Changes in turnover and vacancy rates of care workers in England from 2008 to 2010: panel analysis of national workforce data

Introduction

It is frequently observed that changing demographics and an ageing population are increasing the demand for social care while traditional pools of social care staff are shrinking (Hussein & Manthorpe 2005). At the same time, the need to improve the quality of social care services has led to calls for more staff to be trained (Cavendish 2013). The combination of growing demand and higher expectations needs to be set in the context of fears about the sustainability of recruitment and retention of front-line staff (Skills for Care 2014). In the United Kingdom (UK), concern is long-standing about high vacancy and turnover rates in all settings of the care sector, particularly in relation to front-line staff, such as care workers (Donoghue 2009). In 2011, a survey conducted by the National Care Forum (NCF) (voluntary sector care homes in the main) of its member organisations found that over 40% of care home front-line staff had left their job within a year of taking up their post (National Care Forum (NCF) 2011). High vacancy and turnover rates have been associated with employment context, such as sector of employment (private vs. public) and employment conditions, such as paid sick leave and health insurance (Loan-Clarke et al. 2010, Rosen et al. 2011). The type and nature of setting and the way work is organised may also affect the impact of organisational characteristics on individual workers. For example, in domiciliary or home care, the lack of a fixed workplace means that levels and effects of management and co-workers are likely to be different from those within care homes.

In England, direct care workers comprise 72% of the care sector's workforce, an estimated 1.2 million workers (Skills for Care 2010). Overall, pay rates are among the lowest in the UK with the majority of wages on or near the National Minimum Wage (Hussein 2011a, Low Pay Commission 2011); this is particularly the case in the growing home-care sector (Rubery et al. 2011, Bessa et al. 2013). In England, previous analysis of workforce data revealed that while vacancy rates among direct care staff were relatively low in the private or commercial sector, turnover rates were highest among private care providers, which might be related to pay levels (Hussein 2009).

Several models of staff turnover exist in the economic, human resource and organisational literature. Many highlight the high association of turnover with individual job satisfaction (e.g. Parsons et al. 2003). Organisation and work environment factors, such as management styles, team support and working conditions, play important roles in staff retention (Castle & Engberg 2006). Some of the most commonly cited organisational predictors of turnover include staffing levels (including organisation size), senior management turnover and profit status (public, private and not-for-profit or charitable). Workload level, pay and working conditions are associated with the profits expected by different organisations. The same factors are also linked with care worker job satisfaction and intention to leave. However, United States' (US) research shows that effects of these factors on turnover vary by type of services provided and local economic conditions (Castle 2008, Donoghue 2009).

The consequences of turnover and vacancies among care workers, whether in care homes or home-care services, are extensive (Ball et al. 2013). The economic costs alone of high turnover and vacancy rate are considerable if training, loss of skills and experience, and recruitment and induction costs are taken into account. High levels of vacancies and turnover rates also impact on service quality in different ways. The most apparent one is the level of risk associated with an understaffed service or with staff who do not have enough experience or skills to meet users' needs. In the US, Castle and Engberg (2005) showed a significant association between different indicators of quality of care and turnover rates of nursing aides (generally equivalent to care home workers in the UK).

Methods and data

The National Minimum Data Set for Social Care (NMDS-SC), launched in 2007, gathers standardised workforce information for the social care sector in England. It provides a ‘minimum’ set of information about services and staff across all service user groups and sectors within the social care sector in England (establishment data set). Employers provide information about individual staff members offering a detailed picture of the workforce (individual workers data set). The data holder (Skills for Care) archives the data from time to time; during this process, older records may disappear from newer data sets. To capture as several events, we requested data sets that were 3 months apart during the period of interest. The research team already possessed other data released at additional time points during this period, this resulted in a total of 18 separate NMDS-SC data sets spanning the period December 2007 to March 2011, with the number of records ranging from 13,095 to 25,266 in each data set, covering 421,671 providers' records (including updates). Specifically, we were interested in investigating the experience of the same employers over time. The data were completely anonymised with an identification code allowing the linkage of the experience of the same employer over time. The project was discussed with (King's College London) research ethics panel, and was judged to present no risk as all data are anonymised and thus ethical approval was not required. Following up the same group of care providers has advantages over repeated cross-sectional analysis as it provides a more accurate picture of change over time. For the current analysis, we focused on examining changes occurring within the same group of providers (employers) with at least two updated events over a period of 18 months, which offered sufficient time to observe any patterns of change.

Figure 1 shows the density distribution plot of the updates of this group over time, indicating two clear time points with ‘peak’ number of records. For the current analysis, we defined two data points, March 2008 (T1) and September 2009 (T2), allowing +/− 3 months margin at each time point, with an average of 18 months difference between T1 and T2, thus covering the period from the beginning of January 2008 to January 2010. Using 18 NMDS-SC data sets, we included any provider with two updates in the regions of T1 and T2 resulting in a panel sample of 2964 providers. We conducted several data quality checks on the panel sample, including comparing main organisational characteristics over time. Overall, the panel sample was sufficiently representative of the overall NMDS-SC returns at T1 (Table 1; for full details, see Hussein & Manthorpe 2011). The study is limited in that it reflects the experience of a sample of employers in the social care sector, which had not been constructed as a representative sample of the whole population of employers. However, the sample is large in size and is representative of the larger group of social care employers completing the NMDS-SC, which is the most comprehensive data set on this workforce in England.

We calculate turnover rate as the number of care workers who left their employer within the past 12 months of the data collection point as a proportion of number of care workers in permanent and temporary work arrangements at the time of data collection. Vacancy rate was calculated as the proportion of care work jobs vacant at each of T1 and T2 time points. We focused on key employers' characteristics including region, sector, type of setting and organisation size in relation to changes in vacancy and turnover rates of care workers over time. Changes over time were examined using paired t-test and none of the data items were weighted or corrected during the analysis. C++ programming (GCC compiler) was employed to construct the panel and the statistical analysis was completed on R-Unix statistical environment.

Findings

Care worker turnover and vacancy rates 2008–2010

For the panel of providers examined in this study, the overall turnover rate remained almost unchanged during the period 2008–2010. The mean turnover rate was 22.5% in T1 (median = 14.29) and 22.9% in T2 (median = 14.29). This means that, on average, around a quarter (23%) of the care workforce changed their jobs within the previous 12 months prior to March 2008, with similar workforce traffic 18 months later. The longitudinal analysis indicates that mean care worker vacancy rate declined from 4.6% in T1 to 3.9% in T2.

Regional variations

Table 2 indicates that care worker turnover rates are lowest among providers located in London both in 2008 and 2010; however, they increased by 2.2% over the same period. By January 2010, care worker turnover rate was highest in the East and West Midlands and the South West of England (25%–26%). The largest increases in care worker turnover rates were in the North West and the South West; however, none of these changes from T1 to T2 was statistically significant.

Table 2. Care worker mean turnover and vacancy rates at T1 and T2 by region

Region	Number of providers	Turnover			Vacancy
		Mean turnover rate		Paired t-test	Mean vacancy rate		Paired t-test
		T1	T2		T1	T2
Eastern	255	22.6	21.2	−0.56	4.1	2.6	−1.61
East Midlands	260	26.8	25	−0.53	3.9	2.8	−0.99
London	274	15.5	17.7	0.78	6.0	7.5	0.88
North East	101	20.4	18.5	−0.37	6.3	4.3	−1.62
North West	328	18.9	22.1	1.3	2.5	2.3	−0.27
South East	781	23.2	22	−0.63	5.4	4.4	−1.66
South West	372	22.9	26	1.13	5.1	3.7	−1.48
West Midlands	331	23.9	25.7	0.64	5.0	3.4	−1.74
Yorkshire & Humber	262	23.3	23.9	0.26	3.3	2.4	1.58

• None of the above changes were statistically significant.

Table 2 also shows that care worker vacancy rate was highest in London, particularly at T2; this may be related to the high cost of living in the capital relative to the overall low levels of pay in the sector. On the other hand, it remained lowest in the North West and Yorkshire & Humber regions. Care worker vacancy rates declined in all regions over 2008–2010 except in London where they increased by 1.5%; however, this was not statistically significant. When examining changes in vacancy rates within individual providers in different regions, the most pronounced reductions took place in the West Midlands and the South East. However, such reductions were not statistically significant (P = 0.08 and 0.09 respectively).

Organisational factors

Sector

Sector is one of the most important characteristics associated with different measures of workforce stability in social care as it reflects the profit or non-profit status of providers. Previous analysis using the NMDS-SC revealed turnover rates to be generally higher for different job roles within the private sector than other sectors (Hussein 2009). The private sector generally provides less favourable working conditions and lower levels of pay than the public and voluntary or not-for-profit sectors (Rubery et al. 2011). The analysis showed that care worker turnover rate was highest among private providers, especially in T1 (25% at T1 and 24.8% at T2); this compared to only 10.6% in the public sector (local authorities) at T1 and 8% in T2. Care worker turnover rates among providers from other sectors (including a very small number in the National Health Service) stood at 20% in T1 and increased to 23.9% by T2. The analysis indicated that turnover rates did not significantly change over the period of analysis in the main three sectors: public, private and voluntary. Moreover, in the private sector, the median care worker turnover rate declined slightly from 16.7% to 15.4%.

The current longitudinal analysis, which focuses only on direct care workers (not managers), reveals that care worker vacancy rates declined in all sectors over 2008–2010. By January 2010, the mean vacancy rate among private providers was 3.7%, which is very close to the 3.3% among providers in the voluntary sector and slightly higher than that in local authority (public) providers. Figure 2 shows that vacancy rates varied more widely among individual public sector providers, while differences were much narrower among private and voluntary organisations. The reductions observed in vacancy rates within individual providers were significant for those in both the private and voluntary sector, while not significant for public sector providers.

Organisation size

Turnover rate was significantly lower within larger organisations (with 200 employees or more); however, the number of large organisations in the sample was relatively small (n = 15 at T1 to 18 at T2). We investigated individual changes over time in turnover rates for different groups of providers according to the size of organisation (grouped as: micro employers = less than 10 staff members, small = 10–49 staff members, medium = 50–199 and large = 200 or more staff members). Figure 3 shows that organisations of all sizes experienced reductions in care worker vacancy rates from T1 to T2. Vacancy rates were lowest within large organisations, although reductions in vacancy rates were not significant among this group of providers. Micro organisations (employing fewer than 10 staff members) experienced the largest decline in the mean vacancy rate from 8.2% in T1 to 6.2% in T2. Significant reductions were also observed among small- and medium-sized employers (from 4.1 to 3.5 and 3.9 to 2.6 respectively).

Type of care settings

Care worker turnover rates were considerably lower among settings providing nursing care (such as care homes with nursing or home nursing as part of social care) when compared with other services. Table 3 shows that turnover rate in this part of the social care sector was only 4.2% at T1 with a slight increase to 5.7% by T2. Care worker turnover rate was highest in adult residential care (care homes without nursing care) and adult domiciliary care (home care) settings. Little other change in turnover rate was observable by type of care setting.

Table 3. Care worker mean turnover and vacancy rates at T1 and T2 by type of care settings

Type of care setting	Number of providers	Turnover			Vacancy
		Care worker mean turnover rate		Paired t-test	Care worker mean vacancy rate		Paired t-test
		T1	T2		T1	T2
Adult – residential	2021	23.7	23.8	0.11	4.3	3.5	−2.54
Adult – day care	123	14.1	15.4	0.34	2.1	1	−1.15
Adult – domiciliary	623	21.5	22.8	0.74	6.4	5	−1.46
Health	43	4.2	5.7	0.42	0.7	0.7	−0.01
Other	108	15.2	24.1	0.85	10.5	3.5	−1.2

• None of the above changes were statistically significant.

Vacancy rates varied considerably by type of care setting, with lowest vacancy rates being in adult day care and health services (at T2: 1% and 0.7% respectively). On the other hand, care worker vacancy rates were considerably high in home-care services (at T2: 5%). Table 3 indicates that, in terms of change, all employers within different settings experienced reductions in care worker vacancy rates. However, such reductions were only significant for adult residential providers (care homes) (paired t = −2.54, P = 0.01).

Discussion

The current analysis highlights a number of important findings. The period of 2008–2010 saw considerable economic and policy changes that influenced and continue to influence care workforce stability and possibly quality of care. Overall during the period from January 2008 to January 2010, the mean turnover rate of care workers remained unchangeably high at 23%, while vacancy rates significantly declined from 4.6% to 3.9%. However, care worker vacancy rates remained considerably higher than overall vacancy rates in the UK labour market at just 1.7% in June to August 2011 (CI 1.6%–1.8%, ONS 2011). Similarly, a turnover rate of 22%–23% was considerably higher than experienced by other UK employment sectors, standing at 15.7%; however, it was considerably lower than the 34% turnover rate observed in the catering and leisure industry in the same time period (CIPD 2009). Moreover, this rate was substantially lower than the estimated 56.4% turnover rate among nurse aides (the closest US equivalent job role to care workers) in long-term care facilities in six US states (Castle & Engberg 2006) and an annualised turnover rate of 74.5% using 2004 data among certified nursing assistants in US nursing homes (Donaghue 2009).

These findings raise questions around the quality and motivations of new recruits and methods of reducing specific vacancy levels, which other types of research need to address. The strength of a longitudinal analysis is that it can illuminate continuing patterns as well as changes; our findings suggest that while reductions in care worker vacancy rates have occurred, turnover rates remain on average high. These are strong indications that vacancy rate reductions may have been achieved in the context of the lack of other employment opportunities within the wider labour market and increases in people seeking jobs due to the economic recession. The role of measures to encourage unemployed people to enter the labour market and dissuade them from living on unemployment benefits is unknown as a factor in this trend. Increases in turnover rates are associated with a continuous process of under-funding and marketisation, which took place during the study period (Glendinning 2011). For example, turnover rates remained highest within the private sector; however, private sector employers were over-represented among the group with significantly reduced vacancy rates at T2.

Both turnover rates and vacancy rates were highest among both care homes (residential care) and home care (domiciliary care), when compared with the smaller number of organisations providing day care or nursing care. However, home-care providers were over-represented among the group which experienced significant increases in turnover rates over the study period. Such observations reflect changes associated with fragmentation of services, low wages and increased incidences of non-payments of the National Minimum Wage, particularly when calculating travel time (Hussein 2011a, 2014, Bessa et al. 2013).

As with other studies, we found regional or geographical variations in vacancy levels, and turnover rates are likely to reflect local labour market variations (see Castle 2008 on the US). The capital, London, experienced one of the lowest turnover rates combined with one of the highest vacancy rates of care workers in the UK. The relatively low turnover rate in London (16% at T1 and 18% at T2) is likely to be associated with the high prevalence of non-EEA migrants there compared to other regions (Hussein 2011b, Skills for Care 2011). Non-EEA migrants are usually subject to immigration controls that prevent them from changing employers and they are not generally entitled to social security benefits if unemployed and so may not move jobs readily.

One of the main findings highlighted by this panel analysis is the great variability in the stability of the care workforce by different organisational characteristics, especially sector and type of activity, despite an unchangeable ‘average’ turnover rate over time. Just over half the providers in the data analysed in this study experienced no change in care worker turnover rates. However, nearly a quarter saw significant reductions in turnover rates, with another quarter experiencing significant increases. Using this reconstructed panel data, we were able to investigate and compare stated reasons for leaving a job among the two groups of employers experiencing improvement and deterioration in care worker turnover. These are reported elsewhere (Hussein et al. forthcoming). However, they point to the importance of the suitability of jobs as well as levels of wages among other conditions.

Conclusion

The policy value of these findings is evident at a number of levels. First, social care policy could point to the success of this area in turning round or beginning to halt the severe workforce shortages of the sector and to slow down the pace of turnover. Instead of concentrating on recruitment and retention as endemic to the sector, there may be room to acknowledge that some employers have experienced improvements in their vacancy and turnover rates. Researchers and policy makers might wish to place greater emphasis on differentiating between providers to understand work culture, support mechanisms and other characteristics that are associated with improved turnover and vacancy rates. There may be room for regional rather than national initiatives, especially in the London region and its surroundings. Policy makers may wish to support regulatory bodies to examine vacancy and turnover rates in line with these national findings to ensure that they are paying attention to workforce turnover as a sign of possible poor recruitment and to ensure that proper induction and support for new staff are a reality. At a time of increased emphasis on care at home, it is worthwhile examining why it is that care homes appear to have greater stability of staff and fewer vacancies than home-care agencies. Possibly, some workers may wish to combine care home and home-care work and this too might be explored to see if this enhances staff satisfaction, and leads to good outcomes for service users.