Surgeons' and Anesthesiologists' Perceptions of Turnover Times

Abstract

BACKGROUND: 

Perception of turnovers may be influenced less by actual turnover times per se than by a mental model of factors influencing turnover times.

METHODS: 

A survey was performed at a U.S. academic hospital in 2010. Each of the 78 subjects estimated characteristics of his/her turnover times in 2009. Responses were compared with the actual times.

RESULTS: 

Numbers of comments were not proportional to actual total waiting times experienced. Surgeons with 2 or more comments (n = 10) averaged the same numbers of turnovers as did surgeons who made 1 or no comments (n = 13) (P = 0.62). Four of the 10 surgeons with 2 or more comments averaged <2 turnovers per month (“very few turnovers”). Perceptions of turnover times were influenced by opinion about team activity during shift change. Most (>79%) subjects thought that the time of the day with the subject's largest number of prolonged (>45 minutes) turnovers was at least 2 hours later than actual (P < 0.0001). Although most prolonged turnovers occurred around noon, 8 surgeons mentioned shift change qualitatively, and most (68%, P = 0.002) subjects estimated a time overlapping with shift change. Surgeons overall overestimated their observed percentage of prolonged turnovers (P = 0.020), and anesthesiologists' estimates were overall unbiased. Surgeons' bias cannot be explained by knowing times of a longer interval such as “skin to skin,” because the other surgeons, with very few turnovers, had responses that were essentially identical (P ≥ 0.87). When we corrected for each subject's actual mean turnover time, surgeons' estimates for their averages were longer than were anesthesiologists' estimates (P = 0.002). Responses were again essentially indistinguishable from those of subjects with very few turnovers (P ≥ 0.23).

CONCLUSIONS: 

Managers should not rely on surgeons or anesthesiologists for their expert judgment on turnover times. Managers should also not interpret comments about turnover times as literally referring to the time, but instead as factors perceived as contributing to the time (e.g., attitude about the facility and the activity of its personnel).

In Brief

Published ahead of print January 6, 2011

Many surgeons care about turnover times. Vitez and Macario had surgeons score the importance of 25 attributes of anesthesia groups using a scale from 0 (“no importance”) to 4 (“a factor that would make me switch groups/hospitals”).¹ The mean score was 4.0 for “ability to calmly manage a crisis.” The mean score was only slightly less (3.8) for short turnover times.

Opinion about turnover times can be nonlinearly related to their duration. Stahl and colleagues' found that surgeons working in a short-turnover-dedicated operating room (OR)² had an increased feeling of personal competence and achievement.³ Eappen and colleagues added a regional anesthesia team for orthopedics at a facility with many prolonged (>45 minutes) turnovers.⁴ Surgeons reported reduced incidence of these turnovers, but quantitative review showed no change.⁴ Eappen speculated that surgeons' perceptions of turnovers were influenced less by turnover times per se than by their seeing anesthesia providers actively caring for their patients. Similarly, when surgeons reported an increased incidence of prolonged turnovers in afternoons since the introduction of incremental compensation to anesthesiologists for work after 3:30 PM, Masursky and colleagues performed analysis and found no change.⁵ No change was observed.⁵ Thus, perception of turnovers may be influenced less by actual turnover time per se than by a mental model of how team activity influences turnover times.

We surveyed surgeons, surgery residents, anesthesiologists, and anesthesiology residents, asking them to estimate their mean turnover times, incidences of prolonged turnover times, and time of the day with the most prolonged turnovers. The last was a focus, because time of day is related to team activity (e.g., shift change).^6–8 Perceptions of turnover times were studied by comparing (a) quantitative responses to actual values; (b) quantitative responses between subjects with many turnovers versus those with very few turnovers (i.e., for whom estimates were less likely based on their actual turnovers); and (c) numbers of qualitative comments between subjects with many versus very few turnovers.

METHODS

The IRB of the State University of New York (SUNY) Upstate approved this survey and data analysis project. All subjects gave written consent for their responses to be compared with their OR information system data.

The population studied was all surgeons, anesthesiologists, anesthesia residents, and surgery residents at SUNY Upstate University Hospital. The survey was administered on paper by the first author, a PhD survey researcher. Some participants filled out the paper survey and some were interviewed, with the first author recording qualitative feedback as it was provided. She approached respondents in physician lounges, resident lounges, weekly staff meetings, and surgical suite entrances.

The surgeons were mostly interviewed. The residents and anesthesiologists mostly completed the paper survey on their own. There was room for comments after each question. Qualitative comments were transcribed and classified on the basis of content into thematic categories such as “turnover time should be shorter than it is.” The numbers of comments made by each subject were analyzed quantitatively.

The introduction and survey questions emphasized that the focus was the subject's turnover times (Table 1). For example, the first sentence was, “I want to ask you a few brief questions about turnover times for your cases at SUNY Upstate and then compare your answers to the information in the Upstate data system.” Pretesting was performed at a nearby private hospital in January and February 2010. Surveys were administered in March and April 2010.

Table 1: Comparisons Between Survey Respondents with and Without at Least 4 Turnovers for Each of the 6 Studied 2-Month Periods

From the recorded times of patients' entrance and exits from the ORs in 2009, the hospital's median turnover time was 33 minutes. The developed statistical analysis of prolonged turnovers relies on the definition being any turnover longer than 15 minutes beyond the average turnover time for the facility after excluding turnovers longer than 90 minutes.⁶ However, we aimed to investigate each subject's prolonged turnovers. Therefore, the survey defined prolonged turnovers as longer than 45 minutes, rather than 48 minutes. The potentially pejorative word prolonged was deliberately not used in the survey, just longer than 45 minutes. During data analysis, each turnover was included in analysis for the subject if the subject was present at the end of one elective (scheduled) case and at the start of the second elective (scheduled) case in the same OR on the same day. Presence was known, because all personnel in an OR for at least part of a case is documented in the OR information system.

Sequential turnover times are often correlated (e.g., due to common causes), particularly prolonged turnover times.⁶ We binned by 2-month period and required for analysis that there be at least 4 turnovers of 90 minutes or less for each of the 6 periods. Subjects not meeting this criteria in 2009 are referred to as those with “very few” turnovers. If a subject was present only at the end of a case (e.g., to assist with closure, extubation, transport), they would count as being present at the turnover. Thus, our numbers (deliberately) overestimate presence. Zero surgery residents and only 2 anesthesia residents met our inclusion criteria. Surgery residents go to other ORs, switch with another resident who starts the next case, and work at different hospitals among months. Anesthesia residents have multiple non-OR rotations. We refer to the 2 senior anesthesiology residents included in the anesthesia group (Table 1) as anesthesiologists for convenience of wording.

Confidence intervals for turnover times were calculated as validated using Monte Carlo simulation.^6,9 For each 2-month period, the mean was calculated for turnovers 90 minutes or briefer. Student's t distribution was applied to the n = 6 means. For each 2-month period, the number of prolonged (>45 minutes) and nonprolonged turnovers was calculated. After Freeman–Tukey transformation of the proportion,¹⁰ Student's t distribution was used to calculate the mean and 95% confidence interval, and then the inverse was taken.⁹

All P values are 2-sided and exact (StatXact-9, Cytel Software, Cambridge, MA). The data are reported as mean ± SD, unless otherwise stated. Associations were tested using Kendall's τ_b. Two independent group binary comparisons were tested using Wilcoxon–Mann–Whitney test. Two group comparisons paired by subject were tested using Wilcoxon signed-ranks test. Incidence (for “most”) was compared with 50% using binomial test.

RESULTS

Factors other than turnovers (e.g., attitude about the facility) influenced perception of turnovers. The responses of the 26 subjects with at least 4 turnovers every other month were essentially indistinguishable from those of the 52 subjects with very few turnovers, controlling for the subject being a surgeon or anesthesiologist (Table 1C–E).

Numbers of comments about turnovers were not proportional to total waiting time experienced. Surgeons^a with at least 2 comments averaged the same numbers of turnovers as surgeons who made 0 or 1 comment (54 ± 59 [n = 10] vs. 57 ± 76 [n = 13], P = 0.62). Four of the 10 surgeons with at least 2 comments averaged <2 turnovers per month.

The time of the day with the largest number of prolonged (>45 minutes) turnovers can be calculated with suitable confidence intervals.⁶ The time is important, because interventions to reduce turnover times involve personnel who need to be scheduled. The most common (as well as mean and median) period for the subjects' prolonged turnovers was from 11 AM to 1 PM, principally because that was when there were many turnovers. Calculation would be unnecessary if surgeons and anesthesiologists knew by experience when most of their prolonged turnovers occur. However, that was not true (Tables 1E and 2A). Most (>79%) subjects thought that the time of the day with his or her largest number of prolonged turnovers was at least 2 hours later than actual (P < 0.0001). Surgeons were more accurate than were anesthesiologists (P = 0.036, Table 2A), especially surgeons with larger caseloads (Table 3A). Nevertheless, there remained large bias (Table 2A), suggesting that the important question is why most subjects' estimates were biased toward later in the day. Two explanations for the bias can be rejected: subjects relying disproportionately on either incidence of turnovers by time of day or on mean turnover time by time of day. The incidence of turnovers of any duration was smaller for the 2-hour periods chosen in comparison with the subject's period with the largest number of prolonged turnovers (P < 0.0001; difference −24 ± 23 over the year). In addition, the mean turnover time did not differ between the chosen and actual periods with the largest number of prolonged turnovers (P = 0.58; difference −0.3 minute ± 3.6 minutes; n = 16). The n equaled only 16 subjects because the other 59 subjects all had too few turnovers to estimate the mean. Specifically, all 59 selected a period for which each had <9 turnovers over the entire year. That observation suggests an explanation: bias may be due to a mental model of factors perceived as contributing to prolonged turnovers. Eight surgeons mentioned “shift change” when answering about time of the day with the most prolonged turnovers. Quantitatively, most (68% of 75, P = 0.002) subjects estimated a time overlapping with shift change: either 1 PM to 3 PM or 3 PM to 5 PM. Because the most common period for actual prolonged turnovers was earlier, likely opinion about when most prolonged turnovers occurred was influenced by perception of team activity during shift change.

Table 2: Comparison of Survey Responses to Actual Turnover Times

Table 3: Kendall τ_b Associations Among Subject Characteristics and Differences Between Responses and Actual Turnovers

Perceptions of magnitudes of turnover times were also likely influenced by a mental model of factors influencing turnover times. Approximately half (P = 0.33) of subjects provided inaccurate estimates of their percentage of turnovers that were prolonged (Table 2B). Surgeons overall overestimated their observed percentage of prolonged turnovers (P = 0.020), whereas anesthesiologists' estimates were overall unbiased (Table 2B). Surgeons' biases cannot be explained by accurately knowing times but of a longer interval such as “skin to skin,” because surgeons and surgery residents present during very few turnovers had responses that were essentially identical (Table 1C–E).

Results suggest similar psychological processes for average turnovers. Most (>84%) subjects' estimate of mean turnover time was not within the confidence interval for his or her actual mean turnover (Table 2C). Most absolute errors exceeded 5 minutes, but not 10 minutes. When we corrected for each subject's actual mean turnover time, surgeons' estimates were larger than were anesthesiologists' estimates (P = 0.002). Responses were again essentially indistinguishable from those of subjects having very few turnovers (Table 1C). Providers overestimating mean turnover times also overestimated the percentage of turnovers that were prolonged (Table 3C). This correlation indicates reliability of responses.

DISCUSSION

Our findings show that many surgeons overestimate turnover times, and markedly overestimate the incidence of prolonged turnovers. In contrast, many anesthesiologists provide unbiased estimates of turnover times. Yet both groups' estimates are likely influenced by factors (e.g., team activity) that are perceived as contributing to turnover times. As is summarized in the second paragraph of the Introduction, our findings are consistent with previous survey research^3–5 in OR management. We recommend that OR managers not interpret comments about turnover times as literally referring to the time.

Turnovers are nonvalue-added time and their selective targeting results in reduced cost and increased throughput, making knowledge of them and their timing economically valuable.^7,11,12 On the basis of our findings, OR managers should not rely on surgeons or anesthesiologists for their expert judgment on turnovers. Furthermore, without recommendations, anesthesiologists and OR nurses do not target appropriate ORs for turnover time reduction but make decisions that maintain continuous activity of teams present.^13–16 The physician perceptions of team activity observed in the current paper may sustain that bias.¹⁷

There was absence of significant association of bias or absolute error in estimates of turnover times with estimates of surgeon or anesthesiologist workload: numbers of cases or numbers of turnovers. Higher-volume surgeons and anesthesiologists were just as inaccurate as were those who had lower volume. This finding is consistent with a previous OR management survey finding that working in surgical suites does not result in knowledge of OR management.¹³

One limitation of our study is that it was performed at one hospital. Investigation at other facilities in different health systems and cultures is warranted because our results suggest that the culture influences perception of turnover times. A second limitation is that we did not study the sensitivity of results to surgeon and anesthesiologist education or to immediate and routine on-site communication regarding their turnover times. The fact that there was no such communication in 2009, and through 2010 when our survey was performed, makes our results replicable, but perhaps limited in usefulness only to other such facilities. Third, additional studies are needed to know how to correct perceptions of turnover times and, if so, whether doing so would have value. For example, it is unknown whether increased satisfaction with team activity causes reduction in the estimate of turnover time. Alternatively, surgeon-sponsored recognitions (e.g., gift certificates to nurses) for reductions in turnover times to agreed upon targets¹¹ may result in knowledge of turnover times and, if so, satisfaction with team activity.

DISCLOSURES

Name: Danielle Masursky.

Role: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Conflict of Interest: Danielle Masursky reported no conflicts of interest.

Attestation: Danielle Masursky has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Franklin Dexter.

Role: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Conflict of Interest: Franklin Dexter reported no conflicts of interest.

Attestation: Franklin Dexter has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.

Name: Sheldon A. Isaacson.

Role: This author helped conduct the study and write the manuscript.

Conflict of Interest: Sheldon A. Isaacson reported no conflicts of interest.

Attestation: Sheldon A. Isaacson approved the final manuscript.

Name: Nancy A. Nussmeier.

Role: This author helped design the study and write the manuscript.

Conflict of Interest: Nancy A. Nussmeier reported no conflicts of interest.

Attestation: Nancy A. Nussmeier approved the final manuscript.