Time is of the essence: an observational time-motion study of internal medicine residents while they are on duty

Background The effects of changes to resident physician duty hours need to be measureable. This time-motion study was done to record internal medicine residents’ workflow while on duty and to determine the feasibility of capturing detailed data using a mobile electronic tool. Methods Junior and senior residents were shadowed by a single observer during six-hour blocks of time, covering all seven days. Activities were recorded in real-time. Eighty-nine activities grouped into nine categories were determined a priori. Results A total of 17,714 events were recorded, encompassing 516 hours of observation. Time was apportioned in the following categories: Direct Patient Care (22%), Communication (19%), Personal tasks (15%), Documentation (14%), Education (13%), Indirect care (11%), Transit (6%), Administration (0.6%), and Non-physician tasks (0.4%). Nineteen percent of the education time was spent in self-directed learning activities. Only 9% of the total on duty time was spent in the presence of patients. Sixty-five percent of communication time was devoted to information transfer. A total of 968 interruptions were recorded which took on average 93.5 seconds each to service. Conclusion Detailed recording of residents’ workflow is feasible and can now lead to the measurement of the effects of future changes to residency training. Education activities accounted for 13% of on-duty time.


Introduction
Prolonged on call shifts and long duty hours are linked to concerns for patient safety, quality of care as well as concerns for the health and safety of resident trainees who may be working in excess of 24-hours without relief. 1 influenced the implementation of restrictions on resident duty hours. In addition, academic health centers have implemented various on-call strategies such as specific night shift relief to help ease resident on call duties or decrease length of consecutive duty hours. Recent publications [6][7][8] have highlighted the ongoing debate concerning the duration of duty hours, its potential harms and benefits. [9][10][11] As discussions around the world continue in consideration of further changes to resident physician duty hours, it is important to have an improved description of workflow in order to measure the impact of new systems of workflow and determine how they will affect patient care, resident education, and resident work-life balance. Having an easy to use and reliable tool for workflow determinations is essential to study the effects of future changes to resident duty hours.
Time-motion studies have been used both in industry 12 and in hospitals [13][14][15][16] to provide an objective method to record time for completed tasks. A previous systematic review of time-motion studies of internal medicine residents identified eight studies which lacked detail in categorization or definition of how time was spent with limited ability to detect trends related to restrictions on duty hours. 17 Residents spent the highest percentage of time in patient care ranging from 27-52% though extrapolating to actual number of hours was not possible with the published data. The percentage time spent on education was modest and decreased minimally after implementation of duty hours restrictions. The meta-analysis demonstrated the paucity of information and the need for detailed moment to moment description of resident work flow. 17 The purpose of this study was to determine the exact amount of time Internal Medicine residents spend on specifically defined tasks while on duty on an inpatient clinical teaching unit (CTU) in an academic hospital. The secondary goal of this project was to test the feasibility of an electronic data capture tool to quantify in minutes how residents spend their time. By understanding and measuring resident workflow future changes to training programs and scheduled activities can be designed to optimize resident efficiency, improve physician-patient interactions, and potentially enhance patient care and safety.

Methods
This study was performed at The Ottawa Hospital in Ottawa, Ontario, Canada. The Ottawa Hospital is an academic health center which serves a total population of 1.25M. The Ottawa Hospital has two sites: the General Campus (549 beds) and the Civic Campus (460 beds). All participants were enrolled in the core Internal Medicine residency training program.
This was an observational time-motion study of Internal Medicine residents, including junior (postgraduate year PGY-1), and senior (PGY-2 and 3) residents during their rotation on the internal medicine CTU.
All residents in the core Internal Medicine residency training program were invited to participate. Participants signed informed consent. The list of tasks and their categories was determined a priori by a focus group of 27 participants of which there were eleven PGY-1 residents, seven PGY-2, four PGY-3 and five attending staff. The initial list of tasks and categories were derived from the results of a systematic review. 17 This list was then expanded, refined and finalized via a Delphi process through three iterations. 18  At the start of each observation block a comment could be recorded to provide additional context to the observations, e.g., "Emergency Room pressure is high today." Additionally, the application was able to record interruptions, including their duration and details. Figure 1 shows a screenshot of the mobile application and Figure 2 shows a screenshot of data entry concerning a task interruption. The observer tapped the button next to the corresponding task to be recorded, which captured the date and time of the start of that activity. Tapping the green button (which indicated that an activity was in progress) captured the date and time of the end of that activity. A comment could be entered for each activity recorded. Each activity had to be completed before recording a new one, except in the case of an interruption. Simple descriptive statistics of the data were done using SAS version 9.3 (Cary, N.C., USA). The study protocol was approved by the Ottawa Health Sciences Network Research Ethics Board. The single observer (CL) underwent training, including the use of the mobile data collection tool, understanding what a typical resident's day entailed and piloting the entire system through five trial run observation blocks which covered both day and night shifts. The observer was instructed to limit interactions with residents as much as possible, only asking if a clarification about the current task (e.g., a computer being used for which purpose) was needed.  Table 1 shows what activities filled most of the times spent in the different categories of tasks. Some activities, for example in the education category, happened less often and took more time, such as didactic large group and bedside teaching. The majority of Direct Care activities did not require physicians to be at the bed side and involved using computer applications. During their entire on duty time residents spent 8.8% of their time in the physical presence of their patients, of which 42.1% of this time was spent in patient counseling activities.

Interruptions
An interruption was defined as the recording of an event which caused the resident to switch focus from a current activity to complete a different one. For example, if a resident was paged and stopped the current activity to answer their page, this was considered an interruption. However, if the resident completed the current activity and then answered their page, this was not considered an interruption. The paging event was no longer interrupting the completion of an activity but became the next task on the resident's to-do list. The study recorded 968 interruptions, for a total of 95.76 hours with an average duration of 93.5±198.82 seconds. The most common interruptions were caused by talking to attending staff in person (50.1%) and answering pages (20.7%). Since the time recorded to the original activity that was interrupted includes the time needed to service the interruption, the total observation time of 516 hours exceeded the elapsed total real time of 366 hours (61 blocks x six-hours). The timer for the original task, which was interrupted, was continued since we wished to document whether residents actually returned and completed those tasks. There were a total of 36 tasks which were interrupted and were never resumed to completion (e.g. cardiac arrest code interrupting a patient counselling activity).

Discussion
The time-motion data capture by a single observer using a mobile device application was easy to use and produced detailed, minute-to-minute information. The application enabled the recording of task interruptions and unlike using wearable video recording devices a live observer could ask for clarification (e.g., whether the computer use being observed was associated with patient care or education). In comparison to other time-motion studies as reported in a systematic review by Leafloor et al. 17 this study yielded superior data than those which used paper and stopwatches, relied on resident reporting or recall, or had limited number of observation events or time. Since the list of tasks and categories were determined by consensus a priori it was easy to record the tasks in real time. The entire data capture system was developed and implemented with readily available and inexpensive software. The use of a mobile tablet device and wireless communications made data entry easy and allowed the capture of timestamps in real time. Security, privacy, and confidentiality were maintained since the mobile device served only as a user interface and stored no data. contact. However less time spent with patients has been associated with a decrease in patient satisfaction. [22][23][24] To offset the finding that residents spent a significant proportion of time looking at electronic data, they also spent 17.3% of time in the direct patient care category talking to and counselling patients.

Communication
Duty hour restrictions seem to be associated with an increase in the frequency and total duration of communication activities. This study found that 65.0% (17.0% sign-in rounds, 4.2% sign-out rounds, 41.6% talking to hospital staff) of time was spent in information transfer activities, compared to 16.5% of time reported in studies before 2003 and 22.3% of time reported in studies afterwards. 17 These exchanges of clinical information including handover are an essential aspect of healthcare delivery, and if handover is not standardized there is a potential for information gaps leading to possible adverse events. 25,26 Interruptions during handovers can potentially disrupt efficient information transfer and an analysis of our dataset into this issue will be the subject of a separate study.

Education
Meeting resident educational objectives is an important aspect in the management of workflow.
Our study found the proportion of time the residents spent in educational activities (13.2%) during their entire on-duty time was similar to that found in the systematic review (average 13%) 17  idiosyncrasy related to when the study was done (at the beginning of the academic year) or other unidentified confounders. Further clarification will be explored since bedside rounding has been at the core of medical teaching and has been received positively by learners. [27][28][29] Across multiple studies of medicine and surgery programs there has been a decrease in educational opportunities 30,31 with duty hour changes. Prior studies have shown that while the number of hours residents were able to sleep has increased and perceived fatigue decreased, the quantity and quality of educational opportunities for residents as well as the quality and continuity of care they provided decreased significantly. 21,32 It is important to note that activities were categorized as Education by the residents themselves and for the most part did not consider hours spent in situational, work based or experiential learning environments. Re-categorizing some or a portion of these events may alter the percentage of time spent in learning.

Personal
Time spent in personal activities (14.7%) was similar to the average of 13.7% found by Leafloor,17 of which half was spent in the on-call room presumably providing an opportunity to sleep. Although there were un-recorded events during the time residents were in the on-call room (e.g. answering pages), our hospital has implemented a rule that junior residents are not to be called between 03:00-08:00 for a new consult or admission that requires an in-person visit.
A meta-analysis of the impact of reduced duty hours in surgical residents found an improvement in resident fatigue levels and general well-being. 30 A 2006 study also reported that surgical residents spent more time having lunch after the 2003 duty hour changes and reported a decrease in "emotional burnout." 31 These findings suggest that duty hour changes do support a better life-work balance for residents.

Strengths of the study
To our knowledge, this study reports the results of the most comprehensive and complete time-motion analysis of internal medicine residents to date. The categories and tasks were selected a priori and recorded consistently by a single observer. Time was automatically recorded and the wireless mobile technology ensured complete and reliable real-time data capture. e55

Limitations
Despite precautions taken to minimize interactions between the observer and the residents, the mere presence of an observer may have changed their behavior. The Hawthorne effect 33 may have caused residents who were observed to avoid doing nonwork related activities while being shadowed. Although activities of a personal nature were indeed recorded (conversations, emails, social media, etc.) we would not be able to measure the suppressive effects of having a shadow observer. Other confounders include: resident factors, hospital environment factors, time block factors, and day of week effects. Residents who did not consent or were not observed may be different. This study was done during the beginning of the academic year which may be different. The use of a single observer versus multiple observers may have introduced measurement bias due to a single observer's interpretation of witnessed events. The observer may misclassify some observed events. However, clarification and validation of events were done in real-time if needed. Personal time may have embedded events such as answering pages and issuing orders while residents were in their on-call room. Some communication activities, for example, talking to attending staff by phone, may have embedded learning points and some of that time should contribute to education time. The physical layouts of the Civic and General campuses are different. The observation blocks were not all done in consecutive sequence. Different attending staff physicians may have different scheduled teaching or case review events. A possible "gold standard" for time-motion study by analyzing the continuous recording of a subject while wearing a personal video recording device (e.g. GoPro) was not considered because of the significant privacy issues involved.

Conclusions
Time-motion observation of residents while they are on duty on an internal medicine clinical teaching unit using a mobile data capture tool can generate comprehensive, complete, and detailed records. Understanding how time is spent while residents are on duty can assist discussions with residency program directors, hospital management, quality and safety directorates and resident physicians. Different work patterns can be designed, tried and measured for their effects on patient care and satisfaction, resident learning experiences and resident well-being. Further research studies employing different methodologies are needed to clarify the effects of some of the aboveidentified limitations.
Authorship: All authors contributed to the design of the study. CL and AH were involved with the design of the mobile data collection tool and data collection process. EL and DMR were involved in the data analysis. All authors were involved in the interpretation of the results and writing of the manuscript. All authors read and approved the final text.
Acknowledgments: CL was funded by the Department of Medicine grant for innovation in medical education. None of the other authors received funding for the study.
Availability of data: The datasets collected and analyzed during the current study are available in deidentified form from the corresponding author on reasonable request.
Funding: This research was supported by funds from the Department of Medicine, through a grant for innovation in medical education, and by The Ottawa Hospital. The funding bodies had no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript. Each activity in Tabs parts A, B, C, D have the same flow. Note the header (highlighted above) must be filled prior to clicking on any of the activity. Each activity has 2 buttons, the left one is the start and end time button for the activity, the right one is the interruption start and end time button for the activity. When you click the button, it will go green to indicate the time tracking for the activity has started. The right button remains red until an interruption needs to be tracked. It also turns green. e64 This snapshot shows an activity having started followed with an interruption which results in an additional screen to document the activity that caused the interruption. The Category and Task are the same as the ones listed in each Tab (by design). More free text details can also be provided. It is not mandatory to document anything on the Interruption screen, one can cancel out of it, however, the interruption is still tracked. The interruption must be stopped by clicking on the rightmost green button (will revert back to red) before the activity can be stopped by clicking on the leftmost green button (will revert back to red). Note the same activity can be interrupted more than once as long as the activity hasn't been stopped.
The Main form making up the 4 TABs use the same logic for each activity with different parameters to identify each one. The sample code below is for 1 activity which is made up of 4 buttons, each tied to an Event Procedure. Consequently, there are 4 and they're shown below. Each calls a separate function, however this same function is used for every activity. As noted, only the parameters identifying the activity (highlighted) are specific, this parameter is maintained in the Unbound field. The detailed event procedures and function calls are available on request.