Percentage of complications according to preventability.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Teixeira PGR, Inaba K, Salim A, et al. Preventable Morbidity at a Mature Trauma Center. Arch Surg. 2009;144(6):536–541. doi:10.1001/archsurg.2009.82
To analyze the preventable and potentially preventable complications occurring at a mature level I trauma center.
Academic level I trauma center.
The study included 35 311 trauma registry patients.
Main Outcome Measures
The cause, effect on outcome, preventability (preventable, potentially preventable, or nonpreventable), and loop closure recommendations for all preventable and potentially preventable complications, and clinical data related to each complication retrieved from the trauma registry and individual medical records.
Over the 8-year study, 35 311 trauma registry patients experienced 2560 complications. Three hundred fifty-one patients (0.99% of all patients) had 403 preventable or potentially preventable complications. The most common preventable or potentially preventable complications were unintended extubation (63 patients [17% of complications]), surgical technical failures (61 patients [15% of complications]), missed injuries (58 patients [14% of complications]), and intravascular catheter–related complications (48 patients [12% of complications]). These complications were clinically relevant; 258 (64% of complications) resulted in a change in management, including 61 laparotomies, 52 reintubations, 41 chest tube insertions, and 19 vascular interventions.
The incidence of preventable or potentially preventable complications at an academic level I trauma center is low. These complications often require a change in management and cluster in 4 major categories (ie, unintended extubation, surgical technical failures, missed injuries, and intravascular catheter–related complications) that must be recognized as critical areas for quality improvement initiatives.
Regionalized trauma systems and trauma center designation decrease mortality after severe injury.1,2 Recent studies suggest that the overall risk of death is significantly lower when care is provided at a level I trauma center3 and that a 15% reduction in mortality could be attributed to the implementation of trauma systems.4 A key factor contributing to these improved outcomes is implementation of a system for continuous quality assessment.
Continuous efforts to identify and better understand specific factors that contribute to unfavorable outcomes are critical for the improvement of trauma care.5 In particular, the identification of adverse events that are preventable has become an important performance assessment tool that has been used to evaluate trauma protocols and trauma systems1,6-10 and as a marker of the quality of trauma care provided.11,12
As mortality in a well-established trauma center plateaus, the next step in trauma care improvement requires a thorough understanding of the preventable complications that can be corrected. The overall incidence of complications in the trauma population has been reported13 and has remained stable over time.14 Preventable complications, however, have not been extensively analyzed in this population.
The objective of the present study was to review the rate and causes of preventable or potentially preventable complications occurring in the care of injured patients at a mature, high-volume, academic level I trauma center to identify potential areas for improvement. Our hypothesis is that preventable complications are centered on specific causes that can be identified and used for improvement.
The Los Angeles County–University of Southern California Medical Center is a high-volume academic level I trauma center with 24-hour in-house coverage by a trauma surgeon. All complications are reported daily by the house staff and are independently verified by a trauma registry nurse. All complications are comprehensively reviewed at a mandatory weekly morbidity and mortality conference attended by all attending surgeons and fellows in the Division of Trauma and Critical Care as well as those services such as neurosurgery, orthopedics, anesthesia, and radiology that may have been directly involved in each case. Each complication is reviewed and is classified by consensus according to preventability. All decisions as to whether the complication was preventable, potentially preventable, or nonpreventable are made by the attending surgeons present as a panel review decision at the morbidity and mortality conference. An attending trauma surgeon documents each complication, its root causes, outcomes, and loop closure recommendations.
All morbidity and mortality reports from January 1, 1998, through December 31, 2005 (8 years), were reviewed, and all complications classified as preventable or potentially preventable were identified. Complications that resulted in death during this period have been thoroughly analyzed and published previously and are excluded from this study.15
Each complication was considered a single event and was entered into a computerized database (Excel 2003; Microsoft Corp, Redmond, Washington) and tabulated into 6 functional categories (airway; perioperative; delay in diagnosis or treatment; catheters, drains, and tubes; operative complications; and other complications). Clinical data including patient demographic data, injury history, course during hospitalization, circumstances surrounding the complication, management, and outcome of the complication were retrieved from the trauma registry and from individual medical record review.
Continuous variables are presented as mean (SD) and categorical variables as percentages. Percentages were compared using χ2 for trend.
Over the 8-year study, 35 311 patients with trauma were admitted to the Los Angeles County–University of Southern California Medical Center. During this time, 2560 complications occurred. Of these, 2157 complications (84%) were nonpreventable, 214 (8%) were potentially preventable, and 189 (7%) were preventable.
Beginning in 1998, the overall rate of complications decreased and reached a plateau. The percentage of preventable and potentially preventable complications compared with nonpreventable complications remained constant throughout the 8-year study (P value for trend = .44) (Figure). On average, the preventable and potentially preventable complications represented 16% (2%) of the annual number of complications.
During the study period, 351 patients (1% of total admissions) experienced 403 complications that were classified as preventable or potentially preventable. Thirty-nine patients (11%) had multiple complications. The demographic data for this population are given in Table 1. The preventable and potentially preventable complications occurred in 6 main categories: airway (24%); perioperative (23%); delay in diagnosis or treatment (20%); catheters, drains, and tubes (17%); operative (9%); and other (7%). The preventable and potentially preventable complications were clinically relevant; 258 (64% of complications) resulted in a change in management including 61 laparotomies, 52 reintubations, 41 chest tube insertions, and 19 vascular interventions.
Unintended extubation (17%) was the most frequent preventable or potentially preventable complication. Reintubation was required in 49% of cases. One episode was further complicated by bradycardia and cardiac arrest before reintubation; however, the patient survived to hospital discharge.
Most unintended extubations were self-extubations (93%). Two patients self-extubated during computed tomography; the remainder were in patient care areas. All were due to inadequate sedation or restraint usage. The remainder of the unintended extubations occurred during bathing (n = 2), transfer from stretcher to computed tomographic scanner (n = 2), entrapment in a rotating bed (n = 1), and bronchoscopy (n = 1). All patients were successfully reintubated with the exception of the patient with unintended extubation during bronchoscopy, who required a surgical airway.
Seventeen instances of failed extubation were preventable. Nine episodes of witnessed preventable or potentially preventable aspiration with respiratory compromise occurred. Five episodes were attributed to technical problems during intubation: 3 in the initial resuscitation phase, 1 in the computed tomographic scanner, and 1 during an episode of respiratory distress on the ward. In 3 patients, failure to obtain a swallowing assessment after prolonged intubation was identified as the cause of aspiration.
Overall, perioperative complications were the second most common type of preventable or potentially preventable complication and encompass a wide spectrum. Technical failures represented 15% (n = 62) of the total preventable and potentially preventable complications. Of the technical failures, 29% (n = 18) were due to a hollow viscus leak, with 7 resulting in enterocutaneous fistulas. Urinary leaks occurred in 3 patients with ureteral injury. Fascial dehiscence represented 13% (8) of complications. Technical failures were related to vascular procedures in 19% (n = 12) of complications, including 5 graft occlusions. All graft occlusions required revision, and 2 ultimately resulted in limb amputation.
Another common cause of technical failure was residual pneumothorax attributed to poor technique after chest tube removal, occurring in 7 patients. In all but 1 of these patients, reinsertion of a chest tube was required to evacuate the pneumothorax.
Postoperative bleeding occurred in 19 patients, with 17 requiring a second laparotomy. The source of bleeding was the mesentery (n = 4), short gastric vessels (n = 2), abdominal wall (n = 2), kidney hilum (n = 2), spleen (n = 1), liver (n = 1), gastric suture line (n = 1), gastrostomy site (n = 1), intercostal spaces (n = 1), thyroid gland (n = 1), and unknown source (n = 3).
Surgical site infection occurred in 13 patients (11 incisional infections and 2 intra-abdominal infections). These were considered preventable or potentially preventable because of inadequate antibiotic coverage or wound management. All 5 deep surgical site infections and 1 intra-abdominal abscess resulted in fascial dehiscence. In the 2 intra-abdominal abscesses, the decision to repair a destructive colon injury rather than performing a resection was considered the potential cause of the leak.
Delayed or missed diagnosis, the third most frequent preventable or potentially preventable complication, occurred in 58 patients (14%). A detailed description of all delayed or missed diagnoses is given in Table 2.
Seventeen percent of preventable complications were related to catheters, drains, or tubes; most were associated with intravascular catheters (Table 3). The most frequent complication in this group was pneumothorax secondary to central line placement, accounting for 56% (n = 28) of all catheter-related complications. All pneumothoraces required chest tube insertion. In 1 patient, in addition to chest drainage, intubation and mechanical ventilation were required because of respiratory distress.
Inadvertent removal of tubes and catheters was also common and included gastrostomy tubes (n = 5), chest tubes (n = 4), Jackson-Pratt drains (n = 3), and nephrostomy tubes, pigtail catheters, and pulmonary artery catheters (n = 1 each). In 3 of the percutaneous gastrostomy tube removals, laparotomy was required. Two of the Jackson-Pratt drain removals resulted in formation of a peripancreatic fluid collection that required percutaneous drainage. In 1 patient, breakage of the drain during removal resulted in a retained intra-abdominal drain that required laparotomy for retrieval.
The most frequent operative complication was iatrogenic injury (Table 4). The spleen was the most commonly injured organ, and 8 patients required splenectomy.
Injury to the left ventricle occurred during a resuscitative thoracotomy in 1 patient. This patient had previously undergone a thoracotomy because of trauma without pericardium closure. As a result, the heart was adherent to the chest wall, and the left ventricle was injured during the resuscitative thoracotomy. This was repaired, and the patient survived.
In 3 patients, complications related to access during an interventional radiology procedure occurred. Two patients with aortic rupture underwent endovascular stenting. At the completion of the procedure, during withdrawal of the stent deployment device, the common femoral artery in 1 patient and the external iliac artery in the other patient were avulsed, requiring surgical open bypass. The third patient with pelvic fractures underwent angiography; however, no active bleeding was identified and no embolization was performed. Thrombosis of the access site caused limb ischemia that was managed with surgical thrombectomy after unsuccessful angiographic thrombectomy. During the angiographic thrombectomy, the brachial artery access site also thrombosed, requiring surgical open bypass.
In 12 patients, errors related to drug or fluid administration were the cause of potentially preventable complications (Table 5). Pressure ulcers occurred in 11 patients. In 4 of these patients, the ulcer was related to prolonged cervical collar usage in comatose patients, and the remainder were in the sacral or occipital region.
At the beginning of the 20th century, Ernest Codman, MD, a true advocate of patient safety, challenged the Boston medical community with his “end results” theory, which consisted of keeping a systematic record of diagnostic and treatment errors occurring during the care of patients and identifying the effect of these errors on outcomes.16 As 21st- century physicians are challenged by the increasing complexities of patient care, Dr Codman's visionary theory has become even more relevant. Surgical outcomes assessment and patient safety have become a critical issue, as outlined by Polk17 in his 2005 presidential address to the American Surgical Association emphasizing the importance of quality, safety, and transparency in minimizing surgical error.
Institutional morbidity and mortality conferences remain a critical component of quality improvement in trauma care. Peer review of complications and deaths in these local forums enables identification and thorough evaluation of specific adverse events. The analysis of preventable deaths has been critical in the improvement of trauma care provision1 and demonstrates that even at mature trauma centers, despite these improvements, preventable deaths continue to occur.15,18
Complications also remain an important measure of the quality of trauma care. This study demonstrated that at a mature, high-volume, academic level I trauma center, preventable and potentially preventable complications continue to occur at a low but constant rate. A wide variety of preventable and potentially preventable complications were found. Most complications, however, were clustered in 4 main areas: unintended extubation, surgical technical failures, missed injuries, and intravascular catheter–related complications. The preventable and potentially preventable complications that occurred during the study were clinically relevant; 258 (64% of complications) resulted in a change in management including 61 laparotomies, 52 reintubations, 41 chest tube insertions, and 19 vascular interventions. These findings also suggest that clinically relevant preventable complications occur even when established protocols are in place. For example, even with a rigorous evidence-based sedation protocol in place, unintended extubation can occur. For a protocol to be effective, the patient must be recognized as needing the protocol, the protocol must be initiated and followed up, and the patient must be clinically monitored. This emphasizes the importance of the human interface in the successful application of any protocol. It also underlines that protocols need to be constantly updated and reinforced as new evidence becomes available. The review, updating, and reinforcement of management protocols for problem areas such as sedation and extubation, the tertiary trauma examination, and intravascular catheter insertion and care may have a role in reducing the rate of preventable complications.
This study was performed at an academic teaching center with residents and clinical fellows actively participating in the care of trauma patients. With our study design, we were unable to capture the level of training at which the errors occurred. The effect of surgeons in training on the incidence of preventable and potentially preventable complications remains to be investigated in future studies.
The present study has several limitations inherent in its retrospective design. The classification of the complications is subjective. The panel review system and the classification by consensus help to decrease bias; however, this remains a possibility in the assignment of errors. Capture of the complications relied entirely on the clinical team and nursing staff. Although unlikely with the redundant system, it is possible that some complications were missed. In particular, minor complications may be unreported. The stringent policy of complication identification and reporting by 2 independent groups (clinical team and trauma registry nurses) adopted at this institution, however, makes it unlikely that major complications remain unaccounted for. Although every attempt was made to capture all complications, after discharge, if a patient sought care at another center because of a complication, this would also underestimate the number of complications documented.
In conclusion, despite 24-hour coverage by attending trauma surgeons and an aggressive morbidity and mortality identification and review process with a loop closure program, preventable complications continue to occur at a mature trauma center. This is a clinically relevant problem, frequently resulting in changes in management. These complications tend to cluster in 4 major groups that must be recognized as critical areas for quality improvement initiatives in the care of trauma patients. Ongoing identification and analysis of complications even at a mature trauma center are mandatory.
Correspondence: Kenji Inaba, MD, MS, Division of Trauma Surgery and Surgical Critical Care, University of Southern California, USC+LAC Medical Center, 1200 N State St, Room 10-750, Los Angeles, CA 90033 (email@example.com).
Accepted for Publication: May 8, 2008.
Author Contributions: Dr Teixeira had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Teixeira, Inaba, Rhee, Brown, DuBose, and Demetriades. Acquisition of data: Teixeira, Browder, and Nomoto. Analysis and interpretation of data: Teixeira, Inaba, Salim, and DuBose. Drafting of the manuscript: Teixeira, Inaba, and DuBose. Critical revision of the manuscript for important intellectual content: Teixeira, Salim, Rhee, Brown, Browder, DuBose, Nomoto, and Demetriades. Statistical analysis: Teixeira. Administrative, technical, and material support: Browder, DuBose, and Nomoto. Study supervision: Inaba, Salim, Rhee, Brown, DuBose, and Demetriades.
Financial Disclosure: None reported.
Previous Presentation: This study was presented at the 66th Annual Meeting of the American Association for the Surgery of Trauma; September 27, 2007; Las Vegas, Nevada.