals receiving nutrition orally or via a feeding tube.ClinicalNot specifiedMUSTMNA-SFSingle-center>65 years78Mortality; LOSMNA-SF was a significant predictor of mortality, whereas MUST showed a weak associationSmith et al.(35) 2009AustraliaProspective cohort studySurgical inpatientsSurgicalGastrointestinal diseasesSGAMRCS and ANSSingle-centerNot specified148Risk of postoperative complicationsMRCS had a higher predictive value for postoperative complications than SGASorensen et al.(36) 2008DenmarkProspective cohort studyHospitalized adultsClinical/SurgicalGastrointestinal diseases, oncological; critical careNRS 2002N/AMulticenter≥18 years5,051Hospital complications; mortality; LOSNRS 2002 was an independent predictor of all evaluated outcomesSungurtekin et al.(37) 2008TurkeyProspective observational studyCritically ill adultsICUSurgicalSGAN/ASingle-center≥18 years124ICU mortality; ICU LOSSGA was able to predict the evaluated outcomesTakahashi et al.(38) 2021JapanRetrospective cohort studyPatients with completely resected non-small cell lung cancerSurgicalOncologicalGNRIPNI and CONUTSingle-centerNot specified475Overall Survival Postoperative ComplicationPNI, CONUT, and GNRI showed high predictive value for the evaluated outcomesTakaoka et al.(39) 2017JapanRetrospective observational studyPatients with crohn's disease admitted to gastroenterologyClinicalGastrointestinal DiseasesSGANRS 2002, O-PNI, MUST and CONUTSingle-centerNot specified40LOS >28 daysAll tested tools, except for MUST, were associated with length of stay. SGA was able to predict the studied outcomesTangvik et al.(40) 2014NorwayProspective observational studyInpatientsClinical/Surgical/ICUUninformedNRS 2002Adapted NRS 2002 (using only the first four questions)Single-center≥18 years3,279Mortality; morbidity; readmissionThe adapted NRS proved to be a strong predictor of the evaluated outcomesvan Venrooij et al.(41) 2011NetherlandsCohort studyPatients undergoing coronary artery graft surgery admitted to the cardiothoracic surgical wardSurgicalCardiovascular diseaseMUSTMUST modified for cardiovascular cirurgical and SNAQSingle-center≥18 years325Adverse postoperative outcomesModified MUST for cardiac surgery was associated with longer ICU and hospital staysWatanabe et al.(42) 2021JapanRetrospective observational studyElderly people with renal cell cancerSurgicalOncologicalGNRIN/ASingle-center≥65 years62Postoperative complicationsPatients with postoperative complications had significantly lower GNRI valuesWu et al.(43) 2010ChinaProspective observational studyPatients with newly diagnosed gastrointestinal cancer (without metastasis) undergoing surgerySurgicalOncologicalSGAN/ASingle-center≥18 years505LOS complications and hospital cost predictionSGA classification was significantly associated with LOS and costs but not with complicationsWu et al.(44) 2020ChinaProspective observational studyPatients with cirrhosis without liver cancer or uncontrolled comorbid diseaseClinicalLiver diseasesHFR-TNPNRS, MUST, and HFR-TNPSingle-centerAdults155MortalityMUST showed lower sensitivity. HFR-TNP predicted outcomes better than NRS 2002Xiao et al.(45) 2022ChinaRetrospective observational studyPatients undergoing radical gastrectomySurgicalOncologicalCONUTN/ASingle-center≥18 years106Postoperative complications; quality of life; overall survivalCONUT was an independent risk factor for postoperative complications and quality of life. Higher CONUT scores were associated with shorter survivalXu et al.(46) 2022ChinaRetrospective cohort studyPatients with primary gastric adenocarcinoma before surgerySurgicalOncologicalGLIMPhenotypic criteriaSingle-centerNot specified1,188Survival time; postoperative complicationWeight loss showed the strongest correlation with survival and complications. Greater weight loss and low muscle mass (tomography) were associated with worse outcomesXu et al.(47) 2022ChinaRetrospective analytics studyPatients with gastric cancer undergoing radical gastrectomySurgicalOncologicalGLIM with MUST screening score ≥1SGASingle-centerNot specified895Postoperative complications at 1 month; In-hospital mortality; LOS readmission; mortalityGLIM criteria showed a moderate association with SGA. Patients classified as malnourished by GLIM were predictors of survival and complicationsYu et al.(48) 2021ChinaRetrospective observational studyPatients with cardiovascular disease(acute coronary syndrome)SurgicalCardiovascular diseaseNRS-2002N/AMulticenterNot specified3,185MortalityNRS ≥3 points associated with mortality and the incidence of acute kidney disease in patients with ACSZhang et al.(49) 2021ChinaProspective observational studyPatients with cardiovascular disease and primary diagnosis of strokeClinicalCardiovascular and neurological diseasesESPEN-DCMCONUT GNRI, MUST and NRS-2002Multicenter≥18 years593Mortality at 3 and 12 months after discharge; functional capacity (mRS)NRS 2002 had better predictive value for mortality at 3 and 12 months. NRS-2002 and MUST were significantly associated with mRS at 12 months after dischargeChávez-Tostado et al.(50) 2020MexicoProspective cross-sectional studyPatients hospitalized with gastrointestinal disease(excluding ICU)ClinicalGastrointestinal diseaseNRS-2002, SGA e CONUTN/ASingle-center18-90 years196LOS; complications; mortalityCONUT was the only tool to predict complications. None of the tools predicted mortalityCharlton et al.(51) 2012AustraliaRetrospective observational studyhospitalizedClinicalUninformedMNAN/AMulticenter>65 years2,076Readmission; LOS; mortalityNutritional status assessed by MNA at admission was correlated with LOSBurgel et al.(52) 2021BrazilProspective observational studyhospitalized, expect icuClinical/SurgicalUninformedAND-ASPENSGAMulticenter≥18 years600Mortality; readmission; LOSAND-ASPEN predicted an approximately 1.4-fold greater risk of prolonged hospitalization and hospital readmission. In addition to identifying a 5.0 times. higher risk of death in hospital and within 6 months, in an adjusted analysisBell et al.(53) 2014AustraliaAccuracy studyhip fractureSurgicalHip fractureMNA-SFCID-10AM (malnutrition)Single-centerElderly142Postoperative mobility; mortality; home dischargeMNA-SF predicts probability of home discharge and mortality within 4 months, but not reduction in postoperative mobilityAtalay et al.(54) 2008TurkeyRetrospective observational studycritically ill elderly hospitalized with enteral and parenteral nutritional therapyClinicalNeurological, cardiovascular and oncological diseasesSGAN/ASingle-center≥65 years119Mortality; LOSSGA classification did not influence the mortality rateAlmendra et al.(55) 2022BrazilRetrospective observational studyelderlyClinicalGastrointestinal, renal and oncological diseasesMNANRS-2002 and MNASingle-center≥65 years277LOSMNA was the tool associated with length of hospital stayAlmasaudi et al.(56) 2019UKRetrospective observational studypatients undergoing surgery for colorectal cancerSurgicalOncologicalMUSTN/ASingle-centerNot defined363Postoperative complications; LOS; mortalityMUST demonstrated to be an independent marker for length of hospital stay, but not for complications Medium and high risk by MUST demonstrated higher mortalityAllepaerts et al.(57) 2020BelgiumProspective cross-sectional studyPatients admitted to geriatricsClinicalNeurological, oncological diseasesGLIM with MNA-SF screeninga phenotypic and an etiological criterionSingle-centerEldery791-year institutionalization; mortalityGLIM (one phenotypic and one etiological criterion) was associated with 1-year mortalityAlisgahari et al.(58) 2019IranCross-sectional studyPatients with ischemic strokeClinicalNeurological diseasesMNAN/AMulticenter≥65 years253Biochemical and anthropometric parameters; LOS; mRS PerformanceMNA was significantly associated with poor outcomes in this subgroupAkimoto et al.(59) 2021JapanRetrospective observational studyPatients with ischemic strokeClinicalNeurological diseasesCONUTGNRISingle-center≥65 years218Anthropometric parameters; LOS; mRS PerformanceCONUT was associated with poorer prognosis, higher mRS scores, and more complicationsAcarbaş(60) 2021TurkeyRetrospective observational studyPatients undergoing spinal surgerySurgicalSpinal surgeryPNI, CONUT and GNRIN/ASingle-center≥65 years454Prediction of perioperative adverse events.Higher CONUT scores (pre-surgical) were associated with adverse events. GNRI showed the greatest accuracy for this outcomeAbd-el-gawad et al.(61) 2014EgyptProspective cohort studyPatients admitted to geriatricsClinicalNot specifiedMNAGNRISingle-center≥60 years131LOS; Infection-related complications; mortalityGNRI was an independent predictor of mortality at 3 and 6 monthsAbbass et al.(62) 2020ScotlandRetrospective observational studyPatients with lung cancer prior to radiotherapyClinicalOncologicMUSTModified frailty index (mFI)Single-centerNot specified643Overall survivalMUST was independently associated with overall survival and was a prognostic indicator of 12-month survivalAaldriks et al.(63) 2013NetherlandProspective observational studyPatietns with colorectal cancer prior to chemotherapyClinicalOncologicMNACognitive decline, frailty and mental status questionnairesMulticenter≥70 years143MortalityMNA was associated with mortality and predicted poor tolerance of chemotherapy cyclesLomivorotov et al.(64) 2013RussiaProspective observational studyPatients undergoing extracorporeal circulationSurgicalCardiovascular diseaseSGA, NRS-2002, MUST, MNA-SF and SNAQN/ASingle-centerAdults1,193Mortality; postoperative complications; LOS; ICU LOSOnly MUST and MNA-SF were independent predictors of postoperative complications. SNAQ, MUST, and MNA predicted prolonged hospitalizationLin et al.(65) 2021ChinaRetrospective observational studyPatients with tuberculosisClinicalTuberculosisPG-SGAN/ASingle-center≥65 years128Mortality Liver injuryHigh PG-SGA scores were independently associated with mortality and liver injuryHuang et al.(66) 2021ChinaProspective observational studyOverweight patients undergoing radical gastrectomy for gastric cancerSurgicalOncologicGLIMAddition of muscle strength and gait speed variablesSingle-centerNot specified587Postoperative complications MortalityAdding muscle quality (tomography), strength (handgrip), and gait speed (6 m) to GLIM improved prediction of outcomes, with gait speed showing the strongest contributionHung et al.(67) 2021TaiwanProspective cohort studyPatients with head and neck cancer undergoing concomitant chemoradiotherapyClinicalOncologicMNA-SFN/AMulticenter≥20 years461Overall survivalMNA-SF was correlated with toxicity and adverse events and was associated with a lower hospital readmission rateInoue et al.(68) 2020JapanRetrospective cohort studyPatients undergoing elective endovascular thoracic aneurysm repairSurgicalCardiovascular diseaseCONUTN/ASingle-centerNot specified60Overall survival; non-aneurysm-related mortality; prevalence of reinterventionsMalnutrition identified by CONUT was a negative predictor of survival prognosisInoue et al.(69) 2019JapanRetrospective observational studyPatients with hip fracture after fall undergoing surgical treatmentSurgicalHip fractureMNA-SF, MUST, NRS 2002 and GNRIN/ASingle-center≥65 years205LOS; functional outcomesMNA-SF showed a stronger association with functional outcomes. GNRI was significantly associated with walking speed (10 m). MUST and NRS 2002 were not associated with functional outcomesJayant et al.(70) 2020IndiaProspective observational studyPatients with an oncological diagnosis undergoing elective surgerySurgicalOncologicMUST, SGA and NRIN/ASingle-center≥18 years342LOS; postoperative complications within 30 daysSGA and MUST demonstrated good reliability for the evaluated outcomesJeejeebhoy et al.(71) 2015CanadaCohort study prospectiveInpatients from medical and surgical clinicsClinical/SurgicalCardiovascular, neurological, gastrointestinal, respiratory diseasesNRS 2002SGAMulticenter≥18 years733LOS; hospital readmissionSGA classification, especially category C, predicted length of hospital stayKalaiselvan et al.(72) 2017IndiaProspective observational studyICU patients receiving mechanical ventilation for >48 hICURespiratory failure, septic shock, neurological disorders,m-NUTRICN/ASingle-centerAdults678ICU LOS; time off mechanical ventilation; mortalityHigher m-NUTRIC scores were associated with longer ICU stay and predicted mortalityKang et al.(73) 2020South KoreaProspective observational studyPatients with acute ischemic stroke admitted within 7 days of symptom onsetClinicalNeurological diseaseGNRIN/ASingle-centerNot specified1,9063-month prognosis after hospitalization (mRS)Moderate and severe GNRI risk categories were associated with unfavorable outcomesKatayama et al.(74) 2020JapanRetrospective observational studyPatients undergoing percutaneous coronary intervention with rotational atherectomySurgicalCardiovascular diseaseGNRIN/ASingle-centerNot specified206Major adverse cardiovascular events, including mortalityGNRI was an independent predictor of major adverse cardiovascular eventsKomici et al.(75) 2019ItalyProspective observational studyPatients with acute myocardial infarctionICUCardiovascular diseaseMNAN/ASingle-center≥65 years174MortalityMNA results were an independent predictor of long-term mortalityKootaka et al.(76) 2021JapanRetrospective cohort studyPatients with cardiovascular diseaseClinicalCardiovascular diseaseGLIMESPEN-DCMSingle-center≥20 years921Mortality; physical performanceGLIM and ESPEN-DCM were significantly associated with mortality. However, only GLIM predicted low functionalityKoren-Hakim et al.(77) 2016IsraelProspective cohort studyElderly patients with hip fracture undergoing surgerySurgicalHip fractureMNA-SF; NRS 2002; MUSTN/ASingle-center≥65 years215LOS; postoperative complications; readmission within 6 months; mortality at 36 monthsOnly MNA-SF predicted readmissions and mortalityLee et al.(78) 2021TaiwanRetrospective observational studyPatients with blood cultures positive for non-albicans candida speciesClinicalCandidiasisMUSTN/AMulticenterAdults37828-day mortalityMUST ≥2 was independently associated with a higher risk of 28-day all-cause mortalityLim et al (79) 2014SingaporeProspective cohort studyHospitalized patients (excluding psychiatric, ICU, and maternity)ClinicalNot specified3-MinNSN/ASingle-centerAdults818LOS; hospital readmission; mortalityHigher 3-MinNS scores were associated with longer hospital stay and higher mortality at 1 and 3 yearsNascè et al.(80) 2021SwitzerlandProspective cohort studyPatients with suspected pneumoniaClinicalSuspected pneumoniaMNAGeriatric and functionality assessmentSingle-center≥65 years200MortalityMNA significantly predicted long-term mortality (1 year)Nishioka et al.(81) 2020JapanCohort study prospectiveElderly people with stroke transferred from intensive care hospitalsClinicalNeurological diseaseESPEN-DCMMNA-SF, MUST and GNRISingle-center≥65 years420Functionality; Discharge destinationGNRI showed predictive validity for discharge destination. MNA-SF showed fair concurrent validity but required new cutoff pointsNuotio et al.(82) 2016FinlandProspective observational studyPatients in the perioperative period for hip fractureSurgicalHip fractureMNA-SFN/ASingle-center≥65 years472Institutionalization; mortality and morbidityMNA-SF was an independent predictor of the main outcomes studiedOliveira et al.(83) 2019BrazilLongitudinal studyCritical illy ill patients hospitalized fo at least 48 hoursICUPostoperative complications, septic shock, infection, heart problems, and gastrointestinal complicationsm-NUTRICNUTRIC-PCR and SGASingle-center≥18 years130ICU LOS; 28-day mortalityHigher m-NUTRIC scores were associated with mortality. NUTRIC showed a positive association with SGA, With good agreement between m-NUTRIC and NUTRI-PCROzkalkanli et al.(84) 2009TurkeyProspective studyPatients undergoing elective orthopedic surgeries hospitalized >2 daysSurgicalOrthopedicSGANRS 2002Single-center≥18 years256Postoperative complications; mortality; LOSNRS 2002 better predicted postoperative complications than SGA. No significant differences were observed between tools for the other outcomesBarbosa et al.(85) 2010BrazilProspective cohort studyadults hospitalized in the wardsClinicalNot specifiedNRS 2002, MNA-SF, MUST and SGAN/ASingle-center≥18 years705Complications; LOS; mortalityNRS and SGA better predict outcomes. SGA was more strongly associated with hospital LOSRabito et al.(86) 2017BrazilProspective cohort studyhospitalizedClinicalGastrointestinal, cardiovascular, and oncological diseasesNRS 2002MUST, MST and SNAQSingle-center≥18 years752Morbidity MortalityMUST showed greater identification of mortality risk (2.34-fold). All tools were sensitive for identifying prolonged hospitalizationRasheedy et al.(87) 2020EgyptCross-sectional studyhospitalizedClinicalNot specifiedMNAGNRISingle-center≥60 years150Fragility; sarcopeniaGNRI showed greater accuracy for frailty than MNA. GNRI risk categories were significantly associated with lower muscle strengthRaslan et al.(88) 2010BrazilProspective observational studyhospitalizedClinical/SurgicalMetabolic, infectious, inflammatory, immunological diseases, and oncological diseasesNRS-2002, MNA-SF and MUSTN/ASingle-center≥18 years706Complications; LOS; mortalityNRS 2002 and MNA-SF predicted outcomes better overall; however, NRS 2002 showed the strongest performanceRaslan et al.(89) 2011BrazilProspective observational studyhospitalizedClinical/SurgicalNot specifiedSGA e NRS 2002N/ASingle-center≥18 years705Complications; LOS; mortalityCombining SGA and NRS 2002 improved reliability, and both were positively associated with the evaluated outcomesRodrigues et al.(90) 2015BrazilRetrospective cohort studyPatients with gynecological tumorsClinicalOncologicPG-SGAN/ASingle-center≥18 years146Readmission; mortalityPG-SGA classification was consistent with the evaluated outcomesRuiz et al.(91) 2018ColombiaProspective cohort studyPatients with heart and lung diseasesClinicalHeart and lung diseasesMSTN/AMulticenter≥18 years800LOS Mortality Readmission Hospital costsA positive MST (≥2 points) was associated with the evaluated adverse outcomesSantos et al.(92) 2017BrazilCross-sectional studyOncology patients hospitalized for at least 3 daysClinicalOncologicPG-SGAN/ASingle-center≥20 years333LOS MortalityPG-SGA was an important marker of prolonged hospitalization and higher mortalitySanz-París et al.(93) 2016SpainProspective observational studyPatients with diabetesClinicalDiabetesESPEN-DCMMNA-SFMulticenter≥65 years1,014LOS MortalityAn MNA score of 7 was associated with a 2.7-fold increase in in-hospital deathsSöderström et al.(94) 2014SwedenProspective cohort studyOlder adultsClinical/SurgicalDiabetes; neurological disorders; lung diseases; rheumatoid arthritis; kidney failureMNAN/ASingle-center≥65 years1,767MortalityMNA predicted premature mortalitySaseedharan(95) 2019IndiaProspective cohort studyCritically ill patientsICUClinical and surgicalNUTRICSGA e NRS 2002Single-centerN/A348Mortality LOS in ICU LOSNRS 2002 and SGA correlated significantly with ICU LOS; NRS 2002 also correlated with overall hospital LOS. NUTRIC score was the best predictor of mortalityZhao et al.(96) 2020ChinaObservational retrospective cohortPatients with cardiovascular disease undergoing percutaneous coronary interventionSurgicalNon-ST-segment elevation acute coronary syndromeGNRIN/ASingle-centerN/A2,299Mortality; Non-fatal myocardial infarction; revascularizationLower GNRI was a significant predictor of adverse prognosis Characteristics and main results of the included studies Table 2Tools evaluated in the included studies and the evidence of positive outcomes associated with their useToolsMortalityICU mortalityOverall survivalLOSICU LOSHospital readmissionBad prognosisBad outcomesPostoperative complicationMobilityDisabilityFunctional outcomeTreatment toxicityMNA-SFxxxxxxxxMNA fullxxNRS 2002xxxxMUSTxxxxNUTRICxm-NUTRICxxGNRIxxxxSGAxxxxxPG-SGAxxxNUTRICxMSTxMCRSxCONUTxxxSNAQxHFR-TNPx3-MinNSxxESPEN-DCMxANSPNIxxO-PNIxAND-ESPENxxICU: Intensive Care Unit; LOS: Length of Stay; MNA-SF: Mini Nutritional Assessment Short Form; MNA full: Mini Nutritional Assessment; NRS 2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening; m-NUTRIC: Modified Nutrition Risk in Critically Ill; GNRI: Geriatric Nutritional Risk Index; SGA: Subjective Global Assessment; PG-SGA: Patient-Generated Subjective Global Assessment; NUTRIC: Nutrition Risk in Critically Ill; m-NUTRIC: Modified Nutrition Risk in Critically Ill MST- Malnutrition Screening Tool; MCRS: Malnutrition-related Complications; CONUT: Controlling Nutritional Status; SNAQ: Short Nutritional Assessment Questionnaire; HRF-TNP: Royal Free Hospital-Nutritional Prioritizing Tool; 3-MinNS: 3-Minute Nutrition Screening; ESPEN-DCM: European Society for Clinical Nutrition and Metabolism Diagnostic Criteria for Malnutrition; ANS: Automated Nutrition Score; PNI: Prognostic Nutritional Index; O-PNI: Onodera's Prognostic Nutritional Index; AND-ESPEN: Academy of Nutrition and Dietetics- European Society for Clinical Nutrition and Metabolism. Tools evaluated in the included studies and the evidence of positive outcomes associated with their use ICU: Intensive Care Unit; LOS: Length of Stay; MNA-SF: Mini Nutritional Assessment Short Form; MNA full: Mini Nutritional Assessment; NRS 2002: Nutritional Risk Screening 2002; MUST: Malnutrition Universal Screening; m-NUTRIC: Modified Nutrition Risk in Critically Ill; GNRI: Geriatric Nutritional Risk Index; SGA: Subjective Global Assessment; PG-SGA: Patient-Generated Subjective Global Assessment; NUTRIC: Nutrition Risk in Critically Ill; m-NUTRIC: Modified Nutrition Risk in Critically Ill MST- Malnutrition Screening Tool; MCRS: Malnutrition-related Complications; CONUT: Controlling Nutritional Status; SNAQ: Short Nutritional Assessment Questionnaire; HRF-TNP: Royal Free Hospital-Nutritional Prioritizing Tool; 3-MinNS: 3-Minute Nutrition Screening; ESPEN-DCM: European Society for Clinical Nutrition and Metabolism Diagnostic Criteria for Malnutrition; ANS: Automated Nutrition Score; PNI: Prognostic Nutritional Index; O-PNI: Onodera's Prognostic Nutritional Index; AND-ESPEN: Academy of Nutrition and Dietetics- European Society for Clinical Nutrition and Metabolism. Results of syntheses Results are presented below, highlighting clinical outcomes and the tested instruments. The MNA and MNA-SF nutritional tools were studied in 21% of trials related to clinical outcomes,(20,24,27,53,67,75,80,82,94) length of hospital stays,(51,55) readmission,(77) postoperative complications,(64) functional outcomes,(69) and chemotherapy toxicity.(63) Notably, the MNA-SF screening tool demonstrated strong predictive value for mortality.(22-24,34,53,75,77,93) Other nutritional screening tools also demonstrated an association with mortality in the studies included: 8% NRS 2002,(25,30,36,48,49,88) 5% MUST,(41,56,78,86) 5% NUTRIC,(32,72,95) 5% GNRI,(33,61,68) 3% m-NUTRIC,(72,83) HFR-TNP,(44) 1% AND-ESPEN,(52) 1%MST,(91) 1% NRS modified,(40) and 1% 3-MinNS.(79) Regarding nutritional assessment tools, evidence of mortality prediction was found for 5% PG-SGA,(65,90,92) 5% SGA,(37,65,89) and 1%ESPEN-DCM.(76) In terms of hospital length of stay, the tools that demonstrated an association were 9% SGA,(28,37,39,43,71,85,95) 3% MNA,(51,64) 1% AND-ESPEN,(52) and 1% O-PNI(39) among the assessment tools; and 5% NRS 2002,(39,88,95) 5% MUST,(39,56,64) 3% MST,(86,91) 1% MUST modified,(41) 1%CONUT,(39) 1% 3-MinNS,(79) 1% SNAQ,(86) and 1% m-NUTRIC(72) among the screening tools. CONUT demonstrated predictive value for postoperative complications, survival, performance, and quality of life across six studies.(38,45,50,59,60,68) ANS was the only screening tool that did not show a positive relationship with clinical outcomes in the included studies. Functional capacity was evaluated in four studies, with reported associations for MNA-SF,(69) SGA,(29) MUST,(49) and GNRI tools.(87) When stratifying the analysis for Latin populations, only 13 of the 77 studies (17%) were conducted in this subgroup, with 11 (14%) conducted in Brazil. Of the Brazilian studies, only one was multicenter; the median sample size was 470 patients, and 7 studies were prospective. Mortality and hospital length of stay were the most frequently assessed outcomes. Among tools analyzed in Brazilian populations, SGA (assessment) and NRS 2002 (screening) were the most studied. In two studies,(85,89) the combination of SGA and NRS 2002 provided more reliable results for length of hospital stay, complications, and mortality among hospitalized individuals. Regarding tools commonly used in Brazilian hospitals, such as NRS 2002,(10) MNA,(97) MNA-SF,(98) and SGA,(99) the original validation articles did not validate these tools based on clinical outcomes. Instead, they were validated using clinical parameters such as biochemical and clinical assessments, including both positive and negative studies on the impact of nutritional therapy. These articles were considered in the PRISMA flowchart under "Identification of studies via other methods." Concerning GLIM, it is not a screening or nutritional assessment tool but rather a set of diagnostic criteria. Six articles evaluated different criteria,(31,46,47,57,66,76) all within surgical or clinical populations. Of these, 83% were conducted in Asian populations and 83% in oncological patients. The median sample size was 711 patients, and all studies were single-center. Across studies, the initial screening tool varied (NRS 2002, MUST, MNA-SF, or ESPEN-DCM). The studies also differed in the number and combinations of GLIM variables tested and added to the initial parameters. In two studies, functional measures such as walking speed and handgrip strength were included.(66,76) In one study that analyzed variables independently, weight loss showed the highest correlation with survival, and low muscle mass identified by tomography was associated with worse outcomes.(46) DISCUSSION The included studies covered populations from 31 countries and were predominantly conducted in single-center settings. Across the included evidence, 20 nutritional screening and assessment tools were evaluated. Mortality was the most frequently assessed clinical outcome, whereas hospital readmission was the least frequently reported. Overall, MNA/MNA-SF appeared to be associated with the broadest range of favorable clinical endpoints. When considering mortality specifically, MNA and NRS 2002 were most consistently associated with mortality risk. Among nutritional assessment tools, the subjective instruments PG-SGA and SGA were among the most extensively studied and showed associations with clinical outcomes. Regarding specific populations, relatively few studies evaluated critically ill patients. This gap may reflect an important limitation of applying existing screening and assessment tools in the ICU population. The European Society for Clinical Nutrition and Metabolism (ESPEN) guideline for critically ill patients also highlights limitations of conventional screening and assessment tools in this setting, particularly because most tools were not developed using variables that are sensitive to critical illness.(100) As a consensus recommendation (low level of evidence; expert opinion), ESPEN suggests considering any critically ill patient who remains in the ICU for >48 hours to be at nutritional risk (strong consensus, 96%). For nutritional assessment, ESPEN recommends detailed clinical assessment given the lack of a specific screening tool for this population (strong consensus, 100%).(100) Together, these points support the need for additional prospective, multicenter studies to validate tools for critically ill patients or to develop instruments that incorporate ICU-relevant variables, given the high risk of pre-existing malnutrition and the catabolic response associated with critical illness. In the context of GLIM, the heterogeneity of populations, variables, and tested combinations identified in this review limits direct comparisons across studies and reduces the ability to draw definitive conclusions. Similar concerns have been reported in prior literature, including a review by Fonseca et al.(101) and another review by Correia et al.(102) Establishing standardized research protocols will be important to improve comparability across populations and to support more consistent implementation of GLIM criteria alongside screening tools. An additional limitation is the scarcity of studies involving Latin American populations (17%), particularly from Brazil (14%). This gap raises concerns regarding the reliability and applicability of commonly used tools in these settings, as population-specific socioeconomic and clinical characteristics may not be adequately represented in the broader evidence base.(103) Future studies should prioritize Latin American populations and explicitly consider socioeconomic disparities and context-specific determinants that may influence nutritional risk and clinical outcomes. Finally, further research is needed that prioritizes clinically relevant outcomes, evaluates distinct subpopulations, and assesses both established tools and the development of new instruments. Machine-learning approaches, supported by large clinical databases, may enable the development of simpler and more accurate tools that integrate screening and assessment functions to better guide nutrition interventions. However, clinical studies combining nutritional screening or assessment with machine learning remain limited. Notably, Muñoz Díaz et al.(104) combined MNA-SF with additional features in a logistic regression model and reported improved performance using a novel variable set. Duan et al.(105) used PG-SGA-based diagnoses with selected predictors in an XGBoost model to identify key predictive variables. These initiatives illustrate the potential for more efficient, concise, and accurate nutritional evaluation methods. This study has limitations. First, the exclusive focus on hospitalized populations may limit generalizability. Second, validation studies that did not evaluate clinical outcomes were not described in detail. Third, the inclusion of "clinical outcomes" in the search strategy may have underestimated the total number of validation studies that focused primarily on diagnostic or screening performance rather than clinical endpoints. CONCLUSION Overall, nutritional screening and assessment tools used in routine practice were validated many years ago, often relying primarily on subjective clinical assessments and with relatively limited reporting of clinical outcomes. In addition, most studies were conducted in single-center settings and predominantly in non–Latin American populations. Future research should prioritize multicenter designs, improve population representativeness, and incorporate clinically relevant outcomes when evaluating both established and newly developed tools.