Race sounds like simple information collected from patients-but over time, changes in the way race classification, requirements for patient demographic information, and changes in patients’ perceptions of race make it worthwhile for patients The data point for keeping a grain of salt, the expert said, records.

“We often think of race as a very “noisy” indicator,” said Suchisaria, a professor and director of the Johns Hopkins University Machine Learning and Healthcare Laboratory and the CEO of Bayes Health. Record or collect consistently. A company that develops artificial intelligence for clinical decision support.

Saria said that when race is used in predictive models, Bayesian Health also uses machine learning techniques to integrate the data, while considering the uncertainty of the reliability of the collected data.

Dr. Brooke Cunningham, a physician and sociologist at the University of Minnesota School of Medicine, pointed out that race is not a fixed variable or a set of categories.

Over time, people’s perceptions and classification of American race have changed.The federal census is conducted every 10 years, often with Changed the ethnic grouping it collects.

Cunningham said that race is not a biological variable and should not be used as such in medicine.

People who are considered black in the United States may be labeled differently in Africa, Latin America, or other regions. This can confuse new immigrants to the United States, and people often fill out paperwork without clearly defining when to use a particular label.

This was the case when Dr. Nigam Shah, Deputy Chief Information Officer of Stanford Healthcare Data Science and Associate Dean of Research at Stanford University School of Medicine, moved to the U.S.

“When I came to the United States, I dutifully selected’American Indians’ in the first few forms I filled out,” Shah said. “I mean, I grew up in India and I was in America.”

American Indian is another term for Native Americans.

For certain subgroups (such as Middle Eastern or Latino), the choice of race category can also be confusing, and they may not be sure which label is best for them. Many organizations only offer five ethnic categories to choose from—American Indian/Native Alaska, Asian, Black/African, Native Hawaiian/Other Pacific Islander, and White—and a separate racial question, ask Hispanic/Latino tradition.

The patient’s self-reported race may even change over time, or be reported in different ways in different care locations, depending on their understanding of race and the available categories. Patients who identify with more than one race may choose to choose only one if they feel closer to their identity, or do not know whether they can choose multiple options.

Patients may also be confused as to why they are asked to share racial and ethnic data, and may therefore refuse to do so.

Shah cited a quality improvement project he saw about a year ago, in which researchers asked patients in family medicine clinics about their race and ethnicity, and then compared the patient’s response with data recorded in the EHR system. According to a poster displayed at a conference, the project found that the probability of a patient being misclassified was approximately 37%.

The project did not delve into the cause of the mismatch. But “this mismatch is shocking,” Shah said. “I don’t know how to deal with these tags.”

Earlier learn have suggestion The quality of race and ethnicity data in patient records needs to be improved. Some people have found that Latino and Native American patients are most likely to be misclassified.

Medical institutions They collect racial and ethnic data in different waysRegenstrief Institute President and CEO Dr. Peter Embi said, and the consistency of the information obtained. Embi joined Vanderbilt University Medical Center in January as the head of the Department of Biomedical Informatics.

Some medical institutions may allow patients to self-report the data on paper or electronic forms, while other medical institutions may require registrars to ask patients to determine their race and ethnicity during registration-staff may not like this.In some cases, this is possible Registrar is making assumptions According to the patient’s appearance or name, understand the patient’s race and ethnicity.

Embi said: “I worry that many times, this is not a self-identified race and ethnicity that a person would report.”

Traditionally, demographic data (including race and ethnicity) was collected by registrants, who entered the information into the registration or patient receiving module, and then sent the data to the EHR. However, more and more data is self-reported by patients in patient portals, check-in kiosks, or in intake forms that are electronically sent to patients before appointments.

Hans Buitendijk, chairman of the EHR Association and director of interoperability strategy at Cerner, said: “What I want to say is that this situation is growing.”

Race and ethnicity data is not always collected in the EHR itself, but for the EHR to be certified by the Office of the National Health IT Coordinator of the Ministry of Health and Human Services, it must be able to record data about race and ethnicity-according to the Office of Management and Budget And the standards of the Centers for Disease Control and Prevention define race. Starting from the 2014 certification standard, EHR is expected to allow users to record multiple games.

The CDC standard contains more than 900 categories related to race and ethnicity; although the EHR must be able to record each of these concepts, the developer does not need to show all these concepts to the user.

The hospital can choose to display race and ethnicity categories in different ways, as long as the options can be reorganized to match OMB standard Used for federal reports-including five major race categories and one race category.