We conducted a multicentre cross-sectional observational and analytical study over a 6-month period (February-July 2018). We included patients aged 2–16 years with a duration of T1D of at least 1 year that provided signed informed consent to participate in the study. We excluded patients with any other form of diabetes, a duration of diabetes of less than 6 months or with comorbidities that may impact quality of life or affect comprehension of the questionnaire. Once the questionnaires were submitted, we automatically excluded those with more than 3 items left blank or, in the case of the proxy report, those that were not filled out by the main caregiver.

The study was approved by the Ethics Committee of the Hospital de Málaga, which coordinated the project. We calculated that the sample size needed to achieve a precision of 5.3% in estimating a proportion in a normal asymptotic distribution with a 95% confidence interval (bilateral) considering the known prevalence would be of 220 cases, assuming a loss of 15%.

We selected cases randomly and proportionally to the number of eligible patients managed in the paediatric endocrinology clinics of the 6 participating hospitals in Andalusia: Hospital Serranía de Ronda, Hospital General San Agustín (Linares), Hospital Materno-Infantil de Jaén (Jaen), Hospital Materno-Infantil de Jerez (Jerez), Hospital Universitario Virgen Macarena (Seville) and Hospital Materno-Infantil de Málaga (Malaga). We included a specific number of patients from each centre based on the total number managed at the site. We gave the questionnaires to the 3 patients that had the earliest appointments in the clinic in the morning until we obtained the necessary sample size. We also stratified by age group (2–4, 5–7, 8–12 and 13–16 years) to obtain a homogeneous sample based on the actual prevalence of the disease.

Each patient and/or parent was given 2 questionnaires: the Pediatric Quality of Life Inventory (PedsQL) version 3.0 Type 1 Diabetes Module15 and, to assess adherence to treatment, the Self Care Inventory-Revised Version (SCI-R16). The former is considered the best instrument to assess quality of life in children. There are different versions with the language adapted for age (ages 2–4; 5–7; 8–12 and 13–18). This questionnaire is completed by the patients themselves if they are older than 8 years and otherwise completed by a caregiver. There is no version for children under 2 years. The questionnaire covers a recall period of 7 days and consists of 5 scales with a total of 28 items: diabetes symptoms (11 items), treatment barriers (4 items), treatment adherence (7 items), worry (3 items), and communication (3 items). All items are scored on a 5-point Likert scale ranging from 0 (never) to 4 (almost always). After questionnaires were submitted, we calculated the mean total score and transformed it to a scale from 0 to 100. High scores indicate few problems and therefore a high quality of life.

The SCI-R16 for assessment of adherence was completed by a parent if the patient was aged less than 11 years or by the patient if aged more than 11 years. We analysed the following variables: checking blood glucose levels with a blood glucose meter, checking ketone levels, correct dose of insulin infusion; adherence to adequate carbohydrate portions and reading food labels, response to episodes of hypoglycaemia. The questionnaire comprises 15 items scored on a scale from 1 (never) to 5 (always).

In addition, we collected data on the following:

• -

  Demographic characteristics: age at diagnosis, duration of disease, sex and pubertal stage.

• -

  Metabolic control (data collected in the electronic health records database during previous follow-up visits): treatment modalities (MDI or CSII), use of flash glucose monitoring systems or integrated real-time continuous glucose monitoring/insulin (RT-CGM/CSII) systems. We assessed the degree of metabolic control based on the mean of the two latest HbA1c measurements, whether capillary or venous. Lastly, we recorded the number of episodes of severe hypoglycaemia (defined as episodes that the patient could not resolve independently and required intervention by a third party) in the past year. We also collected data on episodes of diabetic ketoacidosis, comorbidities and complications of diabetes.

We have expressed continuous variables as mean and standard deviation. We used normality tests and multiple linear regression models with stepwise variable selection and the Akaike information criterion (AIC) for all SCI-R and PedsQL groups and for the SCI-R and PedsQL in the total sample with the purpose of assessing the impact on these scores of changes in the following variables: HbA1c concentration (treated as a continuous variable), presence of complications, treatment with MDI or CSII and others such as current age, number of severe hypoglycaemia episodes or the type of glucose monitoring system used. We defined statistical significance using a confidence level of 95%, considering a p-value of less than 0.05 significant. We used the Spearman correlation coefficient (rho) to assess the correlation between adherence to treatment and quality of life. We interpreted coefficients of 0.1−0.29 indicative of a weak correlation, coefficients of 0.3−0.49 indicative of a moderate correlation, and coefficients of 0.5 or greater indicative of a strong correlation. The statistical analysis was performed with the software R (version 3.5.0).

We selected 220 patients, and the final sample consisted of 178 after exclusion of 42 patients.

In the final sample, 51.1% of the patients were male, the mean age at diagnosis was 6.1 years and the mean age at the time of the study was 10.6 years. The mean duration of disease was 4.7 years, with a duration of more than 5 years in 67% of the patients (Table 1).

We found that 12.9% of patients used a CSII system. Also, 83.2% monitored glucose levels exclusively through capillary blood measurements, while 16.8% used some form of interstitial glucose measuring device; among the latter, 9.5% used a flash glucose monitoring system (FreeStyle), while 7.3% used integrated RT-CGM/CSII systems. In this sample, all of the latter used Medtronic MiniMed 640 G pumps connected to an Enlite/Guardian sensor, as this was the only system funded by the public health system of Andalusia.

The mean HbA1c level concentration was 7.1%, and 25% of patients had concentrations of 7.5% or higher.

In the sample, 5.5% of the patients experienced severe episodes of hypoglycaemia. Since the disease duration was short, only 1% of patients had renal complications in the form of microalbuminuria, and when it came to comorbidities, 8.4% had coeliac disease and 8.3% autoimmune thyroiditis.

In the analysis concerning adherence to treatment, we found a mean score of 4.21 ± 0.32. The lowest score corresponded to the preventative/routine aspects of care, while the highest scores corresponded to management of low blood glucose levels and the administration of insulin (Table 2). Overall, the SCI-R score was inversely correlated to the HbA1c concentration and to age, although the correlation coefficient was too low to be considered significant (Table 3).

Table 3 shows detailed results of the analysis by dimension of the SCI-R, of which the salient findings were that the HbA1c concentration was correlated to the diet and exercise dimensions and the use of a CSII system was positively correlated to the routine care dimension, the latter being the strongest correlation found in our analysis.

As for the analysis concerning quality of life, the mean total score in the PedsQL was 72.29 ± 8.29. We ought to highlight that we found a direct association with real time continuous glucose monitoring and with a lower number of episodes of severe hypoglycaemia, and that these correlations were strong and only exceeded in strength by the presence of renal complications (a variable that was difficult to assess due to the small number of patients in this subset, although the association was strong) (Table 2). Table 4 presents detailed results for the analysis by subscale. The most salient finding was that the use of the integrated RT-CGM/CSII system was positively correlated to treatment adherence and worry related to the disease, the latter being a strong correlation. Lastly, age was positively correlated to treatment adherence, although this correlation was weak.

We did not find a statistically significant association between adherence to treatment and quality of life assessed by means of the SCI-R and PedsQL scores (Spearman rho, 0.18; P > .05).