EditorialOriginal Research

Comorbidities, Utilization, and Quality of Care as Predictors of Diabetes Complications

Winston Liaw, Omolola E Adepoju, Carlos G Fuentes, Jiangtao Luo, Bill Glasheen, Ben King, Ioannis Kakadiaris, Todd Prewitt, Pete Womack, Jess Dobbins and LeChauncy Woodard
The Journal of the American Board of Family Medicine January 2026, 39 (1) 157922; DOI: https://doi.org/10.3122/jabfm.2025.250185R1

Abstract

Objective To determine factors associated with diabetes complications, measured by the Diabetes Complications Severity Index (DCSI).

Research Design and Methods This longitudinal analysis used insurer data (2016 – 2020) and included Medicare Advantage beneficiaries aged 65 and older with type 2 diabetes. The dependent variable was DCSI. Independent variables included year, demographics (age, sex, race/ethnicity, language, dual eligibility, rurality), comorbidities (Charlson (CCI) and Functional Comorbidity Indexes (FCI)), utilization (risk adjustment scores, emergency department, urgent care, outpatient, physician, inpatient, and pharmacy claims), and quality measures (hemoglobin A1c and blood pressure control). Four multilevel mixed-effects models were developed: demographics (model 1), comorbidities (model 2), utilization (model 3), and quality measures (model 4).

Results We included 49,843 individuals. Model 1 showed a relationship between year (IRR=1.32, p<0.001, 2020 vs. 2016), sex (IRR=0.86, p<0.001, female vs. male), race/ethnicity (IRR=1.06, p<0.001, Black vs. white), dual eligibility (IRR=1.26, p<0.001 yes vs. no), and rurality (IRR=0.90, p<0.001, yes vs. no). CCI (IRR=1.18, p<0.001) and FCI (IRR=1.08, p<0.001), which share overlapping and distinct comorbidities with DCSI, were associated with higher DCSI. Emergency department visits (IRR=1.01, p<0.05) and physician visits (IRR=1.003, p< 0.05) were associated with higher DCSI. Not meeting the blood pressure quality measure was linked to higher DCSI (IRR=1.10, p<0.05), while hemoglobin A1c control was not.

Conclusions Year, male sex, race/ethnicity, non-rural status, comorbidities, emergency department visits, and not meeting the blood pressure measure were linked to higher DCSI. Future research should develop strategies for high-risk groups in primary care settings.

In the United States (US), chronic diseases are projected to double between 2020 and 2050.1 Diabetes mellitus is a particularly deadly chronic disease, that can diminish quality of life, increase financial burdens, and shorten lifespans.2–4 In 2021, 38.4 million individuals, or 11.6% of Americans, were living with either diagnosed or undiagnosed diabetes.5 This figure is projected to rise to 54.9 million by 2030.6 Individuals with diabetes mellitus have a higher risk for acute and long-term complications, such as hyperglycemia, nervous system damage, kidney disease, eye damage, and cardiovascular events.7 The high cost of treating diabetes complications increases the urgency of predicting and preventing these adverse outcomes.8,9 Diabetes is addressed in one out of every seven office visits in the US, and 84% of individuals with diabetes see a primary care physician annually.10,11 Together, these figures highlight the central role of primary care in managing diabetes and reducing the burden of its complications.

Preventing diabetes complications remains a key, national priority that is codified within the objectives of Healthy People 2030.12 These efforts aim to curb the high costs associated with diabetes mellitus, which now accounts for one out of every four dollars spent on health care in the US.9 Individuals with complications disproportionately contribute to these costs, with complications accounting for over half of an individual’s diabetes-related expenses.13 Among patients with diabetes mellitus, one-third are expected to have two or more complications over their lifetimes.14 Thus, identifying those at high risk for complications and connecting them with evidence-based interventions is critical for meeting national goals.

One way to quantify diabetes-related complications is to use the Diabetes Complications Severity Index (DCSI), a 14-point scale that combines the number and severity of complications.14 Prior studies have found that the DCSI predicts poor outcomes that are important to patients.14 Specifically, each one-point increase in the DCSI is associated with a $2744 increase in health care costs,15 a 34% increase in mortality, and a 29% increase in hospitalizations.14 This index is important because it provides a standardized method for evaluating the severity of diabetes-related complications and facilitates more precise risk stratification. By quantifying and eventually predicting diabetes-associated complications, clinicians can ensure that scarce resources are given to those who will benefit the most.

Less is known about the factors associated with increases in DCSI and the interventions that can halt upward trends. One study of veterans found that having mental health or substance use diagnoses was associated with lower DCSI scores.16 Another concluded that pancreatic cancer, Hispanic ethnicity, and Black race were associated with higher DCSIs.17 Similarly, age, sex, and poverty have been linked to higher DCSIs.18,19 Unfortunately, these studies have limited generalizability because they were cross-sectional,19 involved targeted populations (like veterans),16 or were conducted outside the US.20,21 A limited number of studies explored the relationship between utilization and DCSI and found that enhanced access to outpatient services can mitigate expected increases.20,21 In addition to improved access, a wide range of interventions have the potential to prevent the accumulation of complications, though few studies have examined changes in DCSI. For example, providing patients with tools for self-management, including patient education, can improve glycemic control.22 Newer interventions, such as glucagon-like peptide-1 receptor agonists,23 expanded team members,24 and digital applications,25 have similarly improved health. Unfortunately, these can be out of reach for many, due to costs, availability, and shortages.

To identify those at highest risk, we examine a longitudinal, national cohort of older adults in the US and combine the covariates used in the aforementioned studies into a unified analysis. We also explore the influence of widespread quality measures (for diabetes mellitus, blood pressure, and cholesterol control) on the trajectory of DCSI.26 While improvements in chronic disease control should theoretically reduce the risk of complications, it’s unknown whether meeting these measures is ultimately associated with lower DCSI scores. To this end, this paper aims to identify the demographic, comorbidity, utilization, and quality of care factors associated with DCSI scores to guide interventions that reduce diabetes morbidity and mortality.

Research Design and Methods

Data

This longitudinal analysis assesses the factors associated with DCSI during the study period (2016-2020). Data were obtained from a national health payer, including enrollees with at least 12 months of continuous enrollment. The data included enrollment files, visit-level medical claims, and outpatient pharmacy claims data from Humana Inc. This US-based company provides Medicare Advantage, stand-alone Medicare prescription drug plans, and commercial plan offerings. The data encompassed demographic information, comorbidities, utilization, and member quality measures.

The population studied included individuals (1) aged 65 years and older; (2) who have a primary diagnosis of type 2 diabetes; 3) are enrolled in Medicare Advantage, including those with disabled status, special needs plans, and dual eligibility (which includes those who are eligible for both Medicaid and Medicare); (4) are continuously enrolled for the entire study period; and (5) fully complete information (no missing values). Individuals were excluded if they died during the study period or enrolled in group plans that are contractually excluded. Of the initial 323,306 Medicare Advantage visit-level records, we included 196,126 records representing 49,843 unique individuals who met the eligibility criteria (Figure 1).

Figure 1.
Figure 1. Flow Diagram for the Analytic Sample.

Measurement

The dependent variable was DCSI, measured in 2020.27 First introduced by Young et al, the DCSI score is a tool used to assess the severity of diabetes complications.14 It is based on the presence and severity of seven groups of diabetic complications, including retinopathy (e.g., retinal detachment and blindness), nephropathy (e.g., nephrotic syndrome and renal failure), neuropathy (e.g., nerve palsy and autonomic neuropathy), cerebrovascular disease (e.g., stroke), cardiovascular disease (e.g., myocardial infarction and heart failure), peripheral vascular disease (e.g., thromboembolism and gangrene), and metabolic complications (e.g., ketoacidosis and hyperosmolar hyperglycemic nonketotic syndrome). In computing the score, complications are categorized into two or three levels: 0 for no abnormality, 1 for some abnormality, and 2 for severe abnormality (range 0-13; 14 total levels). The max score is 13 because neuropathy has two rather than three levels (not present = 0, abnormal = 1). The final score represents the sum of the scores across all seven groups of complications, with 2 to 12 individual complications per group. Higher scores indicate more severe or numerous complications.

The primary independent variable was time, captured by year (2016-2020, with each year as a fixed effect). We adjusted for the following sociodemographic characteristics: [(age (65-69 vs. 70-74 vs. 75-79 vs. 80-84 vs. 85-89 vs. 90+), sex (male vs. female), race/ethnicity (Non-Hispanic White vs. Non-Hispanic Black vs. Other vs. Asian vs. Hispanic vs. Native American), primary language (English vs. Spanish vs. Other), dual eligibility status (no vs. yes), and patient geographic residence (rural vs. non-rural)]; comorbidities/risk indices [(Charlson Comorbidity Index (CCI), Functional Comorbidity Index (FCI), Medicare advantage risk scores (Centers for Medicare and Medicaid Services risk adjustment score based on hierarchical condition categories) and prescription risk scores28]; utilization patterns [(for emergency department, all physician visits (including both primary and specialty care visits), and inpatient admits)], and quality, [(captured by the Healthcare Effectiveness Data and Information Set (HEDIS) measures for hemoglobin A1c (HbA1c) and blood pressure control)].29,30 We used International Classification of Diseases codes (ICD-10) for diagnoses, Current Procedural Terminology (CPT) codes for utilization, and Logical Observation Identifiers and Names and Codes (LOINC) for quality measures. The CCI is a composite measure that predicts mortality while FCI predicts functional limitations. Both indices include congestive heart failure, myocardial infarction, stroke, peripheral vascular disease, and chronic obstructive pulmonary disease. FCI includes conditions that affect physical functioning, like arthritis, visual impairment, and hearing impairment. In contrast, CCI includes conditions that affect mortality including metastatic cancer and Acquired Immunodeficiency Syndrome.

Analysis

We used multivariate analysis of variance to assess differences in baseline characteristics. To examine factors associated with accumulating diabetes complications over time in this cohort, we used four multilevel mixed effects models with member ID as a random effect to account for repeated DCSI scores nested within individuals. The models used the following set of variables: year+ demographics only (model 1), year+ demographics + comorbidities (model 2), year+ demographics + prior utilization patterns (model 3), and year+ demographics + prior utilization patterns + quality measures (model 4). These models used a progressively expanding set of variables with one exception. We initially included CCI and FCI as independent variables in all 4 models, tested the variance inflation factor, and found no risk of multicollinearity for FCI and CCI. However, using the Hausman test on the FCI and CCI values, we found that both were endogenous to DCSI in the utilization (model 3) and quality (model 4). Accordingly, we excluded CCI and FCI from models 3 and 4. Considering that DCSI is a count variable, with values from 0-13 only, poison specification was used for the multilevel mixed effects models. An independent institutional review board approved this study. Patients and members of the public were not involved in the design and execution of this study. All data management and analyses were performed using SAS 9.4.31 Findings were considered statistically significant at P < .05.

Data and Resource Availability

The datasets generated during and/or analyzed during the current study are not publicly available because they are proprietary and contain protected health information. These data sets could be made available from the data steward, although transfer of the data would need to be negotiated with the entity from which the data originated.

Results

The analysis included data from 2016 to 2020 for 49,843 individuals. In 2016, over half of the sample was female (Table 1), and nineteen percent of the sample identified as Black, while 2.5% identified as Hispanic. Over sixty percent reported English as their primary language. The mean DCSI in 2016 was 2.17.

View this table:
Table 1. Baseline Characteristics, 2016-2020.

In model 1 (demographics), each subsequent year was correlated with higher DCSI scores relative to the reference year 2016 (Table 2). Females (IRR: 0.86) had significantly lower rates of increasing DCSI scores, compared to their male counterparts. We observed a dose response relationship between age and DCSI score, with progressively higher scores observed in older age groups compared to the reference group (65–69 years), ranging from an IRR of 1.08 (70–74 years) to 1.49 (90+ years, p<0.01). Compared to Non-Hispanic white older adults, Black older adults (IRR: 1.06, p<0.01) had significantly higher rates of increasing DCSI scores, while Hispanic (IRR: 0.94; p<0.01), Asian (IRR: 0.78; p<0.01) and other race older adults (IRR: 0.86; p<0.01) had lower rates of increase. Compared to English speakers, those who spoke Spanish as their primary language (IRR: 1.02, p=0.02) had significantly higher rates of increasing DCSI. Older adults eligible for dual enrollment (IRR: 1.26, p<0.01) had a 26% higher rate of increasing DCSI scores, compared to those not eligible, while residence in a rural area (IRR: 0.90, p=0.02) was associated with lower rates of increasing DCSI scores compared to non-rural residents.

View this table:
Table 2. Factors Associated with Diabetes Complications Severity Index.

In model 2 (Table 2), CCI (IRR: 1.18; p<0.01) and FCI (IRR: 1.08; p<0.01) were the strongest predictors of increasing DCSI scores. Consistent with model 1, year, Black race and dual eligibility remained significantly associated with higher rates of DCSI increase, while female sex, Asian and other races, and rural residence were associated with lower rates of increasing DCSI scores. While age remained significant in model 2, older age groups demonstrated an inverse relationship with DCSI.

In model 3 (Table 2), after accounting for demographics, risk scores, and healthcare utilization variables, several factors were significantly associated with increasing DCSI scores. Each subsequent year remained significantly associated with higher rates of increasing DCSI scores, with IRRs ranging from 1.10 (2017) to 1.19 (2020). Females had lower rates of increasing DCSI scores compared to males (IRR = 0.87, p<0.01). In contrast, progressively higher DCSI scores were observed with increasing age, with the oldest groups (90+) demonstrating the highest rate relative to the 65–69 reference group (IRR = 1.46, p<0.01). Black race remained associated with higher rates of increase (IRR = 1.08), p<0.01), while Asian and other race groups had lower rates (IRR = 0.95 and 0.96, respectively, p<0.01). Spanish as a primary language (IRR: 1.04, p=0.02) was associated with significantly higher rates DCSI increases. Dual eligibility was modestly associated with higher rates (IRR = 1.03, p<0.02), whereas rural residence was associated with lower rates (IRR = 0.95, p<0.01). Among clinical predictors, the Medicare Advantage risk score (IRR = 1.10, p<0.01), prescription risk score (IRR = 1.09, p<0.01), and several measures of healthcare utilization—including annual ED visits (IRR = 1.01, p<0.01), urgent care visits (IRR = 1.01, p<0.01), and inpatient admits (IRR = 1.07, p<0.01)—were significantly associated with DCSI progression.

In model 4 (Table 2), which additionally included quality measures for blood pressure and hemoglobin A1c, the overall pattern remained similar to model 3. Each year remained positively associated with DCSI progression (IRRs: 1.10–1.19, all p<0.01). Female sex (IRR = 0.84, p<0.01), Asian (IRR = 0.89, p=0.02) and other race (IRR = 0.93, p<0.01), and rural residence (IRR = 0.92, p<0.01) were associated with lower rates of DCSI increase, while Black race (IRR = 1.08, p<0.01), Spanish as primary language (IRR: 1.04, p<0.01) and dual eligibility (IRR = 1.06, p<0.01) were associated with higher rates. A clear dose-response relationship was observed with age: compared to adults aged 65–69, those aged 70–74 had an IRR of 1.09, 75–79 had 1.25, 80–84 had 1.38, 85–89 had 1.50, and 90+ had 1.51 (p<0.01 for all). Black race (IRR: 1.08, p<0.01) and dual eligibility (IRR: 1.06–1.26, p<0.01) were associated with greater increases. Clinical and utilization factors largely retained significance, with Medicare Advantage risk score (IRR = 1.11), prescription risk score (IRR = 1.09), annual inpatient admissions (IRR = 1.07), and annual ED and urgent care claims (IRR = 1.01 each) remaining significant predictors. Notably, not achieving the blood pressure <140/90 quality measure was associated with a higher rate of DCSI increase (IRR = 1.10), while hemoglobin A1c <9% not achieved was not significant (IRR = 1.001).

Discussion

In this national, longitudinal study, we identified the demographic, comorbidity, utilization, and quality of care factors associated with DCSI. In general, we found that DCSI increased over time and that individuals who were male, Black, Spanish-speaking, living in urban areas, or those with dual-eligible status had higher DCSI scores. Comorbidities also affected DCSI; a one-point increase in CCI was associated with an 18% increase in DCSI. From the perspective of utilization and quality, higher predicted prescription drug costs, emergency department claims, and not achieving guideline-recommended blood pressure levels were associated with higher DCSI. In contrast, those who identified as Hispanic or lived in rural areas had lower DCSIs. The relationship between age and DCSI was mixed. In the first model, older age was associated with higher DCSI, but these two variables were inversely related when accounting for other covariates. After controlling for comorbidities in model 2, for example, we hypothesize that individuals who have reached older ages may benefit from unmeasured genetic, behavioral, and social factors that are protective. This relationship should be studied in different populations to determine whether it is reproducible. In general, our findings validate and add to the results of others. Like us, others found that more comorbidities,16 rising age,32 and minoritized populations17,19 had higher DCSIs. These findings have important implications for the primary care clinicians delivering holistic, person-centered diabetes care.33

For example, improving quality in primary care could prevent complications, though our findings were mixed. Over the past two decades, there has been a push to transform our healthcare system to one that rewards value rather than volume. As a result, systems dedicated to recording, improving, and reporting the quality measures that drive value have emerged, representing a massive investment that costs primary care practices over $65,000 annually.34 Our research found that one quality measure (blood pressure control) was associated with lower DCSI. This relationship is supported by decades of evidence demonstrating that improved blood pressure control can reduce the incidence of heart disease, strokes, and kidney disease.35,36 Furthermore, other factors may explain why improved blood pressure control protects against the accumulation of diabetes complications. For instance, the ability to meet this quality metric may reflect well-resourced primary care practices with the processes (like electronic health record alerts) and people (like dietitians) needed to control chronic conditions systematically. Additionally, the patients who visit these practices may experience lower barriers to access, live in less impoverished neighborhoods, and have money to pay for the medications, supplies, food, tests, and specialists needed to manage diabetes mellitus effectively.

Paradoxically, meeting the quality measure for diabetes control was not associated with lower DCSI. Despite the investment in quality improvement, rates of amputations and end-stage renal disease remain stubbornly high.37 These trends suggest that systems to improve quality have had an inconsistent and inadequate effect. Alternatively, our finding could be a function of when we assessed the quality measure (2016-2019). Measurements in these years represent diabetes control at one point in time and may not be representative of control throughout the study period. For example, a patient who did not meet the quality measure in 2019 may have made changes that reduced their hemoglobin A1c in 2020. Furthermore, hemoglobin A1c is a surrogate marker strongly tied to but distinct from the complications that ultimately lead to worse health.26 While early optimization of glucose has been shown to reduce long-term complications from diabetes,7,36,38 not all interventions that lower hemoglobin A1c translate into improved clinical outcomes. This discrepancy has led some to question the test’s utility as a quality measure.39 Similar to our other findings, this relationship should be studied in different populations.

Researchers can use these findings to better risk stratify patients. For example, separate groups have developed tools to predict complications using data from the United Kingdom Prospective Diabetes Study40 and the Action to Control Cardiovascular Risk in Diabetes Study (Risk Equations for Complications of type 2 Diabetes).41 To date, no tools predict changes in DCSI. Risk stratification is critical to treatment paradigms for conditions like cardiovascular disease, where risk determines individuals should receive cholesterol-lowering medication.42 The effective pairing of risk and treatment was cited as a factor responsible for declines in cardiovascular mortality.43 For diabetes mellitus, the status quo is that treatment is intensified only after poor outcomes (e.g., high hemoglobin A1c or the emergence of a complication) appear. To reduce complications, a more proactive approach is needed.

Several limitations should be considered when interpreting these results. First, while our cohort is national, it comes from a single insurer. Our results may differ if we included uninsured individuals. Second, because age affects access to Medicare and the natural course of diabetes mellitus, we only included individuals aged 65 and older. Thus, our results may need to be more generalizable to younger individuals. Third, we did not include covariates that affect DCSI, like specific medications and lab results. Social risk factors, including housing, food, loneliness, and neighborhood characteristics, can similarly influence diabetes control but were omitted. We did not include quality measures for microalbumin or cholesterol as these were either retired or modified during the study period. These will be assessed in future papers. Finally, the ICD-10 codes we used to generate DCSI scores have known limitations. For example, the accuracy of these codes is dependent on clinician documentation, financial incentives to code maximally, and access to care, all of which can lead to false positives (e.g., the patient does not have the coded complication) and false negatives (e.g., the patient has a complication that was not coded). We did not assume that complications persisted because we could not distinguish between the resolution of complications and coding oversights. If a complication was coded in one year but not in a subsequent one, the complication was not carried forward, which may underestimate DCSI. This may be particularly impactful in areas with low access, like rural communities. Other methods for capturing complications, such as checklists, have the potential to improve accuracy but are more time-consuming and, thus, less scalable.

Our findings reveal that comorbidities, Black race, prescription drugs, and emergency department visits are associated with increased DCSI. In contrast, rural residence and blood pressure control are linked to lower DCSI. Given the high incidence of complications, these results highlight the critical need for targeted interventions and proactive risk stratification to mitigate the health and financial burdens of diabetes complications. Future research should investigate the contribution of additional factors and validate these findings in diverse populations.

Conflicts of Interest

W.L. received a gift from Humana, Inc. Winston Liaw is a consultant for MedirAI and the American Academy of Family Physicians.

Guarantor

O.A. had full access to all the study data and is responsible for its integrity and the accuracy of the data analysis.

Previous Presentation

North American Primary Care Research Group Annual Meeting; November 20-24, 2024; Québec City, Quebec, Canada.

Corresponding Author

Winston Liaw, MD, MPH, Department of Health Systems and Population Health Sciences, Tilman J. Fertitta Family College of Medicine, University of Houston, Houston, TX, wliaw{at}central.uh.edu

This article was externally peer reviewed.

Acknowledgements

We would like to thank Natalie Ou for her assistance in preparing this manuscript. W.L. wrote the manuscript, designed the study, and interpreted the data. O.A. wrote the manuscript, designed the study, and analyzed and interpreted the data. C.F. wrote the manuscript and interpreted the data. J.L. contributed to the study design, data analysis, and manuscript review. B.G. analyzed and interpreted the data and reviewed/edited the manuscript. B.K. contributed to the study design, data analysis, interpretation, and manuscript review. I.K. contributed to study design, data interpretation, and manuscript review. T.P. contributed to the study design, data interpretation, and manuscript review. P.W. analyzed and interpreted the data and reviewed/edited the manuscript. J.D. analyzed and interpreted the data and reviewed/edited the manuscript. L.W. designed the study, interpreted the data, and reviewed/edited the manuscript.

  • Received for publication May 19, 2025.
  • Accepted for publication September 2, 2025.

References

  1. ↵
    1. Ansah J. P.,
    2. Chiu C. T.
    (2023) Projecting the chronic disease burden among the adult population in the United States using a multi-state population model. Front Public Health 10:1082183, doi:10.3389/fpubh.2022.1082183, https://doi.org/10.3389/fpubh.2022.1082183. .
    OpenUrlCrossRefPubMed
  2. ↵
    1. Kaptoge S.,
    2. Seshasai S.,
    3. Sun L..,
    4. et al.
    (2023) Life expectancy associated with different ages at diagnosis of type 2 diabetes in high-income countries: 23 million person-years of observation. Lancet Diabetes Endocrinol 11(10):731–742. .
    OpenUrlPubMed
  3. ↵
    1. Scollan-Koliopoulos M.,
    2. Bleich D.,
    3. Rapp K. J.,
    4. Wong P.,
    5. Hofmann C. J.,
    6. Raghuwanshi M.
    (2013) Health-Related Quality of Life, Disease Severity, and Anticipated Trajectory of Diabetes. Diabetes Educ 39(1):83–91, doi:10.1177/0145721712467697. .
    OpenUrlCrossRefPubMed
  4. ↵
    1. Parker E. D.,
    2. Lin J.,
    3. Mahoney T..,
    4. et al.
    (2024) Economic costs of diabetes in the U.S. in 2022. Diabetes Care 47(1):26–43, doi:10.2337/dci23-0085. .
    OpenUrlCrossRefPubMed
  5. ↵
    (11 29, 2023) National Diabetes Statistics Report | Diabetes | CDC. . 2024-5-3. https://www.cdc.gov/diabetes/data/statistics-report/index.html.
  6. ↵
    1. Rowley W. R.,
    2. Bezold C.,
    3. Arikan Y.,
    4. Byrne E.,
    5. Krohe S.
    (2017) Diabetes 2030: Insights from Yesterday, Today, and Future Trends. Popul Health Manag 20(1):6–12, doi:10.1089/pop.2015.0181, https://doi.org/10.1089/pop.2015.0181. .
    OpenUrlCrossRefPubMed
  7. ↵
    1. Stratton I. M.
    (2000) Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ 321(7258):405–412, doi:10.1136/bmj.321.7258.405, pmid:10938048, pmcid:PMC27454. .
    OpenUrlAbstract/FREE Full Text
  8. ↵
    1. Ravaut M.,
    2. Sadeghi H.,
    3. Leung K. K..,
    4. et al.
    (2021) Predicting adverse outcomes due to diabetes complications with machine learning using administrative health data. Npj Digit Med 4(1):24, doi:10.1038/s41746-021-00394-8, pmid:33580109, pmcid:PMC7881135. .
    OpenUrlCrossRefPubMed
  9. ↵
    1. Parker E. D.,
    2. Lin J.,
    3. Mahoney T..,
    4. et al.
    (2024) Economic Costs of Diabetes in the U.S. in 2022. Diabetes Care 47(1):26–43, doi:10.2337/dci23-0085. .
    OpenUrlCrossRefPubMed
  10. ↵
    1. Antono B.,
    2. Bazemore A.,
    3. Dankwa-Mullan I..,
    4. et al.
    (2021) The State of Primary Care in the United States: A Chartbook of Facts and Statistics (Robert Graham Center; IBM-Watson Health; American Board of Family Medicine & Center for Professionalism & Value in Health Care), https://www.graham-center.org/content/dam/rgc/documents/publications-reports/reports/PrimaryCareChartbook2021.pdf?utm_source=chatgpt.com. . 2025-5-15.
  11. ↵
    1. S L.,
    2. K K.
    (2023) National Hospital Ambulatory Medical Care Survey: 2019 National Summary Tables (National Center for Health Statistics (U.S.)), doi:10.15620/cdc:123251, https://doi.org/10.15620/cdc:123251. .
    OpenUrlCrossRef
  12. ↵
  13. ↵
    1. Zhuo X.,
    2. Zhang P.,
    3. Hoerger T. J.
    (2013) Lifetime Direct Medical Costs of Treating Type 2 Diabetes and Diabetic Complications. Am J Prev Med 45(3):253–261, doi:10.1016/j.amepre.2013.04.017, https://doi.org/10.1016/j.amepre.2013.04.017. .
    OpenUrlCrossRefPubMed
  14. ↵
    1. Young B. A.,
    2. Lin E.,
    3. Von Korff M..,
    4. et al.
    (2008) Diabetes Complications Severity Index and Risk of Mortality, Hospitalization, and Healthcare Utilization. Am J Manag Care 14(1):15–23. .
    OpenUrlPubMed
  15. ↵
    1. Hazel-Fernandez L.,
    2. Li Y.,
    3. Nero D..,
    4. et al.
    (2015) Relationship of diabetes complications severity to healthcare utilization and costs among Medicare Advantage beneficiaries. Am J Manag Care 21(1):e62–70. .
    OpenUrlPubMed
  16. ↵
    1. Schmidt E. M.,
    2. Barnes J.,
    3. Chen C.,
    4. Trafton J.,
    5. Frayne S.,
    6. Harris A. H. S.
    (2019) Patient and Health Care Factors Associated With Long-term Diabetes Complications Among Adults With and Without Mental Health and Substance Use Disorders. JAMA Netw Open 2(9), doi:10.1001/jamanetworkopen.2019.12060, pmid:31553472, pmcid:PMC6763972. .
    OpenUrlCrossRefPubMed
  17. ↵
    1. Farias A. J.,
    2. Wu A. H.,
    3. Porcel J..,
    4. et al.
    (2020) Diabetes-Related Complications and Pancreatic Cancer Incidence in the Multiethnic Cohort. JNCI Cancer Spectr 4(5):pkaa035, doi:10.1093/jncics/pkaa035, pmid:33134820, pmcid:PMC7583154. .
    OpenUrlCrossRefPubMed
  18. ↵
    1. Weng W.,
    2. Liang Y.,
    3. Kimball E.,
    4. Hobbs T.,
    5. Kong S.
    (2018) Trends in comorbidity burden and treatment patterns in type 2 diabetes: Longitudinal data from a US cohort from 2006 to 2014. Diabetes Res Clin Pract 142:345–352, doi:10.1016/j.diabres.2018.05.031. .
    OpenUrlCrossRefPubMed
  19. ↵
    1. Young B. A.,
    2. Lin E.,
    3. Von Korff M..,
    4. et al.
    (2008) Diabetes complications severity index and risk of mortality, hospitalization, and healthcare utilization. Am J Manag Care 14(1):15–23. .
    OpenUrlPubMed
  20. ↵
    1. Du W.,
    2. Liu P.,
    3. Xu W.
    (2022) Effects of decreasing the out-of-pocket expenses for outpatient care on health-seeking behaviors, health outcomes and medical expenses of people with diabetes: evidence from China. Int J Equity Health 21(1):162, doi:10.1186/s12939-022-01775-5, pmid:36384591, pmcid:PMC9667616. .
    OpenUrlCrossRefPubMed
  21. ↵
    1. Wu C. X.,
    2. Tan W. S.,
    3. Toh M. P. H. S.,
    4. Heng B. H.
    (2012) Stratifying healthcare costs using the Diabetes Complication Severity Index. J Diabetes Complications 26(2):107–112, doi:10.1016/j.jdiacomp.2012.02.004. .
    OpenUrlCrossRefPubMed
  22. ↵
    1. Cochrane Metabolic and Endocrine Disorders Group
    1. Duke S.A.S.,
    2. Colagiuri S.,
    3. Colagiuri R.
    (1 21, 2009) Individual patient education for people with type 2 diabetes mellitus. in Cochrane Database Syst Rev, ed Cochrane Metabolic and Endocrine Disorders Group doi:10.1002/14651858.CD005268.pub2, pmid:19160249, pmcid:PMC6486318. .
    OpenUrlCrossRefPubMed
  23. ↵
    1. Wang S.,
    2. Wang S.,
    3. Wang Y.,
    4. Luan J.
    (2024) Glycemic Control, Weight Management, Cardiovascular Safety, and Cost-Effectiveness of Semaglutide for Patients with Type 2 Diabetes Mellitus: A Rapid Review and Meta-analysis of Real-World Studies. Diabetes Ther 15(2):497–519, doi:10.1007/s13300-023-01520-3, pmid:38175486, pmcid:PMC10838895. .
    OpenUrlCrossRefPubMed
  24. ↵
    1. Nkomazana O.
    1. Egbujie B. A.,
    2. Delobelle P. A.,
    3. Levitt N.,
    4. Puoane T.,
    5. Sanders D.,
    6. Van Wyk B.
    (2018) Role of community health workers in type 2 diabetes mellitus self-management: A scoping review. in PLOS ONE, ed Nkomazana O. 13(6):e0198424, doi:10.1371/journal.pone.0198424, pmid:29856846, pmcid:PMC5983553. .
    OpenUrlCrossRefPubMed
  25. ↵
    1. Baradez C.,
    2. Liska J.,
    3. Brulle-Wohlhueter C.,
    4. Pushkarna D.,
    5. Baxter M.,
    6. Piette J.
    (2022) Brief Digital Solutions in Behavior Change Interventions for Type 2 Diabetes Mellitus: A Literature Review. Diabetes Ther 13(4):635–649, doi:10.1007/s13300-022-01244-w, pmid:35279813, pmcid:PMC8917814. .
    OpenUrlCrossRefPubMed
  26. ↵
    1. Jiang D. H.,
    2. O'Connor P. J.,
    3. Huguet N.,
    4. Golden S. H.,
    5. McCoy R. G.
    (2022) Modernizing Diabetes Care Quality Measures: Analysis examines modernizing diabetes care quality measures. Health Aff (Millwood) 41(7):955–962, doi:10.1377/hlthaff.2022.00233, pmid:35759700, pmcid:PMC9288231. .
    OpenUrlCrossRefPubMed
  27. ↵
    1. Glasheen W. P.,
    2. Renda A.,
    3. Dong Y.
    (2017) Diabetes Complications Severity Index (DCSI)—Update and ICD-10 translation. J Diabetes Complications 31(6):1007–1013, doi:10.1016/j.jdiacomp.2017.02.018. .
    OpenUrlCrossRefPubMed
  28. ↵
    1. Centers for Medicare & Medcaid Services
    (10 24, 2024) Risk Adjustment. . 2024-11-20. https://www.cms.gov/medicare/payment/medicare-advantage-rates-statistics/risk-adjustment.
  29. ↵
    1. Groll D.,
    2. To T.,
    3. Bombardier C.,
    4. Wright J.
    (2005) The development of a comorbidity index with physical function as the outcome. J Clin Epidemiol 58(6):595–602, doi:10.1016/j.jclinepi.2004.10.018. .
    OpenUrlCrossRefPubMed
  30. ↵
    1. Charlson M. E.,
    2. Pompei P.,
    3. Ales K. L.,
    4. MacKenzie C. R.
    (1987) A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis 40(5):373–383, doi:10.1016/0021-9681(87)90171-8. .
    OpenUrlCrossRefPubMed
  31. ↵
    1. SAS
    SAS 9.4. . 2024-8-20. https://www.sas.com/en_us/home.html.
  32. ↵
    1. Hazel-Fernandez L.,
    2. Li Y.,
    3. Nero D..,
    4. et al.
    (2015) Relationship of diabetes complications severity to healthcare utilization and costs among Medicare Advantage beneficiaries. Am J Manag Care 21(1):e62–70. .
    OpenUrlPubMed
  33. ↵
    1. Davies M. J.,
    2. Aroda V. R.,
    3. Collins B. S..,
    4. et al.
    (2022) Management of Hyperglycemia in Type 2 Diabetes, 2022. A Consensus Report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care 45(11):2753–2786, doi:10.2337/dci22-0034, pmid:36148880, pmcid:PMC10008140. .
    OpenUrlCrossRefPubMed
  34. ↵
    1. Casalino L. P.,
    2. Nicholson S.,
    3. Gans D. N..,
    4. et al.
    (2009) What Does It Cost Physician Practices To Interact With Health Insurance Plans?: A new way of looking at administrative costs—one key point of comparison in debating public and private health reform approaches. Health Aff (Millwood) 28(Supplement 1):w533–w543, doi:10.1377/hlthaff.28.4.w533, https://doi.org/10.1377/hlthaff.28.4.w533. .
    OpenUrlAbstract/FREE Full Text
  35. ↵
    1. Blood Pressure Lowering Treatment Trialists' Collaboration
    (2013) Blood pressure lowering and major cardiovascular events in people with and without chronic kidney disease: meta-analysis of randomised controlled trials. BMJ 347(oct03 1):f5680–f5680, doi:10.1136/bmj.f5680, pmid:24092942, pmcid:PMC3789583. .
    OpenUrlAbstract/FREE Full Text
  36. ↵
    1. Uk Prospective Diabetes Study Group
    (1998) Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 38. BMJ 317(7160):703–713, doi:10.1136/bmj.317.7160.703. .
    OpenUrlAbstract/FREE Full Text
  37. ↵
    1. Centers for Disease Control and Prevention
    United States Diabetes Surveillance System. . 2024-8-2. https://gis.cdc.gov/grasp/diabetes/diabetesatlas-surveillance.html.
  38. ↵
    1. The Diabetes Control and Complications Trial Research Group
    (1990) Diabetes control and complications trial (DCCT): update. Diabetes Care 13(4):427–433, doi:10.2337/diacare.13.4.427. .
    OpenUrlAbstract/FREE Full Text
  39. ↵
    1. Rodriguez-Gutierrez R.,
    2. McCoy R. G.
    (2019) Measuring What Matters in Diabetes. JAMA 321(19):1865, doi:10.1001/jama.2019.4310, pmid:30985865, pmcid:PMC7332201. .
    OpenUrlCrossRefPubMed
  40. ↵
    1. Clarke P. M.,
    2. Gray A. M.,
    3. Briggs A..,
    4. et al.
    (2004) A model to estimate the lifetime health outcomes of patients with Type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68). Diabetologia 47(10):1747–1759, doi:10.1007/s00125-004-1527-z, https://doi.org/10.1007/s00125-004-1527-z. .
    OpenUrlCrossRefPubMed
  41. ↵
    1. Basu S.,
    2. Sussman J. B.,
    3. Berkowitz S. A.,
    4. Hayward R. A.,
    5. Yudkin J. S.
    (2017) Development and validation of Risk Equations for Complications Of type 2 Diabetes (RECODe) using individual participant data from randomised trials. Lancet Diabetes Endocrinol 5(10):788–798, doi:10.1016/S2213-8587(17)30221-8, https://doi.org/10.1016/S2213-8587(17)30221-8. .
    OpenUrlCrossRefPubMed
  42. ↵
    1. Diao J. A.,
    2. Shi I.,
    3. Murthy V. L..,
    4. et al.
    (2024) Projected Changes in Statin and Antihypertensive Therapy Eligibility With the AHA PREVENT Cardiovascular Risk Equations. JAMA 332(12):989, doi:10.1001/jama.2024.12537, pmid:39073797, pmcid:PMC11287447. .
    OpenUrlCrossRefPubMed
  43. ↵
    1. Mensah G. A.,
    2. Wei G. S.,
    3. Sorlie P. D..,
    4. et al.
    (2017) Decline in Cardiovascular Mortality: Possible Causes and Implications. Circ Res 120(2):366–380, doi:10.1161/CIRCRESAHA.116.309115, https://doi.org/10.1161/CIRCRESAHA.116.309115. .
    OpenUrlAbstract/FREE Full Text
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools

Jump to section

Keywords