Organising health care services for people with an acquired brain injury: an overview of systematic reviews and randomised controlled trials

Two of the RCTs reported on mortality; of these, one of the trials found that there were reduced odds of mortality in the integrated care group at three, six and 12 months (OR 0.37; 95%CI 0.21 to 0.66 at 12 months) [16] whereas the other RCT found that there were similar levels of mortality between groups [14]. All three included RCTs reported on participants’ level of dependence in activities of daily living following intervention [15‐17]. While one of the RCTs found no significant differences between groups [15], the others both reported more favourable outcomes for the integrated care group; Indredavik and colleagues (2000) reported that a higher proportion of participants were independent six months after stroke as measured using the Rankin Scale (65% vs 52%, odds ratio (OR) = 1.72; 95% CI 1.10 to 2.70) [14]. In addition, Kalra and colleagues found that more participants in the integrated care group had a favourable outcome on the Barthel Index (scores 15–20 out of 20) than those receiving care from the mobile stroke team at three (OR 1.16; 95% CI 1.02 to 1.32) and 12 months (OR 1.27; 95% CI 1.12 to 1.44) [16]. Two of the RCTs reported on the impact on health related quality of life. One of the RCTs found no significant differences between groups [15] whereas the other RCT found that participants in the intervention group reported significantly improved quality of life (mean score 78.9 vs 75.2) [14].

All three RCTs reported on length of stay with variability in the findings. One RCT found that the total duration of hospital stay was similar between groups [16] while the remaining studies both reported a positive effect of the intervention on length of hospital stay. One RCT found that there was a 52% reduction in length of hospital stay for the integrated care group (mean of 14 days vs 29 days) [15] and the other found that the combined stroke unit plus rehabilitation unit length of stay was statistically shorter in the integrated care group (mean of 19 days vs 31 days) [14]. Finally, one of the RCTs assessed participant satisfaction and found that the group receiving integrated care reported statistically significant higher satisfaction with active participation in treatment program planning [15].

In summary, there is Level I evidence from three RCTs that the provision of integrated care may result in similar or reduced levels of mortality and similar or improved functional outcome when compared with a more fragmented approach. Furthermore, integrated care models used between the acute and home rehabilitation setting may result in a significantly reduced length of stay in acute services.

The SR and RCT both reported mortality rates post-intervention. The SR was unable to find any effects on mortality associated with use of an ICP. In contrast, the RCT found that patients receiving the intervention were significantly less likely to be dead or dependent at 90 days than patients receiving usual care (42% vs 58%, difference in absolute change = 15.7% (95% CI 5.8 to 25.4)). The SR and RCT both examined effects on quality of life; the SR identified evidence (based on one RCT [19]) that self-reported quality of life was significantly lower in the ICP group at six months whereas the RCT found that patients in the ICP group were more likely to have higher scores on the physical health component of the quality of life assessment tool (difference in absolute change = 3.4 (95% CI 1.2 to 5.5)) [19].

The SR and RCT both reported on resource use; the SR reported that readmissions to hospital were lower in the ICP group (based on one RCT and one NRCT (OR 0.11, 95% CI 0.03 to 0.39). The data from the SR related to length of stay was conflicting with two RCTs [19, 20] finding increased length of stay and two NRCTs [21, 22] finding shorter length of stay in the ICP group. The RCT conducted by Middleton and colleagues found no significant differences in length of hospital stay between groups [19]. The SR examined the effects of ICP use on patient satisfaction and found some evidence (based on one randomised study [20] that patient satisfaction (as rated on a scale of one to ten) was significantly lower in participants receiving care based on an ICP (Weighted Mean Difference (WMD) -1.1, 95% CI −1.91 to −0.29).

Based on the available evidence from one SR, there is currently a lack of evidence supporting the implementation of ICPs to improve patient outcomes for people with ABI and limited evidence that ICPs may reduce patient satisfaction and quality of life compared to usual care. However, there is conflicting Level I evidence from one high quality RCT that a specific pathway used on an acute stroke unit to manage fever, hyperglycaemia and swallowing can improve patient outcome of death or dependency at 90 days compared with usual care [18].

In the category of case management we included two SRs due to differences in the participants involved; one involved only people with stroke and the other involved only people with TBI. A Cochrane review published by Ellis and colleagues in 2010 included 16 RCTs that examined the effectiveness of stroke liaison workers for patients following stroke [23]. We found one RCT (n = 380) published after the search date of the SR that examined the effects of a six month case management intervention for people with stroke on discharge from acute care[24]. In addition, a systematic review published in 1999 examined the effectiveness of case management for patients following traumatic brain injury [25]. Since then, three RCTs involving people with a TBI have assessed the evidence for a case management intervention [26‐28]. Another RCT examined the evidence for case management focussed on returning people with an ABI to work [29]. The participant group comprised seven people with TBI, six people with cerebral infarcts, seven people with intracranial haemorrhage and two people with other diagnoses.

The SR conducted with stroke survivors found that there were no differences between groups on mortality at follow up [23]. Both SRs [23, 25] and two of the RCTs [26, 28] reported on levels of dependence in ADL function post intervention. Both SRs found that there were no significant differences between groups. However, Ellis and colleagues found in a subgroup analysis that stroke survivors with mild to moderate disability (Barthel score of 15 to 19) who received case management were more likely to be have a significant reduction in dependence (OR 0.62, 95% CI 0.44 to 0.87) in comparison to those who did not receive case management. One RCT conducted with people with TBI found that patients receiving case management had improved functional status (on a composite outcome measure) at follow up however the magnitude of the effect was unclear [28], whereas another RCT involving people with TBI found there were no significant differences between groups in ADL function measured using the Functional Independence Measure and Disability Rating Scale at follow-up [26]. Both SRs and four of the RCTs reported on levels of participation at follow up. The SR involving stroke survivors found that there were no significant differences between groups at follow up. The SR involving people with a TBI reported that there was some evidence (based on two NRCTs) that case management improved vocational status. The three RCTs conducted with people with TBI showed conflicting results; two of the RCTs were unable to find differences between groups in participation outcomes [26, 28] whereas the other RCT found that people in the intervention group reported less impact on work and leisure roles than those in the control group at follow up [27]. The RCT involving people with an ABI provided a case management intervention focussed on returning people to work and found that at six month follow up a higher number of people from the intervention group were employed (64% vs 36%) [29]. One SR [23] and five RCTs [24, 26‐29] examined the effects of case management intervention on quality of life. The SR and four of the RCTs found there were no differences in outcomes between groups [23, 24, 26, 27, 29]. The remaining study found that the intervention group had significantly better adjusted mean scores on the Euroqol (0.78 vs 0.67) [28].

The SR involving stroke survivors reported on participant satisfaction [23]; they found evidence (based on three RCTs included in their review) suggesting that participants receiving case management were more satisfied that someone had listened to them (OR 1.58, 95% CI 1.14 to 2.19).

In summary, one systematic review containing 16 RCTs and one RCT published since the review suggest that there are few overall significant benefits in providing case management services for people with stroke however there may be some benefits in providing case management for people with mild to moderate levels of disability. In addition, case management may increase satisfaction with care. One systematic review and four RCTs provided inconclusive evidence regarding the effectiveness of case management for people with TBI.

We identified a Cochrane review published in 2012 that examined the effect of early supported discharge programs after stroke [30]. Early supported discharge was defined in the review as “any intervention that aimed to accelerate discharge from hospital with the provision of support (with or without a 'therapeutic' rehabilitation intervention) in a community setting”. The review included 14 RCTs. No RCT, NRCT, CBA or ITS studies were identified which were published subsequently to this review and addressed this research question.

The SR reported on level of dependence in ADL and found that based on the results of nine trials there were no significant differences between groups. However, the SR found that patients receiving an ESD program were marginally more independent in extended ADL (standardised mean difference (SMD) 0.12 (95% CI 0.00 to 0.25)). The SR found no significant effect on quality of life or mortality. However, when assessing the outcome of death or dependency they found strong evidence that ESD programs led to similar or improved functional outcome (OR 0.80 (95% CI 0.67 to 0.97)) [30]. Subgroup analysis showed that patients with mild to moderate levels of disability (initial Barthel index >9/20) appeared to benefit most from ESD intervention however, it should be noted that the benefits associated with ESD were not always sustained at longer term follow up assessment [30].

The SR reported on the effect on length of stay and found ESD intervention was associated with a significant reduction in length of stay in acute care equivalent to approx seven days (Mean Difference (MD) -6.84 (95% CI −11.20 to −2.49)). Overall patients receiving ESD services were significantly more likely to report satisfaction with care services (OR 1.60, 95% CI 1.08 to 2.38, P = 0.02).

In summary, there is Level I evidence that ESD programs can improve functional outcomes, reduce length of stay and increase patient satisfaction. The authors of the review reached a consensus that the important factors of the ESD service included delivery of therapy in the home setting and coordinated multidisciplinary care. The greatest benefits were seen where the ESD team coordinated the hospital discharge, post-discharge care and delivery of home rehabilitation and support.

We identified six RCTs in this category [31‐36]; three of these studies involved people with a traumatic brain injury and the remaining involved people with stroke. No SRs, NRCT, CBA or ITS studies were identified which addressed this research question. The purpose of the intervention, and health professionals delivering the intervention in this category varied greatly. Study interventions included: visits from a geriatric rehabilitation physician or physiotherapist [31], assessment conducted by a stroke nurse six months after stroke [32], follow up in a multidisciplinary TBI clinic [33], routine follow up approximately one week post TBI from an occupational therapist or psychologist [34, 35] and follow up one month following discharge after stroke [36].

There were few significant findings in favour of short term programs of continuity of care. One RCT reported on mortality outcomes and found no significant differences between groups [31]. Two RCTs reported on level of dependence in ADL and found no significant differences between groups [31, 32]. Five RCTs reported on participation outcomes; three of these found no significant differences between groups [31‐33]. The remaining two RCTs were conducted with people with TBI and found that people in the control group were significantly more likely to have difficulties in participation (as measured by the Rivermead Head Injury follow up Questionnaire) than those in the intervention group [34, 35]. Two RCTs examined effects on quality of life and were unable to identify significant benefits associated with intervention [33, 36].

One of the RCTs assessed participant satisfaction and found that patients receiving the intervention reported that they were more satisfied with the information provided to them and the planning prior to their return home compared with usual care [31].

In summary, we identified a number of studies evaluating short term programs however, low study quality and clinical heterogeneity means that it is difficult to draw conclusions about these studies.

The SR reported broadly on the effect on patient outcomes (which included mortality, disability, quality of life and patient satisfaction). They found conflicting evidence as approximately half of the sixteen studies included in the review found mostly positive relationships with patient outcome whereas the other half reported either limited or no significant relationship [37]. One good quality RCT reported on mortality following intervention and found no significant differences between groups [38]. Three RCTs reported on participants’ level of dependence in ADLs post intervention [20, 38, 41]. Findings were mixed; one of the RCTs found no significant difference [20] whereas another reported that patients in the intervention group made more improvement on the motor component of the Functional Independence Measure (13.6% absolute difference in percentage of patients gaining more than 23 points) [41]. In contrast, Dirks and colleagues found that although the quality monitoring intervention achieved their goal of increasing thrombolysis rates, patients in the intervention group were less likely to have a good clinical outcome (Modified Rankin Scale <3) at 3 months (adjusted OR, 0.56 (95% CI 0.42 to 0.74) [38]. This same RCT found there were no significant differences between groups in quality of life [37].

Three RCTs reported on resource use. One RCT found that there were some cost savings associated with the intervention due to lower hospital admission costs (difference -$455 USD; 95% CI -$232 to -$679 USD) [38]. However, two RCTs conducted in stroke rehabilitation settings found no significant differences between groups in length of stay [20, 41]. Three RCTs [38‐40] and one NRCT [42] reported on the effect of the quality monitoring intervention on process measures. The study conducted by Dirks and colleagues found that rates of thrombolysis were increased in the intervention group (44.3% vs 39.8% in the control group (difference 4.5%; 95% CI 3.1% to 5.9%) [38]. Johnston and colleagues found that standardised stroke discharge orders increased the rates of ‘optimal treatment’ in individual patients (from 37% to 45%) [39] and Hinchey and colleagues found significant improvements in achieving one of four targeted process outcomes (98% in the intervention group compared to 87% in the control group) which was discharging a patient with atrial fibrillation on warfarin. Only one of the studies was unable to find any significant benefits in favour of the intervention group on process outcomes [40]. One RCT reported on participant satisfaction and found that patients in the intervention group reported significantly lower levels of satisfaction with intervention (mean 7.7 vs 8.8) [20].

In summary, level I evidence from one systematic review and six clinical trials indicates that while some improvements in processes outcomes have been associated with quality monitoring interventions, the effects on patient and health system outcomes are not consistent. There is not clear evidence supporting a particular approach to quality monitoring, and the primary studies evaluating these interventions were of variable methodological quality.