Project 465467

Differences in socioeconomic position (SEP) lead to unfair differences in health across social groups. This is known as health inequity. Despite the importance of SEP on health, this information is not routinely collected from patients, and therefore not available in our health databases. As a result, it is difficult to identify patients at high risk of poor outcomes, and to develop interventions to address these inequities. There is a single electronic health record (EHR) being implemented at all the hospitals in Alberta. This initiative presents a unique opportunity to optimize the collection of SEP indicators for hospitalized patients across the province. We will conduct a mixed methods evaluation to enhance the collection of SEP indicators that are at the root of health inequities. These indicators include gender, income, employment, occupation, education, and ethnicity. Our study has two objectives. First, we will develop natural language processing techniques (a type of artificial intelligence) to extract information about the SEP indicators in clinical documentation in EHR. That is, we will see if the notes and documents created by doctors, nurses, and other healthcare providers can be accurately mined for information about the six SEP indicators. We will evaluate how accurate these algorithms are, by comparing the information that is extracted to the information in the chart, through manual chart review. This approach likely cannot adequately capture all six SEP indicators of interest. Our second objective then is to understand how to optimize collection of SEP information when patients come into hospital, through focus groups with patients, healthcare providers, and administrative staff across the province. These findings will have far-reaching implications that can enhance SEP information in EHRs, so that fair and equitable care can be provided across the population.