PURPOSE: To devise and implement a practice algo rithm that would enable rapid

detection and appropriate furlough of hospital employees with adenoviral conjunc tivitis in
order to prevent healthcareassociated epidemic keratoconjunctivitis.
DESIGN: Evaluation of an ongoing quality assurance/
improvement initiative.

METHODS: Employees of Johns Hopkins Hospital with signs and symptoms of

adenoviral conjunctivitis under went evaluation by nurse practitioners in Occupational
Health and rapid diagnostic testing by realtime polymer ase chain reaction (PCR).
Sequencing was used to deter mine serotype when adenovirus was detected. Signs,
symptoms, diagnosis, and disposition of employees with eye complaints as well as PCR
and serotype results
were recorded.

RESULTS: Over a 36month period approximately18% of initial employee visits

were due to unique, eye related complaints. Viral conjunctivitis was suspected in 542
of 858 employees with eye complaints (62%); adenovirus was detected by PCR in 44
of 542 suspected viral conjunctivitis cases (8%) or 44 of 858 employees with any eye
concern (5%). Fourteen of the 44 em ployees had adenoviral serotypes and clinical
presenta tion consistent with epidemic keratoconjunctivitis (type 37 [n [ 6], 8 [n
[ 4], 4 [n [ 3], 19 [n [ 1]). Other serotypes found in individuals with less severe
conjunctivitis were 3 (n [ 5), 4 (n [ 5), 56 (n [ 4),
1 (n [ 2), 42 (n [ 1), and 7 (n [ 1). No healthcare associated adenoviral
conjunctivitis outbreaks occurred after algorithm implementation, and fewer employees
required furlough than had clinical diagnosis alone been used.

CONCLUSIONS: The algorithm is an effective infection prevention tool that

minimizes productivity loss compared to clinical diagnosis and allows for determina tion
of prevalence and serotype characterization of adeno viral conjunctivitis in hospital
employees. (Am J Ophthalmol 2016;163:38–44. 2016 by Elsevier Inc. All rights
reserved.)

ADENOVIRAL EYE INFECTION IS A PUBLIC HEALTH concern, with

manifestations ranging from self limited conjunctivitis to prolonged ocular
morbidity in the case of epidemic keratoconjunctivitis (EKC). Some adenoviral
infections, however, may not be easily distinguishable from EKC at presentation; the 3
other clinical scenarios are nonspecific follicular conjunctivitis, pharyngeal conjunctival
fever, and acute hemor rhagic conjunctivitis. EKC is most commonly caused by serotypes
8, 19, and 371; less typical serotypes associated. obvious and induces more morbidity than
the other adeno
viral eye conditions and is associated with a longer disease course and significant
injection, chemosis, and keratitis that can impair vision

Outbreaks in hospitals and eye clinics result in substantial morbidity and lost
productivity.6–12 In response to healthcareassociated EKC outbreaks,6–12 one of which
resulted in the temporary closure of a large eye institute,12 a ‘‘redeye room,’’ where persons
with potentially infectious conjunctivitis could be triaged and isolated, was established in
1990 in the Wilmer Eye Institute Emergency Room (ER), Department of Ophthalmology, at
the Johns Hopkins Hos pital (JHH).13 This arrangement was part of an infection
prevention program consisting of ophthalmic instrument decontamination, hand hygiene,
use of singledose eye drops, and employee furloughs.14 When institutional changes led
to the permanent closure of the Wilmer ER and the redeye room at the end of January
2008, a multidis ciplinary team composed of a cornea specialist and stake holders in
infection prevention, employee health, and the clinical virology/molecular infectious
disease clinical diag nostic laboratory was convened in order to devise and imple ment a new
practice algorithm for preventing healthcare associated transmission of adenoviral eye
disease. Twenty month pilot data have been published.14 Here we report
36month data on adenovirus detection rates and the results of adding serotype determination
to this infection preven tion practice algorithm as applied to healthcare workers with red
eye

METHODS

THE INSTITUTIONAL REVIEW BOARDS OF JOHNS HOPKINS Medicine ruled that

publishing of an evaluation of this quality improvement initiative did not require
approval as it ‘‘does not involve human subjects research under the Department of Health
and Human Services or Food and Drug Administration regulations.’’
The new redeye employee practice algorithm, which is
in effect at all times, is shown in Figure 1. It consists of initial evaluation by nurse
practitioners in the Occupa tional Health clinic and rapid diagnostic testing by real time
polymerase chain reaction (PCR) in individuals with signs and symptoms consistent
with adenoviral conjunctivitis. The Occupational Health clinic sees em ployees with
health concerns that arise during work hours and may impinge on their ability to work. In
order to limit the possibility of transmission within the clinic, employees seeking redeye
evaluation are required to call ahead to make an appointment.
A corneal specialist taught nurse practitioners in the Occupational Health clinic to
recognize signs and symp toms of probable viral conjunctivitis (vs conditions like
suspected corneal abrasion or subconjunctival hemor rhage) and how to collect swab
specimens of the inferior conjunctival fornix.14 The nurse practitioners were instructed to
swab only when viral conjunctivitis was suspected. Swabs (polyester, Dacron, or rayon
with plastic or aluminum shafts) were placed in M4 medium and sent at room temperature
for PCR testing performed daily at the JHH Clinical Virology/Molecular Infectious
Disease Diagnostic Laboratory. Screening criteria included dura tion of symptoms,
presence of viral prodrome, discharge or tearing, and unilateral onset. Employees with
suspected viral conjunctivitis were evaluated, swabbed, and discharged home within 30
minutes of intake. Nurse prac titioners notified employees of PCR results and furlough
status by 8 PM if specimens were received in the labora tory by 3 PM. If adenovirus
was detected by PCR, a
2week furlough was invoked. Direct sequencing of spec imens found to contain
adenovirus DNA was performed retrospectively for epidemiologic purposes—to determine
the proportion of employees infected with EKC associated adenoviral serotypes
(most typically types 8,
19, and 37).15–21
Toward the end of the furlough, the nurse case manager in the Occupational Health clinic
contacted employees with adenoviral conjunctivitis by phone to assess symptoms and to
schedule a followup examination at Occupational Health that was required before return
to duty. Furlough was continued if symptoms and signs of conjunctivitis persisted, given
probable infectious potential. Employees were sent to the Department of Ophthalmology
for consul tation for conditions the nurse practitioners deemed an emergency; symptoms
including but not limited to persistent redness, tearing, pain, and blurred vision; or
conditions in which the nurse practitioners were unsure of diagnosis.

ADENOVIRAL DETECTION AND SEROTYPE DETERMINA TION: As

described in the pilot study,14 PCR for adeno
virus in conjunctival specimens of employees was developed and validated in the Johns
Hopkins Hospital molecular microbiology diagnostic laboratory. ‘‘Total nucleic acid was
isolated from conjunctival specimens (processed volume, 400 microliters) using
automated instrumentation (BioRobot M48, using Virus Mini Proto col, version 1.1
software and MagAttract Virus Mini M48 reagents [Qiagen, Germantown, MD];’’14
elution volume,
125 mL). Adenovirus DNA was detected by realtime PCR (primers: 59 GCC ACG
GTG GGG TTT CTA AAC TT39 , 59 GCC CCA GTG GTC TTA CAT GCA
CAT C39 ; fluorogenic probe FAM 59 TGC ACC AGA CCC GGG CTC AGG
TAC TCC GA39
TAMRA)22 using 7500 Real Time PCR instruments
(Life Technologies, Carlsbad, California, USA). The analytical sensitivity (95% detection
rate or limit of detec tion) is 300 copies/mL, and the assay detects at least 16 adenovirus
serotypes, including strains from 6 of 7 adeno virus serogroups (AF). When adenoviruses
were detected, serotype was determined by nested PCR of the hexon gene hypervariable
regions 1–623 using previously extracted to tal nucleic acid (outer primers: AdhexF1
(1913519160)
59 TIC TTT GAC ATI CGI GGI GTI CTI GA39 /
AdhexR1 (2000920030) 59 CTG TCI ACI GCC TGR TTC CAC A39 ; inner primers:
AdhexF2 (1916519187)

59 GGY CCY AGY TTY AAR CCC TAY TC39 / AdhexR2 (1996019985) 59 GGT
TCT GTC ICC CAG AGA RTC IAG CA39 ) followed by bidirectional Sanger
sequencing using AdhexF2/AdhexR2 as sequencing primers in reactions containing
BigDye Terminator v.3.1
(Life Technologies). Sequencing was performed on either a 3100 or a 3500 Genetic
Analyzer (Life Technologies). These reagents correctly identified 50 prototype strains; se
rotypes 15 and 29 have identical sequences in the queried region, cannot be
distinguished,23 and therefore are re ported as 15/29. The ability to identify serotypes
associated with ocular disease (8, 19, 37) and other serotypes that commonly cause ocular
disease (3, 4, 7, 11) was confirmed in house using acquired strains (ATCC, Chantilly, Vir
ginia, USA).14

Result

FROM NOVEMBER 22, 2011 TO OCTOBER 31, 2014, 858 OF 4883 initial employee
Occupational Health visits (18%) were due to unique, eyerelated complaints. The number
of em ployees seen with eye concerns ranged from 6 to 36 per month (Figure 2). Most
employees complained of red eye Of the 858 employees with eye concerns, 542 (62%) under
 went conjunctival swabbing and adenovirus PCR testing (Figure 1). Fortyfour employees
(8% of suspected adeno viral cases, or 5% of all employees with eye concerns) were
positive for adenovirus by realtime PCR. Adenovirus serotype could be determined for 32
employees. Overall, 14 employees had serotypes commonly associated with EKC and/or
signs and symptoms consistent with EKC. EKC associated serotypes that were detected
included type 37 (n ¼ 6), 8 (n ¼ 4), and 19 (n ¼ 1). Three of 8 employees infected with
serotype 4 had signs and symptoms consistent with EKC. The 5 other employees infected
with serotype 4 had less severe conjunctivitis. Other serotypes found were 3 (n ¼ 5), 56 (n ¼
4), 1 (n ¼ 2), 7 (n ¼ 1), and 42 (n ¼ 1). Serotype could not be determined by sequencing for
12 em ployees. Although adenovirus DNA was detected by real time PCR from these 12
samples, no PCR products were obtained with the less sensitive nested PCR reaction used
for serotype determination. A semiquantitative analysis of cycle thresholds at which
adenovirus DNA amplification was detected suggested that samples that could not be sero
typed contained approximately 1000fold less DNA than samples for which serotype was
obtained. The median PCR cycle at which adenovirus DNA was detected by realtime
PCR was 41 for samples that could not be sero typed, compared to 33 for samples for
which a serotype was obtained. None of the 12 patients whose samples could not be
serotyped had EKC.
More than half (469/858; 55%) of employees had condi tions that were determined to be
noninfectious after consultation by a Wilmer or private ophthalmologist; allergic
conjunctivitis was most common. The other diag noses included corneal or conjunctival
foreign body, pinguecula, episcleritis, scleritis, corneal abrasion, contact lens overuse, iritis,
dacryocystitis, and preseptal cellulitis. The remainder (389/858; 45%) had adenoviral
infection confirmed by PCR (44 listed above) or were diagnosed with ‘‘viral’’ or ‘‘bacterial’’
conjunctivitis by an ophthalmol ogist (either at Wilmer or in the community), a primary
care provider, or a doctor at an urgent care center. Conjunctival cultures were not reported
for any of these employees, implying that ‘‘viral’’ or ‘‘bacterial’’ was a clin ical diagnosis.
The majority of employees seen by primary care providers and urgent care doctors were
started on topical antibiotic therapy.
To compare the prevalence of laboratorydiagnosed adenoviral conjunctivitis among
healthcare employees with eye concerns with the prevalence of this condition among
general ophthalmology patients as diagnosed by clinical examination and/or culture, we
examined billing data from the Wilmer General Eye Service. Visits with billing diagnoses
consistent with adenoviral conjunctivitis (including International Classification of
Diseases, 9th revision [ICD9] codes 372.00, 372.03, 372.71, 379.93,
077.1, 077.3, 077.4, and 077.8) were analyzed. In the 2 year period prior to redeye room
closure (January 2006
through January 2008), there were 9609 patient visits with unique diagnoses (ie, followup
visits for the same diagnosis for the patient were excluded). An ICD9 code consistent with
adenoviral conjunctivitis was attached to 986 visits; the vast majority of diagnoses were
made clinically. There fore, the prevalence of clinically diagnosed adenoviral
conjunctivitis in the General Eye Service was 986 of
9609, or 10%
Diskusi :
THE NEW TRIAGE ALGORITHM FOR HOSPITAL EMPLOYEES with red eye has
achieved its goal to isolate and furlough employees with adenoviral conjunctivitis in a
rapid manner, thus preventing spread of disease to patients and other employees. The
causes of red eye in hospital em ployees are now recorded systematically at 1 location
(rather than simple description of red eye and varying levels of evaluation at multiple
clinics), and the prevalence of adenoviral infection, as well as infecting serotypes, can be
monitored for epidemiologic purposes. To our knowl edge, there is no other similar
institutionwide triage sys tem for adenoviral conjunctivitis, nor has there been any report
of adenovirus serotype or prevalence in hospital em ployees, which, by PCR diagnosis, is
lower than clinical diagnosis would indicate.
This algorithm has several potential benefits. First, PCR offers stateofthe art sensitivity and
specificity compared to other methods, such as culture and antigen detection. Second, the
initial evaluation of healthcare workers at a single site by a select group of trained providers
maximizes the likelihood of adherence to EKC prevention policies and decreases the
possibility of healthcareassociated spread. Prior to this algorithm, numerous different clini
cians evaluated employees, adenovirus cultures were infre quently obtained, adenoviral
conjunctivitis was often clinically diagnosed, and identification of EKCrelated se rotypes
was not performed. Moreover, many ophthalmolo gists are reluctant to examine cases of
probable viral conjunctivitis, necessitating this algorithm. Third, given that clinical
diagnosis of this disease is imperfect and that initial screening by nurse practitioners at
Occupational Health is expected to yield falsepositive cases, adjunctive use of PCR testing in
employees with suspected adenoviral conjunctivitis facilitates judicious use of furlough for
infec tion prevention. In the findings above, there was almost a 10fold difference between
diagnostically confirmed and clinically suspicious cases (44 employees with serotype and
disease course consistent with adenoviral conjuncti vitis vs 542 with clinically suspicious
findings); furlough of over 500 employees over a 3year period would have been
impractical.
Determination of adenovirus prevalence and delineation of serotypes in hospital
employees with red eye became important objectives once the triage algorithm was
implemented. Prevalence of adenoviral conjuncti vitis among hospital personnel
with eye concerns (5%) was less than half that seen in General Eye Service patients
(10%) in whom the infection was clinically diagnosed. Possible explanations for this
difference include different prevalence of adenoviral conjunctivitis between the general
public and hospital employees or the comparatively lower specificity of clinical diagnosis
compared with PCR testing. Most likely some of the clinically diagnosed cases in the
General Eye Service were not truly adenoviral, which is supported by findings of the
current study.
Retrospective molecular serotyping showed that approx imately 30% of the employees
with adenoviral conjuncti vitis were infected with an EKCassociated strain, which raised
the question of whether the algorithm should include prospective serotype
determination to tailor furlough duration. For example, employees with EKC associated
serotypes could be assigned a 2week work furlough while those with other serotypes
could be furloughed for a shorter duration until resolution of clinical symptoms; clinical
clearance by Occupational Health nurse practitioners would be required to return to
work in either situation. The 2week furlough commonly recom mended for patients with
presumed or definite adenoviral conjunctivitis may have arisen in part from a report of a
large outbreak of EKC in an ophthalmology department in which adenovirus was isolated
 from conjunctival sur faces up to 2 weeks after the onset of clinical illness.12 In spring
2014, however, based on the low prevalence of
EKC in employees suspected of having viral conjunctivitis and a suspicion that many
employees with adenoviral conjunctivitis recovered well in advance of the 2week mark,
the ophthalmologist involved in developing this pol icy suggested that the nurse case
manager call furloughed employees at the 1week mark. Most employees with nonEKC
serotypes reported not having any tearing, redness, or discomfort at that time point.
Compilation of such reports after 2 months suggested that a 2week furlough might
be unnecessary for the majority of em ployees with adenoviral conjunctivitis. Therefore,
in April
2014 the redeye policy was further modified. In this manner, serotyping is done
prospectively (with results in
2–5 days) and furlough length is tailored depending on whether serotype is typical for
EKC (types 4, 8, 19, 37) and whether signs and symptoms are consistent with EKC. The
nurse practitioners call employees 5–6 days after
presentation to Occupational Health to inquire of their symptoms. If the employee is
still symptomatic, the furlough continues. If the employee is relatively asymptom atic, the
nurse practitioner asks the employee to return to Occupational Health to be cleared for
work; the vast ma jority of employees with nonEKC serotypes are able to work after a
7day furlough. As before, in all cases, em ployees require clearance by Occupational
Health in order to return to work.
Detection of adenovirus by PCR does not imply the eye is infectious, and adenoviral
conjunctivitis can span from mild disease to EKC. Though there are other infectious
conjunctivitides (eg, enteroviral), which typically resolve within 5–7 days,24 adenoviral
conjunctivitis is more com mon and potential complications (such as chronic visual
symptoms from keratitis) are more severe. For these rea sons, the focus of this PCRbased
algorithm was adenovirus detection.
Based on a preliminary analysis, the redeye algorithm has been cost effective for our
institution. This analysis was based on Johns Hopkins Hospital absorbing the cost of PCR
performed on employees with clinical suspicion of adenoviral conjunctivitis (542) and the
cost to the hos pital of furloughing only PCRpositive employees (44) for
2 weeks vs the cost of furloughing all 542 employees, as might have been done prior to
this algorithm. From sero type prevalence and record of employees’ clinical course, it
appears that the 2week furlough is excessive for most employees with adenoviral
conjunctivitis. If trends continue, based on the distribution of serotypes and furlough
lengths (onethird of employees with adenovirus conjunctivitis having EKCassociated
serotypes requiring
2 weeks vs twothirds with nonEKCrelated serotypes requiring 1 week), the number of
furlough days potentially could be reduced by another third. A more flexible policy,
invoking shorter or longer furlough based on community molecular epidemiologic data of
circulating adenovirus se rotypes, as begun in April 2014, may be more appropriate and
deserves further investigation.
While PCR for adenovirus is available, there is currently no commercially available PCR test
for the detection of adenovirus in conjunctival specimens. Therefore, clinical laboratories
must develop their own tests for this use. Laboratorydeveloped tests are regulated by a
federal law entitled Clinical Laboratory Improvement Act, which stip ulates that laboratories
can develop and offer tests clini cally if they ascertain the analytical performance
characteristics of the test. For molecular infectious disease tests, analytical sensitivity is
defined as limit of detection, which is determined in experiments with a dilution series of
concentrations, and is defined as the concentration of the panel member for which 95% of
replicates are detected. Manufacturers seeking Food and Drug Administration (FDA)
approval to market a test are held to the standard of a clinical trial to define the clinical
indication for the test and the clinical performance characteristics relative

to disease, defined by clinical symptoms and the results of a goldstandard diagnostic test
(clinical sensitivity and clin ical specificity). Such determinations are not required of
laboratorydeveloped tests. Demonstrating clinical need for initiatives like the one
described herein may motivate other centers to adopt some or all parts of the algorithm to
diagnose adenoviral conjunctivitis accurately in order to limit transmission of disease.
The perceived benefit from such testing also may provide commercial impetus for
development of an FDAcleared/approved PCRbased test for adenoviral conjunctivitis that
can be used in hospi tals, clinics, and other settings.
The inability to sequester a potentially large number of individuals with possible infectious
conjunctivitis in a designated room in the Department of Ophthalmology led us to
develop an algorithm for rapid diagnosis of adeno viral conjunctivitis in hospital employees,
who have the potential to infect patients and other employees alike. Implementation
required a multidisciplinary effort across several hospital departments and divisions. This
compre hensive institutional policy has allowed for accurate detection and tracking of the
prevalence of adenoviral conjunctivitis, and judicious application of furlough based on
objective criteria. The relative simplicity and functional benefits of this effort suggest that it
merits consideration as a practice model for hospitals and clinics seeking to prevent
healthcareassociated transmission of adenoviral conjunctivitis. The addition of serotyping
furthers our understand ing of the prevalence of adenoviruses that cause EKC. It also
provides preliminary data to support the development of tests (such as a carefully designed
multiplex PCR test) that are simpler to perform but as accurate as direct Sanger sequencing
to detect all adenoviruses and to distinguish those that cause