This is the largest study to date to evaluate the incidence of and risk factors for retinal artery occlusion (RAO) during cardiac surgery. The authors reviewed a national sample of discharge information on 5.8 million cardiac procedures performed between 1998 and 2013 and found the overall incidence of RAO was 7.77 cases per 10,000 procedures.
Conditions associated with significant increased risk were: giant cell arteritis (OR, 7.73), transient cerebral ischemia (7.67), carotid artery stenosis (7.52), embolic stroke (4.43), hypercoagulability (2.90), myxoma (2.43), diabetes with ophthalmic complications (1.89) and aortic insufficiency (1.85).
In contrast, acute coronary syndrome (0.72), congestive heart failure (0.73), type 2 diabetes (0.74), female gender (0.77), thrombocytopenia (0.79), atrial fibrillation (0.82) and smoking (0.82) and were associated with a lower risk.
Procedures requiring surgical opening of the heart, such as septal repair and mitral or aortic valve surgeries were associated greater risk of RAO, whereas less invasive procedures such as coronary artery bypass grafting (CABG) were negatively associated with RAO (0.61).
Though perioperative RAO is well reported in literature, there is little research on the risk factors for patients undergoing cardiac surgery. Unsurprisingly, the authors found several shared risk factors for perioperative RAO and stroke, such as carotid artery stenosis and hypercoagulable states. The suspected pathogenic role of emboli is further confirmed by the finding that open-heart surgeries and postoperative bleeding substantially increased the risk of RAO. Also, thrombocytopenia was protective of RAO, and though the lack of platelets increases bleeding, it also decreases emboli.
Interestingly, the authors did not find use of cardiopulmonary bypass (CPB) to increase risk, despite the significant risk of stroke found previously if CPB is used for more than 2 hours. The lack of association found between well-established risk factors including diabetes mellitus and atrial fibrillation is also noteworthy, yet the authors point out these findings should be interpreted with caution due to coding issues for these diseases.
One limitation of this study was its reliance on discharge data rather than complete patient medical records. Discharge data are more susceptible to undocumented diagnoses, over- or under-diagnosis and coding errors. In addition, this data set cannot provide the valuable information on the severity of vision loss or long-term outcomes that a longitudinal follow-up study would provide.
Because perioperative RAO is difficult to treat and often leads to irreversible vision loss, physicians must understand the pathophysiology so that risk factors may be identified. In turn, this could prompt modification of the informed consent and surgical planning process.