Diabetes Patient

The dataset focused on readmission of Diabetes patients in the U.S. from 1999-2008. As we explored the data, we noticed many unspecified data points, specifically in weight, age, medication, and other categories. Our goal was to evaluate how choices in data formatting/presentation can hinder our ability to understand relationships between crucial factors, reducing a patient to vague statistics, thus neglecting a holistic, comprehensive understanding of their life.

               For our project, we used elements from the data deconstruction methods we learned in class: Poirier’s “Reading Datasets” method, and Koopman’s “Format Anatomies” method. Under Poirer’s method, we used her Connotative and Deconstructive readings. The connotation element makes us question what each category might imply about how hospital patients are defined within the system, and the deconstructive element makes us look for what categories and values are missing or vague, and what that might reveal about the dataset’s design. Through this method, we’re attempting to gain a better understanding of how neglecting important, yet socially sensitive information can interfere with data analysis. (Poirer) Under Koopman’s method, we used his input-level format anatomy element. By doing this, we observe each data point and how it was formatted the way it was, and question why the researchers decided to make that choice (Koopman 155).

Through Poirer’s connotative reading, we found that the prominent categories included were race, gender, age range, medication changes, number of prior visits, and readmission status. By focusing only on these categories, the study frames diabetes as measurable through institutional interaction, where identities are reduced to demographic strata and behavioral patterns only within hospitals.

               Through Poirer’s deconstructive reading, we observed that precise ages and weights, type of diabetes, mental health history, food access, long-term glucose trends, and HbA1c levels were missing. In the absence of these details, we do not get a comprehensive view of the reality of a patient’s life and history with diabetes, which are crucial for uncovering relationships between the variables. From here, we question why we’re missing these data points. 

Under Koopmans format anatomies, we focused on the element of anatomies of input-level formats. Through this, we noticed formatting choices such as age grouped into brackets, hba1c results are only a snapshot, not continuous, what it means that weight is either blank, “?”, or within brackets. Diagnoses were coded numerically, and readmission was only recorded by <30 days, >30 days, or no readmission. From these formatting choices, the patient is reduced to an encounter, the body becomes coded files, and chronic illness becomes episodic readmission risk. Diabetes if formatted as a hospital, compliance, and cost-management problem, instead of a long-term lived condition influenced by structural inequalities, social, and environmental factors.

               In this study, race is recorded, yet socioeconomic context and structural factors are not considered. This may frame diabetes as an individual problem instead of a possible systemic issue, which is unfair to the patients and deters us from possible underlying issues that require different approaches.

Ultimately, we concluded that this dataset doesn’t simply measure diabetes, it creates a specific version of the patient that is institutional, episodic, and demographically classified, which neglects structural inequalities. When these social determinants are missing, inequalities and nuances become invisible.

Works Cited

Poirier, Lindsay. “Reading datasets: Strategies for interpreting the politics of data signification.” Big Data &amp; Society, vol. 8, no. 2, 1 July 2021 https://doi.org/10.1177/20539517211029322.

Koopman, Colin. “Format Anatomies.” Data Equals: Democratic Equality and Technological Hierarchy, The University of Chicago Press, Chicago, Illinois, 2025, pp. 155.