Our results show a significant improvement in both outcome measures (EQ-5D-3L utility score and FIM functional score) as well as a difference in costs in favor of the provision of a rehabilitation service to inpatients on a Saturday in addition to Monday to Friday compared to Monday to Friday usual care rehabilitation alone. The confidence interval ellipses for the ICERs show that all the 50%, 75%, and most of the 95% confidence intervals fall within the bottom right hand quadrant suggesting that the intervention is likely to be dominant over the comparator of usual care. Therefore, from a health service perspective, the provision of a rehabilitation service to inpatients on a Saturday in addition to Monday to Friday compared to Monday to Friday rehabilitation alone, is likely to be cost saving per QALY gained and for a MCID gained in functional independence.
The National Institute for Health and Clinical Excellence (NICE) in the UK
 reports a cost effectiveness threshold range of £20,000 (AUD$34,200) to £30,000 (AUD$51,400) per QALY gained
 and there are similar values reported in Australia for the Australian Pharmaceutical Benefits Advisory Committee
. The results of this clinical trial report an average cost saving of over AUD$40,000 per QALY gained, in contrast to the above mentioned willingness to pay per QALY gained. If the willingness to pay in this study was AUD$50,000 per QALY gained or per MCID gained in functional independence, the probability of cost effectiveness in the intervention group approached 100%. Our results are consistent with a recent systematic review that reported a more intensive inpatient rehabilitation service can result in reduced cost to the health service, while improving patient outcomes
This economic evaluation may have important implications for health services that offer inpatient rehabilitation, with potential to reduce costs per admission, improve patient outcomes and improve patient access. When considering the health service perspective, it is reported that in the inpatient rehabilitation setting the patient length of stay is the largest contributor to health care costs
[38, 39]. This may explain the likely reduction in cost for the intervention group, with an observed three day reduction in patient length of stay, over the rehabilitation admission and the 30 day readmission period. Policy makers may support this model of care with increased efficiency associated with cost savings, as it allows the same number of patient admissions to be managed at a lesser cost or may facilitate improved patient access to rehabilitation beds. This may lead to an improvement in the flow of patients through the health system and have a positive impact on the ‘bed block’ faced by acute wards
[40, 41]. For example, a 30 bed rehabilitation unit with an average length of stay of 29 days would have approximately 380 annual admissions; if the average length of stay reduced to 26 days, then annual admissions could potentially increase to 420.
Despite these benefits, implementation of this model of care also needs to consider workforce redesign. Traditionally, allied health clinicians work Monday to Friday, so the clinicians providing rehabilitation services may be reluctant to change their work practices by working on the weekend
. This may be negated by penalty rates providing an incentive to work on weekends. We note that we did not have a problem staffing the service in our clinical trial. Another issue for implementation involves who will pay and who will make the savings. This is about the redistribution of resources across budgets. Budget silos might mean that the costs will be incurred by allied health departments but the gains will be at the broader hospital level. If the funds come out of the smaller budgets of the departments providing the additional services (in this case physiotherapy or occupational therapy), it must be reconciled that these are not the budgets that accrue the overall savings generated at the hospital level.
The strengths of this economic evaluation are that it was completed alongside a blinded fully powered randomized controlled trial, it used an appropriate alternative intervention as 70% of Australian rehabilitation inpatient health services do not offer a weekend physiotherapy service
, and it was reported according to the CHEERS checklist [see Additional file
]. A limitation of this study includes the differing patient length of stay included in the calculation of health related quality of life and the functional status gained, because these measures were taken at admission and discharge from rehabilitation so that the mean change in clinical outcomes was for a different time period for each group. We have reported the mean change per day to address this limitation. In addition the calculation of incremental cost effectiveness ratios accounted for this variability since length of stay is the largest contributor to cost. While inclusion of patients with a cognitive impairment is considered a strength of this study, we are unable to report on the exact numbers of patients in this group. However, on admission to rehabilitation 5% (n = 54 of the 996 participants) did not complete the health related quality of life questionnaire and the main contributing factor was reduced cognition, as identified by the assessors. Other strengths of this study included access to complete clinical cost data on all patients across the two rehabilitation inpatient services, and inclusion of a range of rehabilitation diagnoses and patients with a language other than English as their first language. Therefore, we are confident that the results are generalizable across public acute phase inpatient rehabilitation settings. There were minor variations to the trial protocol. These included the use of multiple imputation rather than the carry forward technique for missing data, consistent with recent recommendations
, as well as a reduced data collection period due to a higher than expected rate of participant recruitment. This study did not include the wider economic impact from a health system perspective during the rehabilitation inpatient admission, as well as the impact on the community once the patients are discharged from rehabilitation including return to work. As this economic evaluation did not use a health system perspective or report on long term economic outcomes post discharge, this warrants future research, which is planned.