Clinical Risk Groups: Turning Data into Insight to Improve Patient Readmission Rates

The demands placed upon healthcare organisations, payers and providers continue to escalate in the 21st century due to ever-growing populations with longer life expectancy, advances in preventative medicine, interventional healthcare services—and the repercussions of a global pandemic.

In addition to handling the daily demands of healthcare, organisations are faced with a major reoccurring challenge: patient readmissions. Readmissions not only drive up costs for the patient and hospital, but are also linked to longer hospital stays and higher rates of morbidity.

For example, elderly patients are more likely than any other group to be readmitted, with 15% of patients over the age of 65 in England readmitted within 28 days.¹ A systemic review of 34 studies reported an average avoidable readmission rate of 27%.

To combat the rising patient readmission rates, clinicians studied the correlation between readmissions and patient follow ups and post discharge, and the importance of timely follow-ups. Because each patient varies greatly in age, long term conditions, severity of conditions, and associated social determinants of health, clinicians need a way to make apples-to-apples comparisons while factoring in layers of variability. This is where the use of Clinical Risk Groups (CRGs) comes into play.

CRGs are a clinical, categorical classification system that uses readily available administrative data to identify children and adults with chronic health conditions. Health records are used to assign individuals into mutually exclusive clinical categories. These categories take a whole-person approach to measuring a patient’s burden of illness, and the clinical understanding built into CRGs ensures each category is adjusted for severity. This reflects the extent and progression of specific chronic conditions or combination of chronic conditions.

A study published in the Journal of General Internal Medicine looked at the impact identifying high-risk patients in a Medicare accountable organisation (ACO) and found the ACO was able to reduce hospital readmissions by 5.7% and decrease emergency department visits by 6.3% by targeting these high-risk patients identified through CRGs.²

According to the Annals of Family Medicine, “The focus of most research examining the impact of outpatient follow up on preventing readmissions has been narrow, focusing on specific disease states.”³ Researchers analyzed the impact of using CRGs on care coordination and patient outcomes in a large primary care practice, and they found that it led to a 32% reduction in hospital readmissions and a 34% reduction in emergency department visits among high-risk patients.⁴

In a similar study, researchers identified high-risk patients with chronic pulmonary disease (COPD) in a large healthcare system. The patients identified as high-risk had lower healthcare costs and fewer hospitalizations than patients who were not identified as high-risk.⁵

Early outpatient follow ups result in lower rates of readmissions; however, we know that the individuals in these groups will vary widely in their comorbidities, condition complexities and needs. We also know that individuals with multiple chronic conditions, all interacting with each other, will have clinical presentations and patterns of resource usage that are more than the sum of the individual conditions.

Because hospital resources are limited, rather than conducting follow-up care for each patient in a given care pathway in the same way, hospitals are using CRGs to identify patients most likely to benefit from early follow up, thus lowering their risk of readmission.³ Not only were hospitals able to see the groups most impacted from early follow-up care, they could also determine the optimal time for that follow up (e.g. seven days post discharge vs. 21 days). 

Advancements in artificial intelligence (AI) have made it easier for clinicians to track patient data by automating the many processes involved in risk adjustment and data analysis. Some of the key AI features that have advanced patient data tracking include natural language processing (NLP), machine learning (ML) and predictive analysis.

NLP is a type of AI that can analyze and understand human language. It is used to extract data from clinical notes and other unstructured data sources to create more comprehensive patient profiles for risk adjustment purposes. ML is a type of AI that allows computers to learn and improve from experience without being explicitly programmed. This is used to analyze large datasets and identify patterns to help improve patient outcomes and improve risk accuracy.

Lastly, predictive analytics uses statistical models and ML algorithms to analyze data and make predictions about future events. This is used to predict which patients are at the highest risk of developing certain conditions or experiencing adverse outcomes, which helps clinicians prioritize care management resources.

Rising patient readmission rates continue to be a recurring challenge for healthcare organisations. But implementing CRGs paired with AI assistance has been shown to help identify high-risk patients and target early follow-up care, resulting in lower rates of readmissions and emergency department visits.

References

1. Vernon, D., Brown, J. E., Griffiths, E., Nevill, A. M., & Pinkney, M. (2019). Reducing readmission rates through a discharge follow-up service. Future Healthcare Journal, 6(2), 114–117. https://doi.org/10.7861/futurehosp.6-2-114

2. Sood, M. M., Rigatto, C., Komenda, P., Mojica, J., & Tangri, N. (2014). Mortality risk for women on chronic hemodialysis differs by age. Canadian Journal of Kidney Health and Disease, 1, 10. https://doi.org/10.1186/2054-3581-1-10

3. Jackson, C., Shahsahebi, M., Wedlake, T., & DuBard, C. A. (2015). Timeliness of outpatient follow-up: An evidence-based approach for planning after hospital discharge. The Annals of Family Medicine, 13(2), 115–122. https://doi.org/10.1370/afm.1753

4. Han, T. S., Murray, P., Robin, J., Wilkinson, P., Fluck, D., & Fry, C. H. (2021). Evaluation of the association of length of stay in hospital and outcomes. International Journal for Quality in Health Care. https://doi.org/10.1093/intqhc/mzab160

5. Beauvais, B., Richter, J., & Brezinski, P. (2017). Fix these first. Journal of Healthcare Management, 62(3), 197–208. https://doi.org/10.1097/jhm-d-15-00048