Several authors have made comparisons of rehabilitation length of stay (LOS) between countries or across other jurisdictions (Burke et al. 1985; Muslumanoglu et al. 1997; Pagliacci et al. 2003; Chan & Chan 2005). Additionally, others have noted the trend for progressively shorter LOS over the past several decades, especially in the US (DeVivo 2007; DeVivo et al. 1991; Morrison 1999; Eastwood et al. 1999) although there is also data from Israel that shows this as well (Ronen et al. 2004). Stover noted that reductions in the 1970s and early 1980s were likely due to increased efficiency of rehabilitation teams (Stover 1995). More recent reductions in the US have been attributed to restrictions imposed by payers (Morrison 1999). Table 2 summarizes various reports in the literature for LOS organized by jurisdiction and also by the time period for which the data was collected. Data were only included in this table if the underlying sample was deemed representative of an overall heterogeneous population of individuals with SCI (i.e., unselected sample of a single or multi-centre study). Some data was included and grouped for evaluating specific issues and this has been appropriately indicated. In addition, data from studies for which it was not clear that the purpose of admission was for comprehensive inpatient rehabilitation (and may have involved acute care) were not included.
Rehabilitation LOS is also known to vary according to neurological status and data from studies reporting LOS organized by level of injury (i.e., paraplegia vs. tetraplegia) or completeness are shown in Table 3. Again this is organized by jurisdiction (country) and the time period over which the sample was analyzed.
As seen in Tables 2 and 3, rehabilitation LOS varies widely from country to country. While no investigators have systematically analyzed country-by-country variation it is apparent that the US has typically shorter rehabilitation LOS times than other countries reporting data. Most data has originated in the US, bolstered by the development of the US model systems database, with reports from other countries for the most part limited to a handful of descriptions of single-centre experience.
Within the US, it is clear that the trend for progressively shorter rehabilitation LOS has continued to 2009. Across 5 separate reports, the National SCI Statistical Centre (2005, 2009), DeVivo (2007), Morrison (1999) and Eastwood et al. (1999) indicated reduced LOS from the period between 1973 to 2006. Eastwood et al. (1999) examined the large US Model systems database of individuals with traumatic SCI (N=3,904) and reported annual mean LOS values from 1990 to 1997. For these years, the highest value was 80.9 days in 1992 and the lowest was 54.3 days in 1996. Mean LOS values for 1990-1992 seemed fairly stable at higher values, with 1994-1997 values lower and 1993 at an intermediate value. DeVivo (2007) has reported on the same dataset over a longer period of time beginning in 1973 (N=24,333), to extend the trend to a LOS of 45 days in 2006. Morrison (1999) performed a direct comparison of 1991 vs 1995 mean LOS values in the largest SCI rehabilitation in the US in order to assess the effect of shorter rehabilitation LOS on functional outcomes. These authors confirmed an even more striking difference between these 2 years given an average LOS of 95.8 days in 1991 as compared to 54.2 days in 1995 (p<0.001). Other reports have described reductions over earlier periods, most notably multi-centre investigations associated with the US Model Systems databases (De Vivo et al. 1991). These same trends are apparent by looking at the public data available from the US National SCI Statistical Centre (NSCISC 2005, 2009).
It is uncertain if the same patterns have been seen in non-Model System centres or in other countries, although it is clear from a single-centre report from Israel analyzing LOS decade by decade that significantly lower LOS was seen beginning in 1996 as compared to earlier time periods (Ronen et al. 2004). Data from this report and also reports from other countries (Tooth et al. 2003, Burke et al. 1985, Australia; Chan & Chan 2005, China; Pagliacci et al. 2003, Italy; Sumida et al. 2001, Japan; Schonherr et al. 1999, Netherlands) indicated LOS remains significantly longer than reported in US data. LOS data from these reports is displayed over time and across several countries in Figure 1 (Note: US Model Systems data for this figure is that reported in the National SCI Statistical Centre 2005 Annual Report).
A low-cost, low intensity, outpatient rehabilitation program is reported by a Columbian group (Lugo et al 2007; N=42) where in-patient rehabilitation was shortened to an average of 13.5 days and augmented with 18 month, interdisciplinary out-patient rehabilitation follow-up. This low cost intervention achieved adequate functional goals, although these were achieved over a longer period of time due to the lack of accessibility to continuous and intensive therapy. This report might inform payer-directed LOS reduction efforts which may be driven by a focus on costs and may not necessarily circumvent any consequences associated with reductions to LOS by an increased attention to outpatient services.
Also apparent from Table 3 is the relationship of longer LOS associated with higher level of injury and greater severity of injury. Similar patterns were seen in all studies describing rehabilitation LOS for individuals with varying injuries. That is, the greatest mean rehabilitation LOS values were seen for those with complete tetraplegia (especially high level) whereas the shortest mean values occurred for those with incomplete paraplegia (DeVivo et al. 1990; Tooth et al. 2003; Chan & Chan 2005) although this relationship of level and injury severity was only a non-significant trend in the data from Israel (Ronen et al. 2004).
There is level 3 evidence (with predominately US data) that rehabilitation LOS has become progressively shorter during the period of 1973 to 2006. For other countries, only investigators from Israel have published data in a single report that is consistent with this trend.
There is level 3 evidence that those with higher level and more severe injuries have longer rehabilitation LOS.