key: cord-0934108-x22na0t2
authors: Jose, Thulasee; Warner, David O.; O’Horo, John C.; Peters, Steve G.; Chaudhry, Rajeev; Binnicker, Matthew J.; Burger, Charles D.
title: Digital Health Surveillance Strategies for Management of Coronavirus Disease 2019
date: 2020-12-14
journal: Mayo Clin Proc Innov Qual Outcomes
DOI: 10.1016/j.mayocpiqo.2020.12.004
sha: 61586f837c8871e5de160ec8bd4dfd4b94bca4fe
doc_id: 934108
cord_uid: x22na0t2

Objective To describe the design, implementation and utilization of Electronic Health Record (EHR) based digital health surveillance strategies used to manage the coronavirus disease (COVID-19) pandemic and ensure delivery of high quality clinical care such as case identification, remote monitoring, telemedicine services, and recruitment to clinical trials at Mayo Clinic. Setting Mayo Clinic, a large multistate integrated health care system with over 1.5 million annual patient visits that utilizes the Epic© electronic health record (EHR) system. Methods and Results Rule-based live registries were designed in EHR system to classify patients who 1) are currently tested positive for COVID-19; 2) were tested positive but has recovered from COVID-19; 3) are suspected of COVID-19 but did not yet meet clinical diagnostic criteria,; 4) had tested negative for COVID-19; and 5) exceeded a risk score for serious complications from COVID-19. Using registries, custom dashboards and operational reports were developed to provide daily high-level summary for clinical practice use and provide up to date information used to manage individual patients impacted by COVID-19, including support of case identification, contact isolation and other care management tasks. Conclusion we developed and implemented a systematic approach for using EHR patient registries to manage the COVID-19 pandemic that proved feasible and useful for managing a large multistate group clinical practice. The key to harnessing the potential of digital surveillance tools to promote patient-centered care during COVID-19 pandemic was to use the registry data, reports and dashboards as informatics tools to inform decision-making.

The coronavirus disease 2019 (COVID-19) pandemic continues to challenge the health system with the need to make operational decisions based on limited data and rapidly evolving clinical practice guidelines. [1] [2] [3] [4] [5] [6] [7] Digital health surveillance applications based on electronic health record (EHR) data can help address the challenges posed by COVID-19, including needs for health surveillance, screening, triage, diagnosis, and monitoring. New approaches include adoption of novel digital patient-facing self-triage and self-scheduling tools. 8 However, there is little experience with using live -rule based EHR patient registries to manage a pandemic crisis.

A well-executed patient registry can provide a real-time view of the clinical practice, patient outcomes, safety, and comparative effectiveness. 9 In this report, we present an approach to using EHR patient registries in the management of COVID-19 in a large multistate group practice. As the COVID-19 pandemic continues to impact the health system and the clinical practice, we share this experience to provide insight to other institutions that use similar interoperable EHR systems.

Mayo Clinic is a large academic health system with primary campuses located in 

Mayo Clinic Laboratories (MCL) is the laboratory referral center for Mayo Clinic. MCL specializes in esoteric laboratory testing for health care organizations throughout the United States and around the world. Diagnosis of coronavirus disease 2019 (COVID-19) illness is due to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Molecular tests are used to detect genetic material of the SARS-CoV-2 virus using polymerase chain reaction (PCR). An antigen test is a newer diagnostic COVID-19 test that can detect certain proteins that are part of the SARS-CoV-2 virus. In this report, the current FDA-approved diagnostic tests for COVID-19 utilized by Laboratory Medicine and Pathology, Mayo Clinic include both the molecular and antigen techniques.

Currently, five molecular methods are available at Mayo Clinic in Rochester, MN for detection of SARS-CoV-2 RNA in upper respiratory swab samples (e.g., nasopharyngeal swab, nasal swab, throat swab) and/or lower respiratory specimens (e.g., sputum, bronchoalveolar lavage and tracheal secretion). First, a laboratory test targeting the nucleocapsid gene of SARS-CoV-2; samples are extracted on either the bioMérieux easyMAG or eMAG instruments, J o u r n a l P r e -p r o o f Jose 6 followed by PCR amplification on the Roche LightCycler 480. Second, the Roche SARS-CoV-2 real-time PCR assay is performed on the automated cobas 6800 or 8800 instruments, which perform both extraction and amplification. Third, samples may be tested by the Abbott SARS-CoV-2 RealTime assays on the m2000 instrument. Fourth, the fully automated Hologic Panther system is a transcription mediated amplification assay used to detect SARS-CoV-2 RNA in clinical samples. And finally, the Luminex ARIES system is a rapid (2 hour), sample-to-answer system that is offered to detect SARS-CoV-2 RNA in upper respiratory swab specimens. Each of these methods has received emergency use authorization from the FDA, has a limit of detection (LoD) in the range of 100 to 500 copies/mL and an analytical specificity approaching 100%. The diagnostic and management procedures for COVID-19 at Mayo Clinic has been previously described. 4

The registry framework was developed through rapid cycles of iterative processes under the oversight of the Mayo Clinic COVID-19 Research Task Force. 10 Five live COVID-19 registries were developed utilizing EHR data collected from the COVID-19 diagnostic and serological testing strategies already implemented by the health system. 3, 11, 12 Based on pre-determined rules criteria, patients are placed in respective COVID-19 registry (Table 1) 

Dashboard provided a view of the entire clinical practice by listing current and historical statistics of key practice performance indicators for the health system. The information is summarized in a tabular and/or graphical view for conducting performance analysis of the entire clinical practice. Dashboards allow quick detection of negative trend development and help the health system to take immediate action. The COVID-19 registries were used to develop tailored COVID-19 dashboards. Recovered (former cases), COVID-19 Negative (tested, but not positive) and COVID-19 suspected (met some criteria to suggest testing, but diagnosis was not yet finalized).

The COVID -19 patient registries were implemented for clinical practice use starting May 5, 2020. As of December 07, 2020, a total of 765, 324 COVID-19 tests were completed across the health system, with test results and other information that flowed real-time to the COVID-19 registries. The utilization of the new COVID-19 registries by our health system is depicted in Figure 1 .

Customized dashboards were implemented using data from COVID-19 registries. All dashboards used to manage patient care were created in Epic so that the information was available in the workflow. Each dashboard was tailored according to the needs of the end-users and enabled users to slice data for additional insights including current trends on testing capacity, bed capacity, ventilator utilization, and hospitalization statistics. The dashboards enabled clinicians and scientists to visualize data in a self-service capacity for purposes directly relevant to their practices. For example, the surgical services use the COVID-19 serology dashboard Several operational reports were also created using the COVID-19 registries data. These reports provide information that is used at an operational level to identify and manage needs of patients.

For example, a resource is made available for COVID-19 patients who tested positive or had pending test results that were being discharged from the hospital without a means of J o u r n a l P r e -p r o o f Jose 10 transportation. The registry data was used to arrange transportation pro-actively rather than reactively. Another example is the remote monitoring care team COVID-19 report used to notify, monitor, and advise patients with confirmed or probable for COVID-19. Reports generated from these registries were used to manage patient care across various clinical care settings for multidisciplinary purposes including scheduling appointments. For the inpatient practice, a COVID-19 report generated using the registry facilitates nursing leadership assessment of staffing and personal protective equipment required for hospitalized COVID-19 patients. A related subsequent report assists the care team responsible for monitoring the postdischarge status and home care.

A comprehensive report library in EHR system serves as the repository for all the reports designed to provide information on COVID-19 patient populations. Providers use the report library to find COVID-19 status of their respective patient panel or inpatient list. The remote communication and bulk mailing tools within registries assist with the prompt notification, monitoring, and linkage to the required medical and support services for patients. Targeted content is tailored for the various administrative and healthcare delivery teams. The registries are also the primary source used to identify patients eligible for relevant clinical trials, managed centrally through an institutional research COVID-19 task force. For all applications, users are instructed to only utilize the reports that are applicable to their role and business need and observe Mayo Clinic policies for privacy and confidentiality.

Registries have demonstrated utility in the management of many health conditions. 13-24 We rapidly developed implemented and utilized EHR patient registries to manage the daily pandemic. The primary function of the COVID-19 registries was to equip our health system to process real-time data and classify patients into the appropriate diagnostic category. The registries also complimented other innovative contributions from our health system. 2, 3, 10-12, 25-39

The COVID-19 registries are used extensively to manage the daily operations of the clinical practice. For example, surgical services utilize patient registry data and dashboards ( Figure 1) to develop a new clinical workflow process that was implemented at the preoperative evaluation clinic to manage the testing, symptom screening, patient education, and preoperative optimization of surgical patients 39 In addition, artificial intelligence is applied to the registry data to enhance screening and diagnosis for high risk patients utilizing various risk scores. Registry data is also used to coordinate with state and federal health agencies to optimize supply chain management decisions and allocation of staff and resources across sites. 37, 40 COVID-19 dashboards and have been used at the population health level. 41, 42 Many medical specialties have already invested into developing COVID-19 patient registries. [43] [44] [45] [46] Although useful for research, patient registries outside EHR system have limited capacity to receive real-time data on patients. A large regional public academic health center used rapid COVID-19 registry, the primary focus was to support telehealth during the pandemic. 51 To that end, we built dashboards and analytical reports for various stakeholders in the health system according to their current as well as anticipated (e.g. resource allocation) clinical practice needs. Broad access to the EHR self-service analytics tools also enable users create ad hoc reports at their discretion. The remote monitoring type reports (using the registry data) allow clinicians and staff the ability to monitor their specific cohorts of interest. Examples of monitoring include the ability to track status of infection, testing dates, follow-up dates, etc.

Compared to other EHR optimization efforts to manage COVID-19, 41, 51, 52 our digital health surveillance strategies not only supported workflow optimization for tailored health services, but also provided a basis for COVID-19 research to understand long-term effects COVID-19 may have in patients. For example, our registry data can be used to generate cohorts to compare patients that ever tested positive vs. never tested positive vs. COVID-19 cases diagnosed on or after April 1, 2020. The code could only be used for cases diagnosed on or after April 1, 2020. Generally, COVID-19 cases diagnosed before April 1, 2020 were reported using ICD-10 codes that describe the patient's signs, symptoms, or associated illnesses. ICD-10 codes and associated guidelines may continue to evolve as more is learned about the virus clinical manifestations, transmission, and long-term effects. Of note, the COVID-19 registry required frequent maintenance as might be expected during a pandemic. For example, when the CDC changed guidelines, registry rules were changed accordingly. Our methodology employed is generalizable to the extent that other health systems also utilize Epic© can adopt our registry framework. However, the registry methodology and specific application has not been validated in other patient populations.

In conclusion, we developed and implemented a systematic approach for using EHR patient registries to manage the COVID-19 pandemic that proved feasible and useful for managing a large multistate group clinical practice in the U.S. The key to harnessing the potential of digital surveillance tools to promote patient-centered care during COVID-19 pandemic was to use the registry data, reports and dashboards as informatics tools to inform decision-making. J o u r n a l P r e -p r o o f 

Age • 3 points for age > 80 yrs.

• 2 points for age between70-79 yrs.

• 1 point for age between 60-69 yrs.

• 0 points for age < 60 yrs. 

Guide to understanding the 2019 novel coronavirus

Mayo Clinic Strategies for COVID-19 Elements of an Effective Incident Command Center

Drive-Through Testing: A Unique, Efficient Method of Collecting Large Volume of Specimens During the SARS-CoV-2 (COVID-19) Pandemic

Thank you for this suggestion to improve the manuscript. We added several examples in the results section of how the registry was used for both administrative and practice support. 1. Page 6, 2nd paragraph -is very awkward in general, needs edited.We edited the paragraph to improve readability and clarity. 2. Page 9, last paragraph is also awkwardly written and needs edited.We edited and rearranged the paragraph and included this into discussion section to improve readability and clarity. 3 . Table 3 is very difficult to follow. I would suggest using bullet points for listings of the Dashboard content or visual examples of the actual dashboards themselves.Thank you for your assessment. In response, we have deleted table 3 

Reviewer: 2 This is a brief report on an implementation of a module within the EHR system (EPIC). The concept is not original as many efforts are in place to use real-world data to characterize cohorts, assess severity, and so on. To be of interest, I would have like to have seen more information on the implementation, usability and utility, data quality assurance and impacts on decision-making and management of COVID19 cases.Thank you for the opportunity to improve the manuscript. To address, we have added more detailed examples of the implementation, usability and utility, data quality assurance and impacts on decisionmaking and management of COVID19 cases in the results section. Reviewer: 3 J o u r n a l P r e -p r o o f Thank you for the privilege to review your manuscript. I think it is valuable and exemplary of… --Utilizing data and dash boards to immediately manage the pandemic and its complexities with efficiency and speed -clearly a very valuable tool. The manuscript does a nice job bringing the concept to life and sparking the imagination.--Demonstrating how the pandemic has again spurred innovation, catapulting us into the future. To me, this is the most exciting aspect of this manuscript. The authors describe the concept so well that it is easy to see the potential to move from this stage to machine learning/AI, predictive/CART decisiontrees, which would be an advantage for decision-makers (physician wellness), patient-centered decisionmaking (patient experience), improving quality while optimizing efficiency and resource management, thus access and value. It also eludes to the potential to automate and standardize rounding, reduce manual documentation, facilitate communication, coordination, and collaboration between multiple stakeholders -with everyone looking at the same dashboard information. The "self-serve" reports are very exciting. I believe most think we should routinely have access to such dashboards to manage our practices. To my knowledge this is the first paper to convincingly describe that is immediately possible. Like so many things, we needed the pandemic innovate and improve. (As you say pg 10, line 35).Overall, the paper is short and conceptual. Though, in some instances, the factors are described in detail (eg. COVID testing), whereas other factors are not described at all (eg. clinical and lab factors, comorbidities -described elsewhere, ref.4, please refer to the tables). The "registry framework" is also published elsewhere, ref.10. The "additional data collected in each COVID-19 registry" seems complete -please refer to the tables.Thank you for your complimentary comments. We share your excitement for this work and the potential expansion in support of administration, practice and research. We also appreciate your specific questions and suggestions. In an effort to respond and improve the manuscript, we have provided additional information below with corresponding changes in the revised manuscript. ** Page 7, line beginning 49: Could you describe the software used? Were the registries made within Epic or exported from Epic? Were the dashboards made in epic? Or using Tableau, or other software?All the registries are created within Epic to permit access in the normal workflow and promote enterprise wide access and use. The dashboards used by clinicians were also made in Epic. The executive dashboard use registry data but was created outside Epic using Tableau Server. We added these details in the revised manuscript ** page 8, line beginning 11: What was the process of making tailored reports for care teams? What this resource intensive? Would anyone with an informatics background be able to do this, or what skills are needed? Could you give specific examples of reports that were commonly generated and for what purpose?J o u r n a l P r e -p r o o f Enterprise wide, we have a streamlined process to allow end users to create reports based on need. Additional reporting resources are provided by the enterprise on a priority basis. Reporting is a resource intensive process but we have sufficient IT infrastructure already in place to serve the needs of the enterprise. Informaticians are available across enterprise to streamline requests and create reports for end-users. Base reports using the registries were built for each registry. Several columns pulling registry data were also made available. Ideally this provided a sound base for users to more easily customize the reports without requesting new build.An example of a COVID-19 report generated from the registry that we have added to the revised manuscript is the inpatient nursing report to track hospitalized COVID-19 to insure appropriate staffing and PPE. The report follows COVID-19 patients from admission to discharge including serial SARS-CoV-2 testing results. A subsequent related report generated from the COVID-19 registry data was used by the care team to monitor patient health while recovering at home. We added these details to the revised manuscript. ** page 9, line beginning 23:These uses are very exciting. In the discussion could you give a few examples how these uses improved circumstances, outcomes or experience? And how these reports and dashboards were able to do things not possible without the dashboard? Specific examples how the dashboard saved the day?Surgical dashboards were used COVID-19 registry data to insure and monitor testing of patients prior to surgery. Surgical services also used the registry to manage PPE and staffing. The registry provided the necessary clinical and lab results via the dashboards and eliminated tedious manual data collection that may have delayed patient care. Importantly, the availability of the data in Epic permitted use in the daily workflow of the surgical services. We added these examples to the revised manuscript.Our remote monitoring type reports (using the registry) allow clinicians and staff the ability to monitor their specific cohorts of interest. Examples of monitoring include the ability to track status of infection, testing dates, follow-up dates, etc. and resolve the infection directly from the report. Without a live registry, real time analytics of COVID-19 status of patients moving through health system is difficult. An external registry will result in delayed information transmission and would lead to delay in care. We add these examples to the manuscript.Was the process predictive (PPV or NPV), reliable? We know the COVID tests are too often false negatives, how did the additional clinical and laboratory data improve PPV or clinical decisionmaking? Did those factors and comorbidities predict outcomes? Or was it descriptive data?We agree that molecular tests for COVID-19 may be negative in select cases, and therefore, should not be used in isolation to make a diagnosis or predict outcomes. Instead, we recommend that a combination of laboratory testing and clinical criteria be used to determine the likelihood of a patient having COVID-19, or not. Our experience is that the combination of these factors increases the predictive value (both PPV and NPV) of diagnosis and enhances clinical decision-making.J o u r n a l P r e -p r o o f **page 11, line beginning 39: There must have been more obstacles. How accurate was ICD-coding, etc? Were data sets accurate and complete? Reliable? Accurate? Were their complaints or shortcomings, plans for the future? Is there reason to think this methodology would not be generalizable? You describe a large disparate population spanning the nation We appreciate your insight and continue to face several obstacles. For example, as the CDC changed guidelines and retracted rules, corresponding rule changes in the registry were required. A registry built to track a dynamic situation such as a pandemic must be viewed as a "live" resource that requires diligent maintenance and regular updates. Nonetheless, the registry served as a versatile informatics tool for the clinical practice. The methodology employed in this manuscript is generalizable to the extent that other health system that also has Epic can replicate our efforts.We initially had encounter and problem list diagnoses as inclusion criteria, but later removed them leaving only testing data and infection status to identify the cohort. The issue with the encounter diagnosis was the wrong selection of an ICD10 code when placing the SARS-CoV-2 lab test. A subset of users did not select the 'suspected' code when placing the initial order and if the test came back negative the diagnosis might incorrectly add the patient to the registry. The problem list had less issues with inaccurate coding, although was removed from the inclusion rule due to a lack of resolving the problem after the infection was resolved. The active problem list diagnosis kept the patients in our 'Active Confirmed Cases' registry even though they were classified as recovered.I would like to see more manuscripts like this one, so more physicians can advocate that we move faster to make such dashboards routinely available.Thank you for your rigorous review, insight and kind comments!