Antecedents of unfinished nursing care: a systematic review of the literature

Within the overarching UNC term [1], there are three main concepts: Tasks Left Undone, Missed Nursing Care, and Implicit Rationing of Nursing Care. Tasks Left Undone was first conceptualised by Solchalski in 2004 [4], defined it as activities left unfinished during the last shift because nurses lacked the time to undertake them. In this context, nurse workloads and time constrains were both considered antecedents; however, specific factors triggering or hindering tasks left undone were not conceptualized [4].

A few years later, Kalisch [13] introduced the concept of Missed Nursing Care as every aspect of nursing care required by a patient that is partly or totally omitted or significantly delayed. In the first theoretical model, four elements at the nurse level were related to Missed Nursing Care: team norms, decision-making processes, internal values and beliefs, and habits [14]. In the same year, Kalisch and Williams [6] developed the MISSCARE survey to measure Missed Nursing Care. This instrument also includes a set of other reasons of missed care as perceived by nurses, namely deficiencies in communication, material resources, and labour resources. A few years later, Kalisch and Xie [15] advanced their model by including three additional antecedents regarding (a) the hospital (size, teaching status, Magnet status), (b) the unit(s) (case mix index, nurse staffing levels, type of nurse staffing, absenteeism, overtime, and work schedules), and (c) the teamwork characteristics. In the same period, Schubert et al. [3] established the Implicit Rationing of Care concept as the withholding of or failure to carry out necessary nursing measures for patients. In their theoretical model, antecedents considered (a) the organizational variables (namely the budget, policy priorities, resource allocation, the management structure, the culture, and the climate); (b) the nursing work environment variables, including the adequacy of resources and skill mix, interdisciplinary collaboration, nursing management, autonomy, and responsibility; and (c) the philosophy of care variables, such as the priority setting, the cultural values, the standards of care, and local and national guidelines. Moreover, patient variables, including illness severity and co-morbidities, and nurse variables, such as the nurses’ experience(s), education, skills, and knowledge have been established as other factors affecting the occurrence of Implicit Rationing of Nursing Care [3].

According to the state of the research in this field, UNC antecedents were considered in a linear relationship [4, 14], within an input–process–outcome model [16]. In more recent years, the debate has moved to more complex models, the so-called systems approach [16], where UNC has been examined holistically rather than as the sum of different parts. In this context, researchers have considered several factors as interrelated each other [3]. For example, hospital units comprising different sub-systems interacting with each other, are influenced by the nursing philosophy and the work environment that might be different in each of them [3]. Alongside these internal interactions, external factors might affect each unit, as for example, the hospital’s organisational variables [3]. Therefore, a multi-level approach has been introduced in this research debate, examining how upper-level management might affect the clinical nurses and, consequently, the UNC occurrence at the bedside [16]. Thus, factors external to the unit at the hospital, regional, or national levels (e.g., policies, rules) as implementing cost-containing measures in the attempt to increase productivity and efficiency, might affect the UNC occurrence at the bedside [3, 16].

The Sample, Phenomenon of Interest, Design, Evaluation, Research type (SPIDER) [18] methodology was used to establish the review question. Then, according to the elements specified (Table 1), the search terms were identified [19] (Table 2) without considering specific key words as expressing the influence of specific factors (e.g., morning shifts) [20].

In a preliminary phase, the International Prospective Register of Systematic Reviews (PROSPERO) database was checked to determine whether there were ongoing systematic reviews about the antecedents of UNC. Then, MEDLINE, the Cumulative Index to Nursing and Allied Health Literature (CINAHL), and SCOPUS were searched.

There were included those primary studies: (a) investigating antecedents of the UNC in adult care settings (patients ≥ 18 years of age); (b) providing measures of relationships between the investigated antecedent(s) and the occurrence of the UNC; (c) employing quantitative designs and reporting the abstract, (d) published in English, German, Greek or Italian, the languages accessible to the research team (see authors); (e) from 2004 up to 21 January 2020, when the Task Left Undone concept, included in the UNC overarching term, was first established [4, 21]. Reviews, systematic reviews, overviews, or integrative reviews retrieved were scrutinised in their references manually for relevant primary studies potential eligible according to the inclusion criteria.

Specifically, there were included those studies providing inferential statistics such as correlations, associations (odds ratios, relative risks), or other estimations (e.g., beta) to evaluate the relationship between one or more antecedents and the occurrence of the UNC. In some studies, the authors did not provide sufficient data to establish how the investigated variables were associated with UNC [22]. In these cases, the available conceptual models of the UNC were used to assess the study’s eligibility [3, 14, 23].

There were excluded those studies: (a) qualitative in nature; (b) regarding settings caring for patients < 18 years (paediatric) and other settings (obstetrics and psychiatric) due to specific care provided and the relevance of other potential factors in hindering/increasing UNC (e.g. the role of family relatives) [21]; (c) including other health care professionals (e.g., midwives), (d) reporting only descriptive measures of antecedents (e.g., frequencies), thus not assessing associations with UNC; (e) not reporting an abstract and published in languages other than those previously listed. Reviews were also excluded, although their reference lists were checked manually for appropriate studies.

A total of 1,120 sources were identified (Fig. 1). Subsequently, 990 studies, including 291 duplicates, were excluded by two researchers’ experts in the unfinished care field, who evaluated titles and abstracts independently and then agreed upon which studies to evaluate further. The remaining 132 studies were carefully read in their full texts by the same researchers, in an independent fashion. Disagreements emerged regarding four studies; thus, the entire research team (see authors) was involved in multiple meetings, in order to reach consensus regarding the inclusion. At the end of the process, 58 studies were included.

Two researchers developed the data extraction grid by including the following elements: author(s); country; aims; design (if longitudinal: assessment points); when the study was performed (year(s); the setting (hospital, community, number of units or centres involved); the sampling methods used; the participants involved and their main demographic characteristics; the instruments used for data collection (explanatory variables [as antecedents] and the UNC phenomenon); and the main findings regarding the relationships between the antecedent(s) and the occurrence of UNC.

The entire research team then approved the extraction grid via an online meeting. Three researchers piloted the grid independently with three studies, and all authors agreed that no changes were needed at the end of the pilot. Then, the same three researchers independently extracted the data and agreed upon.

The data extracted were thematically analysed [24] using an inductive approach [25]. First, all antecedents have been summarised and categorised; then, the directions of their relationships with UNC were also summarised as increasing, decreasing, or not influencing the UNC occurrence by considering the study design and the data extracted. Researchers worked independently and then agreed upon the findings. The entire research team (see authors) reached consensus about the identified categories and the overall findings in two online meetings.

The 58 studies were evaluated for their methodological quality with the Joanna Briggs Institute Critical Appraisal approach. This was selected according to its capacity to guide the development of high-quality systematic reviews addressing policy and practice interventions [26]. First, different tools were selected according to the design of the studies included, namely the critical appraisal tool for analytical cross-sectional, cohort, and quasi-experimental studies [26‐28]. Then, the evaluators were trained in the use of each tool with an online meeting; during the training, multiple exercises were offered to answer each item included in the tools (Y: yes; N: no; U: unclear; NA: not applicable) in order to ensure rigor in the assessment. Then, there were identified three couples of researchers responsible for around 19 studies/each: the evaluation was performed by one researcher and then cross-checked by a second researcher. In the case of disagreements, the entire research team was involved in multiple meetings, in order to discuss the evaluation and reach a consensus. All 58 studies demonstrated sufficient quality and, therefore, all were included in the review.

The 58 studies included (Additional File 1) collected from 2006 [29, 30] to 2018–19 [31]. The majority were cross-sectional, except for three cohort studies [30, 32, 33] and one quasi-experimental study [34]. Most of the studies aimed to investigate the occurrence of UNC and its antecedents and only one was performed to validate the MISSCARE survey [35].

Twenty-one studies were conducted in the United States [29, 34, 36‐54]; five in Australia [11, 55‐58]; three each in Switzerland [59‐61] and Cyprus [8, 62, 63]; and two each in England [32, 64], South Korea [65, 66], Israel [67, 68] and China [69, 70]. The remaining were conducted in Brazil, Canada, Denmark, Germany, Korea, Kuwait, Iceland, Italy, Ireland, Mexico, the Philippines, South Africa, Sweden, and Jordan. Only four multi-country studies have been conducted to date [37, 71‐73].

Most of the studies involved acute care hospitals, predominantly medical, surgical, rehabilitation, and intensive care units, and only six studies were conducted in nursing homes [33, 44, 45, 54, 60, 61].

The convenience sampling method was used in 47 studies; the remaining 11 studies [29, 33, 35, 42, 45, 54, 60, 61, 64, 73, 74] used random selection. The assessment of both antecedents and the occurrence of UNC was largely based on nursing staff perceptions. Sample sizes ranged from 71 to 33,659 nurses, mostly including registered nurses (RNs) and nursing assistants, with participation ranging from 8.1% to 100%. The majority of them were female (48.9%–100%), with experience in the role ranging between 5.14 and 16.6 years. Only two studies included patients as a target population [32, 70].

Thirty-four studies used a version of the MISSCARE survey [34‐38, 40‐45, 47‐53, 55‐58, 62, 63, 66‐68, 73‐79] while seven used the Basel Extent of Rationing Nursing Care Assessment tool [8, 59‐61, 64, 65, 70]. Only one study used the Perceived Implicit Rationing of Nursing Care survey instrument [39].

Studies were conducted with a variable methodological quality. Failures in reporting confounding factors and the strategies implemented to deal with the confounders identified, were the major deficiency for the cross-sectional studies. Among cohort and the quasi-experimental studies fewer failures have emerged (Supplementary Files 2, 3 and 4).

Antecedents of the UNC have been categorised at the unit, nurse, and patient levels as summarised in Table 3.

As reported in Table 4, staff levels, measured as the adequacy of staff perceived by nurses, the nurse-to-patient ratio, or the hours-per-patient day, have been the most investigated UNC antecedent to date. According to the adopted measure (e.g., adequacy versus inadequacy), studies have documented a positive or negative influence on UNC. Evidence has been accumulated regarding the relationship between a low nurse-to-patient ratio and the increase in the UNC occurrence. Exceptions have been documented by Orique et al. [40] and Zhu et al. [70] in their cross-sectional studies, and by Griffiths et al. [32] underlying a non-linear effect when the nursing hours per patient day were greater than seven. Regarding a component of staff adequacy, as the nursing unit’s number of bedside duty hours out of the number of hours offered by licensed and unlicensed personnel (= skill mix), only Castner et al. [41] documented that an increased skill mix decreased the occurrence of UNC.

The workloads measured with different metrics as the number of patients admitted or discharged [32, 80] have been reported to increase the occurrence of the UNC when measured as a global score by using the MISSCARE survey [41]. Exceptions have been reported by McNair et al. [42] and Orique et al. [40] where workloads were not found to affect the occurrence of UNC. Moreover, performing non-nursing tasks, which might contribute to increased workloads, has also been reported to affect, mainly increasing, the occurrence of UNC [34, 69, 84].

Regarding working shifts, studies have documented conflicting findings, with some reporting that nurses working during the day shift perceived an increase in the occurrence of the UNC [37, 43, 55] while others did not [33, 56]. Moreover, working overtime has also been documented to increase the perceived occurrence of UNC [11, 44, 57, 65], except for the study performed by Phelan et al. [75].

Differently, concerning the quality of the work environment, including aspects ranging from better communication, better grading of caring ethical climate or patient safety, studies have consistently documented that a better environment decreases the occurrence of UNC [8, 11, 29, 31, 38, 41, 44‐47, 49, 52‐54, 57, 59, 60, 64, 66, 69, 71, 76, 77, 84]. Effective teamwork [11, 29, 45, 46, 57, 76] as well as a higher score in all dimensions of the Nursing Teamwork Scale (including for example, team leadership, team orientation) [38] were also reported to decrease the perception of UNC. Conversely, communication issues were documented to increase UNC [41, 55, 77]. Moreover, working according to team nurses’ model care delivery as compared to total patient care have been reported to increase the UNC perceptions among nurses [78]. Conflicting findings have emerged instead regarding the type of unit, with Bragadóttir et al. [76] and Coleman [47], documenting that medical and surgical nurses perceived a higher occurrence of UNC compared with those working in other units such as intensive care units. On the other hand, Papastavrou et al. [8] reported that nurses in surgical units perceived low UNC when compared with those working in medical units.

Studies about the influence of the facility/hospital were also performed. Knopp-Sihota et al. [33] and Blackman et al. [55] documented that nurses working in urban and metropolitan hospitals reported a higher occurrence of UNC compared with those working in private hospitals [11]. By contrast, Knopp-Sihota et al. [33] reported that nurses working in a not-for-profit hospital perceived a low occurrence of UNC, similar to that documented in Magnet hospitals [48].

As reported in Table 4, the age of nurses has been investigated for its role in the perception of UNC with conflicting findings. Some have documented that older age is associated with low levels of UNC [77, 78, 80], while others have reported the opposite [33, 43, 58, 75]. Similarly, professional experience has been documented to have a variable influence on UNC, with more clinical experience associated with a higher perceived occurrence of UNC [37, 38, 41, 43, 57] and others documenting the opposite findings [66, 71, 73, 75, 77]. In addition, also the gender has been investigated with conflicting findings, with some studies reporting that female nurses perceived more UNC [43] or less [71, 78] as compared with male nurses.

Regarding nursing education, RNs were reported as perceiving more UNC compared with other roles [38, 43, 57, 58, 76]. However, some authors [34, 40, 84] reported that higher education as attending multiple educational opportunities, prevents the perception of UNC.

In terms of working profiles, Blackman et al. [11] and Chapman et al. [57] have both documented that working part-time decreases the perception of UNC among nurses while working full time remains unclear [68, 77]. Moreover, nurses reporting a higher ratio of absenteeism have been documented to perceive higher levels of UNC [37, 43], except for Kalisch et al. [36] who did not find any relationship.

Almost all studies, except for the one conducted by Knopp-Sihota et al. [33], have reported similar findings regarding nurses’ dissatisfaction, stress, emotional exhaustion, intention to leave, and other elements of poor professional well-being, all of which increase the perceived occurrence of UNC [35, 40, 84]. Moreover, Drach-Zahavy and Srulovici [67] and Srulovici and Drach-Zahavy [68] documented that nurse’s personal degree of accountability influences their perception of UNC, while Blackman et al. [73] analysed the association between UNC and the country of origin of nurses, documenting that Australian nurses perceived more UNC compared with Italian nurses.

Only two studies have investigated antecedents at the patient levels suggesting that clinical instability may play a role regarding the UNC. Caring for more unstable or critical patients, requiring frequent monitoring or units with higher patient death rate [64, 80] were reported as factors increasing the occurrence of UNC (Table 4).

Several studies were conducted in a short period, mainly in the United States (US). Subsequently, the process of missed nursing care concept development firstly reported among US studies, was researched also across the world, especially to Europe [2]. An important impetus for this development has been a project funded by the European Union in 2016, the RANCARE [86], which brought together scholars and practitioners from 34 countries who had worked for four years, giving an international perspective to a relatively unacknowledged nursing problem. However, there are only a few examples of studies conducted at a multi-country level [37, 71‐73], where antecedents might function differently according to variances in national health services, education systems, cultures, and resources devoted to the nursing care. Moreover, studies have been mainly focused on the hospital/acute care setting, suggesting that more research is needed in community and nursing home settings to accumulate evidence in these settings of care.

To date, both antecedents and the occurrence of UNC have been studied mainly from the point of view of nursing staff as a self-assessment, perceptions that can be influenced by several biases. Moreover, some studies have examined perceptions of both nurses and nurse’s aides [40, 43, 48, 77]; these professionals have a different scope of practice, and this may have influenced their assessments. Only two studies involved patients, a perspective that should be considered in future research to better understand the occurrence of UNC also from their point of view [87].

The study designs were largely cross-sectional in nature, with mainly convenience samples and a great variance in the participation rates, that all might have introduced biases in the evaluation of both antecedents and the UNC occurrence. In addition, antecedents and UNC occurrence have been largely measured at the same time point, thus assuming that the former has influenced the latter whereas control variables and/or confounding factors (e.g., the overtime, as paid or not) were not investigated. These issues have been reported also by Griffiths et al. [88] regarding the state of the art of the evidence about the nursing staffing and outcomes.

Longitudinal, pre- and post-study designs, or comparative studies are encouraged to increase the strength of evidence, by quantifying also the benefits of reducing/minimizing unfinished care and the costs, feasibility and long-term sustainability of implemented interventions. However, study designs should be considered in light of the complexity of the nursing care and the issue under study: UNC occurs in the real world of nursing across the world as a multifactorial phenomenon. Assessing precise antecedents might be difficult—moreover, designing interventional studies manipulating for example, the work environment, or the number of staff might be not feasible given the complexity of the turbulence of environments, and the challenges of the long-term implementation. Therefore, an in-depth discussion regarding the research issues in this field is required, analogously to that already developed in the context of nursing staffing and outcomes [88].

Conceptual articles have highlighted that UNC is influenced by patient care demand, resource allocation, and relationship/communication issues [14] as well as by patient, organisational, nursing work environment, philosophy of care, and nurse variables [3]. In recent years, there has been a more comprehensive consideration of macro-, meso-, and micro-level factors by examining how upper-level management might affect clinical nurses and, consequently, UNC at the bedside [16]. However, according to the findings of this review, primary studies available to date appear to have investigated antecedents only at the unit, nurse, and patient levels. Therefore, despite a clear conceptualisation of the importance of the factors at the system level [16], empirical studies seem to have captured only a limited extent of factors with heterogeneous findings.

At the unit level, the staff adequacy as measured with different methods (e.g., workloads [40, 41, 44, 55, 68, 80] versus nurse-to-patient ratio [43, 51, 59, 65, 67, 69‐71, 81, 82], using subjective or objective data) influences the occurrence of UNC. Moreover, other processes such as patients’ admissions and discharges or caring for patients with complex needs increase the occurrence of UNC, likely because they affect workloads in an unpredictable manner that requires a revision of staff dynamics and resource assignments [89]. On the other hand, performing non-nursing tasks [69, 80, 84] were documented to increase UNC as well as working overtime [11, 44, 57, 65]. Unfinished care might be triggered by the underuse of nurses, constrained to compensate for deficiencies in auxiliary resources thus leaving nursing care undone; conversely, unfinished care might be the consequence of the tiredness and reduced performance of nurses due to the amount of overtime work. A clear direction has not emerged regarding shifts (e.g., morning versus nights) and this might be due to the different patterns of both shifts (e.g., 12 h) and workloads established at the unit level. Specifically, those working morning shifts are required to deal with the high number of concentrated activities, while those working night shifts have few resources to meet care needs.

Within the unit level, the findings mainly reflect the structural variables [90] of the unit, with modifiable factors that might reduce or minimise UNC. These factors include adequate staff levels, preventing nurses from performing non-nursing tasks and working overtime, and implementing strategies to deal with the unpredictability of workloads for some shifts. The findings support the conceptualisation of Jones et al. [16] that factors affecting the occurrence of UNC can be considered in light of micro-economic theories as the efficient allocation of scarce resources to nursing care.

Several studies [8, 11, 29, 31, 38, 41, 44‐47, 49, 52‐54, 57, 59, 60, 62, 64, 66, 69, 71, 76, 77, 84] have concluded that a better work environment leads to a decrease in the UNC. Hence, promoting greater communication, better caring ethical climate, and respect among nurses and across health care professionals, all reduce or minimise the UNC. These factors, mainly reflecting the process variables of the unit [90], suggest that there is a need to invest in good practice environments for nurses, a strategy that can be developed by nurse managers but requires profound support from the entire system and education to work together effectively. Indeed, the findings that emerged regarding Magnet hospitals [48] and some hospitals/units (rural versus urban [11, 33], surgical versus medical [8, 47, 76]) can explain their capacity to minimise or reduce the UNC as work environments where nursing care is supported and valued.

Studies investigating the relationships between some individual characteristics of nurses (e.g., age, gender, and work experience) and the occurrence of UNC have mainly reported conflicting findings. Some authors also included variables that are not usually measured, such as the nurse’s personality and the country of origin [67, 73] and no trends in this dimension were detected. The interest in individual variables seems to be linked with the fact that the UNC has been investigated mainly as nurses’ perceptions; therefore, it is influenced by the profile of the nurse. However, apart from some antecedents (e.g., education), most of them appear to be unmodifiable, thus suggesting that they should be considered by nurse managers while, for example, they compose shifts that mix different nurse profiles (e.g., age, gender, education). Conceptually, authors have emphasised that nurses’ experience [36‐38, 41, 43, 57, 66, 71, 73, 75, 77], education [11, 32, 34, 38, 40, 43, 51, 56‐58, 76, 84], and skill mix [41] may influence the quality of the decision-making processes and, ultimately, the occurrence of UNC. However, when nurses are called to make decisions on how to allocate the limited time available, they desire to provide the best care for their patients and eliminate unfinished care. They also need to be supported in making decisions though a positive ethical climate and organisational guidance [62]. Unfortunately, the mental processes involved in decision-making regarding care that can be left undone has been unexplored and more studies are needed to increase understanding of how nurses set priorities while they are trying to cope with the endless needs of patients in the complex environments of the contemporary care facilities [91]. This reflection might also explain why nurses perceive more UNC [38, 43] compared with nurse’s aides suggesting that in studies investigating the unfinished care perceived, a stratification of the responders according to their educational level, is required. Additionally, some of the nurse variables that have been investigated seem to play a dual role as antecedents and as consequences of UNC. For example, decreased professional satisfaction levels [33, 35, 40, 54, 84, 85] might lead to increase the unfinished care but also might be a consequence of the UNC, as reported in conceptual models [15].

Patient characteristics as antecedents of UNC have been poorly investigated. According to the available studies, clinical instability [64, 80] can increase the need for nursing care due to the additional care patients required in context with limited resources. Moreover, only recently the patients’ perceptions about UNC have been summarised [63]: authors concluded that many of the unmet needs perceived by patients do not always correspond to the perceptions of nurses. It is therefore vital to understand the UNC phenomenon from the patients’ point of view with more research; this endeavour will allow researchers to design appropriate interventions. For example, complex clinical cases might require more care with unpredictable flows that can be provided with flexible processes and models of care delivery.

This systematic review has several limitations. Despite the rigorous approach, some studies might be missed for different reasons and among other, the fact that no quantitative measures (Cohen’s kappa coefficient) were used to evaluate the agreement across researchers regarding the study inclusion. Moreover, according to the limitations applied in the languages (English, Greek, Dutch, and Italian) a potential publication bias might have been introduced. Furthermore, the timeframe was limited by including only studies published after 2004, the year when the first concept pertaining to the UNC was established [4, 21]. However, studies using different key words might have been performed before the establishment of the mentioned conceptual definition; other studies might have been conducted after without using the conceptual definition, leading in both cases to a publication bias. In addition, to map antecedents, the search terms identified were general and designed to capture all studies in the field and not those addressing specific (known) antecedents of the UNC (for example shift patterns) [20]. Finally, there were excluded those studies conducted in specific settings (paediatric, psychiatric) according to the available knowledge [21]: the inclusion of these settings in future reviews might contribute to broaden the evidence available.

In performing the data extraction, some antecedents might have been neglected in favour of providing a comprehensive map of those investigated to date. Commonalities across antecedents by using an inductive approach [25] were searched to summarise the data: although an investigator triangulation was performed [92], researchers might have been influenced in the data analysis process by their previous background and experience regarding the issue. Furthermore, the relationship between antecedents and the UNC as decreasing, increasing or not influencing its occurrence, has not been weighted in its evidence according to the study design or for example the effect size, suggesting an area of improvement in future reviews.