Malnutrition has been diagnosed as a disease in the complication or comorbidity (CC)/major complication or comorbidity (MCC) catalog of China Healthcare Security Diagnosis Related Groups (CHS-DRGS) with disease codes E40-E46 (1). Nutritional intervention was an important part of multidisciplinary treatment to improve the prognosis of patients, including a shorter hospital stay, less post-operative complications, lower mortality and reduced medical costs (2–4). There were several tools available for the assessment of malnutrition, such as the subjective global assessment (SGA), patient-generated subjective global assessment (PG-SGA), and prognostic nutritional index (PNI). However, the diagnostic criteria have not been harmonized in recent years. In September 2018, the Global Leadership Initiative on Malnutrition (GLIM) was issued to reach a global consensus on the clinical diagnosis of malnutrition (5). It made a progress that the nutritional assessment developed from the subjective methods such as SGA and PG-SGA to diagnosis malnutrition by phenotypic and etiological indicators.

Hospitalized patients undergoing hepatobiliary-pancreatic surgery have a high risk of malnutrition (6). Complex changes in nutrient metabolism and varying degrees of malnutrition may occur in hepatobiliary and pancreatic diseases. Nutritional status in turn affects the occurrence, development and prognosis of the diseases to form a vicious circle. Therefore, it was vital to early diagnosis malnutrition and timely implement nutritional intervention in the multidisciplinary. The applicability and effectiveness of the GLIM criteria have been verified among hospitalized patients with stroke (7), chronic pulmonary obstructive diseases (COPD) (8), gastrointestinal surgery (9), and heart failure (10). However, malnutrition prevalence, as determined by GLIM, among hospitalized patients undergoing hepatobiliary-pancreatic surgery remains largely unknown. Therefore, we aimed to (1) assess its prevalence, as defined by the GLIM, among hospitalized patients undergoing hepatobiliary-pancreatic surgery; (2) compare the PG-SGA, SGA, and PNI with the GLIM criteria in diagnosing malnutrition among these patients; (3) observe the nutritional intervention in hospitalized patients; (4) analyze the factors influencing nutritional risk and malnutrition among these hospitalized patients. In summary, we hope that our research can provide a reference for the more accurate nutritional diagnosis and treatment management of this group.

The prevalence of malnutrition among hospitalized patients undergoing hepatobiliary-pancreatic surgery is high (6), but the diagnosis has been controversial (21). In 2018, the American Society for Parenteral and Enteral Nutrition (ASPEN) and the Parenteral and Enteral Nutrition Society of Asia (PENSA) proposed the new GLIM criteria. These criteria provide new insights for the diagnosis of malnutrition. While, its varied prevalence in the same group may occur when diagnosed using different criteria or methods.

In this cross-sectional study, the GLIM criteria showed the lowest prevalence, and PNI defined the highest among the four methods. The prevalence was 36.75, 44.58, and 60.24%, as diagnosed by the GLIM, PG-SGA, and PNI, among 332 tumor patients. Among the 174 non-tumor patients, the prevalence was 9.77, 10.92, and 32.18% as determined by the GLIM, SGA, and PNI, respectively. A similar conclusion was reached by Miwa' study (22), the GLIM also showed the lowest prevalence (21%) among chronic liver disease patients when compared with SGA(35%), and RFH-GA (26%). As we know, the incidence of malnutrition varies relay on how estimator used the assessment tools, but regardless of the assessment tool, clinical outcomes in hospitalized patients with or without malnutrition are quite different.

The results of the Cohen's kappa statistic showed that the PG-SGA and GLIM had a relative high level of agreement (Kappa = 0.814, P < 0.001), the SGA and GLIM had a good level of agreement (Kappa = 0.752, P < 0.001), while the PNI and GLIM had fair level of agreement (Kappa = 0.265, P < 0.001). In the previous study, the PG-SGA and GLIM had a moderate level of agreement (kappa = 0.519, P < 0.001) in 360 esophageal cancer (EC) patients after esophagectomy (8). A possible explanation of the fair level of agreement between the GLIM and PNI may be that the PNI was a completely objective tool, including the level of serum albumin and the total number of peripheral leukocytes, while it did not consider BMI or reduced food intake. Meanwhile, the cutoff value of the PNI was different for different diseases. The widely used PNI cutoff value in gastrointestinal surgery is 45, and previous studies suggested that PNI ≤ 45 was significantly associated with poor prognosis in patients (23, 24). Moreover, Wang et al. (25) excluded patients with PNI ≤ 45, and the results showed that those with PNI≥ 55 had more difficulty obtaining a pathological complete response than patients with PNI >45 and <55. In Li Chen's study (26), the optimal cutoff value of PNI was 51. Therefore, the range of PNI cut-off values should be explored in clinical studies with large samples.

The possible reasons that the GLIM showed the lowest prevalence of malnutrition may be as follows: (1) in our study, most cases met the etiologic criteria because of the apparent reduced food intake, while did not meet the phenotypic one, mainly because that their weight loss did not meet the diagnostic criteria interval; (2) The GLIM criteria recommended the use of the dual-energy X-ray absorptiometry (DEXA), the computed tomography (CT) and the magnetic resonance imaging (MRI) to get the fat free mass index (FFMI), some other measurement of body composition such as bioelectrical impedance analysis (BIA) can be used as a reference. The normal values of the calculated muscle quality and the cut-point standards were still lack in China. So we excluded its measurement in our study, which may cause the lower prevalence of malnutrition diagnosed by the GLIM. In the present, some researchers have attempted to use the 15th percentile (p15) of the CC (A value <p15, male, 30 cm; female, 29 cm) as a positive for reduced muscle mass (8, 27). However, the optimal cut-off values for reduced muscle mass are debated. Other researchers discovered that using body composition measurements can further improve the GLIM criteria performance compared to using anthropometric measurements alone (28). There are racial and gender differences in normal values of the muscle mass and cut-off criteria, so the threshold of the related index of muscle mass reduction requires lager samples from different ethnic groups (e.g., Chinese, European) and higher quality research data.

Nutritional screening, nutritional assessment, nutritional intervention and monitoring were consisted of the standardized nutritional support therapy strategy. Among them, nutritional intervention plays a significant role (6). In our study, a low proportion of nutritional support was provided during hepatobiliary surgery. This may be because of the lack of recommendations for patients with nutritional risk by bedside physicians, and that of high-quality literature reports proving that nutritional support impacts the improvement in outcomes. So the medical staff should focus on the nutritional status and the nutritional intervention of hospitalized patients.

We found that the mean age of malnourished group was significantly higher than the well-nourished group according to all the four diagnostic tools. Moreover, other nutritional characteristics, including albumin and hemoglobin, were significantly lower in the malnourished group. These results are consistent with those of domestic and foreign studies (29, 30). Furthermore, the univariable analysis showed that older age, lower BMI and tumorous were risk factors for nutritional risk and malnutrition. The presence of benign or malignant disease due to the typical characteristics of malignancies may alter nutritional status (e.g., inflammation, central anorexia, metabolic disturbance worsen food intake, muscular status and energy reserve, which allow to define good nutritional status or malnutrition, are affected by tumorous).

Poulter (31) compared the GLIM, ESPEN, and ICD-10 criteria and identified 23.0%, 5.5%, and 12.6% of the cohort as malnourished, respectively. Xiaolin et al. compared the GLIM and PG-SGA criteria among cancer patients. They also found a lower prevalence of malnutrition diagnosed by the GLIM criteria (36.8 vs. 42.5%) (32). Regarding study strengths, this study is one of the first attempts to assess the diagnostic performance of four screening tools for malnutrition diagnosis, including the GLIM, PG-SGA, SGA, and PNI criteria, among hospitalized patients undergoing hepatobiliary-pancreatic surgery. Thus, our results show an important gap in this field.

However, our study has some limitations. First, it excluded reduction in muscle mass, which may have led to bias. Due to the short-staffed and the lack of equipment such as the dual-energy X-ray absorptiometry and bioelectrical impedance analysis in our study, we could not obtain the precise data about the muscle mass. Second, the GLIM calls for the use of the criteria in prospective and retrospective cohort studies to validate their application for clinical practice. However, our study design was retrospective and we used medical records, which inevitable had missing data, and the risk of selection bias was not negligible. Third, many studies observed the relationship between malnutrition and the adverse clinical outcomes of patients (33, 34). We mainly focused on short-term nutritional characteristics, and long-term clinical outcomes, such as complications, hospital readmission and mortality were not observed or reported. In the future, high-quality prospective clinical validation reports and prospective multi-center clinical application data analysis will be the basis for the improvement of the GLIM. The association between nutritional status and clinical outcome of patients assessed by various tools needs to be further explored.

Our study has practical implications. We assessed the prevalence of malnutrition, as defined by the GLIM, among hospitalized patients in hepatobiliary-pancreatic surgery and compared it with the other three criteria, which will help to reduce the bias of early clinical application of GLIM. Our findings provided some guidance and reference for the diagnosis of malnutrition, which may improve scientific and standardized nutrition management in hepatobiliary and pancreatic diseases. Meanwhile, we attempted to lay the foundation for further GLIM prospective clinical validation and health economics research. It is recommended to fill in the first page of the medical record with malnutrition information for disease codes E40-E46 and hopefully our findings can provide objective and valuable information for the diagnosis of malnutrition.

In conclusion, the incidence of nutritional risk and malnutrition was high among patients with hepatobiliary and pancreatic diseases, specifically in those with tumors. Among the four tools, the GLIM defines the lowest prevalence of malnutrition. The PG-SGA and GLIM had a relative high level of agreement. We should pay attention that there was only small part of patients received nutritional support in our current study. Further prospective and well-designed cohort studies were needed to validate the relevance of the GLIM criteria to explore the “gold standard” for clinical validity.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Ethics Committee of the Ningbo Medical Center Lihuili Hospital. The patients/participants provided their written informed consent to participate in this study.

LZ contributed to conception, administration, writing, and editing of the manuscript. KJ, ZD, and RW contributed to nutritional risk screening and assessment of malnutrition. JF and LY contributed to data collection including disease diagnosis, age, sex, height, BMI, LOS, and laboratory data. All authors contributed to the article and approved the submitted version.