The effect of social deprivation on local authority sickness absence rates

Abstract

Background There is an extensive body of research relating to the association between ergonomic and psychosocial factors on sickness absence rates. The impact of deprivation on health indices has also been extensively investigated. However, published research has not investigated the extent of any association between standard measures of deprivation and sickness absence and ill-health retirement rates.

Aim To establish if a relationship exists between standard measures of deprivation, used by the UK central government to determine regional health and social welfare funding, and sickness absence and ill-health early retirement rates in English local government employers.

Methods Local authority sickness absence rates for 2001–02 were regressed against the 2004 Indices of Multiple Deprivation in a multiple regression model that also included size and type of organization as independent variables. A second model using ill-health retirement as the dependent variable was also estimated.

Results In the full regression models, organization size was not significant and reduced models with deprivation and organization type (depending on whether teachers were employed by the organization or not) were estimated. For the sickness absence model, the adjusted R² was 0.20, with 17% of the variation in sickness absence rates being explained by deprivation rank. Ill-health retirement showed a similar relationship with deprivation. In both models, the deprivation coefficients were highly significant: for sickness absence [t = −7.85 (P = 0.00)] and for ill-health retirement [t = −4.79 (P = 0.00)].

Conclusions A significant proportion of variation in sickness absence and ill-health retirement rates in local government in England are associated with local measures of deprivation. Recognition of the impact of deprivation on sickness absence has implications for a number of different areas of work. These include target setting for Local Government Best Value Performance Indicators, history taking in sickness absence consultations and the role of deprivation as a confounding factor in sickness absence intervention studies.

Introduction

There is a substantial research base establishing a correlation between a wide range of adverse health outcomes and socioeconomic deprivation. In recognition of this, scales of deprivation, the Indices of Multiple Deprivation (IMD) have been developed and have underpinned decision making in health care funding in the UK for many years in an attempt to redress these inequalities [1,2]. The need to allow for such deprivation in funding and performance target setting for health and social care providers is a common area of debate, most recently in applying the same ‘hurdle’ for payment to general practitioners, based on the Quality Outcomes Framework, in deprived areas as those in more affluent regions equitable [3]. In clinical medicine, the ASSIGN cardiovascular risk assessment score has recently been amended to improve risk prediction, compared to the Framingham tables, by factoring for deprivation [4]. Research in non-UK populations has found similar links, with half of the differentials in health indices found between Maoris and European New Zealanders attributed to social disadvantage [5].

The impact of inequalities in health within the population of rich countries and the potential benefits of investment in health on economic and social costs such as absence from work attributed to non-communicable ill-health has become an increasingly high-profile issue. The European Union has recently sponsored research on the economic benefits of investment in public health on productivity and sickness absence rates [6]. However, few published studies have investigated any association between absence and deprivation. A Finnish study of sickness absence rates in public sector employees found correlations between increasing medically certified absence, low municipal revenue and high unemployment [7]. A further Finnish study suggested that differences between long-term sickness absence rates in blue-collar workers between three Finnish regions were associated with differing socioeconomic gradients, as measured by subject income and job title [8]. UK studies have found census measures of limiting long-term illness, mortality and permanent sickness benefits correlated with deprivation scores, but any relationship with sickness absence rates in the employed was not investigated [9]. A further UK study correlated the inception of medically certified absence with deprivation indices, but it was not stated whether all subjects were in employment or were unemployed and seeking enhanced health-related state benefits [10].

Research into the associations and causation of the wide variability in sickness absence rates experienced by different employers is hindered by a number of variables. These include differing occupational groups with a range of work-related physical and psychosocial demands, differing terms and conditions of service and non-standardized methods of calculating absence rates [11]. Local government employees in the UK largely undertake or support the delivery of statutory functions with similar employment groupings between geographical areas. Employees largely enjoy defined conditions and terms of service consistent either with those established for all UK local government employees or teachers. UK central government, through the Department of Communities and Local Government, requires local authorities to annually submit Best Value Performance Indicators (BVPIs) for a range of factors relevant to the efficient and effective delivery of services. The ‘quality’ of local authorities and funding are based on these factors which include measures of sickness absence (BVPI12) and ill-health retirement rates (BVPI15). The UK National Audit Office requires these data to be collated using defined standardized definitions and also audit the reliability of the data [12]. These factors help control for a number of confounding factors in comparing differences in sickness absence rates between local government employers.

The adverse health impact of deprivation on health indices has been well established and is used as a factor in determining regional funding decisions for health and social services in the UK. Central government appraisal of the quality of local authority BVPI12 and BVPI15 do not reflect local deprivation indices. We sought to determine whether

• (i) deprivation indices are associated with sickness absence and ill-health retirement rates in UK local authority employees and should be factored into expectations of performance in these parameters and

• (ii) deprivation indices are a potentially useful measure to use in sickness absence intervention studies to control for non-work-related confounding factors.

Methods

Measures of deprivation were derived from the IMD data from 2004. Each English local authority area (n = 353) is geographically divided into ‘Super Output Areas’ (SOAs, n = 32 482) of approximately equal population. The SOAs do not overlap each other or the geographical borders of local authority areas. For each SOA, a number of routinely available indicators of deprivation are weighted under one of seven ‘domains’ (Table 1).

IMD data were derived from the Office of the Deputy Prime Minister website for the period 2001–02 [13]. Each IMD for a geographical area is a composite of seven domains. Each domain is derived from routinely available indicators of deprivation (e.g. income support claimant rates for ‘employment’).

Low scores equate to high deprivation. Ostensibly, health-related factors constitute only 13.5% of the overall IMD, although a high degree of correlation exists between all domains (typically k = 0.9; C. Connolly, personal communication).

BVPI12 is defined as the total full-time equivalent (FTE) days/shifts lost to sickness absence during the year divided by the average number of FTE staff employed during the year in each local government employer for 2001–02 (average days lost per employee). These data were selected as it best reflected the year for the data upon which the 2004 IMD statistics were based. Only local authorities with no geographical overlap with other authorities were included, leading to the exclusion of county councils (n = 12). Data were derived from all district, metropolitan and unitary local authorities (n = 353). BVPI12 data identified by the district auditor as unreliable were omitted from the analysis, leaving a total of 254 available absence rates.

BVPI15 is the ill-health retirement rate from the Local Government Pension Scheme and, if teachers were employed, the Teacher's Pension Scheme for each local government employer for 2001–02 (average number of retirements due to ill-health per 1000 employees). The data set for BVPI15 was complete.

Local authorities (LA) undertake similar statutory functions; however, some are responsible for education (‘unitary’ and ‘metropolitan’ authorities). This results in the employment of teaching and school support staff, who will normally constitute >50% of all authority employees where employed. Consequently, data for local authorities were analysed in two groups dependent on employment of school staff.

The SOA population figures within each LA area were summed and used as a surrogate for LA employee head count.

The rank of IMD rather than the IMD score was used as the independent variable for deprivation (DEP) for two reasons. First, the IMD score is a highly manipulated statistical construct which has undergone logarithmic conversion. Second, because of its manipulation, the results of a regression analysis using the IMD score are difficult to interpret. The interpretation of rank is more straightforward and is more appropriate for the potential use of the results in performance assessment.

Given the number of data points available, parametric analyses were used.

Multiple regression was used to test the hypothesis that sickness absence and ill-health retirement rates are dependent on three independent variables:

• (i) IMD rank (DEP),

• (ii) Type of organization (TYPE) and

• (iii) Local authority population (as surrogate for number of local authority employees) (SIZE).

The regression model was used to calculate R² (the coefficient of determination) to calculate the extent to which the independent variables were associated with the dependent variables (sickness absence and ill-health retirement rates). Adjusted R² was also calculated as it declines in value if the contribution to the explained deviation by additional variables is less than the impact on the increase in the number of degrees of freedom, whereas R² alone is not.

Analysis was conducted using SPSS for Windows, version 10.0.

Results

Details of the variables used in the analysis are given in Table 2.

Size of the organization, as indicated by the size of the population served by a local authority, was not significant in any of the models. The results reported therefore exclude the independent variable SIZE.

The model for sickness absence (BVPI12) using the predictors TYPE and DEPrivation Rank produced an adjusted R² of 0.2 (R = 0.46, R² = 0.21, standard error of the estimate = 2.26) indicating that 20% of the variation in illness loss per employee was associated by the two independent variables. The bulk of this association (17%) was due to deprivation rank.

Local authorities in more deprived areas (lower IMD rank) had greater rates of sickness absence per employee. The UK central government has used quintiles to differentiate local authorities in terms of health improvement targets and funding programmes. The ‘Spearhead’ group of local authorities are the 20% with the highest levels of deprivation and ill-health. In local delivery plans, different performance indicators have been set for Spearhead local authorities compared with others. Dividing the set of local authorities contained in the data set used for this study (the 353 district, metropolitan and unitary local authorities) gave ∼70 authorities in each quintile. Multiplying the beta coefficient of the DEP variable (0.0129) derived in this study by 70 gave a value of 0.9 sickness days per employee (Table 3). This means that for each deprivation quintile, from the most deprived, there was an additional sickness day lost per employee per year.

The TYPE variable also had a significant negative coefficient. This means that there was less sickness absence in unitary local authorities employing teachers. Other things being equal, local authorities with teachers on their payroll experienced one less sickness day per employee per year compared with local authorities that did not employ teachers (Table 3).

The authorities with missing or unreliable absence data (n = 99) were evenly distributed between the deprivation quintiles with ∼20 authorities per quintile.

Ill-health retirement showed a similar relationship with IMD and organization type (TYPE) as for days lost to sickness. Both independent variables were highly significant. However, together the independent variables only explained 7% of the variation in ill-health retirement rates across local authorities. This is likely to be due to the very small numbers and resulting greater random variation between districts. Nevertheless, the results confirm the strong relationship between measures of work-related sickness and ill-health and characteristics of local authorities in respect of level of deprivation and type or organization (Table 4).

Discussion

This study found a correlation of sickness absence in local government employers with deprivation. This finding is unexpected in that local government employees are, by definition, employed and in relatively secure and stable work. The intuitive expectation that this employment group would be buffered from the effects of deprivation was not supported.

The Whitehall studies found an adverse health gradient with diminishing employee grade [14]. This gradient has been proposed as a contributing cause of workplace-related health inequality [8]. The present study could indicate that, where considered at all, the limited socioeconomic factors measured in earlier sickness absence studies may have reflected the breadth of deprivation experienced by subjects in the communities in which they live, rather than a narrow work-related disadvantage. Alternatively, high levels of deprivation may lead to an adverse impact on work-related stress by local authority staff, most of whom are in ‘front-line’ service provision roles such as teaching and social care. Clarification of this issue may help in the design and interpretation of future sickness absence-related studies.

Should the association between deprivation and sickness absence be supported in further studies, further research into the extent to which employee assistance programmes, which provide support to staff with a wide range of non-occupational life problems, may be merited. It is only recently that prospective studies have demonstrated the effectiveness of social interventions on health outcomes even in well-researched areas of deprivation such as the impact of substandard housing on health [15]. The lack of evidence from intervention studies on addressing health inequalities was emphasized in the UK government's Wanless report [16]. The role, if any, for occupational health services in addressing these health inequalities will require well-designed and funded prospective intervention studies. The scale of the personal, social and economic impact of sickness absence rates associated with deprivation found in the current study should emphasize to public bodies funding academic occupational health the potential future value of such research.

Potential weaknesses of this study include the assumption that most local authority employees live within the borders of their employers' geographical boundaries. This may be a reasonable assumption in the larger district authorities, but in urban areas such as the boroughs of London this may not be the case. No routinely available data could be identified to clarify this issue.

The UK central government uses a range of BVPIs to assess the performance of local authorities. In addition, the UK central government funds the research leading to the generation of the IMDs, which it, in part, then uses to inform funding decisions regarding regional health and social welfare budgets. The correlation between IMDs and local authority sickness absence would suggest that recognition of the impact of deprivation on BVPI12 in local authorities in deprived areas may be justified to ensure equity of target setting for BVPI12 improvement.

Sickness absence is recognized as a behaviour with multiple psychosocial and medical determinants. This study further supports the need for this phenomena to be understood within the individual employees' social and personal context. In the social sciences, where analysis of the effects of deprivation and ill-health is well developed, interventions such as coping and control belief are being promoted [17]. This study emphasizes the need for occupational health professional training in history taking in absence consultations, and subsequent advice, to formally identify and address these factors.

This study suggests that deprivation indices are a potentially large confounding factor in research relating to sickness absence interventions. In such future research, it may be appropriate to use the deprivation ranks from the SOAs in which study subjects reside, for instance by the use of home postcodes, to control for this element of confounding.

While a fifth of the variation in sickness absence rates was associated with deprivation, the cause of the remaining variability remains unclear. Other factors such as variability in the occupational make up of different authorities may be a factor. Although the employment of teaching staff was controlled for, the direct employment of social care staff and construction workers varies between authorities and was not independently addressed. The impact of differing sickness absence policies could also not be controlled for.

Conflicts of interest

None declared.

References