Testing the effect of discharge destination on outcomes for people with isolated lower limb fractures

Some patients may not benefit from inpatient rehabilitation, but numerous factors must be considered

In this issue of the MJA, Kimmel and her co‐authors report analysing data from the Victorian Orthopaedic Trauma Outcomes Registry (VOTOR)1 with the aim of determining whether inpatient rehabilitation after isolated lower extremity fracture in working age people might be associated with poorer long term outcomes. For those of us in the rehabilitation services, this investigation further develops a familiar theme, doubt about the value of inpatient rehabilitation services.

The associations observed by Kimmel and colleagues between inpatient rehabilitation and poorer outcomes are strong: the odds of recovery or return to work were reduced by roughly half for people receiving inpatient rehabilitation. Further, the authors note that a randomised controlled study had previously found no evidence for benefit from inpatient rehabilitation after knee arthroplasty.2

There are hypothetical reasons why inpatient rehabilitation may not be beneficial, or even harmful, for some patients. It is recognised that patients in rehabilitation wards are generally inactive for most of the day3 and that the ward environment facilitates rest in bed. The amount of therapy promoting mobility provided each day is limited. In this environment, infection with multiresistant organisms and unnecessary administration of opioid medications are possible; sleep may be disturbed, and there is risk of injury from falls. Inpatient rehabilitation may encourage dependence, prompt passivity, and isolate people from their usual social support. The adverse effects of inpatient rehabilitation have not been examined in detail, but a framework for their investigation has been proposed.4

It is certainly conceivable that patients with lower extremity fractures might be less active, and therefore recover less well, in hospital than at home. However, the data reported by Kimmel and colleagues are from an observational study, and their conclusion is based upon an association. This relationship could be affected by a number of confounding factors. The authors have applied propensity scoring techniques to reduce this problem, primarily correcting imbalances in the data, with the possible exception of compensation status. But, as the authors acknowledge, there may have been unmeasured confounders; most crucially, factors related to socio‐economic disadvantage, psychological reaction to the injury (including catastrophising), unmeasured physiological factors such as obesity, and the differential impact of compensation types.

A clinician considering the findings of Kimmel and her co‐authors would comment that important factors related to socio‐economic disadvantage have not been considered, such as housing instability, food insecurity, transportation needs, utility needs, and interpersonal safety.5 Transfer to an inpatient rehabilitation unit both frees an acute hospital bed and allows time to resolve these problems. In other areas of inpatient rehabilitation, this clearly applies to people with severe acquired disability waiting in these facilities for acceptance and support from the Australian National Disability Insurance Scheme (NDIS).6 Further, deficiencies in developing the NDIS in some states may have meant that rehabilitation for younger people with severe injuries is inadequate in non‐compensable situations, so that such patients have been transferred to residential aged care facilities.7 These examples underline the risk of changing services without appropriate research and policy development.

It could also be argued that the extended Glasgow Outcomes Scale does not adequately measure functional recovery because it was developed for assessing people with neurotrauma. Nevertheless, the scale measures functioning, has face validity, and evidence for its validity outside neurotrauma has been reported.8

Finally, the VOTOR is funded by the Victorian Transport Accident Commission. This funding should perhaps be declared as a competing interest because, as the statutory insurer of third party personal liability for road crashes in Victoria, it has a clear and direct interest in the outcomes of this study.

Propensity scoring in an observational study cannot substitute for a properly conducted clinical trial, particularly for an intervention as complex as inpatient rehabilitation. The findings of an observational study should not change clinical practice. Kimmel and colleagues have acknowledged the potential problems with their analyses and have qualified them appropriately, acknowledging, in particular, the possibility of residual confounding.

The next step is to design a randomised clinical trial for testing whether inpatient rehabilitation is advantageous for a carefully characterised group of people with traumatic lower extremity fractures. The design problems for such an experimental study are well understood, particularly funding, but also health professional barriers and resistance from private hospital management. In Victoria, the Transport Accident Commission and Worksafe Victoria are key stakeholders. However, this is a broader question for all Australia that the Medical Research Future Fund should consider investigating.