Nurses’ knowledge and deliberations crucial to Barcoded Medication Administration technology in a Dutch hospital: Discovering nurses’ agency inside ruling

Aim

The aim of the qualitative empirical study was to determine how, from a standpoint of nurses, the use of BCMA institutionally and textually mediates nurses’ deliberations in the process of decision-making. Our focused interest on ethnographically describing nurses’ use of barcoded medication systems joins a broader interest in information and computer technologies (ICT) in healthcare. Our research into nurses’ use of a barcoded medication systems contributes a rare critique of nurses’ use of the electronic health record (Campbell and Rankin, 2017), of which the barcoded electronic medication administration record (eMAR) is a part. Considered a ‘closed-loop’ system (Rack et al., 2012), the barcoded system is expected to reduce the risk of ‘human factor’ errors (Henrickson Parker, 2015; Werner et al., 2012). Most of the research into digitalized medication systems (cited above) focuses on evaluating claims of improved safety and efficiency. The ‘Science and Technology Studies (STS)’ literature focusses on the relationship between scientific knowledge, technological systems and society. Our IE and practice theory approach partially connects to two broad streams of STS. First, our research focuses on practices and, in particular, the textual and social organization of practices initiated by technology. The technology organizes and structures nurses’ practices discourse and knowledge. Second, the impact of technology on the values of care. This study connects to Verbeek’s view on the invasiveness and mediating aspects of technology and joining the important critical analysis of Berg and Bowker (1995) and subsequent papers of the same ilk (Berg, 2001; Berg et al., 2003). Our interest in the contemporary advances of BCMA as a healthcare technology is to critically look at the technological premises of the closed-loop system (errors reduced, more efficient use of stock, etc.). Our aim is to discover how nurses’ medication administration knowledge works, within the built-in procedural design of the barcoded steps and stops. The ethnographic research we report here details how nurses interact with the electronic systems. We examine nursing knowledge; both knowledge that is included and knowledge left behind, when a central organizer of nurses’ work, medication administration, is digitalized. We contribute to the evaluation of such technologies by expanding this practice to consider more than quantified metrics of errors and stock. We open up the field of errors and human factors science for discussion using an alternative paradigm about the social organization of knowledge. The human factors science discipline rejects the notion that humans are primarily at fault when they make errors using equipment and other technologies. ‘They are invented by humans and should be easy for humans to use’, explains Joseph Cafazzo, an associate professor in the University of Toronto’s faculty of medicine who conducts human factors research. ‘If the user makes an error, it’s usually the result of a design flaw’ (Eggertson, 2014: 25). We connect what nurses’ know about the circumstantial features of their work, its moment-to-moment pace and its risks, with the knowledge they rely on to invent the workarounds that seem an intractable feature of nurses’ work with digitalized systems of medication administration (Voshall et al.,2013).

Design

The study involved 9 months (2011–2012) of qualitative field research in a Dutch hospital that included direct observation of people at work and an examination of documents. The first author (Boonen) performed participative observations of nurses distributing medication to patients. This standpoint data were used to identify the tensions and contradictions that nurses encountered while they worked. Consequently, the field notes were used as a starting point to investigate the social organization of the problems, by empirically tracking how nurses were coordinated within the logics and textual systems of the barcode technologies.

Sample

In IE, descriptions of linked institutional practices constitute the ‘sample’. The researcher samples observed activities that lead into and extend out from the standpoint informants’ work. The sample is guided by a focus on problems – some ordinary and mundane, others more serious – that can be traced to other people’s work. The sample may be limited by practicalities such as access to informants, access to texts, and the time available to follow the ‘threads’. It is possible for the sample to be small and confined to work that is done in close proximity to the standpoint location (e.g. the sample could be confined to how work on a nursing unit is linked into work in a pharmacy). Or the sample could be a broad set of institutional practices (e.g. how nurses’ work is linked into the work of bioengineers and software developers). The adequacy of the sample is determined by whether the problems are seen to be familiar or plausible and whether the researcher’s analysis contributes new insights into the social organization of similar settings.

This study sampled the institutional practices implicated in nurses’ work with the barcoded medication system. The sample included nurses, clerks, pharmacists, medical doctors and managers. The many computer fields and other texts (policies, protocols, guidelines and so forth) were critical features of the data that were sampled and analysed. This article reports only a small view of the broader findings.

Data collection

Data were collected on an orthopaedic unit where 17 nurses were followed during their use of BCMA. Boonen (himself a nurse) performed 63 hours of observation of activity connected to BCMA. Observations were conducted during a range of differently scheduled shifts (several overlapping variations of days, evenings and nights). During the observations, nurses were asked to think out loud and to reflect on their actions while mediating the BCMA into their work. Most observations and reflections were audio recorded and transcribed verbatim. Called ‘spect-acting’ (Gill, 2011), this is an ethnographic approach differentiated from ‘shadowing’. The goal is not to garner the neutral descriptions of an unobtrusive researcher, but rather to reflexively accommodate researcher bias in settings that are inevitably ‘malleable, shifting and temporary’ (p. 133). Four nurses were interviewed. The audio-recorded interviews focused on their experiences of working with BCMA. The observations produced about 151 pages of field notes that were analytically organized into 249 scenes. Approximately 2000 pages of documents were amassed. Analysis was focused on nurses’ use of the technology and how the fields embedded in the technology coordinated what they did. The relevant documents (safety and error reports, policies, electronic drug libraries and so forth) were used to establish the empirical links between the nurses’ work and institutional practices.

The dominant discourses in nurses’ work with the BCMA

To present and develop these findings, we begin by using data from an interview with a nurse. The interview contains several traces of the various ‘discursive practices’ that we identify and track. The nurse described working with a patient with Parkinson’s disease whose personal medication timing, imperative to his wellbeing, differed from the standard hospital medication rounds. The BCMA was not equipped to deviate from the standard prescheduled timetable. The patient was reliant on the nurses’ memory for correct medication administration. To help with this, one of the nurses introduced an old fashioned method – a paper-based medication administration record (MAR), which was placed on the cupboard close to the patient. However, those nurses who had only ever worked with the BCMA were at a loss with the paper-based MAR: for the less experienced nurses, it seemed to complicate rather than clarify the patient’s care.

This example illustrates how nurses’ patient-related knowledge and ‘old fashioned’ practical knowledge on how to administer medication was necessary to mediate the medication technology. The example also has traces of how the BCMA is supposed to work, as a tool to improve patient safety, with established routines and protocols that are built into the system to ‘force’ adherence in line with the knowledge of computer experts, health safety experts, risk managers and so on. Although this knowledge is generated outside the setting, nevertheless it directs how nurses should proceed. Our findings show that nurses routinely (reasonably and sensibly) break these rules. For nurses, the BCMA, including its accompanying systems and ‘stops’, introduces uncertainty about what is ‘good’ nursing practice and whether their practical knowledge of patients can be brought to the fore or must be subordinated. Our data were rife with the sorts of dilemmas that this excerpt represents.

The problems that nurses encounter when working with the BCMA system are organized within a variety of ‘conversations’ that make up the dominant discourses among nurses when they make each moment-to-moment decision about whether, and how, they might mediate the problems the BCMA produces. These are schematically represented in Figure 1. Our findings point to those institutional activities that are implicated in the issues introduced into nurses’ work processes by the BCMA. The dashed line (Figure 1) around the item BCMA/Technology (T) identifies the BCMA as our focus and represents the ‘traffic’ between the ruling relations of the state, safety practices and the design of the BCMA system. Although there is a top-down hierarchical safety discourse in place, this is not our entry point. The schematic positioning of the nurse inside the authoritative system becomes relevant as soon as nurses begin their work within a conceptual, discursive, technological frame that does not fit with what they know about their medication work. This is where a nurse working with BCMA would be positioned as Smith’s (1987) ‘small hero’. As researchers we stand with that nurse – shining a light into the ruling relations, working to better understand how they coordinate the problems and tensions of the many ‘small heroes’ who are subject to the BCMA technologies.

OPEN IN VIEWER

Figure 1.

Model of nurse discourse.

In Figure 1, the institutional discourse related to medication is dominated by the safety concept (SC) – a belief that administering medication is inherently risky, and that nurses’ professional adherence to protocol is crucial in ensuring safe practice and preventing catastrophic events. These beliefs about safety are built into the BCMA technological design (T) which is buttressed by the organizational discourse (OD) which inserts specific rules and policies that are reciprocally embedded into the BCMA technology and also supported by nurses’ education and training. It is at the juncture of the OD that we noted nurses compromising the ruling guidelines by activating an informal discourse that we termed the ‘nurse discourse’ (ND). This discourse is more closely aligned with ‘actual work processes’ – such as the work processes described in the interview excerpt above – that ensured an adequate nursing response to the specific needs of the patient with Parkinson’s disease. The ‘ND’ is based on immediate practical knowledge. We depict this with the dashed line because it is always what the nurse must mediate with the OD (the protocol-based knowledge) that directs a certain standard way of acting. The space between the two dashed lines (between the ND and the OD) alert us to how the ND (that which is known from ‘being there’) coordinates a workaround (or rule breaking). The blurred line between organizational and nursing discourse represents whether or not a nurse’s activity might be institutionally worked on as rule breaking or not. It represents the discursive practices that determine how nurses make decisions about what to do in their medication work. It provides a bigger picture of the various sources of knowledge that nurses must address when trying to decide how to proceed. Our data show a form of informal and ‘renegade’ knowledge that nurses use to determine how they can safely break the rules so that the patient needs are met. Concurrently, work processes are activated so that the nurse does not get reprimanded.

Building on the schemata of the ND that is activated (Figure 1), we elaborate on how nurses routinely break the rules and how their rule breaking work is exacerbated by assumptions about how medication work should proceed.

Covering the traces of agency and ‘flaws’ in the BCMA

Nurses’ activities to accommodate the essential character of their work are systematically subordinated to the ruling relations of the BCMA technology. The on-going series of necessary activities that are episodic, discontinuous but nonetheless accomplish the many lines of work that are required in each distinctive shift of duty are reconstructed into a seamless production of what the BCMA intends. So much so, the nurses’ knowledgeable contributions and agency remain unseen – and thus not valued (and likely being made dispensable with new models of care), within the increasingly taken-for-granted trust that is placed in the computerized organization of nurses’ work. The BCMA flaws that nurses mediate and the safety work they are producing remain outside of the authorized views about how safe practice is being achieved.

A good example of a nurse covering over her important work of addressing flaws in how the BCMA coordinated her practice arose when a nurse was doing the first medication round of her shift and she did not yet know the details about the patient whose medication she was preparing. Prior to the introduction of the computerized medical record and the BCMA, at the start of the shift, each nurse would print out a list with the names of the patients whose care was assigned to them. They would make notes of the important things that are relevant in the care process of each patient. However, nurses are now expected to rely exclusively on the digital patient files as these are considered the most current and ‘safest’ source of information. Opening the patient’s full electronic file while working with the BCMA system is cumbersome and time-consuming. It requires the nurse to click through multiple computer fields. Consequently, some nurses routinely break the rule and continue to use a printout as a pocket tool. In this case, however, the nurse being observed was following the rule; she did not have a printout to provide the detail about the hip surgery the patient had undergone and the patient could not supply those details. The situation that arose related to a prescription for Indocid that is prescribed for its analgesic and anti-inflammatory effects following a total hip replacement. Routine administration of Indocid is part of the hip replacement protocol. However, Indocid is contraindicated for patients with a history of gastric ulcer. The nurse noted that the analgesic on record was Paracetamol. In view of the patient’s incomplete report, the nurse had questions about the Paracetamol. She needed to confirm exactly what kind of a hip surgery this patient had undergone in order to know how to proceed with the medication round. In order to save time and to get the information she needed, she consulted with a colleague who had broken the rule and had a printout at hand to confirm that the surgery was a hip replacement that fitted the protocol for Indocid. The first nurse then queried why Indocid had not been ordered and the second nurse responded that the patient was allergic to Indocid. She did not get this information from her printout, but from her prior knowledge of the patient. The allergy was not noted on BCMA screen. The nurse’s knowledge about Indocid, her questioning about the Paracetamol, her consultation with her colleague and their collaborative efforts to ensure the right medication got to the right patient and the inaccurate (missing) note of allergy is completely obscured when she simply prepares and administers the Paracetamol.

Another set of prudent nursing deliberations that emerged from the data exposed a different built-in limitation of the BCMA. The BCMA technology cannot accommodate patients who take a medication that is not stocked in the hospital pharmacy. In these situations, patients use their personal supply of medications that they bring with them into the hospital. The doctor orders the medication and the nurses must monitor the patient taking it. However, because the pharmacy cannot enter non-formulary medications, the ‘personal’ medication process generates a red dot (as in error) in the medication overview. The system of coloured dots is a feature of the BCMA programming. They are indicators (both for nurses and for the record) that a patient received the right medication in the right time and dose. It is distracting for nurses to consistently be alerted to the red dot – a false indicator of an error that is, in fact, not real. In these cases, the nurses have learned how to open a dropdown menu and select the option ‘medication distributed during network failure’. This is now a routine (though time-consuming practice) of ‘clicking’ several computer fields. It is a solution that nurses have adapted to mediate the barriers that the computer generates. Nurses’ knowledge (and how they share that knowledge informally among one another) is not visible. Moreover, the system anomalies that generate their creative responses are concealed. They are practices that allow the dominant discourse of ‘safety’, including its logics embedded in the technologies to remain unchallenged.

The observational and interview data revealed other consistent ‘error’ messages that are faulty and distracting. For example, a single-dose medication ‘warning’ consistently appears for medications that are ordered: ‘administer on the day of surgery as premedication’. This warning message appears during every medicine round even when the surgery date has long passed. Unable to ‘click’ this message away, nurses appear immune to it; it was eventually just not noticed/ignored by the series of nurses being observed. The nurses could explain the cause of this warning. They indicated that the doctor orders a pre-operative protocol into the system. Once the protocol is completed, the doctor does not re-enter the system to place a ‘stop’ on the order. The next day the BCMA system automatically switches the protocol into the ‘if necessary’ field in order to keep the record of the medication that has been administered but in the absence of a ‘stop’ order. The medication has not been officially (legally) stopped, so the error message cannot be deleted. All the nurses know how this works. Nurses’ judicious work to ignore the error messages is necessary. However, these sorts of issues that nurses both accept and cover over are not entirely without consequence. One nurse who was explaining the consistent error message said, ‘Because you are asking me about this, I’ve become aware again of how these ridiculous things happen’. What nurses see and know as ‘ridiculous’ do not appear in the formal accounts of the BCMA technologies as safer and more efficient (Cescon and Etchells, 2008; Fowler et al., 2009; Poon et al., 2010; Roark, 2004; Young et al., 2010). Moreover, nurses’ work to reduce the demands of the BCMA on doctors’ time goes unremarked and remains hidden.

‘Interruptions’ or the inherent character of the work: episodic discontinuity

The current trend in the research that focuses on ‘interruptions’ (Stamp and Willis, 2010) during a nurse’s medication process upholds an Archimedean view that nurses’ work can be mechanistically broken into discrete tasks that should be isolated and protected from being interrupted. Our reading of Smith’s descriptions of the ‘episodic discontinuities’ (1987) that characterized her mothering work matched the observations of nurses at work; whereby there were always several lines of action that nurses must attend to and nurses are constantly prioritizing and reprioritizing. Like mothering work, whereby each line of action is linked and connected to the broad needs of each individual child, nurses must function within the complexities of each different patient’s needs in an ever changing, dynamic reality of a nursing unit in each different moment of its unfolding reality. The ruling (abstract) assumption on how nurses proceed on a medication round is that medication administration is their exclusive focus. However, our research shows that this is rarely the case. Despite the presence of a protocol that is embedded in the BCMA design (institutional ruling), which entails highly systematic and stepwise practices, most nurses subvert these into a personal system that reduces risk of error and increases their capacity to respond to patients’ individual needs. These are not only a patient’s needs for medication, but the multiple issues that arise during hospitalization.

The data from this ethnography show that contrary to the linear expectations of a conceptually theorized safe medication round, nurses must respond to patients’ conditions in the specific times that they present. To do otherwise would generate risks for patients that are greater than those created by nurses apparently ‘interrupting’ their focus on the medication administration. In the interviews, the nurses explained that they are constantly ‘on guard’ during their medication rounds, just in case they are called away to immediately assist a colleague or respond to an urgent request for assistance by a patient. The BCMA does not provide for this necessary ebb and flow of the needs of multiple patients. It runs within the conceptual idea that a nurse can isolate medication administration for each individual patient from whatever else is going on. Moreover, the BCMA ‘times out’ when the nurse leaves to handle the things that are coming up as she works. Simple things such as supporting a patient from the bathroom to their bed or a family member asking about her relatives’ situation cannot be easily deferred. There are no built-in strategies to provide nurses with a reminder of where they were in the administration process (which patient’s medications were being prepared, which medications had already been taken from the supply and so forth). Nurses must simply remember so that they can pick up the round at the point where they left off. They expressed concerns about how these daily practical realities generate safety risks, especially when they are hurrying and are understaffed. Several of the nurses had generated personal safety systems using post-it notes and/or marks on the paper medication cups to maintain medication safety in the face of other compelling work demands.

There are also institutionally structured times when nurses administering medications must prioritize other issues. For example, on all occasions of day shift observations, at the outset of the shift, there were only two nurses to attend to 15 patients. On these mornings, one nurse started the routine care processes of greeting, assessing, assisting to the toilet, offering mouth care prior to breakfast and so forth, while the other nurse proceeded to administer the 0800 medications. Whenever needed, for example, with the physical work of helping people into chairs or onto the toilet, the medication nurse would stop to help. Moreover, early in the shift is the common time for doctors to commence their rounds. This too required the medication nurse to be drawn away from the medication work to consult with doctors about patient care issues. This necessary ‘episodic discontinuity’ (Smith, 1987) that characterized a medication round also routinely occurred during the observations at the end of the day due to shift handover and/or patients returning from the operating room. On these occasions, the nurses were observed delaying the start of the medication round. During one observation, a newly admitted patient had a great many questions about their medications. This absorbed the nurse’s time and resulted in the BCMA error message being generated on each patient whose medications, at the end of the round, were flagged by the BCMA system as being administered late. On another occasion, the medication nurse, who is often the most visible person (because a great deal of medication work is conducted in the hallway), was drawn into a conversation with the daughter of an elderly patient. The daughter had questions, because her mother told her that she was prescribed new medication and was confused. The nurse was drawn away from her round to explain to the daughter that her mother was being treated for a bladder infection.

These examples show the incongruence between the portrayals of medication administration as a stepwise series of events (Figure 2. The data show that the work actually unfolds in consecutive scenes such as the circuitous work process depicted in Figure 3. In reality, each nurse is connected to the immediate environment that consists of people with needs that are unpredictable. However, the BCMA is designed (in theory) for a predictable and linear institutional organization (Figure 2). Our data show that nurses’ practice rarely follows the linear line and is most of the time characterized by a circular course of events (Figure 3). The institutional control that is ordered by BCMA in the straight line does not always fit the reality of the care process that is at the core of nursing practices. In the 249 data scenes examined, there were only 7 scenes that followed the straight line embedded in the BCMA technology. Most of the time nurses followed the line of necessary episodic discontinuities in order to meet the expectation of good and safe care that they constantly mediate with medication distribution as shown in Figure 3.

OPEN IN VIEWER

Figure 2.

The linear way of BCMA ordered by (S)tate in a safety discourse of medication reconciliation and verification, translated into (T)echnology of BCMA and processed into (O)organizational (D)iscourse.

OPEN IN VIEWER

Figure 3.

Example of scene and the ruling relations (from Figure 2) that are activated in almost every scene.

Rule breaking

An important finding from this study is in the social organization of rule breaking. During observations, Boonen frequently witnessed nurses being stopped by the system but also finding ways to work around these stops. This was the case when a physician had given the order to administer oral pain medication to a patient who was reporting nausea. The nurse knew that the oral medication may trigger vomiting and be an ineffective intervention for the patient’s pain. Although the nurse scanned the barcode of the pill, she actually administered a suppository. The nurse’s decision was complicated. She knew that, at the next dosage, the patient’s nausea would likely be resolved and oral pain medication would be indicated. She also knew that officially changing oral medication into a suppository would require her precious time and effort: the nurse would need to phone the doctor and acquire a new order which would require lots of scrolling and clicking in order to make proper changes in the system. Her safe and simple solution made more sense in that moment of her practice.

Another case of rule breaking was evident in the case of the Parkinson’s medication of our first exemplar whereby the nurses just ignored BCMA but nonetheless clicked on ‘medication administered’ under the pre-programmed BCMA times. They were engaged in falsifying the official record while developing the creative handwritten medication schedule working in close collaboration with the patient and family.

A more routine example of how nurses breech the built-in protocols of the BCMA (the theoretical construction of medication administration that is expected to establish safe practice) is similar to the prior example of the patient with Parkinson’s disease in-so-far as the nurses needed to mediate the pre-programmed rigid BCMA system of medication times. In orthopaedic surgeries, patients are routinely prescribed antibiotics 8 and 16 hours after first incision. When ‘transcribing’ the order in the BCMA, the nurses regularly change the doctor’s direction and put the required antibiotic into the ‘as necessary’ field of the BCMA. ‘As necessary’ (also known as ‘prn’) medications are administered based on patients’ requests and nurses’ judgements. But the antibiotic is required. It must be administered at exactly the right times postoperatively. The nurses must breech the protocol (changing an explicitly required medication to prn) because the exact time of the first incision is not known at the time the doctor is writing the order. Thus, the timing can only be calculated after the patient arrives on the ward after the operation, when the nurse reviews the operating reports and determines the time of the first incision. If nurses did not do this work of mediating the BCMA, they would have to involve a doctor each time a patient arrived back to the ward. The doctor would have to calculate and order the prescribed times of administration. This would be time-consuming and it does not fit the actual flow of the caring process. The doctor is busy and nurses’ time is valuable. Calling doctors, waiting for them to return the call and then waiting for the necessary order is simply not safe or practical – when there is so much else that needs attention in the provision of competent postoperative nursing care. Thus, the nurses have established a ‘routine’ breech. Establishing the antibiotic as a ‘prn’ relies on important knowledge. Nurses need to know that (1) antibiotics are never prn, a patient either needs it or does not; (2) nurses understand that the prn coding of the antibiotic is a signal to the nurse: ‘look out, an exception’ and (3) the improper use of the term ‘if necessary’ is an acceptable (though illegal) transcription practice for surgical orthopaedics. If the nurses had not developed this process the antibiotics would likely not be properly administered. The signalling of the time would not be noticed because the medication cart, when not in use during the pre-programmed medication timings, is located in a closed medication room – thus, unless nurses are alerted, and make handwritten notes for themselves (a practice that, we elaborate later, is discouraged within the new trust placed in the technologies), a real medication error would happen.

This example, as in the previous one, shows how the logics of BCMA leads nurses into circuitous, time-consuming individualized responses to what is needed in the situation and how the traces of nurses’ knowledgeable responses are necessarily covered over, so that it appears as though the ruling relations of the BCMA have accomplished the safe practice. The serious ‘rule breaking’ activity of changing a doctor’s order without consulting a doctor is part of what is needed from nurses to generate proper care. Within the increased technological scrutiny of nurses’ timed and accurate documentation, this socially organized rule breaking is a tension that is built into nurses’ everyday work.