Yvonne N. Boudreau, MS

Setting and Study Design

The study was conducted at Northwestern Memorial Hospital (NMH), a 920‐bed tertiary care teaching hospital in Chicago, IL, and was deemed exempt by the Institutional Review Board of Northwestern University. General medical patients were admitted to 1 of 6 units based on bed availability. Five of the medical units consisted of 30 beds, and 1 unit consisted of 23. Each unit was equipped with continuous cardiac telemetry monitoring. Three units were staffed by teaching service physician teams consisting of 1 attending, 1 resident, and 1 or 2 interns. The other 3 units were staffed by hospitalists without the assistance of resident or intern physicians. As a result of a prior intervention, physicians' patients were localized to specific units in an effort to improve communication practices among nurses and physicians.20

Beginning in March 2010, all general medical units held SIDR each weekday morning. SIDR took place in the unit conference room, was expected to last approximately 3040 minutes, and was co‐led by the unit nurse manager and a medical director. Unit nurse managers and medical directors received specific training for their roles, including 3 hours of simulation‐based exercises designed to enhance their skills in facilitating discussion during SIDR. All nurses and physicians caring for patients on the unit, as well as the pharmacist, social worker, and case manager assigned to the unit, attended SIDR. Attendees used a structured communication tool to review patients admitted in the previous 24 hours. The plan of care for other patients was also discussed in SIDR, but without the use of the structured communication tool.

Importantly, nurse management underwent restructuring in the summer of 2011. Nurse managers, who had previously been responsible for overseeing all nursing activities on a single unit, were redeployed to be responsible for specific activities across 34 units. This restructuring made it very difficult for nurse managers to colead SIDR. As a result, the unit nurse clinical coordinator assumed coleadership of SIDR with the unit medical director. Nurse clinical coordinators worked every weekday and did not have patient care responsibilities while on duty. In addition to their role in coleading SIDR, nurse clinical coordinators addressed daily staffing and scheduling challenges and other short‐term patient care needs.

Teamwork Assessment

We adapted the Observational Teamwork Assessment for Surgery (OTAS) tool, a behaviorally anchored rating scale shown to be reliable and valid in surgical settings.2123 The OTAS tool provides scores ranging from 0 to 6 (0 = problematic behavior; 3 = team function neither hindered nor enhanced by behavior; 6 = exemplary behavior) across 5 domains (communication, coordination, cooperation/backup behavior, leadership, and monitoring/situational awareness) and for prespecified subteams. We defined domains as described by the researchers who developed OTAS. Communication referred to the quality and the quantity of information exchanged by team members. Coordination referred to management and timing of activities and tasks. Cooperation and backup behavior referred to assistance provided among members of the team, supporting others and correcting errors. Leadership referred to provision of directions, assertiveness, and support among team members. Monitoring and situational awareness referred to team observation and awareness of ongoing processes. We defined subteams for each group of professionals expected to attend SIDR. Specifically, subteams included physicians, nurses, social work‐case management (SW‐CM), pharmacy, and coleaders. We combined social work and case management because these professionals have similar patient care activities. Similarly, we combined unit medical directors and nurse clinical coordinators as coleaders. By providing data on teamwork performance within specific domains and for specific subteams, the OTAS instrument helps identify factors influencing overall teamwork performance. We modified OTAS anchors to reflect behaviors during SIDR. Anchors assisted observers in their rating of teamwork behaviors during SIDR. For example, an anchor for exemplary physician communication behavior was listens actively to other team members (looks at other team members, nods, etc). An anchor for exemplary physician leadership was assigns responsibility for task completion when appropriate.

Two researchers conducted unannounced direct observations of SIDRs. One researcher (Y.N.B) was a medical librarian with previous experience conducting observational research. The other researcher (A.J.C.) had observed 170 prior SIDRs as part of a related study. Both researchers observed 10 SIDRs to practice data collection and to inform minor revisions of the anchors. We aimed to conduct 78 independent observations for each unit, and 20 joint observations to assess inter‐rater reliability. All subteams were scored for each domain. For example, all subteams received leadership domain scores because all team members exhibit leadership behaviors, depending on the situation. In addition to teamwork scores, observers recorded the number of patients on the unit, the number of patients discussed during SIDR, attendance by subteam members, and the duration of SIDR. For the SW‐CM and coleader subteams, we documented presence if one of the subteam members was present for each patients' discussion. For example, we recorded present for SW‐CM if the social worker was in attendance but the case manager was not.

Data Analysis

We calculated descriptive statistics to characterize SIDRs. We used Spearman's rank correlation coefficients to assess inter‐rater reliability for joint observations. Spearman's rank correlation is a nonparametric test of association and appropriate for assessing agreement between observers when using data that is not normally distributed. Spearman rho values range from 1 to 1, with 1 signifying perfect inverse correlation, 0 signifying no correlation, and 1 signifying perfect correlation. We used the MannWhitney U test to assess for differences in overall team scores between services (teaching vs nonteaching hospitalist service) and KruskalWallis tests to assess for differences across units, domains, and subteams. The Kruskal‐Wallis test is a nonparametric test appropriate for comparing three or more independent samples in which the outcome is not normally distributed. We used a t test to assess for difference in duration by service, and Spearman rank correlation to assess for correlation between time spent in discussion per patient and overall team score. All analyses were conducted using Stata version 11.0 (College Station, TX).

SIDR Characteristics

We performed 7 direct observations of SIDR for 4 units, and 8 observations for 2 units (44 total observations). Units were at 99% capacity, and SIDR attendees discussed 98% of patients on the unit. Attendance exceeded 98% for each subteam (physicians, nurses, SW‐CM, pharmacy, and coleaders). SIDR lasted a mean 41.4 11.1 minutes, with a mean 1.5 0.4 minutes spent in discussion per patient. SIDR was significantly longer in duration on teaching service units compared to the nonteaching hospitalist service units (1.7 0.3 vs 1.3 0.4 minutes per patient; P < 0.001).

Inter‐Rater Reliability

Inter‐rater reliability across unit level scores was excellent (rho = 0.75). As shown in Table 1, inter‐rater reliability across domains was good (rho = 0.530.68). Inter‐rater reliability across subteams was good to excellent (rho = 0.530.76) with the exception of the physician subteam, for which it was poor (rho = 0.35).

Assessment of Teamwork by Unit, Domain, and Subteams

Teaching and nonteaching hospitalist units had similar team scores (median [interquartile range (IQR)] = 5.2 [1.0] vs 5.2 [0.4]; P = 0.55). We found significant differences in teamwork scores across units and domains, and found differences of borderline statistical significance across subteams (see Table 2). For unit teamwork scores, the median (IQR) was 4.4 (3.94.9) for the lowest and 5.4 (5.35.5) for the highest performing unit (P = 0.008). Across domain scores, leadership received the lowest score (median [IQR] = 5.0 [4.65.3]), and cooperation/backup behavior and monitoring/situational awareness received the highest scores (median [IQR]) = 5.4 [5.05.5] and 5.4 [5.05.7]; P = 0.02). Subteam scores ranged from a median (IQR) 5.0 (4.45.8) for coleaders to 5.5 (5.05.8) for SW‐CM (P = 0.05). We found no relationship between unit teamwork score and time spent in discussion per patient (rho = 0.04; P = 0.79).

Setting and Study Design

The study was conducted at Northwestern Memorial Hospital (NMH), a 920‐bed tertiary care teaching hospital in Chicago, IL, and was deemed exempt by the Institutional Review Board of Northwestern University. General medical patients were admitted to 1 of 6 units based on bed availability. Five of the medical units consisted of 30 beds, and 1 unit consisted of 23. Each unit was equipped with continuous cardiac telemetry monitoring. Three units were staffed by teaching service physician teams consisting of 1 attending, 1 resident, and 1 or 2 interns. The other 3 units were staffed by hospitalists without the assistance of resident or intern physicians. As a result of a prior intervention, physicians' patients were localized to specific units in an effort to improve communication practices among nurses and physicians.20

Beginning in March 2010, all general medical units held SIDR each weekday morning. SIDR took place in the unit conference room, was expected to last approximately 3040 minutes, and was co‐led by the unit nurse manager and a medical director. Unit nurse managers and medical directors received specific training for their roles, including 3 hours of simulation‐based exercises designed to enhance their skills in facilitating discussion during SIDR. All nurses and physicians caring for patients on the unit, as well as the pharmacist, social worker, and case manager assigned to the unit, attended SIDR. Attendees used a structured communication tool to review patients admitted in the previous 24 hours. The plan of care for other patients was also discussed in SIDR, but without the use of the structured communication tool.

Importantly, nurse management underwent restructuring in the summer of 2011. Nurse managers, who had previously been responsible for overseeing all nursing activities on a single unit, were redeployed to be responsible for specific activities across 34 units. This restructuring made it very difficult for nurse managers to colead SIDR. As a result, the unit nurse clinical coordinator assumed coleadership of SIDR with the unit medical director. Nurse clinical coordinators worked every weekday and did not have patient care responsibilities while on duty. In addition to their role in coleading SIDR, nurse clinical coordinators addressed daily staffing and scheduling challenges and other short‐term patient care needs.

Teamwork Assessment

We adapted the Observational Teamwork Assessment for Surgery (OTAS) tool, a behaviorally anchored rating scale shown to be reliable and valid in surgical settings.2123 The OTAS tool provides scores ranging from 0 to 6 (0 = problematic behavior; 3 = team function neither hindered nor enhanced by behavior; 6 = exemplary behavior) across 5 domains (communication, coordination, cooperation/backup behavior, leadership, and monitoring/situational awareness) and for prespecified subteams. We defined domains as described by the researchers who developed OTAS. Communication referred to the quality and the quantity of information exchanged by team members. Coordination referred to management and timing of activities and tasks. Cooperation and backup behavior referred to assistance provided among members of the team, supporting others and correcting errors. Leadership referred to provision of directions, assertiveness, and support among team members. Monitoring and situational awareness referred to team observation and awareness of ongoing processes. We defined subteams for each group of professionals expected to attend SIDR. Specifically, subteams included physicians, nurses, social work‐case management (SW‐CM), pharmacy, and coleaders. We combined social work and case management because these professionals have similar patient care activities. Similarly, we combined unit medical directors and nurse clinical coordinators as coleaders. By providing data on teamwork performance within specific domains and for specific subteams, the OTAS instrument helps identify factors influencing overall teamwork performance. We modified OTAS anchors to reflect behaviors during SIDR. Anchors assisted observers in their rating of teamwork behaviors during SIDR. For example, an anchor for exemplary physician communication behavior was listens actively to other team members (looks at other team members, nods, etc). An anchor for exemplary physician leadership was assigns responsibility for task completion when appropriate.

Two researchers conducted unannounced direct observations of SIDRs. One researcher (Y.N.B) was a medical librarian with previous experience conducting observational research. The other researcher (A.J.C.) had observed 170 prior SIDRs as part of a related study. Both researchers observed 10 SIDRs to practice data collection and to inform minor revisions of the anchors. We aimed to conduct 78 independent observations for each unit, and 20 joint observations to assess inter‐rater reliability. All subteams were scored for each domain. For example, all subteams received leadership domain scores because all team members exhibit leadership behaviors, depending on the situation. In addition to teamwork scores, observers recorded the number of patients on the unit, the number of patients discussed during SIDR, attendance by subteam members, and the duration of SIDR. For the SW‐CM and coleader subteams, we documented presence if one of the subteam members was present for each patients' discussion. For example, we recorded present for SW‐CM if the social worker was in attendance but the case manager was not.

Data Analysis

We calculated descriptive statistics to characterize SIDRs. We used Spearman's rank correlation coefficients to assess inter‐rater reliability for joint observations. Spearman's rank correlation is a nonparametric test of association and appropriate for assessing agreement between observers when using data that is not normally distributed. Spearman rho values range from 1 to 1, with 1 signifying perfect inverse correlation, 0 signifying no correlation, and 1 signifying perfect correlation. We used the MannWhitney U test to assess for differences in overall team scores between services (teaching vs nonteaching hospitalist service) and KruskalWallis tests to assess for differences across units, domains, and subteams. The Kruskal‐Wallis test is a nonparametric test appropriate for comparing three or more independent samples in which the outcome is not normally distributed. We used a t test to assess for difference in duration by service, and Spearman rank correlation to assess for correlation between time spent in discussion per patient and overall team score. All analyses were conducted using Stata version 11.0 (College Station, TX).

SIDR Characteristics

We performed 7 direct observations of SIDR for 4 units, and 8 observations for 2 units (44 total observations). Units were at 99% capacity, and SIDR attendees discussed 98% of patients on the unit. Attendance exceeded 98% for each subteam (physicians, nurses, SW‐CM, pharmacy, and coleaders). SIDR lasted a mean 41.4 11.1 minutes, with a mean 1.5 0.4 minutes spent in discussion per patient. SIDR was significantly longer in duration on teaching service units compared to the nonteaching hospitalist service units (1.7 0.3 vs 1.3 0.4 minutes per patient; P < 0.001).

Inter‐Rater Reliability

Inter‐rater reliability across unit level scores was excellent (rho = 0.75). As shown in Table 1, inter‐rater reliability across domains was good (rho = 0.530.68). Inter‐rater reliability across subteams was good to excellent (rho = 0.530.76) with the exception of the physician subteam, for which it was poor (rho = 0.35).

Assessment of Teamwork by Unit, Domain, and Subteams

Teaching and nonteaching hospitalist units had similar team scores (median [interquartile range (IQR)] = 5.2 [1.0] vs 5.2 [0.4]; P = 0.55). We found significant differences in teamwork scores across units and domains, and found differences of borderline statistical significance across subteams (see Table 2). For unit teamwork scores, the median (IQR) was 4.4 (3.94.9) for the lowest and 5.4 (5.35.5) for the highest performing unit (P = 0.008). Across domain scores, leadership received the lowest score (median [IQR] = 5.0 [4.65.3]), and cooperation/backup behavior and monitoring/situational awareness received the highest scores (median [IQR]) = 5.4 [5.05.5] and 5.4 [5.05.7]; P = 0.02). Subteam scores ranged from a median (IQR) 5.0 (4.45.8) for coleaders to 5.5 (5.05.8) for SW‐CM (P = 0.05). We found no relationship between unit teamwork score and time spent in discussion per patient (rho = 0.04; P = 0.79).

Setting and Study Design

The study was conducted at Northwestern Memorial Hospital (NMH), a 920‐bed tertiary care teaching hospital in Chicago, IL, and was deemed exempt by the Institutional Review Board of Northwestern University. General medical patients were admitted to 1 of 6 units based on bed availability. Five of the medical units consisted of 30 beds, and 1 unit consisted of 23. Each unit was equipped with continuous cardiac telemetry monitoring. Three units were staffed by teaching service physician teams consisting of 1 attending, 1 resident, and 1 or 2 interns. The other 3 units were staffed by hospitalists without the assistance of resident or intern physicians. As a result of a prior intervention, physicians' patients were localized to specific units in an effort to improve communication practices among nurses and physicians.20

Beginning in March 2010, all general medical units held SIDR each weekday morning. SIDR took place in the unit conference room, was expected to last approximately 3040 minutes, and was co‐led by the unit nurse manager and a medical director. Unit nurse managers and medical directors received specific training for their roles, including 3 hours of simulation‐based exercises designed to enhance their skills in facilitating discussion during SIDR. All nurses and physicians caring for patients on the unit, as well as the pharmacist, social worker, and case manager assigned to the unit, attended SIDR. Attendees used a structured communication tool to review patients admitted in the previous 24 hours. The plan of care for other patients was also discussed in SIDR, but without the use of the structured communication tool.

Importantly, nurse management underwent restructuring in the summer of 2011. Nurse managers, who had previously been responsible for overseeing all nursing activities on a single unit, were redeployed to be responsible for specific activities across 34 units. This restructuring made it very difficult for nurse managers to colead SIDR. As a result, the unit nurse clinical coordinator assumed coleadership of SIDR with the unit medical director. Nurse clinical coordinators worked every weekday and did not have patient care responsibilities while on duty. In addition to their role in coleading SIDR, nurse clinical coordinators addressed daily staffing and scheduling challenges and other short‐term patient care needs.

Teamwork Assessment

We adapted the Observational Teamwork Assessment for Surgery (OTAS) tool, a behaviorally anchored rating scale shown to be reliable and valid in surgical settings.2123 The OTAS tool provides scores ranging from 0 to 6 (0 = problematic behavior; 3 = team function neither hindered nor enhanced by behavior; 6 = exemplary behavior) across 5 domains (communication, coordination, cooperation/backup behavior, leadership, and monitoring/situational awareness) and for prespecified subteams. We defined domains as described by the researchers who developed OTAS. Communication referred to the quality and the quantity of information exchanged by team members. Coordination referred to management and timing of activities and tasks. Cooperation and backup behavior referred to assistance provided among members of the team, supporting others and correcting errors. Leadership referred to provision of directions, assertiveness, and support among team members. Monitoring and situational awareness referred to team observation and awareness of ongoing processes. We defined subteams for each group of professionals expected to attend SIDR. Specifically, subteams included physicians, nurses, social work‐case management (SW‐CM), pharmacy, and coleaders. We combined social work and case management because these professionals have similar patient care activities. Similarly, we combined unit medical directors and nurse clinical coordinators as coleaders. By providing data on teamwork performance within specific domains and for specific subteams, the OTAS instrument helps identify factors influencing overall teamwork performance. We modified OTAS anchors to reflect behaviors during SIDR. Anchors assisted observers in their rating of teamwork behaviors during SIDR. For example, an anchor for exemplary physician communication behavior was listens actively to other team members (looks at other team members, nods, etc). An anchor for exemplary physician leadership was assigns responsibility for task completion when appropriate.

Two researchers conducted unannounced direct observations of SIDRs. One researcher (Y.N.B) was a medical librarian with previous experience conducting observational research. The other researcher (A.J.C.) had observed 170 prior SIDRs as part of a related study. Both researchers observed 10 SIDRs to practice data collection and to inform minor revisions of the anchors. We aimed to conduct 78 independent observations for each unit, and 20 joint observations to assess inter‐rater reliability. All subteams were scored for each domain. For example, all subteams received leadership domain scores because all team members exhibit leadership behaviors, depending on the situation. In addition to teamwork scores, observers recorded the number of patients on the unit, the number of patients discussed during SIDR, attendance by subteam members, and the duration of SIDR. For the SW‐CM and coleader subteams, we documented presence if one of the subteam members was present for each patients' discussion. For example, we recorded present for SW‐CM if the social worker was in attendance but the case manager was not.

Data Analysis

We calculated descriptive statistics to characterize SIDRs. We used Spearman's rank correlation coefficients to assess inter‐rater reliability for joint observations. Spearman's rank correlation is a nonparametric test of association and appropriate for assessing agreement between observers when using data that is not normally distributed. Spearman rho values range from 1 to 1, with 1 signifying perfect inverse correlation, 0 signifying no correlation, and 1 signifying perfect correlation. We used the MannWhitney U test to assess for differences in overall team scores between services (teaching vs nonteaching hospitalist service) and KruskalWallis tests to assess for differences across units, domains, and subteams. The Kruskal‐Wallis test is a nonparametric test appropriate for comparing three or more independent samples in which the outcome is not normally distributed. We used a t test to assess for difference in duration by service, and Spearman rank correlation to assess for correlation between time spent in discussion per patient and overall team score. All analyses were conducted using Stata version 11.0 (College Station, TX).

SIDR Characteristics

We performed 7 direct observations of SIDR for 4 units, and 8 observations for 2 units (44 total observations). Units were at 99% capacity, and SIDR attendees discussed 98% of patients on the unit. Attendance exceeded 98% for each subteam (physicians, nurses, SW‐CM, pharmacy, and coleaders). SIDR lasted a mean 41.4 11.1 minutes, with a mean 1.5 0.4 minutes spent in discussion per patient. SIDR was significantly longer in duration on teaching service units compared to the nonteaching hospitalist service units (1.7 0.3 vs 1.3 0.4 minutes per patient; P < 0.001).

Inter‐Rater Reliability

Inter‐rater reliability across unit level scores was excellent (rho = 0.75). As shown in Table 1, inter‐rater reliability across domains was good (rho = 0.530.68). Inter‐rater reliability across subteams was good to excellent (rho = 0.530.76) with the exception of the physician subteam, for which it was poor (rho = 0.35).

Assessment of Teamwork by Unit, Domain, and Subteams

Teaching and nonteaching hospitalist units had similar team scores (median [interquartile range (IQR)] = 5.2 [1.0] vs 5.2 [0.4]; P = 0.55). We found significant differences in teamwork scores across units and domains, and found differences of borderline statistical significance across subteams (see Table 2). For unit teamwork scores, the median (IQR) was 4.4 (3.94.9) for the lowest and 5.4 (5.35.5) for the highest performing unit (P = 0.008). Across domain scores, leadership received the lowest score (median [IQR] = 5.0 [4.65.3]), and cooperation/backup behavior and monitoring/situational awareness received the highest scores (median [IQR]) = 5.4 [5.05.5] and 5.4 [5.05.7]; P = 0.02). Subteam scores ranged from a median (IQR) 5.0 (4.45.8) for coleaders to 5.5 (5.05.8) for SW‐CM (P = 0.05). We found no relationship between unit teamwork score and time spent in discussion per patient (rho = 0.04; P = 0.79).