Purpose
While povidone-iodine (PI) is a commonly used antiseptic solution for postinjection endophthalmitis (PIE) prophylaxis, chlorhexidine (CHX) may be better tolerated by patients. This systematic review and meta-analysis aims to comprehensively compare the efficacy and safety of CHX and PI for PIE prophylaxis.
Design
Systematic Review and Meta-Analysis.
Methods
MEDLINE, Embase, and the Cochrane Library were systematically searched for relevant literature from January 2000 to July 2024. Studies that compared CHX with PI for either PIE prophylaxis, functional outcomes, or safety outcomes were included. A random-effects meta-analysis was performed, using the Mantel-Haenszel method for dichotomous outcomes and inverse variance method for continuous outcomes. The primary outcomes were the comparative PIE incidence after CHX versus PI. Secondary outcomes pertained to best-corrected visual acuity (BCVA), patient intolerance, pain, safety endpoints, and specimens cultured.
Results
A total of seven studies were included. The incidence of presumed PIE ( n CHX = 60/185,849, 0.032% vs n PI = 69/269,491, 0.026%), culture-positive PIE ( n CHX = 30/185,849 0.016% vs n PI = 24/269,491, 0.009%; odds ratio [OR] = 2.03, 95% confidence interval [CI] = [0.74, 5.58], P = .17) and culture-negative PIE ( n CHX = 30/185,849, 0.016% vs n PI = 45/269,491, 0.017%; OR = 0.97, 95% CI = [0.45, 2.06], P = .93) were nonsignificantly different between PI and CHX groups. No significant risk differences were observed for these outcomes. Across three studies reporting on 54 eyes with culture-positive endophthalmitis, the odds of having positive Staphylococcus epidermidis cultures were significantly lower in the CHX group ( n CHX = 9/30, 30.0% vs n PI = 18/24, 75.0%; OR = 0.15, 95% CI = [0.04, 0.54], P = .004); however, the odds were nonsignificantly different for Staphylococcus aureus ( P = .64), Streptococcus viridans ( P = .46), Streptococcus mitis ( P = .83), and Enterococcus faecalis ( P = .46). In cases of endophthalmitis, the mean BCVA was nonsignificantly different between CHX and PI groups at the last study observation ( P = .67). The systematic review revealed minimal patient intolerance to CHX, more favorable reductions in pain with CHX than PI, greater corneal epitheliopathy with PI than CHX, and nonsignificantly different reductions in ocular surface microbial isolates between comparators.
Conclusions
CHX and PI were associated with nonsignificantly different rates of presumed, culture-positive, and culture-negative PIE. Endophthalmitis cases associated with both antiseptic agents had a comparable prognosis with respect to BCVA. However, CHX exhibited a more favorable safety and pain profile compared to PI and was associated with a lower odds of positive S. epidermidis cultures in endophthalmitis cases.
INTRODUCTION
I ntravitreal injections (IVIs) of antivascular endothelial growth factor (anti-VEGF) agents and corticosteroids are widely used in the treatment of retinal vascular disorders, , where multiple injections are usually required to maintain a therapeutic effect. Postinjection endophthalmitis (PIE) is a devastating complication of IVIs, with an estimated incidence of 0.049% (ie, 1 in 1786 IVIs) to 0.056% (ie, 1 in 2041 IVIs). Mitigating the risks of PIE is imperative; in a multicenter cohort study, the PIE Study Group demonstrated that the use of prefilled syringes during IVIs of ranibizumab was associated with reduced rates of culture-positive PIE. No-talking policies during IVI administration may further reduce the risk of PIE. However, antisepsis techniques are most crucial for preventing PIE. Currently, povidone-iodine (PI) remains the most commonly used antiseptic solution in this setting. The routine use of PI as an antiseptic agent for IVIs is supported by its widespread availability, cost-effectiveness, broad-spectrum antimicrobial activity, and low incidence of microorganism resistance.
Chlorhexidine (CHX) may serve as a viable alternative to PI for IVI antisepsis, particularly in patients reporting significant post-IVI pain. For instance, a previous survey of patients undergoing IVIs found that among those reporting high pain levels with 5% PI, self-reported pain scores subsided after switching to 0.1% CHX gluconate (ie, from 8 out of 10 pain to 3 out of 10 pain). Nonetheless, the comparative efficacy and safety of PI and CHX with respect to the risk of PIE remains unclear. In the present systematic review and meta-analysis, we aim to provide a comprehensive analysis of the comparative incidence of PIE attributed to the use of the antiseptic agents PI and CHX.
METHODS
This study was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the identification number CRD42024566575. We abided by the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines. Furthermore, our study adhered to the tenets of the Declaration of Helsinki, and institutional review board approval was not required due to the nature of this review.
Search Strategy and Eligibility Criteria
We searched Ovid MEDLINE, Embase, and Cochrane Library from January 2000 to July 2024 for relevant literature. A sample search strategy for Embase is available in Supplemental Table 1. We included any observational study or randomized controlled trial (RCT) that compared CHX with PI for either (1) PIE prophylaxis, (2) functional outcomes, or (3) safety outcomes. Our systematic review also included studies reporting on the efficacy or safety of only CHX in this context. We excluded (1) non-English studies, (2) in vitro studies, (3) gray literature, and (4) studies reporting on patients receiving CHX or PI antisepsis for any ocular surgery.
Study Selection and Data Collection
Study records were uploaded to Covidence (Veritas Health Innovation). The titles and abstracts of each study were reviewed independently by at least two authors (A.M., B.K.T., R.S.H.). Studies recommended for initial inclusion then underwent full-text screening by the same independent reviewers. Data extraction and risk of bias evaluations for included studies were conducted independently by at least two reviewers (B.K.T., R.S.H., T.D.) using prepiloted standardized forms on Microsoft Excel (v16.62; Redmond, Washington, United States). Throughout these study processes, conflicts between reviewers were resolved through discussion and adjudication by another impartial author (M.M.P.).
From each study, we extracted study characteristics (ie, journal, authors, publication year, study period, country, and study design), participant characteristics (ie, sex, age, injection indications, IVI agents, injection technique, and antisepsis formulations), and outcomes of interest. Our primary outcomes were the comparative odds ratio (OR) of PIE following antisepsis with CHX versus PI. Our secondary outcomes were the differences between the CHX and PI groups in terms of cultured specimens in endophthalmitis cases, best-corrected visual acuity (BCVA), intolerance, pain, safety, and microbial cultures from ocular surface swabs.
Risk of Bias and Certainty of Evidence
Risk of bias was evaluated using Cochrane’s Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I) tool for observational studies and the Risk of Bias tool 2 (RoB-2) tool for RCTs. , The quality of evidence derived from meta-analysis outcomes was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE). Two independent reviewers evaluated bias risks (R.S.H., T.D.) and certainties of evidence (A.M., R.S.H.).
Data Analysis
Study participant characteristics were summarized using descriptive statistics. Continuous data were reported as mean ± standard deviations from individual studies and as weighted mean differences for meta-analysis outcomes. Categorical variables were reported as counts and proportions from individual studies and as ORs for meta-analysis outcomes. 95% confidence intervals (CIs) were reported for both continuous and categorical meta-analysis outcomes. Pairwise, random-effects meta-analyses weighted by the number of IVIs were conducted using Review Manager (RevMan 5.4; The Nordic Cochrane Centre, The Cochrane Collection). Risk differences were also calculated for PIE outcomes as an absolute measure of effect using RStudio (PBC, Massachusetts, United States). The inverse variance and Mantel-Haenszel statistical methods were applied to continuous and dichotomous meta-analysis outcomes, respectively. Studies with zero events in both treatment arms did not contribute to OR effect estimates. Subgroup analyses for studies reporting on 0.05% CHX and 5% PI were performed. Additionally, leave-one-out sensitivity analyses were performed for pairwise meta-analyses. Publication bias was assessed using visual inspection of funnel plots. The Cochrane Handbook I 2 thresholds were used to determine statistical heterogeneity: might not be important (0%-40%), moderate (30%-60%), substantial (50%-90%), and considerable (75%-100%) heterogeneity. A P value of less than .05 denoted statistical significance.
RESULTS
Study Selection and Characteristics
A total of 1,029 articles underwent title and abstract screening, of which 8 full-texts were assessed for eligibility and 7 studies reporting on CHX antisepsis were included ( Figure 1 ). , , Six (86%) studies were observational , , and one (14%) was an RCT. Six (86%) studies compared outcomes between patients receiving CHX and PI, , , , whereas one compared outcomes between patients receiving CHX acetate 0.05% and CHX gluconate 0.1%. All studies were conducted in English-speaking countries: three (43%) in Australia, , , two (29%) in Canada, , and two (29%) in the United States. , Four (57%) studies specified that patients with age-related macular degeneration were included, , , , and three (43%) specified that patients with diabetic macular edema and retinal vein occlusion were included, , , one (14%) of which also included patients with myopic choroidal neovascularization. Five (71%) studies reported using intravitreal ranibizumab, , , , , five (71%) reported using intravitreal aflibercept, , , , four (43%) reported using intravitreal bevacizumab, , , , and two (29%) reported using intravitreal steroids. , One (14%) study also included IVIs of methotrexate and foscarnet (treatment indication not specified).
Six (86%) studies used aqueous CHX , and one (14%) used an alcohol-based CHX. Four (57%) studies used CHX 0.1% , , , and three (43%) used CHX 0.05%. , , Four (57%) studies used PI 5% , , , and one (14%) used PI 10%. Four (57%) studies reported administering IVIs in an office setting , , , and two (29%) in a procedure room. , A detailed summary of study characteristics, including techniques and precautions for IVI administration, is provided in Table 1 . Staphylococcus epidermidis was the most cultured specimen, and an exhaustive list of the causative organisms in endophthalmitis cases can be found in Table 2 .
| Study | Study Period | Country | Antiseptic Agent | Exposure Time | Treatment Regimen | Location | Number of Injections |
|---|---|---|---|---|---|---|---|
| Ton et al | Jan 1, 2019, to Nov 30, 2021 | USA | 0.05% CHX 5% PI | 7 min minimum + 30 s before injection | For intravitreal injections: 1. At least 5 drops of the antiseptic agent for at least 7 min 2. Repeat antisepsis before injection For endophthalmitis cases: 1. Culture sample acquisition via pars plana vitreous tap or anterior chamber paracentesis 2. Intravitreal antibiotics (vancomycin 1 mg in 0.1 mL) and intravitreal ceftazidime (2.25 mg in 0.1 mL); amikacin (400 mg in 0.1 mL) was substituted for ceftazidime if the patient had a penicillin allergy | Office | 67,695 |
| Merani et al | Aug 1, 2011, to Feb 28, 2015 | Australia | 0.05% CHX 0.1% CHX | 30, 60, or 90 s | For endophthalmitis cases: 1. Vitreous tap 2. Injection of vancomycin 1 mg in 0.1 mL and ceftazidime 2.25 mg in 0.1 ml 3. Pars plana vitrectomy (for 1 patient) | Office | 40,535 |
| Mishra et al | May 2019 to Oct 2022 | Canada | 0.05% CHX (4% alcohol base) 10% PI | 30 s before injection | For intravitreal injection: 1. Tetracaine 1% anesthesia (with 1% subconjunctival lidocaine if needed) 2. Antisepsis 3. Postinjection flushing with increased lubricating drops if uncomfortable For endophthalmitis cases: 1. Vitreous tap and injection of intravitreal antibiotics 2. Vitrectomy with injection of intravitreal antibiotics 3. Both treatments | Office | 170,952 |
| Oakley et al | May 2013 to July 2015 | Australia | 0.1% CHX | 10 min | For intravitreal injections: 1. Drop of tetracaine 2. Drop of 0.1% CHX gluconate 3. Lignocaine gel 4. Eye taped shut for 10 min 5. Irrigation with 0.1% CHX gluconate 6. Drop of CHX prior to injection | Dedicated procedure room | 4,322 |
| Oakley et al | – | Australia | CHX 5% PI | 2 min | – | – | 40 |
| Stephenson et al | July 2009 to July 2022 | Canada | 0.1% CHX 5% PI | 2 min minimum and again immediately before injection | For endophthalmitis cases: 1. Vitreous and/or aqueous sample for microbiological analysis 2. Intravitreal injection of broad-spectrum antibiotics (typically vancomycin and ceftazidime) ± dexamethasone 3. Pars plana vitrectomy for severe endophthalmitis | Office | 216,593 |
| Ali et al | May 2018 to Sept 2018 | USA | 0.1% CHX 5% PI | 1 min | For intravitreal injections: 1. Single drop topical anesthesia (0.5% proparacaine hydrochloride) 2. 1 drop of 5% PI or 0.1% CHX gluconate in right eye and left eye receiving other agent | Treatment room | 100 |
| Comparative Study | CHX | PI |
|---|---|---|
| Ton et al | Streptococcus parasanguinis, n = 1/9 Propionibacterium acnes, n = 1/9 | Streptococcus mitis, n = 1/13 Staphylococcus epidermidis, n = 1/13 |
| Mishra et al | Staphylococcus epidermidis, n = 4/20 Streptococcus viridans, n = 2/20 Enterococcus faecalis, n = 2/20 Staphylococcus aureus, n = 0/20 Citrobacter koseri, n = 1/20 Granulicatella adiacens, n = 1/20 Staphylococcus hominis, n = 1/20 Streptococcus dysgalactiae, n = 1/20 | Staphylococcus epidermidis, n = 8/29 Streptococcus viridans, n = 1/29 Enterococcus faecalis, n = 1/29 Staphylococcus aureus, n = 2/29 Citrobacter koseri, n = 0/29 Granulicatella adiacens, n = 0/29 Staphylococcus hominis, n = 0/29 Streptococcus dysgalactiae, n = 0/29 |
| Stephenson et al | Staphylococcus epidermidis, n = 5/31 Staphylococcus aureus, n = 1/31 Staphylococcus capitis, n = 1/31 Streptococcus viridans, n = 1/31 Streptococcus mitis, n = 4/31 Enterococcus faecalis, n = 1/31 Dermabacter, n = 1/31 Corynebacterium, n = 1/31 Serratia, n = 1/31 Haemophilus, n = 0/31 | Staphylococcus epidermidis, n = 9/27 Staphylococcus aureus, n = 0/27 Staphylococcus capitis, n = 0/27 Streptococcus viridans, n = 0/27 Streptococcus mitis, n = 0/27 Enterococcus faecalis, n = 0/27 Dermabacter, n = 0/27 Corynebacterium, n = 0/27 Serratia, n = 0/27 Haemophilus, n = 1/27 |
Risk of Bias and GRADE results
Per the ROBINS-I tool, five (83%) observational studies had a moderate risk of bias and one (17%) had a low risk of bias (Supplemental Table 2). Reasons for downgrading study quality included concerns regarding confounding (50%, n = 3/6), outcome measurement (33%, n = 2/6), selection of participants (17%, n = 1/6), and missing data (17%, n = 1/6). Per the RoB-2 tool, the included RCT had a low risk of bias with respect to randomization process, deviation from intended interventions, missing outcome data, outcome measurement, and selection of reported results. Per the GRADE tool, the certainty of evidence of meta-analysis outcomes ranged from low to moderate, as outlined in Supplemental Table 3.
Meta-Analysis: PIE Rates, BCVA, and Causative Organisms
Four studies compared the incidence of PIE in CHX and PI groups, wherein 129 endophthalmitis events were reported across 455,340 IVIs. , , , The absolute event rate of presumed PIE was comparable between the PI and CHX groups ( n CHX = 60/185,849, 0.032% vs n PI = 69/269,491, 0.026%). However, poor overlap in the 95% CIs was noted for this outcome, revealing variability in the odds of PIE across studies ( Figure 2 , A): Mishra et al had a higher odds of PIE associated with CHX (OR = 3.10, 95% CI = [1.75, 5.48], P < .01), while Ton et al and Stephenson et al demonstrated no significant difference between the two antiseptic agents (OR = 0.65, 95% CI = [0.28, 1.52], P = .32 and OR = 0.93, 95% CI = [0.55, 1.55], P = .78, respectively). Furthermore, there was no significant risk difference between agents for presumed PIE (risk difference = 0.00008, 95% CI = [−0.00020, 0.00035], P = .58).
The absolute event rates for culture-positive PIE were 0.016% in the CHX group ( n = 30/185,849) and 0.009% in the PI group ( n = 24/269,491), while the rates for culture-negative PIE were 0.016% in the CHX group ( n = 30/185,849) and 0.017% in the PI group ( n = 45/269,491). The meta-analysis demonstrated that the odds of culture-positive PIE (OR = 2.03, 95% CI [0.74, 5.58], P = .17, I² = 64%, Figure 2 , B) and culture-negative PIE (OR = 0.97, 95% CI [0.45, 2.06], P = .93, I² = 61%, Figure 2 , C) were nonsignificantly different between the PI and CHX groups. Additionally, there were no significant risk differences observed for these outcomes (risk difference = 0.00007, 95% CI = [−0.00005, 0.00019], P = .25 and risk difference = −0.00001, 95% CI = [−0.00013, 0.00010], P = .80, respectively).
Across three studies reporting on 54 eyes with culture-positive endophthalmitis, , , the odds of any Staphylococcus -positive cultures were significantly lower in the CHX group compared to the PI group ( n CHX = 12/30, 40.0% vs n PI = 20/24, 83.3%; OR = 0.12, 95% CI = [0.03, 0.49], P = .003, I 2 = 0%, Figure 3 , A). This finding was consistent for Staphylococcus epidermidis cultures ( n CHX = 9/30, 30.0% vs n PI = 18/24, 75.0%; OR = 0.15, 95% CI = [0.04, 0.54], P = .004, I 2 = 0%, Figure 3 , B) but did not hold true for Staphylococcus aureus cultures ( n CHX = 1/28, 3.6% vs n PI = 1/22, 4.5%; OR = 0.56, 95% CI = [0.05, 6.40], P = .64, I 2 = 13%, Figure 3 , C). The odds of any Streptococcus -positive cultures ( n CHX = 9/30, 30.0% vs n PI = 2/24, 8.3%; OR = 3.96, 95% CI = [0.72, 21.74], P = .11, I 2 = 0%, Figure 3 , D), including Streptococcus viridans ( n CHX = 3/28, 10.7% vs n PI = 1/22, 4.5%; OR = 2.14, 95% CI = [0.28, 16.03], P = .46, I 2 = 0%, Figure 3 , E) and Streptococcus mitis ( n CHX = 4/18, 22.2% vs n PI = 1/12, 8.3%; OR = 1.48, 95% CI = [0.04, 51.66], P = .83, I 2 = 54%, Figure 3 , F) were nonsignificantly different between culture-positive endophthalmitis cases in the CHX and PI groups. Additionally, the odds of culture-positive endophthalmitis with Enterococcus faecalis were nonsignificantly different between comparators ( n CHX = 3/28, 10.7% vs n PI = 1/22, 4.5%; OR = 2.14, 95% CI = [0.28, 16.03], P = .46, I 2 = 0%, Figure 3 , G).
