Bedside Diagnosis of the ‘Red Eye’:A Systematic Review

ABSTRACT
BACKGROUND: In patients with red eye, traditional teachings suggest that photophobia,
visual blurring, and eye pain indicate serious eye disease; in patients with presumed
conjunctivitis, the finding of purulent drainage traditionally indicates a bacterial cause. The
accuracy of these teachings is unknown.
METHODS: A MEDLINE search was performed to retrieve articles published between 1966
and April 2014 relevant to the bedside diagnosis of serious eye disease and bacterial
conjunctivitis. RESULTS: In patients with red eye, the most useful findings indicating serious
eye disease are anisocoria (with the smaller pupil in the red eye and difference between pupil
diameters >1 mm; likelihood ratio [LR], 6.5; 95% confidence interval [CI], 2.6-16.3) and
photophobia, elicited by direct illumination (LR, 8.3; 95% CI, 2.7-25.9), indirect illumination
(LR, 28.8; 95% CI, 1.8-459), or near synkinesis test (“finger-to-nose convergence test,” LR,
21.4; 95% CI, 12-38.2). In patients with presumed conjunctivitis, complete redness of the
conjunctival membrane obscuring tarsal vessels (LR, 4.6; 95% CI, 1.2-17.1), observed purulent
discharge (LR, 3.9; 95% CI, 1.7-9.1), and matting of both eyes in the morning (LR, 3.6; 95% CI,
1.9-6.5) increase the probability of a bacterial cause; failure to observe a red eye at 20 feet (LR,
0.2; 95% CI, 0-0.8) and absence of morning gluing of either eye (LR, 0.3; 95% CI, 0.1-0.8)
decrease the probability of a bacterial cause.
CONCLUSIONS: Several bedside findings accurately distinguish serious from benign eye
disease in patients with red eye and, in patients with presumed conjunctivitis, distinguish
bacterial from viral or allergic causes.

Evaluation of the “red eye” is a common problem, accounting for up to 1% of primary care
office visits1 and resulting in more than $300 million in annual costs for treating conjunctivitis
alone.2 When evaluating the red eye, clinicians address 2 distinct issues. First, the clinician must
determine whether the red eye is caused by serious eye disease (eg, uveitis, keratitis, corneal
abrasion, or scleritis), diagnoses requiring prompt referral to an eye specialist, or a more benign
disorder of the conjunctiva (eg, conjunctivitis, episcleritis, or subconjunctival hemorrhage).
Second, in patients with suspected conjunctivitis, clinicians want to accurately identify those
most likely to have a bacterial cause, because this group is most likely to benefit from
antimicrobial eye drops. According to traditional teachings, 3 findings indicate serious eye
disorders: significant eye pain, visual blurring, and photophobia. In patients with suspected
conjunctivitis, classic teachings suggest a bacterial cause is more likely in patients with disease
onset during winter months or if there is purulent discharge. Nonbacterial conjunctivitis, on the
other hand, is considered more likely if the patient presents during the summer or there is watery
discharge, conjunctival follicles, or preauricular adenopathy. The purpose of this review is to
identify the accuracy of these and other traditional bedside findings in distinguishing serious
from benign eye disease and, in patients with presumed conjunctivitis, distinguishing bacterial
causes from viral or allergic causes.

MATERIALS AND METHODS

We searched PubMed to identify all English-language studies that evaluated the diagnostic
accuracy of the patient interview and physical examination in adult patients presenting with a red
eye. The specific search strategy is shown in Appendix E1, (available online). We included all
studies meeting the following 3 criteria: (1) The study enrolled either consecutive unselected
adults with red eye, all of whom eventually underwent the diagnostic standard of slit-lamp
examination to distinguish serious disease (uveitis, keratitis, corneal abrasion, or scleritis) from
benign disorders (conjunctivitis, episcleritis, or subconjunctival hemorrhage) or consecutive
adults with red eye and presumed conjunctivitis, all of whom underwent bacterial culture of their
conjunctival secretions. Bacterial conjunctivitis was defined as recovery of a known pathogen
from conjunctival secretions (ie, Streptococcus pneumonia, Haemophilus influenzae, Moraxella
catarrhalis, or Staphylococcus aureus); (2) the clinical findings were defined clearly; and (3) the
study presented sufficient information to create 2 _ 2 tables and calculate sensitivity, specificity,
and likelihood ratios (LRs). Studies enrolling children (age <18 years) and studies of patients
with suspected trachoma were excluded. Both authors independently read all the articles related
to either of the 2 study questions and extracted data to calculate sensitivity, specificity, and
positive and negative LRs using standard definitions. Any differences were settled by discussion.
If any cell in the 2 _ 2 table contained the value of 0, 0.5 was added to all cells before calculating
LRs or pooled estimates. Pooled estimates were calculated using the DerSimonian and Laird
random-effects model.3 Specific definitions of selected findings are shown in Table 1.
RESULTS
Distinguishing Serious From Benign Eye Disease
Five studies enrolling 957 consecutive patients were included in this review. Four studies
recruited patients with red eye,4-7 and 1 study enrolled patients with “miscellaneous eye
complaints.”8 All patients underwent slit-lamp biomicroscopy: Serious disease was found in 4%
to 59% (mean, 27%), mostly anterior uveitis (iritis) and corneal disorders (herpes simplex
infection, corneal abrasion, and miscellaneous causes of keratitis).
Two findings—anisocoria and the presence of pain during maneuvers causing pupillary
constriction—accurately indicated serious disease. All 3 methods of inducing pupillary
constriction were accurate: direct light reaction (“direct photophobia,” LR, 8.3; 95% confidence
interval [CI], 2.7-25.9) (Table 2), consensual light reaction (“indirect photophobia,” LR, 28.8;
95% CI, 1.8-459), and near synkinesis pupillary constriction (“finger-to-nose convergence test,”
LR, 21.4; 95% CI, 12-38.2). The absence of pain during pupillary constriction decreased the
probability of serious disease, especially the negative finger-tonose convergence test (LR, 0.3,
95%CI, 0.1-0.6) and negative direct photophobia test (LR, 0.4; 95% CI, 0.3-0.5). Anisocoria,
defined as the smaller pupil in the affected red eye and a difference greater than 1 mm, increased
the probability of serious disease (LR, 6.5; 95% CI, 2.6-16.3). No studies were identified
investigating the accuracy of abnormal visual acuity or eye pain in patients with the red eye.
Distinguishing Bacterial from Nonbacterial
(Viral, Allergic) Conjunctivitis
Three studies enrolling 281 consecutive patients with presumed conjunctivitis were included in
our review.9-11 All patients underwent bacterial cultures, and 45% had positive bacterial
cultures. Most studies excluded patients with previous trauma, eye surgery, chemical injury,
visual blurring, contact lenses, conspicuous uveitis (ie, perilimbal ciliary flush), or obvious deep
orbital pathology. In these studies, the findings increasing the probability of positive bacterial
culture the most were complete redness of the conjunctiva obscuring tarsal vessels (LR, 4.6; 95%
CI, 1.2-17.1) (Table 3), observed purulent discharge (LR, 3.9; 95% CI, 1.7-9.1), matting of both
eyes in the morning (LR, 3.6; 95% CI, 1.9-6.5), and onset during winter or spring (vs summer;
LR, 1.9; 95% CI, 1.1-3.2). Findings increasing the probability of a nonbacterial cause include
absence of a red eye when observed at 20 feet (LR, 0.2; 95% CI, 0-0.8), absence of morning
gluing of either eye (LR, 0.3; 95% CI, 0.1-0.8), and presentation during summer (vs winter or
spring, LR, 0.4; 95% CI, 0.1-0.9). Of note, the patient’s report of “purulent drainage” was
unhelpful diagnostically (LR, 0.8; 95% CI, 0.3-2.1), in contrast to its value when discovered
during examination. Different qualities of eye discomfort—burning or itching—were unhelpful
diagnostically. The findings of preauricular adenopathy, conjunctival follicles, and conjunctival
papillae also were diagnostically inaccurate (although the point estimate of papillary
conjunctivitis, LR of 4.4, almost reached statistical significance). The Rietveld scoring scheme is
accurate: A score of þ4 or more increases probability of bacterial conjunctivitis (LR, 6.6; 95%
CI, 3-14.6), whereas a score of 0 or less decreases probability (LR, 0.4; 95% CI, 0.2-0.8).
DISCUSSION
Most ophthalmologic diagnosis depends on empiric observation by specialists using slit-lamp
biomicroscopy. However, our results demonstrate that some clinical variables easily observed by
primary providers without a slit lamp also accurately diagnose serious eye disease and bacterial
conjunctivitis. Simple observation of the pupil and the patient’s response to pupillary
constriction provide important clues suggesting serious eye disease. Inflammation of the iris and
spasm of the ciliary body are likely responsible for the miosis and painful pupillary constriction
characteristic of iritis; the cause of relative miosis in corneal disorders is unclear. Although our
results confirm the value of relative miosis, relative mydriasis (ie, anisocoria with the red eye
having a larger pupil) is not a benign finding but may indicate acute angle-closure glaucoma,
another diagnosis requiring prompt referral to a specialist (the studies in this review included
only 1 patient with this diagnosis). Of note, the absence of photophobia or anisocoria does not
exclude the possibility of serious disease: Our results show that 23% to 56% of patients with
serious pathology lack photophobia, and 81% lack anisocoria. In contrast to the proven value of
photophobia, we found no studies evaluating the other 2 traditional signs of serious eye
pathology (abnormal visual acuity and eye pain). These signs are likely insensitive, however; in
other studies of patients with proven uveitis, for example, 53% had a visual acuity of 20/60 or
better.12 Obviously, any patient with red eye and abnormal visual acuity or significant pain must
be referred promptly to a specialist. In patients with conjunctivitis, the findings of bilateral
matted eyes, redness obscuring tarsal vessels, and purulent discharge all increase the probability
of a bacterial cause. These findings likely represent a more severe inflammatory response, which
produces prominent vasodilation of conjunctival vessels and neutrophilic exudate. Other studies
using Giemsa staining of conjunctival secretions have demonstrated a correlation between
neutrophilic predominance and bacterial cause.9 The accuracy of the Rietveld score combines
the value of eye matting from inflammatory infiltrate with the absence of symptoms
characteristic of allergic causes (itching and prior conjunctivitis). Although our study failed to
prove the accuracy of conjunctival follicles or papillae as individual findings, another study of
700 patients with red eye showed that the combination of follicles, preauricular lymph nodes,
and scant water discharge predicted a viral cause, whereas the combination of mucopurulent
discharge without follicles or adenopathy correlated with positive bacterial cultures.13

STUDY LIMITATIONS
Limitations of our review include the small size and number of studies forming the basis of our
conclusions. Also, in the pupillary constriction studies, it is possible that there was lack of
blinding between “test” and “diagnosis,” because the person performing the penlight test was
likely also the clinician later performing slit-lamp biomicroscopy. Finally, it is possible that our
second research question— distinguishing bacterial from nonbacterial conjunctivitis—is
unimportant because most bacterial conjunctivitis spontaneously resolves without complications
within 7 days even without antibiotics.14 Nonetheless, we believe this distinction is important,
because antimicrobial treatment of bacterial conjunctivitis shortens duration of symptoms by 0.5
to 1.5 days on average, and the number-needed-to-treat with antibiotics to get 1 extra patient into
remission in the first 2 to 5 days is only 6.14
CONCLUSIONS
In patients with the red eye, simple examination of the pupils and response to pupillary
constriction accurately identify patients who require immediate referral to an ophthalmologist.
All patients with visual blurring and significant eye pain also require immediate referral,
although these findings have not been evaluated systematically. In patients with conjunctivitis, a
bacterial cause is more likely if there is bilateral matting of the eyes, conjunctival redness
obscuring tarsal vessels, and purulent drainage. These patients are most likely to benefit from
topical antimicrobials.