Using Data To Uncover High-Risk Patient Populations

Added September 12, 2018

Stratifying patient data is the process of taking a complete patient data set and pulling away parts of the dataset to uncover insights not previously seen. Patient risk-stratification allows a data analyst to remove layers within the data one at a time or in groups to better understand their patients’ health. These patient groups are referred to as ‘strata’ or ‘blocks’. Each stratum represents a specific segment of a patient population. These groups are unique and can allow a health data analyst to focus on specific characteristics of a patient group which can be directly addressed by a care management program.

The goal of using strata is to establish different approaches for managing the patient population so ‘tests’ can be applied to each stratum. The most effective approach will provide the greatest impact on the health of the patient population. There are many stratification methods that can be used including: using patient demographics, identifying social determinants of health, uncovering risk factors, and using indexes and models.

Patient Demographic Makeup

Knowing the demographic makeup of a patient population is important. Diseases which are chronic or specialized can have a greater frequency of occurrence in certain types of demographics. For example, the average age of a first heart attack is 66 for men and 70 for women and the risk of a heart attack begins to increase after the age of 45 for men and 55 for women.

By understanding these statistics, cardiovascular units can know which patient types they should focus on based on gender and age.

Social Determinants of Health

Social Determinants play a significant role in the health and wellness of a patient population. The healthcare provided to a patient is necessary for the health of a patient, but it isn’t the leading factor in the overall health and wellness of the patient or patient population. According to The Henry J. Kaiser Family Foundation, “health behaviors such as smoking and diet and exercise, are the most important determinants of premature death. Moreover, there is a growing recognition that a broad range of social, economic, and environmental factors shape individuals’ opportunities and barriers to engage in healthy behaviors.”

As a healthcare organization better understands the social determinants related to the health and risk of their patients, they can organize their data analysis for identifying patients with related social determinants.

Uncover Risk Factors

Healthcare organizations who understand the risk factors related to patients, who have the greatest propensity for requiring additional care after discharge, can apply those risk factors to the patient health data. By applying patient risk factors to patient data, healthcare organizations can stay out in-front of clinical and financial risk within their patient population. There are many risk factors that can be used by the patient analytics team including patient utilization, cost per patient, symptom management, etc.

Using Indexes and Models

There are many ways of tracking the risk of patients. Some of the most common risk scores and quality metrics focus on comorbidity in one way or another. Comorbidity is the presence of two or more chronic conditions with a patient or patient population.

Comorbidity indexes can be used to rank the risk potential of chronic patients. Some of the methods to track comorbidity include:

Comorbidity-Polypharmacy Score (CPS): A simple measure that consists of the sum of all known comorbid conditions and all associated medications.

Elder Risk Assessment (ERA): Patients over the age of 60 encompass the ERA. The assessment includes the patient’s age, gender, marital status, and the number of hospital stays. This index is used in tracking patients who get readmitted within 30 days following admission into a nursing home.

Charlson-Deyo Comorbidity Measure: Originally derived to classify comorbidities affecting one-year mortality in cancer patients, it sums 17 specific conditions. The Charlson predictive model focuses on patients with many comorbid illnesses.

Many of these comorbidity indexes can be used in modeling patient populations. For example, the Charlson-Deyo model allows for a calculation of patient risk based on patient age and the number and types of comorbid conditions of a patient.


Patient health data provides a well of information for a healthcare organization. By stratifying patients and creating custom patient segments and groups, healthcare organizations can find out the demographic makeup of the patients they serve. They can identify social determinants that contribute to a patient’s health and wellness. They can uncover risk factors for patient care adoption and use patient indexes and models to uncover patients with the highest risk.

By utilizing these tools, a healthcare data analyst can discover patients who the healthcare organization can affect change with. They can assist in decreasing financial and clinical risk within a healthcare population and directly complement the care management and care engagement team.

Healthcare data analytics is the first-place healthcare organizations should go to when looking to identify patients who they can help manage care for. The insight provided by patient health analytics can be invaluable for a healthcare organization.