Hypothesis: Patients undergoing pars plana vitrectomy for the indication of diabetic vitreous hemorrhage will have a lower incidence of postoperative vitreous hemorrhaging during the 6-month trial period when vitreous substitution with 20-30% sulfur hexafluoride gas is utilized compared to vitreous substitution with balanced salt solution.

Subjects enrolled into the study are randomized into one of the ensuing treatment cohorts: Group A patients undergo vitreous substitution with 20-30% sulfur hexafluoride (SF6) gas during PPV, whereas Group B patients undergo vitreous substitution with balanced salt solution (BSS) during PPV. Simple randomization allocates patients into treatment groups during PPV. Once all relevant maneuvers are completed by the surgeon prior to vitreous substitution, a simulated coin toss program decides the vitreous substitute, and therefore which group the patient enters. Randomization intraoperatively mitigates against selection bias of utilized surgical maneuvers such as endodiathermy on the part of the surgeon and allows for subjects to not undergo randomization in cases where intraoperative retinal breaks occur that would preclude the subject from receiving vitreous substitution with BSS.

Subjects undergo data collection at 3 postoperative visits: 1) 15 +/- 5 days following PPV, 2) 40 +/- 10 days following PPV, and 3) 185 +/- 15 days following PPV. Study subjects are evaluated at non-study times at the judgment of the examiner.

Inclusion:

1. The patient has medically-managed type I or II diabetes mellitus.

2. The age of the patient is > 18 years old,

3. Snellen best-corrected visual acuity is from 20/40 to hand motions at one foot in the subject's study eye

4. Proliferative diabetic retinopathy with a clinically evident vitreous hemorrhage of duration of at least one month by subjective history is present in the subject's study eye.

5. The vitreous hemorrhage is symptomatic and primarily responsible for the patient's reduced vision in the judgement of the examiner.

6. Grade 0 or I vitreoretinal adhesion according to the classification system described by Ahn et al [@] in the study eye is clinically present.

   • Ahn J, Woo SJ, Chung H, et al. The effect of adjunctive intravitreal bevacizumab for preventing postvitrectomy hemorrhage in proliferative diabetic retinopathy. Ophthalmology. 2011; 118: 2218-2226.

Exclusion:

1. The subject's study eye previously underwent anterior or posterior vitrectomy.
2. A lens or cornea opacity is thought to be responsible for two or more lines of reduced visual acuity in the subject's study eye (cataract, corneal scar, ectasia, etc.).
3. Optic nerve or retina disease otherwise not related to diabetes mellitus is thought to be responsible for two or more lines of reduced visual acuity in the subject's study eye (optic neuritis, macular degeneration, glaucoma, epiretinal membrane, etc.).
4. A non-ocular cause (i.e. cerebrovascular accident) or amblyopia is thought to be responsible for two or more lines of reduced visual acuity in the subject's study eye
5. Neovascular glaucoma with a high intraocular pressure ( > 30 mm Hg) is discovered in the subject's study eye.
6. Medically uncontrolled systemic hypertension (systolic > 200 mmHg or diastolic > 120 mmHg) is present.

The operating surgeons did not participate in the preoperative or postoperative care of the patient, including the data collection.