a b s t r a c t

Objectives: Routine emergency department (ED) HIV or HCV screening may inadvertently capture patients al- ready diagnosed but does not specifically prioritize identification of this group. Our objective was to preliminarily estimate the volume of this distinct group in our ED population through a pilot electronic health record build that identified all patients with indications of HIV or HCV in their EHR at time of ED presentation.

Methods: Cross-sectional study of an urban, academic ED’s HIV/HCV program for previously diagnosed patients August 2017-July 2018. Prevention program staff, alerted by the EHR, reviewed records and interviewed patients to determine if confirmatory testing or linkage to care was needed. Primary outcome was total proportion of ED patients for whom the EHR generated an alert. Secondary outcome was the proportion of patients assessed by program staff who required confirmatory testing or linkage to HIV/HCV medical care.

Results: There were 65,374 ED encounters with 5238 (8.0%, 95% CI: 7.8%-8.2%) EHR alerts. Of these, 3741 were

assessed by program staff, with 798 (21%, 95% CI: 20%-23%) requiring HIV/HCV confirmatory testing or linkage to care services, 163 (20%) for HIV, 551 (69%) for HCV, and 84 (11%) for both HIV and HCV services.

Conclusions: Patients with existing indication of HIV or HCV infection in need of confirmatory testing or linkage to care were common in this ED. EDs should prioritize identifying this population, outside of routine screening, and intervene similarly regardless of whether the patient is newly or previously diagnosed.

(C) 2020

Introduction

human immunodeficiency virus and Hepatitis C (HCV) are high-priority threats to health and key drivers of Healthcare costs [1]. Diagnosis can lead to linkage and adherence to medical treatment, sub- sequently improving health outcomes and reducing new transmissions [2]. Routine screening in medical settings has been recommended by the Centers for Disease Control and Prevention (CDC) since 2006 [3]. The screening process typically involves 1) an initial (i.e. rapid) test for antibodies (HIV and HCV) and antigens (HIV only), 2) antibody dif- ferentiation immunoassay (for HIV only if rapid test is positive to specif- ically detect HIV-1 and HIV-2), and 3) a follow-up confirmatory RNA

* Corresponding author at: 231 Albert Sabin Way ML 0769, Cincinnati, OH 45267-0769, United States of America.

E-mail addresses: [email protected] (R.M. Ancona), [email protected] (C. Hamilton), [email protected] (F.J. Fernandez), [email protected]

(K.A. Faryar), [email protected] (B.H. Lane), [email protected] (M.S. Lyons).

test if rapid results are positive (HCV) or if antibody differentiation re- sults are negative or indeterminate (HIV) [4,5]. Once a patient has been confirmed positive, a diagnosis is made and positive individuals are linked to medical care [6]. The confirmatory test is critical to com- municating a positive diagnosis and initiating medical treatment and behavior modification.

Emergency departments (EDs) increasingly screen patients for HIV and HCV [7,8]. Through this process, EDs also encounter patients who present with prior positive results missing a confirmatory test, and pa- tients who were previously diagnosed who need linkage or re-linkage to medical care. While ED screening programs may inadvertently iden- tify these patients through repeat testing, evidence of a distinct popula- tion in need of these health services (i.e. confirmatory testing and linkage to care) beyond the scope of routine screening is extremely lim- ited. To date, there has been only one report of an ED initiative specifi- cally focusing on the prevention needs of previously diagnosed HIV patients [9]. This preliminary experience occurred before the era of widespread electronic health records (EHR) and depended on ED staff

https://doi.org/10.1016/j.ajem.2020.05.020 0735-6757/(C) 2020

1832 A.H. Ruffner et al. / American Journal of Emergency Medicine 38 (2020) 1831-1833

referral or sequential in-person approach to identify participants [9]. Be- yond this, ED prevention programs have not focused on patients already known to be infected in any intentional or systematic way.

Our objective was to preliminarily estimate the proportion of pa- tients with existing EHR information suggesting HIV or HCV infection in a single safety-net ED population by using an automated EHR alert system to target patients potentially in need of confirmatory HIV/HCV testing or linkage to care.

Methods

Study design

Cross-sectional study using existing EHR of an ED and its HIV/HCV prevention program. The study was approved by the study-site Institu- tional Review Board.

Study setting and population

The study was conducted at a Midwestern, urban, university- affiliated medical center. The ED of this medical center is a primary source of safety-net care for local socioeconomically disadvantaged populations. Of the approximately 75,000 adult patients seen annually, about 50% are Black, 0.5% are Hispanic, and 40% are uninsured.

ED HIV/HCV prevention program

This ED operates an HIV and HCV screening and linkage to care pro- gram that involves EHR prompted screening by nurses at triage as well as adjunct health promotion professionals (i.e. program staff) operating in parallel with usual ED staff to augment screening and assist in Care coordination. This routine HIV/HCV screening program screens all adult patients (aged b65 for HIV and adults born between 1945 and 1965 for HCV) who are not known to be previously positive and who do not have a prior test in the system. Our ED utilizes the HCV Antibody reflex to HCV Quantitative test order. During the study period, program staff were scheduled seven days a week, 16 hours a day (8 am-12 am), with two staff scheduled per shift.

In addition to our routine HIV/HCV screening program, we modified the EHR to identify patients who may be in need of HIV/HCV services, but would not require an initial test and therefore not be identified through routine screening. This additional modification was pro- grammed to search EHR records of all ED patients for any non- negative HIV or HCV test result, any diagnosis of HIV or HCV, or any mention of HIV or HCV on patient problem lists. The presence of any de- fined criteria initiated a text alert and an EHR inbox alert to program staff, who would then evaluate the need for HIV/HCV health services and assist ordering of confirmatory testing or linkage to medical care if needed. This process was added to the program’s existing routine screening workflow, without any increase in staffing. It is important to note that due to lack of preliminary data on this population, our pro- gram was unable to estimate volume of alerts for planning purposes. limited resources remained a challenge as alert volume exceeded staff capacity and funds were unavailable to increase program staffing at that time.

Program staff reviewed alerts when not actively engaged with a pa- tient. Alerts were not reviewed if the patient was no longer in the ED. Prevention staff recorded information about patient assessment in real-time in the program-specific electronic database. Data on patients for whom the alert fired was extracted from the EHR.

Study protocol

EHR and prevention program data were reviewed for a one-year pe- riod from August 2017 thru July 2018. Measures included whether or not an EHR alert was sent to prevention program staff, EHR information triggering the alert, and for alerts that were reviewed by program staff:

barriers to completing HIV and HCV needs assessments, results of com- pleted HIV and HCV needs assessments, receipt of HIV or HCV confirma- tory testing, receipt of linkage to care assistance, HIV and HCV test results, and whether or not the patient was successfully linked to HIV or HCV medical care.

Key outcome measures

The primary outcome was the proportion of ED patients for whom the previous positives EHR program generated an alert, indicating pa- tients with a previously positive HIV/HCV test result who may be in need of confirmatory testing or linkage services. Of the encounters where alerts were reviewed, secondary outcomes were the number and proportion who required confirmatory testing or linkage to care services.

Data analysis

Analysis was descriptive. Frequencies with proportions and 95% con- fidence intervals (CIs) were calculated for all outcomes. The unit of anal- ysis was encounter, since each encounter was considered an opportunity for intervention. Analyses were conducted using R version

3.4.3 for Windows [10].

Results

Details of study flow appear in Fig. 1. During the observation period there were 65,374 ED encounters, with 5238 (8.0%, 95% CI: 7.8%-8.2%) system alerts for EHR evidence of possible HIV or HCV. Of the 5238 pa- tients identified by the EHR alert, reasons for identification included: 3629 (69.3%, 95% CI: 68.0%-70.5%) for non-negative lab result, 3469

(66.2%, 95% CI: 64.9%-67.5%) for prior diagnosis, and 3265 (62.3%, 95% CI: 61.0%-63.6%) for problem lists. Multiple reasons were possible for each patient and 3448 (65.8%) had more than one reason initiating the EHR alert.

Prevention staff evaluated 3741 encounters with an alert. Of these, 798 (21.3%, 95% CI: 20.0%-22.7%) had an identified service need. One hundred forty-three (17.9%) needed an HIV confirmatory test, 104 (13.0%) were previously HIV positive requiring linkage or re-linkage to care, 145 (18.2%) needed an HCV confirmatory test, and 490 (61.4%) were previously HCV positive requiring linkage or re-linkage to care. Patients could require more than one service and we identified 84 (10.5%) encounters where both HIV and HCV services were needed.

Discussion

This study preliminarily estimates the volume of ED patients with EHR evidence of HIV or HCV infection, who require confirmatory testing and linkage to care. In this ED, almost 10% of patients had existing elec- tronic health record test results or diagnosis indicating HIV or HCV

Pages for HIV/HCV in EHR Prior to Current ED Visit

(n = 5,238, 8%)

Evaluated for Services Needed (n = 3,741, 71%)

Service(s) Needed (n = 798, 21%) HIV Only = 163

HCV Only = 551 HIV & HCV = 84

HCV Linkage to Care Needed**

(n = 490, 61%)

HCV Confirmatory Test Needed** (n = 145, 18%)

HIV Confirmatory Test Needed** (n = 143, 18%)

HIV Linkage to Care Needed**

(n = 104, 13%)

Total Pages (N = 5,238)*

Lab History HIV/HCV (n = 3,629) Medical History HIV/HCV (n = 3,469) Problem List HIV/HCV (n = 3,265)

ED Encounters 08/01/2017-07/31/2018 (N = 65,374)

* Prior HIV/HCV in EHR was determined by SNOWMED codes, the categories are not mutually exclusive (e.g. a patient could have HIV/HCV in lab history and medical history)

** These service needs are not mutually exclusive.

Fig. 1. Flow chart of process identifying patients who may require HIV or hepatitis C services (confirmatory testing or linkage/re-linkage to care) when presenting to the emergency department.

A.H. Ruffner et al. / American Journal of Emergency Medicine 38 (2020) 1831-1833 1833

infection. Of those patients assessed comprehensively by prevention program staff, approximately 36% needed testing to confirm their diag- nosis, approximately 13% were not in care for HIV, and approximately 60% were not in care for HCV. These results clearly demonstrate that ED screening for undetected cases is not the only important pathway for identifying ED patients in need of medical care for HIV and HCV.

Comparative evaluation of different methods for targeting the subset of ED patients with known disease has not been conducted. Our previ- ous experience has been that depending on human attention is ineffi- cient and resource intensive even when supplemented by dedicated program staff [9]. In this instance, we designed a basic EHR alert for this program to emphasize sensitivity by considering any prior test re- sult or diagnosis. We did not program the alert to specifically consider whether or not the patient’s diagnosis was confirmed or whether or not the patient was in treatment. As a result, program staff were re- quired to review cases for patients who had confirmed diagnosis and were receiving definitive care or, in the cases of HCV, had completed treatment [9]. A more sophisticated EHR algorithm might be able to tar- get populations with greatest prevention needs more efficiently. Similar EHR algorithms could be designed and used for other disease processes, such as cardiovascular disease or diabetes, in order to assist with identi- fying those who require further testing or linkage to care services.

This study was not designed to rigorously evaluate barriers to inter-

vention for this cohort of patients or the likelihood of successful inter- vention. Published barriers to re-linkage include: substance use, Mental illness, and homelessness; and our experience was similar [9]. Program resources were insufficient to address the volume and breadth of needs. However, we were able to successfully complete 11 confirma- tory HIV tests with 9 (82%) patients diagnosed, and 6 patients (67%) were linked to medical care. For HCV, we were able to successfully com- plete 74 confirmatory tests with 55 (74%) patients diagnosed, and 9 (16%) linked to medical care. We successfully re-linked 33/104 (32%) patients previously diagnosed with HIV and 37/490 (8%) patients previ- ously diagnosed with HCV.

The results of this preliminary program in linking infected individ- uals to care suggests that yields from engaging patients who have previ- ously tested positive are lower than linkage rates when screening for new cases. This suggests that expansion of provided services or working with other entities (e.g. public health departments, social service agen- cies, substance use treatment centers) would improve linkage. A com- prehensive evaluation of barriers for patients previously diagnosed is beyond the scope of this paper.

This report should be considered in the context of several limita- tions. Data on program assessment and intervention were collected in a structured manner but in a Clinical context without the usual controls and quality assurance measures typical of prospective research investi- gation. Results of this program are not generalizable to settings with dif- ferent disease epidemiology or without EHR systems. It would be theoretically possible to use an EHR alert, such as we describe, to alert usual ED providers in settings without dedicated prevention program staff, but we did not study that operational model. The alert used only health records available within our health system.

Conclusions

Individuals with prior positive HIV or HCV test results frequently present to EDs for medical care. Many of these patients are either un- aware of their diagnosis, require confirmatory testing, or if diagnosed, have not accessed definitive medical care or have fallen out of care. Identifying, assessing, and assisting this population to complete HIV and HCV diagnostic algorithms and begin treatment is essential to im- proving individual and public health. By utilizing electronic health

records systems to help EDs automate the identification of this popula- tion, we were able to estimate the number of previously diagnosed ED patients who require additional health services. In addition to screening for new HIV and HCV cases, ED Prevention efforts should prioritize pa- tients who have already been screened with positive or incomplete di- agnostic algorithms and develop systematic identification procedures that go beyond occasional capture through repeat screening.

Financial support

Supported by an investigator-initiated health services award from Gilead Sciences, Inc.’s FOCUS (Frontlines of Communities in the United States) program. FOCUS funding supports HIV, HCV, and HBV screening and linkage to the first medical appointment after diagnosis. Gilead played no role in study design or conduct of the research. The counsel- ing and testing program described in this report was also supported by the Ohio Department of Health via Hamilton County Public Health and by Ryan White funding provided by the Cincinnati Health Network, United States of America.

Presentations

This work was previously presented at the 2018 Society for Aca- demic Emergency Medicine (SAEM) annual meeting in Indianapolis, IN.

CRediT authorship contribution statement Andrew H. Ruffner:Conceptualization, Methodology, Writing - orig-

inal draft.Rachel M. Ancona:Methodology, Formal analysis.Catherine

Hamilton:Methodology, Software.Francisco J. Fernandez:Methodol- ogy, Software.Kiran A. Faryar:Writing - review & editing.Bennett H. Lane:Writing - review & editing.Michael S. Lyons:Conceptualization, Supervision.

Declaration of competing interest

MSL, RMA, KAF, and AHR received investigator-initiated health ser- vices award support paid to the institution from Gilead Sciences, Inc.

References

1. Hsieh YH, et al. Emergency departments at the crossroads of intersecting epidemics (HIV, HCV, Injection drug use and opioid overdose)-estimating HCV incidence in an urban emergency department population. J Viral Hepat. 2018;25(11):1397-400.
2. Menchine M, et al. Moving beyond screening: how emergency departments can help extinguish the HIV/AIDS epidemic. West J Emerg Med. 2016;17(2):135-8.
3. Branson BM, et al. Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. MMWR Recomm Rep. 2006;55(RR- 14):1-17 [quiz CE1-4].
4. Lin X, et al. Routine HIV screening in two health-care settings-New York City and New Orleans, 2011-2013. MMWR Morb Mortal Wkly Rep. 2014;63(25):537-41.
5. Haukoos JS, et al. Acute HIV infection and implications of fourth-generation HIV screening in emergency departments. Ann Emerg Med. 2014;64(5):547-51.
6. Bradley H, et al. Vital signs: HIV diagnosis, care, and treatment among persons living with HIV-United States. MMWR Morb Mortal Wkly Rep. 2011;63(47):1113-7 2014.
7. Haukoos JS, Lyons MS, Rothman RE. The evolving landscape of HIV screening in the emergency department. Ann Emerg Med. 2018;72(1):54-6.
8. Galbraith JW. Hepatitis C virus screening: an important public health opportunity for United States emergency departments. Ann Emerg Med. 2016;67(1):129-30.
9. Lyons MS, et al. A novel emergency department based prevention intervention pro- gram for people living with HIV: evaluation of early experiences. BMC Health Serv Res. 2007;7:164.
10. R Core Team. R: A language and environment for statistical computing. V., Austria: R Foundation for Statistical Computing; 2017 URL https://www.R-project.org/.