Improving Efficiency of Pediatric Hospital Medicine Team Daily Workflow

Improvement efforts targeted 5 HM teams at a large, urban, academic children’s hospital. HM cared for ∼8000 patients per year in 2016 and 2017. One HM team, the complex care team, cares for children with neurologic impairment and/or technology dependence (ie, gastrostomy tube, tracheostomy tube) and has a cap of 10 patients. The other 4 teams care for patients hospitalized with general pediatric diagnoses and have an average daily census of ∼10 patients, with significant seasonal variation. Teams admit patients daily and consist of an HM attending, 2 senior residents, 2 to 4 interns, 1 to 2 medical students, and intermittently, an HM fellow. The complex care team has additional team members: social worker, dietician, pharmacist, care manager, and nurse practitioners. Weekday resident teaching conferences occur from 7:30 am to 8 am and from 12 pm to 1 pm. Rounds start after the morning teaching conference. All documentation and order entry occur in the EHR. Attending physicians addend or cosign resident and nurse practitioner notes. Our HM division has grown from 18 faculty and staff in 2011 to 46 in 2017.

FCRs are the standard of care at our institution.⁶  Most patients on HM service are admitted to a 48-bed inpatient unit, but some patients are placed on other units because of nursing expertise and overflow with higher census. Nurses caring for patients on >1 HM team lacked awareness of when teams would be rounding on specific patients, and simultaneous rounding of HM teams on a single unit made consistent nursing presence on FCRs challenging.

Our multidisciplinary improvement team consisted of HM attending physicians, a chief resident, a senior resident, an intern, and 2 nursing leaders. An analysis of failure modes and effects was conducted, and key drivers were identified and revised over time (Fig 1) and included (1) shared expectations, (2) enhanced physician and nursing buy-in and communication, (3) streamlined rounding process, and (4) transparency of data.

Before the start of the project, we sought to define our problem and build consensus. We surveyed attending physicians and residents to rate satisfaction regarding rounds length, notes completion timing, balance of meetings and other time demands, and amount of teaching while on clinical service. The majority of attending physicians and residents felt that rounds should end by 11 am. We chose 11:30 am as an absolute goal end time to allow for flexibility during high census periods. Both attending physicians and residents rated patient census as the primary driver of inefficiency, also noting patient complexity and team member inefficiencies as major contributors. Most attending physicians were satisfied with resident note completion time.

We subsequently distributed a separate 1-item survey to attending physicians to target note completion preferences. The majority of respondents (50 of 53; 97%) preferred to complete notes before 5 pm.

Nurses were engaged via nursing leader discussions; failure reasons for lack of presence on rounds were reflective of previous studies¹⁵  and included lack of physician perception that the nurse should be notified for rounds, nurses busy caring for other patients, or the team being unable to locate the nurse.

Interventions were developed to address identified key drivers: (1) establishment of a prerounds huddle, (2) visible team expectations and reminders, (3) scheduled rounds on the complex care team, (4) real-time nurse notification of patient rounds via an electronic platform, (5) workflow redesign, (6) attending feedback and data transparency, and (7) implementation of a resource attending.

Key interventions are further described in the sections below.

Our first intervention established rounding expectations, including a daily prerounds huddle, a process for notifying the nurses about rounds, and a clear end time for rounds. The prerounds huddle was led by the senior resident and formalized the discussion of (1) rounding order of all patients including new admissions, prioritizing patients designated as “watchers,” patients at risk for deterioration,¹⁶  and early discharges, (2) sensitive, patient-related psychosocial issues, and (3) afternoon schedules for team members, highlighting which residents would be leaving for continuity clinic. Additional stated expectations were that all nurses be notified as HM teams approach rooms for FCRs and that rounds end by 11:30 am. If all patients had not been seen by 11:30 am, a contingency plan allowed for the majority of the team to depart to complete other tasks (such as calling subspecialty consults) while the senior resident and attending finished seeing remaining patients.

Two study team members (N.I.U., A.M.S.), both hospitalists with residency program leadership roles, sent weekly e-mails that stated rounding expectations and highlighted active interventions to residents and attending physicians on clinical service. Study team members attended monthly resident team orientation meetings to reiterate expectations.

A written expectations sheet (Supplemental Fig 6) used to highlight expectations throughout the day was distributed to physician team members. Additionally, residents were encouraged to sign notes on rounds whenever possible.

Given that rounds on the complex care team require coordination of multiple providers and families, a rounds scheduling tool that assigned specific time slots for each patient was implemented. A template was used to create a daily schedule that included an estimated time for each patient’s rounds. The senior residents created and distributed the schedule to unit leaders and multidisciplinary team members each morning.

To bolster nursing presence on rounds, a team member was designated responsible for nurse notification. Residents identified nursing assignments via EHR assignments, unit electronic whiteboard displays, or printed sheets on some units. Residents communicated rounding time to bedside nurses via Voalte (http://www.voalte.com), a secure texting mobile app that allows bidirectional communication. Furthermore, we developed contingency plans in the event that bedside nurses were unable to participate in rounds because of other patient needs or rounding with another team, such as a charge nurse joining rounds in their absence. If no nurse was available, residents were directed to follow-up with verbal or texted plan communication.

To minimize interruptions to round during high census times, the role of direct admission calls was shifted to a nonrounding attending. Additionally, the attending role during a previously established daily 10 pm huddle was enhanced to promote efficiency the next day by encouraging preparatory task completion, such as sending prescriptions to pharmacies, for patients anticipated to be medically ready for discharge.¹⁷ 

In our review of the self-reported data from Research Electronic Data Capture (REDCap)¹⁸ , we noted attending variation in meeting specific bundle measures and that individual attending physicians performed similarly each time they were on service (eg, attending “D” always signed notes after 10 pm when on service).Thus, feedback was provided to attending physicians on their performance in comparison with their colleagues by sharing individual data on the percentage of days in which all 3, 2 out of 3, or <2 efficiency criteria were met with the corresponding number of days on service.

The aim and overall progress toward achieving the goal was shared with physicians via weekly e-mails and on an intermittent basis during divisional meetings and resident teaching conferences. Feedback was also provided to our primary nursing unit regarding nursing presence on rounds.

On the basis of observations from December 2016 to March 2017 during which HM census was high and rounds frequently extended past 11:30 am, additional attending workforce was added to optimize efficiency. The resource attending, a nonrounding physician, was added on weekdays from December 2017 to March 2018. This attending huddled with all 5 HM teams each morning, saw early discharges, evaluated new patients admitted during rounds, and facilitated direct admission calls.

Our bundled measure of efficiency included 3 components. First, nursing presence on FCRs was recognized as important for partnering to create plans and provide bedside communication with families throughout the day. To determine a goal for this measure, our team reviewed the baseline of nursing presence and noted variation by team and day. Recognizing competing priorities for nurses who cover up to 4 patients on >1 team, a goal was set at a majority of 70% of rounds with nurses present. Second, rounds end time by 11:30 am allowed time for working rounds that incorporate teaching but also ensured that resident teams were able to complete top-priority work before the noon teaching conference. Third, HM attending notes signed by 5:00 pm promoted timely documentation of a patient’s clinical status and allowed supervising physicians to leave the hospital for the day without documentation left to complete at home. A balancing measure of teaching episodes on rounds was intermittently assessed via research coordinator observation to ensure education was not compromised.

Baseline data were obtained from September to November 2016. Data collection relied on a daily REDCap¹⁸  survey completed by attending physicians and available through a phone-compatible link (Supplemental Fig 4). The link was sent via weekly e-mail and text message. Attending physicians recorded rounding census, number of nurses on rounds, rounds end time, and note completion time; a free text box allowed for comments regarding failures. Infrequent missing data (1.9% of data points) were excluded from analysis.

Data were not collected on the weekends because factors that impact efficiency (eg, resident conferences, committee meetings) are more prominent on weekdays.

A statistical process-control p-chart was used to analyze our primary bundled measure (the percent of opportunities in which all 3 efficiency criteria were achieved). Individual p-charts were also used to follow each measure separately: percentage of nurses on rounds, rounds ending by 11:30 am, and attending notes signed by 5:00 pm. Control chart rules for identifying special cause were used for centerline shifts.¹⁹  Failure information provided in surveys or verbally to the study team informed new interventions. We validated self-reported note completion time data by extracting EHR time stamps for attending physicians’ signed notes. A subset of data (1 week’s data each month during May 2017–April 2018) was reviewed, accounting for 14% of total data collected. The self-reported note completion times from each of the 5 attending physicians were crossreferenced with the generated data set, with emphasis on identifying time stamps after 5 pm and determining if attending physicians accurately self-reported failures. When the self-reported note completion time was discordant from a time stamp, chart review was completed. Of 300 data points reviewed, the self-reported data and the EHR data were concordant 97% of the time.

This quality improvement (QI) work was not deemed human subjects research.

The percentage of patients meeting all 3 efficiency criteria increased from 28% to 61% over 6 months. Significant week-to-week variation remained. However, there was improvement from the high census period in 2016–2017 compared to the high census period in 2017–2018 (28% vs 61%). Team member role designation was associated with the first centerline shift to 53%. A downward shift in the centerline to 38% corresponded to the beginning of the 2017 academic year as new residents entered the system. The centerline shifted back to 61% without any new interventions as our teams gained experience through the fall of 2017; variation around this centerline continued weekly thereafter (Fig 2). The efficiency bundle was completed more often in times of low census (7 or less: 70%) versus medium census (8–13: 50%) versus high census (>13: 30%) (Supplemental Fig 5).

Nursing presence on rounds increased from 65% to 82% and was sustained for >12 months (Fig 3). The goal was accomplished through ensuring reliable contacts to notify nurses of rounds and feedback to the nursing units.

The rounds ending by 11:30 am measure was already high, with the exception of the winter respiratory season in 2016 (Fig 3). Clear expectations and contingency plans, as well as the addition of a resource attending during high census times, facilitated an FCR end time of 11:30 am most days, regardless of the census (Fig 3).

Attending note completion time varied weekly, and we were unable to sustain success to shift the centerline. Failure reasons conveyed by attending physicians included lack of timely learner documentation, prioritizing patient care responsibilities, nonclinical demands and/or responsibilities such as administrative meetings, and attending preference to sign notes after 5 pm.

In considering our balancing measure, as expected, there was a wide range of teaching episodes both before and after interventions. Ongoing qualitative data from monthly rotation evaluations were collected from the residents who noted effective teaching as a strength of the HM service.

We sought to improve 3 efficiency criteria established through resident and attending consensus. Through sequential plan-do-study-act cycles, we increased the percentage of HM teams meeting all 3 efficiency criteria from 28% to 61% in 6 months. A broad approach to workflow throughout the day that included proactive planning through team huddles, enhanced nursing communication, and additional high census resources was key to our success.

Centerline shifts were primarily associated with key team member role designation. Clear responsibilities and structured opportunities for open communication improve collaboration and efficiency among team members.²⁰  On our HM teams, delineation of specific roles (ie, senior resident responsible for notifying nurse of rounds) decreased redundancy and ensured task completion consistency. Additionally, incorporating a prerounds huddle improved communication among team members and promoted strategic rounding habits.

Poor workflow associated with daily rounds can lead to inefficiencies that pull physicians from other tasks or may delay clinical decisions.²¹  Barriers include long rounding times and nurse and/or ancillary staff perceptions of not being valued by physicians.²²  To address these barriers, several of our interventions were focused on the rounding period. In previous work on resident efficiency during rounds, authors identified several areas of possible “waste” that may negatively impact efficiency, specifically waiting until formal rounds to discuss care plans and redundant plan discussions.¹⁴  Formalized huddles in the health care setting facilitate communication and empower teams to make decisions that may improve effiency.²³  Indeed, the prerounds huddle allowed for dialogue about key information that could expedite care plan execution before formal rounds and likely contributed to more efficient rounding.

Additionally, focused efforts to include nurses on FCRs can improve communication about ongoing and updated patient care plans, allowing an opportunity for nursing input and clarifications. Nursing involvement in FCRs is linked to improved efficiency of care given nurses can provide relevant information related to clinical status and progress to discharge goals.⁶  Improved physician-nurse communication is a supporting mechanism for effective teamwork and allows for shared mental models, closed-loop communication, and mutual trust.²⁴  Families also identify the important role of nurses in encouraging active participation of caregivers in FCRs.²⁵  In general, nurses value participation in rounds but may perceive coordination of rounding times as a barrier.²⁶  We found that advanced notification and scheduled rounding times (on the complex care team) may facilitate nursing presence on rounds.

Interestingly, note completion by 5 pm was the component of our bundled measure that revealed the most variability. Although this measure was chosen on the basis of attending consensus, the desires of attending physicians and the realities of patient care and life responsibilities were not always aligned. Our findings may represent individual attending preferences for managing work-life integration; some may prioritize note completion before leaving the hospital, whereas others prioritize home responsibilities, deferring note completion until late evening. Although not measured in our providers, burnout (defined as emotional exhaustion and depersonalization that can lead to decreased effectiveness in the workplace²⁷ ) can affect physician job satisfaction and patient care quality²⁸  and can be exacerbated by work-related stress and inefficiencies. Our work to improve HM team efficiency may indirectly contribute to improving provider satisfaction.

We recognized that despite several interventions, the predictable patient census increase each winter negatively impacted HM efficiency and workflow. After census strain noted on the system in winter of 2016–2017, strategic staffing changes were made, and the role of “resource attending” was created to facilitate efficiency during the 2017–2018 winter. Additionally, although efficiency was a major driver of our improvement efforts, preservation of resident education through exposure to attending physicians during rounds was important; additional resources ensured our attending physicians were able to offload tasks and preserve time with learners.

This work has limitations. We relied on attending physicians to complete daily REDCap surveys; hence, the data were subject to recall and social desirability bias. However, the majority of surveys were completed in real time limiting the impact of recall bias, and note time verification revealed high concordance between reported times and EHR time stamps. Attending physicians were also willing to document failures. Additionally, we limited the amount of data we asked our attending physicians to enter into the daily surveys, which limited the number and breadth of the measures we were able to reliably follow. Also, our work occurred at the main campus of our large tertiary care center with QI resources and support and may not be generalizable to other settings.

We used QI methods to increase the percentage of HM teams meeting 3 efficiency criteria from 28% to 61% through interventions focused on establishing clear roles, expectations, and consistent communication practices among providers and staff. The next steps include tailored interventions to improve timely note completion as well as exploration of the relationship between our QI interventions and more distal outcomes such as physician burnout, patient and/or family experience through enhanced nurse-physician communication, and discharge efficiency.