Study population and design
This study is a retrospective study based on the Investigation on Nutrition Status and its Clinical Outcome of Common Cancers (INSCOC) cohort in China; a detailed description of the design, methods and development of the INSCOC study was provided elsewhere [22, 23]. The patients with pathologically diagnosed solid tumor(s) at any stage who met the inclusion criteria were recruited from multiple institutions in China between 2013 and 2020. The inclusion criteria in the present study were: 1) patients aged 65 years or more; 2) a histological diagnosis of the solid malignant tumor; and 3) a hospital stay longer than 48 h. The exclusion criteria were: 1) patients with Acquired immunodeficiency syndrome (AIDS) or transplanted organ(s); 2) patients who were admitted to the intensive care unit (ICU) and were in a critical condition at the beginning of recruitment, 3) patients who refused to participate or would not cooperate with the questionnaire survey. Additionally, as shown in the study schematic (Supplementary Fig. 1), participants who had a missing critical clinical examination, or follow-up data, or more than 10% of all data, were excluded. Finally, 2724 elderly patients were included in the current analysis. The study was conducted in line with the Helsinki declaration; its design was approved by the local Ethics Committees of all participants’ hospitals. All patients signed an informed consent form before participating in the study. The trial was registered at http://www.chictr.org.cn with registration number ChiCTR1800020329.
Body mass index (BMI) was calculated for all participants, categorized using the classifications for the Chinese population: underweight (< 18.5 kg/m2), normal weight (18.5 ~ 23.9 kg/m2), overweight, or obesity (> 24 kg/m2). Malnutrition was assessing used three nutritional assessment tools for all patients. First, participants were evaluated by dietitians using standard PG-SGA to determine their degree of malnutrition. The participants were classified into two categories: non-malnutrition (PG-SGA < 4); malnutrition (PG-SGA ≥ 4). Based on GLIM criteria, at least one phenotypic (weight loss (%) within 6 months, low BMI, and reduced muscle mass) and one etiologic (reduced food intake or assimilation, disease burden, and inflammatory condition of cancer) criterion were required to diagnose malnutrition when participants were screened at risk of malnutrition in NRS 2002 . As all participants with cancer met the etiologic criterion of the GLIM criteria, it was excluded from the GLIM used in this study. The PG-SGA SF consists of four boxes: 1) body weight, 2) food intake, 3) symptoms affecting oral food intake, and 4) activities and function. According to the PG-SGA SF, the optimal cut-off value to determine malnutrition was five by using maximally selected rank statistics (supplementary Fig. 2).
The demographic, anthropometric, and clinical parameters were collected for all participants with the first 48 h after admission, including gender, age, BMI, primary tumor site, TNM stage, Chronic Disease information, lifestyle habits (e.g., alcohol, smoking), Karnofsky Performance Status (KPS). Pathological staging was defined according to the 8th edition of the AJCC TNM staging system. Treatment information and follow-up data were also collected. Fasting blood tests, such as albumin, globulin, creatinine, neutrophil, and lymphocyte, were collected with standard laboratory techniques within 48 h of admission. Albumin-globulin ratio (A/G) and neutrophil-lymphocyte ratio (NLR) were calculated, the NLR ≥ 3 was defined as elevated NLR in this study. The calf-circumference (CC) was measured using flexible and non-elastic tape. Handgrip strength (HGS) was measured in the dominant hand with a Jamar dynamometer.
The primary outcome was overall survival. Through the electronic medical record system, patients with one or more re-admission records could be extracted. The overall survival (OS) time was defined as participants who were followed up from the date of the first-time admission until death from any cause, or the end of follow-up (December 31, 2020), whichever came first. For overall mortality and the event rate per 1000 patient-years of follow-up, individuals alive at the end of follow-up were censored at that time.
Variables are expressed as the means±standard (SD), percentage, or median with interquartile range. Their differences were analyzed using Student’s t-test to see if variables followed a normal distribution or nonparametric tests (Mann-Whitney or Kruskal-Wallis) if variables did not follow a normal distribution. Qualitative variables were analyzed using chi-square tests or Fisher corrections if necessary. Kaplan-Meier curves were used to analyze the survival data, and the Log-rank tests were used to compare survival between groups. Cox regression analysis was used to evaluate the prognostic impact of malnutrition, including those covariates associated with known poor prognosis or p-value < 0.05 in the univariate cox analysis. Adjusted hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) were derived from Cox models after adjusting for covariates.
The possible linear relationship between PG-SGA SF and the all-cause mortality was evaluated using restricted cubic spline regression. The time-dependent receiver operating characteristic (ROC) curves, area under the curve (AUC) analyses,and ROC-AUC values for each time-point were used to evaluate the predictive performance of the malnutrition assessment tools. The Harrell C-statistics, continuous net reclassification improvement (cNRI), and integrated discrimination improvement (IDI) were calculated to assess and compare the discrimination capacity of the PG-SGA SF to predict mortality. Calibration curves were generated by comparing the predicted survival with the observed survival after bias correction. To evaluate the potential clinical net benefit of the model, the researchers performed a decision curve analysis (DCA). The significance level was set at P < 0.05 (two-sided probability). All analysis was performed using R version 3.6.2 (http://www.rproject.org). DCA was performed using the source file “stdca.r”, downloaded from https://www.mskcc.org.