Identifying type 1 and type 2 diabetic cases using administrative data: a tree-structured model

Developing an automated algorithm for identification of children and adolescents with diabetes using electronic health records from the OneFlorida+ clinical research network

AIM

To develop an automated computable phenotype (CP) algorithm for identifying diabetes cases in children and adolescents using electronic health records (EHRs) from the UF Health System.

MATERIALS AND METHODS

The CP algorithm was iteratively derived based on structured data from EHRs (UF Health System 2012-2020). We randomly selected 536 presumed cases among individuals aged <18 years who had (1) glycated haemoglobin levels ≥ 6.5%; or (2) fasting glucose levels ≥126 mg/dL; or (3) random plasma glucose levels ≥200 mg/dL; or (4) a diabetes-related diagnosis code from an inpatient or outpatient encounter; or (5) prescribed, administered, or dispensed diabetes-related medication. Four reviewers independently reviewed the patient charts to determine diabetes status and type.

RESULTS

Presumed cases without type 1 (T1D) or type 2 diabetes (T2D) diagnosis codes were categorized as non-diabetes/other types of diabetes. The rest were categorized as T1D if the most recent diagnosis was T1D, or otherwise categorized as T2D if the most recent diagnosis was T2D. Next, we applied a list of diagnoses and procedures that can determine diabetes type (e.g., steroid use suggests induced diabetes) to correct misclassifications from Step 1. Among the 536 reviewed cases, 159 and 64 had T1D and T2D, respectively. The sensitivity, specificity, and positive predictive values of the CP algorithm were 94%, 98% and 96%, respectively, for T1D and 95%, 95% and 73% for T2D.

CONCLUSION

We developed a highly accurate EHR-based CP for diabetes in youth based on EHR data from UF Health. Consistent with prior studies, T2D was more difficult to identify using these methods.

Experimental approaches for induction of diabetes mellitus and assessment of antidiabetic activity: An in vitro and in vivo methodological review

BACKGROUND

Diabetes mellitus poses a global health challenge, driving the need for innovative therapeutic solutions. Experimental methods play a crucial role in evaluating the efficacy of potential antidiabetic drugs, both in vitro and in vivo. Yet concerns about reproducibility persist, necessitating comprehensive reviews.

OBJECTIVES

This review aims to outline experimental approaches for inducing diabetes and evaluating antidiabetic activity, synthesizing data from authoritative sources and academic literature.

METHODS

We conducted a systematic search of prominent databases, including PubMed, ScienceDirect, and Scopus, to identify relevant articles spanning from 1943 to the present. A total of 132 articles were selected for inclusion in this review, focusing on in vitro and in vivo experimental validations of antidiabetic treatments.

RESULTS

Our review highlights the diverse array of experimental methods employed for inducing diabetes mellitus and evaluating antidiabetic interventions. From cell culture assays to animal models, researchers have employed various techniques to study the effectiveness of novel therapeutic agents.

CONCLUSION

This review provides a comprehensive guide to experimental approaches for assessing antidiabetic activity. By synthesizing data from a range of sources, we offer valuable insights into the current methodologies used in diabetes research. Standardizing protocols and enhancing reproducibility are critical for advancing effective antidiabetic treatments.

Algorithms to define diabetes type using data from administrative databases: A systematic review of the evidence

AIMS

To find the best-performing algorithms to distinguish type 1 and type 2 diabetes in administrative data.

METHODS

Embase and MEDLINE databases were searched from January 2000 until January 2023. Papers evaluating the performance of algorithms to define type 1 and type 2 diabetes by reporting diagnostic metrics against a range of reference standards were selected. Study quality was evaluated using the Quality Assessment of Diagnostic Accuracy Studies.

RESULTS

Of the 24 studies meeting the eligibility criteria, 19 demonstrated a low risk of bias and low concerns about the applicability of the study population across all domains. Algorithms considering multiple diabetes diagnostic codes alone were sensitive and specific approaches to classify diabetes type (both metrics >92.1% for type 1 diabetes; >86.9% for type 2 diabetes). Among the top 10-performing algorithms to detect type 1 and type 2 diabetes, 70% and 100% featured multiple criteria, respectively. Information on insulin use was more sensitive and specific for detecting diabetes type than were criteria based on use of oral hypoglycaemic agents.

CONCLUSIONS

Algorithms based on multiple diabetes diagnostic codes and insulin use are the most accurate approaches to distinguish type 1 from type 2 diabetes using administrative data. Approaches with more than one criterion may also increase sensitivity in distinguishing diabetes type.

Validation of Register-Based Diabetes Classifiers in Danish Data

Purpose

To validate two register-based algorithms classifying type 1 (T1D) and type 2 diabetes (T2D) in a general population using Danish register data.

Patients and Methods

After linking data on prescription drug usage, hospital diagnoses, laboratory results and diabetes-specific healthcare services from nationwide healthcare registers, diabetes type was defined for all individuals in Central Denmark Region age 18-74 years on 31 December 2018 according to two distinct register-based classifiers: 1) a novel register-based diabetes classifier incorporating diagnostic hemoglobin-A1C measurements, the Open-Source Diabetes Classifier (OSDC), and 2) an existing Danish diabetes classifier, the Register for Selected Chronic Diseases (RSCD). These classifications were validated against self-reported data from the Health in Central Denmark survey - overall and stratified by age at onset of diabetes. The source-code of both classifiers was made available in the open-source R package osdc.

Results

A total of 2633 (9.0%) of 29,391 respondents reported having any type of diabetes, divided across 410 (1.4%) self-reported cases of T1D and 2223 (7.6%) cases of T2D. Among all self-reported diabetes cases, 2421 (91.9%) were classified as diabetes cases by both classifiers. In T1D, sensitivity of OSDC-classification was 0.773 [95% CI 0.730-0.813] (RSCD: 0.700 [0.653-0.744]) and positive predictive value (PPV) 0.943 [0.913-0.966] (RSCD: 0.944 [0.912-0.967]). In T2D, sensitivity of OSDC-classification was 0.944 [0.933-0.953] (RSCD: 0.905 [0.892-0.917]) and PPV 0.875 [0.861-0.888] (RSCD: 0.898 [0.884-0.910]). In age at onset-stratified analyses of both classifiers, sensitivity and PPV were low in individuals with T1D onset after age 40 and T2D onset before age 40.

Conclusion

Both register-based classifiers identified valid populations of T1D and T2D in a general population, but sensitivity was substantially higher in OSDC compared to RSCD. Register-classified diabetes type in cases with atypical age at onset of diabetes should be interpreted with caution. The validated, open-source classifiers provide robust and transparent tools for researchers.

Polygenic Scores Help Reduce Racial Disparities in Predictive Accuracy of Automated Type 1 Diabetes Classification Algorithms

OBJECTIVE

Automated algorithms to identify individuals with type 1 diabetes using electronic health records are increasingly used in biomedical research. It is not known whether the accuracy of these algorithms differs by self-reported race. We investigated whether polygenic scores improve identification of individuals with type 1 diabetes.

RESEARCH DESIGN AND METHODS

We investigated two large hospital-based biobanks (Mass General Brigham [MGB] and BioMe) and identified individuals with type 1 diabetes using an established automated algorithm. We performed medical record reviews to validate the diagnosis of type 1 diabetes. We implemented two published polygenic scores for type 1 diabetes (developed in individuals of European or African ancestry). We assessed the classification algorithm before and after incorporating polygenic scores.

RESULTS

The automated algorithm was more likely to incorrectly assign a diagnosis of type 1 diabetes in self-reported non-White individuals than in self-reported White individuals (odds ratio 3.45; 95% CI 1.54-7.69; P = 0.0026). After incorporating polygenic scores into the MGB Biobank, the positive predictive value of the type 1 diabetes algorithm increased from 70 to 97% for self-reported White individuals (meaning that 97% of those predicted to have type 1 diabetes indeed had type 1 diabetes) and from 53 to 100% for self-reported non-White individuals. Similar results were found in BioMe.

CONCLUSIONS

Automated phenotyping algorithms may exacerbate health disparities because of an increased risk of misclassification of individuals from underrepresented populations. Polygenic scores may be used to improve the performance of phenotyping algorithms and potentially reduce this disparity.

Identifying type 1 and 2 diabetes in research datasets where classification biomarkers are unavailable: assessing the accuracy of published approaches

OBJECTIVES

We aimed to compare the performance of approaches for classifying insulin-treated diabetes within research datasets without measured classification biomarkers, evaluated against two independent biological definitions of diabetes type.

STUDY DESIGN AND SETTING

We compared accuracy of ten reported approaches for classifying insulin-treated diabetes into type 1 (T1D) and type 2 (T2D) diabetes in two cohorts: UK Biobank (UKBB) n = 26,399 and Diabetes Alliance for Research in England (DARE) n = 1,296. The overall performance for classifying T1D and T2D was assessed using: a T1D genetic risk score and genetic stratification method (UKBB); C-peptide measured at >3 years diabetes duration (DARE).

RESULTS

Approaches' accuracy ranged from 71% to 88% (UKBB) and 68% to 88% (DARE). When classifying all participants, combining early insulin requirement with a T1D probability model (incorporating diagnosis age and body image issue [BMI]), and interview-reported diabetes type (UKBB available in only 15%) consistently achieved high accuracy (UKBB 87% and 87% and DARE 85% and 88%, respectively). For identifying T1D with minimal misclassification, models with high thresholds or young diagnosis age (<20 years) had highest performance. Findings were incorporated into an online tool identifying optimum approaches based on variable availability.

CONCLUSION

Models combining continuous features with early insulin requirement are the most accurate methods for classifying insulin-treated diabetes in research datasets without measured classification biomarkers.

The Impact of Mandatory Nonmedical Switching From Originator to Biosimilar Insulin Glargine

PURPOSE

This study monitors for early changes in health services utilization after a mandatory policy to switch patients from originator to biosimilar insulin glargine in British Columbia, Canada.

METHODS

We conducted a prospective cohort study of patients treated with originator insulin glargine. The policy cohort included patients treated with originator insulin glargine in the 6 months before the policy change (May 27, 2019). Three historical control cohorts included users of originator insulin glargine during the 6 months before May 27 each year in 2016, 2017, and 2018. Patients who discontinued or switched use of the originator insulin glargine and those without cost coverage by the provincial drug plan were excluded. Using likelihood ratios, we compared the daily use of medications, outpatient visits, and hospitalizations in the 12 months after the policy change with the daily use in 3 historical control cohorts. A sustained likelihood ratio above a predefined threshold of 7.1 was interpreted as an early signal of a possible policy impact.

FINDINGS

Each cohort included 15,344 to 17,310 patients. In the first year of the policy, we observed increases in (1) insulin glargine use (the cumulative incidence increased by 2.5% compared with the mean of the 3 historical cohorts), (2) oral antidiabetic medication use (increased by 2.8%), and (3) outpatient visits (increased by 1.4%). Likelihood ratios greater than the threshold of 7.1 were detected for these 3 outcomes.

IMPLICATIONS

We observed marginal changes in health services utilization without detecting signals of negative health impacts on patients targeted by the British Columbia policy of mandatory switching from originator to biosimilar insulin glargine.

Newer long-acting insulin prescriptions to type 2 diabetes patients: prevalence and practice variation

Background Little is known about the prescription of expensive non-recommended newer long-acting insulins (glargine 300 U/ml and degludec) for type 2 diabetes mellitus (T2DM) patients. Aim To identify practice variation in and practice and patient-related characteristics associated with the prescription of newer long-acting insulins to T2DM patients in primary care. Design and Setting Retrospective cohort study in Dutch general practices (Nivel Primary Care Database). Method The first prescription for intermediate or long-acting insulins in 2018 was identified for patients aged ≥40 using other T2DM drugs. Per practice, the median percentage and interquartile range (IQR) of patients with newer insulin prescriptions were calculated. Multilevel logistic regression models were constructed to calculate intraclass correlation coefficients (ICC) and quantify the association of patient and practice characteristics with prescriptions for newer insulins (odds ratio’s (OR) and 95% confidence intervals (CI)). Results 7,757 patients with prescriptions for intermediate or long-acting insulins from 282 general practices were identified. A median percentage of 21.2% (IQR=12.5-36.4%) of all patients prescribed intermediate or long-acting insulins per practice received a prescription for newer insulins. After multilevel modelling, the ICC decreased from 20% to 19%. Female sex (OR=0.77;95%CI=0.69–0.87), age ≥86 years compared to 40-55 years (OR=0.22;95%CI=0.15-0.34), prescriptions for metformin (OR=0.66;95%CI=0.53-0.82), sulphonylurea (OR=0.58;95%CI=0.51-0.66) or other newer T2DM drugs (OR=3.10;95%CI=2.63-3.66) and dispensing practices (OR=1.78;95%CI=1.03-3.10) were associated with the prescription of newer insulins. Conclusion The interpractice variation in the prescription of newer insulins is large and could only be partially explained by patient and practice related differences. This indicates substantial opportunities for improvement.

A clinical diagnostic model based on an eXtreme Gradient Boosting algorithm to distinguish type 1 diabetes

BACKGROUND

Accurate classification of type 1 diabetes (T1DM) and type 2 diabetes (T2DM) in the early phase is crucial for individual precision treatment. This study aimed to develop a classification model having fewer and easier to access clinical variables to distinguish T1DM in newly diagnosed diabetes in adults.

METHODS

Clinical and laboratory data were collected from 15,206 adults with newly diagnosed diabetes in this cross-sectional study. This cohort represented 20 provinces and 4 municipalities in China. Types of diabetes were determined based on postprandial C-peptide (PCP) level and glutamic acid decarboxylase autoantibody (GADA) titer. We developed multivariable clinical diagnostic models using the eXtreme Gradient Boosting (XGBoost) algorithm. Classification variables included in the final model were based on their scores of importance. Model performance was evaluated by area under the receiver operating characteristic curve (ROC AUC), sensitivity, and specificity. The performance of models with different variable combinations was compared. Calibration intercept and slope were evaluated for the final model.

RESULTS

Among the newly diagnosed diabetes cohort, 1,465 (9.63%) persons had T1DM and 13,741 (90.37%) had T2DM. Body mass index (BMI) contributed the most to the model, followed by age of onset and hemoglobin A1c (HbA1c). Compared with models with other clinical variable combinations, a final model that integrated age of onset, BMI and HbA1c had relatively higher performance. The ROC AUC, sensitivity, and specificity for this model were 0.83 (95% CI, 0.80 to 0.85), 0.77, and 0.76, respectively. The calibration intercept and slope were 0.02 (95% CI, -0.03 to 0.06) and 0.90 (95% CI, 0.79 to 1.02), respectively, which suggested a good calibration performance.

CONCLUSIONS

Our classification model that integrated age of onset, BMI, and HbA1c could distinguish T1DM from T2DM, which provides a useful tool in assisting physicians in subtyping and precising treatment in diabetes.

Validation of a type 1 diabetes algorithm using electronic medical records and administrative healthcare data to study the population incidence and prevalence of type 1 diabetes in Ontario, Canada

INTRODUCTION

We aimed to develop algorithms distinguishing type 1 diabetes (T1D) from type 2 diabetes in adults ≥18 years old using primary care electronic medical record (EMRPC) and administrative healthcare data from Ontario, Canada, and to estimate T1D prevalence and incidence.

RESEARCH DESIGN AND METHODS

The reference population was a random sample of patients with diabetes in EMRPC whose charts were manually abstracted (n=5402). Algorithms were developed using classification trees, random forests, and rule-based methods, using electronic medical record (EMR) data, administrative data, or both. Algorithm performance was assessed in EMRPC. Administrative data algorithms were additionally evaluated using a diabetes clinic registry with endocrinologist-assigned diabetes type (n=29 371). Three algorithms were applied to the Ontario population to evaluate the minimum, moderate and maximum estimates of T1D prevalence and incidence rates between 2010 and 2017, and trends were analyzed using negative binomial regressions.

RESULTS

Of 5402 individuals with diabetes in EMRPC, 195 had T1D. Sensitivity, specificity, positive predictive value and negative predictive value for the best performing algorithms were 80.6% (75.9-87.2), 99.8% (99.7-100), 94.9% (92.3-98.7), and 99.3% (99.1-99.5) for EMR, 51.3% (44.0-58.5), 99.5% (99.3-99.7), 79.4% (71.2-86.1), and 98.2% (97.8-98.5) for administrative data, and 87.2% (81.7-91.5), 99.9% (99.7-100), 96.6% (92.7-98.7) and 99.5% (99.3-99.7) for combined EMR and administrative data. Administrative data algorithms had similar sensitivity and specificity in the diabetes clinic registry. Of 11 499 711 adults in Ontario in 2017, there were 24 789 (0.22%, minimum estimate) to 102 140 (0.89%, maximum estimate) with T1D. Between 2010 and 2017, the age-standardized and sex-standardized prevalence rates per 1000 person-years increased (minimum estimate 1.7 to 2.56, maximum estimate 7.48 to 9.86, p<0.0001). In contrast, incidence rates decreased (minimum estimate 0.1 to 0.04, maximum estimate 0.47 to 0.09, p<0.0001).

CONCLUSIONS

Primary care EMR and administrative data algorithms performed well in identifying T1D and demonstrated increasing T1D prevalence in Ontario. These algorithms may permit the development of large, population-based cohort studies of T1D.

Use of a Patient Portal for Engaging Patients with Type 2 Diabetes: Patterns and Prediction

Background: Patient portals empower patients by providing access to their health information and facilitating communication with care providers. This study aimed to examine the usage patterns of a patient portal offered as part of an electronic health record and to identify predictors of portal use among patients with type 2 diabetes (T2DM). Methods: A 2-year retrospective cohort study was performed using outpatient data from the health care system and its patient portal. Demographic and clinical data from 38,399 T2DM patients were analyzed. Descriptive statistics were used to summarize portal usage patterns. Binary logistic regression was employed to examine predictors and two-way interactions associated with portal use. Results: Almost one-third of patients (n = 12,615; 32.9%, 95% confidence interval: [32.38%-33.32%]) had used the portal for a mean 2.5 ± 1.9 years before the study period. Portal use was higher on weekdays than on weekends (P < 0.001). An increase in portal use was observed in response to e-mail reminders. A nonlinear relationship between age and portal use was observed and depended on several other predictors (P's < 0.05). Patients living in more rural areas with low income were at lower odds to use the portal (P = 0.021), and this finding also applied to nonwhites with low income (P < 0.001). More chronic conditions and a higher initial glycated hemoglobin value were associated with portal use (P = 0.014). Conclusions: The patient portal usage remained relatively stable over the 2-year period. A combination of factors was associated with an individual's patient portal use. Patient engagement in portal use can be facilitated through a proactive approach by health care providers.

Association Between Clinical Practice Group Adherence to Quality Measures and Adverse Outcomes Among Adult Patients With Diabetes

IMPORTANCE

Clinical practice group performance on quality measures associated with chronic disease management has become central to reimbursement. Therefore, it is important to determine whether commonly used process and disease control measures for chronic conditions correlate with utilization-based outcomes, as they do in acute disease.

OBJECTIVE

To examine the associations among clinical practice group performance on diabetes quality measures, including process measures, disease control measures, and utilization-based outcomes.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective, cross-sectional analysis examined commercial claims data from a national health insurance plan. A cohort of eligible beneficiaries with diabetes aged 18 to 65 years who were enrolled for at least 12 months from January 1, 2010, through December 31, 2014, was defined. Eligible beneficiaries were attributed to a clinical practice group based on the plurality of their primary care or endocrinology office visits. Data were analyzed from October 1, 2018, through April 30, 2019.

MAIN OUTCOMES AND MEASURES

For each clinical practice group, performance on current diabetes quality measures included 3 process measures (2 testing measures [hemoglobin A1c {HbA1c} and low-density lipoprotein {LDL} testing] and 1 drug use measure [statin use]) and 2 disease control measures (HbA1c <8% and LDL level <100 mg/dL). The rates of utilization-based outcomes, including hospitalization for diabetes and major adverse cardiovascular events (MACEs), were also measured.

RESULTS

In this cohort of 652 258 beneficiaries with diabetes from 886 clinical practice groups, 42.9% were aged 51 to 60 years, and 52.6% were men. Beneficiaries lived in areas that were predominantly white (68.1%). At the clinical practice group level, except for high correlation between the 2 testing measures, correlations among different quality measures were weak (r range, 0.010-0.244). Rate of HbA1c of less than 8% had the strongest correlation with hospitalization for MACE (r = -0.046; P = .03) and diabetes (r = -0.109; P < .001). Rates of HbA1c control at the clinical practice group level were not significantly associated with likelihood of hospitalization at the individual level. Performance on the process and disease control measures together explained 3.9% of the variation in the likelihood of hospitalization for a MACE or diabetes at the individual level.

CONCLUSIONS AND RELEVANCE

In this study, performance on utilization-based measures-intended to reflect the quality of chronic disease management-was only weakly associated with direct measures of chronic disease management, namely, disease control measures. This correlation should be considered when determining the degree of financial emphasis to place on hospitalization rates as a measure of quality in treatment of chronic diseases.

Construction of a flow chart-like risk prediction model of ganciclovir-induced neutropaenia including severity grade: A data mining approach using decision tree

WHAT IS KNOWN AND OBJECTIVE

Haematological toxicities such as neutropaenia are a common side effect of ganciclovir (GCV); however, risk factors for GCV-induced neutropaenia have not been well established. Decision tree (DT) analysis is a typical technique of data mining consisting of a flow chart-like framework that shows various outcomes from a series of decisions. By following the flow chart, users can estimate combinations of risk factors that may increase the probability of certain events. In our previous study, we demonstrated the usefulness of this approach in the evaluation of adverse drug reactions. Therefore, we aimed to construct a risk prediction model of GCV-induced neutropaenia including severity grade.

METHODS

We performed a retrospective study at the Hokkaido University Hospital and enrolled patients who received GCV between April 2008 and March 2018. Neutropaenia was defined as an absolute neutrophil count (ANC) <1500 cells/mm³ and a decrease to <75% relative to baseline. We classified the patients who developed neutropaenia in three groups (Grades 2-4) based on the National Cancer Institute-Common Terminology Criteria for Adverse Events. Data collection was achieved through the retrieval of medical records. We employed a chi-squared automatic interaction detection algorithm to construct the DT model and compared the accuracies to the logistic regression model (a conventional statistical method) to evaluate the established model.

RESULTS AND DISCUSSION

In total, 396 adult patients were included in the study; 61 (15.4%) developed neutropaenia. Three predictive factors (hematopoietic stem cell transplantation, baseline ANC <3854 cells/mm³ and duration of therapy ≥15 days) were extracted using the DT analysis to produce five subgroups, the incidence of neutropaenia ranged between 1.7% and 52.8%. In each subgroup, patients who developed neutropaenia were categorized based on the severity. The accuracies of each model were the same (84.6%), which indicated precision.

WHAT IS NEW AND CONCLUSION

We successfully built a risk prediction model of GCV-induced neutropaenia including severity grade. This model is expected to assist decision-making in the clinical setting.

Construction of a risk prediction model of vancomycin-associated nephrotoxicity to be used at the time of initial therapeutic drug monitoring: A data mining analysis using a decision tree model

OBJECTIVES

In our previous study, we built a risk prediction model of vancomycin (VCM)-associated nephrotoxicity using decision tree (DT) analysis. However, this has several limitations in clinical applications. Our objective here was to construct a clinically applicable risk prediction model to be used at the time of initial therapeutic drug monitoring (TDM), in patients with uncomplicated infections.

METHOD

A retrospective study was conducted at Hokkaido University Hospital. Subjects that had received VCM were extracted between November 2011 and April 2017. Nephrotoxicity was defined as an increase in serum creatinine of 0.5 mg/dL or 50% or higher from baseline. The additional inclusion criteria in this study were as follows: (1) the target trough level of VCM was set to 10 to 15 mg/L, and (2) the duration of therapy was 7 to 14 days. Patients were assumed to have uncomplicated infections. Risk factors for nephrotoxicity were evaluated, which could be extracted at the initial TDM. In the DT analysis, a chi-squared automatic interaction detection algorithm was constructed.

RESULTS

A total of 402 patients were enrolled, and 56 (13.9%) patients developed nephrotoxicity. In the DT analysis, concomitant medications (furosemide, piperacillin-tazobactam, and vasopressor drugs) and an initial VCM trough concentration ≥ 15.0 mg/L were extracted as predictive variables by which patients were divided into six subgroups. The incidence of nephrotoxicity was 5.2% to 70.0%, with subgroups classified as low to high risk of nephrotoxicity. The accuracy of DT model was favourable (87.1%).

CONCLUSION

We propose that the DT model built in this study is applicable to clinical practice.

Insulin Glargine U100 Utilization in Patients with Type 2 Diabetes in an Italian Real-World Setting: A Retrospective Study

BACKGROUND

This study is aimed at estimating the proportion of type 2 diabetes mellitus (T2DM) patients treated with basal insulin (insulin glargine U100) and at evaluating daily insulin dose, treatment pattern, and adherence to treatment of these patients.

METHODS

Data from administrative and laboratory databases of 3 Italian Local Health Units were retrospectively collected and analyzed. All patients with a diagnosis of T2DM between 01/01/2012 and 31/12/2012 were considered, and those with at least a prescription of insulin glargine between 01/01/2013 and 31/12/2014 were included and followed up for one year. For each patient, we evaluated HbA1c levels both at baseline and during the follow-up period and the daily average dose of insulin. Medication adherence was defined by using medication possession ratio (MPR) and reported as proportion of patients with MPR ≥ 80%.

RESULTS

7,422 T2DM patients were available for the study. According to the antidiabetic medication prescribed, patients were categorized into four groups: insulin glargine only, insulin glargine plus oral glucose-lowering drugs, insulin glargine plus rapid-acting insulin, and insulin glargine plus DPP-4 inhibitors. Median daily dose of insulin among insulin glargine only patients was higher than in other groups (35 IU vs. 20 IU, p < 0.05), and a higher percentage of them achieved a target HbA1c value of less than 7.0% (53.8% vs. 30%, p < 0.001). Adherence to insulin treatment was lowest (41%) in the insulin glargine only group compared to other groups (ranging from 58.4% to 64.4%), p < 0.001.

CONCLUSIONS

A large proportion of T2DM patients treated with insulin fail in achieving the glycemic target of HbA1c level < 7%, irrespective of treatment regimen; however, basal insulin only is associated with lower therapeutic unsuccess. Adherence to antidiabetes medications is also suboptimal in these patients and should be addressed to improve long-term outcomes of reducing and preventing microvascular and macrovascular complications.

Validation of an algorithm for identifying type 1 diabetes in adults based on electronic health record data

PURPOSE

Algorithms using information from electronic health records to identify adults with type 1 diabetes have not been well studied. Such algorithms would have applications in pharmacoepidemiology, drug safety research, clinical trials, surveillance, and quality improvement. Our main objectives were to determine the positive predictive value for identifying type 1 diabetes in adults using a published algorithm (developed by Klompas et al) and to compare it to a simple requirement that the majority of diabetes diagnosis codes be type 1.

METHODS

We applied the Klompas algorithm and the diagnosis code criterion to a cohort of 66 690 adult Kaiser Permanente Colorado members with diabetes. We reviewed 220 charts of those identified as having type 1 diabetes and calculated positive predictive values.

RESULTS

The Klompas algorithm identified 3286 (4.9% of 66 690) adults with diabetes as having type 1 diabetes. Based on chart reviews, the overall positive predictive value was 94.5%. The requirement that the majority of diabetes diagnosis codes be type 1 identified 3000 (4.5%) as having type 1 diabetes and had a positive predictive value of 96.4%. However, the algorithm criterion involving dispensing of urine acetone test strips performed poorly, with a positive predictive value of 20.0%.

CONCLUSIONS

Data from electronic health records can be used to accurately identify adults with type 1 diabetes. When identifying adults with type 1 diabetes, we recommend either a modified version of the Klompas algorithm without the urine acetone test strips criterion or the requirement that the majority of diabetes diagnosis codes be type 1 codes.

Emerging Approaches in Surveillance of Type 1 Diabetes

PURPOSE OF REVIEW

Surveillance of type 1 diabetes provides an opportunity to address public health needs, inform etiological research, and plan health care services. We present issues in type 1 diabetes surveillance, review previous and current methods, and present new initiatives.

RECENT FINDINGS

Few diabetes surveillance systems distinguish between type 1 and type 2 diabetes. Most worldwide efforts have focused on registries and ages < 15 years, resulting in limited information among adults. Recently, surveillance includes use of electronic health information and national health surveys. However, distinguishing by diabetes type remains a challenge. Enhancing and improving surveillance of type 1 diabetes across all age groups could include validating questions for use in national health surveys. In addition, validated algorithms for classifying diabetes type in electronic health records could further improve surveillance efforts and close current gaps in our understanding of the epidemiology of type 1 diabetes.

Influence of Diabetes on Trends in Perioperative Cardiovascular Events

OBJECTIVE

Patients undergoing noncardiac surgery frequently have diabetes mellitus (DM) and an elevated risk of cardiovascular disease. It is unknown whether temporal declines in the frequency of perioperative major adverse cardiovascular and cerebrovascular events (MACCEs) apply to patients with DM.

RESEARCH DESIGN AND METHODS

Patients ≥45 years of age who underwent noncardiac surgery from January 2004 to December 2013 were identified using the U.S. National Inpatient Sample. DM was identified using ICD-9 diagnosis codes. Perioperative MACCEs (in-hospital all-cause mortality, acute myocardial infarction, or acute ischemic stroke) by DM status were evaluated over time.

RESULTS

The final study sample consisted of 10,581,621 hospitalizations for major noncardiac surgery; DM was present in ∼23% of surgeries and increased over time (P for trend <0.001). Patients with DM experienced MACCEs in 3.3% of surgeries vs. 2.8% of surgeries for patients without DM (P < 0.001). From 2004 to 2013, the odds of perioperative MACCEs after multivariable adjustment increased by 6% (95% CI 2-9) for DM patients, compared with an 8% decrease (95% CI -10 to -6) for patients without DM (P for interaction <0.001). Trends for individual end points were all less favorable for patients with DM versus those without DM.

CONCLUSIONS

In an analysis of >10.5 million noncardiac surgeries from a large U.S. hospital admission database, perioperative MACCEs were more common among patients with DM versus those without DM. Perioperative MACCEs increased over time and individual end points were all less favorable for patients with DM. Our findings suggest that a substantial unmet need exists for strategies to reduce the risk of perioperative cardiovascular events among patients with DM.

Usefulness of a decision tree model for the analysis of adverse drug reactions: Evaluation of a risk prediction model of vancomycin-associated nephrotoxicity constructed using a data mining procedure

OBJECTIVES

Several publications concerning decision tree (DT) analysis in medical fields have recently demonstrated its usefulness for defining prognostic factors in various diseases. However, there are minimal reports on the predictors of adverse drug reactions. We attempted to use DT analysis to discover combinations of multiple risk factors that would increase the risk of nephrotoxicity associated with vancomycin (VCM). To demonstrate the usefulness of DT analysis, we compared its predictive performance with that of multiple logistic regression analysis.

METHOD

A single-centre, retrospective study was conducted at Hokkaido University Hospital. A total of 592 patients, who received intravenous administrations of VCM between November 2011 and April 2016, were enrolled. Nephrotoxicity was defined as an increase in serum creatinine of ≥0.5 mg/dL or a ≥50% increase in serum creatinine from the baseline. Risk factors for VCM nephrotoxicity were extracted from previous reports. In the DT analysis, a chi-squared automatic interaction detection algorithm was constructed. For evaluating the established algorithms, a 10-fold cross validation method was adopted to calculate the misclassification risk of the model. Moreover, to compare the accuracy of the DT analysis, multiple logistic regression analysis was conducted.

RESULTS

Eighty-seven (14.7%) patients developed nephrotoxicity. A VCM trough concentration of ≥15.0 mg/L, concomitant medication (vasopressor drugs and furosemide), and a duration of therapy ≥14 days were extracted to build the DT model, in which the patients were divided into 6 subgroups based on variable rates of nephrotoxicity, ranging from 4.6 to 69.6%. The predictive accuracies of the DT and logistic regression models were similar (87.3%, respectively), indicating that they were accurate.

CONCLUSION

This study suggests the usefulness of DT models for the evaluation of adverse drug reactions.

Lower incidence of macrovascular complications in patients on insulin glargine versus those on basal human insulins: a population-based cohort study in Italy

BACKGROUND AND AIM

The aim of this study was to compare the use of insulin glargine and intermediate/long-acting human insulin (HI) in relation to the incidence of complications in diabetic patients.

METHODS AND RESULTS

A population-based cohort study was conducted using administrative data from four local health authorities in the Abruzzo Region (900,000 inhabitants). Diabetic patients without macrovascular diseases and treated with either intermediate/long-acting HI or glargine were followed for 3-years; the incidence of diabetic (macrovascular, microvascular and metabolic) complications was ascertained by hospital discharge claims and estimated using Cox proportional hazard models. Propensity score (PS) matching was also used to adjust for significant differences in the baseline characteristics between the two groups.

RESULTS

Overall, 1921 diabetic patients were included: 744 intermediate/long-acting HI and 1177 glargine users. During the 3-year follow-up, 209 (28.1%) incident events of any diabetic complication occurred in the intermediate/long-acting HI and 159 (13.5%) in the glargine group. After adjustment for covariates, glargine users had an HR (95% CI) of 0.57 (0.44-0.74) for any diabetic complication and HRs of 0.61 (0.44-0.84), 0.58 (0.33-1.04) and 0.35 (0.18-0.70) for macrovascular, microvascular and metabolic complications, respectively, compared to intermediate/long-acting HI users. PS analyses supported these findings.

CONCLUSIONS

The use of glargine is associated with a lower risk of macrovascular complications compared with traditional basal insulins. However, limitations inherent to the study design including the short length of observation and the lack of data on metabolic control or diabetes duration, do not allow us to consider this association as a proof of causality.

Validation of pediatric diabetes case identification approaches for diagnosed cases by using information in the electronic health records of a large integrated managed health care organization

We explored the utility of different algorithms for diabetes case identification by using electronic health records. Inpatient and outpatient diagnosis codes, as well as data on laboratory results and dispensing of antidiabetic medications were extracted from electronic health records of Kaiser Permanente Southern California members who were less than 20 years of age in 2009. Diabetes cases were ascertained by using the SEARCH for Diabetes in Youth Study protocol and comprised the "gold standard." Sensitivity, specificity, positive and negative predictive values, accuracy, and the area under the receiver operating characteristic curve (AUC) were compared in 1,000 bootstrapped samples. Based on data from 792,992 youth, of whom 1,568 had diabetes (77.2%, type 1 diabetes; 22.2%, type 2 diabetes; 0.6%, other), case identification accuracy was highest in 75% of bootstrapped samples for those who had 1 or more outpatient diabetes diagnoses or 1 or more insulin prescriptions (sensitivity, 95.9%; positive predictive value, 95.5%; AUC, 97.9%) and in 25% of samples for those who had 2 or more outpatient diabetes diagnoses and 1 or more antidiabetic medications (sensitivity, 92.4%; positive predictive value, 98.4%; AUC, 96.2%). Having 1 or more outpatient type 1 diabetes diagnoses (International Classification of Diseases, Ninth Revision, Clinical Modification, code 250.x1 or 250.x3) had the highest accuracy (94.4%) and AUC (94.1%) for type 1 diabetes; the absence of type 1 diabetes diagnosis had the highest accuracy (93.8%) and AUC (93.6%) for identifying type 2 diabetes. Information in the electronic health records from managed health care organizations provides an efficient and cost-effective source of data for childhood diabetes surveillance.

Lower rate of cardiovascular complications in patients on bolus insulin analogues: a retrospective population-based cohort study

BACKGROUND

Few studies are available evaluating the impact of rapid-acting insulin analogues on long-term diabetes outcomes. Our aim was to compare the use of rapid-acting insulin analogues versus human regular insulin in relation to the occurrence of diabetic complications in a cohort of diabetic patients through the analysis of administrative databases.

METHODS

A population-based cohort study was conducted using administrative data from four local health authorities in the Abruzzo Region (900,000 inhabitants). Diabetic patients free of macrovascular disease at baseline and treated either with human regular insulin or rapid-acting insulin analogues were followed for a maximum of 3 years. The incidence of diabetic complications was ascertained by hospital discharge claims. Hazard ratios (HRs) and 95% CIs of any diabetic complication and macrovascular, microvascular and metabolic complications were estimated separately using Cox proportional hazard models adjusted for patients' characteristics and anti-diabetic drug use. Propensity score matching was also used to adjust for significant difference in the baseline characteristics between the two treatment groups.

RESULTS

A total of 2,286 patients were included: 914 receiving human regular insulin and 1,372 rapid-acting insulin analogues. During the follow-up, 286 (31.3%) incident events occurred in the human regular insulin group and 235 (17.1%) in the rapid-acting insulin analogue group. After propensity score-based matched-pair analyses, rapid-acting insulin analogues users had a HR of 0.73 (0.58-0.92) for any diabetes-related complication and HRs of 0.73 (0.55-0.93) and 0.55 (0.32-0.96) for macrovascular and metabolic complications respectively, as compared with human regular insulin users. No difference between the two groups was found for microvascular complications.

CONCLUSIONS

Our findings suggest that the use of rapid-acting insulin analogues is associated with a lower risk of cardiovascular and metabolic complications compared with human regular insulin use.

Validation of administrative data sources for endoscopy utilization in colorectal cancer diagnosis

Background

Validation of administrative data is important to assess potential sources of bias in outcome evaluation and to prevent dissemination of misleading or inaccurate information. The purpose of the study was to determine the completeness and accuracy of endoscopy data in several administrative data sources in the year prior to colorectal cancer diagnosis as part of a larger project focused on evaluating the quality of pre-diagnostic care.

Methods

Primary and secondary data sources for endoscopy were collected from the Alberta Cancer Registry, cancer medical charts and three different administrative data sources. 1672 randomly sampled patients diagnosed with invasive colorectal cancer in years 2000–2005 in Alberta, Canada were included. A retrospective validation study of administrative data for endoscopy in the year prior to colorectal cancer diagnosis was conducted. A gold standard dataset was created by combining all the datasets. Number and percent identified, agreement and percent unique to a given data source were calculated and compared across each dataset and to the gold standard with respect to identifying all patients who underwent endoscopy and all endoscopies received by those patients.

Results

The combined administrative data and physician billing data identified as high or higher percentage of patients who had one or more endoscopy (84% and 78%, respectively) and total endoscopy procedures (89% and 81%, respectively) than the chart review (78% for both).

Conclusions

Endoscopy data has a high level of completeness and accuracy in physician billing data alone. Combined with hospital in/outpatient data it is more complete than chart review alone.

Detection of pelvic inflammatory disease: development of an automated case-finding algorithm using administrative data

ICD-9 codes are conventionally used to identify pelvic inflammatory disease (PID) from administrative data for surveillance purposes. This approach may include non-PID cases. To refine PID case identification among women with ICD-9 codes suggestive of PID, a case-finding algorithm was developed using additional variables. Potential PID cases were identified among women aged 15-44 years at Group Health (GH) and Kaiser Permanente Colorado (KPCO) and verified by medical record review. A classification and regression tree analysis was used to develop the algorithm at GH; validation occurred at KPCO. The positive predictive value (PPV) for using ICD-9 codes alone to identify clinical PID cases was 79%. The algorithm identified PID appropriate treatment and age 15-25 years as predictors. Algorithm sensitivity (GH = 96.4%; KPCO = 90.3%) and PPV (GH = 86.9%; KPCO = 84.5%) were high, but specificity was poor (GH = 45.9%; KPCO = 37.0%). In GH, the algorithm offered a practical alternative to medical record review to further improve PID case identification.