Conjunctivitis

Prior Probability of Viral vs Bacterial Conjunctivitis

Once infectious conjunctivitis is clinically suspected, the clinician must assess the likelihood of viral vs bacterial conjunctivitis. A key factor is the age of the patient. In studies that included children (mean age, 4.7 years; range, 1 month to 18 years), bacterial conjunctivitis was present in 71% while viral conjunctivitis was present in only 16%; 13% were attributed to other causes that were culture negative, chlamydia, or ultimately judged to be allergic. Among adults (mean age, 25.7 years), the prevalence of viral conjunctivitis was 78% while bacterial conjunctivitis was present in 16%; 6% were attributable to other causes (including herpes simplex and chlamydia).

Population Among Whom Viral or Bacterial Conjunctivitis Should Be Considered

Conjunctivitis is common and patients usually present for care with ocular redness and discharge. Other common causes of red eye that usually can be distinguished clinically from infectious conjunctivitis are allergic conjunctivitis (though most patients do not seek medical care)¹, subconjunctival hemorrhage, and dry eye disease. Serious but less frequent causes of ocular redness include foreign body, corneal abrasion, corneal infection, scleritis, iritis, and glaucoma. For practical purposes, the presence of a red eye with discharge indicates infectious conjunctivitis in the majority of cases, and it requires clinicians to assess the likelihood of a viral vs a bacterial etiology.

Assessing the Likelihood That the Patient Has Bacterial Conjunctivitis

Most studies of conjunctivitis are reports of large case series where only the sensitivity can be determined. Among the most frequently studied findings, the highest sensitivities for viral conjunctivitis were conjunctival follicles (77%; 95% CI, 36%-96%), watery discharge (77%; 95% CI, 33%-96%), eyelid swelling (59%; 95% CI, 19%-90%), bilateral eye symptoms (44%; 95% CI, 33%-56%), and pharyngitis (43%; 95% CI, 28%-59%). Among the most frequently studied findings for bacterial pharyngitis, the most sensitive findings were the patient reported having their eyelids stuck together on awakening (86%; 95% CI, 67%-95%), mucopurulent ocular discharge (67%; 95% CI, 48%-82%), and bilateral eye symptoms (59%; 95% CI, 44%-72%).

Studies that included patients with either viral or bacterial conjunctivitis can be evaluated to determine which findings distinguished a viral etiology for conjunctivitis from a bacterial etiology. The most useful findings for viral conjunctivitis included the presence of pharyngitis (LR range, 5.4-9.9), preauricular lymphadenopathy (LR range, 2.5-5.6), and contact with another person with red eye (LR, 2.5; 95% CI, 1.6-3.7). For bacterial conjunctivitis, the presence of mucopurulent ocular discharge (LR, 2.1; 95% CI, 1.7-2.6) and otitis media (LR, 2.5; 95% CI, 1.5-4.4) were associated with the presence of bacterial conjunctivitis. The presence of a watery discharge (LR, 0.03; 95% CI, 0.004-0.26) or pharyngitis (LR range, 0.10-0.19) was associated with a lower likelihood of bacterial conjunctivitis. (See Table 111-1.)

Reference Standard Tests

The reference standard tests for viral or bacterial conjunctivitis are polymerase chain reaction studies or microbiological cultures.

Reference

Clinical Scenario

A woman in her 30s presented to her primary care physician with a 3-day history of ocular redness and watery discharge. The symptoms started in the left eye and 2 days later spread to the right eye. A coworker recently had an episode of “pink eye” and was prescribed antibiotic eye drops. On examination, visual acuity was 20/20 in each eye and diffuse redness involving the conjunctiva was present with mild eyelid swelling (Figure 111-1). Is this presentation more consistent with viral or bacterial conjunctivitis? Should antibiotic eye drops be prescribed?

Background

Conjunctivitis is a common condition seen by general practitioners, and patients usually present for care with ocular redness and discharge. Although the incidence and prevalence of conjunctivitis are difficult to estimate, viral conjunctivitis may account for up to 20 million cases per year in the US.¹ Many patients with conjunctivitis initially present to primary care physicians rather than to ophthalmologists.^2,3 Conjunctivitis accounts for approximately half of all eye-related visits to general practitioners⁴ and an estimated one-third of eye-related emergency department visits.⁵

There is a broad differential diagnosis for the patient presenting with red eye; a detailed approach and differential diagnosis of red eye has been described elsewhere.^6,7 Other common causes of red eye include allergic conjunctivitis, subconjunctival hemorrhage, and dry eye disease. More serious causes occur less frequently and include foreign body, corneal abrasion, corneal infection, scleritis, iritis, and glaucoma. Most generalist physicians can differentiate infectious conjunctivitis from other forms of red eye. For practical purposes, the presence of red eye with discharge indicates infectious conjunctivitis in the majority of cases, and requires clinicians to assess the likelihood of a viral vs a bacterial etiology. However, some clinicians may have difficulty differentiating between viral and bacterial conjunctivitis.⁸ For example, in a multicenter study,⁹ only 57 of 157 patients (32%) with clinically suspected bacterial conjunctivitis had positive bacterial cultures. As a result, many generalist physicians may prescribe topical antibiotics for patients with infectious conjunctivitis even though a large proportion of cases are caused by viruses.^10-13 Excessive prescribing of antibiotics has numerous potential adverse effects, including the promotion of bacterial resistance, the possibility of toxic and allergic reactions (which may confound the clinical picture and perpetuate inflammation of the ocular surface), and increased cost.

A previous systematic review from 2003 reported that there was no evidence regarding whether symptoms or signs detectable by the generalist physician could help differentiate viral vs bacterial conjunctivitis.¹⁴ This Rational Clinical Examination article evaluates the relative prevalence of viral vs bacterial conjunctivitis in adults and children, and determines which clinical findings may help differentiate patients with viral conjunctivitis from those with bacterial conjunctivitis.

Methods

Literature Search and Selection

OVID MEDLINE was searched for English-language literature published from January 1946 to March 2022 to identify articles that included series of patients with viral conjunctivitis, bacterial conjunctivitis, or both. For the search, the Medical Subject Heading conjunctivitis (focus) was used, excluding case reports, comments, and letters. All retrieved titles were screened by 2 of the authors (D.J. and D.L.), and abstracts of potentially relevant articles were further reviewed. Any article that reported on the relative frequency of viral vs bacterial conjunctivitis, or on the prevalence of various symptoms or signs in either condition was included. Articles on neonatal conjunctivitis were not included. In addition, citations of retrieved articles as well as relevant review articles were reviewed to identify further articles. Consistent with the Rational Clinical Examination series, all articles were assigned a quality score based on a modified grading system (Supplemental Box). Studies with a methodological quality level ranging from 1 to 4 used a microbiological reference standard for diagnosis, whereas studies with a quality level of 5 (the lowest quality) used a clinical reference standard for diagnosis.

Data Analysis and Statistical Methods

The primary outcomes were analyzed using meta-analysis and included the relative prevalence of viral vs bacterial conjunctivitis and the diagnostic accuracy for various symptoms and signs associated with the 2 conditions.

Because the comparative frequency of viral vs bacterial conjunctivitis is known to differ in adults vs children, only studies that reported prevalence in which data could be stratified by age (adults [aged ≥ 19 years] vs children [aged 1 month-18 years]) were included in the final analysis for estimating relative prevalence of viral vs bacterial conjunctivitis. Prevalence was summarized with random-effects measures using Comprehensive Meta-Analysis version 3 (Biostat).

Data for diagnostic accuracy were reported as sensitivity, specificity, and likelihood ratios (LRs) for various history and examination findings. Studies included in the calculation of summary measures of the sensitivity, specificity, and LRs were required to include an undifferentiated population of patients with infectious conjunctivitis and be of Rational Clinical Examination level 1 or 2 quality. When calculating the LRs and 95% CIs from the individual studies, 0.5 was added to the 2 × 2 table cells with a value of 0.¹⁵ The results were summarized with point estimates (when evaluated in 1 study), range (when evaluated in 2 studies), and univariable random-effects measures (when evaluated in ≥ 3 studies). The univariable random-effects measures were calculated using Comprehensive Meta-Analysis. Univariable measures were used because only 1 finding was evaluated in as many as 4 studies.¹⁶ Quantitative assessment of heterogeneity was not possible because each finding was evaluated in so few studies and the use of the I² statistic is not recommended in diagnostic test meta-analyses.¹⁷

As a secondary outcome, the prevalence (frequency) of various clinical findings was evaluated with a random-effects summary measure in patients with viral and bacterial conjunctivitis using studies with a methodological quality level of 1 to 4 (level 3 or 4 studies were primarily case series). The frequency of a finding in a case series of patients represents the sensitivity, but the specificity cannot be estimated. Studies with a methodological quality level of 5 were not included in the final data analysis.

Results

The literature search identified 1891 articles, of which 80 included series of patients with viral or bacterial conjunctivitis (Supplemental Figure). A review of the references and citations did not yield any additional articles. Thirty-two studies^9,18-49 were included in the analysis; of these, 11 had a methodological quality level of 1 or 2 (details of the included studies appear in Supplemental Table 1).^{21,25,26,28,29,31,33,40,41,45,47} These studies included 24 case series, 5 randomized clinical trials, 1 nonrandomized comparative trial, and 2 cohort studies.

Prevalence of Viral vs Bacterial Conjunctivitis

A total of 5 studies (n = 1088 patients) used a microbiological reference standard, such as polymerase chain reaction or microbiological cultures, suitable for estimating the relative frequency of viral vs bacterial conjunctivitis.^{26,36,39,45,47} All 5 studies reported data for children and only 1 of the studies included data from an adult population.⁴⁷ Among children in these 5 studies (n = 881 patients; mean age, 4.7 years [age range, 1 month-18 years]), bacterial conjunctivitis was more common than viral conjunctivitis (71% [95% CI, 38%-91%] vs 16% [95% CI, 4%-46%], P = .01). Approximately 13% of conjunctivitis cases in children were either culture negative, culture positive for Chlamydia trachomatis, or presumed to be allergic conjunctivitis. Only 1 study from Pakistan from 1992 provided data for only adults (mean age, 25.7 years) on the relative prevalence of microbiologically confirmed viral vs bacterial conjunctivitis.⁴⁷ In this study, 33/207 (16%) cases of conjunctivitis were due to bacterial infection and 161/207 (78%) were from viral infection (P < .001); 13/207 (6%) were due to other causes, including herpes simplex and chlamydial conjunctivitis.

Frequency of Clinical Findings

Most identified studies were case series (ie, level 3 or 4 studies) that were limited to either patients with viral or bacterial conjunctivitis and the frequency of a finding represented the sensitivity. The data were combined with studies that included both viral and bacterial conjunctivitis and the results were summarized to describe the frequency of findings that a clinician may evaluate (Table 111-2).

For the clinical findings evaluated in at least 3 studies among patients with viral conjunctivitis, the most frequent were conjunctival follicles (77% [95% CI, 36%-96%]; 1350 patients from 8 studies), watery discharge (77% [95% CI, 33%-96%]; 287 patients from 3 studies), eyelid swelling (59% [95% CI, 19%-90%]; 441 patients from 4 studies), bilateral eye symptoms (44% [95% CI, 33%-56%]; 943 patients from 12 studies), pharyngitis (43% [95% CI, 28%-59%]; 241 patients from 5 studies), and conjunctival hemorrhages (35% [95% CI, 16%-60%]; 1178 patients from 5 studies).

For the clinical findings evaluated in at least 3 studies among patients with bacterial conjunctivitis, the most frequent were the patient reported having his or her eyelids stuck together on awakening (86% [95% CI, 67%-95%]; 465 patients from 3 studies), mucopurulent ocular discharge (67% [95% CI, 48%-82%]; 991 patients from 7 studies), and bilateral eye symptoms (59% [95% CI, 44%-72%]; 1424 patients from 10 studies).

Diagnostic Accuracy of Clinical Findings: Primary Analysis

The literature search identified 11 studies (6 with a quality level of 1 and 5 with a quality level of 2; n = 1725 participants) that reported on symptoms or signs in patients with both viral and bacterial conjunctivitis for which traditional 2 × 2 tables could be constructed for estimates of diagnostic accuracy (Table 111-3 and Supplemental Tables 2, 3, 4). The presence of pharyngitis (sensitivity range, 0.55-0.58; specificity range, 0.89-0.94; positive LR range, 5.4-9.9), preauricular lymphadenopathy (sensitivity range, 0.17-0.31; specificity range, 0.93-0.94; positive LR range, 2.5-5.6), and contact with another person with red eye (sensitivity, 0.18 [95% CI, 0.14-0.22]; specificity, 0.93 [95% CI, 0.90-0.95]; positive LR, 2.5 [95% CI, 1.6-3.7]) were the clinical findings that best discriminated viral from other forms of conjunctivitis. The presence of watery discharge and conjunctival follicles also were associated with higher positive LRs; however, the 95% CIs were wide and crossed 1.0. The absence of mucopurulent ocular discharge was associated with higher likelihood of viral conjunctivitis (sensitivity, 0.34 [95% CI, 0.27-0.42]; specificity, 0.24 [95% CI, 0.14-0.38]; positive LR, 2.7 [95% CI, 1.8-4.1]).

The clinical findings associated with a higher likelihood of bacterial conjunctivitis included mucopurulent ocular discharge (sensitivity, 0.76 [95% CI, 0.60-0.87]; specificity, 0.66 [95% CI, 0.58-0.73]; positive LR, 2.1 [95% CI, 1.7-2.6]) and otitis media (sensitivity, 0.24 [95% CI, 0.20-0.29]; specificity, 0.91 [95% CI, 0.85-0.94]; positive LR, 2.5 [95% CI, 1.5-4.4]). All studies that evaluated otitis media were in children only (514 participants from 3 studies; age range, 1 month-18 years). The likelihood of bacterial conjunctivitis also was higher when the patient did not have pharyngitis (sensitivity range, 0.06-0.11; specificity range, 0.42-0.45; positive LR range, 2.1-2.14). The LRs for the other clinical findings had wide 95% CIs that crossed 1.0 or were close to 1.0, rendering them not useful.

Discussion

Infectious conjunctivitis is the most common ophthalmic presentation to generalist physicians. The results of this review indicated that the prevalence of bacterial conjunctivitis in children was higher than that of viral conjunctivitis, although the number of studies was small and the 95% CIs around the estimates were wide. The data on the relative causes of bacterial vs viral conjunctivitis in adults were limited but suggested that viral conjunctivitis was more common than bacterial conjunctivitis.

In this analysis, no single symptom or sign reliably differentiated patients with viral conjunctivitis from those with bacterial conjunctivitis. However, a number of clinical findings were associated with moderately high LRs that may assist the clinician. For example, the presence of pharyngitis, lymphadenopathy, or contact with another person with red eye were each associated with a modestly higher likelihood of viral conjunctivitis. Likewise, the presence of mucopurulent discharge and otitis media (in a child) similarly were associated with a higher likelihood of bacterial conjunctivitis.

The nature of discharge (watery vs mucopurulent) has been considered to be highly relevant in differentiating viral conjunctivitis from bacterial conjunctivitis. The results from the meta-analysis indicated that watery discharge was associated with higher LRs for viral conjunctivitis, whereas mucopurulent discharge was associated with bacterial conjunctivitis. However, 21% of 319 cases of viral conjunctivitis were associated with mucopurulent discharge, and 26% of 227 cases of bacterial conjunctivitis were associated with watery discharge, therefore, the nature of discharge alone was insufficient to differentiate the 2 types of conjunctivitis. In 1 study,⁹ patients (n = 177) who reported having had eyelids stuck together on awakening were more likely to have positive bacterial cultures (odds ratio, 11 [95% CI, 3.5-36]); however, the results of this study may have been biased by inclusion criteria requiring the presence of mucopurulent discharge, eyelids stuck together on awakening, or both.

In the meta-analysis, rapid immunoassay testing (Supplemental Table 4) was associated with significant positive (9.5) and negative (0.26) LRs in identifying viral conjunctivitis. However, 2 of 5 studies (which accounted for 66% of 464 patients) were sponsored by the manufacturer of the rapid immunoassay test.^40,41 The other studies did not replicate the high sensitivity and specificity reported in these 2 articles.

A brief clinical history can aid in distinguishing viral conjunctivitis from bacterial conjunctivitis and may be helpful in ruling out other causes of red eye. Although infectious conjunctivitis may be associated with mild ocular irritation, photophobia, or both, more severe symptoms such as decreased visual acuity, pain, or severe photophobia suggest a more serious diagnosis. Important points to elicit from the patient’s history include the timing of symptom onset (ie, acute vs insidious), nature of discharge, whether eyelids are stuck together on awakening, presence of upper tract respiratory infection, and contact with another person with red eye. Contact lens wear is also important to elicit on history because patients may present with rare, atypical infections such as acanthamoeba.⁵²

Many generalist physicians have limited access to the examination equipment necessary to perform an in-depth ocular examination, such as slit lamp biomicroscopy. However, inspection of the eye with the assistance of an appropriate light source can provide valuable information. For example, without a slit lamp, the clinician can evaluate the pattern of redness (monocular or binocular, diffuse or sectoral) and clarity of the cornea. Sectoral redness of the conjunctiva or sclera or whitening of the cornea are not typical of conjunctivitis. If present, the finding of preauricular lymphadenopathy can suggest viral conjunctivitis.

Most cases of infectious conjunctivitis are relatively mild and can be managed by the generalist physician. Many generalist physicians may feel compelled to prescribe topical antibiotics without a clear diagnosis of viral or bacterial conjunctivitis because of concern about undertreatment of bacterial conjunctivitis and potential resultant morbidity. However, most cases of bacterial conjunctivitis are self-limited, and topical antibiotics only reduce recovery time by an estimated 1 to 2 days.⁵⁰ If bacterial conjunctivitis is suspected, it is prudent to prescribe antibiotics to hasten clinical recovery (even if by a short time only) and possibly reduce the risk of transmitting the infection to others. Another proposed treatment strategy is delayed prescribing in which patients are told to start antibiotic eye drops if no clinical resolution is seen after 48 hours.⁷

The clinical findings that warrant referral to ophthalmology care include conjunctivitis in a neonate, severe symptoms, decreased visual acuity, failure to respond to therapy, the presence of conjunctival membranes, and diagnostic uncertainty. A nonurgent referral to an ophthalmologist is appropriate in cases of suspected viral conjunctivitis to examine for herpes-related conjunctivitis or for late complications such as corneal infiltrates. If the clinician is uncertain of the diagnosis, referral to ophthalmology care is warranted in preference to indiscriminate administration of antibiotic eye drops.

Limitations

This review has several limitations. First, the meta-analysis was limited by a relatively small number of studies with a methodological quality level of 1 or 2 that included both patients with viral conjunctivitis and those with bacterial conjunctivitis for which 2 × 2 tables could be created. As such, the frequency of each clinical finding is reported for the 2 conditions to assist the clinician in gauging the diagnostic utility of a given finding. For example, the positive LR range in the study for conjunctival follicles in identifying viral conjunctivitis (based on a sample of only 120 patients from a single level 1 study and a single level 2 study) was 0.88 to 3.1; however, this finding was reported in 77% of patients with viral conjunctivitis vs 32% with bacterial conjunctivitis based on a larger number of patients in studies of with a quality level of 1 to 4.^25,45

Second, considering viral conjunctivitis or even adenoviral conjunctivitis as a single clinical entity is difficult. Even though patients with viral conjunctivitis will not benefit from antibiotics, a diverse group of viruses may result in similar yet varied presentations of conjunctivitis. For example, many serotypes of adenovirus have been reported to cause outbreaks of conjunctivitis, each with varied clinical presentations. The literature may not equally represent all adenovirus serotypes. It is known that serotypes of adenovirus vary across region⁵⁴ and change from year to year.^51,53 In addition, several studies included in the meta-analysis were based on older studies (some > 10 years old) and may not necessarily represent the etiologies of infectious conjunctivitis in 2022.

Third, the meta-analysis only investigated single symptoms or signs of bacterial vs viral conjunctivitis. It is unknown whether combinations of findings could produce a more accurate prediction model such as those used for some other conditions (eg, criteria for identifying streptococcal pharyngitis).⁵⁵

Case Resolution

On further questioning, the patient also reported mild sore throat (positive LR range for viral conjunctivitis, 5.4-9.9) and that her eyelids were not stuck together on awakening (positive LR for viral conjunctivitis, 3.5); no preauricular nodes were palpable on examination (positive LR range for viral conjunctivitis, 0.73-0.89). The findings were most consistent with viral conjunctivitis. The patient was disappointed to not be receiving a prescription for antibiotic eye drops, but was educated about the likelihood of a viral infection and agreed to try over-the-counter eye drops and cold compresses for comfort. She was instructed to limit spread to household members by washing hands thoroughly and not sharing towels. She also was instructed to call back if redness and discharge were not improving within 48 hours to 72 hours.

Conclusion: Clinical Bottom Line

In this review, bacterial conjunctivitis was more common than viral conjunctivitis in children and viral conjunctivitis was more common than bacterial conjunctivitis in adults, although the prevalence estimates were based on limited evidence. Symptoms and signs associated with a higher likelihood of viral conjunctivitis in adults and children included concomitant pharyngitis, an enlarged preauricular node, and contact with another person with red eye, and signs associated with a higher likelihood of bacterial conjunctivitis included the presence of mucopurulent discharge and otitis media, but no single symptom or sign differentiated the 2 conditions with high certainty.

Acknowledgment

The following disclosures were reported at the time this original article was first published in JAMA.

Conflict of Interest Disclosures: Dr Simel reported receiving nonfinancial support from the Veterans Health Administration during the conduct of the study and receiving personal fees from JAMAevidence outside the submitted work. No other disclosures were reported.

Funding/Support: Dr Simel’s work was supported by the Durham Center of Innovation to Accelerate Discovery and Practice Transformation at the Durham VA Health Care System.

Role of the Funder/Sponsor: The funder/sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Dr Simel is Associate Editor of JAMA, but he was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Additional Contributions: We thank Stephanie Baxter, MD (Queen’s University), Isabella Ircher, PhD (Queen’s University), Kelly Muir, MD (Durham VA Health Care System and Duke University), and William Clayton Bordley, MD (Duke University), for their advice and expertise in manuscript preparation. These contributors were not financially compensated for their contributions.

References

Supplemental References