Arthroscopic Rotator Cuff Repair: The Learning Curve Dan Guttmann, M.D., Robert D. Graham, M.D., Megan J. MacLennan, M.S., and James H. Lubowitz, M.D.
Purpose: The purpose of this study was to answer the question: How many cases are required for a surgeon to become proficient in performing arthroscopic rotator cuff repair? We hypothesize that as surgical experienced is gained, learning can be quantitatively shown by a significant decrease in operative time. Type of Study: Prospective case series. Methods: Rotator cuff repair time (RCRT) in minutes (as well as other time components comprising total surgical time) was recorded for 100 consecutive patients having arthroscopic rotator cuff repair performed by a single surgeon beginning with his first case in private practice. Mean RCRTs for consecutive blocks of 10 cases were compared. Learning is graphically represented by plotting the RCRT by case number and generating a logarithmic trend curve. A best-fit linear equation (y ⫽ mx ⫹ b) allows comparison of the initial 10 cases with the subsequent 90 cases, where m, the slope, represents the rate of decrease in RCRT (learning). Results: Mean RCRT decreased significantly (P ⬍ .05) from the first block of 10 cases to the second block of 10 cases. There were no significant changes in mean RCRT when comparing other consecutive blocks of 10 cases. The slope of the line fitting the first block of 10 cases is ⫺8.75; the slope (m) of the line fitting the subsequent 90 cases is ⫺0.23. There is no significant difference in mean RCRT when cases are stratified by tear size. Conclusions: Graphic representation of RCRT by case number generates a learning curve whereby learning is quantitatively shown as a significant decrease in operative time as surgical experience is gained. Clinical Relevance: Qualification of the learning curve for arthroscopic rotator cuff repair provides a guide for orthopaedic surgeons contemplating the expected time line for acquiring proficiency in this technique. Key Words: Rotator cuff—Shoulder—Learning curve.
H
ow many operative cases are required for a surgeon to become proficient in performing arthroscopic rotator cuff repair? The purpose of this study was to answer this question. We hypothesize that as experience is gained, learning can be quantitatively shown by a significant decrease in operative time. We illustrate this process by generating a learning curve, a graphic representation of progressive learning during successive periods of practice.1
From the Taos Orthopaedic Institute Research Foundation (D.G., M.J.M., J.H.L.) Taos, New Mexico; and Hill Country Sports Medicine (R.D.G.), San Marcos, Texas, U.S.A. Address correspondence and reprint requests to James H. Lubowitz, M.D., Taos Orthopaedic Institute Research Foundation, 1219-A Gusdorf Rd, Taos, NM 87571, U.S.A. E-mail:
[email protected] © 2005 by the Arthroscopy Association of North America 0749-8063/05/2104-3958$30.00/0 doi:10.1016/j.arthro.2004.12.006
394
We are of the opinion that (all-) arthroscopic repair of the rotator cuff could offer the advantages of less trauma to the deltoid muscle; better assessment of glenohumeral, subacromial, and rotator cuff pathology; lower risk of infection; less postoperative pain; and easier rehabilitation. The success of this method is well established.2-18 However, a disadvantage of arthroscopic rotator cuff repair may be the technical difficulty of performing the procedure. According to Gartsman et al., “. . . the advantages [of arthroscopic rotator cuff repair] must be considered against the technical difficulty of the method, which limits its application to surgeons who are skilled in both open and arthroscopic procedures on the shoulder.”11 Surgeons performing arthroscopic rotator cuff repair are required to recognize the pattern of the tear through the arthroscope, mobilize rotator cuff tissue in cases of retracted tears, accurately place suture anchors, secure the cuff
Arthroscopy: The Journal of Arthroscopic and Related Surgery, Vol 21, No 4 (April), 2005: pp 394-400
ROTATOR CUFF REPAIR LEARNING CURVE with sutures, manage multiple sutures in the confines of the subacromial space, and tie secure knots. Technical difficulty and the resulting additional surgical time may be a decisive factor for surgeons in avoiding transition from open or mini-open to the arthroscopic rotator cuff repair technique. The pros and cons of transitioning to arthroscopic rotator cuff repair have been debated.19,20 Gartsman et al.9 assume that a “learning curve” affects their results. We quantitate this learning curve as a guide for orthopaedic surgeons contemplating the expected time line for acquiring proficiency in this technique. METHODS After Institutional Review Board approval and patient informed consent, we prospectively analyzed 100 consecutive patients having arthroscopic rotator cuff repair performed by a single surgeon (D.G.) beginning with his first case in private practice. This surgeon has just completed an Orthopaedic Shoulder and Elbow Fellowship, during which he participated in 194 shoulder arthroscopy procedures, including 168 arthroscopically assisted open or mini-open rotator cuff repair procedures, but only 3 all-arthroscopic rotator cuff repair procedures. All procedures were performed on an outpatient basis with the assistance of an Orthopaedic Sports Medicine Fellow or a Physician Assistant. Indications for surgery were symptomatic full-thickness rotator cuff tears in patients who failed nonoperative management. All tears were documented by preoperative magnetic resonance imaging. Patients with additional pathology including labral tears, glenohumeral arthrosis, acromioclavicular joint disease, adhesive capsulitis, and biceps tendon tears were included in the study. Excluded were patients with the intraoperative finding of irreparable full-thickness rotator cuff tear, which we defined as a tear in which tissue friability or tear geometry precludes secure suture fixation of the rotator cuff to the greater tuberosity of the humerus. Evaluation Total surgical time was calculated as the sum of the following time components, which were separately measured and recorded (in minutes) for each patient (by the circulating nurse assigned to the case who was blinded to the purpose of the investigation): (1) general anesthesia time, which is the number of minutes from when the patient is brought into the operating room until induction of general anesthesia is com-
395
pleted; (2) positioning time, which is the number of minutes from when general anesthesia is induced until the patient is positioned in the beach-chair position; (3) preparation and draping time, which is the number of minutes from when the patient is positioned in the beach-chair position until a skin incision is made and includes patient skin preparation, draping, and arthroscopic equipment setup and troubleshooting (e.g., correcting for a fogged camera lens, faulty pump tubing, etc.); (4) glenohumeral arthroscopy time, which is the number of minutes from when a skin incision is made until the arthroscope is moved from the glenohumeral joint to the subacromial space and may include diagnostic arthroscopy and treatment of chondral, labral, biceps, capsular, or loose body pathology; (5) subacromial decompression time, which is the number of minutes from when the arthroscope is placed into the subacromial space until initiation of arthroscopic rotator cuff repair and may include both arthroscopic subacromial decompression (in all cases) and distal clavicle excision (in some cases); (6) rotator cuff repair time (RCRT), which is the number of minutes from when arthroscopic rotator cuff repair is initiated until the repair is completed; (7) and out-of-room time, which is the number of minutes from when rotator cuff repair is completed until the patient is transferred to the recovery room and includes skin incision closure, dressing and sling placement, extubation, and patient transfer. Patient age and sex were recorded. Rotator cuff tears were classified by size by the operating surgeon using a marked probe according to their greatest diameter using the system of De Orio and Cofield as small (⬍1 cm), medium (1 to 3 cm), large (3 to 5 cm), and massive (⬎5 cm).21,22 Operative Technique Surgery was performed in the beach-chair position using both interscalene block and general anesthesia. The glenohumeral joint was inspected initially and intra-articular pathology was addressed. The arthroscope was then redirected into the subacromial space and arthroscopic subacromial decompression was performed. In cases of massive rotator cuff tears where concern over tissue quality created the potential for superior humeral head migration, the coracoacromial ligament was elevated off of the periosteum of the acromion but not resected. A distal clavicle excision was performed in cases of symptomatic arthrosis. Rotator cuff tears were evaluated for repairability by grasping the tendon edges, mobilizing the tendons
396
D. GUTTMANN ET AL.
as needed, and determining the ability of the tissues to be pulled toward the greater tuberosity of the humerus using an arthroscopic grasper or traction sutures. Traction sutures were not used in all cases but were used in cases of massive and retracted tears to facilitate margin convergence23 and to determine tissue quality. Once margin convergence was achieved, sutures from the suture anchors (described later) were able to bring the rotator cuff to the anchors under no tension with the arm at the side and with the traction sutures removed. Adhesions were removed (mobilization) with an arthroscopic elevator, a motorized resector, and/or electrocautery until full excursion was possible. In cases of massive and retracted tears, intraarticular releases were also performed during the glenohumeral arthroscopy portion of the case. A cancellous bed (footprint) was created at the greater tuberosity by removing a thin layer of cortical bone with a power burr or shaver. Suture anchors (5.0 or 6.5-mm doubly preloaded Biocorkscrews; Arthrex, Naples, FL) were inserted into the footprint. As dictated by the geometry of the repair, side-to-side simple sutures facilitated margin convergence technique.23 Linear tears without greater tuberosity avulsion were repaired using only side-to-side sutures. Suture from the anchors were passed through cuff tissue using penetrating instruments or suture lassos (Arthrex). Sutures were then tied arthroscopically using a sliding knot. After repair of the rotator cuff, the shoulder was taken through a full range of motion while visualizing the integrity of the repair. If a knot broke or suture pulled through tissue, or if an anchor pulled out from osteoporotic bone, an additional suture anchor was placed. Statistical Methods Consecutive blocks of 10 cases were analyzed for mean time and standard deviation (in minutes) for total surgical time and individual time components. Mean values were compared using a 2-sample t test with 95% confidence intervals (Minitab, State College, PA). Differences were considered of statistical significance for P ⬍ .05. Figures were created using Excel (Microsoft, Redmond, WA). The primary outcome measure, RCRT, was analyzed using a logarithmic trend curve. RCRT was also analyzed using a best-fit linear equation (y ⫽ mx ⫹ b) allowing comparison of the initial 10 cases with the subsequent 90 cases where x represents case number, y represents RCRT (in minutes), and m, the slope, illustrates the
rate of decrease in RCRT as experience is gained (learning). RESULTS The 100 consecutive arthroscopic rotator cuff repair cases were collected over a 24-month period. These cases represent all operative rotator cuff repairs by this surgeon; no cases were converted to open or miniopen repair. During this period, 3 patients with the intraoperative finding of irreparable full-thickness rotator cuff tear (as defined in Methods) were excluded from the study. In 8 cases, linear tears without greater tuberosity avulsion were repaired using only side-toside sutures. In all other cases, the rotator cuff was repaired to the footprint without tension with the arm at the side. A knot or suture broke 5 times in the first block of 10 cases (case numbers 1-10), 3 times in the second block of 10 cases (case numbers 11-20), and 1 time each in the third, fourth, and sixth blocks of 10 cases. A 5.0-mm anchor pulled out one time in the second block of 10 cases; this anchor was replaced with a 6.5-mm anchor. In 7 cases (case numbers 11-15, 18, and 25), the surgical technique was modified and bioabsorbable headed-screws (Arthrex) were used instead of suture anchors. In 3 of these cases (case numbers 12, 13, and 25), it was observed that when the shoulder was taken through a range of motion while visualizing the integrity of the repair, the screw head could buttonhole through the tendon tissue resulting in loss of fixation. In cases 12 and 13, the screws were removed and replaced with anchors. In case 25, the screw could not be removed, and an anchor was placed. Aside from the technical difficulties noted above (knot or suture breakage, anchor pullout, screw buttonholing), no operative complications occurred. After case number 42, the surgical technique was additionally modified: a cannula was used in the posterior portal in cases 43-100. The posterior cannula is believed to simplify suture management in the subacromial space during arthroscopic rotator cuff repair. The mean patient age was 57 years (range, 33 to 85 years). There were 65 male and 35 female subjects. There were 7 small tears, 39 medium tears, 30 large tears, and 24 massive tears. RCRT by case number is shown in Fig 1. Mean time and standard deviation in minutes for total surgical time and individual time components for consecutive blocks of 10 cases is shown in Table 1. Mean RCRT decreased significantly (P ⫽ .01) from the first block of 10 cases (case numbers 1-10) to the
10 91-100 9 81-90 8 71-80 7 61-70 6 51-60 5 41-50 4 31-40 3 21-30 2 11-20 1 1-10
397
*Significant (P ⬍ .05) change from previous block.
The purpose of this study was to quantitate the number of operative cases required for a surgeon to
General anesthesia time 14.8 ⫾ 3.7 14.8 ⫾ 2.4 16.7 ⫾ 3.1 13.7 ⫾ 4.4 12.0 ⫾ 4.2 10.2 ⫾ 3.5 10.8 ⫾ 3.6 15.3 ⫾ 6.4 11.5 ⫾ 3.7 9.7 ⫾ 2.7 Positioning time 8.9 ⫾ 3.0 8.4 ⫾ 2.3 7.4 ⫾ 1.9 8.2 ⫾ 4.0 11.5 ⫾ 4.3 12.7 ⫾ 6.5 12.7 ⫾ 7.5 11.8 ⫾ 4.2 10.1 ⫾ 2.6 10.8 ⫾ 5.5 Preparation and draping time 13.8 ⫾ 1.9 16.8 ⫾ 4.8 19.4 ⫾ 4.1 17.1 ⫾ 3.0 23.6 ⫾ 10.2 17.5 ⫾ 9.0 15.8 ⫾ 6.1 14.0 ⫾ 4.7 17.5 ⫾ 3.7 16.0 ⫾ 5.1 Glenohumeral arthroscopy time 15.5 ⫾ 3.3 18.3 ⫾ 4.1 13.9 ⫾ 7.5 12.0 ⫾ 2.5 15.3 ⫾ 7.9 13.9 ⫾ 4.5 12.0 ⫾ 6.8 11.7 ⫾ 6.7 12.0 ⫾ 5.9 16.3 ⫾ 7.3 Subacromial decompression time 29.7 ⫾ 5.9 22.9 ⫾ 4.6* 22.5 ⫾ 8.2 23.0 ⫾ 8.1 18.6 ⫾ 6.7 17.1 ⫾ 7.5 16.0 ⫾ 4.9 11.2 ⫾ 4.0 15.0 ⫾ 6.6 12.9 ⫾ 4.4 Rotator cuff repair time 96.5 ⫾ 38.7 48.4 ⫾ 35.5* 36.6 ⫾ 25.9 58.7 ⫾ 38.4 55.8 ⫾ 38.6 30.3 ⫾ 16.0 40.0 ⫾ 27.6 30.5 ⫾ 32.9 32.7 ⫾ 21.0 34.7 ⫾ 21.9 Out-of-room time 16.0 ⫾ 3.2 17.9 ⫾ 6.3 21.5 ⫾ 8.8 17.4 ⫾ 6.3 15.7 ⫾ 7.5 14.4 ⫾ 3.7 17.2 ⫾ 3.4 14.2 ⫾ 6.3 14.8 ⫾ 3.9 14.6 ⫾ 5.9 Total surgical time 195.2 ⫾ 47.8 147.5 ⫾ 32.3* 138.9 ⫾ 24.7 150.1 ⫾ 35.7 151.6 ⫾ 48.5 116.1 ⫾ 20.5 124.5 ⫾ 31.9 109.9 ⫾ 25.8 113.6 ⫾ 29.8 115.0 ⫾ 25.8
DISCUSSION
Block Number Case Number
second block of 10 cases (case numbers 11-20). There were no significant changes in mean RCRT when comparing the other consecutive blocks of 10 cases. With regard to all other time components, there were no significant changes in mean time when comparing the consecutive blocks of 10 cases with the following exceptions: mean total surgical time decreased significantly (P ⫽ .02) from the first block of 10 cases to the second block of 10 cases. Subacromial decompression time decreased significantly (P ⫽ .01) from the first block of 10 cases to the second block of 10 cases. RCRT by case number is again illustrated in Fig 2, in which best-fit linear (as opposed to logarithmic) trend lines are separately applied to the first block of 10 cases and the subsequent 90 cases. The slope (m) of the line fitting the first block of 10 cases is ⫺8.75 (rapid rate of decrease in RCRT [learning]); the slope (m) of the line fitting the subsequent 90 cases is ⫺0.23 (less rapid rate of decrease in RCRT [learning]). There is no significant difference in mean RCRT when cases are stratified by tear size (small v medium, P ⫽ .48; small v large, P ⫽ .69; small v massive, P ⫽ .93; medium v large, P ⫽ .06; medium v massive, P ⫽ .27; large v massive, P ⫽ .42 [Fig 3]). There is no significant difference in mean RCRT when bioabsorbable headed-screw cases are compared with all other cases (P ⫽ .12). There is no significant difference in mean RCRT when the side-to-side suture cases are compared with all other cases (P ⫽ .17).
TABLE 1.
FIGURE 1. RCRT in minutes by case number. Individual data points are connected by straight lines and a logarithmic (best-fit) trend curve.
Mean Time and Standard Deviation in Minutes for Individual Time Components and Total Surgical Time by Consecutive Blocks of 10 Cases
ROTATOR CUFF REPAIR LEARNING CURVE
398
D. GUTTMANN ET AL.
FIGURE 2. RCRT in minutes by case number. Individual data points are connected by best-fit linear trend lines for the first block of 10 cases and the subsequent block of 90 cases.
become proficient in performing arthroscopic rotator cuff repair. Using RCRT as our primary outcome measure, we determined that rapid learning occurred during the first 10 cases (Fig 2). In addition, RCRT was significantly slower when comparing the first block of 10 cases with the second block of 10 cases (Table 1). Learning continued (at a less rapid rate) throughout the remainder of the study (Figs 1 and 2). Our results support the hypothesis that as surgical experience is gained, learning may be shown as a quantitative decrease in operative time. Learning curves have been quantified and graphically illustrated in industrial fields since the 1930s, when the Boeing Company of Seattle, WA, plotted production cost by units of production.24 However, “the statistical methods used for assessing learning effects in health technology assessment have been crude and the reporting of studies poor.”25 We are unaware of previous illustrations of learning curves in the medical literature. However, various orthopaedic, laparoscopic, and endovascular studies do discuss learning.26-32 In addition to surgery time, outcome measures include radiographic appearance of orthopaedic implants, accuracy of orthopaedic screw placement, video analysis of motor skills, conversion rate to open procedures, complications, mortality, morbidity, length of intensive care unit and hospital stay, blood loss, contrast load, fluoroscopy time, radiation exposure, number of images acquired, and finally, clinical outcome measures such as pain ratings or outcome scores. A limitation of our study is that the primary outcome measure, RCRT, does not determine clinical outcome. Faster surgery may represent learning; however, faster surgery does not represent better surgery.
Twenty-four month clinical outcome data will be collected for future evaluation of study patients. These data, when evaluated by case number, may represent another illustration of learning. Other limitations of the study require consideration. The greatest limitation is susceptibility bias. Patients with different tear patterns, tear sizes, and repair techniques may have different prognoses with regard to RCRT (even in the absence of significant differences in RCRT when comparing bioabsorbable headedscrew cases with all other cases, when comparing side-to-side suture cases with all other cases, or when stratifying by tear size). In addition, although performance bias is minimized because all cases were performed by a single surgeon, diverse surgical first assistants and operating room teams may effect RCRT. In our opinion, diverse tear patterns, tear sizes, and surgical teams reflect the real world of surgical practice and the real world of surgical learning. By prospectively including consecutive cases of all patients with full-thickness rotator cuff tears treated operatively by a single surgeon, systematic introduction of study bias is minimized. With regard to modification of the surgical technique, our opinion is that this too reflects the real world of surgical learning. Some modifications (use of a cannula in the posterior portal) seem to be for the better and some (use of bioabsorbable headed-screws instead of Biocorkscrew suture anchors) for the worse. Nevertheless, susceptibility bias and performance bias are limitations of this analysis. Transfer bias is minimal as all data are collected on all patients at the time of the procedure. Recording bias is minimal as the data were collected by a circulating nurse who had no knowledge of the purpose of
FIGURE 3. size.
Mean RCRT and standard deviation in minutes by tear
ROTATOR CUFF REPAIR LEARNING CURVE the study. Reporting bias is minimal as time in minutes is a standard outcome measure used in the scientific literature. However, reporting bias does exist; because we are unaware of previous representations of learning curves in the medical literature, it is difficult to compare our methods or results with those of other investigations. Thus, an additional limitation of the study is that the size of the consecutive blocks of cases analyzed is arbitrary. Block size of 10 cases was selected because 100 cases divide neatly into 10 blocks of 10. In addition, block size of 10 cases was selected because gross visual inspection of the logarithmic (bestfit) trend curve (Fig 1) suggests a notable change in the rate of learning after case 10. Other sizes of consecutive blocks of cases could be considered. A limitation of our statistical methods is that correction for multiple testing was not performed. Regarding the primary outcome measure (RCRT), consideration of 10 blocks of 10 cases results in 9 comparisons. The Bonferroni correction for 10 comparisons would have us divide alpha (0.05) by 9 such that differences would be considered of statistical significance when P ⬍ .0055. Thus, the decrease in RCRT from the first block of 10 cases to the second block of 10 cases would not be considered significant (P ⫽ .01). In the opinion of the authors, this correction is neither customary nor indicated; such stringent adjustment of the alpha level comes at the expense of reduction of the statistical power of the study and increases the risk that true differences will be rejected (type II error).33,34 Our results may contribute to a discussion in the orthopaedic literature with regard to tear size. Although some suggest that arthroscopic rotator cuff repair is easiest in cases of small or medium tears,11,35 our data and experience support the philosophy outlined by Gartsman et al. that “The size of the tear is not the factor; we have more difficulty repairing chronic retracted tears, regardless of size, than we do repairing large or massive mobile tears.”11 The purpose of this report is not to describe the actual time required to perform arthroscopic rotator cuff repair. RCRT may vary among surgeons. Nevertheless, mean RCRT for the last block of 10 cases was 35 minutes. To allow comparison with published literature, we note that mean time from skin incision to skin closure for the last block of 10 cases was 69 minutes. This may be compared with the mean arthroscopic rotator cuff repair operative time of 56 minutes reported by Gartsman et al.11 There is “substantial variability” in the number of
399
repetitions estimated as necessary to achieve proficiency in arthroscopic surgical procedures. 36 We generate a learning curve that serves as a guide for arthroscopic surgeons contemplating the expected timeline for acquiring proficiency in arthroscopic rotator cuff repair. This curve reflects one surgeon’s experience. Surgeons with different experience could have a different learning curve. CONCLUSION Graphic representation of RCRT by case number generates a learning curve whereby learning is quantitatively shown as a significant decrease in operative time as surgical experienced is gained. REFERENCES 1. Horowitz LM. Learning. World Book Online Americas Edition. Available at www.worldbookonline.com/ar?/na/arar317040. htm. Accessed February 3, 2003. 2. Bennett WF. Arthroscopic repair of full-thickness supraspinatus tears (small-to-medium): A prospective study with 2- to 4-year follow-up. Arthroscopy 2003;19:249-256. 3. Bennett WF. Arthroscopic repair of massive rotator cuff tears: A prospective cohort with 2- to 4-year follow-up. Arthroscopy 2003;19:380-390. 4. Burkhart SS. A stepwise approach to arthroscopic rotator cuff repair based on biomechanical principles. Arthroscopy 2000; 16:82-90. 5. Burkhart SS. Arthroscopic treatment of massive rotator cuff tears. Clin Orthop 2001;390:107-18. 6. Burkhart SS, Danaceau SM, Pearce CE. Arthroscopic rotator cuff repair: Analysis of results by tear size and by repair technique—Margin convergence versus direct tendon-to-bone repair. Arthroscopy 2001;17:905-912. 7. Gartsman GM. Arthroscopic management of rotator cuff disease. J Am Acad Orthop Soc 1998;6:259-266. 8. Gartsman GM. All arthroscopic rotator cuff repairs. Orthop Clin North Am 2001;32:501-510. 9. Gartsman GM, Brinker MR, Khan M. Early effectiveness of arthroscopic repair for full-thickness tears of the rotator cuff: An outcome analysis. J Bone Joint Surg Am 1998;80:33-40. 10. Gartsman GM, Hasan SS. What is new in shoulder and elbow surgery? Specialty update. J Bone Joint Surg Am 2003;85:171154. 11. Gartsman GM, Khan M, Hammerman SM. Arthroscopic repair of full-thickness tears of the rotator cuff. J Bone Joint Surg Am 1998;80:832-840. 12. Lo IK, Burkhart SS. Spotlight on surgical techniques. Current concepts in arthroscopic rotator cuff repair. Am J Sports Med 2003;31:308-324. 13. Lyons TR, Savoie FH, Field LD. Arthroscopic repair of partial-thickness tears of the rotator cuff. Arthroscopy 2001;17: 219-223. 14. Murray TF Jr, Snyder SJ. Arthroscopic repair of medium to large full-thickness rotator cuff tears: Outcome of two to six year follow-up. J Shoulder Elbow Surg 2002;11:19-24. 15. Severund EL, Ruotolo C, Abbott DD, Nottage WM. Allarthroscopic versus mini-open rotator cuff repair: A long-term retrospective outcome comparison. Arthroscopy 2003;19:234238.
400
D. GUTTMANN ET AL.
16. Stollsteimer GT, Savoie FH. Arthroscopic rotator cuff repair: Current indications, limitations, techniques, and results. Instr Course Lect 1998;47:59-65. 17. Tauro JC. Arthroscopic rotator cuff repair: Analysis of technique and results at two and three-year follow-up. Arthroscopy 1998;14:45-51. 18. Wilson F, Hinov V, Adams G. Arthroscopic repair of fullthickness tears of the rotator cuff: Two to fourteen year followup. Arthroscopy 2002;18:136-144. 19. Nottage W, Severud E. A comparison of all arthroscopic versus mini-open rotator cuff repair: Results at 45 months. Proceedings of the Summer Institute. Rosemont, IL: American Academy of Orthopaedic Surgeons, 2001. 20. Yamaguchi K, Levine WN, Marra G, Galatz LM, Klepps S, Flatow E. Transitioning to arthroscopic rotator cuff repair: The pros and cons. J Bone Joint Surg Am 2003;85:144-155. 21. Ciepiela MD, Burkhead WZ. Classification of rotator cuff tears. In: Burkhead WZ, ed. Rotator cuff disorders. Baltimore: Williams & Wilkins, 1996. 22. DeOrio JK, Cofield RH. Results of a second attempt at surgical repair of a failed initial rotator cuff repair. J Bone Joint Surg Am 1984;66:563-567. 23. Burkhart SS. Margin convergence: A method of reducing strain in massive rotator cuff tears. Arthroscopy 1996;12:335358. 24. Anthes GH. The learning curve. Computerworld 2001. Available at www.computerworld.com. Accessed July 2, 2003. 25. Ramsay CR, Grant AM, Wallace SA, Garthwaite PH, Monk AF, Russell IT. Statistical assessment of the learning curves of health technologies. Health Technol Assess 2001;5:1-79. 26. Callaghan JJ, Heekin RD, Savory CG, Dysart SH, Hopkinson WJ. Evaluation of the learning curve associated with uncemented primary porous-coated anatomic total hip arthroplasty. Clin Orthop 1992;282:132-44. 27. Pardiwala D, Prabhu V, Dudhniwala G, Katre R. The AO
28. 29.
30.
31.
32.
33. 34. 35.
36.
distal locking aiming device: An evaluation of efficacy and learning curve. Injury 2001;32:713-718. Bagdasarian RW, Bolton JS, Bowen JC, Fuhrman GM, Richardson WS. Steep learning curve of laparoscopic splenectomy. J Laparoendosc Adv Surg Tech 2000;10:319-323. Chan SW, Hensman C, Waxman BP, Blamey S, Cox J, Farrell K, Fox J, Gribbin J, Layani L. Technical developments and a team approach leads to an improved outcome: Lessons learned implementing laparoscopic splenectomy. ANZ J Surg 2002;72: 523-527. Lee WA, Wolf YG, Hill BB, Cipriano P, Fogarty TJ, Zarins CK. The first 150 endovascular AAA repairs at a single institution: How steep is the learning curve? J Endovasc Ther 2002;9:269-276. Lobato AC, Rodriquez-Lopez J, Diethrich EB. Learning curve for endovascular abdominal aortic aneurysm repair: Evaluation of a 277-patient single-center experience. J Endovasc Ther 2002;9:262-268. Rosen J, Solazzo M, Hannaford B, Sinanan ML. Task decomposition of laparoscopic surgery for objective evaluation of surgical residents’ learning curve using hidden Markov model. Comput Aided Surg 2002;7:46-61. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology 1990;1:43-46. Bhandari M, Whang W, Kuo JC, Devereaux, PJ, Sprague S, Tornetta P. The risk of false-positive results in orthopaedic surgical trials. Clin Orthop 2003;413:63-69. Iannotti JP, Bernot MP, Kuhlman JR, Kelly MJ, Williams GR. Postoperative assessment of shoulder function: A prospective study of full-thickness rotator cuff tears. J Shoulder Elbow Surg 1996;5:449-457. O’Neill PJ, Cosgarea AJ, Freedman JA, Queale WS, McFarland EG. Arthroscopic proficiency: A survey of orthopaedic sports medicine fellowship directors and orthopaedic surgery department chairs. Arthroscopy 2002;18:795-800.