U.S. flag An official website of the United States government
  1. Home
  2. Science & Research
  3. Science and Research Special Topics
  4. Advancing Regulatory Science
  5. Assessing disparities in occurrence of type 2 diabetes adverse drug events in minority populations using real world administrative claims and electronic health records
  1. Advancing Regulatory Science

Assessing disparities in occurrence of type 2 diabetes adverse drug events in minority populations using real world administrative claims and electronic health records

CERSI Collaborators: Hadi Kharrazi, MD, PhD, FAMIA, FACMI; Jonathan P Weiner, DrPH; Elham Hatef, MD, MPH; Martin Bishop, PharmD, MS, BCACP; Hsien-Yen Chang, PhD; Chris Kitchen, MSc.

FDA Collaborators: Qi Liu, PhD, MStat; Christine Lee, PharmD, PhD; Hao Zhu, PhD, MStat.; Yueqin Zhao, PhD; Yandong Qiang, MD, PhD, MHS, MPH; Anuradha Ramamoorthy, PhD; Richardae Araojo, PharmD, MS (Formerly of FDA); Justin Penzenstadler, PharmD, MS.

Project Start Date: August 1, 2020
Project End Date: August 31, 2023

Regulatory Science Challenge

Adverse drug events (ADEs) are costly and largely mitigable and preventable. Severe hypoglycemic (SHG) events are serious ADEs experienced by patients with type 2 diabetes (T2D). SHG events trigger the extreme drop of blood sugar level thus leading patients to experience confusion, disorientation, seizure, loss of consciousness, and sometimes coma. SHG events require immediate medical attention and often result in admission to emergency departments (ED). Most of these ED admissions can be avoided if patients can recognize the signs and symptoms of severe hypoglycemia and manage their medications appropriately. The increasing burden of SHG events has been recognized as an important public health issue, potentially affecting millions of patients. For example, in 2008, hypoglycemia was identified as the third most common ADE among Medicare beneficiaries, while nearly all identified cases of hypoglycemia were considered preventable 1 . Moreover, clinically significant hypoglycemia has shown to affect 2 to 5 percent of hospitalized patients with diabetes 2 ,3 .

The National Action Plan for Adverse Drug Event Prevention1 recognizes that specific patient populations may be more vulnerable to ADEs than others. These populations include individuals with deprived socioeconomic status, inadequate health literacy level, limited access to health care services, and ethnic and racial populations. Unfortunately, racial and ethnic minority populations often experience SHG events more than others. Research is needed to understand the factors associated with higher levels of SHG events in minority populations with T2D.

Project Description and Goals

This project aimed to assess potential racial and ethnic disparities in the occurrence and prognosis of T2D patients experiencing severe hypoglycemic (SHG) events. In collaboration with FDA’s subject matter experts, a team of investigators at the Johns Hopkins Center for Population Health IT (CPHIT), used retrospective clinical data extracted from the Johns Hopkins Medical Institute’s electronic health record (EHR) system to improve the identification of SHG, especially among racial and ethnic minority populations such as Black/African American populations. The research team also explored the potential textual patterns captured in clinical notes when racial and ethnic minority populations express SHG events compared to other patients. Findings from this study can support FDA and other regulatory agencies to better inform patients and clinicians on potential differences in preventing SHG events among racial and ethnic minority populations with T2D.

Research Outcomes/ Results

  • The study team used a combination of diagnostic, medication, and laboratory test values to identify patients with T2D. A total of 139k patients, 16+ years old, with T2D were identified using EHR data across three years (2017 to 2019). Close to 34.7% of this population were Black/African American people, while 52.5% were White people, 5.5% were Asian people, and 7.3% were identified as other races or ethnicity. Although social determinant of health data is considerably under captured in clinical settings as structured code, around 2.5% of this population had a code indicating transportation challenges, while 1.8% had housing challenges, and close to .5% expressed food insecurity. The average adapted diabetes complications severity index (aDCSI) of the study population was 1.044 across the years. The mean Charlson comorbidity index (CCI) was close to 2.34 across the years.
  • Using a combination of diagnostic and laboratory test values, an average of 2.3% of the patients were identified experiencing an SHG event in any of the given years. The share of Black/African American patients who experienced SHG was higher (3.2%) than White (2.1%) and Asian (1.5%) patients. Patients in the age groups of 16-24 and 85+ years old had a higher rate of SHG events (4.7% and 3.3%) compared to other age groups in the study population (i.e., SHG prevalence ranged between 2.1% and 2.5% among other age groups).
  • Mining EHR’s clinical notes increased the number of patients who were identified as having experienced an SHG event. Using pattern matching algorithms that are designed to identify sign and symptoms of SHG in clinical notes of patients, increased the population of patients who have experienced SHG by 16%, while using a more advanced natural language processing technique increased the SHG population by 18%.
  • Models predicting the occurrence of an SHG event in the same year (i.e., current) or next year (i.e., prospective) revealed that being in the 16-24 year age group (OR 5.14 (CI 4.28-6.17) for concurrent and & 2.62 (CI 2.02-3.40) for prospective prediction), having food insecurity (OR 2.34 (CI 1.91-2.86) & 3.44 (CI 2.68-4.40)), sustaining housing challenges (OR 1.91 (CI 1.68-2.18) & 2.32 (CI 1.95-2.76)), administrating insulin (OR 4.33 (CI 4.04-4.64) & 1.77 (CI 1.60-1.96)), or taking alpha-glucosidase inhibitors (OR 2.27 (CI 1.43-3.60) & 1.94 (CI 0.97-3.89)) significantly increase the likelihood of an SHG event.
  • The “Black/African American” race variable had an odds ratio of 1.30 (CI 1.18-1.44) in predicting an SHG event in the current year among the study population. This effect was measured after controlling for demographic (e.g., age, sex), clinical (e.g., aDCSI index), medication (e.g., diabetes medications), and social determinant of health (e.g., transportation, housing, food insecurity) variables. The odds ratio increased to 1.46 (CI 1.34-1.58) for the Black/African American race variable when predicting an SHG event in the next year. The odds ratio of the Black/African American race slightly decreased when the SHG population were expanded using the EHR’s clinical notes (e.g., OR in the prospective model slightly decreased to 1.37 when SHG events identified by the natural language processing techniques were added to the overall SHG population).

Research Impacts

  • Supporting a collaboration: This project has facilitated a strategic collaboration between FDA’s subject matter experts and the Johns Hopkins researchers. The topics of collaboration ranges from ADEs, to diabetes, predictive modeling, population health informatics, real world data (e.g., EHRs), minority health, and health equity research.
  • Peer-reviewed publications and conference presentation: The study team has planned two scientific manuscripts for publication in peer-reviewed journals. The first manuscript will describe the value of EHRs in identifying SHG events among T2D patients. The second manuscript will discuss the underlying racial and ethnic disparities in the occurrence of SHG events among T2D patients. The study team has also planned to submit an abstract to the next AMIA (American Medical Informatics Association) symposium discussing the challenges in using real world data to identify SHG events in racial and ethnic minority populations.
  • Future research: This study has led to follow-up discussions between FDA and Johns Hopkins experts to further explore the possibility to secure internal or external funding to support health disparities research in the regulatory science domain using real world data.

 

  • 1U.S. Department of Health and Human Services, Office of Disease Prevention and Health Promotion. (2014). National Action Plan for Adverse Drug Event Prevention. Washington, DC: Author.https://health.gov/sites/default/files/2019-09/ADE-Action-Plan-508c.pdf (Accessed May 5, 2022)
  • 2Cook CB, Kongable GL, Potter DJ, Abad VJ, Leija DE, Anderson M. Inpatient glucose control: a glycemic survey of 126 U.S. hospitals. J Hosp Med. 2009 Nov;4(9):E7-E14. doi:10.1002/jhm.533. PMID: 20013863.
  • 3Wexler DJ, Meigs JB, Cagliero E, Nathan DM, Grant RW. Prevalence of hyper- and hypoglycemia among inpatients with diabetes: a national survey of 44 U.S. hospitals. Diabetes Care. 2007 Feb;30(2):367-9. doi:10.2337/dc06-1715. PMID: 17259511.
 
Back to Top