- Email: [email protected]
Risk Tolerance and Bile Duct Injury: Surgeon Characteristics, Risk-Taking Preference, and Common Bile Duct Injuries
Risk Tolerance and Bile Duct Injury: Surgeon Characteristics, Risk-Taking Preference, and Common Bile Duct Injuries Nader N Massarweh, MD, Allison Devlin, MS, Rebecca Gaston Symons, MPH, Jo Ann Broeckel Elrod, PhD, David R Flum, MD, FACS, MPH Little is known about surgeon characteristics associated with common bile duct injury (CBDI) during laparoscopic cholecystectomy (LC). Risk-taking preferences can influence physician behavior and practice. We evaluated self-reported differences in characteristics and risk-taking preference among surgeons with and without a reported history of CBDI. STUDY DESIGN: A mailed survey was sent to 4,100 general surgeons randomly selected from the mailing list of the American College of Surgeons. Surveys with a valid exclusion (retired, no LC experience) were considered responsive, but were excluded from data analysis. RESULTS: Forty-four percent responded (1,412 surveys analyzed), 37.7% reported being the primary surgeon when a CBDI occurred, and 12.9% had more than one injury. Surgeons reporting an injury were slightly older (52.8 ⫾ 9.0 years versus 51.3 ⫾ 9.8 years; p ⬍ 0.004) and in practice longer (20.8 ⫾ 9.7 years versus 18.9 ⫾ 10.5 years; p ⬍ 0.001). Surgeons not reporting a CBDI were more likely trained in LC during residency (63.3% versus 55.4% injuring) as compared with surgeons reporting a CBDI, who were more likely trained at an LC course (29.8% versus 38.2%). Surgeons in academic practice or who work with residents had lower reported rates of CBDI (7.9% versus 14.5% [academics]; 18.7% versus 25.0% [residents]). Mean risk score was 12.4 ⫾ 4.4 (range 6 to 30 [30 ⫽ highest]) with a similar average between those who did (12.2 ⫾ 4.5) and did not (11.9 ⫾ 4.4) report a CBDI (p ⬍ 0.23). Compared with surgeons in the lowest three deciles of risk score, relative risk for CBDI among surgeons in the upper three deciles was 17% greater (p ⫽ 0.07). CONCLUSIONS: More years performing LC and certain practice characteristics were associated with an increased rate of CBDI. The impact of extremes of risk-taking preference on surgical decision making can be an important part of decreasing adverse events during LC and should be evaluated. (J Am Coll Surg 2009;209:17–24. © 2009 by the American College of Surgeons) BACKGROUND:
Cholecystectomy, when performed laparoscopically, is often an ambulatory procedure with a low rate of major complications. The most substantial risk remains a common bile duct injury (CBDI), which occurs in 1 of every 200 to 400 patients.1,2 Half of the surgeons who perform laparoscopic cholecystectomy (LC) are estimated to cause such an injury during their career, making litigation related to CBDI among the leading sources of medical malpractice claims against surgeons.3-5 Repair of these injuries is often
complex and can require multiple interventions and, not infrequently, biliary reconstruction.3,6,7 Major CBDI also has a considerable impact on quality of life, functional status, and survival.8-11 Among Medicare beneficiaries young and old in the 1990s, a CBDI during cholecystectomy was associated with a nearly threefold increase in the risk of death during followup compared with uninjured patients.11 Routine use of intraoperative cholangiography (IOC) during LC for the purposes of preventing major CBDI has demonstrated benefit, especially early in a surgeon’s career.2 Most of the supportive evidence is observational and the majority of US surgeons do not use IOC routinely.1,12 Reasons for not using an IOC include a perception that it is ineffective, a lack of available ancillary radiology staff (to help perform the procedure in the operating room), and a number of other possibilities.13 Most likely the actual rea-
Disclosure Information: Nothing to disclose. Received October 30, 2008; Revised February 25, 2009; Accepted February 25, 2009. From the Departments of Surgery, Health Services, and the Surgical Outcomes Research Center, University of Washington, Seattle, WA. Correspondence address: Nader N Massarweh, MD, Department of Surgery, University of Washington, 1959 NE Pacific St, Box 356410, Seattle, WA 98195-6410. email: [email protected]
© 2009 by the American College of Surgeons Published by Elsevier Inc.
17
ISSN 1072-7515/09/$36.00 doi:10.1016/j.jamcollsurg.2009.02.063
18
Massarweh et al
Risk Tolerance and Bile Duct Injury
Abbreviations and Acronyms
CBDI IOC JPI LC
⫽ ⫽ ⫽ ⫽
common bile duct injury intraoperative cholangiogram Jackson Personality Index laparoscopic cholecystectomy
son for not using IOC is multifactorial. The reason major CBDIs occur is also likely multifactorial and can include not using an IOC; misinterpretation of the surgical anatomy because an IOC was not done or misinterpreted; or poor decision making, such as not converting to an open procedure when questions about the anatomy arise. Way and colleagues14 proposed heuristic processes (mental rules of thumb the mind constructs by making use of uncertain probabilistic information to help identify the common bile duct) and a willingness to disregard disconfirmatory information, such as a concerning IOC, as contributing factors to CBDI. Improving quality of delivered care is a focus of many health care reform initiatives. Certain aspects of quality, such as the process and structure of care, are studied with relative ease. There are other factors that are more difficult to quantify. For example, a topic not traditionally studied is the influence of an individual’s personality on everyday clinical decision making and practice—more specifically, the role of risk-taking behavior and risk preference. The risk-taking construct became an established part of overall personality assessment in the 1970s.15 In the field of personality psychology, the Jackson Personality Index (JPI) is a well-established assessment tool, with risk-taking as 1 of 15 personality subscales.16 Pearson and colleagues17 adapted the JPI risk-taking assessment scale to evaluate the influence of physician “risk attitudes” on clinical practice. Although never studied among surgeons, our hypothesis was variability in risk-taking preference can bear on clinical practice. In conducting this survey of practicing surgeons, we specifically hoped to identify discrete demographic, practice, or risk-taking characteristics associated with a selfreported history of causing a CBDI.
METHODS All procedures used to conduct this study were approved by the University of Washington Institutional Review Board. A random sample of 5,000 American College of Surgeons members was selected from the roster of “general surgeons” to receive this mailed survey. The survey content and the mailing methodology have been described in an earlier study.13 Although the JPI risk-taking subscale in its entirety is composed of 20 questions, pilot data from the Pearson
J Am Coll Surg
Table 1. Questions Assessing Surgeon Risk Preference I enjoy taking risks. I try to avoid situations with uncertain outcomes. Taking risks does not bother me if the gains involved are high. I consider security an important element in every aspect of my life. People have told me that I seem to enjoy taking chances. I rarely take risks when there is another alternative. Responses to all questions were scored on a 5-point Likert scale ranging from “very much like me” to “not at all like me.” Scores were adjusted for missing responses.
study using this risk-taking subscale revealed the complete assessment was too lengthy and, as such, was shortened to six questions.17 We chose to use these same six questions (Table 1). Risk-taking preference was scored on a 5-point ordinal scale based on surgeon responses. For three of the questions (questions 1, 3, and 5), the highest point value was assigned to the most affirmative responses (“Not at all like me” ⫽ 1 point; “very much like me” ⫽ 5 points). For the other three questions, the scoring scheme was reversed. Scores were calculated in the same manner described by Pearson and colleagues17 (multiplying by a correction factor for any missing responses that assumes the average of available responses applies to missing responses: risk-taking score ⫽ sum of response ⫻ [6/6, number of missing responses]) with stratification into risk categories based on raw score: high risk scored ⬎1 SD more than mean; moderate risk scored between 1 SD more than and 1 SD less than the mean; low risk scored ⬎1 SD less than mean. Descriptive statistics were calculated for the cohort of respondents. Responses from self-identified CBDI injuring surgeon were compared with noninjuring surgeons. Categorical variables were compared using chi-square statistics. Calculated p values refer to the significance of the differences across strata (between injuring and noninjuring surgeons’ responses) and are not adjusted for multiple comparisons. A “missing” category was assigned to nonresponses, as most questions received ⬍100% response. This missing category was excluded from statistical comparisons. Mean values of the continuous variables (age, years since residency, and risk score) were compared using Student’s t-test and are reported with standard deviations. Statistical analyses were performed using STATA version 10.1 (Stata Corp).
RESULTS A total of 4,100 surveys were mailed. One hundred eight were not post-deliverable and were discounted when calculating response rate. Among surgeons who returned an incomplete survey (n ⫽ 303), the most commonly noted reason was retirement (63%), followed by cholecystectomy not being a part of the practice (16.6%), no reason offered
Vol. 209, No. 1, July 2009
Massarweh et al
Risk Tolerance and Bile Duct Injury
19
Table 2. Responding Surgeon Demographic Information Characteristics
% Male (%) Age (y), mean ⫾ SD Years since residency, mean ⫾ SD Training included IOC (%) Practice setting (%) Private practice Academic staff Private/public hospital HMO Federal Other Missing Procedures with residents (%) Yes No Missing Point of care (%) Community hospital Tertiary referral center Outpatient surgery center Other Missing Cholecystectomy training (%) Course Residency/fellowship Partner No formal training Other Missing Cholecystectomies (%) 1–10 11–20 21–50 51–100 ⬎100 Missing†
All (n ⴝ 1,412)
Noninjuring (n ⴝ 878)
Injuring (n ⴝ 534)
p Value*
89.2 51.8 ⫾ 9.6 19.6 ⫾ 10.3 79.3
62.0 88.0 51.3 ⫾ 9.8 18.9 ⫾ 10.5 79.2
37.8 91.2 52.8 ⫾ 9.0 20.8 ⫾ 9.7 79.4
0.15 0.004 ⬍0.001 0.99
55.5 12.0 10.6 1.4 0.9 4.8 14.8
52.5 14.5 10.3 1.4 0.7 5.4 15.3
60.3 7.9 11.2 1.5 1.3 3.8 14.0
22.6 58.7 18.6
25.0 55.9 19.1
18.7 63.5 17.8
61.4 17.6 2.4 3.8 14.9
59.2 19.4 2.2 3.9 15.4
65.0 14.6 2.8 3.6 14.0
33.0 60.3 2.5 1.9 0.6 1.7
29.8 63.3 3.0 1.6 0.5 1.8
38.2 55.4 1.7 2.4 0.8 1.5
4.8 8.3 28.2 36.0 22.1 0.6
6.8 10.3 28.6 33.5 20.2 0.7
1.5 5.1 27.5 40.1 25.3 0.6
0.003
0.01
0.13
0.01
⬍0.001
*Across strata. † Missing category corresponds to the percentage of surgeons not answering the question. IOC, intraoperative cholangiogram.
(13%), and deceased (7.3%). Total response rate to the survey was 1,756 of 3,992 (44%) and included incomplete surveys returned with a valid reason. After discounting incomplete surveys and duplicate responses, the total number of surveys analyzed was 1,412. A total of 534 survey respondents (37.8%) reported having caused at least one CBDI. General demographic and practice data for the survey respondents are presented in Table 2. Surgeons reporting a CBDI were slightly older
(mean age 52.8 years versus 51.3 years; p ⫽ 0.004) and in practice longer (20.8 versus 18.9 years; p ⬍ 0.001). Surgeons in academic practice (7.9% versus 14.5%) or who worked with residents regularly (18.7% versus 25.0%) reported a CBDI less frequently. Surgeons not reporting a CBDI were more likely trained in LC during residency (63.3% noninjuring versus 55.4% injuring) as compared with surgeons reporting a CBDI, who were more likely trained at an LC course (29.8% noninjuring versus 38.2%
20
Massarweh et al
Risk Tolerance and Bile Duct Injury
Table 3. Characteristics of Common Bile Duct Injuries Caused by Injuring Surgeons (n ⫽ 534) Reported ⬎1 CBDI Yes No Missing Biliary reconstruction Yes No Missing Patient death as a result Yes No Missing Suspected cause of CBDI Misidentification of cystic duct Unknown Anatomic reason Traction injury Electrocautery injury More than one reason Other Missing*
n
%
180 333 21
33.7 62.4 3.9
397 119 18
74.3 22.3 3.4
8 509 17
1.5 95.3 3.2
307 56 42 31 17 16 53 12
57.5 10.5 7.9 5.8 3.2 3.0 9.9 2.2
*Missing category corresponds to the percentage of surgeons not answering the question. CBDI, common bile duct injury.
injuring). A smaller proportion of low-volume (⬍20 cholecystectomies/year) as compared with high-volume (⬎100 cholecystectomies/year) surgeons (6.6% lowvolume versus 25.3% high-volume) reported a CBDI. Table 3 summarizes the circumstances surrounding the reported CBDIs and the surgeon-perceived reason for injury. One-third of respondents who self-reported a CBDI stated they had caused more than one such injury (12.7% of the total cohort). A biliary reconstruction was performed for 74.3% of patients with an injured bile duct and 1.5% reported that the injured patients died because of the injury. In the majority of patients, the surgeon cited a problem with the anatomy (eg, aberrant anatomy, adhesions, inflammation, and so forth [7.9%]) or misidentification of the cystic duct (57.5%) as the main reason for CBDI. Injuring and noninjuring surgeons did not report any substantial differences in IOC use (Table 4). A greater proportion of injuring surgeons thought most CBDIs occur in the setting of unclear anatomy (58% versus 49%; p ⫽ 0.01). Within both groups of surgeons, a notable number misinterpreted an IOC image (Fig. 1) clearly demonstrating no filling of the proximal bile ducts, with 16% of injuring and 18% of noninjuring surgeons thinking a CBDI to be unlikely. Based on the information provided by the
J Am Coll Surg
image, similar proportions of both groups reported they were likely or very likely to clip and cut the structure (18.9% injuring versus 15.4% noninjuring) and convert to an open procedure (59.7% injuring versus 61.1% noninjuring). Mean risk-taking preference score for this cohort was 12.0 ⫾ 4.5 (12.2 ⫾ 4.5 injuring and 11.9 ⫾ 4.4 noninjuring; p ⫽ 0.23), with a range from the lowest possible score of 6 to the highest possible score of 30. Several surgeons did not complete all six questions (one question, n ⫽ 21; two questions, n ⫽ 2; three questions, n ⫽ 1; four questions, n ⫽ 1; five questions, n ⫽ 3), and 12 surgeons did not complete any of the questions. We did not find any statistically significant difference in the distribution of risktaking preference scores between injuring and noninjuring surgeons when stratified by 1 SD around the mean (Fig. 2A). When we examined the distribution based on deciles of data, we again did not find any significant differences (Fig. 2B), but we did note a greater proportion of injuring as compared with noninjuring surgeons in the upper three deciles. The absolute percentage difference between each of these three deciles was relatively small (⬃2%), but the relative risk of CBDI for surgeons within these three upper deciles was 17% greater (p ⫽ 0.07) than for those in the lowest three deciles (where the proportion of noninjuring surgeons is greater). Similarly, when we evaluated risk score based on how respondents answered questions pertaining to Figure 1, a greater proportion of surgeons with low risk scores were more likely to think the IOC demonstrated a CBDI (interpreting with caution; Fig. 3A) and were more likely to covert to open based on the cholangiogram results (proceeding with caution; Fig. 3B).
DISCUSSION CBDI during LC is a relatively uncommon complication.1,2 The impact on a patient’s quality of life and the subsequent economic and clinical consequences of such injuries are profound.8-11 Our group recently published results from a survey of US surgeons examining reported differences in practice between routine and selective cholangiographers and found considerable differences with regard to knowledge, use, and opinions about IOC.13 Here we report results of a unique aspect of the survey that describes self-reported differences in characteristics, practice patterns, and risk-taking preference associated with CBDI during LC. When LC was introduced, surgeons learned the procedure at short, 1- to 3-day courses and subsequently used these techniques without initial supervision or proctoring. Another published survey evaluating the effects of residency training on the rates of bile duct injury demon-
Vol. 209, No. 1, July 2009
Massarweh et al
Risk Tolerance and Bile Duct Injury
21
Table 4. Surgeon Cholangiography Use and Characteristics Based on Common Bile Duct Injury Status Type of cholangiographer Routine Selective Missing† Cholecystectomies with IOC (%) 0–10 11–25 26–50 51–75 ⬎75 Missing† Type of cholangiogram Fluoroscope Single-shot x-ray Missing† Time to perform IOC (min) 1–10 11–20 21–30 31–40 Missing†
All (%) (n ⴝ 1,412)
Noninjuring (%) (n ⴝ 878)
Injuring (%) (n ⴝ 534)
p Value*
26.9 72.8 0.3
27.8 71.9 0.3
25.5 74.3 0.2
0.54
48.2 17.1 4.6 4.8 25.0 0.3
46.4 17.5 5.0 4.2 26.5 0.3
51.1 16.5 3.9 5.8 22.5 0.2
79.0 19.8 1.3
79.4 19.3 1.4
78.3 20.6 1.1
0.77
45.6 42.6 9.8 1.4 0.6
44.7 42.7 10.1 1.9 0.6
47.2 42.3 9.2 0.7 0.8
0.25
0.23
*Across strata. † Missing category corresponds to the percentage of surgeons not answering the question. IOC, intraoperative cholangiogram.
strated graduates from more contemporary time periods (between 1992 and 1998 as compared with 1980 to 1990) were less likely to experience laparoscopic learning curve injuries, resulting in a decreased number of CBDIs.3 Although it appears formalized, LC training as a part of general surgery residency programs has contributed to these results,18 major CBDIs (complete transection) still occur and can potentially be preventable. CBDI can also be a manifestation of more years practicing LC. This could account for a greater proportion of those reporting a CBDI being older, longer in practice (had less opportunity for residency training where laparoscopy was an important component), and receiving LC training at a course. Interestingly, academic surgeons and those who reported performing the majority of their procedures with residents were less likely to report a CBDI, but also more frequently reported being low-volume surgeons (32.2% low-volume versus 7.1% high-volume [academics]; 23.8% low-volume versus 12.5% high-volume [work with residents]). This might simply relate to volume, but might also be because surgeons working with residents benefit from having a second set of “critical eyes” in the operating room. It remains to be determined if the dynamics of teaching, with emphasis on reviewing anatomy during dissection, are also involved in these differences.
One point of concern was the relatively high proportion of surgeons from both groups who failed to correctly identify an obvious, impending CBDI presented through a clinical vignette (Fig. 1). When respondents were presented an image of a cholangiogram demonstrating no proximal filling of the hepatic ducts, which would suggest a likely misidentification of the CBD as the cystic duct, 16% and 18% of injuring and noninjuring surgeons, respectively, thought an injury was unlikely. Although studies sufficiently powered to detect a difference in rates of CBDI when an IOC is used consistently demonstrate a benefit,12 implicit within the argument for IOC use is the assumption an IOC is properly performed and properly interpreted. If surgeons are misinterpreting IOCs, educational initiatives aimed at improving basic radiologic interpretation could be useful. Some surgeons can correctly perform an IOC, but disregard the results and rely instead on visual cues, resulting in a misidentification injury.14 The study by Way and colleagues14 describes how heuristics can lure the unsuspecting surgeon into making anatomic errors, so he or she does not believe they are cutting the common bile duct. Greater use of IOC can help decrease slavish reliance on heuristics and might be key to avoiding CBDIs based on faulty cognitive interpretation of familiar visual cues. Tolerance of anatomic uncertainty during cys-
22
Massarweh et al
Risk Tolerance and Bile Duct Injury
J Am Coll Surg
Figure 1. (A) The clinical vignette, (B) intraoperative cholangiogram image demonstrating no proximal filling of the hepatic ducts, and (C) subsequent questions as presented in the survey.
tic duct dissection could account for the relative underuse of IOC before clipping and cutting ductal structures. Differences in cognitive processing of data, divergent thresholds for accepting potential anatomic variants, or willingness to take risks in the setting of disconfirmatory information need to be addressed in more creative ways. We know little about surgeon-level practice variation based on differences in personality and risk assessment. Several studies have attempted to evaluate the role of emotional intelligence as it pertains to the core competencies of residency training.19-21 To our knowledge as yet no study has applied such measures to assess impact on clinical practice. The impact of risk-taking preference on physician practice has been assessed in several studies.17,22-25 Pearson and colleagues17 (using the same methodology here) demonstrated higher risk-taking preference scores among emergency physicians who were less likely to admit patients with chest pain at low or medium risk for myocardial infarction. Similarly, Nightingale and colleagues22-25 used a different risk-assessment method and showed risk-taking preference does come to bear on decisions pertaining to laboratory test selection, admission rates through the emergency department, and practice preference in critical care scenarios. To our knowledge, no study has ever performed risk-taking
Figure 2. Common bile duct injuring and noninjuring surgeon risktaking preference based on (A) low-, moderate-, and high-risk preference scores and (B) risk-preference deciles. Differences between injuring and noninjuring surgeons within either the upper or lower three deciles were not statistically significant.
preference assessment on surgeons. We used one of these previously used methods, asking six questions taken from the JPI risk-taking subscale to assess the influence of risktaking preference on the rate of CBDI.17 Stratifying surgeons into risk groups based on standard deviations around the mean, we did not find any substantial differences based on low-, moderate-, and high-risk categories. Ideally, we would have carried the stratification out to 2 SDs to identify outliers at the extremes of risktaking preference. Unfortunately, because the data were right-skewed, use of standard deviations might not be the best way to represent extremes of personality. Additionally, because this form of personality testing is an emerging area of study, there is little guidance to suggest break points that might be appropriate for stratifying groups. We were concerned the risk intervals were too wide to capture subtle differences or trends in risk-taking preference and so we stratified scores into deciles. Although there was no statis-
Vol. 209, No. 1, July 2009
Figure 3. Based on Figure 1C. Surgeon risk-taking preference based on (A) interpretation of the image and (B) likelihood of converting to an open procedure based on the image. For (A), surgeons were categorized as cautious if they were “likely” or “very likely” to think Figure 1 represented a potential common bile duct injury and were not cautious if they answered “unlikely.” For (B), surgeons were categorized as cautious if they answered “likely” or “very likely” to convert to open based on Figure 1 and were not cautious if they answered “unlikely.”
tically significant difference, a greater proportion of injuring as compared with noninjuring surgeons clustered in the upper three deciles. Because of a lack of data at the high end of the risk scale (because of the distribution of the data), we did not believe expanding our stratification more would yield meaningful results. This six-question risk-assessment method has not been validated for lack of a gold standard and has never been used among surgeons for comparative evaluation. Our analysis of this component of the survey was exploratory, and we found it interesting that within this decile-based stratification, a greater proportion of those with a reported history of a CBDI were at the high end of the risk-taking spectrum. The absolute differences in reported rates of CBDI between groups in the upper (26.2% injuring versus
Massarweh et al
Risk Tolerance and Bile Duct Injury
23
20.6% noninjuring; p ⫽ 0.57) and lower (42.4% injuring versus 45.0% noninjuring; p ⫽ 0.57) three deciles were small, but this small difference actually translated to a 17% increase in the relative risk of CBDI. Findings that suggest a group of surgeons might be at increased risk for a potentially preventable injury merits additional exploration. One consideration is surgeons with higher risk-taking preference are more likely to ignore disconfirmatory evidence in an abnormal IOC or to rely on heuristics to influence or guide surgical decision making. A greater proportion of surgeons interpreting Figure 1 with caution (thinking it “likely” or “very likely” represents a CBDI about to happen) fell within the low-risk category (Fig. 3A). Alternatively, differences in risk-taking preference might manifest as poor clinical decisions during operation. Respondents reporting they would proceed with caution based on Figure 1 (“likely” or “very likely” to convert to open) also fell within the low-risk category (Fig. 3B). A larger cohort might have allowed for better evaluation of these differences and easier identification of outliers at the extremes. Studies targeting nontraditional end points, such as surgeon cognition, data assimilation, perception, and risk assessment would be useful to explore this relationship more. There were several notable limitations to this study. Our response rate fell below the epidemiologic standard of 60%, subjecting our study to nonresponse bias. Our study might also have suffered from normative responses (in which surgeons select response based on what they think is the expected answer as opposed to responses reflecting actual practice or belief ), although this survey was anonymous. Another important limitation is missing data. Not all questions received a 100% response rate, possibly limiting the generalizability of our findings. Assessing between-group differences was the central goal of this study; the overall effect of missing data should be small. Finally, although the JPI is an established personality assessment tool when used as a whole, the use of this six-question “mini” JPI to assess risk-taking preference is uncertain and has yet to be validated. Our results identified surgeons with markers of increased clinical activity and certain practice characteristics as reporting higher rates of CBDI; but the individual contribution of these variables could not be reliably assessed. Although we did not find any statistically significant association between average risk-taking preference and the incidence of CBDI, our analysis of extremes of risk-taking preference (often used in evaluations of behavior and personality) did demonstrate a higher injury rate among those with higher risk-taking preference scores. This finding could inform future studies specifically aiming to explore the relationship between surgeon personality or cognitive processes and clinical practice or outcomes. Other high-
24
Massarweh et al
Risk Tolerance and Bile Duct Injury
risk industries, such as aviation, emphasize standard training and consistency in the application of proved safety interventions that decrease the risk of even rare, catastrophic outcomes. The surgical community is working toward creating a “safety culture” where even rare events, like CBDI, are accounted for by better training and standard use of safety measures. Incorporating modules that simulate potential intraoperative errors in judgment (like misidentification of the CBD), stress safe decision making, and emphasize use of proved safety measures, are likely to be helpful as surgical simulators become a more integral part of residency training. These changes in training and culture might be key to mitigating any risk posed by a higher tolerance for risk. Given that nearly half of all surgeons will cause a CBDI during their career, identifying surgeon characteristics associated with this injury can be important in developing injury-reduction strategies. The current lack of readily modifiable risk factors highlights the importance of identifying other, perhaps less traditional, targets for quality improvement intervention. Author Contributions Study conception and design: Broeckel Elrod, Flum Acquisition of data: Devlin, Broeckel Elrod, Flum Analysis and interpretation of data: Massarweh, Devlin, Gaston Symons Drafting of manuscript: Massarweh, Devlin Critical revision: Massarweh, Flum Acknowledgment: We thank the American College of Surgeons for their gracious support of this research.
REFERENCES 1. Flum DR, Dellinger EP, Cheadle A, et al. Intraoperative cholangiography and risk of common bile duct injury during cholecystectomy. JAMA 2003;289:1639–1644. 2. Flum DR, Koepsell T, Heagerty P, et al. Common bile duct injury during laparoscopic cholecystectomy and the use of intraoperative cholangiography: adverse outcome or preventable error? Arch Surg 2001;136:1287–1292. 3. Archer SB, Brown DW, Smith CD, et al. Bile duct injury during laparoscopic cholecystectomy: results of a national survey. Ann Surg 2001;234:549–558; discussion 558⫺559. 4. Francoeur JR, Wiseman K, Buczkowski AK, et al. Surgeons’ anonymous response after bile duct injury during cholecystectomy. Am J Surg 2003;185:468–475. 5. Laparoscopic Injury Study and Risk Management Review For General Surgery 2000. Exhibit 6. Rockville, MD: Physician Insurers Association of America; 2000:11, 15.
J Am Coll Surg
6. Kern KA. Malpractice litigation involving laparoscopic cholecystectomy. Cost, cause, and consequences. Arch Surg 1997;132:392– 397; discussion 397⫺398. 7. Lillemoe KD, Martin SA, Cameron JL, et al. Major bile duct injuries during laparoscopic cholecystectomy. Follow-up after combined surgical and radiologic management. Ann Surg 1997; 225:459–468; discussion 468⫺471. 8. Boerma D, Rauws EA, Keulemans YC, et al. Impaired quality of life 5 years after bile duct injury during laparoscopic cholecystectomy: a prospective analysis. Ann Surg 2001;234:750–757. 9. Melton GB, Lillemoe KD, Cameron JL, et al. Major bile duct injuries associated with laparoscopic cholecystectomy: effect of surgical repair on quality of life. Ann Surg 2002;235:888–895. 10. Moore DE, Feurer ID, Holzman MD, et al. Long-term detrimental effect of bile duct injury on health-related quality of life. Arch Surg 2004;139:476–481; discussion 481⫺482. 11. Flum DR, Cheadle A, Prela C, et al. Bile duct injury during cholecystectomy and survival in Medicare beneficiaries. JAMA 2003;290:2168–2173. 12. Massarweh NN, Flum DR. Role of intraoperative cholangiography in avoiding bile duct injury. J Am Coll Surg 2007;204: 656–664. 13. Massarweh NN, Devlin A, Elrod JA, et al. Surgeon knowledge, behavior, and opinions regarding intraoperative cholangiography. J Am Coll Surg 2008;207:821–830. 14. Way LW, Stewart L, Gantert W, et al. Causes and prevention of laparoscopic bile duct injuries: analysis of 252 cases from a human factors and cognitive psychology perspective. Ann Surg 2003;237:460–469. 15. Jackson DN, Hourany L, Vidmar NJ. A four-dimensional interpretation of risk taking. J Pers 1972;40:483–501. 16. Jackson DN. Jackson Personality Inventory-Revised. Port Huron, MI: Sigma Assessments Systems, Inc; 1994. 17. Pearson SD, Goldman L, Orav EJ, et al. Triage decisions for emergency department patients with chest pain: do physicians’ risk attitudes make the difference? J Gen Intern Med 1995;10: 557–564. 18. Ferzli GS, Fiorillo MA, Hayek NE, Sabido F. Chief resident experience with laparoscopic cholecystectomy. J Laparoendosc Adv Surg Tech A 1997;7:147–150. 19. Jensen AR, Wright AS, Lance AR, et al. The emotional intelligence of surgical residents: a descriptive study. Am J Surg 2008; 195:5–10. 20. Killgore WD, Kahn-Greene ET, Lipizzi EL, et al. Sleep deprivation reduces perceived emotional intelligence and constructive thinking skills. Sleep Med 2008;9:517–526. 21. Talarico JF, Metro DG, Patel RM, et al. Emotional intelligence and its correlation to performance as a resident: a preliminary study. J Clin Anesth 2008;20:84–89. 22. Nightingale SD. Risk preference and laboratory use. Med Decis Making 1987;7:168–172. 23. Nightingale SD. Risk preference and laboratory test selection. J Gen Intern Med 1987;2:25–28. 24. Nightingale SD. Risk preference and admitting rates of emergency room physicians. Med Care 1988;26:84–87. 25. Nightingale SD, Grant M. Risk preference and decision making in critical care situations. Chest 1988;93:684–687.
Cholecystectomy, when performed laparoscopically, is often an ambulatory procedure with a low rate of major complications. The most substantial risk remains a common bile duct injury (CBDI), which occurs in 1 of every 200 to 400 patients.1,2 Half of the surgeons who perform laparoscopic cholecystectomy (LC) are estimated to cause such an injury during their career, making litigation related to CBDI among the leading sources of medical malpractice claims against surgeons.3-5 Repair of these injuries is often
complex and can require multiple interventions and, not infrequently, biliary reconstruction.3,6,7 Major CBDI also has a considerable impact on quality of life, functional status, and survival.8-11 Among Medicare beneficiaries young and old in the 1990s, a CBDI during cholecystectomy was associated with a nearly threefold increase in the risk of death during followup compared with uninjured patients.11 Routine use of intraoperative cholangiography (IOC) during LC for the purposes of preventing major CBDI has demonstrated benefit, especially early in a surgeon’s career.2 Most of the supportive evidence is observational and the majority of US surgeons do not use IOC routinely.1,12 Reasons for not using an IOC include a perception that it is ineffective, a lack of available ancillary radiology staff (to help perform the procedure in the operating room), and a number of other possibilities.13 Most likely the actual rea-
Disclosure Information: Nothing to disclose. Received October 30, 2008; Revised February 25, 2009; Accepted February 25, 2009. From the Departments of Surgery, Health Services, and the Surgical Outcomes Research Center, University of Washington, Seattle, WA. Correspondence address: Nader N Massarweh, MD, Department of Surgery, University of Washington, 1959 NE Pacific St, Box 356410, Seattle, WA 98195-6410. email: [email protected]
© 2009 by the American College of Surgeons Published by Elsevier Inc.
17
ISSN 1072-7515/09/$36.00 doi:10.1016/j.jamcollsurg.2009.02.063
18
Massarweh et al
Risk Tolerance and Bile Duct Injury
Abbreviations and Acronyms
CBDI IOC JPI LC
⫽ ⫽ ⫽ ⫽
common bile duct injury intraoperative cholangiogram Jackson Personality Index laparoscopic cholecystectomy
son for not using IOC is multifactorial. The reason major CBDIs occur is also likely multifactorial and can include not using an IOC; misinterpretation of the surgical anatomy because an IOC was not done or misinterpreted; or poor decision making, such as not converting to an open procedure when questions about the anatomy arise. Way and colleagues14 proposed heuristic processes (mental rules of thumb the mind constructs by making use of uncertain probabilistic information to help identify the common bile duct) and a willingness to disregard disconfirmatory information, such as a concerning IOC, as contributing factors to CBDI. Improving quality of delivered care is a focus of many health care reform initiatives. Certain aspects of quality, such as the process and structure of care, are studied with relative ease. There are other factors that are more difficult to quantify. For example, a topic not traditionally studied is the influence of an individual’s personality on everyday clinical decision making and practice—more specifically, the role of risk-taking behavior and risk preference. The risk-taking construct became an established part of overall personality assessment in the 1970s.15 In the field of personality psychology, the Jackson Personality Index (JPI) is a well-established assessment tool, with risk-taking as 1 of 15 personality subscales.16 Pearson and colleagues17 adapted the JPI risk-taking assessment scale to evaluate the influence of physician “risk attitudes” on clinical practice. Although never studied among surgeons, our hypothesis was variability in risk-taking preference can bear on clinical practice. In conducting this survey of practicing surgeons, we specifically hoped to identify discrete demographic, practice, or risk-taking characteristics associated with a selfreported history of causing a CBDI.
METHODS All procedures used to conduct this study were approved by the University of Washington Institutional Review Board. A random sample of 5,000 American College of Surgeons members was selected from the roster of “general surgeons” to receive this mailed survey. The survey content and the mailing methodology have been described in an earlier study.13 Although the JPI risk-taking subscale in its entirety is composed of 20 questions, pilot data from the Pearson
J Am Coll Surg
Table 1. Questions Assessing Surgeon Risk Preference I enjoy taking risks. I try to avoid situations with uncertain outcomes. Taking risks does not bother me if the gains involved are high. I consider security an important element in every aspect of my life. People have told me that I seem to enjoy taking chances. I rarely take risks when there is another alternative. Responses to all questions were scored on a 5-point Likert scale ranging from “very much like me” to “not at all like me.” Scores were adjusted for missing responses.
study using this risk-taking subscale revealed the complete assessment was too lengthy and, as such, was shortened to six questions.17 We chose to use these same six questions (Table 1). Risk-taking preference was scored on a 5-point ordinal scale based on surgeon responses. For three of the questions (questions 1, 3, and 5), the highest point value was assigned to the most affirmative responses (“Not at all like me” ⫽ 1 point; “very much like me” ⫽ 5 points). For the other three questions, the scoring scheme was reversed. Scores were calculated in the same manner described by Pearson and colleagues17 (multiplying by a correction factor for any missing responses that assumes the average of available responses applies to missing responses: risk-taking score ⫽ sum of response ⫻ [6/6, number of missing responses]) with stratification into risk categories based on raw score: high risk scored ⬎1 SD more than mean; moderate risk scored between 1 SD more than and 1 SD less than the mean; low risk scored ⬎1 SD less than mean. Descriptive statistics were calculated for the cohort of respondents. Responses from self-identified CBDI injuring surgeon were compared with noninjuring surgeons. Categorical variables were compared using chi-square statistics. Calculated p values refer to the significance of the differences across strata (between injuring and noninjuring surgeons’ responses) and are not adjusted for multiple comparisons. A “missing” category was assigned to nonresponses, as most questions received ⬍100% response. This missing category was excluded from statistical comparisons. Mean values of the continuous variables (age, years since residency, and risk score) were compared using Student’s t-test and are reported with standard deviations. Statistical analyses were performed using STATA version 10.1 (Stata Corp).
RESULTS A total of 4,100 surveys were mailed. One hundred eight were not post-deliverable and were discounted when calculating response rate. Among surgeons who returned an incomplete survey (n ⫽ 303), the most commonly noted reason was retirement (63%), followed by cholecystectomy not being a part of the practice (16.6%), no reason offered
Vol. 209, No. 1, July 2009
Massarweh et al
Risk Tolerance and Bile Duct Injury
19
Table 2. Responding Surgeon Demographic Information Characteristics
% Male (%) Age (y), mean ⫾ SD Years since residency, mean ⫾ SD Training included IOC (%) Practice setting (%) Private practice Academic staff Private/public hospital HMO Federal Other Missing Procedures with residents (%) Yes No Missing Point of care (%) Community hospital Tertiary referral center Outpatient surgery center Other Missing Cholecystectomy training (%) Course Residency/fellowship Partner No formal training Other Missing Cholecystectomies (%) 1–10 11–20 21–50 51–100 ⬎100 Missing†
All (n ⴝ 1,412)
Noninjuring (n ⴝ 878)
Injuring (n ⴝ 534)
p Value*
89.2 51.8 ⫾ 9.6 19.6 ⫾ 10.3 79.3
62.0 88.0 51.3 ⫾ 9.8 18.9 ⫾ 10.5 79.2
37.8 91.2 52.8 ⫾ 9.0 20.8 ⫾ 9.7 79.4
0.15 0.004 ⬍0.001 0.99
55.5 12.0 10.6 1.4 0.9 4.8 14.8
52.5 14.5 10.3 1.4 0.7 5.4 15.3
60.3 7.9 11.2 1.5 1.3 3.8 14.0
22.6 58.7 18.6
25.0 55.9 19.1
18.7 63.5 17.8
61.4 17.6 2.4 3.8 14.9
59.2 19.4 2.2 3.9 15.4
65.0 14.6 2.8 3.6 14.0
33.0 60.3 2.5 1.9 0.6 1.7
29.8 63.3 3.0 1.6 0.5 1.8
38.2 55.4 1.7 2.4 0.8 1.5
4.8 8.3 28.2 36.0 22.1 0.6
6.8 10.3 28.6 33.5 20.2 0.7
1.5 5.1 27.5 40.1 25.3 0.6
0.003
0.01
0.13
0.01
⬍0.001
*Across strata. † Missing category corresponds to the percentage of surgeons not answering the question. IOC, intraoperative cholangiogram.
(13%), and deceased (7.3%). Total response rate to the survey was 1,756 of 3,992 (44%) and included incomplete surveys returned with a valid reason. After discounting incomplete surveys and duplicate responses, the total number of surveys analyzed was 1,412. A total of 534 survey respondents (37.8%) reported having caused at least one CBDI. General demographic and practice data for the survey respondents are presented in Table 2. Surgeons reporting a CBDI were slightly older
(mean age 52.8 years versus 51.3 years; p ⫽ 0.004) and in practice longer (20.8 versus 18.9 years; p ⬍ 0.001). Surgeons in academic practice (7.9% versus 14.5%) or who worked with residents regularly (18.7% versus 25.0%) reported a CBDI less frequently. Surgeons not reporting a CBDI were more likely trained in LC during residency (63.3% noninjuring versus 55.4% injuring) as compared with surgeons reporting a CBDI, who were more likely trained at an LC course (29.8% noninjuring versus 38.2%
20
Massarweh et al
Risk Tolerance and Bile Duct Injury
Table 3. Characteristics of Common Bile Duct Injuries Caused by Injuring Surgeons (n ⫽ 534) Reported ⬎1 CBDI Yes No Missing Biliary reconstruction Yes No Missing Patient death as a result Yes No Missing Suspected cause of CBDI Misidentification of cystic duct Unknown Anatomic reason Traction injury Electrocautery injury More than one reason Other Missing*
n
%
180 333 21
33.7 62.4 3.9
397 119 18
74.3 22.3 3.4
8 509 17
1.5 95.3 3.2
307 56 42 31 17 16 53 12
57.5 10.5 7.9 5.8 3.2 3.0 9.9 2.2
*Missing category corresponds to the percentage of surgeons not answering the question. CBDI, common bile duct injury.
injuring). A smaller proportion of low-volume (⬍20 cholecystectomies/year) as compared with high-volume (⬎100 cholecystectomies/year) surgeons (6.6% lowvolume versus 25.3% high-volume) reported a CBDI. Table 3 summarizes the circumstances surrounding the reported CBDIs and the surgeon-perceived reason for injury. One-third of respondents who self-reported a CBDI stated they had caused more than one such injury (12.7% of the total cohort). A biliary reconstruction was performed for 74.3% of patients with an injured bile duct and 1.5% reported that the injured patients died because of the injury. In the majority of patients, the surgeon cited a problem with the anatomy (eg, aberrant anatomy, adhesions, inflammation, and so forth [7.9%]) or misidentification of the cystic duct (57.5%) as the main reason for CBDI. Injuring and noninjuring surgeons did not report any substantial differences in IOC use (Table 4). A greater proportion of injuring surgeons thought most CBDIs occur in the setting of unclear anatomy (58% versus 49%; p ⫽ 0.01). Within both groups of surgeons, a notable number misinterpreted an IOC image (Fig. 1) clearly demonstrating no filling of the proximal bile ducts, with 16% of injuring and 18% of noninjuring surgeons thinking a CBDI to be unlikely. Based on the information provided by the
J Am Coll Surg
image, similar proportions of both groups reported they were likely or very likely to clip and cut the structure (18.9% injuring versus 15.4% noninjuring) and convert to an open procedure (59.7% injuring versus 61.1% noninjuring). Mean risk-taking preference score for this cohort was 12.0 ⫾ 4.5 (12.2 ⫾ 4.5 injuring and 11.9 ⫾ 4.4 noninjuring; p ⫽ 0.23), with a range from the lowest possible score of 6 to the highest possible score of 30. Several surgeons did not complete all six questions (one question, n ⫽ 21; two questions, n ⫽ 2; three questions, n ⫽ 1; four questions, n ⫽ 1; five questions, n ⫽ 3), and 12 surgeons did not complete any of the questions. We did not find any statistically significant difference in the distribution of risktaking preference scores between injuring and noninjuring surgeons when stratified by 1 SD around the mean (Fig. 2A). When we examined the distribution based on deciles of data, we again did not find any significant differences (Fig. 2B), but we did note a greater proportion of injuring as compared with noninjuring surgeons in the upper three deciles. The absolute percentage difference between each of these three deciles was relatively small (⬃2%), but the relative risk of CBDI for surgeons within these three upper deciles was 17% greater (p ⫽ 0.07) than for those in the lowest three deciles (where the proportion of noninjuring surgeons is greater). Similarly, when we evaluated risk score based on how respondents answered questions pertaining to Figure 1, a greater proportion of surgeons with low risk scores were more likely to think the IOC demonstrated a CBDI (interpreting with caution; Fig. 3A) and were more likely to covert to open based on the cholangiogram results (proceeding with caution; Fig. 3B).
DISCUSSION CBDI during LC is a relatively uncommon complication.1,2 The impact on a patient’s quality of life and the subsequent economic and clinical consequences of such injuries are profound.8-11 Our group recently published results from a survey of US surgeons examining reported differences in practice between routine and selective cholangiographers and found considerable differences with regard to knowledge, use, and opinions about IOC.13 Here we report results of a unique aspect of the survey that describes self-reported differences in characteristics, practice patterns, and risk-taking preference associated with CBDI during LC. When LC was introduced, surgeons learned the procedure at short, 1- to 3-day courses and subsequently used these techniques without initial supervision or proctoring. Another published survey evaluating the effects of residency training on the rates of bile duct injury demon-
Vol. 209, No. 1, July 2009
Massarweh et al
Risk Tolerance and Bile Duct Injury
21
Table 4. Surgeon Cholangiography Use and Characteristics Based on Common Bile Duct Injury Status Type of cholangiographer Routine Selective Missing† Cholecystectomies with IOC (%) 0–10 11–25 26–50 51–75 ⬎75 Missing† Type of cholangiogram Fluoroscope Single-shot x-ray Missing† Time to perform IOC (min) 1–10 11–20 21–30 31–40 Missing†
All (%) (n ⴝ 1,412)
Noninjuring (%) (n ⴝ 878)
Injuring (%) (n ⴝ 534)
p Value*
26.9 72.8 0.3
27.8 71.9 0.3
25.5 74.3 0.2
0.54
48.2 17.1 4.6 4.8 25.0 0.3
46.4 17.5 5.0 4.2 26.5 0.3
51.1 16.5 3.9 5.8 22.5 0.2
79.0 19.8 1.3
79.4 19.3 1.4
78.3 20.6 1.1
0.77
45.6 42.6 9.8 1.4 0.6
44.7 42.7 10.1 1.9 0.6
47.2 42.3 9.2 0.7 0.8
0.25
0.23
*Across strata. † Missing category corresponds to the percentage of surgeons not answering the question. IOC, intraoperative cholangiogram.
strated graduates from more contemporary time periods (between 1992 and 1998 as compared with 1980 to 1990) were less likely to experience laparoscopic learning curve injuries, resulting in a decreased number of CBDIs.3 Although it appears formalized, LC training as a part of general surgery residency programs has contributed to these results,18 major CBDIs (complete transection) still occur and can potentially be preventable. CBDI can also be a manifestation of more years practicing LC. This could account for a greater proportion of those reporting a CBDI being older, longer in practice (had less opportunity for residency training where laparoscopy was an important component), and receiving LC training at a course. Interestingly, academic surgeons and those who reported performing the majority of their procedures with residents were less likely to report a CBDI, but also more frequently reported being low-volume surgeons (32.2% low-volume versus 7.1% high-volume [academics]; 23.8% low-volume versus 12.5% high-volume [work with residents]). This might simply relate to volume, but might also be because surgeons working with residents benefit from having a second set of “critical eyes” in the operating room. It remains to be determined if the dynamics of teaching, with emphasis on reviewing anatomy during dissection, are also involved in these differences.
One point of concern was the relatively high proportion of surgeons from both groups who failed to correctly identify an obvious, impending CBDI presented through a clinical vignette (Fig. 1). When respondents were presented an image of a cholangiogram demonstrating no proximal filling of the hepatic ducts, which would suggest a likely misidentification of the CBD as the cystic duct, 16% and 18% of injuring and noninjuring surgeons, respectively, thought an injury was unlikely. Although studies sufficiently powered to detect a difference in rates of CBDI when an IOC is used consistently demonstrate a benefit,12 implicit within the argument for IOC use is the assumption an IOC is properly performed and properly interpreted. If surgeons are misinterpreting IOCs, educational initiatives aimed at improving basic radiologic interpretation could be useful. Some surgeons can correctly perform an IOC, but disregard the results and rely instead on visual cues, resulting in a misidentification injury.14 The study by Way and colleagues14 describes how heuristics can lure the unsuspecting surgeon into making anatomic errors, so he or she does not believe they are cutting the common bile duct. Greater use of IOC can help decrease slavish reliance on heuristics and might be key to avoiding CBDIs based on faulty cognitive interpretation of familiar visual cues. Tolerance of anatomic uncertainty during cys-
22
Massarweh et al
Risk Tolerance and Bile Duct Injury
J Am Coll Surg
Figure 1. (A) The clinical vignette, (B) intraoperative cholangiogram image demonstrating no proximal filling of the hepatic ducts, and (C) subsequent questions as presented in the survey.
tic duct dissection could account for the relative underuse of IOC before clipping and cutting ductal structures. Differences in cognitive processing of data, divergent thresholds for accepting potential anatomic variants, or willingness to take risks in the setting of disconfirmatory information need to be addressed in more creative ways. We know little about surgeon-level practice variation based on differences in personality and risk assessment. Several studies have attempted to evaluate the role of emotional intelligence as it pertains to the core competencies of residency training.19-21 To our knowledge as yet no study has applied such measures to assess impact on clinical practice. The impact of risk-taking preference on physician practice has been assessed in several studies.17,22-25 Pearson and colleagues17 (using the same methodology here) demonstrated higher risk-taking preference scores among emergency physicians who were less likely to admit patients with chest pain at low or medium risk for myocardial infarction. Similarly, Nightingale and colleagues22-25 used a different risk-assessment method and showed risk-taking preference does come to bear on decisions pertaining to laboratory test selection, admission rates through the emergency department, and practice preference in critical care scenarios. To our knowledge, no study has ever performed risk-taking
Figure 2. Common bile duct injuring and noninjuring surgeon risktaking preference based on (A) low-, moderate-, and high-risk preference scores and (B) risk-preference deciles. Differences between injuring and noninjuring surgeons within either the upper or lower three deciles were not statistically significant.
preference assessment on surgeons. We used one of these previously used methods, asking six questions taken from the JPI risk-taking subscale to assess the influence of risktaking preference on the rate of CBDI.17 Stratifying surgeons into risk groups based on standard deviations around the mean, we did not find any substantial differences based on low-, moderate-, and high-risk categories. Ideally, we would have carried the stratification out to 2 SDs to identify outliers at the extremes of risktaking preference. Unfortunately, because the data were right-skewed, use of standard deviations might not be the best way to represent extremes of personality. Additionally, because this form of personality testing is an emerging area of study, there is little guidance to suggest break points that might be appropriate for stratifying groups. We were concerned the risk intervals were too wide to capture subtle differences or trends in risk-taking preference and so we stratified scores into deciles. Although there was no statis-
Vol. 209, No. 1, July 2009
Figure 3. Based on Figure 1C. Surgeon risk-taking preference based on (A) interpretation of the image and (B) likelihood of converting to an open procedure based on the image. For (A), surgeons were categorized as cautious if they were “likely” or “very likely” to think Figure 1 represented a potential common bile duct injury and were not cautious if they answered “unlikely.” For (B), surgeons were categorized as cautious if they answered “likely” or “very likely” to convert to open based on Figure 1 and were not cautious if they answered “unlikely.”
tically significant difference, a greater proportion of injuring as compared with noninjuring surgeons clustered in the upper three deciles. Because of a lack of data at the high end of the risk scale (because of the distribution of the data), we did not believe expanding our stratification more would yield meaningful results. This six-question risk-assessment method has not been validated for lack of a gold standard and has never been used among surgeons for comparative evaluation. Our analysis of this component of the survey was exploratory, and we found it interesting that within this decile-based stratification, a greater proportion of those with a reported history of a CBDI were at the high end of the risk-taking spectrum. The absolute differences in reported rates of CBDI between groups in the upper (26.2% injuring versus
Massarweh et al
Risk Tolerance and Bile Duct Injury
23
20.6% noninjuring; p ⫽ 0.57) and lower (42.4% injuring versus 45.0% noninjuring; p ⫽ 0.57) three deciles were small, but this small difference actually translated to a 17% increase in the relative risk of CBDI. Findings that suggest a group of surgeons might be at increased risk for a potentially preventable injury merits additional exploration. One consideration is surgeons with higher risk-taking preference are more likely to ignore disconfirmatory evidence in an abnormal IOC or to rely on heuristics to influence or guide surgical decision making. A greater proportion of surgeons interpreting Figure 1 with caution (thinking it “likely” or “very likely” represents a CBDI about to happen) fell within the low-risk category (Fig. 3A). Alternatively, differences in risk-taking preference might manifest as poor clinical decisions during operation. Respondents reporting they would proceed with caution based on Figure 1 (“likely” or “very likely” to convert to open) also fell within the low-risk category (Fig. 3B). A larger cohort might have allowed for better evaluation of these differences and easier identification of outliers at the extremes. Studies targeting nontraditional end points, such as surgeon cognition, data assimilation, perception, and risk assessment would be useful to explore this relationship more. There were several notable limitations to this study. Our response rate fell below the epidemiologic standard of 60%, subjecting our study to nonresponse bias. Our study might also have suffered from normative responses (in which surgeons select response based on what they think is the expected answer as opposed to responses reflecting actual practice or belief ), although this survey was anonymous. Another important limitation is missing data. Not all questions received a 100% response rate, possibly limiting the generalizability of our findings. Assessing between-group differences was the central goal of this study; the overall effect of missing data should be small. Finally, although the JPI is an established personality assessment tool when used as a whole, the use of this six-question “mini” JPI to assess risk-taking preference is uncertain and has yet to be validated. Our results identified surgeons with markers of increased clinical activity and certain practice characteristics as reporting higher rates of CBDI; but the individual contribution of these variables could not be reliably assessed. Although we did not find any statistically significant association between average risk-taking preference and the incidence of CBDI, our analysis of extremes of risk-taking preference (often used in evaluations of behavior and personality) did demonstrate a higher injury rate among those with higher risk-taking preference scores. This finding could inform future studies specifically aiming to explore the relationship between surgeon personality or cognitive processes and clinical practice or outcomes. Other high-
24
Massarweh et al
Risk Tolerance and Bile Duct Injury
risk industries, such as aviation, emphasize standard training and consistency in the application of proved safety interventions that decrease the risk of even rare, catastrophic outcomes. The surgical community is working toward creating a “safety culture” where even rare events, like CBDI, are accounted for by better training and standard use of safety measures. Incorporating modules that simulate potential intraoperative errors in judgment (like misidentification of the CBD), stress safe decision making, and emphasize use of proved safety measures, are likely to be helpful as surgical simulators become a more integral part of residency training. These changes in training and culture might be key to mitigating any risk posed by a higher tolerance for risk. Given that nearly half of all surgeons will cause a CBDI during their career, identifying surgeon characteristics associated with this injury can be important in developing injury-reduction strategies. The current lack of readily modifiable risk factors highlights the importance of identifying other, perhaps less traditional, targets for quality improvement intervention. Author Contributions Study conception and design: Broeckel Elrod, Flum Acquisition of data: Devlin, Broeckel Elrod, Flum Analysis and interpretation of data: Massarweh, Devlin, Gaston Symons Drafting of manuscript: Massarweh, Devlin Critical revision: Massarweh, Flum Acknowledgment: We thank the American College of Surgeons for their gracious support of this research.
REFERENCES 1. Flum DR, Dellinger EP, Cheadle A, et al. Intraoperative cholangiography and risk of common bile duct injury during cholecystectomy. JAMA 2003;289:1639–1644. 2. Flum DR, Koepsell T, Heagerty P, et al. Common bile duct injury during laparoscopic cholecystectomy and the use of intraoperative cholangiography: adverse outcome or preventable error? Arch Surg 2001;136:1287–1292. 3. Archer SB, Brown DW, Smith CD, et al. Bile duct injury during laparoscopic cholecystectomy: results of a national survey. Ann Surg 2001;234:549–558; discussion 558⫺559. 4. Francoeur JR, Wiseman K, Buczkowski AK, et al. Surgeons’ anonymous response after bile duct injury during cholecystectomy. Am J Surg 2003;185:468–475. 5. Laparoscopic Injury Study and Risk Management Review For General Surgery 2000. Exhibit 6. Rockville, MD: Physician Insurers Association of America; 2000:11, 15.
J Am Coll Surg
6. Kern KA. Malpractice litigation involving laparoscopic cholecystectomy. Cost, cause, and consequences. Arch Surg 1997;132:392– 397; discussion 397⫺398. 7. Lillemoe KD, Martin SA, Cameron JL, et al. Major bile duct injuries during laparoscopic cholecystectomy. Follow-up after combined surgical and radiologic management. Ann Surg 1997; 225:459–468; discussion 468⫺471. 8. Boerma D, Rauws EA, Keulemans YC, et al. Impaired quality of life 5 years after bile duct injury during laparoscopic cholecystectomy: a prospective analysis. Ann Surg 2001;234:750–757. 9. Melton GB, Lillemoe KD, Cameron JL, et al. Major bile duct injuries associated with laparoscopic cholecystectomy: effect of surgical repair on quality of life. Ann Surg 2002;235:888–895. 10. Moore DE, Feurer ID, Holzman MD, et al. Long-term detrimental effect of bile duct injury on health-related quality of life. Arch Surg 2004;139:476–481; discussion 481⫺482. 11. Flum DR, Cheadle A, Prela C, et al. Bile duct injury during cholecystectomy and survival in Medicare beneficiaries. JAMA 2003;290:2168–2173. 12. Massarweh NN, Flum DR. Role of intraoperative cholangiography in avoiding bile duct injury. J Am Coll Surg 2007;204: 656–664. 13. Massarweh NN, Devlin A, Elrod JA, et al. Surgeon knowledge, behavior, and opinions regarding intraoperative cholangiography. J Am Coll Surg 2008;207:821–830. 14. Way LW, Stewart L, Gantert W, et al. Causes and prevention of laparoscopic bile duct injuries: analysis of 252 cases from a human factors and cognitive psychology perspective. Ann Surg 2003;237:460–469. 15. Jackson DN, Hourany L, Vidmar NJ. A four-dimensional interpretation of risk taking. J Pers 1972;40:483–501. 16. Jackson DN. Jackson Personality Inventory-Revised. Port Huron, MI: Sigma Assessments Systems, Inc; 1994. 17. Pearson SD, Goldman L, Orav EJ, et al. Triage decisions for emergency department patients with chest pain: do physicians’ risk attitudes make the difference? J Gen Intern Med 1995;10: 557–564. 18. Ferzli GS, Fiorillo MA, Hayek NE, Sabido F. Chief resident experience with laparoscopic cholecystectomy. J Laparoendosc Adv Surg Tech A 1997;7:147–150. 19. Jensen AR, Wright AS, Lance AR, et al. The emotional intelligence of surgical residents: a descriptive study. Am J Surg 2008; 195:5–10. 20. Killgore WD, Kahn-Greene ET, Lipizzi EL, et al. Sleep deprivation reduces perceived emotional intelligence and constructive thinking skills. Sleep Med 2008;9:517–526. 21. Talarico JF, Metro DG, Patel RM, et al. Emotional intelligence and its correlation to performance as a resident: a preliminary study. J Clin Anesth 2008;20:84–89. 22. Nightingale SD. Risk preference and laboratory use. Med Decis Making 1987;7:168–172. 23. Nightingale SD. Risk preference and laboratory test selection. J Gen Intern Med 1987;2:25–28. 24. Nightingale SD. Risk preference and admitting rates of emergency room physicians. Med Care 1988;26:84–87. 25. Nightingale SD, Grant M. Risk preference and decision making in critical care situations. Chest 1988;93:684–687.