Cardiovascular risk management among individuals with type 2 diabetes and severe mental illness: a cohort study.

AIMS/HYPOTHESIS: The aim of this study was to compare cardiovascular risk management among people with type 2 diabetes according to severe mental illness (SMI) status. METHODS: We used linked electronic data to perform a retrospective cohort study of adults diagnosed with type 2 diabetes in Scotland between 2004 and 2020, ascertaining their history of SMI from hospital admission records. We compared total cholesterol, systolic BP and HbA1c target level achievement 1 year after diabetes diagnosis, and receipt of a statin prescription at diagnosis and 1 year thereafter, by SMI status using logistic regression, adjusting for sociodemographic factors and clinical history. RESULTS: We included 291,644 individuals with type 2 diabetes, of whom 1.0% had schizophrenia, 0.5% had bipolar disorder and 3.3% had major depression. People with SMI were less likely to achieve cholesterol targets, although this difference did not reach statistical significance for all disorders. However, people with SMI were more likely to achieve systolic BP targets compared to those without SMI, with effect estimates being largest for schizophrenia (men: adjusted OR 1.72; 95% CI 1.49, 1.98; women: OR 1.64; 95% CI 1.38, 1.96). HbA1c target achievement differed by SMI disorder and sex. Among people without previous CVD, statin prescribing was similar or better in those with vs those without SMI at diabetes diagnosis and 1 year later. In people with prior CVD, SMI was associated with lower odds of statin prescribing at diabetes diagnosis (schizophrenia: OR 0.54; 95% CI 0.43, 0.68, bipolar disorder: OR 0.75; 95% CI 0.56, 1.01, major depression: OR 0.92; 95% CI 0.83, 1.01), with this difference generally persisting 1 year later. CONCLUSIONS/INTERPRETATION: We found disparities in cholesterol target achievement and statin prescribing by SMI status. This reinforces the importance of clinical review of statin prescribing for secondary prevention of CVD, particularly among people with SMI.

Risk factors and prediction of hypoglycaemia using the Hypo-RESOLVE cohort: a secondary analysis of pooled data from insulin clinical trials.

AIMS/HYPOTHESIS: The objective of the Hypoglycaemia REdefining SOLutions for better liVES (Hypo-RESOLVE) project is to use a dataset of pooled clinical trials across pharmaceutical and device companies in people with type 1 or type 2 diabetes to examine factors associated with incident hypoglycaemia events and to quantify the prediction of these events. METHODS: Data from 90 trials with 46,254 participants were pooled. Analyses were done for type 1 and type 2 diabetes separately. Poisson mixed models, adjusted for age, sex, diabetes duration and trial identifier were fitted to assess the association of clinical variables with hypoglycaemia event counts. Tree-based gradient-boosting algorithms (XGBoost) were fitted using training data and their predictive performance in terms of area under the receiver operating characteristic curve (AUC) evaluated on test data. Baseline models including age, sex and diabetes duration were compared with models that further included a score of hypoglycaemia in the first 6 weeks from study entry, and full models that included further clinical variables. The relative predictive importance of each covariate was assessed using XGBoost's importance procedure. Prediction across the entire trial duration for each trial (mean of 34.8 weeks for type 1 diabetes and 25.3 weeks for type 2 diabetes) was assessed. RESULTS: For both type 1 and type 2 diabetes, variables associated with more frequent hypoglycaemia included female sex, white ethnicity, longer diabetes duration, treatment with human as opposed to analogue-only insulin, higher glucose variability, higher score for hypoglycaemia across the 6 week baseline period, lower BP, lower lipid levels and treatment with psychoactive drugs. Prediction of any hypoglycaemia event of any severity was greater than prediction of hypoglycaemia requiring assistance (level 3 hypoglycaemia), for which events were sparser. For prediction of level 1 or worse hypoglycaemia during the whole follow-up period, the AUC was 0.835 (95% CI 0.826, 0.844) in type 1 diabetes and 0.840 (95% CI 0.831, 0.848) in type 2 diabetes. For level 3 hypoglycaemia, the AUC was lower at 0.689 (95% CI 0.667, 0.712) for type 1 diabetes and 0.705 (95% CI 0.662, 0.748) for type 2 diabetes. Compared with the baseline models, almost all the improvement in prediction could be captured by the individual's hypoglycaemia history, glucose variability and blood glucose over a 6 week baseline period. CONCLUSIONS/INTERPRETATION: Although hypoglycaemia rates show large variation according to sociodemographic and clinical characteristics and treatment history, looking at a 6 week period of hypoglycaemia events and glucose measurements predicts future hypoglycaemia risk.

Flash monitor initiation is associated with improvements in HbA1c levels and DKA rates among people with type 1 diabetes in Scotland: a retrospective nationwide observational study.

AIMS/HYPOTHESIS: We assessed the real-world effect of flash monitor (FM) usage on HbA1c levels and diabetic ketoacidosis (DKA) and severe hospitalised hypoglycaemia (SHH) rates among people with type 1 diabetes in Scotland and across sociodemographic strata within this population. METHODS: This study was retrospective, observational and registry based. Using the national diabetes registry, 14,682 individuals using an FM at any point between 2014 and mid-2020 were identified. Within-person change from baseline in HbA1c following FM initiation was modelled using linear mixed models accounting for within-person pre-exposure trajectory. DKA and SHH events were captured through linkage to hospital admission and mortality data. The difference in DKA and SHH rates between FM-exposed and -unexposed person-time was assessed among users, using generalised linear mixed models with a Poisson likelihood. In a sensitivity analysis, we tested whether changes in these outcomes were seen in an age-, sex- and baseline HbA1c-matched sample of non-users over the same time period. RESULTS: Prevalence of ever-FM use was 45.9% by mid-2020, with large variations by age and socioeconomic status: 64.3% among children aged <13 years vs 32.7% among those aged ≥65 years; and 54.4% vs 36.2% in the least-deprived vs most-deprived quintile. Overall, the median (IQR) within-person change in HbA1c in the year following FM initiation was -2.5 (-9.0, 2.5) mmol/mol (-0.2 [-0.8, 0.2]%). The change varied widely by pre-usage HbA1c: -15.5 (-31.0, -4.0) mmol/mol (-1.4 [-2.8, -0.4]%) in those with HbA1c > 84 mmol/mol [9.8%] and 1.0 (-2.0, 5.5) mmol/mol (0.1 [-0.2, 0.5]%) in those with HbA1c < 54 mmol/mol (7.1%); the corresponding estimated fold change (95% CI) was 0.77 (0.76, 0.78) and 1.08 (1.07, 1.09). Significant reductions in HbA1c were found in all age bands, sexes and socioeconomic strata, and regardless of prior/current pump use, completion of a diabetes education programme or early FM adoption. Variation between the strata of these factors beyond that driven by differing HbA1c at baseline was slight. No change in HbA1c in matched non-users was observed in the same time period (median [IQR] within-person change = 0.5 [-5.0, 5.5] mmol/mol [0.0 (-0.5, 0.5)%]). DKA rates decreased after FM initiation overall and in all strata apart from the adolescents. Estimated overall reduction in DKA event rates (rate ratio) was 0.59 [95% credible interval (CrI) 0.53, 0.64]) after FM vs before FM initiation, accounting for pre-exposure trend. Finally, among those at higher risk for SHH, estimated reduction in event rates was rate ratio 0.25 (95%CrI 0.20, 0.32) after FM vs before FM initiation. CONCLUSIONS/INTERPRETATION: FM initiation is associated with clinically important reductions in HbA1c and striking reduction in DKA rate. Increasing uptake among the socioeconomically disadvantaged offers considerable potential for tightening the current socioeconomic disparities in glycaemia-related outcomes.

Development and validation of a cardiovascular risk prediction model in type 1 diabetes.

AIMS/HYPOTHESIS: We aimed to report current rates of CVD in type 1 diabetes and to develop a CVD risk prediction tool for type 1 diabetes. METHODS: A cohort of 27,527 people with type 1 diabetes without prior CVD was derived from the national register in Scotland. Incident CVD events during 199,552 person-years of follow-up were ascertained using hospital admissions and death registers. A Poisson regression model of CVD was developed and then validated in the Swedish National Diabetes Register (n = 33,183). We compared the percentage with a high 10 year CVD risk (i.e., ≥10%) using the model with the percentage eligible for statins using current guidelines by age. RESULTS: The age-standardised rate of CVD per 100,000 person-years was 4070 and 3429 in men and women, respectively, with type 1 diabetes in Scotland, and 4014 and 3956 in men and women in Sweden. The final model was well calibrated (Hosmer-Lemeshow test p > 0.05) and included a further 22 terms over a base model of age, sex and diabetes duration (C statistic 0.82; 95% CI 0.81, 0.83). The model increased the base model C statistic from 0.66 to 0.80, from 0.60 to 0.75 and from 0.62 to 0.68 in those aged <40, 40-59 and ≥ 60 years, respectively (all p values <0.005). The model required minimal calibration in Sweden and had a C statistic of 0.85. Under current guidelines, >90% of those aged 20-39 years and 100% of those ≥40 years with type 1 diabetes were eligible for statins, but it was not until age 65 upwards that 100% had a modelled risk of CVD ≥10% in 10 years. CONCLUSIONS/INTERPRETATION: A prediction tool such as that developed here can provide individualised risk predictions. This 10 year CVD risk prediction tool could facilitate patient discussions regarding appropriate statin prescribing. Apart from 10 year risk, such discussions may also consider longer-term CVD risk, the potential for greater benefits from early vs later statin intervention, the potential impact on quality of life of an early CVD event and evidence on safety, all of which could influence treatment decisions, particularly in younger people with type 1 diabetes.

The association of polypharmacy and high-risk drug classes with adverse health outcomes in the Scottish population with type 1 diabetes.

AIMS/HYPOTHESIS: The aim of this work was to map the number of prescribed drugs over age, sex and area-based socioeconomic deprivation, and to examine the association between the number of drugs and particular high-risk drug classes with adverse health outcomes among a national cohort of individuals with type 1 diabetes. METHODS: Utilising linked healthcare records from the population-based diabetes register of Scotland, we identified 28,245 individuals with a diagnosis of type 1 diabetes on 1 January 2017. For this population, we obtained information on health status, predominantly reflecting diabetes-related complications, and information on the total number of drugs and particular high-risk drug classes prescribed. We then studied the association of these baseline-level features with hospital admissions for falls, diabetic ketoacidosis (DKA), and hypoglycaemia or death within the subsequent year using multivariate Cox proportional hazards models. RESULTS: Not considering insulin and treatment for hypoglycaemia, the mean number of prescribed drugs was 4.00 (SD 4.35). The proportion of individuals being prescribed five or more drugs at baseline consistently increased with age (proportion [95% CI]: 0-19 years 2.04% [1.60, 2.49]; 40-49 years 28.50% [27.08, 29.93]; 80+ years 76.04% [67.73, 84.84]). Controlling for age, sex, area-based socioeconomic deprivation and health status, each additional drug at baseline was associated with an increase in the hazard for hospitalisation for falls, hypoglycaemia and death but not for DKA admissions (HR [95% CI]: falls 1.03 [1.01, 1.06]; DKA 1.01 [1.00, 1.03]; hypoglycaemia 1.05 [1.02, 1.07]; death 1.04 [1.02, 1.06]). We found a number of drug classes to be associated with an increased hazard of one or more of these adverse health outcomes, including antithrombotic/anticoagulant agents, corticosteroids, opioids, antiepileptics, antipsychotics, hypnotics and sedatives, and antidepressants. CONCLUSIONS: Polypharmacy is common among the Scottish population with type 1 diabetes and is strongly patterned by sociodemographic factors. The number of prescribed drugs and the prescription of particular high-risk drug classes are strong markers of an increased risk of adverse health outcomes, including acute complications of diabetes.

Marked improvements in glycaemic outcomes following insulin pump therapy initiation in people with type 1 diabetes: a nationwide observational study in Scotland.

AIMS/HYPOTHESIS: Our aim was to assess the use of continuous subcutaneous insulin infusion (CSII) in people with type 1 diabetes in Scotland and its association with glycaemic control, as measured by HbA1c levels, frequency of diabetic ketoacidosis (DKA) and severe hospitalised hypoglycaemia (SHH), overall and stratified by baseline HbA1c. METHODS: We included 4684 individuals with type 1 diabetes from the national Scottish register, who commenced CSII between 2004 and 2019. We presented crude within-person differences from baseline HbA1c over time since initiation, crude DKA and SHH event-rates pre-/post-CSII exposure. We then used mixed models to assess the significance of CSII exposure, taking into account: (1) the diffuse nature of the intervention (i.e. structured education often precedes initiation); (2) repeated within-person measurements; and (3) background time-trends occurring pre-intervention. RESULTS: HbA1c decreased after CSII initiation, with a median within-person change of -5.5 mmol/mol (IQR -12.0, 0.0) (-0.5% [IQR -1.1, 0.0]). Within-person changes were most substantial in those with the highest baseline HbA1c, with median -21.0 mmol/mol (-30.0, -11.0) (-1.9% [-2.7, -1.0]) change in those with a baseline >84 mmol/mol (9.8%) within a year of exposure, that was sustained: -19.0 mmol/mol (-27.6, -6.5) (-1.7% [-2.5, -0.6]) at ≥5 years. Statistical significance and magnitude of change were supported by the mixed models results. The crude DKA event-rate was significantly lower in post-CSII person-time compared with pre-CSII person-time: 49.6 events (95% CI 46.3, 53.1) per 1000 person-years vs 67.9 (64.1, 71.9); rate ratio from Bayesian mixed models adjusting for pre-exposure trend: 0.61 (95% credible interval [CrI] 0.47, 0.77; posterior probability of reduction pp = 1.00). The crude overall SHH event-rate in post-CSII vs pre-CSII person-time was also lower: 17.8 events (95% CI 15.8, 19.9) per 1000 person-years post-exposure vs 25.8 (23.5, 28.3) pre-exposure; rate ratio from Bayesian mixed models adjusting for pre-exposure trend: 0.67 (95% CrI 0.45, 1.01; pp = 0.97). CONCLUSIONS/INTERPRETATION: CSII therapy was associated with marked falls in HbA1c especially in those with high baseline HbA1c. CSII was independently associated with reduced DKA and SHH rates. CSII appears to be an effective option for intensive insulin therapy in people with diabetes for improving suboptimal glycaemic control.

Quantitative levels of serum N-glycans in type 1 diabetes and their association with kidney disease.

We investigated associations of quantitative levels of N-glycans with hemoglobin A1c (HbA1c), renal function and renal function decline in type 1 diabetes. We measured 46 total N-glycan peaks (GPs) on 1565 serum samples from the Scottish Diabetes Research Network Type 1 Bioresource Study (SDRNT1BIO) and a pool of healthy donors. Quantitation of absolute abundance of each GP used 2AB-labeled mannose-3 as a standard. We studied cross-sectional associations of GPs and derived measures with HbA1c, albumin/creatinine ratio (ACR) and estimated glomerular filtration rate (eGFR), and prospective associations with incident albuminuria and final eGFR. All GPs were 1.4 to 3.2 times more abundant in SDRTN1BIO than in the healthy samples. Absolute levels of all GPs were slightly higher with higher HbA1c, with strongest associations for triantennary trigalactosylated disialylated, triantennary trigalactosylated trisialylated structures with core or outer arm fucose, and tetraantennary tetragalactosylated trisialylated glycans. Most GPs showed increased abundance with worsening ACR. Lower eGFR was associated with higher absolute GP levels, most significantly with biantennary digalactosylated disialylated glycans with and without bisect, triantennary trigalactosylated trisialylated glycans with and without outer arm fucose, and core fucosylated biantennary monogalactosylated monosialylated glycans. Although several GPs were inversely associated prospectively with final eGFR, cross-validated multivariable models did not improve prediction beyond clinical covariates. Elevated HbA1c is associated with an altered N-glycan profile in type 1 diabetes. Although we could not establish GPs to be prognostic of future renal function decline independently of HbA1c, further studies to evaluate their impact in the pathogenesis of diabetic kidney disease are warranted.

Time trends in deaths before age 50 years in people with type 1 diabetes: a nationwide analysis from Scotland 2004-2017.

AIMS/HYPOTHESIS: We aimed to examine whether crude mortality and mortality relative to the general population below 50 years of age have improved in recent years in those with type 1 diabetes. METHODS: Individuals with type 1 diabetes aged below 50 and at least 1 year old at any time between 2004 and 2017 in Scotland were identified using the national register. Death data were obtained by linkage to Scottish national death registrations. Indirect age standardisation was used to calculate sex-specific standardised mortality ratios (SMRs). Poisson regression was used to test for calendar-time effects as incidence rate ratios (IRRs). RESULTS: There were 1138 deaths in 251,143 person-years among 27,935 people with type 1 diabetes. There was a significant decline in mortality rate over time (IRR for calendar year 0.983 [95% CI 0.967, 0.998], p = 0.03), but the SMR remained approximately stable at 3.1 and 3.6 in men and 4.09 and 4.16 in women for 2004 and 2017, respectively. Diabetic ketoacidosis or coma (DKAoC) accounted for 22% of deaths and the rate did not decline significantly (IRR 0.975 [95% CI 0.94, 1.011], p = 0.168); 79.3% of DKAoC deaths occurred out of hospital. Circulatory diseases accounted for 27% of deaths and did decline significantly (IRR 0.946 [95% CI 0.914, 0.979], p = 0.002). CONCLUSIONS/INTERPRETATION: Absolute mortality has fallen, but the relative impact of type 1 diabetes on mortality below 50 years has not improved. There is scope to improve prevention of premature circulatory diseases and DKAoC and to develop more effective strategies for enabling people with type 1 diabetes to avoid clinically significant hyper- or hypoglycaemia. Graphical abstract.

Comparison of serum and urinary biomarker panels with albumin/creatinine ratio in the prediction of renal function decline in type 1 diabetes.

AIMS/HYPOTHESIS: We examined whether candidate biomarkers in serum or urine can improve the prediction of renal disease progression in type 1 diabetes beyond prior eGFR, comparing their performance with urinary albumin/creatinine ratio (ACR). METHODS: From the population-representative Scottish Diabetes Research Network Type 1 Bioresource (SDRNT1BIO) we sampled 50% and 25% of those with starting eGFR below and above 75 ml min-1 [1.73 m]-2, respectively (N = 1629), and with median 5.1 years of follow-up. Multiplexed ELISAs and single molecule array technology were used to measure nine serum biomarkers and 13 urine biomarkers based on our and others' prior work using large discovery and candidate studies. Associations with final eGFR and with progression to <30 ml min-1 [1.73] m-2, both adjusted for baseline eGFR, were tested using linear and logistic regression models. Parsimonious biomarker panels were identified using a penalised Bayesian approach, and their performance was evaluated through tenfold cross-validation and compared with using urinary ACR and other clinical record data. RESULTS: Seven serum and seven urine biomarkers were strongly associated with either final eGFR or progression to <30 ml min-1 [1.73 m]-2, adjusting for baseline eGFR and other covariates (all at p<2.3 × 10-3). Of these, associations of four serum biomarkers were independent of ACR for both outcomes. The strongest associations with both final eGFR and progression to <30 ml min-1 [1.73 m]-2 were for serum TNF receptor 1, kidney injury molecule 1, CD27 antigen, α-1-microglobulin and syndecan-1. These serum associations were also significant in normoalbuminuric participants for both outcomes. On top of baseline covariates, the r2 for prediction of final eGFR increased from 0.702 to 0.743 for serum biomarkers, and from 0.702 to 0.721 for ACR alone. The area under the receiver operating characteristic curve for progression to <30 ml min-1 [1.73 m]-2 increased from 0.876 to 0.953 for serum biomarkers, and to 0.911 for ACR alone. Other urinary biomarkers did not outperform ACR. CONCLUSIONS/INTERPRETATION: A parsimonious panel of serum biomarkers easily measurable along with serum creatinine may outperform ACR for predicting renal disease progression in type 1 diabetes, potentially obviating the need for urine testing.

Biomarker panels associated with progression of renal disease in type 1 diabetes.

AIMS/HYPOTHESIS: We aimed to identify a sparse panel of biomarkers for improving the prediction of renal disease progression in type 1 diabetes. METHODS: We considered 859 individuals recruited from the Scottish Diabetes Research Network Type 1 Bioresource (SDRNT1BIO) and 315 individuals from the Finnish Diabetic Nephropathy (FinnDiane) study. All had an entry eGFR between 30 and 75 ml min-1[1.73 m]-2, with those from FinnDiane being oversampled for albuminuria. A total of 297 circulating biomarkers (30 proteins, 121 metabolites, 146 tryptic peptides) were measured in non-fasting serum samples using the Luminex platform and LC electrospray tandem MS (LC-MS/MS). We investigated associations with final eGFR adjusted for baseline eGFR and with rapid progression (a loss of more than 3 ml min-1[1.73 m]-2 year-1) using linear and logistic regression models. Panels of biomarkers were identified using a penalised Bayesian approach, and their performance was evaluated through 10-fold cross-validation and compared with using clinical record data alone. RESULTS: For final eGFR, 16 proteins and 30 metabolites or tryptic peptides showed significant association in SDRNT1BIO, and nine proteins and five metabolites or tryptic peptides in FinnDiane, beyond age, sex, diabetes duration, study day eGFR and length of follow-up (all at p < 10-4). The strongest associations were with CD27 antigen (CD27), kidney injury molecule 1 (KIM-1) and α1-microglobulin. Including the Luminex biomarkers on top of baseline covariates increased the r2 for prediction of final eGFR from 0.47 to 0.58 in SDRNT1BIO and from 0.33 to 0.48 in FinnDiane. At least 75% of the increment in r2 was attributable to CD27 and KIM-1. However, using the weighted average of historical eGFR gave similar performance to biomarkers. The LC-MS/MS platform performed less well. CONCLUSIONS/INTERPRETATION: Among a large set of associated biomarkers, a sparse panel of just CD27 and KIM-1 contains most of the predictive information for eGFR progression. The increment in prediction beyond clinical data was modest but potentially useful for oversampling individuals with rapid disease progression into clinical trials, especially where there is little information on prior eGFR trajectories.

Glycaemic control trends in people with type 1 diabetes in Scotland 2004-2016.

AIMS/HYPOTHESIS: The aim of this work was to examine whether glycaemic control has improved in those with type 1 diabetes in Scotland between 2004 and 2016, and whether any trends differed by sociodemographic factors. METHODS: We analysed records from 30,717 people with type 1 diabetes, registered anytime between 2004 and 2016 in the national diabetes database, which contained repeated measures of HbA1c. An additive mixed regression model was used to estimate calendar time and other effects on HbA1c. RESULTS: Overall, median (IQR) HbA1c decreased from 72 (21) mmol/mol [8.7 (4.1)%] in 2004 to 68 (21) mmol/mol (8.4 [4.1]%) in 2016. However, all of the improvement across the period occurred in the latter 4 years: the regression model showed that the only period of significant change in HbA1c was 2012-2016 where there was a fall of 3 (95% CI 1.82, 3.43) mmol/mol. The largest reductions in HbA1c in this period were seen in children, from 69 (16) mmol/mol (8.5 [3.6]%) to 63 (14) mmol/mol (7.9 [3.4]%), and adolescents, from 75 (25) mmol/mol (9.0 [4.4]%) to 70 (23) mmol/mol (8.6 [4.3]%). Socioeconomic status (according to Scottish Index of Multiple Deprivation) affected the HbA1c values: from the regression model, the 20% of people living in the most-deprived areas had HbA1c levels on average 8.0 (95% CI 7.4, 8.9) mmol/mol higher than those of the 20% of people living in the least-deprived areas. However this difference did not change significantly over time. From the regression model HbA1c was on average 1.7 (95% CI 1.6, 1.8) mmol/mol higher in women than in men. This sex difference did not narrow over time. CONCLUSIONS/INTERPRETATION: In this high-income country, we identified a modest but important improvement in HbA1c since 2012 that was most marked in children and adolescents. These changes coincided with national initiatives to reduce HbA1c including an expansion of pump therapy. However, in most people, overall glycaemic control remains far from target levels and further improvement is badly needed, particularly in those from more-deprived areas.

The effect of dapagliflozin on glycaemic control and other cardiovascular disease risk factors in type 2 diabetes mellitus: a real-world observational study.

AIMS/HYPOTHESIS: Dapagliflozin, a sodium-glucose cotransporter 2 (SGLT2) inhibitor, is indicated for improving glycaemic control in type 2 diabetes mellitus. Whether its effects on HbA1c and other variables, including safety outcomes, in clinical trials are obtained in real-world practice needs to be established. METHODS: We used data from the comprehensive national diabetes register, the Scottish Care Information-Diabetes (SCI-Diabetes) collaboration database, available from 2004 to mid-2016. Data within this database were linked to mortality data from the General Registrar, available from the Information Services Division (ISD) of the National Health Service in Scotland. We calculated crude within-person differences between pre- and post-drug-initiation values of HbA1c, BMI, body weight, systolic blood pressure (SBP) and eGFR. We used mixed-effects regression models to adjust for within-person time trajectories in these measures. For completeness, we evaluated safety outcomes, cardiovascular disease events, lower-limb amputation and diabetic ketoacidosis, focusing on cumulative exposure effects, using Cox proportional hazard models, though power to detect such effects was limited. RESULTS: Among 8566 people exposed to dapagliflozin over a median of 210 days the crude within-person change in HbA1c was -10.41 mmol/mol (-0.95%) after 3 months' exposure. The crude change after 12 months was -12.99 mmol/mol (-1.19%) but considering the expected rise over time in HbA1c gave a dapagliflozin-exposure-effect estimate of -15.14 mmol/mol (95% CI -15.87, -14.41) (-1.39% [95% CI -1.45, -1.32]) at 12 months that was maintained thereafter. A drop in SBP of -4.32 mmHg (95% CI -4.84, -3.79) on exposure within the first 3 months was also maintained thereafter. Reductions in BMI and body weight stabilised by 6 months at -0.82 kg/m2 (95% CI -0.87, -0.77) and -2.20 kg (95% CI -2.34, -2.06) and were maintained thereafter. eGFR declined initially by -1.81 ml min-1 [1.73 m]-2 (95% CI -2.10, -1.52) at 3 months but varied thereafter. There were no significant effects of cumulative drug exposure on safety outcomes. CONCLUSIONS/INTERPRETATION: Dapagliflozin exposure was associated with reductions in HbA1c, SBP, body weight and BMI that were at least as large as in clinical trials. Dapagliflozin also prevented the expected rise in HbA1c and SBP over the period of study.

Persistent C-peptide secretion in Type 1 diabetes and its relationship to the genetic architecture of diabetes.

BACKGROUND: The objective of this cross-sectional study was to explore the relationship of detectable C-peptide secretion in type 1 diabetes to clinical features and to the genetic architecture of diabetes. METHODS: C-peptide was measured in an untimed serum sample in the SDRNT1BIO cohort of 6076 Scottish people with clinically diagnosed type 1 diabetes or latent autoimmune diabetes of adulthood. Risk scores at loci previously associated with type 1 and type 2 diabetes were calculated from publicly available summary statistics. RESULTS: Prevalence of detectable C-peptide varied from 19% in those with onset before age 15 and duration greater than 15 years to 92% in those with onset after age 35 and duration less than 5 years. Twenty-nine percent of variance in C-peptide levels was accounted for by associations with male gender, late age at onset and short duration. The SNP heritability of residual C-peptide secretion adjusted for gender, age at onset and duration was estimated as 26%. Genotypic risk score for type 1 diabetes was inversely associated with detectable C-peptide secretion: the most strongly associated loci were the HLA and INS gene regions. A risk score for type 1 diabetes based on the HLA DR3 and DQ8-DR4 serotypes was strongly associated with early age at onset and inversely associated with C-peptide persistence. For C-peptide but not age at onset, there were strong associations with risk scores for type 1 and type 2 diabetes that were based on SNPs in the HLA region but not accounted for by HLA serotype. CONCLUSIONS: Persistence of C-peptide secretion varies widely in people clinically diagnosed as type 1 diabetes. C-peptide persistence is influenced by variants in the HLA region that are different from those determining risk of early-onset type 1 diabetes. Known risk loci for diabetes account for only a small proportion of the genetic effects on C-peptide persistence.

Prediction of retinopathy progression using deep learning on retinal images within the Scottish screening programme.

BACKGROUND/AIMS: National guidelines of many countries set screening intervals for diabetic retinopathy (DR) based on grading of the last screening retinal images. We explore the potential of deep learning (DL) on images to predict progression to referable DR beyond DR grading, and the potential impact on assigned screening intervals, within the Scottish screening programme. METHODS: We consider 21 346 and 247 233 people with type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM), respectively, each contributing on average 4.8 and 4.4 screening intervals of which 1339 and 4675 intervals concluded with a referable screening episode. Information extracted from fundus images using DL was used to predict referable status at the end of interval and its predictive value in comparison to screening-assigned DR grade was assessed. RESULTS: The DL predictor increased the area under the receiver operating characteristic curve in comparison to a predictor using current DR grades from 0.809 to 0.87 for T1DM and from 0.825 to 0.87 for T2DM. Expected sojourn time-the time from becoming referable to being rescreened-was found to be 3.4 (T1DM) and 2.7 (T2DM) weeks less for a DL-derived policy compared with the current recall policy. CONCLUSIONS: We showed that, compared with using the current retinopathy grade, DL of fundus images significantly improves the prediction of incident referable retinopathy before the next screening episode. This can impact screening recall interval policy positively, for example, by reducing the expected time with referable disease for a fixed workload-which we show as an exemplar. Additionally, it could be used to optimise workload for a fixed sojourn time.

Impact of COVID-19 and Non-COVID-19 Hospitalized Pneumonia on Longer-Term Cardiovascular Mortality in People With Type 2 Diabetes: A Nationwide Prospective Cohort Study From Scotland.

OBJECTIVE: In this study we examine whether hospitalized coronavirus disease 2019 (COVID-19) pneumonia increases long-term cardiovascular mortality more than other hospitalized pneumonias in people with type 2 diabetes and aim to quantify the relative cardiovascular disease (CVD) mortality risks associated with COVID-19 versus non-COVID-19 pneumonia. RESEARCH DESIGN AND METHODS: With use of the SCI-Diabetes register, two cohorts were identified: individuals with type 2 diabetes in 2016 and at the 2020 pandemic onset. Hospital and death records were linked for determination of pneumonia exposure and CVD deaths. Poisson regression estimated rate ratios (RRs) for CVD death associated with both pneumonia types, with adjustment for confounders. Median follow-up durations were 1,461 days (2016 cohort) and 700 days (2020 cohort). RESULTS: The adjusted RR for CVD death following non-COVID-19 pneumonia was 5.51 (95% CI 5.31-5.71) prepandemic and 7.3 (6.86-7.76) during the pandemic. For COVID-19 pneumonia, the RR was 9.13 (8.55-9.75). Beyond 30 days post pneumonia, the RRs converged, to 4.24 (3.90-4.60) for non-COVID-19 and 3.35 (3.00-3.74) for COVID-19 pneumonia, consistent even with exclusion of prior CVD cases. CONCLUSIONS: Hospitalized pneumonia, irrespective of causal agent, marks an increased risk for CVD death immediately and over the long-term. COVID-19 pneumonia poses a higher CVD death risk than other pneumonias in the short-term, but this distinction diminishes over time. These insights underscore the need for including pneumonia in CVD risk assessments, with particular attention to the acute impact of COVID-19 pneumonia.

Ongoing burden and recent trends in severe hospitalised hypoglycaemia events in people with type 1 and type 2 diabetes in Scotland: A nationwide cohort study 2016-2022.

AIMS: We examined severe hospitalised hypoglycaemia (SHH) rates in people with type 1 and type 2 diabetes in Scotland during 2016-2022, stratifying by sociodemographics. METHODS: Using the Scottish National diabetes register (SCI-Diabetes), we identified people with type 1 and type 2 diabetes alive anytime during 2016-2022. SHH events were determined through linkage to hospital admission and death registry data. We calculated annual SHH rates overall and by age, sex, and socioeconomic status. Summary estimates of time and stratum effects were obtained by fitting adjusted generalised additive models using R package mgcv. RESULTS: Rates for those under 20 with type 1 diabetes reached their minimum at the 2020-2021 transition, 30% below the study period average. A gradual decline over time also occurred among 20-49-year-olds with type 1 diabetes. Overall, females had 15% higher rates than males with type 2 diabetes (rate ratio 1.15, 95% CI 1.08-1.22). People in the most versus least deprived quintile experienced 2.58 times higher rates (95% CI 2.27-2.93) in type 1 diabetes and 2.33 times higher (95% CI 2.08-2.62) in type 2 diabetes. CONCLUSIONS: Despite advances in care, SHH remains a significant problem in diabetes. Future efforts must address the large socioeconomic disparities in SHH risks.

Can deep learning on retinal images augment known risk factors for cardiovascular disease prediction in diabetes? A prospective cohort study from the national screening programme in Scotland.

AIMS: This study's objective was to evaluate whether deep learning (DL) on retinal photographs from a diabetic retinopathy screening programme improve prediction of incident cardiovascular disease (CVD). METHODS: DL models were trained to jointly predict future CVD risk and CVD risk factors and used to output a DL score. Poisson regression models including clinical risk factors with and without a DL score were fitted to study cohorts with 2,072 and 38,730 incident CVD events in type 1 (T1DM) and type 2 diabetes (T2DM) respectively. RESULTS: DL scores were independently associated with incident CVD with adjusted standardised incidence rate ratios of 1.14 (P = 3 × 10-04 95 % CI (1.06, 1.23)) and 1.16 (P = 4 × 10-33 95 % CI (1.13, 1.18)) in T1DM and T2DM cohorts respectively. The differences in predictive performance between models with and without a DL score were statistically significant (differences in test log-likelihood 6.7 and 51.1 natural log units) but the increments in C-statistics from 0.820 to 0.822 and from 0.709 to 0.711 for T1DM and T2DM respectively, were small. CONCLUSIONS: These results show that in people with diabetes, retinal photographs contain information on future CVD risk. However for this to contribute appreciably to clinical prediction of CVD further approaches, including exploitation of serial images, need to be evaluated.

Relation of Incident Type 1 Diabetes to Recent COVID-19 Infection: Cohort Study Using e-Health Record Linkage in Scotland.

OBJECTIVE: Studies using claims databases reported that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection >30 days earlier was associated with an increase in the incidence of type 1 diabetes. Using exact dates of diabetes diagnosis from the national register in Scotland linked to virology laboratory data, we sought to replicate this finding. RESEARCH DESIGN AND METHODS: A cohort of 1,849,411 individuals aged <35 years without diabetes, including all those in Scotland who subsequently tested positive for SARS-CoV-2, was followed from 1 March 2020 to 22 November 2021. Incident type 1 diabetes was ascertained from the national registry. Using Cox regression, we tested the association of time-updated infection with incident diabetes. Trends in incidence of type 1 diabetes in the population from 2015 through 2021 were also estimated in a generalized additive model. RESULTS: There were 365,080 individuals who had at least one detected SARS-CoV-2 infection during follow-up and 1,074 who developed type 1 diabetes. The rate ratio for incident type 1 diabetes associated with first positive test for SARS-CoV-2 (reference category: no previous infection) was 0.86 (95% CI 0.62, 1.21) for infection >30 days earlier and 2.62 (95% CI 1.81, 3.78) for infection in the previous 30 days. However, negative and positive SARS-CoV-2 tests were more frequent in the days surrounding diabetes presentation. In those aged 0-14 years, incidence of type 1 diabetes during 2020-2021 was 20% higher than the 7-year average. CONCLUSIONS: Type 1 diabetes incidence in children increased during the pandemic. However, the cohort analysis suggests that SARS-CoV-2 infection itself was not the cause of this increase.

Deep learning detection of diabetic retinopathy in Scotland's diabetic eye screening programme.

BACKGROUND/AIMS: Support vector machine-based automated grading (known as iGradingM) has been shown to be safe, cost-effective and robust in the diabetic retinopathy (DR) screening (DES) programme in Scotland. It triages screening episodes as gradable with no DR versus manual grading required. The study aim was to develop a deep learning-based autograder using images and gradings from DES and to compare its performance with that of iGradingM. METHODS: Retinal images, quality assurance (QA) data and routine DR grades were obtained from national datasets in 179 944 patients for years 2006-2016. QA grades were available for 744 images. We developed a deep learning-based algorithm to detect whether either eye contained ungradable images or any DR. The sensitivity and specificity were evaluated against consensus QA grades and routine grades. RESULTS: Images used in QA which were ungradable or with DR were detected by deep learning with better specificity compared with manual graders (p<0.001) and with iGradingM (p<0.001) at the same sensitivities. Any DR according to the DES final grade was detected with 89.19% (270 392/303 154) sensitivity and 77.41% (500 945/647 158) specificity. Observable disease and referable disease were detected with sensitivities of 96.58% (16 613/17 201) and 98.48% (22 600/22 948), respectively. Overall, 43.84% of screening episodes would require manual grading. CONCLUSION: A deep learning-based system for DR grading was evaluated in QA data and images from 11 years in 50% of people attending a national DR screening programme. The system could reduce the manual grading workload at the same sensitivity compared with the current automated grading system.

Effect of serum sample storage temperature on metabolomic and proteomic biomarkers.

Prospective biomarker studies can be used to identify biomarkers predictive of disease onset. However, if serum biomarkers are measured years after their collection, the storage conditions might affect analyte concentrations. Few data exists concerning which metabolites and proteins are affected by storage at - 20 °C vs - 80 °C. Our objectives were to document analytes affected by storage of serum samples at - 20 °C vs - 80 °C, and to identify those indicative of the storage temperature. We utilized liquid chromatography tandem mass spectrometry and Luminex to quantify 300 analytes from serum samples of 16 Finnish individuals with type 1 diabetes, with split-aliquot samples stored at - 80 °C and - 20 °C for a median of 4.2 years. Results were validated in 315 Finnish and 916 Scottish individuals with type 1 diabetes, stored at - 20 °C and at - 80 °C, respectively. After quality control, we analysed 193 metabolites and proteins of which 120 were apparently unaffected and 15 clearly susceptible to storage at - 20 °C vs - 80 °C. Further, we identified serum glutamate/glutamine ratio greater than 0.20 as a biomarker of storage at - 20 °C vs - 80 °C. The results provide a catalogue of analytes unaffected and affected by storage at - 20 °C vs - 80 °C and biomarkers indicative of sub-optimal storage.