It’s like sending a message in a bottle: A qualitative study of the consequences of one-way communication technologies in hospitals

Abstract

Objective

We examine how physicians and nurses use available communication technologies and identify the implications for communication and patient care based on the theory of workarounds.

Materials and Methods

We conducted a qualitative study at 4 U.S. hospitals during 2017. Researchers spent 2 weeks at each hospital conducting unit-based observation, shadowing, interviews, and focus groups with nurses and physicians. Using an iterative process, we inductively coded and thematically analyzed data to derive preliminary themes. The theory of workarounds provides an organizational lens on workarounds, consisting of 5 components: antecedents, types, effects, managerial stance, and organizational challenges of workarounds. The first 3 components of the theory helped us to organize and explain our findings.

Results

Communication technologies consisted of pagers and telephones. Antecedents to workarounds included one-way information flow, differential access related to differences in technology types, and technology mismatch. Types of workarounds included bypassing a variety of obstacles and substituting for unavailable resources. Direct effects of workarounds included pager fatigue, interruptions in patient care, and potential errors.

Discussion

One-way communication technologies created an environment where workarounds could flourish. By placing results within the context of the theory of workarounds, we extend what we know about why and how workarounds develop, and offer strategies to minimize workarounds’ adverse effects.

Conclusions

Through the theory of workarounds, we see that there is a trajectory to workarounds with potential consequences for clinicians and patients. Two-way communication technologies could minimize workarounds and gaps in information exchange, and reduce unnecessary interruptions and the potential for adverse events.

INTRODUCTION

The innovative communication technologies that we use in our social lives are slowly making their way into health care, where many clinicians still depend on outdated and potentially inadequate technologies.¹ For example, a national survey of hospital-based clinicians found that pagers were the technology most commonly used.² Nevertheless, with more information and communication technology (ICT) being deployed in hospital settings, clinicians must navigate a growing but fragmented communication space.³

Research suggests that increasing use of ICT affects communication between nurses and physicians, and contributes to more,^4,5 not fewer,⁶ communication difficulties, in part because of alterations in the content and patterns of communication brought about by the introduction of ICT.⁴ Structuring communication exchanges in electronic format can also create ambiguity and reduce flexibility.⁴ Moreover, a recent systematic review of the impact of mobile technology (ie, hand-held devices that facilitate two-way communication or data transfer) on teamwork and communication in hospitals concluded that although mobile technology has the potential to improve communication significantly, there are still key organizational, technological, and security challenges that must be addressed.¹

Understanding how communication technology is used in hospitals and the potential effects on communication practices and patient care delivery thus remains an important issue. In particular, one underexplored issue is the use of workarounds to offset ICT shortcomings. Practically ubiquitous in health care, workarounds are goal-driven adaptations or improvisations that individuals and groups use to reduce barriers to completing tasks or defined routines.^7–9 Workarounds can have positive effects, such as when they lead toward greater efficiency or effectiveness,⁸ and many can be sanctioned by the organization.¹¹ However, workarounds are also associated with significant errors,^7,10 yet how and why ICT-related workarounds develop is not well understood. Theory can help answer questions of how and why phenomena exert their effects by identifying relationships between concepts of interest and organizing them in a manner that facilitates empirical testing.¹¹

As part of a larger study on describing how communication technologies facilitate or hinder communication between physicians and nurses, we sought to examine how physicians and nurses use available communication technologies and understand the implications for communication and patient care. The theory of workarounds acted as a guiding framework to help us understand how and why communication technology is used, and the role it has in creating workarounds. The theory of workarounds was developed using studies that have identified and defined workarounds across various organizational settings, including health care. It defines the environmental phenomenon associated with workarounds, the various types of workarounds that occur in organizational settings, and the effects workarounds have. The theory further emphasizes how workarounds are viewed by those directly and indirectly impacted by them within an organization. We placed our results within the theory of workarounds to organize and better understand our findings.

MATERIALS AND METHODS

Study design, site selection, and recruitment

Detailed descriptions of the methods have been published previously.^12,13 This 4-year study was conducted at 4 hospitals in the Midwest United States, beginning with a quantitative survey and progressing to sequential qualitative methods.¹⁴ Findings from a national survey of 105 hospitals and subsequent telephone interviews with respondents from 8 hospitals were used to identify and recruit the 4 sites for more in-depth qualitative study.¹³ We selected sites to derive a variety of community and academic hospitals of different sizes, and included one hospital from the Veterans Health Administration. Chief Nurse Executives who participated in the initial survey selected the general care unit for our study, in consultation with nurse managers. During our site visits, unit leadership presented the study at staff meetings. However, the study team was available to answer questions and recruit and consent participants.

Data collection

At each hospital [ML] and 1 to 2 research assistants spent 2 weeks on each unit collecting data. We visited hospitals sequentially, with knowledge gained from each hospital used to inform our approach to subsequent site visits. We collected data via observation, shadowing, interviews, and focus groups, usually in that order. Using these methods in sequence allowed us to become familiar with the technology and clinician workflow at each site and build rapport with participants before conducting interviews and focus groups.

Observation

Our team conducted individual observations during various shifts to understand how workflow and communication practices varied depending on the time of day, staffing ratios, and patient volume. Over the course of the first week of data collection at each site, each observation session per researcher lasted between 2 and 4 hours per day, so that on average each researcher spent 18 hours conducting general observation. Observation captured verbal and nonverbal communication and helped identify nurse and physician task behaviors, both of which produced discussion points for focus groups and interviews specific to each site. During observation, we took handwritten unstructured field notes on workflow and communication practices among physicians and nurses, which we subsequently transcribed electronically.

Shadowing

Shadowing is a specific type of observation in which a single individual is followed for a set period of time,¹⁵ and was useful for understanding contextual factors that influenced communication practices. On each unit, observers individually shadowed 4 to 7 nurses and 3 to 6 physicians for approximately 2 hours to understand their role responsibilities and choices for using particular communication technologies. Participants received $50 gift cards for shadowing experiences. We took notes during shadowing and electronically transcribed them into more detailed narrative encounters.

Focus groups and interviews

We conducted focus groups and interviews with nurses and physicians at each site. We recruited nurses from the units to stay after their shift; focus groups lasted between 45 and 60 minutes. If nurses were unable to join a focus group, we conducted individual interviews using the same interview guide. We conducted individual interviews or focus groups with physicians based on their availability, with individual interviews tending to be shorter than focus groups (range 10:00 to 75:37). Interviews and focus groups were semi-structured, allowing clinicians to answer our questions, discuss topics among themselves, and bring forward any additional communication and ICT challenges that they were experiencing. Participants received $40 gift cards for participating in focus groups or interviews. Our questions (the same for nurses and physicians) focused on communication practices and technologies, work relationships, and patient safety. Audio-recorded interviews and focus groups were transcribed verbatim. The use of data from multiple sources (observation, shadowing, interviews, focus groups) added credibility and validity to our findings.¹⁶ Our university’s institutional review board approved the study (HUM00092942). We obtained verbal consent prior to general observation, and written consent prior to shadowing sessions, interviews, and focus groups.

Data analysis

We used an inductive approach to analyze data, which allowed us to identify emergent themes. All authors were involved in data analysis, which consisted of iteratively comparing codes to the transcripts until consensus was reached. We generated code reports, which aggregate the data under the same code, and reviewed them to ensure consistency in coding. Code reports examined specifically for this analysis included health information technology, choosing the medium, manipulation, hierarchy, and relationships. Definitions for these codes are in Supplementary Table 1. Using the code reports, we identified patterns across codes and grouped them to develop themes. For example, we found that, when analyzed, the codes “one-way communication,” “differential access,” and “technology mismatch” led to the use of workarounds and were therefore grouped together to develop the theme, “antecedents to workarounds.” We used NVivo 8 (QSR International, Melbourne, Australia) to manage the data.

Theory of workarounds

Once we completed coding and identified preliminary themes, we sought a theoretical model that would help to interpret and contextualize our findings.¹⁷ We reviewed the literature and found that the theory of workarounds reflected our preliminary themes and could help further conceptualize our findings.⁸ The theory organizes workarounds into 5 “voices” or components that fall on a spectrum: starting with antecedents to workarounds, types of workarounds, direct effects of workarounds, management stance on workarounds, and finally organizational challenges and dilemmas that arise from workarounds.^8,9 Because our data were already organized into preliminary themes, we created a table with our codes, themes, definitions, and supporting quotes (Supplementary Table 2) and compared them with the theory of workarounds. We found that the themes of antecedents, types, and effects of workarounds correlated with our data, as described subsequently. Because we lacked data on management stance on workarounds and organizational challenges that arise from workarounds, we do not include these in our findings.

RESULTS

 

The total number of participants by role and site are in Table 1.

Table 1.

The number of study participants by role and site

	Site 1: Community Hospital	Site 2: Academic Teaching Hospital	Site 3: Academic-Affiliated VA Hospital	Site 4: Community Hospital	Total
Nurses	31	39	22	18	110
Physicians	6	16	18	5	45
Total nurses and physicians	37	55	40	23	155

Commonly used communication technologies consisted of pagers (alphanumeric and/or text) and telephones (hospital-provided mobile and/or personal smartphones). Alphanumeric pagers transmit not only numbers, but also additional relevant information such as patient name and clinical scenario.¹⁸ Text pages are sent from Web-based paging systems and allow messages of up to 240 characters to be sent, providing more detail than is possible via alphanumeric pagers. We found that the types of available communication technologies or devices varied across sites and between units within the same site (Table 2).

Table 2.

Available ICT and communication practices between physicians and nurses

	Site 1: Community Hospital	Site 2: Academic Teaching Hospital		Site 3: Academic-Affiliated VA Hospital	Site 4: Community Hospital
Unit(s)	Unit A (medical-surgical)	Unit A (telemetry)	Unit B (transplant)	Units A and B (medical-surgical)	Unit A (oncology)
EHR platforms
HER	Cerner	Sunrise Clinical Manager (Allscripts)		CPRS	EPIC
Mobile devices
Physician	alphanumeric pager, mobile phone	alphanumeric pager	alphanumeric pager	alphanumeric pager	mobile phone
Nurse	mobile phone	mobile phone	none	none	none
Communication practices via devices
Nurse to physician	-numeric page -text page	-numeric page -text page	-numeric page -text page	-numeric page -text page	−call physician’s mobile phone
Physician to nurse	-call nurse’s mobile phone -call nursing station phone	-call nurse’s mobile phone -call nursing station phone	−call nursing station phone	−call nursing station phone	−call nursing station phone

EHR: electronic health record; ICT: information and communication technology.

When we interpreted our results organized by the theory of workarounds, the implications for communication and patient care became apparent, uncovering some of the complexity surrounding workarounds and offering possible strategies to reduce any potential ill effects. Each of the 3 concepts from the theory of workarounds (ie, antecedents to, types of, and effects of workarounds) was associated with 3 themes, as indicated in Supplementary Table 2. According to the theory, there are numerous antecedents to, or phenomena associated with, workarounds.⁸ Each antecedent described subsequently is followed by a description of a type of workaround. We describe direct effects of workarounds all together, as we did not find that any one antecedent or type of workaround was associated with a specific direct effect.

Antecedents to and types of workarounds

The first and most significant antecedent we found was that messages were limited to a one-way flow of information because nurses could only send texts, not receive them, and this contributed to workarounds. Nurses had no way of knowing if physicians received or read their text pages unless they received a call from the physician or saw an order placed in the electronic health record (EHR) addressing the issue. If a physician did not acknowledge receiving a text page, the nurse was uncertain if the message had been read, as this nurse at site 1 explained:

I can't tell that you've [physician] read my message. I have no idea. I could sit here and keep bugging you all night and just annoy you, when really you've already made your mind up that you're not gonna do anything. But there's not an order that they would put in, like “I'm choosing not to do anything.”

A resident at site 2 also spoke about the uncertainty of messages, linking message uncertainty to numeric pages:

If it's just a number, you have no idea what it's about. (…) There's no way for us to tell if they're [nurses] sending us a number because our patient was having a stroke alert, if they're coding, if they're decompensating, or if their potassium is 3.6.

This antecedent contributed to a workaround in which nurses bypassed the obstacle of physicians not returning their calls by using a less information-rich modality to contact physicians. Instead of using the full range of capabilities of existing technology, nurses sent only numeric pages (even though physicians’ pagers were able to receive alphanumeric messages that could include additional context and content). This clearly achieved an important aim for the nurses by forcing a response or acknowledgement. This workaround developed because of inadequate information technology functionality in all 4 sites. Nurses could send a text page to physicians through Web-based paging systems, but they could not receive confirmation of delivery or a reply text. To deal with the uncertainty, nurses sent numeric instead of text pages to prompt a return phone call and eliminate any ambiguity about whether their messages were received. One nurse at site 3 used the analogy of sending messages in a bottle to describe text paging’s inability to close the communication loop:

I don't like using text page at all just because it does feel like a big question mark. You send the bottle out but you never know if it's going to come back or anything.

Differential access

Another antecedent to workarounds was differential access to communication devices, which also contributed to 1-way communication practices. Nurses were required to use hospital phones or pagers and did not have the option to use personal devices, whereas some physicians listed their personal cell phone numbers in the staff directory and set up their personal smartphones to receive pages. Only physicians had portable devices capable of receiving (but not sending) texts, while nurses had to use unit-bound computers with software only capable of sending texts. Differential access may have compounded perceptions of the occupational hierarchy between physicians and nurses, because while physicians did not have to respond to nurses, nurses did have to respond to physicians. For example, a nurse at site 1 said:

I think the power thing comes into play when they don't respond to you because they feel like they don't have to because they're the ones doing the orders, but if they were to reach out to us, we definitely need to call them back, you know?

A physician at Site 2 also associated differential access to technology with hierarchical differences between clinicians. He said that the hospital’s plan to get access for physicians and nurses to two-way text messaging could help reduce this difference:

Well I think there’s certainly vertical hierarchy. (…) Hopefully one of the things that text messaging does (…) is it’s so informal and so ubiquitous. I think it does break down the vertical hierarchy much more so [than one-way text paging].

Using the chain of command to bypass lack of responsiveness was a workaround that nurses sometimes used because of this antecedent; the lack of responsiveness was an obstacle to nurses’ ability to carry out their routines. Nurses did not have the authority to force a physician to respond because of their lower status in the hospital hierarchy, but their managers did, so nurses would go to their managers when physicians were not responding. During a nurse focus group at site 1, we were told,

So, one other option is to, it’s kind of a last resort, but you can call the house manager and be like ‘Hey, I’m having a really hard time getting a hold of this doctor. Can you try to figure something out?’ and she has kinda like, above resources. Like, she can call their office and their doctor, their boss basically, and try to get a hold of them a different way.

Technology misfits

A third antecedent developed from what the theory calls “technology misfits.”⁹ That is, there was a mismatch between technology and the realities and contingencies of day-to-day work. For example, while nurses sent numeric or text pages to physicians, physicians responded by telephoning the nurses on landlines at nursing stations. Although nurses at sites 1 and 2 were assigned mobile phones, each time a nurse worked, they could be assigned a different phone (and different number), so physicians knew neither which nurse called them, nor which patient they referred to. As a result, although both physicians and nurses had mobile phones in site 1, they did not use the phones to call each other.

Unavailable resources caused another type of workaround in which clinicians had to find a substitute for the resource that was missing.⁸ For example, on one unit at sites 2 and 3 physicians did not have hospital-issued mobile phones, so as a s ubstitute they had to find an available landline to return calls or pages. However, when physicians returned a call, the nurse who called them may have left the nursing station. A resident at site 3 explained,

They [nurses] step away from their phone. If you don't get back to them in a decent time, then you don't know who the nurse was or who the patient was. You have to get back to those kind of [calls] quickly.

Effects of workarounds

There were 3 direct effects of the workarounds that we identified. One was that some physicians and nurses described developing “pager fatigue” as a result of the high volume of pages. Nurses said that the number of pages could affect physicians’ ability to distinguish the importance of a page and their subsequent response time. Nurses also noted how the high volume of pages could desensitize physicians to the messages in the pages, so that the physician might no longer distinguish between the urgency of messages arriving, posing a patient safety concern. A nurse at site 2 said,

They [physicians] get kind of page-fatigue and it's frustrating. If I call you, I have a real issue and you need to listen to me. But because 20 other nurses that night have paged you [about nonurgent issues] the resident's now so flustered that he doesn't see the importance of my critical situation.

A second effect was that workarounds to deal with one-way communication created hazards and potential errors. Some physicians pointed out the potential patient safety risks associated with the expectation of a quick response time. For instance, a resident at site 2 said that he might turn his attention away from a sick patient to return a page, which ended up being about “a diet order or something not as urgent.”

A direct effect of nurses using numeric pages to contact physicians (when text paging was available) was interruptions in patient care. Because numeric pages did not provide any information other than numbers, physicians were unable to determine the level of urgency or prioritize messages that came from numeric pages. Many physicians said they had to consider every numeric page as urgent and return it as soon as possible, causing unnecessary interruptions in patient care.

As one physician at site 2 told us,

Some of those things don't need a call back… If you are out seeing patients or in a patient's room—if you had a way to communicate back, oh this patient, their potassium is 3.5 and you text back and say okay, I'll replace later. That would save a lot of time, then having to walk out of a patient's room, find a phone, and call them back, have that discussion. And then go back in that patient's room—see that patient, trying to establish rapport with that patient you are seeing, but you can't because you keep getting called out to do things.

DISCUSSION

Communication is a complex process, and technologies designed to facilitate communication are similarly complex, yet we lack sufficient evidence of the ability of these technologies to support effective communication.¹⁹ Our findings contribute to the evidence base, and by using the theory of workarounds to interpret our results, insights into how the environmental context influenced technology use to contribute to workarounds emerged. Our findings describe how workarounds are derived from the organizational context, the types of workarounds that are created in response to environmental constraints, and the effects workarounds can have on workload, potential errors, and interruptions to patient care.

Our study confirms previous research that found a unidirectional flow of information via pagers²⁰ and is noteworthy because it suggests that despite ongoing technological advances in the intervening 15 years, we have not found a way to consistently support two-way communication. When viewed through the lens of the theory of workarounds, our results indicate that one-way communication was a specific antecedent that contributed to different types of workarounds and various effects.

Others have characterized workarounds broadly as either formal or informal.⁷ Our framing using the theory of workarounds may be more useful and contribute something new to our understanding of workarounds for 2 reasons. First, it helps to unravel some of the complexity surrounding workarounds by distinguishing workarounds from their antecedents and effects. Second, possible strategies to reduce workarounds may be most effective if they target antecedents, because by eliminating an antecedent, the workaround may not develop, or fall out of use.

Across all 4 sites, one-way communication practices set up an environment in which workarounds flourished. Our study is unique in emphasizing not only the workaround, but also specific antecedents such as one-way communication practices, and in doing so we bring attention to upstream contributors to workarounds. Consistent with other studies, we found that one-way text paging can be ineffective,⁶ increasing the risk of miscommunication and errors that may lead to suboptimal patient care.²¹ Text paging has benefits over numeric paging, such as the ability to convey urgency, thus leading to decreased disruptions to patient care and workflow and increased satisfaction among nurses and physicians.^18,22–24 However, without a message receipt, text paging fails to close the communication loop²⁵ because the sender does not know if the message was received.²⁶ There is also the potential that important patient care information gets lost in transfer, posing a risk to patient safety.

We found that differential access of text pagers contributed to unidirectional messaging that reinforced occupational hierarchies, and created barriers to efficient and effective information exchange. Occupational hierarchies are pervasive in health care.²⁷ Physicians’ higher status relative to nurses was reflected in our study by differential access to technology and by double standards for communication practices such as expected response times based on role.

All types of workarounds have as a common element the need to accomplish some goal that without the workaround would not be possible.²² Yet simply introducing modern technology into health care without considering unique characteristics of healthcare work or clinician roles may not yield desired results.^4,28 In a study in which both nurses and physicians were provided with smartphones for communication purposes, physicians were concerned that emergency text pages might not always be transmitted, so they were reluctant to use the smartphones as their primary device.²⁹ Nurses, although satisfied with the smartphones overall, reported system issues such as receiving smartphones without fully charged batteries and battery charges not lasting for a 12-hour shift.²⁹

Reducing the technology misfits that can lead to workarounds may require capturing variation in preferences and technology use across multiple sites. Our multisite study makes a significant contribution to the literature by demonstrating common workarounds across all 4 sites, despite differences in hospital type, geographic location, EHR platform, and types of technology in use. The types of technology in use in single site studies may reflect administrative decisions based on resource allocation, rather than on clinician preference or technology fit. For example, there is evidence from a single-site study that nurse-carried mobile phones improve nursing workflow and increase reliability of communication and clinician responses over unit-based phones.³⁰ Similarly, in 2 single-site studies, smartphone use led to significantly fewer interruptions, more time with patients, less time spent waiting for calls or on hold, and increased clinician satisfaction.^29,31 A multisite study meanwhile found that smartphones (used only by physicians as in our study) increased interruptions for residents but proved less time-consuming than returning an alphanumeric page.²⁴

Despite significant innovation and widespread use of communication technology in health care, the use of numeric paging has not decreased in 25 years.⁴ Hospital administrators who supply alphanumeric pagers may not be aware that these pagers cause workarounds, and that the majority of issues being communicated via pagers can cause unnecessary interruptions in patient care.²⁴ In one study, half of surgeries involving residents were interrupted at least twice for overwhelmingly nonurgent issues.³² Other studies suggest that Web-based communication tools may reduce physician interruptions by routing nonurgent messages to an inbox within the EHR instead of as a page.^21,33 Additionally, one study found that routine adoption of secured messaging was significantly less disruptive to workflow for nurses and physicians.³ Findings like these suggest that other less disruptive communication devices would better suit the majority of messages and possibly reduce the effect of this workaround, as high levels of interruption have an effect on cognitive functioning and may lead to clinician errors.³⁴

Considerations for future research include expanding the theory of workarounds to develop an understanding of how and why hospital administrators decide to purchase one technology or another. Hospital purchasing decisions and who makes them (and why) would be powerful additions to the theory because they could produce actionable knowledge to help address workarounds. Purchasing decisions that resolve to reduce antecedents to workarounds (ie, eliminating differential access to communication devices, purchasing devices that support bidirectional flow of information) may have the effect of reducing workarounds themselves and any associated adverse effects. More research into the theory of workarounds’ utility for describing or explaining other ongoing challenges in communication via mobile technology is needed because the theory may not be appropriate in every case. For example, the theory describes other types of workarounds that we did not find such as lying or cheating for personal benefit, or pretending to comply.⁸

This study has several limitations. None of the hospital sites in our study used smartphones for secure messaging, which would have provided more variation in ICT and its effect on communication practices. The 4 hospital sites were located in the Midwest, limiting the generalizability of our findings. Because nurse leaders chose participating units, there is a chance of selection bias. Finally, participants may have behaved differently or expressed different opinions than usual because of the presence of researchers. We took steps to minimize this effect by our prolonged engagement on each unit (2 weeks) and deriving our data from multiple sources.

CONCLUSION

Our view of workarounds has largely been limited to understanding only the actions people take to continue their work. The theory of workarounds expands the notion of workarounds and situates them in context from cause all the way through to the challenges that they create organizationally, rather than individually. This shifts the focus away from the individual’s action to understanding that action as a response to the organizational context, and suggests that workarounds should be addressed at the organizational level. The theory of workarounds helps to understand how and why workarounds develop, thereby offering organizational guidance on whether or not to address the workaround and derive better alignment between strategic goals and outcomes.

FUNDING

This project was supported by grant number R01HS022305 from the Agency for Healthcare Research and Quality. The content does not necessarily represent the official views of the Agency for Healthcare Research and Quality.

SUPPLEMENTARY MATERIAL

Supplementary material is available at Journal of the American Medical Informatics Association online.

AUTHOR CONTRIBUTIONS

MM, MH, and SK made substantial contributions to the conception and design of the work, while ML was closely involved in the acquisition of data. ML and MM drafted the work, and all 4 authors revised it critically for important intellectual content. All authors gave final approval of the version to be published and agree to be accountable for all aspects of the work.

CONFLICT OF INTEREST STATEMENT

None declared.

DATA AVAILABILITY STATEMENT

The data underlying this article will be shared on reasonable request to the corresponding author.