Dr. Marian Neuhouser, the Head of Fred Hutch’s Cancer Prevention Program and lead author of the study, emphasized the significance of using the WHI dataset to explore long COVID. “Long COVID remains a serious health problem for many,” she said. “Many prior studies on long COVID use cross-sectional data. Since the WHI has robust health history, demographic, and lifestyle data going back more than 30 years, we were able to leverage this vast data resource using novel machine learning to identify potential risk factors for long COVID in older women.” The study included more than 1,230 postmenopausal women who had been diagnosed with COVID-19, of whom 425 reported experiencing long COVID symptoms. Common symptoms among this group included fatigue, brain fog, memory issues, and musculoskeletal pain. Dr. Neuhouser and her team employed machine learning techniques to identify the top 20 risk factors associated with long COVID in this group, which were further analyzed using statistical models.
The study found that women who had experienced significant weight loss in the two years before their COVID-19 diagnosis were at a higher risk of developing long COVID. In addition, those with mobility issues, such as difficulty bending, kneeling, or needing assistive devices like canes or walkers, were also more likely to suffer from long COVID. Certain pre-existing health conditions, including a history of rheumatoid arthritis, heart valve procedures, and sleep disturbances, were found to increase the likelihood of long COVID. Interestingly, while older age is typically thought to be associated with more severe outcomes, in this cohort, it was linked to a lower probability of long COVID diagnosis. In contrast, factors such as recent weight loss and physical limitations increased the risk.
These findings have important implications for public health, particularly in how we manage the long-term effects of COVID-19 in older populations. Dr. Neuhouser highlighted that these findings open new avenues for clinical practice and future research. “Clinicians may be able to use the information to identify patients who could be at risk for long COVID,” she noted. She further pointed out that a natural next step for research would be to “build a risk prediction model,” which could help doctors more effectively screen and manage patients at risk for long COVID. Such a model would enable healthcare providers to identify vulnerable individuals early and potentially intervene before long COVID symptoms become debilitating.