nursing shortage literature review

• Search Menu
• Accident and Trauma
• Anaesthesia
• Cardiothoracic Surgery
• Cardiovascular Disease
• Child and Adolescent Psychiatry
• Critical Care/Intensive Care/Emergency Medicine
• Dermatology
• Endocrinology
• Environment and Disease
• Gastroenterology
• General Practice
• Geriatric Medicine
• Haematology
• Health Policy
• Health Economics
• Infectious Diseases
• Liver Disease
• Neonate Medicine
• Neurophysiology
• Neurosurgery
• Obstetrics and Gynaecology
• Occupational Medicine
• Ophthalmology
• Otolaryngology
• Paediatric Medicine & Surgery
• Palliative Medicine
• Perioperative Medicine
• Public Health Medicine
• Renal Medicine
• Respiratory Medicine
• Rheumatology
• Sports Medicine
• Transplantation
• Tropical Medicines
• Advance articles
• Editor's Choice
• Author Guidelines
• Submission Site
• Open Access
• About British Medical Bulletin
• Editorial Board
• Advertising and Corporate Services
• Journals Career Network
• Self-Archiving Policy
• Dispatch Dates
• Journals on Oxford Academic
• Books on Oxford Academic

Article Contents

Introduction, the current nursing workforce—what do we know, how are shortages calculated and why do shortages of nurses arise, how can nurse shortages be reduced, gap analysis, concluding comment, acknowledgements, conflict of interest statement.

• < Previous

Global nurse shortages—the facts, the impact and action for change

• Article contents
• Figures & tables
• Supplementary Data

Vari M Drennan, Fiona Ross, Global nurse shortages—the facts, the impact and action for change, British Medical Bulletin , Volume 130, Issue 1, June 2019, Pages 25–37, https://doi.org/10.1093/bmb/ldz014

• Permissions Icon Permissions

Nurses comprise half the global health workforce. A nine million shortage estimated in 2014 is predicted to decrease by two million by 2030 but disproportionality effect regions such as Africa. This scoping review investigated: what is known about current nurse workforces and shortages and what can be done to forestall such shortages?

Published documents from international organisations with remits for nursing workforces, published reviews with forward citation and key author searches.

Addressing nurse shortages requires a data informed, country specific model of the routes of supply and demand. It requires evidence informed policy and resource allocation at national, subnational and organisation levels.

The definition in law, type of education, levels and scope of practice of nurses varies between countries raising questions of factors and evidence underpinning such variation. Most policy solutions proposed by international bodies draws on data and research about the medical workforce and applies that to nurses, despite the different demographic profile, the work, the career options, the remuneration and the status.

Demand for nurses is increasing in all countries. Better workforce planning in nursing is crucial to reduce health inequalities and ensure sustainable health systems.

Research is needed on: the nursing workforce in low income countries and in rural and remote areas; on the impact of scope of practice and task-shifting changes; on the impact over time of implementing system wide policies as well as raising the profile of nursing.

Achieving population health, universal health coverage and equitable access to health care is dependent on having a health workforce that is of sufficient capacity, capability and quality to meet epidemiological challenges and changing demand 1 . The World Health Organization (WHO) predicts increased global demand for health and social care staff with the creation of 40 million new jobs by 2030. 2 Professionally educated nurses are numerically the largest professional group in most countries and comprise about half the global healthcare workforce. 2 In 2014, WHO and the World Bank calculated a global shortage of 9 million nurses and midwives. They predicted this will reduce by 2030 to 7.6 million but it will have a disproportionate impact on regions such as Africa. 3 This scoping review 4 addressed the questions: what is known about the current nursing workforce, how are shortages calculated and why do shortages of nurses arise, what can be done to forestall such shortages in a national and global context and where are the evidence gaps?

A scoping review maps out the breadth of issues, identifying areas for policy and research. 4 The review has drawn on the publications of international organizations with remits for health workforce (WHO, Office of Economic Co-operation and Development[OECD]) and nursing (International Council of Nurses [ICN]), reviews concerned with nursing workforce and shortages identified through database searches (SCOPUS, Medline, CINAHL 1-1-2008–1-12-2018) and follow up of cited literature and key authors. The review excluded literature concerned with midwives and nurse-midwifes.

WHO estimated 21 million nurses/midwives 2 globally in 2014, although there are variations in definition and deployment, which we explain later in this section. Despite the 2008 global financial crisis, the number of nurses has grown in many countries 5 . The absolute numbers range from over 3 million nurses in large countries such China, India and the United States to under 5000 in smaller countries such as Guinea, Iceland and Jamaica 3 . Many countries, however, have very little data on the distribution, types or trends of their nursing workforce that contrasts with information held about the medical workforce 2 .

Nurse shortages have to be considered in the context of international variation in health system development, size of the economy, the population size as well as the presence of other key health professionals, the most important of which is medicine. The international variation in ratio of nurses to population and doctors is illustrated in Table 1 .

Examples of variation by country in ratios of nurses to population and to doctors in 2017 or nearest year data available

Data compiled from four OECD sources 5 , – 8 .

The figures in Table 1 , however, mask some fundamental variations as to who counts as a nurse; not all countries have legislation to protect the title and education level of ‘nurse’ 9 . Registered nurse (RN) academic levels also vary; for example, in Europe RN education is at diploma level in seven countries, e.g. Luxembourg, but at degree level in others, e.g. United Kingdom (UK) 10 . Some countries regulate multiple levels of nurse, such as practical nurses and advanced practice nurses. Regulated practical nurses have different names: enrolled nurse (Zambia), licensed practical nurses (US), nurse assistant (Ghana) and nurse associate (UK). Advanced practice roles are those in which RNs, with additional training, undertake diagnostic and treatment roles traditionally the domain of the doctors. Like practical nurses, these are variously named: nurse clinician (Botswana), nurse officer (Lesotho) and nurse practitioner (US). Advanced practice roles and education are not always regulated, for example as in the UK 11 . Some countries also have nurse anaesthetists who are licensed to provide general and regional anaesthesia independently. Countries that have nurse anaesthetists include: Sweden, 12 the US that has over 42 000 13 and Ghana where, regulated by the General Medical Council, there are 14 nurse anaesthetists to every one doctor anaesthesiologist 14 . The scope of practice of nurses also varies. For example, in some countries, RNs have legal authority to prescribe pharmaceutical drugs independently although there are differences as to which nurses (on registration or with additional education) and which classes of drugs 15 . The extent of the scope of practice and advanced practice roles in any country reflects historical and contemporary issues including shortages or mal-distribution of doctors as well as support or otherwise from medical professional bodies.

Four further points are relevant in considering the global nursing workforce and shortages. Firstly, 90% of nurses are women 2 . Even in countries where there have been active programmes to increase the recruitment of men, such as the US, less than 10% are male 16 . Secondly, in most countries the majority of nurses earn less than the average wage of that country 2 . There is some evidence that male nurses on average earn more than female nurses 16 and that nurses from minority ethnic backgrounds earn less and are under-represented in senior grades than those of majority ethnic origin (see for example evidence from the UK 17 ). Thirdly, the majority of nurses are employed within hospitals (see Table 2 ) despite broader international policy aims of strengthening primary care. Lastly, the majority of nurses are salaried employees although in a few countries some practice as independent, self-employed professionals as in the infirmiers liberales in France (see Table 2 ) 18 . Taking into account these similarities and variations in the education, deployment and scope of practice of nurses, we turn now to consider shortages of nurses and the causes.

Nurse employment in different sectors from five exemplar high-income countries in 2018 or nearest year

Data Sources US Bureau of Labor Statistics 19 , Australian Health Workforce 20 , NHS Digital England 21 , Japan Nursing Association 22 , Ministère des Solidarités et de la Santé 23 .

* National Health Service only, ** Infirmiers liberale in primary care.

Calculating shortages

Definitions of ‘shortages’ in workforce are policy contingent and vary between health care systems. Criteria of ‘hard to fill vacancies’ or trends in ‘volume of current vacant posts’ are often used to describe health systems experiencing financial and demand pressures 24 . The latter measure is used currently in the National Health Service (NHS) in the UK and has been reported to have an upward trend over the previous 3 years 25 . A more prosaic definition of national shortages is whether RNs are on a government’s occupation shortage list for inward migration, as they are for Australia 26 and the UK 27 but not for the US 28 , at the time of writing.

In contrast, there are those definitions of shortages that flow from staff planning projections. These calculate any gap between the numbers of nurses required (demand for) against the future number available to work (the supply). One such example is from WHO and the World Bank in which shortages are defined as lower than the minimum number of doctors and nurses per head of population required to achieve population health targets. 3 The targets in this case being 12 of the infectious disease, child and maternal health and non-communicable health specified in the Sustainable Development Goals 3 . Using national data, WHO/World Bank estimated the 7.6 million global shortfall by 2030 with disproportionate impact on Africa and low-income countries. However, many health care systems have other developments beyond minimum targets that create demand for nurses but few have undertaken nurse staffing planning projections at a national level. Only 5 of 31 high-income OECD member countries have modelled their demand for and supply of nurses to 2025. Of the five, four (Australia, Canada, Ireland and UK) predicted shortages and one, the US, predicted a surplus of qualified nurses 29 .

Models of supply of and demand for nurses

WHO offers a system-wide model for the supply and stock of all types of health professionals 2 . We have adapted the model to focus specifically on the supply and availability of qualified nurses (Fig. 1 ). This illustrates the inflows and outflows to the pipeline of supply, to the pool of RNs and to stock of nurses available for employment as nurses. The model applies at national and sub-national levels, where other patterns become more evident such as internal migration from rural to urban areas. All countries face similar problems in the supply of nurses and other health professionals in remote and rural areas 30 . There are some countries, which have an oversupply of nurses, e.g. Philippines as part of ‘export’ model—whereby working age women (often with children) enter other countries as migrants and work in the health system, sending money home to support their families 31 .

Model of the supply of nurses able and willing to participate in a national nurse labour market (adapted from WHO 2 ).

Model of factors increasing and decreasing the demand for nurses.

Shortages occur when demand for nurses outstrips the numbers of nurses available for employment. An overarching factor influencing demand is the economy; for example, vacant nurse posts were frozen following the 2008 global financial crisis in countries such as Iceland 9 and Kenya 32 . Many factors influence the demand for nurses and we offer a model of these in Figure 2 . However, the extent of the demand for nurses is country and time specific. For example, there was increased demand in Thailand in the late eighties when a strong economy was the catalyst for the growth in private hospitals 33 and currently in the US where state legislation specifies the ratio of RNs to in-patients 34 . The volume of internationally educated nurses in a country maybe an indicator of a shortfall against demand or may be the custom and practice for the supply of nurses. For example, the Netherlands has had less than 1% foreign-trained nurses in its workforce consistently over the past 15 years while others such as the UK, Australia and New Zealand have consistently had over 14% in the same period 5 .

What attracts nurses to jobs and why do they leave?

Individual : skills and interests, career plans, caring/family responsibilities, financial responsibilities.

Job characteristics : remuneration, other financial benefits (e.g. health insurance, pension), hours and pattern of working, type and volume of work, physical and/or emotional intensity of work, variety of work, team working, level of responsibility/autonomy, clinical and managerial support, professional support.

Organization : clinical and employer reputation, type (e.g. private, public), size of organization, size of specialties within an organization, infrastructure to support employees (e.g. child care facilities, meal and social facilities), access to professional and career development activities and/or funding for these.

Location : urban, sub-urban, rural, proximity to family and/or other services such as schools for children.

Macro and meso level factors in the home country with the perceived converse in the country of destination : weak economy, political instability and/or civil unrest, unemployment of nurses, low status of nurses, poor working conditions, few opportunities for nurses for career progression.

Personnel level factors : desire for different cultural, life and/or health system experience, perceived opportunity for better financial rewards, perceived opportunity for improved quality of life for family and children, following already established partner or family network in the destination country, opportunity for career advancement and/or education.

Trying to understand why nurses leave and what retains nurses in their jobs has been a perennial question; the first Lancet Commission into shortages of nurses in the UK was published in 1933 39 . Innumerable literature reviews on the subject in the intervening years demonstrate the interlocking range of factors at individual, organizational and the broader socio-economic level 40 . A recent umbrella review of systematic reviews investigating the determinants of nurse turnover (leaving their jobs) in high-income countries reported that most studies focused on ‘individual’ factors influencing ‘intention to leave’, i.e. plans rather than actual leaving 41 . Most of the evidence reported was at the individual level; high levels of stress and burnout, job dissatisfaction and low commitment was associated with intention to leave. The few studies looking at intentions to remain found this had a strong association with good supervisor support 41 . However, there was an absence of studies that considered the interplay of factors at multiple levels (e.g. individual, job characteristics, organizational characteristics and the wider labour market) on actual leaving rates of RNs or on any subgroups, such as those with caring responsibilities 41 . The International Council of Nurses has also noted there is a paucity of evidence that has considered turnover and retention of nurses in low-income countries 42 . We turn now to consider the evidence for action to solve the shortages and mal-distribution.

The WHO calls for greater investment in all human resources for health and advocates for policy attention across the system of production, regulation and employment 2 . Most commentators on nursing shortages make similar arguments that policy attention needs to be paid to all elements (known as policy bundles) and avoid policy making that relies on oversimplified linear thinking. The evidence to support this comes from high- and low-income countries where programmes that focused only on increasing the numbers entering nurse training, have failed to make an impact on increasing numbers entering the workplace or reducing gaps in priority areas with a history of shortages. Subsequent analysis has identified multiple reasons for this failure including: insufficient infrastructure for clinical education, weak regulation of education standards and few posts to apply for, see for example the review from Sub-Saharan African countries 43 . This is not to argue that increasing numbers entering nurse training is inappropriate, but should be seen as one lever among a policy bundle, including for example, retention measures. The UK and the US provide interesting comparisons, in that one (the UK) has reduced nurse training numbers over the past 15 years, has significant numbers of vacancies and plans to rely on internationally trained nurses over the next few years, while the other (the US) has significantly increased nurse graduates over the past 15 years and does not count nursing as a shortage occupation 44 .

Drawing on commentaries and WHO strategic direction statements for strengthening the nursing and midwifery contribution Table 3 describes policy actions to scale up and sustain the nursing workforce at different levels of the health care system.

Exemplars of policy areas to address improved supply, retention and productivity of nurses

The policy solutions have to attend to the demand as well as the supply side, i.e. to increase RN productivity (for example, working to the full extent of their license, task shifting to assistants, using technologies and community health workers) as well as to produce more and retain more RNs. However, documented evaluations of the impact of the implementation of policy bundles on the nursing workforce are rare not least because of the inter-sectoral nature of enactment and the relatively long period between policy decisions, implementation and outcome. Even where there are evaluations of implementation of policy actions to address shortages, such as the WHO strategy for remote and rural areas, these focus on the medical profession not nurses 45 .

At the micro level (organization/service delivery level), good human resource management practices are known to reduce the rates of voluntary turnover in all industries 46 . An umbrella systematic review considered interventions to reduce turnover rates of nurses (i.e. to retain them in their posts) in high-income countries 47 This review found relatively weak evidence for most interventions but there was strong evidence of positive impact for transition programmes and support for newly qualified nurses 47 . There was also evidence that nurse manager leadership styles that were perceived as encouraging work group cohesion were also effective in reducing turnover 47 . Positive working environments are those that not only ensure the nurses well-being but sustain or increase their motivation in their work. There is some evidence that many RNs (like many physicians) in high-income countries consider that they are not working to the full scope of their training and are undertaking work that could be undertaken by others 44 . However, task shifting, shifts in jurisdiction and changes in skill-mixes in teams raises questions of adequate preparation, patient safety and cost effectiveness—all of which require consideration within specific contexts. For example, a growing body of evidence in high-income countries demonstrates a relationship between RN staffing levels and patient safety in acute in-patient hospital settings. Recent research on in-patient hospital care in the UK demonstrates that lower RN staffing and higher levels of admissions per RN are associated with increased risk of death during an admission to hospital and that use of nursing assistants does not compensate for reduced RN staffing 48 . There are significant gaps in evidence for the most effect ways of increasing RN productivity as well as attracting and retaining RNs in the workplace that requires attention to be given to the macro and overarching issues in every health system.

The first major gap is in relation to nursing workforce planning. Workforce planning at a national level is an inexact science and is often absent for nursing, which is in contrast with medical manpower planning. To plan for solutions, you need to understand the scale of the problem, which in the case of nursing, is limited by the significant evidence gaps. For example, at the national level, nursing workforce data is often incomplete and based on historical activity rather than projections. The 2016 WHO resolution on human resources for health urges all countries to have health workforce-related planning mechanisms and has introduced national health workforce accounts with core indicators, including ratio of nurses to population, for annual submission to the WHO Secretariat 1 . However, this could be considered the minimum requirement for benchmarking rather than proactively modelling the future demand for nurses, the availability and supply of nurses and planning to meet the gap or shortfall. This then flags the next gap—the evidence to base the planning decisions on.

The second major gap is the evidence informing policy decisions about interventions that work to attract, equitably distribute, retain and sustain a nursing workforce against the requirements of any health care system. It is noteworthy that the WHO guidance for scaling up and retaining all health care professionals 2 is predominantly evidenced from studies of doctors, thus further emphasizing this gap in the evidence. The demographic profile, status, education, career options and remuneration levels for these two professions are very different and assumptions that evidence from one professionally is automatically applicable and relevant to the other is contestable and at worst misleading.

Overlaying these gaps in knowledge there is an issue, common across many countries as noted by WHO 2 , that the profession of nursing has not been valued and given the policy attention congruent with its scale. Having a weak voice and influence in national and international health workforce policy development has inevitable consequences, which in this context means that fewer levers are available to address shortages and action is slow. At the time of writing, there is a global WHO-sponsored campaign called Nursing Now (2018–20) ( https://www.nursingnow.org/ ) that supports country-specific campaigns and activities to raise the profile of the nursing profession, develop leaders for governments and to make change at a systems level. This is involving nurses in policy making, particularly with regard to increasing and retaining the nursing workforce. The response to, and impact of, these calls for country-specific campaigns is yet to be evaluated.

This review has demonstrated that that the nature and size of the professionally qualified nurse workforce is shaped by the societal context of individual countries—political choices that influence decisions about resource allocation to health systems, demographics (labour market pressures on working age, particularly of women), image of nursing and its positional power in relation to medicine, demands for care/health and social inequalities. Understanding nursing shortages and acting on them requires attention to the gaps in knowledge and evidence but also the wider societal context of nursing.

This review was undertaken without external funding.

The authors have no potential conflicts of interest.

World Health Organization . Sixty-ninth World Health Assembly. Resolution WHA.69.19: Global Strategy on Human Resources for Health: Workforce 2030 . 2016 . http://apps.who.int/gb/ebwha/pdf_files/WHA69/A69_R19-en.pdf?ua=1&ua=1 (12 December 2018, date last accessed) .

World Health Organization . Global Strategy on Human Resources for Health: Workforce 2030 . https://www.who.int/hrh/resources/pub_globstrathrh-2030/en/ (12 December 2018, date last accessed) .

Scheffler R , Cometto G , Tulenko K , et al.  Health Workforce Requirements for Universal Health Coverage and the Sustainable Development Goals . Background Paper N.1. Human Resources for Health Observer Series No 17 . World Health Organization , 2016 . https://www.who.int/hrh/resources/health-observer17/en/ (23 December 2018, date last accessed) .

Arksey H , O’Malley L . Scoping studies: towards a methodological framework . Int J Soc Res Methodol 2005 ; 8 : 19 – 32 . DOI: 10.1080/1364557032000119616.

Google Scholar

OECD . Health at a Glance. Chapter 8 Health Workforce . 2017 ; 148 – 165 . https://read.oecd-ilibrary.org/social-issues-migration-health/health-at-a-glance-2017_health_glance-2017-en#page160 (12 December 2018, date last accessed) .

OECD/WHO . Health at a Glance: Asia/Pacific 2016: Measuring Progress towards Universal Health Coverage . Paris : OECD Publishing , 2016 , http://dx.doi.org / 10.1787/health_glance_ap-2016-en . (12 December 2018, date last accessed) .

Google Preview

OECD/European Union . Health at A Glance: Europe 2018 Availability of Nurses Section 11.7 ©OECD/EUROPEAN UNION 2018 . https://www.oecd-ilibrary.org/social-issues-migration-health/health-at-a-glance-europe-2018_health_glance_eur-2018-en . (12 December 2018, date last accessed).

OECD . Factbook 2015–2016: Economic, Environmental and Social Statistics Ratio of Nurses to Physicians: 2013 or Latest Available Year . https://read.oecd-ilibrary.org/economics/oecd-factbook-2015-2016/ratio-of-nurses-to-physicians_factbook-2015-graph199-en#page1 (12 December 2018, date last accessed) .

Benton DC , Fernández-Fernández MP , González-Jurado MA , et al.  Analysis of a global random stratified sample of nurse legislation . Int Nurs Rev 2015 ; 62 : 207 – 17 .

Praxmarer-Fernandes S , Maier CB , Oikarainen A , et al.  Levels of education offered in nursing and midwifery education in the WHO European region: multicountry baseline assessment . WHO Europe. Public Health Panorama 2017 ; 3 : 357 – 536 http://www.euro.who.int/__data/assets/pdf_file/0008/348416/PUBLIC_HEALTH_PANORAMA_SEPTEMBER_RUS.pdf#page=64 . (12 December 2018, date last accessed).

Maier C , Aiken L , Busse R . Nurses in Advanced Roles in Primary Care: Policy Levers for Implementation . OECD Health Working Papers, No. 98, OECD Publishing, Paris . 2017 . http://dx.doi.org/10.1787/a8756593-en (12 December 2018, date last accessed) .

Meeusen V , Ouellette S , Horton B , et al.  The global organization of nurses in anesthesia: the International Federation of Nurse Anesthetists . Trends in Anaesthesia and Critical Care 2016 ; 6 : 20 – 25 .

United States Bureau of Labour Statistics . Occupational Employment and Wages, May 2017. 29-1151 Nurse Anesthetists . https://www.bls.gov/oes/current/oes291151.htm (12 December 2018, date last accessed) .

Ghana Association of Certified Registered Anesthetist . Information. https://ifna.site/country/ghana/ (12 December 2018, date last accessed) .

Bhanbhro S , Drennan VM , Grant R , et al.  Assessing the contribution of prescribing in primary care by nurses and professionals allied to medicine: a systematic review of literature . BMC Health Serv Res 2011 ; 11 : 330 doi.org/10.1186/1472-6963-11-330 .

Landivar LN . Men in Nursing Occupations: American Community Survey Highlight Report . US Census Bureau . 2013 . https://www.census.gov/library/working-papers/2013/acs/2013_Landivar_02.html (12 January 2018, date last accessed).

Department of Health and Social Care (England) . NHS Pledges Action to Eliminate Ethnicity Pay Gap . 2018 . https://www.gov.uk/government/news/nhs-pledges-action-to-eliminate-ethnicity-pay-gap (12 January 2019, date last accessed) .

Ministère des Solidarités et de la Santé (France) . Data DRES TABLEAU 1. EFFECTIFS DES INFIRMIERS . 2018 . http://www.data.drees.sante.gouv.fr/TableViewer/tableView.aspx?ReportId=3704 (20 January 2019, date last accessed) .

US Bureau of Labor Statistics . Occupational Outlook. Handbook 2018 . https://www.bls.gov/ooh/healthcare/registered-nurses.htm#tab-3 (12 December 2018, date last accessed) .

Australia Health Workforce . Nurses in Focus . 2013 . Adelaide, Health Workforce Australia , http://www.health.gov.au/internet/main/publishing.nsf/Content/hwa-archived-publications (12 December 2018, date last accessed).

England NHS Digital General Practice Workforce , Final 30 September 2018, Experimental Statistics . https://digital.nhs.uk/data-and-information/publications/statistical/general-and-personal-medical-services/final-30-september-2018-experimental-statistics and NHS Workforce Statistics (10 January 2019, date last accessed) .

Japan Nursing Association . Statistic: Numbers Employed . 2017 . https://www.nurse.or.jp/jna/english/statistics/index.html (10 January 2019, date last accessed) .

Ministère des Solidarités et de la Santé . Data Dres Tableau 1. Effectifs Des Infirmiers . 2018 . http://www.data.drees.sante.gouv.fr/TableViewer/tableView.aspx?ReportId=3704 (20 January 2019, date last accessed) .

Ono T , Lafortune G , Schoenstein M . Health workforce planning in OECD countries: a review of 26 projection models from 18 countries. OECD Health Working Papers , No. 62 , Paris : OECD Publishing , 2013 . http://dx.doi.org/10.1787/5k44t787zcwb-en .

NHS Digital . NHS Vacancy Statistics England February 2015–March 2018, Provisional Experimental Statistics . 2018 . https://digital.nhs.uk/data-and-information/publications/statistical/nhs-vacancies-survey/nhs-vacancy-statistics-england---february-2015---march-2018-provisional-experimental-statistics (12 December 2018, date last accessed) .

Home Office (UK) . Immigration Rules Appendix K: Shortage Occupation List . 2016 updated January 2019 . https://www.gov.uk/guidance/immigration-rules/immigration-rules-appendix-k-shortage-occupation-list (12 January 2019, date last accessed) .

Australia Government, Department of Home Affairs . Temporary Skill Shortage Visa: Skilled Occupation List . https://immi.homeaffairs.gov.au/visas/working-in-australia/skill-occupation-list (12 January 2019, date last accessed) .

US Citizenship & Immigration Services . H-1C Registered Nurse Working in a Health Professional Shortage Area as Determined by the Department of Labor . Expired 2009 . https://www.uscis.gov/working-united-states/temporary-workers/h-1c-registered-nurse/h-1c-registered-nurse-working-health-professional-shortage-area-determined-department-labor . (12 December 2018, date last accessed).

World Health Organization . Health Workforce and Labour Market Dynamics in OECD High-Income Countries: a Synthesis of Recent Analyses and Simulations of Future Supply and Requirements . Human Resources for Health Observer Series No 20. 2017 . https://www.who.int/hrh/resources/health-observer20/en/ . (12 December 2018, date last accessed).

World Health Organization . Increasing Access to Health Workers in Remote and Rural Areas through Improved Retention Global Policy Recommendations . https://www.who.int/hrh/retention/guidelines/en/ (12 December 2018, date last accessed).

Kanchanachitra C , Lindelow M , Johnston T , et al.  Human resources for health in Southeast Asia: shortages, distributional challenges, and international trade in health services . Lancet 2011 ; 377 : 769 – 81 . doi: 10.1016/S0140-6736(10)62035-1 .

Appiagyei AA , Kiriinya RN , Gross JM , et al.  Informing the scale-up of Kenya’s nursing workforce: a mixed methods study of factors affecting pre-service training capacity and production . Hum Resour Health 2014 ; 12 : 47 http://www.human-resources-health.com/content/12/1/47 .

Abhicharttibutra K , Kunaviktikul W , Turale S , et al.  Analysis of a government policy to address nursing shortage and nursing education quality . Int Nurs Rev 2017 ; 64 : 22 – 32 .

Jones T , Heui Bae S , Murry N , et al.  Texas nurse staffing trends before and after mandated nurse staffing committees . Policy Polit Nurs Pract 2015 ; 16 , 79 – 96 . DOI: 10.1177/1527154415616254 .

Rottenberg S . On choice in labor markets . Ind Labor Relat Rev 1956 ; 9 : 183 – 99 .

Trépanier A , Gagnon MP , Mbemba GI , et al.  Factors associated with intended and effective settlement of nursing students and newly graduated nurses in a rural setting after graduation: a mixed-methods review . Int J Nurs Stud 2013 ; 50 : 314 – 25 .

Mbemba GI , Gagnon MP , Hamelin-Brabant L . Factors influencing recruitment and retention of healthcare workers in rural and remote areas in developed and developing countries: an overview . J Public Health Africa 2016 ; 7 : 565 . doi: 10.4081/jphia.2016.565 .

Prescott M , Nichter M . Transnational nurse migration: future directions for medical anthropological research . Soc Sci Med 2014 ; 107 : 113 – 23 .

The Lancet Commission on Nursing . The Lancet 1933 ; 221 : 369 – 70 https://doi.org/10.1016/S0140-6736(00)98375-2 .

Nei D , Snyder LA , Litwiller BJ . Promoting retention of nurses: a meta-analytic examination of causes of nurse turnover . Health Care Manage Rev 2014 ; 40 : 237 – 253 .

Halter M , Boiko O , Pelone F , et al.  The determinants and consequences of adult nursing staff turnover: a systematic review of systematic reviews . BMC Health Serv Res 2017 ; 17 : 824 . doi: 10.1186/s12913-017-2707-0. Review .

Buchanan I , Shaffer FA , Catton H . Policy Brief: Nurse Retention. Geneva, International Council of Nurses . 2018 . https://www.icn.ch/news/new-policy-brief-nurse-retention-released (12 January 2019, date last accessed) .

Bvumbwe T , Mtshali N . Nursing education challenges and solutions in sub Saharan Africa: an integrative review . BMC Nurs 2018 ; 17 : 3 . doi: 10.1186/s12912-018-0272-4 .

Colombo F . The Nursing Workforce: Past Trends, Future Developments . 2016 . http://www.oecd.org/health/health-systems/Nursing-workforce-February2016.pdf (20 January 2019, date last accessed) .

Dolea C , Stormont L , Braichet JM . Evaluated strategies to increase attraction and retention of health workers in remote and rural areas . Bull World Health Organ 2010 ; 88 : 379 – 385 .

Jiang K , Lepak DP , Hu J , et al.  How does human resource management influence organizational outcomes? A meta-analytic investigation of mediating mechanisms . Acad Manag J 2012 ; 55 : 1264 – 94 .

Halter M , Pelone F , Boiko O , et al.  Interventions to reduce adult nursing turnover: a systematic review of systematic reviews . Open Nurs J 2017 ; 15 : 108 – 23 . doi: 10.2174/1874434601711010108. www.ncbi.nlm.nih.gov/pmc/articles/PMC5725565/ .

Griffiths P , Maruotti A , Recio Saucedo A , et al.  Nurse staffing, nursing assistants and hospital mortality: retrospective longitudinal cohort study . BMJ Qual Saf 2018 . bmjqs-2018-008043 . doi: 10.1136/bmjqs-2018-008043 . [Epub ahead of print]

Email alerts

Citing articles via.

• Recommend to your Library

Affiliations

• Online ISSN 1471-8391
• Print ISSN 0007-1420
• Copyright © 2024 Oxford University Press
• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

A systematic review study on the factors affecting shortage of nursing workforce in the hospitals

Affiliations.

• 1 Vanuatu College of Nursing Education, Ministry of Health, Port Vila, Vanuatu.
• 2 School of Nursing and Healthcare Leadership, University of Bradford, Bradford, UK.
• PMID: 36303066
• PMCID: PMC9912424
• DOI: 10.1002/nop2.1434

Aim: This study aimed to determine factors that influence the nursing workforce shortage and their impact on nurses.

Design: This study applied a systematic review design.

Methods: Using Cochrane library guidelines, five electronic databases were systematically searched (Research 4life-PubMed/Medline, Scopus, Embase, CINAHL) from 2010-2021. The remaining articles with pertinent information were presented in a data extraction sheet for further thematic analysis. A Reporting Items for Systematic Reviews and Meta-Analysis Flow Diagram was adopted and used. The studies published from 2010-2021 and in English language were examined and included in the systematic review.

Results: Four themes were identified as factors influencing the nursing workforce shortage, including Policy and planning barriers, Barriers to training and enrolment, Factors causing nursing staff turnover and Nurses' stress and burnout. Nursing workforce shortage is a global challenge that roots in multiple causes such as individual, educational, organizational and managerial and policy-making factors.

Keywords: nurse burnout; nurse retention; nursing shortage; nursing workforce; systematic review.

© 2022 The Authors. Nursing Open published by John Wiley & Sons Ltd.

Publication types

• Meta-Analysis
• Systematic Review
• Job Satisfaction
• Nursing Staff, Hospital*

• ISI Impact Factor: 1.042
• [email protected]
• Volume 15 - Issue 4
• Download PDF [ P: 441-443 ]

Article Citation

Manana Machitidze. Impact of The Nurses Education and Shortage on The Patients Care Outcomes-Literature Review. Am J Biomed Sci & Res. 2022 - 15(4). AJBSR.MS.ID.002135. DOI: 10.34297/AJBSR.2022.15.002135.

Navigation Menu

Introduction, acknowledgment, conflict of interest, share this article.

Impact of The Nurses Education and Shortage on The Patients Care Outcomes-Literature Review

Manana machitidze.

*Corresponding author: Manana Machitidze, School of Health Sciences, University of Georgia, Georgia.

Received: January 23, 2022; Published: February 25, 2022

DOI: 10.34297/AJBSR.2022.15.002135

Aim/Objective: The reasons for the shortage of nurses and the factors affecting it varies according to the development of the countries. However, the Covid-19 pandemic has exacerbated the problem globally. The aim of this article is to analyze the problem and its triggers, as well as the consequences.

Background: Qualified nurses are needed to ensure proper care for patients and the general population. Therefore, the lack of well-trained and educated nurses is a rather big problem, especially if the deficit is widespread. The shortage of nurses has reached a critical level for healthcare services, both locally and globally. The global shortage of nurses has been further exacerbated by the impact of the Covid-19 pandemic. Lack of human resources and professional nurses affects patient care outcomes. With sufficient human resources, qualified nurses are needed. The quality of patient care can only be ameliorated if there is sufficient nursing staff in the hospital. Lack of nurses hinders the implementation of nursing education, as there is a shortage in this area. Shortage of teachers limits the number of students admitted to nursing programs.

Design and Methods: The article is based on secondary research and limited to descriptive analysis. The article is based on the analysis of publications and analysis of international organizations (WHO, the International Council of Nurses (ICN)), as well as a review on the shortage of nurses and COVID -19 pandemic impact on Nurses shortage (SCOPUS, PubMed).

Results and Conclusions: Nursing shortages are worrisome, because effects the safety of both-the nurse and the patient. Nurses are busier, more stressed, and less focused on the details during working. The result of this is less communication to patients/ colleagues, making mistakes during working, which increases risks for errors and unsafety during patient care, treatment, and recovery processes. The global shortage of nurses has been further exacerbated by the impact of the Covid-19 pandemic.

Keywords: Nurse, Education, Nursing Shortage, Triggers; Well-Trained; Human Resources; Ameliorated; Communication; Stressed; International Organizations

Abbreviations: IRF - Idiopathic retroperitoneal fibrosis; RPF- retroperitoneal fibrosis; CT - computed tomography; KUB - kidney, ureter and bladder; AMA - anti-smooth muscle antibody; ANA - antinuclear antibody; RF - rheumatoid factor; Anti-TPO - antithioperoxides; ESR - erythrocyte sedimentation rate; Ig4-RD - immunoglobulin 4 related disease; MRI - magnetic resonance imaging; IVU - intravenous program

The nursing shortage is a critical issue not only in Georgia but in different parts of the world. For instance, in the United States, the shortage of nurses in the hospitals happens to cause serious problems since it affects the overall quality of patient care [1]. What’s more, as early research demonstrates, the lack of nursing staff is directly linked to patient complications and even death [2]. Several organizations in the United States, including the IOM, AHRQ and CMS focus on improving the standards and safety of patient care on a national scale. These organizations understand that to reach this goal, they need to first work on increasing the quantity of highly qualified nursing workforce in the hospitals [1]. In other words, the quality of patient care can only be ameliorated if there is sufficient nursing staff in the hospital (ibid).

Moreover, Hassmiller and Cozine [3] explore different ways to address the nurse shortage. By taking the Robert Wood Johnson Foundation’s (RWJF’s) example of supporting nurses, the authors outline how crucial it is to invest in the nursing programs, including leadership and mentoring programs for nurses. It is equally important to change the working processes for nurses-instead of spending more time on administrative tasks, the nurses should focus on the patient and improve care in their units. Furthermore, Hassmiller and Cozine emphasize the role of investing in research. In the case of RWJF, the foundation supported new studies at New York University and the University of Buffalo by tracking the career paths of newly graduated nurses over four years and identifying the factors impacting their work decisions. It is important to note that the problem with the nursing shortage should be addressed by bothprivate and public organizations and invest in the qualified nursing workforce at local and national levels (ibid). Correspondingly, Hahn and Truman’s [6] study describes the importance of public health interventions through educational programs and policies.

While it is important to look at the implications of the shortage of highly qualified nursing staff, it is equally crucial to examine the reasons behind this problem. The nursing shortage is both-a local and a global multifaceted problem. Based on the World Health Organization (WHO) report, 7.2 million healthcare workers are in shortage regarding the existing health needs. This problem, which is becoming more and more serious in Europe, Asia, and North America, might regulate the future world health policy [8]. It’s important to note that population aging and changes in global demographics become one of the factors standing behind the global undersupply of nursing staff. According to the UN, there will be a double quantity of people aged 60 and over, increasing up to 2.1 billion seniors by 2050 [9]. Aside from the demographic reasons, there is also a strong factor of migration involved in the processes. For instance, as Poland became part of the EU, highly qualified nurses started to move to different countries to ameliorate their living conditions. Economic migration of qualified nurses is not only the source of undersupplied nursing staff in Poland but in various countries [8]. It is evident, that to avoid further shortage of the nursing staff in the future, the healthcare system needs to develop better employment conditions for nurses, implement programs oriented on their professional growth and regulate their salary system (ibid). As there is increasing number of nurses leaving their profession [7] it is crucial to urgently address the problem through governmental and non-governmental interventions.

The shortage of nurses became particularly acute at the onset of the Covid-19 pandemic. The current situation has exacerbated the problem. The statistics are disturbing. According to the State of the World’s Nursing (SOWN) report, there was a global deficit of almost 6 million nurses before the pandemic, with the majority (89%) concentrated in low-and low-middle-income countries. Added to this is the fact that for the next 10 years, one in six nurses in the world is expected to retire. This means that 4.7 million new nurses will have to replace them; The migration of nurses puts at risk some country’s health requirements. By the same report, every eighth nurse is in migration. [7]. Countries should seriously consider maintaining nursing attractiveness as a profession. It is important for them to be provided with pay and working conditions, career prospects, continuing postgraduate education. Which will help ensure that the periodic and long-term supply of new nurses will not be delayed.

According to the American Nurses Association (ANA), by 2022 there will be available more registered nurses working than in other professions. According to an article in the Nursing Times, based on a study by the U.S. Bureau of Labor Statistics, 11 million additional nurses are needed to prevent deficits in the future. It is also expected that from 2016 to 2026, the employment of nurses will grow at a faster rate (15%) than in other professions. (NCBI, 2020).

The global shortage of nurses has been further exacerbated by the impact of the Covid-19 pandemic. It is likely that in countries where the impact of different waves of pandemics was significant and severe, it also affected nursing staff. Because nurses ‘’burned out’’, infected, and ‘’post-COVID’’ disability helped to isolate them from work for a short or long period of time. 90% of NNAs report inadequate busy work schedules and a lack of human resources due to pandemics. Low pay, fatigue, stress, and busy work schedules are the major determinants that make nurses think to leave working during pandemics. ICN member NNAs 20% indicates that in 2020 the number of nurses who left the job increased. 70% of associations says their countries are ready to increase the number of nursing students, but still be an interval of few years before new graduate nurses are ready to enter the workforce. It is unfortunate that due to the lack of nurses, the aging of staff, and pandemic result, the ICN estimates that up to 13 million nurses will be needed to fill the global shortage of nurses in the future. (International Council of Nurses, 2021). The data is really alarming.

Addressing the problem of lack of nurses is important for improving the treatment outcomes of patients. Much research is still needed to explore, analyze, and better determine the relationship between nursing care and patient outcomes. When talking about the shortage of nurses, it is interesting to have a consistent analysis of where the shortage of nurses begins. Where is the shortage of nurses coming from-lack of nursing students or graduates? While there is a shortage of nurses in the country, how easy is it for a graduate to get a job? Why do nurses leave their jobs? Is the nursing faculty ready to accept students and ensure a quality teaching process? What is the distribution of this deficit in the different specializations of nursing, can we equate the shortage/deficit with the number of primary care workers and the number of nurses in the clinical sector? Consequently, the factors that contribute to the shortage of nurses are different, not only globally, but also nationwide, and therefore, effective strategies to eliminate the shortage must be different. Unfortunately, due to the lack of nursing staff for a variety of reasons, the nursing profession continues to be in short supply. This is due to the lack of potential nursing educators/mentors, unequal distribution of labor, and migration. Nursing shortages are worrisome because effects the safety of both-the nurse and the patient. Nurses are busier, more stressed, and less focused on the details during working. The result of this is less communication to patients/colleagues, making mistakes during working, which increases risks for errors and unsafety during patient care, treatment, and recovery processes.

I thank the International organizations and the authors for their work whose research findings were used in the article.

Table 1: The cause of the wound must be identified before the initiation of treatment so that it can be controlled accordingly. There are different causes of the wound which are listed.

Figure 1: Homeostasis is the first response generated within minutes. Platelets play their role in clotting. The period of 1-4 days of injury is Inflammation. during this period neutrophils and macrophages remove cell debris and involve in the process of phagocytosis. Proliferation is the period of the 4th-21st day of injury. Proliferation involves the re-establishment of skin function and different cells (Macrophages, Lymphocytes, Fibroblasts, Neurocytes, and Angiocytes) perform their function to re-establish the function of the skin.

Consanguinity Worldwide

• Peter I Buerhaus, Karen Donelan, Beth T Ulrich, Linda Norman, Mamie Williams, et al. (2005) Hospital RNs' and CNOs' perceptions of the impact of the nursing shortage on the quality of care. Nurs Econ 23(5): 214-221 .
• Jack Needleman, Peter Buerhaus (2003) Nurse staffing and patient safety: Current knowledge and implications for action. Int J Qual Health Care 15(4): 275-277.
• Susan B Hassmiller, Maureen Cozine (2006) Addressing the nurse shortage to improve the quality of patient care. Health Affairs 25(1): 268-274.
• Anita Atwal, Kay Caldwell (2006) Nurses' perceptions of multidisciplinary team work in acute health-care. Int J Nurs Pract 12(6): 359-365.
• Lisa M Haddad, Pavan Annamaraju, Tammy J Toney-Butler (2020) Nursing Shortage. Stat Pearls [Internet].
• Robert A Hahn, Benedict I Truman, (2015) Education improves public health and promotes health equity. Int J Health Serv 45(4): 657-678.
• International Council of Nurses (2020) COVID-19 and The International Supply of Nurses. Report For The International Council of Nurses.
• Marć M, Bartosiewicz A, Burzyńska J, Chmiel Z, Januszewicz P (2018) A nursing shortage - a prospect of global and local policies. Int Nurs Rev 66(1): 9-16.
• United Nations (2015) World Population Prospects: The 2015 Revision.
• World Health Organization (2013) A Universal Truth: No Health Without a Workforce.

Sign up for Newsletter

Sign up for our newsletter to receive the latest updates. We respect your privacy and will never share your email address with anyone else.

American Journal of Biomedical Science & Research (ISSN: 2642-1747) is an Open access online Journal dedicated in advancing the latest scientific knowledge of science, medicine, technology and its related disciplines.

Contact info

BiomedGrid LLC, 333 City Boulevard West, 17 th Floor, Orange, California, 92868, USA   +1 (626) 698-0574   [email protected]

Quick Links

Aim & scope

Editorial Committee

Open Access

Peer Review Process

Special Issues

Online Submission

Register for Editorial Committee

Recommended a Librarian

Advertise with us

Refer a friend.

Track Your Article

Suggested by

Referrer Details

Disseminate your Research today!

We offers a wealth of free resources on academic research and publishing. Sign up and get complete access to a vibrant global community of researchers.

If you forgot / lost your password, please click on this link to reset it.

Not a member? Sign Up

Forgot Password?

Already have an account? Log In

Unsubscribed

Your request for unsubscribing was successful. Unsubscription will be finalized in next 24 hours. During this time you may receive some previously planned communications.

Join as Reviewer We welcome academicians to join our committee

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Elsevier Sponsored Documents

Beyond ratios - flexible and resilient nurse staffing options to deliver cost-effective hospital care and address staff shortages: A simulation and economic modelling study

Peter griffiths.

a Faculty of Environmental and Life Sciences, University of Southampton, Southampton, UK

b National Institute for Health Research Applied Research Collaboration (Wessex), Southampton, UK

c Portsmouth Hospitals University NHS Trust, Portsmouth, UK

Christina Saville

Jane e. ball, jeremy jones, thomas monks.

d University of Exeter Medical School, Exeter, UK

Associated Data

This paper draws on research and data reported in more detail in the NIHR Journals Library Health Services and Delivery Research. The data for this paper consist of anonymous ward and hospital parameters and simulation results. All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review. The simulation model and accompanying documentation are available from the author on reasonable request.

In the face of pressure to contain costs and make best use of scarce nurses, flexible staff deployment (floating staff between units and temporary hires) guided by a patient classification system may appear an efficient approach to meeting variable demand for care in hospitals.

We modelled the cost-effectiveness of different approaches to planning baseline numbers of nurses to roster on general medical/surgical units while using flexible staff to respond to fluctuating demand.

Design and setting

We developed an agent-based simulation, where hospital inpatient units move between being understaffed, adequately staffed or overstaffed as staff supply and demand (as measured by the Safer Nursing Care Tool patient classification system) varies. Staffing shortfalls are addressed by floating staff from overstaffed units or hiring temporary staff. We compared a standard staffing plan (baseline rosters set to match average demand) with a higher baseline ‘resilient’ plan set to match higher than average demand, and a low baseline ‘flexible’ plan. We varied assumptions about temporary staff availability and estimated the effect of unresolved low staffing on length of stay and death, calculating cost per life saved.

Staffing plans with higher baseline rosters led to higher costs but improved outcomes. Cost savings from lower baseline staff mainly arose because shifts were left understaffed and much of the staff cost saving was offset by costs from longer patient stays. With limited temporary staff available, changing from low baseline flexible plan to the standard plan cost £13,117 per life saved and changing from the standard plan to the higher baseline ‘resilient’ plan cost £8,653 per life saved.

Although adverse outcomes from low baseline staffing reduced when more temporary staff were available, higher baselines were even more cost-effective because the saving on staff costs also reduced. With unlimited temporary staff, changing from low baseline plan to the standard cost £4,520 per life saved and changing from the standard plan to the higher baseline cost £3,693 per life saved.

S hift-by-shift measurement of patient demand can guide flexible staff deployment, but the baseline number of staff rostered must be sufficient. Higher baseline rosters are more resilient in the face of variation and appear cost-effective. Staffing plans that minimise the number of nurses rostered in advance are likely to harm patients because temporary staff may not be available at short notice. Such plans, which rely heavily on flexible deployments, do not represent an efficient or effective use of nurses.

Study registration: ISRCTN 12307968

Tweetable abstract: Economic simulation model of hospital units shows low baseline staff levels with high use of flexible staff are not cost-effective and don't solve nursing shortages.

What is already known about the topic?

• • Because nursing is the largest staff group, accounting for a significant proportion of hospitasl' variable costs, nurse staffing is frequently the target of cost containment measures
• • Staffing decisions need to address both the baseline staff establishment to roster, and how best to respond to fluctuating demand as patient census and care needs vary
• • Flexible deployment of staff, including floating staff and using temporary hires, has the potential to reduce expenditure while meeting varying patient need, but high use of temporary staff may be associated with adverse outcomes.

What this paper adds

• • Low baseline staff rosters that rely heavily on flexible staff provide cost savings largely because units are often left short staffed, leading adverse patient outcomes and increased non-staff costs.
• • A staffing plan set to meet average demand is cost effective compared to a plan with a lower baseline.
• • A staffing plan with a higher baseline, set to meet demand 90% of the time, is more resilient in the face of variation and may still be highly cost effective

1. Introduction

In the face of pressure to contain costs and to use nursing staff, who are in short supply, as efficiently as possible, it is important to understand how best to plan staffing on hospital units. Key decisions relate to the balance between the baseline staffing level to routinely roster (schedule) and the use of flexible staffing (floats and temporary hires) to meet variation in demand caused by variation in census and the needs of patients. The goal is to ensure that the unit staffing system is able to meet fluctuating demand while avoiding wasteful use of human resources and the associated costs. Flexible approaches to staffing deployment to meet variable demand for care have been advocated as a way of ensuring staffing levels are maintained in the face of nursing shortages ( Aiken et al., 2013 ). Some studies have claimed that flexible staffing plans are superior to fixed plans ( Kortbeek et al., 2015 ) but concerns have been raised about potential adverse effects on quality from high use of temporary staff (e.g. Bae et al., 2015 , 2010 ; Dall'Ora et al., 2019b ).

In addressing staff shortages, it is important to test assumptions about efficient and effective staff deployment. In a recent review of staffing tools, we found that there is a dearth of evidence about the performance of staffing methods in practice and, in particular, little evidence of the costs and effectiveness of different approaches to determining nurse staffing requirements ( Griffiths et al., 2020a ). Although there is some evidence that tools in use can measure demand, it is not clear that they identify an optimal staffing level, nor do tools intrinsically address how to schedule staff in advance to meet anticipated variation in demand.

Managing staffing to address variation in demand is a major challenge. Studies have shown substantial variation in demand for nursing care between different hospital units but also from day to say within a unit ( Davis et al., 2014b ; Griffiths et al., 2018 ; Van den Heede et al., 2009 ). Rather than operating with a high baseline staff to accommodate anticipated peaks, flexible deployment of staff is often assumed to be the most efficient approach to meeting such variable demand but there is a lack of evidence for cost-effectiveness or the appropriate balance between the core establishment and flexible deployments ( Dall'ora and Griffiths, 2018 ). Because the adverse effect of low nurse staffing has been demonstrated in many studies and is now widely accepted, there has been much focus on mandatory staffing policies and minimum staffing ratios ( Driscoll et al., 2018 ; Griffiths et al., 2016 ; Kane et al., 2007 ). However, the use of ratios is often considered inflexible and inefficient ( Buchan, 2005 ) and even when such a policy is in operation the challenge remains to ensure the proper balance between permanent staff who are rostered in advance, and flexible staffing in order to maintain the required staffing level as demand varies.

In a previous publication, Saville et al. (2021) we explored different staffing policies guided by the Safer Nursing Care Tool ( The Shelford Group, 2014 ). The Safer Nursing Care Tool (often referred to by initials SNCT or as the ‘Shelford Tool’) is a patient classification system. It is used in most English National Health Service Hospitals to guide baseline establishments (that is the number of nurses to employ) and, increasingly, daily staff deployments ( Ball et al., 2019 ) informing decision about redeployment of staff between units (floating) or the hiring of temporary staff from the hospital internal pool (bank) or external staffing agencies. In an extensive literature review we found no evidence to determine the cost effectiveness of different ways to use such tools to guide staffing decisions ( Griffiths et al., 2020a ).

In our previous study, we used a simulation model to compare a standard staffing plan, following Safer Nursing Care Tool recommendations, where baseline staffing was set to meet average demand, with two alternatives. Firstly, we considered a staffing plan in which fewer staff are rostered routinely, where the emphasis is on the use of flexible deployments, anticipating that most fluctuations in demand would be met by internal redeployment and use of temporary staff. Secondly, we considered a staffing plan in which the baseline staff to be rostered is set at a level that is higher than the mean and is designed to be sufficient to cope with most peaks in demand. While still using flexible staffing, this plan emphasised the resilience of the baseline roster in the face of varying levels of demand.

We found that if the number of staff from the permanent establishment rostered on each shift was set at a low level, costs were reduced, but this apparent efficiency was achieved by leaving many shifts understaffed, largely because of the limited availability of staff to float between units or fulfil short notice requests for additional temporary staff ( Saville et al., 2021 ). Both the levels of understaffing and cost savings were highly dependent on our assumptions about the availability of temporary staff. When more temporary staff were available, understaffing was less common but consequently cost savings were much reduced. In this paper, we consider this further, extending our models to consider the cost-effectiveness of the different approaches.

Using data from an observational study in general inpatient units (wards) of acute care hospitals, we developed a simulation model of demand for unit based inpatient nursing. We used this to test various staffing plans guided by the Safer Nursing Care Tool patient classification system ( Griffiths et al., 2020b ; Saville et al., 2021 ), simulating the staffing levels achieved on each unit and for each shift in the face of variable demand and variable supply of staff. We then estimated the costs and consequences of the resulting staffing levels in an economic model, calculating cost per life saved using estimates of the effects of low staffing on length of stay and risk of death derived from a recent study ( Griffiths et al., 2018 ).

2.1. Staffing plans

We considered and compare three staffing plans. In the ‘standard’ plan, a baseline number of staff are rostered (scheduled) to work on each unit, set at a level designed to meet average demand observed on that unit, as measured by the patient classification system. This reflects the typical approach to using staffing tools where the mean average of staffing requirements is used to guide decision-making ( Griffiths et al., 2020a ).

All staffing plans incorporate flexible staffing guided by the demand presented by the patients on each unit on each day, again measured by the patient classification system. If on any given shift demand for staff on any unit is relatively low (for example because of low census or lower than usual patient acuity), excess staff are available to float other units. If the required hours of nursing care on any given shift exceeds the hours that are available from rostered staff on a unit, then ‘excess’ staff can be floated from another unit within the same broad specialty, defined by the hospitals organisational structure (e.g. general medical / surgical). If it is not possible to make up the shortfall with float staff, then temporary staff can be hired from a pool of internal ‘bank’ staff or an external agency. Bank staff include staff employed by the hospital but without fixed hours or assignments, or staff available to work voluntary overtime and so bank hours includes voluntary overtime.

The first alternative staffing plan aims to make more use of flexible staffing, and so has a lower number of staff on the baseline roster than the standard plan. We term this plan ‘flexible’ because the low baseline staff means that it is anticipated that most variation in demand would require the use of these flexible temporary assignments while baseline rosters are set to meet minimal routinely observed demands.

In the second alternative plan, rosters are set at a higher level than the standard plan, anticipating that most upward variation in demand can be met by the staff who are rostered. Although this plan also uses flexible staffing, we term this plan ‘resilient’ because the emphasis is on having enough staff available to meet variation in needs within the routinely scheduled rosters, without recourse to additional measures. See Table 1 for details.

Staffing scenarios tested in the simulation model.

2.2. Simulation model

We developed a simulation in the software AnyLogic researcher edition version 8.3.2. ( The AnyLogic Company, 2019 ). The simulation is an example of a Monte Carlo simulation since many input parameters (such as absence rates and demand for nurses) are stochastic, so are modelled as random variables following probability distributions. Fig. 1 shows the main simulation steps, and both a video of the simulation in action ( Saville and Monks, 2019 ) and a detailed model description following the STRESS reporting guidelines for agent-based simulations ( Monks et al., 2019 ) are available (see Supplementary material Appendix 1).

Flowchart of simulation steps adapted from Griffiths et al. (2020b ) with permission.

Daily demand for nursing care was simulated using the Safer Nursing Care Tool acuity dependency measure ( The Shelford Group, 2014 ). To provide parameters for our models, we undertook an observational study over one year (2017) in 81 general (adult) medical/surgical inpatient units in three hospitals, in London, South East and South West England ( Griffiths et al., 2020b ; Saville et al., 2021 ). Because estimates for mortality and the effect of low staffing on length of stay used in the present analysis (see below) applied to general medical and surgical populations, we excluded a specialist cancer hospital that participated in the parent study from this aspect of the study.

The base number of staff to be scheduled on each unit / shift was determined by taking a 20-day sample of daily patient acuity and dependency measures from each unit, as recommended when using the Safer Nursing Care Tool for establishment setting ( The Shelford Group, 2014 ). The number of full-time nurses to be employed (the establishment) was converted into daily hours, removing the uplift for annual leave, staff education and other time away from the unit. The daily staff hours were distributed across shifts based on the distribution observed on each unit. Similarly the skill mix between registered nurses and assistants was based on the observed skill mix for each unit.

In the simulation, units are the ‘agents’ that move between being understaffed, adequately staffed or overstaffed as supply of staff and demand from patients varies. The daily census and the acuity/dependency of patients, varied based on parameters derived from the observational study, was used to simulate variable demand for nursing care. Additionally we included one-to-one specialing requirements in the demand measure ( Wood et al., 2018 ), based on the observed use of such staff in each unit. To reduce computational complexity, we did not model the year in order but rather took a random draw from the demand data for each ward for each shift. Where units had experienced substantial changes that might necessitate different numbers of staff scheduled, e.g. extra beds in winter, we only used the data from the unit before the change.

Starting with a base number of staff rostered on each unit on each shift, the model simulates short-notice staff absence due to sickness. The model then attempts to address any staffing shortfalls relative to that day's demand from patients, firstly by redeploying (floating) staff from overstaffed to understaffed units, secondly by hiring staff from an internal bank of hospital employees and thirdly hiring from an external agency.

In all cases, registered nurses are substituted for registered nurses, and assistants are substituted for assistants. A threshold of 15%, corresponding to the tolerance used in the RAFAELA staffing tool, was used to trigger attempts to fill staff shortfalls ( Fagerström et al., 2014 ). We assumed rates of unanticipated absence through short notice sickness of 4% for assistants and 3% for registered nurses, approximating known differences in sickness absence between these groups ( Dall'Ora et al., 2019a ). Current national sickness rates for all nursing staff are 4.5% (Moberly, 2018), but longer term sickness absence can be anticipated and temporary staff or overtime added to the roster and so this was not considered separately. The availability of temporary staff was, initially, based on the empirical availability to fill short notice requests as reported by one of the participating hospitals. Temporary staff availability varied by time of day and day of the week but was always less than 50%, often considerably so (see supplementary material Table 5). Because such constraints are likely to be dependent on local labour market conditions, we also considered both a higher and unlimited availability of temporary staff as well as a scenario where there was no availability of temporary staff and the only flexible staff availability was floating staff from over staffed units (see Table 1 ). All allocations, hires and redeployments are subject to constraints about whole people, who must be deployed for a half/whole shift. For each shift, we calculated the achieved staffing relative to patient need.

2.3. Model validation

Throughout model development, we performed verification (checking correct implementation of the model in simulation software) and validation (checking that we built an appropriate and sufficiently accurate model). Full details of the validation are given in Griffiths et al. (2020b) . We worked closely with nurses with responsibilities for workforce at the participating hospitals, who agreed assumptions and sense-checked results. We also presented and discussed early versions of the simulation model and results with the project steering group, which included nursing research, mathematical modelling and nursing workforce experts. We tested the model's sensitivity to a number of key assumptions and found that neither staffing costs nor the rate of over / understaffed shifts appeared to be greatly affected by most assumptions such as the sickness / absence rates or the relative efficiency of temporary staff. The only assumption that made very large differences to the parameters estimated by the model related to the availability of temporary staff.

The estimated required staffing levels corresponded closely to the staffing actually deployed in the participating hospitals ( Griffiths et al., 2020b ). Estimated daily staff costs were similar (£140–150 per patient day) to actual costs. Since we assumed that hospitals were able to employ the staff needed to fill baseline rosters and did not consider staff rostered to cover longer term sickness, the level of temporary staff actually used in the participating hospitals was generally higher than in our models, as were the staff costs. We ran the model 10 times for each hospital and for a range of staffing scenarios, and calculated 95% confidence intervals around the means to assess the errors around the estimates. The confidence intervals were narrow, for example, for the standard staffing scenario, the intervals for costs were <£0.25 per patient day.

2.4. Staffing scenarios and cost effectiveness

We used the model to conduct a series of simulated experiments to explore the effect of different staffing plans on achieved staffing levels, the costs of staffing and patient outcomes by varying a number of parameters. The staffing plans and scenarios for temporary staffing availability are detailed in Table 1 .

We took a limited perspective on costs, focussing on nurse staffing costs and reduced resources in terms of bed utilisation. The costs were for 2017 and no discounting was applied because of the short-term time horizon. Prices are in Great British Pounds (£) sterling. Using the 2017 Organisation of Economic Cooperation and Development (OECD) purchasing power parities, £1 had equivalent purchasing power to $1.46 US ( https://data.oecd.org/conversion/purchasing-power-parities-ppp.htm ). Hourly employment costs for substantive staff at each pay band were estimated using the mean costs for each band reported in the Unit Costs of Health and Social Care ( Curtis and Burns, 2017 ). Salary and additional employer costs including pension contributions were included to derive a total cost for substantive and bank staff. Costs for agency staff were estimated using NHS guidance applicable to the study period, which set a cap on payment rates designed to reduce costs ( NHS Improvement, 2018a ). This therefore represents a low estimate. The daily unit staffing costs include allowances for working unsociable hours, in accordance with the Agenda for Change framework, using multipliers applicable to the study period ( Nursing Times, 2008 ). See supplementary material Table 6.

The effects of each alternative staffing plan on length of stay and the risk of death relative to the standard plan were estimated using regression coefficients from a recent longitudinal study of the effect of variation in nurse staffing on patient outcomes (see supplementary material Table 7) undertaken in one of the participating hospitals ( Griffiths et al., 2018 , 2019 ). This study was chosen because of the direct connection to staffing data for the current study and the robust longitudinal design using patient level exposure to variation in nurse staffing as the independent variable. All alternative estimates we identified were from cross-sectional studies or different countries. Although direct comparisons with other studies of patient level exposure with longitudinal design is not possible due to different measures of low staffing, the observed effects were broadly similar in magnitude (e.g. Needleman et al., 2011 ).

We used data from the ‘standard’ staffing plan to determine a unit mean staffing level against which low staffing was judged. We summed the days of low staffing across all units. We calculated the mean change in staffing levels associated with each staffing plan. As there is some evidence that using high levels of temporary staff (>1.5 h per patient day) can have an adverse effect on mortality ( Dall'Ora et al., 2019b ) we also summed days with high temporary staffing. From data for each unit on each day, we calculated the overall risk of exposure to low staffing and high temporary staffing. We estimated both staff costs and net costs after taking into account the value of bed days saved. The cost of changes in length of stay were estimated using the 2017/18 national average reference cost for a non-elective excess bed day (£337) ( NHS Improvement 2018b ), a likely conservative assumption as it assumes that there are no specific treatments costs associated with extended stay.

We identified mortality rates and average length of stay for the hospitals from published data ( NHS Digital 2018a ; NHS Digital, 2018b ) and used these as the assumed baseline for the standard staffing plan. We calculated the change in the number of deaths by subtracting the number of deaths associated with the ‘standard’ staffing plan from the number of deaths estimated in the new scenario ((Δexposure * Risk Ratio * baseline risk * population) – baseline death rate). We estimated ‘numbers needed to treat’ (NNT) or ‘numbers needed to harm’ (NNH) ( Cook and Sackett, 1995 ) associated with each alternative plan using the formulae NNT=1/ARR or NNH=1 /ARI (where ARR and ARI are the absolute risk reduction and increase respectively). These figures represented the number of patients who would need to be exposed to a given staffing plan (on average) to ‘save’ (or lose) one additional life. We calculated the incremental cost effectiveness of alternative plans (Δcost/Δmortality) relative to the ‘standard’ plan. Our primary analysis considered the effect of low staffing only on mortality. Our secondary analysis additionally considered the effect of exposure to high levels of temporary staffing.

Given the narrow confidence intervals for parameters produced by the simulation coupled with the computational time involved in running the models for multiple hospitals and multiple scenarios, data were generated from a single model run (365 days) for each of three different hospital configurations and we report results as an unweighted average of the three with incremental cost effectiveness ratios calculated as average change in outcomes / average change in costs. The underlying data generated by the model comprises 677,809 patient days and 29,565 unit days. We undertook sensitivity analyses to determine the impact of a number of parameters used in the economic model, including relative staff costs and the estimated effect of low staffing on mortality.

2.5. Approvals and permissions

The study was prospectively registered (ISRCTN 12,307,968), ethical approval was granted by the University of Southampton (ergo ID 18,809) and permission to undertake the research was granted by the Health Research Authority (IRAS ID 190,548).

When availability of temporary staff was limited to the empirically observed level, the estimated staff cost per patient day for the standard staffing plan was £133, with a mean achieved staffing level of 3.6 registered nurse hours and 3.5 nursing assistant hours per patient day. Across all units achieved staffing was, on average, 9% below the measured requirement, with 90% of wards between 2% understaffed and 16% understaffed on average. Distributions of both required and achieved staffing varied by unit, but typically showed a slight positive skew (median 0.6). Sixteen per cent of units (13/81) showed a strong positive skew (>1) in required staffing and 14% (11/81) showed a strong positive skew in achieved staffing. Only one ward showed a strong negative skew in achieved staffing. Both required (median 0.7) and achieved staffing (median 1.1) tended to positive kurtosis, meaning there were more values in the tails of the distribution (further from the mean) than expected under a normal distribution, with 52% (42) of units having kurtosis of >1 for achieved staffing.

The achieved staffing level was lowest for the low baseline flexible staffing plan and highest for the high baseline resilient plan ( Table 2 ) although differences between the plans reduced as the assumed availability of temporary staff increased. The achieved staffing levels and costs for the high baseline staffing plan were much less sensitive to changes in the availability of temporary staff than were the other plans.

Achieved staffing levels and daily costs for different staffing plans with varying temporary staff availability.

RN Registered Nurse, NA nursing assistant, HPPD hours per patient day.

Staffing plans with higher baseline staff (standard vs low baseline and high baseline vs standard) were associated with higher costs but shorter lengths of stay and fewer deaths ( Table 3 ). As the assumed availability of temporary staffing increased, differences in outcomes and costs between staffing plans reduced. Staffing plans with higher baseline staffing became more cost effective (less additional cost per improved outcome) as temporary staffing availability increased.

Changes in Costs, effects and cost-effectiveness for standard vs low baseline (flexible) and high baseline (resilient) vs standard plans for varying levels of temporary staff availability.

Where the availability of temporary staff was limited (based on the observed availability), the high baseline resilient staffing plan increased staffing costs by 5.5% whereas the low baseline flexible staffing plan reduced staff costs by 10.8% compared to standard plan. When the availability of temporary staff was unlimited, the high baseline plan was associated with a 2.9% increase in staff costs while the low baseline plan was associated with a reduction of 1.6%.

However, while lower baseline staffing plans were associated with reduced staff costs, they were also associated with worse outcomes because the achieved staffing levels were lower, despite the use of flexible deployments. With limited temporary staff availability, the high baseline resilient staffing plan was associated with a 1.2% reduction in the average length of hospital stay, and a 4.5% reduction in the relative risk of death, equating to one life saved for every 665 patients admitted to a hospital (number needed to treat). By contrast, with the low baseline flexible plan, many shifts were left critically understaffed, and consequently there was a 1.7% increase in the average length of hospital stay and an 8.3% increase in the risk of death, equating to one additional death for every 361 patients.

The outcomes for the low baseline flexible staffing plan were more sensitive to the availability of temporary staff than were those for the resilient plan. If no temporary staff were available, then deaths were increased by 13.4% with a flexible staffing plan (relative to the standard plan), whereas with unlimited temporary staff availability deaths were increased by only 1.9%. For the resilient plan the equivalent range reduction in death ranged from 4.5% with no temporary staff availability to 3% with unlimited temporary staff.

Compared to the low baseline flexible plan the standard staffing plan staff cost £ 21,766 per life saved when availability of temporary staff was limited. Much of the additional staff cost is offset by the value of reduced hospital stays. The net cost per life saved was £13,117. Similarly staff costs per life saved associated with the higher baseline resilient plan were £19,437 (compared to the standard plan). More than 50% of this cost was offset by the value of the reduced length of stay, leading to a net cost of £9506 per life saved.

Although the adverse effects of plans with lower baseline staffing were reduced with higher temporary staff availability, the relative cost-effectiveness of higher staffing was more favourable under these circumstances. For example, with unlimited temporary staff availability the net cost per life saved for the standard plan relative to the flexible plan is only £4250, while the cost per life saved for the resilient plan relative to the flexible plan was reduced to £3963.

Our primary analysis assumed that temporary staff are as effective as permanent staff and so adverse outcomes result from understaffing alone. There is some evidence that high levels of temporary staffing can have an adverse effect on patient outcomes and so we also considered these additional adverse effects of temporary staffing on mortality (see Table 3 ). The estimated mortality associated with lower baseline staffing was increased when considering an adverse effect from high levels of temporary staff. In the primary analysis higher availability of temporary staff tended to reduce the difference between plans and mitigate the adverse effects of lower baseline staffing. This was not the case when an adverse effect of high temporary staff was included. When unlimited temporary staff were available the low baseline flexible staffing plan was associated with a 15.1% increase in mortality compared to the standard plan.

Consequently, if a negative effect from high temporary staff was assumed, the cost effectiveness of higher baseline staffing was further improved, particularly when comparing standard staffing to the low baseline ‘flexible’ staffing plan and when temporary staff availability was higher. For example, the net cost per life saved for the standard plan compared the low baseline plan was only £515 per life saved (compared to £4250 in the primary analysis with no adverse effect from high temporary staff).

3.1. Sensitivity analyses

We undertook a series of sensitivity analyses. Although the pattern of results was largely unchanged by variation in model parameters, the magnitude of differences between staffing plans was sensitive to core parameters in the model, although these differences were generally unlikely to change substantive conclusions about cost effectiveness. Table 4 illustrates this by showing the change in net cost per life saved for the high baseline resilient staffing plan relative to the standard plan for limited and unlimited availability of temporary staff associated with alteration of some core parameters.

Effect of changing model parameters on net cost per life saved for high baseline resilient vs standard staffing plan with limited and unlimited availability of temporary staff.

The most significant sensitivity was the estimated effect of low staffing on mortality. Taking the upper bound of the 95% confidence interval for the mortality effect considerably reduced the cost per life saved, whereas taking the lower bound increased it. With limited availability of temporary staff net cost per life saved was increased by £18,377 if the low bound estimate of the effect on mortality was used.

In our original model, we assumed that bank staff were cheaper than permanent staff (because pension costs were reduced) and that agency staff were paid at the rate capped by NHS Improvement guidance. Changing these assumptions, including changing the assumptions so that bank and agency staff had similar costs (equivalent to sourcing all temporary staff from the bank) made most difference when unlimited availability of temporary staff was assumed and only small differences when availability was limited. A 25% increase in agency staff costs substantially reduced the cost per life saved associated with higher baseline staffing when temporary staff availability was unlimited.

If no floating of staff from overstaffed units to understaffed staffed units was permitted, the cost per life saved associated with the resilient staffing plan increased, although the magnitude of the difference was small relative to the impact of assumptions about the availability of temporary staff (see supplementary material Appendix 1).

The main model results are based on the unweighted average of models built with data and unit configurations of three different hospitals. Individual hospital results varied in their magnitude but the pattern of results for the relative costs or effects of different staffing plans and the impact of varying availability of temporary staff was generally consistent (supplementary material Table 9). However, for one hospital, the staff costs under the flexible staffing plan were marginally more expensive than for the standard plan if there was unlimited availability of temporary staff. In all cases the net cost per life saved for the resilient staffing plan was <£15,000 when temporary staffing was available.

4. Discussion

In this study, we undertook a series of simulation experiments to understand the relative cost-effectiveness of three staffing plans. The plans differed in the level of baseline nurse staffing planned on the roster (schedule) but all used flexible deployments to meet daily variation in demand. Although low baseline staffing and use of flexible staff deployment to meet peaks in demand is often assumed to be economically efficient, we found that staffing plans with higher baseline staffing appeared to be cost-effective compared to plans with lower baseline staffing.

Under a staffing plan with baseline staffing set at a low level, where the use of flexible staff to meet varying need was emphasised, we found that variation in demand could not be effectively addressed by floating staff from overstaffed units or hiring temporary staff whose availability is limited. Although staff costs were reduced when compared to a standard plan, economic efficiency was achieved in large part because many shifts were left understaffed, leading to adverse consequences for patient care. Patient deaths were increased. Because length of stay was also increased the net cost savings from reduced staff were much less than indicated by the saving on direct staff costs. As the availability of temporary staff increased, understaffing associated with the low baseline staffing plan was reduced but so were the cost savings compared to the standard staffing plan. Conversely, a higher baseline staffing plan, which emphasised resilience and was set to meet most peaks in demand, was more expensive but delivered better outcomes. Additional costs were partly offset by savings from reduced hospital stays.

In the face of ongoing nurse shortages, a flexible staffing policy that relies on a minimal core staff and makes heavy use of temporary assignments to deploy scarce staff to meet need seems highly appealing. The priority is, by implication, to avoid the apparent inefficiency of ‘overstaffing’, treating it as a risk to be avoided that is equivalent to understaffing ( Saville et al., 2019 ). However, our findings challenge a number of key assumptions underlying such an approach. Distributions of demand are not symmetrical for many units, and understaffing was far more common than overstaffing in both our simulation and our empirical observations ( Griffiths et al. 2020 ; Saville et al., 2021 ). With a low baseline, the risk of understaffing is higher and there is less likelihood that any given unit is overstaffed. Therefore fewer staff are available to float to understaffed units. Consequently, the demand for temporary staff from other sources (in this study an internal bank or an external agency) is increased, as is the risk that shifts remain unfilled.

Any strategy designed to balance the risk of overstaffing and understaffing must recognise that there is considerable ‘salvage value’ associated with an apparent excess of staff ( Davis et al., 2014a ). Any excess staff clearly have value because they could be deployed elsewhere, but when not redeployed they will still be adding value directly on their home unit even though apparently ‘surplus’ to requirements. This is because the criteria used to identify the required staffing level is not set at a level beyond which there is zero or a diminishing effect from adding additional staff. Instead, the target staffing level is defined to meet a subjectively defined level of quality ( Fagerström et al., 2014 ; The Shelford Group, 2014 ) and there is evidence that outcomes and quality of care improve further when staffing is at levels above those prescribed by current norms ( Fagerstrom et al., 2018 ; Griffiths et al., 2020b ; Junttila et al., 2016 ). Furthermore, higher nurse staffing is likely to contribute more value beyond that considered here, where we focus on a single effect, the risk of death. There are other benefits associated with higher staffing that are of value to patients, including avoidance of adverse events that fall short of death and improved patient experience (e.g. Bridges et al., 2019 ; Shang et al., 2019 ).

With many countries experiencing staff shortages, the goal of higher baseline staffing may seem simply unachievable. However, it is only when the availability of temporary staff is not limited at all that the low baseline flexible staffing plan appears to be viable in terms of avoiding a substantial negative impact on patient outcomes. Under these circumstances, the cost savings are largely eroded and the cost per life saved associated with staffing plans using higher baseline staff deployed on units is modest. Furthermore, if adverse effects from high levels of temporary staff are considered, adverse outcomes could still remain at a high level. Perhaps most striking of all in the context of labour market shortages, under these circumstances the average achieved staffing levels under the flexible staffing policy come close to those of the other policies and so there is little reduction in overal demand for staff. In our models, we assumed that temporary staff were equally as efficient as permanent staff. In some cases, for example where the flexible staff are staff from the home unit undertaking overtime, this assumption may be warranted but in most circumstances, this is unlikely to be the case ( Duffield et al., 2020 ). There is little indication that this flexible staffing plan achieves a more efficient deployment of a scarce staffing resource or substantially reduces overall demand for that scarce resource. Rather it may simply be an inefficient way of deploying much the same resource.

Superficially, our findings appear to contradict those of Kortbeek et al. (2015) and others who have modelled the benefits of flexible staffing. However, many previous modelling studies have been based on achieving particular nurse to bed or patient ratios without taking into account variation in need at the patient level or the impact on quality and outcomes ( Saville et al., 2019 ). Kortbeek et al. described a flexible staffing model based on hourly bed census predictions as ‘efficient’, although the productivity gains relative to a fixed staffing solution were modest for most of the scenarios considered ( Kortbeek et al., 2015 ). In our study all the scenarios considered were ‘flexible’ in so far as they all call on redeployments and temporary staff to meet varying demand. Our study focussed on the best approach to determining the baseline staff levels in the face variation and real life constraints on staff availability. Similar to Kortbeek et al. (2015) we show that floating of staff between units makes a modest but important contribution, which appears to improve the cost-effectiveness of higher baseline staffing levels. Crucially, neither Kortbeek's result nor ours can be used to support low baseline staffing levels as a means to achieve an efficient flexible staffing policy.

Even if the only consideration was reduced staff costs, savings from lower baselines are smaller than assumed because of increased hospital stays, which result from complications and delays in discharge preparation ( Needleman et al., 2006 ). Little saving is made when there is unlimited temporary staff availability, but outcomes remain worse in part because of the ‘trigger effect’ whereby short staffing has to be of sufficient magnitude to warrant an additional staff member being scheduled for a significant proportion of a whole shift.

Our findings are more consistent with those of Harper et al. (2010) , who modelled demand for nursing care with a similar measure to that used in the current study. They concluded that in order to minimize costs, a hospital should employ more nurses than the average need would indicate. We found that although a staffing plan with higher base staffing incurred additional staff costs, much of the increased cost was offset by savings from reduced hospital stays and such plans are potentially cost-effective. Differences may arise because Harper et al. (2010) considered only permanent and agency staff, whereas our models assumed that the first call would be on floating staff and then staff from the internal bank, who are potentially cheaper than permanent staff because of reduced contributions to pensions for what is, in effect, overtime. Furthermore, Harper et al.’s study preceded attempts to constrain what were regarded as excessive charges to the National Health Service for such staff in the UK ( NHS Improvement, 2018a ). Perhaps most significantly, Harper's study calculated the full cost of meeting need whereas our model included real world constraints, which meant that some shifts were understaffed. Understaffing associated with lower base staffing rosters accounts for most of the differences in staff costs between the staffing plans.

At the outset, we described a staffing plan with a higher baseline establishment as ‘resilient’ because it was designed to ensure that enough staff were rostered to meet predictable peaks in demand. A key characteristic of a resilient system is its ability to cope with stressors ( Berg et al., 2018 ). In this respect our findings show that the resilient staffing plan does indeed cope with the stress of varying demand more successfully than the alternatives. While our findings do not support the use of a flexible staffing plan with a low baseline staff roster, all our plans make use of flexible staffing to some extent. It is clear that some degree of flexibility has the potential to benefit patients and is likely to be superior to a fixed staffing plan and so flexible staffing does contribute to resilience.

Floating of staff between units in our simulation made a positive contribution to cost effectiveness. Other studies have highlighted that modest use of internal redeployments have the potential to be beneficial, although unconditional use of floating staff between units is problematic and may lead to poor outcomes ( Maenhout and Vanhoucke, 2013 ). The presence of staff who are unfamiliar with the unit means that such staff are unlikely to be as efficient as permanently assigned staff ( Duffield et al., 2020 ). The potential harms associated with use of temporary staff might be minimised by the use of dedicated float pools of staff who are specifically trained to work in a range of areas and rostered with the intention of being available to float between units, although there is limited direct evidence ( Dall'ora and Griffiths, 2018 ). Assuming that such float pools are themselves routinely rostered there is still a need to determine the home unit on which a staff member is deployed unless floated. The resilient staffing plan modelled here could provide the basis of such a float team system, with some staff from the float team rostered on to units to deal with the anticipated peaks in demand but available to be floated to other units if conditions permitted and required it. Depending on the overall level of demand, baseline rosters under such a system might be even higher, reducing the requirement for additional temporary hires.

Unless very pessimistic assumptions about life expectancy or utility of life gained were made for our estimates, it seems likely that the cost-effectiveness estimates for the cost per life saved for the resilient staffing plan would lead to costs per quality adjusted life year that sit well below generally accepted thresholds for defining cost-effectiveness ( Marseille et al., 2014 ). An intervention that costs less that the annual per capita gross domestic product per disability adjusted life year is regarded as highly cost effective ( Marseille et al., 2014 ). The UK per capita gross domestic product in 2017 was £39,977. If we assume that each death averted in the current model achieves at least one disability adjusted life year, higher baseline staffing was always cost effective, even under the most pessimistic assumptions about the effect of higher staffing on mortality.

Other more stringent thresholds have been suggested for defining acceptable cost-effectiveness ( Claxton et al., 2015 ). The National Institute for Health and Care Excellence, the independent body charged with evaluating costs and effectiveness for treatments to be provided by the publically funded universal National Health Service, categorises drugs that cost £10,000 per quality adjusted life year as providing ‘exceptional value for money’. Depending on the achieved life expectancy it is possible that higher baseline staffing meets even these more stringent thresholds for cost effectiveness. Additionally, if the potential adverse effects of temporary staff are considered, the economic argument for the higher baseline resilient staffing plan, with its lower use of temporary staff, becomes compelling.

On the other hand, because the ‘flexible’ staffing plan is associated with worse outcomes than the standard, it is easy to reject it simply because it is not an effective strategy. Given that the standard approach to determining baseline staff is the norm, the relative cost-effectiveness does not support a case for disinvestment and a move to a lower baseline. Under circumstances where the low baseline flexible staffing plan comes closest to providing an acceptable alternative, the relative cost-effectiveness of the standard staffing plan approaches the level at which the standard plan would dominate in terms of economic decision making, because the cost per life saved associated with it is so low. Indeed, in one of our hospital models the standard plan proved to be both cheaper and more effective.

4.1. Limitations

The use of simulation allows experiments about different staffing configurations on a scale that would be simply unfeasible in real life. However, although our results appear to be robust to variation in many of the assumptions in our model and the simulation was extensively validated, the results are, nonetheless, simulated. We considered only a limited range of costs and it is possible that adverse events associated with low staffing generate additional costs for the hospital, which would tend to further reduce the cost per life saved for staffing plans with higher baseline rosters. The pattern of results was consistent across the three hospitals in the model but the magnitude of the differences between staffing plans is sensitive to the hospital configurations and so cannot be generalised to other hospitals. The estimates of the effects of nurse staffing came from a robust longitudinal study undertaken in one of the hospitals that provided data for the models. While the evidence that nurse staffing plays a causal role in variation in patient outcomes seems compelling, unmeasured confounding and other shortcomings nonetheless potentially bias the results of observational studies. However, the substantive conclusions about the relative benefits and cost-effectiveness of higher baseline staffing were not altered when assuming much lower adverse effects of low staffing.

We did not directly model any efficiency or effectiveness loss associated with use of temporary staff, although there is some evidence that suggests that there is some loss of productivity and / or quality when temporary staff are deployed ( Dall'ora et al., 2018 ). Our secondary analysis considered the potential adverse outcomes associated with heavy use of temporary staff, producing results that were substantially more favourable for higher baseline staffing levels.

The underlying data on bank staff used to develop the models made no distinction between staff who were working voluntary overtime on their own unit and other staff employed through the bank, who would be less familiar with the host unit. Therefore we were unable to account for differences in these approaches to flexible staffing in our model. The adverse effect of high levels of temporary staffing we modelled may be somewhat sensitive to the mix of staff, although use of overtime has also been associated with adverse outcomes. However our results are not qualitatively altered by either this effect or the relative costs of staff groups. This limitation seems unlikely to change substantive conclusions.

As our model did not seek to predict patterns of demand, rather to estimate the average effect of a fixed baseline staffing level, we sampled data for each day at random and did not consider serial correlations. This simulates the average effect over a period with a fixed permanent roster with short term responses to varying demand by deploying temporary staff. If typical demand levels changed in the medium to long term, for example due to a seasonal pattern, baseline staff levels would need to be altered accordingly. Thus our model is consistent with guidance from the National Institute for Health and Care Excellence in England, which recommends a review of unit staffing plans at least twice a year, with additional reviews when changes such as patient case mix, which could alter demand, occur ( National Institute for Health and Care Excellence (NICE), 2014 ).

5. Conclusions

Flexible staffing plans that attempt to make the best use of a scarce nurse staffing resource by minimising the staff that are routinely rostered are likely to harm patients because temporary staff may not be available at short notice. Such plans are not efficient or effective solutions to nurse shortages. When sufficient temporary staff are available, there is little reduction in costs or the overall number of staff required, and so this approach does little to address nursing shortages.

A staffing plan using flexible deployments with a low number of staff on the baseline roster is not resilient, because it is unable to properly meet varying demand. In contrast, a plan with a higher baseline staffing set to meet predictable peaks in demand is both resilient and more flexible, because the ‘excess’ staff are productive. In the context of a sufficient baseline establishment, flexible staffing, including floating staff between units, can be guided by shift-by-shift measurement of patient demand, but proper attention must be given to ensuring that the baseline number of staff rostered is sufficient to meet at least average demand or higher. The apparent risk of overstaffing is unlikely to materialise because the additional staff contribute to flexible staffing when available to be floated between units and enhance quality and safety if they remain in the home unit. Staffing plans with higher baseline staff levels are highly cost effective.

CRediT authorship contribution statement

Aspden: Investigation, Resources; Ball: Conceptualization, Writing - Review & Editing, Funding Acquisition; Cassar: Investigation, Resources; Chable: Conceptualization, Resources, Funding Acquisition ; Dimech: Conceptualization, Resources, Funding Acquisition ; Griffiths : Conceptualization, Methodology, Software, Validation, Formal Analysis, Writing – original draft, Visualization, Project Administration, Funding Acquisition ; Hunter: Investigation, Resources; Jeffrey: Investigation, Resources, Funding Acquisition; Jones : Conceptualization, Methodology, Data Curation, Writing - Review & Editing, Funding Acquisition ; Lambert: Project Administration; Maruotti: Methodology, Funding Acquisition ; Monks: Conceptualization, Methodology, Writing - Review & Editing, Project Administration, Funding Acquisition ; Pattison: Conceptualization, Funding Acquisition, Resources; Recio Saucedo: Conceptualization, Investigation, Funding Acquisition ; Saville: Methodology, Software, Validation, Formal Analysis, Investigation, Data Curation, Writing – original draft, Project Administration ; Sinden: Conceptualization, Resources, Funding Acquisition

Data availability statement

Declaration of competing interest.

PG is a member of the National Health Service Improvement (NHSI) safe staffing faculty steering group. The safe staffing faculty programme is intended to ensure that knowledge of the Safer Nursing Care Tool (SNCT), its development and its operational application is consistently applied across the NHS.

Acknowledgments

The Safer Nursing Care Team study team comprise: Griffiths, Peter 1,2,3 *; Saville, Christina 1,2 ; Ball, Jane E 1,2 ; Jones, Jeremy 1 ; Monks, Thomas 4; Chable, Rosemary 5 ; Dimech, Andrew 6 ; Jeffrey, Yvonne 7 ; Mauruotti, Antonello 8 ; Pattison, Natalie 9,10 ; Recio Saucedo, Alejandra 1 ; Sinden, Nicola 3 Additionally, Aspden, Clare 5 , Cassar, Tracey 3 and Hunter, Shirley 7 contributed to data collection and Lambert, Francesca contributed project management and advised on patient and public involvement.

1 Health Sciences, University of Southampton, UK

2 National Institute for Health Research Applied Research Centre (Wessex), Southampton, UK

3 Portsmouth Hospitals University NHS Trust, UK

4 University of Exeter, UK

5 University Hospital Southampton NHS Foundation Trust, Southampton, UK

6 The Royal Marsden NHS Foundation Trust, London, UK

7 Poole Hospital NHS Foundation Trust, Poole, UK

8 Libera Università Maria SS. Assunta, Rome, Italy

9 University of Hertfordshire, UK

10 East and North Herts University NHS Trust, UK

An extended account of this study was published as part of the NIHR Journals Library as: Griffiths, P., Saville, C., Ball, J.E., Chable, R., Dimech, A., Jones, J., Jeffrey, Y., Pattison, N., Saucedo, A.R., Sinden, N., Monks, T., 2020. The Safer Nursing Care Tool as a guide to nurse staffing requirements on hospital wards: observational and modelling study. Health Serv Deliv Res 8, 16.10.3310/hsdr08160

This report presents independent research funded by the UK's National Institute for Health Research (NIHR) Health Services and Delivery Research Programme (award number 14/194/21). The views and opinions expressed in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the Health Services and Delivery Research Programme or the Department of Health and Social Care.

Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.ijnurstu.2021.103901 .

Appendix. Supplementary materials