Abstract
Distinguishing urgent from non-urgent communication is critical. We aimed to understand how hospital staff choose what communication technologies to use. A mixed methods study was conducted with four focus groups, with staff working on postpartum care units, and log data from a hands-free communication system was analyzed. We found that urgent communications were appropriate for the hands-free device, with less urgent communications sent through secure chat in the electronic health record. Exceptions included when the intended recipient was in the operating room and during sensitive discussions. The most common duration of calls on the hands-free device was 16–60 seconds, with few more than 5 minutes. The most frequent reason for incomplete calls was the user not logging in (36%), which could be reduced by eliminating logging in and training. We recommend an interruptive communication technology for urgent information and an electronic health record chat for less urgent information that does not require an immediate response. We recommend forwarding and sending calls to a hospital-provided cellphone from the hands-free communication system for provider roles that do not align with the intended workflow for hands-free communications.
Introduction
Characteristics of Communication in Healthcare
In any complex socio-technical setting, real-time communication is critical for safe operations; timely, efficient, and reliable communications are crucial in any high-consequence environment. Clinical environments are dynamic and typically event-driven (Richardson & Ash, 2010). Often, registered nurses providing bedside care are the central coordinators, clinical providers, unit clerks, administrative staff, patients, and their family members. Healthcare communication is more fragmented, non-linear, and interruptive than other domains (Brixey et al., 2008) and often involves waiting on others to respond to communication requests (Jennings et al., 2022). In one study, over half of information exchanges occur through clinician–clinician communication via synchronous channels with an interruption rate of 42% (Edwards et al., 2009). Asynchronous communication channels (e.g., pagers, electronic health record chat, voicemail, email) rely upon the recipient returning the communication, which may add delays in care and potentially impact patient care adversely (Edwards et al., 2009). An essential aspect of healthcare is the need to have protected health information (PHI) not overheard by others, which is regulated by the Health Insurance Portability and Accountability Act (HIPAA). Therefore, in-person and hands-free communication devices containing PHI should not be overheard by other patients, family members, or visitors (Richardson & Ash, 2010).
Gap in Research on Hands-free Communication Devices in Postpartum Care Units
No known research has been conducted on hands-free communication devices in postpartum care units. Perinatal care is the time from when a patient becomes pregnant through a year after birth. In the hospital, the units where care is provided for laboring patients who do not receive a C-section are labor and delivery and postpartum care units for following delivery. The typical time spent in postpartum care units is 24–48 hours following an uncomplicated vaginal birth. In contrast, with the model of couple care, the same nurse provides care to the mother-baby dyad in the same room. C-section deliveries include operating room care and usually require an additional day or two on the postnatal unit to recover (Hassan et al., 2022). Complicated deliveries can consist of care provided in the Neonatal Intensive Care Unit (NICU) for newborns and the adult Intensive Care Unit (ICU) for birthing persons.
Our study fills this gap, which is critical as there are several significant contextual differences in perinatal care as compared to intensive and emergency care, where most research on communication technology has been conducted. For example, when a nurse cares for a dyad of a mother and a newborn simultaneously, they communicate with both specialized health care providers for post-delivery care for adults (obstetrics and gynecology and family medicine physicians, midwives and advanced practice providers) and newborns (pediatricians, family medicine, and lactation consultants). When one or both dyad members require placement in an intensive care unit (ICU or NICU, respectively) or the newborn nursery, the nurses and physicians on the postpartum units must coordinate with personnel from those units. Interactions with family members include diagnosing and treating patients and providing education to support the in-home care for both postpartum individuals and their infants. Both during the birth and afterward, emergency surgery may need to be conducted in operating rooms, which generates a need to coordinate with staff impacted by the surgery while also maintaining communications about care for other patients, even while in the operating room. The physicians and APPs responsible for postpartum patients are typically caring for laboring patients and may be receiving communications during delivery or a c-section. For medical residents, communication to organize and coordinate care for resident physicians and to support training resident physicians continues even while in the operating room. Finally, communications about distressing events such as severe morbidity or loss of a newborn or mother during inpatient care are exceedingly challenging and require in-person, focused attention with minimal interruptions from communication technologies.
Conceptual Focus: Work System Resilience
We conducted this research to identify work system resilience gaps using three communication technologies on perinatal care units: a hands-free communication device, chat in an electronic health record, and individual pagers that display free text, such as phone numbers or room numbers. We focused our study conceptually on how communication technology supports work system resilience (Neville et al., 2022). Work system resilience is the intrinsic ability of a work system to adaptively adjust its functioning before, during, or following events, changes, disturbances, and opportunities and thereby sustain operations required to achieve the system’s goals and mission (adapted from Nemeth et al., 2011). Distinguishing urgent from non-urgent communication through the select of the communication medium has been identified to influence response time for physicians in response to nurse requests (Manojlovich et al., 2021). Our study setting exemplifies a complex, high-consequence environment where real-time, synchronous communication is critical for safe and resilient operations (Roth et al., 2009).
Theoretical Concepts Informing Our Analysis and Recommendations
Our study is informed by a variety of theoretical concepts drawn from the naturalistic decision-making (Patterson & Hoffman, 2012), human factors (Carayon et al., 2006; Patterson et al., 2022; Roth et al., 2002), and resilience engineering (Neville et al., 2022; Roth et al., 2009) literature. Based on the naturalistic decision-making literature, we define healthcare communication through technology as the macrocognition function of coordination, where communication is an activity conducted in a complex setting with experts in specialized roles (Patterson & Hoffman, 2012). Short, coded communications occur between hospital staff specializing in perinatal care based upon a shared common ground for relevant medical terminology, policies, technologies, and social norms for responding to requests for interaction across roles (Clark & Brennan, 1991). Based on studies of how clinicians use artifacts in hospitals that do not always fully meet users’ needs (Patterson, 2018), we expect that where there are poorly met needs, informal “covert” communications technologies are used as workarounds in addition to formally provided and sanctioned communications technologies. Finally, we build upon a theoretical framework of how communications and alarm technologies can make design choices that clearly distinguish between more and less urgent communications, particularly when communications interrupt ongoing work (Patterson et al., 2022). For this study, we were particularly interested in how emergency broadcasts are made simultaneously to a group of practitioners expected to respond immediately. If not responded to immediately, these broadcasts are escalated. Different communication technologies, modes, escalation pathways, and filters are employed based on the urgency of the outgoing communication and the current urgency of ongoing work for the recipients.
Contextual Factors in Coordination in Perinatal Care Using Communication Technologies
In our study setting, clinicians do not always know the individuals serving in a particular role at the hospital or the best way to quickly communicate with them while minimizing interruptions to ongoing patient care activities. For example, an attending physician specializing in obstetric and gynecologic care on a postnatal care unit might need to talk with the nurse caring for a particular mother and child dyad. To identify the individual assigned to provide nursing care to the patient, the physician could search for a paper artifact that has the current shift’s assignments of nurses to patients or log in to the electronic health record and identify that a nurse logged in to care for that patient. If the hospital has a hands-free communication device, a voice command could query where the nurse caring for a particular patient is on the unit.
Prior Research on Hands-free Communication Devices in Intensive Care and Emergency Care
Several studies have assessed the effectiveness and impact of hands-free communication devices in hospitals. Cooney and colleagues (2018) found that implementation of a hands-free communication device reduced communication delays with staff in the Pediatric Intensive Care Unit (from a median of 120 seconds to 9 seconds to respond). Ernst and colleagues (2013) found that introducing a hands-free communication device was associated with shorter interruptions (from an average of 9 minutes before the hand-free device vs. 4 minutes after). Kuruzovich and colleagues (2008) found a decreased time to respond to patient requests following the integration of a hands-free communication system with a nurse call system; in addition, they unexpectedly found that nurses organized meetings and conducted conversations with other nurses through the system. Richardson and colleague (2010) conducted guided observations and interviews at an academic medical center and community hospital. They identified that the organization’s resilience, specifically the ability to adapt to changes from introducing new communications technologies, was necessary for effective use. They found that hands-free communication devices improved access to communication in general. However, there were challenges relating to workflow integration, increased interruptions to ongoing patient care interactions, and patient privacy concerns.
Contextual Factors with Hands-free Device Use in Perinatal Care at the Study Site
By a hospital protocol specific to perinatal care areas at the study site, communications not received through the hands-free communication device are forwarded to alternate recipients specified in a hospital-defined communication chain. Each chain has a primary, secondary, and tertiary step. Typically, the steps go up an administrative chain. For nurses, the communication chain is from a nurse to a charge nurse, from a house supervisor to a nurse manager. For physicians, the communication chain is from a resident physician, advanced practice provider, or fellow physician to an Ob Chief resident or an attending physician in a specialty service. Certified Nurse Midwives’ next step is the attending physician in obstetrics and gynecology.
Contextual Factors with Secure Chat in Electronic Health Records in Perinatal Care
Chat communications conducted through the electronic health record are asynchronous and initiated by logging in to the electronic health record and sending a message to a selected clinician; the clinician’s status is displayed on the messaging interface based on whether the user has logged in or selected options indicating that they are not available. These messages can be linked to specific patient records if desired.
Guiding Research Question
In summary, with this mixed methods study, we aim to identify resilience gaps with the appropriate and inappropriate use of three communication technologies in perinatal care and the unintended impacts of communication technology use on the clinician-patient relationship. Our guiding research question is: What are the characteristics of information and contextual factors that inform the choice of communication technology to use when providing perinatal care in the hospital?
Methods
Study Design Overview
This mixed methods study aimed to understand how hospital staff on perinatal units choose from three possible communication technologies for urgent and non-urgent communications, how design and training decisions facilitate or hinder effective use, and how they impact building and maintaining rapport with patients. Focus groups with clinical staff were conducted, and call logs were analyzed. The data from these two data collection methods were analyzed concurrently and in a mutually converging fashion; additionally, the focus groups included both exploration of problems and social norms as well as participatory co-design to identify characteristics of desired solutions to problems. Log data and transcripts of focus groups were triangulated to assess the frequency and extent of issues, as well as to gain context for reasons for issues and possible solutions. The Institutional Review Board approved this research.
Study Design Rationale
We used a mixed methods approach, combining focus groups with log data. First, we conducted focus groups with perinatal care staff (nurses and physicians) to understand from their perspectives about when and how to use and not use communication technologies. Then, we analyzed behaviors using automatically logged events from the hands-free communication device used as the primary communication technology by nurses, physicians, and other clinical stakeholders. We augmented focus groups with another method because it is unlikely that participants in a focus group would feel sufficiently safe to self-report not using a hands-free communication device at all or using it in unintended ways. The log data also provided insight into practice on the unit at a staff population level, which is impossible with focus groups. Partly due to the pandemic, we did not conduct direct observations. In addition, only some individual clinicians are observed during direct observations, limiting the ability to generalize insights. In addition, observed clinicians may use communication technology more as intended rather than how they would usually use it in practice. This distinction has been framed as the difference between work-as-done (WAD) and work-as-intended (WAI) by Hollnagel and Woods (1983), who were influenced by Leplat and Hoc (1983). Before the COVID-19 pandemic, we piloted direct observations of various clinicians, patients, and family members on perinatal care units, which informed our study design. In particular, we observed one clinician having difficulty with voice recognition to call another clinician and negative impacts on a clinician-patient-support person discussion immediately before discharge home with their newborn. During our pilot observations, we found that it was not easy to understand the context around health care team communications, particularly when we could not observe prior clinician interactions and interactions after our observations.
Setting
We conducted our study in a southeastern quaternary care hospital in the United States on the perinatal care units of Labor and Delivery, Postnatal Care, and Antepartum, which cared for patients admitted to antepartum and for more complex postpartum patient needs.
Three Communication Technologies
Three communication technologies were used in the hospital, which had sophisticated information technology support. The formal expectations, as depicted in training artifacts, for when to use the hands-free communication device as compared to a pager were provided for emergency (immediate response), urgent situations (less than 15 minutes to respond), and non-urgent situations in the perinatal care units of the hospital (see Figure 1). Communications conducted through the hands-free communication device were real-time and synchronous, which were transmitted on a speaker worn around the neck or in a pocket of clinical staff members who received calls and initiated calls using voice commands or by pushing buttons (call, hold/do not disturb, scroll, select) on the device. Pagers assigned to individual clinicians supported asynchronous communication by displaying alpha-numeric messages transmitted to the recipient, typically a phone or hospital room number. There were also role-specific pagers, such as an antepartum pager given to the resident responsible for covering the antepartum unit to wear in addition to their pager. Pages not returned within 30 minutes were repaged, and after 30 more minutes, the next person in the communication chain was contacted by page. Escalation protocol for emergency, urgent, and non-urgent communications in perinatal care.
Focus Groups Data Collection
A series of one-hour semi-structured virtual sessions with two to seven participants each were planned for the focus groups. Inclusion criteria were being a healthcare practitioner using a hands-free communication device to communicate with others in a hospital setting, being 18 years or older, and having been employed at the hospital for more than 2 weeks. Four 60-minute virtual sessions were scheduled on a remote meeting platform, two with nurses and two with physicians. Participants were recruited by posting an announcement on internal hospital listservs and sharing the announcement via an email from investigators at the participating hospital. Inclusion criteria were: (1) currently a healthcare practitioner who uses a hands-free device to communicate with others providing perinatal care in a hospital setting, (2) 18 years or older, and (3) employed at the hospital for more than 2 weeks. We used a moderator’s guide and dedicated note-taker during the focus group with semi-structured questions. The interdisciplinary team developed the moderator’s guide focusing on the appropriate and inappropriate use of the three communication technologies used in perinatal care units (hands-free device, EHR chat, and pager) and how communication technologies impact clinician-patient relationships. A single investigator with significant experience moderated all the focus groups [E.P.] while other investigators took real-time notes. Sessions were video-recorded, and Descript software Version 66.1.1 was used to transcribe the sessions automatically. Participants provided written consent and received a $50 gift card for participation.
The semi-structured questions asked during the focus group were: (1) When is the system useful or a communication is appropriate to have on the [hands-free communication device]? (a) What about problematic uses or features of the [hands-free communication device]? (2) When is the system useful or a communication is appropriate to have on [electronic health record chat]? (a) What about problematic uses or features of [electronic health record chat]? (3) When is the system useful or a communication is appropriate to have via pagers? (a) What about problematic uses or features of the pager system? (4) Do you have any thoughts about how we might reduce interference with in-person discussions, such as a nurse talking with a patient preparing to be discharged when receiving an incoming [hands-free communication device] call? (5) One possibility for system improvement is to reduce unnecessary communications. For example, with email, some companies discourage use of “reply all” for comments such as “Thank you!” Are there any opportunities to reduce or eliminate communications along these lines? (6) Overall, do you have any other thoughts on what we have talked about today? (7) Is there anything that we didn’t ask, that might be helpful for us to understand from your perspective?
Focus Groups Data Analysis
A thematic analysis was employed to analyze the focus group data (Garvin et al., 2019; Guest et al., 2011). Responses to focus group questions about the three technologies were pooled concerning valence (positive and negative responses), and emergent themes were identified.
After initial familiarization with the data, we used memos to generate initial themes related to the conceptual focus. Three investigators developed memos using the comments feature in a word processing program on transcriptions from the two focus groups with nurses. Each investigator generated an initial codebook independently, and then a series of three one-hour meetings were held to reconcile the differing perspectives through discussion.
During a series of meetings, three investigators agreed on the initial codebook using the data from the first two focus groups. One investigator created a spreadsheet with self-identified parsed statements, meaning that the investigator selected what transcript segments equated to a single code from the codebook. Two investigators coded the set of two transcripts, who had participated in real-time in both focus groups as observers. Differences in coding were identified and discussed, resulting in a modified codebook. Following the two additional focus groups, the process was repeated with an investigator with medical expertise to generate codebooks and resolve differences independently. The initial two focus groups were then recoded using the revised codebook. Three investigators reached consensus regarding the final set of codes from the nurse focus groups. The same approach was used for the physician focus groups, although one of the two investigators had medical expertise. The final codebook (In the Supplementary Material Table 1) was developed after a series of meetings with three investigators and applied to all parsed quotes from all four focus groups. Inter-rater reliability was assessed using percent agreement among two raters (one a graduate student in medical informatics, one a medical student) for a 20% randomly selected (using = RAND function in Microsoft Excel v2401) sample of parsed phrases, with a required minimum of 70% agreement.
Log Data Collection from Hands-free Communication Devices
The data from March 12, 2020 until January 08, 2021 for the call logs includes 137,956 rows of incoming and outgoing calls from the hands-free communication devices used on the postnatal care unit. The selection of the time frame for the hands-free communication device data is based on when we started our data collection for pragmatic reasons as soon as we obtained approval from the Institutional Review Board. Data fields have the date, time, call duration, user role, and event type.
Log Data Analysis
The automatically logged reasons were analyzed with descriptive statistics for event types. We manually reviewed all other data to determine its structure and created a data dictionary. The data elements that were initially identified were: (1) Accepted Call from X, (2) Added Nurse to X (Department), (3) Broadcasted OB Fast less than 3 seconds, (4) Broadcasted to a care unit, (5) Called X, (6) Left Message for X, (7) Missed Call from X, (8) Logged In, (9) Logged Out, 10) Do not Disturb (DND) On, 11) Do Not Disturb (DND) Off, and 12) incomplete calls (multiple types). For the incomplete calls, the categories were: (1) Accepted, (2) Incomplete due to not logged in, (3) Incomplete due to not answered/rejected, (4) Incomplete due to call canceled by originator, (5) Incomplete due to the call wait being rejected, (6) Incomplete due to phone not answered, (7) Incomplete due to busy, (8) Incomplete due to not online, (9) Incomplete due to DND/blocked call, (10) Incomplete due to call timed out and (11) Incomplete due to leaving a message. Two investigators [LW and EP] created a spreadsheet with all the call categories. We reviewed training manuals from the hands-free communication device company and consulted with hospital information technology personnel. Two interdisciplinary team meetings were conducted with all investigators to review and group the categories into analytic tables. Tables were further modified following peer review. The final tables are presented in this manuscript.
Findings
Focus Groups
Two focus groups were conducted with nurses (four participants on May 2, 2022; two participants on May 4, 2022), one focus group with resident physicians (two from Obstetrics and Gynecology on November 17, 2022), and one mixed focus group (one nurse, one attending physician from Family Medicine, one attending physician from Obstetrics and Gynecology, one fellow physician from Maternal Fetal Medicine, and one hospitalist pediatric physician on November 15, 2022).
Number of Nurse and Physician Statements for Codebook Categories.
Two categories were identified unique to the physician statements: (1) communications in the operating room and (2) burnout. For communication in the operating room, resident physicians stated some concerns on behalf of the Ob Chief resident, where the perception was that the Ob Chief resident should never turn off the hands-free communication device, even while in the operating room, due to the importance of the Ob Chief resident’s duties. In addition, by not using the DND function while in the operating room, the resident physicians worried that they would not as fully address urgent concerns for patients not in the operating room as compared to if they were in other locations when receiving the call, including having immediate access to an electronic health record. When asked directly by a facilitator during a focus group, nurse participants were not willing to take on any hands-free communication devices from physicians for several reasons, one of which would be challenges with knowing which device was signaling an incoming call if multiple devices were worn at one time and the difficulties of targeting only one device to send an outgoing call with a voice command if wearing multiple devices. It is possible that placing hospital-provided phones in a dedicated location in an operating room with Vocera calls forwarded to them would facilitate the willingness of the circulating nurse to respond to incoming calls.
Log Data: Duration
Duration, Number, and Percentage of Accepted Incoming and Outgoing Calls on Hands-free Device.
Log Data: Data Exclusion Protocol
We first excluded 13,399 administrative events when analyzing the log data. These events are automatically logged by the hands-free communication device to enable information technology staff to troubleshoot issues with the system but do not represent communications between users.
Log Data: Reasons for Incomplete Calls
Reasons for Incomplete Calls and How to Reduce Them.
Triangulation on Incomplete Calls When Not Logged in: Focus Groups and Log Data
Each clinician has two requirements to successfully log in at the beginning of a work period in the hospital. (1) Log in as an individual by vocally stating a first and last name, and (2) Log in to a role by vocally stating to be added to a group (e.g., “Add me to OB OR Attending”). Individual calls to a person are based on the person logging in to their name. Broadcast messages to a role are received based on the person logging in to that role. Users log out by saying, “Log me out,” and can find out whether someone else remains logged in on their device because they did not log out by saying, “Who am I?” Therefore, there are multiple potential explanations for unsuccessful calls relating to logging in, including issues with voice recognition and logging in to one, but not both, of the required ways to log in. One physician noted in a focus group, “You can’t necessarily like just say call [room number’s nurse] because sometimes they’re not logged in, or it doesn’t link,” and “the nurses who sign in to their patient’s room, that’s really helpful. But when they don’t do that, then I’m, like, very lost sometimes.”
There are several possibilities to reduce these incomplete calls. Users could receive refresher training, reminding them to log in at the beginning of their shift and log out at the end before handing over the device to the next shift. Hospitals can add reminders to the hands-free communication devices to log in before allowing the first call to go out. Hospitals could also use financial and other incentives to reward logging in within 2 minutes of the start of a work shift.
The most frequent category was logging in (5480 events) and logging out (5460 events). These events indicate that users generally do not forget to log out after logging in since the number is nearly the same. These findings suggest that the issues raised with voice recognition for logging in are likely highly frustrating to the affected individuals. Although likely not experienced by many individuals, inclusive design requires accommodating diversity in voice sounds and accents, including providing alternate means to log in and issue commands (Pattison & Stedmon, 2006). We propose recommendations to address these issues and note that the majority of clinicians do not experience them. The large number of events associated with logging in and out also ensures that the hands-free communication device is used routinely in practice by many or most of the clinicians expected to use it.
Triangulation on Incomplete Calls When Need to Add a Nurse to a Department: Focus Groups and Log Data
One other administrative event type was infrequent, adding a nurse to a department (41 events). Although none of the focus group participants mentioned challenges with contacting a user not assigned to the correct department, this finding suggests that there might be an opportunity to streamline or improve the reliability of assigning users to departments. Future research could explore the typical and unexpected workflows with creating users and assigning them to departments and when the administrative event of adding a nurse to a department would occur in those workflows. For example, suppose it is impossible to contact a user not assigned to a department through a hands-free communication device. How is that communicated to the person initiating a call? What does a user experience when assigned to the wrong department? Do they receive escalated and broadcasted calls that are not intended for them?
Triangulation on Incomplete Calls When Do Not Disturb is on: Focus Groups and Log Data
Our remaining event log data indicate that the Do Not Disturb (DND) function is rarely used, with the same relative frequency for turning it on (650 events) and turning it off (643 events). This insight is crucial because it indicates that the hands-free communication device is used most of the time by all users, in that users are not frequently setting the “Do Not Disturb” feature on and then fail to set it back to off later. This insight aligns with the comments from the nurses in the focus groups that they are afraid to use the “Do Not Disturb” feature or lower the volume at night to avoid waking patients in rooms because they do not want to forget to turn it back on later.
The data on the blocking function for DND being on adds to our understanding of how frequently using that feature impacts the ability to reach a user immediately. The common reason for avoiding using DND was forgetting to turn it off later when incoming calls would be wanted. Both physicians and nurses expressed concerns about low-priority incoming calls (e.g., do you like us to grab you a drink from the cafeteria?) interrupting extraordinarily sensitive ongoing conversations (e.g., a fetal demise). Two night shift nurses wanted to avoid waking sleeping patients when receiving incoming calls. A physician wanted to avoid constantly turning down requests for calls through the hands-free communication device when in the hands-on portion of delivery. Multiple physicians expressed concern about disturbing the operating room by answering incoming calls during an operation. One attending physician wanted hands-free communication devices not to be allowed in the operating room to enhance patient safety and possibly hand the devices over to someone in another role, such as an operating room nurse.
Based on the focus group data, it seems likely that physicians would be more satisfied being able to get nurses, even if it interrupts a nurse providing care to a patient in a room or disturbs a patient who is sleeping. Nurses would likely be more satisfied that the incoming call was blocked while using the function to indicate they were unavailable to take a call. One design option would be to do a callback when the receiver is back online to the person who initiated the call, similar to automated callback features for customer support services. One nurse focus group participant said, “Maybe I’m bad. So, if I’m talking to a patient and it goes off, I just say no, and then it usually transfers to the phone at the unit [clerk], and they’ll answer, and they’ll take a message for me.” Another approach is to provide support to negotiate an interruption, either by adding buttons representing a request to delay the acceptance of an incoming call (e.g., a “wait” button) or with communications training (e.g., recipient says “please wait”… followed by “go ahead” within 60 seconds). Finally, the design could support prospective memory to make reinitiating a task after task switching more robust and safer for patients. For example, the state of the situation at the time of the interruption could be provided, such as by highlighting differences between data at that time versus the present time and showing what task(s) had been initiated but not yet completed when the interruption occurred.
Triangulation on Emergency Broadcasts (OB Fast): Focus Groups and Log Data
The log data provide insight regarding the issues raised in the focus groups with accidentally initiating an “OB Fast” broadcast, where the OB FAST contacts a team of clinical personnel assigned to respond to obstetric emergencies, similar to teams responsible for responding to Code Blue broadcasts. There are a total of 606 OB Fast broadcasts, which represent 4.5% of the logged events. Although it is unknown how many broadcasts are genuine emergencies and how many are accidental, this finding implies a relatively small number of events. Therefore, the criticality of the type of broadcast suggests a need to improve the system, but the frequency of improperly broadcasted emergency events is somewhat low in general. Nevertheless, related research demonstrated that increasing the Positive Predictive Value for broadcasted Code Blue auditory alarms reduced response times in hospitals by nurses (Hansen et al., 2023). Therefore, we recommend changing the design to require a more purposeful approach to setting off an OB Fast, such as holding a button for 4 seconds, since a double-click trigger was described as going off frequently in focus groups and was an easy mistake to make. Similarly, there are relatively few instances (519 events) where a broadcast to the Postnatal Care Unit was made following both the primary person and the second assigned person not responding to a request to talk on the hands-free communication device.
Triangulation on Reasons for Incomplete Calls: Focus Groups and Log Data
The following most frequent categories might not be amenable to interventions to reduce them: (1) Receiver did not answer, (2) Originator canceled call, (3) Left message, and (4) Receiver is busy. One possible exception is when the originator cancels a call due to voice recognition issues. For example, one focus group participant said, “Well, I would say, this is anecdotal, and I would say probably 20% of the time it doesn’t understand me or maybe a quarter of the time it doesn’t understand me on the first try.” Another focus group participant said, “Getting other people sometimes is an issue, probably when I’m not pronouncing the name correctly.” Further research with direct observation might provide insight into how often and when voice recognition does not work.
In some cases, an alternative communication strategy was used, such as finding the staff member in person when the voice recognition was problematic; one focus group participant said, “Oftentimes, the [hands-free communication device] doesn’t understand you. It takes three to 4 minutes to try to call someone that is in a room, you know, three hundred feet away. And so by the time that you’ve dealt with the frustration saying ‘call this person, call this person’, you could have actually gone and talked to them face to face.”
Triangulation on Incomplete Calls when Not Connected to the Wireless Network: Focus Groups and Log Data
The next category is a new concept from the log analysis that was not identified in the focus groups, in that a user was not on the wireless network in the hospital (703 calls, 3.8%). Although relatively infrequent, it is relatively easy to identify “dead spots” in the hospital where there is no network connectivity, address them, and increase connectivity reliability through hardware and software upgrades. Further research with stakeholders knowledgeable about this category could inform the reasons for being offline and the impacts on patient safety and quality of care when connectivity is unavailable or lost during a conversation.
Discussion
Overall, we learned that the most urgent communications requesting immediate assistance from hospital staff are appropriate for untargeted, emergency auditory broadcasts to everyone in a shared physical space, whereas the least-urgent communications that are primarily informational without any associated request for action are appropriate for individual messages in a non-interruptive threaded chat.
Appropriate Use of Hands-free Communication Device: Emergency Broadcasts
From this study, we learned that the most urgent communications, emergency broadcasts for immediate assistance in inpatient maternity care were deemed most appropriate for the synchronous channel to the unit with the hands-free communication device. As found in other domains, front-line clinical staff highly value having a dedicated community channel for emergency broadcasts with few false alarms (e.g., Hansen et al., 2023; Militello et al., 2007; Welch et al., 2013).
Appropriate Use of Hands-free Device: Urgent Communications Needing an Immediate Response
In addition, urgent communications needing an immediate response, such as requesting admission to the Neonatal Intensive Care Unit, were similarly appropriate for the hands-free communication device. However, during the research period, the pediatric medicine service did not use the device other than inclusion in OB Fast teams for delivery emergencies requiring team members from the Neonatal Intensive Care Unit (NICU). Different communication technologies for various types of clinicians align with a historical “guild and workshop model” for independent decision-making within a physician specialty area (Wears & Perry, 2002). Alternative communication paths for urgent situations during the research period, where a response was expected within 15 minutes, included nurses calling the nurse practitioner on the service and having them walk around to find the specific clinician, using a pager, and walking to their care unit to see them for an in-person conversation. After the research period, the pediatric medicine service used Vocera for some time, followed by hospital-provided phones that received and sent Vocera calls. Using phones prevented missing Vocera calls when the pediatricians were in other buildings providing care and reduced disruptions to ongoing discussions with patients, support persons, and hospital staff on postpartum care units and in the Neonatal Intensive Care Unit. An essential feature of the phones is that they do not require signing in to receive calls. So incomplete calls due to not logging in would likely be addressed, as well as redirected calls because the person calling did not realize that the sign-out had already been completed. The following physician is now taking care of patients.
The typical call duration using the hands-free communication device was 16–60 seconds, with calls longer than 2 minutes rare and a small number of calls up to 10 minutes long. These findings are shorter than in a 2009 study, where interruptions were an average of 4 minutes using a hands-free communication device in the emergency department (Ernst et al., 2013). The duration may reflect the different communications in emergency departments and perinatal care units.
We found that some calls were incomplete due to unreliable connectivity on the wireless server. A prior study of secure chat similarly found that dropped connectivity hampered using secure chat for communications (Feinberg et al., 2019).
Appropriate Use of Secure Chat in Electronic Health Record: Informational, Non-urgent Communications
Electronic health record (EHR) chat was appropriate for informational, non-urgent communications without immediate response. This insight aligns with findings from a prior study that the number of non-urgent phone calls was significantly reduced with the implementation of EHR chat in an emergency department (Luu et al., 2022). Our findings are similar to a 2019 study of secure chat on phones (Feinberg et al., 2019), where they found that tailoring had to be made after implementation to ensure that the message was received, including adding message delivery and read receipt functionality. This mode of communication avoided a potential HIPAA violation (although see Freundlich et al., 2018 for a discussion of whether secure chat complies with HIPAA) from someone overhearing the communication, was safe, was not entered into the permanent medical record, and supported one-click access to access a patient’s chart. For example, if a nurse requests that a physician write an order via chat, the physician can click on the request to open the patient’s chart, which provides immediate access to writing an order.
Conversely, if a nurse uses a hands-free communication device to call a physician to request an order be written, the provider needs to remember the request, go to a computer, log in, and remember the patient’s name and what was requested to be ordered. However, more complex trade-offs were made. For example, contacting a provider more quickly could result in violations by overhearing information protected by the HIPAA. In some situations, using the hands-free communication device negatively affected efforts to build patient rapport during sensitive conversations. In some situations, needing to respond to urgent incoming calls violated social norms to avoid answering calls when performing or assisting an operation.
Communication Technology Impacts on Clinician-patient Relationships
Interestingly, although a prior study found that one in four nurses felt receiving many phone calls from families in the Intensive Care Unit disrupted their work activities (Gurses & Carayon, 2007), we did not identify this as a theme.
Resilience Gaps, Recommendations, and Scientific Contributions
Resilience Gap, Recommendations for Technology Design and Use, and Conceptual Contribution.
Similar to a prior study following the implementation of a hands-free communication device, there was little routine use of pagers for communication between providers except for unusual workflows (Richardson & Ash, 2010). One surprising insight was that hospital pagers are used by policy for some urgent communications. Specifically, pagers are the sole formally approved means for contacting physician services not on the hands-free communication system beyond communicating in person; pagers are used to contact a physician service that does not reliably use the hands-free communication system if a communication request is not answered within 15 minutes. A related finding was that some individual physicians arranged for all paged communications to be redirected to their personal smartphone devices through the hospital’s information technology infrastructure. Based on that insight, we recommend the continued use of pager devices in hospitals, similar to the findings of Best (2021). From a resilience perspective, the pager system can be viewed as a backup means of communication if other means are unavailable (Neville et al., 2022).
An unexpected insight was that using the hands-free communication device in the operating room setting was problematic for supporting perinatal care in the hospital. Attending physicians wanted to eliminate interruptions during pregnancy and childbirth-related operations. In contrast, resident physicians, particularly the Ob Chief resident, did not feel they had the authority to give their hands-free communication devices to nursing staff during the operation. With multiple pagers to monitor at once from physicians, a nurse can triage incoming communication requests. With multiple hands-free communication devices in addition to their own worn around a neck or in a pocket, it is unclear how a voice command could be targeted to a specific device or how interruptions could be triaged given the frequent use of the system for nursing personnel. Providing additional support for negotiating whether to receive an incoming call (an interruption of ongoing activity) aligns with research on how to support negotiations about interruptions when using a human-computer interface (McFarlane, 2002).
We were surprised that a nurse focus group participant stated that calls on the hands-free communication device would be routed to the unit clerk, who would then take a message if the response was “No” to an incoming call. Suppose this interpretation by the focus group participant is incorrect, and indeed, calls are routed to the next nurse on the unit rather than the clerk. In that case, we recommend targeted training about when to use the “no” command with the hands-free communication device and the impacts on others on the unit when it is used. Suppose this participant was correct and this strategy was viewed as beneficial. In that case, we recommend treating this as an instance of “positive deviance” where resilience is enhanced through an innovation not formally documented in policies and procedures, thus a workaround. Suppose the strategy helps strengthen respectful conversations between providers and patients that build and maintain rapport by reducing interruptions and indicating that the current patient is the priority during the conversation. In that case, we suggest formally spreading this strategy to all units and incorporating it into training, policies, and procedures.
The findings regarding the usefulness of the DND feature were mixed, with some expressing a desire to use the feature if it was easier to remember to turn it back on again after a short time in DND. The DND feature would be needed in the operating room for an extended period. One possibility for the operating room would be to add a design option to make incoming calls private, such as by routing them through an earpiece (Jacques et al., 2006); in the operating room, it would be essential to do this option with a hands-free interaction to avoid touching the device. A possible issue with using an earpiece is confusion, where staff, patients, or the support person mistakenly think that a verbal statement intended for the person through the earpiece is meant for them. There appear to be several possibilities to enhance the design and use of the DND feature, which would likely be used more often if it was easier to remember to turn it off several minutes later or to route incoming calls while in DND mode to a particular person or role that is different than the default setting. One possibility would be to send a reminder to turn off Do Not Disturb every 10 minutes. Ideally, the time for the reminder could be specified by the user to tailor it to the current situation. These suggestions align with a finding that, for Automated Teller Machines (ATMs), when the card is returned after the money, there is a 96–100% greater rate of users forgetting to take their card out of the machine (Zimmermann & Bridger, 2000).
Many hospitals use somewhat outdated technology to facilitate communications and patient care (Best, 2021). However, there are opportunities to enhance these legacy systems and workflows to be more effective. In other domains, such as military command and control structures like the Army, communications technologies have been designed more recently, and there are fewer regulatory and bureaucratic barriers to change. Our design and training recommendations can potentially provide valuable insights even for these domains with more sophisticated communication technologies. For example, when many specialized roles have team-based and individual-based preferences for their primary means of communication, it can be challenging to broadcast emergencies to all in real-time without considering this variation. In addition, the urgency of incoming communications needs to be assessed in the context of the criticality of ongoing work and communications, similar to a prior study where NASA Johnson mission controllers listened in on the conversations of a work team before using Voice Loops to interrupt their communications (Patterson et al., 1999). Similarly, socially aware robots are taught to interrupt following social norms. Specifically, they interrupt the ongoing activity of the receiver without permission as a last resort and only for urgent requests. The social norm is to signal a desire to interrupt and wait for an invitation. The invitation is based on the perceived urgency of the request based on gaze, angle of the head, and other body language cues, relative roles in a hierarchy for the requester and person providing the invitation to interrupt, and the current activity of the person being interrupted (Avelino et al., 2021). Social norms for how quickly to respond to messages left for the recipient are influenced by communication technology; for example, messages on a personal device are typically responded to more rapidly than emailed communications (Kaju & Maglio, 2018). Hands-free communication technologies could add features such as providing insight into the location of an intended call recipient and whether an intended call recipient is engaged in another conversation on the hands-free device. These features could then allow the social norm of gauging the correlation between the intended communication and the recipient’s ongoing tasks before interrupting.
In healthcare, there is a novel element of the need to build and maintain patient rapport with uninterrupted communication during highly emotionally charged discussions. For example, a physician explained that a request from another physician asking if they wanted anything from the cafeteria on the hands-free communication device was potentially traumatizing for the patient who learned that their fetus was unlikely to survive.
Limitations
This study has several limitations. First, the data used in this study was obtained from a single site. Call log data were collected during the COVID-19 pandemic, and the hands-free communication device data, a small subset of all communications performed at the site, was captured. For this study, we did not analyze electronic health record chats, pagers, in-person conversations, phone conversations, or patient perspectives. The focus groups did not represent every user at the site who uses the hands-free communication device for their job role. Because there were no direct observations of behavior in the work setting targeting communication activities, some interpretations may reflect the ideal use of communication technologies more than the actual use. On the other hand, comments included descriptions of less-than-ideal choices about technology selection for communication, particularly by others. Finally, none of the recommended strategies have been evaluated for effectiveness in enhancing resilience.
Implications for Future Research
Our insights have some implications for future research. First, our recommendations can be studied for impact and effectiveness when implemented in a hospital on postpartum care units and subsequently broadened to other care settings. The use of hospital-provided phones by the pediatric medicine service after the research period could be assessed for effectiveness and possibly spread to other users of the hands-free communication device. Pager technologies are used partly to fill gaps in communication needs that are not well-served by other available technologies, mainly when a particular communication channel is used so often that it is hard to identify that an incoming communication needs a quick response. Revealing the continued use is essential to converting those purposes to different technologies, and it is challenging without directly observing practitioners in the work. While infrequent, the administrative event of assigning a nurse to a department raises questions that additional research could answer about how assigning users to organizational areas could be streamlined or made more reliable and what the consequences are for “wrong unit” nurses’ assignments. As large language models (LLMs) improve, there are opportunities to convert one-by-one communications, such as chats through the electronic health record, into summarized threads highlighting key aspects. Information fusion technologies could be improved to the point where filtering alarms and communications inappropriately labeled urgent could remove “noise” from channels dedicated to emergency broadcasts and urgent messages that are appropriate to interrupt ongoing activity. Extending flexibility in modifying the volume and timing of incoming calls, particularly during emotional conversations, could be studied concerning an anticipated reduction of burnout and improvement in relationships between providers and between providers and patients. Finally, we anticipate that dashboard displays could be used to support direct communications and identify missing or delayed urgent communications, which can reduce patients “falling through the system’s cracks.”
Conclusions
We conclude that hands-free communication devices are helpful for urgent real-time communications except for personnel currently working in operating rooms, that secure chat in the electronic health record is useful for non-urgent communication, and that variable use of the pager system by physician services makes it challenging to remove safely as an option for communications. We provide recommendations for design and training strategies based on better aligning the communication technologies with social norms for when to interrupt, when to delay an interruption, and what modality to use to communicate the appropriate level of urgency of the communication request. We recommend using interruptive communication technology for urgent information and an electronic health record chat for less urgent information that does not require an immediate response. We recommend forwarding and sending calls from a hospital-provided cellphone to the hands-free communication system for provider roles that do not align with the intended workflow for hands-free communications.
Supplemental Material
Supplemental Material - Distinguishing Urgent From Non-urgent Communications: A Mixed Methods Study of Communication Technology Use in Perinatal Care
Supplemental Material for Distinguishing Urgent From Non-urgent Communications: A Mixed Methods Study of Communication Technology Use in Perinatal Care by Laura Wilson, Alison M. Stuebe, Marina Pearsall, Megan Mansour, Kristin P. Tully, Jennifer H. Garvin, Kevin Jones, and Emily S. Patterson in Journal of Cognitive Engineering and Decision Making.
Footnotes
Author’s Note
The authors are solely responsible for this abstract’s contents, findings, and conclusions, which do not necessarily represent the views of AHRQ.
Acknowledgments
We thank Dr Alison Stuebe for designing the study and providing the de-identified dataset used in this thesis with support from her analytics team and clinical colleagues. We thank Dr Alison Sweeney and Jenny Jensen for their critical clinical insights, Noah Green for extracting and de-identifying the data, Fernanda Ochoa Toro for de-identifying the data, Dr Maria Brunette for insights related to disparities considerations for mothers with feeling comfortable and getting equitable care for urgent clinical needs in a hospital setting, Dr Susan O’Hara, Anne Dew, Dr Austin Mount-Campbell, Dr Dana Womack, and Dr Laurie Wolf for insights with the use of hands-free communication devices by nurses, physicians, and clerks at other hospitals, and C.J. Hansen for support with strategies for the analysis of the log data.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by grant number R18HS027260 from the Agency for Healthcare Research and Quality (AHRQ), U.S. Department of Health and Human Services (HHS).
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
