The Malnutrition in Polytrauma Patients (MaPP) study is an observational prospective cohort study that was performed on 100 adult severely injured patients at five Level-1 trauma centers, three in the United States (Massachusetts General Hospital and Brigham and Women’s Hospital at Boston, and Ryder Trauma Center in Miami) and two in the Netherlands (Leiden University Medical Center at Leiden and Haaglanden Medical Center Westeinde at The Hague). The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the local Institutional Review Boards (protocol number Netherlands: NL64016.058.17, approved on February 21, 2018; protocol number USA: 2018P000202/PHS, approved on April 3, 2018). This study is reported in line with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement [12]. The study methods are described in detail in the published study protocol [13].

All consecutive adult (≥ 18 years) patients with severe injuries (defined as Injury Severity Score, ISS ≥ 16) caused by blunt trauma, admitted to the ICU of one of the participating centers, were eligible for inclusion. Patients needed to be admitted to the ICU for more than 48 h and were not primarily managed in another hospital. Patients with burn wounds and penetrating injuries were excluded.

Trauma patients newly admitted to the ICU were screened for inclusion criteria upon admission by the investigators at the participating hospitals between July 2018 and April 2022. Eligible patients were asked to provide written informed consent for participation in the study. If the patient was unable to provide consent (e.g., due to unconsciousness), a legal representative was asked to provide informed consent. If a legal representative gave consent, and the patient became able to provide consent later in the study, they were asked to confirm it themselves. In cases where the patient did not have a legal representative, data was collected prospectively, and the patient was asked for consent once they could do so. If the patient declined to participate in the study, their data was removed from the electronic database. The patient and/or their legal representative could withdraw consent and exit the study at any time.

As described in the study protocol, the a priori sample size calculation showed that 195 patients were needed to answer the primary question of the MaPP study [13]. Due to the low inclusion rate during the COVID-19 pandemic, it was decided to prematurely end the inclusion at 100 patients.

Our exposure of interest was high nutritional risk defined by modified Nutrition Risk in the Critically Ill (mNUTRIC) ≥ 5 [8]. Our comparator was low nutritional risk defined by mNUTRIC < 5. The mNUTRIC score was determined by trained personnel within 24 h after ICU admission. This score is based on five items: age, Acute Physiology and Chronic Health Evaluation II (APACHE II) score [14], Sequential Organ Failure Assessment (SOFA) score [15], the number of comorbidities, and number of days in-hospital prior to ICU admission. The APACHE II score measures ICU mortality based on a number of laboratory values and patient signs. The SOFA score uses measurements of major organ function to determine the degree of organ failure. The mNUTRIC, APACHE II, and SOFA scores are listed in the Appendix.

The primary outcome was malnutrition, defined as a Subjective Global Assessment (SGA) score ≤ 5. The SGA score was assessed at ICU admission, every five days during ICU stay, at ICU discharge, weekly during admission to the ward, and at hospital discharge [16]. The SGA is a nutritional assessment tool that has been validated for the acute hospital setting, surgical patients, and patients admitted to the ICU requiring mechanical ventilation [17,18,19]. The SGA score is shown in the Appendix.

Secondary outcomes were complications, including systemic complications (sepsis, Acute Respiratory Distress Syndrome (ARDS), Systemic Inflammatory Response Syndrome (SIRS), multiple-organ failure), pneumonia, urinary tract infection (UTI), deep venous thrombosis (DVT), and pulmonary embolism (PE). Mortality was analyzed as a separate outcome parameter. This was also described in the study protocol [13].

Patient demographics, including age, sex, and body mass index (BMI), were recorded, along with trauma characteristics such as the Abbreviated Injury Scale (AIS) for all body regions and the ISS. Information on nutritional support was collected, and patients were categorized based on whether they received oral feeding or (par)enteral feeding. For the patients who received (par)enteral nutrition, the timing of its administration was documented, and whether it was initiated within 48 h or after 48 h of admission. Target energy goals were calculated through a weight-based predictive Eq. (25 kcal/kg/day). In overweight patients (BMI > 25 kg/m2), the adjusted body weight was used, which is calculated through the ideal body weight. The ideal body weight is calculated by the following equation: 0.9 × height in cm—100 (male) (or—106 (female)). To account for the metabolic demand of adipose tissue and muscle, an additional 25% of the excess weight (actual body weight minus ideal body weight) is added to the ideal body weight to calculate the adjusted weight [20]. According to the ESPEN guidelines, target energy goals should be met after 3–7 days of admission. It was documented whether goals were met after < 48 h, 3–7 days, and after > 7 days of admission. Surgical procedures that required patients to go to the operating room were documented. Other in-hospital outcomes included hospital length of stay (LOS), ICU LOS, and ventilator days.

All analyses were performed using IBM SPSS Statistics for Windows, version 25 (IBM Corp., Armonk, N.Y., USA) and R version 4.2.2. P-values < 0.05 were considered statistically significant. The baseline characteristics of the patients with low and high nutritional risk were compared using the Chi-square test or Fisher’s exact test (in case of expected cell counts < 5) for categorical variables, the independent samples T-test for normally distributed continuous variables, and the Mann–Whitney U test for skewed continuous variables.

The prevalence of high nutritional risk was calculated as the proportion with a 95% confidence interval (CI) of patients with a mNUTRIC score ≥ 5. The malnutrition rate was calculated as the proportion of patients well-nourished at admission who developed malnutrition during admission as diagnosed with the SGA. The patients who were already malnourished at admission were excluded from this analysis. The incidences of malnutrition, complications, and mortality during ICU and total hospital stay were compared between the patients with high and low nutritional risk using the Chi-square test. Furthermore, a cause-specific Cox regression model was fitted to analyze whether high nutritional risk was related to developing malnutrition during hospital admission. In this model, receiving (par)enteral feeding was added as a binary time-dependent covariate.