Stroke is a significant cause of death and disability worldwide, and its impact is growing over time (1). According to estimates, approximately 13.7 million new cases of stroke occur worldwide each year (2). The Kingdom of Saudi Arabia (KSA) has a significant burden of stroke, with the incidence rate expected to double by 2030 (3). Stroke is the second most common cause of death in KSA (4). One of the major complications after stroke is dysphagia, affecting over 50% of stroke survivors (5). Furthermore, stroke survivors are at a high risk of developing malnutrition and nutritional issues due to various factors, mainly associated with neurological complications (6).

Malnutrition is prevalent among stroke survivors and has been associated with a higher risk of complications, prolonged hospital stays, reduced functional improvement during rehabilitation, and elevated mortality rates (7). Malnutrition affects 6 to 62% of stroke survivors (7). Furthermore, stroke survivors face a high risk of malnutrition, which significantly impedes recovery and long-term health (7, 8). Therefore, accurate and early identification is critical for timely nutritional intervention and improved outcomes (9).

Nutritional assessment is the first and most crucial step in nutrient management (10). In recent decades, several nutritional screening tools have been developed and put into practice. The European Society for Clinical Nutrition and Metabolism (ESPEN) endorses the Nutrition Risk Screening 2002 (NRS-2002) as the standard tool for screening, recognizing it as a reliable method for evaluating the risk of malnutrition (11–13). In addition, the subjective global assessment (SGA) is a clinical tool designed to evaluate nutritional status by integrating anthropometric, biochemical, and other indicators (14). It is a widely recognized tool for identifying disease-related malnutrition due to its safety, simplicity, and cost-effectiveness (14). While it has some limitations, the SGA is regarded as the gold standard for validating other nutritional assessment methods (15).

Recently, a core leadership committee composed of representatives from major global clinical nutrition societies, including the American Society for Parenteral and Enteral Nutrition (ASPEN), ESPEN, the Federación Latinoamericana de Terapia Nutricional, Nutrición Clínica y Metabolismo (FELANPE), and the Parenteral and Enteral Nutrition Society of Asia (PENSA), was formed to standardize adult malnutrition criteria and create a globally accepted tool for diagnosing malnutrition across various clinical settings (16). The Global Leadership Initiative on Malnutrition (GLIM) was established in 2019 to address the increasing global burden of malnutrition, which includes undernutrition, overweight, and obesity. The GLIM approach involves two steps: the first step is risk screening to identify at-risk patients using any validated screening tool, such as the Nutritional Risk Screening 2002 (NRS2002), and the second step is the assessment (diagnosis) of malnutrition. The GLIM outlines specific phenotypic criteria (e.g., weight loss, low BMI, reduced muscle mass) and etiological criteria (e.g., reduced food intake, inflammation) that must be met for a malnutrition diagnosis. GLIM represents a consensus that requires validation for clinical application (16). Additional evidence is necessary to confirm the validity of the GLIM criteria. To explore this, we hypothesized that GLIM is a reliable tool for diagnosing malnutrition among Saudi stroke survivors when compared to the Subjective Global Assessment (SGA). Therefore, the primary objective of this study was to validate the use of GLIM criteria as a diagnostic tool for malnutrition in stroke survivors in comparison to SGA (16).

The present study aimed to validate the use of GLIM criteria as a diagnostic tool for malnutrition in stroke survivors, in comparison to the SGA. Our findings indicate that the GLIM criteria demonstrate acceptable performance in diagnosing malnutrition, with an area under the receiver operating characteristic curve (AUC) of 0.819. The sensitivity was found to be 78.2%, and the specificity was excellent at 85.7% when compared to the SGA. Furthermore, both GLIM and SGA showed significant correlations with various nutritional parameters, including weight, BMI, MUAC, CC, and MAMC.

The SGA has been widely used worldwide and validated across various clinical settings with different populations. The GLIM criteria group considers it a “semi-gold” standard (16). In addition, several studies have validated GLIM criteria compared to SGA among different populations (24–26).

Our study revealed a significant prevalence of malnutrition among stroke survivors, with rates of 47.1% according to the GLIM criteria, comparable to Brito et al. (26) at 41.6%. Concordance between GLIM and SGA was strong, with 51.9% malnutrition identified by SGA, consistent with Balci et al. (25) (κ = 0.804). However, our study’s findings differ from Allard et al. (27), who reported lower GLIM sensitivity (61.3%). On the other hand, our findings indicated greater effectiveness in identifying malnourishment among stroke survivors, underscoring the significance of GLIM in advancing prompt nutritional therapies and enhancing recuperation results (16).

When comparing the well-nourished and malnourished stroke survivors by GLIM and SGA, there was a significant difference in weight, BMI, MUAC, CC, and MAMC (p < 0.05). For instance, the weight of a well-nourished survivor using GLIM was 75.24, while that of the malnourished survivor was 64.80, p = 0.001. This means that underweight stroke survivors would have weighed significantly less compared to those who were well-nourished. These findings are further supported by the BMI readings: well-nourished subjects had a BMI of 27.08 kg/m², while those malnourished had a BMI of 23.66 kg/m² (p = 0.000). This agrees with Speranza et al. (28), who, after conducting their research, reported that lower values of both BMI and CC were found to be significantly associated with higher nutritional risk, thus underlining once again the importance of this measure in the identification of malnutrition, as well as placing emphasis on the nutritional status concerning recovery and rehabilitation outcomes post-stroke. The consistent findings between the GLIM and SGA criteria further underscore their clinical utility for the diagnosis of malnutrition, with both tools indicating significant differences in important anthropometric measures of nutritional status. These findings indicate that a gross compromise in muscle mass and general physical health, characterized by compromised anthropometric measures such as MUAC and CC, is a key determinant of recovery following stroke.

Evaluation of biochemical aspects of stroke survivors based on GLIM and SGA criteria provides interesting insights into the diagnosis of malnutrition. The minor differences observed in biochemical markers, especially in Table 3, align with the results reported by Fiorindi et al. (29), stating that simple, routine nutritional indicators are sometimes insufficient to determine malnutrition among patients. As shown in Table 2, most of the biochemical markers recorded were similar between well-nourished and malnourished groups (p-values > 0.05), with a slight trend in total cholesterol, the values for which were lower among malnourished (3.33 ± 0.78 mmol/L) compared to well-nourished (3.75 ± 1.32 mmol/L). This aligns with other studies that have demonstrated that low cholesterol levels can indicate malnutrition in stroke patients, potentially resulting from insufficient nutrient intake or changes in cholesterol metabolism. In contrast, Table 3 shows a significant difference in total cholesterol levels based on SGA (p = 0.049), which is consistent with the findings of Galindo Martín et al. (30), who reported a correlation between malnutrition identified by GLIM criteria and adverse clinical outcomes. Their findings indicate that cholesterol levels may play a crucial role in evaluating nutritional status and predicting complications across different patient groups. Consequently, albumin and hemoglobin, which are biochemical markers of malnutrition, might not effectively distinguish between stroke patients with malnutrition and those who are well-nourished. Combining GLIM and other clinical criteria with comprehensive assessments could enhance nutritional interventions.

According to the GLIM criteria, there were no significant differences in the incidence of diabetes mellitus (DM), hypertension (HTN), and dyslipidemia (DLP) between malnourished and non-malnourished patients, with p > 0.05. This suggests that the GLIM criteria may not effectively capture the complex relationships between malnutrition and specific comorbidities in this patient population. However, the SGA tool showed a p-value of 0.050 in HTN and 0.003 in DLP between the malnourished and well-nourished groups. These findings show that SGA is more useful in determining the relationship between nutrition and these diseases. The strong association with DLP corresponds with a previous study, which notes that malnutrition may worsen disorders in lipid metabolism and contribute to worse cardiovascular outcomes (7). Thereby, despite the promising findings regarding the application of GLIM for the diagnosis of malnutrition, SGA may offer a better understanding of the connection between malnutrition and comorbidities, including DLP and HTN, which supports the necessity for the use of multiple screening tools in clinical practice to enhance the approach to the patient.

The agreement level between the GLIM criteria and the SGA was κ = 0.635, indicating substantial concordance. These findings are consistent with earlier studies across different populations that have validated the GLIM criteria against the SGA. For instance, a retrospective study by Zhang et al. (31) involving 3,777 cancer patients reported an agreement of κ = 0.54 and a low sensitivity of 70%. Similarly, a retrospective cohort study of 784 hospitalized patients found that while the GLIM criteria demonstrated good specificity at 89.8%, their sensitivity was lower at 61.3% compared to the SGA (27).

Few studies have examined the prevalence of malnutrition in stroke patients using GLIM criteria. A cross-sectional study involving 304 stroke patients in the rehabilitation phase reported that 25.3% were diagnosed with malnutrition, with 67.5% categorized as severely malnourished (32). Additionally, a retrospective cohort study of 122 stroke patients revealed a higher prevalence of 64.8% (33). In contrast, our study found that 47.1% were diagnosed with malnutrition. The elevated percentage of malnourished patients in the retrospective cohort study may be attributed to the characteristics of the enrolled population, which primarily consisted of older adults (33). Furthermore, a cross-sectional study of 115 acute stroke patients indicated that 28.7% were diagnosed as malnourished (34).

It is well documented that BMI is the main significant factor influencing the risk, outcomes, and rehabilitation of stroke patients. The present study found that about 14.3% of malnourished survivors were underweight. The findings are consistent with a retrospective cohort study that observed the BMI of the majority of malnourished patients with strokes underweight range. Another study found that 22% of the patients had a low BMI (35). These findings highlight the significant proportion of stroke patients who are at risk of malnutrition due to low BMI. Additionally, another study found that 28.7% of the patients had malnutrition (34).

A key indicator of malnutrition is non-volitional weight loss, a criterion strongly highlighted by the GLIM consensus (16). In the present study, non-volitional weight loss was the predominant phenotypic criterion observed in 91.8% of malnourished survivors. This finding is consistent with a previously mentioned retrospective cohort study, which found that weight loss was the most common phenotypic criterion, affecting 83.9% of patients with stroke (33).

Another important phenotypic criterion in the GLIM evaluation of malnutrition is reduced muscle mass. In our study, reduced muscle mass was identified in 41.8% of the well-nourished group, while this percentage increased to 87.8% among malnourished individuals. This aligns with findings from a single-center cohort study of 189 patients with acute stroke, which reported low muscle mass in 46% of males and 56% of females (36). Therefore, early evaluation of muscle mass in post-stroke patients may be essential.

Regarding etiological criteria, reduced food intake was identified as the most common factor among malnourished survivors, occurring in 89.8% of cases, while disease or inflammation was present in 49% of these individuals. Inflammation was assessed using the neutrophil-to-lymphocyte ratio (NLR) in instances where albumin levels were not available. NLR is a widely recognized and easily accessible inflammatory marker. However, it is important to emphasize that low albumin is the preferred indicator of inflammation, as recommended by the GLIM group (16).

In addition to their accuracy in identifying malnutrition, clinicians must consider the strengths and limitations of each nutrition screening tool. This study is among the few that have assessed the validity of various malnutrition screening instruments, including the latest GLIM, within the rehabilitation context in Saudi Arabia. Consequently, we believe our findings provide valuable insights into the critical selection of appropriate malnutrition screening tools in stroke rehabilitation settings.

However, this study has several limitations. Since this study was conducted at a single center in Riyadh City, the generalizability of the results to the broader Saudi population is limited. Additionally, food intake was evaluated through self-reporting rather than direct dietary assessment methods. Furthermore, men and women were pooled in the analysis without sex-specific adjustments, and the proportion of female participants was relatively low compared with males, which may limit the applicability of the findings to female patients. Therefore, caution is warranted when interpreting the results in females. Future studies should apply sex-stratified analyses for more precise interpretation. Future research is needed to explore the feasibility and predictive validity of these malnutrition screening tools.

In conclusion, the GLIM criteria demonstrated good sensitivity and specificity, as well as acceptable criterion validity, and substantial agreement with the SGA was observed. Validation studies comparing GLIM with SGA on a larger nationwide scale are strongly recommended to further confirm these findings in stroke survivors. Additionally, it is advisable to conduct GLIM validation studies in Saudi Arabia and abroad with various patient populations to explore the potential of adopting GLIM criteria as a standardized nutritional assessment tool in different institutions and hospitals.