The Role of Telehealth and Clinical Informatics in Data Driven Primary Care Redesign

Abstract

Clinical informatics linked inpatient and emergency department use to clinical data to evaluate utilization for population segments. Trend analysis demonstrates how remote registered nurse care management and the COVID-79 pandemic reduced emergency department utilization in adult populations with high social needs.

Background

Continual innovations in remote patient monitoring and telehealth technology have created an opportunity to better manage medically complex patients using remote care management. Telehealth has been touted as having the potential to improve quality of care and access for high-need high-cost patients promising both integration and reduced costs (American Hospital Association (AHA), n.d.). Data is increasingly able to be easily transmitted to providers and care coordinators which can then be analyzed and acted upon allowing for appropriate interventions to be implemented to improve value-based care. By combining clinical information from the electronic health record (EHR) and data on hospital admissions and emergency department visits from the health information exchange (HIE), primary care offices can identify the high-risk population in their practice and compare effectiveness of chronic care management with usual care. To successfully implement these technological driven interventions, standards and best practice examples are needed to guide data flow, in form analysis and documentation, and improve interoperability.

As an example, the HIE in the Buffalo, NY, area is currently able to provide utilization data in real-time to primary care practices, creating an opportunity to evaluate the outcomes of interventions designed to improve care to high-need or complex patients. The Buffalo Interprofessional Advanced Primary Care (BIAPC) project is a collaboration between nursing researchers and an academic family medicine practice that aims to utilize data analytics to divide the patient roster into segments, target the complex population, and evaluate the impact of newly implemented interventions on outcomes. The goal of the 5-year BIAPC project is to improve care quality, reduce low-value care, and improve the patient’s and provider’s experience of care meeting the quadruple aim (Bodenheimer & Sinsky, 2014). The BIAPC project objectives included integrating HIE data, developing improved data analytics, and adding telephonic care management to improve continuity and utilization value.

As the BIAPC project was beginning to make advances in care coordination, the onslaught of the COVID-19 pandemic required providers to close their doors to all non-urgent healthcare needs resulting in decreases in inpatient, emergency department, and outpatient visits (Jeffery et 2020). Telehealth then served as the primary way for many hospitals and clinics to continue to manage patients with complex needs (Hollander & Carr, 2020). However, prior to the onset of the current pandemic, only 24% of healthcare organizations had telehealth programs in place (Finnegan, 2020). Noted barriers to widespread adoption of teleheatth have included reimbursement limitations (which were subsequently lifted), the high cost of technology, the limited ability of health care to adopt workflow changes, and lack of data analytic capabilities (AHA, n.d.). These challenges to adopting telehealth technologies have only been accentuated by the COVID-19 pandemic. A recent scoping review examined the use of telehealth during the COVID-19 pandemic and found most of the articles included were opinions and perspectives that expressed positive sentiment around the exponential growth of telehealth during the pandemic (Doraiswamy et al., 2020). Yet, Kaplan (2020) warns while the pandemic has created an explosion in telehealth use, the quick implementation rushed patients and clinicians into using technologies without considering access and usability issues, and further suggests more evidence is needed to address safety, research gaps, and emerging questions around the effectiveness of telehealth use.

The BIAPC interventions serve as examples of the challenges and advantages to implementing telehealth outreach and using clinical informatics to monitor outcomes. The coordinating transitions project provided the evidence-base for the BIAPC interventions to improve value of care to complex patients (Hewner et al., 2017). Critical factors in that study included real-time notification of discharges from the acute care setting; dividing the population into segments based on disease complexity and social risk factors; and timely outreach by clinical staff to improve continuity and value of care (Hewner et al., 2018). The BIAPC project leveraged existing technology, staff, expertise, and data from the EHR and HIE to compare the effectiveness of these interventions on reducing inpatient and emergency department utilization and adapting their care management practices to improve outcomes for high-need high cost individuals. This paper presents trends in utilization during 2018, 2019, and 2020 to evaluate the impact of implementation of a nurse managed telehealth outreach program and provider telehealth visits, both before and during the pandemic.

Methods

The focus of this observational study is utilization in the final 3 years of the 5-year project. The BIAPC team implemented the project in July 2016 in an academic family medicine practice that encompassed three primary care practice sites in the Buffalo metropolitan area. Physician providers had academic appointments and rotated through the sites. Clinical staff at the satellite practices included physicians, physician assistants, and medical assistants. Because this academic family practice setting included medical residents, they did not employ nurse practitioners. The central office staff included three practical nurses who provided care coordination as well as a full-time office manager and data analyst who provided data analytics and administrative services.

Clinical Workflow Redesign

Initially the practice relied on practical nurses in the central office to make outreach phone calls to recently discharged individuals; however, there was no effort to target the population at greatest risk for readmission, and low value inpatient and emergency department utilization remained at high levels. As the team moved toward developing a care plan for complex patients, the practice tried to recruit registered nurses into the care manager role; however, retention was an issue in this new role. In quarter two of 2019, the practice decided to outsource management of complex patients to a remote registered nurse (RN) care management program with a focus on the high-need high-cost population based on Medicare eligibility.

Information Management Redesign

Concurrent with the changes in the clinical workflow, the data aggregation and analytic capabilities of the practice Increased. At the start of the BIAPC project, the practice’s data analyst was extracting data from the EHR for quality metric reporting. The practice anticipated getting admission, discharge, and transfer (ADT) data from the HIE, but was unsure how to incorporate that data into the existing framework. The analytic team lead used practice-specific data extracted from the New York State (NYS) Medicaid Data Warehouse to demonstrate the utility of combining utilization data and chronic disease diagnoses to identify and target the population for care coordination. The practice was able to add utilization and the chronic conditions to their table of data extracted from the EHR.

The outcomes analysis aimed to facilitate the practice’s transition to data-driven care for high-need high-cost patients. This goal required enhancement of the clinical summary database, integration of new data sources such as ADT notification, redesigning clinical workflow to facilitate transitional care after discharge from the acute setting, and data analytics that facilitate evaluation of the impact of interventions on clinical outcomes. Over time, the practice evolved from sporadic telephone outreach after discharge to an approach that targeted a high-need high-cost population using outsourced RN care coordination as an adjunct to the interprofessional team. The results section will illustrate how – prior to the pandemic – clinical practice redesign had begun to influence utilization outcomes of the high-cost high-need population and will also present the changes in patterns that resulted from the pandemic.

Clinical Complexity Algorithm

Hewner and colleagues developed the clinical complexity algorithm in 2005 while employed as a population health analyst at a regional managed care organization. The specifics of the algorithm are described elsewhere (Hewner et al., 2017; 2018). Essentially, the algorithm classifies individual patients for the presence of 12 chronic conditions induding chronic kidney disease, heart failure, coronary artery disease, diabetes, chronic obstructive pulmonary disease, substance use disorder, mental health diagnosis, asthma, obesity, hypertension, lipid disorder, smoking (listed in order of complexity from highest to lowest), and then ranks patients based on the complexity of the condition (with heart failure and chronic kidney disease as most severe) and presence or absence of comorbid chronic conditions (such as diabetes and coronary artery disease). Persons with major chronic conditions (asthma through chronic kidney disease) are considered in the chronic cohort and the remainder (non-chronic cohort) can be divided into those with at-risk conditions and those without. Each case on the roster is assigned to their disease classification and cohort with their total number of encounters (inpatient, emergency department, and outpatient) for the time frame (quarter years). Results are aggregated to age groups, insurers, and cohorts for this analysis.

Results

At present, the family medicine practice serves a roster of 17,570 patients that includes children under age 18 (6%), adults aged 18–64 (70%), and older adults (24%). Since 2017, when one inner city practice relocated to a new site, the majority (77%) of patients identify as White non-Hispanic. lndividuals with at least one major chronic condition (chronic cohort) make up 57% percent of the roster. Because the majority of the population is in the 18–64 age group, we will highlight the results for that age group and will consider those eligible for Medicaid and the disabled population (Medicare recipients under age 65) as those with high social need.

Complexity by Insurer

The vast majority of the 12,229 patients ages 18–64 included in the analysis are commercially insured (73%) with 19% insured by Medicaid and 8% insured by Medicare (see Table 1). The clinical complexity algorithm divides the population into two cohorts with 55% classified as chronic and 45% classified as non-chronic. The percentage of Medicaid or Medicare patients in the chronic cohort (n = 6,723) increased to 35%, while only 65% are commercially insured. In the non-chronic cohort (n = 5,506) only 20% are insured by Medicaid or Medicare. However, 50% of those with commercial insurance have chronic conditions which represents the largest portion of patients on the roster. Thus, the large commercially insured population has a major impact on utilization rates in both the chronic and non-chronic cohort.

Table 1.

Percentage of Patients by Complexity and Insurer Age 18–64, N = 12,229

	Chronic Cohort	Non-Chronic Cohort	Total Percent
Medicare	1,478	899	8%
Medicaid	802	180
Commercial	4,443	4427	73%
Total Percent	55%	45%

Inpatient Utilization

In 2018, when the project began, rates of inpatient utilization in the chronic cohort (ages 18–64) were 2.6% compared to 1% in the non-chronic cohort (see Figure 1). Therefore, BIAPC interventions focused on the chronic cohort of patients where inpatient utilization rates decreased over the project period dropping to 2.2% in the fourth quarter of 2019. Over time the utilization patterns of the chronic cohort began to merge with that of the non-chronic cohort with the non-chronic cohort inpatient use surpassing that of the chronic in the fourth quarter of 2019. Within the chronic cohort, rates for persons with commercial insurance were 1.4% in 2018, compared to 4% for Medicaid, and 6.6% for those with Medicare insurance.

Figure 1.

Percent of IP Utilization Chronic Cohort vs Non-chronic Cohort Age 18–64, N=12, 229

Outpatient Utilization

Outpatient utilization mirrored that of inpatient utilization at the onset of the project with the chronic cohort being seen at much higher rates that than the non-chronic cohort (see Figure 2). Similarly, chronic cohort outpatient use trended down over time going from a high of 55%, down to 43% by the fourth quarter of 2019 while non-chronic use increased eventually merging with that of the chronic.

Figure 2.

Percent of OP Utilization Chronic Cohort vs Non-chronic Cohort Age 18–64, N=12,229

Emergency Department Utilization

In the chronic cohort, 8.5% of the (18–64) population had an emergency visit in quarter one in 2018, and this rate was more than double that of the non-chronic cohort (3.5%) (see Figure 3). Emergency department utilization was of particular concern in the non-commercially insured population. Figure 4 shows that the emergency department utilization rate was about three times larger in the Medicaid (15.6%−20%) and Medicare (between 15.6%−22.6%) population in the chronic cohort than in the commercially insured populations (4.8%−5.4%). However, Medicaid and Medicare emergency department utilization dropped to near commercial levels in quarter three in 2019.

Figure 3.

Percent of ED Utilization Chronic Cohort vs Non-chronic Cohort Age 18–64, N=12,229

Figure 4.

Percent of Chronic Cohort ED Utilization by Insurer Age 18–64, N=6,723

Impact of Telehealth

A sharp decrease in emergency department utilization in the non-commercially insured chronic cohort was seen beginning in the third quarter of 2019 when the practice began to outsource the care coordination of a subset of chronic patients to remote RN care management (see Figure 4). In contrast, the non-chronic cohort emergency department use (see Figure 3) trended up over time merging with that of the chronic and peaking in the first quarter of 2020 at 7.5%. The remote RN care management started making outreach phone to non-commercial complex cases in June of 2019. During the pandemic the remote RN care management continued with 238 of the complex patients ages 18–64 on the practice roster receiving outreach in 2020.

Impact of COVID-19

In 2020 when the region shut down, the practice swiftly transitioned to telehealth visits with 3,678 telehealth visits taking place for patients ages 18–64 (see Figure 2). Tetehealth visits were made to both the chronic and non-chronic patients that were scheduled for routine or follow up visits or were experiencing medical issues that needed addressing.

There was surprisingly little change in the inpatient utilization in both cohorts during 2020 when the pandemic caused cancellation of elective procedures (see Figure 1). The emergency department utilization dropped significantly by the second quarter of 2020 with the non-chronic patient use (6.5%) either merging with or surpassing the chronic cohort (5.9%) (see Figure 3). Emergency department use in the chronic cohort – which one might have expected to increase during the pandemic when usual sources of care were not an option – decreased and remained trending down in 2020; however, the non-chronic cohort continued to use the emergency department at higher than previous rates. The outpatient utilization data (see Figure 2) showed similar patterns over the course of the COVID-19 pandemic with the most abrupt decrease seen in the first and second quarter of 2020 when the outpatient offices were shut down and the chronic and non-chronic cohorts being seen at similar rates.

Upon closer examination of patients ages 18–64 who used the emergency department during the pandemic in 2020, 1,742 patients had a total of 3,026 emergency department visits which is comparable to the 3,338 visits seen in 2018 (see Table 2). However, emergency department visits increased during the pandemic for the non-chronic population, while emergency department visits in the chronic population decreased. Of the 1,742 patients who had an emergency department visit in 2020, 114 had four or more visits. The disease categories of healthy, obese (BMI > 30), mental health, and substance use disorder accounted for 68% of the total visits in these frequent users (see Table 3). Only 30 of the 114 patients with four or more emergency department visits in 2020 received either a telehealth visit or were part of the remote care management group. Of the 30 high utilizers who participated in telehealth during 2020, 26 utilized telehealth appointments and only four of the patients were being managed by the RN care management.

Table 2.

Total ED Utilization per Year by Cohort Age 18–64

ED Visits	2018	2019	2020
Chronic Cohort	2,492	2,465	1,641
Non-Chronic Cohort	846	1,154	1,385
Total Visits	3,338	3,619	3,026

Table 3.

2020 Frequent ED Use (4+ visits) by Disease Category

Disease Category	Number of Patients
Healthy (without chronic conditions)	24
Obese (SMI > 30)	23
Mental Health Diagnosis	17
Substance Use Disorder	13
Diabetes	10
Asthma	8
Chronic Obstructive Pulmonary Disease	7
History Of Smoking	6
Chronic Kidney Disease	2
Heart Failure	2
Coronary Artery Disease & Heart Failure	1
Hypertension	1
Total Patients	114

Discussion

In the first 3 years of the project, few changes in utilization were noted as the practice learned how to understand their patient population and analyze their data. Small decreases in inpatient utilization in the chronic cohort in 2018 and the first half of 2019, can be attributed to the redesign in clinical workflow at the practice which included sporadic outreach by the interprofessional team to follow up after discharge. However, the large drop seen in inpatient and emergency department utilization in the third quarter of 2019 was when the practice – after ongoing analysis of patient utilization data – made the decision to outsource the telephone outreach to an RN care management company with a focus on the complex chronic cohort. As described by the practice manager and supported by McConnell (2019), decreases noted in outpatient visits prior to COVID-19 can be attributed to reimbursement changes in primary care implemented at that time requiring practices to show improvements in low value utilization and away from the fee for service model.

Limitations of the study include only having access to the data that the practice collects to analyze the impacts of interventions on utilization outcomes. For example, we are unable to obtain a count of how many outreach efforts each patient has received from the remote care management, but only that they are a patient that is a part of that intervention and has received at least one call. The observational study design also limits the availability of a control population and results in the inability to account for other external variables that may have had an impact on utilization patterns.

The data clearly demonstrates the combination of continued RN care management outreach during the pandemic, combined with telehealth visits implemented by the practice, resulted in keeping chronic patients out of the hospital and the emergency department during a period when the office was closed. During the COVID-19 pandemic, cohort differences in utilization patterns disappeared with the non-chronic utilization mirroring that of the chronic. Furthermore, differences in inpatient, emergency department, and outpatient utilization by type of insurer were reduced over the course of the project, although it is important to remember than the non-commercially insured populations combined make up only 27% of the 1864 aged population.

The practice’s implementation of telehealth supported care management of chronic conditions and may have prevented emergency department visits or an inpatient stay. Frequent utilization is defined by the number of emergency department visits or hospitalizations and four or more visits in a 12-month period has been noted in many analyses to be the definition of frequent emergency department use (Vinton et al., 2014). Patients with frequent emergency department visits often did not have major chronic conditions or had behavioral health comorbidities (mental illness or substance use). Frequent utilizers may have been socially complex and were not having their social complexities addressed. Studies have found 4%−8% of emergency department patients account for 21%−28% of all emergency department visits and at least 30% of all emergency department visits are non-urgent (Uscher-Pines et al., 2013). The BIPAC utilization data supports this with only 1% of the roster accounting for 26% of the overall emergency department use during the pandemic and non-chronic disease categories driving utilization.

Future Recommendations

Future studies aimed to understand low value utilization must consider how to better use clinical informatics to screen for and address the care management burdens and social determinants that impact patient’s healthcare choices. Based on the results of this project, future studies should consider both telehealth visits and remote care management as potential interventions. Although we cannot conclude that low value high-cost utilization decreased overall due to interventions, the pandemic has allowed us to see how much of emergency department utilization is truly excess especially for those with social complexity.

To capitalize on the opportunity to base practice redesign decisions on utilization data, nurse informaticists need to obtain data from a source that includes all hospitalizations and emergency department visits in the region, such as the HIE. Next, it is important to clearly define the target population and use an appropriate algorithm to segment the population of interest. The complexity algorithm (Hewner et al., 2017; 2018) applied in this study not only allows for analysis by specific disease categories, but also allows to segment the population into chronic and non-chronic cohorts. These elements in addition to consistently tracking data over time — allow the informatics nurse, or researcher, to understand the impact of interventions on utilization.

Conclusion

Data sharing with the primary care practice and quarterly analysis of patient data showed the impact of both remote care management and telehealth on utilization patterns. Emergency department utilization trends over time demonstrated those under 65 with Medicaid and Medicare, and the non-chronic super-utilizers, had more visits than would be expected based on medical complexity alone. This would suggest there is excess utilization due to social complexity.

A CDC report addressing health disparities during the COVID-19 pandemic noted an increase in emergency department visits related to social and behavioral determinants of health in the complex patient population that has no other options for care (Hartnett et al., 2020). Although Hartnett and colleagues (2020) recommend continued use of telehealth visits and triage help lines to minimize transmission risk, persons with complex social and behavioral needs often lack the technology and skill required to seek alternative healthcare avenues. This small population of socially complex patients account for a large proportion of healthcare costs, and failure to address the complex needs of persons with multiple chronic conditions and/or socioeconomic disadvantage contributes to frequent emergency department and inpatient utilization (Long et 2017).

Evaluation of utilization during the COVID-19 pandemic demonstrated there is untapped potential to improve patient outcomes using telehealth in the primary care setting (Jetty et al., 2021). The CDC predicted pandemic-related increases in telehealth in primary care may have long-term effects on low value emergency department use (Koonin et al., 2020). Interventions in both the primary care and acute care settings are critical to efforts to reduce emergency department and inpatient utilization. However, nurses working in informatics are often employed in the acute care setting and are focused primarily on institutional data, and primary care offices often do not have the resources to employ informatics specialists. Thus, nurse informaticists must take a systems-thinking approach that integrates both acute and primary care settings to address the issue of readmissions and emergency department recidivism.