Economic evaluation of peritoneal dialysis and hemodialysis in Thai population with End-stage Kidney Disease

We conducted the study using real-practice data across the country to perform a cost-utility analysis of dialysis therapies for ESKD patients in Thailand. Comparing HD with PD in age groups from 20 to 60 years indicated that the ICERs were 19,434 to 24,796 USDs and 31,913 to 39,192 USDs per QALY gained in governmental and societal perspectives, respectively. As a result, the HD was not cost-effective as an initial treatment relative to the PD from both governmental and societal perspectives given the current willingness to pay threshold in Thailand of 4,766 USD per QALY. The results were also validated using a probabilistic sensitivity analysis, and were consistent with the deterministic results. One-way sensitivity analysis revealed that the direct medical cost of HD with chronic complications was the most sensitive parameter for the cost-effectiveness of PD versus HD. These findings have substantial implications regarding optimising the quality of ESKD care and preventing chronic complications, which would have a paramount impact on cost-effectiveness outcomes.

Our findings were similar to previous cost-effectiveness analyses, which found that PD was more cost-effective than HD [10, 25‐29] including the study conducted in Thailand in 2007; although this study found that both PD and HD were not cost-effective when compared with palliative care, while PD was cost-effective relative to HD. In addition, a few studies conducted in South East Asia also found consistent results that PD was more cost-effective than HD from both government [30] and societal perspectives [25, 28]. However, a few studies reported contrary results, which indicated HD was almost as cost-effective as PD [31] or more cost-effective than PD [32]. Those inconsistent findings may be explained by the differences in country context and the source of input parameters used. For decades, the effectiveness of dialysis modalities in terms of patients’ survival has been debated due to the difficulty of conducting randomised control trials [15, 33]. The sources of input parameters regarding patient survival were also varied in the economic evaluation, which may have impacted the outcome. Some cost-effectiveness analyses used survival input parameters from a single dialysis modality and assumed that HD and PD had similar survival outcomes [10, 25], while others analysed data from national registry databases and used distinct input parameters [28, 29, 31]. A prior economic evaluation of kidney replacement therapy in Thailand that compared HD and PD to palliative care assumed that PD and HD had equivalent survival outcomes. The results revealed that the PD strategy dominated the HD strategy when PD survival was assumed to be equal to HD survival, which was obtained from the Thai Renal Replacement Therapy Registry⁷. Our study used transitional probabilities derived from real-practice data of 10,252 adult ESKD patients who initiated dialysis with PD or HD between 2008 and 2018 in 15 hospitals located throughout Thailand. We also used a Weibull regression with the adjustment of covariables (i.e., age, gender, reimbursement, region, type of hospital, diabetes, hypertension, and chronic complications related to chronic kidney disease (i.e., cardiovascular disease and chronic kidney disease-related mineral and bone disorders)) [34] that could affect patients’ survival indicating HD gained more life years and quality-adjusted life years than PD. We were concerned that the dialysis modality selection was not randomised due to reimbursement policy and doctor/patient preferences, which could result in selection bias and invalid findings. Therefore, we estimated treatment effects using the propensity score method with inverse probability weighting regression adjustment to minimise selection bias. Interestingly, despite real-practice data in our country showing that HD results in a higher QALY gain than PD due to better patient survival and utility, as shown in Table 2, PD has remained more cost-effective than HD due to PD’s significantly lower lifetime treatment cost.

Our analysis demonstrated that the cost-effectiveness acceptability curve would shift to HD is more cost-effective than PD when willingness to pay of Thailand reaches 25,320 USD per QALY, i.e., approximately five times higher than the current Thailand’s willingness to pay and 3.5 times Thailand’s GDP per capita in 2021 at 7,186 USD. Those findings provide information on the cost-utility of ESKD treatments in Thailand, while the decision depends upon the healthcare system’s willingness to absorb incremental costs associated with HD.

The strength of our study was the newly developed Markov model, which included acute complications such as vascular access infection/dysfunction or peritonitis, as well as chronic complications such as cardiovascular disease, cerebrovascular disease, and chronic kidney disease-related mineral and bone disorders in ESKD patients [34]. This model more accurately represented the natural history of diseases than previous studies, which only included dialysis mode transition and death. More updated data across the country were used, including cost, disease progression (i.e., CKD-related chronic complications and mortality) from 15 hospitals, utility data from a cross-sectional survey of 11 hospitals plus the data from Thai Renal Replacement Therapy Registry. Although these data were only in some parts of the whole country, they mostly reflect the current practices and setting of Thailand. Third, the analyses were conducted based on age at dialysis initiation, which is a significant component in the cost-utility analysis outcomes.

Nonetheless, our study had several limitations due to the following factors: first, we did not consider other modalities (e.g., kidney transplantation, automated peritoneal dialysis, and hemodiafiltration) that are currently applied for kidney replacement therapy in ESKD patients in Thailand. However, KT rate was only 3.6% [35] and automated PD accounts for approximately 3% of peritoneal dialysis in Thailand [36]. We also needed to estimate some parameters not included in hospital billing, such as erythropoiesis-stimulating agents and PD solution from Thailand renal replacement therapy registry, in addition to individual-level cost data from hospital-electronic records due to the complexity of ESKD management’s cost structure and the Thai reimbursement system. However, we conducted a one-way sensitivity analysis on all parameters to ensure the robustness of our results.