Results Summary
What was the project about?
Quality measures assess how well healthcare providers deliver care. Examples of these measures include patient health outcomes like blood sugar control or how often patients go to the emergency room, or ER. These measures usually take into account how sick patients are by looking at how many health conditions a patient has and how severe they are.
Social risks also affect patient health outcomes. These risks may include unstable housing, income, or food. But quality measures don’t currently account for social risks.
In this study, the research team wanted to learn how health problems and level of social risk affect quality measures. To do this, the team combined patients’ health data with social risk data about their neighborhoods.
What did the research team do?
The research team used patient health data from community health centers. The data included health records and insurance claims for 73,328 patients with diabetes and 988,106 patients with Medicaid. Using these data, the team looked at the number and severity of patients’ health problems; they also looked at blood sugar levels and ER visits.
Then the research team linked the patient health data to data about social risks in the patients’ neighborhoods. Using data on seven social risks, such as unemployment, poverty, and overcrowded housing, the team calculated a social deprivation index, or SDI, score. Patients who lived in an area with more social risks had higher SDI scores.
Next, the research team looked at how patients’ health problems and SDI affected their blood sugar control and number of ER visits.
Clinicians, healthcare staff, and patients helped design the study.
What were the results?
Patients who had multiple or severe health problems were likely to have more ER visits but not poor blood sugar control.
After accounting for the number and severity of health problems, patients who lived in neighborhoods with more social risks were more likely to have poor blood sugar control. They were also more likely to have ER visits.
What were the limits of the project?
Patient health data came from community health centers. Social risk data were from the neighborhoods where patients lived and not from individual patients. Results may differ with other types of data.
Future studies could look at how including patients’ own social risk affects assessments of quality of care.
How can people use the results?
Health systems can include data on social risks to improve how well measures assess quality of care.
| How this project fits under PCORI’s Research Priorities The PCORnet® Study reported in this results summary was conducted using PCORnet®, the National Patient-Centered Clinical Research Network. PCORnet® is intended to improve the nation’s capacity to conduct health research, particularly comparative effectiveness research (CER), efficiently by creating a large, highly representative network for conducting clinical outcomes research. PCORnet® has been developed with funding from the Patient-Centered Outcomes Research Institute® (PCORI®). |
Professional Abstract
Background
Healthcare providers are increasingly evaluated based on performance measures for quality of care, including their patients’ health outcomes and frequency of emergency department (ED) utilization. These healthcare quality measures usually account for patients’ medical complexity. However, social risk factors like unstable housing or food insecurity, which may also contribute to patient outcomes, are not usually considered when evaluating the performance of healthcare providers.
Objective
To assess the combined effects of medical and social risk on two healthcare quality measures, glycemic control and ED visits
Study Design
| Design Element | Description |
|---|---|
| Design | Empirical analysis |
|
Data Sources and |
EHR data for 906,000 patients from the ADVANCE Clinical Research Network and 1.2 million patients from the OneFlorida Clinical Research Network; American Community Survey data; Medicaid claims data |
| Analytic Approach | Linkage of EHRs with geocoded community-level data from the American Community Survey and Medicaid claims data, predicted probabilities from multivariable logistic regression |
| Outcomes | Glycemic control using HbA1C levels, ED utilization |
Methods
Researchers used data from two PCORnet® Clinical Research Networks, ADVANCE and OneFlorida, which include many safety-net clinics. Data included electronic health records (EHRs) linked to Medicaid claims and geocoded community-level data from the American Community Survey. For 73,328 patients with diabetes and 988,106 patients with Medicaid coverage, researchers measured medical complexity using the Charlson Comorbidity Index (CCI) and the number of mental and behavioral health conditions. To measure patients’ level of social risk, researchers used patient addresses to identify the social deprivation index (SDI) for the geographic area where patients lived. The SDI is a composite score of that quantifies levels of disadvantage across small areas using seven domains from the American Community Survey, such as the percentage of people who are unemployed, living in poverty, or living in overcrowded housing. SDI scores range from 0 to 100 with higher scores representing higher levels of social risk.
Using logistic regression models, researchers analyzed the data with and without the SDI to assess the separate and combined effects of medical complexity and social risk factors on the odds of ED utilization and poor glycemic control. Regression models estimated how patient outcomes would change with variations in CCI and SDI.
Health system administrators, patients, healthcare providers, and clinic staff helped develop and test the methods.
Results
After adjusting for social risk factors, medical complexity was associated with higher odds of ED utilization (p<0.05) but not poor glycemic control.
Higher levels of social risk were associated with higher odds of poor glycemic control and ED utilization, even after adjusting for medical complexity. A 10-point increase in SDI was associated with 3%–5% increased odds of poor glycemic control (p<0.05) and 4% increased odds of ED utilization (p<0.05).
Limitations
Patient data were from community health centers. Results may differ for patients in other settings. The data on social risk factors were community-level data and may not apply to individual patients.
Conclusions and Relevance
Social risk factors affected measures of glycemic control and ED utilization even after accounting for medical complexity. Including data on patients’ social risk factors could improve the accuracy of healthcare quality assessments.
Future Research Needs
Future research could examine the effect of including individual-level social risk data on healthcare quality assessments.
| How this project fits under PCORI’s Research Priorities The PCORnet® Study reported in this results summary was conducted using PCORnet®, the National Patient-Centered Clinical Research Network. PCORnet® is intended to improve the nation’s capacity to conduct health research, particularly comparative effectiveness research (CER), efficiently by creating a large, highly representative network for conducting clinical outcomes research. PCORnet® has been developed with funding from the Patient-Centered Outcomes Research Institute® (PCORI®). |
Final Research Report
View this project's final research report.
Journal Citations
Related Journal Citations
Peer-Review Summary
Peer review of PCORI-funded research helps make sure the report presents complete, balanced, and useful information about the research. It also assesses how the project addressed PCORI’s Methodology Standards. During peer review, experts read a draft report of the research and provide comments about the report. These experts may include a scientist focused on the research topic, a specialist in research methods, a patient or caregiver, and a healthcare professional. These reviewers cannot have conflicts of interest with the study.
The peer reviewers point out where the draft report may need revision. For example, they may suggest ways to improve descriptions of the conduct of the study or to clarify the connection between results and conclusions. Sometimes, awardees revise their draft reports twice or more to address all of the reviewers’ comments.
Peer reviewers commented and the researchers made changes or provided responses. Those comments and responses included the following:
- The reviewers noted that when the researchers assessed the effect of social determinants of health (SDH), there were different effects whether the analytic model used community or individual SDH. The reviewers said that it was clear that if the analytic model already included individual SDH, the addition of community SDH did not have much effect. The reviewers asked for the researchers to add similar information about whether individual SDH had an effect on the results of the analytic model even when community SDH were already included in the analyses. The researchers added language to clarify that this was the case and agreed with the reviewers that individual SDH may possess a greater ability to provide actionable information to clinicians treating those patients.
- The reviewers asked the researchers to explain how they operationalized number of emergency department visits: whether that was calculated as one or more visits, or number of visits. The researchers acknowledged that the measurement of emergency department visits was not consistent, mainly because the data were collected differently at the different sites and across different phases of the study.
- One reviewer asked the researchers to clarify in the report that their research focused on one type of provider, community health centers. Therefore, the reviewer noted, the researchers’ advocacy for system-wide changes to ratings of provider performance metrics could lead to more resources for higher-income patients rather than resulting in more system resources going to patients with greater social complexity that could interfere with their health. The researchers agreed and clarified in the report their recommendations that system-wide reimbursement changes should not only take into account which providers had the best clinical outcomes, but also patients’ SDH, which were found to have a greater effect on outcomes than clinical complexity.
Conflict of Interest Disclosures
Project Information
Key Dates
Study Registration Information
^The original principal investigator for this project was Scott Fields, MD, MHA.