EHR Conversion on the PreOp Care: A Pre-Post Workflow Comparison

Abstract

The impact of EHRs conversion on clinicians’ daily work is crucial to evaluate the success of the intervention for Hospitals and to yield valuable insights into quality improvement. To assess the impact of different EHR systems on the preoperative nursing workflow, we used a structured framework combining quantitative time and motion study and qualitative cognitive analysis to characterize, visualize and explain the differences before and after an EHR conversion. The results showed that the EHR conversion brought a significant decrease in the patient case time and a reduced percentage of time using EHR. PreOp nurses spent a higher proportion of time caring for the patient, while the important tasks were completed in a more continuous pattern after the EHR conversion. The workflow variance was due to different nurse’s cognitive process and the task time change was reduced because of some new interface features in the new EHR systems.

1. INTRODUCTION

With the rapid development of new information technology, the electronic health record (EHR) has continually expanding functions and is integral to clinical workflow. Clinical workflow is the process of clinicians working in teams collaboratively, sharing responsibilities and communicating effectively to achieve the goal of providing excellent patient care (1). Providing excellent clinical care is a complex process that involves multiple information sources, many situational variations and the seamless integration of different roles and responsibilities, using effective interdisciplinary teamwork, in a fast-paced, dynamic clinical environment. To deal with the complexity, modern medical facilities are increasingly applying cutting edge information technology and decision support in healthcare delivery. The EHR and its integrated decision support and management algorithms play a vital role in the decisionmaking process of interdisciplinary teams of providers through providing data, information and knowledge. For example, the EHR has an underlying effect of directing attention and shaping the behavior of providers by structuring the work tasks through the interface design. Necessarily, clinical workflow is profoundly mediated by EHRs. We refer to this as “EHR-mediated workflow”.

EHR use during preoperative care. Compared with other hospital departments, perioperative care has a direct and crucial impact on patient safety and patient outcomes (2). The time pressure and workload of caring for patients who are receiving operative procedures exceed the capabilities of many operative settings, bringing higher risks of errors. Evidence showed that one in 20 perioperative medication administrations results in an error, with more than one third leading to adverse events (3). As the last step before the operation room, PreOp care is highly time-sensitive. During the limited time, the work includes reviewing and executing orders, reconciling medications, tracking patient’s condition, and coordinating multiple roles. PreOp nurse, who is the primary PreOp caregiver, would encounter a set of tasks of information gathering, screening, and verification. PreOp nurses rely on EHR to do documentation. “Clinical documentation can provide the evidence to show how perioperative nurses are implementing evidence-based practices, supporting professional development, and even providing patient safety, quality, and satisfaction,” according to Janice Kelly, the president of AORN Syntegrity (4). Only after getting the accurate information can the surgical team be fully aware of the patient’s current condition, and then use this information to make the right decisions (5). The importance of high-quality and efficient PreOp nursing care raises the needs for embedding EHR with features clinging to the clinician’s practical workflow. EHR designers and those involved in implementation should not expect that ‘one size fits all’ and should focus on clinician-oriented workflow because there are various roles of clinicians whose information needs and tasks differ significantly (6).

Workflow analysis and EHR use. EHR and clinical workflow become deeply intertwined. On one hand, the provider’s routine workflow decides what functions and features are required for EHR. For physicians and nurses, order management, medication administration, history information review, health data documentation, and communication are necessary functions in EHR. On the other hand, EHR, as a component in the clinical environment, shapes workflow. Ash and colleagues reported in a study that the implementation of computerized provider order entry (CPOE) “enables shifts in power related to work redistribution and safety initiatives and causes a perceived loss of control and autonomy by clinicians” (7). EHR transition makes a change in how people do their work, how staffs feel and clinical practice requires customization of EHR to support the specific organizational needs (8). There is no standard method to assess EHR transition. Parameters can vary with different purposes, such as looking at how the transition changes communication with patients, how experienced clinicians use different EHRs for the same work in the same setting. These factors impacting practice can help to forge better EHR practices (9).

Workflow analysis challenges. There is no standard method for workflow analysis. It could be conducted with different approaches based on the goal. Typical goals included looking at the time efficiency, modeling the workflow, describing the information flow (10–14) or pursuing the problem-solving strategies for targeting barriers (15,16). As for the study design, observational study has been widely used with the advantage of not interrupting and affecting the clinicians’ work (1,12,16). The observation can be practiced in a variety of approaches. A comprehensive literature review on methods of workflow analysis in healthcare reported that the ethnographic observation and interviews were the most two common methods and most results were descriptive (17). The conventional Time and Motion (T&M) studies collect details regarding how clinicians spend time performing each clinical task and have been widely used in workflow analysis research (18). Workflow analysis studies using T&M have mostly collected data by shadowing clinicians for a work shift, recording the time for each task, and annotating the observed tasks. T&M measures the time on each task and yields reliable quantitative results for workflow analysis. However, evaluating workflow just by time is insufficient. An efficient EHR-mediated workflow for users, especially nurses, should not only take less time, but be capable of providing system effectiveness, managing cognitive load, minimizing fragmentation, reducing complexity, and facilitating smooth coordination (19,20). Therefore, qualitative and descriptive information from context is also needed to inform the analyses further.

The goal of this study was to assess the impact of different EHR systems on PreOp nurse’s workflow. The human-computer-interaction (HCI) is a discipline devoted to the study of the usability of computer interfaces and user-centered design (21). However, many HCI studies are conducted in a controlled laboratory environment. This approach is limited in understanding the clinical workflow. The context of a hospital, especially the perioperative environment necessitates a different kind of approach. Computational ethnography is an emerging family of methods leveraging automated means for collection in situ data reflective of real end user’s actual, unaltered behaviors using a software system or a device in real-world settings’ (22). By using the digital traces of technology during the EHR use, one can obtain objective and rich data without spending as much time, energy and human resources.

In this study, we employed a video ethnographic approach, recording the screen activities of nurses performing various tasks. That is using some screen capture software to record videos of screen activities automatically. The recorded screen capture video is thick data that presents the interface, the behavior of users and the sequence of tasks, which allows for much detailed qualitative and quantitative analyses. Cognitive engineering has been used to study nurses’ use of health information technologies (23–25). The application of cognitive engineering methods helps to identify and model the logistic flow and information needs, thus potentially achieve workflow improvement and optimal human performance (26). We cannot fully understand the formation and variant of the workflow, derive the cause of problems and errors without considering cognitive factors on patient safety. By integrating cognitive engineering, we conducted a microanalysis to investigate PreOp nurse’s performance at a fine level of granularity (27). Cognitive task analysis is a technique that was extended from traditional task analysis and yields information about the knowledge, thought processes and goal structure that underlie observable task performance (28). It is a useful tool to analyze complex clinical workflows by characterizing clinician’s activities into a set of tasks. At the task-level analysis, one can reveal crucial facets of interactions with the working system, explore variation and identify underlying reasons (29). There have been comparatively few micro-analytic studies in the clinical workflow field.

2. METHODS

2.1. Data collection

Mayo Clinic has undergone a three-year EHR conversion processes for each of its clinical settings between 2016 and 2019. We aim to find out if the EHR conversion is making a positive impact or not by assessing the difference of PreOp nursing workflow before and after the EHR transition. We conducted direct observation and collected video data prior to and subsequent to the conversion. Data were collected at the Mayo Clinic Hospital (Rochester) in Minnesota. In 2016 when the hospital was using the previous EHR (pre-conversion), the baseline data was collected. The post data in the PreOp department were collected in December 2018, six months after the system had gone live. The observed subjects were PreOp nurses, who provide the majority of PreOp care and use the EHR intensively. Using the same data collection method, PreOp cases were video captured by the usability evaluation software Morae™. A defined case capture started when a patient entered the PreOp room and the nurse logged in the EHR to start documenting the patient; ended when the nurse finished all the needed care and documentation for this patient and logged out of the EHR. Five PreOp cases were recorded before and after the EHR conversion each. The nurse managers helped to select nurses with more than two years of working experience and assisted in arranging the data collection slots based on the availability of the nurses and devices. We collected ten cases involving ten different nurses providing PreOp care for different types of surgeries, including partial knee replacement, elbow biceps tendon repair and valve-sparing aortic root replacement.

Video data is the direct in-situ observational data collected using software called Morae. Morae^TM provided a screen capture of EHR use, recorded mouse movements, clicks and screen transitions. Besides, the software is also linked to a webcam to perform a video and audio capture of the user’s behaviors and voices before the screen. When collecting data, the webcam was adjusted to an angle covering the PreOp nurse’s computer desk surface, so that to record the nurse’s actions such as note-taking, typing, and scanning medication labels (30). A “Think-aloud” protocol is a technique used in the cognitive study encouraging individuals to speak out loud what they are doing, thinking and seeing when performing a task (31). Before each case, we met the nurse and introduced the research purpose briefly, assuring them that it was not for quality management so they can do the work as usual. But they were encouraged to use the “think-aloud” technique especially during documentation to help us understand their interaction with the EHR. Their “think-aloud” utterance and conversation with patients and colleagues were also recordeded.

2.2. Data analysis

The video data for 10 cases were analyzed at the micro-analytic level using both quantitative and qualitative approaches. The Morae software had the main area showing the nurse’s screen capture, and a small window at the bottom right corner shows the nurse’s desktop behaviors recorded by the webcam. We sought to characterize the PreOp nursing workflow patterns and analyzed and compared the data of two EHR systems in the following steps.

Task Coding. The first step was segmenting the video by tasks, reducing the dense video data into quantifiable segments. The EHR interfaces had labels on each tab and button, which a nurse can choose to do related documentation. To start with, the initial definition of tasks was usually named after the tab names. However, the challenge was deciding the granularity of tasks as there were high-level tasks consists of smaller tasks, forming a hierarchy relationship between tasks. Therefore, after listing all the initially defined tasks, we revised the task list by merging some small tasks which took only a few seconds with tasks of a similar property. Finally, we got a task list with a reasonable number of tasks that were catered for PreOp care (32).

Task Classification. Since the video data also captured the nurse’s desktop behaviors and the conversation with the patient, we identified certain situations when a nurse was giving patient bedside care like giving medication or leaving the room for supplies. In other words, no computer use was observed. That duration was classified as doing ‘off-screen tasks’. One the other hand, the time when the nurse was using the EHR were all considered as doing ‘EHR tasks’. Separating ‘EHR tasks’ and ‘off-screen task’ helped to reflect how the nurse’s time was distributed to computers and patients respectively. This matters because clinician’s computer use was reported to sometimes adversely impact clinician-patient interaction during the care delivery when the provider’s attention was too focused on the screen (33) A preferred care delivery, especially for PreOp care, would be a higher proportion of direct patient care time. So we calculated the ‘case time’, the ‘EHR task’ time and the ‘off-screen task’ time as metrics for care quality.

Task Fragmentation. Traditional time and motion techniques can show how time is distributed to different tasks, as well as the task sequence and task time. A visualization of timebelt used in another study (18) was introduced to represent how much time was used for what task in which sequence. However, our observation was that there was a recurrence of certain tasks. A nurse may need to stop and start a task to or more times before finally completing it. This raised the concept of task fragmentation, which affects work continuity. Completing one task during a continuous period allows people to focus attention on one thing, minimizing the cognitive load (34). But this is almost impossible in the complex clinical environment full of interruptions and coordination. Naturally, the level of the task fragmentation is worth measuring as the more fragmented a task is, the more cognitive load it brings. And cognitive load was reported to play a negative role in care quality (35). A previous research paper used the concept of Average Continuous Time (ACT) and Proportion Per Instance (PPI) to measure the task fragmentation (36). The calculation formula is shown below. In the formula, the task instance refers to the number of occurrences of each type of EHR task. The assumption was that the smaller the ACT and PPT are, we can say the more fragmented a task is. Based on the total time duration, we selected tasks that consumed a relatively long time in cases and calculated the task fragmentation of these tasks. We assume understanding what leads to the fragmentation of these important tasks and raise solutions accordingly would make the biggest difference in the time efficiency.

$$ProportionPerInstance(PPI)=\frac{\text{Total\hspace{0.17em}Task\hspace{0.17em}Time}}{\text{Number\hspace{0.17em}of\hspace{0.17em}Task\hspace{0.17em}Instances\hspace{0.17em}\hspace{0.17em}}*\text{\hspace{0.17em}\hspace{0.17em}Total\hspace{0.17em}EHR\hspace{0.17em}Time}}$$

Cognitive Engineering Process. The quantitative data analysis had to be complemented with qualitative contextual data, which explained the user’s cognitive process, the environmental factors, interruption occurrence and so on. Otherwise, merely the quantitative data would be inadequate to investigate the myriad of challenges the nurses are facing. The thick video data, which provided the nurse’s desktop capture and the audio record of conversations and ‘think aloud’ protocols, was adequate to help researchers understand what was happening in the patient room. We can consolidate the cognitive process to understand the decision-making process of the nurses, thus finding out how the cognitive process differed before and after the EHR conversion.

3. RESULTS

After the task coding of five pre-conversion cases and five post-conversion cases, we got a task list consisted of 20 task types: ‘active patient worklist reminder’, ‘allergies’, ‘clinician start/stop’, ‘discharge planning/social services’, ‘off-screen tasks’ ‘IV charting’, ‘medication administration record’, ‘medication reconciliation’, ‘nurse assessment’, ‘nurse report’, ‘pain assessment’, ‘patient order management’, ‘pre-op checklist’, ‘pre-surgical screening’, ‘psychosocial assessment’, ‘selecting patient’, ‘surgical logistics’, ‘surgical patient education’, ‘vital signs charting’ and ‘writing’. Among them, all tasks except for ‘off-screen tasks’ were regarded as ‘EHR tasks’. This task list is formed specifically for the EHR use for PreOp care. Next, the case time, EHR task time and off-screen task time of each patient case were calculated as is shown in Table1. After the conversion, there was a significant decrease in case time, from an average of 68 mins (SD = 23 mins) to 38 mins (SD = 16 mins). Of the total time devoted to PreOp, the time using EHR decreased from 55 mins (83%) to 20 mins (53%). As for the off-screen tasks like providing bedside care and interacting with the patients or family face-to-face increased from 17% to 47% after conversion. Because of the limited case number, the average value could be impacted by an extreme minimal or maximum value. Thus, we decided to visualize the time distribution of ‘EHR tasks’ and ‘off-screen tasks’ by percentage in every case, as is shown in figure 1. It is interesting to find that the post-conversion cases consistently had a higher percentage of time doing off-screen tasks, which eliminated the influence of extreme minimal or maximum values.

Table 1.

The case time, EHR tasks time, and off-screen tasks time spent in minutes of each case

Time Spent in Minutes	PRE						POST
1	2	3	4	5	Avg.	1	2	3	4	5	Avg.
Case time	39.2	50.2	75.1	80.9	94.1	67.9	49.5	54.7	21.6	19.9	44.6	38.1
Off-screen time	3.6	4.5	24.2	14.6	15.3	12.4	26.6	21.8	13.5	5.8	23.3	18.2
EHR tasks time	35.5	45.7	51.0	66.3	78.8	55.4	22.9	32.9	8.1	14.1	21.3	19.9

Figure 1.

Time Distribution of EHR tasks and off-screen tasks by percentage

The timebelt visualization in figure 2 demonstrated the pattern of the task types and sequences of the five pre-conversion cases and five post-conversion cases. One single horizontal belt represented one patient case, and the length of the belt indicated the case duration in seconds. Each belt showed the sequence of tasks performed by the nurse, where each task was color-coded. The five pre-conversion cases had much longer belts than the post cases, with several nursing assessment sections within each belt. The off-screen tasks, represented as long black sections, consisted of longer continuous time segments than the EHR tasks for both pre and post-conversion cases. The EHR tasks showed a more fragmented pattern in the middle of the pre cases, while at the beginning of the post cases.

Figure 1.

Timebelt of the patient cases. Each belt represents one patient case. The different colors represent different tasks. The length of each sections is proportional to the time duration in the unit of second. The sequence of sections is aligned with the real observed task sequence from the videos.

To measure the task fragmentation using ACT and PPI, we chose the commonly recurring tasks, nursing assessment and PreOp-checklist. We started by counting the recurrence number of these two types of tasks in each case. The number of nursing assessment instances were substantially less after conversion (6 instances per pre-conversion case and 169.8s per instance compared to 3.2 instances per post-conversion case and 100s per instance). The PreOp checklist also decreased in time and instance (3 instances per case with 117.1s per instance before conversion as opposed to 2.2 instances per case with 88.5s per instance after conversion). Table 2 showed the calculated ACT and PPI of nursing assessment and PreOp checklist tasks in each case. Please note that the EHR time was in the unit of seconds for calculation. The result showed that the nursing assessment task was more fragmented with a smaller average PPI of 5.9% before the EHR conversion. The PreOp checklist task also showed a more fragmented pattern with a smaller average PPI of 4% before the EHR conversion. Overall, the post-conversion cases were proved to be able to complete these two important tasks in a continuous manner, which implies fewer interruptions and lower cognitive load.

Table 2.

The measurement of task fragmentation through Average Continuous Time (ACT) and Proportion Per Instance (PPI) on the tasks of nursing assessment and PreOp checklist.

	Pre 1	Pre 2	Pre 3	Pre 4	Pre 5	Post 1	Post 2	Post 3	Post 4	Post 5
EHR time (s)	2132.99	2742.69	3057.09	3975.78	4725.7	1375.05	1975.65	487.78	845.48	1280.62
Nursing Assessment
ACT	188.18	228.3	173.8	121.1	163	42	255.6	22.5	55.5	65
PPI	8.8%	8.3%	5.7%	3.0%	3.4%	3.1%	12.9%	4.6%	6.6%	5.1%
Average	5.90%					6.40%
PreOp Checklist
ACT	58.66	163.6	152.58	128.5	146.3	148	84	N/A	23	119
PPI	2.8%	6.0%	5.0%	3.2%	3.1%	10.8%	4.3%	N/A	2.7%	9.3%
Average	4%					6.70%

Based on the recorded conversation and nurse’s think aloud, a qualitative analysis was conducted to describe the cognitive process of the nurse after consolidating different patient cases. In the pre-conversion cases, the nurse went to the assigned room, making sure the room was prepared and everything was set up. Before seeing the patient, the nurse would log in to the computer and pull up that patient’s record to get an understanding of the patient’s case, check for documents like consent form and make a to-do list in the head. When the patient arrived in the room and took height and weight by the patient care assistant, the first task would be verifying the identity and letting the patient change the gown. After that, the nurse started to do a series of routine documentation tasks consisted of ‘plan of care’, ‘vital signs flow sheet’, ‘reminders and restraints’, ‘admission flow sheet’, ‘PreOp checklist’, and ‘home medication list’. These documentation tasks were shown as separate tabs in the previous EHR system. Nurses may or may not follow the tab sequence displayed in the interface, depending on their working habits and experience. The fact that the documentation process was practiced in different sequence contributed to the workflow variance. For example, one nurse was observed to start with the PreOp checklist so that she can page whom she needed for the missed service or items early and not cause any delays before the patient entered the operating room. Then, the patient case may begin with some off-screen sections during which the nurse was coordinating with other clinicians by phone. As the EHR documentation process was essentially gathering information and entering the information to the right place, the nurse was mostly making decisions on what charts should be documented for different surgery types, and how she can get the needed information for each question. Before the EHR conversion, multiple electronic systems needed to be accessed to gather different aspects of a patient’s data. We observed nurses searching for information jumped to another platform to fill in the answer to the question in EHR. The nurse needed to go through the cognitive process of deciding which platform to go and locating the needed information efficiently. In some cases when the information was found to be outdated according and this created additional distractions for the nurse.

In the post-conversion cases, the preparing process was generally similar: verify the identity of the patient, take the height and weight, and get the patient changed into a gown. The nurse would go through the EHR information to understand the case before seeing the patient, but it was not until she saw the patient when she can click the nurse start button in the new EHR system and capture when the care physically started. Meanwhile, the status board in the hallway projecting the EHR tracking system would change the status, notifying other clinicians which stage the patient was at. The next step was checking the orders. The current EHR system would present the orders categorized by PreOp, IntraOp and PostOp so that the nurse can identify the corresponding orders at the right stage. With the listed orders in mind, the nurse would plan accordingly to do the orders. Essentially, the documentation work content after the EHR conversion was mostly the same as those in the previous EHR system like inform consent, IV charting, vital signs charting, admission questions and medication list. But the documentation tasks were formatted into a one-level PreOp checklist listing all the required charts on the main page with checkboxes allowing nurses to tick off after finishing each item. The sequence in which the nurses complete the checklist varied by person. Some nurses were likely to follow the default sequence, while some may jump to do some certain charting first. This leads to a variance in the workflow. But during each documentation task, the behaviors of different nurses were similar, directly asking the patient or simply verifying with the patient. This was because some fields were pre-filled in previous visits. Even if a nurse needed to review the patient’s history information, the nurse can retrieve information from the summarization page within the EHR system, not /from another platform. Compared with the cases before conversion, there was no system or platform transition observed after conversion as features and functions were merged into one system.

4. DISCUSSION

Applying video ethnographic methods in a complex clinical environment provides solid data for quantitative time and motion analysis and qualitative contextual analysis. The significant difference was identified in the time distribution of the PreOp nurse and the change in their workflow. A considerable proportion of time was spent using EHR during the PreOp care, but after the EHR conversion, the proportion decreased significantly. Despite the shortened case time, the measurement of task fragmentation analysis found that tasks were conducted in a more continuous pattern after the conversion. Moreover, the visualization of the workflow showed some variance across different users. But qualitative cognitive process analysis showed a routine pattern of PreOp care. The main variance was due to the different documentation sequences, which was impacted by the EHR interface, the habit of users and the task priority within each case based on the nurse’s experience.

Task pattern and task fragmentation. The time and motion results showed that even after the EHR conversion, the EHR time occupied over 50 percent of the total case time and greatly affected the time spent face-to-face with the patient. Because the focus of PreOp care was doing the last -step information screening and input before entering the operating room, the high proportion of EHR time was understandable. Generally, it was a prominent fact that EHR had taken a considerable amount of clinician’s time in some time and motion studies across different roles and facilities (37–39). To dig deeper into the numeric time distribution results, the timebelt visualization and task fragmentation calculation were found helpful by providing a glance at the overall pattern. These techniques have been gaining popularity in clinical workflow studies (18,39). Our study found the new EHR facilitated the completion of the important tasks with more continuity and less disruption, implying less cognitive load. Interestingly, a comparison study focusing on physician’s workflow before and after the implementation of the computerized provider order entry (CPOE) found that the working pattern was more fragmented after calculating the task fragmentation despite modest changes in the time distribution by tasks (18). The contradictory findings may be explained by the fact that the challenges from paper to the first-generation electronic system was greater. However, EHR conversion usually means from one electronic system to another electronic system. Since the motivation of conversion was usually out of quality improvement, the user who had dealt with electronic system before would face fewer challenges in transition. Generally, what the numeric time and motion results can show were relatively limited. By combining with some visualization tools and task fragmentation calculation, the future researcher can investigate clinical workflow at a deeper level.

Factors impacting documentation sequence. The qualitative analysis showed that nurses had different sequence completing the documentation. Despite the different habits and experiences of nurses, in many cases, the documentation sequence was essentially a decision made by nurses according to the case, which means it was directed by the nurse’s cognitive process, not necessarily by the EHR interface design. Several factors impacted the documentation sequence. First, tasks requiring further coordination with other clinicians should be prioritized. In the PreOp checklist, there were documents such as H&P requiring actions from the physician and needing to wait. So some nurses would start with the rocky task to avoid holding things up. Second, tasks requiring information from the patient directly should be finished early. For some charts, the nurse was able to answer based on the history information and her observation. While some other charts require the patient’s answer or involvement like physical examination. For example, some nurses would tend to give IV placement early, so they will do IV charting as soon as they gave the IV. During the PreOp care, there were many interruptions as surgeons, residents, and anesthesiologists would come and talk to the patient. Completing patient-involving tasks first would give the nurse some flexibility when the patient was talking to others. Third, based on the surgery type, some physician’s orders of special medication or devices needed to be conducted first. If a sedative medication was ordered, which took an hour to take effect, the nurse would give the medication and do the medication administration charting first. Using cognitive task analysis to investigate the clinician’s thoughts was helpful to identify user’s needs, thus unfold the gap between the cognitive process and electronic system (40).

The new interface features improving efficiency. The video capture showed that there were new features in the current EHR system that greatly shortened the task time. The communication feature of the system allows the nurse to page other clinicians directly by clicking the name in the EHR, while it previously needed the nurse to search the intranet by name, found the pager number and make a phone call. A more beneficial point is that the information is now all in one place, the EHR system, unlike the previous time when the nurse needed to hover between several systems and manually searching and copying information from one to another. The time spent on information searching, retrieval was greatly reduced, and repetitive documentation was avoided to a large extend. Such features tackled the availability and convenient access to information in EHR, had proved to be crucial in improving patient care (41).

This study has some limitations. First, the observed workflow was specifically the preoperative care focusing on the PreOp nurse. Therefore, this nurse-centered workflow did not describe much about other clinician roles although there was considerable coordination during the PreOp care. The PreOp care we covered was only partial PreOp workflow. This was due to the needs of the comparison study. To control as many conditions as possible under the complex clinical environment, we build a clear boundary of the scope we want to focus on and excluded elements that can bring excessive variance. Another limitation was due to the device arrangement. In practice, a nurse can have access to different computers, including the computers in the patient room and the nurse station. We were mainly collecting data from the computer a nurse would use in the patient room. This was because it was impossible to predict which computer the nurse would use other than this computer. Besides, the facility usually assigned a mobile computer station to each nurse, so the nurse can roll the computer to wherever she wanted to do the documentation. The possibility of documentation on several computers was minimized. Lastly, there was a limitation of the sample size. The generalization of the results requires further research. This is a common disadvantage of a qualitative study with thick video-analytic data. Instead of going too broad, we went deep into each case to explore patterns. However, the strong and consistent differences across case strengthen our belief that the differences are substantive and possibly generalizable, despite the small sample size.

5. CONCLUSION

Understanding the impact of EHRs conversion on clinicians’ daily work is crucial to evaluate the success of the intervention and to yield valuable insights into quality improvement. To assess the impact of different EHR systems on the preoperative nursing workflow, we used a structured framework combining quantitative time and motion study and qualitative cognitive analysis to characterize, visualize and explain the differences before and after an EHR conversion. The results showed that the EHR conversion brought a significant decrease in the patient case time and a reduced percentage of time using EHR. PreOp nurses spent a higher proportion of time caring for the patient, while the important tasks were completed in a more continuous pattern after the EHR conversion. The workflow variance was due to different nurse’s cognitive processes and the task time change was reduced because of new functions and enhanced interface features in the new EHR systems.