Investigating the dynamics of nurse migration in early

0 downloads 0 Views 291KB Size Report
A longitudinal questionnaire survey of variation in regional retention of ... Information that does exist on migration patterns ...... other activities (Table 6, columns 3 and 4). Nurses in ..... system that can record job movements of all NHS staff.
ARTICLE IN PRESS

International Journal of Nursing Studies 45 (2008) 1064–1080 www.elsevier.com/ijns

Investigating the dynamics of nurse migration in early career: A longitudinal questionnaire survey of variation in regional retention of diploma qualifiers in England Sarah Robinson, Trevor Murrells, Peter Griffiths Nursing Research Unit, King’s College London, James Clerk Maxwell Building, 57 Waterloo Road, London SE1 8WA, UK Received 4 April 2007; received in revised form 25 June 2007; accepted 2 July 2007

Abstract Background: Increasing mobility of healthcare professionals has led to concerns that certain countries or regions are depleted of sufficient staff to meet healthcare needs. In formulating appropriate strategies to ensure better retention locally, human resource managers are hindered by lack of information about migration patterns. Purpose and aims: Purposes included studying movement of diploma nurses qualifying in England and contributing to literature on developing methods for obtaining migration data. Specific aims ascertained: regional variation in retention of locally trained nurses; associations between nurses’ profile and retention in training region; and impact on each region of inter-regional movement of nurses. Method: Questionnaires sent to a nationally representative cohort of adult branch nurses at qualification (n ¼ 1596) and at subsequent intervals thereafter provided data on all employment and other activities and geographical location of each. Event histories constructed from chart data were used to analyse length of retention in region of training and movements between regions. Retention was operationalised through developing the construct ‘engagement with nursing’. Results: Older entrants and those with children were more likely to nurse in their training region than younger and childless counterparts. Regions differed in retention of locally trained nurses and in the impact on their diplomate workforce of inter-regional movement. Regional variations were insufficiently explained by differences in nurses’ profiles; hence influences of regional characteristics were also considered. Conclusion: Retention strategies should include maintaining the policy of recruiting greater diversity of entrants, particularly mature entrants and those who have children. In developing local strategies, each region needs: information about retention of different components of their workforce; and an understanding of how regional characteristics can facilitate or constrain retention. National and international workforce organisations need to plan how best to obtain accurate and comparable nurse migration data. r 2007 Elsevier Ltd. All rights reserved. Keywords: Workforce; Employment; Migration; Employment mobility; Retention; Tracking systems

What is already known about the topic?

 Migration Corresponding author. Tel.: +44 020 7848 3057.

E-mail address: [email protected] (S. Robinson). 0020-7489/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijnurstu.2007.07.001

of nurses between and within countries leaves some areas depleted of sufficient staff, a situation exacerbated by global shortages of nurses.

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

 There is widespread national and international recog

nition of a lack of systems to obtain data to track migration patterns of nurses that is accurate, complete and comparable across local and national boundaries. Information that does exist on migration patterns and plans of nurses has been used to inform workforce-planning strategies.

What this paper adds

 The study adds to the international literature on nurses’  

migration patterns by providing hitherto unavailable information about the movements of diplomate nurses in early career from their training region in England. Retention in training region was positively associated with groups included in campaigns to increase diversity of entrants into nursing, in particular those who were older and those who had children. The study contributes to the development of methods to obtain accurate and comparable data on nurse migration through developing the construct ‘engagement with nursing’ and demonstrating the level at which information should be obtained about geographical location of workplaces.

1. Introduction Aspects of the increasing mobility of the healthcare workforce, nationally and internationally, are of growing concern; in particular, the adverse effects that may be created for regions and countries depleted of staff recruited to work elsewhere. Recommendations to ensure better distribution of staffing resources are long-standing, as is the recognition that this requires accurate and timely data, comparable within and across national borders. Such data are, however, notable by their inadequacy or absence. The study reported here focused on movement from region of training of diploma nurses qualifying in England. As well as contributing to the international literature on migration and providing specific information for UK workforce planners, the paper also considers problems that may be entailed in obtaining data about healthcare professionals’ patterns of migration. The aims of the study were formulated in the context of staff shortages, the development of strategies to counteract this problem and a specific UK dimension of concern about loss of healthcare staff from areas with high costs of living. 1.1. Nursing retention and recruitment strategies At a time of growing demand globally for healthcare, many countries are experiencing staff shortages

1065

(O’Brien-Pallas et al., 2005) and particularly of nurses (WHO, 2000; International Council of Nurses, 2002). These shortages, usually expressed as the proportion of posts remaining vacant within a given period (Zurn et al., 2005), often conceal regional and service differences (Buchan, 2000a). In England for example, nursing vacancy rates vary considerably by healthcare administrative region and particular services. While nursing shortages have been cyclical (International Council of Nurses, 2002; Antonazzo et al., 2003), it is thought unlikely that supply will again outpace demand (Buchan, 2000a). Moreover, modernisation of healthcare systems in countries such as the UK, the US, Australia and Canada depend on substantial increases in nurse numbers (Department of Health, 2000; Kline, 2003; O’Brien-Pallas et al., 2005) and improving retention and recruitment are policy priorities worldwide. In relation to the former, strategies include improving opportunities for professional development and the quality of working conditions. Success has been evidenced for example, in the US by improved retention rates in Magnet hospitals (Aiken and Havens, 1999) and in the UK by falling vacancy rates (Department of Health, 2001a; Health and Social Care Information Centre, 2005). Turning to recruitment, then strategies at national and sub-national level have been two-fold; attracting more people into nursing and attracting qualified staff practising elsewhere. In the UK, the former includes increasing the numbers of conventional entrants, typically female school leavers, but also targeting a wider group: men and mature entrants; members of ethnic minorities; those without traditional academic qualifications; and those who have worked in other occupations and/or raised a family (UKCC, 1986, 1987, 1999; NAO, 1992; Department of Health, 1999). Encouraging this wider group into nursing has been driven by declining numbers of conventional entrants, attracting groups thought more likely to stay, and having a workforce that better represents patient populations. Changes in workforce composition and comparative retention rates by demographic profile will indicate the success or otherwise of this strategy. The strategy of recruiting staff working in healthcare elsewhere has been facilitated by increased labour mobility generally (Diallo, 2004; O’Brien-Pallas et al., 2005; NHS National Workforce Projects, 2006) with international mobility of healthcare professionals in particular, emerging as a major feature of economic globalisation (Kingma, 2001; Bach, 2004; Diallo, 2004). Intensive recruiting campaigns, particularly for nurses, have caused concern, however, since some countries, or regions of countries, have been deprived of resources for their own needs with consequent loss of professional and financial investments made in healthcare education (International Council of Nurses, 2002; Kline, 2003;

ARTICLE IN PRESS 1066

S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

Diallo, 2004). Hence recommendations for planning initiatives to promote better long-term distribution of healthcare staff are long-standing (Mejia, 1978; Simmonds, 1989; Bach, 2004; Diallo, 2004; O’BrienPallas et al., 2005). Formulating initiatives that are appropriate, however, requires adequate statistical data. Information is needed on: region or country of origin from which workers originate; their numbers and demographic profile, career stage and employment status; length of time worked before moving; reasons for moving; destinations and length of stay in each; and subsequent moves. Most such information however is unavailable; either it is missing or it is incomplete, inaccurate and not comparable across local and national boundaries (Buchan, 2000a; Diallo, 2004; O’Brien-Pallas et al., 2005; Buchan and O’May, 1999; Buchan et al., 2006). Recommendations to address this problem include making better use of existing databases and developing permanent systems to trace staff movements (Mejia, 1978; Diallo, 2004; O’Brien-Pallas et al., 2005) with accompanying studies investigating reasons for migration and subsequent plans (Buchan, 2000a). Consideration has been given to what follow-up systems might entail, such as a unique identifier for each health professional and how their location might best be identified (O’Brien-Pallas et al., 2005). 1.2. Developing retention strategies through understanding trends in migration Existing data about migration from international and national organisations and from specific research projects, demonstrate how these can inform workforce planning. Various studies have indicated the extent to which regions retain the health professionals that they have trained. In Canada for example, analysis of national datasets showed that although over 80% of general and mental health nurses remained in their training province, nurses were moving away from rural to urban areas (Baumann et al., 2004). Strategies to counter this trend included financial and professional incentives to work in rural areas (Baumann et al., 2004). In the UK, research shows that London is a net exporter of healthcare professionals. A study by Parkhouse and Lambert (1997) of UK doctors qualifying between 1974 and 1993 showed that while Belfast retained 79% of its qualifiers and Glasgow 79%, this fell to 54% for Cardiff and still lower for London at 42%. More recent work (Buchan et al., 2003) has shown that London is also a net exporter of nurses and allied health professionals as well as doctors. Initiatives such as affordable housing schemes are being introduced to counteract the high living costs that drive people away from London and, increasingly, the South-East (Stock, 2000; Buchan, 2000b; Hutt and Buchan, 2005) and proposals have

been mooted for an improved cost of living allowance to attract health workers to the capital (Hutt and Buchan, 2005). Younger employees are generally regarded as more mobile than their older counterparts (NHS National Workforce Projects, 2006) and certainly research indicates higher turnover among the former; for example for all healthcare staff in the British NHS (Gray and Phillips, 1996) and for nurses in Ireland (Murray, 1999; McCarty et al., 2002). Yet research on international mobility suggests that the age profile of migrators may vary by country (Buchan et al., 2006); information that indicates those age groups at which particular countries might most usefully direct retention strategies. Understanding nurses’ plans about migration can also inform workforce planning. In the UK for example, 20% of registered general nurses trained in England plan to work abroad after qualifying but most intend returning to UK nursing and are not permanently lost (Murrells et al., 1996; Robinson et al., 2001). On the other hand, although 60% of overseas nurses based in London were planning to stay for 5 years, just under half (43%) were also considering a subsequent move elsewhere, most likely the US or Canada (Buchan et al., 2006). The latter finding contributed to recognition that contributions of overseas nurses to decreasing vacancy rates may not be sustained and that continuing focus on retention of UK qualifiers is essential (Royal College of Nursing, Queen Margaret University College and Edinburgh, 2004). Developing policies to ensure that national and local workforces provide a balance of experience, skill and potential longevity of service requires understanding of migration patterns of component groups in the workforce: established and temporary post holders; those at different career stages; and qualifiers from different courses (Buchan, 2000a; O’Brien-Pallas et al., 2005). In the UK, information to this effect for nursing is scarce. As noted, London is a net exporter of nurses (Buchan et al., 2003) and an earlier study showed that just under half the 1994–1996 qualifiers from the degree course at Birmingham University remained working in the West Midlands region in the early post-qualification period (Wheeler et al., 2000). 1.3. Studying the movement of diplomate nurses: purpose and aims The overarching aim was to quantify and understand nursing movements during early career; to ascertain the impact of nursing movements on each regions workforce and suggest explanations for these movements. By focusing on the movement from region of training of a nationally representative cohort of nurses in early career, this study has relevance for several stakeholders in the planning of the UK nursing workforce. Those who commission training places and invest in nurse

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

1067

education (Workforce Development Confederations) require information about: the proportion of nurses remaining within their borders; the length of time they stay; and whether retention varies by demographic profile. The movement of recently qualified nurses are of interest to those in Strategic Health Authorities who have responsibility for acute, elderly and general services; these have higher than average vacancy rates (4.6% vs. 3.9% for all specialties in England 2000) (Department of Health, 2001) but are the most likely destinations of newly qualified nurses. As noted at the outset of the paper, the study also sought to contribute to the wider literature on problems in tracing the migration patterns of healthcare professionals. Of particular note is the definition of retention in a manner that can be operationalised for research purposes; the construct developed here to this end is referred to as ‘engagement with nursing’. The specific research aims were to ascertain:

whether this indicated moving away from established nursing in the region by using other information; for example, agency nursing whilst seeking an F grade post was not regarded as ‘disengagement from nursing’, whereas agency nursing for a month before working abroad was thus regarded. If maternity or sick leave was the last event and proceeded by a nursing job, these events were included in the period of ‘engagement with nursing’ on the assumption of a subsequent return. A period of ‘engagement with nursing’ in region ended when an activity other than established nursing was started; for brevity referred to as other activities. These included: starting a nursing job in another region; leaving the UK; starting non-healthcare work or education; taking a break for childcare; or working as an agency/bank nurse for longer than 4 months. Length of time of each period of ‘engagement with nursing in region’ was calculated from dates provided for each activity.

1. Whether regions varied in the extent to which nurses ‘engage with nursing’ in their training region. 2. Whether nurses’ engagement with nursing in training region was associated with profile variables and whether such associations contributed to explaining regional variations. 3. The impact on each region of nurses moving between regions.

2.2. Identifying place of work

2. Operational definitions The research was undertaken by means of questionnaires sent to diplomate nurses between qualification and 3 years; details of design and methods are discussed in Section 3. Data on activities and geographical movement were obtained from career charts completed by each respondent at each time-point. Here we set out the two operational definitions developed for the study that enabled the above aims to be addressed using the information provided in the career charts. 2.1. Defining engagement with nursing The construct ‘engagement with nursing in region’ was developed to provide a meaningful and operational definition of retention in region. This was defined as a period of working continuously in established nursing jobs or holding such jobs interspersed with other activities not regarded as indicating ‘disengagement with nursing in region’. These included: full-time training and maternity leave followed by nursing, or periods of agency/bank nursing of 3 months or less between nursing jobs or training. When agency/bank nursing was a person’s last career chart event and was less than 4 months, we assessed

Tracking movements requires decisions about level of detail requested about places of residence or work and criteria for grouping details into larger units for analysis. If current administrative area is requested, problems arise if boundaries subsequently change. In England, administrative boundaries have long been counties for residence and regional health authority (RHA) for healthcare work but there have been several name and boundary changes to both. Respondents were therefore asked to name the town or city in which each activity was based since these rarely change. As interest focused on movement from training region, the town/city of each activity was grouped into the eight RHAs and inter-regional movement then analysed. Subsequently, these RHAs have been replaced by nine regional government offices. Comparisons revealed an exact match for the first five regions listed (London, South East, South West, Eastern and West Midlands) and a near match for the sixth (North West). The other two (Trent and Northern & Yorkshire), however were divided into three regional government offices (East Midlands, North East and Yorkshire & Humber); implications for interpreting findings are included in the discussion. 3. Research design and methods 3.1. Research design The research design was correlational and longitudinal (Robinson et al., 2003a, b). Subjects were surveyed prospectively from qualification onwards and at three subsequent time-points (6, 18 months and 3 years).

ARTICLE IN PRESS 1068

S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

3.2. Research methods 3.2.1. Sample selection The study population consisted of all individuals who qualified from the diploma programme in England in 1997–1998. This programme was and remains the predominant initial training in England. A multistage sampling approach was used to select nurses. Strata were formed from the eight RHAs that existed at the time. Regions contained a variable number of colleges. Half (or close to it) of all colleges were selected from each region with further sub-sampling of intakes from larger colleges. Sample size was based on information collected from the English National Board and from our own enquiries. We estimated that a sample of 2109 nurses was required which represented a third of those expected to qualify. More details are found in Marsland and Murrells (2000). 1832 (87%) agreed to participate, 1596 (76%) responded to the first questionnaire and 1055 (50%) at 3 years (includes tracer follow up of nonrespondents). There was some variation in response rates between regions that became more noticeable as time went on with lower rates in London and Eastern regions and higher rates in the South West and Trent. 3.2.2. Data collection Data were collected by questionnaire that covered a range of topics using both open and closed questions. A question was developed specifically for collecting information on career histories from which data on geographical movements were drawn. Start and end dates were requested for each activity since qualification. For nursing jobs location, employing organisation, specialty, grade and type of contract (established or temporary post) was requested. Instructions were provided about how to include periods of secondment, maternity or sick leave, unemployment and full-time study. Each participant had an identifying number in order to: construct individual event histories; link data about expectations with subsequent events; and send reminders to non-respondents. A large pilot group of 600 qualifiers (from the previous year) from the four branches of the diploma programme was used to develop the questionnaire starting with in-depth interviews, then testing drafts and piloting postal versions. Addresses were provided by nurses who attended face-to-face recruitment sessions prior to qualification. Regular contact was maintained with the cohort. Questionnaires were sent initially to the main address, then a second time to the same address if no response before sending to an alternative address (typically the parents) provided by participants. The final option if there was no response after the three mailings was to trace the nurse via the United Kingdom Central Council (UKCC) for Nursing, Midwifery and Health Visiting (now the Nursing and Midwifery

Council) and for the UKCC to send a questionnaire on our behalf. 3.2.3. Data analysis Mean time engaged in nursing before exiting to another region or activities other than established nursing was calculated using the Kaplan and Meier (1958) method for censored data. The Log-Rank test (Mantel–Cox) was used to test the equality of survival distributions across different categories (e.g. regions, ethnic group) (note the SPSS version 13 procedure is unable to utilise sample weights). Dichotomous (binary)-dependent responses (whether first nursing job was in region of training) were modelled using multivariate logistic regression and time to event data (time in engaged in nursing in training region before an exit to another region or non-nursing activity) were modelled using Cox regression. In each case, a reduced set of variables were selected that were significantly associated at the 10% level with the dependent variable. The backward and forward stepwise (Wald) method was used to determine which variables entered the ‘best’ model. Once the ‘best’ model was identified this was refitted to the data. This increased the analysis sample by incorporating respondents previously excluded because of missing data (on those variables that did not enter the model). To make maximum use of data available, longitudinal analyses (Aims 1 and 2) were based on the 1339 respondents returning at least the 6-month questionnaire. The 900 returning all three questionnaires, the 215 returning the first two but not the third, and the 222 returning the first only, formed the three categories of a new variable called respondent group that was included in logistic regression models to partly adjust for nonresponse. Those subsequently traced were only asked for details of activity at 3 years and so are only included in cross-sectional analyses (Aim 3). The analyses associated with each aim were as follows.

 Aim

1. 2. 3. 4. 5.

1: Regional variation in nurses’ engagement with nursing in training region Longitudinal analyses (Tables 1 and 2) demonstrated overall figures and regional variation in the proportion who: nursed continuously in region; nursed in region and left for nursing job elsewhere; nursed in region and left for activities other than established nursing jobs; started a first nursing job outside training region; and mean time spent in region before exiting to 2 or 3 above.

 Aim

2: Explaining variations in engagement with nursing in training region

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

100 90 80 70 60 % 50 40 30 20 10 0

6 mths

London

18 mths

South East

1069

3 years

South West Eastern West Midlands

Trent

North West

Northern & Yorkshire

Fig. 1. Proportion of nurses nursing in their training region at 6, 18 months and 3 years.



Univariate and multivariate analyses investigated associations of profile variables with 1–5 above (Tables 3 and 4). Profile variables included those discussed in the background section as contributing to increasing diversity of entrants, as well as route into nursing, plans and first job information. Regression modelling investigated whether variations between regions in the proportion of nurses whose first nursing job was in another region and variation in the length of time in first period in training region could be attributed to differences in profile variables (Tables 5 and 6). The former was modelled using logistic regression and the latter using Cox regression. Two Cox regression models were fitted; firstly, where the period ended with moving to a nursing job in another region, and secondly, where it ended with moving to other activities. Variables that relate to specific sub-groups are greyed out in Tables 5 and 6 when they do not apply to the group being analysed. For both types of response (first nursing job not in training region, time to exit) models were fitted to: all nurses, and to two sub-groups of nurses (those who had obtained their first nursing job when asked at qualification and started that job, and those who had a spouse/ partner at qualification). These two sub-groups were asked additional questions that were incorporated into the modelling. Note that none of the variables relating specifically to spouse or partner were significantly associated with time engaged in nursing ending in a move to another region or non-nursing activity. Aim 3: Impact on each region of movement of nurses between regions Cross-sectional data showed, for each region, the proportion of ‘home trained’ nurses still working in the region (Fig. 1). For each time-point, diplomate’s region of training was cross-tabulated by region in which working. For each region, this showed the overall difference between numbers who trained in the region and numbers working in the region i.e. an

indication of overall loss or gain on investment made in training (Table 7) and patterns of movements between regions. Change in Table 7 has been expressed in absolute numbers and as a percentage of responders who trained in the region. 3.3. Reliability and validity Extensive pilot work was conducted so that the definition of what constituted an event (to be included on the career chart) was clear and unambiguous. All career charts were inspected and inconsistencies were resolved sometimes by making reference to other information contained within the questionnaire. The last event from the preceding questionnaire was always written on the first line of the next questionnaire to orientate the nurse. 3.4. Ethical considerations This study pre-dated the requirement of MREC approval, guidance was followed from staff of the university from which students were recruited as to the internal procedures required for ethical approval. At no time were participants or Colleges identified. The only addresses held on local databases were those provided by the participant.

4. Findings Findings are presented for each aim. 4.1. Regional variation in diplomate engagement with nursing in training region The majority of nurses (64%) nursed continuously in their training region (Table 1). Others nursed in their training region and then moved to nurse in another region (8%) or into other activities

1070

Table 1 Nursing career history after qualification by region Training region

Engaged in established nursing in training region

Started nursing in training region before exit to Nursing job in another region

Activities other than established nursing

Never obtained a nursing job

Unable to classify

First nursing job not in training region

Total

No.

%

No.

%

No.

%

No.

%

No.

%

No.

%

London South East South West West Midlands Eastern Trent North West Northern & Yorkshire

104 124 60 78 91 79 163 162

55 53 63 69 62 65 79 68

14 21 11 8 11 11 8 16

8 9 12 7 8 9 4 7

18 27 17 6 12 7 9 9

10 12 18 5 8 6 4 4

1 8

o1 3

2 2

1 1

1

1

6 1 1 6

4 1 o1 3

2

1

49 51 6 21 27 23 25 42

26 22 7 18 18 19 12 18

188 233 94 114 146 120 207 237

100 100 100 100 100 100 100 100

Total

860

64

101

8

106

8

22

2

6

o1

244

18

1339

100

Table 2 Engagement in nursing in training region by region Training region

First nursing job in training region

Time engaged in established nursing in training region before exit to Another region

No. London (n ¼ 155) South East (n ¼ 260) South West (n ¼ 96) West Midlands (n ¼ 116) Eastern (n ¼ 141) Trent (n ¼ 129) North West (n ¼ 202) Northern & Yorkshire (n ¼ 233) All (n ¼ 1332) (w2, p) 7 d.f.

121 193 89 93 108 104 174 181 1063

%

95% CL

78 (72, 74 (69, 93 (88, 80 (73, 77 (70, 81 (74, 86 (81, 78 (72, 80 (78, (21.86, .003)

85) 80) 98) 87) 84) 87) 91) 83) 82)

Activities other than established nursing

Rank

Mean

95% CL

Rank

Mean

95% CL

Rank

3 1 8 5 2 6 7 4

32.74 33.15 33.14 34.12 32.84 32.92 35.00 34.07 33.60

(30.99, (31.92, (31.29, (32.65, (31.12, (31.30, (34.22, (33.05, (33.13, (10.89,

1 5 4 7 2 3 8 6

31.98 32.14 31.22 34.32 32.88 34.56 34.78 34.71 33.43

(30.19, (30.85, (29.14, (32.99, (31.20, (33.51, (34.01, (33.84, (32.95, (33.83,

2 3 1 5 4 6 8 7

34.49) 34.38) 34.98) 35.58) 34.55) 34.54) 35.77) 35.08) 34.07) .14)

33.77) 33.43) 33.29) 35.66) 34.55) 35.61) 35.56) 35.57) 33.90) o.001)

ARTICLE IN PRESS

%

S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

No.

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

(8%). Almost a fifth (18%) however, started nursing in another region. Some striking regional differences emerged; London (55%) and South East (53%) appeared least able to retain their nurses; North West appeared best able to do so (79%); and South West had a high proportion initially commencing work in the region (93%) but conversely the highest proportion of movers to a nursing job elsewhere (12%) and to other activities (18%). Survival distributions for time engaged with nursing in training region before taking a nursing job elsewhere did not differ significantly overall between regions (Table 2). Mean time (months) varied from 32.7 in London to 35.0 in North West. Survival distributions for time engaged in nursing before moving to other activities, however, differed significantly overall; mean time (months) varied from 31.2 in South West to 34.8 in North West. Rankings (Table 2) identified regions with highest rates of loss, either at point of first nursing job or after a period of nursing in training region. London consistently had the highest ranks (3,1,2) followed by South East (1,5,3) and Eastern (2,2,4). North West (7,8,8) had the lowest ranks. 4.2. Explaining regional variations in engagement with nursing in training region Associations between profile variables and retention in training region were investigated first. Statistical modelling then ascertained how simultaneous effects of profile variables affected regional estimates and identified key predictors of nursing retention overall and regionally. 4.2.1. Profile variables and retention in region Tables 3 and 4 show overall profile information and how different variables related to: proportion of nurses who took a first nursing job elsewhere; how long (mean time in months) nurses remained engaged in established nursing in their training region prior to moving to nurse elsewhere or to other activities. Variables statistically associated with at least one of the three outcomes at the 10% level of significance are shown. 4.2.1.1. Demographics. Men and women did not differ in the proportion starting nursing outside their training region. For those who did start nursing in their training region, the average time engaged with nursing before moving into other activities was shorter for men than women (p ¼ .046). Younger rather than older nurses were more likely to have a first job outside their training region (p ¼ .003; linear association po.001). For those who started nursing in their training region, a U-shaped relationship emerged for age in relation to time engaged in

1071

established nursing before moving into other activities (p ¼ .002). White Irish nurses had shorter periods engaged in nursing and this most likely reflects a decision to nurse in Eire. 4.2.1.2. Having a partner. Those stating that partner’s current geographical location did not affect their own plans were those most likely to have a first nursing job elsewhere (p ¼ .007). Those who had a partner, spent longer on average engaged in established nursing in their training region before moving to a nursing job elsewhere compared with those who did not (p ¼ .004); however, none of the variables concerned with specific effects of partner’s work were significant. Partner variables were not significantly associated with moving to other activities. 4.2.1.3. Children living with diplomate. Those with children living with them were more likely to contribute to nursing in training region than those without. They were less likely to take a first nursing job elsewhere (po.001). Of those who did start nursing in their training region, those without children spent less time doing so before departing to a nursing job elsewhere, than those with children (p ¼ .007). Those who did not have children stayed in training region for a shorter time on average than those with children before moving into other activities (p ¼ .014). The effect of children on work plans was not associated with engagement in established nursing. 4.2.1.4. Routes into nursing and pre-course education. Those for whom nursing was not their first choice of occupation were more likely to contribute to nursing in their training region than the other two groups (Table 4). These nurses were least likely to start nursing outside their training region (p ¼ .052) and those who did start nursing in their training region spent longest doing so before moving to other activities (p ¼ .053). Highest educational qualification, however, was not associated with any of the key outcomes. 4.2.1.5. Aspects of plans. Employment intentions were associated with first nursing job not in training region, without quite achieving statistical significance (p ¼ .060). Associations with time engaged with nursing, having started first nursing job in training region, were far weaker. Time planning to stay in first nursing job was not associated with whether first nursing job was in training region but was associated with length of time engaged in established nursing in training region before moving. Firstly, a longer time was spent before moving to a nursing job elsewhere by those who were unsure how long they would stay in their first job than other respondent groups (p ¼ .001). Secondly, a longer time was spent before moving to other activities by those

ARTICLE IN PRESS 1072

S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

Table 3 Engagement with nursing: demographics, partner and children Overall profile (n ¼ 1339)

First nursing job not in training region (n ¼ 244)

Started first job in region of training and exit to

Another region

Age (continuous) (w2, p) 1 df Age (ordinal) 20–22 23–24 25–29 30–34 35–39 40–44 45+ (w2, p) 6 df

Mean (SD)

Mean (SD)

26.9 (6.9)

25.3 (5.9) (16.23, .001) % 22 22 17 13 9 12 10 (19.62, .003)

% 36 18 20 9 9 5 2

Sex Female Male (w2, p) 1df

93 7

Ethnicity White British White Irish Other white nationality Black, Asian or Chinese (w2, p) 3 df

87 5 3 4

18 27 6 25 (11.07, .011)

61 38

17 20 (3.00, .083)

6 21 22 8 42

18 10 16 20 20 (9.18, .057)

12 26 61

18 24 14 (9.80, .007)

Partner Has partner Yes No 2 (w , p) 1df Effect of partner’s level of income (n ¼ 815) I could choose not to work I could choose to work part-time rather than full-time Means I have to work full-time Means I have to work at least part-time Has no effect (w2, p) 4 df Effect of current geographical location of partner’s work (n ¼ 815) Affects own plans Does not affect own plans Not influenced by partner’s work (w2, p) 2 df Children No Yes (w2, p) 1 df All

76 24 100

planning to stay longest in first job (p ¼ .002). This, to some degree, confirms the link between plans and actual behaviour. Those expressing uncertainty about plans at three years were more likely than the other groups to have

21 10 (19.92, o.001) 18

Mean (months)

Activities other than established nursing SE

34.09 32.75 (8.14, .004)

0.27 0.46

33.26 34.52 (7.18, .007) 33.6

0.30 0.38 0.24

Mean (months)

SE

33.73 32.57 31.95 34.14 34.68 35.24 34.83 (20.61, .002)

0.37 0.67 0.67 0.68 0.59 0.54 1.15

33.54 31.53 (4.00, .046)

0.24 1.21

33.63 29.73 32.67 33.43 (10.37, .016)

0.25 1.84 1.47 1.41

33.09 34.33 (6.07, .014) 33.43

0.30 0.38 0.24

started a first nursing job elsewhere (p ¼ .020) and those who stated that they would very likely be nursing in the UK at 3 years spent a significantly longer time engaged with nursing in their training region before moving to other activities (po.001).

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

1073

Table 4 Engagement with nursing: routes into nursing, plans and aspects of first nursing job

Routes in Nursing Nursing was first choice Nursing was not first choice Unsure 2 (w , p) 1 df Plans immediately after qualifying re. Employment in the UK (n ¼ 1322) Not take up paid employment Full-time paid employment and part-time study Part-time paid employment and part-time study Full-time paid employment Part-time paid employment Part-time paid employment and full-time study (w2, p) 5 df

Overall profile (n ¼ 1339)

First nursing job not Obtained first nursing job in training region when asked at qualification and started that job before exit to in training region (n ¼ 244) Another region Activities other than established nursing

(%)

(%)

51 24 24

18 15 22 (5.92, .052)

3 8 1 82 5 1

29 15 13 19 5 24 (10.58, .060)

Time planning to stay in first nursing job (n ¼ 1133) For rest of my life o6 months after qualifying 6 months after qualifying 46 months after qualifying Not sure (w2, p) 4 df

3 1 13 51 30

Likelihood of working in UK nursing at 3 years Unable to say at this stage Very unlikely Unlikely Quite likely Very likely (w2, p) 4 df

6 2 4 17 68

30 27 14 21 17 (11.72, .020)

Obtained job at qualification which was started Yes No 2 (w , p) 1 df

85 15

17 24 (5.82, .016)

90 6

18 10 (2.98, .085)

First nursing job working preference achieved (full or part-time) (n ¼ 1133) Working as preferred Not working as preferred 2 (w , p) 1 df First nursing job specialty preference achieved (n ¼ 1133) Yes No Did not mind 2 (w , p) 2 df

75 9 14

First nursing job in trust in which placements were based (n ¼ 1133) Yes No (w2, p) 1 df All

66 29 100

4.2.1.6. Aspects of first nursing job. Two aspects of first nursing job were significantly associated with first nursing job not in region of training; starting (first)

Mean (months)

26.81 33.10 33.40 All censored 34.59 (18.24, .001)

54 44 (252.73, o.001) 18

33.60

SE

4.55 0.73 0.37 0.36

0.24

Mean (months)

SE

33.40 34.28 32.53 (5.88, .053)

0.34 0.39 0.60

27.90 32.25 34.02 35.04 32.65 (17.40, .002)

4.00 0.83 0.32 0.94 0.51

28.94 30.03 28.71 32.54 34.25 (49.46, o.001)

1.76 2.66 1.63 0.68 0.24

33.00 33.63 34.70 (5.99, .050)

0.32 0.84 0.51

33.43

0.24

nursing job that was obtained at qualification (p ¼ .016) and first nursing job being based in the trust where placements were based. The latter is partly a question

ARTICLE IN PRESS 1074

S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

artefact since it was expected that those who took up this option would remain in their training region. Those who were working as preferred (full or part-time) were more likely to take a first nursing job elsewhere than those who were not (p ¼ .085). Not minding specialty of first nursing job was associated with longer engagement with nursing before moving to other activities (p ¼ .050). 4.2.2. Statistical modelling Age, ethnicity, children living with diplomate, routes into nursing, time planning to stay in first nursing job and likelihood of working in UK nursing at 3 years were

the emergent predictors of engagement with nursing in training region. When and under what circumstances this occurred is described in greater detail below. The only profile variable to enter all three logistic regression models was having children living at home (Table 5). Those with children were more likely to start a nursing job in their training region than those who did not. Starting the nursing job offered at qualification, ethnicity, and partner’s influence on geographical location emerged as important predictors depending on the sub-group analysed. Those who started the nursing job they were offered at qualification and

Table 5 Logistic regression models: first nursing job not in training region Variable

All b

SE (b)

Started job obtained at qualification

Had a spouse or partner

b

SE (b)

b

SE (b)

Respondent group 6 months 18 months 3 years (w2, p) 2 df

0.44 0.22 0.29 0.19 0.00 (7.87, .020)

0.47 0.20 0.00 (5.43, .066)

0.24 0.22

0.53 0.13 0.00 (3.08,.21)

0.30 0.27

Children No Yes (w2, p) 1 df

0.89 0.21 0.00 (18.20, .001)

0.91 0.00 (13.83, o.001)

0.25

0.83 0.00 (11.40, .001)

0.25

0.86 0.22 2.43 0.00 (13.00, .005)

0.35 0.45 0.92

0.44 0.77 0.00 (12.16, .002)

0.31 0.22

Obtained job at qualification which was started Yes 0.54 0.19 No 0.00 (w2, p) 1df (8.43, .004) Ethnic group White British White Irish Other white nationality Black, Asian or Chinese (w2, p) 3 df Effect of current geographical location of partner’s work Affects own plans Does not affect own plans Not influenced by partner’s work (w2, p) 2 df Training region London South East South West West Midlands Eastern Trent North West Northern & Yorkshire (w2, p) 7 df

0.50 0.39 1.01 0.11 0.09 0.11 0.25 0.00 (18.42, .010)

NagelKerke R2

0.072

0.24 0.24 0.45 0.30 0.28 0.29 0.28

0.10 0.31 1.46 0.09 0.06 0.11 0.44 0.00 (13.93, .052) 0.082

0.28 0.26 0.58 0.34 0.31 0.31 0.31

0.70 0.13 1.93 0.66 0.03 0.30 0.72 0.00 (25.68, .001) 0.126

0.32 0.31 0.73 0.46 0.35 0.43 0.40

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

respondents who were White British or from another white nationality (column 2), and those whose current geographical location was not influenced by their partner’s work were more likely to start their first nursing job in training region (column 3). Turning to relationships between profile variables, training region and time engaged in nursing before moving to a nursing job elsewhere, findings (Table 6) show that having a spouse or partner emerges as an important predictor for all nurses and for the sub-group of nurses who had obtained their first nursing job when asked at qualification (and started that job). Having children living at home was associated, but not significantly (p ¼ .088), in the model for those who started their first nursing job in their training region. Time planning to stay in first nursing job was strongly associated with time engaged with established nursing, before moving to a nursing job elsewhere for those who started the nursing job they obtained at qualification (in their training region). While not entirely unexpected, this again provides evidence that plans, to some extent, are operationalised. Finally, relationships were investigated between profile variables; training region and time engaged in nursing before moving to other activities. Both age (linear and quadratic components) and likelihood of UK nursing at three years were significantly associated with time in first nursing job for all nurses (column 3) and for nurses who said they had obtained first nursing job when asked at qualification and started that job (column 4). Movement into other activities peaks around age 28 and declines thereafter. Two additional variables were significant in the model for the sub-group who obtained first nursing job when asked at qualification and started that job (column 4). Those who did not mind which specialty they wanted to work in, were less likely to move into other activities, as were those who said they intended staying in their first nursing job for more than 6 months. 4.2.3. Regional variation remaining after removing profile variable effects Statistical modelling was used to produce regional estimates having accounted for the simultaneous effects of the profile variables. Regional estimates varied significantly in two of the three logistic regression models and two of the Cox regression models (time in first nursing job before moving to other activities). London and South East nurses were the most likely to start a first nursing job elsewhere and South West and North West the least likely. Region was not significantly related to moves into nursing into other regions (Table 6, columns 1 and 2) but was related to movement into other activities (Table 6, columns 3 and 4). Nurses in North West and Northern and Yorkshire took longer to move into other activities in comparison with other

1075

regions, whereas the shortest times were associated with London, South East and South West. Profiles variables typically had a minor impact on regional estimates. The one exception was the Cox regression model for time engaged with nursing before moving to other activities (Table 6, column 3). The effect was to reduce overall variability and for South West to move into second place behind South East. 4.3. Impact on each region of movement of diplomates between regions Fig. 1 shows that at 6 months, South West and North West regions had retained the highest proportion of their qualifiers at around 80%. London had retained just over 70% with South East and Eastern falling slightly below this figure. By 18 months and 3 years, London had only retained just under half its qualifiers. Retention was on the low side for South East and Eastern regions, while North West and Northern and Yorkshire regions retained just over 70% of their qualifiers. These findings, however, do not necessarily result in shortages of recently trained nurses, since recruitment from other regions could have compensated for losses. The cross-tabulation of nurses’ training region by region of work provided an overall position at the three timepoints for each region in terms of gain or loss (Table 7) and details of inter-regional movements. At 6 months, the three southern regions differed from each other: there was a loss for London (15), a small gain for South East (6) and a large gain (28) for South West. The three ‘middle’ regions (West Midlands, Eastern and Trent) each made a small overall gain. The two northern regions each suffered an overall loss, with Northern and Yorkshire the hardest hit (24 lost). The overall gain or loss figures for each region at 18 months and 3 years are shown in columns 2 and 3, respectively of Table 7 and reveal differences between regions in the way their relative positions changed over time. London’s position improved at 3 years, whereas that for South East worsened considerably. South West continued to enjoy an overall increase although at a slightly lower level at each time-point. The three ‘middle’ regions showed no change or small gains. North West had an improved position, although still a loss, while Northern and Yorkshire had a higher loss at 3 years than at 6 months. Largest flows from training region to region of work at 6 months were: London to South East (net gain 13), from South East to South West (+16) and from Northern and Yorkshire to Trent (+13); at 18 months: London to South East (+8), London to Eastern (+10), South East to South West (+17), West Midlands to South West (+8) and Northern and Yorkshire to Trent (+10); at 3 years: South East to South West (+11), Trent to North West (+7) and Northern and Yorkshire

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

1076

Table 6 Cox regression model: time engaged with nursing in training region before moving Variable

Time in first nursing job in training region before moving to Another region

Activities other than established nursing

Started first nursing job in training region

Started job obtained at qualification

Started first nursing job in training region

Started job obtained at qualification

b

b

b

SE (b)

b

SE (b)

0.55 (9.16,.002) 0.01 (9.96, .002)

0.18

0.47 (6.69, .010) 0.008 (7.44, .006)

0.18

SE (b)

SE (b)

Age Linear (w2, p) 1 df Quadratic (w2, p) 1 df Spouse or partner Yes No (w2, p) 1 df

0.43 0.00 (4.21, .035)

0.20

Children No Yes (w2, p) 1 df

0.47 0.00 (2.91, .088)

0.28

Time planning to stay in first nursing job o6 months after qualifying 6 months after qualifying 46 months after qualifying Not sure (w2, p) 3 df

0.003

0.49 0.00 (5.09, .024)

0.22

2.10

0.64

0.92

0.61

0.61

0.36

0.11

0.29

0.67

0.29

0.57

0.24

0.00 (12.49, .006)

0.00 (31.62, o.001)

Likelihood of working in UK nursing in 3 years Unable to say at this stage Very unlikely Unlikely Quite likely Very likely (w2, p) 4 df

1.43

0.34

1.30 0.52 1.46 0.30 0.58 0.25 0.00 (36.93, o.001)

First job specialty preference achieved Yes No Did not mind (w2, p) 2 df Trainining region London South East South West West Midlands Eastern Trent North West Northern & Yorkshire (w2, p) 7 df

0.35 0.34 0.47 0.14 0.32 0.26 0.73 0.00 (8.41, .30)

0.36 0.32 0.39 0.41 0.38 0.37 0.45

0.003

0.63 0.47 0.63 0.41 0.47 0.53 0.75 0.00 (10.04, .19)

0.40 0.37 0.44 0.44 0.44 0.41 0.53

1.28 1.48 1.31 0.56 1.01 0.41 0.25 0.00 (28.13, .001)

0.42 0.38 0.42 0.53 0.44 0.49 0.48

1.40

0.36

1.34 1.43 0.59 0.00 (6.24, .044)

0.61 0.32 0.26

1.01 0.79 0.00 (9.93, .019)

0.41 0.51

1.10 1.31 1.38 0.40 0.83 0.16 0.01 0.00

0.44 0.41 0.44 0.58 0.47 0.52 0.52

(26.22, o.001)

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080 Table 7 Net gain or loss at 6, 18 months and 3 years by region Training region

Net gain or loss of diplomates at 6 months

18 months

3 years

No.

No.

No.

%

%

%

London 15 8 18 15 5 5 South East +6 +3 +3 +2 12 8 South West +28 +32 +26 +40 +24 +37 West Midlands +1 +1 +2 +2 +3 +4 Eastern +4 3 0 0 2 2 Trent +1 +1 0 0 0 0 North West 11 5 8 5 3 2 Northern & Yorkshire 24 5 18 10 30 18

to Trent (+14). These flows confirm the overall net gains in Table 7 for South West and net losses for Northern and Yorkshire, whereas for Trent gains from Northern and Yorkshire were made up by losses to several other regions.

5. Discussion Implications of the findings for workforce planning are considered first. These are followed by challenges in developing systems to trace migration, as indicated by strengths and limitations of the methods employed in this study. 5.1. Implications of migration patterns in early career By devising a career chart covering activities since qualification and developing the construct of ‘engagement with nursing’, we were able to trace movements of diplomate nurses in early career. Overall patterns varied by profile and region and both have implications for regional retention strategies. 5.1.1. Variations in migration patterns by profile Findings suggest that recruiting greater diversity of entrants to nursing, particularly those who are older and have families, should be encouraged. Older nurses were less likely than their younger counterparts to obtain their first nursing job elsewhere and to move into other activities. Those who had a partner spent longer on average in their training regions before moving elsewhere. Those with children compared to those without, spent longer on average nursing in training region before moving to other activities and were less likely to start their first nursing job elsewhere. For this group, moving geographical location may have been resisted so as not to destabilise schooling, whereas younger nurses without

1077

partners and/or children felt free to move as desired. The small number of men with children in the cohort limited investigation of the combined effects of gender and children on mobility. Although universities are increasing the proportion of mature entrants into nursing courses (Hutt and Buchan, 2005) staff should ensure that this group are supported through training to successful qualification. Subsequently, employers need to ensure that career aspirations of older entrants are recognised and supported (Hutt and Buchan, 2005), especially as these may only emerge as confidence develops (Robinson and Bennett, 2007). Uncertainty about nursing in the future was associated with moving out of training region or into other activities and it is suggested that regular discussions about career plans and reasons for leaving might identify strategies that would encourage regional retention. That nurses who were flexible about the specialty of their first nursing job were less likely to move into other activities suggest that career guidance sessions should emphasise that specialty of first job does not necessarily determine future directions and that experience in diverse specialties can contribute to career progress. 5.1.2. Variation in migration patterns by region Regions varied considerably in migration patterns of their nurses but associations between nurses’ profile with retention in region of training were insufficient to explain these variations. The study was limited in that nurses’ reasons for staying in or leaving their region were not explored: partly an outcome of basing the research on an existing project with a wide brief that precluded such questions. Reasons for migration identified in other studies may, however, illuminate regional differences revealed here. Professional advancement is a major reason for national and international migration of health professionals generally (Diallo, 2004) and for nurses in particular (WHO, 2000; Kingma, 2001; International Council of Nurses, 2002; Buchan et al., 2006; Kline, 2003). The specialist facilities available in London’s wide range of teaching hospitals make the capital a popular destination for taking nurse training and for gaining post-qualification experience. Specialist experience available in cities such as Sheffield and Nottingham may have contributed to the move out of Northern and Yorkshire to Trent while cities such as Bristol may have played the same role in attracting nurses from all regions to move to the South West. Quality of life may also be influential in nurses’ movements. For some, the social and cultural life of large cities may have been a magnet. Family networks may be important and high retention of home qualifiers in Northern and Yorkshire, North West and South West may reflect strong family ties associated with these

ARTICLE IN PRESS 1078

S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

regions. Quality of life is affected by living costs and perceptions that accommodation may be more affordable in the Northern regions may have contributed to retention of home qualifiers, while high costs in London and South East may have had the opposite effect. It remains to be seen whether the various initiatives introduced, such as key worker housing schemes will encourage newly qualified staff to remain in high-cost regions, particularly London. Regions also varied considerably in the extent to which nurses movement between regions resulted in each having a net gain or loss in relation to the number trained (Table 7). The net loss for London supports findings from studies of healthcare professionals as a whole (Buchan et al., 2003) and of medical staff specifically (Parkhouse and Lambert, 1997). That the two northern regions (North West and Northern and Yorkshire) should also emerge as net losers of nurses was unexpected. Although figures for net gain or loss of nurses are not the same as vacancy rates, these two regions had lower than average (acute, elderly, general) vacancy rates for their nursing workforces (2.0% and 2.8%, respectively vs. 4.6% for England in 2000) (Department of Health, 2001). For a region to be a net loser of nurses suggests that investment made in nurse education is not returned and highlights the importance for each region of understanding migration patterns of the separate components of their workforces and tailoring retention strategies accordingly (Gould, 2006). 5.2. Implications of methods for developing nurse tracking systems Others (e.g O’Brien-Pallas et al., 2005) have considered aspects of follow-up systems such as unique identifiers and decisions about information to be obtained on location. These and other aspects of such systems were addressed in this study and highlight the various challenges that they present. Allocating an identifier to each diplomate enabled tracing of movements, linking of plans to subsequent events and sending reminders to non-respondents. Trust had to be established over measures to ensure confidentiality, anonymity and security of data before agreement to participate could be secured; and was facilitated by personal contact with each participant. This might be more difficult to achieve on a larger scale. Moreover, the question of voluntary versus compulsory participation would have to be addressed. Although defining retention in a specific location might appear straightforward, this study revealed its potential complexity. We developed the construct ‘engagement with nursing’ in this respect although accept that other researchers may have developed a different definition when faced with the same problem. If

different approaches are published and discussed it may prove possible to develop a common definition. Benefits of comparable findings were revealed here in that by applying the same definition to event histories of nurses in every regional health authority in England, regions could be compared and inter-regional movements analysed. This study demonstrated the importance of obtaining information about place of residence/work at a level that is unlikely to change and then grouping these into larger units for analysis. Currently, the best strategy for a national follow-up system may be to use postcodes since these are least likely to change. The change from Regional Health Authority to government office had little impact on interpreting findings other than if movements in and out of Trent and Northern and Yorkshire continued, East Midlands would likely benefit at the expense of North East and the northern part of Yorkshire and Humber. 5.3. Contributions and limitations This study has added to the knowledge base of internal migration and to our understanding of nurse migration in early career. Much of the work to date has focused on large geographic countries (e.g. Canada, Australia) or in the developed world where migration is typically from rural to urban areas (e.g. Baumann et al., 2004), whereas the overall tendency here was for movement in the other direction (e.g. away from London and the South East). There has been little research on migration of nurses immediately after qualification and what there is has been quite small scale and local (e.g. Wheeler et al., 2000). This is the first national survey of newly qualified nurses that has used event history analysis to investigate movement geographically and between different types of events (e.g. agency, non-healthcare jobs etc). The British Household Panel Survey lends itself to event history analysis however the sample of nurses is small (Gage, 2001) and not all nurses are newly qualified. The sample was representative of those qualifying with a diploma, the majority entry-level qualification into the profession. It is possible that those qualifying with a bachelor’s degree are subject to different factors, which influence their movements. However, it is more likely that they are subject to the same factors although the demographic profile of this group may differ and may comprise a higher proportion of the more mobile nurses identified here. Thus, the study identifies the implication of changes in initial qualification and differing recruitment strategies in the future. The response rates achieved in this study involved substantial administrative work; nonetheless the study was limited by attrition increasing over time and differing by region. Since non-respondents were more

ARTICLE IN PRESS S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

likely to have left nursing, this might have affected estimates for loss to nursing overall and varying regional response rates might have caused bias for some regional estimates. Using methods incorporating right censoring, adjusting for non-response patterns and including variables that predict non-response helped reduce some of these biases. The study was limited by resources in that it was only possible to cover the first 3 years after qualification and it is unknown how long observed patterns subsequently persisted. Our experience indicates that national follow-up systems would require substantial administrative and data analysis resources. One particular definition of attrition was developed for this study. We have started a process that hopefully will lead to a common definition or set of definitions of attrition that will allow for comparability of findings across studies. 5.4. Recommendations We believe that a system that tracks nurses (and other healthcare professionals) is essential for workforce planning. To date these systems have not existed apart from the Nursing and Midwifery Council database. Nurses are required to re-register every 3 years and therefore from a workforce planning perspective information is incomplete. This database has been used to monitor international nurses becoming eligible to practice in the UK. The new Electronic Staff Record (ESR) for England has raised the bar. We now have a system that can record job movements of all NHS staff between NHS organisations. By April 2007 the Electronic Staff Records (ESR) (http://www.esrsolution.co.uk/) system had been rolled out to over 350 NHS organisations and will eventually be rolled out to the remaining 250 plus NHS organisations. ESR is an integrated human resources and payroll systems that will replace 67 different systems that currently or previously existed and that will enable a consistent approach to NHS staff administration. Staff can update their own records, request training courses and progress their careers through Personal Development Plans.

6. Conclusion Effective systems to trace and understand nurses’ movements will not solve worldwide nurse staffing shortages; that can only be done through improving recruitment and retention. Such systems, however, provide information about where and why nurses move and this, in turn, may contribute to national and regional retention strategies. This paper has begun to build a picture of the movements of diplomate nurses in England in early career and as such may be of interest to UK workforce planners. The methodological issues

1079

raised, however, are likely to be relevant to workforce planners in the UK and elsewhere concerned with migration and retention of all healthcare professionals.

Acknowledgements Thanks to members of the pilot cohort for help in developing questionnaires, members of the main cohort for completing them, and past team members (Rachel Hardyman, Gary Hickey, Louise Marsland and Alison Tingle) for contributions to the study. This work was undertaken by the Nursing Research Unit, which receives funding from the Department of Health. The views expressed in the publication are those of the authors and not necessarily those of the Department of Health. Disclaimer: This is an independent research publication commissioned and funded by the Policy Research Programme in the Department of Health. The views expressed are not necessarily those of the department.

References Aiken, L., Havens, D., 1999. Shaping systems to achieve desired outcomes. Journal of Nursing Administration 29 (2), 14–20. Antonazzo, E., Scott, A., Skatum, D., Elliott, R., 2003. The labour market for nursing: a review of the labour supply literature. Health Economics 12, 465–478. Bach, S., 2004. Migration patterns of physicians and nurses: still the same story? Bulletin of the World Health Organisation 82 (8), 624–625. Baumann, A., Blythe, J., Kolotylo, C., Underwood, J., 2004. Mobility of Nurses in Canada. The Nursing Sector Study Corporation, Ottawa, Canada. Buchan, J., 2000a. Planning for change: developing a policy framework for nursing labour markets. International Nursing Review 47 (4), 199–206. Buchan, J., 2000b. Helping nurses bridge the house price gap. Employing Nurses and Midwives 45, 11–12. Buchan, J., O’May, F., 1999. Globalisation and healthcare labour markets: a case study from the United Kingdom. Human Resources for Development Journal 3 (3), 34–47 (WHO, Geneva, Switzerland). Buchan, J., Finlayson, B., Gough, P., 2003. In Capital Health? Meeting the Challenge of London’s Healthcare Workforce. The King’s Fund, London. Buchan, J., Jobanputra, J., Gough, P., Hutt, R., 2006. Internationally recruited nurses in London: a survey of career paths and plans. Human Resources for Health 4 (14), 1–10. Department of Health, 1999. Making a Difference; Strengthening the Nursing, Midwifery and Health Visiting Contribution to Health and Healthcare. London Department of Health. Department of Health, 2000. The NHS Plan, A Plan for Investment, A Plan for Reform. The Stationery Office, London (Cm 4818-1).

ARTICLE IN PRESS 1080

S. Robinson et al. / International Journal of Nursing Studies 45 (2008) 1064–1080

Department of Health, 2001. Department of Health Vacancies Survey 2000. Department of Health, London. Diallo, K., 2004. Data on the migration of healthcare workers: sources, uses and challenges. Bulletin of the World Health Organisation 82 (8), 601–607. Gage, H., 2001. Keeping nurses nursing: a quantitative analysis. Nursing Times 97 (7), 35–37. Gould, D., 2006. Locally targeted initiatives to recruit and retain nurses in England. Journal of Nursing Management 14 (4), 255–261. Gray, A., Phillips, V., 1996. Labour turnover in the British national health service: a labour market analysis. Health Policy 36 (3), 273–289. Health and Social Care Information Centre, 2005. Vacancy Survey for 2005. Health and Social Care Information Centre, London. Hutt, R., Buchan, J., 2005. Trends in London’s NHS Workforce: An Updated Analysis of Key Data. The King’s Fund, London. International Council of Nurses, 2002. Career Moves and Migration: Critical Issues. International Council of Nurses, Geneva, Switzerland. Kaplan, E.L., Meier, P., 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53 (282), 457–481. Kingma, M., 2001. Nursing migration: global treasure hunt or disaster in the making? Nursing Inquiry 8 (4), 205–212. Kline, D., 2003. Push and pull factors in international nurse migration. Journal of Nursing Scholarship 35 (2), 107–111. Marsland, L., Murrells, T., 2000. Sampling for a longitudinal study of the careers of nurses qualifying from the preregistration diploma course. Journal of Advanced Nursing 31 (4), 935–943. McCarty, G., Tyrell, M., Cronin, C., 2002. National Study of Turnover in Nursing and Midwifery. Department of Health and Children, Dublin, Ireland. Mejia, A., 1978. Migration of physicians and nurses: a worldwide picture. International Journal of Epidemiology 7 (3), 207–215. Murray, B., 1999. A Report on the Analysis of the Pattern of Nurses Entering and Exiting Employment in Three Dublin Maternity Hospitals 1996–1998. Health Service Employers Agency, Dublin, Ireland. Murrells, T., Robinson, S., Marsland, L., 1996. Nursing abroad: where, when and for how long? Nursing Management 9 (2), 21–22. National Audit Office, 1992. Nursing Education: Implementation of Project 2000 in England. HMSO, London. NHS National Workforce Projects, 2006. A Guide to Labour Market Information and Intelligence for NHS Workforce Planners. NHS National Workforce projects, Manchester, UK.

O’Brien-Pallas, L., Duffield, C., Tomblin-Murphy, G., Birch, S., Meyer, R., 2005. Nursing Workforce Planning: Mapping the Policy Trail. International Council of Nurses, Geneva, Switzerland (Issue paper 2). Parkhouse, J., Lambert, T., 1997. Home, training and work: mobility of British doctors’. Medical Education 31 (6), 399–407. Robinson, S., Bennett, J., 2007. Career Choices and Constraints: Influences on Direction and Retention in Nursing. Nursing Research Unit, King’s College London, London. Robinson, S., Marsland, L., Murrells, T., Hardyman, R., Hickey, G., Tingle, A., 2001. Adult Branch Diplomates: Starting Out. Nursing Research Unit, King’s College London, London. Robinson, S., Murrells, T., Cox, S., 2003a. Adult Branch Diplomates: The First Six Months. Nursing Research Unit, King’s College London, London. Robinson, S., Murrells, T., Hickey, G., Clinton, M., Tingle, A., 2003b. A Tale of Two Courses: Comparing Careers and Competencies of Nurses Prepared via Three-Year Degree and Three-Year Diploma Courses. Nursing Research Unit, King’s College London, London. Royal College of Nursing, Queen Margaret University College, Edinburgh, 2004. Fragile Future? A Review of the UK Nursing Labour Market in 2003. Royal College of Nursing, London. Simmonds, S., 1989. Human resource development: the management, planning and training of health personnel. Health Policy and Planning 4 (3), 187–196. Stock, J., 2000. Effect of house prices on recruitment tackled. Employing Nurses and Midwives 46, 12–13. UKCC, 1986. Project 2000: A New Preparation for Practice. United Kingdom Central Council for Nursing, Midwifery and Health Visiting, London. UKCC, 1987. Project 2000, The Final Proposals: Project Paper 9. United Kingdom Central Council for Nursing, Midwifery and Health Visiting, London. UKCC, 1999. Fitness for Practice: The UKCC Commission for Nursing and Midwifery Education. United Kingdom Central Council for Nursing, Midwifery and Health Visiting, London. Wheeler, H.H., Cross, V., Anthony, D., 2000. Limitations, frustrations and opportunities: a follow-up study of nursing graduates from the University of Birmingham, England. Journal of Advanced Nursing 32 (4), 842–856. World Health Organisation, 2000. Nursing and Midwifery Services Facing Crisis, Experts Say. WHO, Geneva, Switzerland. Zurn, P., Dolea, C., Stilwell, B., 2005. Nurse Retention and Recruitment: Developing a Motivated Nursing Workforce. Global Nursing Review Initiative. Issue 4. World Health Organisation and International Council of Nurses, Geneva, Switzerland.