One researcher observed a team of obstetricians and midwives providing prenatal care services at an outpatient clinic in a Japanese university hospital. In the observed clinic, a total of 5 obstetricians and 10 midwives provide prenatal care services. After obtaining the approval of 3 obstetricians to observe checkups during their shifts, the researcher conducted the observations by visiting the prenatal care outpatient clinic twice a week over a period of 3 weeks.

At the beginning of the checkups, the obstetricians explained to the pregnant women and their companions the reasons for the researcher’s presence in the clinic. The obstetricians also asked the pregnant women and their companions if they accept having the researcher observe and take notes during the checkup.

After the pregnant women and their companions granted their approval, the researcher directly observed the prenatal care checkups and took notes using pen and paper. The notes contained descriptions of the interactions that took place around the EMR system, key information relating to the women’s course of pregnancy (pregnancy week, pregnancy type, and pregnancy number), sketches of the room layout and the EMR screen, and quotes and impressions from the conversations that took place around the EMR system.

In the observed clinic, there were 2 desks with computer terminals, as shown in Figure 3. One desk was used by the obstetrician and the other by the midwife. The room layout was semi-inclusive patient controlled. The pregnant women could see the EMR screen by moving the direction of their gaze [35].

During the observations, the researcher did not engage in conversations with the present parties. After the pregnant women and their companions left the clinic, the researcher asked the clinical staff questions to clarify certain occurrences. The researcher asked for information about the software that the staff used in addition to the EMR software. The researcher also asked for explanations as to why certain things were done in certain ways, for example, (1) using an image snipping tool, (2) bolding and changing the color of certain text, or (3) copying information from EMR notes and pasting them in other notes. In addition, the researcher inquired about artifacts that were used during the checkups, such as reference books that the staff used and paper files that the pregnant women exchanged with the staff.

In total, the researcher observed a team of 3 obstetricians and 6 midwives performing 37 prenatal care checkups for 35 different pregnant women between the eighth and 33rd week of their pregnancy.

After each observation, the field notes were transcribed and imported into QDA Miner, a qualitative data analysis tool made by Provalis Research [36]. After the observations were completed, the data were analyzed by 3 researchers following the 6 phases of thematic analysis described by Braun et al [37]:

While familiarizing ourselves with the data, we found that the interactions with the EMR system fall into 4 main categories: (1) interactions that support the communication, (2) interactions that hinder the communication, (3) interactions that support the clinical process, and (4) interactions that hinder the clinical process. These categories were mapped into 4 initial codes and were used in the first coding iteration.

After the first coding iteration, we noted more specifically how the EMR system supports/hinders the communication/process. These more detailed descriptions were used to code the data in the second coding iteration.

After the second coding iteration, we extended the codes to reflect aspects that could not be captured in the original codes and we merged codes together when their contents overlapped. Using these extended and refined codes, we conducted our third coding iteration.

After the situated roles were defined and named through the thematic analysis process, we administered a survey for all the prenatal care staff working at the observed clinic. In total, 15 surveys were sent out to 5 obstetricians and 10 midwives. The survey participants included the obstetricians and midwives that we observed in the field study.

First, the purpose of the survey was to validate the situated roles through the experiences of the users. Second, we wanted to understand how frequently the users want to experience the EMR roles and how important is each role to them. Therefore, for each situated role, we asked 3 questions:

The purpose of the first question was to validate the situated roles through the users’ current experiences. Answers to this type of question were reported using a 6-point Likert scale ranging from very frequently (1) to not at all (6).

The purpose of the second question was to understand how frequently the users wanted to experience each role. Answers to this type of question were reported using a 6-point Likert scale ranging from very frequently (1) to not at all (6).

The purpose of the third question was to understand the importance of each situated role. These statements were formulated based on the roles’ nature. For favorable roles, we asked about the importance of their presence. For unfavorable roles, we asked about the importance of their absence. Answers to these statements were reported using a 4-point Likert scale ranging from strongly agree (1) to disagree (4).

In addition to the questions regarding the situated roles, the respondents were asked for their job title and the number of years they had used the target EMR system.

The survey was designed over several iterations. It was pretested by researchers in medical informatics and human-computer interaction who evaluated the structure, understandability, scales, and formulation of the questions. Finally, the survey was pilot tested with 2 graduate students and refined based on their experience and feedback. The final survey was administered in Japanese.

We received a total of 15 survey responses, 5 from obstetricians and 10 from midwives. Table 1 shows the experience of the participants with the EMR system.

The purpose of the survey was threefold: (1) validating the situated roles through the experiences of the users, (2) understanding how often the users want to experience each situated role, and (3) understanding how important each situated role is to the users.

As a wingman, the EMR system supports the care providers in the explanation process.

During the checkups, the clinical staff verbally communicated clinical information to the pregnant women and their companions. This communication helps the pregnant women and their companions understand the current state of the pregnancy and the logic behind clinical decisions. In the observations, the obstetricians used the EMR system as a support tool to provide clinical information and explanations. We observed the obstetricians pointing toward the screen while reading their EMR notes and explaining them. The obstetricians also used automatically generated charts and ultrasound images from their EMR notes to visually communicate information to the pregnant women and their companions.

However, the obstetricians did not always automatically employ this strategy. In one case, while the obstetrician was explaining, the pregnant woman started leaning toward the EMR system’s screen to see the image that the obstetrician was looking at. Only after realizing that the pregnant woman was interested in seeing the image did the obstetrician rotate the monitor in her direction.

As an accomplice, the EMR system helps pause communication with the pregnant women.

We found that the obstetricians used the EMR system as a tool to pause communication with the pregnant women, a strategy which proved particularly useful when their workload was high or in highly emotional situations.

One of the obstetricians expressed the need for a moment to think in which they do not have to maintain a conversation with the pregnant women. In such cases, the EMR system served as a tool to pause the conversation and provide them with the needed moment to think.

Moreover, talking about pregnancies, especially complicated ones, could result in highly emotional situations. In this case, the EMR system provided the obstetricians with a bubble allowing them to distance themselves from the interaction. In one observed case, the obstetrician had to tell the pregnant woman that her pregnancy must be terminated. This woman had already experienced a pregnancy termination. After receiving the information, the pregnant woman started crying. At that moment, the obstetrician resorted to the EMR system to avoid looking at the pregnant woman and allow her to privately wipe her tears and stop herself from crying. When the obstetrician turned to the EMR system, the midwife left her desk and went toward the pregnant woman with a tissue box in hand. The midwife continued standing next to the pregnant woman while the obstetrician was working on the EMR system.

As a third wheel, the EMR system distracts the care providers from communicating with the pregnant women.

We found that the obstetricians spent a major part of the checkup time keyboarding and facing the EMR screen. During the obstetricians’ data input time, the pregnant women waited silently in their chair, looked closely at the EMR screen to see what their obstetrician was typing, or tried to initiate a conversation with the obstetrician or their companions.

While inputting data, the obstetricians responded to the pregnant women in various ways. Most of the time, they responded by turning their heads slightly away from the screen toward the pregnant woman. When the pregnant woman continued to ask questions or tried to engage in conversation, the obstetricians either started to alternate quickly between the screen and her or stopped inputting data and turned their chair away from the desk to face and respond to her. In some cases, they fully rotated their chair, but in most cases, they turned it halfway between their desk and the pregnant woman.

As a bouncer, the EMR system creates an exclusive environment by physically excluding the pregnant women and their companions.

On multiple occasions, we found that the pregnant women and their companions showed interest in looking at the EMR. However, the pregnant women had to actively get closer to the screen while their companions’ assigned chairs were placed too far from the screen, leading most of them to stop trying to look at the screen after a while.

On one occasion, the companion of the pregnant woman stood up to get a better view of the EMR screen. After standing up and realizing that he still cannot get a clear view, he tilted his head and body forward in the direction of the screen. When he realized that, even in this position, he cannot clearly see the contents of the EMR, he went back to his seat. After some time, he got up again, tilted forward toward the screen and went back to his seat, clearly feeling disappointed. A while later, he repeated the same sequence: he stood up, tilted forward, and sat down again. After sitting down, he gazed at the floor, bored and frustrated. Finally, he stood up, moved closer to the pregnant woman and to the EMR screen and remained standing there until the end of the checkup.

As a messenger, the EMR system enables the communication of information between the care providers.

In the case of the observed clinic, every pregnancy was cared for by multiple obstetricians and midwives. The rotation of the clinical staff required them to communicate the pregnant women’s health data. The EMR system was the main tool for communicating clinical information to ensure continuity of care. The EMR system, in this case, provided seamless communication between the clinical staff over time and staff rotations.

Conversely, when pregnant women were transferred from other clinics, the team only had access to the paper records that they had brought with them. In this case, the team created new EMRs for the women. However, the previous notes existing in the paper records were not transferred to the newly created EMRs.

Even though the pregnant women and their family members are involved in communicating health-related information to the care providers, they did not have the ability to directly add information into the EMR. During the checkups, through conversations with the pregnant women and notes from the women’s MCH handbooks, the providers gathered information and added them into EMR memos. However, what went into the EMR remained under the full control of the care providers.

In one examination, a pregnant woman, with a history of high blood pressure, brought along a paper containing a list of blood pressure measurements that she self-monitored and recorded. The obstetrician reviewed the measurements and handed the paper back to the woman. Then, the obstetrician wrote a note in an EMR memo regarding the measurements. However, the full list of blood pressure measurements remained out of the woman’s EMR.

As a summarizer, the EMR system provides the care providers with a summary of the pregnancy’s current state and care course.

The EMR system allows the prenatal care providers to have all the health information in one place. On multiple occasions, before calling a pregnant woman into the clinic, the obstetricians quickly navigated through the previous EMR notes to form a mental summary of her current course of pregnancy.

However, the EMR system did not allow for a quick understanding of the current state of the pregnancy. One obstetrician noted, “we would like to see the course of care in one glance. With paper records, it was easier to do that. However, with this system, it takes a lot of clicking and scrolling to get the full image.”

The staff needed the EMR system to act as a summarizer. To achieve that, the obstetricians employed a workaround. To give themselves and the other providers a quick understanding of the care course, the obstetricians emphasized certain parts of their EMR notes by changing the size, boldness, and color of the text.

As an assistant, the EMR system facilitates the management and preparation of the checkups.

At the beginning of their shift, using the EMR system, the obstetrician and the midwife viewed the list of scheduled checkups. Knowing the number of checkups, they could estimate the workload for the day. Based on that information, they adapted the speed of their work and the duration of checkups. Moreover, using the scheduled checkups list, the obstetrician and midwife knew who they were examining next and had access to her records before the checkup. Before calling the woman in, they reviewed the previous notes and discussed the current state of the woman’s pregnancy. Using this information, they could form a picture of what care actions they needed to perform once the pregnant woman was called in. By allowing for previous preparation, the EMR system makes the checkups run more smoothly. It eliminates the need for the staff to orient themselves and for the pregnant woman to explain the reason for her visit at the beginning of her visit.

As a gossip, the EMR system is not completely trusted with sensitive information. In our analysis, we found that the clinical staff hesitate to include highly sensitive information in the pregnant women’s records because of privacy and legal concerns. In one of our discussions with the staff, one midwife stated,

Employing this sort of strategy to document sensitive information implies that the EMR system is not completely trusted by the staff with information that is usually considered private or could be used for legal purposes.

As an alien, the EMR system (1) has low learnability, (2) requires a high level of recall, and (3) has an interface that is not optimized for routine tasks.

The difficulty of learning how to use the EMR system was one of the problems noted by the obstetricians. One obstetrician mentioned that it took them at least 1 month to get used to the system. Moreover, during a checkup, a staff member walked into the clinic and asked the obstetrician a question regarding the use of the EMR system, which the obstetrician answered by guiding them through the interface.

Furthermore, the EMR system appeared to require a high level of memory recall. The obstetricians frequently paused and tried to recall in which tab a specific field, or note, was placed. The inconvenience of the manual data input was further amplified by an interface design that was not optimized for routine data input and data retrieval tasks.

In addition, the EMR system suffered from performance-related issues. In certain cases, the system would temporarily stop responding or have a slow response time. These issues occurred particularly when the providers opened a new EMR. Even though 90% of the common database queries are usually cached, and the list of patients is previously compiled, opening a new EMR required more than 10 seconds in certain observed cases. This poor performance resulted in obvious frustration and time loss. To counter this issue, some midwives employed a sort of manual caching where they anticipated the need to open the records, opened the records, and let them load before they actually needed them.

As a bureaucrat, the EMR system requires the care providers to halt the care process to input data.

During the checkups, the providers continuously collect data from multiple sources and add it into the EMR. Those sources include conversations with the pregnant women and their family members; ultrasound imaging devices; and measuring devices such as blood pressure meters, weighing scales, and measuring tapes. The lack of integration between the medical devices and the EMR system required the providers to manually input most of the data that they collect. To do so, they had to intermittently pause their clinical flow. Below are some examples of occurrences encountered during the observations.

Before entering the clinic, the pregnant women use a blood pressure meter and a weighing scale located in the waiting room. The machines print the measurements on small paper receipts. Once they enter the clinic, the women hand the paper receipts and their MCH handbooks to the midwives. During the checkup, the midwives copy the measurements into the MCH handbook and then input them into the EMR. The process of copying the data could take up to 3 min. After they copy the measurements, the midwives throw the small paper receipts in a trash bin under their desks. On 2 different occasions, during ongoing checkups, the midwives had to look in the trash bin for receipts that they had previously thrown away. In one of those occasions, a nurse had to come in, put gloves on, and help the midwife look inside the trash bin.

In addition, the midwives routinely measure the belly circumference before the obstetricians start to conduct the ultrasound. Using a measurement tape, they measure the belly twice, vertically and then horizontally. After the second measurement, the midwives sometimes retake the first measurement, as that they might have forgotten the first measure. Once they finish measuring, some midwives prepare the women for the ultrasound, turn off the lights and then head back to their desks to input the measures. As this increases the risk of forgetting the measures, other midwives prefer to head fast to their desks, input the measures in the MCH handbook and the EMR, and then return to the woman to prepare her for the ultrasound.

As for the obstetricians, they routinely use 2 ultrasound devices to collect data. After they finish conducting the ultrasounds, they reflect on the results and summarize them inside free-text EMR notes. Then, they add the ultrasound images to the notes. To do so, they manually copy the information from the output of the ultrasound devices into the EMR. In addition, they use an image snipping tool to take screenshots of the ultrasound images and then they paste the images inside the EMR notes.

It is important to note that similarly to the third wheel, this role is manifested when the providers input data into the EMR system. However, as a third wheel, the EMR system hinders the communication between the providers and the pregnant women, whereas as a bureaucrat, the EMR system hinders the clinical workflow.

To validate the situated roles, we analyzed the responses regarding the users’ current experience. We looked at the midwives’ and obstetricians’ answers separately. All the situated roles were validated as at least one respondent reported experiencing them occasionally. Tables 3 and 4 show the responses of the obstetricians and midwives, respectively. The numbers in the table indicate the number of respondents that chose the option.

To better understand the extent to which different roles are experienced, we assigned them to 3 different categories:

The extent to which the respondents experience the situated roles is shown in Table 5.

To understand the desired frequency of each situated role, we analyzed the users’ responses for how frequently they would like to experience the roles. We looked at the midwives’ and obstetricians’ answers separately.

Tables 6 and 7 show the responses of the obstetricians and midwives respectively. The numbers in the table indicate the number of respondents that chose the option.

To better understand the extent to which different situated roles are wanted, we assigned them to 3 different categories:

The extent to which the respondents want to experience the situated roles is shown in Table 8.

We assumed that favorable roles are ones that users want to experience more frequently than they currently do. Unfavorable roles are ones that users want to experience less frequently than they currently do. Accordingly, the summarizer, the messenger, the assistant, and the wingman were regarded as favorable by both obstetricians and midwives. However, the obstetricians wanted to experience the wingman more frequently than the midwives. The alien, the third wheel, the bureaucrat, and the bouncer were regarded as unfavorable by both groups.

As for the gossip and accomplice, we considered them to be ambivalent. Their ideal frequency is the same as their current frequency. The midwives considered the gossip unfavorable, but the obstetricians wanted to experience it occasionally. Conversely, the obstetricians considered the accomplice unfavorable, but the midwives wanted to experience it occasionally.

As mentioned in the Methods section, for favorable roles, we asked the staff about the importance of their presence. For unfavorable roles, we asked about the importance of their absence. The accomplice role was treated as favorable to the prenatal care staff as it is created by them. The gossip role was treated as unfavorable as it denotes a lack of trust in the system.

To understand the importance of the situated roles, we looked at how much the respondents agreed that the roles’ presence or absence was important. We also looked at the midwives’ and obstetricians’ answers separately. Tables 9 and 10 show the responses of the obstetricians and midwives, respectively. The numbers in the table indicate the number of respondents that chose the option.

To better understand the importance of the different situated roles, we assigned them to 4 different categories:

The situated roles’ levels of importance are shown in Table 11.