Cost-effectiveness analysis of liver transplantation in biliary atresia according to the severity of end-stage liver disease

Biliary atresia (BA) is the most common cause of neonatal cholestasis, which could lead to end-stage liver disease (ESLD) in children [1]. BA Children with delayed diagnosis or failed hepatic portoenterostomy (HPE) generally develop several complications of ESLD [2], requiring both emergency and long-term management, utilizing public health care resources, and having financial burdens to the public health system.

Liver transplantation (LT) is a life-saving procedure with a high successful rate in BA with ESLD [3]. The Thai Pediatric LT registry has reported a 90-day and 1-year survival rates of 93.7% and 90.5%, respectively [4]. Although children who underwent LT have a satisfactory health-related quality of life, they need lifelong immunosuppressive therapy [5]. Some LT children may suffer from post-LT complications, including graft rejection, vascular and bile duct complications, opportunistic infection, and adverse effects from immunosuppressive therapy [3, 6].

Common indications for LT are liver deterioration, particularly decompensated cirrhosis, and complications of ESLD [7‐9]. Finding an optimal timing to perform LT can be challenging. ‘Delayed’ LT can affect children's health, as they may have serious sequelae of the disease or die before the operation. In contrast, children who undergo ‘early’ LT could receive a longer duration of high-cost immunosuppressive therapy and a greater risk of its adverse effects.

The United Network for Organ Sharing (UNOS) has developed Pediatric End-stage Liver Disease (PELD) score to prioritize pre-LT candidates aged < 12 years by using the following data: age, growth status, serum albumin, total bilirubin and international normalized ratio [10]. Patients with a higher PELD score have a greater rate of pre-LT mortality [11]. According to our preliminary study, which has not been published, PELD score ≥ 15 was associated with an increase of pre-LT mortality, similar to a finding from de Vries et al. [12] Moreover, Arnon et al. [13] have found that LT for moderate liver disease has increased expected life-years and survival rates in BA children when compared to LT for severe disease. Hence, early registration before the severe disease may have some benefits among these children. Unfortunately, the cut-off value of PELD score for LT registration varies among health care institutions [11, 14].

LT is associated with high health care expenditures [15]. A previous study has mentioned that a higher PELD score is related to greater health care cost in the first year after LT [16]. To our knowledge, no cost-effectiveness analysis has been performed to evaluate an optimal disease severity for LT registration. Even though most of health care expenditures for pediatric LT can be reimbursed from Thailand’s Universal Coverage Scheme (UCS) [4], many LT centers must absorb any cost which exceeds the government’s reference price. From a transplant center perspective, we aimed to study the cost-effectiveness of LT which could prevent all-cause death among BA children registered on the waiting list with different severity of ESLD. The secondary objective was to estimate the break-even point of LT among children on the waiting list with different severity of ESLD.

According to our findings, initial registration for LT at PELD score < 15 had lower all-cause mortality and less total direct medical cost when compared to registration at a higher PELD score, suggesting that early registration was more cost-effective to prevent death in BA children with ESLD. Certain direct medical cost was different among children registered with the low and with the high PELD scores in each period of LT. Moreover, the cumulative cost of initial registration with the low PELD score was not only less at the beginning, but also over a 5-year period.

Previous studies have examined the relationship between PELD score and mortality in the pre-LT and post-LT periods. According to a large study in children with chronic liver disease from the UNOS [11], PELD score is related to 90-day pre-LT mortality, similar to our findings. While a study reported that children underwent LT at a lower PELD score have less mortality after LT [18], the association is not significant according to Bourdeaux et al. [19], comparable to our finding in the post-LT period. Nevertheless, the number of post-LT death was quite low in this study, which could limit the robustness of statistical analysis.

Since children registered with the low PELD score had less all-cause mortality which was calculated from both pre- and post-LT periods, early registration for LT may be beneficial in BA children with ESLD. As the majority of our children already had high PELD score at the referral and thus were initially registered with the high score, this may emphasize the necessity of earlier referral in BA children, which should be addressed in those with inadequate clearance of jaundice at the age of 6 to 9 months [7].

This study revealed that the total direct cost was less in the low PELD score group. In the pre-LT period, the cost during IPD admission was also significantly lower in the low PELD score group, although the duration of IPD admission was not statistically different. Unfortunately, we had limited information regarding the details of the cost in this period. We also observed that the cost in the OPD was higher in the group with the low PELD score, which may be contributed by a longer duration of follow-up through the OPD visits. Patients in the low PELD score group were older at LT in our study, probably due to the slower progression of the disease. A larger proportion of children in this group underwent HPE, which could delay the necessity for LT in several reports [20‐22], even in patients who underwent the operation at age > 60 days [23]. The waiting time for LT in the low PELD score group was longer than the high PELD score group. This was due to the policy of our center to prioritize pre-LT evaluation for donors of patients with a very high PELD score (> 20) at the registration, as well as the LT operation, to minimize the waiting list mortality.

The cost of LT is usually peak within 3 months after LT [24], which was defined as the LT-period in this study. From our tornado diagram, the cost of the high PELD score group was the most sensitive parameters affecting the ICER. Our study found that the low PELD score group had less cost in this period (730,398 THB vs 863,488 THB). The cost in both groups was higher than the National Health Security Office’s reference price that the hospital could reimburse under UCS, which was the major health insurance scheme in this study (almost 95% of the participants). Currently, the reference price is approximately 470,000–625,000 THB during this period, consisting of 410,000–565,000 THB for the admission during the transplant operation and 60,000 THB for the outpatient care within 3 months after LT. Nevertheless, the cost of LT in this study was lower than the reported cost in North America [24, 25].

Associated factors with the cost of LT have been reported in previous studies. The cost is known to be correlated with the duration of admission during the LT operation and peri-operative care [26], which was significantly less among the low PELD score group in this study. While one study described that living donor LT had more cost within 1 year after LT when compared to deceased donor LT [16], the difference is not demonstrated in another study [26]. Nevertheless, the donor types were not significantly different between the groups of our children.

In post-LT period, the total direct medical cost was peak during the first year after surgery and gradually decreased, similar to previous studies in the US [24] and Canada [25]. Drug cost was the highest among all categories in this study which were mostly contributed by immunosuppressive therapy. According to the study from the US [24], the cost of drugs is the second highest in the OPD visits, followed the cost of laboratory and radiologic investigations. However, the combination of cost in the investigations was less than the cost of drugs in this study. Moreover, we found that the high PELD score group had higher cost of investigation from 90 days to 2 years after LT. Therefore, avoiding over-investigation may be considered in these children.

This study has some limitations. First, the direct medical cost was only retrieved from a single LT center and did not include the cost from other hospitals. Additionally, we did not study indirect cost and quality of life. The application of PELD score also have some limitations in infants [14]; some children may have ESLD-related complications requiring LT, regardless of the PELD score [7]. Nevertheless, the numbers of children with the PELD score-exception conditions were not different between both groups. Moreover, the study was conducted over a long period of time, which details of management may change, although the main protocol for LT was consistent during the study period. In addition, we did not explore direct non-medical cost which could be associated with the number of visits and duration from registration to LT.

Initial registration for LT at PELD score < 15 was more cost-effective to prevent the mortality among BA children with ESLD. The cumulative cost of initial registration at the low PELD score was not only less at the beginning, but also over time. For this reason, early referral and registration for LT should be emphasized to prevent death in BA children with ESLD.