ED as a Clinic

By John Frias Morales, Dr. BA, MS (johnfmorales@gmail.com, 408.914.5765 cell)

Summary

Using the ED as a clinic costs 12 times more than a primary care visit (United Health Group, 2019). Fortunately, there’s a 17.8% ROI opportunity to shift costs using prognostics (Mercer, 2016). After an ED visit occurs, there’s another 30% readmission reduction opportunity from post-discharge transitions follow-up (Burton, 2012). This article describes how to capture value with end-to-end ED measurement modeling, with further detail noted in the citation sources. These are the secret formulas used by managed care top performers; also see PHM ROI article

Social Health

Many vulnerable low-income populations have more Emergency Department (ED) visits than primary care visits and are two times more likely to use the ED as a clinic (Udalova, et. al., 2022). The ED is seen as a last resort one-stop-shop by poor people who lack access to proximal clinics, transportation, and available appointments (Udalova). For example, pregnant Medicaid patients with limited access to primary care are 7 times more likely to used emergency care; heart disease patients with limited access to healthcare are 10 time more likely to use the ED (Prealize, 2023). Furthermore, newly diagnosed patients are often in denial about the level of self-care required to stop a chronic condition (e.g., diabetes, hypertension, asthma, & heart disease) from flaring up and progressing to a higher stage of acuity.

Value-Based ROI

Under new Value-Based Payment (VBP) methodologies (i.e., capitation and FFS incentivized quality performance), managing ED encounters is critical for achieving a favorable ROI. There are many opportunities to practice prevention in the continuum of care, but developing the capability manage ED utilization is a skillset that is transferable to IP and SNF utilization. This value proposition requires proactively shifting acute emergency interventions to outpatient clinics using a preventive model of care.? But how do providers and payers detect acute flare-ups that are avoidable and preventable?

The answer to ED redirecting requires integrating several sets of ED algorithms to manage the entire end to end cycle before and after the ED encounter. This method blends retrospective and prospective ED metrics, as well as use of social health and chronic condition segmentation strategies. This combination method allows managers understand risk factors, as well as protective factors that can reverse unfavorable trends. Protective factors are interventions that have the potential to manage a clinical condition and avoid care in expensive settings.

Proprietary software like the Johns Hopkins ACG and 3M PPV/EAPG are excellent world-class tools for carriers and providers who have industrial-strength analytic platforms. Fortunately, there are free evidence-based and open-source algorithms that can help detecting patters of clinic underutilization and ED overutilization. Open-source alternatives, however, require data scientists who can perform gymnastics with OP & ED encounter data.

Preventable & Avoidable ED

The New York University Emergency Department Algorithm (NYU-EDA) is an open-source algorithm that has been validated (Lemke, 2020). NYU-EDA applies 74,329 ICD-10-CM diagnosis codes to any retrospective ED encounters of interest. The algorithm then classifies these records as Nonemergent, Emergent, Injury, Mental Health Substance Use, and Unclassified. Once the algorithm is run, it’s possible to make inferences about access to primary care. This measure has been so widely used that it became a NCQA HEDIS measure (Avoidable Emergency Department Visits). A Mercer study (2016) demonstrated that potentially preventable ED visits can save 12.6% of dollars spent and low-acuity non-emergent ED costs can be reduced by 5.2%. The avoidable/preventable data set is most useful if it can be segmented by chronic/complex conditions or segmented by severity. Breaking out the data into subgroups helps with understand what additional case management referrals are needed for post-discharge follow up and longitudinal planning.

If an organization isn’t satisfied with a retrospective analysis, Hastings & Howison (2022) have shown that avoidable and preventable ED encounters can be predicted using data science techniques. Another useful AI model to predict avoidable ED visits identified risk factors such as age, # of chronic diseases, and digestive symptoms (Yang, et. al., 2022).

Before ED Visit

The goal of analyzing retrospective data before an ED encounter is to assess if the primary model of care is being utilized appropriately. Is the ED being over used for services that are readily available in the outpatient clinic?

To predict ED encounters, there is a model that predicts when an encounter will occur 30 days in the future (Gao, Pellerin, and Kaminsky, 2018). The Gao model leverages the open-source AHRQ CCS grouper to classify diagnoses and procedures, and the model leverages prior encounter history and patient demographics to enhance prognostics. ?Other prognostics that are predictive of an ED encounter include PCP visits, telehealth experience, new diagnoses, and more ED visits than OP visits. Stakeholders need to know if patients with a newly diagnosed condition have seen the PCP within 90 days of the ED encounter; the patient should not have more ED utilization than outpatient clinic visits.

Since low-income patients are most likely to be high ED utilizers, it’s important to analyze social health factors. ?To measure social health, there are many sources such as SDoH ICD-10-CM Z-Codes and program eligibility; additional sources are the Area Deprivation Index (ADI), Social Deprivation Index (SDI), and Neighborhood Deprivation Index (NDI). The number of office visit No-Show/Cancellation and the number of secured messages & call center contacts are also predictive of ED visits and should be analyzed using prognostic models.

After ED Visit

After an ED encounter, it is essential to measure post-discharge follow up visits within 7 days. The goal of this follow-up care is to ensure the primary care provider is aware of acute events are occurring so a longitudinal chronic care plan can be created for the patient. These follow up visits have been show to reduce readmissions. Follow-Up After Emergency Department Visit for People With High-Risk Multiple Chronic Conditions (FMC) is a HEDIS measure. This is a good place to start with analyzing ED encounters retrospectively.

Some organization have the capability to forecast ED patients who will return or be readmitted within 30 days post discharge. These are extremely valuable for transitions planning and prioritizing which patients need immediate follow up. The severity of each visit can be determined using a set of procedure codes (Jeffery, et. al., 2022). ?This type of prognostic model of ED encounters into the future can help with adjusting clinical follow up workflows to triage patients at risk of readmission.

And finally, if a high-risk group has experienced a break in insurance coverage or changed their PCP, it is important to identify these patients. Disruption in care most impacts cancer patients, pregnant women, dialysis patients, the developmentally disabled and tracheostomy and ventilator dependent patients. Anytime there has been a break in coverage or PCP change, there is a need to coordinate care to ensure the transition period does not result in an ED event. Using data to monitor transitions of care have been shown to reduce utilization in the ED.

Sources

Burton, R. (2012). Improving care transitions. Downloaded from

https://www.healthaffairs.org/do/10.1377/hpb20120913.327236/full/healthpolicybrief_76.pdf

Gao, K, Pellerin, G, Kaminsky, L. (2018). Predicting 30-day emergency department revisits. 24, 11. Downloaded from https://www.ajmc.com/view/predicting-30day-emergency-department-revisits

Hastings, JS, and Howison, M. (2022). Predicting divertible Medicaid emergency department costs. Digital Government Research & Practice, 3,3,19, 1-19. Downloaded from https://dl.acm.org/doi/full/10.1145/3548692.

Jeffery, MM, Bellolio, MF, Wolfson, J., Abraham, JM, Dowd, BE, and Kane, RL. (2016). Validation of an algorithm to determine the primary care treatability of emergency department visits. British Medical Journal, 6, 8. Downloaded from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5013457/

Lemke, KW, Pham, K, Ravert, DM, and Weiner, J. (2020). A revised classification algorithm for assessing emergency department visit severity of populations. American Journal of Managed Care, 26, 3, 119-125. Downloaded from https://www.ajmc.com/view/a-revised-classification-algorithm-for-assessing-emergency-department-visit-severity-of-populations

Mercer (2016). Oklahoma emergency department utilization July 2012 through June 2015. Downloaded from https://oklahoma.gov/content/dam/ok/en/okhca/documents/a0301/19763.pdf

Prealize (2023). State of health 2023 spotlight, social determinants of health. Downloaded from https://www.prealizehealth.com/2023-sdoh-report

Udoloya, V., Powers, D., Robinson, S., and Notter, I. (2022). Most Vulnerable More Likely to Depend on Emergency Rooms for Preventable Care. US Census Bureau. Downloaded from https://www.census.gov/library/stories/2022/01/who-makes-more-preventable-visits-to-emergency-rooms.html

United Health Group (2019). The High Cost of Avoidable Hospital Emergency Department Visits. Downloaded from https://www.unitedhealthgroup.com/content/dam/UHG/PDF/2019/UHG-Avoidable-ED-Visits.pdf

Yang, Y., Yu, J., Liu, S., Wang, H., Dresden, S., Lou, Y. (2022). Predicting Avoidable Emergency Department Visits Using the NHAMCS. AMIA Joint Summits Translational Science Proceedings, 514-523. Downloaded from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9285177/.