PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of blackwellopenThis ArticleFor AuthorsLearn MoreSubmit
Health & Social Care in the Community
 
Health Soc Care Community. 2017 September; 25(5): 1607–1619.
Published online 2016 April 24. doi:  10.1111/hsc.12354
PMCID: PMC5574010

The factors associated with care‐related quality of life of adults with intellectual disabilities in England: implications for policy and practice

Stacey Rand, MAcorresponding author 1 and Juliette Malley, MA MPhil 1 , 2

Abstract

Over the last three decades, quality of life (QoL) has been advocated as an indicator of social care outcomes for adults with intellectual disabilities. In England, the Adult Social Care Survey (ASCS) is conducted annually by local authorities to contribute to the evidence base of the care‐related QoL of people receiving publicly funded adult social care. This study explores relationships between QoL and non‐care‐related factors to identify relationships that could inform social care policy and practice. Cross‐sectional data collected from 13,642 adults who participated in the 2011 and 2012 ASCS were analysed using regression to explore the factors associated with QoL measured using the Adult Social Care Outcomes Toolkit (ASCOT). Self‐rated health, rating of the suitability of home design and anxiety/depression were all found to be significantly associated with ASCOT. Other individual and survey completion factors were also found to have weak significant relationships with ASCOT. The models also indicate that there was an increase in overall ASCOT‐QoL and in five of the eight ASCOT domains (Personal comfort and cleanliness, Safety, Social participation, Occupation and Dignity) between 2011 and 2012. These findings demonstrate the potential value of QoL data for informing policy for people with intellectual disabilities by identifying key factors associated with QoL, the characteristics of those at risk of lower QoL, and QoL domains that could be targeted for improvement over time. Future research should establish causal relationships and explore the risk adjustment of scores to account for variation outside of the control of social care support.

Keywords: adult learning disability services, Community Care and Learning Disabilities, intellectual disabilities, learning disabilities, quality of life

What is known about this topic

  • Quality of life (QoL) has been advocated as an indicator to assess social care for adults with intellectual disabilities.
  • Variation in QoL is associated with individual, environmental or survey‐related factors that are unrelated to the effect of social care.
  • Interpreting and applying QoL data to improve policy and practice would be supported by an understanding of the associations between QoL and these non‐care‐related factors.

What this paper adds

  • After controlling for other factors, there is evidence for an improvement over time in five of the eight domains of care‐related QoL.
  • Health‐related variables are key predictors of care‐related QoL.
  • Associations between individual characteristics and QoL identify areas for further exploration of causal relationships.

Introduction

Over the last three decades, quality of life (QoL) has been advocated as an indicator of the quality and effectiveness of social care policy and practice for adults with intellectual disabilities (Schalock et al. 2002, 2008, Schalock 2004, Verdugo & Schalock 2009). The measurement of QoL has been advanced to enable those who use social care services to have a voice in defining the quality of care, driving provider‐level change and shaping policy strategy (Schalock et al. 2002, Schalock 2004, Brown & Brown 2005, Verdugo & Schalock 2009, Verdugo et al. 2012). This trend is situated within the wider movement towards outcomes‐based performance management across the public sector (Heinrich 2003, Bovaird 2012).

The UK has been at the forefront of this outcomes‐based approach (Bovaird 2012) and has developed an ‘outcomes framework’ for the adult social care sector in England, known as the ‘Adult Social Care Outcomes Framework’ (ASCOF) (Department of Health, 2011). Within the ASCOF, the Adult Social Care Outcomes Toolkit (ASCOT) provides a measure of care‐related QoL (Netten et al. 2011, 2012, Malley et al. 2012). Social care in England includes a range of support services designed to enable individuals to undertake everyday activities: for example, personal hygiene and social participation. The intention is that the indicators within ASCOF are used to provide a social care evidence base to support local and central policy making, as well as to improve the quality of services by guiding local management decisions and to increase transparency of performance in each locality (Department of Health, 2012a, 2013). The inclusion of the ASCOT measure in the ASCOF aims to ensure that the voice of service users is at the heart of policy decisions and innovations designed to improve social care policy and practice (Department of Health, 2011).

Although there are strong arguments in favour of using QoL to inform policy and drive systems‐level and organisational change (Schalock et al. 2002, 2008, Verdugo & Schalock 2009, Verdugo et al. 2012, van Loon et al. 2013), there are measurement challenges associated with measuring individual QoL, for example, response bias and potential exclusion of people with severe intellectual disabilities (Hatton 1998, Perry & Felce 2002, Janssen et al. 2005, Schmidt et al. 2010). There is also the ‘attribution problem’ of how to determine the impact of social care on QoL to the exclusion of other (non‐care‐related) influences. The attribution problem is particularly important when considering how to use cross‐sectional survey‐based outcome data. Without some understanding of the part that social care services play in determining individual QoL, we cannot hope to identify ways of improving QoL through the provision of social care support and services (Malley & Fernández 2010). Indeed, understanding the impact of services on QoL was one of the areas that organisations struggle with when trying to find ways to use the ASCOT measure to improve services (Heath et al. 2015).

This production of welfare framework is useful for exploring the influence of various factors in determining outcomes for service users (Davies & Knapp 1981, Malley & Fernández 2010). This framework posits that the final outcome of social care (i.e. care‐related QoL) is causally influenced by a range of ‘non‐resource inputs’ including socio‐demographic and environmental characteristics. Many non‐resource inputs can be considered ‘non‐care‐related’ factors outside of the direct control of social care. Studies have shown the importance of non‐resource inputs in predicting final outcomes in social care (Forder et al. 2014, Malley & Fernandez 2014), but, to our knowledge, few studies have specifically explored the relationship between these factors and QoL outcomes in people with intellectual disabilities.

One such study compared the effects of various care‐related and non‐care‐related characteristics on QoL of adults with intellectual disabilities in the Netherlands (Claes et al. 2012) and found that social care directly accounted for only 10% of the variance. By contrast, 44% of the variance was accounted for by personal characteristics and a further 8% was associated with environmental factors that were beyond the direct control of services (Claes et al. 2012). Although this particular study did not explore survey‐related factors, such as the effect of help to complete questions, these are also known to influence self‐reported care‐related QoL (van Leeuwen et al. 2014). The relatively small contribution of social care to QoL found in the Dutch sample highlights the importance of developing a better understanding of the relationship between non‐care‐related factors and QoL to support the interpretation and application of QoL to policy and practice in the English and other contexts.

As the outcomes‐based approach is adopted by organisations and governments, it seems likely that QoL measures will play an ever more important role in informing quality and performance improvement. Very little research has looked specifically at how governments can foster the use of QoL data for this purpose (Razik & Beecham 2014). The current literature has, however, identified that, alongside organisational change, a factor promoting the adoption of QoL data for performance improvement is detailed analysis of the relationship between various factors and QoL (Razik & Beecham 2014). Given the limited evidence about the relationship between ASCOT‐QoL and other factors, a more complete understanding of the association between survey‐related factors and individual or environmental characteristics on QoL would seem to be a first step towards applying the outcomes evidence to drive systems‐ or organisational‐level improvement.

In this paper, we use data from the 2011 and 2012 Adult Social Care Survey (ASCS) to explore the factors that affect overall and individual domain scores of ASCOT‐QoL for adults with intellectual disabilities who use publicly funded community‐based social care services in England. Our choice of independent variables is driven by the theoretical framework we have outlined and the availability of data. The aim of this analysis is to support the interpretation and use of the ASCS data to inform local and national policy and practice by adding to the evidence base around the factors associated with the QoL of adults with intellectual disabilities. Furthermore, as a case study, it will add to the emerging evidence on how organisations may use QoL data to drive systems‐ and organisational‐level improvement.

Methods

The Adult Social Care Survey (ASCS)

In England, ASCOF data are collected and reported at the level of local authorities (LAs), as it is these organisations who commission (and in some cases provide) local social care services. A number of outcome measures in the ASCOF are collected from a sample of users of publicly funded social care in the annual Adult Social Care Survey (ASCS). The ASCS was first introduced in 2011 and is conducted by LAs according to guidance set by the Health and Social Care Information Centre (HSCIC) on sampling, data collection and management processes (HSCIC, 2010, 2011). For this study, we use the data collected in 2011 and 2012.

The ASCS questionnaire includes the ASCOT measure of care‐related QoL (Netten et al. 2011, 2012, Malley et al. 2012), which is a self‐report instrument with one item for each of eight QoL domains (see Table 1). An easy‐read version of the questionnaire was developed by the Personal Social Services Research Unit and the Tizard Centre at the University of Kent (Caiels et al. 2010a, Malley et al. 2010). The development, construct validity and internal reliability of the easy‐read version of the ASCOT instrument are reported elsewhere (Caiels et al. 2010b, Rand 2014).

Table 1
Definition of ASCOT‐QoL domains

Each QoL attribute is measured by a single item with four response options that correspond to: the ideal state, which is the preferred situation for the respondent; no needs, where the individual's needs are met but not to the desired level; some needs, where the needs do not have immediate or long‐term health implications; and, high‐level needs, where the individual's needs have immediate or long‐term health implications. Each item is scored as either the ideal state (3), no needs (2), some needs (1) or high‐level needs (0) and combined to create a continuous scale score from 0 to 24. Although utility weights are available for ASCOT (Potoglou et al. 2011, Netten et al. 2012), we do not apply these here because the weights have not been developed for the easy‐read version and the ASCOF reports the equally‐weighted rather than the utility‐weighted score.

The ASCS also includes items to capture factors identified as being theoretically important in determining QoL to support the interpretation of ASCOT‐QoL; these items were selected based on insights from the production of welfare model (Davies & Knapp 1981, Malley & Fernández 2010) and discussions with practitioners from the field of adult social care. These items include socio‐demographics, health status (self‐rated health on a 5‐point scale, pain/discomfort and anxiety/depression on 3‐point scales), disability [difficulty with eight activities of daily living (ADLs) or instrumental activities of daily living (IADLs)] and appropriateness of the design of the home for the individual's needs on a 4‐point scale. Data were also collected on receipt of informal help from someone inside or outside of the household and payments for additional social care support. There are also items capturing whether the questions were completed with assistance from someone else, as well as the source and type of assistance. The survey guidance and questionnaires are available online (HSCIC, 2010, 2011).

Ethics approval for the ASCS was obtained from the Social Care Research Ethics Committee (REC) in England. This is a REC outside of the UK National Health Service responsible for reviewing adult social care research involving sites in England and another UK country.

Sampling

The survey sample was drawn from adults who receive publicly funded social care stratified by client group and age. For this study, a sub‐sample was extracted from the data set using the following inclusion criteria: individuals whose social care records have intellectual disabilities as their primary or secondary client group, who were sent an easy‐read version of the ASCS questionnaire, and were not in residential or nursing care at the time of the survey. Of the 32,395 people sampled with these characteristics, 13,642 (42.1%) completed the questionnaire and were included in the sample analysed in this article. This excludes cases where the primary and secondary client groups (n = 19), the response status (n = 1) or residential status (n = 2) were missing.

Statistical modelling

The data were analysed using Stata version 12. Multivariable regression was used to explore the individual characteristics and survey completion factors associated with QoL while controlling for LA‐level variation. The dependent variables are overall ASCOT‐QoL score (treated as a continuous variable) and the score for each of the eight ASCOT domains (treated as categorical variables). The predictor variables were selected from those available in the ASCS (see Table 2). These were considered according to the theoretical framework provided by the Production of Welfare model (Davies & Knapp 1981), which has been further developed to provide a theoretical basis for understanding the factors that influence social care outcomes other than the quality or quantity of social care support (Malley & Fernández 2010, Forder et al. 2014, Malley & Fernandez 2014). As survey administration factors have been found to affect responses among older adults with physical and sensory impairment (van Leeuwen et al. 2014), these were also considered. Dummy variables for the survey year and local authorities were included to capture changes by geographical location or over time.

Table 2
Predictor variables considered in regression models (n = 13,642)

Including both home design and the number of I/ADLs completed with difficulty meant that the relationship between I/ADLs and QoL was positive. This relationship suggests that I/ADL with difficulty variable, when included with home design, was potentially capturing the effect of social care support, which we would expect to have a positive relationship with QoL. This scenario is plausible as social care support in England is allocated on the basis of social care need, which is partly determined according to the ability to undertake I/ADLs (Department of Health, 2010a). As the primary aim of this analysis was to explore the non‐care‐related factors associated with QoL, it was decided to omit the I/ADL variable from the models.

The overall ASCOT‐QoL score was modelled using OLS estimation. Ordered logit (ologit) regression was initially employed for the models with the eight ASCOT domains as the outcome variables. For these analyses, the responses for the lowest two response categories (0, 1) were collapsed into a single category (1) due to the small proportion of responses in the ‘high needs’ category (see Table 3). Only those variables significant at the 1% level were retained in the final OLS and ologit models.

Table 3
Responses to the ASCOT by QoL domain(Easy Read community‐based version from the 2011 and 2012 ASCS) (n = 13,642)

Due to relatively high rates of missingness (see Table 2), which can cause issues such as a loss of precision in regression models or non‐response bias, multiple imputation using chained equations was applied using MI Impute Chained to generate an imputed data set (van Buuren et al. 1999, Statacorp, 2011). The analysis of the imputed data set generated results similar to the complete case data; therefore, the analyses presented in the article are based on the complete case analysis only.

Results

The sample characteristics are reported in Table 2. Of the respondents, 54.9% were male, 8.1% aged 65 years or older, and 89.7% reported ethnicity as white British or another white background. Of the cases included in the analysis, 10.3% were completed by proxy. The ASCOT‐QoL responses by domain are shown in Table 3. The mean ASCOT‐QoL score was 20.76 (2.87 SD). The Shapiro–Wilk's test of normality (P < 0.001) indicates that the ASCOT‐QoL score was not statistically normal; this was confirmed by visual inspection of the distribution, which was negatively skewed.

The OLS model with equally weighed ASCOT‐QoL score as the dependent variable is shown in Table 4. The model failed diagnostic tests for homoscedasticity, so we report robust standard errors using the Huber–White sandwich estimators (Huber 1967, White 1980). The OLS model also failed the Ramsey RESET test (Ramsey 1969) and Pregibon's link test (Pregibon 1980), which indicates possible model specification error and omitted variable bias. As the ASCOT score was negatively skewed, which is typical of multidimensional health‐related QoL measures (Austin 2002, Basu & Manca 2012, Pullenayegum et al. 2012), we explored forms such as beta regression (Ferrari & Cribari‐Neto 2004) with transformation of the dependent variable at 0 and 1 (Smithson & Verkuilen 2006) and fractional response models (Papke & Wooldridge 1996, Baum 2008) in an attempt to improve the fit of the model. However, the results were similar to those from the OLS and did not improve model fit statistics, so are not presented here.

Table 4
Multiple regression (OLS) with ASCOT quality of life score as the outcome variable

The results of the eight models with the rating of QoL in each ASCOT domain as the dependent variable are shown in Table 5. The models were initially calculated using ordered logistic regression. The assumption of parallel regression, which was tested for using the user‐written omodel Stata command (Wolfe & Gould 1998), was found to fail for all eight models. Generalised ordered logit (gologit) models, which do not assume parallel regression (Fu 1998), were therefore estimated using the user‐written gologit2 command (Williams 2006). The gologit models can be interpreted as two logistic regression models: response option 1 (some/high needs) versus response option 2/3 (no needs/ideal state) and response option 1/2 (some/high or no needs) versus response option 3 (ideal state). For parsimony, we used a partial proportional odds model and, where warranted, constrained some independent variables to meet the parallel regression assumption.

Table 5
Estimation results from the gologit models by ASCOT domain

Three of the four variables capturing underlying health condition‐related and environmental characteristics (self‐perceived health, rating of anxiety/depression and rating of the suitability of the home design) had both a strong negative relationship with QoL score in the OLS model and across all eight gologit models. Those with poorer self‐reported health, worse rating of home design or higher levels of anxiety/depression are more likely to report lower QoL. In instances where the parallel lines assumption was violated, the coefficient was more negative for the 1 vs. 2, 3 contrast than for the 1, 2 vs. 3 contrast. This indicates that those who reported poorer health, less adequate home design for their needs and higher levels of anxiety and depression are especially likely to report poor QoL. The remaining health‐related variable (self‐reported pain) did not reach significance in the OLS model (Table 4) and was only found to be significantly negatively associated with QoL in the gologit model with Personal safety as the outcome variable (Table 5).

The socio‐demographic variables had weak associations with overall ASCOT‐QoL (Table 4). Being male was associated with lower QoL, whereas being aged 65 years and over, or from a white ethnic background were associated with higher QoL. These variables only had significant associations with some of the QoL domains, although the direction of the relationship was consistent across the gologit and OLS models (see Table 5).

Receipt of practical help from someone inside the household, which is taken as an indicator of unpaid care by a partner or relative, was found to have a significant positive association with overall ASCOT score in the OLS model (Table 4). Those who reported that they had practical help from someone inside of the household were more likely to report higher QoL for Dignity. Likewise, there were significant positive relationships between practical help from someone inside the household and Accommodation and Social participation; the coefficients indicate that those who received practical help from someone in the home were less likely to report the lowest QoL states. There was also a significant positive coefficient for the 1 vs. 2, 3 contrast in the model with feeling clean and comfortable (Personal cleanliness) as the outcome variable, which suggests that those who received practical help from someone in the same household were less likely to rate their QoL as poor in this domain. Unlike the other five domains, where significant positive coefficients were observed, there was a significant negative coefficient for the 1, 2 vs. 3 contrast for Control over daily life; those who received help from someone inside the household were less likely to rate their QoL at the top level (‘ideal state’).

Practical help from someone outside of the household did not meet the inclusion criteria in the OLS model. It was, however, included in the gologit model with Social participation as the outcome variable. A significant positive association between receipt of help and QoL was observed with a larger coefficient for the 1 vs. 2, 3 than for the 1, 2 vs. 3 contrast. Respondents who received help from outside of the household were less likely to rate their QoL as poor for Social participation.

Completion of the survey by proxy was positively associated, and help to complete the survey from a care worker was negatively associated with, overall QoL in the OLS model (Table 4). In the gologit models, responses by proxy were more likely to report worse QoL for Occupation and Control over daily life, but had a significantly increased likelihood of reporting better QoL for Accommodation, Personal safety and Social participation. The negative association with overall ASCOT score observed in the OLS model is likely due to the comparatively large effect size for proxy response in the Control over daily life domain. Help to complete the survey from a care worker had both significant positive associations (Food and drink, Accommodation, Personal safety and Dignity) and negative associations (Social participation and Control over daily life) with QoL score across the gologit models. The significant positive relationship in the OLS model is likely to be due to the moderate‐large effect size in the Dignity domain. Varying associations, some of different sign, were also observed in the other survey completion variables included only in the gologit models (see Table 5).

The dummy variable for survey year with 2011 as the base category was found to have a significant positive association with overall ASCOT‐QoL score in the OLS model. This variable was also included in five of the gologit models with the following ASCOT domains as the outcome variables: Personal cleanliness, Personal safety, Social participation, Occupation and Dignity. In all five models, it was found that those who completed the survey in the 2012 data collection were more likely to rate their QoL as better than those who completed the 2011 survey. The survey year variable only failed the parallel regression assumption in one model (Social participation), where the coefficient for the 1, 2 vs. 3 is smaller than the 1 vs. 2, 3 contrast. This suggests that respondents were less likely to report low QoL for Social participation in the 2012 survey.

Discussion

This study explores the individual, environmental and survey‐related characteristics associated with QoL of people with intellectual disabilities who access publicly funded social care in England using the Adult Social Care Survey (ASCS) data. The ASCS aims to capture the service users’ perspective of the outcome of LAs’ activities as commissioners (and in some cases providers) of services, as well as strategic managers of the availability and diversity of social care provision through local policy (Department of Health, 2012b). To this end, it is difficult to disentangle the role of LAs and formal social care from other influences on the QoL of people with intellectual disabilities and, although the analysis presented here does not solve the attribution problem, it does establish the associations between individual, environmental and survey‐related factors that could support the interpretation and use of the ASCS data to inform local and national policy and practice.

An important finding is that, after controlling for a range of variables that capture individual‐level and survey completion factors that may explain variation in ASCOT‐QoL scores, three health‐related variables evaluated in this study are found to be strongly associated with QoL. This is consistent with previous studies that have found a relationship between health and QoL (Schalock et al. 1994, Schalock 2004, Lehmann et al. 2012). This is not surprising; health conditions may contribute to social care needs, and indeed many measures of QoL include physical and/or psychological health as a domain (Townsend‐White et al. 2012). The suitability of the design of the home for an individual's needs may not only capture health and social care need (as evidenced by the relationship with I/ADLs) but also the quality of housing or the home environment in relation to those needs. Housing quality has been found in previous research to be an important indicator associated with QoL; well‐designed housing enables individuals to live in a comfortable, safe and accessible environment (Evans et al. 2002, Wahl et al. 2009). These strong associations between QoL and overall health, psychological health and home design provide tentative support for a policy strategy that recognises the interrelationships between health, social care and housing needs and outcomes.

The findings could be used to identify broad groups who may be at risk of lower QoL or may benefit from strategically targeted support at the local or national level. To illustrate this we draw on two examples. First, the finding that informal care from a co‐resident carer is associated with higher QoL in five of the eight ASCOT domains supports the premise of the carers’ strategy in England (Department of Health, 2010b, 2014) that unpaid care‐giving, particularly by co‐resident carers, helps to support the QoL of the people they care for. However, the negative relationship with practical help from a co‐resident informal carer and Control over daily life, particularly at the ‘ideal state’, does raise the question of whether there may be scope for social care to work alongside unpaid carers to promote independence and increased control over everyday choices. Second, the association between non‐white ethnic backgrounds and worse ratings of Control over daily life and Occupation highlight areas of QoL that could be targeted by policy or practice interventions for non‐white groups. Further research is, however, warranted to understand causal relationships and the effectiveness of such interventions.

The ASCS data collection could also be used to evaluate and drive systems‐level change in QoL over time. After controlling for individual characteristics and survey administration factors, there was a small significant increase in aggregate user‐reported QoL from 2011 to 2012 in five of the eight ASCOT‐QoL domains, particularly at the ‘ideal state’ for Social participation and Occupation domains. This provides tentative evidence for an increase in the QoL of users of publicly funded social care services over time. It should, however, be noted that there has been a 0.2% real term reduction in spending on publicly funded social care for adults under the age of 65 in England between 2010/2011 and 2012/2013 that has meant a reduction in the availability of publicly funded care (National Audit Office, 2014). Due to the nature of the ASCS sample, it is not possible to draw conclusions about the QoL of the wider population of adults with intellectual disabilities in England, or for those individuals who may have lost access to publicly funded services as a result of a reduction in public spending over this period.

A key requirement for outcomes data to drive systems‐ or organisational‐level change is that the data are routinely applied to create an organisation or system that targets resources to maximise positive effects (Schalock et al. 2002). Such outcomes‐based organisational change is a complex process with various steps at which there may be challenges or facilitators (Schalock & Verdugo 2012, Razik & Beecham 2014). In the English context, a considerable challenge is the limited availability of resources at the local level to conduct in‐depth analysis of outcomes data to inform managerial and organisational practice (Heath et al. 2015). If local authorities are to use QoL to drive change, there is a need for investment to support them in their analysis, interpretation and use of data.

Alongside the use of QoL data to inform outcomes‐based strategy, for example through the identification of ‘at‐risk’ groups and targeting of particular areas of QoL for improvement, the development of a risk‐adjusted QoL indicator would allow a fairer comparison of QoL scores between councils in England or over time to evaluate the impact of social policy strategy and local interventions as intended by policy makers (Department of Health, 2011, 2013) and practised by LAs (Heath et al. 2015). Indeed, the analysis presented in this article resembles the types of models used for ‘risk adjustment’ (Iezonni 2013) and could form the basis of a risk‐adjusted indicator for the outcomes framework. Further work would be needed to explore other relevant factors, and to decide whether to exclude some factors on theoretical grounds. For example, it could be argued that ethnicity and home design should be excluded from risk adjustment because of difficulty determining, in the former's case, whether differences are due to disparities in survey administration or reporting rather than experience (Gray et al. 2014) and, in the latter's, its plausible designation as a factor that could be within the control of social care services through the delivery of equipment and adaptations. Our findings also raise the question of whether risk adjustment should be carried out by QoL domain rather than on the overall QoL scale level due to the differences in the pattern of relationships between some of the variables explored here (particularly the survey completion variables) and the different ASCOT‐QoL domains (Ara et al. 2011).

Limitations

There are some limitations to this study. First, due to the cross‐sectional data collection, the associations cannot be interpreted as causal relationships. Evidence of causality would further support the interpretation and application of these findings (Bovaird 2012). Second, the analysis was limited to those variables available in the ASCS data set, and the Ramsey RESET test (Ramsey 1969) and Pregibon's link test (Pregibon 1980) indicated omitted variable bias. One potential source of bias is the omission of service intensity and type of service in the analysis due to a lack of good quality data of the care received by each individual in the sample. As LAs allocate publicly funded social care support to individuals on the basis of social care need, which includes assessment of health conditions, ability to perform ADL and availability of informal care (Department of Health, 2010a), the omission of intensity may mean the regression coefficients for indicators of social care need in the model, such as the health, home design and practical help variables, are underestimated where the sign is negative and overestimated where positive. Furthermore, due to the relatively modest response rate of 42.1%, some caution is required in application of conclusions to the wider population of publicly funded social care service users. Finally, the analysis only focuses on the estimated 12% of adults with intellectual disabilities who receive publicly funded community‐based social care services (Emerson et al. 2012). It does not include individuals who do not meet the eligibility criteria for publicly funded social care, who only purchase social care support privately, or adults who are in residential or nursing care homes.

Despite these limitations, the analysis presented in this article is based on a large, national, randomly selected sample. It provides results with good generalisability to the population of people who use publicly funded adult social care support in England that should be of particular interest to local and national policy makers, and as a study of some of the measurement challenges inherent in an outcomes‐based approach to policy and planning.

Conclusion

This study demonstrates the potential value of the ASCS to inform and evaluate local and national policy and practice in England. Health, psychological well‐being and the design of an individual's home are all possible targets for further research to establish causal relationships and potential targets for local and national policy makers to improve outcomes for people with intellectual disabilities. This analysis has also highlighted groups of adults with intellectual disabilities who may be at risk of lower QoL, and identified particular domains of QoL that could be strategically targeted to improve the QoL of users of publicly funded social care support. There is also the potential to use these data to compare local policy and practice. This study has identified potential factors that could be considered to generate a risk‐adjusted ASCOT‐QoL measure for such a purpose, although further development of the models proposed here is required.

Acknowledgements

The research on which this article is based was funded by the Policy Research Programme in the Department of Health and undertaken by researchers at the Quality and Outcomes of person‐centred care Research Unit (QORU). The views expressed here are those of the authors and are not necessarily shared by any individual, government department or agency. We acknowledge and thank Professor Julien Forder, Dr Karen Jones and the two anonymous peer reviewers for their advice and comments on earlier drafts of the manuscript.

References

  • Ara R., van Hout B., Kearns B. & Brazier J. (2011) Case‐Mix Methodology for the NHS Outcomes Framework. GP Patient Survey Questionnaire Data. Policy Research Unit in Economic Evaluation of Health and Care Interventions, University of Sheffield and University of York, Available at www.eepru.org.uk.
  • Austin P.C. (2002) A comparison of methods for analyzing health‐related quality‐of‐life measures. Value Health 5, 329–337. [PubMed]
  • Basu A. & Manca A. (2012) Regression estimators for generic health‐related quality of life and quality adjusted life years. Medical Decision Making 32, 56–69. [PubMed]
  • Baum C. (2008) Stata tip 63: modeling proportions. Stata Journal 8, 299–303.
  • Bovaird T. (2012) Attributing outcomes to social policy interventions: ‘gold standard’ or ‘fool's gold’ in public policy and management? Social Policy & Administration 48, 1–23.
  • Brown R.I. & Brown I. (2005) The application of quality of life. Journal of Intellectual Disability Research 49, 718–727. [PubMed]
  • van Buuren S., Boshuizen H.C. & Knook D.L. (1999) Multiple imputation of missing blood pressure covariates in survival analysis. Statistics in Medicine 18, 681–694. [PubMed]
  • Caiels J., Fox D., McCarthy M. et al (2010a) Developmental Studies for the National Adult Social Care User Experience Survey: Technical Report. PSSRU Discussion Paper 2724. Personal Social Services Research Unit, University of Kent, Cantebury.
  • Caiels J., Fox D., McCarthy M. et al (2010b) Developmental Studies for the National Adult Social Care User Experience Survey: Technical Report. Personal Social Services Research Unit, The University of Kent, Canterbury.
  • Claes C., Van Hove G., Vandevelde S., van Loon J. & Schalock R. (2012) The influence of supports strategies, environmental factors, and client characteristics on quality of life‐related personal outcomes. Research in Developmental Disabilities 33, 96–103. [PubMed]
  • Davies B. & Knapp M. (1981) Old people's Homes and the Production of Welfare. Routledge and Keegan Paul, London.
  • Department of Health (2010a) Prioritising Need in the Context of ‘Putting People First’: A Whole System Approach to Eligibility for Social Care – Guidance on Eligibility Criteria for Adult Social Care, England 2010. Department of Health, London.
  • Department of Health (2010b) Recognised, Valued and Supported: Next Steps for the Carers Strategy. Department of Health, London.
  • Department of Health (2011) Transparency in Outcomes: A Framework for Quality in Adult Social Care. Department of Health, London.
  • Department of Health (2012a) The 2012/13 Adult Social Care Outcomes Framework. Department of Health, London.
  • Department of Health (2012b) Caring for Our Future: Reforming Care and Support. HM Government, London.
  • Department of Health (2013) The Adult Social Care Outcomes Framework 2014/15. Department of Health, London.
  • Department of Health (2014) Carers Strategy: Second National Action Plan 2014–2016. Department of Health, London.
  • Emerson E., Hatton C., Robertson J., Baines S., Christie A. & Glover G. (2012) People with Learning Disabilities in England 2011. Learning Disabilities Observatory: Public Health England, Available at http://www.improvinghealthandlives.org.uk/publications/1063/People_with_Learning_Disabilities_in_England_2011.
  • Evans G., Kantrowitz E. & Eshelman P. (2002) Housing quality and psychological well‐being among the elderly population. Journal of Gerontology B: Psychological Sciences 57, 381–383. [PubMed]
  • Ferrari S.L.P. & Cribari‐Neto F. (2004) Beta regression for modeling rates and proportions. Journal of Applied Statistics 31, 799–815.
  • Forder J., Malley J., Rand S., Vadean F., Jones K. & Netten A. (2014) Interpreting Outcome Data for Use in the Adult Social Care Outcomes Framework (ASCOF). Discussion Paper 4633. Personal Social Services Research Unit, University of Kent, Canterbury.
  • Fu V. (1998) Estimating generalized ordered logit models. Stata Technical Bulletin 44, 27–30.
  • Gray M., Lepps H. & Blake M. (2014) Exploring Satisfaction with Social Care Services amongst Pakistani, Bangladeshi and White British Populations: Findings from Cognitive Interviews. NatCen Social Research, London.
  • Hatton C. (1998) Whose quality of life is it anyway? Some problems with the emerging quality of life consensus. Mental Retardation 36(2), 104–115. [PubMed]
  • Heath C., Malley J., Razik K. et al (2015) How Can MAX Help Local Authorities to Use Social Care Data to Inform Local Policy? Maximising the Value of Survey Data in Adult Social Care [MAX] Project. Discussion Paper 2896. Personal Social Services Research Unit, The University of Kent, Canterbury.
  • Heinrich C.J. (2003) Measuring public sector performance and effectiveness In: Peters G.B., editor; & Pierre J., editor. (Eds) The SAGE Handbook of Public Administration, 2nd edn, pp. 24–38. Sage Publications, London.
  • HSCIC (2010) Adult Social Care Survey Guidance 2010–11. HSCIC, Available at http://www.hscic.gov.uk/socialcare/usersurveyguide1011.
  • HSCIC (2011) Adult Social Care Survey Guidance 2011–12. HSCIC, Available at http://www.hscic.gov.uk/socialcare/usersurveyguide1112.
  • Huber P.J. (1967) The behavior of maximum likelihood estimates under non‐standard conditions. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability 1, 221–233.
  • Iezonni L. (2013) Risk Adjustment for Measuring Healthcare Outcomes. Health Administration Press, Chicago, IL.
  • Janssen C.G., Schuengel C. & Stolk J. (2005) Perspectives on quality of life of people with intellectual disabilities: the interpretation of discrepancies between clients and caregivers. Quality of Life Research 14, 57–69. [PubMed]
  • Katz S., Down T.D., Cash H.R. & Grotz R.C. (1970) Progress in the development of the index of ADL. Gerontologist 10, 20–30. [PubMed]
  • van Leeuwen K.M., Malley J., Bosmansc J.E. et al (2014) What can local authorities do to improve the social care‐related quality of life of older adults living at home? Evidence from the Adult Social Care Survey. Health & Place 29, 104–113. [PubMed]
  • Lehmann B.A., Bos A.E., Rijken M. et al (2012) Ageing with an intellectual disability: the impact of personal resources on well‐being. Journal of Intellectual Disability Research 57(11), 1068–1078. [PubMed]
  • van Loon J.H., Bonham G.S., Peterson D.D., Schalock R.L., Claes C. & Decramer A.E. (2013) The use of evidence‐based outcomes in systems and organizations providing services and supports to persons with intellectual disability. Evaluation and Program Planning 36, 80–87. [PubMed]
  • Malley J. & Fernandez J. (2014) Generating Adjusted Indicators from Social Care Survey Data. PSSRU Discussion Paper 2873. Personal Social Services Research Unit, London School of Economics and Political Sciences, London.
  • Malley J. & Fernández J.‐L. (2010) Measuring quality in social care services: theory and practice. Annals of Public and Cooperative Economics 81, 559–582.
  • Malley J., Caiels J., Fox D. et al (2010) A report on the Developmental Studies for the National Adult Social Care User Experience Survey. PSSRU Discussion Paper 2721. Personal Social Services Research Unit, University of Kent, Canterbury.
  • Malley J., Towers A.‐M., Netten A., Brazier J., Forder J. & Flynn T. (2012) An assessment of the construct validity of the ASCOT measure of social care‐related quality of life with older people. Health and Quality of Life Outcomes 10, 21. [PubMed]
  • National Audit Office (2014) Adult Social Care in England: Overview. National Audit Office, London.
  • Netten A., Beadle‐Brown J., Caiels J. et al (2011) ASCOT Adult Social Care Outcomes Toolkit: Main Guidance v2.1. University of Kent, Canterbury.
  • Netten A.P., Burge P., Malley J. et al (2012) Outcomes of social care for adults: developing a preference‐weighted measure. Health Technology Assessment 16, 1–166. [PubMed]
  • Papke L. & Wooldridge J. (1996) Econometric methods for fractional response variables with an application to 401 (k) plan participation rates. Journal of Applied Econometrics 11, 619–632.
  • Perry J. & Felce D. (2002) Subjective and objective quality of life assessment: responsiveness, response bias, and resident:proxy concordance. Mental Retardation 40, 445–456. [PubMed]
  • Potoglou D., Burge P., Flynn T. et al (2011) Best‐worst scaling vs discrete choice experiments: an empirical comparison using social care. Social Science & Medicine 72, 1717–1727. [PubMed]
  • Pregibon D. (1980) Goodness of link tests for generalized linear models. Applied Statistics 29, 15–24.
  • Pullenayegum E.M., Wong H.S. & Childs A. (2012) Generalized additive models for the analysis of EQ‐5D utility data. Medical Decision Making 33, 244–251. [PubMed]
  • Ramsey J.B. (1969) Tests for specification errors in classical linear least‐squares regression analysis. Journal of the Royal Statistical Society, Series B 31, 350–371.
  • Rand S. (2014) The Construct Validity and Internal Reliability of the ASCOT Easy Read Measure of Social Care‐related Quality of Life. PSSRU Discussion Paper 4858. Personal Social Services Research Unit, University of Kent, Canterbury.
  • Razik K. & Beecham J. (2014) Review of Factors Important in Use of Survey Data for Quality and Performance Improvement. QORU Discussion Paper. Quality and Outcomes of person‐centred care policy Research Unit (QORU), Canterbury.
  • Schalock R.L. (2004) The concept of quality of life: what we know and do not know. Journal of Intellectual Disability Research 48, 203–216. [PubMed]
  • Schalock R. & Verdugo M.A. (2012) A conceptual and measurement framework to guide policy development and systems change. Journal of Policy and Practice in Intellectual Disabilities 9, 63–72.
  • Schalock R.L., Lemanowicz J.A., Conroy J.W. & Feinstein C.S. (1994) A multivariate investigative study of the correlates of quality of life. Journal on Developmental Disabilities 3, 59–73.
  • Schalock R.L., Brown I., Brown R. et al (2002) Conceptualization, measurement, and application of quality of life for persons with intellectual disabilities: report of an international panel of experts. Mental Retardation 40, 457–470. [PubMed]
  • Schalock R.L., Bonham G.S. & Verdugo M.A. (2008) The conceptualization and measurement of quality of life: implications for program planning and evaluation in the field of intellectual disabilities. Evaluation and Program Planning 31, 181–190. [PubMed]
  • Schmidt S., Power M., Green A. et al (2010) Self and proxy rating of quality of life in adults with intellectual disabilities: results from the DISQOL study. Research in Developmental Disabilities 31, 1015–1026. [PubMed]
  • Smithson M. & Verkuilen J. (2006) A better lemon squeezer? Maximum likelihood regression with beta‐distributed dependent variables. Psychological Methods 11, 54–71. [PubMed]
  • Statacorp (2011) Stata Data Analysis Statistical Software: Release 12. StataCorp LP, College Station, TX.
  • Townsend‐White C., Pham A.N. & Vassos M.V. (2012) Review: a systematic review of quality of life measures for people with intellectual disabilities and challenging behaviours. Journal of Intellectual Disability Research 56, 270–284. [PubMed]
  • Verdugo M.A. & Schalock R.L. (2009) Quality of life: from concept to future applications in the field of intellectual disabilities. Journal of Policy and Practice in Intellectual Disabilities 6, 62–64.
  • Verdugo M.A., Navas P., Gomez L.E. & Schalock R.L. (2012) The concept of quality of life and its role in enhancing human rights in the field of intellectual disability. Journal of Intellectual Disability Research 56, 1036–1045. [PubMed]
  • Wahl H.‐W., Schilling O., Oswald F. & Iwarsson S. (2009) The home environment and quality of life‐related outcomes in advanced old age: findings of the ENABLE‐AGE project. European Journal of Ageing 6, 101–111. [PubMed]
  • White H. (1980) A heteroskedasticity‐consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48, 817–830.
  • Williams R. (2006) Generalized ordered logit/partial proportional odds models for ordinal dependent variables. The Stata Journal 6, 58–82.
  • Wolfe R. & Gould W. (1998) An approximate likelihood‐ratio test for ordinal response models. Stata Technical Bulletin 42, 24–27.

Articles from Wiley-Blackwell Online Open are provided here courtesy of Wiley-Blackwell, John Wiley & Sons