Study design

A randomised controlled clinical trial was developed in patients without diabetes mellitus admitted due to stroke and with an indication for nutritional support by nasogastric tube. We randomly assigned patients to receive either a diabetes-specific enteral formula or a standard formula.

The primary outcome of the study was the development hyperglycaemia associated with enteral nutrition formula during admission. The secondary outcomes were to observe the influence of enteral nutrition formula on comorbidity of stroke (oral route recovery, death, readmission and length of stay) and in complications of enteral nutrition formula.

This study was developed between January 2018 and September 2021 in patients admitted to the Stroke Unit, Department of Neurology of Clinic University Hospital of Valladolid. All patients signed informed consent before recruitment and randomisation.

The study was carried out in accordance with the Declaration of Helsinki and all procedures were approved by the Medical Research Ethics Committee (CEIm) of East Valladolid under code CASVE-NM-17-315. This clinical trial was registered in www.clinicaltrials.gov with code NCT03422900.

Study subjects

Patients with a first-ever ischaemic or haemorrhagic stroke, without history of diabetes mellitus, admitted to the Stroke Unit of our hospital were included in the study. The patients had complete dysphagia and required total enteral nutrition via a nasogastric tube (for at least 7 days). The inclusion criteria were the start of enteral nutrition during the first 72 h after stroke and acceptance by the patient or their legal representative to participate in the study. The main reason to select 72 h was that there is no evidence of the use of early enteral nutrition by nasogastric tube could ameliorate the outcomes and if the patient has no dysphagia could be dangerous [3]. We select 72 h to ensure the stability of patient and the protocol to confirm dysphagia in these patients in our hospital.

The exclusion criteria were enteral nutrition not feasible; severe gastrointestinal pathology; diagnosed diabetes mellitus; current glycated haemoglobin (HbA1c) >6.5%; allergy or intolerance to ingredients of the formula; pregnant women; patients who required an Intensive Care Unit (ICU) stay; previous neurodegenerative disease; and did not sign the informed consent.

Randomisation was done by a randomisation seed generated by http://www.randomization.com/ with a block assignation AABB, ABAB, ABBA, BAAB, BBAA and BABA.

The investigators were aware of the allocation sequence and treatment allocation. The trial personnel who worked in hospital plant only administrate the enteral nutrition prescribed by investigators. There was no blinding because it was a trial in real clinical practice with the formulas commonly used in the hospital. Nevertheless, the investigator who made the randomization is not involved in the clinical follow-up of the patient. This investigator informed the allocation of treatment to the physician of clinical nutrition services that prescribed the indicated treatment. This physician is not involved in data collection or analysis. At the end of the study the physician who had followed the patient in a multidisciplinary team delivered a report of variables to the investigator that includes it in the database. Another investigator took care of the data analysis. In this way, the investigator who collected and analyzed the variables was not the physician who followed the patient.

There was no predefined stopping rule. In the development of study, we have seen a better glycaemia evolution in patients with diabetes-specific formula. For this reason, we consider the primary outcome was solved and all patients could benefit from the use of diabetes-specific formula.

Medical nutrition therapy

A standard enteral nutrition formula (isocaloric, isoproteic formula without fibre) was compared with a diabetes-specific enteral formula (isocaloric and isoproteic formula with carbohydrates with a lower glycaemic index and the presence of mixed soluble and insoluble fibre) (Table 1).

Table 1 Distribution of calories, macronutrients and micronutrients of control and experimental formula.

Energy requirements for patients were estimated by Harris Benedict equation with a stress factor of 1.2. The amount of enteral nutrition was calculated in the function of energy density of the formula and energy requirements of the patient.

Enteral nutrition was delivered with a continuous infusion pump for 24 h without rest.

Study variables

Sociodemographic variables: Age (years), gender (male/female).

Clinical variables: Stroke type (ischaemic/haemorrhagic); stroke clinical severity (National Institutes of Health Stroke Scale (NIHSS)); Rankin scale.

Anthropometric Variables: Usual weight (kg); actual weight (kg); height (m); body mass index (kg/m2).

Metabolic variables:

Biochemical: performed with a Cobas c-711 autoanalyser (Roche Diagnostics): glucose (mg/dL); creatinine (mg/dl); albumin (g/dL); C-reactive protein (CRP) (mg/dL), prealbumin (mg/dL); CRP/prealbumin ratio; glycaemia (mg/dL); sodium (mEq/l); potassium (mEq/l); urea (mg/dL); total cholesterol (mg/dL); and triglycerides (mg/dL).

Diagnosis of hyperglycaemia: fasting plasma glucose >126 mg/dl before the beginning of enteral nutrition or glycaemia >180 mg/dl during the administration of the enteral formula.

Capillary blood glucose (Abbott® Lifestyle Measurer): this was considered the average of capillary glycaemia (mg/dl) every 8 h for 7 days, from the start of enteral nutrition until discharge from hospital.

Gastrointestinal Tolerance: diarrhoea, constipation, gastric emptying, abdominal distension, presence of vomiting. All of these variables were collected daily in the form of a dichotomous variable yes/no by the same investigator.

Complications: length of stay (days); mortality during admission; readmission; presence of pneumonia; dysphagia recovery (negative method of clinical volume-viscosity test (MECV_V)) [19].

Statistical analysis

The data were stored in a database on the statistical package SPSS 23.0 (SPSS Inc. Il, USA) with an official license from the University of Valladolid. An analysis of normality of the continuous variables was performed with the Kolmogorov-Smirnov test.LIe.

The calculated sample size was 150 patients (75 per branch with 10% loss). The study had to be stopped at 60 patients because the primary aim was achieved, and we could not continue due to ethical considerations, these considerations were to obtain the main outcome in patients with diabetes-specific formula, so all the patients could benefit from the use of diabetes-specific formula.

Continuous variables were expressed as mean (standard deviation), parametric variables were analysed with the unpaired and paired t-Student test, and non-parametric variables with the Friedman, Wilcoxon, K Kruskal, and U-Mann Whitney tests. If it was necessary to compare variables in more than two groups, the ANOVA U test (with the Bonferroni post-hoc test) was used. The analysis of variables at the different times of the study was performed using multivariable analysis of variance (MANOVA).

A multivariable analysis was performed adjusted by variables which show differences between groups in a descriptive analysis and with variables which can influence glycaemia.

Qualitative variables were expressed as percentages (%) and analysed using the Chi-square test (with Fisher and Yates corrections when necessary). A p-value < 0.05 was considered significant.