Beyond the Comfort Zone: Residents Assess Their Comfort Performing Inpatient Medical Procedures

Transcription

1 The American Journal of Medicine (2006) 119, 71.e17-71.e24 CLINICAL RESEARCH STUDY Beyond the Comfort Zone: Residents Assess Their Comfort Performing Inpatient Medical Procedures Grace C. Huang, MD, a,b C. Christopher Smith, MD, a Craig E. Gordon, MD, c David J. Feller-Kopman, MD, a Roger B. Davis, ScD, a Russell S. Phillips, MD, a Saul N. Weingart, MD, PhD a,d a Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School; b Carl J. Shapiro Institute for Education and Research at Harvard Medical School and Beth Israel Deaconess Medical Center; c Department of Medicine, New England Medical Center; d Center for Patient Safety, Dana-Farber Cancer Institute, and Harvard Medical School, Boston, Mass. ABSTRACT PURPOSE: Resident physicians learn to perform inpatient bedside procedures in a manner that is neither standardized nor rigorous. As a result, residents may be unskilled and uncomfortable performing procedures. This study characterizes residents comfort performing medical procedures and identifies factors associated with lack of comfort. SUBJECTS: Study subjects were internal medicine resident physicians who performed one of four medical procedures (central line, lumbar puncture, paracentesis, or thoracentesis) on adult medical inpatients between July 1, 2003, and June 30, METHODS: This prospective cohort study was conducted at a 556-bed Boston teaching hospital. Resident physicians evaluated their comfort with 9 aspects of 4 medical procedures, recording this information in an electronic log. We also abstracted operator characteristics and patient demographic data. We analyzed residents comfort with each aspect of the procedure and defined overall comfort as comfort with each of the 9 aspects. RESULTS: A majority of resident physicians reported lack of comfort with at least one aspect of the procedure. Residents reported lack of comfort with 37% of unsupervised procedures. They also reported lack of comfort with the prospect of managing complications in 35% of procedures. In the multivariable analysis, overall comfort was associated with the use of a dedicated medical procedure service (odds ratio [OR] 1.9, 95% confidence interval [CI] ) and inversely associated with postgraduate year 1 status (OR 0.3, CI ), first time performing the procedure (OR 0.4, CI ), thoracenteses (OR 0.4, CI ), and emergent procedures (OR 0.6, CI ). CONCLUSIONS: Many resident physicians are uncomfortable performing common bedside procedures. Experience and supervision mitigate some, but not all, discomfort Elsevier Inc. All rights reserved. KEYWORDS: Diagnostic techniques and procedures; Clinical competence; Performance assessment This study was supported in whole by grants from the Risk Management Foundation of the Harvard Medical Institutions and the Stoneman Center for Quality Improvement in General Medicine and Primary Care. Funding supported collection, management, analysis, interpretation of data, and preparation of the article. Dr. Weingart was supported in part by a K08 clinical scientist career development award from the US Agency for Healthcare Research and Quality. Requests for reprints should be addressed to Grace C. Huang, MD, Carl J. Shapiro Institute for Education and Research at Harvard Medical School and Beth Israel Deaconess Medical Center, 330 Brookline Avenue, E/ES- 212, Boston, MA address: ghuang@bidmc.harvard.edu Invasive bedside procedures, such as central venous catheter placement, lumbar puncture, paracentesis, and thoracentesis, are often performed in teaching hospitals by the least experienced physicians. 1,2 Although the philosophy of see one, do one, teach one seems deeply ingrained in medical training, there is little evidence that this approach ensures adequate technical skills. In fact, the apprenticeship model of instruction, based on vertical transmission of potentially unreliable knowledge and practice, runs counter to the evidence-based model of training needed to prepare the next generation of physicians. Little is known about how residents develop competence performing common bedside procedures or about the errors they make, despite the reality that procedural complications are a significant cause of inpatient morbidity and mortal /$ -see front matter 2006 Elsevier Inc. All rights reserved. doi: /j.amjmed

2 71.e18 The American Journal of Medicine, Vol 119, No 1, January 2006 ity. 3,4 By using experience as a proxy for competence, the American Board of Internal Medicine (ABIM) requires that residents perform a specified number of each procedure during their training to become eligible for board certification. 5 Unfortunately, there is not an empirical basis for this requirement. In fact, one study showed that these standards underestimate the number of procedures needed for residents to be CLINICAL SIGNIFICANCE comfortable with their performance. 6 Another study demonstrated that complication rates decrease by half after clinicians place 50 central venous catheters, well above the ABIM minimum requirement of 5 central lines. 7 Absent a research base, we do not know whether these consensusbased requirements bear any relationship to professional mastery. Residents comfort when performing procedures has been proposed as an alternative marker of competence in several studies. 6,8,9 This perspective is based on the untested assumption that a resident who is uncomfortable performing a procedure is unlikely to be competent at that task. Comfort or confidence has been defined as an all-or-none phenomenon and ascertained by residents recall of past procedures. Comfort, however, may represent a more complex phenomenon. Trainees may be comfortable with certain aspects of particular procedures in some situations or settings, but not in others. Understanding the attributes of a procedure or its context that erode residents comfort may facilitate the design of educational interventions that improve patient care and resident satisfaction. To better understand residents experiences performing medical procedures, we studied internal medicine residents at a Boston teaching hospital. We hypothesized that many residents would be uncomfortable performing these procedures and that comfort would increase with experience and supervision. METHODS Study Site We studied inpatient medical procedures performed by internal medicine residents at a 556-bed tertiary care Boston teaching hospital. The internal medicine training program in 2003 and 2004 included 63 postgraduate year 1 (PGY1) residents (16 in a 1-year preliminary program), and 46 postgraduate year 2 (PGY2) and 46 postgraduate year 3 (PGY3) categorical residents. Residents completed required and elective rotations in general medicine, cardiology, oncology, emergency medicine, intensive care, ambulatory care, and subspecialty services. Internal medicine residents are more uncomfortable performing bedside procedures than previously described. Being unsupervised and the prospect of managing complications were the aspects most difficult for residents. Thoracenteses are particularly challenging. Direct supervision by more experienced staff physicians doubles the odds of being comfortable performing procedures. We introduced a medical procedure service in July The medical procedure service is staffed by hospitalists and pulmonary/critical care specialists with experience teaching and performing common inpatient procedures; it is available 24 hours per day. Medical procedure service faculty supervise residents during 4 bedside procedures: central venous catheter placement, lumbar puncture, paracentesis, and thoracentesis. During the study, the internal medicine program director strongly recommended that house officers consult the medical procedure service when performing procedures, but the consultation was neither mandatory nor universally used. Residents who used the service were required to review an online multimedia curriculum and complete a self-assessment quiz before performing the procedure under direct supervision by a medical procedure service physician. Study Design and Data Sources We conducted a prospective cohort study using data from online resident procedure logs for 527 procedures reported from July 2003 to June Residents were required to complete the log, which was used to document board eligibility in internal medicine. Residents entered the following information in preformatted fields: name of operator, year of training, date of procedure, patient s medical record number, name of attending supervisor if applicable, procedure, number of needle passes used to complete the procedure, immediate complications, urgency of procedure, time of day, location of procedure, and the operator s self-reported prior experience with this procedure. By using a 5-point Likert scale, they also reported comfort with 9 aspects of the procedure: understanding indications and contraindications, obtaining consent, recognizing anatomic landmarks, using equipment components, applying sterile technique, positioning the patient, performing the procedure itself, interpreting results, and managing complications. Responses were selected from drop-down menus or entered into free-text fields. We required all questions to be answered before the procedure could be submitted. Investigators coded free-text complication entries into the following categories: arrhythmia, arterial puncture, bleeding or hematoma, bowel perforation, no fluid obtained (dry tap), hypotension, pneumothorax, and pneumothorax requiring chest tube. We assigned gender and resident type (categorical or preliminary) using records from the graduate medical education office and determined whether the medical procedure service was involved based on the supervisor s name. By using medical record number and procedure

3 Huang et al Beyond the Comfort Zone 71.e19 date, we abstracted patient characteristics (age, race/ethnicity, type of insurance, length of stay) electronically from the hospital s registration system. Because of missing identification information or inaccuracies in procedure logs, we were able to identify patient-specific information for only 418 (79%) of 527 procedures. Data Analyses We tabulated the characteristics of resident physicians by the training year, gender, type, and self-reported number of procedures completed before the index procedure. The unit of analysis for the study was the procedure. We analyzed operator (resident level, resident gender, resident type) and procedure characteristics (procedure location, level of urgency, time of day, use of the procedure service, mean number of needle passes, reported complications) by type of procedure. We defined comfort as a response of 4 or 5 ( somewhat comfortable or extremely comfortable ) on a 5-point Likert scale and stratified comfort by procedure and by year of training. We reasoned that a resident who lacked comfort with any aspect of the procedure was uncomfortable with the procedure as a whole and called this the overall comfort with the procedure. To compare our results with previous studies, we also calculated average comfort as the unweighted mean comfort score across all aspects of the procedure, defining comfort as a mean score of 4 or greater. We used the chi-square test of independence for comparisons using nominal variables and the Wilcoxon rank-sum test for ordinal variables. We used the Kruskal- Wallis statistic for comparisons involving more than two procedures or training levels. We also created a multivariable logistic regression model with backward elimination (P.10) to analyze factors associated with overall comfort, clustering by resident. Independent variables were PGY1 status (with PGY2 and PGY3 as referent group), female gender, categorical status (preliminary as referent group), emergent procedures (nonemergent as referent group), type of procedure (lumbar puncture, thoracentesis, paracentesis were separate dummy variables, with central line as referent group), procedures performed after hours, intensive care unit procedures (compared with elsewhere), and first time performing the procedure. Analyses used Stata 7.0 (StataCorp, College Station, Tex). The study protocol was approved in advance by the hospital investigational review board. RESULTS Resident Characteristics A total of 106 (68%) of 157 residents logged procedures during the academic year (Table 1). Fifty-five residents (52%) were PGY1, 53 (50%) were female, and 94 (89%) were enrolled in the 3-year categorical program. Fourteen residents (13%) performed the index procedure for the first time, and 56 (53%) had performed the procedure 1 to 5 times previously. Table 1 Resident physicians characteristics Total n 106 Training year, n (%) PGY1 55 (52%) PGY2 29 (27%) PGY3 22 (21%) Gender, n (%) Female 53 (50%) Male 53 (50%) Type, n (%) Categorical 94 (89%) Preliminary 12 (11%) Self-reported prior procedures, n (%)* 0 14 (13%) (53%) (21%) (9%) 15 4 (4%) PGY postgraduate year. *At the time of the index procedure, residents reported the number of procedures performed previously. Patient Characteristics Patients had a mean age of 62.7 years (range years). More than 50% were male, and 69% were white. Medicare (48%) was the principal insurer, followed by private insurers (30%), and Medicaid (16%). The mean length of stay was 18.6 days (range days). Procedures Residents completed 527 procedures (Table 2), including 268 central lines, 95 lumbar punctures, 81 paracenteses, and 83 thoracenteses. PGY1 residents performed procedures most often (54% of cases). Central lines were most frequently performed in the intensive care unit (74%), lumbar punctures were divided equally between the emergency department and hospital ward (40%), and paracenteses (52%) and thoracenteses (46%) were most commonly performed on the hospital ward. Most procedures (57%) had urgent indications, except for thoracenteses (58% were elective). The majority of procedures were performed during the workday (8 AM to 5 PM). Residents used the medical procedure service for 39% of procedures. However, residents were especially likely to use the medical procedure service for thoracenteses (78%). Residents reported an average of 1.8 needle passes per procedure. Bleeding was the most frequently cited immediate complication, although it occurred in only 17 of 527 procedures (3%); 13 occurred during central line placements. Comfort with Procedures We asked residents to identify their level of comfort with 9 aspects of each procedure (Table 3). Residents reported lack of comfort performing procedures because of the lack of supervision in 37% of cases. In particular, lack of comfort related to supervision was cited in 67% of unsupervised

4 71.e20 The American Journal of Medicine, Vol 119, No 1, January 2006 Table 2 Procedure characteristics Total number All procedures n 527 Central line n 268 Lumbar puncture n 95 Paracentesis n 81 Thoracentesis n 83 Resident level, n (%) PGY1 283 (54%) 139 (52%) 45 (47%) 48 (59%) 51 (61%) PGY2 140 (27%) 81 (30%) 32 (34%) 17 (21%) 10 (12%) PGY3 104 (20%) 48 (18%) 18 (19%) 16 (20%) 22 (27%) Gender, n (%) Female 273 (52%) 133 (50%) 49 (52%) 47 (58%) 44 (53%) Male 254 (48%) 135 (50%) 46 (48%) 34 (42%) 39 (47%) Trainee type, n (%) Categorical 478 (91%) 242 (90%) 87 (92%) 71 (88%) 78 (94%) Preliminary 49 (9%) 26 (10%) 8 (8%) 10 (12%) 5 (6%) Location, n (%) Ward 149 (28%) 31 (12%) 38 (40%) 42 (52%) 38 (46%) Emergency unit 86 (16%) 37 (14%) 38 (40%) 9 (11%) 2 (2%) Intensive care unit 270 (51%) 198 (74%) 18 (19%) 30 (37%) 24 (29%) Other 22 (4%) 2 (1%) 1 (1%) 0 (0%) 19 (23%) Urgency, n (%) Elective 174 (33%) 61 (23%) 27 (28%) 38 (47%) 48 (58%) Urgent 302 (57%) 174 (65%) 53 (56%) 40 (49%) 35 (42%) Emergent 51 (10%) 33 (12%) 15 (16%) 3 (4%) 0 (0%) Time of day, n (%) 8 AM-5 PM 331 (63%) 146 (54%) 54 (57%) 54 (67%) 77 (93%) 5 PM-11 PM 116 (22%) 66 (25%) 27 (28%) 19 (23%) 4 (5%) 11 PM-8 AM 80 (15%) 56 (21%) 14 (15%) 8 (10%) 2 (2%) Use of procedure service, n (%) Yes 206 (39%) 89 (33%) 27 (28%) 25 (31%) 65 (78%) No 321 (61%) 179 (67%) 68 (72%) 56 (69%) 18 (22%) Mean number of passes, n (SD, range) 1.8 (1.5, 1-20) 1.9 (1.5, 1-15) 2.2 (2.4, 1-20) 1.4 (0.8, 1-5) 1.2 (0.4, 1-3) Reported complications or problems, n Any event Arrhythmia Arterial puncture Bleeding/hematoma Bowel perforation Dry tap Hypotension Pneumothorax Pneumothorax requiring chest tube SD standard deviation. thoracentesis cases. In contrast, residents reported lack of comfort in only 2.4% of supervised procedures. Residents reported lack of comfort with their ability to manage complications in 35% of procedures overall and in 47% of thoracenteses. Residents infrequently reported lack of comfort with other aspects of the procedure, such as equipment (8.5%), anatomy (5.3%), interpretation of results (4.7%), indications/contraindications (2.1%), informed consent (1.9%), and sterile technique (1.3%). We use comfort throughout this article to denote absence of a low comfort score on any single comfort dimension. Because it is possible that an operator s overall comfort is related to the average score (rather than comfort on all dimensions), we also calculated a mean comfort score, where average comfort represented an unweighted mean of 4 or greater (ie, somewhat or extremely comfortable ). Residents reported lack of average comfort performing 15% of central lines placements, 15% of lumbar punctures, 16% of paracenteses, and 10% of thoracenteses. However, the average comfort may misrepresent the comfort level with particular aspects of the procedure. In fact, residents reported lack of comfort with at least one aspect of 51% of procedures and lack of comfort with at least two aspects of 21% of cases. Comfort by Training Level Stratified by training level (Table 4), PGY1 residents were less comfortable than more senior residents in performing all aspects of procedures. PGY1 residents were uncomfortable with the lack of supervision in 59% of unsupervised procedures. Among PGY1 residents, 46% were uncomfortable managing procedural complications, and 13% were uncomfortable with the equipment used for the procedure. PGY1 residents reported lack of comfort with at least one aspect of the procedure in 68% of cases, whereas PGY2 and PGY3 residents reported lack

5 Huang et al Beyond the Comfort Zone 71.e21 Table 3 Lack of comfort, by procedure All n 527 Central line n 268 Lumbar puncture n 95 Paracentesis n 81 Thoracentesis n 93 Lack of comfort with P value* Indications and 11 (2.1%) 6 (2.2%) 0 (0.0%) 0 (0.0%) 5 (6.0%).017 contraindications, n (%) Obtaining consent, n (%) 10 (1.9%) 5 (1.9%) 2 (2.1%) 2 (2.5%) 1 (1.2%).935 Anatomy, n (%) 28 (5.3%) 19 (7.1%) 3 (3.2%) 1 (1.2%) 5 (6.0%).040 Equipment, n (%) 45 (8.5%) 19 (7.1%) 5 (5.3%) 10 (12.3%) 11 (13.3%).001 Sterile technique, n (%) 7 (1.3%) 6 (2.2%) 0 (0.0%) 1 (1.2%) 0 (0.0%).494 Patient positioning, n (%) 14 (2.7%) 6 (2.2%) 4 (4.2%) 2 (2.5%) 2 (2.4%).191 Interpreting the results, n 25 (4.7%) 16 (6.0%) 2 (2.1%) 3 (3.7%) 4 (4.8%).493 (%) Being supervised, n/total 5/206 (2.4%) 3/89 (3.4%) 0/27 (0.0%) 0/25 (0.0%) 2/65 (3.1%).884 (%) Being unsupervised, 119/321 (37.1%) 70/179 (39.1%) 23/68 (33.8%) 14/56 (25.0%) 12/18 (66.7%).001 n/total (%) Managing complications, 186 (35.3%) 87 (32.5%) 32 (33.7%) 28 (34.6%) 39 (47.0%).003 n (%) Mean number of aspects with lack of comfort, n (SD, range) 0.85 (1.1, 0-8) 0.88 (1.2, 0-8) 0.75 (0.1, 0-4) 0.75 (1.0, 0-4) 0.98 (1.1, 0-6).313 SD standard deviation. *Kruskal-Wallis on the Likert scores. Based on the number of supervised procedures. Based on the number of unsupervised procedures. of comfort with at least one aspect in 29% and 39% of procedures, respectively (P.001). Factors Associated with Comfort PGY1 residents were least likely to be comfortable performing procedures, compared with the experience of PGY2 and PGY3 residents (34% vs 71% and 61%, respectively, P.001) (Table 5). Most first-time operators (80%) lacked comfort compared with residents who had performed at least one prior procedure of that type (P.001). Comfort was not associated with gender, resident type, location, urgency, or time of day. In the multivariable analysis, residents overall comfort performing a procedure (ie, at least somewhat comfortable with all aspects of the procedure) was associated with use of the medical procedure service (odds ratio [OR] 1.9, 95% confidence interval [CI] ). Factors associated with lower comfort were status as a PGY1 resident compared with PGY2 and PGY3 residents (OR 0.3, CI ), Table 4 Lack of comfort, by training level PGY1 n 283 PGY2 n 140 PGY3 n 104 Lack of comfort with P value* Indications and contraindications, 10 (4%) 0 (0%) 1 (1%).001 n (%) Obtaining consent, n (%) 6 (2%) 1 (1%) 3 (3%).001 Anatomy, n (%) 14 (5%) 5 (4%) 9 (9%).001 Equipment, n (%) 36 (13%) 4 (3%) 5 (5%).001 Sterile technique, n (%) 6 (2%) 0 (0%) 1 (1%).003 Patient positioning, n (%) 8 (3%) 3 (2%) 3 (3%).001 Being supervised, n/total (%) 3/118 (3%) 1/44 (2%) 1/44 (2%).001 Being unsupervised, n/total (%) 98/165 (59%) 13/96 (14%) 8/60 (13%).001 Interpreting the results, n (%) 20 (7%) 2 (1%) 3 (3%).001 Managing complications, n (%) 127 (45%) 29 (21%) 30 (29%).001 Mean number of aspects with lack of comfort, n (SD, range) SD standard deviation. *Kruskal-Wallis on the Likert scores. Based on the number of supervised procedures. Based on the number of unsupervised procedures. 1.2 (1.2, 0-6) 0.41 (0.7, 0-3) 0.6 (1.1, 0-8).001

6 71.e22 The American Journal of Medicine, Vol 119, No 1, January 2006 Table 5 Factors associated with comfort performing a procedure All Lack of comfort n 270 Comfort n 257 P value* Procedure, n (%).05 Central line 133 (50%) 135 (50%) Lumbar puncture 45 (47%) 50 (53%) Paracentesis 38 (47%) 43 (53%) Thoracentesis 54 (65%) 29 (35%) Training year, n (%).001 PGY1 188 (66%) 95 (34%) PGY2 41 (29%) 99 (71%) PGY3 41 (39%) 63 (61%) Gender, n (%).58 Male 126 (50%) 128 (50%) Female 144 (53%) 129 (47%) Type, n (%).24 Categorical PGY1 152 (65%) 83 (35%) Preliminary PGY1 36 (73%) 13 (27%) Location, n (%).26 Ward 84 (56%) 65 (44%) Emergency unit 38 (44%) 48 (56%) Intensive care unit 135 (50%) 135 (50%) Other 13 (59%) 9 (41%) Urgency, n (%).66 Elective 90 (52%) 84 (48%) Urgent 151 (50%) 151 (50%) Emergent 29 (57%) 22 (43%) Time of day, n (%).61 8 AM-5 PM 175 (53%) 156 (47%) 5 PM-11 PM 57 (49%) 59 (51%) 11 PM-8 AM 38 (48%) 42 (53%) Use of procedure service, n (%).07 Yes 97 (47%) 109 (53%) No 173 (54%) 148 (46%) Self-reported prior procedures, n (%) (80%) 13 (20%) (58%) 114 (42%) (34%) 82 (66%) (42%) 22 (58%) 15 4 (13%) 26 (87%) *Chi-square. At the time of the index procedure, residents reported the number of procedures performed previously. Wilcoxon rank-sum. first time performing the procedure (OR 0.4, CI ), thoracenteses in relation to other procedures (OR 0.4, CI ), and emergent procedures (OR 0.6, CI ). Comfort by Procedure Number We analyzed the relationship between self-reported procedural experience and the percentage of procedures associated with overall comfort (Figure 1). The likelihood that a procedure was associated with comfort increased with the number of prior procedures, peaking at 92% for central line placements, 83% for paracenteses, and leveling off at 67% for lumbar punctures. DISCUSSION We studied 527 procedures performed by 106 internal medicine residents at a Boston teaching hospital and found that more than half of resident physicians were uncomfortable performing at least 1 aspect of 4 common bedside procedures. First-year residents and those with the least prior experience most often said that they lacked comfort. Lack of comfort was also strongly associated with lack of supervision. In contrast, supervision by medical procedure service physicians doubled the odds that the resident was comfortable performing the procedure. We found that resident physicians were particularly uncomfortable performing thoracenteses, perhaps because these procedures were performed relatively infrequently or because of the perceived risk of pneumothorax. Finally, even with significant experience, some residents remained uncomfortable performing procedures, in particular, lumbar punctures. Our findings regarding the impact of experience on comfort are consistent with conventional wisdom; although

7 Huang et al Beyond the Comfort Zone 71.e23 Figure 1 Comfort by number of procedures performed, by procedure. The percentage of procedures associated with comfort is graphed against the number of prior procedures reported by residents, by individual procedure and in aggregate. Thoracentesis was omitted because there were few procedures in the 11 to 15 and 15 range. practice does not make perfect, it makes performance somewhat less harrowing. Experience fails to mitigate the anxiety completely. A significant number of residents remain apprehensive about their ability to perform procedures adequately and to handle complications that may arise. Of note, one half of residents are uncomfortable performing procedures even after having reported performing the minimum number of 5 required by the ABIM, a similar conclusion to that of Hicks and colleagues. 6 This finding may reflect the inherent challenge of performing invasive bedside procedures, but it may also indicate the need for more rigorous training in bedside procedures, including simulators, which are commonly used to teach bronchoscopic procedures. 11 The importance of supervision also has good face validity. Residents may be uncomfortable without supervision because they receive conflicting messages about the role of faculty physicians in clinical care. Whereas attending physicians oversee clinical decision making, residents are often left alone to perform procedures. Our definition of comfort differs from previous studies that assessed average comfort rather than its components. By using the method of previous studies, our residents reported average comfort with 85% of central lines, 85% of lumbar punctures, 84% of paracenteses, and 90% of thoracenteses. These values are comparable to earlier studies. Wickstrom and colleagues 8 reported that 81% of senior internal medicine residents were confident ( 5 or 6 on 6-point scale) performing central line placement, 87% with paracentesis, and 91% with thoracentesis. Sharp and colleagues 9 found that 72% of family practice chief residents perceived themselves to be competent performing paracenteses, 86% with lumbar punctures, 70% with thoracenteses, and 69% with central lines. We found that the use of an average comfort level underestimated residents level of comfort. In addition, earlier studies relied on recollection of past procedures and may be subject to recall bias. Furthermore, residents surveyed at the end of their training may be more confident than they were earlier in training, altering their perspective on past events. Our study design may yield a more accurate, timely, and granular representation of trainees experiences. Several studies support our findings about the contribution of experience and supervision for house staff education. Hicks and colleagues 6 showed that resident comfort varies directly with the number of previous central line placements. Our results are also consistent with studies showing that attending physicians are less vulnerable than surgical residents to procedural errors in simulated laparoscopic surgery 12 and are also less prone to prescription writing errors than junior house officers. 13 Our study is subject to several limitations. First, we may not have captured all inpatient procedures. Some residents may have neglected to complete an online form after every procedure or may have delayed entering information, resulting in recall bias. Most (87% of 63) PGY1 residents, but only half of PGY2 and PGY3 residents, recorded a procedure; the remaining residents may have performed no procedures at all. Because the evaluation instrument doubled as the procedure log, residents may have had little incentive to record procedures beyond the minimum ABIM requirements. It is also possible that we sampled residents whose documentation was most meticulous or a subset of procedure-avid residents. Our findings reflect a single teaching hospital s experience and may not generalize to other settings. Our observation that the medical procedure service increased comfort may be because of self-selection bias,

8 71.e24 The American Journal of Medicine, Vol 119, No 1, January 2006 although one hypothesizes that those inherently uncomfortable with procedures would elect the medical procedure service. In addition, we did not survey the residents regarding their future practice intent, as did Sharp et al. 9 Residents planning to enter interventional fields may be more apt to be comfortable with procedures. Finally, we did not validate residents self-reported comfort level with independent, objective assessments of performance and complications; we plan to explore this connection in future research. To improve resident physicians comfort performing bedside procedures, residency training programs should introduce formal instruction in performing bedside procedures. Interventions should target PGY1 residents, particularly during their first procedures, and teaching sessions should specifically address management of procedural complications. Educational materials should be easily accessible to learners to ensure consistent practice, and the use of multimedia may help residents visualize anatomic landmarks and gain familiarity with equipment. Supervision by faculty-level mentors should be available 24 hours per day, and in particular, thoracenteses should be closely supervised. Procedural training could be of significant benefit to trainees in specialties other than internal medicine, including surgery and obstetrics. Surgeon and author Atul Gawande observed, In surgery, as in anything else, skill and confidence are learned through experience haltingly and humiliatingly. 14 Clinicians and medical educators should recognize that special skills require special training. Failure to prepare properly the next generation of physicians does a disservice to both our patients and our future colleagues. ACKNOWLEDGMENTS The authors acknowledge Dr. Brett Kalmowitz for his assistance creating the procedure log and Dr. Mitchell Rabkin for reviewing a draft of this article. References 1. Wigton RS, Blank LL, Nicolas JA, Tape TG. Procedural skills training in internal medicine residencies. A survey of program directors. Ann Intern Med. 1989;111: Wigton RS. Training internists in procedural skills. Ann Intern Med. 1992;116: Leape LL, Brennan TA, Laird N, et al. The nature of adverse events in hospitalized patients. Results of the Harvard Medical Practice Study II. N Engl J Med. 1991;324: Nettleman MD, Bock MJ, Nelson AP, Fieselmann J. Impact of procedure-related complications on patient outcome on a general medicine service. J Gen Intern Med. 1994;9: American Board of Internal Medicine. Internal medicine policies requirements for certification, procedures required for internal medicine. Available at: Accessed on April 29, Hicks CM, Gonzalez R, Morton MT, et al. Procedural experience and comfort level in internal medicine trainees. J Gen Intern Med. 2000; 15: Sznajder JI, Zveibil FR, Bitterman H, et al. Central vein catheterization: failure and complication rates by three percutaneous approaches. Arch Intern Med. 1986;146: Wickstrom GC, Kelley DK, Keyserling TC, et al. Confidence of academic general internists and family physicians to teach ambulatory procedures. J Gen Intern Med. 2000;15: Sharp LK, Wang R, Lipsky MS. Perception of competency to perform procedures and future practice intent: a national survey of family practice residents. Acad Med. 2003;78: Smith CC, Gordon CE, Feller-Kopman D, et al. Creation of an innovative inpatient medical procedure service and a method to evaluate house staff competency. J Gen Intern Med. 2004;19: Ost D, DeRosiers A, Britt EJ, et al. Assessment of a bronchoscopy simulator. Am J Respir Crit Care Med. 2001;164: Dominguez CO, Flach FM, McKellar PD, Dunn M. Using videotaped cases to elicit perceptual expertise in laparoscopic surgery. Proceedings of the Third Annual Symposium on Human Interaction with Complex Systems, Dayton, Ohio, August 25-28, Los Alamitos (CA): IEEE Computer Society Press; Lesar TS, Briceland LL, Delcoure K et al. Medication prescribing errors in a teaching hospital. JAMA. 1990;263: Gawande A. Complications: A Surgeon s Notes on an Imperfect Science. 1st ed. New York: Metropolitan Books; 2002:18.