Insulin Injection Behaviors and Association with Glycemic Outcomes in Patients with Diabetes: A Real-World Survey in Spain Authors: Alba Galdón Sanz-Pastor, Alicia Justel Enríquez, Ana Sánchez Bao, Rosa P. Quílez Toboso, Beatriz González Aguilera, Mercedes Noval Font, Jessica Ares Blanco, Inés Modrego Pardo, Beatriz Voltas Arribas, Francisco Javier Ampudia-Blasco Published: 2026 ABSTRACT IntroductionOptimal glycemic control is essential to prevent complications in diabetes, and insulin administration practices play a key role. This study assessed insulin injection behaviors and related glycemic outcomes among individuals using multiple daily injections in Spain.MethodsA cross-sectional online survey was conducted via an independent patient platform (Canal Diabetes). Adults (18–65 years) with type 1, type 2, or other forms of diabetes (e.g., maturity-onset diabetes of the young (MODY) or secondary diabetes) requiring daily insulin injections were included. Demographic, clinical, and behavioral data were analyzed using SPSS Statistics, version 23 (IBM Corp., Armonk, NY, USA).ResultsAmong 288 participants (mean age: 43.7 ± 14.7 years; 28.8% male, 71.2% female), 74.7% had type 1 diabetes. Most were white (95.8%) with a mean diabetes duration of 17.8 ± 14.5 years. While 83.3% injected prandial insulin before meals, only 71.2% did so at least 15 min before eating. Errors were frequent: 38.5% reported underdosing and 26.7% overdosing more than five times in the past month. Participants injecting before meals had significantly better outcomes than those injecting at other times, including higher rates of glycated hemoglobin (HbA1c) ≤ 7.5% (83.3% versus 62.5%, p = 0.001), greater time in range (TIR) (74.4% versus 66.3%, p = 0.032), and lower mean glucose (144 versus 159 mg/dL, p = 0.019). Smart pen users (41.2%) reported fewer mild-to-moderate hypoglycemic events (p = 0.021), though no differences were observed in HbA1c or TIR.ConclusionInsulin injection practices in Spain reveal significant variability and opportunities for improvement. Promoting timely premeal insulin administration and leveraging technologies such as smart pens may improve glycemic outcomes.Trial registrationClinicalTrials.gov identifier, NCT06421467.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s13300-026-01897-x. Key Summary Points Why carry out this study?Insulin injection practices including prandial timing, dose accuracy, and adherence are critical determinants of glycemic control in people with diabetes on multiple daily injections, yet real-world data on these behaviors and their clinical impact in Spain are lacking.This study aimed to describe insulin administration practices among adults with diabetes on multiple daily injections in Spain and to explore their association with glycemic outcomes.What was learned from this study?Prandial insulin timing before meals was significantly associated with better glycemic control, including higher rates of glycated hemoglobin ≤7.5% (83.3% versus 62.5%, p = 0.001), greater time in range (74.4% versus 66.3%, p = 0.032), and lower mean glucose (144 versus 159 mg/dL, p = 0.019), while dosing errors and prandial omissions were also highly prevalent and associated with worse outcomes.Smart pen use was associated with fewer mild-to-moderate hypoglycemic events, supporting their potential role in reducing hypoglycemia risk in people with diabetes on multiple daily injections.Premeal insulin timing, carbohydrate counting, and insulin-to-carbohydrate ratio use are modifiable behaviors associated with better glycemic control, suggesting that structured therapeutic education targeting these practices could meaningfully improve outcomes without changes in pharmacological treatment. Why carry out this study?Insulin injection practices including prandial timing, dose accuracy, and adherence are critical determinants of glycemic control in people with diabetes on multiple daily injections, yet real-world data on these behaviors and their clinical impact in Spain are lacking.This study aimed to describe insulin administration practices among adults with diabetes on multiple daily injections in Spain and to explore their association with glycemic outcomes.What was learned from this study?Prandial insulin timing before meals was significantly associated with better glycemic control, including higher rates of glycated hemoglobin ≤7.5% (83.3% versus 62.5%, p = 0.001), greater time in range (74.4% versus 66.3%, p = 0.032), and lower mean glucose (144 versus 159 mg/dL, p = 0.019), while dosing errors and prandial omissions were also highly prevalent and associated with worse outcomes.Smart pen use was associated with fewer mild-to-moderate hypoglycemic events, supporting their potential role in reducing hypoglycemia risk in people with diabetes on multiple daily injections.Premeal insulin timing, carbohydrate counting, and insulin-to-carbohydrate ratio use are modifiable behaviors associated with better glycemic control, suggesting that structured therapeutic education targeting these practices could meaningfully improve outcomes without changes in pharmacological treatment. Introduction Achieving and maintaining individualized glycemic control is critical to preventing chronic complications in individuals with type 1 diabetes (T1D) and type 2 diabetes (T2D). Traditional diabetes management strategies emphasize the adjustment of insulin doses on the basis of fasting and premeal blood glucose levels. However, postprandial hyperglycemia has emerged as a crucial target for optimizing glycemic control and minimizing complications [1–3]. Fast-acting insulin analogs (FAIAs) have been developed to mimic physiological insulin secretion in response to meals, playing a pivotal role in managing postprandial glucose levels. Despite their benefits, first-generation FAIAs, including insulin lispro, aspart, and glulisine, have limitations in replicating the rapid onset and precise duration of endogenous insulin secretion. These constraints require careful timing of prandial insulin administration, ideally 10–15 min before meals. When injected 15 min prior to eating, these insulins typically reach their peak effect around the time blood glucose levels begin to rise, which occurs shortly after food consumption. If insulin is injected during or after meals, the peak action may not coincide with the peak glucose levels, leading to inadequate insulin action and resulting in elevated postprandial blood glucose levels [4]. Second-generation FAIAs, such as faster aspart and ultrarapid lispro, address these limitations by offering quicker onset and shorter duration of action, enhancing postprandial glycemic control. Evidence supports their efficacy and safety, especially when used shortly before the start of meals [5, 6]. However, in real-world settings, adherence to this recommendation remains suboptimal, with many individuals delaying or skipping doses altogether [7, 8]. Optimal management of the postprandial period continues to be influenced by various factors, including timing, dose accuracy, and patient adherence [9–11]. Moreover, injecting insulin at inappropriate times can lead to glycemic variability, which has been associated with adverse health outcomes, including an increased risk of cardiovascular complications [12]. In addition, studies have shown that patients who do not adhere to the recommended injection timing may experience greater fluctuations in their blood glucose levels, complicating diabetes management and increasing the risk of both hyperglycemia and hypoglycemia [13, 14]. For instance, injecting after meals can result in a delayed insulin response, causing blood glucose levels to remain elevated for longer periods, while injecting during meals may lead to insufficient insulin action when it is most needed. Technological advancements, particularly the advent of smart insulin pens, hold promise for addressing these challenges. These devices, equipped with dose-tracking and reminder functions, improve adherence and allow for more precise data collection. Integration with continuous glucose monitoring (CGM) systems offers a comprehensive approach to diabetes management, identifying barriers to optimal glycemic control and guiding individualized treatment plans [15–17]. In this context of evolving pharmacological and technological innovations, understanding the patterns of insulin administration in real-world settings is essential, particularly given that behaviors such as injection timing and dose accuracy are potentially modifiable through structured therapeutic education. This study aimed to characterize real-world insulin administration practices among adults with diabetes on multiple daily injections (MDI) in Spain, and to assess their association with glycemic outcomes. Methods Study Design This was a cross-sectional, observational study designed to evaluate insulin administration practices among individuals with type 1 diabetes (T1D), type 2 diabetes (T2D), or other forms of diabetes in Spain (ClinicalTrials.gov, ID NCT06421467). The study was conducted through an online survey, ensuring a broad reach and accessibility for participants from diverse backgrounds. By capturing real-world practices, the study aimed to provide insights into the variability and challenges of insulin administration in this population. Study Population Participants were adults aged 18–65 years who were diagnosed with T1D, T2D, or other forms of diabetes and currently using insulin therapy involving at least one daily injection of basal and prandial insulin. Other forms of diabetes referred to less common types not classified as T1D or T2D, such as MODY or secondary diabetes (e.g., resulting from chronic pancreatitis), as reported by participants in the self-administered questionnaire; no further clinical specification was available. The study excluded individuals using advanced insulin delivery technologies such as continuous subcutaneous insulin infusion (CSII), sensor-augmented pump therapy, or automated insulin delivery (AID) systems. These exclusions ensured a focus on individuals relying on traditional injection-based insulin therapy, providing a clearer understanding of practices in this group. Recruitment and Data Collection Recruitment was conducted via the Canal Diabetes platform (https://www.canaldiabetes.com), a recognized diabetes information and outreach resource in Spain. The survey was available online for a 3-month period, from April to June 2024, to maximize participation. Potential participants were informed about the voluntary nature of the study and the confidentiality of their responses. Participants who met the inclusion criteria provided informed consent electronically before accessing the survey. The survey was designed to be user-friendly, allowing participants to complete it at their convenience. Data were collected anonymously and securely stored for subsequent analysis. The questionnaire captured a comprehensive range of data, including demographic, clinical, behavioral, and treatment-related variables. To ensure transparency and reproducibility, the exact wording of all survey questions, translated into English, is provided in Supplementary Appendix 1. Study Variables The survey captured a variety of variables relevant to insulin administration practices and diabetes management:Sociodemographic and anthropometric variables: These included age, gender, height, weight, body mass index (BMI), and race, providing a contextual understanding of the study population.Diabetes-related variables: Information included age at diagnosis, duration of diabetes, history of severe hypoglycemia or hyperglycemia requiring medical attention within the past year, and episodes of diabetic ketoacidosis (DKA). Participants also reported their most recent glycated hemoglobin (HbA1c) value and the date it was measured.Insulin treatment variables: Data were collected on: types of basal and prandial insulin used, number and timing of insulin injections, categorized as before, during, or after meals, along with specific time intervals (minutes) for each category, use of smart pens or connected devices, including device type and associated features, frequency of missed insulin doses and errors in insulin correction, such as overestimation or underestimation of doses, injection sites (e.g., abdomen, thigh, arm), rotation practices, and the frequency of needle changes, and additional antidiabetic therapies, if any. Injection timing data were collected exclusively for prandial insulin, as basal insulin is not administered in relation to meals and its timing is therefore not subject to meal-proximity recommendations.Glycemic monitoring variables: Participants provided information on: self-monitoring of blood glucose (SMBG) and continuous glucose monitoring (CGM) practices, CGM parameters over the past 28 days, including time in range (TIR), time above range (TaR), time below range (TbR), mean glucose levels, and coefficient of variation (CV), and frequency of glucose monitoring and any associated lapses, such as failure to act on abnormal glucose values. Sociodemographic and anthropometric variables: These included age, gender, height, weight, body mass index (BMI), and race, providing a contextual understanding of the study population. Sociodemographic and anthropometric variables: These included age, gender, height, weight, body mass index (BMI), and race, providing a contextual understanding of the study population. Diabetes-related variables: Information included age at diagnosis, duration of diabetes, history of severe hypoglycemia or hyperglycemia requiring medical attention within the past year, and episodes of diabetic ketoacidosis (DKA). Participants also reported their most recent glycated hemoglobin (HbA1c) value and the date it was measured. Diabetes-related variables: Information included age at diagnosis, duration of diabetes, history of severe hypoglycemia or hyperglycemia requiring medical attention within the past year, and episodes of diabetic ketoacidosis (DKA). Participants also reported their most recent glycated hemoglobin (HbA1c) value and the date it was measured. Insulin treatment variables: Data were collected on: types of basal and prandial insulin used, number and timing of insulin injections, categorized as before, during, or after meals, along with specific time intervals (minutes) for each category, use of smart pens or connected devices, including device type and associated features, frequency of missed insulin doses and errors in insulin correction, such as overestimation or underestimation of doses, injection sites (e.g., abdomen, thigh, arm), rotation practices, and the frequency of needle changes, and additional antidiabetic therapies, if any. Injection timing data were collected exclusively for prandial insulin, as basal insulin is not administered in relation to meals and its timing is therefore not subject to meal-proximity recommendations. Insulin treatment variables: Data were collected on: types of basal and prandial insulin used, number and timing of insulin injections, categorized as before, during, or after meals, along with specific time intervals (minutes) for each category, use of smart pens or connected devices, including device type and associated features, frequency of missed insulin doses and errors in insulin correction, such as overestimation or underestimation of doses, injection sites (e.g., abdomen, thigh, arm), rotation practices, and the frequency of needle changes, and additional antidiabetic therapies, if any. Injection timing data were collected exclusively for prandial insulin, as basal insulin is not administered in relation to meals and its timing is therefore not subject to meal-proximity recommendations. Glycemic monitoring variables: Participants provided information on: self-monitoring of blood glucose (SMBG) and continuous glucose monitoring (CGM) practices, CGM parameters over the past 28 days, including time in range (TIR), time above range (TaR), time below range (TbR), mean glucose levels, and coefficient of variation (CV), and frequency of glucose monitoring and any associated lapses, such as failure to act on abnormal glucose values. Glycemic monitoring variables: Participants provided information on: self-monitoring of blood glucose (SMBG) and continuous glucose monitoring (CGM) practices, CGM parameters over the past 28 days, including time in range (TIR), time above range (TaR), time below range (TbR), mean glucose levels, and coefficient of variation (CV), and frequency of glucose monitoring and any associated lapses, such as failure to act on abnormal glucose values. Statistical Analysis Data analysis was conducted in two phases. In the descriptive phase, quantitative variables were summarized using measures of central tendency (mean, median, and geometric mean as appropriate) and dispersion (standard deviation, minimum, maximum, and interquartile range). Categorical variables were presented as absolute frequencies and percentages. The analytical phase involved exploring associations between variables of interest using chi-squared tests for categorical variables and independent samples t-tests for continuous variables. Given the cross-sectional and exploratory nature of the study, no multivariable models were applied: the primary aim was descriptive and hypothesis-generating rather than confirmatory. The absence of a formal sample size calculation represents a limitation of this study, as no single primary endpoint on which to base such a calculation was prespecified. The recruitment period was defined a priori (April to June 2024), and all eligible respondents who completed the survey within that timeframe were included. A confidence level of 95% was used to determine statistical significance. For variables with missing responses, analyses were restricted to participants with available data; the number of valid responses is reported for each analysis. The statistical analysis was performed using SPSS Statistics, version 23 (IBM Corp., Armonk, NY, USA), and all data were anonymized to ensure participant privacy. Ethical Considerations This study was conducted in accordance with the Declaration of Helsinki (version Fortaleza, 2013) and approved by the Ethics Committee for Research with Medicines of the Hospital Clínico Universitario de Valencia (approval no. 2024/007, 25 January 2024). All participants provided informed consent electronically before completing the survey. Results General Characteristics of Participants The study included 288 participants with a mean age of 43.7 ± 14.7 years (Table 1). Among them, 28.8% were male and 71.2% were female. The racial/ethnic distribution was as follows: 95.8% white, 0.3% Black, 0.3% Asian, 1.4% other, and 2.1% preferred not to answer. The mean body mass index (BMI) was 24.6 ± 4.28 kg/m2. Most participants had type 1 diabetes (T1D) (74.7%), followed by type 2 diabetes (T2D) (17%), with the remainder (8.3%) having other forms of diabetes. The mean age at diagnosis was 25.9 ± 15.2 years, with a diabetes duration of 17.8 ± 14.5 years. Mild to moderate hypoglycemia in the last year was reported by 73.6%, while severe hyperglycemia was less frequent (4.9%). A history of diabetic ketoacidosis (DKA) was reported in 10.8% of participants in the last year. Table 1Participant demographics and clinical characteristicsCharacteristicValueDemographics Mean age (years)43.7 ± 14.7 Male/female (%)28.8/71.2Ethnicity (%) White95.8 Black0.3 Asian0.3 Other1.4 Prefer not to answer2.1 BMI (kg/m2)24.6 ± 4.28Diabetes characteristics Diabetes type: T1D/T2D/other (%)74.7/17.0/8.3 Age at diagnosis (years)25.9 ± 15.2 Duration of diabetes (years)17.8 ± 14.5 History of DKA in the past year (%)10.8 Mild–moderate hypoglycemia in the past year (%)73.6 Severe hyperglycemia requiring medical attention (%)4.9Glycemic monitoring and outcomes CGM use (%)89.9 Mean TIR (%)73.2 ± 16.2 Mean glucose (mg/dL)146.2 ± 29.6 HbA1c ≤ 7.5% (≤ 58 mmol/mol) (%)79.9 Therapeutic education Carbohydrate counting (%)69.1 Insulin-to-carbohydrate ratio use (%)68.4 Correction factor use (%)63.9BMI body mass index; CGM continuous glucose monitoring; DKA diabetic ketoacidosis; T1D type 1 diabetes; T2D type 2 diabetes; TIR time in range (70–180 mg/dL), HbA1c: glycated hemoglobin Table 1Participant demographics and clinical characteristicsCharacteristicValueDemographics Mean age (years)43.7 ± 14.7 Male/female (%)28.8/71.2Ethnicity (%) White95.8 Black0.3 Asian0.3 Other1.4 Prefer not to answer2.1 BMI (kg/m2)24.6 ± 4.28Diabetes characteristics Diabetes type: T1D/T2D/other (%)74.7/17.0/8.3 Age at diagnosis (years)25.9 ± 15.2 Duration of diabetes (years)17.8 ± 14.5 History of DKA in the past year (%)10.8 Mild–moderate hypoglycemia in the past year (%)73.6 Severe hyperglycemia requiring medical attention (%)4.9Glycemic monitoring and outcomes CGM use (%)89.9 Mean TIR (%)73.2 ± 16.2 Mean glucose (mg/dL)146.2 ± 29.6 HbA1c ≤ 7.5% (≤ 58 mmol/mol) (%)79.9 Therapeutic education Carbohydrate counting (%)69.1 Insulin-to-carbohydrate ratio use (%)68.4 Correction factor use (%)63.9BMI body mass index; CGM continuous glucose monitoring; DKA diabetic ketoacidosis; T1D type 1 diabetes; T2D type 2 diabetes; TIR time in range (70–180 mg/dL), HbA1c: glycated hemoglobin Participant demographics and clinical characteristics BMI body mass index; CGM continuous glucose monitoring; DKA diabetic ketoacidosis; T1D type 1 diabetes; T2D type 2 diabetes; TIR time in range (70–180 mg/dL), HbA1c: glycated hemoglobin Insulin Administration Practices Participants exclusively used multiple daily injections (MDI). The most common basal insulin was insulin degludec (46.5%), followed by insulin glargine U300 (37.2%). A majority (94.8%) administered basal insulin as a single daily dose, with evening administration being the most common (52.4%). Prandial insulin types varied, with faster aspart (44.4%) and insulin aspart (24.7%) being the most common (Fig. 1).Fig. 1Distribution of insulin types and administration times. Graph visualizing the distribution of basal and prandial insulin types and timing preferences. a Distribution of prandial insulin. b Distribution of basal insulin. c Timing preferences of prandial insulin injection. d Injection time among those injecting before meals Distribution of insulin types and administration times. Graph visualizing the distribution of basal and prandial insulin types and timing preferences. a Distribution of prandial insulin. b Distribution of basal insulin. c Timing preferences of prandial insulin injection. d Injection time among those injecting before meals The timing of prandial insulin injection was predominantly before meals (83.3%), while 9.7% reported indistinct timing, 3.8% injected after meals, and 3.1% during meals. Among those injecting before meals, 71.2% did so at least 15 min prior to eating, representing approximately 59% of the total sample. Errors and Omissions in Insulin Dosing Errors in insulin administration were prevalent (Table 2): 26.7% reported excessive doses more than five times in the last month and 38.5% reported insufficient doses more than five times in the same period. Dose omissions were also noted by 24% of participants, with 27.5% forgetting basal doses at least once in the last week and 55.1% forgetting prandial doses at least once in the last week. Of note, participants who forgot at least one prandial dose in the past week had lower time in range (TIR) (66.4% versus 72.7%, p = 0.006) and higher mean glucose levels (155 versus 141 mg/dL, p = 0.014) than those who did not. Table 2Frequency of insulin dosing errorsError typeFrequency (%)Excessive doses in the last month ≤ 5 times73.3 > 5 times26.7Insufficient doses in the last month ≤ 5 times61.5 > 5 times38.5Basal insulin dose omissions ≥ 1 time in the last week27.5Prandial insulin dose omissions ≥ 1 time in the last week55.1Data represent the percentage of participants reporting each error type. Frequencies refer to the past month (dosing errors) or past week (dose omissions) Table 2Frequency of insulin dosing errorsError typeFrequency (%)Excessive doses in the last month ≤ 5 times73.3 > 5 times26.7Insufficient doses in the last month ≤ 5 times61.5 > 5 times38.5Basal insulin dose omissions ≥ 1 time in the last week27.5Prandial insulin dose omissions ≥ 1 time in the last week55.1Data represent the percentage of participants reporting each error type. Frequencies refer to the past month (dosing errors) or past week (dose omissions) Frequency of insulin dosing errors Data represent the percentage of participants reporting each error type. Frequencies refer to the past month (dosing errors) or past week (dose omissions) When stratified by number of daily prandial injections (1–2, 3, or ≥ 4 injections/day), patients with a higher injection frequency showed a significantly greater prevalence of overdose errors (65.2% versus 77.1% versus 87.5%, p = 0.017) and underdose errors (73.9% versus 82.6% versus 91.7%, p = 0.029). No significant differences were observed in dose omissions, severe hypoglycemia, or severe hyperglycemia across injection groups (Table 3). Table 3Injection errors and hypoglycemic events by number of daily prandial injectionsVariable1–2 inj/day (n = 23)3 inj/day (n = 109) ≥ 4 inj/day (n = 120)p-ValueOverdose errors, n (%)65.2%77.1%87.5%0.017Underdose errors, n (%)73.9%82.6%91.7%0.029Dose omissions, n (%)17.4%26.6%25.0%0.651Severe hypoglycemia, n (%)a0.0%12.8%10.0%0.182Overdose errors: at least one episode of insulin overdosing in the past month. Underdose errors: at least one episode of insulin underdosing in the past monthThe absence of severe hypoglycemic events in the 1–2 injections/day group (n = 23) should be interpreted with caution, as the small sample size limits the statistical power to detect differences in low-frequency events Table 3Injection errors and hypoglycemic events by number of daily prandial injectionsVariable1–2 inj/day (n = 23)3 inj/day (n = 109) ≥ 4 inj/day (n = 120)p-ValueOverdose errors, n (%)65.2%77.1%87.5%0.017Underdose errors, n (%)73.9%82.6%91.7%0.029Dose omissions, n (%)17.4%26.6%25.0%0.651Severe hypoglycemia, n (%)a0.0%12.8%10.0%0.182Overdose errors: at least one episode of insulin overdosing in the past month. Underdose errors: at least one episode of insulin underdosing in the past monthThe absence of severe hypoglycemic events in the 1–2 injections/day group (n = 23) should be interpreted with caution, as the small sample size limits the statistical power to detect differences in low-frequency events Injection errors and hypoglycemic events by number of daily prandial injections Overdose errors: at least one episode of insulin overdosing in the past month. Underdose errors: at least one episode of insulin underdosing in the past month The absence of severe hypoglycemic events in the 1–2 injections/day group (n = 23) should be interpreted with caution, as the small sample size limits the statistical power to detect differences in low-frequency events Glycemic Monitoring and Outcomes Self-monitoring of blood glucose (SMBG) was reported by 59.7% of participants, with 76.3% achieving fasting blood glucose levels within the target range of 80–130 mg/dL, and 75% maintaining 2-h postprandial glucose levels between 100–180 mg/dL. Continuous glucose monitoring (CGM) was utilized by 89.9% of participants, yielding a mean TIR of 73.2 ± 16.2% and an average glucose level of 146.2 ± 29.6 mg/dL. Additionally, 79.9% of participants achieved glycated hemoglobin (HbA1c) levels ≤ 7.5% (≤ 58 mmol/mol). Participants who administered prandial insulin before meals showed significantly better glycemic outcomes compared with those not injecting before meals, including a higher proportion achieving HbA1c ≤ 7.5% (83.3% versus 62.5%, p = 0.001), greater TIR (74.4% versus 66.3%, p = 0.032), and lower mean glucose levels (144 versus 159 mg/dL, p = 0.019) (Fig. 2).Fig. 2Glycemic outcomes by timing of prandial insulin administration. Comparison of glycemic outcomes between participants injecting prandial insulin before meals (n = 240) versus those not injecting before meals (including during, after, or indistinct timing) (n = 48). a Proportion achieving HbA1c ≤ 7.5% (p = 0.001). b Time in range (TIR, 70–180 mg/dL) (p = 0.032). c Mean glucose levels (p = 0.019) Glycemic outcomes by timing of prandial insulin administration. Comparison of glycemic outcomes between participants injecting prandial insulin before meals (n = 240) versus those not injecting before meals (including during, after, or indistinct timing) (n = 48). a Proportion achieving HbA1c ≤ 7.5% (p = 0.001). b Time in range (TIR, 70–180 mg/dL) (p = 0.032). c Mean glucose levels (p = 0.019) Among patients injecting before meals, a progressive trend was observed with increasing injection–meal interval: the proportion achieving HbA1c ≤ 7.5% increased from 76% at 0 min to 90% at 15 min (Kruskal–Wallis p = 0.004), while mean glucose levels showed a nonsignificant decreasing trend (153 mg/dL at 0 min versus 141 mg/dL at 15 min, p = 0.253) (Fig. 3).Fig. 3Glycemic outcomes by prandial insulin injection timing interval. Bar charts showing mean time in range (TIR) (a), mean glucose (b), and proportion of participants achieving glycated hemoglobin (HbA1c) ≤ 7.5% (c) according to the time interval between prandial insulin injection and meal initiation, among patients injecting before meals (n = 240) Glycemic outcomes by prandial insulin injection timing interval. Bar charts showing mean time in range (TIR) (a), mean glucose (b), and proportion of participants achieving glycated hemoglobin (HbA1c) ≤ 7.5% (c) according to the time interval between prandial insulin injection and meal initiation, among patients injecting before meals (n = 240) Therapeutic Education and Impact Therapeutic education was well-integrated, with 69.1% of participants practicing carbohydrate counting, 68.4% applying insulin-to-carbohydrate ratios, and 63.9% using correction factors. The use of carbohydrate counting and insulin-to-carbohydrate ratios was associated with better HbA1c control with only 16.7% of users versus 27.4% of nonusers achieving an HbA1c above 7.5% (p = 0.035). Injection site preferences varied by insulin type: for basal insulin, 63.9% used the thigh or buttock, while 36.1% used other areas; for prandial insulin, 54.9% used the abdomen or arm, with 45.1% preferring other regions. Most participants (96.2%) reported regularly rotating injection sites to prevent lipodystrophy, while 3.8% did not. Additionally, 71.9% of participants reported using each needle 1–2 times before changing it, while the remaining 28.1% changed needles less frequently. Insulin Connected Pens Of the 272 participants who responded to this item (94.4% of the total sample), a significant portion (41.2%, n = 112) reported using insulin connected pens, including smart pens (36.8%) and smart caps (4.4%). Among connected pen users, the most commonly used devices were NovoPen Echo Plus® (Novo Nordisk A/S, Bagsværd, Denmark) (77.7%) and NovoPen® 6 (Novo Nordisk A/S, Bagsværd, Denmark) (20.5%), while Insulclock® (Agucisa Medical Technologies, Salamanca, Spain) and InPen® (Companion Medical, San Diego, CA, USA) were used less frequently (0.9% each). The adoption of connected pens was associated with a lower prevalence of mild-to-moderate hypoglycemic events compared with nonusers (68.8% versus 81.3%, p = 0.021) (Table 4). Table 4Impact of smart pen use on glycemic eventsOutcomeSmart pen users (n = 112)Nonusers (n = 160)p-ValueMild–moderate hypoglycemia (%)68.881.30.021Severe hyperglycemia (%)8.03.10.071DKA (%)10.710.00.206Prandial dose omissions (%)19.626.30.849Prandial injection during/after meals (%)16.116.90.861HbA1c > 7.5% (%)17.921.30.490TIR (%)73.973.40.914Mean glucose (mg/dL)1471440.450DKA diabetic ketoacidosis; TIR time in range (70–180 mg/dL). HbA1c glycated hemoglobin. Bold values indicate statistically significant differences (p < 0.05). (n = 272 of 288; 16 participants did not respond to this item) Table 4Impact of smart pen use on glycemic eventsOutcomeSmart pen users (n = 112)Nonusers (n = 160)p-ValueMild–moderate hypoglycemia (%)68.881.30.021Severe hyperglycemia (%)8.03.10.071DKA (%)10.710.00.206Prandial dose omissions (%)19.626.30.849Prandial injection during/after meals (%)16.116.90.861HbA1c > 7.5% (%)17.921.30.490TIR (%)73.973.40.914Mean glucose (mg/dL)1471440.450DKA diabetic ketoacidosis; TIR time in range (70–180 mg/dL). HbA1c glycated hemoglobin. Bold values indicate statistically significant differences (p < 0.05). (n = 272 of 288; 16 participants did not respond to this item) Impact of smart pen use on glycemic events DKA diabetic ketoacidosis; TIR time in range (70–180 mg/dL). HbA1c glycated hemoglobin. Bold values indicate statistically significant differences (p < 0.05). (n = 272 of 288; 16 participants did not respond to this item) These devices provide features such as dose tracking and reminders, which may contribute to better adherence and precision in insulin administration. However, no differences were found in HbA1c, TIR, mean glucose, or frequency of missed doses between smart pen users and nonusers. Discussion This study reveals important trends and challenges in insulin administration among individuals with diabetes in Spain. All participants followed a multiple daily injection (MDI) regimen, with notable variability in the timing and type of insulin used. The predominant use of basal insulin degludec and prandial insulin faster aspart reflects current clinical preferences and aligns with evidence supporting their effectiveness in achieving glycemic targets [18]. The availability of second-generation fast-acting insulin analogs (FAIAs)—such as ultrarapid lispro (still not available in Spain) and faster aspart—represents a significant advancement. These formulations, characterized by a faster onset and shorter duration of action, reduce the dependency on precise premeal timing [19]. Despite these pharmacological improvements, adherence to recommended prandial insulin timing remains suboptimal. Only 71.2% of those injecting before meals reported administering insulin ≥ 15 min prior to meals, and 55.1% admitted missing at least one prandial dose in the past week. Such delays and omissions negatively impact postprandial glycemic control, increasing the risk of complications, including microvascular and cardiovascular disease [3, 8]. Studies have consistently demonstrated that prandial insulin administration aligned with physiological insulin secretion improves glycemic metrics such as glycated hemoglobin (HbA1c), time in range (TIR), and postprandial glucose excursions [20, 21]. In line with our findings, Datye et al. [22] reported that, among individuals with type 1 diabetes (T1D), only 21% administered prandial insulin several minutes before the meal, while the majority did so immediately before (44%), during (10%), or even after eating (24%). In their study, earlier insulin administration was significantly associated with lower HbA1c levels (8.4% or 68 mmol/mol versus 8.8% or 73 mmol/mol, adjusted p < 0.001). Together, these findings highlight the persistent gap between pharmacological recommendations and real-world insulin practices, and the potential metabolic consequences of suboptimal timing. Suboptimal use of prandial insulin including missed and mistimed doses remains a well-established barrier to optimal glycemic control in patients on MDI [23]. Peyrot et al. found that intentional or unintentional omission of insulin doses was common and significantly associated with higher HbA1c levels [24]. This study emphasized that even occasional missed doses could substantially impact long-term glycemic control, especially among individuals on MDI therapy. In line with our findings, a systematic review by Robinson et al. reported that postprandial administration occurs in 25–30% of cases, and bolus omission is highly prevalent in both T1D and type 2 diabetes (T2D) [25]. These behaviors are strongly associated with worse glycemic outcomes, including higher HbA1c and increased risk of hypoglycemia. Thus, consistently administering insulin—particularly before meals—is essential for mimicking physiological insulin response, minimizing postprandial glucose excursions, and achieving target HbA1c levels. Although many patients acknowledge the consequences of such deviations, they continue to miss or delay doses owing to factors such as daily life disruptions, social constraints, and fear of hypoglycemia—underscoring the complex behavioral dynamics of insulin therapy [22]. It is worth noting that faster aspart, used by 44.4% of participants, has a more flexible recommended administration window, which may have contributed to the overall distribution of injection timing observed in this study. Another key finding of this study is the high frequency of dosing errors. Approximately 27% of participants reported overdosing insulin, while 38% admitted to underdosing. Both errors contribute to glycemic variability, a known independent predictor of adverse outcomes [26, 27]. The behaviors identified here (mistimed injections, missed doses, and dosing errors) share a common characteristic: they are potentially correctable through structured therapeutic education, and our data provide direct evidence of this. Participants who administered prandial insulin before meals achieved significantly better glycemic outcomes across all measured parameters. Similarly, the use of carbohydrate counting and insulin-to-carbohydrate ratios was associated with better HbA1c control, with only 16.7% of users versus 27.4% of nonusers achieving an HbA1c above 7.5% (p = 0.035). Carbohydrate counting, insulin-to-carbohydrate ratio use, and appropriate injection timing are skills that can be incorporated into structured diabetes education programs [28], and their association with better glycemic control in our sample supports the value of reinforcing these tools in routine clinical practice. These findings suggest that meaningful improvements in glycemic outcomes may be achievable without changes in pharmacological treatment. Encouragingly, 41.2% of responding participants reported using smart insulin pens, a figure that reflects the growing adoption of these devices in clinical practice in Spain. These devices offer bolus reminders, dose tracking, and data integration that support clinical decision-making. A Spanish pilot study further supports the potential of connected insulin devices in improving glycemic control. In this study, adults with poorly controlled T1D used Insulclock®, a smart cap that automatically records insulin injection timing and syncs with glucose data. After just 4 weeks, participants showed a significant reduction in missed and mistimed bolus doses (decrease per month by −3.9 (p = 0.135) and −5.4 (p = 0.032), respectively), along with improvements in TIR (+7%; p = 0.038), without an increase in hypoglycemia risk [29]. Further evidence from a real-world prospective study by Adolfsson et al. demonstrated that using the connected NovoPen® 6 in individuals with T1D on MDI and continuous glucose monitoring (CGM) resulted in a 43% reduction in missed bolus doses (from 25% to 14% of meals), an 8.5% increase in TIR, and reductions in hyperglycemia and level 2 hypoglycemia—without worsening mean glucose or time in level 1 hypoglycemia [30]. Integration with CGM systems has further amplified these benefits, enabling real-time feedback on insulin dosing and glycemic trends [31]. However, in our study, smart pen use was not associated with improvements in HbA1c or TIR, which may reflect the observational nature of the analysis and the presence of confounding factors such as disease complexity or duration. While these technological solutions hold great promise, their success hinges on user engagement and equitable access. Cost and availability remain major barriers, especially within public health systems or among uninsured populations [32]. Addressing these challenges will be essential to ensure widespread and equitable adoption of these innovations. Finally, the study emphasizes the critical role of behavioral patterns in diabetes management. Non-adherence to prandial insulin timing is closely linked to psychological factors such as treatment burden and injection-related distress [33]. Interventions that are patient-centered—such as motivational interviewing and individualized support—may enhance adherence and outcomes. Beyond pharmacological interventions, behavioral and emotional dimensions significantly affect insulin use. High levels of diabetes-related distress, fear of hypoglycemia, and negative perceptions surrounding injections (e.g., pain or social embarrassment) are frequently cited as reasons for delayed or omitted prandial insulin [24]. These barriers often override patients’ knowledge of proper management and contribute to glycemic instability. Additional factors such as stigma or complex daily routines can further disrupt insulin administration, even in highly motivated individuals [34]. Real-life situations such as eating at restaurants or social gatherings represent a particular challenge: when the exact timing of a meal is unpredictable, patients face a difficult dilemma between injecting too early—risking hypoglycemia—or waiting until food is served, which may result in suboptimal postprandial control. Practical strategies that should be incorporated into structured education include: coordinating with family members or companions to receive advance notice before meals are served, selecting familiar restaurants where meal timing is more predictable, carrying a fast-acting carbohydrate snack to manage unexpected delays, and leveraging the pharmacokinetic advantages of second-generation FAIAs such as faster aspart, which can be administered at the start of a meal without compromising postprandial glycemic control. To promote long-term adherence and metabolic stability, psychosocial support and personalized education should be embedded into standard care. The behaviors identified in this study (delayed or omitted prandial insulin and frequent dosing errors) are gaps that structured educational interventions can target directly. Integrating individualized diabetes education focused on injection timing and dosing accuracy into routine clinical practice could therefore translate into meaningful glycemic improvements at a population level. This study has several notable strengths. It is one of the first investigations to provide detailed insights into real-world insulin administration practices among adults with diabetes using MDI in Spain. The use of a comprehensive and structured online survey allowed for the collection of diverse behavioral, clinical, and technological data from a sizable sample across various regions. Furthermore, the study addressed emerging areas of interest, such as the adoption of smart insulin pens and their association with glycemic outcomes, enhancing their relevance to current clinical practice. The inclusion of CGM-derived metrics such as TIR adds robustness to the analysis of glycemic control. However, certain limitations must be acknowledged. Its cross-sectional design does not allow for causal conclusions between insulin practices and glycemic outcomes. Data were self-reported, which may introduce recall or reporting bias, especially regarding insulin dosing and glucose values. Additionally, recruiting participants through an online platform may have led to selection bias, with overrepresentation of individuals who are more engaged or health-literate. All analyses were based on unadjusted bivariable comparisons, and potential confounding by variables such as diabetes type, disease duration, or CGM use cannot be excluded. The use of unadjusted analyses precludes assessment of the independent contribution of each variable to glycemic outcomes. Additionally, although sex distribution was recorded (28.8% male, 71.2% female), sex-disaggregated analyses were not performed; future studies should explore whether insulin administration practices and outcomes differ by sex and gender. The small number of T2D participants precluded any formal statistical comparison between diabetes types. The predominance of T1D participants (74.7%) may limit the generalizability of findings to individuals with other types of diabetes. Finally, the exclusion of insulin pump users limits the generalizability of the findings to those using MDI. Furthermore, as the survey was disseminated openly through social media channels rather than distributed to a predefined participant list, a formal response rate could not be calculated, which precludes quantification of nonresponse bias. Although the inclusion criteria required at least one daily insulin injection, all participants who completed the survey were using MDI regimens, likely reflecting the profile of the Canal Diabetes user base. This discrepancy is acknowledged as a potential source of selection bias. The predominance of female respondents (71.2%) is a recognized limitation of online survey research, as women are known to participate in web-based surveys at higher rates than men regardless of topic [35]. This sex imbalance may limit the generalizability of our findings to male patients with diabetes on MDI. Conclusions This study provides important insights into insulin administration practices in Spain and highlights key opportunities for improving diabetes care. Optimizing adherence, reducing dosing errors, and ensuring equitable access to new technologies will be critical in advancing patient outcomes. Our data specifically show that appropriate premeal insulin timing and the use of carbohydrate counting and insulin-to-carbohydrate ratios were associated with significantly better glycemic outcomes, suggesting that structured therapeutic education represents a practical, accessible, and underutilized tool in routine clinical care. Future research should explore strategies to enhance behavioral adherence and assess the long-term impact of technological and educational interventions on clinical outcomes and quality of life in people with diabetes. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 133 KB) Supplementary file1 (PDF 133 KB)