Surgeon Perceptions and Reported Complications in Spine Surgery

Abstract

Study Design. 

Systematic review.

Objective. 

To define the term “complications” from the spine surgery literature and contrast this with definitions from other federal institutions, to summarize the incidence of adverse events in cervical, thoracic, and lumbar spine surgery, to include the factors that contribute to these events, and to determine the relationship between complications and patient centered outcomes.

Summary of Background Data. 

Efforts to understand and reduce complications in medicine, and spine surgery in particular have been hampered as a result of the lack of a meaningful and universally acceptable definition. The complex field of spine surgery has been a particularly challenging area for the development of a consensus to constructively describe these “undesirable/unanticipated developments arising during or out of the delivery of health care.” Furthermore, an overall understanding of expected complication rates after major spine surgery is lacking.

Methods. 

A systematic review of the English literature was undertaken for articles published between 1990 and December 2008. Electronic and federal databases and reference lists of key articles were searched to identify articles defining complications and reporting rates of spine surgical complications. Two independent reviewers assessed the level of evidence quality using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria and disagreements were resolved by consensus.

Results. 

The definitions for complications in the spine literature and federal agencies are inconsistent and at times conflicting. Mortality rates for cervical spine and lumbar spine surgery are <1%. For thoracic spine surgery, rates range from 0.3% to 7%. Complication rates range from 5% to 19%, 7% to 18%, and 4% to 14% after cervical, thoracic, and lumbar spine surgery, respectively. Findings from a single study indicate that major complications may have an impact on 1-year self-perceived general health. However, minor complications may not.

Conclusion. 

We define a complication as an unintended and undesirable diagnostic or therapeutic event that may impact the patient's care. Complications should be recorded and analyzed relative to disease severity, patient comorbidities, and ultimately their effect on patient outcomes. Further work needs to be done to develop a complication risk impact index that has the ability to help us assess and communicate the interaction of patient cormobidities and complication severity on patient centered outcomes.

In Brief

We define a complication as an unintended and undesirable diagnostic or therapeutic event that may impact the patient's care. Complications should be recorded and analyzed relative to disease severity, complication severity, and ultimately their effect on patient outcomes. Further work needs to be done to help us understand the ultimate effect on patient centered outcomes.

Despite significant accomplishments in medicine in the areas of research, technology, care delivery, and organizational systems implementations, patients will still experience complications as a result of well-intentioned efforts to deliver high-quality health care. Efforts to understand and reduce complications in medicine, and spine surgery in particular, have been hampered as a result of the lack of a meaningful and universally acceptable definition. After delivering a widely disseminated rallying cry toward enhancing “Patient Safety” in its seminal turn-of-the century article titled “To err is human,” the Institute of Medicine, unfortunately, did not use this opportunity to define complications beyond circumscribing them as “undesirable developments arising during or out of the delivery of health care.”¹ There remains an ongoing juxtaposition of terms such as “complication,” “adverse event,” “adverse occurrence,” and “sentinel event” to name but a few options, with an ongoing battle for the semantic high ground being carried by special interest groups who all profess to endorse the same goal to enhance patient safety.

The complex field of spine surgery has been a particularly challenging area for the development of a consensus to constructively describe these “undesirable/unanticipated developments arising during or out of the delivery of health care.” We sought to provide tangible answers to this ongoing conundrum for the specialty of spine surgery by answering the following 3 questions with this systematic review:

1. How are “complications” defined in spine surgery? How do these definitions compare with other definitions in the literature or those used by other institutions such as the Joint Commission (JC), Medicare, the National Institutes of Health (NIH), and the Food and Drug Administration (FDA)?
2. What is the overall incidence of complications in cervical, thoracic, and lumbar spine surgery? Which factors contribute to these events?
3. What is the impact of complications in spine surgery on patient centered outcomes?

Materials and Methods

Electronic Literature Database

The literature search is outlined in detail elsewhere.1a Briefly, a systematic search was conducted in MEDLINE, EMBASE, the Cochrane Collaboration Library, FDA, JC, and NIH websites for literature published from 1990 to December 2008. We limited our results to humans and to articles published in the English language. Reference lists of key articles were also systematically checked. For our first study question, we identified studies in the spine literature that focused on defining complications or adverse events. We also obtained the definitions of adverse events from the JC, Medicare, NIH, and FDA websites to compare and contrast these. To define an overall incidence of complications in spine surgery (study question #2), we identified large cohort studies (>1000 subjects) whose primary focus was to summarize complications associated with cervical, thoracic, and lumbar spine surgery. To determine the impact of spine surgery complications on patient centered outcomes (study question #3), we identified studies that reported the association between perioperative complications and some type of patient reported measure (e.g., health related quality of life or health status). Exclusions included studies specifically evaluating surgery for spinal cord injury or spinal cord tumors, studies that did not report complications, editorials, review articles without quantitative data, case reports, and non English written studies, Figure 1.

Figure 1: Inclusion and exclusion criteria.

Data Extraction

Each retrieved citation was reviewed by 2 independently working reviewers (D.C.N. and J.R.D.). Most articles were excluded on the basis of information provided by the title or abstract. Citations that appeared to be appropriate or those that could not be excluded unequivocally from the title and abstract were identified, and the corresponding full text reports were reviewed by the 2 reviewers. Any disagreement between them was resolved by consensus. Details of the literature search can be found in Figure 2. From the included articles, the following data were extracted: population, diagnosis, surgical treatment, complications, risk factors, and complication rates (including mortality and reoperation).

Figure 2: Flow chart showing results of literature search.

Study Quality

Level of evidence ratings were assigned to each article independently by 2 reviewers using criteria set by The Journal of Bone and Joint Surgery, American Volume (J Bone Joint Surg Am)² for prognostic studies and therapeutic studies and modified to delineate criteria associated with methodologic quality and described elsewhere, (Supplemental Digital Content, Tables 1–4, available at: https://links.lww.com/BRS/A415).

Analysis

Mortality, reoperation, and complication rates were based on the number of events divided by the number of surgeries. Data were summarized in tables, and qualitative analysis³ was performed considering the following 3 domains: quality of studies (level of evidence), quantity of studies (the number of published studies similar in patient population, condition treated, and outcome assessed), and consistency of results across studies (whether the results of the different studies lead to a similar conclusion).⁴ We judged whether the body of literature represented a minimum standard for each of the 3 domains using the following criteria: for study quality, at least 80% of the studies reported needed to be rated as a level of evidence I or II; for study quantity, at least 3 published studies were needed, which were adequately powered to answer the study question; and for study consistency, at least 70% of the studies had to have consistent results. The overall strength of the body of literature was expressed in terms of the impact that further research may have on the results. An overall strength of “HIGH” means that further research is unlikely to change our confidence in the estimate of effect. The overall strength of “MODERATE” is interpreted as further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. A grade of “LOW” means that further research is likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate, whereas “VERY LOW” means that any estimate of effect is very uncertain.1a

Results

Definition of Adverse Events From the Spine Literature

There is no consistent definition of a complication in the spine surgical literature. In a article published in an attempt to standardize the measurement of adverse events in spine surgery, Mirza et al⁵ preferred the term “adverse occurrence,” which was defined as any medical event in the course of a patient's treatment that had the potential for causing harm to the patient, Table 1. This term was chosen to avoid the connotation of blame, which may associate with the term “complication.” This group reserves the term “adverse event” for the subset of adverse occurrences where the patient experiences harm or requires additional monitoring or intervention. They developed an “Adverse Occurrence Severity Score” similar to the Index for Categorizing Medication Errors developed by the National Coordinating Council for Medication Error Reporting and Prevention.⁶ This is a 0 to 10 point scale ranging from “no effect, no risk” to “life threatening effect” and “death.” Some authors attempt to be more specific and define adverse events in spine surgery as “device-related,” “major,” or “preventable.” These judgments are not always obvious and can have a significant impact on the interpretation of treatment safety.⁵

Table 1: Definitions of Adverse Events and Complications Corresponding to the Source

Patel et al⁷ also discuss the challenges in assessing adverse events with the unclear definition of “complication” being a major issue. This group used a broad definition of perioperative complications to “limit the bias” of retrospective reviews by separately assessing both minor adverse events and significant complications. Rampersaud et al⁸ suggest that although the words “complication” and “adverse event” have significantly different meanings, they are often used interchangeably. They define an adverse event as any unexpected or undesirable events occurring as a result of spinal surgery, and a complication as a disease or disorder, which, as a consequence of a surgical procedure, will negatively affect the outcome of a patient. They hypothesized that most adverse events would not result in complications that would be normally flagged through traditional practice or audit procedures. They tested their hypothesis in a prospective observational study and reported an overall adverse event rate of 14% and an overall complication rate of 3.2%. Ciol et al⁹ and Martin et al¹⁰ summarized reoperation rates as important adverse events when evaluating the safety of lumbar spine surgery. This group relied on reoperation as an important marker of a poor outcome after surgery because it is objective despite being more a reflection of health care utilization rather than patient health status. Reoperations reflect adverse events that occur months after surgery. However, there remains no general agreement as to specific time periods for spinal reoperations, which would constitute a complication. Furthermore, there remains ambiguity as to anatomic limitations of spinal reoperations, for instance restricting the assessment of complications to same level procedures or expanding the definition to other levels in the same general anatomic region. Without doubt the definition of both dimensions, timeline and anatomic level, and impact data interpretation considerably. Despite the concerns expressed by Mirza and Rampersaud, the term “complication” is the most common event reported in the literature and typically reflects adverse events that occur intraoperatively or immediately after surgery.

Definition of Adverse Events From Government Agencies

The JC has developed an adverse event severity categorizing scheme as part of their Sentinel Event Reporting Policy.¹¹ By design, this scheme does not distinguish quality of care concerns from patient outcomes, or real effects from potential effects, requiring the institutions to define “sentinel event” specifically for their own purpose. A sentinel event is an unexpected occurrence involving death or serious physical or psychological injury, or the risk thereof. Serious injury specifically includes loss of limb or function. The phrase “or the risk thereof” includes any process variation for which a recurrence would carry a significant chance of a serious adverse outcome,¹¹Table 1.

Medicare defines an adverse event as an unintended harm, injury, or loss that is more likely associated with a patient's interaction with the health care delivery system than from an attendant disease process, Table 1. Adverse events are tracked by the Medicare patient safety monitoring system (MPSMS). This is a national surveillance program designed to identify the rates of select adverse events within the Medicare population. The approach uses explicit, rule-based chart review to count events. The intent of the system is to provide a tool to measure the safety of hospital care delivery systems. There are no measures for individual provider performance or provider error. The MPSMS identifies adverse events in discrete clinical topics only, and there is no attempt to identify adverse events beyond the list of MPSMS topics. Currently, the MPSMS is counting the national rates for Medicare beneficiaries in the following adverse event categories: central venous catheter (CVC) associated blood stream infections, CVC associated mechanical adverse events, CVC associated blood stream infection adverse drug events, inpatient acquired pressure ulcers and ventilator associated pneumonia, hip and knee replacement adverse events, as well as the postoperative rates of venous thromboembolic events, cardiac events, and pneumonia. The following adverse event measures were added in 2005: in-hospital patient falls, hospital acquired methicillin resistant staphylococcus aureus and vancomycin resistant enterococcus, adverse events associated with femoral artery puncture for catheter angiographic procedures, contrast nephropathy associated with catheter angiography, and catheter associated urinary tract infections replaced the postoperative urinary tract infection measure. There is no indication that spine surgery adverse events are currently part of this system.

In a response to a congressional request to streamline and reduce unnecessary Federal regulations that govern the conduct of extramural scientific research, the NIH published a report “NIH Initiative to Reduce Regulatory Burden” following extensive interviews and focus group meetings with the research community (http://grants.nih.gov/grants/policy/regulatoryburden/index.htm). Among the 5 major areas of focus, the report identified the reporting of adverse events to the IRB for multicenter clinical trials as burdensome and confusing. Some of the confusion stems from the different regulations governing the NIH and the FDA in this area. Federal regulations (45 CFR Part 46, Subpart A), shared by 17 Departments and Agencies as the Common Rule, require written procedures and policies for ensuring reporting of “unanticipated problems” involving risks to participants to the IRB, appropriate institutional officials, and the Department or Agency Head. Under a different set of regulations, 21 CFR 312, the FDA requires the sponsor to notify the FDA and participating investigators of any adverse event associated with the use of a test article that is “both serious and unexpected.” “Serious” is defined as an event that results in death, is life-threatening, requires inpatient hospitalization or prolongation of existing hospitalization, results in persistent or significant disability/incapacity, or necessitates surgical reintervention. “Unexpected” is defined as an event that is NOT listed in a study protocol, an Investigator's Brochure, or specific list established by an overarching agency such as the NIH, Table 1. The reporting of adverse events is in addition to, and does not supplant, periodic reports to the IRB at intervals appropriate to the degree of risk in the study.

The definitions and reporting requirements for adverse events differ between the NIH and the FDA. Generally, the NIH funding Institutes and Centers establish operational definitions of adverse events that apply to the particular trial, and these should be reviewed appropriately, Table 1. FDA regulations contain specific reporting requirements, timeframes for reporting, and agents responsible for reporting adverse events, unexpected adverse drug experiences, and unanticipated adverse device effects. For medical devices, according to 21 CFR 812.3, an unanticipated adverse device effect is any serious adverse effect on health or safety or any life-threatening problem or death caused by, or associated with, a device if that effect, problem, or death was not previously identified in nature, severity, or degree of incidence in the investigational plan or application, or any other unanticipated serious problem associated with a device that relates to the rights, safety, or welfare of subjects. After FDA clearance (i.e., postmarketing phase), one must report deaths and serious injuries that a device has or may have caused or contributed to. Cause or contributed means that a death or serious injury was or may have been attributed to a medical device or that a medical device may have been a factor.

What is the overall incidence of complications in cervical, thoracic, and lumbar spine surgery to include the factors that contribute to these events?

Complications Reported From Large Cervical Spine Studies

We accepted 7 published studies reporting complications after cervical spine surgery. All 7 studies were retrospective cohort studies with >1000 subjects, (Supplemental Digital Content, Table 5, available at: https://links.lww.com/BRS/A415). Five of these studies used data from the Nationwide Inpatient Sample, which is part of the Agency for Healthcare Research and Quality, one used data from the Cervical Spine Research Society registry, and one retrospectively collected medical record and radiographic data on a large cohort of subjects.

Five retrospective cohort studies by Boakye et al,¹² Cook et al,¹³ Wang et al,¹⁴ Patil et al,¹⁵ and Shamji et al¹⁶ used the National Inpatient Sample Database. Boakye and Cook evaluated 58,115 and 34,300 patients, respectively, undergoing fusion. Cook's population was limited to cervical spondylotic myelopathy, whereas Boakye included patients with spondylosis with and without myelopathy. Wang, Patil, and Shamji looked at a broader surgical treatment population, which included fusion and decompression with anterior, posterior, and combination approaches for patients with degenerative disc disease (DDD) or spondylosis. Fountas et al¹⁷ evaluated 1015 patients with radiculopathy and myelopathy because of DDD or spondylosis using retrospective medical record data from inpatient, outpatient, and radiographic reports. Patients were treated with anterior cervical discectomy and fusion. Zeidman et al¹⁸ evaluated 4589 accessed patients with a host of cervical spine conditions through the Cervical Spine Research Society registry.

The mortality rate across these 7 large studies ranged from 0.1%¹⁷ to 0.8%,¹⁸Table 2. The overall complication rate ranged from 5%^14,18 to 19.3%.¹⁷ Boakye reported a complication rate of 13.4% and 6.3% among patients with and without myelopathy with higher rates among patients aged 85 years and older, with 3 or more comorbidities, or posterior fusion (compared with anterior fusion). Wang et al¹⁴ reported an overall complication rate of 5% with an increased complication rate with increasing age. Patients aged 75 years and older had a 12% complication rate among this population. Cook et al¹³ reported an overall complication rate of 10% and specifically looked at patients with and without diabetes reporting complication rates of 15.2% and 9.5%, respectively.

Table 2: Summary of Complications Associated With Each Anatomic Area in Spine Surgery

Mortality rates for cervical spine surgery are <1%. Complication rates range from 5% to 19%, and preoperative nonsurgical risk factors include the following: myelopathy, older age, 3 or more comorbidities, and diabetes.

Complications Reported From Large Thoracic Spine Studies

We accepted 3 published articles reporting complications after thoracic spine surgery. Two were systematic reviews and 1 was a retrospective cohort study (Supplemental Digital Content, Table 6, available at: https://links.lww.com/BRS/A415).

Faciszewski et al¹⁹ used a spine center registry to evaluate 1152 adult patients who had surgery for the thoracic spine during a 23-year period. Diagnoses included scoliosis, DDD, trauma, spondylolisthesis, failed surgery, Scheuermann's kyphosis, spinal stenosis, and infection. Surgical procedures included anterior fusion, posterior fusion, or a combination. The mortality rate was 0.3% and an overall complication rate of 11.5%. Complication rates were greater in patients ≥60 years (relative risk = 2.0; P = 0.002), female versus male (relative risk = 1.3; P = 0.05), and ≥2 comorbidities (relative risk = 1.3; P = 0.04).

Taylor et al²⁰ and Verlaan et al²¹ published systematic reviews evaluating complications associated with the treatment of vertebral compression fractures and traumatic thoracic and lumbar fractures, respectively. They evaluated 3029 and 4304 patients undergoing balloon kyphoplasty/vertebroplasty and several anterior, posterior, or combination techniques, respectively. Taylor reported patients undergoing balloon kyphoplasty had a mortality rate of 7.4%, an overall cement leakage rate of 8%, and a new fracture rate of 20.3%. Patients experiencing cement leakage were asymptomatic. Patients undergoing vertebroplasty had a mortality rate of 6.4%, an overall cement leakage rate of 40% (3% symptomatic), and a new fracture rate of 10.2%. Verlaan reported higher complication rates among patients undergoing posterior long segment and anterior procedures.

Mortality rates for thoracic spine surgery range from 0.3% to 7%, Table 2. Complication rates range from 7% to 18% and preoperative nonsurgical risk factors include the following: older age, female gender, and 2 or more comorbidities.

Complications Reported From Large Lumbar Spine Studies

We accepted 11 published studies reporting adverse events after lumbar surgery. The 11 studies were retrospective cohort studies with >1000 subjects (Supplemental Digital Content, Table 7, available at: https://links.lww.com/BRS/A415). Two of these studies used data from the Medicare Inpatient Health Care Financing Administration, 4 from the Washington State Commission Hospital Abstract Reporting System, 1 from the National Hospital Discharge Survey, 1 from the Nationwide Inpatient Sample, 1 from the Canadian Institute for Health Information, 1 collected medical record and radiographic data on a large numbers of subjects undergoing lumbar surgery, and 1 was a systematic review of randomized trials and observational studies.

Two retrospective cohort studies by Ciol et al⁹ and Oldridge et al²² used the Medicare Inpatient Health Care Financing Administration database during 2 subsequent years (1985 and 1986, respectively) to evaluate different complications (reoperations and mortality, respectively). Ciol and Oldridge evaluated 27,111 and 34,418 patients, respectively. Four retrospective cohort studies by Deyo et al,²³ Taylor,²⁴ Martin,¹⁰ and Malter²⁵ used the Washington State Commission Hospital Abstract Reporting System database during different time periods (1986–1988, 1988–1991, 1990–1993, and 1988, respectively). Taylor and Martin evaluated 4176 and 24,882 patients, respectively, to establish reoperation rates among patients for a myriad of lumbar spine conditions. Malter evaluated 6376 patients to establish reoperation, mortality, and in-hospital complication rates. Deyo evaluated 18,122 patients to establish mortality and complication rates. Kardaun et al²⁶ used data from the National Hospital Discharge Survey during the period 1980 to 1985 to evaluate 3289 patients with disc herniation treated with discectomy. Li et al²⁷ evaluated the largest sample (N = 471,215) identified in this systematic review by evaluating patients who underwent laminectomy without fusion for spinal stenosis from the National Inpatient Sample Database (1993–2002). Wiese sought to determine the influence of the surgeon's experience on the intraoperative complication rate in lumbar microscopic disc surgery by performing a retrospective medical record review on 1872 patients. Patients were included if they exhibited motor paralysis or pain unresponsive to conservative management. Hu et al analyzed discharge data (N = 4722) from the Canadian Institute for Health Information, which included a broad range of diagnoses and lumbar spine surgeries in an effort to see whether complication rates differed by spine surgery. Turner was the single systematic review, which pooled mortality rates across ∼4000 patients being treated with lumbar fusion for a myriad of spine conditions.²⁸

The mortality rate across all studies reporting this event ranged from 0.07%²³ to 0.52%,²²Table 2. In-house mortality rates were higher in males versus females (0.72% and 0.37%, respectively), and 1-year mortality rates increased significantly at 80 years of age, with decompression and excision, male gender, and an increase in comorbidities.²² Li et al²⁷ also reported higher mortality rates in patients over the age of 85 and patients with 3 or more comorbidities.

The overall complication rates across studies ranged from 3.7%²⁶ to 12.8%.²⁴ Complication rates increased with age,²⁷ comorbidities,²⁷ obesity,²⁶ hypertension,²⁶ diabetes,²⁶ and surgeon's with less experience (50–100 surgeries = 10.7% vs. >500 surgeries = 2.2%).²⁹

The overall reoperation rates across all studies reporting this event ranged from 9.5%³⁰ to 19%.¹⁰ Reoperation rates were greater in patients undergoing fusion, in patients with previous back surgery, younger age, recent hospitalization, white race, and diagnosis of herniated disc (compared with other diagnoses).⁹ Taylor et al²⁴ reported higher reoperation rates among patients receiving worker's compensation (18%) compared with patients not receiving worker's compensation (10%) and higher in patients younger than 60 years of age.

Mortality rates for lumbar surgery are <1%. Complication rates range from 3.7% to 12.8%. However, the majority of studies report a complication rate <10%. Reoperation rates range from 10% to 19% and preoperative nonsurgical risk factors include the following: age, previous surgery, white race, obesity, hypertension, diabetes, comorbidities, worker's compensation, and less surgeon experience.

What Is the Impact of Complications on Patient Centered Outcomes?

We identified 1 study in the literature that was designed to evaluate the impact of complications on quality of life in adult deformity surgery by comparing several quality of life measures in those patients with and without postoperative complications,³¹Table 3. The purpose of the study by Glassman et al was to determine whether perioperative complications effect subsequent clinical outcomes measured by several patient centered outcomes: Scoliosis Research Society, Outcomes Instrument, Medical Outcomes Short Form-12 (SF-12), Oswestry Disability Index (ODI), and the Numerical Rating Scales for back and leg pain. The authors used case-control methods to create 3 matched groups using propensity scores based on 5 parameters known to be associated with 1-year clinical outcomes. The following 3 groups were formed from a prospectively accrued multicenter database for adult spine deformity: major complications (n = 46), minor complications (n = 46), and no complications (n = 46). Complications were classified as major or minor based on consensus agreement of the study group surgeons. Comparison between the 3 complication groups revealed that 1-year Scoliosis Research Society outcomes, SF-12 outcomes, ODI outcomes, and Numerical Rating Scales outcomes were not statistically significant. The rates of change between preoperative and postoperative scores were also not significant between groups, except for the SF-12 general health subscale. For the group with major complications, SF-12 general health deteriorated by 2.1 points. During the same period, the minor complication group experienced an improvement of 4.2 points, and the no complication group improved by 1.5 points. The authors concluded that the risk for minor complications may have less of an impact then previously assumed for surgical treatment in adult spinal deformity. In contrast, major complications were reported in ∼10% of cases and adversely effected outcome based on the deterioration of SF-12 general health scores 1-year after surgery. However, all other total scores and subscale scores changes did not demonstrate statistically significant differences between groups.

Table 3: Summary of Study Reporting the Association Between Complications and Change in Patient-Centered Outcome Scores

Unfortunately little work has been done in this area. Many studies collect both complications and quality of life measures but based on our review of the literature, no one in spine surgery except Glassman has attempted to quantify the impact of complications on quality of life. Our search explored both acute complications such as wound infections and more chronic complications such as postoperative DVT, radiculopathy, and reoperations.

Evidence Summary

The definitions for complications in the spine literature and federal agencies are inconsistent and at times conflicting. Complications and adverse events are often used interchangeably. The overall strength of the evidence to establish a definition for complications in spine surgery is “very low,” that is, any estimate of effect is very uncertain, Table 4. This topic should be further explored with an emphasis on standardization. The overall strength of the evidence to establish expected mortality and complication rates after cervical spine surgery is “low,” meaning that further research is likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. The overall strength of the evidence to establish expected mortality and complication rates after thoracic spine surgery is “moderate,” that is, further research is likely to have important impact on our confidence in the estimate of effect and may change the estimate. Mortality rates are less consistent than complication rates. The overall strength of the evidence to establish expected mortality, complication, and reoperation rates after lumbar spine surgery is “high,” meaning that further research is unlikely to change our confidence in the estimate of effect. Given the lack of research, the overall strength of the evidence to establish the impact of postoperative complications on patient centered outcomes after spine surgery is “very low.” A complication impact scale cannot be derived and should be the focus of future prospectively conducted research using large patient cohorts.

Table 4: Rating of Overall Strength of Evidence for Each Key Question

Discussion

The purposes of this systematic review were to define the term “complications” from the spine surgery literature and contrast this with definitions from other institutions such as the JC, Medicare, NIH, and FDA and to summarize the incidence of adverse events in cervical, thoracic, and lumbar spine surgery to include the factors that contribute to these events. We also sought to determine whether complications were associated with patient centered outcomes.

Clearly, the spine literature has its own definitions of “complications.” The term “complication” is typically used with an emphasis on events that occur intraoperatively or immediately after surgery. However, others focus on reoperation as the primary event of importance weeks to months after surgery. Several authors agree that we lack an adequate and standardized definition for adverse events or complications.^5,7,8 Mirza et al⁵ propose a more careful use of these definitions to avoid “blame” when reporting adverse events. Furthermore, they have developed an adverse occurrence severity score to better measure the severity of the adverse events. Rampersaud et al differentiate between adverse events, which are common occurrences after surgery and complications, which develop from adverse events and cause harm to the patient. None of these definitions agree with the governmental agencies. A system of definitions that take into account all these sources should be considered to promote more consistent standardization in the literature for improved safety assessment of surgical interventions.

Mortality rates among patients undergoing a myriad of cervical and lumbar spine surgeries are <1%. Though death events are rare in the cervical and lumbar spine, they are more common after thoracic spine surgery with rates as high as 6.4% among vertebroplasty patients and 7.4% among balloon kyphoplasty patients. Complication rates in the spine range from 4% to 19% for the cervical, thoracic, and lumbar spine. Rates are typically higher in older patients, patients with multiple comorbidities, diabetes, obesity, and hypertension. Rates are also higher when performed by surgeons with less experience or in patients who are receiving worker's compensation. Reoperations were more commonly reported in the lumbar spine and were associated with patients undergoing fusion, previous back surgery, younger age, recent hospitalizations, white race, diagnosis of herniated disc, and receiving worker's compensation. The types of complications are vast and depend on the surgical procedure, the definition, and the method of identification. Fortunately, most complications are not life threatening.

An important question to consider that has tremendous implications to care providers, patients, and insurers is how these complications ultimately affect patient centered outcomes. We identified 1 study in this systematic review that addressed this question and concluded that major complications may affect general health. However, minor complications may not have the impact that many assume.

This systematic review has limitations. First, we did not include all published studies on surgical procedures in the spine. This was beyond the scope of the article. Furthermore, it was not our intent to compare complication by treatment (i.e., efficacy assessment) but rather to establish a broad overview of the safety and possible risks of spine surgery. To accomplish this, we limited our selection of studies to large cohort studies or review articles that pooled complication rates. These studies were retrospective in nature and generally relied on established administrative databases. In general, administrative databases contain data that have been gathered as a by-product of some other process; the data may be collected and entered by hundreds of individuals at many locations; usually, there are few, if any, quality checks on the data; records may have different lengths and structures within the same database; and missing data are common.^32,33 One of the most obvious disadvantages is that these systems were not created for research purposes and, in most cases, researchers did not have input into the design or types of information collected by the systems. They may lack some of the details that researchers might want.³⁴ These characteristics of large databases lead to the controversy over their use in epidemiologic and health services research and point to the need to consider validity and reliability issues.^35,36 Numerical coding systems (e.g., ICD-9-CM) are often used and these have their own peculiarities. There may not be a code for a particular hospital diagnosis and the level of detail in coding may vary or change over the course of time. For the purposes of establishing safety among spine surgery, studies that obtained their data in this fashion were our best alternative. The studies selected for this systematic review were observational studies. Under normal circumstances, when summarizing data for a systematic review, this level of evidence may not be acceptable. This is particularly important when establishing efficacy of a treatment where randomized controlled trials are essential. However, randomized trials may be an insufficient study design to adequately measure patient safety and report adverse events.^37,38 These experimental studies cannot contribute as much to our understanding of the adverse effects associated with therapies.³⁸ Hunter reports the following reasons why randomized controlled trials may not be the best method for assessing adverse events associated with treatment: (1) trials powered for efficacy may be too small to detect adverse events, (2) monitoring of adverse events may not be sensitive or specific for the actual events caused, (3) duration of trials may be too short for detection of events requiring longer exposure, (4) stopping rules in clinical trials may further shorten the duration of exposure after randomization, (5) enrollment criteria may exclude susceptible subgroups, (6) for industry sponsored trials, head-to-head comparison of adverse events caused by drugs (or devices) from different manufacturers may not be available, (7) follow-up studies to detect adverse events that involve the denial of an efficacious treatment to patients may be deemed unethical—patients may not wish to enroll in such a study, and (8) funding to conduct trials solely to quantify adverse events may be difficult to obtain. Hence, our use of observational studies with large study populations was the most appropriate method for establishing a broad overview of adverse events in spine surgery.

In this age of emerging concepts of performance-based medicine and an attempts to measure virtually every conceivable aspect of health care, there is obvious uncertainty regarding the operational definition of the term “complication” in spine surgery. This uncertainty raises important concerns and questions about interpretation of outcomes studies. Although maneuvering in a complex legal environment with questions of culpability and blame looming large over health care providers, a more frank and proactive approach towards identifying and reporting “complications” is in the best interest of all participants focused on patient safety and achieving meaningful progress in health care. In concert with a more systematic recording of “complications” emanates the need to increase our sophistication in interpreting these events.

A standardized definition for complications needs to highlight both the complication severity and the patient's profile of comorbidities that enhance the likelihood of a complication and ultimately the effect on patient centered outcomes. For providing physicians a tool in treatment planning and providing prognostic information to patients and health care administrators, a three-dimensional tool that considers the following is recommended for future development: (1) disease severity and patient comorbidities, (2) complication severity, and (3) the patient centered Quality of Life impact of these factors. The dimensional composite of this would represent an individualized, patient centered, and complication risk impact index (CRII). CRII would highlight the spectrum of disease states and individualized risks of a complications and their effect on patient centered quality of life. For example, when comparing 2 patients who have undergone a similar surgery, Patient A has a hypothetical index of 1.0 and a complication severity of 2.0 whereas Patient B has a hypothetical comorbidity index of 8.0 and a complication severity of 2.0 (both on 0–10 point scales). Twelve months after surgery, Patient A has an ODI score of 95 points whereas Patient B has a score of 30 points, Figure 3. In another example, 2 patients have the same complication severity (6.0). However, Patient A is healthy and has a comorbidity index of 1.0 whereas Patient B has a comorbidity index of 8.0. Patient A's ODI score is 70 points and Patient B's ODI score is 10 points 12 months after surgery, Figure 4. We do not currently have data to produce such a model. However, this is the direction we need to be moving in future spine research and clinical application.

Figure 3: Hypothetical ODI scores (points) for 2 patients based on their hypothetical baseline comorbidity index and severity of complication.

Figure 4: Hypothetical ODI scores (points) for 2 patients based on their hypothetical baseline comorbidity index and severity of complication.

In preoperative discussions with patients and families, it is customary to discuss complications with representation as to the percentage of patient's that may incur a specific complication or outcome. The challenge to this discussion lies in the appreciation that a 20% risk of failure to relieve pain can be interpreted in quite divergent a myriad of ways. Even more concrete, for instance, a 1% risk of paralysis associated with a routine procedure may represent a dreadfully risky proposition to the treating spine surgeon, which in other statistical systems such as cards would be seem to present favorable odds. Because of media sensationalism derived perceptions, a patient's family may, however, associate a disclosure of a 1% risk of paralysis associated with a given procedure as an estimate of “botched” surgeries. How would a CRII based discussion help our patients? A hypothetical example of this three-dimensional system is presented in Figure 5. In this hypothetical example, a patient undergoing a routine spinal fusion procedure without major comorbidities or other risk factors may face a realistic 2.1% chance of perioperative complications based on a hypothetical dichotomous outcomes calculation. In contrast, another patient with concurrent moderate obesity and poorly controlled diabetes mellitus may expect a 45% cumulative risk of perioperative complications. Finally, a patient with 3 major concurrent comorbidities, such as morbid obesity, uncontrolled diabetes mellitus, and chronic anticoagulation who presents with emerging neurologic deterioration may have a 95% chance for a complication. Analogous to the increasing risk of a complication, chances for an unfavorable outcome rise disproportionally. With this recognition of the patient's preexisting comorbidities, a shared informed discussion can proceed. However, the concept of a shared decision-making process ideally encompasses a multifactorial consideration of comorbidities and invasiveness of the procedure. The discussion of potential benefits versus relative risks can be individualized because they have significant impact on morbidity and mortality. The balancing of perceived benefits of a suggested procedure and its relative risks relative to the individual patient and their specific circumstances remains a complicated undertaking with rather vague guidance readily available to date. With the future CRII tool, this can be the interrelation of patient-related factors and the severity of a complication, which would be plotted according to defined parameters, and can be plotted out to display the eventual patient reported outcome illustrated in addition to the discussed probabilities of their outcome. Helping the family to understand that the likelihood of complications is intrinsic to the patient's condition and comorbidities is essential to a collaborative decision-making process. A factually based discussion of correlation of the type and severity of a patient's spinal condition with anticipated effect of patient comorbidities and the potential impact of a potential complication based on its severity can provide the underpinnings of a transparent shared-decision-making process.

Figure 5: Complication Impact Scale (CRII) for predicting treatment failure after spinal fusion based on hypothetical data.

In a trauma example, if a 63-year-old man, diabetic, smoker, with COPD presents to the emergency room following a motor vehicle accident with a thoracic fracture dislocation, incomplete paraplegia, and flail chest it can help the Care Provider Team to communicate proactively the probability of specific outcomes and complications. To our payers and hospital administrators, this fracture patient has substantial risk for several “never” events such as hospital acquired pneumonia, catheter sepsis, wound infection sepsis, and pressure sores. The CRII would help identify the risk of a poor outcome that maybe in turn could help identify revealing modifiable risk factors through further more in-depth analysis with the goal of ultimately positively affecting patient centered outcomes such as quality of life. How would this work? In the polytrauma patient with long bone injuries, acute stabilization has been shown to reduce pulmonary complications. Assessment of the impact of a similar protocol for thoracic fractures would allow for individual patient based risk stratification, and the CRII could estimate the probability of a patient centered outcome such as quality of life.

Generation of the first dimension of the CRII tool will require standardization of the rating of severity of “complications.” Some have classified these as simply minor or major,³¹ whereas others have devised more detailed yet complicated systems (0 to 10 point ratings).^5,6 These systems are still missing 2 additional dimensions. The occurrence of a “complication” and its associated severity should ideally be considered in light of the individual patients disease severity formulated on factors that include demographic characteristics, disease condition, cognitive and psychosocial status, general health including comorbidities and individualized habits such as smoking and alcohol use. The interaction of complication severity and disease severity should then assist in predicting the third dimension, the probability of a failed outcome, the impact of a potential complication on a patient centered outcome, Figure 5. Developing such a model will take additional research; however, such a model is the logical next step in this evolutionary process to study the impact of complications on patient outcomes. Surprisingly, studies correlating patient centered outcomes with complication severity have not been performed with 1 notable exception.³¹ This topic clearly deserves further detailed research but such a system could be invaluable. Despite the best of available medical care, complications will continue to occur in spine surgery, and it behooves us to better understand their impact on future patient outcomes.

Clinical Recommendations

Although many terms are used to describe unintended/undesirable or adverse events that occur during the delivery of health care, we recommend use of the comprehensive term “complication” in an attempt to simplify terminology and focus on patient outcomes rather than issues of blame. We define a complication as an unintended and undesirable diagnostic or therapeutic event that may impact the patient's care. Complications should be recorded and analyzed relative to disease severity, complication severity, and ultimately their effect on patient centered outcomes. We propose a three-dimensional assessment of “complications,” which takes into account patient comorbidity/disease severity and complication severity to determine the probability of a poor patient centered outcome. Further work needs to be done to develop this measure and on integrating these factors as they relate to patient centered outcomes while acknowledging the presence of other potentially significant factors such as health care utilization, socioeconomic impact, medical legal considerations, and operative indications.

Key Points

• The definitions for complications in the spine literature and federal agencies are inconsistent and at times conflicting. Complications and adverse events are often used interchangeably. We recommend the term complication and define it accordingly.
• We propose the future development of a simple and easy to apply three-dimensional complication scale that takes into account complication severity, disease severity, and their impact on patient outcomes (CRII). This CRII scale should be used to help understand long-term patient centered outcomes.
• Mortality rates for cervical and lumbar spine surgery are <1%. Ranges are higher after thoracic surgery. Complication rates for all levels range from 4% to 19%.
• Major complications may have an impact on 1-year self-perceived general health. However, minor complications may not. This important correlation needs further study.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.spinejournal.com).

Acknowledgment

The authors are indebted to Ms. Nancy Holmes, RN, for her administrative assistance.