Duffield, C., Roche, M. A., Diers, D., Catling-Paull, C., & Blay, N. (2010). Staffing,
skill mix and the model of care. Journal Of Clinical Nursing, 19(15-16), 2242-2251.
Aims and objectives
The study aimed to explore whether nurse staffing, experience and skill mix influenced the model of
nursing care in medical-surgical wards.
Methods of allocating nurses to patients are typically divided into four types: primary nursing, patient
allocation, task assignment and team nursing. Research findings are varied in regard to the relationship
between these models of care and outcomes such as satisfaction and quality. Skill mix has been
associated with various models, with implications for collegial support, teamwork and patient outcomes.
Secondary analysis of data collected on 80 randomly selected medical-surgical wards in 19 public
hospitals in New South Wales, Australia during 2004-2005.
Nurses (n=2278, 80.9% response rate) were surveyed using The Nursing Care Delivery System and
the Nursing Work Index-Revised. Staffing and skill mix was obtained from the ward roster and other
data from the patient record. Models of care were examined in relation to these practice environment
and organisational variables.
The models of nursing care most frequently reported by nurses in medical-surgical wards in this study
were patient allocation (91%) and team nursing (80%). Primary nursing and task based models were
unlikely to be practised. Skill mix, nurse experience, nursing workload and factors in the ward
environment significantly influenced the model of care in use. Wards with a higher ratio of degree
qualified, experienced registered nurses, working on their ‘usual’ ward were more likely to practice
patient allocation while wards with greater variability in staffing levels and skill mix were more likely to
practice team nursing.
Models of care are not prescriptive but are varied according to ward circumstances and staffing levels
based on complex clinical decision making skills.
Relevance to clinical practice
Variability in the models of care reported by ward nurses indicates that nurses adapt the model of
nursing care on a daily or shift basis, according to patients’ needs, skill mix and individual ward
Models of care, nursing care delivery systems, skill mix, staffing levels, nursing workload, practice
The term ‘model of care’ is often used in the health literature but not always with an agreed-on
meaning. It can describe broadly how the delivery of care is provided in the context of the whole health
system (Queensland Health 2000) or a care delivery model at a ward or unit level which is supported by
the organisation, has clearly delineated standards and a foundation of evidence-based practice and
principles (Davidson et al. 2006). Nursing models of care have traditionally been concerned with
nursing care delivery systems – a method of allocating workload (patients) to staff (nurses).
In this paper, we present findings from a study which had as one of its key objectives, to examine the
relationships between models of care (here defined as a nursing care delivery system), staffing, skill
mix and the practice environment. The models included in the study were primary nursing, total patient
care (patient allocation), team nursing and task-oriented or functional nursing.
Nursing Models of care
Historically in Australia, the method of allocating patients to staff has been determined by the nurse in
charge of staffing and standards of care on the ward – the Nursing Unit Manager (NUM). Traditionally,
four methods have been described and used here and overseas in countries such as the UK (Thomas
& Bond 1990), USA (Hyams-Franklin et al. 1993), Canada (Shamian et al. 1988), Hong Kong (Chan et
al. 2006) and Finland (Makinen et al. 2003) to name a few.
The first of the four ‘classic’ models of delivering nursing care is ‘patient allocation’ or ‘total patient care’
which has one nurse taking responsibility for the complete care of a group of patients. It is the oldest
model (Marquis & Huston 1992), believed to be based on one used by Irish nurses in the Crimean war
and adapted by Nightingale (Meehan 2003). Total patient care remains popular with patients due to the
high quality and consistency of care that is provided by qualified staff and more importantly, by the
same nurse (Wagner & Bear 2009, Wolf et al. 2008). It is also popular with nurses because of the
degree of autonomy and control they have over their work and responsibility for a particular patient’s
care is easily identified (Gullick et al. 2004, Tiedeman & Lookinland 2004). It does, however, require an
almost all registered nurse (RN) staff (Gullick et al. 2004).
Functional or task-oriented nursing emerged in the 1940s due to a war-related shortage of registered
nurses resulting in the employment of more support staff in hospitals (Marquis & Huston 1992). The
model divides work into tasks assigned to nursing and ancillary personnel based on the complexity of
the task. The RN addresses more complex tasks, whereas lesser skilled staff undertake more routine
tasks. It has fallen out of favour recently due to its regimentation, lack of recognition of patients as
individuals and little time devoted to patients’ psychosocial and spiritual needs (Tiedeman & Lookinland
2004). Patient satisfaction is low in this model as patients cannot establish who is ‘their nurse’ (Ringl
Team nursing emerged following World War II (Tiedeman & Lookinland 2004) in response to the
impersonal nature of task-oriented nursing and in an effort to improve both patient and staff satisfaction
(Marquis & Huston 1992). It is a model whereby a ‘team’ of nurses provides all the care to a specified
group of patients, often allocated by geography. This model allows for supervision of aides, orderlies
and licensed practical or enrolled nurses by a smaller number of RNs. In this model, the RN team
leader supervises lesser-trained patient care providers and performs direct patient care that lesser-
skilled staff are not qualified or licensed to provide. Other team members provide direct patient care to
assigned patients (Tiedeman & Lookinland 2004). In the team nursing model, task-oriented or even
primary nursing elements of care may exist.
Primary nursing, developed in 1968 (Manthey 1980), focuses on providing care in a one patient to one
nurse relationship that promotes continuity of care. Patients are assigned a specific nurse on admission
who assumes 24-hour responsibility for their care for the duration of their stay and their care is
therefore not shift dependent. The primary nurse has the authority to assess, plan, organise,
implement, coordinate and evaluate care for his/her patients and delegates other nurses to follow
through when off duty (Lyon 1993, Manthey 1980, Tiedeman & Lookinland 2004). Should this patient
be readmitted, the same nurse (assuming s/he is still employed) would again undertake responsibility
for their care. This model has been shown to be used widely in Magnet-designated hospitals and linked
to positive patient outcomes (Aiken et al. 1999, Mondino 2005) and staff satisfaction (Allen & Vitale-
Nolen 2005, Garon et al. 2009). Primary nursing, recently renamed Relationship-Based Care, is based
on the philosophy that the development of a nurse-patient-family relationship is paramount to continuity
of care. This model has been shown to empower nurses, promote autonomy and contrary to popular
belief, does not always require an all-RN staffing mix (Manthey 2003). Despite the evidence from
overseas this model is rarely used in Australian hospitals (Middleton et al. 2008) with the exception of
some critical care units (Watts et al. 2007).
There is little research which evaluates these models of care and findings in terms of satisfaction,
quality and cost of care are contradictory (Tiedeman & Lookinland 2004). A comparative study of team
nursing and patient allocation in Australia found no difference in patient satisfaction (Wu et al. 2000).
Generally, patient satisfaction is positively associated with the number of RNs, a better nurse to patient
ratio and level (years) of experience (Tervo-Heikkinen et al. 2008) but often, patients are unaware
which model is being used (Gullick et al. 2004). There is also a suggestion that patient satisfaction is
greater when continuity and communication are maintained rather than as the consequence of a
specific model per se (Tiedeman & Lookinland 2004).
McGillis-Hall and Doran (2004) in a Canadian study of the effect of skill mix and delivery model on
nurses’ perception of care quality found ‘total patient care’ to be a negative influence on these
perceptions. Recently in Australia, Walker, Donoghue and Mitten-Lewis (2007) conducted a natural
experiment. Work sampling studies evaluated a change from a patient allocation model to team nursing
incorporating a change from an all RN workforce to one which included enrolled nurses and assistants
in nursing. To accomplish this change, bed numbers were reduced and more complex patients were
admitted to another clinical unit, thereby reducing caseload and acuity. Following implementation of the
team model, the authors found that there had been an overall decrease in unit related activities and an
increase in direct and indirect patient care activities and hence a more effective and efficient utilisation
of staff. In another Australian study, clinical nurses initially supported implementation of a ‘shared care’
model incorporating aspects of patient allocation and team nursing. Following implementation,
concerns over ineffective communication and reduced availability of experienced staff were expressed
and the incidence of infection rates, incidents (unspecified) and accidents increased (Fowler et al.
2006). Gullick and Shepherd (2004) found that nurses felt more satisfied and autonomous when using
the patient allocation model and that team nursing could leave senior staff exhausted by the need for
constant supervision. In a systematic review on team nursing, Pearson et al. (2006) found that social
support from supervisors or peers in nursing teams increased job satisfaction.
Skill mix plays a role in the choice of which method of care delivery is used. Many years ago Durbin
(1981) suggested that poor skill mix may turn a team-oriented model into a more task-oriented care
delivery system. More recently Adams and Bond (2003a, b) found wards with a poorer skill mix (lower
RN proportion) were more likely to be characterised by a more task-oriented, hierarchical approach.
Gullick et al. (2004) suggested that patient allocation was preferable to maintain quality of care but may
not be possible where RN numbers are low or the skill mix (fewer RNs) demands greater supervision, in
which case team nursing may be more practical. Interestingly, in team nursing a good climate in a team
may offset the reality of inadequate staff dealing with heavy workloads (Sexton et al. 2006). There is a
suggestion that wards using the ‘total patient care’ delivery system have a team nursing focus when
skill mix is thin (McGillis-Hall & Doran 2004). The link between better skill mix (more hours of care
provided by RNs) and positive patient outcomes has now been established in many countries (Kane et
In Australia more recently, there has been a move away from patient allocation models towards a more
team-oriented approach. Using focus groups with over 1000 nurses in rural and metropolitan New
South Wales (NSW), Chiarella and Lau (2006, 2007) found a range of team-based models of nursing
care implemented with consistently positive outcomes, measured as self-reported improved
communication, service integration and comprehensive patient care.
This paper presents findings from a portion of a larger study (Duffield et al. 2007). The present report
examines the relationships between staffing, skill mix and the model of care. The study was conducted
in an environment in which the length of stay of patients was 3.23 days, patients moved on average
2.26 times per episode of care, 1.25 patients went through each medical-surgical bed over a 24 hour
period and where staff ‘churn’ (turnover plus staff and student rotations) averaged 50% (Duffield et al.
2009b, Duffield et al. 2009c).
The component of the study reported here was conducted on 80 randomly selected general
medical/surgical units at 19 public hospitals in NSW, Australia during 2004-2005. Each ward was
studied for seven days. Ethics approval was gained from the Human Research Ethics Committee,
University of Technology, Sydney, 13 health service committees and the NSW Department of Health.
This is a secondary analysis of these data.
Nurses included in data collection were:
• Clinical nurse specialists (CNS) – a personal grade awarded to individual nurses on the basis
of expertise in a specialty demonstrated by qualifications and/or experience (NSW Health
• Registered nurses (RN) – responsibility for preparing RNs resides in the university sector,
leading to registration at bachelor degree level; years of experience range from less than one
year (RN1) to eight or more (RN8);
• Trainee enrolled nurses (TEN) and enrolled nurses (EN) – requires one year paid vocational
training incorporating 15 weeks at a technical college for theoretical training and the balance in
clinical units and is equivalent to LVN/LPN;
• Assistants in nursing (AIN) – equivalent to patient care assistants.
Each nurse participating in the study was asked to complete a Nurse Survey. This survey captured a
wide range of individual nurse data, including the Nursing Care Delivery System instrument (NCDS)
(Murray 1995) to determine the model of care in use on the ward and the 49 item Nursing Work Index-
Revised (NWI-R) (Aiken et al. 2001, Estabrooks et al. 2002, Sochalski et al. 1999). Nurses also
reported their qualifications and whether they considered themselves expert (defined as those whom
other nurses ask for their clinical judgment or who act as preceptors).
Trained nurse data collectors obtained staffing data, including skill mix and hours worked from ward
rosters. Nursing workload (acuity) data were captured for each patient per day using the PRN-80
(Chagnon et al. 1978a, b, Tilquin et al. 1981) from the contemporaneous medical record. PRN-80 and
staffing data were used to calculate the ‘Nursing Demand/Supply’ variable by dividing the required
hours of care per patient from the PRN-80 by the hours of care provided (O'Brien-Pallas et al. 2004).
This provided a measure of the balance between the work required and nursing hours available.
‘Staffing hours’ provided details of the proportion of hours worked by each employment category
(including full or part-time and casual and agency staff) and level of nurse per ward and also the
staffing hours available for all shift periods. Patient characteristics such as admissions for respite care,
or waiting for a care facility (such as a nursing home) were also collected. Data were cleaned and units
with incomplete staffing data were excluded from the analysis. Complete data were available for 77
Each respondent to the NCDS indicated on a 4-item questionnaire the degree to which each of the
following models (primary, patient allocation, team nursing and task nursing) described the organisation
of patient care on that unit, on a four-point scale from ‘a lot like’ to ‘not at all like’:
• I have overall 24-hour responsibility for a group of patients and manage their care from
admission to discharge
• I provide almost all of the care for our assigned patients throughout the shift but I may not have
the same patients from their admission to discharge
• We work as a team or group of nursing staff over the shift and together work to provide care to
a group of patients
• We look at the tasks that need to be completed on the shift for patients and assign staff various
tasks (Murray 1995)
Responses to the NCDS therefore reflect the nurses’ perception of the degree to which each delivery
system is in operation on that ward, thereby acknowledging the variation that may occur with different
mixes of staff and patients, or at different times of the day. Data were treated as categorical and
calculated as a proportion of each response per question. Analysis of responses as the highest
proportion per ward provided an indication of the models in use for each ward.
Data were analysed using Stata analytic software (StataCorp LP 2007). Nurses’ responses to the
NCDS were first calculated as the proportion of each response per question across the sample
(n=2278) and then aggregated to the ward level (n =77) to provide the proportion of responses per
ward for each model. This provided an indication of the degree to which each model of care was
perceived to be in use on that ward. Linear regression analyses were performed using NCDS variables
as dependent variables and staffing, environmental and nurse variables as independent variables. That
is, we reasoned that nursing models were potentially the creation of operational contingencies. To
compare the relative contributions of the independent variables in the models, standardised coefficients
(beta [β] weights) were calculated and reported where significant at the 0.05 level.
Data obtained included survey responses from 2278 nurses (response rate 80.9%) ward staffing on
13,442 nurse-shifts and 22,497 patient-days of information.
Reponses to the NCDS reflected the variation in care delivery systems in operation. Table 1 shows the
reported nursing models of care per nurse (n=2278). Only 15.3% of nurses reported having 24-hour
responsibility for their patients from admission to discharge (primary nursing) and 55.7% reported that
their care delivery system did not operate this way at all on the ward. Over two-thirds of nurses (69.3%)
reported they looked after assigned patients throughout their shift (patient allocation model), but did not
always have the same patients from admission to discharge (as with primary nursing). An additional
22.4% reported that this sounded a little like their system, so effectively nearly 92% of nurses indicated
that this system was in use on that ward at some time. Nurses who reported working in team nursing
models totalled 46.4%, while 34.1% thought it sounded a little like their system. Therefore, 80.5%
nurses considered themselves to be working in a team-related model. The task assignment model of
care was the least likely with 18.7% nurses reporting they worked this way. However, nearly a quarter
(23.2%) thought the description sounded a little like their system. When calculated as the greatest
proportion of responses per ward, the data indicated that either patient allocation or team nursing were
in operation, with no wards favouring task assignment or primary nursing.
The skill mix varied considerably across wards. The proportion of RN staff hours worked ranged from
less than 45-100%. Twelve of the 80 wards had fewer than 60% of RN hours worked while 19 had more
than 80%. On average, one-third of hours (31.3%) were worked by part-time nurses and 13.9% by
casual or hospital pool staff. The lowest percentage of hours worked by full time staff was 24.5% on
one ward and the highest percentage was 84.6%. There were ten wards which had less than 40% and
only three wards that had over 80% full-time hours worked. One ward had a high of 47.8% of casual
and agency hours worked and only one ward in the sample employed no casual or agency staff at all.
Twenty- three wards had greater than 20% casual and agency hours worked.
An examination of staffing hours by nurse grade/year found that RN8s (8 years of experience or more)
provided most of the staffing hours (32%), with ENs providing 22.1% and RN1s (first year of practice)
10.2%. The distribution of staffing across 24 hours showed that most staffing hours were allocated
during the day and evening periods. Considering only RN grades, it was found that more experienced
RNs were working at night, while nurses with one year of experience or less (RN1s) were more likely to
work during the day and evening shifts (see Figure 1).
Correlation of patient, nurse, skill mix and ward environment characteristics with the model of care
found several statistically significant associations (Table 2). Skill mix variables (proportion of hours
worked by RNs) in particular were associated with patient allocation and team nursing approaches. A
higher proportion of expert nurses correlated negatively with team and task models but positively with
patient allocation. The number of hours nurses worked away from their usual ward was correlated with
a task assignment model. The proportion of nurses with a bachelors degree or higher was positively
associated with patient allocation and negatively with team nursing. A higher proportion of patients on
the ward waiting for a care facility (perhaps older and more dependent patients) were more likely to be
cared for in a team nursing rather than patient allocation model. Heavier workload (i.e. a greater
difference between hours of care required by patients and that supplied), more nurses working on their
usual ward, more experienced nurses on the ward and a higher proportion of part time nurses were all
associated with a lower likelihood of primary nursing as the model of care.
Linear regressions were then conducted on the delivery model variables, to determine the strongest
predictors of the various approaches (Table 3). Skill mix, together with nurse characteristics such as
experience and expertise, were found to be significant. The workload of nurses was a negative
predictor of both primary nursing and team nursing, while the hours that nurses were working on wards
other than the one to which they were usually assigned was linked to the task assignment model.
As skill mix had been found to be significant, a further analysis was undertaken with the proportion of
hours of RN by years of experience as predictor variables (Table 4). A higher proportion of most
categories of RNs was related positively to the patient allocation approach, while lower proportions
were negative predictors of team nursing. This supports the previous analysis that the proportions of
RN and EN staff may influence the model used, but does not provide further information on the
relationship between years of experience of RNs and the model of care in use.
In summary the results indicate that there was no predominant model of care (care delivery system) in
operation on any of the wards. Patient allocation or team nursing were most likely and no wards
favoured task assignment or primary nursing. Skill mix and nurse characteristics such as experience
and expertise were found to be significant predictors in the model of care in use. A higher proportion of
RNs predicted a patient allocation model.
The prevailing view in Australia has been that the model of care (nursing care delivery system) used on
a ward has been a ‘philosophical’ decision determined by the nursing unit manager. However the
results from this study challenge this notion. Nurses on the same ward and even on the same shifts on
a ward indicated that they worked to different models. This arises because of the extraordinary level of
variability which occurs now in staffing nursing units and the demanding and unpredictable working
environment (Duffield et al. 2009a, Duffield et al. 2009b). The basis on which a nursing unit manager or
shift team leader decides to allocate staff to patients is now more likely to be a pragmatic rather than
philosophical decision designed to make best use of the staffing resources available and patient
requirements at any given time and made daily or by shift rather than as a unit policy.
To staff their unit or ward nursing unit managers must know and understand their patient population to
use available staff resources to advantage. Shortened lengths of stay, high usage of part-time, casual
and agency staff together with a workforce with significant numbers of newly graduated registered
nurses provide a context where staffing decisions and the allocation of patients to staff have become
more complex. It is not surprising that decisions about how to allocate patients to staff relate to skill mix
and staffing stability (staff working on their usual wards). Task assignment is applied when there is a
poorer skill mix and when nurses are moved to another ward and the ward and patients are unknown to
them. Under these circumstances, the nurse in charge of the shift may feel that task assignment is the
only safe method of allocating patients to staff.
In contrast, patient allocation is associated with richer skill mix (more RNs), more RNs with degrees or
higher and more expert nurses. Better skill mix is predictive of patient allocation even if the RNs were in
their first year of employment. Newly registered nurses (RN1) comprise a significant proportion of the
workforce now on acute medical and surgical wards. Working alone (as occurs with patient allocation),
may not be the best method for inexperienced nurses to learn how to provide care. This method of
caring for patients may predominate as it is the ‘model’ taught in most undergraduate nursing curricula.
It is clear from the results that the model of care used is no longer a decision based on the
philosophical beliefs of the nursing unit manager about how they would like to manage patient care on
their ward or unit. It is a pragmatic decision made by whomever is in charge of the day or shift about
what is the best way to allocate patients to staff to ensure safety. To do this successfully requires skills
not previously expected of managers, certainly not in Australia.
The nursing unit manager needs to understand the casemix of their unit; patient flow daily, weekly and
annually; the skills and knowledge of staff, many of whom may be relatively unknown; which method of
allocating patients produces the best outcomes for their patient mix. More importantly, some of these
same decisions are made on a shift-by-shift basis by team leaders and registered nurses, many of
whom may be very inexperienced first year registered nurses.
Nursing unit managers require the authority, responsibility and accountability for determining
appropriate staffing assignments. To do this successfully they require data which is accurate and
timely. Most importantly they need data on what has worked on their unit with a particular patient mix