Postoperative visual loss (POVL) is a low-incidence, but potentially devastating perioperative complication. Studying such a complication with randomized controlled trials is hindered by enormous economic, logistical, and ethical barriers. Case reports, case series (literature reviews), and databases of adverse events are typically used to describe common features of rare, yet serious adverse events to guide future research. Despite frequent speculation, these forms of investigation cannot determine risk factors, because they are limited by the lack of a comparison control group. Rather, they provide a profile of patients who develop these complications.
The methods used to assess risk factors for low-incidence events primarily consist of retrospective analyses of large databases from either single or multiple institutions, or from national databases. Cases with the adverse event are compared with similar cases without the adverse event. Institutional databases have the advantage over national databases of allowing the investigators to obtain detailed information from any portion of the medical record and verify the accuracy of data. However, institutional databases are limited by relatively small numbers and may lack sufficient power to detect significant differences between the affected and unaffected cases. Furthermore, they may be biased by regional variations in both practice and prevalence of disease, as has been noted for POVL.1–3 Therein lies the strength of national databases, i.e., both strength in numbers and sampling.
In this month’s issue of Anesthesia & Analgesia, Shen et al.4 are the second group to use the Nationwide Inpatient Sample (NIS) to determine the prevalence of POVL and its associated risk factors. They examined different time periods and comparison groups than did Patil et al.5 The NIS collects data from more than 1000 randomly selected, nonfederal hospitals in the United States. It is the largest all-payer inpatient database in the United States, and currently comprises approximately 90% of all US hospital discharges. The NIS is one of many databases that are under the Healthcare Cost and Utilization Project sponsored by the Agency for Healthcare Research and Quality. These databases combine the data collection efforts of state data organizations, hospital associations, private data organizations, and the federal government to create a centralized national information resource on patient care. The NIS was created in 1988 and contains more than 100 elements per hospital stay including primary and secondary diagnoses and procedures, patient demographics, admission and discharge status, expected payment source, total charges, length of stay, and hospital characteristics. It is most commonly used to examine trends in health care costs, access, practice variation, quality of care, and impact of health policy changes.*
Both Shen et al.4 and Patil et al.5 used the NIS to examine associations with POVL and acknowledge many of the significant limitations of this database including incomplete data entry (e.g., missing age and gender in more than 20% of patients), lack of any intraoperative data from the anesthesia record, inability to verify POVL diagnosis or other data, and inability to determine whether a condition was preexisting (e.g., anemia or POVL). An example of the limitations of the database can be found in the article by Patil et al.,5 in which they noted an odds ratio (OR) of 10.1 for the association of hypotension with ischemic optic neuropathy (ION) after spinal fusion surgery. Although it is entirely possible that there is an association, the process by which the NIS data are collected and entered into the database merits scrutiny. Typically, hospital employees, rarely involved in direct patient care, scour the hospital and discharge records. They search for specific codes that may upgrade a patient’s acuity status, for the purpose of improving reimbursement and adjusting outcomes for acuity of care. Thus, entries in the medical record can result in ICD-9 diagnoses without verification. Hypothetically, a trainee or other provider from a consultant service could note that ION was caused by intraoperative hypotension. This diagnosis could be arbitrarily made without consideration of the baseline arterial blood pressure or other associated factors during the operation. More importantly, the medical records of all other spinal fusion patients who did not develop ION will not be scrutinized for hypotension because there was no adverse outcome. Nonetheless, the ICD-9 diagnosis code of hypotension may get entered into the national database for the ION patient, but not for patients without ION whose perioperative arterial blood pressure was never evaluated. The search for associated factors in the medical record is much more intense when there is a complication. Consequently, using this database for associations with ICD-9 codes that are not routinely documented for every patient will lead to questionable results.
Shen et al.4 were very selective and cautious in their use of the NIS database. They examined associated factors that are commonly documented on all patients, regardless of complications. They confirmed previous smaller studies by demonstrating that the prevalence of POVL is highest for cardiac surgery followed by spine surgery. They also demonstrated for the first time that patients undergoing hip and femur surgery and knee arthroplasty had an increased risk of POVL compared with patients undergoing abdominal surgery. Not only does this provide guidance for risk disclosure on these subgroups of patients, but it may provide further insight into the etiology of POVL. These 3 subgroups of surgical procedures share many characteristics: a high proportion of middle-aged to elderly patient population with coexisting diseases; large blood loss procedures with associated anemia, blood transfusion, and high fluid resuscitation; high embolic loads with associated cytokine release and potential for vascular occlusion; and a higher frequency of intentional or unintentional hemodynamic alterations or derangements. Additional physiologic perturbations may result from the prone position for spine surgery and cardiopulmonary bypass for cardiac surgery.
Shen et al.4 noted a decrease in prevalence of POVL between 1996 and 2005. This finding is unique but could be a result of exclusion of high-risk categories of spine surgery such as laminectomy with fusion and/or inclusion of very low-risk categories of spine surgery such as anterior cervical discectomy and fusion. Although the 2 NIS studies on POVL examined different time periods with different inclusion and exclusion criteria, it is striking that Patil et al.5 identified more than 4.7 million patients between 1993 and 2002 who underwent inpatient spinal procedures with 4134 POVL cases, whereas Shen et al.4 had only 0.47 million patients in their spinal fusion subgroup with 140 POVL cases. The most relevant group in which to track POVL cases over time would be the posterior spinal fusion cases, because these are the highest risk category.
Perhaps the most interesting finding from the study by Shen et al.4 was the alarmingly high risk of pediatric patients developing POVL after all surgical procedures (OR 6.9); POVL after spinal fusion surgery (OR 18.3); and cortical blindness for all procedures (OR 64). Similar results were noted in the study by Patil et al. Despite these very high ORs from a national database, there is little published information in the orthopedic, anesthesia, or pediatric literature regarding the association of POVL and pediatric patients. Furthermore, there were only 2 submissions to the American Society of Anesthesiologists (ASA) POVL Registry of pediatric patients 18 yr or younger with perioperative visual loss after spine surgery.6 However, 43% of all POVL cases in the spinal fusion surgery subgroup of Shen et al.4 in the NIS database from 1996 to 2005 were younger than 18 yr of age.
Although this finding is alarming, it may be explained by the fact that the vast majority of pediatric spinal fusion procedures are prolonged multilevel fusions performed for scoliosis. The ASA POVL Registry found that 94% of spinal surgery cases with ION were associated with an anesthetic duration of 6 or more hours and 82% were associated with 1 or more liters of estimated blood loss.6 Therefore, surgical duration and blood loss may easily be confounding associations, data that the NIS database does not collect. Herein lies an opportunity for newly formed data collection organizations such as Wake Up Safe (an official Patient Safety Organization and a component of the Society for Pediatric Anesthesia) and the American Society of Anesthesiologists’ Quality Institute to collect intraoperative data in such a manner that more reliable associations with outcomes can be assessed. For example, numerous ischemic complications are attributed to hypotension in the perioperative period. If we fail to collect at least minimal information on hemodynamic management for all patients, then we will be left with data no better than the NIS and will not be able to effectively evaluate the relationship between hypotension and these ischemic events. The evolution of the electronic medical record should facilitate this type of data collection. We have an opportunity to be leaders in patient safety yet again by collection of high quality data on intraoperative anesthetic management and outcomes on all patients.
Until we can obtain better data in sufficient number, the ASA practice advisory on perioperative visual loss associated with spine surgery can provide some guidance for management of high-risk spine surgery cases.7 Practitioners should carefully weigh the risks and benefits of using anesthetic techniques that may cause extreme physiologic derangements in those procedures that are considered high risk for POVL.
1. Stevens WR, Glazer PA, Kelley SD, Lietman TM, Bradford DS. Ophthalmic complications after spinal surgery. Spine 1997;22:1319–24
2. Warner ME, Warner MA, Garrity JA, MacKenzie RA, Warner DO. The frequency of perioperative vision loss. Anesth Analg 2001;93:1417–21
3. Roth S, Thisted RA, Erickson JP, Black S, Schreider BD. Eye injuries after nonocular surgery. A study of 60,965 anesthetics from 1988 to 1992. Anesthesiology 1996;85:1020–7
4. Shen Y, Drum M, Roth S. The incidence of perioperative visual loss in the United States: a ten year study from 1996 to 2005 of spinal, orthopedic, cardiac, and general surgery. Anesth Analg 2009;109:1534–45
5. Patil CG, Lad EM, Lad SP, Ho C, Boakye M. Visual loss after spine surgery: a population—based study. Spine 2008;33:1491–6
6. Lee LA, Roth S, Posner KL, Cheney FW, Caplan RA, Newman NJ, Domino KB. The American Society of Anesthesiologists’ Postoperative Visual Loss Registry: analysis of 93 spine surgery cases with postoperative visual loss. Anesthesiology 2006;105:652–9
7. American Society of Anesthesiologists Task Force on Perioperative Blindness. Practice advisory for perioperative visual loss associated with spine surgery: a report by the American Society of Anesthesiologists Task Force on Perioperative Blindness. Anesthesiology 2006;104:1319–28
*Overview of the Nationwide Inpatient Sample: http://www.hcup-us.ahrq.gov/nisoverview.jsp, accessed June 4, 2009.