Validation of the type 1 diabetes distress scale (T1-DDS) in a large Danish cohort: Content validation and psychometric properties.

Aim: To validate the Type 1 Diabetes Distress Scale (T1-DDS) in a large sample of adults with Type 1 diabetes (T1D) from diabetes clinics in Denmark. Methods: Altogether 40 adults with T1D were interviewed to explore the content of T1-DDS in a Danish setting and to validate the translation of the T1-DDS into Danish. Subsequently, a survey including T1-DDS, the Problem Areas In Diabetes scale (PAID-20), fear of hypoglycemia, social support, and diabetes duration was answered by 2201 people with T1D. Other person characteristics were collected from the National Patient Register. HbA1c was obtained from the Clinical Laboratory Information System. Data distribution, internal consistency, convergent and construct validity, factor structure, three weeks retest, and cut-points were explored. Results: Interview data supported the relevance of all T1-DDS items for the assessment of diabetes distress among adults with T1D. The T1-DDS showed good content and acceptable construct validity, and the ability to detect high diabetes distress levels. A high correlation between T1-DDS and PAID-20 (rho = 0.91) was found. The retest scores showed a good reliability (all rho ≥0.68) with the highest variability in the Friends/Family Distress and Physician Distress subscales and the lowest variability in the Powerlessness and Eating Distress subscales of the T1-DDS. Qualitative findings pointed out relevant concerns of people with T1D, which were not included in the T1-DDS. Conclusion: The study supports the use of the Danish T1-DDS, but also highlights that existing diabetes distress questionnaires including T1-DDS do not cover all potential diabetes stressors and worries.

Achieving a useful and person-centred diabetes consultation is a shared responsibility between diabetologists and people with diabetes: a qualitative study of perspectives from people with type 1 diabetes.

AIMS: To explore perceptions of useful routine consultations with diabetologists from the perspective of adults with type 1 diabetes, including preferences for discussing psychosocial issues. METHODS: We conducted semi-structured interviews in 2018/2019 with 33 people with type 1 diabetes (age 22-75 years, 20 men and 13 women, median diabetes duration 25 years) recruited from two diabetes clinics in the capital region of Denmark. Interviews were audio recorded, transcribed verbatim and analysed using thematic text condensation. RESULTS: Achieving a useful consultation was perceived as a shared responsibility between people with diabetes and diabetologists. Participants' perspectives of what constitutes a useful consultation and expectations for both consultation and diabetologist varied in relation to perceptions of (1) the interaction between the person with diabetes and diabetologist, including being prepared, being honest, experiencing good rapport and preferring a partnership with the diabetologist or 'keeping it clinical' and (2) the diabetologist's approach to diabetes care, including providing up-to-date knowledge and listening and showing understanding. CONCLUSIONS: Both content and style of diabetes consultations need to be adapted to the individual person with type 1 diabetes. People with diabetes have an important role in expressing their needs and preferences related to both content and style. Diabetologists need to be aware of and attentive to the many individual needs and expectations among people with diabetes, including the desire and need to discuss psychosocial issues. Dialogue tools for preparation and in consultations may enable people with diabetes to voice their needs and expectations and diabetologists to juggle these diversities.

Psychosocial factors associated with HbA1c in adults with insulin pump-treated type 1 diabetes: a systematic review.

AIMS: To identify currently available studies on the association between psychosocial factors and HbA1c in adults with insulin pump-treated type 1 diabetes, by performing a systematic review of the literature. METHODS: MEDLINE, Embase, CINAHL and PsycINFO were searched for original studies on the association between psychosocial factors and HbA1c in ≥ 50 adult, non-pregnant, insulin pump users with type 1 diabetes. RESULTS: The search resulted in 1777 unique records, of which eight were eligible for inclusion. All identified studies were observational, with sample sizes ranging from 51 to 214. Seven different psychosocial factors were investigated in the eight studies. Study analysis suggested that HbA1c may be associated with diabetes numeracy and quality of life. There were no indications of associations between HbA1c and fear of hypoglycaemia or self-efficacy. Results regarding associations between HbA1c and coping style, diabetes distress and locus of control were inconsistent. CONCLUSIONS: This systematic review summarizes the currently limited information on the association between psychosocial factors and HbA1c during insulin pump therapy. The evidence base of the included studies was weak, and this review highlights the need for more research in these areas, with improved methodological and theoretical frameworks, including exploration of a broader spectrum of psychosocial variables and their potential association with HbA1c and other metabolic outcomes. (PROSPERO International prospective register of systematic reviews registration no: CRD42020145705).

Glycaemic variability and hypoglycaemia are associated with C-peptide levels in insulin-treated type 2 diabetes.

AIM: The aim of the study was to evaluate the association between C-peptide levels, glycaemic variability and hypoglycaemia in patients with insulin-treated type 2 diabetes (T2D). METHODS: A total of 98 patients with T2D treated with basal-bolus insulin were enrolled in a cross-sectional study. Glycaemic variability and hypoglycaemia were assessed from continuous glucose monitoring (CGM) data recorded over 6 days: Glycemic variability was assessed by calculating the mean coefficient of variation (CV), while hypoglycemia was defined as sensor glucose levels ≤ 3.9 mmol/L or < 3.0 mmol/L. Fasting C-peptide and fasting glucose were measured on day 1. RESULTS: Low levels of fasting C-peptide correlated with higher CV (r = -0.53, P < 0.0001). In a multivariate regression model with HbA1c, body mass index, diabetes duration and total daily insulin dose, only C-peptide was significantly associated with CV. Patients with ≥ 1 episode of hypoglycaemia had significantly lower median C-peptide levels than patients without hypoglycaemia (274 (136-620) pmol/L vs. 675 (445-1013) pmol/L, respectively; P = 0.0004). Also, 17 patients clinically diagnosed with T2D had detectable glutamic acid decarboxylase (GAD) antibodies (≥ 5 U/mL). These GAD-positive patients had significantly lower fasting C-peptide, higher CV and greater frequency of hypoglycaemia than GAD-negative patients. CONCLUSION: In patients with insulin-treated T2D, low levels of C-peptide are associated with greater glycaemic variability and higher risk of hypoglycaemia, suggesting that C-peptide levels should be taken into consideration when optimizing insulin treatment and assessing hypoglycaemia risk.

Effects of alcohol on plasma glucose and prevention of alcohol-induced hypoglycemia in type 1 diabetes-A systematic review with GRADE.

Because ethanol is thought to be a risk factor for severe hypoglycemia, patients with type 1 diabetes (T1D) are recommended to limit ethanol intake. However, little is known on how ethanol affects plasma glucose and how ethanol-induced hypoglycemia can be prevented. In this study, we systematically reviewed the literature for ethanol effects on plasma glucose and for prevention strategies on ethanol-induced hypoglycemia. Electronic searches on PubMed and Google were conducted in February 2017. Randomized clinical trials and observational studies were included. Studies involved patients with T1D with no history of ethanol abuse. The primary aims were changes in plasma glucose after ethanol intake and prevention strategies for ethanol-induced hypoglycemia. Quality of the studies was assessed by GRADE. Additionally, we searched for guidelines from diabetes associations on their suggested prevention strategies. We included 13 studies. Eight studies reported that ethanol, regardless of administration intravenously or orally, were associated with an increased risk of hypoglycemia due to decrease in plasma glucose, impaired counter-regulatory response, awareness of hypoglycemia, and cognitive function. Five studies did not report an increased risk of hypoglycemia. None of the studies investigated prevention strategies for ethanol-induced hypoglycemia. Recommendations from 13 diabetes associations were included. All associations recommend that ethanol should only be consumed with food intake. The majority of included studies showed that ethanol intake increased the risk of hypoglycemia in patients with T1D. However, the evidence for how to prevent ethanol-induced hypoglycemia is sparse, and further investigations are needed to establish evidence-based recommendations.

Impact of lifestyle intervention for obese women during pregnancy on maternal metabolic and inflammatory markers.

BACKGROUND: Offspring of obese mothers have increased risk of developing obesity and related short- and long-term disease. The cause is multifactorial and may partly be explained by the unfavorable intrauterine environment. Intervention during pregnancy leading to a healthier lifestyle among obese may alter this. OBJECTIVE: To assess the effect of lifestyle intervention on markers of maternal metabolism and inflammation in 'the TOP (Treatment of Obese Pregnant Women) study', a randomized controlled trial. METHODS: In the TOP-study 425 participants with body mass index ⩾30 kg/m2 were randomized to intervention with dietary advices and physical activity assessed by pedometer (PA+D), physical activity assessed by pedometer (PA) or control (C). Of 389 participants completing the study 376 had available blood samples. Serum was analyzed for insulin, c-peptide, lipid profile, leptin, high-sensitivity CRP (hsCRP) and Soluble urokinase Plasminogen Activator Receptor (suPAR), in week 18-20 and 28-30, and simultaneously a 2-h oral glucose-tolerance-test was performed. Diet was assessed in gestational week 11-14 and 36-37 using a validated 360-item Food Frequency Questionnaire. RESULTS: Median levels of hsCRP in gestational week 28-30 were lower in each of the intervention groups (8.3 mg/l in PA+D group, P=0.03; and 8.8 mg/l in PA group, P=0.02) versus the control group (11.5 mg/l). Obtaining 11 000 steps per day as aimed for resulted in a 21% lower hsCRP compared to non-compliant women. Women reporting high carbohydrate intake had around 30% higher hsCRP concentrations in late gestation than women reporting the lowest intake. There were no differences in lipid profile or any of the metabolic markers in gestational week 28-30 when comparing the intervention and control groups. CONCLUSIONS: Lifestyle intervention in obese women can reduce hsCRP representing a marker of inflammation during pregnancy. The effect may partly be mediated by more physical activity and partly by changes in intake of carbohydrates and the glycaemic load.

Comparing effects of insulin analogues and human insulin on nocturnal glycaemia in hypoglycaemia-prone people with Type 1 diabetes.

AIMS: To assess the difference between analogue and human insulin with regard to nocturnal glucose profiles and risk of hypoglycaemia in people with recurrent severe hypoglycaemia. METHODS: A total of 72 people [46 men, mean ± sd age 54 ± 12 years, mean ± sd HbA1c 65 ± 12 mmol/mol (8.1 ± 1.1%), mean ± sd duration of diabetes 30 ± 14 years], who participated in a 2-year randomized, crossover trial of basal-bolus therapy with insulin detemir/insulin aspart or human NPH insulin/human regular insulin (the HypoAna trial) were studied for 2 nights during each treatment. Venous blood was drawn hourly during sleep. Primary endpoints were nocturnal glucose profiles and occurrence of hypoglycaemia (blood glucose ≤ 3.9 mmol/l). RESULTS: During insulin analogue treatment, the mean nocturnal plasma glucose level was significantly higher than during treatment with human insulin (10.6 vs 8.1 mmol/l). The fasting plasma glucose level was similar between the treatments. Nocturnal hypoglycaemia was registered during 41/101 nights (41%) in the human insulin arm and 19/117 nights (16%) in the insulin analogue arm, corresponding to a hazard ratio of 0.26 (95% CI 0.14 to 0.45; P < 0.0001) with insulin analogue. CONCLUSIONS: Treatment with insulin analogue reduces the occurrence of nocturnal hypoglycaemia assessed by nocturnal glucose profiles in people with Type 1 diabetes prone to severe hypoglycaemia. Nocturnal glucose profiles provide a more comprehensive assessment of clinical benefit of insulin regimens as compared to conventional recording of hypoglycaemia.

Effects of advanced carbohydrate counting guided by an automated bolus calculator in Type 1 diabetes mellitus (StenoABC): a 12-month, randomized clinical trial.

AIMS: To test whether concomitant use of an automated bolus calculator for people with Type 1 diabetes carrying out advanced carbohydrate counting would induce further improvements in metabolic control. METHODS: We conducted a 12-month, randomized, parallel-group, open-label, single-centre, investigator-initiated clinical study. We enrolled advanced carbohydrate counting-naïve adults with Type 1 diabetes and HbA1c levels 64-100 mmol/mol (8.0-11.3%), who were receiving multiple daily insulin injection therapy. In a 1:1-ratio, participants were randomized to receive training in either advanced carbohydrate counting using mental calculations (MC group) or advanced carbohydrate counting using an automated bolus calculator (ABC group) during a 3.5-h group training course. For 12 months after training, participants attended a specialized diabetes centre quarterly. The primary outcome was change in HbA1c from baseline to 12 months. RESULTS: Between August 2012 and September 2013, 168 participants (96 men and 72 women) were recruited and randomly assigned to the MC group (n = 84) and the ABC group (n = 84). Drop-out rates were 23.8 and 21.4%, respectively (P = 0.712); 130 participants completed the study. The baseline HbA1c was 75 ± 9 mmol/mol (9.0 ± 0.8%) in the MC group and 74 ± 8 mmol/mol (8.9 ± 0.7%) in the ABC group. At 12 months, change in HbA1c was significant within both groups: MC group: -2 mmol/mol (95% CI -4 to -1) or -0.2% (95% CI -0.4 to -0.1; P = 0.017) and ABC group: -5 mmol/mol (95% CI -6 to -3) or -0.5% (95% CI -0.6 to -0.3; P < 0.0001), but HbA1c reductions were significantly greater in the ABC group (P = 0.033). No episodes of severe hypoglycaemia were reported. CONCLUSIONS: People with Type 1 diabetes initiating advanced carbohydrate counting obtained significantly greater HbA1c reductions when guided by an automated bolus calculator (NCT02084498).

Effect of insulin analogues on frequency of non-severe hypoglycaemia in patients with type 1 diabetes prone to severe hypoglycaemia: The HypoAna trial.

AIM: Insulin analogues reduce the risk of hypoglycaemia compared with human insulin in patients with type 1 diabetes (T1D) and minor hypoglycaemia problems. The HypoAna trial showed that, in patients with recurrent severe hypoglycaemia, treatment based on insulin analogues reduces the risk of severe hypoglycaemia. The present study aims to assess whether this also applies to non-severe hypoglycaemia events during the day and at night. METHODS: This 2-year investigator-initiated multicentre, prospective, randomized, open, blinded endpoint (PROBE) trial involved patients with T1D and at least two episodes of severe hypoglycaemia during the previous year. Using a balanced crossover design, patients were randomized to basal-bolus therapy based on analogue (detemir/aspart) or human (NPH/regular) insulins. A total of 114 participants were included. Endpoints were the number of severe hypoglycaemic events and non-severe events, including documented symptomatic and asymptomatic episodes occurring during the day and at night (ClinicalTrials.gov number: NCT00346996). RESULTS: Analogue-based treatment resulted in a 6% (2-10%; P=0.0025) overall relative risk reduction of non-severe hypoglycaemia. This was due to a 39% (32-46%; P<0.0001) reduction of non-severe nocturnal hypoglycaemia, seen for both symptomatic (48% [36-57%]; P<0.0001) and asymptomatic (28% [14-39%]; P=0.0004) nocturnal hypoglycaemia episodes. No clinically significant differences in hypoglycaemia occurrence were observed between the insulin regimens during the day. The time needed to treat one patient with insulin analogues to avoid one episode (TNT1) of non-severe nocturnal hypoglycaemia was approximately 3 months. CONCLUSION: In T1D patients prone to severe hypoglycaemia, treatment with analogue insulin reduced the risk of non-severe nocturnal hypoglycaemia compared with human insulin.

Effects of subcutaneous, low-dose glucagon on insulin-induced mild hypoglycaemia in patients with insulin pump treated type 1 diabetes.

AIM: To investigate the dose-response relationship of subcutaneous (s.c.) glucagon administration on plasma glucose and on counter-regulatory hormone responses during s.c. insulin-induced mild hypoglycaemia in patients with type 1 diabetes treated with insulin pumps. METHODS: Eight insulin pump-treated patients completed a blinded, randomized, placebo-controlled study. Hypoglycaemia was induced in the fasting state by an s.c. insulin bolus and, when plasma glucose reached 3.4 mmol/l [95% confidence interval (CI) 3.2-3.5], an s.c. bolus of either 100, 200, 300 µg glucagon or saline was administered. Plasma glucose, counter-regulatory hormones, haemodynamic variables and side effects were measured throughout each study day. Peak plasma glucose level was the primary endpoint. RESULTS: Plasma glucose level increased significantly by a mean (95% CI) of 2.3 (1.7-3.0), 4.2 (3.5-4.8) and 5.0 (4.3-5.6) mmol/l to 6.1 (4.9-7.4), 7.9 (6.4-9.3) and 8.7 (7.8-9.5) vs 3.6 (3.4-3.9) mmol/l (p < 0.001) after the three different glucagon doses as compared with saline, and the increase was neither correlated with weight nor insulin levels. Area under the plasma glucose curve, peak plasma glucose, time to peak plasma glucose and duration of plasma glucose level above baseline were significantly enhanced with increasing glucagon doses; however, these were not significantly different between 200 and 300 µg glucagon. Free fatty acids and heart rates were significantly lower initially after glucagon than after saline injection. Other haemodynamic variables, counter-regulatory hormones and side effects did not differ between interventions. CONCLUSIONS: An s.c. low-dose glucagon bolus effectively restores plasma glucose after insulin overdosing. Further research is needed to investigate whether low-dose glucagon may be an alternative treatment to oral carbohydrate intake for mild hypoglycaemia in patients with type 1 diabetes.

Cost-effectiveness of continuous subcutaneous insulin infusion versus multiple daily injections of insulin in Type 1 diabetes: a systematic review.

AIM: Continuous subcutaneous insulin infusion (CSII) is increasingly used in clinical practice for the management of selected patients with Type 1 diabetes. Several cost-effectiveness studies comparing CSII vs. multiple insulin injections (MDI) have been reported. The aim was systematically to review these analyses and test the hypothesis that CSII is a cost-effective use of healthcare resources across settings. METHODS: A literature review was performed using MEDLINE, Cochrane Library and other databases. No time limit or language restrictions were applied. After two rounds of screening, 11 cost-effectiveness analyses were included in the final review, of which nine used the CORE Diabetes Model. A narrative synthesis was conducted and mean cost effectiveness calculated. RESULTS: CSII was considered cost-effective vs. MDI in Type 1 diabetes in all 11 studies in 8 countries, with a mean (95% CI) incremental cost effectiveness ratio of €30 862 (17 997-43 727), US$40 143 (23 409-56 876) per quality-adjusted life year (QALY) gained. CSII was associated with improved life expectancy and quality-adjusted life expectancy (0.4-1.1 QALYs in adults), driven by lower HbA(1c) and lower frequency of hypoglycaemic events vs. MDI. CSII was associated with higher lifetime direct costs due to higher treatment costs but this was partially offset by cost-savings from reduced diabetes-related complications. CONCLUSIONS: Published cost-effectiveness analyses show that in Type 1 diabetes CSII is cost-effective vs. MDI across a number of settings for patients who have poor glycaemic control and/or problematic hypoglycaemia on MDI, with cost-effectiveness highly sensitive to the reduction in HbA1c and hypoglycaemia frequency associated with CSII.

Dual-hormone treatment with insulin and glucagon in patients with type 1 diabetes.

Intensive insulin treatment in type 1 diabetes reduces the incidence and slows the progression of microvascular and macrovascular complications; however, it is associated with an increased risk of hypoglycaemia and weight gain. In this review, we propose dual-hormone treatment with insulin and glucagon as a method for achieving near normalization of blood glucose levels without increasing hypoglycaemia frequency and weight gain. We briefly summarize glucagon pathophysiology in type 1 diabetes as well as the current applications of glucagon for the treatment of hypoglycaemia. Until now, the use of glucagon has been limited by the need for reconstitution immediately before use, because of instability of the available compounds; however, stabile compounds are soon to be launched and will render long-term intensive dual-hormone treatment in type 1 diabetes possible.

Newborn regional body composition is influenced by maternal obesity, gestational weight gain and the birthweight standard score.

AIM: This study investigated whether newborn body composition is influenced by prepregnancy obesity and gestational weight gain (GWG) and explored any associations between body composition and birthweight standard score (z-score), categorised by size for gestational age. METHODS: We recruited 231 obese and 80 normal weight mothers and their newborn infants and assessed the babies' body composition using dual-energy X-ray absorptiometry. RESULTS: The total and abdominal fat masses of infants born to mother who were obese before pregnancy were 135 g (p < 0.001) and 18 g (p < 0.001) higher than the offspring of normal weight mothers. The infants' fat mass increased by 11 g (p < 0.001) for every kilogram of GWG. There were no associations between prepregnancy obesity and fat-free mass. The fat percentage was significantly higher in infants who were large for gestational age (15.3%) than small for gestational age (5.2%) and appropriate for gestational age (9.8%) (p < 0.001). Lower birthweight z-score was associated with a higher proportion of abdominal fat mass (p = 0.009). CONCLUSION: Infants born to obese mothers had higher fat mass at birth, with abdominal fat accumulation. Low birthweight was associated with a lower crude abdominal fat mass, but a higher proportion of total fat mass placed abdominally.

Effects of advanced carbohydrate counting in patients with type 1 diabetes: a systematic review.

AIM: Advanced carbohydrate counting, a systematic method for insulin bolus calculation, is recommended in the management of type 1 diabetes. The aim of this systematic review was to summarize all available evidence from randomized and observational studies of the effects of advanced carbohydrate counting on glycaemic control (HbA(1c)), psychosocial measures, weight and hypoglycaemic events in patients of all age groups with type 1 diabetes on a basal-bolus insulin regimen. METHODS: An electronic search of Scopus, MEDLINE and The Cochrane Library conducted in January 2013 identified 27 relevant articles. Six were randomized controlled trials and 21 were observational studies. Large heterogeneity existed across studies with regard to study design and patient populations. Reporting of statistical measures was insufficient to serve as a basis for a meta-analysis. RESULTS: Overall, the studies demonstrated a positive trend in change in HbA(1c) after introduction of advanced carbohydrate counting. Reductions in HbA(1c) ranged from 0.0 to 13 mmol/mol (0.0-1.2%). Most psychosocial measures improved; however, only few improvements were considered clinically relevant. Both weight gain and reduction were registered, but most studies found no significant weight changes. The majority of studies assessing the incidence of hypoglycaemic events found a significant reduction in the event rate and none reported an increase in the incidence. CONCLUSIONS: In summary, the currently available literature does not provide sufficient evidence to definitively determine the effects of advanced carbohydrate counting on HbA(1c), psychosocial measures, weight or hypoglycaemic events. Nevertheless, the method still appears preferable to other insulin dosing procedures, which justifies continued use and inclusion of advanced carbohydrate counting in clinical guidelines.

Physical activity during pregnancy in normal-weight and obese women: compliance using pedometer assessment.

The objectives of this prospective study were to compare physical activity in 70 normal-weight women with a body mass index (BMI) 20-25 kg/m(2), and 70 obese with a BMI ≥ 30 kg/m(2), before and during pregnancy, and to compare compliance using the pedometer. Physical activity before pregnancy was assessed by questionnaires and during pregnancy by a pedometer worn on 7 consecutive days every 4th week. Obese women were less physically active than normal-weight women both before (p <0.05) and during pregnancy (p <0.0012). Both the compliance and the physical activity gradually declined during gestation. The change in physical activity could be described by a significant interaction between BMI group, gestational age (p <0.007) and the day of the week (p <0.001) when using ANOVA and interaction analysis. Maternal weight gain was larger in the normal-weight than in the obese women, but lower in the non-compliant obese women compared with the compliant (p <0.05).

Insulin analogues and severe hypoglycaemia in type 1 diabetes.

INTRODUCTION: The effect of insulin analogues on glycaemic control is well-documented, whereas the effect on avoidance of severe hypoglycaemia remains tentative. We studied the frequency of severe hypoglycaemia in unselected patients with type 1 diabetes treated with insulin analogues, human insulin, or mixed regimens. METHODS: A questionnaire was posted from six Danish diabetes clinics to 6112 unselected patients with type 1 diabetes and filled in by 3861 patients (63.2%). Primary endpoint was number of episodes of severe hypoglycaemia in the preceding year. Mild hypoglycaemia was also reported. RESULTS: The frequency of severe hypoglycaemic episodes per patient-year in patients receiving long-acting insulin analogues was 1.47±0.18 versus 1.09±0.10 in patients on long-acting human insulin (p=0.01). The frequency of severe hypoglycaemic episodes per patient-year was 1.09±0.11 in patients on short-acting insulin analogues versus 1.26±0.13 in patients on short-acting human insulin (p=0.15), which was statistically significant in an adjusted analysis. CONCLUSIONS: Severe hypoglycaemia is more frequent in patients with type 1 diabetes treated with long-acting insulin analogues. Confounding by indication may be involved. Clinical intervention trials using insulin analogues in patients prone to severe hypoglycaemia are highly needed.

Retinal function in relation to improved glycaemic control in type 1 diabetes.

AIMS/HYPOTHESIS: To study long-term changes in retinal function in response to sustained glycaemia reduction in participants with type 1 diabetes. METHODS: Prospective study using objective measures of retinal function in 17 participants with type 1 diabetes mellitus and minimal to moderate retinopathy who switched from conventional subcutaneous injection to continuous subcutaneous infusion of insulin (CSII). RESULTS: Glycated haemoglobin HbA(1c) gradually decreased from 9.1% at baseline before CSII to 7.4% after 1 year on CSII. Glycaemia was markedly reduced within 1 week after initiation of CSII and remained stable thereafter. Dark adaptation and retinal electroretinographic function at 1, 4 and 16 weeks after initiation of CSII were comparable with baseline values, whereas a significant improvement in rod photoreceptor dark adaptation and dark-adapted b-wave amplitudes were seen after 52 weeks (time to rod-cone break -25% [p < 0.0001], time to a standardised rod intercept -13% [p < 0.0001], dark-adapted rod b-wave full-field amplitude +15% [p = 0.0125], standard combined rod-cone b-wave amplitude +8% [p = 0.049]). No detectable change was observed in cone adaptation, electroretinographic cone function or retinopathy. CONCLUSIONS/INTERPRETATION: After initiation of CSII, the retinal visual pathway of the rods improved with a delay of more than 4 months, over a time scale comparable with the duration of the diabetic retinopathy early worsening response to sustained glycaemia reduction. This indicates that glycaemia has a long-term effect on the disposition of functional capacity in the retinal visual pathway of rod photoreceptors, the cells that appear to be driving the development of diabetic retinopathy.