At the First People’s Hospital of Yunnan Province, a randomized, open-label, inside-subject crossover trial was conducted. Before enrolling in the trial, each patient aged 18 years or older submitted written informed consent. The study was registered with the Chinese Clinical Trial Registry (ChiCTR2100049763) and endorsed by the First People’s Hospital Medical Ethics Committee of Yunnan Province.

Patients between 18 and 45 who had been diagnosed with T1D for more than a year and received multiple insulin injection (MDI) therapy were included. Furthermore, standard body mass index (BMI), normal thyroid function, and recent hemoglobin A1c (HbA1c) levels ranged from 6.5 to 9% before recruitment was needed.

The exclusion criteria were as follows: (i) any comorbid illnesses, such as coeliac disease and autoimmune thyroid disease; (ii) fasting blood glucose on trial day greater than 10 mmol/L or lower than 3.9 mmol/L; (iii) using corticosteroid or other drugs that might impair gastric emptying; (v) having any dietary restrictions, such as food allergies; (vi) having ketoacidosis within 24 h of consuming the test meals.

The FreeStyle Libre Flash Glucose Monitoring System (FGM, Abbott) was implanted in the subcutaneous tissue of the abdomen area or upper buttocks on the first trial day. The insertion was placed one day before the first meal to avoid bias. All patients were educated and trained on how to handle their FGMS daily.

Patients randomly consumed four test meals (NPM-a, NPM-b, HPFM-a, and HPFM-b) on four different occasions, each separated by three days. NPM-a was a normal protein meal based on CC; NPM-b was a normal protein meal based on the modified FPU method; HPFM-a was a high protein-fat meal based on CC; and HPFM-b was a high protein-fat meal based on the modified FPU method. Each meal was prepared at 07:00 AM and made in the hospital kitchen.

Because the influence of the first meal on glucose levels lasts longer depending on the content of the meal, the breakfast meal was chosen as the test meal to eliminate any confounding second-meal effect. Participants were advised to refrain from vigorous exercise and high-fat, high-protein meals the day before the test meal and to fast for 10 h before the test meal. Controlled circumstances were used throughout the trial, and glycemic response factors were minimized. Each meal was consumed within thirty minutes; no food or drink was permitted throughout the 5-hour postprandial period unless symptomatic hypoglycemia occurred.

NPM contains the following ingredients: [milk (250 mL), egg (50 g), beef (50 g), whole wheat bread (75 g); 53 g carbohydrate, 32 g protein, 17 g fat], HPFM contains the following ingredients: [milk (250 mL), egg (50 g), beef (150 g), whole wheat bread (37.5 g);35 g of carbohydrate, 49 g of protein, 18.5 g of fat]. The insulin dosages for NPM-a and HPFM-a were estimated based on the carbohydrate content of the meals; for NPM-b and HPFM-b, the modified FPU was used to determine the insulin dose. The composition of NPM and HPFM meals and the total insulin dosage for each meal are detailed in Table 1.

Before the test meal, the FGMS was tested for proper performance and adherence to the study protocol. For patients receiving MDI therapy, a short-acting insulin bolus was injected subcutaneously at the start of each meal.

All patients utilized the insulin–carbohydrate ratio for mealtime boluses for CC meals (NPM-a, HPFM-a). The current ICR was determined by dividing the total daily insulin dosage by 500 and remained constant throughout the control and test meals. The CC algorithms did not account for fat or protein.

The insulin-to-fat-protein ratio and ICR were employed to administer mealtime boluses for the modified FPU counting meal (NPM-b, HPFM-b). The modified FPU is defined as one FPU was 200 kcal of fat or protein that requires the same quantity of insulin as 10 g of carbohydrates. The mealtime insulin dose was determined and delivered depending on the meal’s carbohydrate, lipid, and protein content. For the NPFM meal, the protein and fat content was one FPU, and HPFM was two FPU.

The test was terminated when hypoglycemia was detected using capillary blood glucose and was repeated the next day. Patients suffering hypoglycemia (glucose levels less than 3.9 mmol/L) were instructed to take juice containing 15 g of carbohydrates. During the research, no patient experienced severe hypoglycemia.

FGM was used to monitor interstitial fluid glucose levels, and only the 5-hour postprandial period FGMS data were utilized for analysis. Aside from FGMS data, capillary blood glucose levels were assessed using the Abbott blood glucose monitoring system at the beginning of the meals (T = 0), 120 min after meals, and when symptomatic hypoglycemia occurred.

FGMS measurements yielded the following outcome parameters: (1) mean glucose levels, which were recorded every 15 min with the FGMS; (2) peak glucose level, which was the highest level recorded during the 5-h postprandial period; the time of its occurrence was used to determine the time to peak glucose; (3) incremental area under the glucose excursion curve, which was determined as the area under the glucose curve during the 5-hour postprandial period with the glucose level at T = 0 as the baseline; (4) hypoglycemic episodes, which were defined as glucose levels less than 3.9 mmol/L measured by FGMS, at which time the onset of hypoglycemia was recorded; (5) time above range (> 10 mmol/L) during the 5-hour postprandial period; and (6) glucose excursions, which were defined as variations in glucose levels measured every thirty minutes.

The study size was determined by a prospective clinically significant difference in mean glucose level of 2.5 mmol/L and a 2 mmol/L within-subject SD in glucose levels when using FGMS. The predicted study size was 16 based on a power of 80% and a two-sided significance level of 5%.

Baseline data for categorical variables are presented as counts and percentages, while the mean and SD are used to represent continuous variables with normal distribution and the median and interquartile range (IQR) are used for nonnormal continuous variables.

SPSS (25.0, IBM Corp., Armonk, NY, USA) was used to perform all statistical analyses. The Mann–Whitney U test or t test was used to compare continuous variables. One-way repeated-measures analysis of variance was used to compare test meals. Generalized linear mixed models accounted for repeated measurements within the same individual, such as glucose levels, excursions, and time above range.