Background Matching demand and supply in nursing work continues to generate debate. Current approaches focus on objective measures, such as nurses per occupied bed or patient classification. However, staff numbers do not tell the whole staffing story. The subjective measure of nurses’ perceived adequacy of staffing (PAS) has the potential to enhance nurse staffing methods in a way that goes beyond traditional workload measurement or workforce planning methods.
Objectives To detect outcomes associated with nurses’ PAS and the factors that influence PAS and to review the psychometric properties of instruments used to measure PAS in a hospital setting.
Design and methods A scoping review was performed to identify outcomes associated with PAS, factors influencing PAS and instruments measuring PAS. A search of PubMed, Cumulative Index to Nursing and Allied Health Literature (CINAHL), Business Source Complete and Embase databases identified 2609 potentially relevant articles. Data were independently extracted, analysed and synthesised. The quality of studies describing influencing factors or outcomes of PAS and psychometric properties of instruments measuring PAS were assessed following the National Institute for Health and Care Excellence quality appraisal checklist and the COnsensus-based Standards for the selection of health Measurement INstruments guidelines.
Results Sixty-three studies were included, describing 60 outcomes of PAS, 79 factors influencing PAS and 21 instruments measuring PAS. In general, positive PAS was related to positive outcomes for the patient, nurse and organisation, supporting the relevance of PAS as a staffing measure. We identified a variety of factors that influence PAS, including demand for care, nurse supply and organisation of care delivery. Associations between these factors and PAS were inconsistent. The quality of studies investigating the development and evaluation of instruments measuring PAS was moderate.
Conclusions Measuring the PAS may enhance nurse staffing methods in a hospital setting. Further work is needed to refine and psychometrically evaluate instruments for measuring PAS.
- organisation of health services
- quality in health care
- human resource management
- organisational development
- health economics
Data availability statement
Data are available in a public, open access repository.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
- organisation of health services
- quality in health care
- human resource management
- organisational development
- health economics
Strengths and limitations of this study
This scoping review is the first to assess (1) the relationship between nurses’ expert opinion of staffing adequacy and outcomes, (2) factors influencing nurses’ perceived adequacy of staffing, and (3) the reliability and validity of instruments measuring perceived adequacy of staffing.
The literature search was extensive, and designed and conducted with the help of a clinical librarian.
Study selection, data extraction and quality appraisal of included studies and instruments were performed by two researchers.
Limitations of this review include the potential that we have missed original literature on influencing factors or outcomes, because we excluded grey literature and qualitative studies.
Since the early 1970s, both researchers and practitioners have been searching for the best way to match demand for nursing work with nursing supply. Societal developments have made adequate staffing more relevant today than ever. Driven by an ageing population and technological progress, demand for care is rising. At the same time, the WHO expects a worldwide shortage of over 7 million nurses and midwives by 2030,1 putting continued pressure on staff. Previous research has indicated an association between nurse staffing levels and nurse-sensitive outcomes such as mortality, adverse events, fall rates, failure-to-rescue and missed care.2–4 Inadequate staffing is also related to burn-out and job dissatisfaction among nurses.5 Not only quantity but also quality in terms of skill mix matters; a higher proportion of registered nurses (RNs) is associated with better outcomes.6 7 Inadequate staffing ultimately threatens safety, quality, affordability and accessibility of care. Therefore, a thorough understanding of staffing adequacy is needed.
The concept of adequacy of staffing can be divided into ‘staffing’ and ‘adequacy’. ‘Staffing’ has been defined in multiple studies. Jelinek and Kavois8 defined nurse staffing as the process of determining the appropriate number and mix of nursing resources necessary to meet workload demand for nursing care at the unit or departmental level. Burke et al9 described hospital staffing as determining the number of personnel with the required skills to meet predicted requirements. Both of these definitions include balancing demand for nursing work with the adequate number and skill mix of nurses. Adding the word ‘adequacy’ to the concept of staffing, the meaning shifts from the process of staffing to a condition in which staffing is adequate. The American Nurse Association defined staffing adequacy as a match between RN expertise and recipient needs within the practice setting,10 but details on what this match entails were omitted. Kramer and Schmalenberg11 asked nurses if their staffing was adequate and received ambiguous answers: ‘That depends – adequate for what? Safe care to all patients? (…) Quality care? (…) Or comprehensive care?’ (p.194).
In the absence of an explicit clarification of what adequate staffing means,12 nurses and managers continue to search for staffing measures that can objectify staffing requirements.13 These measures need to facilitate different inter-related staffing decisions, for example, how many nurses to employ, staff-shift schedule, nurse roster and nurse-ward allocation.14 Many workload and resource planning tools are available related to demand for nursing work, resource planning and workload evaluation.
Demand for nursing work
Demand for nursing work has been estimated by a volume-based approach, that is, patient counts multiplied by an administrative measure of work. This has been expressed as the nursing hours per patient day (HPPD),15 nurse-to-patient ratios2 and full-time equivalent numbers.4 These have been criticised as measures for staffing decisions because different patient needs are ignored.16 The workload-based approach takes different patient care requirements into account and is categorised into activity-based and dependency-based methods.17 The activity-based method is based on how long nursing tasks take and the dependency-based method relies on patient classification of patients’ needs based on indicators, based on which the amount of nursing time can be derived. Disadvantages of the workload-based approach include lack of reliability, validity and flexibility, and the need for time-consuming manual registration.17–19
Resource planning tools
Other resource planning tools indirectly measure adequacy of staffing by quantifying demand and supply. One example is the RAFAELA patient classification system.20 It estimates optimum levels of nursing intensity by balancing demand for care with nursing resources available. The tool is used on a large scale in Finland, but preimplementation in the Netherlands encountered issues of validity and acceptability.21
Workload evaluation tools
Other workload tools evaluate nurses’ workload. Tools to evaluate workload can be objective indirect measures of mental workload, such as brain activity and cardiac responses, or subjective tools such as the NASA Task Load Index and the Subjective Workload Assessment Technique.22 These subjective instruments involve short questionnaires with items that reflect experiences (eg, mental demand, physical demand, temporal demand). Those type of measures are commonly used to evaluate workload or validate measures of staffing requirements,13 reflecting on a broader definition than adequacy of staffing.
In 2010, Fasoli and Haddock18 reported reliability and validity issues with the available workload measurement systems. Nine years later, another review13 concluded that available systems were still highly uninformative. Scientists dispute whether nursing work can be accurately quantified. Hughes23 states that ‘it appears that nursing is more concerned with knowledge processing and nurses’ intentions than just with the activities of caring’ (p.317). Griffiths et al13 describe that ‘there is a limit to what can be achieved through measurement, both because of the fallible nature of the measures, but also because of the complex judgements that are required’ (p.9). In the absence of applicable tools, professional judgement was identified as the nearest to a gold standard workload measurement.13
The match between nurse demand and supply can be measured using the nurses’ perceived adequacy of staffing (PAS). This measure relies on nurses’ expert opinion in which nurses take the unquantifiable fluctuating patient needs and context and situation into account in assessing adequacy of staffing.24 This direct approach to measuring adequacy of staffing contrasts traditional tools that measure staffing adequacy according to demand and supply. Nurses’ perceptions have been accepted as a significant indicator of quality of care,2 while nurse-perceived quality of care was highly associated with objectively measured nurse-sensitive outcomes, showing the validity of the measure.25 Regarding nurse staffing tools, relying on nurses’ perceptions is less common as most approaches attempt to objectify staffing needs.13 However, a reliable and valid measure of PAS may be the optimal approach to helping head nurses and managers make nurse staffing decisions. A positive association of PAS with outcomes for patient, staff and organisation enables evidence-based staffing decision making. Staffing adequacy can potentially be predicted by associating structure and process factors of PAS. Data science techniques may minimise nurse effort by analysing these factors in hospital information systems. However, these techniques have not been explored in nurse staffing literature.26 27
The concept of PAS potentially enhances nurse staffing methods, going beyond traditional workload measurement or workforce planning tools.
To explore this alternative to objective workload measurement tools, we conducted a scoping review to study the potential relevance of nurses’ PAS in the setting of hospital wards. We asked the following research questions:
How is PAS associated with outcomes for the patient, nurse and organisation?
Which factors influence PAS?
If these findings show PAS to be a potentially relevant measure for a new staffing method, we will go on to answer the following research questions:
Which PAS measurement instruments are available in the literature?
What is the reliability and validity of those instruments?
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses—Extension for Scoping Reviews checklist and guidelines to ensure our review was robust and replicable.28 We did not publish a protocol for this review.
PubMed, CINAHL, Business Source Complete (through EBSCOhost) and Embase were searched from inception to November 2019. The following free-text and database subject headings were combined to search for peer-reviewed articles: nursing staff, nurses, nurse, staffing adequacy, inadequate staffing, staffing inadequacy, adequate staffing, requirements for nursing resources, attitude of health personnel, perception and perceive, and truncation symbols, for example, nurs*, were used if suitable. Additionally, we screened reference lists of included studies and reviews on nurse staffing for other relevant studies. No limits regarding publication status, date or language were imposed. The complete search strategy for each database is presented in online supplemental appendix 1. The search was designed and conducted with the help of a clinical librarian.
References from the databases were combined and downloaded into a reference manager, and duplicates were removed. Articles were screened in two phases. First, two reviewers (CM and CO) independently screened all titles and abstracts and selected articles that met the inclusion criteria (table 1). For the measurement instruments that were applied, the primary development and evaluation study was included. The screening resulted in a Cohen’s κ of 0.80. Disagreements about inclusion of studies between the two reviewers (CM and CO) were resolved by discussion. Next, full-text versions were independently screened by the two reviewers and excluded if articles did not meet the inclusion criteria (table 1). Authors were contacted for irretrievable articles.
Data were independently extracted by two reviewers (CM and CO) using a predefined, structured data abstraction form. The form included the author, year of publication, country, journal, aim, research design, population, test setting, sample size, staffing measures, instruments (including subscales), measurement type, validity, reliability, associations between PAS and outcomes, and associations between influencing factors and PAS. Full details of associations were documented and expressed as correlation coefficients (r), β-coefficients (β) derived from linear regression analysis or ORs derived from logistic regression analysis, including their p values and 95% CIs. We also documented whether the associations were corrected for other factors by multivariate analysis.
Quality of the study outcomes associated with PAS and the factors influencing PAS were evaluated according to the National Institute for Health and Care Excellence quality appraisal checklist for quantitative studies reporting correlations and associations,29 adapted from Griffiths et al.3 The checklist assesses bias across four categories—population, confounding factors, measures and analyses—using five response options (++, +, -, not reported, not applicable). The resulting score indicates whether the external validity (ie, the generalisability) and the internal validity (ie, the validity of the associations) are strong, moderate or weak.
The methodological quality of the included PAS instruments was appraised using the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) Risk of Bias checklist.30 31 This checklist, which has been developed to assess the methodological quality of patient-reported outcome measure studies, is suitable for assessing the risk of bias of PAS instruments. Instrument development, structural validity, internal consistency and other measurement properties in the included studies were assessed. Quality was judged as very good, adequate, doubtful or inadequate, and the overall quality was the lowest item rating in the COSMIN boxes.31 Measurement properties were rated sufficient (+), insufficient (-) or indeterminate (?) following the criteria for good measurement properties.31
Quality was appraised by one reviewer (CM) and crosschecked by a second reviewer (CO). Disagreements between reviewers were solved by consensus.
Outcomes for each research question were summarised. With regard to the influencing factors and outcome studies, variables analysed by t-tests, (multivariate) analysis of variance ((M)ANOVA), χ2, correlation or regression were judged significant if the value of p was <0.05 or their CI did not enclose the value of 0 or 1. We judged the structural validity and internal consistency of measurement instruments based on the original development study.
Data for outcomes/influencing factors and measurement instruments were structured separately. The structure-process-outcome model32 was used to structure the influencing factors and outcomes. Influencing factors are factors related to (1) Structure, that is, the physical and organisational context of care delivery, and (2) Process, that is, the technical and interpersonal process of care delivery. Outcomes reflect the impact of those factors demonstrating the result of structure and process. Following the patient care delivery model,33 the influencing factors and outcomes of PAS were clustered into patient, staff and organisation categories. Models including PAS as a dependent variable are described separately.
Both single-item and multi-item measurement instruments were included.
Patient and public involvement
No patient was involved.
Study selection and characteristics
The search identified 3120 studies. After removing duplicates and screening titles and abstracts, 135 eligible studies were included for full-text review, including 6 studies that were identified in the reference lists of included studies. Full-text review excluded a further 59 studies. The main reasons for exclusion were no instrument development or associations with influencing factors or outcomes (24/59), no measurement of PAS (10/59) and staffing measures that were not PAS (8/59). For 13 studies, the full text was not available and the authors did not respond to our request for the full text. In total, 63 studies were included in the analysis (figure 1).
The included studies (tables 2 and 3) were published between 1975 and 2019 worldwide. Most studies (28/63) were carried out in North America,11 24 34–59 25 studies were conducted in Europe,60–84 5 in Asia,85–89 4 in Oceania90–93 and 1 in multiple continents.94
Fifty-two studies included outcomes influenced by PAS or factors that influence PAS.24 35 37 39 40 42–47 49 52–54 56–60 62 63 65–94 Twenty-one studies described the development and evaluation of PAS instruments.11 34 36 38 41 43 44 46 48 50 51 54–56 58 61 64 82 86 87 91 Forty-nine studies used a cross-sectional research design,24 35 37 39 40 42–47 52–54 56 57 59 60 62 63 65–76 78–94 two studies used a longitudinal research design49 77 and one study used a cross-sectional and longitudinal design.58 Complete extracted outcomes and influencing factors are provided in online supplemental appendix 2.
Quality assessment of studies investigating influencing factors and outcomes
The methodological quality of most studies was moderate to good (table 4). We revealed serious methodological flaws (weak internal and external validity) in six studies. The risk of bias was increased by cross-sectional research designs, omitting confounding factors, and the lack of multilevel studies and objective measures. External validity was weak because the source population was not clearly described and because of the use of single sites. An overview of the compete quality appraisal is presented in online supplemental appendix 3.
Outcomes influenced by PAS
Our first research question was to explore the associations between PAS and outcomes for the patient, nurse and organisation. Sixty outcomes were found to be influenced by PAS—27 of these were patient-related, 26 were nurse-related and 7 were organisation-related (table 2). Job satisfaction was investigated in nine studies,39 46 47 52 66 72 75 78 86 quality of care in eight studies,35 47 66 72 75 85 86 94 safety in four studies,71 73 75 77 and missed care,40 62 87 emotional exhaustion,66 68 75 and occupation dissatisfaction39 52 75 in three studies. Forty-nine outcomes were investigated in two or fewer studies. Most outcomes were positively associated with PAS.
Associations with PAS were found for the patient outcomes pain,84 pressure ulcers24 and patient-centred care.60 Williams and Murphy44 asked nurses to rate 10 aspects of care, (including basic hygiene, feeding and medication) from poor to good in six units. Scores for each category were generally higher when staffing was adequate, but results were inconsistent within individual units. Patient safety associated positively with PAS in all studies71 73 77 except for one,75 which reported mixed results. Associations with PAS were also mixed for adverse events,87 infections,49 74 survival,73 patients' ability to manage care after discharge,76 communication with nurses44 87 and missed care.40 45 62 70 87 Cho et al87 found that missed communication and basic care mediate the association between patient-perceived staffing and adverse events and communication with nurses.
PAS had a personal effect on nurses. It affected job satisfaction,39 46 47 52 66 72 75 78 86 burn-out,78 86 effort-reward imbalance,67 depersonalisation, personal accomplishment,68 feelings of being a safe practitioner and workplace cognitive failure,77 psychosocial attention,75 and change efficacy.81 The reported effects of satisfaction with the occupation,39 52 75 intention to leave the occupation,76 intention to leave employment,80 86 89 94 emotional exhaustion,66 68 75 depressive symptoms,67 pain,53 blood pressure and total cholesterol level82 were inconsistent. Pain in the neck, shoulder, arm, lower extremities and musculoskeletal system53 as well as low-density lipoprotein cholesterol levels82 and change commitment81 were not influenced by PAS.
PAS affected organisational outcomes, including nurses’ turnover,42 47 absenteeism,45 quality of nursing73 and quality improved within the last year.75 Mixed results were reported for quality of care.35 47 66 72 75 85 86 94 Patients’ hospital rating was associated with patient-perceived staffing adequacy but not with nurse-perceived staffing adequacy.87 Anzai et al85 found no association between PAS and nurses’ ability to provide quality nursing care.
Influencing factors of PAS
For the second research question, we identified the structural and process factors that influence PAS.
Fifty-two structural factors that influence PAS were identified. These were categorised into demand for care (11 factors), nurse supply (30 factors) and organisation of care delivery (11 factors). The setting type was investigated in seven studies44 47 75 83 84 91 92 and patients-per-nurse in three studies.24 59 87 The remaining 50 factors were investigated in two or fewer studies. Associations were mainly positive, that is, higher scores on structural factors led to more positive PAS.
With regard to demand for care, no consistent results were found for factors associated with PAS. Inconsistent results were found for census,43 44 number of maximum care patients43 and patient classification category.43 58 69 New admissions, transfers, discharges, postoperative patients, specialised nursing procedures43 and crowding scores in the emergency department54 were not related to PAS.
Nurse supply factors influencing PAS were full-time equivalent RNs per patient day,58 HPPD,24 nursing hours,43 patients-per-nurse,24 59 86 (RN) skill mix,24 58 educational level,83 assistive personnel,59 causal/relief staff,90 mental stress,69 90 nurses’ psychological capital46 and life orientation.47 Mixed results were reported for staff hours available,44 presence of students,69 90 nursing role,67 85 gender,75 85 work experience75 83 90 and nurses’ work capacity.69 90 Nursing HPPD, non-RN HPPD,24 59 temporary nursing-care HPPD,49 age75 83 and part-time nurses75 were not related to PAS. Louch et al77 found that levels of agreeableness and conscientiousness moderated the association between PAS and whether nurses feel they can act as a safe practitioner, and that emotional stability moderated the association between PAS and patient safety.
Organisation of care delivery factors unit size, number of beds and number of high-technology hospital services58 affect PAS. Spence et al90 reported that organisation of the clinical manager’s work and the shift schedules was the most important of nine factors that increase workload. In contrast, Rauhala and Fagerström69 found no relationship between managerial planning, work organisation, work rota planning and Professional Assessment of Optimal Nursing Care Intensity Level (PAONCIL) Scores. Mixed results were found for the setting,44 47 75 83 84 91 92 case mix index,58 59 and meetings and training during shifts.69 90 Substitute resources did not correlate with PAONCIL Scores.69
Twenty-seven process factors were investigated in relation to PAS. Most process factors were positively associated with PAS, that is, higher process factor values were related to more positive PAS.
Teamwork was investigated in three studies, and other factors were examined in two or fewer studies. Ward morale,85 error reporting culture, governance, nurse participation in hospital affairs, nurse manager ability, leadership and support, foundations for quality nursing care,88 trust, shared mental models, team leadership, backup,37 79 structural empowerment,46 nurses’ feeling of respect,56 organisational and professional commitment, professional practice climate,47 and unexpected rise in patient volume or acuity,59 all influenced PAS. An increase in positive patient perceptions of staffing was related to an increase in positive perceptions of nurse staffing.87 Intraprofessional and interprofessional cooperation69 88 90 and teamwork37 57 79 showed inconsistent associations with PAS. The perceived influence of nurse leaders was associated with PAS in four out of six leadership domains.35 PAS was not associated with role support.93
Three studies explained PAS using regression models. Kalisch et al59 reported four different models with variables HPPD, case mix index, nursing education, unexpected rise in patient volume and acuity, and inadequate number of assistive personnel. The model including all variables explained most variance in PAS (33.8%). Mark et al58 studied three models explaining between 33% and 51% of the variance in PAS. Patient technology, number of beds, growing admissions, and case mix index were relevant in all three models. Rauhala and Fagerström69 built models for 22 wards including patient classification and non-patient questions as independent variables. The median variance explained by patient factors alone was 45%. Adding non-patient factors increased the median variance to 55%, indicating that patient factors contributed to PAS more strongly than non-patient factors did.
Measurement instruments of PAS
The third research question investigated instruments used to measure the PAS. We found 21 studies that described PAS measurement instruments (table 3),11 34 36 38 41 43 44 46 48 50 51 54–56 58 61 64 82 86 87 91 91 20 of which were found in the development studies. Most instruments were developed in the last two decades, except for two that were developed in the 1970s.43 44 Most instruments (12/19) were developed in the USA.11 34 36 38 41 43 44 48 51 54 55 58
The measurement aim, items and response options of the different instruments varied considerably. Instruments with a direct practical purpose of balancing nurse demand and supply were the head nurse questionnaire,43 PAONCIL,64 assessment of real-time demand for the emergency department54 and the unit staffing/care evaluation form.44 These instruments are used on a daily basis.
PAS is measured in the different questionnaires by single items,36 38 44 46 48 54 56 58 64 82 86 87 multiple items43 91 and multi-item subscales to evaluate safety culture55 and nursing work environment.11 34 41 50 51 61 Some items assess the adequacy of staffing numbers (eg, ‘Enough staff to get the work done’),36 41 43 46 51 55 61 82 86 87 91 and some assess the skill mix (eg, ‘Enough registered nurses on staff to provide quality patient care’).41 43 50 51 61 91 Some instruments attempt to specify the purpose of adequate staffing (eg, adequate ‘for quality care’,11 41 51 56 61 86 87 ‘to handle the workload’,36 55 ‘to meet your patient/clients' needs’,46 91 ‘to get the work done’41 51 61 and ‘to maintain patient safety’50) while other instruments just measure adequacy of staffing without specifying what this entails.38 44 48 58 82
The target respondents of all instruments are nurses in general, head nurses,43 critical care nurses,50 51 charge nurses44 or new graduates.46 One study asked both nurses and patients to assess PAS.87 Most instruments used a 4-point or 5-point Likert Scale.11 34 36 38 41 43 44 46 48 50 51 55 56 58 61 82 86 87 91 Real-time demand for the emergency department54 was assessed using a dichotomous scale: exceed or not exceed. The PAONCIL includes a 7-point scale, and estimates can be made with an accuracy of 0.25 points.64
Reliability and validity
The fourth research question assessed the reliability and validity of PAS measurement instruments. We found methodological flaws in most studies. With regard to the single-item instruments, construct validity of PAONCIL was tested by hypothesising a correlation between PAONCIL scores and patient classification scores.64 No other studies of single-item or multi-item measures reported reliability or validity testing. The Nursing Work Index - Revised development study did not use a staffing subscale,34 so we could not assess psychometric properties. For the remaining six subscales,11 41 50 51 55 61 the methodological quality of structural validity and internal consistency were adequate, except for structural validity of the American Association of Critical-Care Nurses Healthy Work Environment. However, while internal consistency was sufficient in most studies, structural validity was sufficient in only one study.
Our scoping review found that mostly positive perceptions of staffing adequacy (measured using the PAS) are related to positive outcomes for patient, nurse and organisation, confirming the importance of the measure. We identified many factors that influence PAS, but the associations were inconsistent. Twenty-one instruments were identified that measure PAS, and these different instruments had different measurement aims.
Most studies reported that positive perceptions of staffing adequacy are related to positive outcomes for the patient, nurse and organisation. Effects on patient outcomes were inconsistent, mainly because of severe methodological flaws in one study.44 The positive relationship between staffing and outcomes was confirmed by different staffing measures, such as nurse-to-patient ratios.13 95 However, studies explained more of the variation in patient outcomes of PAS than staffing measures such as nurse-to-patient ratios and HPPD,24 60 indicating the informative value. Kalisch et al59 found moderate correlations between nurse-reported staffing adequacy, nurse-to-patient ratios and nursing HPPD, clarifying that these measures ‘may capture different elements of the unit context to explain nurse staffing’ (p.775). It seems that adequate staffing depends on more than just staff numbers and skill mix elements, and that nurses take these additional factors into account when assessing PAS.24 96 In agreement with this, we identified many factors that influence PAS in the present study, including demand for care, nurse staffing, and organisation and process factors. Whether outcomes are improved by objective measurement of workload on a daily basis is unclear.12 The RAFAELA system has provided some evidence that patient safety and mortality are associated with workload level.97 Our finding that measuring the PAS is associated with positive outcomes indicates that measuring the PAS will strengthen nurse staffing tools, which will in turn improve staffing decisions. Measuring the PAS was also found to be relevant in research areas other than nurse staffing. For example, PAS was one of the eight essential factors of magnetism. Magnetism refers to elements that are essential for a work environment that can attract and retain nurses while providing a high level of job satisfaction and quality of care.98
We identified a variety of factors that influence PAS, but were unable to define a valid set of factors that were relevant to nurse staffing. Most factors were investigated in one study and results were inconsistent between studies. There appear to be many factors affecting PAS, including patient-related and nurse-related factors and how care delivery is organised. Factors related to the work environment were also important, such as cooperation, leadership and teamwork. This is in agreement with other studies of factors that influence demand for care.99–102 Hence, patient, nurse and organisation factors were recommended to consider in a staffing model.101 Nurses have disputed traditional instruments for measuring workload because they involve time-consuming manual registration and cannot forecast staffing adequacy.17–19 96 100 103 Including influencing factors in a staffing model can solve these issues, enabling decision makers to align nursing resources in a timely fashion. The study by Trivedi and Warner104 was one of the first attempts to predict staffing adequacy using data. They designed a multivariate regression model that predicted head nurse perceptions of staffing adequacy and used this model to allocate float nurses at the beginning of the shift. Nowadays, more advanced techniques are available. Machine learning and artificial intelligence can be used to analyse hospital data and potentially explain and forecast PAS, supporting staffing decisions. These methods are a prerequisite for reliable and valid measurement of PAS.
Most of the PAS measurement instruments we found were single items, and they did not include psychometric testing. However, multiple psychometric tests can be performed on single items, including tests for content validity, inter-rater variability and responsiveness.105 Although a single item is suitable in some situations,106 multiple items are more reliable. Multiple items should be used for complex constructs as they define the meaning of the construct for the rater.105 Kramer and Schmalenberg found that multiple items are needed to measure PAS.107 However, the downside of administrative burdens have been shown to inhibit successful implementation.21 Most relevant shortcomings of multiple-item instruments of PAS are a lack of information on subscale development, omitting to fully determine structural validity by confirmative factor analysis and confirm other psychometric properties such as reliability, criterion validity, hypothesis testing, measurement error and responsiveness.
Overall, development and evaluation of PAS instruments has been moderate; this reflects the varying use of the measure. There is no established definition of staffing adequacy. Most instruments reflect the adequacy of staff numbers, and some include skill mix (which is becoming increasingly relevant).3 108 In addition, the measurement aims differ between instruments. For some measurements such as safety55 and work environment,34 41 it is sufficient to grade adequacy of staffing, while for nurse staffing decision making understaffing or overstaffing need to be graded. Moreover, instruments measure PAS by referring to the adequacy of full-time equivalent numbers11 or team composition.41 This tactical/strategic decision level of staffing differs from instruments on operational decision levels of capacity management, where decisions involve the staff schedule of a specific shift. Just as for workload measurement tools,12 the decisions supported by the PAS instrument are mostly unspecified. As a result, there are a variety of available instruments, so practical use of PAS in the nurse staffing process is still limited. Decision makers continue to search for objective staffing measures and rely only moderately on nurses’ opinions, so there is still a significant gap between managers and nurses in daily operations.
Strengths and limitations
The strengths of our review includes that our review was set up systematically and assessed the quality of included studies, something which is not mandatory for a scoping review.109 But, there are some limitations to our study. First, we were unable to assess the full text of some studies (0.5%) because of no access and failing requests to researchers. However, because of the small amount of inaccessible studies we consider these studies of minimum impact on our results and conclusions. Second, we searched for studies that developed and validated PAS instruments, which could have affected our results as other publications discussing psychometric properties of included instruments were not included. Finally, we excluded qualitative studies and grey literature, which may have included potential influencing factors or outcomes. Because these studies are often followed up by quantitative studies to determine influencing factors,102 it is likely that these factors and outcomes already are included in the quantitative studies included in this review. Nevertheless, in future research qualitative data should be explored as an extension of the results reported in this review.
Adequate staffing is essential for the patient, nurse and organisation.110 In an ideal situation, PAS would be evaluated daily on the hospital ward to identify inadequate staffing either at the beginning of a shift or in upcoming shifts. Using existing patient and nurse data avoids additional administrative work and incorporating nurses’ judgement potentially generates valid and reliable information acceptable to nursing staff. Measuring PAS in this way is in accordance with existing design principles.101 The information is input for a mutual dialogue and decision making on a team, ward or cross-departmental level. Nursing managers should recognise that staff numbers do not tell the whole staffing story and avoid investing in traditional patient classification systems. Machine learning and artificial intelligence will provide new opportunities for measuring adequacy of staffing in the near future. For adequate and practical measurement of PAS, a balance should be found between using multiple items for reliability and limiting the effort needed to use them. For this to work, practitioners need to be involved in developing adequate PAS measures.
This scoping review found that PAS is positively associated with outcomes for patient, nurse and organisation, supporting the relevance of PAS as a measure for nurse staffing decisions. Many factors were identified that influence PAS, but associations were inconsistent. Instruments used to measure PAS were found to have moderate reliability and validity. Measuring PAS could enhance nurse staffing methods by predicting staffing adequacy based on existing patient and nurse data using machine learning and artificial intelligence techniques. This approach goes beyond traditional workload measurement or workforce planning methods. Further work is needed to refine and psychometrically evaluate instruments measuring PAS.
Data availability statement
Data are available in a public, open access repository.
The authors thank On Ying Chan, information specialist, Radboud University Medical Center, the Netherlands, for her help with developing and refining the search strategy.
Contributors All authors (CM, HV, PH and CO) were involved in planning the scoping review. CM and CO were involved in the search strategy, extraction, quality appraisal and synthesis of data, and wrote the first draft of the manuscript. All authors revised the manuscript drafts and approved the final manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.