The clinical implications and cost-effectiveness of the provision of medical in addition to surgical catheter insertion for peritoneal dialysis in people with kidney failure

Target population and setting

UK Cath was a prospective multi-centre cohort study of incident PD patients at first catheter insertion conducted in 44 of the 72 dialysis centres in the UK, 20 of which were simultaneously participating in the Peritoneal Dialysis Outcomes and Practice Patterns Study (PDOPPS).^9,16 The PDOPPS is an international prospective cohort study designed to identify optimal practices for people treated with maintenance PD. The UK Cath protocol was published in 2017, prior to study closure and analysis. ⁹ Individuals aged 18 or older with irreversible kidney failure who intended to initiate PD as their initial treatment modality and had plans for PD catheter insertion within the following 30 days were given the opportunity to participate in the study by providing consent. Exclusions were made for patients who had already initiated PD, had their PD catheter inserted at a different healthcare facility, or were incapable of providing informed consent. Recruitment began in July 2015 and concluded at the end of December 2017, with data collection ending in April 2019. All subjects received a patient information sheet and gave written consent. Patient partners were involved at each stage of the research process including the design of patient-facing questionnaires.

Cost-effectiveness model

We constructed a decision model to evaluate the costs and health benefits associated with two distinct service approaches for providing PD. One approach (dual pathway) involves two possible PD catheter insertion techniques: medical (requiring percutaneous, radiological, peritoneoscopic techniques) or surgical (requiring open surgical, laparoscopic techniques) while the other approach (surgical pathway) exclusively insert PD catheters surgically. Health benefits are expressed in terms of Quality Adjusted Life Years (QALYs). Costs focus on those that are incurred by the health care system.

The model was developed in Simul8 (Simul8 Corporation, Boston, US) and is summarised in Figure 1. A summary of the key features is provided here with a full account of assumptions made provided in Supplemental Table 1. Patients, representative of those observed in UK Cath, are tracked, one at a time, through the simplified treatment pathway represented by the model. The characteristics of each patient on entering the model and their subsequent treatment experience impact the route and time taken through the various health states the model describes. Patients’ characteristics are age (in 5-year increments from 20 to 90), sex and comorbidities (diabetes, atherosclerotic cardiovascular disease, other cardiovascular disease and cancer). Each patient is simulated having their PD catheter inserted in a service with a dual pathway and again in the service which offers only surgical insertion of PD catheters. A portion of patients (31%) undergo surgical catheter insertion in the dual-pathway simulation. The outcomes for an individual receiving a surgically inserted catheter in the dual-pathway simulation are identical to the outcome for that individual in the simulation of the service exclusively offering surgical insertion for PD catheters. A large number of patients are tracked over a lifetime and the mean costs and health benefits then calculated. The characteristics that inform each patient's pathway through model health states are sampled from incident dialysis patients in the renal registry.

OPEN IN VIEWER

Figure 1.

Health states and possible transitions in the health economic model. All transitions informed by UK Cath study data, with the exception of those into death (informed by age/sex life-tables from USRDS data).

The model first assigns the sampled patient to a PD catheter insertion modality, conditional on the service type, with the patient in the surgical pathway always receiving their PD catheter surgically. At the time of PD catheter insertion, the model then checks if there will be a catheter failure or a catheter-related event in the first 6 weeks based on the probability of failure in the study data. If there is a catheter event, the time at which it occurred is set based on the mean time to a catheter event in the study data conditional on patient characteristics. The model makes a distinction between catheter-related event occurring in the first 6 weeks (primary failure) and those occurring after 6 weeks (secondary failure) to allow for any differences in types of events or the management of events in the early compared to later post-catheter insertion period (Supplemental Table 1).

Patients who do not encounter primary catheter failure remain on PD until they experience a catheter event beyond 6 weeks, transition to haemodialysis (HD) without experiencing a catheter event, receive a kidney transplant or die. Informed by the study data, transition to HD also occurs if the patient experiences three catheter failures, three instances of PD-associated peritonitis (PD peritonitis), five exit-site infections requiring antibiotics or three hospitalisations due to complications (informed by clinical opinion). Once patients initiate HD, they cannot move back to PD, therefore, any subsequent catheter events (failure or catheter complications) will lead to reestablishing the patient on haemodialysis. The UK Cath Study did not provide a means to reliably estimate this specific probability. All catheter reinsertions are performed surgically. Catheter-related events incur both health service costs and reductions in quality of life, based on the UK Cath study data.

The times at which each individual patient will receive a transplant (if at all), or die, are both calculated at the beginning of the model. For those patients that receive a transplant, time to mortality is adjusted to reflect its mortality benefit. The times for entry into these health states (catheter failure, death, PD, HD, kidney transplant) were estimated using parametric survival models including patient characteristics as explanatory variable. The time to catheter events is also a function of the insertion technique (medical or surgical). Separate models were fitted for the duration of PD following second and subsequent catheters, reflecting that these catheters have worse outcomes. We present adjusted Cox model results using the same predictors as our previously published analysis of UK Cath, to allow for a direct comparison between analyses.

Health utilities were assigned to specific health states. The SF-12, patient-reported outcome, was administered as part of UK Cath. Patients in different modality states: PD, HD, transplant (received within one year) and transplant (received one year or longer) – responded to this quality-of-life instrument. These states align with the health states in the model and their responses were transformed into EQ5D-3L utility scores. ¹⁷ Both costs and benefits were discounted at a 3.5% annual rate. These costs are calculated for each patient based on the average annual cost of each event state, adjusted for the duration of an event. NHS Reference Costs were used as the source for the cost of dialysis treatment, hospitalisation for catheter events and catheter insertion costs, stratified based on whether the insertion was medical or surgical. NHS reference costs are published by the NHS and represent the average expenses incurred by the NHS for specific services within a fiscal year. This encompasses all aspects of NHS expenditure associated with service delivery, covering direct treatment costs, facility overheads, staffing expenses and patients’ hospital stays. It does not include patient-borne costs. Therefore, from the cost perspective, we consider only the expenses borne by the health service and exclude patient out-of-pocket expenditures and indirect costs, such as loss of earnings when travelling to a hospital appointment. When unit costs were not available in the 2018/2019 NHS reference costs, all expenses were adjusted to 2018/2019 prices using the NHS Cost Inflation Index Pay and Prices from the Unit Costs of Health and Social Care 2020 publication. ¹⁸ A detailed list of parameters, assumptions and information sources can be found in Supplemental Table 2. All events have EQ5D-3L utility scores, with sex and age also influencing health utility.

Uncertainty in the inputs to the model is reflected using standard methods. Probabilistic sensitivity analysis is used to demonstrate the impact of the combined uncertainty in model inputs. We present the probability that each service delivery type is cost-effective, across a range of values for a QALY that decision makers may find acceptable. The graphical representation is called the Cost-Effectiveness Acceptability Curve (CEAC). We also conducted a series of sensitivity analyses, varying inputs within their 95% CIs.

Duration of treatments and the patient journal through the health states, by catheter insertion pathway

In services offering a dual insertion pathway, 69% of catheters were inserted medically and 31% surgically. Due to the lower catheter event rate associated with the medical catheter insertion technique, the patient journey though the model and the associated number of events and duration of treatment modalities differed between catheter insertion pathways (Table 2). Dual pathways (medical in addition to surgical) had fewer PD complications and initiations on HD, and longer time on PD (in those transplanted and those not transplanted).

OPEN IN VIEWER

Table 2.

Differences in duration of modality/state, costs and quality-adjusted life years between surgical and dual catheter insertion pathways.

		Surgical	Dual	Difference
Mean number of events	Stable PD initiations	2.12	2.03	0.1
	PD without complications	0.16	0.19	−0.03
	PD complications	3.34	3.18	0.16
	HD initiations	0.28	0.25	0.03
	Transplants	0.61	0.61	0
	Age at transplant (years)	53.11	53.17	−0.06
Mean duration (years) in each health state	Time in model	16.53	16.7	−0.21
	Post-transplant	21.0	21.3	−0.31
	Time in model – No transplant	5.30	5.33	−0.03
	PD	3.00	3.07	−0.08
	HD	0.76	0.69	0.07
	PD – No transplant	4.07	4.25	−0.19
	HD – No transplant	1.23	1.08	0.16
Constituent costs (£)	Insertion	3184	2493	690
	Dialysis	84,110	83,839	271
	HD transport	2576	2361	215
	Hospital care and mortality	31,271	31,334	−64
	Transplant	94,250	95,365	−1115
	Pharmaceuticals	8392	8377	15
	Complications	2981	2778	203
Cost–utility (costs, £ and QALYs)	Costs (£)	226,764	226,549	£214
	QALYs	6.995	7.119	−0.124
	ICER			Dual dominates

Cost-effectiveness of medical in addition to surgical catheter insertion (dual pathway) compared to surgical-only catheter insertion (surgical pathway)

Results, reported in Table 2, show that, in the base case analysis the dual (medical and surgical) catheter pathway set-up dominates the surgical approach (i.e., is both cost-saving and QALY gaining). The availability of both insertion techniques results in an average gain of 0.12 QALYs over a lifetime and marginally lower mean lifetime costs of £214. This difference is approximately 0.1% of the mean overall costs borne by patients throughout their kidney care journey. More detail on the source of these cost differences is provided in the constituent costs in Table 2.

The CEAC is shown in Supplemental Figure 1 for values of the maximum acceptable cost per QALY gained between £0 and £50,000. The CEAC reveals that when the maximum willingness to pay for a QALY reaches or exceeds £10,000, the probability of the dual pathway option being cost-effective surpasses 99%. The dual pathway can be considered cost effective at all conventional decision maker thresholds for the value of a QALY.

There is a 60% probability that services offering only surgical insertion are the least costly option. However, there is little uncertainty that services offering dual catheter insertion pathway generate greater health gains. Varying parameters within the ranges of their upper and lower 95% CIs in a series of one-way sensitivity analyses did not change these general conclusions.