Research

Kofi Atiemo, Nikhilesh R. Mazumder, Juan C. Caicedo, Daniel Ganger, Elisa Gordon, Samantha Montag, Haripriya Maddur, Lisa B. VanWagner, Satyender Goel, Abel Kho, Michael Abecassis, Lihui Zhao, and Daniela Ladner
Despite lower socioeconomic status, Hispanics in the United States have equal or higher survival rates compared to non-Hispanic whites (NHW). The authors attempted to understand whether this “Hispanic Paradox” applied to survival rates of liver cirrhosis. Using retrospective data from 20,121 patients in a Chicago-wide database linked together through Datavant technology, the researchers established that, controlling for age, sex, comorbidities, etiology of cirrhosis, and insurance status, Hispanics did indeed have higher survival rates than NHW.

Brandon A. Francis, Jennifer Beaumont, Matthew B. Maas, Eric M. Liotta, David Cella, Shyam Prabhakaran, Jane Holl, Abel Kho, and Andrew M. Naidech
Depressive symptoms in patients with intracerebral hemorrhage (ICH) are common and are associated with worse outcomes. However, those symptoms are often not accurately screened for or diagnosed. By linking together data from multiple EHR centers in Chicago, the researchers were able to see how often patients were treated for depressive symptoms; simultaneously, they performed depression screenings on study patients. After comparing the two data sets, the researchers concluded that depressive symptoms are significantly more common than the ICD codes or subsequent medical treatment would suggest.

Lindsay P. Zimmerman, Satyender Goel, Shazia Sathar, Charon E. Gladfelter, Alejandra Onate, Lindsey L. Kane, Shelly Sital, Jasmin Phua, Paris Davis, Helen Margellos-Anast, David O. Meltzer, Tamar S. Polonsky, Raj C. Shah, William E. Trick, Faraz S. Ahmad, and Abel N. Kho
This paper outlined a novel workflow for recruiting potential trial patients. Members of the community were identified, surveyed, and then assigned an encrypted and hashed identifier. Concurrently, data from a variety of hospitals was linked together at the patient level. Via the encrypted and hashed identifier, the investigators could connect data from the hospital systems to understand whether someone was eligible for the study. The method of recruitment was significantly more efficient than the typical process for most clinical trials.

Theresa L. Walunas, Kathryn L. Jackson, Anh H. Chung, Karen A. Mancera-Cuevas, Daniel L. Erickson, Rosalind Ramsey-Goldman, and Abel Kho
By linking data across six different Chicago health institutions, the researchers were able to understand the extent to which patients with systemic lupus erythematosus (SLE) receive fragmented care and the impact of said care. In identifying 4,276 patients with SLE, 20 percent received care from more than 1 institution; those patients were more likely to have complications, including increased risk of infections, cardiovascular disease, and stroke.

Faraz S. Ahmad, Cheeling Chan, Marc B. Rosenman, Wendy S. Post, Daniel G. Fort, Philip Greenland, Kiang J. Liu, Abel N. Kho, and Norrina B. Allen
The authors sought to understand the degree of agreement of electronic data research networks as compared to data collected by standardized research approaches in a cohort study. The comparisons were made by linking data from MESA (Multi-Ethnic Study of Atherosclerosis), a community-based cohort, with EHR’s from six Chicago-area hospitals. Ultimately, nearly 70 percent had data in both systems and demonstrated mixed results. For some measurements, such as BMI, the correlations between the MESA and EHR data were quite high. For others, such as systolic blood pressure, the correlation coefficient was only 0.39.

Dustin D. French, Jess J. Behrens, Kathryn L. Jackson, Abel N. Kho, Theresa L. Walunas, Charlesnika T. Evans, Michael Mbagwu, Curtis E. Margo, and Paul J. Bryar
By linking together medical encounter data from a number of different Chicago hospital systems, the authors identified 150,661 patients with diabetes (out of a total of two million patients). Using that information, the study estimated the geographic distribution of undiagnosed diabetic retinopathy and found that low-income and minority areas had a disproportionate rate of undiagnosed diabetic retinopathy. Based on that, a new screening mechanism that is not currently funded by Medicare or Medicaid was suggested.

Jamie Fishman, Sara McLafferty, and William Galanter
The study attempted to understand the utilization of emergency departments for preventable conditions at the individual and neighborhood level by linking together emergency department admissions from four Chicago hospitals. By combining the data, the authors showed that individuals in medically underserved areas (MUAs) had a larger chance of emergency department use that could have been prevented.

Kathryn L. Jackson, Satyender Goel, Abel N. Kho, and Rajesh N. Keswani
Using the HealthLNK Data Repository, the authors sought to understand the relationship between hospital utilization and distance from the index hospital for patients with adverse events following an ambulatory endoscopy. Using mortality as the endpoint, the authors found that of the 86 patients (of the 22,898 in the study) who had early mortality, nearly half did not return to the index hospital. In total, there was a statistically significant relationship between a patient’s utilization of their index hospital and their proximity.

James A. Mays, Kathryn L. Jackson, Teresa A. Derby, Jess J. Behrens, Satyender Goel, Mark E. Molitch, Abel N. Kho, and Amisha Wallia
Some patients with diabetes are hospitalized repeatedly for diabetic ketoacidosis (DKA), which is typically associated with worse clinical outcomes. The authors analyzed recurrent DKA and fragmentation of care across six institutions in the Chicago area, using Datavant technology to understand when the same patient would have different DKA episodes in different hospitals. The study found that of the 3,615 patients, 21.6% had recurrent DKA; of the 780 recurrent patients, 125 were hospitalized at more than one hospital.