Secondary Logo

Journal Logo


The Frequency of Perioperative Vision Loss

Warner, Mary E. MD*,; Warner, Mark A. MD*,; Garrity, James A. MD†,; MacKenzie, Ronald A. DO*,; Warner, David O. MD*

Author Information
doi: 10.1097/00000539-200112000-00013
  • Free


The frequency of perioperative vision loss, especially for spinal surgery, has been increasing recently (1,2). In 1999, the American Society of Anesthesiologists formed a task force to review this issue and, if appropriate, to develop a practice advisory for clinicians. Unfortunately, the frequency of permanent vision loss associated with surgery and anesthesia has only infrequently been reported. The published rates vary considerably (3–5), with the most recent estimate being 1:61,000 by Roth et al. (3) in a referral center surgical population that was not well defined. Therefore, we undertook the following study to determine the frequency of new-onset vision loss persisting for longer than 30 days after surgery in a large population undergoing all types of noncardiac surgical procedures and receiving general or central neuraxis regional anesthetics.


With IRB approval, the Mayo Surgical Index Record System (MSIRS) database was queried to identify all patients who were diagnosed with new-onset vision loss after noncardiac surgical procedures performed from January 1, 1986 to December 31, 1998. Patients undergoing cardiac surgical procedures were excluded because their frequency of perioperative vision loss using similar criteria recently had been studied at this institution (6). Based on an earlier report of eye injuries in a heterogenous surgical population by Roth et al. (3), we assumed that patients undergoing more invasive surgical procedures, thereby requiring general or central neuraxis regional anesthetics, might have an increased frequency of perioperative vision loss. Therefore, we limited our review to only patients who received these major anesthetic techniques, excluding those who received peripheral nerve blocks, monitored anesthesia care, or local anesthesia only. Vision loss was defined as the new onset of visual deficit in any part of the visual field or an entire loss of vision in one or both eyes. A number of patients underwent more than one surgical procedure during this study period. We considered each procedure and anesthetic to represent new potential risk for perioperative visual loss and thereby included all cases. Information for patients who died within 30 days of their final procedures was considered separately.

Description of Database

The MSIRS database provides extensive information on the perioperative care and outcomes of the large Mayo Clinic surgical population in Rochester, Minnesota (7). It has been continuously developed since 1910, funded primarily by the Mayo Foundation and the National Institutes of Health. More than 1000 studies related to perioperative care have been published using information from this well-validated database. In brief, trained nurse abstractors collect a large number of patient, surgical, and anesthetic characteristics and measurements. This collection process is continually reviewed and validated. In general, data are classified using the criteria of the International Classification of Diseases for medical conditions (ICD-9-CM) (8) and hospital procedures (HICDA-2) (9). For characteristics not included in this classification scheme, definitions and classifications designed and validated over time by Mayo Clinic are used.

The complete medical records of Mayo Clinic patients are readily available to investigators. Mayo Clinic collects all information for its patients in single-unit medical records. These records have information on all aspects of patients’ medical care, including patient demographic, health, function, laboratory, imaging, surgical, and medical data. Less than 100 medical records of more than 6 million Mayo Clinic patients since 1910 are missing.

Postoperative surveillance and data capture by this system is quite good. A 1993 study of 45,090 ambulatory surgical patients had a 30-day postoperative surveillance rate of 95.9%(10). A 1994 study of perioperative ulnar neuropathy in 1,129,692 patients who received anesthetics during a 35-yr period had a 1-yr postoperative surveillance rate of 95%(11). Internal validation tests of the reliability of the computer database compared with manual extraction of data from a random sample of medical records were performed in 1994 (12) and 1999. Both comparisons showed excellent correlation between MSIRS database information and manual review of medical records.

Identification of Cases of Vision Loss

Potential cases of perioperative vision loss were preliminarily identified by using an ICD data code for postoperative complications (997.99). Secondary diagnostic codes for visual loss, visual field disturbances, ischemic optic neuropathy, vascular occlusions affecting vision, and other unspecified causes of blindness were subsequently queried. These secondary codes range from those specific for vision loss to those that are less specific but often related to visual symptoms (e.g., blurriness affecting vision). One data abstractor who was not one of the original coding nurse abstractors reviewed the medical records of all patients identified by this process. All patients in whom new-onset vision loss initially was reported within 30 days of surgery also were evaluated at Mayo Clinic at least once beyond the 30-days period.

The records of all patients who had new-onset vision loss persisting longer than 30 days after surgery were reviewed by the study ophthalmologist to develop initial impressions if direct surgical trauma or resection of ocular or cerebral tissues may have contributed to these cases of visual loss. A second physician, one of the study anesthesiologists, then consulted the operating surgeons to confirm or refute the study ophthalmologist’s initial impressions that direct surgical trauma or tissue resection may have contributed to these cases of vision loss. The multidisciplinary nature of the Mayo Clinic practice and the long-term stability of its surgical staff facilitated this consultation process. This consultation process, including careful reviews of all pertinent medical records and surgical notes, was successfully completed for 177 patients, with agreement in each case. There were 12 additional cases in which the operating surgeons had retired and were unable to be contacted. For these 12 cases, this study anesthesiologist contacted the current (year 2000) practice committee chairs of the appropriate surgical departments and facilitated reviews of all pertinent medical records. In these 12 cases, the study ophthalmologist’s initial impressions matched the assessments of the practice committee chairs.

A similar process was used to determine the likely causes of new-onset vision losses persisting >30 days after surgery. The study ophthalmologist reviewed all pertinent medical records and developed an initial impression of the most likely cause of the visual loss for each potential case. The study anesthesiologist identified above then contacted an ophthalmologist who did not work at Mayo Clinic Rochester to also review these cases independently and assess the likely cause of visual loss. In each case, the study ophthalmologist and independent ophthalmologist agreed substantially on the most likely cause.

No statistical analyses were performed other than a simple calculation of frequency of this event and a general summary of characteristics of the Mayo Clinic surgical population during the study period. There were too few identified cases to allow any other statistical assessments or a case-control comparison for risk factors.


During the 12-yr study period, 414,701 patients underwent 505,854 surgical procedures and anesthetics. Of these, 4512 patients (1.1%) died within 30 days of their final procedures. Therefore, this study represents data on the remaining 410,189 patients who underwent 501,342 procedures and who survived at least 30 days after their final procedures. The mean age of this final subset of patients was 49.2 ± 13.1 yr (range, 0–108 yr). Nine percent were 12 yr of age or younger, and 18% were 65 yr of age or older. Fifty-two percent were female and 58% were ASA physical status classifications I or II. General anesthesia was used in 95% of all cases (Table 1).

Table 1
Table 1:
Screening Results for Patients with Vision Loss >30 Days, 1986–1998

The initial screening identified 405 patients who had new-onset vision loss (Table 1). Two hundred sixteen patients regained full vision within 30 days. In general, these patients were found to have nonspecific visual changes or small field deficits that resolved within 30 days after surgery. The most common symptom in these patients was postoperative blurriness that resulted in temporary visual loss or change in visual acuity. Only 12% of these 216 patients noted their visual symptoms within 24 h of the completion of their surgical procedures. Nearly 80% first noted their symptoms between 24 and 96 h after their procedures. The symptoms were sufficient to warrant evaluation by ophthalmologists in only 106 of these 216 patients (49%). Nearly 85% of these nonspecific vision changes or small field deficits resolved spontaneously within 7 days of the time that the patients first noted them. Eight patients had hemoglobin levels <8.0 g/dL and were transfused with red blood cells. Their symptoms improved, but it is unclear whether the improvement was a result of the transfusion. Eighteen patients were administered steroids and/or osmotic diuretics, but again, it was difficult to ascertain retrospectively if these treatments contributed to their eventual improvement in vision. In most instances, it was not possible to identify potential causes of these transient changes.

Of the 189 patients who developed visual deficits of longer than 30 days, 185 underwent ophthalmologic or neurologic procedures in which ocular or cerebral tissues were surgically damaged or resected. The remaining 4 patients (1 per 125,234 overall; 0.0008%) developed prolonged visual loss without direct surgical trauma to optic or cerebral tissues. None of the 9285 patients who underwent 11,942 spinal procedures while positioned prone (n = 10,886) or sitting (n = 1056) developed visual loss persisting for longer than 30 days, and only 8 of these patients were identified to have visual changes or small visual field deficits for 30 days or less.

Of the 216 patients who had transient vision loss of <30 days’ duration, we were able to evaluate the visual outcomes of 188, including all 8 who underwent spinal surgeries, at least 1 yr after their final procedures. None of these patients had vision loss recur or occur afresh. At least 27 patients died within 1 yr (14 with cancer, 3 with congestive heart failure, 3 with pneumonia, 1 with cerebral vasculitis, and 6 from unknown causes). We were unable to contact the remaining 11 patients.

Report of Four Cases

Case 1.

A 67-yr-old woman underwent elective tracheostomy, left parotidectomy, and bilateral neck dissections for metastatic squamous cell cancer with general anesthesia while positioned supine. Her preoperative medical history included noninsulin-dependent diabetes mellitus. The left neck dissection was complicated with a sudden loss of approximately 750 mL of blood. The left internal jugular vein was ligated, and the patient was transfused with red blood cells. Her lowest perioperative hemoglobin level was 9.3 g/dL. Her systolic blood pressure was <90 mm Hg for approximately 10 min during this hemorrhage. After an uneventful initial recovery period, on the second postoperative day, she noted the loss of all light perception in her right eye over a 1-h period. A dense afferent papillary defect was present. A funduscopic examination was normal. The vision in her left eye was unaffected. An initial ophthalmologic diagnosis of unilateral posterior ischemic optic neuropathy was made. Two months later, a repeat funduscopic examination found the right optic disk to be pale. Her vision deficit persisted until her death 4 mo later.

Case 2.

A previously healthy 52-yr-old man injured in an automobile accident required urgent thoracotomy for hemothorax. He underwent a general anesthetic while resting in a left lateral decubitus position. He bled excessively before and during surgery, receiving 33 U of red blood cells in the perioperative period. His lowest hemoglobin level during this hospitalization was 7.5 g/dL. During the procedure, he had 3 periods of 10 min or longer when the systolic blood pressure was <80 mm Hg. Upon awakening 6 h after surgery, he noted blindness in both eyes. An initial examination found his pupils to be unreactive and his fundi normal. He did not appear to have suffered head trauma during the accident; both imaging and neurologic examinations preoperatively were negative. He was diagnosed to have bilateral posterior ischemic neuropathy. A repeat funduscopic examination 4 mo later found both optic discs to be pale. His blindness has persisted at least 10 yr.

Case 3.

A 77-yr-old man with an extensive smoking history underwent elective tracheostomy, total laryngectomy, and right radical neck dissection for metastatic squamous cell cancer with general anesthesia while positioned supine. His intraoperative course was uneventful except for <5 min when the systolic blood pressure was approximately 85 mm Hg. His lowest perioperative hemoglobin level was 11.4 g/dL. On the second postoperative day, he developed atrial fibrillation, remaining normotensive. This abnormal rhythm resolved spontaneously in 1 h. On the third postoperative day, he complained of vision loss in his left eye. On examination, he had incomplete right homonymous hemianopsia. His visual acuity was 20/100 in each eye, with macular degeneration bilaterally. Magnetic resonance scanning showed a left occipital infarction, presumably from an embolic event. He was immediately anticoagulated, but there was no improvement in his vision. He died 3 mo after his procedure.

Case 4.

A 73-yr-old woman with controlled hypertension, mild congestive failure, and coronary artery disease underwent elective left total hip arthroplasty with general anesthesia while resting in a lateral decubitus position. Her intraoperative course was uneventful except for <5 min when the systolic blood pressure was approximately 95 mm Hg. Her lowest perioperative hemoglobin level was 9.2 g/dL. Six hours after her procedure, she complained of vision loss in her right eye. On examination, she had incomplete left homonymous hemianopsia. Her visual acuity was 20/25 in each eye. There was no evidence of embolic disease in the fundus. Computed tomography scanning showed a right occipital infarction. The cause, specifically considering hypoperfusion versus embolism, was unclear. Her vision loss persisted for at least 3 yr.


The most important finding of this study is the very small frequency of vision loss in this large surgical population. Only 1 per 118,783 patients (0.0008%) who underwent surgery with general anesthesia during this study period developed a permanent (>30 days) loss of some or all vision. None of the 26,212 patients who underwent surgery with central neuraxis regional anesthesia during this study period developed this complication. Given the recent attention associated with this issue, especially regarding patients undergoing surgical procedures on the spine, this small frequency is quite reassuring. However, four observations that may impact or further describe the relative value of our reported frequency must be considered.

First, this study did not include patients who underwent cardiac surgical procedures. Previous reports have suggested that 0.1% to 2% of these patients may have vision loss (13,14). Shaw et al. (14) specifically compared neurologic outcomes in cardiac surgical patients and those undergoing peripheral vascular surgery. They found that 8 of 312 cardiac surgical patients developed some vision loss compared with 0 of 50 peripheral vascular surgical patients. Nuttall et al. (6), from our institution, used criteria similar to ours to determine the frequency of vision loss in cardiac surgical patients from 1976 to 1994. They found that 0.06% of our cardiac surgical patients developed perioperative vision loss, a rate 75 times more than we found in noncardiac surgical patients. The reason for this very large difference in frequency of vision loss between cardiac and noncardiac surgical patients at our institution (0.06% vs 0.0008%) is unclear and remains to be elucidated. However, many reports have suggested that characteristics likely to be present at cardiac surgery may contribute. Patient characteristics (e.g., elderly age and the presence of generalized arteriosclerosis and other diseases that may affect the cerebral vasculature), procedural issues (e.g., pump perfusion and surgical disruption of particulate matter), and practice patterns (e.g., deliberate postoperative anemia and intraoperative hypotension) may play roles in the increased frequency of vision loss in cardiac surgical patients. A more detailed description of this issue is beyond the scope of this study. However, our deliberate exclusion of this patient population clearly impacted our reported frequency, and our results must be interpreted with consideration of this methodology.

Second, we found that none of our 9285 patients who underwent 11,942 spinal surgical procedures developed perioperative vision loss persisting for longer than 30 days. Much of the recent attention to perioperative vision loss has centered on patients undergoing spinal procedures, primarily while positioned prone. The majority of our spinal surgical patients were positioned prone during this study period, yet we did not find a single patient who lost vision for >30 days. Previous studies have suggested that intraoperative anemia, hypotension, prolonged surgical duration, or combination of these may be associated with an increased frequency of vision loss (4,15,16). Unfortunately, we cannot contrast our spinal surgery practice with those in these previous reports.

Third, our report is only as good as the validity of our recognition of perioperative vision loss. As noted in our methodology, we undertook a variety of processes to detect this complication and to provide validation of the reliability of our detection methods. We were fortunate to be able to use a well validated database, the MSIRS, as our primary data instrument. The ready availability of all related medical records within Mayo Clinic’s unit medical dossier per patient greatly facilitated the review. These were available for all 405 patients initially identified using the HICDA codes related to vision loss. We used a single, experienced nurse abstractor to review cases detected using the broad range of vision-related HICDA codes. Our final finding of only four cases of 405 initially potential cases suggests that our initial sweep of the database was sufficiently broad to capture cases of vision loss. Our methodology seems to have been sensitive but not specific, a characteristic of data sweeps that is desirable when searching for rare outcomes. Finally, we used a single study ophthalmologist in conjunction with all operating surgeons to determine whether damage or loss of ocular or cerebral tissues may have contributed to perioperative vision loss. For cases in which there was disagreement, an independent ophthalmologist was consulted. Although these processes hopefully contributed to the sensitivity and validation of finding all cases of perioperative vision loss in this study, we fully recognize that no process is likely perfect in a retrospective study format, and acknowledge that this frequency should be considered an approximation.

Fourth, although our overall frequency of 1 per 125,234 is very small, it is sufficiently frequent that, if generally applicable to noncardiac surgical populations, many anesthesia providers will see one or more cases in their careers. For example, we provide general anesthesia or central neuraxis regional anesthesia to approximately 60,000 noncardiac surgical patients annually in our institution. Therefore, the group of anesthesia providers at the Mayo Clinic might expect to have one patient with perioperative vision loss every other year. Many anesthesia providers in the United States provide anesthesia services to 30,000 or more patients in an average career (American Society of Anesthesiologists, personal correspondence, December 2000).

Our affected patients were typical, albeit not normal, patients. Two of the four patients who developed vision loss underwent extensive head and neck procedures. Several previous reports have suggested this type of surgery to be a risk factor for perioperative blindness (17–19). Each of our four patients had medical conditions ranging from diabetes mellitus to atrial fibrillation to recent trauma that may have increased the opportunity for phenomenon (e.g., hypoperfusion or embolic events) that previously have been associated in general populations with vision loss.

In summary, we found a very small incidence of perioperative vision loss persisting for longer than 30 days in patients undergoing noncardiac surgical procedures. Importantly, no patients in a large population undergoing spinal procedures while positioned prone developed prolonged vision loss.


1. Posner KL. Committee on professional liability forms a new registry to investigate postoperative blindness. ASA Newsl 1999; 63: 25.
2. Lee LA. Postoperative visual loss data gathered and analyzed. ASA Newsl 2000; 64: 25–7.
3. Roth SE, Thisted RA, Erickson JP, et al. Eye injuries after nonocular surgery. Anesthesiology 1996; 85: 1020–7.
4. Brown RH, Schauble JF, Miller NR. Anemia and hypotension as contributors to perioperative vision loss. Anesthesiology 1994; 80: 222–6.
5. Williams E, Hart W, Tempelhoff R. Postoperative ischemic optic neuropathy. Anesth Analg 1995; 80: 1018–29.
6. Nuttall GA, Dearani JA, Garrity JA, et al. Risk factors for ischemic optic neuropathy following cardiopulmonary bypass: a matched case/control study. Anesth Analg 2001; 93: 1410–6.
7. Kurland LT, Molgaard CA. The patient record in epidemiology. Sci Am 1981; 245: 54–63.
8. International classification of diseases, 9th revision, clinical modification (ICD-9-CM). Vol 2. Diseases. Ann Arbor: Commission on Professional and Hospital Activities, 1968.
9. Hospital international classification of diseases: adaptation 2 (HICDA-2). 2nd ed. Ann Arbor: Commission on Professional and Hospital Activities, 1968.
10. Warner MA, Shields SE, Chute CG. Major morbidity and mortality within 1 month of ambulatory surgery and anesthesia. JAMA 1993; 270: 1437–41.
11. Warner MA, Warner ME, Martin JT. Ulnar neuropathy: incidence, outcome, and risk factors in sedated or anesthetized patients. Anesthesiology 1994; 81: 1332–40.
12. Warner MA, Martin JT, Schroeder DR, et al. Lower-extremity motor neuropathy associated with surgery performed on patients in a lithotomy position. Anesthesiology 1994; 81: 6–12.
13. Sweeney PJ, Breuer AC, Selhorst JB, et al. Ischemic optic neuropathy: a complication of cardiopulmonary bypass surgery. Neurology 1982; 32: 560–2.
14. Shaw PJ, Bates D, Cartlidge NEF, et al. Neurologic and neuropsychological morbidity following major surgery: comparison of coronary artery bypass and peripheral vascular surgery. Stroke 1987; 18: 700–7.
15. Cheng MA, Sigurdson W, Tempelhoff R, Lauryssen C. Visual loss after spine surgery: a survey. Neurosurgery 2000; 46: 625–31.
16. Stevens WR, Glazer PA, Kelley SD, et al. Ophthalmic complications after spinal surgery. Spine 1997; 22: 1319–24.
17. Gotte K, Riedel F, Knorz MC, Hormann K. Delayed anterior ischemic optic neuropathy after neck dissection. Arch Otolaryngol Head Neck Surg 2000; 126: 220–3.
18. Marks SC, Jacques DA, Hirata RM, Saunders JR. Blindness following bilateral radical neck dissection. Head Neck 1990; 12: 342–5.
19. Chutkow JG, Sharbrough FW, Riley FC. Blindness following simultaneous bilateral neck dissection. Mayo Clin Proc 1973; 48: 713–7.
© 2001 International Anesthesia Research Society