Materials and Methods

To establish acceptable and unacceptable success rates for fetoscopic laser photocoagulation (FLP), we performed a systematic review of published articles of consecutive case series ( Appendix 1 ). A retrospective study was conducted from the review of our online database of cases that underwent FLP from June 2005 through September 2006 at University of North Carolina (UNC) at Chapel Hill and from November 2006 through August 2009 at the Texas Children’s Fetal Center (TCFC) in Houston, TX. The study was approved by institutional review boards of both institutions. Fetal team members prospectively collected patient outcome data by direct patient contact and by contact with the referring physician.

There were 3 operators in the group. Operator I (O-I) was a board-certified maternal-fetal medicine specialist who was in active clinical and academic practice for >25 years with an equivalent number of years of experience in ultrasound-guided procedures. O-I was the first to learn the procedure in the group after an observation period in laser centers in Paris, France, and Leuven, Belgium, for 1 month. Twelve live cases were observed between the 2 units, along with multiple hours of reviewing videotapes from previous cases. In addition, this operator visited a laser center in Tampa, FL, for a 2-day course on diagnosis and management of TTTS. The course included a series of lectures and videotape presentations on selective fetoscopic laser photocoagulation in the treatment of TTTS. O-I performed all procedures at UNC as the primary surgeon after an institutional review board approved performing the procedure. The patients were counseled about the experience of the surgeon during the consenting process for the procedure. Operator II (O-II) was a maternal-fetal medicine specialist in active clinical and academic practice for >25 years with an equivalent number of years of experience in ultrasound-guided procedures. O-II assisted O-I for FLP by providing ultrasound guidance during the procedure for the first 30 cases. O-II started performing the laser procedure at TCFC as the primary surgeon, initially performing only cases with a posterior placenta and then advancing to cases with an anterior placenta. O-I assisted in all cases during O-II’s learning phase. Operator III (O-III) was a graduate from an American Board of Obstetrics and Gynecology–approved obstetrics and gynecology residency program and had performed >300 laparoscopic gynecologic procedures. O-III joined the fetal intervention team in July 2008 as a clinical fellow and observed 10 cases and assisted in the next 15 cases, before starting to perform FLP as the primary surgeon. In the beginning, O-III performed FLP on cases with posterior placentation (n = 15) and gradually transitioned to perform the procedure on cases with anterior placentation. O-I or O-II participated in all of O-III’s cases and in the majority of cases both were present in the operating suite when the fellow was performing the procedure. The details and the evaluation of the TTTS patients and the procedure at our center have been previously described.

The data for maternal demographics and for preoperative, intraoperative, and postoperative variables were extracted to software (SPSS version 11.0; SPSS Inc, Chicago, IL). Comparison of parameters was performed using the χ ² test for categorical variables and Fisher’s exact test when an expected frequency was <5. Analysis of variance was performed for continuous variables. Kruskal-Wallis test was used to compare nonparametric variables among the 3 operators. A P value < .05 was considered as significant. The data needed for calculating LC-CUSUM and CUSUM included surgeon’s identity, consecutive case series number for each surgeon, and number of surviving fetuses to birth for each pregnancy. These data were extracted to software (R-Software; R Foundation for Statistical Computing, Vienna, Austria).

The LC-CUSUM was developed to determine when a trainee has reached a predefined level of performance. The LC-CUSUM sequentially tests the null hypothesis “performance is unacceptable” against the alternative “performance is acceptable.” It computes a score from the successive outcomes, with successes yielding an increase in the score and failures yielding a decrease in the score. Once the score reaches a predefined limit (h), the test rejects the null hypothesis in favor of the alternative and performance is deemed acceptable.

Graphically the LC-CUSUM score is plotted on the y-axis against the successive procedures on the x-axis. As long as the score remains in the continuation region, namely between the x-axis and the decision limit h, performance cannot be considered as acceptable and monitoring continues. With accumulation of successes the score increases until it crosses the limit h where proficiency is declared. A particular feature of the LC-CUSUM is that it incorporates a holding barrier at 0 that cannot be crossed and the score St thus remains at 0 if the trainee accumulates numerous successive failures. In this way, the LC-CUSUM remains responsive at all times and if, for instance, poor performance resulted from poor technique, with improvement in skills the trainee will not have to compensate unnecessarily for all the accumulated failures and may be able to show acceptable performance in due course ( Appendix 1 ). For the LC-CUSUM for at least 1 survivor (ALOS), a limit of h = 0.95 was chosen so that the risk to declare an operator proficient when his or her performance was unacceptable was limited to 16% over 75 procedures and the risk not to declare a trainee proficient although his or her performance was acceptable was 21%.

Once an operator demonstrates competency, his or her performance is monitored with a CUSUM test. The CUSUM sequentially tests the null hypothesis “performance is acceptable” against the alternative “performance is unacceptable.” Therefore the CUSUM test, as opposed to the LC-CUSUM, is designed to detect when performance deviates from the acceptable level. Graphically, the CUSUM score increases with summation of failures until it crosses the limit h (which is usually not equal to that of the LC-CUSUM) where inadequate performance is declared. In the present study, the CUSUM was used after the operators had shown proficiency to ensure that the performance was maintained at an acceptable level. For the CUSUM, a limit h = 3.75 was chosen so that the risk to declare unacceptable performance when performance is in fact acceptable was limited to 5% over 100 procedures and the risk not to declare unacceptable performance, when it was indeed unacceptable, was limited to 20% over 100 procedures.