Researchers from Michigan State University and Corewell Health used de-identified electronic health records of more than 1.5 million patients to analyze incidence rates and risk factors of mild cognitive impairment, or MCI, in rural and urban areas in West Michigan. They are the first researchers in the state to use this technique.
Results showed that many cases of MCI could be going undetected among those living in rural communities in the area. Future studies will use these findings to develop artificial intelligence, or AI, tools for earlier identification of the condition.
The retrospective study, which included 10 years of historical patient data, has been published in Alzheimer’s & Dementia: Translational Research & Clinical Interventions. It is the first large-scale analysis representing most of the population of West Michigan, with some of its findings surprising study authors.
“While we had our suspicions about what we would find, we did not expect the potential rate of underdiagnosis of MCI in some of the rural areas in West Michigan to be so high,” said Bin Chen, associate professor in the MSU College of Human Medicine and co-principal investigator of the study.
According to Chen, individuals typically experience MCI before developing dementia. Yet, the study found that patients who progressed directly to dementia without a prior MCI diagnosis, also referred to in the study as MCI skippers, were three times more prevalent than those identified with MCI initially.
“This tells us MCI may be going unreported with some patients,” Chen said.
Understanding the underreporting
David Chesla, co-principal investigator and senior director of research data management at Corewell Health Research Institute in Grand Rapids, agreed and said that this underreporting is what may be causing the MCI incidence rates to be so much lower.
“Our hypothesis from the beginning of this work was that we would have underreporting of cognitive impairment in communities across West Michigan; we just didn’t know to what extent,” Chesla said. “Our suspicion was initially derived from national data that reports a growing incidence rate of MCI within our aging U.S. population. Our patient data mirrors a subset of the national data; however, our patient MCI incidence rate is lower than national averages.”
National averages can range from 10% to 18% depending on race, age and timeframe in which the data was collected, and Chesla said the study showed their numbers demonstrated a significantly lower incidence in West Michigan.
Chesla also indicated that the research team decided to dive deeper into the geographic distribution of patients, allowing them to separate whether patients had an urban or rural location, something he said has not been done before. Doing this provided further evidence that potential underreporting exists with the ratio of MCI skippers to diagnosed MCI cases being 4.3 times higher in rural areas compared to 2.8 times in urban areas.
While lack of access to care in these communities along with other reasons could be driving the higher rate of underreporting, Chesla said that a limitation of the study was having to use information from 10 years ago when electronic record systems were in their earlier stages.
“Today, electronic health records are integrated across most health systems; however, with our work going back in time, there could be fragmentation of records that may be driving the underreporting as well,” Chesla said.
Urban areas show a larger array of MCI risks
Additional findings showed that while risk factors for MCI were similar between the rural and urban populations, the urban areas exhibited a larger array of contributing factors including being African American as well as having hearing loss, inflammatory bowel disease, obstructive sleep apnea and insomnia. The most common contributing factors of MCI include diabetes, stroke, Parkinson’s disease and older age.
According to the researchers, the massive amount of data now gives them the ability to leverage AI to build high-performance machine learning models that can identify higher-risk patients earlier across the state and potentially across the country. It has been shown that early diagnosis is key to potentially reversing or delaying progression of cognitive impairment.
“The goal is to integrate this tool into health care systems everywhere so it can assist physicians in detecting and managing MCI patients more effectively,” Chen said.
But for now, Chesla suggests that if individuals are experiencing symptoms such as hearing loss, mood swings or some of the other more common symptoms, they should not hesitate to reach out to their physician or a health care provider to help.
“We are in an era where there are care plans and rehabilitation services that can aid in slowing, if not reversing, cognitive impairment when caught early,” Chesla said.
The study was co-led by Xiaodan Zhang, a data scientist in the College of Human Medicine, and Martin Witteveen-Lane, a data engineer at Corewell Health, and supported by the Corewell Health-MSU Alliance and the National Institutes of Health.
