Secondary analysis and literature review of community rehabilitation and intermediate care: an information resource

Article history

The research reported in this issue of the journal was funded by the HS&DR programme or one of its proceeding programmes as project number 10/1011/51. The contractual start date was in January 2011. The final report began editorial review in November 2012 and was accepted for publication in June 2013. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HS&DR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Stuart Parker declares National Institute for Health Research (NIHR) grants for Health Services and Delivery Research 12/5003/02 and was Deputy Director for the NIHR South Yorkshire Collaboration for Leadership in Applied Health Research and Care.

Copyright statement

© Queen’s Printer and Controller of HMSO 2015. This work was produced by Ariss et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Introduction to the chapter

This chapter begins with an exploration of the literature regarding patients who are most or least likely to benefit from intermediate care (IC) services. The chapter considers descriptions of some commonly found types of IC services and some specific clinical needs of service users. The review moves on to look at reported patient characteristics, including age and sex, cognitive impairment, living arrangements and functional status on admission and examines how these impact on outcome. This chapter then focuses on the secondary analysis of data from the two previous studies. 1,2

Background

If IC is to contribute to the current NHS agenda by using resources in the most effective way while maintaining service quality, then it is essential that it is offered to those patients who are most likely to benefit from these services.

In order to target services appropriately, evidence is clearly needed to inform the development of appropriate criteria in order to identify both those patients who are most likely to benefit from IC and to identify which patients are least likely to benefit in terms of both physical functioning and quality of life (QoL). The reasons for this are outlined below.

Literature review

Secondary analysis of data

Results

Remaining or returning home

In all 4556 out of 7620 patients (60%) remained or returned home following an episode of IC. The distribution of this outcome is presented in Figure 1.

FIGURE 1.

Frequency of patients remaining or returning home.

The likelihood of returning home shows a large amount of variation between teams. Team COOP PB had the greatest percentage of patients who remained or returned home, with 88% of their patients remaining or returning home. Team COOP SG had the smallest percentage of patients who remained or returned home, with 20% of their patients remaining or returning home.

Level of care at admission

The chance of remaining or returning home is greatest for patients with a LoC, at admission, of 4 (needs regular rehabilitation programme) and 3 (needs slow-stream rehabilitation). Patients with these LoCs, at admission, have a 66.0% and 63.7% chance of remaining or returning home, respectively.

The chance of remaining or returning home is lowest for patients with a LoC, at admission, of 0 (does not need any intervention) for which there was only a 20.4% chance of remaining or returning home. However, many of the patients with a LoC, at admission, of 0 were marked as being an inappropriate referral. Those with a LoC, at admission, of 7 (needs medical care and rehabilitation) had a 34.8% chance of remaining or returning home.

Key point 11: the likelihood of remaining or returning home is related to the LoC at admission to IC. It is greatest for levels 3 and 4 and is lowest (20.4%) for patients with a LoC, at admission, of 0 (does not need any intervention).

Age and sex

For every 10-year increase in age, there was a 6% decrease in the odds of returning home. The chance of returning home was greater for females than males. Females had a 51.8% chance of returning home and males a 47.4% chance.

Key point 12: for every 10-year increase in age there was a 6% decrease in the odds of returning home. The chance of remaining or returning home was greater for females than males.

Relationship between patient characteristics and therapy outcome measures

For all domains of the TOM scale, around half of all patients with outcome data did not improve or worsen. More patients showed improvement in TOM impairment and activity than in participation or well-being. In the imputed data set, the proportion of patients who failed to improve was slightly higher, as patients with missing outcomes had worse predicted outcomes than those for whom data were available. Table 5 summarises the improvement in these scores.

TABLE 5 - Summary of improvement in TOM scores

SD, standard deviation.

Figures relate to the mean of the five imputed data sets. Individuals may have both positive and negative imputed changes in the different data sets and, therefore, the total number of patients with an improvement is not defined.

Outcome	Complete case			After data imputation (n = 7291)a
	Number of patients with data	Mean (SD) improvement	% (n) with any improvement	Mean (SD) improvement	% with any improvement
TOM impairment	5337	0.44 (0.71)	50% (2659)	0.28 (0.83)	43.1%
TOM activity	5339	0.46 (0.71)	51% (2735)	0.31 (0.84)	44.4%
TOM participation	5340	0.38 (0.71)	43% (2279)	0.22 (0.85)	37.0%
TOM well-being	5330	0.30 (0.68)	37% (1975)	0.13 (0.87)	32.1%

The magnitude of these changes is displayed graphically in Figure 2.

FIGURE 2.

Change in TOMs by domain.

Key point 13: nearly half of all patients with outcome data did not improve on any of the domains of TOM.

Baseline scores

Patients with higher TOM scores on admission were less likely to improve. The relative increase in odds of improvement per 1-point increase in score on admission ranged from 0.53 (TOM well-being) to 0.75 (TOM impairment) using imputed data (see Appendix 2). The extent that this is as a result of a ceiling effect (i.e. because the potential to improve is less for those who already have high scores) is not known.

Age

The magnitude of a TOM score increase was related to patient age, with smaller changes being observed in older patients. Although statistically significant, the decreases were modest. On average, the change in TOM scores decreased by 0.02–0.03 units per additional 10 years of age (see Appendix 2).

Sex

Fewer men than women improved for all domains of the TOM. However, the difference in mean scores between men and women was small, i.e. ≤ 0.1 units for all domains. The odds of improvement were significant for all domains except participation (see Appendix 2).

Level of care on admission

Level of care at admission was associated with change in all TOM domains. Fewer than 20% of patients with a LoC, on admission, of 0 improved their TOM scores for all domains (using imputed data). The patients most likely to improve were those with admission LoCs of 3, 4 or 5 [> 40% for impairment, activity and participation, > 30% for well-being (see Appendix 2)]. However, the magnitude of change was modest, with the average (adjusted) mean changes in the highest (levels 4 and 5) and lowest (levels 0 and 6) groups being within 0.06 units of each other. The probability of any improvement as calculated from the imputed data was between 64% (level 6) and 77% (level 4) (see Appendix 2).

Compared with level 0, levels 1–8 all had a higher probability of having any improvement for impairment, activity and participation. The mean difference between the best and worst LoC group was around 0.3 to 0.4 TOM units. The magnitude of the changes in impairment, activity and participation and well-being, adjusted for other covariates, are illustrated graphically in Figures 3–6.

FIGURE 3.

Change in impairment by LoC at admission (model adjusted).

FIGURE 4.

Change in activity by LoC at admission (model adjusted).

FIGURE 5.

Change in participation by LoC at admission (model adjusted).

FIGURE 6.

Change in well-being by LoC at admission (model adjusted).

Key point 14: patients most likely to improve were those with admission LoCs levels of 3, 4 or 5 (> 40% for impairment, activity and participation; > 30% for well-being). The probability of any improvement as calculated from the imputed data was between 64% (level 6) and 77% (level 4).

Route of referral

Improvement was greatest among patients referred in from acute settings [accident and emergency (A&E), ambulance, rapid response, acute hospital, day clinics and fall clinics]. Of particular note is that baseline TOM scores in these subgroups were similar to those referred in from other settings and so these differences are not obviously an artefact of regression to the mean. Patients referred via a community nurse, social care or social work had the smallest improvements in TOMs. Again, despite statistical significance, the effects were modest with the highest and lowest group means being around 0.2 TOM units different.

Key point 15: improvement was greatest among patients referred in from acute settings (A&E, ambulance, rapid response, acute hospital, day clinics and fall clinics) despite similar baseline scores to those admitted from other settings.

Place of care

The patient location was also associated with outcome for TOM and EQ-5D scores, with better outcomes observed among those who initially live more independently (i.e. in their own home, an IC facility or resource centre), who have bigger changes than those admitted from residential or hospital settings.

As was found with the TOM data, the relationship between the EQ-5D score and different locations of care was affected by the fact that 3738 patients (46%) had missing data.

Smaller proportions of the patients in residential/nursing home or acute hospital settings improved than those receiving care in more independent care settings, such as at home or in an IC facility or resource centre. However, the percentage showing an improvement became more consistent across the groups when the imputed data set was analysed.

The proportion of patients showing any improvement was significantly lower among people receiving care in a residential/nursing home, A&E or acute hospital compared with those in other settings for all domains of the TOM (see Appendix 2). The magnitude of the changes was modest (< 0.4 units in all domains) and smallest for TOM well-being scores. Mean changes ranged from < 0.1 (residential/nursing) to 0.2 units (IC facility) (see Appendix 2).

The chance of remaining or returning home was greatest for patients receiving care at home who had previously lived at home unaided prior to their health complication; these patients had a 71.9% chance of returning home. This variable also suggests that the chance of returning home is smallest for patients receiving IC at residential or nursing homes; these patients had only a 10.1% chance of returning home.

Key point 16: smaller proportions of the patients in residential/nursing home or acute hospital settings improved compared with those receiving care in more independent care settings, such as at home or in an IC facility or resource centre.

Key point 17: the chance of returning home was greatest for patients receiving care at home who had previously lived at home unaided (71.9%).

Key point 18: the chance of returning home is smallest for patients receiving IC at residential or nursing homes (10.1%).

Usual living arrangements and changes during intermediate care

After data imputation, the patients’ usual living arrangements were found to be highly associated with level of improvement across all four TOM domains. Patients who normally live in their own home or in sheltered accommodation had greater improvements than those living with relatives, who in turn had greater improvements than those in residential, nursing or other settings.

The mean improvement in TOM impairment was > 0 for all subgroups, but ranged from 0.32 (patients living alone in their own home) to 0.01 (other settings). The association was stronger for TOM activity, ranging from 0.36 (own home) to –0.06 (nursing home). Changes in TOM participation and well-being were less pronounced, but patients living in their own home fared, on average, 0.4 units better than patients who are institutionalised prior to IC. The patterns observed in the complete case data were less clear cut, but broadly echoed these.

The greatest level of improvement was observed among patients who were transferred to IC facilities, followed by those who did not leave their own home during IC. Patients who were transferred to hospital or nursing home were the least likely to improve during IC. Figure 7 illustrates these changes, which account for health status.

FIGURE 7.

Change in activity by location of care (for those not institutionalised pre entry; model adjusted).

Key point 19: patients’ usual living arrangements were found to be highly associated with level of improvement across all four TOM domains. The greatest level of improvement was observed among patients who were transferred to IC facilities, followed by those who did not leave their own home during intervention.

Relationship between patient characteristics and European Quality of Life-5 Dimensions scores

The EQ-5D scores of 38% of patients did not improve during their episode of treatment. This is less than the proportion who did not improve on any of TOM domains. This may be a reflection of the fact that EQ-5D is effectively a continuous measure and is a patient-completed measure (reflecting the patient’s view their own health status), whereas the TOM scale has only 11 possible levels and reflects the health-care professional’s view. Improvements in EQ-5D scores were found to have statistically significant relationships with other variables. For instance, the greatest improvements were seen for LoCs 4, 5 and 7 (approximately 0.04 increase), patients receiving care in their own homes (0.06 increase), referrals from A&E/ambulance service or acute wards (0.05 increase) and lower baseline EQ-5D score (0.05-unit increase per 0.1-unit decrease in baseline). Tables 6 and 7 summarise the improvements in these scores.

TABLE 6 - Changes in EQ-5D scores

SD, standard deviation.

Figures relate to the mean of the five imputed data sets. Individuals may have both positive and negative imputed changes in the different data sets and, therefore, the total number of patients with an improvement is not defined.

Outcome	Complete case			After data imputation (n = 7291)a
	Number of patients with data	Mean (SD) improvement	% (n) with any improvement	Mean (SD) improvement	% with any improvement
EQ-5D	4332	0.18 (0.28)	62% (2684)	0.15 (0.25)	66.4%

TABLE 7 - Summary of findings

Characteristic	Relationship with TOM		Relationship with EQ-5D
	Statistically significant?	Nature of relationship	Statistically significant?	Nature of relationship
Age	Yes	Minimal: TOM reduced by 0.02–0.03 units per 10-year increase in age	No
Sex	Yes	Minimal: on average, females had approximately 0.1 unit greater improvement than males	No
LoC at admission	Yes	Levels 3, 4 and 5 improved greatest. Impairment and activity: levels 3–5 typically 0.2–0.3 units higher than lowest (0 and 6), with other levels between. Participation and well-being: levels 3–5 around 0.15–0.25 higher than other levels	Yes	Levels 7, 4 and 5 had greatest improvement, around 0.04 units more than the remainder
Normal living arrangements	Yes	Greatest improvement for patients receiving care in their own home, with an average increase of 0.15–0.3 units more than living in a residential home or relatives home (which showed the least improvement)	Yes	Greatest improvement for patients receiving care in their own home, with an average increase of 0.06 units more than other locations
Effect of leaving own home during IC	Yes	Greatest improvement for patients in IC facilities, showing an improvement of around 0.5 greater than acute hospitals, which show the smallest improvement. Patients remaining at home or transferred to resource centre have next greatest improvement overall	Yes	Greatest improvement among patients who transfer to IC facility, resource centres or remain in own home. Improvement approximately 0.05–0.15 greater than other locations. Those transferred to acute settings had least change
Who made the referral	Yes	Greatest improvement for wards in acute hospitals of around 0.1–0.25 units more than the remainder	Yes	Greatest improvement for A&E/ambulance service and wards in acute hospitals, this showed an increase of around 0.05 units more than the remainder
Baseline score	Yes	On average, improvement in TOM scores reduced by approximately 0.2 units per unit increase in baseline TOM score	Yes	Improvement in EQ-5D reduced by 0.05 units per 0.1-unit increase in baseline EQ-5D

The magnitude of these changes is displayed graphically in Figure 8.

FIGURE 8.

Change in EQ-5D score.

Discussion

This chapter has explored which patients are most likely to benefit from IC and those which are least likely to benefit and, therefore, may be best placed to receive care elsewhere. A summary of the main findings is detailed in Table 7.

In order to use resources appropriately, it is important to select patients carefully to receive interventions that are most likely to be of benefit. The data presented here provide information to improve the selection of patients most likely to be assisted by IC. The analysis adds to the evidence in the literature review that the most very elderly patients may be less likely to benefit from IC in terms of rehabilitation outcomes. However, the differential benefits associated with reduced age were small and, therefore, the clinical significance of this is tenuous. The data analysis supports the findings of the literature review that neither age nor sex is likely to be useful criteria on which to select patients for IC.

However, both the literature review and analysis of these data indicate that severity of impairment and health-related QoL (HRQoL) appear to affect outcomes. Patients with higher scores on the TOM scale and EQ-5D questionnaire, on admission, are less likely to benefit, although this may be because patients with higher score are healthier to begin with and have less room for improvement. This assertion is supported by the finding in the data analysis that those with a LoC of 0 (‘client does not need any intervention’) showed less change than other levels. However, it is important to note that many of these clients were discharged on the same day that they had been admitted to the service as they were deemed as not requiring intervention. Patients considered ‘inappropriate referrals’ or who did not need the intervention that they had been referred for are a poorly described group and require further investigation.

A similar lack of change in TOM and EQ-5D scores was found for those who need relatively little care (convalescence/respite or prevention/maintenance) rather than those who need slow-stream, regular or intensive rehabilitation, indicating that use of the LoCs may be a useful tool for resource allocation.

The literature provides information about the types of patients included in studies evaluating IC service models using RCTs. The systematic reviews and meta-analyses essentially capture those models for which there has been extensive interest, leading to funding and the conducting of several RCTs.

A key message from the literature review is that RCTs include older people with specific medical [stroke, COPD, congestive heart failure (CHF)] and surgical conditions (fractured neck of femur, CABG). Trials have not included the general population of service users. There is evidence that the rates of improvement in patients admitted to IC are often modest, with two studies showing that only around one-third of patients improve at all. Given the frailty of most of the patients admitted to IC, it could be that no decline in health status is also a positive outcome and perhaps the parameters for success of IC services need to be reconsidered.

The literature review showed evidence that age, cognitive impairment, living alone at admission and functional status at admission may all have an influence on the outcomes of patients using IC; however, the strength and direction of these findings were not consistent enough to draw conclusions for this study.

Analyses of data from our two studies focus on two main sets of characteristics: the assessed need of the subjects in receipt of IC services and the location of care. Age and sex are also included in the analysis but the effects, although statistically significant, were numerically rather small.

For each of the five measures, the extent of change was negatively associated with the baseline score: higher scores at baseline tended to have smaller changes. Increasing age was associated with lower gains in all TOM scores, but the association was small in absolute terms (< 0.05 units of TOMs per 10-year increase). When compared with females, males had marginally higher improvements in impairment and activity, but lower improvements in participation and well-being. EQ-5D score was not associated with age or sex.

The overall outcomes of this study showed that, on average, 60% of patients remain or return home following an episode of IC. Rates of improvement varied according to the outcome measure. On average, 43% of patients improved on the measure of TOM impairment, 44% on TOM activity, 37% as measured by TOM participation and 32% on TOM well-being. Two-thirds of patients (66%) improved on the EQ-5D score measurement after data imputation.

Factors that were statistically associated with a change in TOM scores were patient age (improvement declines with age), sex (females more likely to improve), LoC at admission, living in own home, receiving care in own home or IC facilities, referrals made by acute hospitals and having a lower score at admission.

With respect to all the outcomes (return home, TOM score and EQ-5D score), the analyses generally support the same conclusion: those patients who are more likely to have the most positive outcomes (return to home, large improvement in TOM parameters and EQ-5D score) are more likely to have been assessed as in need of rehabilitation (according to LoC need) by the admitting team.

Levels of care 3, 4 and 5 provide the springboard for most improvement and these represent an assessment that the patient is in need of slow-stream, regular or high-intensity rehabilitation.

Those patients who are likely to do best receive the IC service in their own home. This implies that those who do worst receive IC elsewhere, although it should be noted from the literature review that residential and nursing home settings are largely unevaluated.

However, there are clear indications from the secondary data analysis that place of care is important and that those who improve the least are those who receive care in residential/nursing homes and acute hospital settings. This finding is not surprising given that the level of impairment tended to be more severe for these patients, indicating a higher burden of chronic disease.

Conclusion

The literature review and data analysis can be seen to agree. Those patients for whom IC may be of benefit include older people with medical and selected long-term conditions for whom a needs assessment suggests that there is potential for rehabilitation, which should be provided, when possible, in the patient’s own home.

Gaps in the evidence base concern the location of care (specifically the use of residential and nursing home settings) and the specific clinical concerns of meeting the needs of patients with dementia, or palliative and end-of-life care needs.

Introduction

This chapter deals with the question of whether or not the use of IC services changes the use of hospital inpatient facilities by reducing hospital admissions and what factors (if any) are associated with increased rates of hospital admission (and/or readmission) during or after the use of IC services.

Background

In the first instance, there are some conceptual issues which need to be clarified. Admission to acute hospital care is an event that IC services are frequently designed to avoid completely. For example, HITH services can be constructed to bypass the acute hospital, providing substitute acute and/or subacute care in the community and resulting in the total avoidance of hospital admission. A person in receipt of such a service, whose condition deteriorated to require admission to hospital, would be experiencing a first admission to inpatient hospital care for the episode of care in question. Conversely, someone who deteriorated in an IC scheme designed to support early discharge from hospital will be readmitted to hospital (i.e. will experience a second period of inpatient care during the same episode of illness). For example, nurse-led, post-acute care for frail older people was set up to deliver quality transition care (rather than change episode duration). RCTs showed that the nurse-led inpatient units tended to increase the length of inpatient hospital stay. 46–48 However, the subsequent use of health-care resources was different between intervention and control.

In one unit, this was shown to be largely at the expense of community hospital transfers, i.e. total resource use was similar between the intervention group, who had longer inpatient stays, and the control group, who had shorter inpatient stays but made more use of community resources after discharge. These examples are given to illustrate the potential interdependence of the relationship between duration of inpatient stay and subsequent use of services after discharge. 49 Hospital length of stay has become a simplistic metric used as an indicator of efficiency with associated political overtones. Thus, there are incentives for patients to be moved out of hospital, or to avoid hospital, when in fact they may receive more efficient rehabilitation within a hospital setting.

For the purpose of this chapter, the notion of avoiding first admission and avoiding readmission will be conflated, so that we can consider the issue of reducing hospital admissions without worrying about somewhat arbitrary definitions concerning the type of hospital admission.

Admission/readmission as a policy focus

During the first decade of this century, policy-led initiatives focused on reducing recurrent hospital use by patients at risk of multiple admissions. Examples of this include the idea of case management in the community and using specialised nurses or health-care teams to identify at-risk individuals and provide them with additional support in the management of their long-term conditions. The aim of such interventions would be to reduce the use of secondary care and improve health and well-being through encouraging enhanced self-management and providing support in the community for acute or subacute episodes of deterioration.

Although there was some early enthusiasm for these forms of care in England, the potential for them to reduce secondary care use was somewhat overstated in the first instance. Evaluation of the potential for reductions in hospital admissions,50 and the impact of one form of enhanced nurse practitioner working with older people in England,51 was followed by a re-evaluation of the potential for case management to solve the problem of rising numbers of hospital admissions. Case management approaches have been shown to be appreciated by users and carers52 and, in some configurations, to impact on functional health and well-being of recipients. 53 Another approach to targeting interventions to help prevent hospital admissions has involved the use of predictive modelling to identify people at high risk of readmission54 and target additional support in the community. Both of these approaches (anticipatory case management in primary care and predictive modelling) have typically considered hospital admission risk over a 12-month period.

More recently, there have been a series of shifts in policy, directly and indirectly, addressing the issue of unplanned acute hospital readmissions in England. In particular this has included a focus on community care of chronic conditions,8 payment by results, practice-based commissioning55 and, more recently, the setting up of Clinical Commissioning Groups and alterations to the tariff concerning readmissions. 56 The current government has confirmed a policy of altering the remuneration available to NHS trusts (the tariff) for patients readmitted to acute care within a month of discharge. This has been applied to elective care, but in the future may include a proportion of acute care admissions. The key point about this change in policy is that the trust responsible for the initial episode of inpatient care becomes responsible for the cost of subsequent inpatient care if the patient is readmitted within 30 days of the initial discharge. This places the responsibility for managing discharge across the interface between secondary and primary care (i.e between hospital and community care) firmly in the domain of the acute trust and is probably intended to stimulate re-engineering of services at the crucial interface between primary and secondary care so that they become more closely integrated than they may have been in the past.

Literature review

In conducting this review of the relevant literature, we have focused on the evidence from RCTs, evidenced mostly from systematic literature reviews. These sources sometimes report on admission rates and sometimes report on readmission rates, depending on the objectives of the services being reviewed. We have treated these terms as interchangeable for our purpose of describing the literature evidence on interventions affecting hospital admission rates.

Secondary analysis of data

Discussion

The literature does not suggest that any particular IC model (admission avoidance, discharge support, community rehabilitation, community hospital or other clinical facility for rehabilitation or recuperation, or both, including nurse-led inpatient units) will achieve a specific and consistent impact on admission or readmission to inpatient hospital care.

However, recent changes in policy may have affected the viability of nurse-led discharge units. These hold the potential to transform the experience of discharge, but at the expense of somewhat increased hospital inpatient resource use. These schemes were shown to reduce readmission rates significantly in some analyses. Essentially, this was achieved by taking time to prepare patients for discharge, effectively managing the timing of discharge to suit their needs by increasing duration of inpatient stay. In the new policy environment, in which early readmission is becoming the financial responsibility of the discharging hospital, it is possible that the economics of such NLUs may be more favourable than when originally evaluated.

There is some evidence from analysis of the characteristics of interventions associated with reduced (re)admission rates that a key issue is the effectiveness of working across the interface between primary and secondary care,59 or providing community-based care after hospital discharge. 61 These observations imply that it is important, when transferring patients with complex care needs across the interface between inpatient hospital and community-based care, to transfer the capacity to meet their complex needs in the community.

Our data suggests significant variation between clinical teams in the rates of admission to hospital (range 0–21%). This is consistent with a view that factors other than the specific care model (such as the degree of integration across the acute and secondary care interface) may be important in determining (re)admission risk. The main patient characteristic associated with increased transfer to hospital was a complex presenting need as defined on LoCs, which included the requirement for both rehabilitation and elements of medical care.

The main patient characteristics associated with low rates of transfer to inpatient hospital care were low LoCs need (no need and preventive need only) and being resident in nursing home care. In the latter case, although care needs may be high, or complex, the surrounding environment and (possibly) care plans and policies, may be conducive to receiving such care in situ.

Conclusion

Taken together, a consistent message that emerges from the literature and the secondary analysis of the data from these two studies is that a central concept in minimising hospital admissions in IC in the community is the capacity of the service to meet the client’s, often complex, care needs. That is to say that hospital admissions will be minimised when the services are sufficiently integrated across the interface between hospital and the community that the patient’s care needs can be matched to the available service in the setting to which the client is assigned.

Introduction

Making the transition into long-term care is difficult for many older people and is frequently associated with loss of identity, fractured social networks and discontinuity. The literature surrounding admission to long-term/institutional care for older people suggests that entry into a nursing or residential home often results from a crisis. In such situations, rather than managing the process and ‘making a decision’ about care, older people and their families are more likely to be find themselves ‘realising the inevitable’ and entering the care network with little information or choice. 67 Alongside the personal costs associated with admission to long-term care, considerable public costs have also been identified and a long-standing policy commitment has been made to prevention, with IC at its centre. 4,5,68 The downward trend in the numbers of people in care year on year might suggest some success in relation to this and other objectives. 69

Background

In recent years, a raft of community-based complex interventions has evolved to help prevent admission, with evidence to suggest that these, in part, may have contributed to delays in admission to care for some older people. Nevertheless, around 220,000 people are supported to live in residential or nursing home care in England on a long-term basis, with over three-quarters (77%) of those being aged ≥ 65 years, with over two-fifths (43%) being aged ≥ 85 years. 69 These figures do not include all of those individuals in private residential and nursing home facilities or those with continuing care requirements.

This chapter sets out to identify those factors associated with admission to institutional care for users of IC. This will be achieved through addressing three objectives. First, relevant literature concerning the role of IC in reducing the risk of admission to long-term care will be highlighted. Second, the evidence concerning those factors that have shown to predict admission to institutional care in older people will be presented. Finally, the findings that have emerged from the secondary analysis of the COOP¹ and EEICC² data sets and that address factors that predict admission to institutional care are presented.

Literature review

Secondary analysis of data

Results

A total of 85 out of 6550 patients (1.3%) were discharged into permanent residential or nursing home care. Ten teams reported no institutionalisations, with only two teams reporting over 4% [COOP E, 3 out of 46 patients (6.5%); EEICC PB, 6 out of 109 patients (5.5%)]. Analyses based on complete case data were similar to those based on imputed data and, for simplicity, we present only the former.

The age of the patient, their usual living arrangements and leaving home for IC were all highly statistically significant (p < 0.001) predictors of institutionalisation on univariate analysis. Age was associated with a 93% increase in the odds of institutionalisation per 10-year increase (95% CI 48% to 151%; p < 0.001). All four TOM and EQ-5D scores at admission were also associated with a reduced probability of institutionalisation.

Out of all 1006 patients who received IC in an institutional setting, 56 (5.6%) were discharged to institutions. Out of all 5544 patients who received IC in their usual place of residence (excluding those whose usual residence was an institution), 29 (0.5%) were discharged to institutions.

As with the hospitalisation outcome, the TOM and EQ-5D scores at admission were entered into five separate models because of their collinearity. In each case, the baseline measurement added to the model and all four TOM scores had an associated OR of around 0.7, in other words, the relative odds of admission were reduced by 30% for each unit increase in TOM scores at admission. The gradient in the reduction of probability with increase in the TOM participation score at admission is illustrated in Figure 9. A higher EQ-5D score at admission also reduced the likelihood of being institutionalised following IC (OR 0.92 per 0.1-unit increase).

FIGURE 9.

Institutionalisation for TOM participation score.

Key point 32: on average, for every 1-point increase in TOM participation score, at baseline, there is a 32.8% decrease in the likelihood of admission to long-term care (OR 0.67; p < 0.0001).

Key point 33: increasing age and poor TOM or EQ-5D scores at admission are associated with admission to care for IC service users. Receiving IC services at home is strongly associated with not being admitted to long-term care.

Discussion

This chapter addressed the question concerning those factors that might predict discharge from IC to long-term/institutional care. As such, it set out to review evidence concerning the effectiveness of IC in preventing admission, as well as literature that has highlighted factors, which might predict admission to care for older people. It should be noted that with the exception of Langhorne et al. 71 there exists little analysis on the interaction of the use of IC services, individual factors and admission to institutional care as an outcome.

Factors identified as being statistically significant include residential care home as a place of IC provision, increasing age and lower TOM participation score at admission.

Notable non-significant factors include LoCs of 6 and 2 and being female. It should be noted that discharge to a care setting was likely in only 2% of cases, adding credence to the potential for IC to provide a preventative service,61,70 particularly in the light of more recent UK research, which would suggest that living alone is a significant predictor of admission to long-term care. 75

Findings relating to increasing age concur with previous studies, which have explored those factors that predict admission to care. 73,74 There is a consistent theme within the existing evidence in relation to poor ADLs and admission to care. 74 TOM participation score on admission to IC is identified as a significant finding here, and although conceptually distinct to ADLs, the relationship between function and participation is clear.

Conclusion

Again, the findings here resonate with existing evidence. Langhorne et al.,71 in a review of the effectiveness of early supported discharge, noted that patients with poor ADLs scores at admission were more likely to enter long-term care. It is important, however, to note that by focusing on individual characteristics, the opportunity to explore the relationship between contextual factors and likely outcome is missed. The variations in team composition, setting, skill mix, size and integration and, in particular, the relationship with discharge destination are variables that could explain differences in the data. These areas of investigation demand further research.

Introduction

This question is significant because of substantial political and economic pressure for more elderly patients to be managed in their own homes, whereas previously they had been cared for in hospitals and other institutions. It is important that such changes (at the very least) do not put patients at an increased risk of adverse outcomes, such as death.

Background

There is evidence that IC and community rehabilitation is beneficial to many older people. However, it is important to be able to identify those who are at risk of deteriorating health who may benefit from hospitalisation or increased length of stay prior to discharge.

Intermediate care and rehabilitation should be available to those individuals who are likely to benefit from such services and palliative care services should be available to those who are likely to die. Although we recognise that prediction of death in the elderly is not a precise science, it is important to address this question and examine differences between different teams so we can determine whether or not patients are being correctly referred to appropriate services and whether or not changes have occurred over time.

Knowledge about survival and prognosis is important to clinicians, patients, family and other caregivers and for the planning of long-term facilities and home care. Indeed, in the acute care sector, NHS trusts regularly see their mortality statistics, appropriately corrected for case mix, as one of the routine metrics about them on public information systems. 76 Such data are not available for those receiving community-based NHS services.

Literature review

A copy of the search strategy used to support the literature review for this chapter is located in Appendix 1.

There have been several studies using different algorithms to identify the community-dwelling vulnerable older people who are more likely to die at an earlier point in time but none specifically studying those receiving IC. Therefore, we describe here some key studies that focus on those at increased risk of death in old age.

Saliba et al. 77 developed a screening tool that was tested on more than 6000 community-dwelling older people, who were defined as persons aged ≥ 65 years.

A complex data analysis investigated a method for identifying older people at risk of losing physical function and death. This was used to identify a vulnerable group that comprised 32% of the ≥ 65-year-olds studied. This group had a fourfold greater risk of death when compared with the rest of the population.

Unsurprisingly, increased mortality is associated with increased age and clinical instability. A study by Guerini et al. 78 provided a definition of clinical instability which used routine clinical measures (such as blood pressure and heart rate) to predict increased likelihood of death on admission to a rehabilitation and aged care unit.

Chronic kidney disease, which is more common in old age, was investigated by Roderick et al. 79 who showed that declining kidney function is significantly associated with an increase in all-cause and cardiovascular mortality in those > 75 years of age.

After a study of more than 9000 patients in the USA living in the community aged between 65 and 102 years of age, Kazanjian et al. 80 concluded that pulse pressure appears to be the best measure for predicting mortality in older people.

The many physiological and social factors that have been investigated to determine their contribution to mortality often give rise to symptoms which increase the likelihood of hospitalisation and may still be present on early discharge and transfer to IC.

The study by Kazanjian et al. 80 found that the odds of death increased with institutionalisation and with increasing cognitive and physical impairment. Although vision and hearing problems and the presence of heart disease, stroke and diabetes were all strongly related to 5-year mortality in univariate and unadjusted analyses, their contributions were minimal in the multivariate analyses. Increased body mass index was associated with lower mortality in both univariate and multivariate analyses.

Ostbye et al. ,81 in a similar population-based study in Canada, confirmed the importance of sex, age, functional status, cognition and health status in predicting 5-year mortality. Accounting for cognitive status, physical status and specific disease variables led to the difference in mortality between older people in the community and in institutions being reduced.

A population-based prospective study of more than 1000 older individuals (64–85 years) in Finland by Hirvensalo et al. 82 ranked participants into four groups: (1) intact mobility and physically active (mobile–active), (2) intact mobility and sedentary (mobile–sedentary), (3) impaired mobility and physically active (impaired–active) and (4) impaired mobility and sedentary (impaired–sedentary). The analysis adjusted for age, marital status, education, chronic conditions, smoking and physical exercise earlier in life. The study found a twofold increase in the risk of death in impaired–active and a three times greater risk in impaired–sedentary groups than in mobile–active groups. However, the risk of death did not differ between mobile–active and mobile–sedentary groups.

The authors concluded that mobility impairments predicted mortality and dependence. However, among people with impaired mobility, physical activity was associated with lower risks, whereas the risk did not differ according to activity level among those with intact mobility. Despite their overall greater risk, mobile-impaired people may be able to prevent further disability and mortality by physical exercise. Interestingly, increasing independent activity is one of the key aims of IC.

Secondary analysis of data

The statistical methodology was undertaken as described in Appendix 2. Reported probabilities of mortality were calculated using a model, which included age, sex, LoC at admission, route of referral and the location where the patient was receiving care.

Discussion

There has been an increase in deaths of patients referred to intermediate/community care over time. This is associated with a greater number of patients receiving these services who have more complex and severe health and social care needs. These findings are in line with the literature related to deaths within the community of older people. However, this needs to be placed in the context of the evidence that even the best geriatric medical acute unit expects a mortality rate of around 15–20% of patients at 3 months after admission. 83 The determinants of death in inpatient care are the same as we found in these community studies (i.e. severity and complexity of health condition and previous functional health status). Although IC services may have seen a rise in mortality, they are receiving patients at a lower risk of death compared with the population they are likely to have come from (i.e. mortality rates in the region of 1–3% could be interpreted as indicating appropriate selection of patients for IC services).

Another important issue is related to whether or not IC and community rehabilitation are appropriate or inappropriate services in caring for those coming to the end of their life.

Conclusion

Many surveys have concluded that, generally, people would prefer to die in their own home,84 but unfortunately people with palliative and supportive care needs are generally explicitly excluded from the IC trials, so we have little or no evidence about providing palliative care in IC services to work with. However, there is a trend in developing services, which integrate rehabilitation and IC services with those of palliative care for older people with a broad range of disabilities and health-care challenges. Moving more of this care into the community may well be appropriate but it is important to consider that a recent meta-analysis (of more than 10,000 patients) indicated the value of comprehensive geriatric assessment to reducing deterioration of health, and yet costs increased. 85 Consideration needs to be given to the skill mix to support these patients appropriately.

Introduction

Over recent years, it has been recognised that about 85% of health-care expenditure is spent on treating 15% of the population. Much of this expenditure is focused on patients with chronic conditions, an increasing number of which are older patients. These patients are much more frequent users of all health services, but in particular, they are much more frequently admitted to hospital for episodes of acute care.

Background

As a result of the NHS Plan4 and NSF for older people,5 IC and rehabilitations services have been developed to rehabilitate patients more effectively after episodes of acute care, assist earlier discharge, promote the greater independence of older people with acute conditions and provide support in the community to prevent hospital readmissions. IC services are generally, but not always, community-based, interdisciplinary teams, which are staffed by a range of health and social care professionals. Some are located in NHS organisations and others in social care. The development of community rehabilitation and IC/services (CRAIC/S) service has followed no particular formula and, in reality, services across England and Wales differ enormously in their size, configuration and working practices. This has led to some researchers questioning whether or not IC really exists as a stable concept. 1 Furthermore, there is little real understanding about how most effectively to configure IC services for optimum impact. Knowledge about what team factors are associated with the greatest benefits for patients in terms of improving and maintaining their health status, and their independence, is therefore vital.

Literature review

Secondary analysis of data

In keeping with the question aims, the available data set and the findings of the literature review, the relationships between a number of structural team-level variables and patient outcomes, hypothesised as potentially significant, were tested.

Results

Skill mix

The full results of the analysis can be seen in Figure 10. Overall, the analysis of the combined data set found few significant relationships. The results do suggest that TOM impairment improves more among teams that have a higher skill mix (i.e. larger number of different disciplines), with TOM impairment change scores increasing by 0.029 units with each additional discipline represented in the team. This relationship is represented in the scatterplot in which the size of the circle represents the magnitude of the teams’ weight in the analysis.

FIGURE 10.

Scatterplot showing the relationship between skill mix and change in TOM impairment. Circles represent the individual teams, with the size of the circle representing the magnitude of the teams’ weight in the analysis.

Ratio of support staff to professionals

It was found that having more domiciliary or clinical support staff in teams was associated with a small improvement in TOM impairment scores. For every unit increase in clinical support staff, TOM impairment scores increased by approximately 0.01 units; this increase was consistent whether or not the complete case data set or a data set with imputed data was used. There was also a similar relationship between TOM impairment and number of domiciliary support workers but, unlike the other findings, this was heavily influenced by the data from one team. The largest standardised mean TOM impairment change (0.6 units greater than predicted by its case mix) was observed in the team with the highest number of domiciliary staff but, although interesting, removing this data point from the analysis resulted in a substantially reduced (and non-significant) relationship. The relationships are shown in the scatterplots Figures 11 and 12.

FIGURE 11.

Scatterplot showing the relationship between numbers of clinical support workers and change in TOM impairment. Circles represent the individual teams, with the size of the circle representing the magnitude of the teams’ weight in the analysis.

FIGURE 12.

Scatterplot showing the relationship between numbers of domiciliary support workers and change in TOM impairment. Circles represent the individual teams, with the size of the circle representing the magnitude of the teams’ weight in the analysis.

Key point 47: there is consistent evidence that more clinical support staff in teams were associated with a small improvement in TOM impairment scores. Similar results were found for domiciliary staff. However, this latter finding was heavily influenced by data from one team.

Discussion

A literature review was undertaken examining papers published between 2008 and 2012. Only five papers were found to be relevant to the question ‘what team-level factors are associated with the greatest benefits for patients in terms of health status?’. The results of the review showed that there was some empirical evidence that interdisciplinary teamworking in IC with older patients was more effective than usual care in reducing length of stay and costs. 86 Using integrated care facilitators to improve co-ordination between interdisciplinary team members proved effective with COPD and CHF patients in reducing emergency readmissions presentations, admissions and length of stay. 88 However, these studies do not directly examine team-level factors associated with better outcomes.

The study by Dixon et al. 87 did find that increased skill mix was significantly associated with a 17% reduction in service costs and a high proportion of support workers were associated with better improvement in patient QoL.

Qualitative studies in the review found indicative evidence that a number of team process variables contribute to better patient care. However, the variables identified were not within the scope of this study.

The results of the statistical analysis of this combined data set found some weak empirical evidence to support the findings of the above empirical study. Increased skill mix and higher proportions of clinical and domiciliary support workers in CRAIC/S teams were significantly associated with improvements in impairment scores using the TOM tool. However, it must be noted that the study by Dixon et al. 87 utilised one of the two original data sets which form the combined data set for this study. 1 In addition, owing to the number of hypothesis tests undertaken, there is a high possibility of at least one of the findings being a spurious association.

Conclusion

The literature review and secondary analysis of this combined data set does provide additional evidence that interdisciplinary teamworking in IC may be associated with better outcomes for patients. However, relatively few significant results were found overall and a relatively wide number of variables were analysed. Therefore, we urge that caution is taken when considering these results.

Introduction

Owing to limited resources, it is important that health care is provided in the most efficient way possible so that the health produced by those services that are provided is maximised. The standard framework by which this is investigated is cost-effectiveness analysis. This examines the costs and effects of different ways of treating patients and summarises this in an incremental cost-effectiveness ratio.

Earlier in the report, we described teams that have been involved in the two studies. Only two were integrated health and social care teams and the others had a broad range of different professional and support staff. The analysis here is restricted to examining the configurations of the teams involved, which may not represent all the different models of IC available.

Different skill mixes across IC teams can potentially have an effect on patient outcomes and costs. Therefore, for example, a greater number of professions and a greater number of professionally qualified staff could improve care by allowing a fuller range of therapies and support to be provided, but it could increase costs because of higher salaries and a greater number of visits.

Background

The simplest approach to cost-effectiveness analysis compares two randomised groups and then calculates the additional costs and additional benefits associated with the most effective treatment for each group. However, this is not possible here as different services have different patient groups. Controlling for these differences, in the presence of patient clustering and cluster-level costs, makes assessment of cost-effectiveness problematic. The analysis plan for this study starts with simple descriptive analysis, then adds in multivariate analyses of costs then ends with an exploratory analysis of cost-effectiveness. Specifically, we sought to produce:

• descriptive analysis of mean resource use, costs and quality-adjusted life-years (QALYs) per patient for each team

• multivariate analyses examining the relationship between skills mix and patient-level costs

• mean costs and QALYs gained will be plotted for each team on the cost-effectiveness plane to highlight which teams are the most cost-effective

• multivariate analysis examining the relationship between skills mix and team cost-effectiveness.

Literature review

The COOP study1 identified several important relationships between team characteristics and costs. Using a sample of 1167 patients, it was shown that the average cost per patient increased as the number of different practitioners increased, but fell with a rise in the proportion of qualified staff. Strong relationships were also identified relating to teamworking as measured by the Workforce Dynamics Questionnaire. Exploratory analyses showed no relationship between cost-effectiveness and skills mix variables across the 20 teams studied. However, another study of 403 patients produced different findings, which indicated the opposite effect with respect to the number of different practitioners and no effect relating to qualified staff. 1 No assessment of cost-effectiveness was undertaken in the latter study.

The EEICC study2 did not include a formal evaluation of cost-effectiveness but did collect some of the same data as COOP. 1 Combining these two data sets is therefore possible and, with the greater number of patient observations and IC teams, it could give us a clearer picture of the relationship between team characteristics, costs and cost-effectiveness.

Secondary analysis of data

Our analysis undertaken within COOP1 and the subsequent study (EEICC)2 had different objectives and collected different sets of data. For example, within COOP1 we collected patient-level data on number and length of all patient contacts broken down by staff professional group. As such, we were able to estimate detailed costs for each patient and explain how these varied by patient and team characteristics. These data were not collected for EEICC,2 which looked more broadly at how overall cost changed with the implementation of an interprofessional management tool. 90 Consequently, the COOP1 analysis cannot be directly replicated using data from both studies. The analysis is necessarily restricted to data that were collected in both studies using the same methods.

Results

The teams show a wide variation in average duration of care (ranging from 1 to 138 days) and average cost per patient (ranging from £318 to £11,511) as shown in Table 15. All teams saw, on average, an increase in the health of their patients as shown by positive QALY gains.

When patient-level costs are examined, clear relationships are seen with respect to HRQoL (as measured by the EQ-5D), impairment and activity (both as measured by the TOM) (Table 16). The nature of the relationship is the same for all three variables, i.e. costs initially increase with improving health (or reduced impairment) and then fall for higher levels of health improvement. Considering the staffing variables, there is only good evidence to suggest that increased numbers of different types of practitioners are associated with higher costs.

An examination of Figure 13 shows a massive variability in costs and outcomes. The three services depicted by the points joined by the kinked line are relatively more efficient than the others are. Points vertically above the three points defining this efficiency frontier are services that have higher costs for the same outcomes. The frontier represents combinations of the efficient services and, therefore, in theory, all services above the frontier are generating higher costs for patient outcomes than could be produced by the efficient services.

FIGURE 13.

Cost-effectiveness of the different teams. Note: data points represent mean cost and mean QALY gain for each of the teams. The slope of an imaginary line drawn from the origin to the data point equates to cost divided by QALY gain and represents the average cost-effectiveness ratio. The three teams linked by the two lines are relatively more cost-effective than the others.

An explanation of this variability in statistical terms was undertaken using the multivariate analysis on average NMB, with the results shown in Table 17. These results show no clear explanatory effect of different staffing patterns (i.e. number of practitioners and proportion of skilled staff). However, there is weak evidence of economies of scale, with larger services (proxied by annualised patient days) generating a higher NMB.

Discussion

The results indicate highly heterogeneous services. Costs per patient range from £318 to £11,511, with patient characteristics being the dominant explanatory effects. Cost-effectiveness is also highly variable with only size of service showing any sign of an explanatory effect. Staffing patterns as described by the proportion of qualified staff and number of different types of staff have little or no influence on costs and relative cost-effectiveness.

These results add to a growing body of literature that show highly variable costs, outcomes and inconsistent relationships relating to staffing patterns. 1,87 The massive heterogeneity of services is likely to have an impact on our ability to identify any relationships that exist. Indeed, the variability may also suggest that comparisons between all the services is not sensible and they appear to differ in terms of their purpose. A service with an average duration of care of 1 day cannot be designed to deliver the same care packages as a service with an average duration of care of 138 days. A further qualitative assessment of the different types of teams may be helpful to identify more useful comparisons; however, with only 33 teams, subdividing the sample will make any statistical exploration of differences problematic.

Conclusion

This study represents the most comprehensive attempt to explain differences in costs and cost-effectiveness across different IC teams. This work, in tandem with other comparable studies, is unable to identify consistent and clear relationships relating to staffing. This suggests that efficiency savings are possible by many services by reducing staff costs to levels seen in comparable teams. However, the identification of comparable teams may be problematic, as there appear to be quite profound differences between them that are not readily explained quantitatively.

Introduction

There are two interrelated reasons why it is important to explore how IC services have changed over time. First, IC services are widely assumed to be flexible regarding the services they provide and how these are delivered as they need to respond to changes in national policy and to fit with local variations in health-care provision.

At a national level, they have been widely implemented as a mechanism to deal with policy changes, such as emergency care reform. 92,93 At a local level, IC entry thresholds are inconsistent and any entry guidance that does exist is locally determined. IC services often receive patients on the basis that there is ‘nowhere else for them to go’, rather than because they demonstrate a real potential for improvement in the IC setting. 1 IC is, therefore, uniquely sensitive to changes in the health-care system as a whole. Qualitative data arising from projects 1 and 2 found that IC teams perceive that they are under growing pressure to help meet hospital targets regarding lengths of stay and waiting times, by accepting patients with increasingly complex needs.

Second, service- and team-level characteristics have been associated with patient outcomes. Indeed, one national evaluation of IC found that service characteristics were a better predictor of service costs and patient outcomes than patient characteristics. 10 Therefore, in order to provide appropriate care it would be expected for services to change in line with the changing demands placed on them.

Background

There has been rapid growth in the use of support workers, rather than qualified practitioners, to deliver much of the care within IC. 94 Project 1 found that the level of patient impairment and patient needs were unrelated to skill mix. 1 Evidence is urgently needed to ensure that hospital avoidance schemes such as IC benefit the patient and that IC facilities have the appropriate skill mix to address the increasingly complex needs of the patients they take on. However, there has been little research to date to inform decisions regarding what constitutes appropriate staffing in IC in terms of staff type, skill mix and intensity. In the context of ongoing changes to case mix and workforce, commissioners and managers have a distinct lack of evidence to support decision-making regarding staffing of, or referrals to, IC.

The investigation of staff type and skill mix changes in IC services over time may demonstrate whether there are trends across the data set regarding a direction of development of these services, or whether local considerations continue to determine changes in the sector leading to increasing heterogeneity.

Literature review

Literature searches for this chapter (see Appendix 1) were undertaken on MEDLINE (via EBSCOhost) and Cumulative Index to Nursing and Allied Health Literature (CINAHL; via EBSCOhost) between December 2011 and June 2012. Articles were included as an update for an earlier set of extensive literature searches (COOP1 and EEICC2). The search used terms relating to IC and then facets relating to service models, for example, skill mix or management structures. These searches identified 2176 papers of potential interest.

These papers were explored to establish their relevance to the study. The main inclusion criteria were that they provided either empirical evidence or commentary on any of the following themes:

• drivers for change in English community care

• organisational and interorganisational models of community care in the UK

• workforce planning and team configurations for community care in the UK

• the effect of team configurations on patient outcomes for community care or rehabilitation in the UK.

Papers were excluded in cases in which findings were not transferable to the English health-care setting or if they were concerned with aspects of service configuration which were not recorded in the data for this study (for instance, methods of case management, non-mandatory training or details of leadership and supervision).

After titles were examined, 294 papers remained. The abstracts were read and further papers were eliminated, leaving 18 papers of possible interest. After examination of full copies, four further papers were excluded. These were concerned with models and methods for workforce planning, indicators of interorganisational integration, a qualitative study of stepped care and prevalence of mental health problems in IC patients. The remaining 14 papers were then reviewed in depth and a main aim of the review was to explore the range of evidence. Therefore, although the quality of studies was taken into consideration, papers were not excluded on the grounds of quality.

No papers directly addressed the issues raised by the study question (i.e. changes in IC service configuration over time). A main theme in the literature, with limited importance for this study, was models of services at a macro level including aims of the service, place of treatment or description of type of service (e.g. supported discharge or admission prevention, geriatric day hospital, community hospitals, NLUs, HAH, home-based rehabilitation). These papers rarely described the make up of such services, although it is recognised that staffing requirements for the specific aims of services (e.g. facilitated discharge and admission avoidance) should be distinct. 95 Papers describing drivers for change provide useful background information, as do snapshot descriptions of IC service configurations (including the relationship of configurations with patient outcomes).

Secondary analysis of data

The purpose of these analyses was to investigate whether or not IC teams had undergone changes in their configuration or their patient’s characteristics over the time period between the first and second studies. A simplistic approach would be to compare the data collected for all the teams involved in study 1 with data for all the teams involved in study 2. The major drawback of this approach is that many of the teams were involved in only one of the two studies. As there is considerable heterogeneity among IC teams, any differences observed could be due to the underlying differences in the teams included in the two studies, as opposed to a temporal effect. Although we report these findings below, this caveat will need to be borne in mind when interpreting the figures. An alternative approach, and one which allows a more direct comparison, is to compare data taken from teams included in both studies. This is not without difficulty, as doing so still relies on a concept of unchanging, well-defined IC teams. Service evaluations carried out as part of study 2, demonstrated that members of IC teams often had disparate and poorly defined definitions of their work organisation. They were subject to almost constant change, with very little managerial focus on maintaining a cohesive approach to joint working. On occasions, it was found that while managerial definitions and aims of the service had changed substantially (including changes to admission criteria), these changes had not been communicated to staff members. Therefore, the assumption that comparing the same team at different time points is actually comparing like with like cannot be taken for granted and must be explored in some depth at various levels of the organisation.

It is notable that shortly following study 2, two of these teams were disbanded (with staff being allocated to GP practices over a large geographical area) and one team stopped working with clients in their homes. Many of the other teams had uncertain futures and were preparing for unpredictable changes. All teams were wholly or partly funded by PCTs, which are being rapidly dismantled as a result of the recent white paper. 56 The heterogeneity and instability of IC services presents the researcher in this field with a challenging environment, particularly regarding conducting longitudinal studies and in drawing conclusions and making generalisable recommendations.

However, although some teams changed beyond recognition owing to major service reorganisation, seven teams remained fairly stable between project 1 (COOP)1 and project 2 (EEICC). 2 The majority of the analyses focused on these seven teams.

Results

Team characteristics

Table 18 and Figure 14 demonstrate some important differences between the ways that services are organised. Although most aim to prevent acute/long-term care admissions and facilitate discharge from acute services, there are some differences to priorities. Services range from medically focused nurse-led, community hospital-based teams funded by PCTs (e.g. EEICC team PB) to teams based in local authority establishments, jointly funded by social services and the NHS, with a focus on providing social care in clients’ own homes and small residential units (e.g. teams F and Q).

TABLE 18 - Description of teams taking part in both studies

LA, local authority; OT, occupational therapy.

Team	Service aim	Service description	Primary and secondary setting	Initial age of service	Background of team leader	Funding organisation	Organisation setting
1	Supported discharge	Multidisciplinary team	Resource centre, client’s home	5 years	Social worker	PCT/support staff	Support staff
2	Admission prevention	Rapid response and community rehabilitation service	Client’s home, nursing home	4 years	Nurse	PCT/LA	PCT
3	Admission prevention and facilitate early discharge	Step-up and step-down multidisciplinary team	Client’s home, hospital (inpatient/outpatient)	3 years	Dietitian	PCT	PCT
4	Admission prevention and facilitate early discharge	Nurse-led step-down/step-up facility	Community hospital	8 years	Nurse	PCT	PCT
5	Community stroke rehabilitation	Community multidisciplinary team	Client’s home, hospital outpatient	No data	Nurse	PCT/support staff	PCT
6	Admission prevention and facilitate early discharge	Primarily step down, social services assessment and co-ordination	Client’s home, reablement unit	8 years	Social worker	PCT/support staff	Support staff
7	Admission prevention and facilitate early discharge	IC	Client’s home, hospital (inpatient)	8 years	OT	PCT	PCT

FIGURE 14.

Changes in team configurations. (a) Number of clinical staff; (b) number of clinical support staff; (c) number of management staff; (d) number of social care staff; (e) number of non-clinical support staff; (f) number of domiciliary support staff; (g) total number of staff; and (h) total number of staff types. a, The average of the six teams; and b, the average of all teams combined.

Patient characteristics

Demographics

The following patient characteristics changed slightly. Overall, patients were 1 year younger in the EEICC study2 than in the COOP study. 1 Although the majority of patients were females, the percentage of males rose by 4% from COOP1 to EEICC. 2 Comparisons for the seven teams involved in both studies are shown in Figures 15 and 16. These demonstrate a stable average age of patients and slightly less of an increase in the percentage of male patients for these teams.

FIGURE 15.

Changes in ages of patients. T1 represents the COOP study and T2 the EEICC study. a, The average of the seven teams; and b, the average of all teams combined.

FIGURE 16.

Changes in sex of patients. T1 represents the COOP study and T2 the EEICC study. a, The average of the seven teams; and b, the average of all teams combined.

In the following summaries, data are presented for the seven teams that appeared in both COOP1 and EEICC,2 the average of these (‘Total^a’) and the average of all teams combined (‘Total^b’).

Inappropriate referrals

The extent to which teams had inappropriate referrals was measured through two routes. First, there was a response on the patient record form to indicate patients who were considered to have been inappropriately referred. Second, it can be assumed that patients who are assessed as having a LoC need of 0 (not requiring service) at admission have not been appropriately referred.

There was considerable overlap between patients considered as inappropriately referred when using both methods of identification as demonstrated in Table 24. Therefore, we have shown findings using both methods individually and also included findings that take the overlap into account.

TABLE 24 - Percentage of inappropriate referrals by LoC at admission

LoC at admission	Number (%) of inappropriate referrals
0	252 (45)
1	69 (4)
2	8 (8)
3	26 (2)
4	35 (2)
5	6 (1)
6	11 (4)
7	21 (11)
8	3 (4)

For teams that took part in both study 1 and study 2, LoC 0 at admission and inappropriate referrals both increased (from 5.0% to 8.5% and from 5.0% to 13.0% respectively). The increase in the mean for all teams was similar (LoC 0 = 6.7% to 9.5%, inappropriate referrals = 4.1% to 10.4%). The trend was upwards; some teams demonstrated a slight decrease or remained almost the same, while three teams showed large increases. There was no uniform increase across all teams (Figures 17 and 18).

FIGURE 17.

Changes in LoC = 0, patients not needing an intervention on admission. a, The average of the seven teams; and b, the average of all teams combined.

FIGURE 18.

Changes in patients indicated as inappropriately referred. a, The average of the seven teams; and b, the average of all teams combined.

Teams 1 and 5 had a small reduction in LoC = 0. Team 7 had the highest LoC = 0 in study 2 and this rose considerably from 0% in study 1 to 20% in study 2.

Teams 1, 4 and 6 had very low numbers of inappropriate referrals in study 2. This represented a small reduction for team 1 and a nearly steady situation for teams 4 and 6. Teams 2, 5 and 7 saw large increases. Team 3 remained fairly steady, above the mean for study 1 and below the mean for study 2.

Although the variation between teams demonstrates that the specific reasons for the overall increase in inappropriate admissions are likely to be different for each team that contributes to the increase, the upward trend should be noted. Qualitative evidence from study 2 indicates that this may be due to a lack of alternative appropriate services for people with low-level needs or little rehabilitation potential. The lack of clear service identities and admission criteria may also be a contributing factor.

Key point 70: for the teams taking part in both studies, the percentage of patients deemed not to require the service (LoC = 0) or inappropriately referred rose (from 5.0% to 8.5% and from 5.0% to 13.0%). There were similar increases for all study data (from 6.7% to 9.5% and from 4.1% to 10.4%, respectively).

Figure 19 makes allowance for the overlap between patients identified as inappropriately referred and with a LoC of 0. Across the teams involved in both studies, the percentage of patients either identified as inappropriate referrals or with a LoC = 0 increased from 7.7% to 14.8%, and from 8.6% to 14.4% across all 22 teams. Although the trend is upwards, these averages mask the high variability between teams. Three of the teams (teams 2, 5 and 7) demonstrated a large increase, but a fourth (team 3) showed only very minor increase and the remaining three teams reported small decreases, albeit on very few inappropriate referrals, between COOP1 and EEICC. 2

FIGURE 19.

Patients either identified as inappropriate referrals or LoC = 0. a, The average of the seven teams; and b, the average of all teams combined.

Location of care

The location where teams deliver care is an important predictor of patient outcomes, and has an important bearing on the culture of the team and the type of care they are able to provide. For the sake of comparison, the percentage of patients receiving care at home was compared with patients receiving care in hospital. Patients receiving care in other locations account for the differences between these.

Overall, the percentage of patients receiving care at home increased; however, when considering the teams that participated within both study 1 (COOP1) and study 2 (EEICC2), this decreased. Teams 1,3, 4 and 6 all had increases in the percentage of patients receiving care at home, while teams 2, 5 and 7 all had decreases. Regarding the number of patients receiving care in hospital, teams 1, 2, 5 and 6 were quite stable, whereas teams 3 and 7 saw large increases and team 4 had a large decrease. However, differences between teams varied by up to 100%.

Key point 71: perhaps the most striking feature of changes in the location of care is the variation between teams (up to 100%), rather than the variation over time.

Overall, comparing both studies, the number of patients receiving care at home increased (Figures 20 and 21). However, looking just at teams who participated within both study 1 and study 2, this decreased. A similar story is told by the changes in patients receiving care in hospital. Although there was a reduction for the study as a whole, for teams taking part in both studies there was a slight increase. These findings suggest that the teams themselves had different referral patterns and provided different types of services in a varying diversity of locations. The types of locations for which services are provided are mostly stable and, therefore, do not seem to be influenced by national issues such as changes in patients’ needs or policy directives. However, there seem to be strong local determinates incorporating structural changes, which redefine how the service is incorporated into the local health economy.

FIGURE 20.

Changes in patients receiving care at home. a, The average of the seven teams; and b, the average of all teams combined.

FIGURE 21.

Changes in patients receiving care at hospital. a, The average of the seven teams; and b, the average of all teams combined.

Teams 1, 3, 4 and 6 had a slight increase in the percentage of patients receiving care at home, but teams 2, 5 and 7 saw a slight decrease. Regarding the number of patients being treated in hospital, this increased for teams 3 and 7, reduced for team 4, and the other teams were stable.

Team 4 had a reduction in the amount of care provided in hospital from 100% in study 1 to a little over 80% in study 2. However, only a small amount of this is accounted for in providing care in patients’ homes. This ward-based service seems to have diversified to provide more care in other locations and, again, these findings do not demonstrate a uniform change. Instead, they suggest relative stability for the majority of teams and changes in both directions for a small number of teams, which (when data are aggregated) tend to suggest small trends.

Most teams provide care in a mix of locations, which makes the distinction between bed- and home-based care difficult to sustain at a service level. Teams also seem to be diversifying regarding their locations of care. Future studies would therefore benefit from the ability to distinguish between bed- and home-based care at a patient level rather than a service or team level. This would be particularly important for the purposes of benchmarking services.

Discussion

An interesting finding regarding patient characteristics is related to the age and sex of the patients across the two projects. It is widely accepted that the population of the UK is aging and that life expectancy is higher for females. Therefore, we would expect to see a slightly older mean patient age and slightly more females than males in the second project. However, across the whole data set, the reverse was observed. It is difficult to make assumptions about why it might be the case that patients admitted to these IC services do not follow demographic trends.

There were some consistent trends between the data from the seven teams and the whole data set, for example increased management and social care staff, decreased clinical support staff and increased total number of staff types. However, these changes were small and variation from team to team meant that there was no discernible pattern of changes for team characteristics.

Overall, and for the teams that took part in both projects, there were increases in management, social care and domiciliary support staff. Clinical support staff saw drastic reductions for the teams that took part in both studies and for the studies overall.

The total number of staff increased slightly for the studies as a whole, but decreased substantially for the teams that took part in both studies. The total number of different types of staff increased overall and for teams from both studies. These findings would suggest that, on the whole, team size is increasing slightly, the number of management staff is increasing slightly and there are a greater variety of roles and increased skill mix in IC services. The number of social care staff increased from a very low baseline, while clinical support staff numbers were noticeably reduced. However, when viewed on a team-by-team basis, the picture is much more complicated. Although some teams remained quite stable on these measures, others changed drastically and not in a uniform direction. Therefore, slight mean increases shown in the aggregated data sets disguise the radical and unpredictable changes experienced by some individual teams.

Conclusion

The changes highlighted in the teams that took part in these projects continue to reflect the diversity of the delivery of care through IC services. Far from being any uniformity regarding models of service organisation, there is not even any coherent direction of service development. The large variation in this small sample hints at the difficulty of categorising IC services and the complexity of predicting how they might develop. A useful area for further investigation would be to investigate the complex drivers for change: what evidence (if any) informs decisions about the development of IC services, how are these decisions taken, who do they involve, what are the triggers for change and what are the impacts for service users and members of staff?

Introduction

Earlier in this report detail was given on the number of policy changes and directives published over the last 10 years by the Department of Health aimed at promoting and influencing the character of IC. These changes are likely to directly influence the nature and types of patients admitted to IC.

Background

Our qualitative data from the two studies suggests that IC services are receiving referrals of more chronically ill and dependent patients in recent years. This needs further interrogation as it has important resourcing implications. It is known that there are many influences associated with referring patients to particular services. Those who refer to services are often seen as the gatekeepers to the services. Investigating these issues provides insight into issues associated with policy change, equity of access and local provision.

Literature review

The systematic reviews of the literature (derived from the search strategy in Appendix 1 for the COOP1 and EEICC2 studies) did not elicit any studies relating to changes in referral patterns to community-based health-care provision for older people over time. However, studies of referral patterns to hospital-based services and mental health care services at particular points in time identify issues affecting local variation. For example, Bickell et al. 120 identified a sex bias in referral patterns for coronary heart disease treatment, which had increased over time and led to more appropriate referral for females than males.

The impact of change in style of service delivery having an effect on referrals was identified by Beck et al. 121 who established that the shift from an inpatient- to an outpatient-based service resulted in fewer patients being investigated and treated in hospital and more patients treated as outpatients. This had the impact of reducing inpatient related costs, while outpatient-related costs increased. The overall contribution of drug costs to the total cost had increased greatly over time.

Clinical practice often diverges from clinical practice guidelines. Espeland and Baerheim’s122 study of GP referrals for plain radiography of the lumbar spine in different countries found that 60% did not conform to documented guidelines. The factors that affected their decisions included clinical ordering criteria, patient wishes, the practitioner’s response, uncertainty, professional dignity, access to services, perception of patient needs, sense of pressure from other health-care providers and expectations about the consequences. The authors suggested that there are several attitude-related and external barriers, which change over time.

A longitudinal observational study123 of a tertiary care psychiatric service over a 10-year period (1988–97), including 4429 consecutive referrals, exposed significant changes during the study period. Not only were changes in the diagnosis of patients observed over time (e.g. levels of psychosocial and somatic functioning of referred patients decreased), but management recommendations were also altered. Interestingly, these authors note that owing to a decrease in length of stay over the 10-year observation period, the correlation of lag time (the time from admission to the hospital until referral to community psychiatry service) and length of stay decreased.

Secondary analysis of data

Seven teams were involved in both studies (COOP1 and EEICC2). In general, these teams showed small increases in the number of referrals over the period of the two studies. It is important to remember when considering these results that the referral agent was not noted on the records of all patients.

For the seven teams taking part in both studies, the most notable changes are the decrease in referrals from AHPs (from 34.2% to 12.5%) and the increase in referrals from acute settings (from 18% to 32%). Referrals from GPs over the two study periods rose slightly (from 10.9% to 16.3%) and there has been a reduction in referrals by social workers (from 8.1% to 4.4%). There has been little change in referral patterns from informal carers, friends and family and A&E (see Table 19).

When considering teams as a whole, i.e. including data from teams that were not in both studies, referrals from acute settings remain fairly stable (from 18.8% to 24.1%). Referrals from AHPs are slightly decreased (from 26.9% to 21.7%) and there is a slight increase in referrals from community nurses (from 8.7% to 11.1%). One referral route that showed similar changes to the data from the seven teams was social workers. Referrals through this route declined (from 9.9% to 6.2%) (see Table 19).

Comparing teams exposes the great variation in referral rates and demonstrates that these have not changed in a similar fashion over the period. Some teams show greater changes in referral routes between the two study periods. For example, team 2 shows a referral rate from acute hospitals changing from 13.2% to 40.5% as compared with team 3, which changed from 4.7% to 10.7%. Team 1 shows an increased referral rate, again from acute hospitals, from 23.1% to 33.1% (see Table 19) and team 5 shows the opposite trend with a reduction in referral rates from acute hospitals (from 77.8% to 50.2%). Most teams have shown a broadening of referral routes between the two study periods; team 5 showed only three referral routes in the first study and this had broadened to seven referral routes in the second study.

We investigated whether or not the referral route has an impact on patient outcome. It is possible that some referrers will be able to refer patients who would be more likely to benefit from IC, whereas others may not be identifying patients who would be the most appropriate for these services.

Figure 22 relates the referral route with changes in impairment on the TOM. This indicates that referrals from A&E and acute hospitals benefit most with the least benefit coming from those directly referred by nurses and social services.

FIGURE 22.

Change in TOM impairment by referral route.

This finding is confirmed by examining referral routes against the EQ-5D (Figure 23).

FIGURE 23.

Change in EQ-5D score by referral route.

Changes in patient characteristics at admission

Table 23 in Chapter 7 shows the percentage of patients with each LoC at admission for each of the two studies. This is shown individually for each of the seven teams, that were in both studies, the average of the data from the teams, that were in both studies and for all the data.

Key point 78: the nature of referrals has changed over time in different ways for different teams.

The average LoC at admission, when considering the teams that were included in both studies (see Table 23), indicates a slightly higher percentage not requiring IC and more requiring slow-stream rehabilitation. However, when reviewing the data from all teams in both studies, it is noticeable that there has also been an increase in the numbers of patients not needing any intervention (but from a higher baseline) which may lead one to consider whether or not there has been an increase in the number of inappropriate referrals over time. Across the whole data set, there is also a large increase in LoC = 1 (needs prevention/maintenance programme).

Key point 79: there is some indication of an increase in inappropriate referrals to IC in more recent years.

Therapy outcome measures

Despite this increase in inappropriate referrals, there is some indication from the two studies that the proportion of patients being referred to IC who have more severe impairments has increased over time with 23% having scores of 2.5 or lower in COOP1 and 29.7% having scores of 2.5 or lower in EEICC2 (Table 26). Many patients with severe impairments would not be seen as likely to benefit from enablement or rehabilitation; their needs being judged as requiring more social care support alone.

TABLE 26 - Therapy outcome measure impairment score at admission, categorised and split by study

TOM impairment score on admission	Study
	COOP,1 % (n) of patients (N = 456)	EEICC,2 % (n) of patients (N = 2172)
0–2.5	23 (105)	29.7 (645)
3–5	77 (351)	70.3 (1527)

However, these averages fail to demonstrate the complexity at an individual team level, as demonstrated in Figure 24. Although patients in some teams (teams 2, 3, 4 and 5) show more impairment in the second study, patients in teams 1, 6 and 7 are (on average) less severely impaired.

FIGURE 24.

Therapy outcome measures impairment at admission for study 1 and 2. a, The average of the seven teams; and b, the average of all teams combined.

Although the trend is towards more severely impaired patients, the activity scores of the TOM do not show much difference between the studies (Table 27).

TABLE 27 - Therapy outcome measure activity score at admission, categorised and split by study

TOM activity score on admission	Study
	COOP,1 % (n) of patients (N = 460)	EEICC,2 % (n) of patients (N = 2172)
0–2.5	25.2 (116)	27.9 (605)
3–5	74.8 (344)	72.1 (1567)

Key point 82: a higher proportion of patients with more severe impairments are being referred to IC.

EQ-5D

There is little overall change between the studies for the health-related QoL measure on admission for patients from the seven teams included in both studies. However, for data from the whole study, there is a slight improvement (Figure 25). Three of the seven teams show an improvement in the health status of their patient population on admission for this measure, whereas four teams show a decline over time.

FIGURE 25.

The EQ-5D score at admission. a, The average of the seven teams; and b, the average of all teams combined.

Discussion

We were surprised that so little research had been conducted into the impact of health policy on service provision for care of the elderly. There is a lot to be learned from monitoring change over time in service provision and its impact on outcomes.

All new services take some time to be integrated into the cohesive structure of integrated care so it is not overly surprising that the numbers of referrals to IC have increased over time. Furthermore, this trend has been encouraged by explicit Department of Health policy and NHS incentives to care for the frail elderly within the community. However, there remains substantial variation between IC teams in the number of referrals they receive through different routes and the increase in numbers over time. The greatest change has been in referrals from acute hospitals and we suspect that this was associated with greater emphasis on reducing length of stay. This could be one of the reasons why a greater percentage of more impaired patients are being referred to IC.

The number of referral routes to IC has increased, which is likely to be associated with greater recognition of the services provided by a broader community of health- and social-care professionals.

The finding that those referred by A&E and hospital departments improved to a greater extent than those referred directly by nurses, many of whom would be community based, is not surprising. It is likely that these patients would have had acute disabling conditions, whereas a greater proportion of those referred from the community would have chronic/frail and declining conditions.

Implications for clinicians, purchasers and researchers

Contributions of authors

The authors of this report are listed in agreed order related to their contribution. The following were involved in the construction of the original bid, including the research design: Pamela Enderby, Ann McDonnell, Tony Smith, Susan Nancarrow, Stuart Parker, Michael Campbell, Steven Ariss, Mike Bradburn and Anthony Ryan.

All authors were part of the project-working group, who met regularly throughout the project. A Project Management Committee consisted of Professor Pamela Enderby, Dr Steven Ariss and Mr Mike Bradburn and was responsible for the day-to-day management of the project. Professor Pamela Enderby was the chief investigator, Dr Steven Ariss was the project manager and Mr Mike Bradburn managed the data cleaning and statistical analyses.

The specific contribution of each author is listed below.

Dr Steven M Ariss (research fellow, Health Service Research) was the project manager for this project (EEICC 2) and for study 2 (EEICC). 2 He led the analysis and writing of Chapter 7. He led the collection and analysis of qualitative data and managed the quantitative data for study 2 (EEICC). 2 He carried out the initial data cleaning of the EEICC2 data set, prior to input to Stata. He also assisted Professor Pamela Enderby in compiling and editing this final report.

Professor Pamela M Enderby (Professor of Community Rehabilitation) was the chief investigator of this project. She contributed to both previous projects (COOP1 and EEICC). 2 She assisted with data cleaning, analyses and interpretation of the findings. She provided leadership to the working group and compiled the final report. She was lead author on Chapter 8.

Dr Tony Smith (senior lecturer, Health and Social Care Leadership) was a codesigner of the proposal for EEICC 12 and project manager, and a contributor to the EEICC 2 proposal. He led the analysis and writing of Chapter 5. He led the collection and input of qualitative data (EEICC 1). 2 He also assisted Professor Pamela Enderby in compiling and editing this final report.

Professor Susan A Nancarrow was the principal investigator on the COOP1 project. She reviewed and edited two drafts of the final compiled report, reformatted and restructured Chapter 1 (combined two original sections into one) and developed an outline of the proposed benchmarking framework.

Mike J Bradburn was the lead statistician. He undertook the merger of the databases and data manipulation, led the quantitative analysis and contributed to the writing of the report.

Deborah Harrop co-ordinated and carried out the literature reviews and was responsible for the citations and references in this report.

Professor Stuart G Parker was the joint lead author of Chapter 1 and lead author of Chapter 2.

Professor Ann McDonnell (Reader in Health and Social Care Research) led the literature review and the analysis of data to explore the characteristics of patients whose health does not improve or declines in IC. She led the writing up of this work and was joint lead for Chapter 1.

Professor Simon Dixon was responsible for the economic analysis and was the lead author for Chapter 6.

Dr Tony Ryan was the lead author of Chapter 3.

Alexandra Hayman undertook additional quantitative analyses and contributed to the writing of the report.

Professor Michael Campbell (Professor of Medical Statistics), was a co-applicant and attended management meetings and advised on data collection, form design and statistical analysis.

Project advisory group

The research advisory group was chaired by David Quinney and included representatives with extensive experience of care service management, service delivery, statistical analysis and a lay member. Members were Fiona Shield, Stuart Parker, Michael Campbell and Christine Allmark. This group provided valuable direction for the project and a knowledgeable external perspective.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HS&DR programme or the Department of Health.

Results summary

All searches have been written up for MEDLINE using the EBSCOhost interface.

Explanation of search terms used: / = medical subject heading (MeSH); exp = exploded MeSH; * = denotes any character/s; ? = denotes any character; ti = title word; ab = abstract word; N = adjacency of words; N3 = adjacency within 3 words; "" = phrase search

Chapter 1 search strategy

Numbers in brackets are the number of hits.

1. intermediate care.ti,ab. (1402)

2. patient* N5 hotel*.ti,ab. (111)

3. transition* N3 car*.ti,ab. (13,142)

4. halfway N3 home*.ti,ab. (11)

5. halfway house*.ti,ab. (192)

6. subacute N3 care.ti,ab. (365)

7. community N5 rehabilitation.ti,ab. (1533)

8. rehabilitat* N5 cent*.ti,ab. (4794)

9. home* N5 hospital*.ti,ab. (7991)

10. facilitat* N3 discharge*.ti,ab. (504)

11. support* N3 discharge*.ti,ab. (700)

12. expedit* N3 discharge*.ti,ab. (55)

13. earl* N3 discharge*.ti,ab. (3758)

14. home* N3 car*.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N5 hospital*.ti,ab. (20,421)

17. communit* N3 care.ti,ab. (7703)

18. intermediate care facilities/ (594)

19. subacute care/ (701)

20. exp rehabilitation centers/ (10,711)

21. hospitals, community/ (9603)

22. exp home care services/ (36,693)

23. or/1-22 (71,742)

24. old* N3 people.ti,ab. (16,058)

25. old* N3 person*.ti,ab. (10,293)

26. elder* N3 people.ti,ab. (12,278)

27. elder* N3 person*.ti,ab. (7412)

28. old* N3 generation*.ti,ab. (811)

29. elder* N3 generation*.ti,ab. (115)

30. pensioner*.ti,ab. (709)

31. geriatric*.ti,ab. (27,648)

32. exp aged/ (2,026,009)

33. geriatrics/ (25,588)

34. or/24-33 (2,056,204)

35. therap* N3 outcome* N3 measure*.ti,ab. (758)

36. quality N2 life.ti,ab. (114,333)

37. qaly*.ti,ab. (3666)

38. benefi*.ti,ab. (417,322)

39. return* N3 home*.ti,ab. (2386)

40. exp outcome assessment (health care)/ (547,012)

41. exp treatment outcome/ (510,872)

42. quality-adjusted life years/ (5185)

43. or/35-42 (998,477)

44. 23 and 34 and 43 (3503)

45. publication date 20080101-20110131 (981)

Chapter 2 search strategy

Numbers in brackets are the number of hits.

1. intermediate care.ti,ab. (1402)

2. patient* N5 hotel*.ti,ab. (111)

3. transition* N3 car*.ti,ab. (13,142)

4. halfway adj3 home*.ti,ab. (11)

5. halfway house*.ti,ab. (192)

6. subacute N3 care.ti,ab. (365)

7. community N5 rehabilitation.ti,ab. (1533)

8. rehabilitat* N5 cent*.ti,ab. (4794)

9. home* N5 hospital*.ti,ab. (7991)

10. facilitat* N3 discharge*.ti,ab. (504)

11. support* N3 discharge*.ti,ab. (700)

12. expedit* N3 discharge*.ti,ab. (55)

13. earl* N3 discharge*.ti,ab. (3758)

14. home* N3 car*.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N5 hospital*.ti,ab. (20,421)

17. communit* N3 care.ti,ab. (7703)

18. intermediate care facilities/ (594)

19. subacute care/ (701)

20. exp rehabilitation centers/ (10,711)

21. hospitals, community/ (9603)

22. exp home care services/ (36,693)

23. or/1-22 (71,742)

24. old* N3 people.ti,ab. (16,058)

25. old* N3 person*.ti,ab. (10,293)

26. elder* N3 people.ti,ab. (12,278)

27. elder* N3 person*.ti,ab. (7412)

28. old* N3 generation*.ti,ab. (811)

29. elder* N3 generation*.ti,ab. (115)

30. pensioner*.ti,ab. (709)

31. geriatric*.ti,ab. (27,648)

32. exp aged/ (2,026,009)

33. geriatrics/ (25,588)

34. or/24-33 (2,056,204)

35. tom.ti,ab. (1888)

36. toms.ti,ab. (230)

37. "therapy outcome measure*".ti,ab. (71)

38. EQ5D.ti,ab. (358)

39. "quality of life".ti,ab. (115,094)

40. qaly*.ti,ab. (3832)

41. "quality adjusted life years".ti,ab. (2513)

42. exp outcome assessment (health care)/ (565,906)

43. exp treatment outcome/ (528,785)

44. quality-adjusted life years/ (5408)

45. or/35-44 (662,574)

46. 23 and 35 and 45 (5024)

47. publication date 20080101-20120431(1512)

Chapter 3 search strategy

Numbers in brackets are the number of hits.

1. intermediate care.ti,ab. (1402)

2. patient* N5 hotel*.ti,ab. (111)

3. transition*N3 car*.ti,ab. (13,142)

4. halfway N3 home*.ti,ab. (11)

5. halfway house*.ti,ab. (192)

6. subacute N3 care.ti,ab. (365)

7. community N5 rehabilitation.ti,ab. (1533)

8. rehabilitat* N5 cent*.ti,ab. (4794)

9. home* N5 hospital*.ti,ab. (7991)

10. facilitat* N3 discharge*.ti,ab. (504)

11. support* N3 discharge*.ti,ab. (700)

12. expedit* N3 discharge*.ti,ab. (55)

13. earl* N3 discharge*.ti,ab. (3758)

14. home* N3 car*.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N5 hospital*.ti,ab. (20,421)

17. communit* N3 care.ti,ab. (7703)

18. intermediate care facilities/ (594)

19. subacute care/ (701)

20. exp rehabilitation centers/ (10,711)

21. hospitals, community/ (9603)

22. exp home care services/ (36,693)

23. or/1-22 (71,742)

24. old* N3 people.ti,ab. (16,058)

25. old* N3 person*.ti,ab. (10,293)

26. elder* N3 people.ti,ab. (12,278)

27. elder* N3 person*.ti,ab. (7412)

28. old* N3 generation*.ti,ab. (811)

29. elder* N3 generation*.ti,ab. (115)

30. pensioner*.ti,ab. (709)

31. geriatric*.ti,ab. (27,648)

32. exp aged/ (2,026,009)

33. geriatrics/ (25,588)

34. or/24-33 (2,056,204)

35. institutionali*.ti,ab. (9711)

36. exp institutionalization/ (7391)

37. or/35-36 (15,212)

38. 23 and 34 and 37 (252)

39. publication date 20080101-20110131(30)

Chapter 4 search strategy

Numbers in brackets are the number of hits.

1. intermediate care.ti,ab. (1402)

2. patient* N5 hotel*.ti,ab. (111)

3. transition* N3 car*.ti,ab. (13,142)

4. halfway N3 home*.ti,ab. (11)

5. halfway house*.ti,ab. (192)

6. subacute N3 care.ti,ab. (365)

7. community N5 rehabilitation.ti,ab. (1533)

8. rehabilitat* N cent*.ti,ab. (4794)

9. home* N hospital*.ti,ab. (7991)

10. facilitat* N discharge*.ti,ab. (504)

11. support* N discharge*.ti,ab. (700)

12. expedit* N discharge*.ti,ab. (55)

13. earl* N discharge*.ti,ab. (3758)

14. home* N3 car*.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N5 hospital*.ti,ab. (20,421)

17. communit* N3 care.ti,ab. (7703)

18. intermediate care facilities/ (594)

19. subacute care/ (701)

20. exp rehabilitation centers/ (10,711)

21. hospitals, community/ (9603)

22. exp home care services/ (36,693)

23. or/1-22 (71,742)

24. old* N3 people.ti,ab. (16,058)

25. old* N3 person*.ti,ab. (10,293)

26. elder* N3 people.ti,ab. (12,278)

27. elder* N3 person*.ti,ab. (7412)

28. old* N3 generation*.ti,ab. (811)

29. elder* N3 generation*.ti,ab. (115)

30. pensioner*.ti,ab. (709)

31. geriatric*.ti,ab. (27,648)

32. exp aged/ (2,026,009)

33. geriatrics/ (25,588)

34. or/24-33 (2,056,204)

35. "meals on wheels".ti,ab. (113)

36. community N4 nurs*.ti,ab. (7804)

37. neighbo*rhood N4 nurs*.ti,ab. (55)

38. district N4 nurs*.ti,ab. (1682)

39. public N4 health N4 nurs*.ti,ab. (5120)

40. home* N4 nurs*.ti,ab. (24,661)

41. home* N4 car*.ti,ab. (30,147)

42. domicil* N3 therap*.ti,ab. (210)

43. home* N3 therap*.ti,ab. (3682)

44. home* N3 rehabilitat*.ti,ab. (1272)

45. food services/

46. community health nursing/

47. home care services/

48. house calls/

49. or/35-48

50. follow*.ti,ab. (1,997,134)

51. after*.ti,ab. (2,993,989)

52. subsequent*.ti,ab. (466,724)

53. or/50-52 (4,494,480)

54. 23 and 34 and 49 (8019)

55. 53 and 54 (1721)

56. publication date 20080101-20120131 (408)

Chapter 4 search strategy

Numbers in brackets are the number of hits.

1. intermediate care.ti,ab. (1402)

2. patient* N5 hotel*.ti,ab. (111)

3. transition* N3 car*.ti,ab. (13,142)

4. halfway N3 home*.ti,ab. (11)

5. halfway house*.ti,ab. (192)

6. subacute N3 care.ti,ab. (365)

7. community N5 rehabilitation.ti,ab. (1533)

8. rehabilitat* N5 cent*.ti,ab. (4794)

9. home* N5 hospital*.ti,ab. (7991)

10. facilitat* N3 discharge*.ti,ab. (504)

11. support* N3 discharge*.ti,ab. (700)

12. expedit* N3 discharge*.ti,ab. (55)

13. earl* N3 discharge*.ti,ab. (3758)

14. home* N3 car*.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N5 hospital*.ti,ab. (20,421)

17. communit* N3 care.ti,ab. (7703)

18. intermediate care facilities/ (594)

19. subacute care/ (701)

20. exp rehabilitation centers/ (10,711)

21. hospitals, community/ (9603)

22. exp home care services/ (36,693)

23. or/1-22 (71,742)

24. old* N3 people.ti,ab. (16,058)

25. old* N3 person*.ti,ab. (10,293)

26. elder* N3 people.ti,ab. (12,278)

27. elder* N3 person*.ti,ab. (7412)

28. old* N3 generation*.ti,ab. (811)

29. elder* N3 generation*.ti,ab. (115)

30. pensioner*.ti,ab. (709)

31. geriatric*.ti,ab. (27,648)

32. exp aged/ (2026009)geriatrics/ (25,588)

33. or/24-33 (2,056,204)

34. mortalit*.ti,ab. (373,017)

35. death.ti,ab. (380,908)

36. dying.ti,ab. (21,582)

37. mortality/ (31,253)

38. death/ (10,887)

39. or/35-39 (717,348)

40. 23 and 34 and 40 (1548)

41. publication date 20080101-20110131 (332)

Chapter 5 search strategy

Numbers in brackets are the number of hits.

1. interdisciplinary.ti,ab. (16,908)

2. interprofessional.ti,ab. (2059)

3. multiprofessional.ti,ab.(663)

4. multidisciplinary.ti,ab. (34,399)

5. inter-disciplinary.ti,ab. (290)

6. inter-professional.ti,ab. (393)

7. co-operat*.ti,ab. (8157)

8. multi-professional.ti,ab. (459)

9. multi-disciplinary.ti,ab. (2574)

10. "inter disciplinary".ti,ab. (290)

11. "inter professional".ti,ab. (393)

12. "multi disciplinary.ti,ab. (2574)

13. "multi professional".ti,ab. (459)

14. cooperat*/ti,ab. (83,511)

15. collaborat*.ti,ab. (64,743)

16. or/1-15 (202,352)

17. team*.ti,ab. (74,350)

18. patient care team/ (47,188)

19. or/18-19 (107,956)

20. team* N1 size*.ti,ab. (59)

21. grade* N1 mix*.ti,ab. (190)

22. skill* N1 mix*.ti,ab. (577)

23. intensity.ti,ab. (208,889)

24. workload*.ti,ab. (14,122)

25. leadership.ti,ab. (16,538)

26. satisf*.ti,ab. (181,055)

27. role* N2 flexibility.ti,ab. (247)

28. integrat*.ti,ab. (226,942)

29. team* N2 work*.ti,ab. (4199)

30. management.ti,ab. (575,321)

31. "career progression".ti,ab. (159)

32. quality.ti,ab. (435,142)

33. empower*.ti,ab. (9366)

34. communicat*.ti,ab. (14,2904)

35. "sense of direction".ti,ab. (139)

36. uncertainty.ti,ab. (31,353)

37. culture*.ti,ab. (701,693)

38. conflict*.ti,ab. (63,560)

39. training.ti,ab. (197,867)

40. workload/ (13,250)

41. leadership/ (23,694)

42. communication/ (55,282)

43. conflict psychology/ (13,018)

44. culture/ (24,064)

45. or/20-44 (2,573,427)

46. treatment* N1 outcome*.ti,ab. (31,821)

47. mortality.ti,ab. (377,671)

48. "length of stay".ti,ab. (22,662)

49. patient* N1 admission*.ti,ab. (8163)

50. patient* N1 discharge*.ti,ab. (17,686)

51. patient* N1 readmission*.ti,ab. (580)

52. patient* N1 transfer*. ti,ab. (5271)

53. "quality of health care".ti,ab. (2419)

54. outcome* N1 assessment*.ti,ab. (4605)

55. treatment* N1 fail*.ti,ab. (22,580)

56. "cause* of death*".ti,ab. (43,437)

57. "child mortalit*.ti,ab. (1577)

58. fatal* N1 outcome*.ti,ab. (6508)

59. "fetal mortalit*".ti,ab. (1171)

60. hospital* N2 mortalit*.ti,ab. (22,155)

61. "infant mortalit*".ti, ab. (5952)

62. "maternal mortalit*". ti,ab. (5111)

63. "perinatal mortalit*". ti,ab. (7125)

64. surviv* N1 rate*.ti,ab. (87,735)

65. patient* N1 admitted*.ti,ab. (37,601)

66. patient* N1 readmit*.ti,ab. (966)

67. patient* re-admit*.ti, ab. (147)

68. patient* re-admission*.ti,ab. (56)

69. length of stay/ (49,269)

70. patient admission/ (163,82)

71. patient discharge/ (16,155)

72. patient readmission/ (6227)

73. patient transfer/ (4987)

74. quality of health care/ (50,269)

75. outcome and process assessment/ (19,442)

76. treatment failure/ (22,713)

77. cause of death/ (31,025)

78. child mortality/ (921)

79. fatal outcome/ (43,961)

80. fetal mortality/ (190)

81. hospital mortality/ (17,663)

82. infant mortality/ (23,300)

83. maternal mortality/ (7059)

84. perinatal mortality/ (474)

85. survival rate/ (107,386)

86. quality of health care/ (50,269)

87. process assessment (health care)/ (2570)

88. or/46-87 (852,356)

89. "intermediate care".ti,ab. (902)

90. patient* N3 hotel*.ti,ab. (71)

91. transition* N3 car*.ti,ab. (13,149)

92. halfway N3 home*.ti,ab. (11)

93. "halfway house*".ti,ab. (191)

94. subacute N3 care.ti,ab. (367)

95. community N3 rehabilitation.ti,ab. (1210)

96. rehabilitat* N3 cent*.ti,ab. (4298)

97. home* N4 hospital*.ti,ab. (7155)

98. facilitat* N3 discharge*.ti,ab. (506)

99. support* N3 discharge*.ti,ab. (702)

100. expedit* N3 discharge*.ti,ab. (56)

101. earl* N3 discharge*.ti,ab. (3758)

102. home* N3 car* N3 service*.ti,ab. (2283)

103. home* N3 treatment*.ti,ab. (4359)

104. communit* N3 hospital*.ti,ab. (17,855)

105. communit* N3 care.ti,ab. (7705)

106. primary N3 care.ti,ab. (70,690)

107. "general practitioner*".ti,ab. (32,292)

108. GP*.ti.ab. (97,306)

109. "family physician*".ti,ab. (10,135)

110. community N3 nurs*.ti.ab. (7058)

111. "post acute care".ti,ab. (234)

112. "step up".ti,ab. (1367)

113. "step down".ti,ab. (195)

114. community N3 rehab*.ti,ab. (1260)

115. home* N3 nurs*.ti,ab. (24,072)

116. home* N3 car*.ti,ab. (27,538)

117. neighbo*rhood N3 nurs*.ti,ab. (52)

118. district N4 nurs*.ti,ab. (1681)

119. public N3 health N3 nurs*.ti,ab. (4935)

120. domicil* N3 therap*.ti,ab. (210)

121. home* N3 therap*.ti,ab. (3682)

122. home* N3 rehabilitat*.ti,ab. (1275)

123. care N3 home* N3 manager*.ti,ab. (4)

124. nursing N3 home* manager*.ti,ab. (2)

125. commission*.ti,ab. (21,902)

126. provider* N3 care*.ti,ab. (28,430)

127. nursing homes/ (25,961)

128. primary health care/ (45,727)

129. physicians, family/ (14,001)

130. general practitioners/ (599)

131. intermediate care facilities/ (594)

132. subacute care/ (702)

133. rehabilitation centers/ (6051)

134. hospitals, community/ (9603)

135. exp home care services/ (36,767)

136. hospitals, community/ (9608)

137. community health nursing/ (17,613)

138. home nursing/ (7642)

139. house calls/ (2035)

140. health personnel/ (18,038)

141. community health services/ (24,665)

142. community health centers/ (5352)

143. or/89-142 (480,040)

144. 16 and 19 and 45 and 88 and 143 (922)

145. publication date 20080101-20120331 (293)

Chapter 5 (additional) search strategy

Numbers in brackets are the number of hits.

1. support* N2 staff*.ti,ab. (379)

2. support* N2 worker*.ti,ab. (200)

3. non-professional*.ti,ab. (43)

4. or/1-3 (613)

5. 4 N5 30 (41)

6. professional*.ti,ab. (17,637)

7. management.ti,ab. (33,065)

8. leader*.ti,ab. (5364)

9. leadership/ (23,694)

10. or/6-9 (75,679)

11. 5 N5 10 (91)

12. 10 N5 30 (6664)

13. team* N1 size*.ti,ab. (59)

14. grade* N1 mix*.ti,ab. (190)

15. skill* N1 mix*.ti,ab. (577)

16. or/13-16 (1243)

17. length of stay/ (49,269)

18. patient admission/ (16,382)

19. patient discharge/ (16,155)

20. "length of stay".ti,ab. (22,662)

21. patient* N1 admission*.ti,ab. (8163)

22.  patient* N1 discharge*.ti,ab. (17,686)

23.  patient* N1 readmission*.ti,ab. (580)

24.  patient* N1 transfer*. ti,ab. (5271)

25. dur* N3 stay*.ti,ab. (2303)

26. or/17-25 (37,350)

27. or/16, 26 (38,518)

28. patient*.ti,ab. (138,013)

29. patients/ (146,346)

30. or/27-28 (271,334)

31. 16 N10 30 (28)

32. staff* N10 ratio*.ti,ab. (829)

33. patient* N10 ratio*.ti,ab. (8422)

34. interdisciplinary.ti,ab. (16,908)

35. interprofessional.ti,ab. (2059)

36. multiprofessional.ti,ab.(663)

37. multidisciplinary.ti,ab. (34,399)

38. inter-disciplinary.ti,ab. (290)

39. inter-professional.ti,ab. (393)

40. co-operat*.ti,ab. (8157)

41. multi-professional.ti,ab. (459)

42. multi-disciplinary.ti,ab. (2574)

43. "inter disciplinary".ti,ab. (290)

44. "inter professional".ti,ab. (393)

45. "multi disciplinary.ti,ab. (2574)

46. "multi professional".ti,ab. (459)

47. cooperat*/ti,ab. (83,511)

48. collaborat*.ti,ab. (64,743)

49. or/34-48 (202,352)

50. or/ 5, 11, 12, 32, 33, 31 (15,676)

51. 27 and 49 and 50 (57)

52. publication date 20080101-20120331 (44)

Chapter 7 search strategy

Numbers in brackets are the number of hits.

1. team* N1 size*.ti,ab. (59)

2. grade* N1 mix*.ti,ab. (190)

3. skill* N1 mix*.ti,ab. (577)

4. work* N1 distribut*. ti,ab. (765)

5. role* N1 distribut*.ti,ab. (395)

6. “role* structure*”.ti,ab. (43)

7. team* N2 distribut*.ti,ab. (81)

8. team* N2 organi?*.ti,ab. (902)

9. team* N2 structur*.ti,ab. (378)

10. team* N1 model*.ti,ab. (421)

11. team* N2 style*.ti,ab. (34)

12. team* N2 strategy* (89)

13. team* N2 strategies* (199)

14. team* N2 plan*.ti,ab. (586)

15. service* N2 distribut*.ti,ab. (680)

16. service* N2 organi?*.ti,ab. (4767)

17. service* N2 structur*.ti,ab. (1061)

18. service* N1 model*.ti,ab. (1951)

19. service* N2 style*.ti,ab. (122)

20. service* N2 strategy*.ti,ab. (252)

21. service* N2 strategies.ti,ab. (540)

22. management* N1 style*. ti,ab. (518)

23. management* N1 strategy*. ti,ab. (4078)

24. “management* strategies.” ti,ab. (9601)

25. management* N1 structur*.ti,ab. (1207)

26. management* N1 model*.ti,ab. (1538)

27. manager* N2 style*.ti,ab. (112)

28. manager* N2 strategy*.ti,ab. (50)

29. manager* N2 strategies.ti,ab. (164)

30. manager* N2 structur*.ti,ab. (110)

31. manager* N1 model*.ti,ab. (120)

32. leadership N1 style*.ti,ab. (520)

33. leadership N1 strategy*.ti,ab. (49)

34. leadership N1 strategies.ti,ab. (95)

35. leadership N1 structur*.ti,ab. (137)

36. leadership N1 model*.ti,ab. (218)

37. “service provision*”.ti,ab. (3106)

38. service N2 goal*.ti,ab. (217)

39. service N2 aim*.ti,ab. (395)

40. service N2 vision*.ti,ab. (101)

41. service N2 mission*.ti,ab. (117)

42. “organi?ation* provision*”.ti,ab. (30)

43. organi?ation* N2 goal*.ti,ab. (734)

44. organi?ation N2 aim*.ti,ab. (466)

45. organi$ation N2 vision*.ti,ab. (180)

46. organi$ation N2 mission*.ti,ab. (334)

47. service* N2 commission*.ti,ab. (575)

48. service* N2 “policy driver*”.ti,ab. (2)

49. service* N2 re-organi$*.ti,ab. (26)

50. service* N2 reorgani$*.ti,ab. (346)

51. service* N2 fund*,ti,ab. (1446)

52. service* N2 restructur*.ti,ab. (224)

53. service* N2 re-structur*.ti,ab. (3)

54. organi?ation* N2 commission*.ti,ab. (159)

55. organi?*ation* N2 re-organi$*.ti,ab. (576)

56. organi?ation* N2 reorgani$*.ti,ab. (99)

57. organi?ation* N2 fund*,ti,ab. (827)

58. organi?ation* N2 restructur*.ti,ab. (188)

59. organi?ation* N2 re-structur*.ti,ab. (2)

60. organi?ation* N2 style*.ti,ab. (123)

61. organi?ation* adj2 strategy*.ti,ab. (494)

62. “organi?ation* stategies*”.ti,ab. (272)

63. organi?ation* N2 structur*.ti,ab. (12,489)

64. organi?ation* N1 model*.ti,ab. (1238)

65. “financial policy*”.ti,ab. (23)

66. “financial policies”.ti.ab. (25)

67. “economic policy*”.ti,ab. (168)

68. “economic policies”.ti,ab. (154)

69. /or 1-68 (31,734)

70. "intermediate care".ti,ab. (902)

71. patient* N3 hotel*.ti,ab. (71)

72. transition* N3 car*.ti,ab. (13,149)

73. halfway N3 home*.ti,ab. (11)

74. "halfway house*".ti,ab. (191)

75. subacute N3 care.ti,ab. (367)

76. community N3 rehabilitation.ti,ab. (1210)

77. rehabilitat* N3 cent*.ti,ab. (4298)

78. home* N4 hospital*.ti,ab. (7155)

79. facilitat* N3 discharge*.ti,ab. (506)

80. support* N3 discharge*.ti,ab. (702)

81. expedit* N3 discharge*.ti,ab. (56)

82. earl* N3 discharge*.ti,ab. (3758)

83. home* N3 car* N3 service*.ti,ab. (2283)

84. home* N3 treatment*.ti,ab. (4359)

85. communit* N3 hospital*.ti,ab. (17855)

86. communit* N3 care.ti,ab. (7705)

87. primary N3 care.ti,ab. (70,690)

88. "general practitioner*".ti,ab. (32,292)

89. GP*.ti.ab. (97,306)

90. "family physician*".ti,ab. (10,135)

91. community N3 nurs*.ti.ab. (7058)

92. "post acute care".ti,ab. (234)

93. "step up".ti,ab. (1367)

94. "step down".ti,ab. (1957)

95. community N3 rehab*.ti,ab. (1260)

96. home* N3 nurs*.ti,ab. (24,072)

97. home* N3 car*.ti,ab. (27,538)

98. neighbo*rhood N3 nurs*.ti,ab. (52)

99. district N4 nurs*.ti,ab. (1681)

100. public N3 health N3 nurs*.ti,ab. (4935)

101. domicil* N3 therap*.ti,ab. (210)

102. home* N3 therap*.ti,ab. (3682)

103. home* N3 rehabilitat*.ti,ab. (1275)

104. care N3 home* N3 manager*.ti,ab. (4)

105. nursing N3 home* manager*.ti,ab. (2)

106. provider* N3 care*.ti,ab. (28,430)

107. nursing homes/ (25,961)

108. primary health care/ (45,727)

109. physicians, family/ (14,001)

110. general practitioners/ (599)

111. intermediate care facilities/ (594)

112. subacute care/ (702)

113. rehabilitation centers/ (6051)

114. hospitals, community/ (9603)

115. exp home care services/ (36,767)

116. hospitals, community/ (9608)

117. community health nursing/ (17,613)

118. home nursing/ (7642)

119. house calls/ (2035)

120. health personnel/ (18,038)

121. community health services/ (24,665)

122. community health centers/ (5352)

123. or/70-122 (460,761)

124. 69 and 123 (4192)

125. publication date 20080101-20120431 (1552)

Chapter 8 search strategy

Numbers in brackets are the number of hits.

1. "intermediate care".ti,ab. (902)

2. patient* N3 hotel*.ti,ab. (71)

3. transition* N3 car*.ti,ab. (13,149)

4. halfway N3 home*.ti,ab. (11)

5. "halfway house*".ti,ab. (191)

6. subacute N3 care.ti,ab. (367)

7. community N3 rehabilitation.ti,ab. (1210)

8. rehabilitat* N3 cent*.ti,ab. (4298)

9. home* N4 hospital*.ti,ab. (7155)

10. facilitat* N3 discharge*.ti,ab. (506)

11. support* N3 discharge*.ti,ab. (702)

12. expedit* N3 discharge*.ti,ab. (56)

13. earl* N3 discharge*.ti,ab. (3758)

14. home* N3 car* N3 service*.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N3 hospital*.ti,ab. (17,855)

17. communit* N3 care.ti,ab. (7705)

18. primary N3 care.ti,ab. (70,690)

19. "general practitioner*".ti,ab. (32,292)

20. GP*.ti.ab. (97,306)

21. "family physician*".ti,ab. (10,135)

22. community N3 nurs*.ti.ab. (7058)

23. "post acute care".ti,ab. (234)

24. "step up".ti,ab. (1367)

25. "step down".ti,ab. (1957)

26. community N3 rehab*.ti,ab. (1260)

27. home* N3 nurs*.ti,ab. (24,072)

28. home* N3 car*.ti,ab. (27,538)

29. neighbo*rhood N3 nurs*.ti,ab. (52)

30. district N4 nurs*.ti,ab. (1681)

31. public N3 health N3 nurs*.ti,ab. (4935)

32. domicil* N3 therap*.ti,ab. (210)

33. home* N3 therap*.ti,ab. (3682)

34. home* N3 rehabilitat*.ti,ab. (1275)

35. care N3 home* N3 manager*.ti,ab. (4)

36. nursing N3 home* manager*.ti,ab. (2)

37. provider* N3 care*.ti,ab. (28,430)

38. nursing homes/ (25,961)

39. primary health care/ (45,727)

40. physicians, family/ (14,001)

41. general practitioners/ (599)

42. intermediate care facilities/ (594)

43. subacute care/ (702)

44. rehabilitation centers/ (6051)

45. hospitals, community/ (9603)

46. exp home care services/ (36,767)

47. hospitals, community/ (9608)

48. community health nursing/ (17,613)

49. home nursing/ (7642)

50. house calls/ (2035)

51. health personnel/ (18,038)

52. community health services/ (24,665)

53. community health centers/ (5352)

54. or/1-53 (460,761)

55. team*,ti,ab. (75,913)

56. patient care team/ (47,188)

57. or/55-56 (109,851)

58. refer*.ti,ab. (446,319)

59. pattern*.ti,ab. (786,253)

60. structur*.ti,ab. (1,247,720)

61. decision*.ti,ab. (126,552)

62. behav*.ti,ab. (677,616)

63. decision*.ti,ab. (171,876)

64. or 58-63 (2,625,621)

65. 57 N1 64 (4191)

66. 54 and 57 and 64 (42)

67. publication date 20080101-20120431 (14)

Chapter 9 search strategy

Numbers in brackets are the number of hits.

1. "intermediate care".ti,ab. (902)

2. patient* N3 hotel*.ti,ab. (71)

3. transition* N3 car*.ti,ab. (13,149)

4. halfway N3 home*.ti,ab. (11)

5. "halfway house*".ti,ab. (191)

6. subacute N3 care.ti,ab. (367)

7. community N3 rehabilitation.ti,ab. (1210)

8. rehabilitat* N3 cent*.ti,ab. (4298)

9. home* N4 hospital*.ti,ab. (7155)

10. facilitat* N3 discharge*.ti,ab. (506)

11. support* N3 discharge*.ti,ab. (702)

12. expedit* N3 discharge*.ti,ab. (56)

13. earl* N3 discharge*.ti,ab. (3758)

14. home* N3 car* N3 service*.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N3 hospital*.ti,ab. (17,855)

17. communit* N3 care.ti,ab. (7705)

18. primary N3 care.ti,ab. (70,690)

19. "general practitioner*".ti,ab. (32,292)

20. GP*.ti.ab. (97,306)

21. "family physician*".ti,ab. (10,135)

22. community N3 nurs*.ti.ab. (7058)

23. "post acute care".ti,ab. (234)

24. "step up".ti,ab. (1367)

25. "step down".ti,ab. (1957)

26. community N3 rehab*.ti,ab. (1260)

27. home* N3 nurs*.ti,ab. (24,072)

28. home* N3 car*.ti,ab. (27,538)

29. neighbo*rhood N3 nurs*.ti,ab. (52)

30. district N4 nurs*.ti,ab. (1681)

31. public N3 health N3 nurs*.ti,ab. (4935)

32. domicil* N3 therap*.ti,ab. (210)

33. home* N3 therap*.ti,ab. (3682)

34. home* N3 rehabilitat*.ti,ab. (1275)

35. care N3 home* N3 manager*.ti,ab. (4)

36. nursing N3 home* manager*.ti,ab. (2)

37. commission*.ti,ab. (21,902)

38. provider* N3 care*.ti,ab. (28,430)

39. primary health care/ (45,727)

40. physicians, family/ (14,001)

41. general practitioners/ (599)

42. intermediate care facilities/ (594)

43. subacute care/ (702)

44. rehabilitation centers/ (6051)

45. hospitals, community/ (9603)

46. exp home care services/ (36,767)

47. hospitals, community/ (9608)

48. community health nursing/ (17,613)

49. home nursing/ (7642)

50. house calls/ (2035)

51. health personnel/ (18,038)

52. community health services/ (24,665)

53. community health centers/ (5352)

54. or/1-53 (437,544)

55. benchmark*.ti,ab. (12,724)

56. service* N2 configur*.ti,ab. (137)

57. service* N2 audit*.ti,ab. (390)

58. "performance indicator*".ti,ab. (1401)

59. performance* N2 measure*.ti,ab. (17,460)

60. monitor* N2 service*. ti,ab. (937)

61. compar* N2 service*.ti,ab. (2355)

62. quality N2 indicator*.ti,ab. (5274)

63. benchmarking/ (8670)

64. clinical audit/ (438)

65. exp quality indicators, health care/ (9573)

66. or/56-65 (52,354)

67. tool N3 kit*.ti,ab. (360)

68. toolkit*.ti,ab. (1488)

69. guide*.ti,ab. (276,818)

70. checklist*.ti,ab. (14,962)

71. "check list*".ti,ab. (1988)

72. guidance.ti,ab. (46,127)

73. referral*.ti,ab. (54,407)

74. recommendation*.ti,ab. (113,722)

75. checklist/ (858)

76. practice guidelines as topic/ (64,122)

77. referral and consultation/ (45,783)

78. or/68-77 (527,630)

79. 54 and 66 and 78 (1413)

Chapter 9 (with UK specification) search strategy

Numbers in brackets are the number of hits.

1. "intermediate care".ti,ab. (902)

2. patient* N3 hotel*.ti,ab. (71)

3. transition* N3 car*.ti,ab. (13,149)

4. halfway N3 home*.ti,ab. (11)

5. "halfway house*".ti,ab. (191)

6. subacute N3 care.ti,ab. (367)

7. community N3 rehabilitation.ti,ab. (1210)

8. rehabilitat* N3 cent*.ti,ab. (4298)

9. home* N4 hospital*.ti,ab. (7155)

10. facilitat* N3 discharge*.ti,ab. (506)

11. support* N3 discharge*.ti,ab. (702)

12. expedit* N3 discharge*.ti,ab. (56)

13. earl* N3 discharge*.ti,ab. (3758)

14. home* N3 car* N3 service.ti,ab. (2283)

15. home* N3 treatment*.ti,ab. (4359)

16. communit* N3 hospital*.ti,ab. (17,855)

17. communit* N3 care.ti,ab. (7705)

18. primary N3 care.ti,ab. (70,690)

19. "general practitioner*".ti,ab. (32,292)

20. GP*.ti.ab. (97,306)

21. "family physician*".ti,ab. (10,135)

22. community N3 nurs*.ti.ab. (7058)

23. "post acute care".ti,ab. (234)

24. "step up".ti,ab. (1367)

25. "step down".ti,ab. (1957)

26. community N3 rehab*.ti,ab. (1260)

27. home* N3 nurs*.ti,ab. (24,072)

28. home* N3 car*.ti,ab. (27,538)

29. neighbo*rhood N3 nurs*.ti,ab. (52)

30. district N4 nurs*.ti,ab. (1681)

31. public N3 health N3 nurs*.ti,ab. (4935)

32. domicil* N3 therap*.ti,ab. (210)

33. home* N3 therap*.ti,ab. (3682)

34. home* N3 rehabilitat*.ti,ab. (1275)

35. care N3 home* N3 manager*.ti,ab. (4)

36. nursing N3 home* manager*.ti,ab. (2)

37. commission*.ti,ab. (21,902)

38. provider* N3 care*.ti,ab. (28,430)

39. primary health care/ (45,727)

40. physicians, family/ (14,001)

41. general practitioners/ (599)

42. intermediate care facilities/ (594)

43. subacute care/ (702)

44. rehabilitation centers/ (6051)

45. hospitals, community/ (9603)

46. exp home care services/ (36,767)

47. hospitals, community/ (9608)

48. community health nursing/ (17,613)

49. home nursing/ (7642)

50. house calls/ (2035)

51. health personnel/ (18,038)

52. community health services/ (24,665)

53. community health centers/ (5352)

54. or/1-53 (437,544)

55. benchmark*.ti,ab. (12,724)

56. service* N2 configur*.ti,ab. (137)

57. service* N2 audit*.ti,ab. (390)

58. "performance indicator*".ti,ab. (1401)

59. performance* N2 measure*.ti,ab. (17,460)

60. monitor* N2 service*. ti,ab. (937)

61. compar* N2 service*.ti,ab. (2355)

62. quality N2 indicator*.ti,ab. (5274)

63. benchmarking/ (8670)

64. clinical audit/ (438)

65. exp quality indicators, health care/ (9573)

66. or/56-65 (52,354)

67. tool N3 kit*.ti,ab. (360)

68. toolkit*.ti,ab. (1488)

69. guide*.ti,ab. (276,818)

70. checklist*.ti,ab. (14,962)

71. "check list*".ti,ab. (1988)

72. guidance.ti,ab. (46,127)

73. referral*.ti,ab. (54,407)

74. recommendation*.ti,ab. (113,722)

75. checklist/ (858)

76. practice guidelines as topic/ (64,122)

77. referral and consultation/ (45,783)

78. or/68-77 (527,630)

79. 54 and 66 and 78 (1383)

80. UK*.ti,ab. (56,026)

81. united kingdom*.ti,ab. (21,109)

82. england*.ti.ab. (27,344)

83. "northern ireland*.ti,ab. (3061)

84. scotland*.ti,ab. (10,036)

85. wales*.ti,ab. (13,984)

86. english*.ti,ab. (40,592)

87. britain*.ti.ab. (11,214)

88. british.ti.ab. (31,370)

89. scottish.ti,ab. (5387)

90. "northern irish”.ti.ab. (149)

91. welsh.ti.ab. (1216)

92. or/81-91 (189,972)

93. 54 and 66 and 78 (1410)

94. 54 and 66 and 78 and 92 (150)

Missing patient characteristics data

Of the 8070 patients included, 3731 (46%) had missing data for at least one of the data items in the table above. The most common missing fields were the baseline questionnaires for EQ-5D (29%) or one of the four TOM items (16%). Of the remaining factors, 2024 (25%) patients had one missing data item and a further 724 (9%) had two or more missing. A total of 225 patients (3%) were excluded as a result of having excessive missing data (including age or sex), as there was too few observed data to base imputations on.

The remaining patients had missing data imputed by multiple imputation incorporating chained estimation. 45 Convergence issues were encountered when fitting models based on multiple logistic regression and as a result the number of predictive covariates had to be limited for the LoC, usual living arrangements, place receiving care and referral route. Although the four TOM domains were highly correlated, these were generally all present or all absent and so the domains could not be used to predict each other. Furthermore, although EQ-5D outcomes tended to be associated with the TOM outcomes they were more frequently missing and so TOM scores (when present) were used to predict EQ-5D but not vice versa.

Missing outcome data

Outcome data were imputed independently of the patient characteristic data, with the exceptions of the IC team and the baseline score (the latter was itself imputed when necessary). Along with these, additional data concerning the destination of the patient at discharge were also used to predict the missing TOM or EQ-5D outcome scores. Patients with missing data were categorised as having either died, discharged home, discharged to residential/nursing home or similar, admitted to hospital or an inappropriate referral. This categorisation was particularly relevant, as discharge to nursing home was associated with worse outcomes than patients discharged to their own home among patients for whom data were available. Patients who died were assigned a TOM or EQ-5D score of 0 at discharge; otherwise, the missing items were predicted based on the above along with the (possibly imputed) baseline score and team. Other covariates (e.g. age, sex) were deliberately omitted from the imputation of the outcome data, as the aim of the study was to assess whether or not there was an association and their inclusion would directly influence this.

Change between baseline and outcome measures

The relationship between baseline characteristics and change in the TOM and EQ-5D scores was assessed in two ways. First, by the magnitude of change in each of the TOM dimensions and EQ-5D scores and, second, by whether or not there was a positive or negative change in those scores. These analyses were done with complete case data and with a data set in which the missing values were imputed. A comparison of the analyses of the complete case data set and the imputed data set demonstrated that the results were generally in agreement.

Model coefficients for therapy outcome measures change: impairment

Model coefficients for therapy outcome measure change: activity

Model coefficients for therapy outcome measure change: participation