Undernutrition and associated factors among adult cancer patients in northwest Ethiopia: a global leadership initiative on malnutrition-based approach Authors: Tizita Mesfin, Samuel Dagne, Abebaw Alemu, Hunegnaw Almaw, Adane Nigusie, Netsanet Fentahun Published: 2025 ABSTRACT AbstractBackgroundUndernutrition contributes to nearly half of global cancer-related deaths. In Ethiopia, previous studies using the Subjective Global Assessment (SGA) have reported an undernutrition prevalence of 30–60% among cancer patients. The Global Leadership Initiative on Malnutrition (GLIM) criteria offer a standardized, evidence-based approach by combining phenotypic and etiologic factors. However, their application and related evidence remain limited in Ethiopia, particularly in Bahir Dar City.ObjectiveTo assess the prevalence of undernutrition and its associated factors among adult cancer patients receiving follow-up care at public referral hospitals in Bahir Dar City, Northwest Ethiopia, in 2024.MethodsAn institution-based cross-sectional study was conducted among 403 cancer patients on follow-up at Felege Hiwot and Tibebe Ghion Comprehensive Specialized Hospitals from June 8 to July 8, 2024. Participants were selected using a simple random sampling technique. Data were collected via face-to-face interviews using a structured questionnaire and through chart reviews. Anthropometric measurements, dietary intake, and clinical data were used to assess undernutrition based on the GLIM criteria. Data were entered using the KoboToolbox and analyzed with SPSS version 26. Bivariable and multivariable binary logistic regression analyses were conducted to identify factors associated with undernutrition. Statistical significance was determined at a 95% confidence interval (CI) and p-value < 0.05.ResultsThe study included 393 participants, yielding a response rate of 97.8%. The prevalence of undernutrition among adult cancer patients was 50.3% (95% CI: 44.4%–55.5%), with 31.5% classified as severely undernourished and 18.8% as moderately undernourished. Multivariable analysis identified four factors significantly associated with undernutrition: loss of appetite (AOR = 18.21, 95% CI: 10.01–33.12), presence of chronic comorbidities (AOR = 3.86, 95% CI: 1.67–8.94), female sex (AOR = 2.17, 95% CI: 1.17–4.03), and larger tumor size, with the highest risk observed in T4 tumors (AOR = 17.98, 95% CI: 6.45–29.19).ConclusionUndernutrition was common among adult cancer patients in Bahir Dar City, with loss of appetite, female sex, tumor size, and comorbidities as significant predictors. Early application of the GLIM criteria, management of treatment-related symptoms, and targeted nutrition interventions for high-risk groups, particularly women, are recommended to improve patient outcomes.Clinical trial numberNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12885-025-15488-9.Keywords: Cancer patients, Undernutrition, GLIM, Public referral hospitals, Bahir dar, Ethiopia Introduction Cancer imposes a major and growing global burden, with nearly 20 million new cases reported in 2022 and projections indicating a rise to 35 million by 2050 [1, 2]. This escalation is particularly concerning in low- and middle-income countries, where healthcare systems often face substantial resource constraints. In Africa, gastrointestinal cancers, including stomach, colorectal, and liver cancers, constitute a significant proportion of cancer diagnoses, influenced by regional risk factors and gaps in prevention and treatment services [3, 4]. Ethiopia mirrors this trend, with cancer incidence rising by approximately 32% over the last decade [5]. Stomach and colorectal cancers are among the most prevalent types, and malnutrition, particularly undernutrition, frequently complicates the clinical course, affecting patient outcomes [6, 7]. Globally, undernutrition among cancer patients is highly prevalent. Studies show that severe undernutrition affects over one-third of patients worldwide, with even higher rates in low- and middle-income countries. For example, undernutrition rates range from 35% to 65% in some European and Latin American countries [8–13]. In Africa, studies utilizing nutritional screening tools such as the Patient-Generated Subjective Global Assessment (PG-SGA) and the Malnutrition Universal Screening Tool (MUST) have reported undernutrition prevalence rates as high as 76% among cancer patients [14, 15], while data from Ethiopia indicate undernutrition rates between 48% and 61% among cancer patients, depending on cancer type and study setting [16–18]. Such a high burden has serious clinical implications, as undernutrition is known to impair treatment tolerance, prolong hospitalization, reduce quality of life, and increase mortality among cancer patients [7, 19, 20]. Malnutrition is a pervasive and critical concern among cancer patients globally. It involves an imbalance in nutrient intake and can manifest as either undernutrition, characterized by weight loss, reduced muscle mass, and micronutrient deficiencies, or overnutrition [21]. In cancer patients, malnutrition most commonly presents as disease-related undernutrition, driven by inflammatory responses and metabolic alterations associated with the tumor and its treatment. These changes include systemic inflammation, muscle wasting, and disrupted energy metabolism, all of which are linked to impaired treatment response, decreased quality of life, increased morbidity, and higher mortality rates [7, 22]. The pathophysiology of cancer-related malnutrition is multifactorial. Tumor-induced inflammation elevates cytokines such as interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α), which contribute to muscle catabolism and fat loss. Simultaneously, reduced food intake, due to symptoms such as anorexia, nausea, vomiting, taste changes, and malabsorption, creates a hypermetabolic state that accelerates nutritional decline [23–27]. This leads to cancer-related undernutrition or cachexia, a syndrome that affects up to 80% of patients with advanced disease and is strongly associated with poor prognosis [17, 26]. Given the substantial impact of malnutrition (predominantly undernutrition) on clinical outcomes, early identification and effective management are imperative [7, 28, 29]. However, assessing nutritional status in cancer care is challenging, particularly in low-resource settings [30]. In many clinical environments, including Ethiopia, assessment primarily relies on body mass index (BMI) and biochemical tests [17, 31]. While biochemical markers such as serum albumin and pre-albumin offer objective data on nutritional status, they are influenced by factors like inflammation and hydration status [32–34]. Moreover, routine use of these biochemical tests is often constrained by cost, accessibility, and technical requirements [14]. BMI, though simple and widely used, lacks sensitivity and may fail to detect muscle wasting or inflammation‑driven malnutrition, as it does not reflect body composition or the distribution of fat and muscle tissue, and patients with normal BMI may still have significant muscle loss and associated poor outcomes [29, 35, 36]. In research settings, the Subjective Global Assessment (SGA) is commonly employed as a gold standard tool, valued for its comprehensive approach that includes dietary history, functional status, and physical examination [16, 18]. However, SGA’s reliance on clinician expertise and patient communication, coupled with its subjective nature, limits its universal applicability and reproducibility [37]. To address these challenges and provide a standardized diagnostic approach, the Global Leadership Initiative on Malnutrition (GLIM) introduced international consensus criteria. The GLIM framework combines phenotypic criteria, such as low BMI, reduced muscle mass, and unintentional weight loss, with etiologic criteria, including reduced intake/absorption and systemic inflammation [29]. GLIM aims to harmonize malnutrition diagnosis across populations and clinical contexts, enhancing comparability and guiding interventions. Importantly, GLIM is primarily recommended to assess the magnitude and severity of malnutrition, representing an advancement in nutritional assessment, especially in settings where comprehensive biochemical or functional testing is not feasible [28, 29]. Despite its importance, nutritional screening remains inadequately implemented. Studies show that up to 37% of hospitalized patients are not routinely screened for malnutrition [38, 39]. In Ethiopia, BMI and biochemical assessments are still the primary tools in clinical use, even though they often fail to detect undernutrition effectively [16, 18]. Although SGA is commonly used in research, its limitations restrict widespread clinical application. This underscores the urgent need for standardized, context-appropriate, and practical assessment methods to guide timely nutritional interventions in oncology care. Accordingly, this study aims to evaluate the extent and severity of undernutrition among cancer patients using the GLIM criteria for malnutrition, a standardized and comprehensive approach to nutritional assessment. Although the GLIM framework defines malnutrition broadly, this study focuses on undernutrition, as overweight and obese individuals were excluded from analysis. Due to the known limitations of commonly used methods like BMI and biochemical markers in clinical practice, GLIM offers a more reliable alternative for identifying malnutrition and guiding appropriate nutritional interventions. Implementing this framework in the Ethiopian healthcare context will help generate meaningful data to enhance nutritional screening, support early detection of undernutrition, and improve patient outcomes. In low-resource settings, such evidence is essential for integrating nutritional support into routine cancer care and minimizing the impact of nutrition-related complications. Methods and materials Study area and period This study was conducted from June 8 to July 8, 2024, at two public referral hospitals in Bahir Dar City, the capital of the Amhara Regional State in northwestern Ethiopia. According to a 2018 report by the Amhara Bureau of Finance and Economic Development, the population of the Amhara Region was projected to reach approximately 22.5 million in 2020 [40]. The study was carried out at Felege Hiwot Comprehensive Specialized Hospital (FHCSH) and Tibebe Ghion Comprehensive Specialized Hospital (TGCSH), the two main public referral hospitals in the city. Together, they serve as the primary cancer treatment centers for Bahir Dar and the surrounding areas, with a referral population of approximately 7 million. FHCSH is a major referral hospital in the Amhara Region, offering services in surgery, internal medicine, pediatrics, obstetrics and gynecology, psychiatry, dentistry, orthopedics, and oncology. Its oncology unit was established in July 2017 and provides both inpatient and outpatient services. The hospital is equipped for cancer diagnosis, including tissue biopsy and imaging, and offers services such as diagnosis, surgery, and chemotherapy. During the six and three months preceding the study, approximately 2,793 outpatients and 1,470 inpatients received oncology services at FHCSH, respectively. It has 22 beds and is staffed by two senior oncologists, nine nurses, four general practitioners, and three support staff. TGCSH is a large teaching and referral hospital affiliated with Bahir Dar University. Located about 10 km south of the city center, the hospital has more than 500 beds and offers a wide range of specialized services, including oncology care. In the three months prior to the study, its oncology unit served around 1,021 cancer patients. The unit includes eight beds and is staffed by three senior oncologists and six nurses. Study design An institution-based cross-sectional study was employed. Source and study population All cancer patients aged 18 years and older enrolled and receiving follow-up care at referral public hospitals in Bahir Dar city. Inclusion and exclusion criteria All Cancer patients who were aged 18 years and older and enrolled, and had follow-up care at referral public hospitals in Bahir Dar city for at least six months. Whereas, patients who were critically ill & could not communicate were excluded from the study. Sample size determinations The sample size was calculated based on both objectives of the study. For the first objective, to estimate the prevalence of undernutrition among adult cancer patients, a single population proportion formula was used, assuming a 48.1% prevalence of malnutrition from a previous study at Hawassa University Comprehensive Specialized Hospital [18], with a 5% margin of error and 95% confidence level. This yielded a minimum required sample size of 384. Since the annual patient population exceeds 10,000, the finite population correction (FPC) was not applied, as the population can be considered effectively large. For the second objective, to identify factors associated with undernutrition, a two-population proportion formula was applied using EPI Info version 7.0, with the following assumptions: 95% confidence level, 80% power, and a 1:1 ratio of exposed to unexposed. Variables such as loss of appetite [18] and stage IV cancer [41], based on previous studies, produced a maximum required sample size of 112. Since this was smaller than the 384 calculated for the first objective, the larger sample size was retained. After adding a 5% non-response rate, the final sample size was increased to 403 participants. Sampling technique and procedures Participants were selected using simple random sampling, with patient registration lists from each hospital’s oncology unit serving as the sampling frame. Proportional allocation was applied based on patient flow over the previous three months. Felege Hiwot Comprehensive Specialized Hospital recorded 1,470 cancer patient visits (approximately 490 per month), resulting in a sample size of 238. Tibebe Ghion Comprehensive Specialized Hospital had 1,021 visits (about 340 per month), contributing 165 participants. Outcome variable: undernutrition Undernutrition was the primary outcome of this study and was assessed using the Global Leadership Initiative on Malnutrition (GLIM) criteria [29], a globally recognized framework for diagnosing undernutrition in clinical populations. GLIM requires the presence of at least one phenotypic and at least one etiologic criterion. Since this study focused exclusively on undernutrition, overweight and obese participants were excluded from the analysis, in line with GLIM recommendations to apply criteria contextually. Phenotypic indicators included Unintentional weight loss: Calculated as the percentage difference between the patient’s current weight and their weight six months prior, as documented in medical records. Although these data were obtained from routine clinical records and thus beyond direct researcher control, all facilities used regularly calibrated digital scales and standardized weighing procedures. To enhance data reliability, previous weights were verified by reviewing multiple weight records from the same patient’s chart within the six-month period to confirm consistency and accuracy. Any previous weight that was clearly inconsistent or implausible based on clinical judgment was excluded from analysis. Unintentional weight loss: Calculated as the percentage difference between the patient’s current weight and their weight six months prior, as documented in medical records. Although these data were obtained from routine clinical records and thus beyond direct researcher control, all facilities used regularly calibrated digital scales and standardized weighing procedures. To enhance data reliability, previous weights were verified by reviewing multiple weight records from the same patient’s chart within the six-month period to confirm consistency and accuracy. Any previous weight that was clearly inconsistent or implausible based on clinical judgment was excluded from analysis. Low body mass index (BMI): Measured as weight in kilograms divided by height in meters squared (kg/m²). Low body mass index (BMI): Measured as weight in kilograms divided by height in meters squared (kg/m²). Reduced muscle mass: Assessed using mid-upper arm circumference (MUAC) at the midpoint of the non-dominant arm. Although advanced techniques such as dual-energy X-ray absorptiometry (DEXA) or bioelectrical impedance analysis (BIA), CT, or MRI are preferred for assessing muscle mass, the GLIM consensus supports the use of proxy measures such as MUAC or calf circumference in resource-limited settings. In this study, MUAC was used as a practical and validated proxy to assess reduced muscle mass, as recommended when more advanced body composition measurements are not feasible. MUAC cut-off points were adopted from a large, multicenter study among adult cancer patients [42], which proposed thresholds of < 24.5 cm for men and < 23.2 cm for women as indicative of muscle mass depletion. Although these thresholds were not specifically validated for Ethiopian populations, they were selected due to the absence of national reference values for MUAC among adult oncology patients in Ethiopia. The cited study provides cancer-specific reference values derived from a diverse patient population and aligns with GLIM recommendations. In this context, these cut-offs were considered both clinically relevant and methodologically appropriate for application in our setting. Reduced muscle mass: Assessed using mid-upper arm circumference (MUAC) at the midpoint of the non-dominant arm. Although advanced techniques such as dual-energy X-ray absorptiometry (DEXA) or bioelectrical impedance analysis (BIA), CT, or MRI are preferred for assessing muscle mass, the GLIM consensus supports the use of proxy measures such as MUAC or calf circumference in resource-limited settings. In this study, MUAC was used as a practical and validated proxy to assess reduced muscle mass, as recommended when more advanced body composition measurements are not feasible. MUAC cut-off points were adopted from a large, multicenter study among adult cancer patients [42], which proposed thresholds of < 24.5 cm for men and < 23.2 cm for women as indicative of muscle mass depletion. Although these thresholds were not specifically validated for Ethiopian populations, they were selected due to the absence of national reference values for MUAC among adult oncology patients in Ethiopia. The cited study provides cancer-specific reference values derived from a diverse patient population and aligns with GLIM recommendations. In this context, these cut-offs were considered both clinically relevant and methodologically appropriate for application in our setting. Etiologic indicators included Reduced food intake: Evaluated using the Dietary Diversity Score (DDS), derived from a qualitative 24-hour dietary recall (FAO, 2021; see data collection procedure). A DDS below five food groups was considered indicative of poor dietary intake [43]. Although GLIM typically allows for quantitative assessment, DDS was used here as a practical qualitative proxy to capture dietary adequacy in a resource-limited setting. Reduced assimilation was additionally identified through the presence of nutrition-impact symptoms, such as nausea, vomiting, loss of appetite, dysphagia, diarrhea, constipation, or abdominal pain, that could interfere with digestion or absorption. Reduced food intake: Evaluated using the Dietary Diversity Score (DDS), derived from a qualitative 24-hour dietary recall (FAO, 2021; see data collection procedure). A DDS below five food groups was considered indicative of poor dietary intake [43]. Although GLIM typically allows for quantitative assessment, DDS was used here as a practical qualitative proxy to capture dietary adequacy in a resource-limited setting. Reduced assimilation was additionally identified through the presence of nutrition-impact symptoms, such as nausea, vomiting, loss of appetite, dysphagia, diarrhea, constipation, or abdominal pain, that could interfere with digestion or absorption. Reduced assimilation: Identified through the presence of nutrition-impact symptoms, such as nausea, vomiting, loss of appetite, dysphagia, diarrhea, constipation, or abdominal pain, which may interfere with digestion or absorption. Reduced assimilation: Identified through the presence of nutrition-impact symptoms, such as nausea, vomiting, loss of appetite, dysphagia, diarrhea, constipation, or abdominal pain, which may interfere with digestion or absorption. Disease burden/inflammation: Confirmed based on an active cancer diagnosis, associated with chronic metabolic stress. Disease burden/inflammation: Confirmed based on an active cancer diagnosis, associated with chronic metabolic stress. Severity of Undernutrition: Participants were classified according to GLIM phenotypic thresholds as having moderate undernutrition (Stage 1) or severe undernutrition (Stage 2), as detailed in Table 1. Table 1Severity classification of undernutrition based on GLIM criteriaSeverityWeight loss over 6 monthsBMI (Kg/m2) (< 70 years/≥70 years)MUAC (cm) (Men/Women)Normal< 5%≥ 20/≥ 22≥ 24.5/≥ 23.2Moderate undernutrition (Stage 1)5–10%18.5–20/20–2220.6–24.5/19.4–23.2Severe undernutrition (Stage 2)> 10%< 18.5/< 22< 20.6 cm <19.4 cmSource: Adapted from GLIM consensus recommendations: recommendations [29] and population-specific MUAC thresholds [42] Severity classification of undernutrition based on GLIM criteria Source: Adapted from GLIM consensus recommendations: recommendations [29] and population-specific MUAC thresholds [42] Independent variables A range of potential predictors of undernutrition were assessed based on previous studies and biological plausibility. These were grouped as follows: Socio-demographic factors: age, sex, marital status, educational status, occupation, place of residence, and household wealth index. Socio-demographic factors: age, sex, marital status, educational status, occupation, place of residence, and household wealth index. Clinical factors: cancer type, cancer stage and grade, duration since diagnosis, treatment modality (chemotherapy, radiotherapy, or surgery), comorbidities, and performance status (ECOG scale). Clinical factors: cancer type, cancer stage and grade, duration since diagnosis, treatment modality (chemotherapy, radiotherapy, or surgery), comorbidities, and performance status (ECOG scale). Dietary and nutrition-related factors: dietary diversity score (DDS), household food insecurity status (HFIAS), and nutrition impact symptoms such as anorexia, nausea, vomiting, diarrhea, constipation, and dysphagia. Dietary and nutrition-related factors: dietary diversity score (DDS), household food insecurity status (HFIAS), and nutrition impact symptoms such as anorexia, nausea, vomiting, diarrhea, constipation, and dysphagia. Lifestyle and anthropometric factors: BMI, MUAC, and unintentional weight loss over the past six months. Lifestyle and anthropometric factors: BMI, MUAC, and unintentional weight loss over the past six months. Disease burden and inflammation: presence of active malignancy, which is inherently associated with chronic inflammation and metabolic stress. Disease burden and inflammation: presence of active malignancy, which is inherently associated with chronic inflammation and metabolic stress. Standard and operational definitions Cancer stage: The stage of a cancer describes the size of the cancer and how far it has spread [44]. Tumor grade: Cancer cells are given a grade according to how different they are from normal breast cells and how quickly they are growing [44]. Food security status: Households were considered food secure if they did not experience any food-related hardships or only occasionally worried about having enough food. Otherwise, they were classified as food insecure [45]. Undernutrition: It is diagnosed using the GLIM criteria, which require at least one phenotypic criterion and one etiologic criterion [29]: Phenotypic criteria: Weight loss > 5% in the past 6 months or Low BMI (< 20 for individuals < 70 years, < 22 for those ≥ 70 years) [29] or Reduced muscle mass: defined using MUAC values of < 24.5 cm for men and < 23.2 cm for women [42]. Phenotypic criteria: Weight loss > 5% in the past 6 months or Low BMI (< 20 for individuals < 70 years, < 22 for those ≥ 70 years) [29] or Reduced muscle mass: defined using MUAC values of < 24.5 cm for men and < 23.2 cm for women [42]. Etiologic criteria: Presence of gastrointestinal or systemic factors that hinder food intake or absorption (e.g., anorexia, dysphagia, nausea, vomiting, diarrhea, constipation, abdominal pain, short bowel syndrome, acute infection, or chronic comorbidities) [29]. Etiologic criteria: Presence of gastrointestinal or systemic factors that hinder food intake or absorption (e.g., anorexia, dysphagia, nausea, vomiting, diarrhea, constipation, abdominal pain, short bowel syndrome, acute infection, or chronic comorbidities) [29]. Severity of undernutrition: It was classified according to GLIM phenotypic thresholds, as presented in Table 1. Dietary Diversity Score (DDS): Assessed by counting the number of different food groups consumed over the previous 24 h, using 10 food groups as per FAO guidelines (2021). A DDS < 5 indicates poor dietary diversity [43]. Data collection tools and procedures Data were collected using a combination of structured, interviewer-administered questionnaires and a review of patient medical records. The questionnaire was specifically developed for this study based on relevant literature and clinical guidelines, including the components required to operationalize the GLIM criteria. It captured information on socio-demographic characteristics, food intake, weight loss history, nutrition impact symptoms (e.g., anorexia, nausea), physical activity, and clinical data. An English-language version of the questionnaire is provided as Supplementary File 1. Socio-demographic and wealth index questions were adapted from the 2016 Ethiopian Demographic and Health Survey (EDHS) [46]. The wealth index was constructed using Principal Component Analysis (PCA) based on variables such as household asset ownership, type of latrine, and water source, and households were categorized into three groups: poor, middle, and rich. Clinical information, such as tumor size, cancer grade and stage, time since diagnosis, comorbidities, treatment history, and treatment goals, was extracted from patients’ medical records. The household food insecurity status was estimated using the Household Food Insecurity Access Scale (HFIAS), a standardized tool developed by the Food and Nutrition Technical Assistance (FANTA) project, a set of questions associated with the experience of food access that appeared to distinguish food-secure from food-insecure households across different cultural contexts. HFIAS has nine occurrence questions (yes or no), followed by a frequency-of-occurrence question to determine whether the condition happened rarely (once or twice), sometimes (three to ten times), or often (more than ten times) in the past four weeks. The HFIAS occurrence questions relate to three different domains of food insecurity: anxiety and uncertainty of food supply, insufficient food quality, and insufficient food intake and its physical consequences. The HFIAS tool is validated in developing countries, including Ethiopia, to measure the access components of food insecurity in urban and rural settings [47]. The continuous HFIAS score is categorized into four levels of household food insecurity (access) prevalence: food secure, mildly, moderately, and severely food insecure [45]. Dietary diversity among adult cancer patients was evaluated using the standardized guidelines developed by the Food and Agriculture Organization (FAO) of the United Nations [43]. The DDS is a qualitative indicator used to assess the variety of foods consumed over 24 h, serving as a proxy measure for nutrient adequacy at the individual level. Data were collected through a 24-hour dietary recall, during which participants were asked to report all foods and beverages consumed in the preceding day. To minimize recall bias, the 24-hour recall was conducted qualitatively following FAO (2021) recommendations. The recall period was clearly defined (“from waking yesterday to waking today”), and participants were asked to list all foods and drinks consumed without estimating portion sizes. Data collectors used a standardized script and probing questions for commonly forgotten items to ensure completeness and reduce interviewer influence. Reported items were then classified into predefined food groups, as recommended by the FAO. Each food group consumed was assigned a score of 1, while groups not consumed received a score of 0. The DDS was calculated as the sum of these scores, with higher values indicating a more diverse diet. According to the FAO guidelines, the consumption of five or more food groups was considered indicative of adequate dietary diversity. Performance status was determined using the Eastern Cooperative Oncology Group (ECOG) Performance Status Scale, which ranges from 0 (fully active) to 4 (completely bedridden), providing insight into the patient’s functional capacity. Scores of 0–1 indicate good performance status, whereas scores of 2–4 indicate poor performance status [48]. Anthropometric measurements were collected by trained data collectors following standardized procedures. Weight was measured using a calibrated digital scale on a flat surface, with participants wearing light clothing and no shoes, and recorded to the nearest 0.1 kg. Height was measured using a wall-mounted stadiometer with participants standing in the Frankfort plane. BMI was later calculated from these measurements. Mid-upper arm circumference (MUAC), measured at the midpoint between the acromion and olecranon of the non-dominant arm. All measurements were performed twice, and the average value was recorded to ensure accuracy. Data quality assurance The questionnaire was originally prepared in English, then translated into Amharic and back-translated to maintain consistency in meaning. A pretest was carried out on 5% of the total sample (20 participants) at the same data collection sites to ensure contextual relevance and operational feasibility of the questionnaire. The pretest participants were excluded from the main study to avoid contamination. Revisions were made based on the pretest findings to enhance clarity and comprehension. Data collection was conducted by five trained BSc nurses under the supervision of two MPH professionals. Both data collectors and supervisors participated in a three-day training that covered study protocols, ethical considerations, and effective communication with participants. During the training, measurement standardization was performed to ensure accuracy and consistency of anthropometric measurements (weight, height, and MUAC) across data collectors. A standardized weighing scale was used for weight measurement and was adjusted to zero before each reading. Supervisors were responsible for ensuring adherence to procedures, overseeing real-time data collection, and addressing any challenges encountered during the process. The principal investigator reviewed all completed questionnaires daily to check for completeness, accuracy, and internal consistency. Data processing and analysis Data were initially coded and entered using KoboToolbox and then exported to SPSS version 26 for analysis. The dataset was checked for completeness, inconsistencies, and missing values, with any discrepancies addressed before analysis. Tumor size and tumor grade were missing in over 70% of participants; therefore, these variables were excluded from the primary multivariable analysis. To assess the potential impact of this missingness on model estimates, a sensitivity analysis was conducted by refitting the logistic regression model without these variables. The results remained consistent, with no significant change in model fit (Hosmer–Lemeshow test: p = 0.154 in the sensitivity model vs. 0.886 in the original model), indicating that the missing data did not materially affect the findings. All other analyses were performed using complete-case data for the respective variables. Descriptive statistics, including frequencies, percentages, and means with standard deviations (SD), were used to summarize socio-demographic and clinical variables. Nutritional status was categorized into three groups, and corresponding dummy variables were generated for analysis. Socioeconomic status was assessed using principal component analysis (PCA) based on household assets and housing conditions, following the approach commonly used in national demographic and health surveys. Initially, households were categorized into five wealth quintiles (from the poorest to the richest). For analysis and presentation purposes, these quintiles were subsequently regrouped into three categories: poor (the lowest two quintiles), medium (middle quintile), and rich (top two quintiles). Undernutrition was initially categorized into three GLIM-based severity levels (no undernutrition, moderate, severe). For the purpose of statistical analysis, these categories were dichotomized into undernourished versus not undernourished to allow binary logistic regression. To examine factors associated with undernutrition, binary logistic regression was employed. Variables with a p-value less than 0.25 in the bivariable analysis were included in the multivariable logistic regression model to control for potential confounders. This threshold was selected to avoid prematurely excluding variables that may become significant in the multivariable context, as recommended by Bendel and Afifi (1977) [49] and Hosmer and Lemeshow (2013) [50]. All variables included in the multivariable model were tested for multicollinearity. Collinearity diagnostics were assessed using the tolerance and variance inflation factor (VIF). All tolerance values were above 0.6, and all VIF values ranged from 1.038 to 1.603, which are well below the commonly accepted cutoff of 5, indicating no significant multicollinearity among the independent variables. Model fitness was assessed using the Hosmer–Lemeshow goodness-of-fit test, which was not significant (p = 0.886), indicating a good fit of the model to the data. Results were presented as adjusted odds ratios (AORs) with 95% confidence intervals (CIs). Result Socio-demographic characteristics A total of 393 participants took part in the study, with a response rate of 97.8%. The mean age was 47.24 ± 13.64 years (range 18–94 years). The majority of respondents were female, 268 (68.2%). About 201 (51.1%) of the respondents were urban residents. The majority, 298 (75.8%), were married. Regarding occupation, 185 (47.1%) were farmers, and 57 (14.5%) were merchants. Additionally, one-third of the participants (33.3%) belonged to the lowest wealth quintile (Table 2). Table 2Socio-demographic characteristics of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 393)VariablesCategoryFrequencyPercent (%)Age18–304411.231–408321.141–5012932.851–607920.1> 605814.8SexMale12531.8Female26868.2ResidencyRural20151.1Urban19248.9ReligionOrthodox34688.0Muslim399.9Protestant82.0Educational statusNo formal education22056.0Primary and secondary9925.2Preparatory school123.1College and above6215.8Marital statusMarried29875.8Single338.4Divorced215.3Widowed4110.4OccupationGovernment employee5514.0Merchant5714.5Farmer18547.1Non employed9624.4Family size< 533084.0>=56316.0HFIASFood secure21454.5Mildly food insecure246.1Moderate food insecure11629.5Severe food insecurity399.9Wealth IndexPoor13133.3Medium14035.6Rich12231.0 Socio-demographic characteristics of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 393) Clinical characteristics of participants Among the 393 cancer patients included in the study, 13.2% reported a family history of cancer. Digestive system cancers were the most prevalent (35.9%), followed by breast cancer (25.7%) and genitourinary cancers (20.1%). The majority (87.5%) had been diagnosed within the past year. Regarding cancer stage, 33.3% were at Stage II, 29.3% at Stage III, and 26.5% at Stage IV. Tumor size and grade data were missing in over 70% of cases; among reported values, tumors measuring 2–5 cm (15.5%) and Grade II tumors (13.0%) were most common. Palliative chemotherapy was the treatment intent for 63.1% of participants, while 36.9% were receiving treatment with curative intent. Consistently, 82.4% of patients were currently receiving chemotherapy, with 43.5% undergoing their second to fourth cycle. Additionally, 13.2% had comorbid chronic illnesses, and 19.6% had a poor performance status (score < 2) (Table 3). Table 3Clinical characteristics of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 393)VariableCategoryFrequencyPercent (%)Family historyNo34186.8Yes5213.2Type of cancerBreast cancer10125.7Digestive tract14135.9Genitourinary system7920.1Hematologic235.9Head & neck153.8Respiratory184.6Musculoskeletal82.0Neurologic41.0Skin41.0Duration of illness< 1 year34487.51–2 years379.42–3 years51.33–4 years61.55–6 years10.3> 6 years34487.5Stage of cancerStage I41.0Stage II13133.3Stage III11529.3Stage IV10426.5Not reported399.9Tumor sizeTx (Tumor not assessed)28171.5T1(< 2 cm)174.3T2(2–5 cm)6115.5T3(> 5 cm)225.6T4(extension to chest wall)123.1Tumor gradeGrade I41.0Grade II5113.0Grade III338.4Grade IV256.4Not reported28071.2Treatment intentPalliative24863.1Curative14536.9Comorbid diseaseNo34186.8Yes5213.2Current treatmentChemotherapy32482.4Anti-pain5313.5Chemotherapy and surgery10.3Hormonal therapy143.6Surgery10.3Cycle of treatmentFirst cycle4511.5Second -fourth cycles17143.5Fifth- sixth cycles6316.0Greater than six cycles5113.0Not specified6316.0Performance status< 27719.6≥ 231680.4 Clinical characteristics of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 393) Dietary habits of participants Table 4 summarizes key the dietary habits of the participants. The majority (68.7%) had a poor dietary diversity score, suggesting limited variety in food intake. Most participants (64.4%) reported eating three times a day, with smaller proportions eating either twice (24.2%) or four times daily (11.5%). Regarding feeding independence, 86.3% of participants were able to eat without any support, while 12.0% required partial assistance, and 1.8% were completely dependent on others for feeding. Table 4Dietary habits of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 393)VariableCategoryFrequencyPercent (%)DDSPoor27068.7Good12331.3Frequency of mealsTwice9524.2Three times25364.4Four times4511.5Eating independenceI cannot eat without support71.8Partially supported by others4712.0I can eat without any support33986.3 Dietary habits of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 393) Nutritional status Out of the 393 participants initially enrolled, 18 (4.6%) were overweight and 7 (1.8%) were obese; these individuals were excluded from GLIM-based analysis to focus on undernutrition. The final analytic sample consisted of 368 participants. Phenotypic assessment revealed a high burden of undernutrition: 38.9% of participants experienced unintentional weight loss exceeding 10% over six months, and 40.8% were severely underweight based on BMI. MUAC measurements indicated that 20.4% had moderate muscle depletion and 1.4% had severe depletion. Regarding etiologic criteria, 50% of participants reported loss of appetite, while vomiting (15.5%), pain during swallowing (9.8%), and comorbid diseases (13.3%) were also common (Table 5). Table 5Distribution of phenotypic and etiologic criteria based on GLIM framework and nutritional status among study participants (N = 368)GLIM componentCriteriaCategoryFrequency N (%)Phenotypic criteriaUnintentional weight loss over 6 months< 5%104(28.3)5–10%121(32.9)> 10%143(38.9)BMINormal142(38.6%)Moderate underweight76(20.7)Severe underweight150(40.8)MUACNormal288(78.3)Moderate malnutrition75(20.4)Severe malnutrition5(1.4)Etiologic criteriaChewing problemYes13 (3.5)No355(96.5)Pain during swallowingYes36(9.8)No332(90.2)VomitingYes57(15.5)No311(84.5)DiarrheaYes10(2.7%)No358(97.3)ConstipationYes26(7.1)No342(92.9)Loss of appetiteYes184(50)No184(50)Comorbid diseaseYes49(13.3)No319(86.7) Distribution of phenotypic and etiologic criteria based on GLIM framework and nutritional status among study participants (N = 368) Based on the requirement of at least one phenotypic and one etiologic criterion, 50.3% (95% CI: 44.4%–55.5%) of participants were classified as undernourished. Among these, 31.5% had severe malnutrition (Stage 2) and 18.8% had moderate malnutrition (Stage 1), while the remaining 49.7% had normal nutritional status Fig. 1). Nutritional status of adult cancer patients attending follow-up care at Public Referral Hospitals in Bahir Dar City, Northwest Ethiopia, 2024, based on GLIM criteria. Categories are classified according to the Global Leadership Initiative on Malnutrition (GLIM) criteria, assessing malnutrition severity as moderate or severe based on phenotypic and etiologic factors Disease–Treatment factors and undernutrition The prevalence of undernutrition varied significantly by type of cancer, with the highest rates observed in patients with musculoskeletal (75.0%) and digestive cancers (62.2%), and the lowest in neurologic cancer (25.0%). Advanced cancer stage was associated with increased undernutrition, rising from 25.0% in Stage 1 to 61.9% in Stage 4. Similarly, higher tumor grades (III and IV) showed greater undernutrition prevalence (56.3% and 60.9%, respectively). Tumor size also correlated with nutritional status; patients with larger tumors (T3 and T4) had higher rates of undernutrition (57.1% and 66.7%, respectively) compared to those with smaller tumors (T1: 13.3%). Regarding treatment, over half of the patients undergoing chemotherapy were undernourished (52.0%). Those receiving anti-pain medications had a slightly lower prevalence of undernutrition (44.0%), while patients on chemotherapy combined with surgery showed no undernutrition. Hormonal therapy patients had the lowest undernutrition rate among treatments (36.4%). Undernutrition tended to increase with the number of chemotherapy cycles, from 50.0% in the first cycle to 61.7% in those receiving more than six cycles (Table 6). Table 6Disease -and treatment-related factors and nutritional status of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 368)VariablesCategoryUnder nutritionNormalType of cancerBreast cancer28(32.2%)59(67.8%)Digestive84(62.2%)51(37.8%)Genitourinary37(48.7%)39 (51.3%)Hematologic12(52.2%)11 (47.8%)Head & neck5(38.5%)8(61.5%)Respiratory10(55.6%)8(44.4%)Musculoskeletal6(75.0%)2(25.0%)Neurologic1 (25.0%)3(75.0%)Skin2(50.0%)2(50.0%)Cancer stageStage11(25.0%)3(75.0%)Stage 245(37.2%)76 (62.8%)Stage 364(58.2%)46(41.8%)Stage 460 (61.9%)37(38.1%)Stage 515(41.7%)21(58.3%)Tumor gradeGrade I1(25.0%)3(75.0%)Grade II14(31.1%)31(68.9%)Grade III18(56.3%)14(43.8%)Grade IV14(60.9%)9(39.1%)Not reported138(52.3%)126(47.7%)Tumor sizeT12(13.3%)13(86.7%)T220(36.4%)35(63.6%)T312(57.1%)9(42.9%)T48(66.7%)4(33.3%)Tx (Tumor not assessed)143(54.0%)122(46.0%)Current treatmentChemotherapy159 (52.0%)147 (48.0%)Anti-pain22 (44.0%)28 (56.0%)Chemotherapy and surgery0 (0.0%)1 (100.0%)Hormonal therapy4 (36.4%)7 (63.6%)Cycle of treatmentFirst cycle22 (50.0%)22 (50.0%)Second – Fourth cycles78 (48.8%)82 (51.2%)Fifth – Sixth cycles34 (55.7%)27 (44.3%)Greater than six cycles29 (61.7%)18 (38.3%)Not specified22 (39.3%)34 (60.7%) Disease -and treatment-related factors and nutritional status of adult cancer patients on follow-up care at referral public hospitals in Bahir Dar city, Northwest Ethiopia, 2024 (n = 368) Factors associated with undernutrition among adult cancer patients In the bi-variable analysis, variables with p-values less than 0.25, including female sex, pain during swallowing, gastrointestinal (GI) cancer type, chronic comorbidities, loss of appetite, poor performance status, and tumor size, were included in the multivariable logistic regression model. After adjustment, four factors remained statistically significant predictors of malnutrition. Female patients were more than twice as likely to be malnourished compared to male patients (AOR = 2.17; 95% CI: 1.17–4.03). Patients with chronic comorbidities had nearly four times higher odds of malnutrition than those without comorbidities (AOR = 3.86; 95% CI: 1.67–8.94). Loss of appetite was the strongest predictor, with affected patients having over eighteen times higher odds of being malnourished than those without appetite loss (AOR = 18.21; 95% CI: 10.01–33.12). The odds of malnutrition increased with tumor size. Compared with T1 tumors (< 2 cm), T2 tumors (2–5 cm) had over eight times higher odds of malnutrition (AOR = 8.46; 95% CI: 1.17–16.18), T3 tumors (> 5 cm) had over fifteen times higher odds (AOR = 15.31; 95% CI: 1.77–32.11), T4 tumors (extension to chest wall) had nearly eighteen times higher odds (AOR = 17.98; 95% CI: 6.45–29.19), and tumors of unknown size (Tx) had nearly fourteen times higher odds (AOR = 13.93; 95% CI: 2.01–25.29). Other factors, including pain during swallowing, GI cancer type, and poor performance status, were not statistically significant in the adjusted model. Discussion This study revealed a high burden of malnutrition among oncology patients attending follow-up care in Northwest Ethiopia. Using the Global Leadership Initiative on Malnutrition (GLIM) criteria, 50.3% of participants were identified as malnourished (95% CI: 44.4%–55.5%). Of these, 31.5% were categorized as severely malnourished and 18.8% as moderately malnourished, while the remaining 49.7% had a normal nutritional status (Fig. 1). This is the first study in Northwest Ethiopia, and among the few in sub-Saharan Africa, to apply the GLIM criteria to assess cancer-related malnutrition, addressing a critical knowledge and practice gap and offering evidence that may inform local clinical nutrition protocols, early intervention strategies, and health policy. Multivariable analysis identified loss of appetite, chronic comorbidities, female sex, and larger tumor size as independent predictors of malnutrition. Among these, appetite loss was the strongest predictor, increasing the odds of malnutrition by more than eighteen times (Table 7). Table 7Bi-variable and multivariable logistic regression analysis of factors associated with malnutrition among adult cancer patients attending follow-up care at public referral Hospitals, Northwest Ethiopia (n = 368)Variables with categoryUnder Nutrition95% CI AOR95% CI AORP valueYesNoSex Male (Ref)77 (61.6%)48 (38.4%)11 Female108 (44.4%)135 (55.6%)1.72 (1.09–2.73)*2.17 (1.17–4.03)*0.014Pain during swallowing Yes27 (75.0%)9 (25.0%)2.73 (1.16–6.42)*2.47 (0.93–6.57)0.071 No (Ref)158 (47.6%)174 (52.4%)11Type of cancer GI cancer84 (62.2%)51 (37.8%)2.18 (1.37–3.45)*1.77 (0.98–3.19)0.059 Others (Ref)101 (43.3%)132 (56.7%)11Chronic comorbid Yes33 (67.3%)16 (32.7%)2.27 (1.43–6.52)**3.86 (1.67–8.94)*0.002 No (Ref)152 (47.6%)167 (52.4%)11Loss appetite Yes144 (78.3%)40 (21.7%)12.56 (7.67, 20.61)**18.21 (10.01–33.12)**< 0.001 No (Ref)41 (22.3%)143 (77.7%)11Performance status Poor46 (60.5%)30 (39.5%)1.69 (1.01–2.82)*1.38 (0.67–2.86)0.379 Good (Ref)139 (47.6%)153 (52.4%)11Tumor size T1 (< 2 cm) (Ref)2 (13.3%)13 (86.7%)11 T2 (2–5 cm)20 (36.4%)35 (63.6%)3.71 (0.76–18.15)*8.46 (1.17–16.18)*0.034 T3 (> 5 cm)12 (57.1%)9 (42.9%)8.67 (1.55–21.46)*15.31 (1.77–32.11)*0.013 T4 (extension to chest wall)8 (66.7%)4 (33.3%)13 (1.92–23.99)*17.98 (6.45–29.19)*0.001 Tx143 (54.0%)122 (46.0%)7.62 (1.69–24.42)*13.93 (2.01–25.29)*0.008Reference categories are explicitly labeled as "Ref"; AOR Adjusted Odds Ratio, CI Confidence Interval; *p < 0.05; *p < 0.001 Bi-variable and multivariable logistic regression analysis of factors associated with malnutrition among adult cancer patients attending follow-up care at public referral Hospitals, Northwest Ethiopia (n = 368) Reference categories are explicitly labeled as "Ref"; AOR Adjusted Odds Ratio, CI Confidence Interval; *p < 0.05; *p < 0.001 A comparable prevalence of malnutrition has been reported at Ujian Cancer Hospital, China (48.4%) [51]. Lower prevalence rates were observed at Daping Hospital, China (33.3%) [52], in Brazil (23%) [53], and in Australia (22.6%) [54], whereas higher prevalence was reported in Spain (80%) [55] and at the Cancer Hospital of Zhengzhou University, China (75.7%) [56]. Variations across studies may reflect differences in patient characteristics, including cancer type, disease stage, and treatment status, as well as clinical settings (inpatient versus outpatient), timing of nutritional assessment, and availability of nutritional support services [57–60]. Socioeconomic factors and regional healthcare infrastructure may further influence malnutrition prevalence among cancer patients. By applying the GLIM criteria, our study employed a standardized approach that combines phenotypic and etiologic indicators, allowing for meaningful comparisons with international studies. In Ethiopia, studies using other validated tools, such as PG-SGA, also indicate a substantial malnutrition burden: 48.1% in Hawassa [18], 61.4% in Jimma [17], and 58.2% in Addis Ababa [41], while lower prevalence of 12.2% was documented in Gondar [61]. The similarity or difference in prevalence compared to GLIM may be due to differences in tool sensitivity and diagnostic criteria. PG-SGA incorporates patient-reported symptoms and functional status, which may detect malnutrition differently than GLIM’s combined phenotypic and etiologic criteria. Therefore, results can be surprisingly similar in some populations where both muscle loss and intake reduction are pronounced, or differ where subtle nutritional deficits are only captured by one tool. Differences in sensitivity and diagnostic thresholds among these tools may contribute to variability in reported malnutrition prevalence. While direct sensitivity and specificity comparisons between GLIM, SGA, and BMI have not been conducted in this Ethiopian population, international studies indicate that GLIM offers moderate to strong diagnostic performance. A meta-analysis reported a pooled sensitivity of 0.72 (95% CI: 0.64–0.78) and specificity of 0.82 (95% CI: 0.72–0.88); using SGA as the reference further improved accuracy (sensitivity 0.81; specificity 0.80) [62]. When compared with SGA, GLIM tends to be more sensitive but sometimes less specific, as demonstrated in a prospective validation study where specificity was as low as 33% despite high sensitivity [63]. In ambulatory cancer patients, GLIM shows high specificity (98%) but moderate sensitivity (60%) when muscle mass is included, with performance decreasing if muscle mass is excluded [64]. Another study showed sensitivity around 61% and specificity around 90% compared to SGA [65]. These variations underscore the importance of context and available diagnostic parameters when interpreting GLIM’s performance. Loss of appetite emerged as the strongest independent predictor of malnutrition. This finding is consistent with previous studies in Hawassa, Addis Ababa, Malaysia, Spain, India, and Vietnam [18, 41, 66–69], all of which highlight anorexia as a common and significant contributor to cancer-related undernutrition. The underlying mechanisms are multifactorial and include tumor-induced metabolic alterations, systemic inflammation, and the release of pro-inflammatory cytokines (e.g., TNF-α, IL-1, IL-6), which disrupt hunger signaling, alter taste and smell, and reduce food intake [70, 71]. Cancer treatments such as chemotherapy and radiotherapy may further exacerbate appetite loss by inducing nausea, mucositis, or gastrointestinal disturbances [72]. The presence of chronic comorbidities was also significantly associated with malnutrition. Similar findings were reported from Turkey [73] and China [74]. Comorbid conditions, such as diabetes, cardiovascular disease, or chronic infections, may compound the nutritional burden through mechanisms like chronic inflammation, metabolic dysregulation, dietary limitations, or increased catabolic activity. Moreover, the added clinical complexity may divert focus from nutritional care, particularly in resource-constrained settings. Female sex was significantly associated with higher odds of under-nutrition, with women being more than twice as likely to be undernourished compared to men. This finding is consistent with some previous studies from Jordan [75] and Bangladesh [76] that reported a higher risk among female cancer patients, but it contrasts with others from Kenya [14, 15] and France [77], where male patients were found to be equally or more affected. Several factors may explain the higher risk of undernutrition among female patients. Biologically, women generally have lower baseline muscle mass and are more prone to sarcopenia and catabolic stress. Sociocultural roles, including caregiving responsibilities and prioritizing family needs, combined with limited financial resources and gender-based disparities in healthcare access, may reduce women’s ability to seek timely nutritional care. In food-insecure households, women may further restrict their own intake to preserve food for others. Together, these factors likely increase the nutritional vulnerability of female cancer patients in this setting. The odds of malnutrition increased with tumor size. Patients with larger tumors (T2–T4) and those with unknown tumor size (Tx) had progressively higher odds of being undernourished. Similar findings were reported from Kenya [15] and Japan [78]. Larger tumors are typically associated with greater metabolic demands and heightened systemic inflammatory responses. In addition, they may cause mechanical obstruction, pain, or other local effects that interfere with chewing, swallowing, or digestion. From a clinical standpoint, tumor size may serve as a simple but effective screening indicator to help prioritize nutritional assessment and early intervention, especially in settings where comprehensive screening tools are not routinely available. This study underscores the need for routine, standardized nutritional screening in oncology care in Ethiopia. The use of GLIM criteria allows for early detection and better risk stratification compared to BMI alone, which remains the most commonly used tool in the region. Given the identified risk factors, appetite loss, comorbidities, female sex, and tumor size, targeted interventions should be incorporated into cancer care guidelines, including nutrition counseling, supplementation, and interdisciplinary care. Policymakers and hospital administrators should prioritize capacity building for clinical nutrition services and consider validating GLIM tools for local settings. Strengths and limitations This study has several strengths. It applied the Global Leadership Initiative on Malnutrition (GLIM) criteria, which combine phenotypic and etiologic indicators, providing a comprehensive and standardized approach to diagnosing malnutrition. The relatively large sample size and inclusion of patients from one of the largest oncology centers in Northwest Ethiopia enhance the representativeness and generalizability of the findings in similar tertiary healthcare settings. Additionally, the analysis of tumor size and socioeconomic status (wealth index) adds value by identifying clinically and socially relevant predictors of malnutrition. Some limitations should be acknowledged. Muscle mass was assessed using mid-upper arm circumference (MUAC), with cut-offs adopted from a multicenter study of adult cancer patients outside Ethiopia. Although these thresholds were considered clinically relevant, their use in this population may introduce measurement bias or misclassification error, as they have not been formally validated in Ethiopian oncology patients. MUAC is also less precise than advanced techniques such as dual-energy X-ray absorptiometry (DEXA) or bioelectrical impedance analysis (BIA). Weight loss relied on patient self-report, potentially introducing measurement bias. The study population was restricted to hospital-based follow-up patients, which may not fully represent newly diagnosed or community-based cancer patients. GLIM criteria have not yet been formally validated in Ethiopia, and variations in healthcare infrastructure and socioeconomic factors may affect applicability. Missing data on tumor size and grade could have influenced the analysis, although sensitivity analyses indicated that model fit and conclusions remained robust. Despite these limitations, the study provides valuable insights into the prevalence and severity of malnutrition among cancer patients in Ethiopia and offers a foundation for improving nutritional assessment and interventions in resource-limited settings. Future research should focus on formally validating the GLIM criteria in diverse Ethiopian cancer populations and conducting implementation trials to evaluate the effectiveness and cost-benefit of integrating structured nutritional support into standard oncology care. Conclusion This study revealed a high prevalence of malnutrition among cancer patients attending follow-up care in public referral hospitals in Northwest Ethiopia, with over half of the participants affected. Using the GLIM criteria, four factors were independently associated with malnutrition: loss of appetite, chronic comorbidities, female sex, and increased tumor size, with appetite loss being the strongest predictor. These findings highlight the multifactorial nature of cancer-related malnutrition and the importance of early identification of at-risk patients. Incorporating routine nutritional screening using structured tools such as the GLIM criteria into standard oncology care may support timely detection and management of malnutrition. Expanding access to dietetic services and providing targeted nutritional support, particularly for women and patients with appetite loss, comorbidities, or larger tumors, could potentially improve clinical outcomes and quality of life. Further studies are needed to validate the GLIM criteria in the Ethiopian context and to evaluate the long-term effects of nutritional interventions in diverse cancer populations. Supplementary Information Supplementary Material 1 Acknowledgements We would like to thank Bahir Dar University, data collectors, supervisors, and study participants. Abbreviations Adjusted Odds Ratio Body Mass Index Dietary Diversity Ethiopia Demographic and Health Survey Eastern Cooperative Oncology Group European Society for Clinical Nutrition and Metabolism Food and Nutrition Technical Assistance Felege Hiwot Comprehensive Specialized Hospital Food and Agriculture organization Global Leadership Initiative on Malnutrition General Practitioner Household Food Insecurity Access Scale Interferon Gamma Interleukin Malnutrition Universal Screening Tool Mid-upper arm circumference Patient Generated Subjective Global Assessment Principal Component Analysis Performance Status Subjective Global Assessment Statistical Package for the Social Sciences Tibebe Ghion Comprehensive Specialized Hospital World Health Organization Authors’ contributions TM, SD and NF assisted in the conception of the study, developed the proposal, involved in the method development, analysis, interpretation of the findings and in development of manuscript. AA, HA and AN involved in the method development, interpretation of the findings and in development of manuscript. All the authors read, revised and approved the final manuscript. Funding The authors declare that there is no funding source. Data availability Data will be available upon request from the corresponding author. Declarations Ethics approval and consent to participate Ethical clearance was obtained from the Institutional Review Board of the College of Medicine and Health Sciences, Bahir Dar University. Official support letters were also obtained from each participating public referral hospital. Before data collection, all participants were informed about the purpose, procedures, and expected outcomes of the study. Written informed consent was obtained from participants with formal education. For participants without formal education, study information was read aloud in their local language. Verbal consent was obtained in the presence and with the assistance of caregivers, who helped explain and clarify the procedures and purpose. In cases where participants were unable to independently provide informed consent due to illiteracy or other limitations, consent was obtained from a legally authorized representative or appropriate caregiver on their behalf, in accordance with ethical approval. Participants who showed signs of undernutrition during data collection, such as low MUAC, low BMI, visible wasting, or reported poor food intake, were immediately referred to the attending clinicians for further evaluation and appropriate nutrition counseling and management, in accordance with each hospital’s clinical protocol. To ensure confidentiality, only coded identifiers (not personal names) were used on the questionnaires. Ethical principles, including voluntary participation, informed consent, and data privacy, were emphasized during the training of data collectors and supervisors. This study was conducted following the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. This research is original and has not been submitted or published elsewhere. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement Data will be available upon request from the corresponding author.