Acute Conjunctivitis, Bacterial: Likelihood Ratios

Introduction: Acute conjunctivitis can be either viral or bacterial. One study in children had a 54% prevalence of bacterial etiology (42% H flu, 12% S pneumo) and 20% prevalence of adenoviral, with ~30% culture negative [J Pediatr. 1981 Apr;98(4):531-6. PMID: 6970802]. Another study in adults showed that 32% of patients had a bacterial etiology, with 47% S pneumo, 23% S aureus, 16% H flu [BMJ. 2004 Jul 24;329(7459):206-10. PMID: 1520119].

[Edit Diagnosis] [Merge dx] [Add prevalence]

Tags: Infection Ophthalmology Tag this Diagnosis.

The likelihood ratios of findings for Acute Conjunctivitis, Bacterial are listed below. See the left navigation bar to change the display.

Pre-Test Probability (Prevalence): %.	Post-Test Probability (Predictive Value): %.	Switch to display mode.
Note that calculating probabilities from more than one** finding is inherently inaccurate because findings are not independent. (For example, using two positive findings that share a common pathogenesis is likely to overestimate the true probability.)

Good Negative Likelihood Ratio
Finding	Result	+LR	-LR
[ + ] Glued Eyes	+ 0 -	1.3	0.3
either unilateral or bilateral, in the morning Study: BMJ. 2004 Jul 24;329(7459):206-10. PMID: 15201195. [Note that I have recalculated the PLR and NLR from their raw data - I disagree with the way they calculated certain things.]
[ + ] No History of Conjunctivitis	+ 0 -	1.2	0.4
Study: BMJ. 2004 Jul 24;329(7459):206-10. PMID: 15201195. [Note that I have recalculated the PLR and NLR from their raw data - I disagree with the way they calculated certain things.]

Good Positive Likelihood Ratio
Finding	Result	+LR	-LR
[ + ] Bilateral Glued Eyes	+ 0 -	3.6	0.7
in the morning Study: BMJ. 2004 Jul 24;329(7459):206-10. PMID: 15201195. [Note that I have recalculated the PLR and NLR from their raw data - I disagree with the way they calculated certain things.]
[ + ] Bilateral Secretion	+ 0 -	2.3	0.8
Study: BMJ. 2004 Jul 24;329(7459):206-10. PMID: 15201195. [Note that I have recalculated the PLR and NLR from their raw data - I disagree with the way they calculated certain things.]

Poorly Diagnostic Findings
Finding	Result	+LR	-LR
[ + ] Purulent Secretion	+ 0 -	0.8	1.1
(note that this is not sens / spec at all - in fact it may be slightly less prevalent in bacterial!) Study: BMJ. 2004 Jul 24;329(7459):206-10. PMID: 15201195. [Note that I have recalculated the PLR and NLR from their raw data - I disagree with the way they calculated certain things.]