PROJECT PR-PRU-1217-21101.22
Related Output(s):
Background:
Workforce capacity is a key issue in Adult Social Care (ASC). With the number of filled posts declining for the first time and demand for care further increasing (Skills for Care 2022), there are concerns about the sector’s capacity to sustainably meet future demand as well as continue to provide good quality care. Emerging evidence points to pay and contractual conditions as important drivers of labour supply and stability (Moriarty et al. 2018, Vadean and Saloniki 2021) as well as care quality (Towers et al. 2021). There is, however, limited understanding of the extent to which improving employment conditions, such as increasing the wage floor or improving wage progression for skills and experience, are effective in improving workforce and care outcomes.
Aims:
This project aims to understand the influence that employment conditions have on workforce and care outcomes in adult social care (ASC). The project will aim to examine:
The relationship between pay, pay distribution and pay compression in ASC and avoidable hospitalisations (e.g. hip fractures and pulmonary infections) as well as ASCOT social care related quality of life.
Methods:
We will build on findings from work in the Retention and Sustainability of Social Care Workforce (RESSCW) project as well as a previous ASCRU study on wage differentials and wages elasticities in ASC to examine the relationship between pay and workforce outcomes (e.g. job separation). We will use data from the Adult Social Care Workforce Data Set (ASC-WDS) (i.e. years 2016 to 2022). In addition to worker hourly wages, we also aim to assess employer wage distributions (e.g. percentiles) and pay compression (i.e. year-on-year difference in the wage distribution) and their relationship to job separations.
As the ASC-WDS does not include information on the destination of job leavers, our previous analysis focused mainly on estimating wages elasticities of labour supply to the firm. We will explore methods of estimating of wages elasticities of labour supply to the sector based on wage elasticities of all job separation and the share of job separations out of ASC (as reported by Skills for Care and The Health Foundation). We will also examine variation in wage elasticities by different characteristics (e.g. qualifications, experience and training), job roles (e.g. registered nurses and care workers) as well as over time.
There are no secondary data for England linking social care workforce/job characteristics to service users’ outcomes. We will create a relevant dataset by linking SCRQoL scores from the Adult Social Care Survey (ASCS) – aggregated at upper-tier local authority (LA) level by user group (e.g. age, service type and care setting) – to staffing, job quality (e.g. pay and pay distribution) and care establishment data from the Adult Social Care Workforce Data Set (ASC-WDS) – aggregated at LA-level by the same groups. Skills for Care (SfC) have agreed to provide staffing estimates at LA-level for seven years (2015 to 2021). These data will be supplemented with data on care needs, income and wealth, deprivation, and labour market characteristics from various national returns (e.g. Census 2011/2021, the Index of Multiple Deprivation, Department of Work and Pensions, Valuation Office Agency, Care Quality Commission).
An additional care outcome measure for social care users is the count and length of avoidable hospitalisations (e.g. hip fractures and pulmonary infections). The frequency and length of hospitalisation episodes from the Hospital Episodes Statistics (HES) can be aggregated at LA-level by age group. As HES does not collect information on social care recipient status and care setting, we cannot link these hospital indicators directly to social care service user populations and in turn social care provider/workforce characteristics. Rather we will focus on hospital populations that are more likely to have social care needs (e.g. frail older populations, people with multiple conditions etc.).
Statistical methods will be used to estimate the above relationships, with controls for a range of confounding factors, and panel fixed effects models to account for unobserved heterogeneity. The SCRQoL measures will be adjusted using techniques previously developed at PSSRU to reflect the ‘added value’ by social care services (Forder, Malley et al. 2015; Yang, Forder and Nizalova 2017).
We will build on previous and current work of the research team on: a) supply of social care services and avoidable hospitalisations using linked HES with CQC data (Forder, Vadean and Allan, forthcoming); b) the ‘adjustment’ of SCRQoL to derive the impact of care services (IIASC report 2018); c) the impact of staffing, turnover and vacancies on care home quality (Allan, Vadean 2020), d) the impact of wages on care home quality ratings (Allan, Vadean, forthcoming); and e) measuring the productivity of residential long-term care (Yang, Forder and Nizalova 2017).
As a longer-term plan, we will begin to explore analysis on individual-level service-user data (from ASCS) linked to establishment-level data from the ASC-WDS. For this data-linkage to be possible, establishment ID information will need to be collected in ASCS. We have started discussions in this respect with NHS Digital and DHSC.
Limitations
LA-level analysis may not adequately identify causal relationships.
Florin Vadean (Co-Lead), Stephen Allan (Co-Lead), Julien Forder, Olena Nizalova
