Systematic Review and Cost-Benefit Analysis of Radial Artery Access for Coronary Angiography and Intervention

Abstract

Background

Radial artery access for coronary angiography and interventions has been promoted for reducing hemostasis time and vascular complications compared to femoral access, yet it can take longer to perform and is not always successful, leading to concerns about its cost. We report a cost-benefit analysis of radial catheterization, based on results from a systematic review of published randomized controlled trials (RCTs).

Methods and results

The systematic review added five additional RCTs to a prior review, for a total of 14 studies. Meta-analyses, following Cochrane procedures, suggested that radial catheterization significantly increased catheterization failure (OR 4.92, 95% CI 2.69–8.98), but reduced major complications (OR 0.32, CI 0.24–0.42), major bleeding (OR 0.39, CI 0.27–0.57) and hematoma (OR 0.36, CI 0.27–0.48) compared to femoral catheterization. It added approximately 1.4 minutes to procedure time (CI −0.22 to 2.97), and reduced hemostasis time by about 13 minutes (CI −2.30 to −23.90). There were no differences in procedure success rates or major adverse cardiovascular events.

A stochastic simulation model of per-case costs took into account procedure and hemostasis time, costs of repeating the catheterization at the alternate site if the first catheterization failed, and the inpatient hospital costs associated with complications from the procedure. Using base-case estimates based on our meta-analysis results, we found the radial approach cost $275 (95% CI: −$374 to −$183) less per patient from the hospital perspective. Radial catheterization was favored over femoral catheterization under all conditions tested.

Conclusion

Radial catheterization was favored over femoral catheterization in our cost-benefit analysis.

Introduction

There is an increasing interest in the use of radial artery access for coronary angiography and interventions, with proponents arguing that the technique is associated with lower rates of bleeding and major vascular complications, improved patient comfort, and shorter times to hemostasis and ambulation than femoral access.

Despite these possible benefits, detractors have pointed to the longer procedure times and access failures associated with radial artery access. As a result, the use of radial access varies widely by practitioner, institution, and region. Radial artery access is the primary mode of access in several European countries, Canada and Japan, while used in the minority of catheterizations in the United States.^{1, 2}

Using a systematic review of randomized studies comparing radial and femoral artery vascular access, we performed a cost-benefit analysis from the hospital perspective to evaluate whether the reduction in procedure-related complications and reduction in hemostasis times attributable to radial artery catheterization would offset the increased procedure time and failure rate of radial catheterization. This information can then be used to support a local decision to switch to radial catheterization or continue using femoral catheterization.

Methods

Overview

We updated a previously published systematic review³ to obtain the most up-to-date probability and cost estimates for our cost-benefit analysis. We then developed a stochastic simulation model to determine the incremental per-case cost of radial versus femoral catheterization. The component costs of radial and femoral catheterization were identified, and the incremental costs were calculated as the product of an estimated unit cost and an estimated frequency at which the cost would be incurred. So for example, the estimated cost per case of procedure time is the per-minute cost of procedure time (C_pt) multiplied by the procedure time difference between radial and femoral catheterization (T_pt); and the estimated per-case cost of major bleeding complications is the cost of a major bleed (C_mb) multiplied by the difference in the probability of a major bleed (P_mb) between radial and femoral catheterization. The model outcome, the incremental cost of radial versus femoral catheterization, is shown in Equation 1 and the component variables in Table 1.

Table 1.

Variables included in cost-benefit model

Variable	Definition	Baseline	Range	Source
*Costs (US$ per event)
Cpt	Cost of procedure time (US$/minute)	15.55	15.55 to 62.20	Local data
Cht	Cost of hemostasis time (US$/minute)	1.75	0 to 15.55	FAME4
Cas	Cost of converting between radial and femoral access (catheterization at alternate site)	35	0 to 120	FAME4
Cmc	Cost of a minor complication	2,282	1,325 to 3,238	ACUITY5
Ch	Cost of a hematoma	3,779	1,890 to 7,558	GUSTO6
Cvc	Cost of a vascular complication	6,377	3,189 to 12,754	Kugelmass7
Cmb	Cost of a major bleeding event	6,739	5,585 to 8,784	Mean and range of 5 studies
Durations (minutes)
Tpt	Difference in procedure time with radial access	1.38	−0.22 to 2.97	Meta-analysis
Tht	Difference in hemostasis time with radial access	13.07	−23.9 to −2.3	Meta-analysis
Complication Probabilities with Femoral Catheterization
Pas	Probability of failed catheterization (catheterization at alternate site)	1.89%	–	Meta-analysis
Pmc	Probability of minor complication	5.70%	–	Meta-analysis
Ph	Probability of hematoma	2.02%	–	Meta-analysis
Pvc	Probability of vascular complication	2.53%	–	Meta-analysis
Pmb	Probability of major bleeding	1.11%	–	Meta-analysis
†Complication Probabilities with Radial Catheterization
Pas	Probability of failed catheterization (catheterization at alternate site)	9.30%	5.08% to 16.97%	Meta-analysis
Pmc	Probability of minor complication	2.22%	1.54% to 3.25%	Meta-analysis
Ph	Probability of hematoma	0.73%	0.55% to 0.97%	Meta-analysis
Pvc	Probability of vascular complication	0.81%	0.61% to 1.06%	Meta-analysis
Pmb	Probability of major bleeding	0.43%	0.30% to 0.63%	Meta-analysis

^*Cost values were varied in one-way sensitivity analyses; all other variables were modeled as triangular distributions

^†Meta-analysis used femoral catheterization times as baseline and calculated incremental time for corresponding radial times

$$\text{Incremental}\text{cost}\text{per}\text{case}=({\text{C}}_{\text{pt}}•{\text{T}}_{\text{pt}})+({\text{C}}_{\text{ht}}•{\text{T}}_{\text{ht}})+({\text{C}}_{\text{as}}•{\text{P}}_{\text{as}})+({\text{C}}_{\text{mc}}•{\text{P}}_{\text{mc}})+({\text{C}}_{\text{h}}•{\text{P}}_{\text{h}})+({\text{C}}_{\text{vc}}•{\text{P}}_{\text{vc}})+({\text{C}}_{\text{mb}}•{\text{P}}_{\text{mb}})$$ Equation 1

Systematic review and meta-analysis

To estimate the cost of complications in catheterization procedures and the cost of unsuccessful catheterizations, we first must have evidence-based estimates of rates for these events. A systematic review providing these estimates was published by Jolly et al.³ in 2009. We updated the systematic review and meta-analysis with data published from randomized controlled trials (RCTs) comparing radial and femoral catheterization.

Our review followed procedures set out in the PRISMA statement. Systematic searches of the Medline, EMBASE, and Cochrane Central Register of Controlled Trials (CENTRAL) databases were carried out in May 2011. Search strategies combined various indexing terms for coronary angiography and percutaneous coronary intervention (PCI) using an OR statement, and then combined those results with a term for the radial artery. The Medline search was limited to articles published since 2008.

Studies were included in the review if they assigned patients to catheterization groups randomly and reported any of the outcomes of interest, and included data from at least 25 patients per group. Studies were excluded if they selected for specific patient subgroups, as their results might not be representative of the whole. We did not exclude articles from consideration on the basis of language.

The literature searches yielded a total of 566 hits which was reduced to 43 after screening of titles and abstracts, and then reduced to 25 after elimination of duplicate records. All articles under consideration were successfully retrieved, and five met inclusion criteria. These were added to the 9 RCTs cited by Jolly to yield a total of 14 studies included in the meta-analyses (Figure 1).

Figure 1. PRISMA flow diagram.

PRISMA diagram of search results and article flow

We assessed the quality of each included trial using a nine-point scale based on scales from Jadad⁸ and Chalmers⁹ (see Supplemental Table 1). The overall quality of the evidence base for each outcome of interest was assessed using the GRADE system.^10–12 Data from each trial was abstracted by an experienced research analyst and was not done in duplicate.

Meta-analyses were performed using Cochrane methods and RevMan version 5.1 software. For dichotomous outcome variables such as complication rates, summary odds ratios and confidence intervals were calculated. For continuous variables such as procedure time, we analyzed mean differences between radial and femoral groups. All analyses used fixed-effects models as the default, changing to random-effects if significant heterogeneity across studies was detected. Thresholds for use of random-effects methods were conservative: significant heterogeneity was defined as a Q statistic with a p-value less than 0.1 or an I² statistic greater than 50%.

Predefined variables for subgroup analysis included year of publication (before or since 2008), quality rating (5 or greater), use of closure devices in femoral catheterization (all or some patients vs. no use), and trials involving only those undergoing PCI. We also tested the robustness of the conclusions by eliminating individual studies from the meta-analyses to determine if our conclusions were sensitive to an individual study.

Cost-benefit model

The stochastic simulation model was developed in TreeAge Pro 2009 (TreeAge Software Inc., Williamstown, MA) and Table 1 displays its inputs. Each patient entering the model had a risk of complications from their catheterization, depending on location (i.e., femoral or radial). The analysis was from the hospital perspective, and included the costs of catheterization lab and recovery room equipment and staffing (per minute), cost of a replacement catheter needed in the event access from the initial site failed, and the cost of added hospital stay, diagnosis, and treatment resulting from procedure complications. Input data for costs including those for catheterization procedure time (C_pt), hemostatis time in the recovery room (C_h), and cost of supplies needed for catheterization at the alternate site (C_as) were obtained from our local hospitals. These costs were in agreement with results of the FAME trial,⁴ a large-scale multicenter clinical trial of coronary artery catheterization and stenting. We found estimates for complication costs (C_mc, C_h, C_vc, and C_mb) in studies identified in our meta-analysis of procedure times and risks, and in a supplemental systematic Medline search limited to economic studies of coronary catheterization procedures published in English since 2007. These cost estimates include peri-procedure costs as well as costs of extended hospital stays, tests, and treatments associated with complications.

The systematic review yielded meta-estimates for T_pt and T_ht: the differences in minutes for procedure time and hemostasis time respectively for radial catheterization versus femoral catheterization. It also provided summary risk ratios and confidence intervals comparing probability variables P_as: the access success rate at radial and femoral sites, and P_mc, P_h, P_vc, and P_mb: the risk of minor complications, hematoma, vascular complications and major bleeding, respectively. These risk ratios were applied as multipliers on the baseline probabilities of each type of femoral access complication to determine the corresponding probabilities of complications from radial access (Table 1).

Each simulation sent 1,000 patients through the model 1,000 times for a total of 1 million trials with unique outcomes to calculate the incremental cost per case. Monte Carlo probabilistic sensitivity analyses simultaneously varied the time and complication probability variables throughout the ranges listed in Table 1 (confidence intervals on the meta-analysis results were used as the range). In addition, one-way sensitivity analyses varied the cost of complications (C_mc, C_h, C_vc, and C_mb) and procedure costs (C_pt, C_h, and C_as) to determine the effect of each of these cost estimates on the incremental cost.

We used the mean cost across the studies as our baseline for each complication cost variable, and the maximum and minimum costs as the range for sensitivity analysis (Table 1). Since only one trial reported specific costs of hematomas,⁶ we multiplied and divided that cost by two to create a high and low range respectively for sensitivity analysis. For the procedure time cost and hemostasis time cost variables, we had to select arbitrary ranges for sensitivity analysis, ensuring we had at least a factor of two difference between baseline values and each extreme.

Results

Systematic review and meta-analysis

Nine RCTs were included in Jolly’s previous meta-analysis. Our systematic searches for additional trials (see PRISMA flow diagram, Figure 1) yielded five additional trials meeting the inclusion criteria for a total of fourteen. The trials are described in Table 2. Not all of the trials reported every outcome under consideration, and definitions of bleeding and other complications varied from trial to trial.

Table 2.

Table of studies

Study	Quality (0–9)	Patients	Closure devices for femoral group	Major complication definition	Major bleeding definition	Other definitions
† RIVAL 201113	7	2,361 diag 4,660 interv.	*Some (26%)	Pseudoaneurysm requiring closure, AV fistula, “large” hematoma, or ischemia requiring surgery (major bleeding reported separately)	Death, transfusion of 2+ units, hypotension requiring inotropes, disability, hemoglobin decr. 5 g/dL, intracranial or intraocular bleed
Hou 201014	4	100 interv.	None	Reported individually	Hemoglobin decrease ≥ 2 g/dL or transfusion
Brueck 200915	5	654 diag. 370 interv.	*Most (93%)	Transfusion, hemoglobin dec. 3 g/dL, pseudoaneurysm, hematoma, etc.	Hemoglobin decrease > 3 g/dL or transfusion
Santas 200916	5	533 diag. 572 interv.	None	Retroperitoneal bleed, transfusion, hemoglobin decrease 5 g/dL, vascular complication req. surgery		Minor complications: pseudoaneurysm, ischemia, hematoma > 5 cm
Achenbach 200817	4	228 diag. 79 interv.	None	Reported individually		Minor complications reported individually
Lange 200618	2	195 diag. 102 interv.	Not reported	Not reported	Not reported
† OUTCLAS 200519	5	644 interv.	None	Reported individually	Hemoglobin decrease > 3 g/dL or transfusion
OCTOPLUS 200420	5	193 diag. 186 interv.	*Some (7%)	Reported individually	Hemoglobin decrease ≥ 3 g/dL or transfusion
Reddy 200421	4	75 diag.	Study variable	Reported individually	Transfusion
CARAFE 200122	5	108 diag. 92 interv.	Interventional patients only	Reported individually	Transfusion
† Cooper 199923	5	200 diag.	None	Transfusion, pseudoaneurysm, ischemia	Transfusion	Hematoma defined as any measurable discoloration
Mann 199824	3	142 interv.	All	Reported individually	Transfusion
BRAFE 199725	5	‡150 interv.	All	Reported individually	Not reported
† ACCESS 199726	4	900 interv.	None	Reported individually	Hemoglobin decrease > 3 g/dL or transfusion

^*–Closure devices used at physician’s discretion

^†–Authors report funding or sponsorship by device or drug manufacturer

^‡–Female patients excluded

Quality of articles assessed using modified Jadad scale (see Supplemental Table 1) where scores can range from 0 (lowest quality) to 9 (highest quality)

Most trials, particularly the recent ones, combined patients undergoing purely diagnostic catheterization with those having PCI. Published results did not report outcomes separately in diagnostic and interventional procedures. Where femoral closure devices were included in study protocols, their use was usually at the discretion of the cardiologist performing the procedure, and outcomes were not separately reported for cases with and without the devices. The one exception was Reddy’s study²¹ where patients were randomized in three groups: radial catheterization, femoral catheterization with closure device, and femoral catheterization without closure device. However, results of that study were confounded by differences in the catheters used for the closure device and control patients.

Catheterization, procedure success, and cardiovascular events

Detailed results of our meta-analyses are presented in the online data supplement, while the results are summarized in Table 3. Meta-analysis of catheterization success rates (Supplemental figure 1A) found that patients randomized to radial catheterization were five times more likely to need conversion to the other access site than patients randomized to femoral catheterization (summary odds ratio [OR] 4.92, 95% confidence interval [CI] 2.69–8.98). However, the higher failure rate for initial catheterization did not affect overall success rates for the diagnostic or interventional procedure (OR 1.03, 95% CI 0.93–1.13, Supplemental figure 1B). Major adverse cardiovascular events also did not differ between radial and femoral catheterization groups (Supplemental figure 2).

Table 3.

Summary of meta-analysis results

Outcome	N trials	N patients	Summary odds ratio [95% CI]	Favors	Significance	Effect of recent studies
Catheterization failure	12	11,273	4.92 [2.69–8.98]	Femoral	p < 0.001	No change
Procedure success	11	10,579	0.97 [0.89–1.07]	No difference	NS	No change
MACE	11	10,531	0.96 [0.77–1.21]	No difference	NS	No change
Major complications	13	11,913	0.32 [0.24–0.42]	Radial	p < 0.001	No change
Major bleeding	12	10,908	0.39 [0.27–0.57]	Radial	p < 0.001	No change
Hematoma	10	9,661	0.36 [0.27–0.48]	Radial	p < 0.001	No change
Minor complications	Insufficient data for meta-analysis
Mean difference (min) [95% CI]
Procedure time	9	3,086	1.38 [−0.22–2.97]	Femoral	p = 0.09	More uncertainty
Hemostasis time	3	1,255	13.1 [2.3–23.9]	Radial	p = 0.02	Smaller difference

Complications

While definitions of “major complications” varied from study to study, and some included bleeding or hematoma in their figures for major complications, the meta-analyses showed remarkable consistency in the relative complication risks of radial and femoral catheterization (Supplemental figure 3). Complication rates were reduced 60 to 70 percent with radial catheterization compared to femoral catheterization, and the risk reduction had high statistical significance. We could not meta-analyze results for minor complication rates because one of the included studies¹⁶ reported only the percentage reduction and not the absolute number of minor complications in each group (it did report absolute numbers for the other types of complications). Therefore, to be conservative in our estimate of the safety benefits of radial catheterization, we chose the smallest summary effect size result from the other meta-analyses to use in the cost-benefit analysis for minor complications.

While significant heterogeneity was found in some of the variables meta-analyzed, the meta-analysis results were robust. None of our subgroup analyses, selecting more recent trials, higher-quality trials, trials involving only patients undergoing PCI, or trials where femoral closure devices were used, made any substantive changes in conclusions. In addition, none of the conclusions were dependent on a single trial.

Time-related variables

Many of the clinical trials reported mean times for completing the catheterization procedure as well as fluoroscopy time. Our meta-analysis (Supplemental figure 4) found a trend towards longer procedure times with the radial approach, but the statistical significance (p = 0.09) did not reach the standard threshold. However, the time difference was significant for fluoroscopy, supporting a hypothesis that the observed effect for procedure time is real and not a chance finding. Since the time results favor femoral catheterization, we presumed for the sake of our cost model that there was a real time cost for the radial approach.

Meta-analysis of hemostasis time data found a significant reduction in time with radial catheterization, though we are less certain of the magnitude of the effect due to heterogeneity of the study results.

Cost-benefit analysis

Using baseline figures for the costs of catheterization (Table 1), we estimated net hospital cost savings of $275 (95% CI: −$374 to −$183; negative values imply cost savings) per patient using radial catheterization instead of femoral catheterization. The large cost savings was driven primarily by complication costs as the procedure costs of radial catheterization were on average $1.52 (95% CI: $0.52 to $2.56) more than procedure costs of femoral catheterization.

Figure 2 shows the results of our sensitivity analysis. None of the changes to the input variables tipped the balance of costs in favor of femoral catheterization (i.e., all 95% CIs were negative, implying cost savings for radial catheterization).

Figure 2. Sensitivity analysis.

Tornado diagram showing the effect of changes in each cost component variable on the net cost savings of radial catheterization as compared to femoral catheterization. Baseline values from our meta-analyses are indicated by the vertical dotted line. None of the changes to the component variables tip the balance of costs in favor of femoral catheterization.

Figure 3 shows the sensitivity of the net cost difference to changes in the time it takes to perform a radial catheterization. Our meta-analysis of clinical trials found that on average, the radial procedure takes 1.4 minutes longer than the femoral procedure (see dotted line in Figure 3). That difference would have to increase to approximately 20 minutes in order for the added cost of catheterization lab time to fully offset the other cost differences and make the overall cost of femoral catheterization less than the cost of radial catheterization.

Figure 3. Effect of radial access procedure time on cost savings.

Figure 3. Relationship of procedure time difference between radial and femoral catheterization and cost savings with radial catheterization. The dotted line represents the actual difference in procedure time as determined by the meta-analysis. If radial and femoral procedures take the same time, the model concludes that radial catheterization will be $297 less costly. The cost advantage of radial catheterization disappears only if it takes approximately 20 minutes longer than femoral catheterization.

Similarly, Figure 4 shows the effect of reducing complication rates in femoral catheterization on the net cost results. The left end of the graph represents our baseline rates for each of the four complication types modeled. Moving to the right on the x axis simultaneously reduces all four rates by the same proportion, with no change to the complication rates of radial catheterization. In order to overturn the finding that radial catheterization is less costly, the rates of all those complications of femoral catheterization would have to be reduced by approximately 60%.

Figure 4. Effect of reducing femoral access complication rate on cost-savings of the radial access strategy.

Figure 4. Using baseline risks of femoral and radial catheterization from the meta-analyses, radial catheterization saves $275 per patient. Risks of femoral catheterization must be reduced by approximately 60% with no corresponding change to radial risk in order for the net costs to be equal.

Discussion

Our cost-benefit analysis suggests that radial catheterization lowers hospital costs at the same time that it reduces adverse effects to patients. In addition, none of the changes to cost variables brought the net cost savings to a point which would favor femoral catheterization. Widespread adoption of radial catheterization could have substantial savings for the US health care system given that over a million coronary catheterizations are performed in the US annually.^{27, 28}

Since the first cases were reported in the late 1980’s,²⁹ radial artery access for diagnostic coronary angiography and PCI has gained momentum in many hospitals in Europe and Asia.² Advocates of the radial approach argue that it reduces vascular complications and rates of major bleeding, which has been demonstrated clearly in a recently published large randomized clinical trial: RIVAL.¹³ In addition, without the need for lengthy post-procedural bedrest, earlier ambulation and therefore discharge are possible, potentially reducing hospital costs and improving patient satisfaction.³⁰ They also report greater patient satisfaction with radial catheterization than with femoral.^{13, 26}

In the United States, however, radial artery cases account for less than 10% of the diagnostic cases and approximately 1% of PCI cases.¹ In addition, the radial approach is used less commonly in elderly patients, women and patients with acute coronary syndromes.¹ This discrepancy appears to stem from concerns about increases in procedure time, radiation exposure, and access failure in patients undergoing radial artery catheterization offsetting the benefits of decreased vascular complications. We estimate that the procedure time would need to increase by 20 minutes per case to make a femoral artery access strategy less costly than radial access, while our meta-analysis found a difference of only 1 minute and 23 seconds between them. There was a small but statistically significant increase in fluoroscopy time associated with radial access, but the increase was less than 1 minute.

Due to the steeper learning curve for gaining proficiency with the radial approach, we recognize that the benefits we demonstrate may not be observed early as practitioners develop facility with radial arterial access. One trial has shown that procedural time for the radial approach was longer than for the femoral approach at an interim analysis, but that at the end of the learning curve, there were no significant differences in procedure time.²⁶ Performing enough procedures to attain proficiency may pose a practical challenge for many cardiologists in the United States who have been trained to use the femoral artery approach, but do not have the opportunity to learn radial arterial access. Nevertheless, our systematic review and meta-analysis indicate that it would be to the benefit of patients for cardiologists to obtain training so they can use the radial approach over the long term.

Our study demonstrated that the savings from reduced vascular complications outweighed the increased costs of longer procedure times and access failure associated with radial artery access by a large margin. Although procedure outcomes as measured by major adverse cardiovascular events were equivalent between the two groups, radial artery access was associated with significantly reduced vascular access site complications and bleeding. Despite using procedure times and cost estimates unfavorable to radial artery access, a radial artery strategy was still superior due to the high cost of vascular complications associated with femoral access. Even if the difference in hemostasis time were equivalent between the techniques (such as if a femoral artery closure device were used on every case), the reduction in vascular complications would still make radial access less costly.

The pooled vascular event rate in the patients undergoing femoral access was 3.3% versus 1.0% for radial access. This estimation is somewhat higher than the major femoral vascular event rate of 2.1 to 2.8% published in a large clinical registry.³¹ In our study, a more than 60% reduction in all complications of femoral catheterization, with no corresponding reduction in complications of radial catheterization, would be needed to bring the net cost of femoral catheterization to that of radial catheterization. Recent retrospective studies have demonstrated a very low femoral vascular complication rate with the use of an arterial closure device,³² however no randomized studies have shown that femoral artery catheterization, with or without a vascular closure device, lower rates of vascular complications compared to radial access. If there were no differences at all between the vascular access site complications of radial versus femoral access, the increased cost for the radial artery strategy would only be $1.18 per patient, primarily due to the cost of access site failure. On the other hand, even if the absolute reduction in the rate of hematomas, major bleeding, and major vascular complications from radial access was only 0.1% for total events, there would still be a $18.07 per case cost savings from the radial artery access strategy. Therefore, even a slight reduction of vascular complications would result in a cost savings from using a radial artery access strategy.

Perspective

It needs to be emphasized that our cost study is analyzed from the hospital perspective, as costs were calculated based on reports of actual hospital expenditures to cover catheterization lab time, hospitalization, blood transfusions, and treatment of vascular complications. In particular, we considered major bleeding and vascular complications separately, therefore treating the costs as additive. While the extra cost to the hospital of a blood transfusion is independent of the cost for prolonged hospitalization, imaging studies, and vascular repair, from the payer’s perspective a blood transfusion would likely be subsumed under a diagnosis related group for a patient who undergoes a vascular repair. Since the cost of complications was the primary driver of increased costs of the femoral artery approach, the cost savings of a radial artery strategy to the payer and to society may not be as evident. Importantly, we did not consider the comfort or preference of patients and physicians, which could arguably be the primary reason a particular vascular access site is chosen.

Limitations

An important objection to the published literature comparing these access strategies is that many studies evaluate outcomes among operators who are already highly proficient with the procedure. The radial technique has a steeper learning curve than femoral artery access.³³ Failed access and radial artery complications are high early on, and decrease as operators became more skilled at the procedure.¹³ Furthermore, the patients enrolled in clinical trials comparing these strategies must be suitable candidates for either procedure, which might exclude high risk patients with complicated access issues. Therefore our estimates of complication rates, procedure and recovery times may be vulnerable to a publication bias that favors the radial access strategy. While one early trial²⁵ excluded female patients because of their smaller arteries, subsequent trials have included a broad range of patients.

Another limitation is the estimation of costs assigned to clinically-significant hematomas, which may necessitate vascular imaging, surgical consultation, possible intervention, and longer hospitalizations. It is unclear whether small or moderate sized hematomas that would not necessarily lead to these costly measures were counted in the clinical trials. But our conclusions are not solely dependent on the cost impact of hematomas. They contribute just $40 to $56 to the $275 per-case savings of radial versus femoral catheterization.

There are complications of radial access that did not factor into our analysis. Radial artery occlusions occur in approximately 5% of radial cases,³⁴ but it is difficult to attribute a cost to this complication since there may be no immediate consequences. A radial occlusion, however, can limit vascular access and conduits for surgical bypasses, which could significantly affect the kinds of therapies available to patients in the future.

Finally there was significant heterogeneity of study results in some of the meta-analyses used to estimate procedure times and success rates for our cost-benefit analysis (see Supplemental figure 1 and Supplemental figure 4). To test whether or not heterogeneity could be driving our conclusions, we excluded the studies most favorable to radial catheterization in analyses with significant heterogeneity (p < 0.10 or I² ≥ 50%) and repeated those analyses. The resulting changes to the meta-analytic results were small and did not change our conclusions. For example, excluding the Reddy and Santas trials from the meta-analysis of procedure time (Supplemental figure 4A) reduced I² from 80% to 6%, but raised the summary time difference only from 1.38 minutes to 2.30. That change would reduce the net savings of radial catheterization from $275 to $251. Heterogeneity was not significant in the complication rate differences, which make the largest contribution to cost savings.

Conclusion

A strategy of routine radial artery access for coronary angiography is associated with an overall reduction in hospital costs and vascular complications if operators have the proficiency with radial artery access to approximate the outcomes of clinical trials.

What is Known

• In the US, radial artery catheterization is performed in the minority of diagnostic angiograms and rarely in percutaneous coronary interventions.

• Radial artery catheterization can reduce hemostasis time and vascular complications, but can take longer to perform and may require conversion to the femoral site.

What this Article Adds

• Cost savings from reducing complications appear to outweigh additional direct procedure costs of radial catheterization.

• On average, the radial approach saved $275 in direct hospital costs per patient as compared to the femoral approach.

• None of the changes to cost variables brought the net cost savings to a point which would favor femoral catheterization.