Influence of access site choice for cardiac catheterization on risk of adverse neurological

events: A systematic review and meta-analysis

Short title: Access site and neurological events

Alex Sirker1*, Chun Shing Kwok2,3*, Rafail Kontronias2, Rodrigo Bagur4, Olivier Bertrand5,

Robert Butler3, Colin Berry6, James Nolan3, Keith Oldroyd7, Mamas Mamas1,2

*AS and CSK contributed equally to this work

1. Departments of Cardiology, University College London Hospitals and St. Bartholomew’s

Hospital, London, UK.

2. Cardiovascular Research Group, Institutes of Science and Technology in Medicine,

University of Keele and Institute of Cardiovascular Sciences, Stoke-on-Trent, UK

3. Department of Cardiology, Royal Stoke University Hospital, Stoke on Trent, UK

4. Division of Cardiology, University Hospital, London Health Sciences Centre, Western

University, London, Canada

5. Quebec Heart-Lung Institute, Laval University, Laval, Canada.

6. Institute of Cardiovascular and Medical Sciences, University of Glasgow, UK.

7. West of Scotland Regional Heart and Lung Centre, Golden Jubilee National Hospital,

Glasgow, UK.

No relationships with industry

Corresponding author:

Mamas A. Mamas

Professor of Cardiology / Honorary Consultant Cardiologist

Keele Cardiovascular Research Group,

1

University of Keele

Stoke-on-Trent, United Kingdom

Tel: +44 1782 671654 Fax: +44 1782 674467

Email: mamasmamas1@yahoo.co.uk

Keywords: Radial access, femoral access, stroke, neurological complications

Word count: 4,363

2

Abstract

Background: Stroke is a rare but potentially catastrophic complication of cardiac

catheterization. While some procedural aspects are known to influence stroke risk, the impact

of radial versus femoral access site use is unclear. Early observational studies and limited

randomized trial data suggested more frequent embolic events with radial access.

Subsequently, larger pooled analyses have shown no clear differences in stroke risk but were

limited by low event rates. Recent publication of relevant new data prompted our re-

evaluation of this concern. Therefore, we conducted a systematic review and meta-analysis

to evaluate stroke complicating cardiac catheterization with use of transradial versus

transfemoral access.

Methods and Results: A search of MEDLINE and EMBASE was undertaken using OVID

SP with appropriate search terms. RevMan 5.3.5 was used to conduct a random effects meta-

analysis using the inverse variance method for pooling risk ratios (RR) or the Mantel-

Haenszel method for pooling dichotomous data. Pooled data from over 24,000 patients in

randomized controlled trials (RCTs) and over 475,000 patients from observational studies

were used. The risk ratio for (any) stroke, using RCT data, was not significant (RR 0.87, 95%

CI 0.58-1.29). Using observational data, a significant difference favoring radial access was

seen (RR 0.71; 95% CI 0.52-0.98).

Conclusions: Radial access site utilization for cardiac catheterization is not associated with

an increased risk of stroke events. This data provides reassurance and should remove another

potential barrier to conversion to a ‘default’ radial practice among those who are currently

predominantly femoral operators.

3

Introduction

While stroke is a rare complication of percutaneous coronary procedures, it ranks

among the most feared, due to the potentially catastrophic consequences. Contemporary

series indicate that stroke complicating percutaneous coronary intervention (PCI) increases

30-day mortality by between 4- and 10- fold.1 In those who survive, there is a high incidence

of chronic morbidity and loss of independent living.2

The mechanism of stroke in the setting of cardiac catheterization appears to relate

principally to arterial embolism. Studies employing transcranial Doppler demonstrate

intermittent cerebral emboli during such procedures.3,4 The etiology of these emboli may be

gaseous (from microbubbles, particularly during catheter flushing) or particulate.5 The latter

are of greater concern, tend to occur particularly during catheter exchanges, and may

potentially represent thrombus, atheromatous material or calcium.5 The risk of PCI-related

ischemic stroke/transient ischemic attack (TIA) has been associated with various modifiable

procedure-related factors, including the number of catheter exchanges and the caliber of PCI

guide catheter, in a retrospective observational study.6

Use of radial access, in preference to femoral, for coronary procedures has risen

sharply in recent years. The principal driver has been recognition that radial use avoids the

risk of femoral access site related bleeding complications and this appears to translate into a

decrease in mortality and MACE events, at least in higher risk settings.7,8 However, the

impact of increased radial use on rare complications such as stroke (or asymptomatic cerebral

infarcts) is less clearly established. Early observational data suggested that radial access for

left heart catheterization was associated with more Doppler-detected cerebral emboli and this

was supported by a subsequent small randomized controlled trial (RCT) in which stenotic

aortic valves were crossed.3,9 The most recently published aggregate data in this area are from

4

meta-analyses in 2009 and 2013, the latter with a total of 11,273 patients.10,11 These did not

demonstrate any significant difference between radial and femoral use in terms of clinical

stroke events but there was a recognized limitation in power due to the rarity of events.

Since 2013, there has been a significant increase in the volume of published data in this field,

including that from the MATRIX randomized trial (with over 8000 patients) and a national

PCI registry with over 400,000 patients.1,8 Hence we have re-appraised this important issue

with an updated meta-analysis.

5

Methods

Eligibility criteria

This analysis included only studies that i) compared use of radial and femoral access

for cardiac catheterization (with or without PCI), and ii) evaluated stroke or examined other

markers of subclinical neurological events, namely the demonstration of new lesions on brain

magnetic resonance imaging or the detection of microemboli on transcranial Doppler testing.

Pooled analysis was performed only for studies reporting stroke. To be eligible, studies also

had to report quantitative event data so that assessment of outcomes could be compared

between radial and femoral access. There was no restriction based on study design, cohort

type or language of study report. Studies that did not have any stroke events were excluded

because these were underpowered to detect differences in events between radial and femoral

access groups.

Search strategy

A search of MEDLINE and EMBASE was undertaken in September 2015 using

OVID SP with no date or language restriction. The exact search terms were: (transradial or

transfemoral or radial artery or femoral artery or radial access or femoral access) AND

(angiography or angiogram or catheterization or catheterization or PCI or percutaneous

coronary intervention) AND (stroke or microemboli or cerebral infarction or dementia or

cognitive impairment). The bibliographies of relevant original studies and reviews were also

checked for additional studies that met our inclusion criteria.

Study selection and data extraction

Two reviewers (CSK, RK) screened all titles and abstracts retrieved from the search

for studies that met the inclusion criteria. The full manuscript of studies that potentially met

6

all inclusion criteria were reviewed and the final decision to include or exclude studies were

made with two other reviewers (AS, MAM). Independent double extraction was performed

from source publications by two reviewers (CSK, RK). Data was collected on study design,

year, country, number of participants, mean age, % male, participant inclusion criteria,

outcomes evaluated, timing of assessment and results.

Risk of bias assessment

Quality assessment of studies was conducted with consideration of whether the study

was prospective cohort or randomized controlled trial in design, whether both radial and

femoral treatment arms had more than 100 participants, whether there were reliable methods

for outcome ascertainment, whether loss to follow-up was greater than 10%, and whether

there were more than 3 variables with significant baseline differences for randomized trials or

use of adjustments or propensity score matching in cohort studies. Publication bias was

evaluated using asymmetry testing if there were more than 10 studies in the meta-analysis,

and if there was statistical heterogeneity <50%.12

Data analysis

RevMan 5.3.5 (Nordic Cochrane Centre) was used to conduct a random effects meta-

analysis using the inverse variance method for pooling risk ratios (RR) or the Mantel-

Haenszel method for pooling dichotomous data. Where possible, adjusted risk estimates

from primary studies were pooled. Where this data were not available, raw data were used to

calculate unadjusted risk estimates. The primary outcome measure was (all) stroke but

analysis of ischemic and hemorrhagic stroke types was also performed. The analysis was

stratified by whether individual studies were higher quality randomized trials or cohort

observational studies. Statistical heterogeneity was assessed by determining the I2 statistic

7

where I2 values of 30-60% represent a moderate level of heterogeneity.13 We performed

additional analysis considering the effect of year of publication on stroke rates and sensitivity

analysis by stratifying according to diagnostic procedures, PCI procedures or both diagnostic

and PCI procedures.

Results

Study selection and description

Figure 1 provides a flowchart of study selection. Our search results yielded 979

studies, of which 36 met inclusion criteria.1,3,5,7-9,14-45 The reasons for exclusion are shown in

Figure 1. Three studies from the most recent meta-analysis were not included because they

had no stroke events.46-48

The study design, participant characteristics and clinical settings are shown in Table

1. The 36 studies comprised 15 randomized controlled trials, 8 prospective cohort studies, 10

retrospective cohort studies and 3 cohort studies of unclear design. The studies were reported

between 1997 and 2015. There were a total of 1,112,343 participants who underwent 438,238

radial procedures and 643,469 femoral procedures (the crude number of participant for each

access site was not reported for 30,636 participants). The mean age was 65 years across 29

studies that reported age and 70% of participants were male from 31 studies that reported

gender.

Quality assessment of included studies

The quality assessment of included studies is shown in Table 2. Of the 36 included

studies, 23 (64%) were RCTs or prospective cohort studies. There were more than 100

participants in both radial and femoral arms of the studies in 25 (69%) studies and reliable

8

methods for ascertaining outcomes were also present in 25 (69%) of studies. Low loss-to-

follow up was likely or reported in half the studies (n=24) and there were 3 or fewer variables

with significant baseline differences in RCTs, or use of adjustment for confounding in cohort

studies, in 16 (44%) studies.

Comparison of stroke and neurological outcomes by access site

The study outcomes and results are shown in Table 3. The rate of stroke events from

28 studies (one study did not report crude events) was 0.14% (358 events/249,073 total) in

the radial group and 0.19% (472 events/251,095 total) in the femoral group. The rates of

stroke according to publication year are shown in Supplementary Figure 1. The pooled

results showed no significant difference in stroke from RCTs (RR 0.87 95%CI 0.58-1.29,

I2=0%) but a significant difference in cohort studies in favor of radial access (RR 0.71 95%CI

0.52-0.98, I2=26%) (Figure 2).

The risk of ischemic stroke from RCTs and cohort studies is shown in Figure 3. The

rates of ischemic stroke among 5 studies were 0.16% (92 events/57,548 total) and 0.26% (95

events/36,675 total) for the radial and femoral access groups, respectively. The pooled results

showed no difference in ischemic stroke from RCTs (RR 0.92 95%CI 0.41-2.08, I 2=57%)

but a significant difference in cohort studies again favoring radial access (RR 0.58 95%CI

0.41-0.82, I2=0%).

The risk of hemorrhagic stroke among RCTs and cohort studies is shown in Figure 4.

The rates of hemorrhagic stroke among 6 studies were 0.026% (15 events/58,050 total) and

0.032% (12 events/37,136 total) for the radial and femoral access groups, respectively. The

pooled results indicate no significant difference in hemorrhagic stroke rates between radial

and femoral access, from both RCT (RR 0.65 95%CI 0.19-2.15, I2=0%) and cohort studies

(RR 0.93 95%CI 0.22-3.93, I2=24%).

9

We also performed sensitivity analysis considering whether the procedures were

diagnostic, PCI or both. Our findings are similar to those reported in the main analysis,

withof the results suggested no statistical significance any differenceobserved between radial

and femoral access for stroke events by procedure type (Supplementary Figure 2).

Assessment of publication bias

There were more than 10 studies and low levels of heterogeneity in the analysis of

any stroke so we constructed a funnel plot to assess publication bias (Figure 5). There

appears to be some evidence of reporting bias.

Additional studies not included in meta-analysis

Results from a few of the 34 studies that met our inclusion criteria were not pooled as

they reported outcomes measures that did not include stroke events alone. Lund et al studied

42 patients and found 5 cases of new cerebral lesions on diffusion-weighted magnetic

resonance imaging (DWI) in the radial but not femoral access group.3 Jurga et al found a

higher number of microemboli on transcranial Doppler in the radial compared to the femoral

group (10.9±6.3 vs 6.9±4.7, respectively) .5

The observational study by Madiera et al, which reported a composite stroke or TIA

endpoint, found a significant reduction in such events with radial access compared to femoral

(OR 0.86 95%CI 0.27-2.7).38 For more broadly defined ‘neurological complications’, another

paper by Jurga et al demonstrated no significance difference between radial and femoral

access groups (OR 0.94 95%CI 0.75-1.17) (34), as was also the case for Sarno et al (OR 1.22

95%CI 0.91-1.64).43

10

Discussion

Stroke remains a feared, although fortunately rare, complication of invasive cardiac

procedures due to high mortality and significant chronic morbidity and loss of independent

living in survivors. The aggregated events in our analysis indicate an overall stroke rate of

only 0.52%, based on data from 13 RCTs, which is comparable to that found in a preceding

meta-analysis (0.43%).11 Nevertheless, given the repercussions of stroke on quality of life and

functional status, attempts to better understand risks for peri-procedural stroke are certainly

warranted. Additionally, whilst overt stroke is rare, the same fundamental processes are likely

to underlie development of new white matter lesions on brain MRI after cardiac

catheterization, which can occur in the absence of clinical stroke. Such findings are much

more frequent than overt stroke in studies that systematically search for them.49,50 These

lesions are known to persist at follow-up months later and hence seem to represent permanent

injury.51 Such initially ‘silent’ infarcts have been associated with steeper rates of cognitive

decline and increased risks for dementia.52 Previous studies from North America that have

studied temporal trends in stroke complications following PCI have shown that stroke rates

remained fairly constant between 1998-2008, although survival rates have improved over

time. In contrast, an analysis for the British Cardiovascular Interventional Society has shown

increases in national ischemic stroke rates between 2007-2012 due to changes in PCI activity

from elective to ACS indications and higher risk clinical patient demographics, although

haemorrhagic stroke complications decreased significantly due to decreased utilisation of GP

IIb/IIIa use. Our current analysis suggests that stroke rates have trended downwards both for

radial and femoral procedures in both randomised controlled trials and registry studies,

although it is not clear what has driven these changes, although may relate to advances in

pharmacotherapy and procedural techniques. Nevertheless whilst stroke remains a rare

complication of PCI, its outcomes are potentially devastating .

11

Of particular importance for preventing stroke are potentially modifiable technical

aspects of the cardiac catheterization procedure. There are various theoretical mechanisms by

which access site choice might alter the risk of cerebral emboli. For example, radial (or

brachial) access (particularly from the right upper limb) involves less catheter contact with

the aortic arch, a known rich source of potential athero-emboli.53 However, upper limb access

routes pass adjacent to the ostia of either the innominate or vertebral arteries, and tortuosity in

this region (as frequently encountered in older patients) might predispose to inadvertent guide

wire passage into these ostia, or to mechanical pressure from the angulated catheter on

atheromatous plaque in the more proximal subclavian artery. Other factors, such as procedure

duration, number of catheter exchanges and use of anticoagulation, also impact on risks of

embolism and may differ between radial and femoral approaches. Hence the net effect of

radial versus femoral access use on neurological events is difficult to predict a priori. Meta-

analyses of RCTs to date have not shown any significant difference between access site

choices in terms of stroke risk but they have been hampered in the strength of these

conclusions by the rarity of events.

Interpretation of findings from the present study

We used information from over 24,000 patients in RCTs and over 475,000 patients

from higher quality observational studies. This provided statistical power to search for

differences in a rarely occurring endpoint (stroke). For our principal outcome measure (all

stroke), pooled RCT data yielded a non-significant risk ratio of 0.87 (95% CI 0.58-1.29.

When looking at (all) stroke in the pooled cohort studies (which are dominated by 2 large

series), a significant result favoring radial use emerged, in contrast to the RCT data.

Although these observational studies have greater power in terms of far higher subject

12

numbers, they are of course subject to the usual issues of selection bias, and the particular

issue of confounding, that may not be fully corrected by measures such as propensity

adjustment. It is therefore possible that the more favorable stroke outcomes with radial use in

the observational studies might reflect a lower risk cohort. We have also undertaken

sensitivity analyses that study outcomes for diagnostic cardiac catheterisation and PCI

separately, and these sensitivity analyses show concordance with the findings of our main

analyses, that there are no significant differences in stroke risk associated between cases

undertaken through the radial and femoral approach, even in elective diagnostic cases which

are undertaken in the absence of anti-coagulation with heparin or bivalirudin.

Given other highly compelling reasons for choosing radial access (discussed earlier),

these aggregate all-stroke data are reassuring and should encourage the continuing increase in

radial use now being seen in the United States and elsewhere.

For the specific endpoint of ischemic stroke, a lower frequency of reporting for this

outcome measure led to lower subject numbers for meta-analysis. For the RCTs, this

unsurprisingly resulted in a non-significant difference between access sites and wide

confidence intervals, limiting the strength of any conclusions. For the cohort data, the

findings are driven by the results favoring radial use from a single large series. For

hemorrhagic stroke, the issue of low patient numbers and events, with consequent low

statistical power, was even more evident.

Limitations

This study is subject to certain limitations common to all meta-analyses, in terms of

susceptibility to reporting and publication bias. Although providing power to comment

13

usefully on all-stroke events, there was insufficient reporting (and hence inadequate patient

numbers) to draw any firm conclusions regarding possible subgroups of particular interest –

including, for example, comparing those with right versus left radial access (in whom contact

with the aortic arch by the catheter is known to differ). Available data on subclinical lesions

(new white matter cerebral changes on MRI), with radial versus femoral access, are scant and

hence (despite their potential long-term relevance) could not be pooled in the meta-analysis.

Finally, whilst our analysis has not shown any statistically significant differences in

neurological events between radial and femoral access site use in 24,000 patients derived

from RCTs and close to 1/2 million patients in cohort studies, this does not prove equivalence

as neurological events are rare complications in PCI and given that the absolute differences in

neurological events between the two access sites are small (0.14% and 0.19%), our analysis

may still be underpowered to detect small but statistically significant differences.

Conclusions

The increase in popularity of use of radial access for coronary procedures has been

driven by a reduction in both access site related bleeding complications and mortality in

several RCTs and large national registries, particularly in high-risk patient subgroups.

Previously expressed concerns about possible increased risks of neurological problems have

been difficult to fully allay until now, given the low frequency of overt events. This present

analysis takes advantage of the totality of high quality published data, including recently

published major RCT and observational work. It provides important reassurance regarding

the relative risk of such events associated with radial versus femoral approaches. This should

remove another potential barrier to conversion to a ‘default’ radial practice among those who

are currently predominantly femoral operators.

Acknowledgement

14

None

Funding Sources

None

Disclosures

None

15

References

1. Kwok CS, Kontopantelis E, Myint PK, Zaman A, Berry C, Keavney B, et al. Stroke following percutaneous coronary intervention: type-specific incidence, outcomes and determinants seen by the British Cardiovascular Intervention Society 2007-12. Eur Heart J 2015;36(25):1618-1628.

2. Dukkipati S, O'Neill WW, Harjai KJ, Sanders WP, Deo D, Boura JA, et al. Characteristics of cerebrovascular accidents after percutaneous coronary interventions. J Am Coll Cardiol 2004;43(7):1161-1167.

3. Lund C, Nes RB, Ugelstad TP, Due-Tonnessen P, Andersen R, Hol PK, et al. Cerebral emboli during left heart catheterization may cause acute brain injury. Eur Heart J 2005;26(13):1269-1275.

4. Bladin CF, Bingham L, Grigg L, Yapanis AG, Gerraty R, Davis SM. Transcranial Doppler detection of microemboli during percutaneous transluminal coronary angioplasty. Stroke 1998;29(11):2367-2370.

5. Jurga J, Nyman J, Tornvall P, Mannila MN, Svenarud P, van der Linden J, et al. Cerebral microembolism during coronary angiography: a randomized comparison between femoral and radial arterial access. Stroke 2011;42(5):1475-1477.

6. Hoffman SJ, Routledge HC, Lennon RJ, Mustafa MZ, Rihal CS, Gersh BJ, et al. Procedural factors associated with percutaneous coronary intervention-related ischemic stroke. JACC Cardiovasc Interv 2012;5(2):200-206.

7. Jolly SS, Yusuf S, Cairns J, Niemela K, Xavier D, Widimsky P, et al. Radial versus femoral access for coronary angiography and intervention in patients with acute coronary syndromes (RIVAL): a randomised, parallel group, multicentre trial. Lancet 2011;377(9775):1409-1420.

8. Valgimigli M, Gagnor A, Calabro P, Frigoli E, Leonardi S, Zaro T, et al. Radial versus femoral access in patients with acute coronary syndromes undergoing invasive management: a randomised multicentre trial. Lancet 2015;385(9986):2465-2476.

9. Hamon M, Lipiecki J, Carrie D, Burzotta F, Durel N, Coutance G, et al. Silent cerebral infarcts after cardiac catheterization: a randomized comparison of radial and femoral approaches. Am Heart J 2012;164(4):449-454 e1.

10. Jolly SS, Amlani S, Hamon M, Yusuf S, Mehta SR. Radial versus femoral access for coronary angiography or intervention and the impact on major bleeding and ischemic events: a systematic review and meta-analysis of randomized trials. Am Heart J 2009;157(1):132-40.

11. Patel VG, Brayton KM, Kumbhani DJ, Banerjee S, Brilakis ES. Meta-analysis of stroke after transradial versus transfemoral artery catheterization. Int J Cardiol 2013;168(6):5234-5238.

12. Ioannidis JP, Trikalinos TA. The appropriateness of asymmetry tests for publication bias in meta-analyses: a large survey. CMAJ 2007;176(8):1091-1096.

13. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327(7414):557-560.

14. Achenbach S, Ropers D, Kallert L, Turan N, Krahner R, Wolf T, et al. Transradial versus transfemoral approach for coronary angiography and intervention in patients above 75 years of age. Catheter Cardiovasc Interv 2008;72(5):629-635.

16

15. Benit E, Missault L, Eeman T, Carlier M, Muyldermans L, Materne P, et al. Brachial, radial or femoral approach for elective Palmaz-Schatz stent implantation. Cathet Cardiovasc Diagn 1997;41(2):124-130.

16. Bernat I, Horak D, Stasek J, Mates M, Pesek J, Ostadal P, et al. ST-segment elevation myocardial infarction treated by radial or femoral approach in a multicenter randomized clinical trial. J Am Coll Cardiol 2014;63(10):964-972.

17. Brueck M, Bandorski D, Kramer W, Wieczorek M, Holtgen R, Tillmanns H. A randomized comparison of transradial versus transfemoral approach for coronary angiography and angioplasty. JACC Cardiovasc Interv 2009;2(11):1047-1054.

18. Chodor P, Krupa H, Kurek T, Sokal A, Swierad M, Was T, et al. Radial versus femoral approach for percutaneous coronary interventions in patients with Acute Myocardial Infarction (RADIAMI): A prospective, randomized, single-center clinical trial. Cardiol J 2009;16(4):332-340.

19. Chodor P, Kurek T, Kowalczuk A, Swierad M, Was T, Honisz G, et al. Radial vs femoral approach with StarClose clip placement for primary percutaneous coronary intervention in patients with ST-elevation myocardial infarction. RADIAMI II: a prospective, randomised, single centre trial. Kardiologia Polska 2011;69(8):763-771.

20. Cooper CJ, El-Shiekh RA, Cohen DJ, Blaesing L, Burket MW, Basu A, et al. Effect of transradial access on quality of life and cost of cardiac catheterization: a randomized comparison. Am Heart J 1999;138(3 Pt 1):430-436.

21. Cruden NLM, Teh CH, Starkey IR, Newby DE. Reduced vascular complications and length of stay with transradial rescue angioplasty for acute myocardial infarction. Catheter Cardiovasc Interv 2007;70(5):670-675.

22. Dangoisse V, Guedes A, Gabriel L, Jamart J, Chenu P, Marchandise B, et al. Full conversion from transfemoral to transradial approach for percutaneous coronary intervention results in a similar success rate and a rapid reduction in in-hospital cardiac and vascular major events. Eurointervention 2013;9(3):345-352.

23. de Andrade PB, de Andrade MVA, Barbosa RA, Labrunie A, Hernandes ME, Marino RL, et al. Femoral versus radial access in primary angioplasty. Analysis of the ACCEPT Registry. Arq Bras Cardiol 2014;102(6):566-570.

24. Deftereos S, Giannopoulos G, Raisakis K, Kauoukis A, Kossyvakis C, Pappas L, et al. Transradial access as first choice for primary percutaneous coronary interventions: Experience from a tertiary hospital in Athens. Hellenic J Cardiol 2011;52(2):111-117.

25. Eyupkoca F, Cutlan B, Yasar E, Bayramotlu A, Karakus Y, Ozdemir R. Which approach for diagnostic coronary angiography in elderly patients: Radial and femoral? Turk Kardiyoloji Dernegi Arsivi 2012;40:298.

26. Genereux P, Mehran R, Palmerini T, Caixeta A, Kirtane AJ, Lansky AJ, et al. Radial access in patients with ST-segment elevation myocardial infarction undergoing primary angioplasty in acute myocardial infarction: the HORIZON-AMI trial. Eurointervention 2011;7(8):905-916.

27. Hamon M, Coste P, van't Hof A, Ten Berg J, Clemmensen P, Tabone X, et al. Impact of arterial access site on outcome after primary percutaneous coronary intervention. Circ Cardiovasc Interv 2015:8(6):e002049.

28. He P, Yang Y, Qiao S, Xu B, Yao M, Wu Y, et al. Comparison of short- and medium- term clinical outcomes between transradial approach and transfemoral approach in a high-volume PCI heart center in China. PLoS ONE 2015;10(3):e0118491.

29. Jaffe R, Hong T, Sharieff W, Chisholm RJ, Kutryk MJ, Charron T, et al. Comparison of radial versus femoral approach for percutaneous coronary interventions in octogenarians. Catheter Cardiovasc Interv 2007;69(6):815-820.

17

30. Jen HL, Yin WH, Chen KC, Feng AN, Ma SP, Cheng CF, et al. Transradial approach in myocardial infarction. Acta Cardiol 2011;66(2):239-245.

31. Jin C, Wai F. Comparison of costs between transradial and transfemoral percutaneous coronary intervention: data from one large single center in China. Eur Heart J 2014;35:1075.

32. Johnman C, Pell JP, Mackay DF, Behan M, Slack R, Oldroyd KG, et al. Clinical outcomes following radial versus femoral artery access in primary or rescue percutaneous coronary intervention in Scotland: retrospective cohort study of 4534 patients. Heart 2012;98(7):552-557.

33. Mehta SR, Jolly SS, Cairns J, Niemela K, Rao SV, Cheema AN, et al. Effects of radial versus femoral artery access in patients with acute coronary syndromes with or without ST-segment elevation. J Am Coll Cardiol 2012;60(24):2490-9.

34. Jurga J, Sarkar N, Romanitan MO, von Euler M, Lagerqvist B, Tomvall P, et al. Periprocedural stroke is not independently affected by the access site during coronary angiography and PCI. Eur Heart J 2014;35:860.

35. Louvard Y, Benamer H, Garot P, Hildick-Smith D, Loubeyre C, Rigattieri S, et al. Comparison of transradial and transfemoral approaches for coronary angiography and angioplasty in octogenarians. Am J Cardiol 2004;94(9):1177-1180.

36. Pristipino C, Trani C, Nazzaro MS, Berni A, Patti G, Patrizi R, et al. Major improvement of percutaneous cardiovascular procedure outcomes with radial artery catheterisation: results from the PREVAIL study. Heart 2009;95(6):476-482.

37. Raposo L, Madeira S, Teles RC, Santos M, Gabriel HM, Goncalves P, et al. Neurological complications after transradial or transfemoral approach for diagnostic and interventional cardiac catheterization: a propensity score analysis of 16,710 cases from a single centre prospective registry. Catheter Cardiovasc Interv 2015;86(1):61-70.

38. Madiera S, Raposo L, Goncalves P, Santos R, Teles H, Gabriel J, et al. Incidence of periprocedural stroke and transient ischemic attack after transradial or transfemoral approach for diagnostic and interventional cardiac catheterization. Eur Heart J 2013;34:P1260.

39. Rodriguez-Leor O, Fernandez-Nofrerias E, Carrillo X, Mauri J, Oliete C, Rivas Mdel C, et al. Transradial percutaneous coronary intervention in cardiogenic shock: a single-centre experience. Am Heart J 2013;165(3):280-285.

40. Romagnoli E, Biondi-Zoccai G, Siciahbasi A, Politi L, Rigattieri S, Pendenza G, et al. Radial versus femoral randomized investigation in ST-segment elevation acute coronary syndrome. J Am Coll Cardiol 2012;60(24):2481-2489.

41. Romagnoli E, de Vita M, Burzotta F, Cortese B, Biondi-Zoccai G, Summaria F, et al. Radial versus femoral approach comparison in percutaneous coronary intervention with intraaortic balloon pump support: The RADIAL PUMP UP Registry. Am Heart J 2013;166(6):1019-26.

42. Sabbah MM, Kadota K, Fuki Y, Goto T, Habara S, Tanaka H, et al. Comparative effectiveness of different arterial approaches "Transbrachial, Transradial and transfemoral" in percutaneous coronary interventions: a real-world experience. J Am Coll Cardiol 2013;62:B89.

43. Sarno G, Lagerqvist B, Frobert O, Olivecrona G, Nilsson J, Tornvall P, et al. Outcomes of radial and femoral access for coronary angiography and PCI over time: a report from the nation-wide Swedish coronary angiography and angioplasty registry. Eurointervention 2012;8N.

18

44. Sciahbasi A, Fischetti D, Picciolo A, et al. Transradial access compared with femoral puncture closure devices in percutaneous coronary procedures. Int J Cardiol 2009;137(3):199-205.

45. Tizon Marcos, Barbeau GR, Roy L, Gleeton O, Boudreault JR, Proulx GR, et al. Are neurological complications following percutaneous coronary diagnostic or interventions more common in transfemoral versus transradial approach? Eur Heart J 2009;30:873.

46. Ziakas AG, Koskinas KC, Gavrilidis S, Giannoglou GD, Hadjimiltiades S, Gourassas I, et al. Radial versus femoral access for orally anticoagulated patients. Catheter Cardiovasc Interv 2010;76(4):493-499.

47. Louvard Y, Lefevre T, Allain A, Morice M. Coronary angiography through the radial or the femoral approach: The CARAFE study. Catheter Cardiovasc Interv 2001;52(2):181-187.

48. Cantor WJ, Puley G, Natarajan MK, Dzavik V, Madan M, Fry A, et al. Radial versus femoral access for emergent percutaneous coronary intervention with adjunct glycoprotein IIb/IIIa inhibition in acute myocardial infarction--the RADIAL-AMI pilot randomized trial. Am Heart J 2005;150(3):543-549.

49. Hamon M, Gomes S, Oppenheim C, Morello R, Sabatier R, Lognone T, et al. Cerebral microembolism during cardiac catheterization and risk of acute brain injury: a prospective diffusion-weighted magnetic resonance imaging study. Stroke 2006;37(8):2035-2038.

50. Busing KA, Schulte-Sasse C, Fluchter S, Fuselbeck T, Haase KK, Neff W, et al. Cerebral infarction: incidence and risk factors after diagnostic and interventional cardiac catheterization--prospective evaluation at diffusion-weighted MR imaging. Radiology 2005;235(1):177-183.

51. Omran H, Schmidt H, Hackenbroch M, Illien S, Bernhardt P, von der Recke G, et al. Silent and apparent cerebral embolism after retrograde catheterisation of the aortic valve in valvular stenosis: a prospective, randomised study. Lancet 2003;361(9365):1241-1246.

52. Vermeer SE, Prins ND, den Heijer T, Hofman A, Koudstaal PJ, Breteler MM. Silent brain infarcts and the risk of dementia and cognitive decline. N Engl J Med 2003;348(13):1215-1222.

53. Khoury Z, Gottlieb S, Stern S, Keren A. Frequency and distribution of atherosclerotic plaques in the thoracic aorta as determined by transesophageal echocardiography in patients with coronary artery disease. Am J Cardiol 1997;79(1):23-27.

19

Table 1: Study design, participants and clinical settingStudy ID Design; Country,

YearNo. of participants; radial; femoral.

Age % Male Clinical setting

Achenbach 200814

RCT; Germany; Unclear.

307; 152; 155. 78 45 Participants had in-patient invasive angiography of age at least 75 years.

Benit 199715 RCT; Belgium; 1994-1995.

105; 50; 55. 58 56 Elective angioplasty with the Palmaz-Schatz stent.

Bernat 201416 RCT; International; 2009 to 2012.

707; 348; 359. 62.1 77 Participants had STEMI in STEMI-RADIAL trial.

Brueck 200917 RCT; Germany; 2006-2008.

1,024; 512; 512. 63.6 58.7 Patients with diagnostic or interventional cardiac catheterization

Chodor 200918 RCT; Poland; 2005-2006.

100; 50; 50. 59.5 68 Participants had STEMI in RADIAMI trial.

Chodor 201119 RCT; Poland; 2006-2008.

108; 49; 59. 59.6 64 Participants had STEMI in RADIAMI II trial.

Cooper 199920 RCT; USA; 1996-1997.

200; 101; 99. 60 69 Participants underwent diagnostic cardiac catheterization.

Cruden 200721 Retrospective cohort study; UK; 2003-2005.

287; 44; 243. 59 82.9 STEMI patients undergoing rescue PCI.

Dangoisse 201322

Prospective cohort study; France; 2002-2007.

3,600; 1,672; 1,928.

65 76 Participants had primary PCI.

De Andrade 201423

Prospective cohort study; Brazil; 2010 -2011.

588; 178; 410. 61.8 75.2 Participants had STEMI in ACCEPT registry.

Deftereos 201024

Retrospective cohort study; Greece; 2009-2010.

98; 65; 33. 64 74.4 Participants had STEMI and underwent primary PCI.

20

Eyupkoca 201225

Cohort study; Turkey; 2010-2012.

450; 264; 186. NA NA Participants were elderly undergoing coronary angiography.

Genereux 201126

RCT; International; 2005-2007.

3,334; 200; 3,134.

60 77 Participants had STEMI in HORIZON-AMI trial.

Hamon 20129 RCT; France; Unclear.

160; 83; 77. 74.5 54.3 Participants had cardiac catheterization and had aortic stenosis.

Hamon 201527 RCT; International; Unclear.

2,152; 1,012; 1,140.

Median 60 and 63.

69 Participants had STEMI in EUROMAX trial.

He 201528 Retrospective cohort study; China; 2006-2011.

2,728; 1,364; 1,364.

59 68 Participants had PCI.

Jaffe 200729 Prospective cohort study; Canada; 2000-2004,

228: 97; 131 82.5 59.2 Octogenarians undergoing elective PCI.

Jen 201130 Retrospective cohort study; China; 2005-2007.

117: 85; 37. 66 76.9 Patients with STEMI undergoing primary PCI.

Jin 201031 Retrospective cohort study; China; 2010.

5,539; 4,888; 651.

NA NA Participants had PCI.

Johnman 201232 Retrospective cohort study; Scotland; 2000-2009.

4,534; 2,392; 2,142.

NA 75 Participants had primary or rescue PCI in the Scottish Coronary Revascularisation Register.

Jolly 20117 & Mehta 201233

RCT; International; 2006-2010.

7,021; 3507; 3,514.

62 73 Participants had ACS in RIVAL trial.

Jurga 20115 RCT; Sweden; 2007-2008.

43; 20; 23. 64 86 Participants had coronary angiography.

Jurga 201434 Cohort study; Sweden; 2003 to 2010.

259,045;77,155; 151,086.

66 66 Participants had coronary angiography or PCI in the SCAAR registry.

Kwok 20151 Retrospective cohort 403,646; 65 73 Participants had PCI in the BCIS dataset.

21

study; UK; 2007-2012.

177,458; 226,188.

Louvard 200435 RCT; International; 2001-2003.

377; 192; 185. 83 53 Participants had coronary angiography or PCI and were octogenarians in OCTOPLUS study.

Lund 20053 Prospective cohort study; Norway; Unclear.

42; 33; 9. 59.3 78.7 Participants underwent elective left heart catheterization.

Pristipino 200936

Prospective cohort study; Italy; 2006.

1052; 509;543 66 71 Participants in the PREVAIL study of primary PCI.

Rapsoso 201537 & Madiera 201338

Prospective cohort study; Portugal; 2006- 2012

16,710; 4,231; 12,479.

66 66.6 Participants underwent diagnostic or interventional cardiac catheterization.

Rodriguez-Leor 201239

Prospective cohort study; Spain; 2007-2012.

122; 80; 42. 66 83.6 Participants had cardiogenic shock and were undergoing PCI.

Romagnoli 201240

RCT; International; 2009-2011

1001; 500; 501. 65 73.3 Participants had early revascularization due to STEACS, participants of the RIFLE STEACS trial.

Romagnoli 201341

Retrospective cohort study; Italy; 2007-2011.

321; 112; 209. 70 72.9 Participants had intraaortic balloon pump and were undergoing PCI.

Sabbah 201342 Retrospective cohort study; Japan; 2006-2012.

3,851; 2,797; 1,054.

NA NA Participants underwent PCI.

Sarno 201243 Cohort study; Sweden; 2003-2010.

299,786; 101,060; 198,726.

NA NA Participants underwent coronary angiography or PCI in the Swedish coronary angiography and angioplasty registry (SCAAR).

Sciahbasi 200944 Prospective cohort study; Italy; 2006 to 2007.

1,492; 604; 888. 67 67 Participants underwent PCI with radial access or femoral puncture with closure device.

Tizon Marcos 200945

Retrospective cohort study; Canada;

83,409; 52,191; 31,218.

NA NA Participants underwent coronary angiography or PCI.

22

1990-2007.Valgimigli 20158 RCT; International;

2011-2014.8,404; 4,197; 4,207.

66 73.4 Participants had ACS in the MATRIX trial.

23

Table 2: Study quality assessmentStudy ID RCT or

prospective design.

Sample size ≥100 participants in each arm.

Reliable ascertainment of outcomes.

Low lost to follow up.

No more than 3 baseline differences between groups (RCT) or use of adjustments (cohorts).

Achenbach 200814

RCT Yes Yes, after intervention patients were contacted again by clinical visit or telephone interview.

Yes, no loss to follow up.

Yes, RCT, difference in time to cine angiographic acquisition and duration of diagnostic angiography.

Benit 199715 RCT No Yes, follow up in outpatient clinic at 1 month.

Unclear. Yes, RCT, difference in maximal pressure exerted by balloon and arterial time of the procedure.

Bernat 201416 RCT Yes Yes, clinical follow up for 30 days, and telephone follow-up for 6 months.

Yes, no loss to follow up.

Yes, RCT, difference in hypertension and BMI only.

Brueck 200917 RCT Yes Unclear methods but prospective collection.

Yes, no loss to follow up.

Yes, RCT, difference in access failure, median procedural time and median fluoroscopy time.

Chodor 200918 RCT No Unclear methods but prospective collection.

Yes, no loss to follow up.

Yes, RCT, no difference in baseline characteristics.

Chodor 201119 RCT No Unclear methods but prospective collection.

Yes, no loss to follow up.

Yes, RCT, difference in age.

Cooper 199920 RCT No Yes, 1 day and 1 week physical examination was performed.

Yes, no loss to follow up.

No, RCT, difference in male sex, aortic valve disease, hypertension and smoking.

Cruden 200721 Retrospective No Yes, data from interventional database, hospital records and catheter

Unclear. No, cohort, unadjusted results.

24

laboratory records.Dangoisse 201322

Prospective Yes Yes, procedural, clinical and laboratory data entered into PCI database using hospital's electronic medical record.

Unclear. No, cohort, unadjusted results.

De Andrade 201423

Prospective Yes Yes, data from electronic system.

Yes, 52 cases 8.1% loss to follow up.

No, cohort, unadjusted results.

Deftereos 201024

Retrospective Yes No, retrospective review of clinical and angiographic data.

Yes, no loss to follow up.

No, cohort, unadjusted results.

Eyupkoca 201225

Unclear Yes Unclear Unclear. No, cohort, unadjusted results.

Genereux 201126

RCT Yes Yes, clinical events were adjudicated by blinded events committee.

Unclear. No, RCT, difference in history of hyperlipidemia, aspirin, thienopyridine, Ca2+ channel blocker, Abciximab, Eptifibatide, given with Bivalirudin for bail out, and ACEi or ARB on discharge.

Hamon 20129 RCT No Yes, MRI scan within 24 hours before and 48 hours after and some patients had transcranial Doppler.

Yes, 5 patients refused 2nd MRI or MRI no interpretable.

Yes, RCT, difference in catheters used >3 and anticoagulation per catheter.

Hamon 201527 RCT Yes Yes, events were adjudicated by independent, blinded clinical events

Yes, 20 patients no included in intention-to-treat population.

No, RCT, difference in P2Y12 inhibitor loading dose/maintenance dose, time from initiation of anticoagulation

25

committee. to angiography, left main stem disease, right coronary artery disease, left circumflex disease, drug eluting stent, thrombectomy, pre-PCI TIMI flow, CABG during hospitalization, aspirin on discharge, P2Y12 inhibitor on discharge, ACEi/ARB on discharge.

He 201528 Retrospective Yes Yes, death from death certificates.

Yes, 328 loss to follow up 1.5%.

Yes, propensity matched analysis.

Jaffe 200729 Prospective No Yes, clinical data from database confirmed by chart review.

Unclear. No, cohort, unadjusted results.

Jen 201130 Retrospective No Yes, endpoints from outpatient visits.

Unclear. No, cohort, unadjusted results.

Jin 201031 Retrospective Yes Unclear. Unclear. No, cohort, unadjusted results.Johnman 201232

Retrospective Yes Yes, data from Scottish Morbidity Record and the General Registrar for Scotland.

Unclear. Yes, cohort, adjusted for age, smoking, diabetes, hypertension, non-cardiac valvular disease, renal impairment, left ventricular dysfunction, cardiogenic shock, comorbidity, disease severity, deprivation quintile, indication for PCI, year of procedure and hospital.

Jolly 20117 & Mehta 201233

RCT Yes Yes, outcomes adjudicated by masked central committee.

Yes, 2 loss to follow up.

Yes, RCT, difference in arterial sheath size and number of diagnostic catheter used.

Jurga 20115 RCT No Yes, multifrequency Yes, no loss to No, RCT, difference in diabetes

26

Doppler system. follow up. mellitus, number of coronary stenosis, fluoroscopy time, number of particulate microemboli, right MCA.

Jurga 201434 Unclear Yes Yes, two neurologist evaluated neurological complications.

Yes, 12 cases unverifable.

Yes, cohort, adjusted for age, gender, previous CABG, previous stroke, diabetes, vascular access site.

Kwok 20151 Retrospective Yes No, data from BCIS database for in-hospital events.

Yes, 12,103 loss to follow up 2.8%.

No, cohort, unadjusted results.

Louvard 200435

RCT Yes Unclear methods for data collection.

Yes, no loss to follow up reported.

Yes, RCT, difference in diabetes and previous PCI.

Lund 20053 Prospective No Yes, Transcranial Doppler and clinical examination performed by neurologist and MRI.

Yes, no loss to follow up reported.

No, cohort, unadjusted results.

Pristipino 200936

Prospective Yes Yes, data prospectively collected with case report form which was transferred to database. Data was cross checked for consistency by database manager.

Yes no loss to follow up reported.

No, cohort, unadjusted results.

Rapsoso 201537 & Madiera 201338

Prospective Yes Yes, data from institutional database and clinical files from each case were

Yes, no loss to follow up reported.

Yes, cohort, adjusted for baseline characteristics.

27

reviewed by clinician with preference for neurologist.

Rodriguez-Leor 201239

Prospective No Yes, prospective data collection for the patients' clinical evolution.

Yes, no loss to follow up reported.

No, cohort, unadjusted results.

Romagnoli 201240

RCT Yes Yes, end points were adjudicated by blinded central independent clinical events-committee.

Yes, no loss to 1 month follow up.

Yes, RCT, difference in artery puncture-to-balloon time and arterial sheath size.

Romagnoli 201341

Retrospective Yes Yes, retrospective from electronic database and checking of hospital charts and direct visit.

Yes, no loss to follow up reported.

No, cohort, unadjusted results.

Sabbah 201342 Retrospective Yes Unclear. Unclear. No, cohort, unadjusted results.Sarno 201243 Unclear Yes Unclear, data from

SCAAR registry.Unclear. Yes, cohort, adjusted for

differences in clinical, vessel, lesion characteristics, year of procedure and enrolling center.

Sciahbasi 200944

Prospective Yes Unclear. Unclear. No, cohort, unadjusted results.

Tizon Marcos 200945

Retrospective Yes No, retrospective review of charts.

Unclear. No, cohort, unadjusted results.

Valgimigli 20158

RCT Yes Yes, all outcomes were adjudicated by blinded and masked committee.

Yes, 30 patients lost to follow up or refused follow up.

Yes, RCT, difference in unfractionated heparin in catheter lab, left main coronary artery disease, bypass graft.

28

Table 3: Study outcomes and resultsStudy ID Participants;

DesignOutcome and

follow upResults

Achenbach 200814 In-patient invasive angiography of age at least 75 years; RCT.

Stroke at up to 96 hours post-procedure.

Stroke: radial group 0/152 vs 1/155.

Benit 199715 Elective PCI for symptomatic IHD; RCT.

In-hospital stroke In-hospital stroke: radial group 0/50 vs femoral group 1/55.

Bernat 201416 STEMI; RCT. Peri-PCI stroke, 48h stroke, 30 day stroke.

Peri-PCI stroke: radial group 0/348 vs femoral group 1/359.48h stroke: radial group 1/348 vs femoral group 1/359.30 day stroke: radial group 1/348 vs femoral group 1/359.

Brueck 200917 Patients receiving diagnostic or interventional cardiac catheterization; RCT.

Cerebrovascular accidents

Peri-PCI stroke: radial group 0/512 vs femoral group 2/512.

Chodor 200918 STEMI, RCT. In-hospital stroke. In-hospital stroke: radial group 0/50 vs femoral group 1/50.Chodor 201119 STEMI; RCT. In-hospital stroke. In-hospital stroke: radial group 0/49 vs femoral group 1/59.Cooper 199920 Diagnostic cardiac

catheterization; RCT.Peri-procedural stroke.

Peri-procedural stroke: radial group 0/101 vs femoral group 1/99.

Cruden 200721 Rescue PCI for AMI; Retrospective cohort study.

In-hospital stroke. In-hospital stroke: radial group 0/44 vs femoral group 3/243.

Dangoisse 201322 PCI; Prospective cohort study.

In-hospital disabling stroke.

In-hospital disabling stroke: radial group 5/1,672 vs femoral group 3/1,928.

De Andrade 201423 STEMI; Prospective cohort study.

6 month stroke. 6 month stroke: radial group 0/178 vs femoral group 3/410.

Deftereos 201024 STEMI primary PCI; Retrospective cohort

In-hospital stroke. In-hospital stroke: radial group 2/65 vs femoral group 2/33.In-hospital intracranial bleeding; radial group 1/65 vs femoral group

29

study. 1/33.In-hospital ischemic CVA: radial group 1/65 vs femoral group 1/33.

Eyupkoca 201225 Coronary angiography in elderly; Cohort study.

Stroke unclear follow up.

Stroke: radial group 0/254 vs femoral group 1/186.

Genereux 201126 STEMI; RCT. 1 year stroke. 1 year stroke: radial group 1/200 vs femoral group 30/3,134.Hamon 20129 Aortic stenosis with

coronary catheterization; RCT.

Cerebral infarct at 48 hours after catheterization.

ITT: Cerebral infarct: radial: 15/83 vs femoral: 9/77.

Hamon 201527 STEMI; RCT. 30 day stroke. 30 day any stroke: radial group 3/1,012 vs femoral group 13/1,140.30 day ischemic stroke: radial group 3/1,012 vs femoral group 11/1,140.30 day hemorrhagic stroke: radial group 0/1,012 vs femoral group 2/1,140.

He 201528 Elective PCI; Retrospective cohort study.

In-hospital stroke. In-hospital stroke propensity matched: radial group 1/1,634 vs femoral group 3/1,634.

Jaffe 200729 Primary PCI in octogenarians; Prospective cohort study.

In-hospital stroke In-hospital stroke: radial group 1/97 vs femoral group 3/131.

Jen 201130 Primary PCI, Retrospective cohort study.

Peri-procedural intracranial hemorrhage.

Peri-procedural intracranial hemorrhage: radial group 0/85 vs femoral group 1/37

Jin 201031 PCI; Retrospective cohort study.

In-hospital stroke. In-hospital stroke: radial group 0/508 vs femoral group 3/508.

Johnman 201232 PCI; Retrospective cohort study.

30 day and 1 year stroke.

Fatal or non-fatal stroke at 30 day: radial vs femoral OR 0.55 (0.18-1.66), at 1 year OR 0.29 (0.1-0.81).

Jolly 20117 & Mehta 201233

ACS; RCT. 48h stroke,30 day stroke.

48h stroke: radial group 7/3507 vs femoral group 6/3514 OR 1.17 (0.39-3.48)

30

30 day stroke: radial group 20/3,507 vs femoral group 14/3,514, OR 1.43 (0.73-2.80).30 day stroke in STEMI: radial group 5/955 vs femoral group 4/1,003, OR 1.30 (0.35-4.84).30 day stroke in NSTEMI: radial group 15/2,552 vs femoral group 10/2,511, OR 1.48 (0.67-3.30).

Jurga 20115 Stable angina; RCT. Number of emboli on transcranial Doppler.

Number of particulate microemboli in radial and femoral group were 10.9±6.3 (10 (1-120)) and 6.9±4.7 (6 (1-19)), respectively.

Jurga 201434 Coronary angiogram and PCI; Cohort study.

Periprocedural neurological complication.

Periprocedural neurological complications for radial group 77/77,155 vs 302/151,086. OR 0.94 (0.75-1.17).

Kwok 20151 PCI; Retrospective cohort study.

In-hospital stroke. In-hospital stroke: radial group 222/177,458, femoral group 288/226,188.

Louvard 200435 Coronary angiography and PCI; RCT.

Cerebrovascular accident.

Cerebrovascular accident: radial group 0/192 vs 1/185.

Lund 20053 Elective catheterization; Prospective cohort study.

New cerebral DWI lesions.

New cerebral lesions: radial group 5/33 vs 0/9.

Pristipino 200936 Primary PCI in PREVAIL study; Prospective cohort study.

Peri-procedural stroke

Peri-procedural stroke: radial group 0/509 vs femoral group 1/543

Rapsoso 201537 & Madiera 201338

Diagnostic and interventional procedures; Prospective cohort study.

Stroke or TIA 48 hours after the procedure.

Stroke or TIA: radial group 6/4,195 vs femoral group 20/12,515, adjusted OR 0.86 (0.27-2.7).

Rodriguez-Leor 201239

Emergency PCI and cardiogenic shock;

In-hospital stroke. In-hospital stroke: radial group 3/80 vs femoral group 2/42.

31

Prospective cohort study.

Romagnoli 201240 Suspected STEACS planned for early revascularizaton; RCT.

30 day stroke 30 day stroke: radial group: 4/500 vs femoral group 3/501Cerebral bleeding: radial group: 1/500 vs femoral group 0/501

Romagnoli 201341 PCI with intraaortic balloon pump; Retrospective cohort study.

30 day stroke. 30 day stroke: radial group 4/112 vs femoral group 4/209.

Sabbah 201342 PCI; Retrospective cohort study.

Stroke. Stroke: radial group 3/2,797 vs femoral group 5/1,054.

Sarno 201243 Coronary angiogram and PCI; Cohort study.

In-hospital neurological complications.

In-hospital neurological complication: radial group 202/101,060 vs femoral group 397/198,726, OR 1.22 (0.91-1.64).

Sciahbasi 200944 PCI; Prospective cohort study.

Stroke. Stroke: radial group 1/604 vs femoral group 0/888.TIA: radial group 2/604 vs femoral group 2/888.

Tizon Marcos 200945 Coronary angiogram and PCI; Retrospective cohort study.

Non-hemorrhagic and hemorrhagic stroke.

Stroke: radial group 71/52,191 vs femoral group 66/31,218.Ischemic stroke: radial group 61/52,191 vs femoral group 63/31,218.Hemorrhagic stroke: radial group 10/52,191 vs femoral group 3/31,218.

Valgimigli 20158 ACS; RCT. Stroke, ischemic and hemorrhagic at 30 days.

30 day stroke: radial group 16/4,197 vs femoral group 16/4,207.30 day ischemic stroke: radial group 12/4,197 vs femoral group 11/4,207.30 day hemorrhagic stroke: radial group 3/41,97 vs femoral 5/4,207.

32

Figure 1: Flow diagram of study selection

33

Figure 2: Risk of stroke with radial and femoral access by study design

34

Figure 3: Risk of ischemic stroke with radial and femoral access by study design

35

Figure 4: Risk of hemorrhagic stroke with radial and femoral access by study design

36

Figure 5: Funnel plot for publication bias for the analysis of the risk of stroke with radial and

femoral access by study design

37

Supplementary Figure 1: Stroke rates by year of publicationRCT only

38

Cohort studies only

RCT or Cohort studies

39

Supplementary Figure 2: Risk of stroke with radial and femoral access by study design and procedure type

40