Key results
The majority of the participants were willing to travel a longer time of 75 min to a transplantation center to improve their outcomes. Preferences emerged when improvements in the outcomes at the regional center were not highly distinctive in the DCE (4% and 6%, respectively).
The participants recruited at the hospital seemed to be slightly more quality sensitive than those recruited via registration offices. Out of all characteristics, only the recruitment strategy was found to be associated with decision-making. The proportion of participants recruited at the hospital categorized as quality sensitive was higher than in the subgroup of participants recruited via registration offices. These findings corresponded with the literature [
15‐
21]. Nevertheless, statistically significant influence of the recruitment strategy on decision-making could only be found in the outcome in-hospital mortality rate and not in three-year survival. We did not identify other statistically significant patient characteristics in our multivariate logistic regression models. However, overall model quality was poor.
Analyzing two different outcomes in our study enabled comparison of the participants’ perceptions between in-hospital mortality and 3-year survival. The overall group of participants who were survival rate-sensitive was only 2.22 percentage points higher than the mortality risk-sensitive group. This gradual difference led to the assumption that the overall participants rated mortality risk and 3-year survival rates equally, leading to the assumption that there is no big difference in the rating of short- and long-term mortality outcomes. In the subgroup of participants recruited via registration offices, the difference between mortality rate sensitivity and survival rate sensitivity was more dominant. Similar differences were found in the subgroup of participants in mostly urban and rural areas, with a higher share of participants classified as survival-rate sensitive. This could be a hint, that people outside the context of a medical treatment perceive the improvement of a long-term outcome more important than the improvement of a short-term outcome. Nevertheless, the significance of these findings could not be proven due to the small sample size. Still, the impact of single decisions should be kept in mind.
Generalizability
Comparing our findings with similar studies, differences in healthcare systems of respective countries, differences in applied methods such as recruiting strategies, execution of the DCE, investigated diseases and outcomes, and driving distances to local and regional healthcare centers have to be kept in mind.
Shalowitz et al. asked participants in a DCE to imagine being diagnosed with ovarian cancer [
27]. Overall, 80% of their 60-person sample were willing to travel longer distances to have a 6% higher 5-year survival rate after initial treatment.
Chang et al. [
28] used a DCE to find that 80.06% (
N = 103) of interviewed parents were willing to travel to a regional hospital with a 3% lower mortality rate (vs 6% in a local hospital) when they imagined their child had to undergo heart surgery.
Landau et al. [
29] found via a DCE, in cases of abdominal aortic aneurysm, that 91% (
N = 67) of patients preferred treatment at a regional hospital when mortality risk at a local hospital was higher than at the regional hospital (3% vs 2%).
Finlayson et al. [
23] used a DCE and found that a regional hospital was preferred by the majority (55%,
N = 100) in cases of pancreas cancer treatment when mortality was lower than at a local hospital (3% vs 6%). Applied regression models to identify factors influencing decision-making found older age and fewer years of formal education were associated with preferences for local hospitals with worse outcomes.
In a previous study, we used the same population and methods and investigated patient preferences for elective total knee arthroplasty between driving distance and better outcomes (lower mortality risks and lower revision rates). Overall, 71.7% and 86.11%, respectively, were willing to accept longer travel times to a hospital to have lower mortality and revision rates, respectively. Lower school qualifications were identified as being associated with preferences for local treatment [
30].
As summarized by Bühn et al., there was a mutual trend in all known studies analyzing the trade-off between shorter travel times and lower medical risks of surgical treatment. Participants tended to accept longer distances or travel times in order to lower surgical risks of their treatment. However, decision-making seems to be not only be determined by rational reasons such as information about outcomes and distance to the hospital. In Finlayson et al.’s study, the share of patients preferring local treatment was rather high when outcomes were worse than in the regional hospital. Even when the risk of dying in the local hospital was 100%, 10% (
n = 10) still preferred local treatment [
23].
In some studies, the proportions of participants choosing regional treatment when outcomes in local and regional hospitals were equal were rather high (Landau et al. 40%, Chang et al. 17.5%, and Shalowitz et al. 32.00%), except for Finlayson et al. (0%) and Burkamp et al. (0–1%). Bühn et al. concluded that factors other than medical outcomes and distance to hospital also may influence decision-making [
11]. The regression analysis findings by Finlayson et al. and Burkamp et al. were inconsistent [
23,
30]. Our recent results confirmed the general trend that better outcomes are preferred over shorter travel times. When comparing shares of patients accepting longer travel times, sample sizes should be taken into account. Most studies had small sample sizes, increasing the impact of a single decision.
Limitations
Certain sample sizes (
n ≥ 25) are required to analyze decision-making in relevant subgroups with logistic regression [
26]. Relevant sample sizes are too small to perform further logistic regression models.
Quality improvement steps in each iteration in the DCE are rather large and might not represent real quality improvements in treatment when choosing a distant center. Nevertheless, Nijboer et al. found that the mean in-house mortality in case of liver transplantation in Germany between 2007 and 2010 was 17.6% with a range between 0 and 71.4%. The mean 3-year survival was reported with a mean of 66.0% and a range from 0 to 100% [
31]. The wide range in outcomes confirms significant improvement in outcomes are possible when the transplantation center can be chosen.
Travel times were selected because the study population also underwent a DCE for outcomes of elective total knee arthroplasty [
30]. Total knee arthroplasty is a far more common procedure than liver transplantation. A hospital performing total knee arthroplasty can be reached by shorter driving times. The time setting chosen represents a compromise in order to enable analysis of both issues with one survey and dataset. Further, since the participants are not actual patients that will undergo a liver transplantation, our results might not represent the decision-making of actual patients for liver transplantation.
In Germany, minimum volume thresholds did not promote further centralization of the distribution of liver transplantation centers [
32]. Another limitation is that most participants lived in urban areas, therefore generalizability of the answers, e.g., for people in rural areas is limited. Because of drop out we recruited 107 instead of 90 participants in the hospital. The hospital was in an urban area. Eighty-nine (83.2%) of the participants recruited in the hospital lived in an urban area. Because of that overall the most participants (62.2%) lived in urban areas.
Our results showed that the majority of the participants were willing to trade short travel times for better outcomes. However, it remains uncertain how supply structure changes when thresholds are raised, and a travel time of more than 100 min becomes necessary. It is unclear if a travel time of more than 100 min to a center with better outcomes is tolerated in the same expression. Chang et al.’s findings promoted a lower preference for better treatment outcomes when the travel time to the regional center doubled [
28]. This implies that better outcomes are not preferred unconditionally over travel times. Regarding the performance of transplantation centers above and below the minimum volume thresholds, Nimptsch et al. found that differences in mortality rates could not be affirmed [
33]. Using linear regression, Nijboer et al. found that a higher 1-year overall survival correlated with a higher number of transplantations. They could not affirm this for in-house mortality and 3-year overall survival [
31]. Based on the mixed findings and lack of further studies, it is uncertain if outcomes such as mortality rate and 3-year survival improve when centralization is extended by raising the minimum volume threshold.