Abstract
The shift towards the greater personalisation of care has seen a rise in personal assistants (PAs) employed by individuals using their own funds or receiving public support. Despite their growing importance, the recruitment and retention of PAs can be challenging. In this article, we empirically explore what affects the turnover and vacancies of PAs. We find that PA turnover and vacancies are affected by care need, type of support and local labour market factors. The new evidence from this article shows the difficulties of employing PAs and potential policy levers that could be used to improve PA recruitment and retention.
Introduction
Demand for social care is increasing over time with an ageing population. This is associated with policy in the UK and elsewhere moving towards increased choice, personalisation and care at home (Pavolini and Ranci, 2008). This includes ‘cash for care’ policies, which enable individuals to direct the support for their own care independently, deciding how best to allot funding to meet their care needs (Ungerson, 2004; Da Roit and Le Bihan, 2010). In England, this occurs through direct payments for social care needs and personal health budgets for healthcare. These budgets can be spent on services, equipment and staff.
Many of those who receive cash for care will use it to directly employ staff rather than buy in services from home care agencies. By employing staff themselves, people in need of support can maintain choice and control over their care, but they also take on the responsibilities that come with becoming an individual employer, such as recruiting and paying staff and other considerations. Staff employed directly by a person in need of care are known as personal assistants (PAs), and they provide person-centred care in the home, workplace or community.1 For England, there are estimated to be 65,000 DP holders who employ staff directly themselves. This corresponds to 90,000 PAs working in 120,000 mainly part-time jobs (Skills for Care, 2022b), accounting for around 7 per cent of the estimated 1.79 million jobs in adult social care nationally. In addition, PAs are employed by people who use their own funds and by those in receipt of National Health Service (NHS) personal health budgets, though estimates of how many are employed are not available (Skills for Care, 2022a).
Despite the growing importance of PAs, there are still challenges with their recruitment and retention, as is the case for staff elsewhere in adult social care (Scourfield, 2005; OECD, 2020). Recent survey and qualitative evidence suggest that many individual employers find it hard to recruit a PA, and vacancy rates for PAs are among the highest among social care workers (Woolham et al, 2019b; Skills for Care, 2022b; TLAP, 2022). PAs also have a substantial turnover rate, though lower compared to other care workers (Skills for Care, 2022b). Although these trends are well documented, the factors that affect them are not well understood. Most of the existing evidence on the factors that affect the recruitment and retention of adult social care staff focus on different sectors, such as domiciliary or residential care, not PAs per se (Morris, 2009; Butler et al, 2010; Banijamali et al, 2014; Hussein et al, 2016; Moriarty et al, 2018). Any evidence on the factors that affect PA employment is either qualitative or refers to different countries and settings (Ungerson, 1999; Woolham et al, 2019a; 2019b; Melchiorre et al, 2022). To date, quantitative evidence on the determinants of PA turnover and vacancies in England is limited. However, understanding which factors affect PA employment from a quantitative point of view is important for policy purposes, as it allows us to understand their relative importance and therefore design targeted interventions. The aim of this article is to fill this evidence gap and quantitatively assess how individual employer and local market characteristics impact PA turnover and vacancies.
Background
Direct payments and personal health budgets
The increased employment of PAs in England has been driven largely by policy change towards the direct payments of cash (Glasby and Littlechild, 2022). Current legislation (the Care Act 2014) requires all local authorities to assign a personal budget to those eligible for public support (through needs and means tests). All those eligible are offered the chance to receive a direct payment; otherwise, the budget is managed by the local authority or a third party, for example, a service provider (Age UK, 2023). Personal health budgets are the NHS version of direct payments and are similar in format; from 2013, certain groups of service users, generally those with long-term mental or healthcare needs, had the right to receive a personal health budget. This budget can be notionally managed on their behalf by a third party or the client can receive the budget as a direct payment and manage it themselves to meet their ongoing health needs (NHS England, 2022).
The use of direct payments and personal health budgets has increased over time. Whereas just over 3,500 people were receiving direct payments in 2000, 220,000 people received direct payments in 2021/22 (Glasby and Littlechild, 2022; Skills for Care, 2022b). The use of personal health budgets is also rapidly increasing. There were 88,953 individuals in receipt of personal health budgets at the end of the third quarter of 2019/20 compared to 40,344 at the end of the third quarter of 2018/19 (NHS Digital, 2020). Significant positive effects on outcomes have been found for the use of direct payments and personal health budgets (Netten et al, 2012; Jones et al, 2013). The advantages to individual employers from using direct payments or personal health budgets to employ PAs are usually linked to increased choice and control. This can include the selection of the PA(s), care continuity and greater flexibility in both timings and what tasks can be fulfilled (Glendinning et al, 2000b; Ungerson, 2004; McGuigan et al, 2016; Glasby and Littlechild, 2022). The negative aspects include the burden of increased administration and the potential for increased anxiety resulting from this at the individual level (Poole, 2006; Netten et al, 2012; McGuigan et al, 2016).
The overlap between health and social care needs suggests that the help PAs offer can often extend across this divide (Glendinning et al, 2000a). PAs funded under a personal health budget are more likely to provide skilled and clinically focused care compared to PAs funded through direct payments. As the funds come from the NHS, there is a legal obligation to fund all assessed health needs and to ensure that PAs providing health-related tasks receive appropriate training to carry them out (Butler et al, 2010; Skills for Care, 2015; 2022a). With personal health budgets, funding for training is protected. This is not the case for direct payments, where PA training depends on employers’ discretion. Many individual employers, however, are reluctant to provide training for their PAs, even though they may often require help with health-related tasks (Woolham et al, 2019b). This is usually because they fear that this would imply less funding for care and a shift towards a more ‘task-focused’ care model, as opposed to one matching their specific needs (Woolham et al, 2019b). At the same time, many PAs would be willing to carry out such tasks, subject to appropriate training (Norrie et al, 2020). Overall, employers will still find difficulties in recruiting suitable PAs with either type of funding. People in receipt of a personal health budget may find it difficult to attract skilled assistants to carry out clinical tasks at the same pay rate that one would pay a more general care assistant. Individual employers in receipt of direct payments will also face recruitment challenges, as the lack of training and skill development opportunities is often a disincentive for PAs to take on such roles (Woolham et al, 2019b).
PAs and their role
A PA in England is defined as someone who is employed directly by a person who is in need of support or by a family member or representative of that person. The care PAs provide is person centred and aims at enabling the person in need to live their lives according to their wishes and interests (Skills for Care, 2022a). Given the personalised nature of care, PAs and their job role will vary markedly depending on the person who is employing them. The role could be taken by friends or family or someone unknown to the direct payment recipient. However, only in exceptional circumstances will it be a close relative (Skills for Care, 2019). Roles can range from being part of a team of PAs helping to support someone who has 24-hour needs through to helping someone for a couple of hours a week. The responsibilities will also vary, from personal care, such as bathing and feeding, to supporting the person in need of care with leisure activities, shopping, driving and cleaning (Skills for Care, 2019; Woolham et al, 2019b).
Although traditionally associated with direct employment, the term ‘personal assistant’ has evolved over time to include a wider range of employment relationships (Skills for Care, 2017c). This reflects the wide range of needs of individual employers that can be met through a variety of job roles. Direct employment remains the most common type of working relationship for PAs (84 per cent), followed by self-employment (7 per cent) and temporary employment (4 per cent) (DH, 2011; Skills for Care, 2017c; 2023a; Woolham et al, 2019a). Less frequently, people will employ workers from a home care agency to perform PA tasks (Skills for Care, 2017c; 2023b).2 Under direct employment, PAs are employed directly by a person in need of care and are entitled to certain employment rights and statutory payments. By employing a PA this way, individual employers have not only greater control over the work that PAs do but also greater responsibility for payments, taxes, insurance and other considerations (Skills for Care, 2019; Woolham et al, 2019b). If PAs work as self-employed, then individual employers contract them as a ‘business’ to provide a service. Self-employed PAs will therefore not have employment rights and responsibilities under this contractual relationship (Skills for Care, 2017c; Woolham et al, 2019a). Individual employers may also directly employ PAs under a more casual arrangement, where work may not always be guaranteed, for example, to provide sickness or annual leave cover. In other cases, care workers may be recruited from a home care agency to perform a PA role. In those cases, the agency will have more control over the type of work that the PA will do and be responsible for that worker (Skills for Care, 2017c).
The type of employment relationship will depend, to a large extent, on the degree of complexity of the PA role. Support with complex needs is more likely to be linked to direct employment, with individual employers having a greater degree of both control and responsibility, while support with simpler tasks is likely to be associated with less direct forms of employment (for example, through an agency) (Skills for Care, 2017c). Still, for the majority of cases, the PA job will be characterised by a more direct relationship with the individual employer compared to other care workers in adult social care. This means that the care PAs provide will be more personalised and that individual employers will have a greater degree of responsibility over the contractual relationship.
PAs are generally low paid, with an estimated average wage of £10.21 per hour. This is comparable to the mean hourly wage of care workers in the independent sector (voluntary and for-profit care providers) of £9.66 (Skills for Care, 2022b). Both PAs and other care workers have similar levels of qualifications (Level 2 or above), and the same percentage of PAs and other care workers (66 per cent) have (or are working towards) the Care Certificate, an agreed set of standards that define the knowledge, skills and behaviours expected of specific job roles in the health and social care sectors (Skills for Care, 2022b).
However, the comparability of pay and qualifications does not reflect that there are differences in employment conditions. For example, care workers working for home care agencies usually have a period of notice, (statutory) sick pay and holiday entitlements (Francis and Netten, 2003; Glendinning, 2012). There may also be opportunities for promotion to senior care worker roles. On the negative side, there are issues with paying for travel time or cost reimbursements and other aspects of pay, for example, unsociable hour rates (Fleming and Taylor, 2007; Rubery et al, 2015; Hall et al, 2017). On the other hand, aspects of PA employment conditions can be challenging. As outlined earlier, training opportunities for PAs can be limited within the job due to employer reluctance (Ahlström and Wadensten, 2012; Woolham et al, 2019b). Further, many PAs do not have a written contract and are not paid overtime for longer hours, hours outside the usual working day, bank holiday work or weekends (Woolham et al, 2019a). Sick leave is not always available, and when it is, many PAs will not use it (Woolham et al, 2019a; Roland et al, 2022). According to existing evidence, only a small number of PAs will contribute to a pension scheme, and some are not paying National Insurance, sometimes due to very low wages (Woolham et al, 2019a). Lack of support for PAs in the event of a dispute with their employer is also widespread, while the role can also be lonely, with some PAs not being in contact with their peers (Ahlström and Wadensten, 2012; Woolham et al, 2019b).
Despite these challenges, the PA role can be rewarding. Due to the direct support they provide to an individual, PAs can develop close bonds and form relationships with employers that are like family (Ungerson, 2004; Shakespeare and Stöckl, 2018; Manthorpe et al, 2020). However, this type of relationship could easily mask problems in the employee–employer dynamic. It can create unhealthy pressure on the PA and blur the distinction between defined job tasks and ‘helping out’ (Ungerson, 1999; Christensen and Manthorpe, 2016; Manthorpe et al, 2020).
Recruitment and retention of PAs
Recruitment practices of PAs vary, including word of mouth, adverts in local shops, job centre adverts and local authority registers (Ungerson, 2004; Skills for Care, 2015; 2017b; Woolham et al, 2019a). Many PAs are recruited from other forms of social care, but others will have no prior social care experience (Skills for Care, 2019; Woolham et al, 2019a). There is also overlap with other job roles: nearly 30 per cent of PAs have additional jobs with independent sector or local authority providers (Skills for Care, 2022b).
Recruiting and maintaining suitable PAs can be difficult for many individual employers. Vacancy rates for PAs are among the highest in social care, and turnover rates are substantial (Woolham et al, 2019b; Skills for Care, 2022b; TLAP, 2022). The reasons behind these trends are not yet well understood. The bulk of the qualitative and empirical literature that looks at the factors associated with the recruitment and retention of social care workers has focused on other sectors, such as domiciliary or residential care. From this literature, we know that turnover and vacancy rates of the adult social care workforce are higher in the private sector and in residential and home care settings and lower in large organisations and nursing care settings (Hussein et al, 2016). The wage and skills gap between social care and the NHS and other retail and hospitality employers further contributes to the recruitment and retention challenges faced by the adult social care sector (Moriarty et al, 2018). Competition for the workforce can also take place between different care settings, with domiciliary care providers facing challenges in recruiting because of alternative social care providers (Ware et al, 2001). International evidence has shown that age, lack of access to health insurance, low status and poor pay increase staff turnover in home care (Morris, 2009; Butler et al, 2010; Banijamali et al, 2014). Turnover intention is also positively influenced by higher job demands, health, temporary contracts and the likelihood of employment elsewhere in social care and negatively influenced by job resources, job satisfaction and caring for clients (King et al, 2013; Jang et al, 2017; Barken et al, 2018).
Although employment challenges are widespread across the adult social care sector, the distinctive characteristics of the PA role suggest that the factors affecting PA employment may differ from other social care workers (Scourfield, 2005). For example, in recent qualitative studies, poor behaviour by employers towards PAs, a lack of role clarity and a lack of training opportunities are often cited as the main reasons for high PA turnover and vacancy rates (Woolham et al, 2019b). Evidence also suggests that PA recruitment and retention are more difficult in full-employment (high-wage) areas and that staff will switch from PA roles to other social care providers (Glendinning et al, 2000b; Carmichael and Brown, 2002; Ungerson, 2004; Woolham et al, 2019b). Despite this substantial body of qualitative work, the quantitative evidence base on the determinants of PA employment remains limited. This restricts our ability to understand the relative importance of those factors and make targeted policy interventions. In this article, we try to address this evidence gap and bring new quantitative evidence on the factors that affect PA recruitment and retention. We used a new data set of individual employers with information on individual employer characteristics and PA turnover and vacancies and matched it to data on local authority characteristics. With the use of regression analysis, we then estimated the effect of individual employer characteristics, such as age, need type and funding arrangements, and local market characteristics, such as measures of supply and demand for social care, on the probability of PA turnover and vacancies.
Data and methods
Data and measures
The data for this study come from the Skills for Care survey of individual employers who employ their own care and support staff. Skills for Care is the strategic workforce development and planning body for adult social care in England and works with employers, PAs, national and local government, and partners to report on the status of the social care workforce and to increase understanding of the key drivers of workforce change. Skills for Care carried out two anonymous surveys of individual employers across England in 2017 and 2019. Using two national support organisations and an online survey, they surveyed over 10,000 and 18,000 individual employers in 2017 and 2019, respectively (Skills for Care, 2017a; 2019). Individual employers responded on a voluntary basis. The response rate was just over 10 per cent, and the final sample size was 1,043 and 1,960 individual employers for 2017 and 2019, respectively.3 The pooled sample size corresponds roughly to 4.3 per cent of the total number of people who employ PAs in the country, and the sample may not therefore be representative of the entire individual employer population. The individual employers contacted could be funding their PA support through a direct payment from a local authority, a personal health budget from the NHS or using their own funds as self-funders. The survey included individual employers who directly employed a PA or received services from self-employed PAs but did not include employers who employed PAs via an agency (Skills for Care, 2017a; 2019). Data were anonymised by Skills for Care before they were used by the authors of this study, and the use of these data received ethical approval from the University of Kent’s Social Research Ethics Committee (Ref: SRCEA240).
In terms of outcome measures, the survey asked individual employers how many workers had left their employment in the past 12 months and how many staff vacancies they currently had. From these data, we constructed two binary indicator variables taking the value of 1 when employers had at least one leaver and at least one vacancy, respectively, and 0 otherwise.
The survey also collected data about employers’ age band, type of primary care need,4 council area of residence, type of funding, number of current staff5 and access to information and training. From these data, we constructed the variables that we use as covariates in the empirical analysis. These include the total number of staff employed and binary variables taking the value of 1 for whether an individual employer was over 65 years old, had learning disability support needs, had mental health needs, had access and mobility needs, had physical support with personal care needs (such as eating, bathing, dressing, toileting and transferring), had memory and cognition needs, had sensory needs, had social support needs, was funding care via a personal health budget or own money, and had undertaken training. Training could be a formal qualification, a structural awareness course or specific subject awareness training (for example, around issues of employment law, payrolls, managing and supervision, and so on). The reference categories in the empirical analysis were personal care for needs and direct payments for funding.
We used the information on the council areas that employers lived in to match employers to data on the characteristics of their local areas. We controlled for the following contextual factors at the local authority level: the proportion of people entitled to a Personal Independence Payment (PIP); the proportion of the population aged over 65; the unemployment rate of people aged over 16; the count of alternative social care employers (care homes and home care providers) per square kilometre; and the total number of direct care job roles in social care per square kilometre.6 Data on individual employer training and data on the number of direct care jobs in social care were available only for 2019.
Sample and summary statistics
We excluded from the analysis any observations that were coded as not being individual employers and any respondents who reported no care needs and no staff employed. This provided a sample size of n = 2,995. Due to missing values in the turnover and number of vacancies variables, particularly in the 2019 data, the complete case samples for analysis were reduced to n = 2,463 for turnover and n = 2,467 for job vacancies.
Table 1 presents the sample summary statistics. We summarise the means and standard deviations of all the variables included in the analysis. Mean values of the categorical variables are presented as both proportions and percentages, and the standard deviation provides a measure of how dispersed the data are around the mean. Sample summary statistics are calculated across all non-missing observations for each variable, which are shown in the last column of the table. Over 28 per cent of individual employers had at least one PA leaving the job in the previous 12 months, and 15 per cent had at least one vacancy at the time of the interview. Employers required PA support with learning disability (71 per cent), personal care (67 per cent), mental health (57 per cent), access and mobility (53 per cent), sensory (56 per cent), and social (43 per cent) needs. Just over 23 per cent of employers were 65 years old or older. About 86 per cent of employers received a direct payment, 4 per cent received a personal health budget and 8 per cent used their own money to fund a PA. In 2019, about 40 per cent of employers had undertaken some form of training to help them in their role. The average employer in the sample was located in a local authority with an unemployment rate of 4.7 per cent, an older population percentage of 19 per cent, 4 per cent of people entitled to PIP and 0.685 social care providers and 25.7 front-line social care staff per square kilometre.
Sample summary statistics
| Variable | Mean | SD | N | |
|---|---|---|---|---|
| Mean/proportion | Percentage | |||
| Individual employer characteristics | ||||
| Leavers (at least one) | 0.284 | 28.4% | 45.1 | 2,795 |
| Vacancies (at least one) | 0.152 | 15.2% | 35.9 | 2,795 |
| Total number of staff | 2.046 | 1.498 | 2,941 | |
| Care need | ||||
| Learning disability | 0.709 | 70.9% | 0.454 | 2,971 |
| Personal | 0.674 | 67.4% | 0.469 | 2,971 |
| Mental health support | 0.573 | 57.3% | 0.495 | 2,971 |
| Access and mobility | 0.523 | 52.3% | 0.500 | 2,971 |
| Memory and cognition | 0.528 | 52.8% | 0.499 | 2,971 |
| Sensory support | 0.558 | 55.8% | 0.497 | 2,971 |
| Social support | 0.425 | 42.5% | 0.494 | 2,971 |
| Over 65 | 0.232 | 23.2% | 0.422 | 2,950 |
| Funding | ||||
| In receipt of a direct payment | 0.859 | 85.9% | 0.348 | 2,910 |
| In receipt of a personal health budget | 0.044 | 4.4% | 0.204 | 2,883 |
| Own money | 0.084 | 8.4% | 0.277 | 2,910 |
| Any training to help as an employera | 0.396 | 39.6% | 0.489 | 1,947 |
| Local authority characteristics | ||||
| Unemployment rate | 0.045 | 4.5% | 1.995 | 2,868 |
| Population over 65 | 0.189 | 18.9% | 0.042 | 2,871 |
| PIP entitlement | 0.040 | 4% | 0.012 | 2,871 |
| Total number of social care providers (per km2) | 0.685 | 0.789 | 2,922 | |
| Total number of social care jobs (per km2)a | 25.66 | 31.42 | 1,919 | |
Notes: a 2019 only data. Means for categorical variables are presented as proportions and percentages. Sample summary statistics are calculated across all non-missing observations for each variable (N). Regression samples vary by year and model specification. Regression sample size for turnover was n = 2,463 for both years and n = 1,599 for 2019. Regression sample size for vacancies was n = 2,467 for both years and n = 1,603 for 2019. Training as an employer could include formal qualification (M = 0.047, SD = 0.212), structured awareness training (M = 0.051, SD = 0.22), subject awareness training (M = 0.064, SD = 0.24) and ‘other’ (M = 0.257, SD = 0.43).
Sources: Skills for Care (2017a; 2019) individual employer surveys (for individual employer characteristics), Office for National Statistics (ONS) and Care Quality Commission (CQC) (for local authority characteristics).
Table 2 presents the sample distribution of the number of PAs who left the job in the previous 12 months and the number of current PA vacancies. The vast majority of individual employers had either no PA (72 per cent) or just one PA leaving the job (21 per cent). The remaining 7.5 per cent of observations reported over two PAs leaving the job, distributed in the following way: 5 per cent reported two, 1.7 per cent reported three and 0.8 per cent reported four or more leavers. Likewise, the majority of individual employers had no vacancy (85 per cent) or only one vacancy (13 per cent) for a PA. A smaller number (2.3 per cent) of observations had two or more vacancies, with the following distribution: 1.8 per cent reported two, 0.4 per cent reported three and 0.07 per cent reported four vacancies. Given the high skew of outcomes, we proceeded by grouping turnover and number of vacancies into binary indicators measuring whether an individual employer had at least one PA who left employment and at least one PA vacancy.
Sample distribution of the number of PAs who left employment and the number of PA vacancies
| Turnover | Vacancies | ||||
|---|---|---|---|---|---|
| Frequency | Percentage (%) | Frequency | Percentage (%) | ||
| 0 | 2,000 | 71.56 | 0 | 2,370 | 84.79 |
| 1 | 585 | 20.93 | 1 | 361 | 12.92 |
| 2 | 142 | 5.08 | 2 | 51 | 1.82 |
| 3 | 46 | 1.65 | 3 | 11 | 0.39 |
| 4+ | 22 | 0.79 | 4 | 2 | 0.07 |
| Total | 2,795 | 100 | Total | 2,795 | 100 |
Source: Skills for Care (2017a; 2019) individual employer surveys.
Methods
We used multivariate regression analysis to estimate the association between individual employer and local authority characteristics and the probability of having at least one PA leaving the job and at least one vacancy.7 To model the probability of our binary outcomes occurring conditional on other characteristics, we used a Probit estimator, which assumes that this probability follows a standard normal cumulative distribution. We pooled all data from 2017 and 2019 together, and the analysis was done on the entire sample. Although the data come from different points in time and the same model might therefore not apply in each time period, pooling different cross-sections of data allows a larger sample size and greater statistical power. We tried to mitigate these limitations by including in the regression models year dummies to account for any structural changes over time. The models also include regional dummies to account for systematic differences in the supply and demand of social care across regions. We also ran the analysis separately for 2019 only because these data included measures of employer training and social care supply not available in the 2017 data. Standard errors were robust to heteroscedasticity.
Estimation results
Turnover
Table 3 presents the estimation results for turnover. The findings for the pooled sample are shown in Column 1. Columns 2 and 3 present the findings for the 2019 sample, with local social care supply measured as the weighted count of care homes and home care providers and the weighted number of social care jobs in the area, respectively.
The impact of employer and market characteristics on the probability of PA turnover: estimated marginal effects and confidence intervals
| Turnover | ||||||
|---|---|---|---|---|---|---|
| 1 | 2 | 3 | ||||
| Marginal effect | 95% CI | Marginal effect | 95% CI | Marginal effect | 95% CI | |
| Total number of staff | 0.066*** (0.006) | 0.055–0.078 | 0.065*** (0.007) | 0.051–0.079 | 0.065*** (0.007) | 0.051–0.078 |
| Care need (ref: personal care) | ||||||
| Learning disability | –0.043* (0.023) | –0.089–0.003 | –0.035 (0.037) | –0.107–0.037 | –0.035 0.037 | –0.106–0.037 |
| Mental health support | –0.029 (0.026) | –0.080–0.021 | –0.044 (0.029) | –0.101–0.014 | –0.044 0.029 | –0.101–0.014 |
| Access and mobility | 0.004 (0.021) | –0.037–0.045 | 0.006 (0.025) | –0.043–0.054 | 0.006 0.025 | –0.043–0.054 |
| Memory and cognition | –0.013 (0.025) | –0.062–0.036 | –0.007 (0.028) | –0.062–0.048 | –0.006 0.028 | –0.062–0.048 |
| Sensory support | 0.043 (0.031) | –0.018–0.103 | 0.046 (0.033) | –0.019–0.111 | 0.045 0.033 | –0.016–0.079 |
| Social support | 0.015 (0.022) | –0.028–0.058 | 0.033 (0.024) | –0.015–0.081 | 0.032 0.025 | –0.016–0.080 |
| Over 65 | –0.022 (0.021) | –0.063–0.020 | –0.027 (0.026) | –0.077–0.024 | –0.028 0.026 | –0.079–0.023 |
| Funding (ref: direct payment) | ||||||
| Personal health budget | 0.040 (0.041) | –0.039–0.119 | 0.098* (0.059) | –0.017–0.214 | 0.101* (0.059) | –0.015–0.217 |
| Own money | 0.019 (0.031) | –0.041–0.080 | 0.017 (0.033) | –0.047–0.082 | 0.017 (0.033) | –0.047–0.082 |
| Undertaken training | 0.034 (0.021) | –0.007–0.076 | 0.034 (0.021) | –0.007–0.075 | ||
| Local authority characteristics | ||||||
| Unemployment rate | –0.016** (0.008) | –0.031 to –0.001 | –0.023** (0.010) | –0.043 to –0.003 | –0.022** (0.010) | –0.042 to –0.002 |
| Population over 65 (rate) | –0.149 (0.375) | –0.885–0.586 | 0.264 (0.425) | –0.570–1.098 | 0.471 (0.431) | –0.374–1.316 |
| PIP entitlement (rate) | –1.924 (1.561) | –4.984–1.136 | –0.215 (1.785) | –3.713–3.283 | –0.581 (1.800) | –4.109–2.946 |
| Total social care supply | 0.043* (0.024) | –0.003–0.089 | 0.037 (0.025) | –0.012–0.087 | ||
| Total social care jobs | 0.001** (0.001) | 0.000–0.003 | ||||
| Constant | 0.271*** (0.0085) | 0.254–0.287 | 0.257*** (0.0104) | 0.236–0.277 | 0.257*** (0.0104) | 0.236–0.277 |
| Regional dummies | Yes | Yes | Yes | |||
| Year dummies | Yes | |||||
| Pseudo R2 | 0.0693 | 0.0732 | 0.0747 | |||
| Observations | 2,463 | 1,599 | 1,599 | |||
Notes: *** p < 0.01; ** p < 0.05; * p < 0.1. Marginal effects from probit estimations reported. Robust standard errors reported in parentheses. Total social care supply: total number of local authority Care Quality Commission (CQC)-registered care homes and home care providers per square kilometre. Model 1: pooled 2017 and 2019 data with total social care supply as control. Model 2: 2019 data with total social care supply as control. Model 3: 2019 data with total local authority social care jobs per square kilometre as control.
There is a positive association between the total number of staff and PA turnover in all three models. PA turnover is lower for individual employers with learning disability needs by 4 percentage points compared to individual employers with personal care needs in the pooled data specification. Personal health budgets are associated with a significant 10 percentage-point higher probability of PA turnover compared to direct payments in the 2019 model. This association is statistically significant and sizeable.
Local authority characteristics also affect PA turnover in a statistically significant way. Higher unemployment rates are negatively associated with the probability of turnover. A 1 percentage-point increase in the unemployment rate reduces the probability of a PA leaving by 2 percentage points. The local supply of social care, whether in terms of total providers or the total number of jobs available, is positively associated with turnover. In the pooled model, a unit increase in the total supply of social care providers is associated with a 4 percentage-point increase in the probability of turnover, and in the 2019 model, one more job per square kilometre in the local social care sector increases the probability of PA turnover by 0.1 percentage points. Overall, the conditions of the local market and the availability of alternative opportunities within the wider social care sector can place constraints on the retention of PAs.
Vacancies
Table 4 presents the estimation results for vacancies. As with turnover, the first column shows the results from the pooled model, while the following two columns show the results from the 2019 model, each with an alternative measure of social care supply.
The impact of employer and market characteristics on the probability of PA vacancies: estimated marginal effects and confidence intervals
| Vacancies | ||||||
|---|---|---|---|---|---|---|
| 1 | 2 | 3 | ||||
| Marginal effect | 95% CI | Marginal effect | 95% CI | Marginal effect | 95% CI | |
| Total number of staff | 0.011** (0.004) | 0.002–0.019 | 0.014** (0.006) | 0.003–0.025 | 0.014** (0.006) | 0.003–0.025 |
| Care need (ref: personal care) | –0.088 to –0.012 | |||||
| Learning disability | –0.050*** (0.019) | –0.090*** (0.027) | –0.144 to –0.036 | –0.090*** (0.027) | –0.144 to –0.036 | |
| Mental health support | –0.040* (0.020) | –0.080–0.000 | –0.037 (0.023) | –0.083–0.009 | –0.036 (0.023) | –0.082–0.009 |
| Access and mobility | 0.033** (0.017) | 0.001–0.066 | 0.034 (0.021) | –0.007–0.074 | 0.034 (0.021) | –0.007–0.075 |
| Memory and cognition | –0.001 (0.020) | –0.041–0.039 | 0.005 (0.023) | –0.040–0.051 | 0.006 (0.023) | –0.039–0.051 |
| Sensory support | –0.033 (0.025) | –0.082–0.015 | –0.057** (0.026) | –0.107 to –0.006 | –0.057** (0.026) | –0.108 to –0.007 |
| Social support | 0.021 (0.017) | –0.013–0.055 | 0.038* (0.021) | –0.003–0.079 | 0.038* (0.021) | –0.003–0.079 |
| Over 65 | –0.058*** (0.018) | –0.093 to –0.023 | –0.068*** (0.023) | –0.113 to –0.024 | –0.070*** (0.023) | –0.114 to –0.025 |
| Funding (ref: direct payment) | ||||||
| Personal health budget | 0.101*** (0.030) | 0.042–0.160 | 0.096** (0.048) | 0.001–0.190 | 0.097** (0.048) | 0.002–0.191 |
| Own money | 0.003 (0.025) | –0.047–0.052 | 0.0002 (0.028) | –0.055–0.056 | 0.001 (0.028) | –0.055–0.056 |
| Undertaken training | 0.044*** (0.017) | 0.011–0.077 | 0.045*** (0.017) | 0.011–0.078 | ||
| Local authority characteristics | ||||||
| Unemployment rate | –0.012* (0.006) | –0.025–0.000 | –0.020** (0.009) | –0.038 to –0.002 | –0.019** (0.009) | –0.036 to –0.001 |
| Population over 65 (rate) | 0.018 (0.288) | –0.546–0.582 | 0.083 (0.338) | –0.580–0.745 | 0.073 (0.344) | –0.601–0.747 |
| PIP entitlement (rate) | –0.456 (1.188) | –2.784–1.872 | 0.240 (1.426) | –2.555–3.034 | 0.201 (1.419) | –2.581–2.983 |
| Total social care supply | 0.039** (0.017) | 0.005–0.073 | 0.041** (0.019) | 0.003–0.080 | ||
| Total social care jobs | 0.001**0.0005 | 0.000–0.002 | ||||
| Constant | 0.142*** (0.0069) | 0.129–0.156 | 0.146*** (0.0085) | 0.129–0.163 | 0.146*** (0.0085) | 0.129–0.163 |
| Regional dummies | Yes | Yes | Yes | |||
| Year dummies | Yes | Yes | Yes | |||
| Pseudo R2 | 0.0417 | 0.0605 | 0.0602 | |||
| Observations | 2,467 | 1,603 | 1,603 | |||
Notes: *** p < 0.01; ** p < 0.05; * p < 0.1. Marginal effects from probit estimations reported. Robust standard errors reported in parentheses. Total social care supply: total number of local authority Care Quality Commission (CQC)-registered care homes and home care providers per square kilometre. Model 1: pooled 2017 and 2019 data with total social care supply as control. Model 2: 2019 data with total social care supply as control. Model 3: 2019 data with total local authority social care jobs per square kilometre as control.
The total number of staff employed by individual employers is positively associated with the probability of having at least one PA vacancy in all three specifications. Learning disability needs are associated with a statistically significant reduction in the probability of vacancies compared to personal care needs. The estimated reduction in probability is 5 percentage points in the pooled model and 9 percentage points in the 2019 model. Mental health needs have a negative and significant association with vacancies of 4 percentage points in the pooled model. Sensory support needs are significant only in the 2019 model and are associated with a 6 percentage-point reduction in the probability of a vacancy. Access and mobility needs, on the other hand, have a positive association with vacancies in the pooled model of 3 percentage points. Being over 65 years old has a negative and statistically significant association of 6 to 7 percentage points with the probability of a PA vacancy. Funding via a personal health budget has a positive and sizeable statistically significant association of 10 percentage points with the probability of a PA vacancy compared to funding via direct payments. This is similar to the findings on turnover. Training to help with their roles as employers is associated with an increase in the probability of a vacancy of 4 percentage points.
In terms of local market characteristics, a 1 percentage-point increase in the unemployment rate at the local authority level has a significant negative association with the probability of a PA vacancy of 1.2 per cent, which increases to 2 per cent for the 2019-only model. On the other hand, measures of demand for social care, such as the percentage of the local authority population over 65 and the percentage of the local authority population with a PIP entitlement, do not have a significant association with vacancies. The supply of alternative social care providers is another local market factor with a statistically significant association with the probability of PA vacancies in the pooled and 2019 models. One more supplier of social care per square kilometre in the local authority is associated with a 4 percentage-point increase in the probability of having at least one PA vacancy. Finally, using an alternative measure of local social care supply does not change the results, with one more job per square kilometre in the wider social care sector in the local authority where the individual employer lives increasing the chances of a PA vacancy by 0.1 percentage points.8
Discussion
Increasing demand for social care and a shift towards the greater personalisation of care have seen a rise in the demand for PAs in recent years. Despite their growing importance in delivering much-needed care, little is known about their recruitment and retention. This article has looked at how PA turnover and vacancies are affected by individual employer and local market characteristics. We found that the total number of staff employed by individual employers is positively associated with both turnover and vacancies in a statistically significant way, as can be expected from bigger employers. Individual employers with a higher number of assistants may also have a variety of different support needs and thus be more likely to experience PA turnover and have staff openings more frequently.
The type of needs also matters for turnover and vacancies. Employers with a learning disability are less likely to have a PA leaving the job or a PA vacancy compared to people with personal care needs. On the other hand, people in need of support with access and mobility are more likely to have vacancies compared to people with personal care needs. Although we cannot know with certainty the severity of individual employers’ needs from the data, we can hypothesise that learning disability needs and personal care needs are more complex on average compared to other types of need. The greatest long-term care support at an aggregate level (both in terms of the number of people receiving care and gross current expenditure) is provided to people with a learning disability in the 18–64 age group and to people with physical support needs in the 65 and above age group (NHS Digital, 2022). If that is the case, PAs providing complex care are more likely to have higher skills that are not easily substitutable or have clearer roles and be under direct employment as opposed to more flexible working arrangements (Skills for Care, 2017c). This suggests that these PAs may be more likely to receive a higher pay rate due to their specialised skills, have a higher workload in terms of guaranteed hours and have an increased feeling of responsibility to remain in the role. Greater guaranteed hours may reduce negative aspects of the job as seen by PAs, for example, lack of overtime pay (Woolham et al, 2019a), and all of these factors in combination are likely to contribute to the stability of the employment relationship and the reduction in turnover and vacancies for this type of support. Overall, differences in the complexity of care between types of need may explain the stability in the relationship between employers and PAs. However, as we do not have very detailed data on need, more research is required into the differences in staff recruitment and retention based on need levels.
In terms of age, we found that older employers were less likely to have open vacancies, but employer age had no statistically significant association with turnover. This could be indicative of the fact that relationships between employers and PAs can often be long-lasting once an initial match is established (Barken et al, 2018; Woolham et al, 2019b).
Our results also showed that turnover and vacancies were more likely when the funding of PAs was via personal health budgets as opposed to direct payments. This association was also sizeable. Although the recruitment and retention of PAs can be challenging under either funding type (Woolham et al, 2019b), our findings suggest that these challenges are relatively higher for those employers in receipt of personal health budgets. In those cases, employing PAs can be challenging due to the difficulty of recruiting people with higher skills who can provide health-related and nursing care but at a similar pay level to other PAs (Woolham et al, 2019b). Thus, pay seems to be an important factor affecting PA recruitment, particularly for overlapping health and social care needs. Further research is required to assess the importance of pay in PA recruitment and retention.
Training support for employers was associated with higher vacancies and turnover. It is possible that the offered training does not provide the necessary help required for more challenging recruitment issues. Equally, though, we acknowledge that this positive association of employer training with turnover and vacancies could be due to the simultaneity in the observed relationship, whereby employers with high levels of staff turnover or need for a larger support team are more likely to seek training. Employers could also become more aware of their needs and thus more specific and stricter regarding their recruitment requirements (Woolham et al, 2019b).
Our study has also shown the importance of local contextual factors in the retention and recruitment of PAs. We found that higher local authority unemployment was associated with lower PA turnover and vacancies. This finding is in accordance with previous qualitative research findings that PA employment is more difficult in the more affluent areas where pay is low in comparison to average local wages (Carmichael and Brown, 2002; Woolham et al, 2019b). Furthermore, we found that PA turnover and vacancies were more likely in areas with a higher supply of social care providers and jobs. This is similar to findings from the home care literature, which show that turnover among home care workers is higher when there are alternative social care employment opportunities in the local market (Ware et al, 2001; Zeytinoglu et al, 2009). Thus, this effect seems to extend to the pool of potential PAs. We could hypothesise that PAs find alternative roles in other social care settings as more attractive substitutes due to higher wages, increased job security or better career opportunities. The ‘poaching’ of PAs has previously been noted in qualitative research (Glendinning et al, 2000b). In such areas, individual employers will find it more difficult to retain assistants or fill vacancies due to the higher local competition for social care workers.
There are several limitations to this study. First, the findings may not be representative of the individual employer population given that this was a small sample of the population of people who employ PAs in the country. It is possible that individual employers with certain characteristics, such as more severe physical and/or mental health needs, might not have been able to participate in the survey and are therefore under-represented. Furthermore, the survey did not include rich demographic information about the participants, other than their age. We acknowledge these limitations of the data. However, the Skills for Care survey is used to inform their national estimates of PAs’ work characteristics and is the only available data source with information on individual employers and their PA recruitment. Overall, the data are therefore useful for analysing this under-researched area. The data additionally did not allow us to distinguish between directly employed and self-employed PAs. If directly employed and self-employed PAs behave differently in their employment decisions, then there could be heterogeneous effects that we have not explored in this study. However, to the extent that any differences between PA types relate to the complexity of their role, then accounting for need type in our analysis will partially control for such differences. In any case, we do not expect this lack of distinction to affect the average estimates presented in this study. Finally, due to the importance of matching between individual employers and PAs in establishing a successful working relationship, employer personality traits, such as leadership style or willingness to compromise, are likely to be important predictors of turnover and vacancies in this market. These measures are rather difficult to capture with the use of such surveys, however, and we were thus unable to include them in the analysis.
A natural extension of this work is to explore the effect of PA characteristics on recruitment and retention, including their socio-economic characteristics and pay. Low pay, (lack of) training and sick pay, and contracts have all been raised as areas for potential improvement by PAs, whereas the bond and close relationship formed with those they are supporting is seen as a positive (Ahlström and Wadensten, 2012; Shakespeare and Stöckl, 2018; Woolham et al, 2019a; 2019b; Manthorpe et al, 2020). Many of these are likely to impact the turnover and recruitment of PA staff. A corresponding survey of PAs took place at the same time, but only current members of staff and not PAs who left employment were surveyed, so data could not be included in this analysis. We have left this for future research.
Conclusion
This study has provided new findings on important aspects of the PA market. Overall, our results highlight that there are challenges in terms of staff recruitment and retention for those who are taking advantage of the greater personalisation of care through the use of cash policies like direct payments and personal health budgets. These challenges are related to the specific features of the PA role, such as the direct relationship between employers and PAs. They are also related to local labour market conditions, which have important implications for local adult social care policy.
These findings therefore have important implications for local government policy and market sustainability. High staff turnover and vacancies can negatively affect the quality of care (Allan and Vadean, 2021), and greater personalisation is often associated with concerns over market stability in other social care markets (Thomas and Hollinrake, 2014; Stevens et al, 2019). In areas with greater competition for the social care workforce and larger pay differentials between PAs and alternative employers, employing PAs could be difficult. This is likely to be even harder for people with personal care and mobility needs and people with more complex social care and healthcare needs. Difficulty in recruiting and retaining PAs is likely to lead to subsequent pressures on the use of NHS resources if people’s needs are unmet. Local government will need to carefully consider adult social care policy, particularly for the local PA market. For example, the alignment of PA pay to local market conditions may be important for recruitment. However, this may not be feasible given the large pay differentials and pressures on social care funding. Thus, policy may need to focus on other aspects of PA employment that can make this market more sustainable. Workers find PA work more fulfilling and rewarding than other forms of social care work (Woolham et al, 2019b; Manthorpe et al, 2020). In this context, raising the profile of PA employment through greater accredited training opportunities, better career progression and better contracting and employment conditions can be equally important in ensuring that the right incentives are in place for PAs to take on and continue in these roles.
Notes
In other countries, PAs are known as ‘personal care assistants’, ‘personal care workers’, ‘domestic care workers’, ‘caregivers’ and other similar names. The definition of their roles will also differ by country but generally includes paid workers who provide help with activities of daily living (ADLs) at home or in institutions (OECD, 2020).
Although we know that some PAs will be recruited from home care agencies (Skills for Care 2017c), we do not have specific evidence on how prevalent that is. In a recent survey, the employment status for 4 per cent of PAs was classified as ‘other’ than direct employment, self-employment or temporary employment (Skills for Care, 2023b), which could include recruitment from agencies.
While, in principle, it is possible that the same employer responded in both surveys, the data collection was not designed as panel data (that is, following the same employers over time). Therefore, we cannot identify them in the data. If there are any employers who responded to both surveys, we expect this number to be relatively small and treat the two samples as cross-sectional and each observation as unique.
To describe needs, the Skills for Care survey follows the Adult Social Care Outcomes Framework (ASCOF) disaggregated classification used by councils in England to measure outcomes of social care (known as primary support reason): physical support, sensory support, support with memory and cognition, learning disability support, mental health support, and social support (DH, 2014).
Currently employed staff members could be friends and family members or others.
Local authority data on PIP entitlement, older population and unemployment rate were taken from the Office for National Statistics (ONS); local authority data on the number of care homes and home care providers were taken from the Care Quality Commission (CQC); and local authority data on the number of social care jobs were taken from Skills for Care estimates.
As a robustness check, we also ran count models on the number of leavers and vacancies, but the results were not qualitatively different. Results are available upon request.
For both the turnover and vacancies outcomes, we ran models in which we tried to estimate the effect of different types of social care, whether that was care homes or home care providers, but we did not find statistically different effects. Results are available upon request.
Funding
This study was supported by the Health Foundation’s Efficiency Research Programme under Award ID 1325587 as part of the Retention and Sustainability of Social Care Workforce project. The Health Foundation is an independent charity committed to bringing about better health and healthcare for people in the UK.
Acknowledgements
The authors would like to thank Skills for Care for sharing with us the Skills for Care Individual Employer and PA Survey data set and providing helpful assistance. The authors are also grateful to participants at the Retention and Sustainability of Social Care Workforce (RESSCW) webinar on 30 September 2021 and the International Long-Term Care Policy Network (ILPN) International Conference in 2022, as well as two anonymous referees, for valuable feedback. The views expressed are solely those of the authors.
Conflict of interest
The authors declare that there is no conflict of interest.
References
Age UK (2023) Personal Budgets and Direct Payments in Social Care, Factsheet 24, London: Age UK.
Ahlström, G. and Wadensten, B. (2012) Enjoying work or burdened by it? How personal assistants experience and handle stress at work, Journal of Social Work in Disability and Rehabilitation, 11(2): 112–27. doi: 10.1080/1536710x.2012.677648
Allan, S. and Vadean, F. (2021) The association between staff retention and English care home quality, Journal of Aging and Social Policy, 33(6): 708–24. doi: 10.1080/08959420.2020.1851349
Banijamali, S., Jacoby, D. and Hagopian, A. (2014) Characteristics of home care workers who leave their jobs: a cross-sectional study of job satisfaction and turnover in Washington State, Home Health Care Services Quarterly, 33(3): 137–58. doi: 10.1080/01621424.2014.929068
Barken, R., Denton, M., Sayin, F.K., Brookman, C., Davies, S. and Zeytinoglu, I.U. (2018) The influence of autonomy on personal support workers’ job satisfaction, capacity to care, and intention to stay, Home Health Care Services Quarterly, 37(4): 294–312. doi: 10.1080/01621424.2018.1493014
Butler, S.S., Simpson, N., Brennan, M. and Turner, W. (2010) Why do they leave? Factors associated with job termination among personal assistant workers in home care, Journal of Gerontological Social Work, 53(8): 665–81. doi: 10.1080/01634372.2010.517236
Carmichael, A. and Brown, L. (2002) The future challenge for direct payments, Disability and Society, 17(7): 797–808. doi: 10.1080/0968759022000039082
Christensen, K. and Manthorpe, J. (2016) Personalised risk: new risk encounters facing migrant care workers, Health, Risk and Society, 18(3–4): 137–52. doi: 10.1080/13698575.2016.1182628
Da Roit, B. and Le Bihan, B. (2010) Similar and yet so different: cash-for-care in six European countries’ long-term care policies, Milbank Quarterly, 88: 286–309. doi: 10.1111/j.1468-0009.2010.00601.x
DH (2014) The adult social care outcomes framework 2014/15 handbook of definitions, www.gov.uk/government/publications/adult-social-care-outcomes-framework-handbook-of-definitions/.
DH (Department of Health) (2011) Working with personalised care: a framework for supporting personal assistants working in adult social care, www.gov.uk/government/publications/working-for-personalised-care-a-framework-for-supporting-personal-assistants-working-in-adult-social-care.
Fleming, G. and Taylor, B.J. (2007) Battle on the home care front: perceptions of home care workers of factors influencing staff retention in Northern Ireland, Health & Social Care in the Community, 15: 67–76. doi: 10.1111/j.1365-2524.2006.00666.x
Francis, J. and Netten, A. (2003) Home Care Workers: Careers, Commitments and Motivations, PSSRU Discussion Paper 2053, Canterbury: Personal Social Services Research Unit, University of Kent.
Glasby, J. and Littlechild, R. (2022) Social Work and Direct Payments, Bristol: Policy Press, doi: 10.51952/9781847425508
Glendinning, C. (2012) Home care in England: markets in the context of under-funding, Health and Social Care in the Community, 20(3): 292–9. doi: 10.1111/j.1365-2524.2012.01059.x
Glendinning, C., Halliwell, S., Jacobs, S., Rummery, K. and Tyrer, J. (2000a) Bridging the gap: using direct payments to purchase integrated care, Health and Social Care in the Community, 8(3): 192–200. doi: 10.1046/j.1365-2524.2000.00242.x
Glendinning, C., Halliwell, S., Jacobs, S., Rummery, K. and Tyrer, J. (2000b) New kinds of care, new kinds of relationships: how purchasing services affects relationships in giving and receiving personal assistance, Health and Social Care in the Community, 8(3): 201–11. doi: 10.1046/j.1365-2524.2000.00242.x
Hall, P., Jefferson, L., Dale, V., Bennett, L., Birks, Y., Bloor, K., et al (2017) Understanding Domiciliary Care in England, London and York: Partnership for Responsive Policy Analysis and Research, King’s Fund and University of York.
Hussein, S., Ismail, M. and Manthorpe, J. (2016) Changes in turnover and vacancy rates of care workers in England from 2008 to 2010: panel analysis of national workforce data, Health & Social Care in the Community, 24(5): 547–56. doi: 10.1111/hsc.12214
Jang, Y., Lee, A.A., Zadrozny, M., Bae, S.H., Kim, M.T. and Marti, N.C. (2017) Determinants of job satisfaction and turnover intent in home health workers: the role of job demands and resources, Journal of Applied Gerontology, 36(1): 56–70. doi: 10.1177/0733464815586059
Jones, K., Forder, J., Caiels, J., Welch, E., Glendinning, C. and Windle, K. (2013) Personalization in the health care system: do personal health budgets have an impact on outcomes and cost?, Journal of Health Services Research & Policy, 18(2): 59–67. doi: 10.1177/1355819613503152
King, D., Wei, Z. and Howe, A. (2013) Work satisfaction and intention to leave among direct care workers in community and residential aged care in Australia, Journal of Aging and Social Policy, 25(4): 301–19. doi: 10.1080/08959420.2013.816166
Manthorpe, J., Woolham, J., Norrie, C. and Samsi, K. (2020) Family matters: personal assistants’ experiences of engaging and working with their employers’ families, International Journal of Care and Caring, 4(4): 497–511. doi: 10.1332/239788220x15917993343915
McGuigan, K., McDermott, L., Magowan, C., McCorkell, G., Witherow, A. and Coates, V. (2016) The impact of direct payments on service users requiring care and support at home, Practice, 28(1): 37–54. doi: 10.1080/09503153.2015.1039973
Melchiorre, M.G., Quattrini, S., Lamura, G. and Socci, M. (2022) Role and characteristics of personal care assistants of frail older people with functional limitations ageing in place in Italy, International Journal of Environmental Research and Public Health, 19(7). doi: 10.3390/ijerph19073969
Moriarty, J., Manthorpe, J. and Harris, J. (2018) Recruitment and Retention in Adult Social Care Services, London: Social Care Workforce Research Unit, King’s College London.
Morris, L. (2009) Quits and job changes among home care workers in Maine: the role of wages, hours, and benefits, Gerontologist, 49(5): 635–50. doi: 10.1093/geront/gnp071
Netten, A., Jones, K., Knapp, M., Fernandez, J.L., Challis, D., Glendinning, C., et al (2012) Personalisation through individual budgets: does it work and for whom?, British Journal of Social Work, 42(8): 1556–73. doi: 10.1093/bjsw/bcr159
NHS Digital (2020) Personal health budgets – Quarter 3 2019–20, https://digital.nhs.uk/data-and-information/publications/statistical/personal-health-budgets/2019-20-q3/personal-health-budgets-q3-2019-20/.
NHS England (2022) Guidance on direct payments for healthcare: understanding the regulations, www.england.nhs.uk/publication/guidance-on-direct-payments-for-healthcare-understanding-the-regulations/.
Norrie, C., Woolham, J., Samsi, K. and Manthorpe, J. (2020) Skill mix: the potential for personal assistants to undertake health-related tasks for people with personal health budgets, Health and Social Care in the Community, 28(3): 922–31. doi: 10.1111/hsc.12923
OECD (Organisation for Economic Co-operation and Development) (2020) Who Cares? Attracting and Retaining Care Workers for the Elderly, Paris: OECD Health Policy Studies, OECD Publishing.
Pavolini, E. and Ranci, C. (2008) Restructuring the welfare state: reforms in long-term care in Western European countries, Journal of European Social Policy, 18(3): 246–59. doi: 10.1177/0958928708091058
Poole, T. (2006) Direct Payments and Older People, London: King’s Fund.
Roland, D., Allan, S., Chambers, E., Smith, D. and Gousia, K. (2022) Personal assistants in England and the factors associated with absenteeism, Frontiers in Public Health, 10: art 970370. doi: 10.3389/fpubh.2022.970370
Rubery, J., Grimshaw, D., Hebson, G. and Ugarte, S.M. (2015) ‘It’s all about time’: time as contested terrain in the management and experience of domiciliary care work in England, Human Resource Management, 54(4): 753–72. doi: 10.1002/hrm.21685
Scourfield, P. (2005) Implementing the Community Care (Direct Payments) Act: will the supply of personal assistants meet the demand and at what price?, Journal of Social Policy, 34(3): 469–88. doi: 10.1017/s0047279405008871
Shakespeare, T. and Stöckl, A. (2018) Metaphors to work by: the meaning of personal assistance in England, International Journal of Care and Caring, 2(2): 165–79. doi: 10.1332/239788218x15187915600658
Skills for Care (2015) Being a Personal Assistant: A Guide for People Who Want to or Are Already Working, Leeds: Skills for Care.
Skills for Care (2017a) Individual Employers and Personal Assistants Workforce Report, Leeds: Skills for Care.
Skills for Care (2017b) Recruitment and Retention in Adult Social Care: Secrets of Success, Learning from Employers What Works Well, Leeds: Skills for Care.
Skills for Care (2017c) Understanding the Employment Status of Personal Assistants, Leeds.
Skills for Care (2019) Being a Personal Assistant, Leeds: Skills for Care.
Skills for Care (2022a) Individual Employers in Receipt of Personal Health Budgets Employing Personal Assistants Report, Leeds: Skills for Care.
Skills for Care (2022b) The State of the Adult Social Care Sector and Workforce in England, Leeds: Skills for Care.
Skills for Care (2023a) Individual Employers and the Personal Assistant Workforce, Leeds: Skills for Care.
Skills for Care (2023b) The State of the Adult Social Care Sector and Workforce in England, www.skillsforcare.org.uk.
Stevens, M., Moriarty, J., Harris, J., Manthorpe, J., Hussein, S. and Cornes, M. (2019) Social care managers and care workers’ understandings of personalisation in older people’s services, Working with Older People, 23(1): 37–45. doi: 10.1108/wwop-11-2018-0022
Thomas, W. and Hollinrake, S. (2014) Economic and demographic challenges for social care, Journal of Health Organization and Management, 28(5): 653–73. doi: 10.1108/jhom-10-2013-0223
TLAP (Think Local Act Personal) (2022) The Forgotten Workforce: Recruiting and Retaining Personal Assistants: A Survey of People Supported by a Personal Assistant, London: TLAP, www.thinklocalactpersonal.org.uk/Latest/Personal-Assistant-Survey-The-Forgotten-Workforce/.
Ungerson, C. (1999) Personal assistants and disabled people: an examination of a hybrid form of work and care, Work, Employment and Society, 13(4): 583–600. doi: 10.1017/s0950017099000422
Ungerson, C. (2004) Whose empowerment and independence? A cross-national perspective on ‘cash for care’ schemes, Ageing and Society, 24(2): 189–212. doi: 10.1017/S0144686X03001508
Ware, P., Matosevic, T., Forder, J., Hardy, B., Kendall, J., Knapp, M., et al (2001) Movement and change: independent sector domiciliary care providers between 1995 and 1999, Health and Social Care in the Community, 9(6): 334–40. doi: 10.1046/j.1365-2524.2001.03111.x
Woolham, J., Norrie, C., Samsi, K. and Manthorpe, J. (2019a) The employment conditions of social care personal assistants in England, The Journal of Adult Protection, 21(6): 296–306. doi: 10.1108/jap-06-2019-0017
Woolham, J., Norrie, C., Samsi, K. and Manthorpe, J. (2019b) Roles, Responsibilities, and Relationships: Hearing the Voices of Personal Assistants and Directly Employed Care Workers, London: NIHR Policy Research Unit in Health and Social Care Workforce, The Policy Institute, King’s College London. doi: 10.18742/pub01-005
Zeytinoglu, I.U., Denton, M., Davies, S. and Plenderleith, J.M. (2009) Casualized employment and turnover intention: home care workers in Ontario, Canada, Health Policy, 91(3): 258–68. doi: 10.1016/j.healthpol.2008.12.004
