In our large patient population who underwent major non-cardiac surgery, postoperative complications were observed in more than one third of the cases and increased costs on average by US$11,824 per patient (+75 %). These findings are consistent with those reported by previous and smaller studies. In 1008 patients who underwent general and vascular surgery, Dimick et al. (Dimick et al. 2006) reported a US$10,178 cost difference between patients with and without complications. In a similar surgical population, Boltz et al. (Boltz et al. 2012) showed in 2250 patients that the excess costs were US$6358, US$12,802, and US$42,790 for patients developing 1, 2, 3 or more complications, respectively. In the present study, the occurrence of postoperative complications was also associated with prolonged length of stay (+3 days) and increased hospital readmission rates at 30 days (+5.3 % absolute increase, +44 % relative increase). These findings are consistent with previous reports (Eappen et al. 2013; Lawson et al. 2013) emphasizing the dramatic impact of complications on length of stay and readmission rates. This highlights a relevant savings capacity for major surgical procedures.
According to a recent meta-analysis of 38 randomized controlled trials, GDFT has the potential to decrease postoperative morbidity by 17–29 % (Pearse et al. 2014). In our study population, such a decrease in postoperative morbidity would translate into cost-savings ranging between US$754 and US$1286 per patient. Interestingly, potential savings were highly variable from one surgical procedure to the other (Table 3). Two factors affect savings per patient: the actual morbidity rate (the higher the morbidity rate, the higher the savings when all patients receive GDFT) and the cost of complications (the higher the cost, the higher the savings). The actual morbidity rates ranged from 20.2 % for gastrectomies to 67.7 % for esophagectomies (Table 3), and cost of complications ranged from US$5413 for femur and hip-fracture repair to US$30,542 for esophagectomies (Table 3). This large range in morbidity rates and costs of complications explain why the range of potential savings with GDFT was also wide, from US$354–604/patient for femur and hip-fracture repair to US$3515–5996/patient for esophagectomies (Table 3). A third factor affects savings at the hospital level: the volume of surgeries. When taking into account the volume of procedures, the total potential savings per year were the most significant (US$32–55 million), for colectomies, by far (Fig. 2). This finding supports the notion that GDFT should be implemented as a priority in this surgical population.
Three recent studies tried to estimate potential savings related to the use of GDFT. The first study (Bartha et al. 2012) from Sweden is a decision analytic model where assumptions were made regarding morbidity rates before and after GDFT implementation, as well as on hospital costs. This study focused on elderly hip-fracture patients, and the model estimated a 1882 € (around US$2000) cost reduction per patient with GDFT. The second study from the UK was based on a small population of 122 patients who underwent major surgery (Ebm et al. 2014). Morbidity rates before and after GDFT implementation were real, but assumptions were made regarding hospital costs. This study suggested a cost reduction of £2631 (around US$4000) per patient with GDFT. Differences between US and European healthcare systems and costs, as well as the fact that costs were not real but estimated in the Swedish and the UK study, may explain why they both reported potential savings higher than our projections. In the third study, Manecke et al. (Manecke et al. 2014) used real morbidity rates and real costs extracted from the UHC database, which is a large administrative database containing clinical and economic data from over 120 US academic hospitals. As in our study, the only assumption made was related to the reduction in postoperative morbidity with GDFT. Manecke et al. (Manecke et al. 2014) reported potential cost-savings ranging from US$569 to US$970 per patient, i.e., slightly lower than ours. The UHC database contains only 11 possible postoperative complications and is known to underestimate postoperative morbidity rates (Steinberg et al. 2008). For instance, they did not take into account postoperative paralytic ileus, which is not a major complication but a frequent one (Fig. 1), known to have a significant impact on hospital length of stay and costs (Iyer et al. 2009). Finally, the economic evaluation of Manecke et al. (Manecke et al. 2014) did not include 30-day readmission costs, and was limited to academic hospitals. Our analysis was based on a larger number of patients and considered 26 different postoperative complications (including paralytic ileus). We took into account 30-day readmission costs, and 58.2 % of our study population came from non academic centers. For these reasons, we believe that our study provides a more accurate estimation of potential savings associated with the implementation of GDFT at a national level.
To assess a return of investment, our projected savings must be balanced with costs related to GDFT implementation. Assuming average cardiac output-monitoring-related costs of US$300 per patient (US$250 for disposable sensor + US$48 for the amortization of a US$15,000 monitor used two times a week over 3 years), our findings suggest that for each dollar spent to implement GDFT, hospitals should save in return between US$2.5 and US$4. GDFT implementation costs may vary from one hospital to the other, but each hospital could easily forecast the return on investment using its own costs and our model.
Another major burden of complications is the opportunity cost of lost beds for increased length of stay and re-admissions in patients with complications. In busy hospitals, these are beds not taken by new patients with new DRGs and the accompanying payments. With an increased length of stay of 3 days for patients with complications (Table 2), the 76,807 patients with complications represent 230,421 (3 × 76,807) days lost. With the assumption that GDFT would decrease the number of patients with one or more complications to 54,533–63,750 (a 17 to 29 % decrease), it would now represent only 163,599–191,250 days lost. In other words, GDFT has the potential to save between 39,171 and 66,822 days. With the average hospital length of stay across all studied surgical cohorts being 5 days, the implementation of GDFT could result in 7834–13,364 (39,171 and 66,822 divided by 5) new patients admitted in our 541 hospitals over the 2.5-year study period, or 3134–5346 new patients per year. Since payments to hospitals vary so widely, each hospital can use this approach to calculate the increased payment and profit to the bottom line. This is the lost opportunity cost of having patients with complications take up hospital beds needed for other patients and represents the potential additional profit to the hospital. This may easily outweigh the savings from decreased complications.
Our study has certain limitations that should be considered. The analysis was limited to specific major surgeries in which outcome has already conclusively been shown to be improved by the use of GDFT. There are other surgeries, such as major spine and gynecologic surgeries, in which this approach would likely be associated with fewer complications (Mythen et al. 2012; Gan et al. 2002). Our study assumes complete implementation of GDFT, which may be an unrealistic goal. Also, we considered the same postsurgical morbidity reduction with GDFT for all surgical procedures, which may not always be the case. However, previous meta-analysis (Pearse et al. 2014; Hamilton et al. 2011) did not find any interaction between the type of surgery and the effect of GDFT. The article by Pearse et al. (Pearse et al. 2014) was chosen to estimate the effect of GDFT on postoperative morbidity because it is the most recent meta-analysis on the topic. It is important to note that it included studies published many years ago. Because both anesthesia and surgical practices changed over time, the assumption that GDFT would reduce postoperative morbidity by 17–29 % may be questioned in 2015. A meta-analysis by Hamilton et al. (Hamilton et al. 2011) studied the clinical effects of GDFT over time. If GDFT had no effect on mortality in studies published after 2000, the reduction in postoperative morbidity was still highly significant, ranging between 50 and 71 % (average odds ratio 0.38, CI 0.29 to 0.50). Interestingly, if we had used these odds ratios for our calculations, the projected savings with GDFT would have been much higher than those we have reported. Having said that, we must acknowledge that enhanced recovery programs have gained acceptance only recently, as well as changes in the type of fluid administered during the perioperative period (starches and unbalanced crystalloid solutions are used less often). Therefore, the clinical effects of GDFT in this new perioperative medicine era remain to be evaluated by large studies. Finally, we did not have access to reimbursement data, so we were unable to study the effects of morbidity reduction on hospital profit or profit margin, another very important economic driver for the hospital adoption of any new therapeutic strategy (Dimick et al. 2006; Flynn et al. 2014).