|Home | About | Journals | Submit | Contact Us | Français|
To identify factors in the nursing work domain that contribute to the problem of inpatient falls, aside from patient risk, using cognitive work analysis.
A mix of qualitative and quantitative methods were used to identify work constraints imposed on nurses, which may underlie patient falls.
Data collection was done on a neurology unit staffed by 27 registered nurses and utilized field observations, focus groups, time–motion studies and written surveys (AHRQ Hospital Survey on Patient Culture, NASA-TLX, and custom Nursing Knowledge of Fall Prevention Subscale).
Four major constraints were identified that inhibit nurses' ability to prevent patient falls. All constraints relate to work processes and the physical work environment, opposed to safety culture or nursing knowledge, as currently emphasized. The constraints were: cognitive ‘head data’, temporal workload, inconsistencies in written and verbal transfer of patient data, and limitations in the physical environment. To deal with these constraints, the nurses tend to employ four workarounds: written and mental chunking schemas, bed alarms, informal querying of the previous care nurse, and informal video and audio surveillance. These workarounds reflect systemic design flaws and may only be minimally effective in decreasing risk to patients.
Cognitive engineering techniques helped identify seemingly hidden constraints in the work domain that impact the problem of patient falls. System redesign strategies aimed at improving work processes and environmental limitations hold promise for decreasing the incidence of falls in inpatient nursing units.
The greatest number of non-fatal injuries in the acute care setting are caused by patient falls.1 2 In addition to causing injuries, patient falls can lead to increased healthcare expenditures, and are a major motive behind lawsuits.3 The direct financial burden to hospitals increased in October 2008 when the US Center of Medicare and Medicaid Services began to disallow reimbursement charges associated with falls that incurred injuries during a hospitalization.4 Others speculate that private insurers will institute similar ‘no pay for bad performance’ policies in the future.5
Patient falls are adverse events that are largely preventable.6–8 The magnitude of adverse events in US hospitals was described in the Institute of Medicine's (IOM) report of healthcare errors, which estimated that 48 000 to 98 000 patient deaths occur yearly due to preventable medical errors.9 Since the release of this landmark report, many experts have advocated the application of systems engineering tools to transform a complex and often error-laden health system into a safer healthcare system.10–12 Although systems engineering approaches have addressed other adverse events in healthcare,12–14 research is needed that explores the problem of patient falls.
The true impact of patient falls in the inpatient setting is unknown as current reporting mechanisms significantly underestimate the incidence of patient falls.15 However, research over the last decade underscores the significance of the problem. Reports estimate between 4 and 12 falls per 1000 patient days in US hospitals.6 15–17 It is estimated that 15–30% of patient falls cause fractures requiring a cast, traction or surgery, but these estimates do not include other serious injuries and, in some cases, death related to falls.1 2 16–20
Most patient falls are predictable and simple patient risk assessment tools can predict over 70%.6 8 Identified risk factors include poor health status, gait instability, balance problems, age, agitation, confusion, urinary frequency and incontinence, patient history of falls, and certain sedative and hypnotic drugs.8 21 22 While many of these risk factors are well known and documented by nursing staff, the assessment tools and prevention strategies have made little headway in reducing the incidence of falls.6 7 23
In addition to patient-related factors, three elements of the system—nurse staffing, nurse knowledge, and safety culture—are emphasized in the adverse event literature. Studies of nurse staffing (nurse-to-patient ratios) and skill mix (higher percentage of Bachelor of Science prepared and registered nurses than nursing assistants and licensed practical nurses),16 17 24 and the expansion of nursing knowledge in risk assessment for falls25–27 have found mixed results on fall rates. Although the relationship between fall prevention and safety culture has not been rigorously studied, safety culture is often cited in the safety literature as being an important component of error prevention.9 28 29 A culture of safety acknowledges the error-prone nature of hospital work, establishes a blame-free environment, and dedicates resources to safety promotion.30–32 In recent studies, staff perception of a blame-free, safety culture and the emphasis on safety promotion has been shown to be associated with improved falls reporting.22 29 33 Increases in fall reporting offer opportunities for system improvements that may prevent falls which would not exist if the falls went unreported. Despite considerable research, patients continue to fall at unacceptable rates. Therefore, research is needed to uncover other factors that could help prevent falls within the acute care environment, but which are poorly understood at present.
The specific aims of this research were to identify constraints in work processes and the work environment (physical, organizational systems and culture, individual, and technical) imposed on acute care nurses that may increase the likelihood of patient falls, and to identify workarounds employed by nurses to deal with these constraints.
Cognitive work analysis (CWA) is a formal approach in systems and cognitive engineering for examining the functional structure of the work domain, including tasks performed, processes followed, environmental constraints, and information followed available.11 34 Such analysis helps system designers, managers, re-designers, and accident investigators understand how workers interact with their environment and how modifications might create more productive work or reduce errors. In this study we used CWA as an umbrella approach for data collection and analysis. Data were collected both by observation of workers and time motion studies to identify constraints placed on the workers, and by surveys and focus groups to attain input from workers to identify workarounds.
The study took place at a 572-bed academic medical center on the east coast. The center is recognized by the American Nurses Credentialing Center as a magnet organization for excellence in nursing service. The specific unit studied was a 26-bed, general neurological unit that treated stroke and epileptic patients, with an average daily census of 23 patients. The unit nursing staff consisted of 27 full-time equivalent (FTE) registered nurses (RNs), 15 FTE nursing assistants (NA), and 4 FTE nursing unit clerks. The nurse to patient ratio was 1:4 for the stroke population and 1:5 for other patients. Figure 1 shows the unit layout.
Table 1 presents focus group and survey demographic data. Data were collected in categorical form to protect the identity of each participant. The majority of respondents (74%) in the Hospital Survey on Patient Culture reported working 40–59 h per week. Although the survey did not include questions on shift length, the nurse manager reported that the majority of nurses work a combination of 12 and 8 h shifts (full-time nurses are scheduled for two 12 h shifts and two 8 h shifts each week).
The study was approved by the local Institutional Review Board. Participants had the option of participating in all of the data collection methods detailed above or could choose to only to participate in some of the methods. Participants were informed via written consent that the study sought to identify and analyze factors related to the risk of patient falls. There were no direct benefits to individual subjects. The primary risk of participation was the potential loss of confidentiality; however, this risk was considered minimal and unlikely. Participants were informed that all data collected would be published only in aggregate form. Completed surveys were returned anonymously to a box near the nursing station.
A different method of analysis was used for each data source. Focus group discussion notes were summarized during the session by investigators. Included in the analysis were major points confirmed by participant consensus. Single or minority opinion points and points with minor relation to patient falls were not included. The Hospital Survey on Patient Culture was analyzed by computing composite frequencies of the positive responses for items within the 14 safety culture dimensions using Likert response categories (‘neutral’ responses were omitted). The number of annual events reported was averaged by individual respondent. NASA-TLX survey scores were derived from ‘X’ marks on a visual analog scale (0–100) and individual scores were averaged. Likert items on the Nursing Knowledge of Falls Prevention Subscale were also averaged per item (1=strongly disagree, 2=disagree, 3=agree, 4=strongly agree). Field observations findings were discussed between the two observers, following their separate observations. A comparison of notes was used to reach a consensus regarding relevant data to include. Time–motion data were added by category, across the four data collection periods.
All data analysis was conducted iteratively with the data collection process. Preliminary conclusions from the data sources were synthesized into the abstraction–decomposition (A–D) hierarchy (figure 2), to help uncover relationships between disparate components and sub-components of the work domain. The A–D hierarchy is a cognitive engineering model that describes the work domain in a single representation and serves to focus on identifying constraints and workarounds.
Overall, 33 of the 46 (71.7%) members of the nursing unit personnel, including registered nurses (N=27), nurse assistants (N=15), and unit clerks (N=4) participated in at least one phase of the study. There were 25 h of direct work observation. RNs and NAs present were observed in two field observations (9 total hours, 4.5 h per shift); four RNs were more intimately shadowed in the time–motion studies (4 h each). Nineteen of 27 RNs completed the surveys (70.4% response rate). Sixteen of the 46 RNs, NAs, and unit clerks participated in the focus groups. Table 2 summarizes focus group data.
Data from field observation were used to understand how the RNs operationalized fall prevention in their daily work flow. The typical procedure an RN followed to assess and communicate the fall risk for each patient is detailed below.
The time–motion study found that 55% of time was spent on indirect patient care (16% of overall time on clerical duties, 15% communication with unit staff, 9% shift report, 9% using MIS, 6% equipment gathering), with 45% on direct patient care (33% time in patient room, 12% preparing and administering medication).
Table 3 shows results of the survey subscale. In addition to items reported in table 3, 100% of respondents also reported that they were aware of fall prevention protocols at the medical center. All but 2 of the 19 participants reported having had a patient of theirs fall during their shift.
Figure 3 shows results from the NASA-TLX workload instrument. The nurses reported their perceptions of subtypes of work demands (mental demand, physical demand, temporal demand, perception of own performance, effort, and frustration level) related to fall prevention, using a 1–100 visual analog scale.37 The average of the nurses' responses is shown on the y-axis. While ‘high’ and ‘low’ appear on the y-axis for brevity, the anchors used in the survey were: frustration (very low, very high), temporal demand (very leisurely, very frantic), effort (very low, very high), physical activity (very easy, very strenuous), mental demand (very simple, very complex), and view of success (very unsatisfied, very satisfied). As shown, the nurses perceived high levels of frustration, effort, and temporal demand, but little mental demand.
Table 4 reports the results, which were compared with 518 hospitals that voluntarily shared their results with the AHRQ in 2008. The unit met or surpassed benchmarks from a national sample in seven of the 12 comparable dimensions. In one dimension, non-punitive response to error, this unit exceeded the maximum reported in the AHRQ database. The nursing unit had four additional scores at or above the 75th percentile of all reporting hospitals (teamwork within unit, teamwork across units, manager actions, and organizational learning). Manager actions included the staff's perception that their nurse manager seriously considered staff suggestions for safety improvement, did not pressure nurses to work faster or take short cuts, and did not overlook safety problems. Organizational learning included the staff's perception that they were actively working toward improving safety, mistakes led to positive change, and improvements to safety were evaluated. Within-unit teamwork included the perception that the nurses work together when the unit is busy. In contrast, the prominent weaknesses were perceptions of staffing adequacy (<10th percentile) and overall perception of safety (<10th percentile). The low perception of staffing adequacy may relate to greater demand in temporal and effort areas found in the results of the NASA-TLX assessment.
Currently, promoting a safety culture, nurse staffing, and expanding nursing knowledge are the primary system factors emphasized for fall prevention in the nursing literature. Somewhat surprisingly, our findings suggest that these extrinsic system factors were not major causes for concern. The nurses perceived a favorable safety culture in the majority dimensions of the Hospital Survey on Patient Safety Culture and their nurse-to-patient staffing ratios were within the range reported in other studies.46 47 In contrast, our findings indicated that work processes and the physical environment presented the major limitations for fall prevention on this nursing unit. We also found from the NASA-TLX workload assessment, that the nurses perceived a high level of frustration related to fall prevention. Specifically, they perceived that they were exerting a great deal of time, attention, and effort to complete fall prevention tasks that required little mental demand.
Using the CWA approach that integrated findings from all of the data sources, we identified four constraints that appeared to be inhibiting the nurses' ability to prevent falls as well as nurse-developed workarounds for each constraint (table 2).
The nurses tracked and executed several tasks in parallel, some with numerous sub-tasks, by mentally noting their checklist. This notion, which we call ‘head-data’ here, has been described elsewhere.48 The paths toward completion were generally straightforward. However, the challenge was in remembering the list of tasks, the number of sub-tasks remaining, the temporal timeframe(s) remaining, and dealing with new entries into the list. The nurses viewed the activities related to fall prevention as imposing a high temporal demand, but low mental demand.
The nurses' workaround to this constraint was to develop written and mental chunking strategies. For example, several experienced RNs had developed schemas to permit them to retain relatively more information and to bypass limitations on short-term, working memory.49–51 Strategies used were visual reminders (eg, lists written on pocket cards) to outline planned tasks and patient assessment to be completed52 and means of chunking (eg, in their mind enumerating the overall tasks, ordering the list, and then trying to complete one block of sub-tasks before moving to another task).53 54 Given a lack of formal training in these strategies, some nurses appear to have greater skill in developing these schemas than others.
Throughout this study the nurses reported high levels of temporal demand, effort, and frustration in their workload. One component of workload is staffing. While the evidence base on nurse staffing does not call for standard nurse to patient ratios,55–57 the staffing ratios on the studies unit falls within the range mandated in the State of California (1 RN to 5 patients)58 and the range studied by healthcare analysts and economists for non-ICU staffing (1 RN to 4–8 patients) in other studies.46 47 Another component of workload in this analysis is how and where the nurse spends her time. The majority of nurse time on this unit was consumed providing indirect care or medications (67%), which positions nurses in the proximity of the nursing station, away from patient rooms. The task of monitoring for patient falls, however, is one in which the nurse must be in intimate proximity of the patient to predict, detect, and intervene an impending fall. Under a heavy workload with competing parallel demands and a large proportion of time away from the patient, the task of monitoring falls becomes increasingly difficult for the RN. An automation of the monitoring task may alleviate the perceived workload demands.
The nurses' workaround to the workload constraint was to use bed alarms, but these were employed inconsistently and their success varied. Bed alarms are an automated means to address constant monitoring, by notification of patient movement which offers an opportunity to intervene. The bed alarms used on this unit included a monitor with user interface, a speaker built into the bed rail, and a mat placed on the center of the bed, edge of the bed, or the floor. While ideal in theory, alarms used on this unit were either too sensitive or too insensitive. Alarms frequently provided feedback that was so sensitive that normal movement in bed triggered a response. Alarms also had problematic user interfaces whereby the operational modes were restrictive and involved sequences of typing that were difficult for a clinician to comprehend. Activated alarms were difficult to hear in the presence of competing tones (ie, with similar frequencies) and background noise. Similar findings associated with excess auditory stimulation in hospitals have been a problem documented by others.59 60 Given the difficulty that nurses faced in utilizing the alarms, it was not surprising that non-nurses (eg, therapists, family members)—on interacting with a patient—also struggled to reset them.
As a result of inadequate written data (lack of capability in the MIS and paper chart) and verbal communication (between RNs and NAs), the fall risk status of a patient was often communicated only via RN verbal report at shift change, and not consistently done. This lack of structure in the verbal report introduces the possibility for additional human error, in this case inconsistent reporting of fall risk status. We found there to be inconsistency between the MIS data and the nursing care plan that had been compiled from the paper chart, shift change information, and patient assessment. Fall risk status and other pertinent patient information in the MIS were rarely updated. The MIS does not have a field or prompt to complete falls data, and therefore was not conducive to falls tracking. Nurses can enter a general comment, but this information is not readily displayed to the subsequent nurses. The paper chart was the source of more reliable data, but it was not always read at the beginning of the shift, as the nurse would first receive shift report and begin to meet with patients. Therefore, typically only those nurses who had worked the previous day, who were assigned the same patient, could rapidly and readily appreciate the patient's level of risk for falls and be able to select appropriate fall prevention interventions. In most incidences, however, nurses only had consistent patient assignments for two consecutive days.
A second issue surrounds the lack of formal, verbal communication between RNs and NAs and between RNs and the patient/family. While RNs implement fall prevention protocols, the NAs conduct the routine monitoring that ensures their effectiveness. There was an overwhelming consensus in the focus group that the RN–NA partnership was vital to preventing falls, but a lack of communication hindered its effectiveness. The observation data (shadowing and time motion) also showed that in contrast to RNs, the NAs were not involved in the formal exchange of information between RNs at hand-off or with a prior NA. As they were not informed of patient fall risk or RN fall prevention interventions, they relied instead on visual cues (eg, bracelets, slippers, stickers) to determine fall risk.
The nurses' workaround to decipher patient fall risk was to rely on informal querying of the previous care nurse about fall status. This reliance was evident in the TMS, where 24% of time was spent in communication either between nurses at shift change (9%) or with unit staff (15%). Conversely, only 9% of time was spent interacting with the MIS. It is possible than an inadequate understanding of or lack of confidence in the current MIS may have led to its underutilization. It is also important to note that informal querying contributes to human error; formalized documentation systems would better inform fall risk while decreasing the time spent obtaining a patient's status.
The physical layout of the unit removed nurses from physical proximity to their patients and did not permit direct patient visibility. It is quite difficult for a nurse to prevent a fall when positioned far from the patient–while at the nursing station, for example, to mix medication, access equipment, or update documentation. In this study, it was found that nurses spent only 33% of time in the patient rooms, divided among four to five patients. There were no means of formal surveillance to deal with this constraint. In particular, from the location of the nursing station (figure 1) there was a direct viewing angle into only 3 of 17 patient rooms.
The nurses' workaround was to add informal video and audio surveillance to improve patient visibility. Many nurses sought to monitor patients by using closed-circuit video (or relying on video assistants to watch the monitors) to patient rooms. However, most cases where an impending fall might be detected would be caught essentially by chance when a nurse happened to be in position near the camera. Also, video cameras were used only in those rooms (five in total, apart from the three with direct visual angles) where epileptic patients were assigned. Other nurses relied on reports from the unit clerk, the person to whom patients in all rooms could call over the audio intercom. In one observed case, a nurse was able to use this means to talk a patient back into bed, but this likely is a rare occurrence. In the focus group, several nurses reported that more surveillance was necessary via sitters and/or cameras.
While increased audio and video surveillance of patients may alert nurses to an impending fall, the knowledge often occurs too late to prevent it. Patients at risk for falling benefit from having people nearby who can quickly respond to an impending fall. Some redesign strategies to overcome this constraint include relocating nurses' indirect care tasks to be in close physical proximity to the patients' bedside and to allow more space for patient family members to stay overnight.61
The study methodology yields a few limitations. First, as with all single-site research, the findings are unlikely to generalize completely to all inpatient nursing units. The similarity of the other nursing units and their work processes to the one studied must be considered before applying the results of this research. Second, a convenience sample of nursing unit volunteers was used and as such the potential for bias exists. Third, there is a potential for the ‘halo effect’ (to perform with greater vigilance when being observed) in the observed nurses who were aware that the study focused on the prevention of patient falls. Fourth, there are some inconsistencies in the data from different data sources. For example, respondents rated themselves highly on the Nursing Knowledge of Fall Prevention Subscale while in focus groups, and they discussed inadequate understanding of how to use the MIS for fall prevention. Perhaps part of the reason nurses reported this constraint is because there are no prompts or fields related to fall prevention in the MIS. Additional reasons for differences between data sources are likely due to the differences between open ended (focus group) and closed response (survey) forms of questioning. Fifth, there is a possible ‘Lake Wobegon effect’ in the survey data, such that nurses may have inflated their individual above average knowledge base for fall prevention techniques, while also reporting that others had a low overall concern for safety. The surveys used in this study did not directly assess nursing knowledge, and self-report data of this nature can have systematic self-report bias. Finally, the void of psychometrically validated tools for extrinsic systems factors led us to develop the Fall Prevention Survey. While its face validity was assessed by cognitive engineers and advanced practice nurses, there is a need for full psychometric testing.
This study indicates that constraints in the acute care environment may lead to increased patient falls. Traditionally, extrinsic work system factors, such as nurse staffing, knowledge of fall prevention, and safety culture, have been emphasized along with intrinsic patient risk factors. Despite the emphasis in these areas, patients continue to fall. This study sheds greater light on how nursing work processes and the physical work environment contribute to the risk of patient falls. Some of the factors identified can be rectified rather simply, as in the case of standardizing hand-offs between nurses and nurse assistants. Other factors—such as limitations in the physical environment, lack of capability in the MIS, ineffective bed alarms, and unique aspects of nursing workload—are more systemic and require complex solutions. Solutions must consider the low mental but high temporal demands of the work domain, that nurses track and execute numerous tasks in parallel, and are often physically separated from individual patients for large chunks of time. Moving forward, when nursing work processes and the physical work environment are considered together with traditional factors, broader solutions can be developed with greater potential for preventing patient falls.
Funding: The work was supported in part by a grant from the National Library of Medicine (Grant Number T15LM009462). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Library of Medicine or the National Institutes of Health.
Competing interests: None.
Ethics approval: The study was approved by the local Institutional Review Board.
Provenance and peer review: Not commissioned; externally peer reviewed.