3. Materials and Methods
The data for this cross-sectional correlational study were extracted from a web-based survey on nurses’ perceptions of their working environment, quality of nursing care, patient outcomes, and nurse outcomes among a province-wide sample of Canadian nurses. Institutional Review Board ethics was obtained (approval number: H14-00789). A proportionate stratified random sample of RNs and licensed practical nurses (LPNs) was drawn from the provincial nurses’ union database based on geographic region (i.e., health authority) and employment status (full-time, part-time, and casual). In Canada, RN and LPN classifications are distinguished by differences in formal education and scopes of practice. Registered nurses receive more theoretical education and are prepared to care for complex, unstable patients, while LPNs are prepared to care for stable, predictable patients. The survey was content validated by union member focus groups. Unique, password-protected FluidSurvey email invitations were sent out by the nurses’ union on behalf of the research team.
The study sample consisted of all direct care nurses working in medical, surgical or medical-surgical areas in the four largest health authorities. All direct care nurses in acute care settings in British Columbia (BC) are unionized; therefore, we had a complete sample frame. Our final sample (
N = 472) consisted of 354 RNs and 118 LPNs with an estimated response rate of 22.4%. Precise response rates were difficult to determine due to the nature of the union’s database (e.g., active versus inactive members). A similar issue is noted by Ball and colleagues [
5].
3.1. Measures
Adverse Patient Outcomes were measured using RN4CAST questions that asked nurses to estimate the frequency of adverse events (i.e., medication errors, patient falls, and urinary tract infections) “involving you or your patients” on a scale ranging from 0 (
never) to 6 (
everyday) during the last year [
6]. For this study, we recoded data as
occurred less than weekly (0) versus
occurred weekly or more often (1).
Emotional Exhaustion among nurses was measured with the 9-item subscale of the Maslach Burnout Inventory–Human Service Scale (MBI-HSS) [
28]. The emotional exhaustion subscale asks participants to rate their work-related feelings of psychological depletion on a scale of 0 (
never) to 6 (
daily). For this study, the total scores (ranging from 0–54) were dichotomized with scores of 27 and higher indicating high emotional exhaustion or burnout per developer instructions [
28].
Nurses’ Job Satisfaction was measured as the sum of three variables that asked about satisfaction with current job, intent to leave current job during the next year (reverse coded), and recommending the hospital to colleagues as a good place to work. Each item was measured on a 4-point scale. Total scores ranged from 3–12 with higher scores indicating greater job satisfaction. These items were derived from the validated Canadian National Survey on the Work and Health of Nurses [
29].
RN Staffing Levels were measured by computing a patient-to-RN ratio based on two questions that asked nurses to identify the total number of patients and total number of direct care nursing staff on the unit during their last shift. Patient-to-RN ratio was used rather than the patient-to-nurse (RN or LPN) ratio for consistency purposes, as many units did not utilize LPNs. This staffing level method is described in Sermeus et al. [
6].
Patient Acuity and Patient Dependency were measured with one item each based on the American Association of Critical Care Nurses’ Synergy Model™ [
30]. Patient acuity was defined as the instability, complexity, and unpredictability of the patient: participants were asked to rate the average acuity of their patients during the prior month from 1 (
not at all acute) to 4 (
very acute). For this study, we dichotomized acuity levels as
not at all or somewhat acute (0) versus
moderately or very acute (1). Patient dependency was defined as a patient’s ability to do their own activities of daily living, rated from 1 (
very independent) to 4 (
very dependent). These scores were dichotomized as
very or somewhat independent (0) to
very or somewhat dependent (1).
Perceptions of Nurse Workload were measured as the mean score of three items that asked about the frequency of arriving early/staying late, working through breaks to complete work, and perceptions of “too much work” during the past year, measured on a scale of 0 (
never) to 6 (
every day). The mean scores were dichotomized as
never to a few times a week (0) versus
occurring every day (1). These items were taken from the Canadian National Survey on the Work and Health of Nurses [
29].
Nursing Tasks Left Undone was measured by asking nurses to identify, from a list of 14 activities, all the activities that were necessary but left undone during their most recent shift due to lack of time; for a possible range of scores from 0 to 13. Thirteen nursing tasks were identified by Ball et al., including administering medications on time, preparing patients and families for discharge, and adequate patient surveillance [
5]. We added an “other” option to our survey tool.
Compromised Professional Nursing Standards was measured with a single item that asked nurses the frequency of compromised professional nursing standards over the past year due to workload, measured on a scale of 0 (
never) to 6 (
everyday). Scores were dichotomized as
never to a few times a week (0) versus
occurring everyday (1). This item was added to reflect nurses’ “meaning of work” [
31]. Our researcher-developed question was content-validated with nursing focus groups.
Interruptions to workflow were measured as the mean score of three items that asked about the frequency of interruptions over the past month during patient treatments, during documentation, and when receiving patients at shift change, measured on a scale of 0 (never) to 6 (everyday). The mean scores were dichotomized as never to a few times a week (0) versus occurring every day or almost every day (1). These items were based on a focused literature review and content validation with nurse focus groups.
Factor Structure of each of the four measures that involved a mean score (perceptions of workload, interruptions to workflow, emotional exhaustion, and job satisfaction) were examined using exploratory factor analyses with principal components analysis; the results indicated a unidimensional factor structure and satisfactory internal consistency for all measures. Cronbach’s alphas ranged from .67 for job satisfaction (with only 3 items) to .93 for emotional exhaustion (with 9 items). The percentage of variance explained by the single factor ranged from 62% for job satisfaction to 75% for interruptions to workflow.
3.2. Data Analysis
Data were analyzed using hierarchical logistic regression and hierarchical ordinary least squares regression according to the nature of the outcome variable, using the Statistical Package for Social Sciences for Windows 23.0 (SPSS Inc., Chicago, IL, USA). Mediation effects were tested using the Sobel Test [
32], with adjustments made to the coefficients [
33] for the inclusion of dichotomous mediator and outcome variables.
5. Discussion
This study drew on cross-sectional survey data from 472 acute care nurses from one Canadian province. We considered seven indicators of workload: RN staffing levels, patient acuity and patient dependency, nurses’ perceptions of heavy workload, nursing tasks left undone, compromised professional nursing standards, and interruptions to workflow. Similar to other research, patient acuity was found to be strongly associated with each of the three adverse patient outcomes [
35] and RN staffing levels showed a weaker association [
26,
36]. Patient dependency was not found to be associated with patient or nurse outcome measures. This may be because patient dependency in this study reflected activities of daily living only. In reality, patient dependency may reflect expanded aspects of patient functionality. In addition, within many acute care contexts in BC, patient activities of daily living are managed by non-nurses. Patient acuity refers to characteristics such as complexity and unpredictability that require nurse surveillance and intervention [
30].
After accounting for unit-level workload measures, patient acuity and RN staffing levels, nurse perceptions of frequent, heavy workloads and interruptions to work flow showed strong associations with two patient outcomes, falls and UTIs, and a more modest association with the frequency of medication errors. This study’s heavy workload measure includes items associated with nurse perceptions of time pressure, or not enough time to get work done (e.g., arriving early/leaving late, missing breaks, too much work to do). In one simulated study of nurses’ decision-making performance, time pressure negatively influenced nurses’ capacity to detect the need for intervention, resulting in failure to rescue [
37]. Of note is that under conditions without time pressure, nurses with clinical expertise performed better than novice nurses; the positive effects of clinical expertise, however, were negated when time pressure was introduced to clinical simulations [
37]. The European Nurses’ Early Exit study surveyed over 61,000 nurses [
38]. The survey included intent to leave questions, actual turnover and work-related and personal reasons for leaving. The main work-related reason to leave was “time pressure”, chosen as the primary work factor for 70% of the sample population. Our findings suggest that nurses are aware of harmful outcomes associated with time pressure; they may compensate for these job-level heavy workload demands by coming in early, staying late and working through breaks.
At the task-level, interruptions divert nurses from their planned activities [
39] resulting in decreased performance [
40] and increased patient adverse events, such as medication errors [
41]. Whether at the job-level (i.e., heavy workload demands) or at the task-level (i.e., interruptions), deleterious consequences from these workload factors can be averted through administrative actions such as implementation of nurse resource teams to cover shift changes and break times [
42]; and work redesign initiatives that designate dedicated time for essential tasks, such as medication preparation [
43].
Tasks left undone, either partially or fully, mediated the relationships between two workload factors (i.e., perceptions of heavy workloads, interruptions) and patient outcomes. Ball et al. found that care left undone was strongly associated with nurse perceptions of quality, safe care delivery, suggesting that care left undone is a leading, job-level indicator for unsafe staffing [
5]. Although unit-level measures, such as staffing adequacy, add to our appreciation of workload demands, job-level measures, such as leaving tasks undone, may provide administrators with a more accurate depiction of how nurses gauge effective workload management.
With respect to nurse outcomes, patient acuity was associated with higher emotional exhaustion, but it did not influence job satisfaction. A major source of emotional exhaustion is heavy workload demands that are often outside the control of nurses; nurses have “too little time and too few resources to accomplish the job” [
44] (p. 260). For job satisfaction, however, a systematic review of hospital nurse job satisfaction found that nurses derived satisfaction from interesting and rewarding work [
45]. Care of high acuity patients, therefore, may satisfy nurses by optimizing their professional competencies. In their human factors study of nurses’ workloads, Holden et al. found that at the task-level, there were internal and external types of workload demands [
7]. External demands, such as interruptions and divided attention, were associated with nurse reports of increased patient safety concerns. Internal demands, such as mental concentration and problem-solving, were not associated with nurses’ concerns for patient safety outcomes. As stated by Holden et al. “Perhaps in nursing, some amount of this [mental effort] makes work more satisfying, buffers against burnout and improves patient outcomes through superior performance” [
7] (p. 21). Administrators, therefore, need to differentiate between external and internal workload demands; their focus should be on reductions of external factors, such as interruptions, that have deleterious effects on nurses.
After accounting for RN staffing levels and patient acuity, nurses’ perceptions of frequent heavy workloads and interruptions were independent predictors of emotional exhaustion. For job satisfaction, perception of frequent heavy workloads was a significant predictor. Baethge and Rigotti found that work interruptions had negative effects on nurses’ satisfaction with their performance and their irritation with work [
39]. Work irritation is a concept associated with emotional and cognitive strain [
46]. Cross-sectional and longitudinal studies have shown that irritation mediates the relationship between workplace stressors and eventual decreases in well-being [
47,
48]. Baethge and Rigotti further found that time pressure and mental demands fully or partially mediated the relationships between work interruptions and satisfaction with performance [
39].
There is evidence, therefore, that job-level heavy workload demands and task-level interruptions involve externally imposed time pressures and mental exertion that negatively influence patient and nurse outcomes. As stated by Baethge and Rigotti, research on workplace demands and stressors is adding to “promising directions for interventions in the field of occupational health promotion” [
39] (p. 59). Administrators need to work in collaboration with occupational health and safety officers to utilize best practices that reduce damaging workload factors. Proactive strategies for work interruptions were mentioned above. Health circles are an intervention to address the mental and emotional strain of workloads [
49]. Health circles are workplace discussion groups where employees are encouraged to discuss and identify opportunities to decrease workload demands—giving control to employees who are the experts in their workplace.
A significant finding from our study was that the strongest predictor of both nurse outcomes (i.e., emotional exhaustion and job satisfaction) was compromised professional nursing standards due to workload. Moreover, compromised standards were also found to be a significant mediator of both heavy perceived workload and interruptions for both nurse outcomes. Mediation testing is used to test hypothesized casual chains where predictor variables influence intervening variables (i.e., the mediator) that, in turn, influence outcome variables. If the predictor variable influences the outcome variable only through the mediator variable (i.e., indirectly), this is considered full mediation. On the other hand, if the predictor variable influences the outcome variable directly and indirectly through the mediator variable there is partial mediation. In this instance, our findings suggest that heavy workloads and interruptions influence nurse outcomes both directly and indirectly through the mediator variables (i.e., nursing tasks left undone, and compromised standards).
Nursing is a caring profession built upon nurse-patient relationships. When nursing is reduced to “task and time” mechanistic approaches to care delivery, nurses suffer from emotional and moral distress [
50,
51]. Compromised nursing standards are a source of emotional distress and moral distress, with deeper ethical roots. “…moral distress occurs when the internal environment of nurses—their values and perceived obligations—are incompatible with the needs and prevailing views of the external work environment” [
52] (p. 1). Outcomes from emotional and moral distress include emotional exhaustion/burnout, job dissatisfaction and eventual exit from the profession [
52,
53,
54,
55]. Epstein and Delgado [
52] recommended that administrators engage nurses in discussions around values conflicts, while Pendry [
56] advocated for informal team discussions and formal ethics committees.
Van Bogaert et al. studied the relationships between the nurse practice environment and job outcomes and nurse-assessed quality of care [
31]. Job outcomes included job satisfaction, intent to stay in the hospital, and intent to stay in nursing. Mediators included nurse perceptions of workload; decision latitude (i.e., ability to make decisions and use personal/professional skills); social capital (i.e., shared values and perceived team/organizational trust); and three dimensions of burnout (i.e., emotional exhaustion, depersonalization and personal accomplishment). There were direct and indirect effects for workload on job outcomes. Workload, decision latitude and social capital mediated the relationship between practice environment and outcomes variables via the burnout variables. A key finding was that unit-level nursing management had a strong, direct impact on the study’s outcomes. The researchers concluded that unit-level leaders, in particular, can influence job outcomes and nurse perceptions of quality of care by monitoring and responding to nurses’ workload demands, involving nurses in decisions related to patient care delivery, and promoting shared professional values among interdisciplinary team members.
In our study, we examined two potential mediators with respect to patient outcomes, and four potential mediators with respect to nurse outcomes. Tasks left undone was found to be a significant mediator of perceived heavy workload and interruptions for all three patient outcomes, suggesting that in addition to their direct effects, heavy workload and interruptions influence patient outcomes indirectly through their influence on nurses’ ability to complete essential tasks. Similarly, we found indirect effects from perceived heavy workloads and interruptions on both nurse outcomes through tasks left undone and compromised professional nursing standards. These two mediators, therefore, should serve as critical indicators for administrators to monitor and track: these mediators may be the “litmus test” for nurses’ capacity to effectively deliver care within their work environments. Nursing is a unique profession where essential tasks left undone and compromised professional standards signify the potential for adverse patient and nurse outcomes.
Study Limitations
A major strength of this study was that its sample consisted of both RNs and LPNs drawn from multiple hospitals across the four largest health authorities in the province. In Canada and globally, a trend in health care is to use teams of RNs and LPNs to deliver patient care. Health services research, therefore, needs to include the perspectives of RNs and LPNs [
57]. Second, the explanatory model included seven indicators of workload so that independent effects of each could be investigated. However, causal inferences are limited by the cross-sectional data. Other limitations are the low response rate and inconsistency in the time dimension of the some of the measures used in the study. For example, nursing tasks left undone were measured over the last shift, but patient adverse events were measured over the last year and later recoded as less than weekly versus weekly or more often. This inconsistency may have confounded the study findings. Asking nurses’ perceptions of a phenomenon over the last year or last month also increases the possibility of measurement error due to recall bias. The low response rate of the study leads to concerns of sample bias and generalizability of the findings. High response rates, however, do not guarantee representation and vice versa: researchers need to look beyond survey response rates to factors such as non-response error. Non-response error occurs when a significant number of people in the survey sample do not respond and have different characteristics from those who do respond [
58]. As cited in Havaei et al., the total study sample was compared with Canadian Institute for Health Information reports of provincial nurse demographics [
59]. We found that this study sample is similar to the BC nursing workforce with respect to age, gender, and employment status [
60].