Type 1 diabetes mellitus (T1D) is a chronic metabolic condition affecting millions of children and adults worldwide. T1D incidence and prevalence are increasing, with over 60,000 new cases annually in the United States alone.1,2 Daily comprehensive care and periodic surveillance can minimize the risk of long-term multisystem organ complications associated with poor metabolic control, including retinopathy, nephropathy, neuropathy, cardiovascular disease, and acute complications like diabetic ketoacidosis and severe hypoglycemia. Direct medical costs related to T1D exceed $16 billion and are rising.3 Despite numerous advances in insulin delivery and glucose monitoring, achieving adequate glycemic and metabolic control remains challenging, particularly in youth, as evidenced by worsening hemoglobin A1c (HbA1c) trends.4 Adherence to the recommended intense insulin regimens remains a significant barrier,4,5 and a recent analysis concluded that only a minority of youth with T1D achieve American Diabetes Association (ADA) and International Society for Pediatric and Adolescent Diabetes (ISPAD) goals for HbA1c.4
Although HbA1c is universally recognized as the reference marker for glycemic control and a risk factor for complications,6 this metric alone does not truly reflect overall management. For example, HbA1c reflects neither glycemic variability nor hypoglycemia frequency or severity, both known to affect acute and chronic complications.6,7 In addition to quarterly HbA1c assessments, current ADA and ISPAD guidelines recommend several surveillance and risk reduction measures, including regular quarterly blood pressure monitoring, annual urine screening for microalbuminuria, and periodic lipid profile analysis.8,9 Due to the multicomponent nature of such monitoring, our objective was to create a multifaceted and nuanced metric to reflect diabetes care outcomes beyond HbA1c.
The ADA recommends that “efforts to assess the quality of diabetes care and create quality improvement (QI) strategies should incorporate reliable data metrics, to promote improved processes of care and health outcomes.”10 In creating an aggregate metric to evaluate care quality and outcomes, we aimed to incorporate a population health approach to T1D by using current clinical practice recommendations to optimize diabetes care and achieve treatment goals. The Type 1 Diabetes Composite Score (T1DCS) concept began with an interest in developing a QI metric that complements the use of HbA1c in diabetes care management but also better reflects overall achievement of clinical goals as defined by current guidelines and best practices. This metric can also grade clinical diabetes program performance on both patient and population levels. With clinical informatics systems and a population health data management approach, we first built a T1D patient registry to help us develop and deliver this new metric.
QI approaches using metrics such as clinical care scores have successfully improved clinical performance and decreased errors in pediatric cancer, chronic kidney disease, pulmonology, and otolaryngology.11–13 More recently, a similar approach has been reported in pediatric diabetes using process measures and shown to improve care delivery.14 This QI initiative leveraged our electronic medical record (EMR) with a diabetes patient registry to design, build, and launch a T1DCS that would be a robust yet straightforward aggregate metric built using outcome measures. We hypothesize that this aggregate score could serve as a more global diabetes clinical program metric. Our primary objective was to create an easily scored metric per nationally accepted outcome goals that is trackable over time yet available in real time for use in the clinic. This first-of-its-kind “metric combining metrics” utilizes outcome measures to assess diabetes clinic performance.
This QI project was performed at Nationwide Children’s Hospital, a large free-standing academic pediatric medical center, serving communities in Ohio and surrounding Midwest states. Endocrinology at Nationwide Children’s Hospital regularly follows approximately 1,900 children and adolescents with T1D and manages >300 new-onset T1D patients yearly.
A robust electronic health information infrastructure allowed us to design and launch a diabetes patient registry utilizing population health modules within the EMR. A multidisciplinary “Diabetes QI” team, including the medical team (physicians, advanced practitioners, diabetes nurse educators, and social workers), clinical informaticists, data analysts, and QI specialists, created the T1D Health Registry in stages. After initial design work to assemble the registry conceptual framework, we identified and defined the target patient population. The Clinical Informatics team surveyed medical providers to identify key clinical performance indicators. They developed and built the corresponding metrics, including optimizing the processes for incoming, storing, and retrieving data within the EMR. The Clinical Informatics team, in conjunction with the providers, then validated each metric over several iterative cycles. Final planning and execution of “Go-live” included optimizing the user experience, as well as training and implementation. Additional refinements postrelease helped optimize the registry, data collection, and display. Finally, the QI support team facilitated downstream analysis. The T1D Health Registry went live in the summer of 2017.
Composite score development began by identifying clinical elements associated with the best outcomes in T1D care. The multidisciplinary team assembled and proposed key drivers. We identified 9 critical elements reflecting optimal diabetes management that would minimize risk for future complications and other adverse outcomes, such as severe hypoglycemia and diabetic ketoacidosis. These individual elements and their relative weighted scoring were determined and refined by a combination of consensus expert opinion and national and international guidelines per ADA and ISPAD. These elements (Table 1) fall under the domains of management tools (HbA1c, Continuous Glucose Monitoring [CGM] use), diabetes care assessment (diabetes clinic visit attendance, acute care visits [emergency department or inpatient hospitalization]), and complications risk (low-density lipoprotein level, urine microalbumin ratio, blood pressure, episodes of severe hypoglycemia, complication diagnosis code in the medical record). Each element contributes to a positive score based on an optimal goal, and the aggregate sum of the element scores gives a total score. The 9 elements lead to a total possible composite score of 18 based on the scoring rubric (Fig. 1). A higher score would be associated with better compliance with guideline targets and better clinical outcomes. Based on modeling and rapid improvement cycles, score ranges were used to define categories denoting low (15−18 points), moderate (10−14 points), or high (0−9 points) risk of adverse outcomes and color coded for easy viewing within the EMR.
Table 1. -
The Type 1 Diabetes Composite Score
||Most recent HbA1c
||ADA goal of 7.5% or below*
| Continuous Glucose Monitoring
||Recent CGM use
||Active regular CGM use
Diabetes care assessment
| Diabetes clinic visits
||No. clinic visits attended in previous rolling 12 months
||At least 4 visits per year
| Acute care visits
||No. emergency department visits or inpatient admissions in previous rolling 12 months
||No ED/IP visits required
| Lipids: LDL
||LDL value in most recent lipid profile (up to 5 years look-back)
||LDL ≤ 100 mg/dL
| Urine microalbumin ratio
||Urine microalbumin ratio value in most recent urine screening (up to 12 months look-back)
||Negative microalbuminuria (Urine microalbumin <0.030 mg/mg Cr)
| Hypertension screening: blood pressure
||Most recent blood pressure measurement
||BP results <95th percentile for age, sex, and height
| Severe hypoglycemia
||Episodes of severe hypoglycemia (as defined by seizures, loss of consciousness, or requiring assistance with treatment) since the previous clinic visit
||No episodes of severe hypoglycemia
| Overt complication diagnosis
||ICD-10 visit diagnosis-related to diabetes complications of retinopathy, nephropathy, neuropathy
||No complication-related diagnoses
*HbA1c <58 mmol/mol.
BP, blood pressure; Cr, creatinine; ED, emergency department; ICD-10, International Classification of Diseases, 10th revision; IP, in-patient; LDL, low-density lipoprotein.
The score was built based on EMR functionality as part of the population health module. Population health modules encompass functionality within the EMR that allows population-level analytics, to summarize and report population health information within a patient registry database. This approach allows healthcare systems to more efficiently monitor and coordinate health care delivery, quality, and cost and, specifically, make comparisons about demographics, such as age, gender, geography, and/or reimbursement/insurance. The registry helped compile the relevant data across clinic encounters and used the laboratory results database to make the appropriate comparisons and analyses. We then built the scoring mechanism into the EMR and deployed the T1DCS in our clinics via interfacing tools such as a health registry dashboard (“Diabetes Dashboard”) and specialized clinic schedule views to display the pertinent data. EMR informatics specialists further optimized the interface to display the score in dashboards and clinic schedule views for clinic previsit planning (PVP) and clinic use in real time.
This work was identified as QI and did not involve human subjects research; thus, it was exempt from Institutional Review Board review and approval.
We successfully launched this metric build in the fall of 2018 and applied the scoring analysis to our large T1D population. The T1DCS reflects the relative value of each clinical element when evaluating each patient. The score quickly provides the patient and provider with a summary of the current diabetes management status in the clinical setting (Fig. 2). As the score also displays any shortcomings in complying with recommended guidelines per ADA and ISPAD and attaining their goal outcomes, targeted education and intervention plans can be appropriately devised through PVP and in real time. The scoring platform’s intended use was as an “in-clinic dashboard” in addition to PVP functionalities with the clinic/schedule view (Fig. 3). Also, single patient snapshot views were built for use during the clinic visit to show the strengths of current care and any opportunity for improvement, and it updates in real time (“Hover bubble,” Fig. 3).
The registry cohort (n = 1,880 per our initial reporting in March 2019) has a mean age of 14.8 years (range: 2.0−34.6), is 54.5% male, and self-identifies as 77.1% White, 14.2% Black/African American, 4.6% biracial/multiracial, 2.2% Latino/Hispanic, and 1.3% Asian. Of note, 4% (n = 66) of our population indicated Somali ethnicity, reflecting our sizeable Somali immigrant population in central Ohio. T1DCSs ranged from 3 to 18 points, with a mean and median score of 11.9 and 12, respectively, out of 18 possible points.
Individual scores can be viewed over the entire T1D population to demonstrate trends in score data distribution and population health information. Following the launch in November 2018 and a 4-month registry enrollment period, we evaluated the scoring in our cohort. The T1DCS applied over the registry cohort population demonstrates a normal distribution (Fig. 4A). We observed a right shift in scores on a population level over 6 months (March to September 2019), consistent with clinical care improvement and progress in QI efforts (Fig. 4B).
Following this initial proof-of-principle phase, we continued the project study period. We have applied this metric in our institution to improve our care delivery system via enhanced decision support and customized PVP. This approach has also helped track progress in our QI efforts, such as CGM adoption, annual laboratory screening, and multidisciplinary clinic support.
Numerous metrics are typically monitored and tracked to help evaluate glycemic control and risk for metabolic complications in patients with T1D. We designed and built a multicomponent score to more globally reflect an individual’s diabetes management status beyond the HbA1c metric: the T1DCS. This novel metric helps evaluate the clinical status of our patients with T1D in complying with nationally and internationally accepted guidelines for care. Here, we aimed to provide proof of principle that an aggregate score comprised T1D outcome metrics could be built leveraging the EMR and diabetes patient registry systems to track QI project progress longitudinally.
The compilation of individual T1DCS scores across our registry cohort allows us to view our T1D population at a glance and track them over time. The initial population score data showed a normal distribution, and over 6 months, the score curve shifted to the right (higher T1DCS), suggesting an improvement in clinical care outcomes (Fig. 4B). We hypothesize that this right shift may reflect multiple process improvements in clinical care, including increased patient CGM adoption within our population. More importantly, this example demonstrates how the T1DCS can be used in population tracking to identify and resolve care gaps and as a metric for diabetes clinical program performance.
This data analytics approach to QI in the diabetes care arena is novel and has broader implications for population health. As this metric can gauge improvements longitudinally within an institution and across diabetes clinics in multiple institutions, it can develop into a new standard in monitoring diabetes care.
This scoring model does have limitations. Among the major parameters tracked by current ADA and ISPAD guidelines, the only component not addressed is ophthalmology examinations (or retinal photography), which was deferred in this iteration of the score as we did not have an adequate clinical and EMR framework to track this individual component. In addition to retinal/ophthalmologic examination, the score can potentially be improved by adding additional elements considered relevant in monitoring diabetes care and preventing complications, such as measures of psychological burden related to diabetes, including quality of life or diabetes-related distress.5,15 Other indicators that reflect the burden of care can be considered, such as missed school days related to diabetes.16,17 Time in range and other similar CGM metrics can potentially be incorporated as well, as accepted clinical targets to guide diabetes management are now formalized.18 This pilot project aimed to demonstrate that this metric concept is feasible and has the potential for further use in diabetes clinical care and improvement toward best outcomes. As this score build requires a robust informatics infrastructure for data capture, data integrity, and EMR build, this scoring paradigm’s generalizability may not be immediate, particularly at institutions without population health registries or dedicated informatics staff.
In addition to exploring additional elements worthy of inclusion into the T1DCS, the next steps from a population health perspective include sorting and analyzing data by patient demographics to identify healthcare disparities in our region. A breakdown of score distribution by race, ethnicity, age, sex, and zip code (home address) can quickly be generated. With the ability of the T1DCS to update in real time with new data such as current HbA1c, we would anticipate that this metric can be used for PVP and patient education and counseling during clinic visits to address gaps and opportunities in earned points. Also, as best practices guide the individual elements comprising the T1DCS, there may be applications for patient/family-facing use, to help further guide diabetes education and self-management.
The T1DCS is the first description of a “metric combining metrics” using patient outcome measures (including HbA1c) to assess patients with T1D and diabetes clinic performance. Our institution has begun to use this score in PVP, to document outcome improvement in our QI projects, and to assess our clinical program in achieving optimal diabetes outcomes across a broad range of well-accepted domains.
The T1DCS is a novel metric to evaluate the clinical status of diabetes care in our patients with T1D in complying with nationally and internationally accepted guidelines for care. Besides its use in monitoring and tracking progress in diabetes care QI initiatives, this metric can also assess a clinical program’s capability to achieve optimal diabetes outcomes and document outcome improvement.
This population health and data analytics approach, guided by best practices, can be further developed into a potential new standard in monitoring diabetes care. Further applications may include identifying health care disparities and addressing social determinants of health. There may also be broader applications for patient/family-facing use in patient self-management.
The authors have no financial interest to declare in relation to the content of this article.
We would like to acknowledge the hard work and support of the greater T1D Quality Improvement team at Nationwide Children’s Hospital, and our wonderful patients with T1D and their families.
1. Mayer-Davis EJ, Lawrence JM, Dabelea D, et al.; SEARCH for Diabetes in Youth Study. Incidence trends of type 1 and type 2 diabetes among youths, 2002-2012. N Engl J Med. 2017; 376:1419–1429.
2. Rogers MAM, Kim C, Banerjee T, et al. Fluctuations in the incidence of type 1 diabetes in the United States from 2001 to 2015: a longitudinal study. BMC Med. 2017; 15:199.
3. American Diabetes Association. Economic costs of diabetes in the US in 2017. Diabetes Care. 2018; 41:917–928.
4. Foster NC, Beck RW, Miller KM, et al. State of type 1 diabetes management and outcomes from the T1D exchange in 2016-2018. Diabetes Technol Ther. 2019; 21:66–72.
5. Hood KK, Beavers DP, Yi-Frazier J, et al. Psychosocial burden and glycemic control during the first 6 years of diabetes: results from the SEARCH for Diabetes in Youth study. J Adolesc Health. 2014; 55:498–504.
6. Danne T, Nimri R, Battelino T, et al. International consensus on use of continuous glucose monitoring. Diabetes Care. 2017; 40:1631–1640.
7. American Diabetes Association. 6. Glycemic targets: standards of medical care in diabetes-2019. Diabetes Care. 2019; 42suppl 1S61–S70.
8. American Diabetes Association. 13. Children and adolescents: standards of medical care in diabetes-2019. Diabetes Care. 2019; 42suppl 1S148–S164.
9. DiMeglio LA, Acerini CL, Codner E, et al. ISPAD clinical practice consensus guidelines 2018: glycemic control targets and glucose monitoring for children, adolescents, and young adults with diabetes. Pediatr Diabetes. 2018; 19suppl 27105–114.
10. American Diabetes Association. 1. Improving care and promoting health in populations: standards of medical care in diabetes-2019. Diabetes Care. 201942suppl 1S7–S12.
11. Barron CL, Elmaraghy CA, Lemle S, et al. Clinical indices to drive quality improvement in otolaryngology. Otolaryngol Clin North Am. 2019; 52:123–133.
12. Crandall W, Davis JT, Dotson J, et al. Clinical indices can standardize and monitor pediatric care: a novel mechanism to improve quality and safety. J Pediatr. 2018; 193:190–195.e1.
13. Olshefski R, Vaughan M, YoungSaleme T, et al. The cancer care index: a novel metric to assess overall performance of a pediatric oncology program. J Patient Saf. 2016. doi:10.1097/PTS.0000000000000267.
14. Obrynba KS, Indyk JA, Gandhi KK, et al. The diabetes care index: a novel metric to assess delivery of optimal type 1 diabetes care. Pediatr Diabetes. 2020; 21:637–643.
15. Young-Hyman D, de Groot M, Hill-Briggs F, et al. Psychosocial care for people with diabetes: a position statement of the American Diabetes Association. Diabetes Care. 2016; 39:2126–2140.
16. Fleming M, Fitton CA, Steiner MFC, et al. Educational and health outcomes of children treated for type 1 diabetes: Scotland-wide record linkage study of 766,047 children. Diabetes Care. 2019; 42:1700–1707.
17. Jackson CC, Albanese-O’Neill A, Butler KL, et al. Diabetes care in the school setting: a position statement of the American Diabetes Association. Diabetes Care. 2015; 38:1958–1963.
18. Battelino T, Danne T, Bergenstal RM, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care. 2019; 42:1593–1603.