Analyzing Big Data Could Help Tailor Health Care for Elderly Patients
In an evolving health care environment that is increasingly focused on improving patient care, increasing value, and efficiently managing resources, patient data and how that data is used and analyzed becomes even more important. The elderly, particularly those in nursing homes, is a rapidly growing population that could benefit from analyzing these data sets to help develop benchmarks to guide clinical care and long-term care decisions to improve care planning and outcomes.
Three professors in George Mason’s Department of Health Administration and Policy, Farrokh Alemi, Phan Giang, and Janusz Wojtusiak, recently co-authored three research studies in The Gerontologist, the journal of The Gerontological Society of America, that examine data from Veterans Affairs (VA) nursing homes and use predictive modeling to help inform long-term health care decisions. According to the Congressional Budget Office, nearly half of the 8.6 million veterans in the Veterans Health Administration are older than 65, and the fastest growing group is older than 85.
“Benchmarks have not been available to guide clinical care and inform long-term health care decisions,” said Alemi. “The data we have available, in this case for patients in VA nursing homes, could be used to optimize health care planning and delivery and personalize care for patients to improve their quality of life.”
One study analyzed Activities of Daily Living (ADLs) for nearly 300,000 residents in VA nursing homes to provide benchmarks for the likelihood, time until, and sequence of functional decline and recovery. Using data from the Resident Assessment Instrument Minimum Data Set, which is a standardized, federally mandated process for clinical assessment that is used in Medicare- and Medicaid-certified nursing homes, the authors analyzed data from 16 domains related to ADLs. The analysis demonstrated that 57 percent of nursing home residents followed four different paths of ADL impairments; the most common order was loss of independence in bathing, grooming, walking, dressing, toileting, bowel continence, urinary continence, transferring, and feeding.
“Understanding the sequence of ADL impairments allows patients, families, and clinicians to set priorities and care plan,” Alemi said.
In another study, Alemi, Wojtusiak, and colleagues used predictive modeling to determine methods of anticipating a change in ADLs for VA nursing home patients after they had been hospitalized. Electronic Health Record data for more than nearly 5,600 VA Community Living Center residents was analyzed to establish patterns of recovery and loss of ADL functions. The models the authors developed were able to accurately predict ADLs for 14 days after hospitalization, and although accuracy declined from 14 days to one year post-hospitalization, the models were still able to predict a high number of ADLs.
“This study demonstrates that predictive modeling can be used to identify patients who may be at risk of experiencing a temporary or permanent decline in ADL functions and to identify patients who will improve after hospitalization. Using this information, health care professionals can tailor care plans to address individual patient needs,” Wojtusiak said. “Predictive modeling and big data analysis means that we no longer need to talk about an ‘average’ patient. Instead, we can build accurate individualized models that are tailored to individual patients.”
The third study compared hospitalization rates for veterans who were in shared homes as part of the VA Medical Foster Home (MFH) program to those who were in VA nursing homes. The VA MFH program allows patients to reside in a community-based living arrangement as opposed to a nursing home. MFH residents live with a community caregiver who can also address the patient’s specific care needs. Access to big data allowed the authors to compare hospitalization rates for MFH patients to those of VA nursing home patients for 14 common conditions, including falls, bacterial infections, and adverse medication effects. Using a matched case control study to account for differences in the data, the authors determined that MFH patients did not have increased hospitalization rates for common medical conditions compared to VA nursing home patients.
“The results of our analysis suggest that ‘shared homes’ and home-based health care is a safe alternative to nursing home care and does not negatively impact the quality of care or the patient’s quality of life,” Alemi said. “Our access to big data allowed us to make these comparisons since smaller data sets would not have such exact matches on age, gender, and co-morbidities, which we used in our study.”