In most cases, hospital nurses provide 24-hour care that can encompass a variety of roles and set procedures that are required to support patients during each nursing shift. Undoubtedly, time is a limited resource, it is dynamic and it costs healthcare services money (Jones, 2010). The Australian Institute for Health and Welfare (2016) identified the need to improve the productivity of the workforce and the need to address growing service demands as key areas for healthcare reform and cost reduction.
Historically, hospital-based nursing shifts traditionally worked on an 8-hour shift pattern (Ball et al, 2014) with a day/morning (7am–3.30pm), afternoon/evening (2.45pm–11.15pm) and night shift (10.45pm-7.15am). However, over the past 25 years, there has been a change in the nursing shift pattern to include 12-hour shifts, which were seen as a means of improving work satisfaction, increasing continuity of care (Battle and Temblett, 2018) and lowering healthcare costs (Estabrooks et al, 2009). However, Griffiths et al's (2014) review of quality of care and patient safety relative to nursing shift pattern across 12 European countries found that not all countries were adopting the 12-hour shift pattern, with many opting instead for the more traditional 8-hour shift pattern. For example, in surveying 33 659 nurses from 488 hospitals, only 14% of nurses were contracted to work 12-hour shifts; in just two countries, the Republic of Ireland and Poland, 12-hour shifts were the norm (Griffiths et al, 2014). Additionally, Ball et al's (2015) review of nursing patterns in the UK identified a 20% increase over a 4-year period in the use of 12-hour shift patterns, similar to the reasons identified previously—a shortage of nurses, continuity of care, managing home life more effectively and shorter shift rotations.
The day/morning shift has traditionally been seen as having the highest degree of nursing workload because this is the time that patients' activities of daily living are attended to by the nursing team (Deberg et al, 2012). Additionally, medical ward rounds take place during the day/morning shift and specialist and multidisciplinary teams attend and review patient progress, which therefore increases the time burden experienced by both patients and nurses alike (Armstrong et al, 2015). However, there is a culture of work distribution patterns with nurses communicating that they do not want to burden the next shift with incomplete ‘tasks’, which is particularly evident on the day/morning shift where the unwritten rule is that all patient care needs should be completed in this shift period (Chan et al, 2013). Indeed, in a cross-sectional survey of NHS hospitals in the UK, Ball et al (2014) found that ‘care left undone’ accounted for deficits in patient monitoring and assistance with activities of daily living, which were attributable to staffing levels. Moreover, the authors found that missed care or care left undone was predominantly on the morning and afternoon shifts.
Perhaps what is most striking is that little is known about how nurses make effective use of their time, because workload and the types of nursing practices being performed have implications for the quality of patient care and the nurse's wellbeing (Farquharson et al, 2013). Indeed, Griffiths et al (2014) found that, on average, nurses were unable to complete three key nursing activities that were required for patient care during their shift. Moreover, nurses reported higher rates of care left undone when working 12-hour shifts than those nurses working the traditional 8-hour shift pattern.
Background
The Australian Institute of Health and Welfare (AIHW) (2016) reported that, in 2014–2015, there were approximately 10.2 million episodes of care in Australia's public and private hospitals. Between 2010 and 2015 the care episodes increased 3.5% on average each year. More recently, the AIHW (2019) report on Australian hospital statistics showed that episodes of care exceeded 11 million a year, with more than 30 million patient care days—a 2.1% increase in patient care days. As a result, the average length of stay was 5.4 days. This is not dissimilar to other countries where patient care episodes are set to increase from 1.55 to 5%, with the average hospital length of stay averaging 6.5 days (range 1–34 days) (Organisation for Economic Cooperation and Development, 2019) (Table 1). In the same period, the average length of stay for public and private hospitals decreased from 3.0 days to 2.8 days. This means that, with an increase in the number of hospital admissions and a decrease in the number of bed days, nursing workloads are increasing and this has been noted in research into the increasing numbers of nurses leaving the profession (Chaboyer et al, 2008). For example, an Australian, New Zealand and UK cohort study exploring the perceptions of nurses' workload found that nurses leaving the profession described the nursing workload as being excessive (Bogossian et al, 2014). More concerning is that this is becoming a global pattern, with 92% of nurses across 11 countries stating that they face substantial time constraints in delivering effective patient care. Conversely, within this same cohort, 96% suggested that spending more time with patients would have a positive impact on a patient's health outcomes (DeCola and Riggins, 2010). However, what is not clear from the literature is how nursing workflow is distributed over the traditional nursing shift pattern and this is perhaps related to how nursing workload is measured.
| 2010 | 2017 | Change in length of stay | |
|---|---|---|---|
| Australia | 5.0 | 4.1 | 18% (D) |
| Belgium | 7.2 | 6.6 | 8% (D) |
| Canada | 7.7 | 7.4 | 3% (D) |
| France | 5.8 | 5.4 | 6% (D) |
| Germany | 8.1 | 7.5 | 7% (D) |
| Japan | 18.2 | 16.2 | 11% (D) |
| New Zealand | 6.2 | 5 | 19% (D) |
| Russia | 10.8 | 9.3 | 13% (D) |
| UK | 6.1 | 5.9 | 3% (D) |
| USA | 5.4 | 5.5 | 2% (I) |
D=decrease; I=increase
Aims and objectives
The aim of this review was to ascertain the methods in which nursing workload across three traditional nursing shifts is determined. In particular, its key objectives were to:
Method
A preliminary literature search was conducted to identify optimal search terms and the history of the topic. These results were limited, with no articles relating directly to time studies and specific nursing shifts. There were a number of studies on direct care time; how ever, these focused predominately on workload trends and skill mix, rather than distributions of workloads. A more specific search was conducted in key databases including CINAHL, Medline and Scopus. This search yielded a large number of irrelevant articles and therefore was limited to specific subject areas. The subject areas that produced the most relevant articles were: personnel staffing and scheduling, workload, nursing care, job characteristics and task performance and analysis. Search terms used were ‘nurs*’, ‘nurs* workload’ and ‘nurs* work activities.’ These were integrated with the Boolean operator ‘and’. The Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) checklist was used to search and limit records (Moher et al, 2009) (Figure 1). Search results were refined through the inclusion criteria of full text, references and abstract availability. Articles were sourced only from peer-reviewed journals for which English was the primary language so as not to cause errors with translation, and publications were limited to the past 10 years (2007–2017) to identify current research trends. After this initial search, 26 articles were selected from which the authors individually screened the abstracts until consensus was met and these were reduced to the 10 that were included in the review (Table 2). Using a modified version of Cooper's Guide for Literature Reviews (1998), this review summarised past research by drawing conclusions from the many separate investigations that addressed the related or identical hypotheses, with a focus on research outcomes. In addition, the manuscript abstracts were reviewed using the list for qualitative studies and the STROBE checklist for observational studies (von Elm et al, 2007). None of the qualitative papers sourced met the inclusion criteria.
| Authors | Study design | Sample | Data collection | Time of data collection | Site | Findings |
|---|---|---|---|---|---|---|
|
Abbey et al (2012)
|
Time and motion observation methodology | n=76 hours of observation | Time and motion study |
|
12-bed private ICU | Direct care: 40.5% |
|
Armstrong et al (2015)
|
Prospective cohort study | n=87 patients | Nursing activity score (NAS) |
|
MCU and ICU | Mean NAS from MCU |
|
Chaboyer et al (2008)
|
Structured observational study |
n=482 hours of data |
Work sampling |
|
Four medical wards | Indirect activities: 47.3% |
|
Cornell et al (2010)
|
Direct observation |
n=27 nurses |
Direct observation |
|
2 medical-surgical units | Assessment/treatment: 18.5% |
|
Debergh et al (2012)
|
Prospective observational study | n=all patients admitted over a 4-week period | NAS |
|
Surgical paediatric ICU |
Quantifying the nursing work load per shift using NAS Mean NAS per shift |
|
Desjardins et al (2008)
|
Prospective observational study | n=30 nursing shifts | Time and motion study |
|
Surgical wards | Direct patient care: 32.8% |
|
Farquharson et al (2013)
|
A real-time, repeated measure design | 67 nursing staff |
Work observational method by activity timing (WOMBAT) |
|
Medical and surgical wards | Direct care patient care (median=37.5%, IQR 27.8) |
|
Hendrich et al (2008)
|
Time and motion study | n=767 nurses | Time and motion study |
|
15 states 36 medical-surgical units |
Documentation: 35.3% |
|
Westbrook et al (2011)
|
Prospective observational study |
n=57 nurses |
Work observation method by activity timing (WOMBAT) |
|
Two wards in a 400-bed major hospital 2005 and 2006, and then repeated in 2008 | Direct care time: 37% |
|
White et al (2015)
|
Function analysis observational study | n=35 RN and n=17 HCA shifts | Nursing role effectiveness model |
|
Two medical units | Documentation RN unit A 20.9 and RN unit B 21.4% |
HCA=healthcare assistant; ICU=intensive care unit; IQR=interquartile range; MCU=medium care unit NAS=nursing activity scoring; RN=registered nurse
Results
The overall outcome of this review of the literature was to ascertain the key characteristics by which nursing workload is often determined; in the majority of cases this is through direct observation (Table 2). Therefore, it was deemed appropriate to include those aspects of data collection that contribute to the knowledge base surrounding nursing workload. This identified three specific approaches:
Self-reporting
The methods of self-reporting in the literature were: diarising work activities, completing a scoring system and responding to a prompt from a personal digital assistant (PDA) to input data. In four articles, self-reporting was the most common form of work-sampling technique used (Hendrich et al, 2008; Debergh et al, 2012; Farquharson et al, 2013; Armstrong et al, 2015). One benefit of this method is cost-effectiveness because there is no requirement for an observer. Another consideration for the use of self-reporting rather than a direct observational method is that the low cost can allow for a larger sample size, as seen in Hendrich et al's (2008) study. Self-reporting by nurses cannot exclude the possibility of bias and the retrospective data entry may be less reliable because of variations in staff recall (Farquharson et al, 2013). Other self-reporting models include the Nursing Activity Score (NAS), used predominately in intensive care units (ICUs) as a measure of workload with the ability to assess nursing workload per shift (Debergh et al, 2012; Armstrong et al, 2015). The NAS comprises a list of 23 items and expresses time as a percentage of total time, where all weighted time activities are based on real-time assessment of duration of activities. However, its use was designed for nursing work activities in the (ICU) and therefore its suitability outside this clinical environment may be of limited use.
Farquharson et al (2013) used the real-time work-sampling technique with an electronic diary version of the Work Observational Method by Activity Timing (WOMBAT). WOMBAT was delivered on a handheld personal digital assistant (PDA) and nurses were prompted every 90 minutes to input a diary entry detailing their major activity during the previous 10 minutes. In comparison, Hendrich et al's (2008) study used a variety of collection techniques consisting of four separate protocols. The study sample of 763 nurses from 36 different medical-surgical units were randomly assigned to participate in protocol A or B, all nurses participated in protocol C and only those nurses who volunteered participated in D. PDAs were used in protocol A to examine documentation-related activities. In protocol B, the activity in which the nurses were engaged at random times of the shift (an average of 17 times per shift) from a categorical data set were recorded. Protocol C recorded the nurses' location and movement via a radio frequency identification tag worn by the nurse. Protocol D examined the nurses' physiological responses by wearing specialised SenseWear Pro Armbands measuring skin temperature, body temperature, galvanic skin response, heat flux, and motion. The multiple forms of data collection allowed for cross-validation between the protocols for variables such as distance travelled and a nurse's geographical position.
Work-sampling technique: direct observation
In work sampling, the observer takes snapshots of activity via direct observation of the nurse (Finkler et al, 1993). A work-sampling technique was used by Chaboyer et al (2008) on four medical wards to provide multiple, random, intermittent observations of the work activities to provide an estimate of the time spent on each activity. Nurse sampling, where nurses were prompted by a pager to report activities, resulted in a larger sample; however, in one study this approach missed the detail of flow, sequence and demand of work (Cornell et al, 2010). Direct observation gives more detail but is data intensive. Direct observation work sampling is perhaps a balance between the two (Cornell et al, 2010).
Three studies were identified from the literature review that included direct observation and varied in observation times. Chaboyer et al (2008) used randomised 2-hour periods of observation during the perceived busiest nursing times, with activities recorded every 10 minutes. Cornell et al (2010) used 3–4-hour blocks of observation with nurses from two medical-surgical wards who were randomly selected to be continuously observed during the period. Westbrook et al (2011) were more methodical, calculating the time each nurse would be observed using representative sampling. In this study, the WOMBAT method was used to shadow nurses for 1-hour blocks and data were recorded for all work tasks performed. The alternative would be work sampling including many work-sampling data points and a larger sample size, although the degree of precision regarding the data and conclusions would be subject to question (Finkler et al, 1993:594).
Direct continuous observation
Time and motion studies, in which nurses are observed continuously, allow for the most accurate rate of measurement of time spent on various activities and have been used for many years (Abdellah and Levine, 1954; Finkler et al, 1993; Desjardins et al, 2008). A time-and-motion study is able to capture the level of interruptions experienced by nurses (Abbey et al, 2012). Abbey et al (2012) used this method to document the activities undertaken in the ICU and were able to capture all simultaneous activities. Nurses have often undertaken multiple tasks, frequently switching from one activity to another with a high level of interruptions. White et al (2015) used a direct observation method to document the nurses' use of time second by second, using a palm pilot to input the information using functional analysis. This allowed for the capturing of all simultaneous activities that nurses performed. Desjardins et al (2008) opted for a prospective observational time-and-motion study. Data were collected by direct observation using three observers who were experienced registered nurses and was recorded on a PDA with a list of over 200 activities to choose from. Activities were categorised into: direct care, indirect care, non-nursing and personal (Table 3). If resources are devoted to a time study, it will likely be limited by the high cost, with the consequence being a smaller sample size but with very good data (Finkler et al, 1993). Finkler et al (1993) still gave preference to this sampling method. A potential downfall of direct continuous observation is the observer effect, where the presence of the observer alters the behaviour of those being observed, although Monahan and Fisher (2010) challenge the fallacy that people will self-censor in the presence of an observer and state that, over time, the observer becomes integrated into the environment.
| Morning | Evening | Night | |
|---|---|---|---|
| Debergh et al (2012) ICU | 47.0 | 46.3 | 41.6 |
| Armstrong et al (2015) MCU | 43.9 | 44.3 | 37.0 |
| Armstrong et al (2015) ICU | 41.8 | 43.1 | 32.7 |
ICU-intensive care unit; MCU-medium care unit
Acute care nursing is a 24-hour service. However, many of the research articles have restricted work sampling to peak periods. Chaboyer et al (2008) identified this as a limitation of their study. Four of the studies (Desjardins et al, 2008; Hendrich et al, 2008; Debergh et al, 2012 and Armstrong et al, 2015) (Table 2) looked at nursing over a 24-hour period and found no differentiation between the three nursing shifts in terms of direct care time and nursing workload. The literature search identified two articles that discuss workload variations over these shifts (Hendrich et al, 2008 and Farquharson et al, 2013). In an extensive US study, one of the protocols implemented compared the distance travelled by nurses per 10-hour day and per night shift. The median for day shift was on average 3.0 miles and the median for night shift was 2.2 miles, a variance by a factor of four or more (Hendrich et al, 2008:30). The authors suggested that a change in patient-related tasks and acuity is the likely cause for this (Hendrich et al, 2008).
Quantitative analysis of the differences in nursing workload per patient for a 24-hour shift was recorded using the NAS in two separate studies (Table 3) (Armstrong et al, 2015; see also Debergh et al (2012). Each of the 23 items in the NAS checklist is allocated a numerical score and a weight, the sum of which gives a numerical score correlating to the percentage of work time spent caring for the patient (Armstrong et al, 2015). There was no significant variation in the evening and day shifts; the night shift variation was significant within P<0.001. Armstrong et al (2012) found these results surprising, expecting the evening shift to yield a lower NAS score than the day shift. On reflection, a high number of admissions occurred in the evening when administrative and support staff were not present (Armstrong et al, 2015). Another observation and possible explanation was the return of postoperative patients from recovery, increasing the mean workload in the evenings (Armstrong et al, 2015).
Desjardins et al's (2008) study directly observed 30 surgical nursing shifts as a prelude to work reorganisation. They wanted to see the distribution of time with the focus of streamlining non-nursing care activities in the future. Direct care activities on the day shift represented significantly more time relative to the evening or the night shifts (Table 4). The variation was deemed statistically significant with a P value of 0.006 using a one-way analysis of variance (ANOVA). Desjardins et al (2008) identified the overall percentage of indirect care time as unexpectedly high, with the main indirect care activities being communication between health professionals, medication verification/preparation and documentation. Limitations of this study is the small sample size (Desjardins et al, 2008).
| Direct care (P=0.006) | Indirect care (P=0.002) | Non-nursing (P=0.022) | Personal/Nil (P=0.05) | |
|---|---|---|---|---|
| Day (n=14) | 37.2% | 52.3% | 9.1% | 1.3% |
| Evening (n-=10) | 28.8% | 55.6% | 10.7% | 4.9% |
| Night (n=6) | 29.2% | 63.4% | 5.9% | 1.5% |
Categorisation of workload
There is a significant variation in the categorisation of nursing tasks, often directed by the focus of the study. A long list of categories to define tasks makes the data collection tools difficult to use, although too narrow a list means that valuable information is not captured (Cornell et al, 2010). A small set of broad categories such as direct versus indirect care does not provide the detail necessary for process improvement and evaluation (Cornell et al, 2010). The activity sampling and classification systems presented in the literature are varied, with from as little as four to 200 activities being recorded. The predominant focus of these studies was on capturing workload distribution and not on the differentiation of workload across the three shifts. Common themes for categorisation of nursing time identified in the literature are direct care, indirect care, unit-related care and personal. Often each category would vary and encompass a selection of subcategories. Direct care was defined consistently as activities that took place in the presence of the patient (Chaboyer et al, 2007; Desjardins et al, 2008; Westbrook et al, 2011; Abbey et al, 2012; Farquharson et al, 2013). Indirect care was most commonly defined as activities relating to patients that took place away from the patient, for example medication preparation (Chaboyer et al, 2007; Desjardins et al, 2008; Westbrook et al, 2011; Abbey et al, 2012; Farquharson et al, 2013). Unit-related activities included activities that were unrelated to the patient, such as counting narcotic medications, answering telephone calls and activities associated with patient flow (Chaboyer et al, 2007; Westbrook et al, 2011; Abbey et al, 2012; Farquharson et al, 2013). Personal time was less defined and had the widest variation of inclusions, with staff breaks and personal conversations placed in this category (Chaboyer et al, 2008; Desjardins et al, 2008). The focus of categorisation was to cluster significant increments of time to make comparisons and identify targets for change (Hendrich et al, 2008).
Discussion
Understanding nursing work flow is an important consideration with any quality improvement activity, including understanding efficiencies and resource planning (Cornell et al, 2010). An increase in the number of patients with complex medical needs is a driver for re-evaluating nursing workload distribution patterns (Chaboyer et al, 2008; Cornell et al, 2010). An example of this is the use of the NAS to identify workloads and determine optimal nurse-to-patient ratios in the medium care unit (MCU) (equivalent to the high-dependency unit in the UK) (Armstrong et al, 2015). Every patient has different care requirements and these can fluctuate through a nursing shift. The NAS, or a similar scoring system, would have benefits to staffing levels not only in the ICU and MCU environments. However, what is often missing from these types of models is the omission of the mental workloads of nurses and the complex emotional needs of patients and their family members. Cornell et al's (2010) study demonstrated the chaotic pace of nursing work and the rapid switching that takes place from one activity to another. As a result, there are identifiable risks with this pattern of behaviour, including the impact on cognition, slowed performance and an increase in errors, especially medication errors (Cornell et al, 2010).
Moreover, the nature of a nurse's work differs from other industries and it is difficult to document because of its dynamic, non-linear and complex nature (White et al, 2015). Studies examining nursing across the three shifts found that nurses walk more on the day shift than the night (Debergh et al, 2012; Armstrong et al, 2015) and that direct care activities on the day shift represented significantly more time relative to the evening or the night shifts (Desjardins et al, 2008). In all studies reviewed, although direct patient care accounted for less than 50% of all nursing activity, more time was spent on indirect patient care such as handovers, ward rounds and documentation and non-nursing activities such as cleaning and restocking (Desjardins et al, 2008; Chaboyer et al, 2008). On the whole, the majority of nursing activities are spent away from the patient and often out of sight of the patient (Chaboyer et al, 2008). It is argued that taking nurses away from what is deemed essential nursing work can impact on the quality of patient care, and can influence a nurse's job satisfaction negatively and therefore staff retention (Duffield et al, 2008). It can also, in some cases, lead to compassion fatigue. Nursing a patient is undoubtedly a holistic role and to fragment and divide this role into a list of skilled and unskilled tasks in order to be cost-effective seriously devalues the patient and the nursing role (Jones, 2010). There is evidence that suggests that having an increased nursing presence in direct patient care areas can and does improve safety, reduces the risk of falls, identifies the deteriorating patient earlier and puts patients at ease (Jones, 2010).
Implications for practice
Nurses and patients want more time devoted to patient care (Jones, 2010). The literature reviewed in this study showed that, in all the studies, direct nursing care accounted for less than 50% of all nursing time. Therefore, the implications for practice are consistent in establishing:
Conclusion
Nursing surveillance is important for the detection of errors and the prevention of adverse events (Duffield et al, 2008). Inadequate nursing time contributes to poor quality care and excess nursing time contributes to high costs of care (Jones, 2010; Debergh et al, 2012). Parallel to this is the nurse's job satisfaction. Quality of care and job satisfaction are important factors for the sustainability of the nursing workforce.
There are few high-quality nursing studies that detail the workload distributions across the three nursing shifts and this is a potential area for further research. For example, the findings from one study suggested that nurses walk more on the day shift than the night, the mean NAS in the ICU and MCU was comparative between the day and evening shifts and lower on the night (Armstrong, et al, 2015; Debergh et al, 2012). In addition, Desjardins et al's (2008) study showed that direct care activities on the day shift represented significantly more time relative to the evening or the night shifts. Therefore, it is easily seen both empirically and anecdotally that more nursing time is expended in the morning shift than the other two; the night shift has its own exceptions. However, what is often forgotten is that nursing offers 24-hour care and therefore the mindset that everything has to be done in the morning shift has to change not only for the good of nursing care but for patient wellbeing.