Zhang, Tayarani, Wang 2021

“ In a retrospective observational study, we characterized the types and frequency of prenatal care events that are associated with the various built environment factors of the patients’ residing neighborhoods.” 

Using previously collected data, we examined how built environment characteristics of a patient’s neighborhoods— such as walk-ability and land use— are associated with the types and frequency of prenatal care visits. 

“ We tested our hypotheses by linking patients’ health data extracted from de-identified electronic health records (EHR) with publicly available census-tract level data on the built environment.” 

We combined anonymous patient health records with public data about their neighborhoods to see how their surroundings affect their health. 

“Table 3 presents a post-hoc analysis of the distribution of demographics, medications, diagnoses, and built environment factors that were significantly different across the three clusters.”

Table 3 shows the main differences between the three groups regarding patient background, medications, health issues, and their surroundings.

Additionally, the application of natural language processing on unstructured clinical notes may allow us to elicit under-diagnosed and missed PPD as well as other conditions.

By using computers to read through doctors’ written notes, we can find cases of postpartum depression (PPD) and other health issues that might have been missed or not officially recorded.

Future research should address several limitations of this study. Because nearly 15% of racial information was missing from electronic health record data, incorporating patient- reported outcomes may help improve data completeness. Additionally, access to pharmacy claims data would allow future studies to analyze medication dose and frequency in greater detail. Applying these methods to a larger and more diverse patient cohort could further clarify how the built environment is associated with different types of postpartum depression outcomes.

While I understood the overall study design, the Methods section was the most difficult to follow. Specifically, understanding how the authors extracted and standardized information from the EHR data— such as diagnoses, medications, and gestational timing— was challenging. The use of medical coding systems and unstructured clinical notes made it harder to visualize how the data were prepared for analysis. Additional explanation of these processes would help clarify how the results were generated.