Comparing cases with controls with initial univariate analysis to suggest possible risk factors for endophthalmitis following vitrectomy
| Variable (potential risk factor) | Cases (total n=28) | Controls (total n=272) | p Value |
|---|---|---|---|
| Age | Median average age 61 | Median average age 63 | 0.95* |
| Gender (male sex) | 67% | 50% | 0.07† |
| Diabetes mellitus | 9/27 (33.3%) | 51/268 (19.0%) | 0.078† |
| Immunosuppression | 3/25 (12.0%) | 5/267 (1.9%) | 0.023‡ |
| Preop oral steroid use | 2/27 (7.4%) | 2/267 (0.7%) | 0.043‡ |
| Preop topical steroid use | 12/27 (44.4%) | 4/267 (1.5%) | <0.001‡ |
| Recent intraocular surgery | 6/21 (28.6%) | 8/270 (3.0%) | <0.001† |
| Lens status (phakic) | 15/27 (55.6%) | 190/269 (70.6%) | 0.11† |
| Previous glaucoma filtration surgery | 2/27 (7.4%) | 0/267 (0%) | 0.008‡ |
| Surgeon grade (non-consultant) | 5/27 (18.5%) | 78/270 (28.9%) | 0.25† |
| Port size (23G or 25G) | 18/28 (64.3%)§ | 104/265 (39.2%) | 0.011† |
| One or more ports sutured | 16/26 (61.5%) | 205/255 (80.4%) | 0.025† |
| Conjunctiva displaced | 14/17 (82.4%) | 68/84 (81.0%) | 0.89† |
| Subconjunctival antibiotic | 26/27 (96.3%) | 255/271 (94.1%) | 0.64† |
| Intra-cameral or intra-vitreal antibiotic | 1/27 (3.7%) | 7/271 (2.6%) | 0.73† |
| Postoperative hypotony | 0/27 (0%) | 2/271 (0.7%) | 0.65† |
| Postoperative wound leak | 0/27 (0%) | 0/271 (0%) | n/a |
| Retinal detachment | 6/27 (22.2%) | 132/271 (48.7%) | 0.015† |
| Macular hole | 2/27 (7.4%) | 45/271 (16.6%) | 0.28‡ |
| Epi-retinal membrane | 8/27 (29.6%) | 40/271 (14.8%) | 0.055‡ |
| Diabetic vitreous haemorrhage | 6/27 (22.2%) | 19/271 (7.0%) | 0.017† |
| Non-diabetic vitreous haemorrhage | 2/27 (7.4%) | 12/271 (4.4%) | 0.37‡ |
| Diabetic delamination | 3/27 (11.1%) | 14/271 (5.2%) | 0.19‡ |
| Vitreo-macular traction syndrome | 4/27 (14.8%) | 7/271 (2.6%) | 0.011‡ |
| Vitreous opacity | 2/27 (7.4%) | 3/271 (1.1%) | 0.067‡ |
| Other indication for vitrectomy | 2/27 (7.4%) | 7/271 (2.6%) | 0.052‡ |
p Values <0.05 are highlighted in bold and these risk factors were subsequently entered into the logistic regression analysis. Possible risk factors from this analysis were then investigated with multivariate analysis to establish if they were an independent risk factor (see Table 2).
*Mann–Whitney test.
†χ2 test.
‡Fisher's exact test.
§19 cases were 23/25 gauge, but only 18 cases that were 23/25 gauge were included in the logistic regression since one case had insufficient data to be included in the multivariate analysis.