An Economic Evaluation of rt-PA Locking Solution in Dialysis Catheters

Abstract

In a recent randomized trial, weekly recombinant tissue plasminogen activator (rt-PA), 1 mg per lumen, once per week, and twice-weekly heparin as a locking solution (rt-PA/heparin) resulted in lower risks of hemodialysis catheter malfunction and catheter-related bacteremia compared with thrice-weekly heparin (heparin alone). We collected detailed costs within this trial to determine how choice of locking solution would affect overall health care costs, including the cost of locking solutions and all other relevant medical costs over the course of the 6-month trial. Nonparametric bootstrap estimates were used to derive 95% confidence intervals (CIs) and mean cost differences between the treatment groups. The cost of the locking solution was higher in patients receiving rt-PA/heparin, but this was partially offset by lower costs for managing complications. Overall, the difference in unadjusted mean cost for managing patients with rt-PA/heparin versus heparin alone was Can$323 (95% CI, −$935 to $1581; P=0.62). When the costs were extrapolated over a 1-year time horizon using decision analysis, assuming ongoing rt-PA effectiveness, the overall costs of the strategies were similar. This finding was sensitive to plausible variation in the frequency and cost of managing patients with catheter-related bacteremia, and whether the benefit of rt-PA on catheter-related bacteremia was maintained in the long term. In summary, we noted no significant difference in the mean overall cost of an rt-PA/heparin strategy as a locking solution for catheters compared with thrice-weekly heparin. Cost savings due to a lower risk of hospitalization for catheter-related bacteremia partially offset the increased cost of rt-PA.

An increasing number of patients undergoing hemodialysis are managed with permanent hemodialysis catheters.^1,2 In North America, approximately 80% of patients initiate hemodialysis with a catheter, and 20%–60% of prevalent hemodialysis patients receive dialysis with a catheter.^3,4 Most hemodialysis catheters fail within 1 year of insertion; up to two thirds of these failures are due to thrombosis, and most of the remainder are removed because of catheter-related bacteremia.^5,6 The need for catheter thrombolysis, catheter replacement, and management of catheter-related bacteremia contributes to significant costs.⁷ As such, hemodialysis patients managed with a catheter have significantly higher costs than those who have a fistula.^8,9

Given these adverse outcomes and high costs, further research on strategies to optimize the use of hemodialysis catheters is required, particularly with regard to preventing the most common adverse catheter outcomes of malfunction (i.e., inability to initiate or continue dialysis because of to inadequate bloodflow, often the result of catheter thrombosis) and bacteremia.^5,6 Heparin has traditionally been used as an intradialytic catheter locking solution, although other agents are now used, including citrate^10,11 and recombinant tissue plasminogen activator (rt-PA).¹² Intradialytic locking with a thrombolytic agent has many appealing features. For example, it may reduce intraluminal thrombosis and improve catheter function while also eliminating the nidus for biofilm formation and infection.

We recently completed a randomized clinical trial (the Prevention of Dialysis Catheter Malfunction with Recombinant Tissue Plasminogen Activator [PreCLOT]) in hemodialysis patients with a newly inserted tunnelled central venous catheter. The study compared rt-PA once per week and twice-weekly heparin as a locking solution (rt-PA/heparin) with thrice-weekly heparin (heparin alone) as the catheter-locking solution.¹³ During the 6-month study period, the catheter malfunctioned in 40 (34.8%) patients assigned to heparin alone and 22 (20.0%) patients assigned to rt-PA/heparin (hazard ratio, 1.91; 95% confidence interval [95% CI], 1.13 to 3.22; P=0.02), and catheter-related bacteremia occurred in 16 (13.9%) patients assigned to heparin alone and 5 (4.6%) assigned to rt-PA/heparin (hazard ratio, 3.30; 95% CI, 1.18 to 9.22; P=0.02).

Although rt-PA is significantly more expensive than heparin as a locking solution, the consequences of catheter malfunction and catheter-related bacteremia must be considered because they may contribute to substantial costs. We a priori identified the need to collect costs throughout the trial,¹² enabling a detailed comparison of the overall health care costs of rt-PA/heparin versus those of heparin alone. Because the treatment period within the clinical trial was only 6 months in duration, we also created a decision analytic model to estimate the costs and clinical effectiveness of using rt-PA/heparin versus heparin alone as the locking solution over a 1-year period.

Results

Costing Analysis Alongside the Clinical Trial

Baseline patient characteristics are reported in Table 1. For most patients in both the rt-PA/heparin and heparin-only groups, this was their first hemodialysis catheter (67% and 70%, respectively). The mean age in the rt-PA group was slightly lower (61.6 years versus 64.8 for heparin; P=0.13). The total mean duration of patient follow-up for assessment of costs (including ongoing follow-up after clinical outcomes) was 133.0 and 126.2 days in the rt-PA/heparin and heparin groups, respectively.

Table 1.

Baseline characteristics of the PreCLOT study patients

Characteristic	rt-PA (n=110)	Heparin Only (n=115)
Age (yr)	61.6±16.6	64.8±15.2
Women, n (%)	39 (35.5)	49 (42.6)
First catheter ever, n (%)	67 (60.9)	70 (60.9)
Indication for current catheter, n (%)
Catheter-related bacteremia	4 (3.6)	10 (8.7)
Failed access, or awaiting arteriovenous access	21 (19.1)	27 (23.5)
Starting dialysis without arteriovenous access	60 (54.5)	58 (50.4)
Transfer from peritoneal dialysis or transplant	13 (11.8)	8 (7.0)
Other	12 (11.0)	12 (10.4)
Duration of hemodialysis (yr)
Median	0.5	1.0
Interquartile range	0–1.0	0–6.0
Comorbid illnesses, n (%)
Diabetes mellitus	60 (54.5)	64 (55.7)
Ischemic heart disease	26 (23.6)	21 (18.3)
Congestive heart failure	28 (25.5)	22 (19.1)
Cerebral vascular disease	12 (10.9)	17 (14.8)
Hypertension	104 (94.5)	102 (88.7)
Prior pulmonary embolism or DVT	5 (4.5)	6 (5.2)
Prior gastrointestinal bleed	12 (10.9)	9 (7.8)

Values expressed with a plus/minus sign are the mean±SD. DVT, deep venous thrombosis.

The unadjusted costs per patient, by treatment groups, over the 6-month study are displayed in the Supplemental Material. The mean total cost for managing patients with rt-PA/heparin and heparin were Can$1872 (95% CI, Can$1142 to Can$2601) and Can$1538 (95% CI, Can$503 to Can$2573), with no difference in the mean costs for the rt-PA/heparin and heparin groups (Can$323; 95% CI, Can$−935 to Can$1581; P=0.62) (Table 2). In a secondary analysis including the cost of bleeding complications, the mean total cost for managing patients with rt-PA/heparin and heparin were Can$2015 (95% CI, Can$1268 to Can$2763) and Can$1872 (95% CI, Can$803 to Can$2941), with the mean difference in cost similar for the rt-PA/heparin and heparin groups ($134 ;95% CI, Can$−2346 to Can$2286; P=0.66). To assess the robustness of the bootstrapping method for comparing costs, we also used generalized linear regression and a Gaussian linear model to compare costs, noting a similar difference in costs between rt-PA/heparin and heparin (Can$334; 95% CI, Can$−929 to Can$1597; P=0.60). Acknowledging that the price of rt-PA is higher in the United States ($107 per 2 mg compared with Can$64 in Canada, and that hospitalizations and procedures are typically 50% higher in the United States than in Canada,¹⁴ we recalculated the mean total costs for the rt-PA/heparin and heparin strategies, noting no difference in the mean costs for the rt-PA/heparin and heparin groups ($689; 95% CI, $−1187 to $2566; P=0.47).

Table 2.

Unadjusted mean cost of managing patients in the PreCLOT study with rt-PA or heparin over the 6-month study period

Variable	rt-PA (n=110)	Heparin Only (n=115)
Drug (rt-PA/ heparin) costs	1206±598	130±85
Cost of managing patients experiencing catheter malfunction	47±197	113±446
Cost of managing patients with catheter-related bacteremia	619±3957	1295±5629
Total cost	1872±3861	1538± 5605
Difference in total costsa (rt-PA minus heparin) (95% CI)	323 (−935 to 1581)
P value	0.616

Results reported as mean±SD unless otherwise indicated. Costs are expressed in 2011 Can$.

^aMean difference, 95% CI, and P value were calculated with the bootstrap method.

Decision Analytic Modeling

Model Validity

We established internal and predictive validity of our model by comparing the proportion of patients who developed catheter malfunction and catheter-related bacteremia at 6 months, noting similar results between the model and the randomized trial. Moreover, at 6 months, we noted similar relative cost differences compared with the trial (Table 3). Using our model, we noted that 50.0%, 28.8%, 8.0%, and 13.2% of patients were using a catheter, had a fistula, had received a transplant, or had died at 1 year, respectively.

Table 3.

Clinical inputs and cost of treatments and managing complications in the base-case decision analysis

Outcome Variables	Values	Range	Source
Clinical variables
Risk of catheter malfunction rt-PA over 3 mo	0.145	0.093–0.211	PreCLOT
Risk of catheter-related bacteremia rt-PA over 3 mo	0.040	0.013–0.092	PreCLOT
Hazard ratio heparin			PreCLOT
Catheter malfunction	1.910	1.130–3.220	PreCLOT
Catheter-related bacteremia	3.300	1.180–9.220	PreCLOT
Probability of needing a new catheter			PreCLOT
After catheter malfunction	0.034	0.009–0.085	PreCLOT
After catheter-related bacteremia	0.854	NA	PreCLOT
Probability of patients with catheter malfunction returning to a functional catheter, without need for new catheter			PreCLOT
First 3 mo	0.677	NA	PreCLOT
After 3 mo	1.000a	NA	PreCLOT
Risk of hospitalization for catheter-related bacteremia	0.550	NA	PreCLOT
Probability of a establishing a functioning fistula enabling catheter removal every 3 mo	0.115	0.080–0.158	PreCLOT
Risk of fistula failure
First 6 mo	0.0086	NA	Ravani et al.15
After 6 mo	0.0005	NA	Ravani et al.15
Risk of transplant for age <65 y every 3 mo	0.047	NA	CORR35
Risk of mortality over 3 mo
Patient on dialysis age <65 yr	0.019	0.018–0.021	CORR35
Patient on dialysis age >65 yr	0.052	0.049–0.054	CORR35
Cost variables (in 2011 Can$)
Cost of treatment over 3 mo (Can$)
rt-PA	897	449–1346	PreCLOT
Heparin	98	73–122	PreCLOT
Mean cost of hospitalization for catheter-related bacteremia (Can$)	16,440	7297–25,585	PreCLOT
Mean cost of outpatient care for managing catheter-related bacteremia (Can$)	1731	865–2596	PreCLOT
Cost of catheter replacement (Can$)	1409	704–2123	PreCLOT
Cost of rescue rt-PA for patients who develop catheter malfunction (Can$)	533	NA	PreCLOT
Mean cost of managing patients who had a bleeding complication (Can$)	6775	NA	PreCLOT
Cost of maintaining a functioning fistula (per 3 mo) (Can$)	928	NA	Manns et al.9
Discount rate	0.05	NA

NA, not applicable; CORR, Canadian Organ Replacement Registry.

^aOf those who continued to have a poorly functioning catheter at 3 months, all patients who did not require a new line returned to a functional catheter state by the end of the second 3-month period.

Base-Case Analysis

Over a 1-year time horizon, the cost of rt-PA/heparin and heparin prophylaxis that would be required to manage 100 hemodialysis patients with a recently inserted catheter was $173,700 and $19,000, respectively. The overall cost, excluding dialysis costs but including the cost of managing complications (catheter malfunction and catheter-related bacteremia), to manage 100 hemodialysis patients with a recently inserted catheter was $316,300 with rt-PA/heparin and $328,300 with heparin alone. On the basis of the results of the clinical trial, and assuming ongoing efficacy, the use of rt-PA/heparin reduced episodes of catheter-related bacteremia by 62% and catheter replacements by 68% (Table 4). Additional analyses suggested that the discrepancy between the results of the decision analysis (slightly lower costs for rt-PA/heparin compared with heparin alone) and the costing analysis over the trial period (slightly higher costs for rt-PA/heparin compared with heparin alone) was due to the longer observation period for the rt-PA/heparin patients in the trial,¹³ as well as the fact that patients with catheter malfunction (more than one third of patients in the heparin group) were censored after 30 days and therefore were not “eligible” during the trial observation period to develop bacteremia (the more expensive of the adverse outcomes).

Table 4.

Base-case and sensitivity analysis results for a cohort of 100 hemodialysis patients with a recently inserted catheter, using decision analysis over a 1-year period

Variable	Drugs Costs (Can$)		Overall Costs (Can$)		Patients with Bacteremia (n)		New Catheters (n)
	rt-PA	Heparin Only	rt-PA	Heparin Only	rt-PA	Heparin Only	rt-PA	Heparin Only
Base case analysis	173,700	19,000	316,300	328,300	13	34	6	19
Variation in rt-PA hazard ratio for catheter malfunction
Lower 95% CI	173,700	19,000	310,200	328,300	13	34	6	19
Upper 95% CI	173,700	19,000	328,200	328,300	14	34	6	19
Variation in rt-PA hazard ratio for catheter-related bacteremia
Lower 95% CI	173,700	19,000	269,800	328,300	4	34	4	19
Upper 95% CI	173,700	19,000	440,400	328,300	31	34	23	19
Ongoing effectiveness of rt-PA over time: varying rt-PA hazard ratio for catheter malfunction and catheter-related bacteremia after 6 mo
HRs after 6 mo based on upper 95% CIs	173,700	19,000	357,100	328,300	15	34	10	23
HRs after 6 mo set at 1.0 (i.e., assumes rt-PA not effective after 6 mo)	173,700	19,000	367,600	328,300	17	34	10	23
Variation in risk of catheter-related bacteremia (base case, 0.04 episode of bacteremia per 1000 patient-days)
0.013 episode of bacteremia per 1000 patients-days (lower 95% CI)	173,700	19,000	182,700	266,400	3	13	3	10
0.092 episode of bacteremia per 1000 patients-days (upper 95% CI)	173,700	19,000	407,900	557,700	26	53	19	33
Variation in cost of rt-PA
Decreased by 50%	93,500	19,000	226,100	311,800	13	34	6	19
Increased by 50%	255,100	19,000	406,600	344,800	13	34	6	19
Scenario analysis
Using United States costs for rt-PA and hospitalization (in $)	280,400	16,700	469,000	450,500	13	34	6	19
Risk of fistula reduced by 50%	194,800	21,200	328,000	342,200	14	36	9	25

Costs in 2011 Can$, unless otherwise noted. HR, hazard ratio.

Sensitivity Analysis

In general, the results of our analysis were robust to the uncertainty in most variables. For example, the overall costs were robust to plausible variation in the cost of rt-PA and heparin, cost of outpatient care, cost of catheter replacement, exclusion of rescue rt-PA cost, or the hazard ratio of catheter malfunction. However, the model was sensitive to plausible variation in the relative risk of developing catheter-related bacteremia, and whether the relative risk of catheter-related bacteremia was maintained after the 6-month period in which it was tested within the trial (Table 4). In addition, in scenarios where the rate of catheter-related bacteremia was assumed to be lower, or the relative risk of catheter-related bacteremia with heparin use was lower, the cost of the rt-PA/heparin strategy became less attractive (Figure 1). When we assumed that rt-PA was less effective after 6 months, the cost of the rt-PA strategy became higher than that of the heparin strategy (Table 4).

Figure 1.

Two-way sensitivity analysis assessing the impact of changing the baseline rate and hazard ratio of catheter-related bacteremia in the heparin group on overall costs over a one-year time horizon.

We also considered a scenario that might be more likely in the United States, where the cost of rt-PA and hospitalization is higher. In this scenario, we noted that the cost of rt-PA/heparin and heparin drug prophylaxis regimens that would be required to manage 100 hemodialysis patients with a recently inserted catheter was $280,400 and $16,700 over 1 year, respectively. Including the cost of managing complications (catheter malfunction and catheter-related bacteremia), the overall cost to manage these 100 hemodialysis patients with a recently inserted catheter was $469,000 with rt-PA/heparin and $450,500 with heparin alone over 1 year.

Discussion

In the original publication of the PreCLOT study,¹³ we estimated that use of rt-PA/heparin would be associated with much higher costs, although this prior “back of the envelope” analysis assumed that all patients would use their locking solution for 6 months, and we based the cost of managing patients hospitalized for catheter-related bacteremia on outdated costing estimates.⁸ In this updated analysis, using estimates of clinical effectiveness and health care cost data from trial patients exclusively, we noted similar costs for patients managed with rt-PA/heparin and heparin alone as a locking solution, regardless of the method used to compare costs. Specifically, we noted that the higher cost of rt-PA in the rt-PA locking strategy was partially offset by lower costs for managing catheter-related bacteremia and rescue rt-PA for managing catheter malfunction. Importantly, when we considered a scenario that might be more relevant in the United States, where the cost of rt-PA and hospitalization for catheter-related bacteremia is substantially higher,¹⁴ we noted similar results.

When we used decision analysis to extrapolate to a 1-year timeframe, we noted that the results of the decision analysis were sensitive to the frequency and cost of managing catheter-related bacteremia, and the relative risk of catheter-related bacteremia. Of note, the ongoing rate of bacteremia that we considered within our analysis (based on the PreCLOT trial data) was lower than that observed 6 months after catheter placement in a large analysis of the Dialysis Outcomes and Practice Patterns Study,¹⁵ and the costs of hospitalization we observed were lower than other published estimates¹⁴ suggesting that our results may be generalizable to other countries. However, as noted, the results of this analysis were largely driven by cost savings associated with lower rates of catheter-related bacteremia in the rt-PA/heparin group. Given the infrequent rate of catheter-related bacteremia in both groups, further evidence of long-term treatment efficacy is required to obtain a more precise estimate of the risk of catheter-related bacteremia associated with different locking strategies.

Some limitations of this analysis require consideration. Although this study was based largely on one randomized trial of moderate size,¹³ it is important to note that no other large randomized trials have compared rt-PA locking solutions in hemodialysis patients. However, a randomized trial of patients with indwelling central venous catheters and hematologic malignancies also observed a 39% lower risk of catheter-related coagulase-negative staphylococci infections.¹⁶ Among hemodialysis patients, a recent randomized trial comparing gentamicin/citrate and heparin locking solutions reported a significant decrease in the risk of catheter-related bacteremia for gentamicin/citrate locking solutions,¹⁷ consistent with earlier related studies.¹⁸ However, the potential side effects of antibiotic locking solutions have reduced their routine use in clinical practice.¹⁹ Further, studies comparing higher concentrations of trisodium citrate (46.7%) with heparin report no reduction in rates of catheter-related bacteremia.^20,21 Another limitation is that our results in part relate to a small number of patients experiencing very high costs. Although it is not appropriate to exclude these patients (because the costs associated with their care were real), this does reduce the generalizability of our findings. While the cost of managing inpatients with catheter-related bacteremia (>Can$16,000) may seem high, our estimates are lower than those reported in other Canadian and United States studies.^14,22 Finally, PreCLOT only compared rt-PA/heparin with heparin alone, although citrate is now used as the catheter-locking solution by many dialysis units in North America. The cost-effectiveness of rt-PA/heparin in comparison to other locking agents, such as citrate,¹⁰ and in comparison to other prophylactic strategies, such as the use of topical polysporin triple, or mupuricin¹⁹ is uncertain.

How might renal programs use the results of this study? From our costing analysis, we noted numerically higher costs associated with the use of rt-PA (Can$323; 95% CI, −Can$935 to Can$1581; P=0.62). In our decision analysis using a longer time horizon, we noted similar overall costs, although costs for the prophylaxis strategy itself were higher (Can$173,700 versus Can$19,000) if 100 hemodialysis patients with a newly inserted catheter received rt-PA/heparin versus heparin alone as a locking solution. Assuming that the effectiveness of rt-PA is maintained in the long term, our decision analysis suggests that dialysis units with a rate of catheter-related bacteremia >1 per 1000 catheter-days might not experience a significant increase in costs overall (Figure 1). However, although the additional costs associated with use of rt-PA as a locking solution are certain, potential downstream cost savings are less certain. Some additional considerations are needed in the United States, where the cost of rt-PA is borne by dialysis units, whereas any potential cost savings due to lower hospitalization rates would be a benefit to Medicare but not directly to the dialysis units.

In summary, we noted no significant difference in the mean overall cost of a strategy using once-weekly rt-PA, combined with twice-weekly heparin, as a locking solution for catheters compared with thrice-weekly heparin only. Cost savings due to lower risk of hospitalization for catheter-related bacteremia partially offset the increased cost of rt-PA. Further information on the long-term effectiveness of an rt-PA/heparin strategy as a locking solution for catheters is required to guide practice.

Concise Methods

Costing Analysis

We first conducted a costing analysis with the primary objective to determine the cost associated with using rt-PA (1 mg per lumen, administered once per week with heparin used for the remaining two sessions [rt-PA/heparin] or heparin [5000 U/ml after each dialysis session (heparin only)]), taking into account the cost of the locking solutions and all other relevant medical costs. This analysis was conducted alongside the clinical trial¹² as detailed below.

Patient Population and Costing Time Period

All patients randomly assigned in the randomized controlled trial were included, with costs collected from study initiation until study completion (6 months for patients not censored for study outcomes or loss to follow-up). For patients experiencing catheter malfunction (the primary study outcome), the costing period was extended for at least an additional month, or until either six consecutive successful dialysis sessions were achieved (defined as a mean blood flow of ≥300 ml/min at each run) or the catheter was removed. Patients experiencing catheter-related bacteremia were followed until hospital discharge (if admitted) or completion of their outpatient antibiotics, ensuring complete capture of downstream health care costs related to all adverse events.

Costs Considered

In addition to drug costs for rt-PA and heparin (Can$299 and Can$33 per month, respectively), the resources to maintain a hemodialysis catheter were divided into relevant categories, including the cost of managing catheter malfunction (including catheter replacement when necessary) and the cost of managing catheter-related bacteremia (whether managed on an outpatient or inpatient basis). Following an episode of catheter malfunction, the cost of treating any repeat malfunction with rt-PA, or a diagnostic imaging procedure (catheter replacement, fibrin sheath stripping, or catheter re-wiring) was documented.

For patients with catheter-related bacteremia managed without hospital admission, management costs were estimated, including the cost of blood cultures, antibiotics used to treat the bacteremia (based on dose and duration of treatment), laboratory monitoring of antibiotic levels, and catheter replacement (where indicated). For patients requiring hospital admission, we determined (through blinded chart review using a priori definitions) whether the admission was directly attributed to the catheter (included hospitalization for catheter-related bacteremia or complications from the bacteremia [i.e., endocarditis, osteomyelitis, septic arthritis, and epidural abscess]) and complications related to catheter insertion.¹²

Study coordinators (blinded to treatment assignment) collected detailed information on hospital admissions for catheter-related bacteremia in all participating centers. Because costing methods differ across sites, the cost of admissions was based on data from the largest sites (Calgary and Edmonton, Alberta). However, recent studies suggest that hospital costs are similar to those within the sites in Ontario, Canada, where all other patients within this trial were admitted.²² The cost of inpatient care in Alberta was estimated using a microcosting method,^23–25 combining allocation of all direct and overhead costs associated with an inpatient encounter during the entire hospital stay. The costs included represent the cost of treating the patient rather than charges or payments; the quality of the costing data has been ranked highly.²⁴ A similar approach was taken for patients admitted for bleeding complications, the cost of which was considered in secondary analyses. The cost of physician’s services for all inpatient and outpatient encounters was recorded as per the Alberta provincial billing list in 2011.

Analytic Approach for Costing Analysis

In the primary analysis, we considered the cost of the study drugs (rt-PA and heparin) and all costs related to managing catheter malfunction and catheter-related bacteremia. Because there were few bleeding events with no significant difference in frequency between groups,¹³ we excluded the cost of bleeding in the primary analysis, but we did consider bleeding events in a secondary analysis.

Given the variable follow-up in the two study groups, and the expected non-Gaussian distribution of costs, there were unique issues we took into account when comparing costs across treatment groups.^26,27 We used established methods to enable comparisons of mean costs because these are easily interpretable and relevant to health care payer. This included use of nonparametric bootstrap estimates to derive 95% CIs and mean cost differences between the treatment groups.^28,29 Using 1000 bias-corrected bootstrap replications, and based on sampling with replacement from the original data, we estimated the distribution of a sampling statistic to derive 95% CIs.³⁰ In sensitivity analyses, we used generalized linear models to compare total costs across groups.³² We built these models considering three family distributions (Gaussian, inverse Gaussian, and gamma) and specifying two link functions (identity and log). We studied the performance of these models and selected the final model considering the information criteria, goodness of fit, and graphical and formal tests based on residuals. We did analyses with Stata software, version 11 (StataCorp., College Station, TX). We repeated our costing analysis using United States cost estimates, acknowledging that the price of rt-PA is higher in the United States ($107 per 2 mg compared with Can$64 in Canada ($1=Can$1.01 in 2011), and that the cost of hospitalization and procedures is typically 50% higher in the United States than in Canada.^14,32

Decision Analytic Modeling

Overview of Decision Analysis

Because the randomized trial was only 6 months in duration, we also used decision analysis to extrapolate the results from the trial to a 1-year time horizon. Using the results from the trial¹³ and the costing analysis described herein, we performed an incremental cost-effectiveness analysis comparing the study interventions. Model outcomes included cost, number of new catheters, and cases of catheter-related bacteremia.

Target Population

The target population included hemodialysis patients with a recently inserted catheter. To inform the decision analysis, we used costing and clinical data from randomly assigned patients from the time of enrollment throughout the 6-month study period.

Treatment Comparators

We modeled the interventions—rt-PA/heparin and heparin alone as locking solutions—as noted earlier.

Clinical and Costing Inputs

We derived most of the clinical and costing estimates from the Pre-CLOT trial¹³ and the costing analysis described herein (Table 3). The exception was the likelihood of transition from using a catheter to a fistula, which was informed by the trial results (22% of patients were transitioned to a fistula during the 6-month trial). No patients in the trial had an arteriovenous graft placed, which is consistent with current practice in Canada, where grafts are infrequently used³³; as such, we only modeled transition to a fistula. Given the limited number of patients admitted for catheter-related bacteremia (and making the conservative assumption that the duration of hospitalization for catheter-related bacteremia would be similar irrespective of treatment assignment), for the decision analysis we based the mean cost of hospitalizations on all events, regardless of treatment arm.

Analytical Approach

In the baseline analysis, we used decision analysis to model the cost and consequences of using rt-PA/heparin or heparin as the locking solution in a simulated cohort of hemodialysis patients with a recently inserted central venous catheter, categorized into two age groups: those <65 years and those >65 years. The perspective of the economic evaluation was that of the health care purchaser, and we only considered access-related costs (cost of locking agent, cost of managing catheter-related complications, and the cost associated with fistula use in those who subsequently used a fistula).³⁴ We did not include the cost of dialysis or transplantation because these costs were equal in both strategies.

A Markov process was used to model transitions, over recurring 3-month cycles, between the different clinical states considered (Figure 2). Because the effect of rt-PA/heparin on quality of life and survival is uncertain, expressing the results as a cost per quality-adjusted life-year would add unnecessary uncertainty. Therefore, in addition to the cost of each strategy, we report outcomes that are relevant for patients, including the expected number of new catheter insertions and catheter-related bacteremias over a 2-year time horizon. To facilitate interpretation, we present the analyses for a cohort of 100 hemodialysis patients with a recently inserted catheter. We discounted costs and clinical outcomes at an annual rate of 5%.³⁴ Costs were calculated based on 2011 Canadian dollars (1Can$=$1.006). Decision analysis was performed with the use of TreeAge Pro 2012 (TreeAge Software, Inc.).

Figure 2.

Markov model structure.

Consistent with published guidelines,³⁴ we established face, internal, and predictive validity of our model comparing model outputs (a function of both input variable and model structure) with reported data from the trial.

Sensitivity Analysis

We performed extensive sensitivity analyses to test the robustness of our findings. These analyses assessed the effects of varying baseline estimates within the 95% CI observed within the trial, or within clinically plausible ranges for other estimates on access-related cost and clinical outcome, including the duration of benefit of rt-PA. In addition, because the risk of catheter-related bacteremia varies across dialysis units from 0.5 to 8 episodes per 1000 catheter-days,¹⁹ this estimate was also varied widely.

Disclosures

B.J.M.: Investigator-initiated research grant from Hoffman La Roche. N.S.D.: Research grants from Hoffman La Roche, Janssen, Inc., Chief Medical, and Amgen; honorarium for lectures from Takeda, Hoffman La Roche, Chief Medical. M.T.: Investigator-initiated research grants from Abbott and Baxter. M.D.: Speaker/consultant fee and/or research funds from Amgen, Gambro, Baxter, Merck, Novartis. R.H.: Investigator-initiated research grant from Amgen. L.M.: Investigator-initiated research grant from Hoffman La Roche; honorarium for lectures or advisory boards from Hoffman La Roche. B.R.H.: Investigator-initiated research grant from Hoffman La Roche.