Patients expect their physicians to be competent. It is not at all clear, however, what competence is or how it can be best achieved. Our dictionary defines "competent" as "properly qualified" or "adequate for the purpose" [1]. Although competence is a concern throughout medicine, the attainment of competence is of particular urgency in the performance of technical procedures because of their potential for sudden, often serious injury. Patients undergoing procedures expect their physicians to be effective and efficient.

As members of the Standards of Training Committee of the American Society for Gastrointestinal Endoscopy (ASGE), we encountered first hand the difficulty in establishing standards for endoscopic training. The committee was divided on the number of procedures that a trainee should be expected (or required) to perform to attain competence. With the exception of a study on training residents to perform sigmoidoscopies [2], there was little objective evidence on which to base guidelines.

In the absence of information, the committee suggested "threshold" levels below which it would be unlikely that most trainees could attain sufficient skill and experience to be considered competent. For example, the suggested threshold for esophagogastroduodenoscopy (EGD) was set at 100, as was the threshold for colonoscopy, including at least 20 snare polypectomies. Only exceptionally gifted trainees should be considered for certification after fewer procedures. The guidelines proposed that certification should not be automatic on achievement of these thresholds. Rather, certification would require broader evaluation of actual performance, including the appropriateness of indications, recommendations based on endoscopic findings, and the recognition and management of complications.

These recommendations have now been accepted by the ASGE [3]. Because the "threshold" numbers exceeded previous recommendations by the ASGE and other training organizations [4-9], they provoked considerable criticism as being unnecessarily high. Yet a study in this issue of the Annals by Cass and colleagues provides support for these high thresholds [10]. They report the first objective information on the evolution of endoscopic competence in training gastroenterology fellows and surgical residents in the performance of EGD and colonoscopy. The number of procedures that each trainee had performed during the study period ranged from 54 to 162 EGDs (mean, 113) and 39 to 127 colonoscopies (mean, 49). Success is defined in terms of the ability to reach anatomic end points, recognition of an abnormality seen by the attending physician, and the procedure's duration. The authors conclude that "more than 100 supervised esophagogastroduodenoscopies or colonoscopies are necessary to achieve competence in gastrointestinal endoscopy." Actually, this represents a minimum figure; most of their trainees had not achieved these numbers by the end of the study and not all of those who had done so were achieving acceptable success in the extent of intubation or lesion recognition. They also documented a phenomenon familiar to endoscopy teachers, a plateauing or even a temporary decline in success rate as trainees attempt more difficult procedures. After 100 colonoscopies their trainees could reach the cecum without assistance over 80% of the time, the same as after 50. After 50 EGDs, their trainees could intubate the esophagus 95% of the time but could only do so in 75% after 100 cases. With further experience, the success rate returned to its previous high level. By comparison, an experienced endoscopist is expected to reach the cecum in 90% to 95% of colonoscopies and the duodenum in virtually 100% of EGDs in the absence of abnormal anatomy that would make these targets difficult or unattainable.

The findings of Cass and associates [10] represent an important first step in that they used criteria that were easily and objectively measured. However, competence in endoscopy assumes a combination of technical and cognitive skills. A few objective criteria, such as reaching specific anatomic landmarks (for example, the duodenum or cecum) and recognition of the presence of abnormalities, although critically important, provide at best a crude basis for a scoring system. They are by no means sufficient to establish competence, which requires correct interpretation of abnormal findings as well. In addition, the trainee must learn when endoscopy is indicated, how to minimize risk, and how to use endoscopically derived information in clinical practice.

Despite the dramatic technologic developments in endoscopy, our understanding of how endoscopic skills are acquired remains rudimentary. Technical training in any procedure involves three elements: didactic learning, supervised hands-on training, and unsupervised experience [11]. Didactic learning is brief and comprises more or less formal discussions, presentations, and the use of audio-visual aids. Although didactic training methods can accelerate the evolution of procedural skills, they cannot substitute for the hands-on experience through which technical skills, including critical hand-eye coordination, develop. During supervised hands-on training, the trainee performs tasks of increasing difficulty under expert supervision. Unsupervised experience, a critical element of learning, begins during the training period, with expert help available on an increasingly selective basis. Unsupervised experience obviously continues throughout an individual's professional career. The period of supervised hands-on training starts with "easy" procedures, gradually progressing to more difficult cases. In the "real world," endoscopists are faced with an unpredictable mix of cases, from straightforward diagnostic procedures to complex therapeutic ones. In practice, therefore, hands-on training often begins with limited segments of procedures, gradually extending to entire diagnostic procedures and more difficult portions of therapeutic procedures.

Even when competence is achieved, its persistence cannot be assumed. Some hospitals have developed quality assurance programs to monitor competence as part of their recredentialing process. Prospective data for individual endoscopists are collected regarding indications for endoscopy, extent of examination, endoscopic diagnosis, complications, and effect on therapeutic decisions. This can help identify individuals whose skills fall below the accepted level of competence either through disuse, inadequate training, or personal impairment. Sadly, few opportunities currently exist for remedial training because of limited faculty time and available resources that are largely reserved for participants in formal training programs. Research is needed to answer fundamental questions about endoscopic training and competence. However, the absence of incentives for research into questions about the achievement of competence and for teaching endoscopic procedures reflects the limited incentives—and sometimes disincentives—toward education in general. The training of endoscopists demands considerable time and effort for uncertain rewards in terms of prestige, promotion, or financial gain. It is hardly surprising, then, that research into training is regarded as "unprofitable" by aspiring academics. A related problem is that traditional funding sources do not recognize endoscopic training as a legitimate research interest. A case in point has been the troubled development of computer simulation techniques for endoscopic teaching [12-15]. These techniques offer opportunities for repetitive practice of basic hand-eye coordination skills without the need for patients to suffer through the slow learning curve. Computer simulation is also an ideal vehicle for interactive learning and progressive evaluation of skills. However, although these techniques appear promising, formal scientific evaluation has been delayed by lack of funding.

These issues are not limited to gastrointestinal endoscopy and undoubtedly apply to other procedure-oriented specialties—both medical and surgical—where trainees must develop comparable technical and cognitive skills. Progress toward practical and effective procedural training has been painfully slow, but the climate is changing. Public demand for assurances about physician competence is increasing as are questions about how many proceduralists should be trained and what volume and complexity of procedures are needed to maintain skills [16]. A number of "advanced training programs" have already been established in regional centers of excellence to assure competence in highly technical procedures, and others are evolving [17]. This all suggests that the prevailing winds may soon change direction, fostering renewed interest in how competence is best achieved.