PURPOSE

Benchmark statistics are used in quality assurance/quality improvement processes. The purposes of the present report are to 1) review the rationale for a new specialty-specific benchmark study, 2) summarize the methods to create a practice-based research collaborative (P-BRC) designed for collecting data to create benchmarks, and 3) describe the characteristics of the P-BRC surgeon participants.

MATERIALS AND METHODS

The study was designed as a prospective cohort study. We created a P-BRC composed of randomly selected American Association of Oral and Maxillofacial Surgeons (AAOMS) members in private practice in the United States, who agreed to enroll patients scheduled to receive anesthesia of any type in the office-based ambulatory setting. The study variables included clinician demographics and their P-BRC status, grouped as 1) invited, active participants, 2) invited, inactive participants, and 3) uninvited AAOMS members. The P-BRC participants collected data for dozens of variables from their patients related to anesthesia. If the procedure was third molar (M3) surgery, additional M3 procedure-specific data were collected. Data analyses were composed of computing descriptive and bivariate statistics. Preliminary sample size estimates suggested that the P-BRC should include 300 surgeons to produce estimates with a ±5% error.

RESULTS

During the 1-year study interval, 642 surgeons (11.8%) were invited to join the P-BRC from a population of 5,455 eligible AAOMS members. The 124 active participants in the P-BRC contributed 6,344 subjects to the anesthesia data set and 2,978 subjects who had had 9,207 M3s removed to the M3 data set. The active participants in the P-BRC were younger and more likely to be board-certified than were the inactive participants (P < .05). Details of the anesthesia and M3 variables will follow in future reports.

CONCLUSIONS

Despite vigorous efforts, we did not achieve our stated goal of creating a P-BRC composed of a random sample of 300 AAOMS members. With the current P-BRC sample, variables with very high (>93%) or very low (<7%) frequency estimates will produce estimates with the desired range of ±5% error. The P-BRC includes a sample of self-selected, not random, participants and is well-characterized in terms of age, gender, board-certification status, academic degrees, and geographic distribution.