t h e s u r g e on 9 ( 2 0 1 1 ) 2 4 5e2 4 8

avai lable at www.sciencedirect .com

The Surgeon, Journal of the Royal Collegesof Surgeons of Edinburgh and Ireland

www.thesurgeon.net

Wound infection after reconstructive arterial surgeryof the lower limbs: Risk factors and consequences

Rachel O’Brien, Nicholas Pocock, Francesco Torella*

Department of Vascular Surgery, University Hospital Aintree, Lower Lane, Liverpool, L9 7AL, United Kingdom

a r t i c l e i n f o

Article history:

Received 8 September 2010

Received in revised form

7 October 2010

Accepted 9 October 2010

Available online 13 November 2010

Keywords:

Wound infection

Peripheral vascular disease

Vein grafts

* Corresponding author. Tel.: þ44 151 529 49E-mail address: f.torella@liv.ac.uk (F. Tor

1479-666X/$ e see front matter ª 2010 RoyalSurgeons in Ireland. Published by Elsevier Ldoi:10.1016/j.surge.2010.10.005

a b s t r a c t

Objective: To identify predictors, and clinical consequences, of postoperative wound

infection after peripheral vascular surgery.

Materials and methods: Retrospective cohort study. Potential predictors of wound infection

were sought among patient related factors and procedure related factors. Patient outcome

was then analysed according to the incidence of wound infection.

Results: Following 209 procedures, 20 (9.6%) patients suffered a wound infection. On

univariate analysis, infrainguinal surgery, use of vein graft, and tissue loss were associated

with wound infection. On multivariate regression, however, only the association with use

of a vein graft remained (OR ¼ 4.2; 95%CI ¼ 1.2e14.1; P ¼ .022): the incidence of wound

infection was 16/96 (17%; 95%CI ¼ 9.2e24.1) when a vein graft was used and 4/113 (3.5%;

95%CI ¼ .1e7%) when a prosthetic or no graft was used. Wound infection was associated

with increased mortality (4/20 versus 9/189; P ¼ .025) but not limb loss (2/20 versus 7/189;

P ¼ .208). Median (IQR) postoperative hospital stay in patients with wound infection was 22

(15e45) days and 8 (5e15) days in those without wound infection (P ¼ .001).

Conclusions: In lower limb arterial surgery, wound infection is associated with the use of

vein grafts, and results in delayed recovery and increased mortality. Measures to reduce

wound infection should be focussed on such patients.

ª 2010 Royal College of Surgeons of Edinburgh (Scottish charity number SC005317) and

Royal College of Surgeons in Ireland. Published by Elsevier Ltd. All rights reserved.

Introduction used.3 Among patients undergoing lower limb arterial surgery,

Surgical wounds can be classified according to the likelihood

of bacterial contamination and, consequently, of wound

infection (WI). A clean wound implies lack of infection within

the surgical field, no entry into the respiratory, urogenital or

gastrointestinal tracts, and absence of external contamina-

tion: in these circumstances, the risk of WI is very low.1

Surgery on the arterial tree of the lower limbs is an excep-

tion, with a significantly higher incidence of WI than the

average clean wound,2 so much so that routine antibiotic

prophylaxis is indicated even when prosthetic material is not

57; fax: þ44 151 529 6457.ella).College of Surgeons of Ed

td. All rights reserved.

increasedWI risk has been linked with several factors such as

sex, diabetes, presence of tissue loss, previous surgery,

surgery on the lower extremities, use of vein grafts and groin

incision, with little agreement among published studies.4e8

Presently, it is thus very difficult to identify, preoperatively,

high risk patients.

The principal aim of this studywas to identify preoperative

risk factors for WI in patients undergoing reconstructive

arterial surgery of the lower limbs. The secondary aim was to

detect the potential influence of WI on overall patient

outcome, measured by mortality, limb loss and hospital stay.

inburgh (Scottish charity number SC005317) and Royal College of

Table 2 e Grafts.

n %

Ipsilateral reversed GSV 64 30.6

In situ GSV 6 2.9

Contralateral GSV 3 1.4

Arm vein 19 9.1

Composite vein 3 1.4

Anterolateral thigh vein 1 .5

PTFE 11 5.3

Dacron 87 41.6

Composite PTFE-vein 1 .5

Composite Dacron-vein 10 4.8

None 4 2

GSV ¼ greater saphenous vein.

t h e s u r g e on 9 ( 2 0 1 1 ) 2 4 5e2 4 8246

Methods

This was a retrospective cohort study on consecutive surgical

procedures performed over a five year period (2004e2009)

under the care of a single specialist surgeon in a tertiary centre

serving, largely, a deprived urban population. All patients

undergoing surgery for occlusive or aneurysmal disease below

the level of the renal arterieswere included,with the exception

of aortic or aortoiliac aneurysm repairs. Further exclusions

were non reconstructive procedures (such as thromboembo-

lectomy), amputations, endovascular procedures, vascular

access procedures and operations performed through non-

clean wounds. All wounds were closed in layers with poly-

dioxanone or polycaproic acid sutures for the deep layers and

subcuticular polycaproic acid sutures for skin. Data on each

procedure, including outcome, were collected prospectively at

the time of discharge or death and entered in a dedicated

database (SPSS inc., Chicago, Ill, USA).

The main endpoint was a clearly documented diagnosis of

WI in the casenotes and/or electronic records during the

postoperative hospital stay and/or at the 1st postoperative

review (at six weeks), leading to antibiotic treatment with or

without surgical debridement, with or without confirmation

by positive blood or wound cultures.

Potential predictors of WI were selected a priori, in light of

previously published evidence, among patient related factors

(diabetes, sex, presence of tissue loss) and procedure related

factors (use of vein graft, groin dissection, redo operation, type

of surgery).4e8 Variableswith a possible associationwithWI on

univariate tests (P< .2) were entered in amultivariatemodel to

identify independent predictors.

The impact of WI on clinical outcome was investigated by

comparing mortality, limb loss rates and hospital stay in

patients with and without the complication.

Statistical analysis

Continuous variables were describedwithmean and standard

deviation (SD) if normally distributed, and median and quar-

tiles (IQR) if skewed; thesewere comparedwith the unpaired t-

test and ManneWhitney U test respectively. Categorical vari-

ables were compared with the c2 test, with a continuity

Table 1 e Comorbidity.

n %

Ischaemic heart disease 72 35.6

Left ventricular failure 25 12

Atrial fibrillation 31 14.8

Chronic obstructive pulmonary disease 57 27.3

Asthma 5 2.4

Pulmonary fibrosis 2 1

Diabetes 45 21.4

diet controlled 5 2.4

on tablets 29 13.9

on insulin 11 5.3

Hypertension 129 61.7

Cerebrovascular disease 46 22

Renal impairment 17 8.1

correction for 2� 2 tables, or the Fisher’s Exact test when cell

numbers were low. Logistic regression was used to identify

independent predictors of WI: in this analysis, the occurrence

of WI was used as the dependent variable, and potential risk

factors as covariates. For each of these, the odds ofWI and 95%

confidence intervals were calculated. Analysis was performed

with SPSS 16.0 (SPSS inc., Chicago, Ill., USA).

Results

Two hundred and nine procedures were included in the study,

of which 107 were elective and 102 unplanned. Therewere 136

men and 73 women with a mean (SD) age of 70 (11) years,

undergoing surgery for critical ischaemia (143), Claudication

(45), peripheral aneurysms (10) or asymptomatic graft stenosis

(3). Seventy-four patients had major tissue loss. Patients’

comorbidity was typical of patients with arterial disease

(Table 1).

Surgery was performed under general anaesthesia in 191

cases (combined with epidural anaesthesia in 19), regional

anaesthesia in 12 and local infiltration in 6. There were 34

“proximal” reconstructions (aortofemoral or ilio-femoral), 27

extraanatomical bypasses (femoro-femoral, ilio-femoral cross-

over or axillo-femoral), 146 infrainguinal procedures, one

combined extraanatomical and extra-anatomical procedure

and one combined proximal and infrainguinal procedure.

Table 3 e Culture results.

n

Staphylococcus aureus 7

MRSA 5

Diphteroids 2

Streptococcus (b-haemolytic) 1

Pseudomonas aeruginosa 1

Klebsiella 1

Enterobacter 1

No organism growth 5

MRSA: Methycillin Resistant Staphylococcus aureus. In three

patients, cultures showed more than one organism: Klebsiella and

Enterobacter (1), S. aureus and Streptococcus (1), MRSA and Diphte-

roids (1).

Table 4 e Factors associated with wound infection onunivariate analysis.

n WI incidencen (%)

P

Patient related factors

Diabetes 45 4 (8.9) 1

Tissue loss 74 11 (14.9) .093

Female sex 73 6 (8.2) .623

Procedure related factors

Groin incision 193 20 (10.3) .374

Redo procedure 42 6 (14.3) .385

Use of vein graft 96 16 (16.7) .003

Infrainguinal surgery 162 19 (11.7) .051

t h e s u r g e on 9 ( 2 0 1 1 ) 2 4 5e2 4 8 247

Seventy patients hadundergone previous vascular surgery, and

42underwent “re-do” procedures, definedas surgery performed

through a previous scar. In 20 cases, patients underwent foot

debridement and/or minor amputation at the same time of the

arterial reconstruction and, in all but 16, surgery involved

a groin incision. Autologous or prosthetic graftswere used in all

but four procedures (Table 2). All patients received antibiotic

prophylaxis with a single dose of a second generation cepha-

losporin andmetronidazole (or with a penicillin plus clavulanic

acid) at induction, repeated in case of prolonged surgery (>3 h)

or large blood loss (>1 L, estimated).

Wound infections and clinical outcome

Twenty patients suffered a WI (9.6%; 95%CI ¼ 5.58e13.56),

which was caused by staphylococci in more than half of the

cases (Table 3). One of these patients also had the only early

graft infection by Staphylococcus aureus (Dacron graft). Univar-

iate analyses suggested that infrainguinal surgery, use of vein

grafts, and tissue loss had a potential association with the

occurrence of WI (Table 4). On multivariate analysis however,

the only variablewith a significant associationwithWIwas the

use of vein grafts (OR ¼ 4.2; 95%CI ¼ 1.2e14.1; P ¼ .022): the

incidence of WI was 16/96 (17%; 95%CI ¼ 9.2e24.1) in proce-

dures where a vein graft was used and 4/113 (3.5%; 95%

CI ¼ .1e7%) in those where prosthetic or no graft was used.

Among those patientswith vein graftswho suffered aWI, nine

had ipsilateral greater saphenous vein grafts (1 in in situ, eight

reversed), three had arm vein grafts, two had composite vein

grafts and one had a composite Dacron-vein graft. Tissue loss

(OR ¼ 1.9; 95%CI ¼ .74e5.1, P ¼ .176) and infrainguinal surgery

(OR ¼ 2.1; 95%CI ¼ .23e19, P ¼ .519) did not predict WI.

Patients who suffered a WI had a significantly higher

mortality (4/20 versus 9/189; P ¼ .025) and their median (IQR)

hospital stay was 22 (15e45) days, much longer than in

patients without this complication, at 8 (5e15) days (P ¼ .001).

However,WI did not appear to increase the risk of limb loss (2/

20 versus 7/189, P ¼ .208).

Discussion and conclusions

Our study confirms that patients undergoing surgery for

peripheral arterial disease are at high risk ofWI; however, this

risk appears to be largely confined to those receiving vein

grafts, as the incidence of WI after reconstructions with pros-

thetic (or no) grafts was compatible with that of other clean

surgical procedures.1 If confirmed, this finding would prove

invaluable to vascular surgeonsandpatients in several areasof

clinical management such as the informed consent process

and the choice of most appropriate treatment modality. It

would also provide a “target” populationwith a high incidence

of this complication. As previously suggested,4 the study also

showed that patients with WI have a higher mortality than

those without the complication, and that they have a signifi-

cantly longer postoperative hospital stay, thus indicating that

a reduction inWI rates would be desirable both on clinical and

financial grounds, to allow reallocation of hospital resources.

In agreement with other authors,2 we found that the

majority of WIs were caused by staphylococci, which are often

carried by thepatient. As carriers of staphylococci have ahigher

risk of wound infection,9 future attempts at decreasing WI

infection rates in arterial surgery should be directed towards

these micro-organisms.

Our findings arenot in complete agreementwithpreviously

published literature. Available evidence, however, is found in

a very heterogeneous group of studies, investigating different

variables, using various definitions of WI, including diverse

groups of patients in different settings.2e8 Furthermore,

several studies included patients treated over a decade ago, at

a time when clinical practice would have been different from

current standards2,5,7: such heterogeneicity makes interpre-

tation of the findings difficult. In particular, it may seem

counterintuitive that WI may occur more frequently when

autologous tissue is used. In general, however, vein harvest

increases the number of wounds, prolongs operative time and

interrupts larger numbers of subcutaneous lymphatic vessels,

all potential causes of increased risk of WI; these possible

associations cannot, obviously, be confirmed by our study.

The strength of our findings is mitigated by several factors,

one being the inclusiveness of the analysis. Our data were not

collected specifically for this study, therefore some variables

with potential associationwithWI (for example obesity7) were

not included. Some postoperative factors (for example, the

occurrence of wound haematoma2) have also been linkedwith

WI, but we chose to investigate only preoperative variables, as

only these can be used preoperatively as predictors. The

retrospective nature of the study and its setting (single

surgeon’s practice, specialist unit, urban hospital) are further

limitations, the latter suggesting that our results may not

necessarily be reproducible elsewhere. Routine preoperative

screening for Staphylococcus aureus and MRSA would also have

added valuable information but, unfortunately, this was not

carried out throughout the study period.

With the exception of prophylactic antibiotic therapy,

there is no single measure proven to decrease WI rates in

peripheral arterial surgery. Several interventions have been

investigated by randomised controlled trials in other settings,

including preoperative eradication of nasal Staphylococci,9,10

high flow oxygen11 and skin preparation with chlorhexidine

in alcohol,12 with mixed results. Further studies in arterial

surgery are obviously necessary, regardless of the type of

grafts used. Our results, however, suggest that the potential

t h e s u r g e on 9 ( 2 0 1 1 ) 2 4 5e2 4 8248

effect of any positive intervention would be easier to

demonstrate in patients receiving vein grafts.

r e f e r e n c e s

1. Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR,Emori TG, et al. Surgical wound infection rates by woundclass, operative procedure, and patient risk index. Am J Med1991;91:152Se7S. Natl Nosocomial Infections Surveill System.

2. Lee ES, Santilli SM, Olson MM, Kuskowski MA, Lee JT. Woundinfection after infrainguinal bypass operations: multivariateanalysis of putative risk factors. Surg Infect 2000;1:257e63.

3. Stewart AH, Eyers PS, Earnshaw JJ. Prevention of infection inperipheral arterial reconstruction: a systemiatic review andmeta-analysis. J Vasc Surg 2007;46:148e55.

4. Nguyen LL, Brahmanandam S, Bandyk DF, Belkin M,Clowes AW, Moneta GL, et al. Female gender and oralanticoagulants are associated with wound complications inlower extremity vein bypass: Analysis 140 Operations forCritical Limb Ischemia. J Vasc Surg 2007;46:1191e2119.

5. Richet HM, Chidiac C, Prat A, Pol A, David M, Maccario M, et al.Analysis of risk factors for surgical wound infectionsfollowing vascular surgery. Am J Med 1991;91:170Se2S.

6. Ploeg A, Lange C, Lardenoye JW, Breslau P. Nosocomialinfections after peripheral arterial bypass surgery. World JSurg 2007;31:1687e92.

7. Chang JK, Calligaro KD, Ryan S, Runyan D, Dougherty MJ,Stern JJ. Risk factors associated with infection of lowerextremity revascularization: analysis of 365 proceduresperformed at a teaching hospital. Ann Vasc Surg 2003;17:91e6.

8. Bandyk DF. Vascular surgical site infection: risk factors andpreventive measures. Semin Vasc Surg 2008;21:119e23.

9. Perl TM, Cullen JJ, Wenzel RP, Zimmerman MB, Pfaller MA,Sheppard D, et al. Intranasal mupirocin to preventpostoperative Staphylococcus Aureus infections. N Engl J Med2002;346:1871 (MARS study team).

10. Bode LG, Kluytmans JA, Wertheim HF, Bogaers D,Vandenbroucke-Grauls CM, Roosendaal R, et al. Preventingsurgical-site infections in nasal carriers of StaphylococcusAureus. N Engl J Med 2010;362:9e17.

11. Qadan M, Akca O, Mahid SS, Hornung CA, Polk Jr HC.Perioperative supplemental oxygen therapy and surgical siteinfection: a meta-analysis of randomized controlled trials.Arch Surg 2009;144:359e66.

12. Darouiche RO, Wall Jr MJ, Itani KM, Otterson MF, Webb AL,Carrick MM, et al. Chlorhexidine-Alcohol versus Povidone-Iodine for surgical-site antisepsis. N Engl J Med 2010;362:18e26.