Integrating assessments of older people: examining evidence and impact from a randomised controlled trial

SIR—The needs of older people are often multiple, complex and are at times experienced in combination, which require closely targeted services [1] and necessitate the collaboration of several professional groups. However, care has often been disjointed and compartmentalised [2]. One response to this has been a call for integration, between health and social care. Integrated care can operate at the organisational level, in terms of structures [3, 4] or the professional level, in terms of shared working arrangements, multidisciplinary teams and case management [5, 6]; an issue being that integrated structures, although conducive to inter-professional working, do not necessarily guarantee it [7]. Integration is assumed to produce benefits in terms of better co-ordination of services, better outcomes and greater efficiency [8]. Integration remains an aim pursued by successive governments [9]. Policies in several countries call for integration at the assessment stage; a key phase in terms of ensuring care closely matches older people's needs [10, 11]. In England, examples are the single assessment process (SAP), offering a common structure through shared tools and processes [12], and the common assessment framework (CAF) for adults, advocating technical solutions for information sharing [13]. However, while evidence identifies potential benefits of integrating assessment information from different professionals [5, 14] and of more comprehensive geriatric assessment [15], there remains a need to examine the cost-effectiveness of such approaches to guide future policy. We reanalysed data from a randomised controlled trial of integrated assessment between specialist clinicians and social services care managers in England [14] to offer evidence of the costs and benefits of integration. We evaluated the experiment's impact on changes in functioning, admissions to care homes and use and costs of health and social services for frail older people at risk of entering care homes. Consecutive referrals from social services care management teams through 1998–2000 who were being assessed for substantial levels of care, were randomised into an experimental group, receiving both care managers’ and additional clinicians’ assessments (with a standard reporting process back to care managers) or a control group, receiving the usual care managers’ assessments only. The aim was to enable potentially treatable health conditions to be identified, which might obviate the need for care home admission given the proper intervention [16, 17]. A primary objective was therefore avoidance or delay in care home admissions. Changes in functioning were examined by standardised measures of physical functioning [18], behaviour [19], cognitive function [20], depression [21], social networks [22] and self-rated health [23] between baseline (T1) and 6-month follow-up (T2). Costs of NHS and social services’ inputs were calculated by applying national unit costs [24, 25] to the number of units (days/visits) each older person received. Costs of the medical assessment itself and costs to social services of the extra care manager time involved were also taken into account. The informal costs, in terms of personal consumption and housing costs, incurred by carers and the older people themselves were also calculated. The impact of the integrated assessment was measured by: For functioning; analysis of covariance comparing mean change scores (T2 − T1) for each group. For time to first admission to care homes; comparison of Kaplan–Meier survival curves for each group. For predictors of care home entry; a Cox's proportional hazard model. For cost differences; Mann–Whitney test and median values describing the ‘typical’ cost for an individual. Comparisons were of costs per week alive, standardised for differences in survival rates between groups. For variations in costs; regression to unravel the determinants of the care received for each group. A P < 0.10 level was chosen for significance. Supplementary data are available in Age and Ageing online, see Appendix 1 for variables used in these analyses. There were few significant differences in outcomes with most changes being due to random fluctuations. Participants were a very frail group and all indices, with the exception of depression, deteriorated. However, those receiving the integrated assessment experienced a less marked deterioration in their physical functioning (Barthel, −2 versus −6%; F = 3.97, P = 0.04). Controlling for other predictors, the experimental group was predicted to have an advantage of 3.14 points (P = 0.08) over the control group (see Supplementary Data in the Supplementary data available in Age and Ageing online). The experimental group gained an initial advantage, in terms of remaining at home. It maintained this advantage for up to 50 days. Thereafter, only a relatively moderate group difference remained (see Supplementary Data in the Supplementary data available in Age and Ageing online). Table 1 shows the results of the Cox regression model. Accounting for other predictors, there was a tendency for the intervention to reduce the risk of care home admission and in particular by removing the age effect. However, the intervention increased the risk of care home entry for the frailest individuals (seen in interaction effect with respondent's death). Cox regression model of permanent care home entry risk Notes: n = 237. For visual assessment of model fit see Supplementary data available in Age and Ageing online, Figure AS1 (plot of observed and predicted survival curves). The violation of the proportional hazards assumption (significant rho) for cognitive impairment is due to the positive effect only becoming apparent after approx. 30 days and then growing stronger with time. The inclusion of this coefficient does not alter the remaining model, however, and it was therefore retained. Referral time was tested as an additional covariant in order to control for potentially informative left-censoring of risk exposure (experimental group experienced a bigger delay), but was not retained in the model. Cox regression model of permanent care home entry risk Notes: n = 237. For visual assessment of model fit see Supplementary data available in Age and Ageing online, Figure AS1 (plot of observed and predicted survival curves). The violation of the proportional hazards assumption (significant rho) for cognitive impairment is due to the positive effect only becoming apparent after approx. 30 days and then growing stronger with time. The inclusion of this coefficient does not alter the remaining model, however, and it was therefore retained. Referral time was tested as an additional covariant in order to control for potentially informative left-censoring of risk exposure (experimental group experienced a bigger delay), but was not retained in the model. Comparison of costs between the groups revealed few significant differences. Of note, total NHS costs were lower in the experimental group, even after taking into account the cost of the specialist assessment. On aggregate, the integrated assessment accounted for 11% less NHS expenditure than the control; an estimated saving of £1,131. This arose, in part, from reduced hospital costs (£7,666 versus £8,951; 14%) including accident and emergency visits (£55 versus £106; 48%). Total social services costs (apart from additional care manager time), all informal cost components and total costs were not significantly different between the groups (see Supplementary Data in the Supplementary data available in Age and Ageing online). Controlling for other determinants, the cost models (Table 2) contain only significant or borderline insignificant predictors. We found, from interaction terms, that for the frailest individuals the integrated assessment led to increased costs for the NHS and social services but a reduction in costs to older people and their carers. For those with severe cognitive impairment, the integrated assessment had the opposite effect, reducing NHS and social services costs while raising informal costs. The remaining group effects were as follows: (i) the cost of the integrated assessment itself is reflected in the NHS cost model; (ii) social services costs did not change significantly beyond the identified shifts and (iii) after controlling for baseline differences the experimental group appears to have experienced slightly higher informal costs. Regression cost models Notes: n = 237, 243, 237. 95% confidence intervals can be calculated as ±1.96 × SE. Normal error distributions and homoscedasticity were confirmed by formal tests (Shapiro-Wilk and Breusch-Pagan/Cook-Weisberg). The error distributions of models 1 and 2 are approximately normal. Regression cost models Notes: n = 237, 243, 237. 95% confidence intervals can be calculated as ±1.96 × SE. Normal error distributions and homoscedasticity were confirmed by formal tests (Shapiro-Wilk and Breusch-Pagan/Cook-Weisberg). The error distributions of models 1 and 2 are approximately normal. The evidence shows that integrating health and social care assessments offers benefits, principally a reduction in physical deterioration, for the very frail (those at risk of entering care homes). The integrated assessment acted to delay care home admissions while ensuring a greater likelihood that the frailest individuals entered care, a positive targeting effect. The pathway of impact was through clinicians contributing to more accurate identification of health needs, particularly for those most frail and cognitively impaired. The consequences were more appropriate care home admissions, slightly increasing NHS expenditure (through taking remedial action) for those most frail, without overall, shifting the cost burden to social services. These effects translated into higher informal costs for older people and their carers. The experiment offers particular rather than global benefits as it exemplified integration at only one level; that of more precise linkage [26] between specialist clinical and community care assessments. Placement decisions, still the responsibility of care managers, were not based solely on clinical recommendations and this may explain why admissions were not reduced further. The delay in admissions, however, shows the potential for such assessments to contribute to more accurate needs identification and more appropriate care planning. To realise wider benefits may necessitate integration at a higher level of coordination [26], where explicit structures require organisations to work together through the use of single assessment documentation or shared decision making [27]. Evidence from Australia, where placement decisions were made conditional on multidisciplinary assessments by Aged Care Assessment Teams [28], with associated reductions in nursing home admissions, suggests possible benefits. This was essentially the intention of the SAP, which aimed to share information in a planned manner [12]. However, specialist clinicians were largely uninvolved in SAP implementation [29] with it being viewed as predominantly a social services concern; perhaps reflecting a view of national policy as ephemeral to clinical practice. Realising the benefits of integration, however, requires the participation of all involved, perhaps through incentives for effective information sharing [8]. The recent CAF represents an attempt to do so through the medium of common assessment records [13]. A fully integrated [26] system would pull together disparate funding, assessment and care arrangements into one organisation responsible for the medical and social care of defined populations. While it is assumed that this necessarily enhances patient care, in countries where such integration exists, such as Northern Ireland, shared assessment tools and procedures for professionals have still been developed [30]. One reason may be that integration, particularly of assessments, is a professional concern with different occupational cultures and training still influencing practice [14]. Integrated structures, of themselves, may therefore not guarantee better outcomes, such as those detailed here. In light of recent policy advocating integrated assessments, the concern remains as to how this may be appropriately achieved. Future policy should focus on the precise manner by which such integration is implemented. Integrating assessments between health and social care has been a policy aim in England and a number of other countries. Recent policies, including the SAP and CAF in England, have assumed that such integration can confer significant benefits, although the research evidence for this is scarce. From a randomised controlled trial of integrating specialist health clinicians’ assessments with those of social services care managers there were benefits including reduced physical deterioration of older people and delay in care home admission. Evidence here shows that integrating assessments may lead to more appropriate care home admissions through providing assistance to those most frail without shifting the overall cost burden to social services. The precise manner of integrating assessments should be the focus of future enquiry, as integrated structures do not necessarily lead to enhanced multidisciplinary working. None declared. The study on which this analysis was based was funded by the Department of Health in England under the Community Health Services Research Initiative. The sponsors played no role in the design, execution, analysis and interpretation of data, or writing of the study.