Recent and current low food intake – prevalence and associated factors in hospital patients from different medical specialities

Malnutrition adversely affects organ function, outcomes, length of stay, and costs in hospital patients. Low food intake is a major contributor to malnutrition and has been linked to increased short-term mortality. Multiple disease-, hospital-, and patient-related factors can reduce intake. Prior work showed that reduced intake before admission is associated with reduced intake in hospital, but no data had described patients with both recent and current low intake (LIRC), which may better capture sustained inadequate intake and risk for malnutrition than a single time-point measure. Malnutrition prevalence and risk vary across medical specialities, and little contemporary evidence exists for German hospitals. This study aims to quantify the prevalence of LIRC and to identify factors associated with LIRC overall and within specific medical disciplines in German hospitals using nutritionDay data.

Previous studies reported speciality-specific differences in malnutrition risk. Using the Malnutrition Screening Tool, oncological, long-term care, and infectious disease units showed higher nutritional risk. Subjective Global Assessment identified higher malnutrition risk in oncological and gastroenterological patients. Surgical patients more often reported reduced intake in the week prior to nutritionDay. Global nutritionDay analyses showed about half of hospitalized patients had low intake at a single hospital meal and highlighted regional variability. In Germany, a 2006 multicentre study found 27% malnutrition by SGA, ranging from below 10% in Gynaecology to over 50% in Geriatrics, but contemporary German data were lacking. The literature also links weight loss, poor health status, and certain diseases (e.g., digestive, cancer) to malnutrition or reduced intake.

Design and data source: Cross-sectional analysis of adult patients (≥18 years) from German hospital units participating in nutritionDay surveys 2016–2019 and three additional units in August 2020. nutritionDay is an annual, one-day, standardized international survey with hospital, unit, and patient questionnaires; 30-day outcomes are collected. Ethical approvals were in place in Vienna and Erlangen-Nürnberg. Inclusion/exclusion: Included adult patients from German units not using the hospital express survey. Exclusions: lack of consent, missing sex or speciality, and units with <75% 30-day outcome completeness. Final sample: 1865 patients from 127 units in 44 hospitals. Variables: Outcome LIRC defined as reduced intake both on nutritionDay and during the week before admission, based on patient questionnaire. NutritionDay lunch intake categorized as normal (“about all”) vs reduced (“1/2,” “1/4,” “nothing”). Pre-admission intake categorized as adequate (“more than normal,” “normal”) vs reduced (“3/4 of normal,” “about half,” “about a quarter to nearly nothing”). If patients could not self-complete, staff assisted. Independent variables: Speciality groups (Internal Medicine; Gastroenterology incl. Hepatology; Geriatrics; Oncology incl. Radiotherapy; Surgery [General, Cardiac/Vascular/Thoracic, Orthopaedic, Neurosurgery, Trauma]; Neurology; Others). From staff questionnaire: age (≥70 vs <70), sex, BMI (<20, 20–30, >30 kg/m²), admission type (emergency, planned, unknown), length of stay before nutritionDay (≤4 vs >4 days), number of medications (≤5 vs >5), previous surgery (yes/no), prior ICU admission (yes/no), terminally ill (yes/no/unknown), admission diagnoses (digestive disease; endocrine, nutritional and metabolic disease), comorbidities (cancer, dementia, cardiac insufficiency, infection, chronic liver, lung, kidney disease). From patient questionnaire: weight loss in prior 3 months (yes, no, unknown), ability to walk without assistance (yes/no), self-rated health (fair or better vs poor/very poor). Statistical analysis: Descriptive statistics with frequencies/percentages and means±SD. Univariate binary logistic regressions (LIRC as dependent variable) in the total sample, with continuous predictors categorized. All univariate models not inherently containing age or sex were adjusted for both. Patients with missing LIRC were excluded. For independent variables with <10% missing, cases with missing values were excluded; for ≥10% missing, a separate missing category was created. Variables with p<0.1 in univariate analyses were entered into a multivariate binary logistic regression in the total sample. The multivariate model was then run separately within each speciality. Odds ratios (ORs) with 95% CIs were reported. Model fit was assessed by omnibus chi-square test; effect sizes by Cohen’s f².

- Sample: 1865 patients; mean age 67±17 years; specialities included Gastroenterology (23.2%), Surgery (17.3%), Oncology (15.0%), Neurology (13.0%), Internal Medicine (11.4%), Geriatrics (9.2%), Others (11.0%). Half (49.9%) reported weight loss in prior 3 months; 46.1% were emergency admissions. - Reduced intake prevalence: Before admission, 31.0% reported reduced intake (lowest Neurology 20.2%, highest Gastroenterology 37.3%). On nutritionDay, 49.5% ate half or less of the served meal (highest Gastroenterology 56.9%; lowest Neurology with most full meals 58.7%). - LIRC prevalence: 21.1% overall; highest Gastroenterology 26.6%; lowest Neurology 11.2%. In contrast, 27.7% had low intake only on nutritionDay, and 8.2% only before admission. Among those with reduced intake before admission (n=578), 68.0% also had reduced intake on nutritionDay; among those with reduced intake on nutritionDay (n=924), 42.5% also reported reduced pre-admission intake. - Multivariate associations in total sample (N=1410): Female sex OR 1.98 (95% CI 1.50–2.61); weight loss in prior 3 months OR 2.62 (1.93–3.56); unknown weight loss OR 1.90 (1.03–3.51); poor/very poor self-rated health OR 2.17 (1.62–2.91); inability to walk without assistance OR 1.55 (1.14–2.12); digestive disease OR 1.90 (1.40–2.56); cardiac insufficiency OR 0.55 (0.37–0.83). Admission type “I do not know/missing” OR 1.38 (1.02–1.86); emergency vs planned not significant OR 1.41 (0.87–2.29). Medical speciality (vs Internal Medicine): Others OR 0.51 (0.28–0.92); other speciality comparisons not significant in the total model. Model p<0.001; Cohen’s f²=0.230. - Speciality-specific multivariate models: Gastroenterology: female OR 1.88 (1.08–3.29), weight loss OR 4.41 (2.29–8.52), poor health OR 2.23 (1.25–3.97), inability to walk OR 2.17 (1.16–4.05). Oncology: female OR 3.77 (1.71–8.30), weight loss OR 3.35 (1.45–7.73), poor health OR 6.76 (3.09–14.8); admission type unknown OR 5.76 (1.29–25.7). Surgery: weight loss OR 3.78 (1.68–8.51), digestive disease OR 3.43 (1.64–7.17), inability to walk OR 2.33 (1.06–5.10). Internal Medicine: inability to walk OR 3.11 (1.23–7.87), cardiac insufficiency OR 0.16 (0.05–0.55). Others: digestive disease OR 3.85 (1.41–10.5). Neurology and Geriatrics: no significant associations; overall models non-significant or low effect size. - Additional descriptive insight: Among patients with LIRC, 50.9% reported further decrease in intake during hospital stay up to nutritionDay; 22.9% remained constantly reduced; 9.9% increased but still did not finish the meal on nutritionDay.

The study addressed whether sustained low food intake (both before admission and on nutritionDay) is prevalent in German hospitals and which factors are associated, overall and by speciality. LIRC was common (one in five patients), indicating a prolonged period of inadequate intake likely contributing to malnutrition risk. The findings support screening at admission for recent intake to identify patients needing nutritional support early. Associations with female sex, recent weight loss, poor self-rated health, impaired mobility, and digestive disease are consistent with mechanisms of disease-related anorexia, symptom burden, and functional decline influencing intake. The protective association with cardiac insufficiency may relate to different BMI distributions or unmeasured disease severity; stage data were unavailable. Speciality-specific analyses revealed stronger associations in Gastroenterology and Oncology, emphasizing the need for discipline-tailored strategies. Neurology showed the lowest LIRC prevalence and no significant associations, suggesting lower relevance or limited power for this speciality; in Geriatrics, missing data impaired detection of associations despite moderate prevalence. Overall, the results reinforce the role of sustained low intake as a meaningful marker linked to malnutrition risk and potential adverse outcomes and highlight where targeted interventions may be most impactful.

LIRC affected approximately 21% of German hospital patients, with higher prevalence in Gastroenterology and lower in Neurology. Female sex, recent weight loss, poor subjective health, impaired mobility, and digestive disease were associated with LIRC. These results support routine assessment of pre-hospital food intake at admission and continuous monitoring during hospitalization, particularly in gastroenterology and oncology, to enable timely nutritional interventions and malnutrition assessment. Future work should evaluate whether nutritional therapy tailored to high-risk subgroups improves outcomes, extend analyses to international cohorts, incorporate dynamic intake trajectories, and relate LIRC to clinical outcomes compared with current low intake only.

- Self-reported intake introduces perception and recall bias; retrospective assessment of pre-admission intake may be inaccurate. - Cross-sectional design with length bias (patients with longer stays more likely included). - Potential selection bias: overrepresentation of older patients and of digestive and endocrine/metabolic diseases; possible higher participation from nutrition-focused units; severely ill patients less represented. - Missing data reduced effective sample sizes for multivariate analyses (notably in Geriatrics with high missingness in comorbidities), limiting power to detect associations. - Country-specific focus improved system homogeneity but reduced sample size and generalizability beyond Germany. - Disease severity (e.g., cardiac insufficiency stage) was unavailable, limiting interpretation of some associations.