USING SOCIAL NETWORK ANALYSIS TO UNDERSTAND PERIOPERATIVE STAFF TOPOLOGY AT A HIGH-VOLUME SPECIALTY HOSPITAL

Purpose:

   The aim of this study was to characterize the perioperative staff topology of a high-volume specialty hospital using social network analysis.

Method:

   Staff and perioperative electronic medical record data on total hip and knee joint replacement (TJR) surgeries from April 1, 2013 to March 31, 2015 were analyzed. Individual perioperative team members for each case were identified. Five team roles were mapped: surgeon, anesthesiologist, scrub technician, circulating nurse, and surgical assistant. The data was analyzed using social network analysis software (Gephi 0.8.2) and descriptive statistics.

Result:

   16,411 total TJR cases were performed during the study period. We found 10,351 distinct teams with 512 staff members. In only 11.9% (1,953) of cases, a complete 5-member team worked together more than 3 times across the two years. In contrast, 55.2% of the core three-role teams (limited to the surgeon, circulating nurse and scrub technician) operated together four or more times during the study period. Overall, the network of teams (Figure 1) was found to be highly connected; it consisted of 1 connected component with diameter 4, radius 2, graph density 0.189, average path length 1.9, and mean edge weight 6.61 (SD=17.25). For each team member, the mean for degree was 96.66 (SD=78.62), weighted degree was 638.53 (SD=888.05), clustering coefficient was 0.59 (SD=0.18), Eigenvector centrality was 0.39 (SD=0.3), eccentricity was 2.98 (SD=0.59), closeness centrality was 0.55 (SD=0.10), and betweenness centrality was 32.07 (SD=57.92).

Conclusion:

   Most surgeries were performed by teams with a relatively stable triad of surgeon, circulator, and scrub tech, but few had stable 5-member teams.  This variability coexists with high overall connectivity: with a network diameter of 4, each person in the network was at most separated by only two people. Also, two individuals are most likely only separated by one other person (average path length of 1.9) and each person had, on average, worked with 96.7 other people over the two-year period, close to one-fifth of total staff. Our findings illustrate how social network analysis can be used to characterize the interactions of the perioperative staff as a system rather than as isolated individuals or teams. These results have important implications for perioperative staffing, but first need to be analyzed with regard to both patient outcomes and operating room efficiency.