Articles
Surgical site infection after gastrointestinal surgery in
high-income, middle-income, and low-income countries:
a prospective, international, multicentre cohort study
GlobalSurg Collaborative*
Summary
Background Surgical site infection (SSI) is one of the most common infections associated with health care, but its
importance as a global health priority is not fully understood. We quantified the burden of SSI after gastrointestinal
surgery in countries in all parts of the world.
Methods This international, prospective, multicentre cohort study included consecutive patients undergoing elective
or emergency gastrointestinal resection within 2-week time periods at any health-care facility in any country. Countries
with participating centres were stratified into high-income, middle-income, and low-income groups according to the
UN’s Human Development Index (HDI). Data variables from the GlobalSurg 1 study and other studies that have been
found to affect the likelihood of SSI were entered into risk adjustment models. The primary outcome measure was
the 30-day SSI incidence (defined by US Centers for Disease Control and Prevention criteria for superficial and deep
incisional SSI). Relationships with explanatory variables were examined using Bayesian multilevel logistic regression
models. This trial is registered with ClinicalTrials.gov, number NCT02662231.
Findings Between Jan 4, 2016, and July 31, 2016, 13 265 records were submitted for analysis. 12 539 patients from
343 hospitals in 66 countries were included. 7339 (58·5%) patient were from high-HDI countries (193 hospitals
in 30 countries), 3918 (31·2%) patients were from middle-HDI countries (82 hospitals in 18 countries),
and 1282 (10·2%) patients were from low-HDI countries (68 hospitals in 18 countries). In total, 1538 (12·3%) patients
had SSI within 30 days of surgery. The incidence of SSI varied between countries with high (691 [9·4%] of
7339 patients), middle (549 [14·0%] of 3918 patients), and low (298 [23·2%] of 1282) HDI (p<0·001). The highest SSI
incidence in each HDI group was after dirty surgery (102 [17·8%] of 574 patients in high-HDI countries; 74 [31·4%] of
236 patients in middle-HDI countries; 72 [39·8%] of 181 patients in low-HDI countries). Following risk factor
adjustment, patients in low-HDI countries were at greatest risk of SSI (adjusted odds ratio 1·60, 95% credible interval
1·05–2·37; p=0·030). 132 (21·6%) of 610 patients with an SSI and a microbiology culture result had an infection that
was resistant to the prophylactic antibiotic used. Resistant infections were detected in 49 (16·6%) of 295 patients in
high-HDI countries, in 37 (19·8%) of 187 patients in middle-HDI countries, and in 46 (35·9%) of 128 patients in
low-HDI countries (p<0·001).
Lancet Infect Dis 2018
Published Online
February 13, 2018
http://dx.doi.org/10.1016/
S1473-3099(18)30101-4
See Online/Comment
http://dx.doi.org/10.1016/
S1473-3099(18)30118-X
*Collaborating members are
listed in the appendix
Correspondence to:
Dr Ewen M Harrison, NIHR Unit
on Global Surgery (Universities
of Birmingham, Edinburgh and
Warwick), University of
Edinburgh, Clinical Surgery,
Royal Infirmary of Edinburgh,
Edinburgh EH16 4SA, UK
ewen.harrison@ed.ac.uk
or
enquiry@globalsurg.org
See Online for appendix
See Online for podcast
Interpretation Countries with a low HDI carry a disproportionately greater burden of SSI than countries with a
middle or high HDI and might have higher rates of antibiotic resistance. In view of WHO recommendations on SSI
prevention that highlight the absence of high-quality interventional research, urgent, pragmatic, randomised trials
based in LMICs are needed to assess measures aiming to reduce this preventable complication.
Funding DFID-MRC-Wellcome Trust Joint Global Health Trial Development Grant, National Institute of Health
Research Global Health Research Unit Grant.
Copyright © The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
Introduction
Surgical site infection (SSI) is a large health burden for
patients and health-care providers. It is the most common
postoperative complication and causes pain and suffering
to patients.1,2 SSI is universally expensive3 and could result
in catastrophic health expenditure and impoverishment
to patients who are required to pay for their own
treatment.4 In low-income and middle-income countries
(LMICs), single-centre retrospective series have suggested
that SSI could be the most common infection associated
with health care.1 However, prospective, standardised,
and internationally comparable data on the incidence of
SSI and adverse events associated with SSI are lacking.5–8
These knowledge gaps make allocation of resources
to tackle SSI in LMICs challenging. The WHO
Guideline Development Group recently published
29 preoperative, intraoperative, and postoperative
recommendations about SSI prevention.6,9,10 These
recommendations are welcomed but are necessarily
based in large part on data extrapolated from highincome countries and, consequently, might lack validity
in resource-limited settings. Strategic planning to tackle
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
1
Articles
Research in context
Evidence before this study
We searched for evidence of multinational research assessing
surgical site infection (SSI) after abdominal surgery, focusing on
low-income and middle-income countries (LMICs). We searched
PubMed, MEDLINE, Google Scholar, and ClinicalTrials.gov for
articles published between Jan 1, 1997, and June 1, 2017, with the
terms “wound infection” OR “surgical site infection” AND
“developing countries” OR “low income” OR “middle income”
OR “low and middle income”, without language restrictions.
We reviewed related articles, references, and citations of eligible
texts. Several low-volume, single-centre studies to characterise
SSI in LMICs have been done in the past 20 years, but the
research quality is low to medium. These studies were
systematically reviewed in 2011 and included 57 studies of
abdominal surgery, with reported SSI incidence ranging
from 0·4% to 30·9% (between 1·5% and 81·0% for
clean–contaminated surgery, 0·5% to 65·5% for contaminated
surgery, and 0·2% to 100% for dirty surgery). The methodological
quality of individual studies was low and heterogeneity was high,
preventing meta-analysis. One multinational study has been
done since 2010, and included patients from seven high-income,
17 upper-middle-income, and six lower-middle-income
countries. The low observed SSI incidence (4·1%) after abdominal
SSI has been hindered by a lack of high-quality global
data. Microbiological data describing antimicrobial
resistance in SSI and information on the likely origin
of causative organisms are also needed to help
refine prevention strategies and quality-improvement
interventions.6,10
The GlobalSurg Collaborative designed and conducted
an international, multicentre, prospective cohort study
aimed at closing knowledge gaps in the incidence of SSI
in global health settings. The primary aim was to
determine variability in SSI rates in high-income,
middle-income, and low-income settings.
Methods
Study design and participants
This international, multicentre, prospective cohort study
used a published protocol11 and was done by teams of
local investigators who were coordinated by a national
lead investigator. Investigators were recruited via the
GlobalSurg network, social media, and personal contacts.
Any health-care facility in any country treating patients
who fulfilled the inclusion criteria could participate. The
collaborative network methodology has been described
in detail elsewhere.12 Ethical and institutional approval
was sought and obtained by each contributing institution
as per local regulations. A UK National Health Service
Research Ethics review considered this study exempt
from formal research registration (South East Scotland
Research Ethics Service, reference NR/1510AB5).
Individual centres obtained their own audit or
2
surgery could relate to the passive 30-day follow-up strategy;
additional limitations include a lack of data from lowest-income
countries and exclusion of children.
Added value of this study
We identified the burden and clinical impact of SSI in patients
undergoing gastrointestinal surgery in multiple income
settings. We used standardised, validated, prospective
methodology to provide global, contemporaneous data. SSI is
most common after dirty surgery in LMICs. Even after casemix
adjustment, patients in LMICs have a disproportionate burden
of infection. A large proportion of SSIs are caused by
organisms resistant to prophylactic antibiotics with the
greatest apparent burden in LMICs.
Implications of all the available evidence
The burden of SSI is disproportionately greater on patients and
health services in LMICs. Recent WHO recommendations on
preoperative and intraoperative measures for SSI prevention
highlight an absence of high-quality evidence. Urgent,
pragmatic, randomised trials based in LMICs are needed to
assess measures aiming to reduce this preventable
complication and associated antibiotic use.
institutional approval, and ethical approval was obtained
in countries where local research ethics committees
deemed it a requirement. This study is reported
according to the STROBE and SAMPL guidelines.
Investigators included patients from at least one
2-week period that was chosen a priori by the local team.
Consecutive sampling of patients undergoing elective or
emergency gastrointestinal resection was done during
the chosen 2-week period or periods. Consecutive
sampling is a common non-probability sampling strategy
in which all patients fulfilling the inclusion criteria
within a defined time period are enrolled. A 2-week
period was chosen to balance sample size requirements
and pragmatism for the working clinicians who were
enrolling patients and contributing data. The inclusion
criteria were based on two considerations. First, the
procedures were required to be relevant to the general
surgeons who form the collaborative. Second, a
reasonable baseline incidence of SSI was required so
meaningful comparisons could be made with the
predicted cohort size. So-called clean general surgery
cases, such as simple hernia repair, were excluded on
this basis. There was an absolute requirement for all
cases in the chosen period to be included, but no
minimum number was set to avoid bias against smaller
centres. Gastrointestinal resection was defined as
complete transection and removal of a segment of the
oesophagus, stomach, small bowel, colon, rectum,
appendix, or gallbladder and included formation or
reversal of a gastrointestinal stoma. Emergency
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
Articles
procedures were defined as unplanned, non-elective
operations and included procedures for trauma and
reoperation after previous surgery. Open or minimally
invasive procedures (eg, laparoscopic or robotic)
were eligible. No age restrictions were included. Patients
were excluded if the primary indication for surgery
was vascular, gynaecological, obstetric, urological, or
transplantation because the gastrointestinal tract is not
typically opened.
Data variables from the GlobalSurg 1 study13 and other
studies that have been found to affect the likelihood of
SSI were entered into risk adjustment models. Patient
variables included age, sex, physical status according to
the American Society of Anesthesiologists classification
system, existence of immune suppression (eg, HIV
status, active malarial infection, diabetes, use of steroid
therapies, chemotherapy or other immunosuppressive
drugs), and smoking status. Disease-related variables
included diagnostic category, timing of surgery (elective
vs emergency), use of the WHO surgical safety checklist,14
use of laparoscopy, use of epidural anaesthesia,
use of prophylactic antibiotics, and intraoperative
contamination. Contamination level2,15 was defined by the
operating surgeon as clean (an incision in which no
inflammation is encountered in a surgical procedure,
without a break in sterile technique, and during which
the respiratory, alimentary, and genitourinary tracts are
not entered; inclusion criteria for this study excluded this
group), clean–contaminated (an incision through which
the respiratory, alimentary, or genitourinary tract is
entered under controlled conditions but with no
contamination encountered), contaminated (an incision
undertaken during an operation in which there is a
major break in sterile technique or gross spillage from
the gastrointestinal tract, or an incision in which acute,
non-purulent inflammation is encountered; open
traumatic wounds more than 12–24 h old also fall into
this category), or dirty (an incision undertaken during an
operation in which the viscera are perforated or when
acute inflammation with pus is encountered during the
operation [eg, emergency surgery for faecal peritonitis],
and for traumatic wounds where treatment is delayed,
and faecal contamination or devitalised tissue is present).
Data were collected on the incidence and length of
antimicrobial treatment before and after surgery.
A pragmatic view was taken in the use of local
protocols and techniques for collecting and processing
microbiological specimens. Antimicrobial resistance was
defined as resistance in the species presumed to be
pathological to the antimicrobial used for prophylaxis. To
aid in the communication of findings, organisms were
broadly categorised as bowel-derived if cultures contained
only Gram-negative bacilli, Enterococcus species, or
anaerobic organisms, as skin-derived if cultures only
contained skin-derived organisms such as Staphylococcus
species, and as of mixed origin if cultures contained both
bowel-derived and skin-derived cultures.
Data variables were selected to be objective,
standardised, easily transcribed, and internationally
relevant to maximise record completion and accuracy.
Local investigators uploaded records to a secure online
website, provided using the Research Electronic Data
Capture (REDCap) system.16 The lead investigator at each
site checked the accuracy of all cases before data
submission. The submitted data were then checked
centrally and when missing data were identified, the
local lead investigator was contacted and asked to
complete the record. Once vetted, the record was accepted
into the dataset for analysis. Records that were vetted but
remained incomplete were included in the patient
flowchart but excluded from analysis.
Data validation was done in three parts across a
representative sample of centres according to a prespecified protocol (appendix). First, centres self-reported
the key processes used to identify and follow up patients.
Second, independent validators (ie, doctors, nurses, or
medical students who were not part of the recruiting
teams) quantitatively reported case ascertainment and
sampled data accuracy. Third, teams were interviewed to
qualitatively assess collaborator engagement and data
collection processes.
Outcomes
The primary outcome measure was the 30-day SSI
incidence, defined using the US Centers for Disease
Control and Prevention criteria for superficial and deep
incisional SSI.2 These criteria require the patient to have
at least one of the following: (1) purulent drainage from
the superficial or deep (fascia or muscle) incision but
not from within the organ or space component of the
surgical site; (2) pain or tenderness, localised swelling,
redness, heat, or fever, or several of these symptoms, and
the incision is opened deliberately or spontaneously
dehisces; or (3) abscess within the wound (clinically or
radiologically detected).
Organ space infections were recorded separately and
defined as intra-abdominal or pelvic infections detected
clinically or symptomatically, radiologically, or intraoperatively. A mandatory online SSI training module was
completed by all collaborators before data collection.
The secondary outcome measures were designed to
describe the clinical effect of SSI and included: (1) 30-day
postoperative mortality, defined as death any time after
skin closure until 30 days after surgery;17 (2) prevalence
in perioperative antibiotic administration; (3) 30-day
postoperative reintervention incidence (operative, radiological, or endoscopic reintervention any time after skin
closure until 30 days after surgery); (4) the prevalence of
antimicrobial resistance for SSI (microbiological culture
of wound swabs from site of SSI done according to local
protocols, with a pragmatic definition of antimicrobial
resistance defined a priori as resistance to the antimicrobial drug used for prophylaxis for that procedure in
that particular patient); and (5) in-hospital SSI incidence
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
3
Articles
(patients were reviewed for SSI during their stay and at
the time of hospital discharge); and (6) overall 30-day SSI
incidence (patients were assessed at 30 days to determine
whether an SSI had occurred; follow-up was done
in person, by telephone, or by review of medical/
readmission records, dependent on local practices).
Statistical analysis
For the Human Development
Index see http://hdr.undp.org/
en/content/humandevelopment-index-hdi
As described in the protocol,11 consideration was given
to the sample size needed to compare HDI groups. This
was approximated because data describing SSI
incidence internationally are lacking. Taken with data
from the GlobalSurg 1 study,13 for a baseline SSI
incidence of 15%, 550 patients per group (1350 patients
in total after accounting for potential missing data and
loss to follow-up) would allow for a 6·5 percentage point
difference to be detected with a power of 80% at an α
significance level of 0·05.
Variation between different international health settings
was assessed by stratifying countries with participating
centres into tertiles according to the Human Development
Index (HDI). The HDI is the UN’s composite statistic of
life expectancy, education, and income indices. Differences
between HDI tertiles were tested with the Pearson χ² test
for categorical variables and with the Kruskal-Wallis test
for continuous variables. Bayesian multilevel logistic
regression models were constructed to account for casemix
(differing patient, disease, and operative characteristics), as
previously described.18 Briefly, non-informative priors were
used with sensitivity analyses done on alternative priors
and different chain initiation points or chain lengths.
Models were constructed using the following principles:
(1) variables associated with outcome measures in previous
studies were accounted for; (2) demographic variables were
included in model exploration; (3) population stratification
by hospital and country of residence was incorporated as
13 265 participant records included in the
analysis
712 unknown SSI outcome
14 missing SSI outcome
12 539 participants with primary surgical
site infection outcome available
12 539 participants from 343 hospitals
in 66 countries
High HDI
7339 participants
from 193 hospitals
Middle HDI
3918 participants
from 82 hospitals
Figure 1: Patient flowchart
SSI=surgical site infection. HDI=Human Development Index.
4
Low HDI
1282 participants
from 68 hospitals
random effects with constrained gradients; (4) all firstorder interactions were checked and included in final
models if found to be influential; (5) final model selection
was done using a criterion-based approach by minimising
the widely applicable information criterion (WAIC) and
discrimination determined using the c-statistic (area under
the receiver operator curve). Model coefficients are
presented as odds ratio (OR) and 95% credible intervals
(CI; analogous to confidence intervals in frequentist
statistics, but philosophically distinct). In a further analysis,
a restricted cubic spline transformation was applied to the
continuous representation of the HDI to account for
potential non-linearity (three knots distributed equally
across the range of HDI rank). This was substituted into
the final multilevel model (generalised additive model) and
posterior predictions were made for specified covariate
levels with 95% CI determined. All analyses were done
using the R Foundation Statistical Program version 3.1.1
and Stan A C++ Library for Probability and Sampling
version 2.10.0. This trial is registered with ClinicalTrials.
gov, number NCT02662231.
Data sharing
The dataset can be explored using an online visualisation
application at http://ssi.globalsurg.org.
Role of the funding source
The funder of the study had no role in study design, data
collection, data analysis, data interpretation, or writing of
the report. The corresponding author had full access to
all the data in the study and had final responsibility for
the decision to submit for publication.
Results
Between Jan 4, 2016, and July 31, 2016, 13 265 patient
records were submitted for analysis. 726 (5·5%) records
remained incomplete after quality control, leaving
12 539 records for the final analysis (figure 1). These
patients were from 343 centres across 66 countries
(15 countries in Africa, 16 countries in Asia, 22 countries
in Europe, eight countries in North America, one country
in Oceania, and four countries in South America; table 1).
7339 (58·5%) patients were from countries with high
HDI, 3918 (31·2%) patients were from countries with
middle HDI, and 1282 (10·2%) patients were from
countries with low HDI. 1291 (10·3%) patients were
children (aged 16 years or younger). Missing data were
uncommon (appendix p 12), and no patterns were seen
when comparing included and missing data (appendix
pp 13, 14).
The most common operations were cholecystectomy
(4412 [35·2%] of 12 539 patients) and appendicectomy
(4179 [33·3%]; appendix p 1). 6117 (48·8%) patients had
emergency surgery, 5887 (46·9%) patients had an open
approach, and a surgical safety checklist was used before
8843 (70·5%) cases (table 1). Overall, 9922 (79·1%)
operations were clean–contaminated, 1540 (12·3%)
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
Articles
High HDI (n=7339)*
Mean age (SD), years
Middle HDI (n=3918)†
Low HDI (n=1282)‡
48·7 (21·5)
37·3 (18·5)
32·4 (19·1)
Male
3248 (44·3%)
1508 (38·5%)
Female
3683 (50·2%)
2215 (56·5%)
Missing
408 (5·6%)
195 (5·0%)
I
2498 (34·0%)
II
3191 (43·5%)
III+
Total (n=12 539)§
p value
43·5 (21·3)
<0·001
678 (52·9%)
5434 (43·3%)
<0·001
562 (43·8%)
6460 (51·5%)
··
42 (3·3%)
645 (5·1%)
··
2299 (58·7%)
687 (53·6%)
5484 (43·7%)
<0·001
1106 (28·2%)
409 (31·9%)
4706 (37·5%)
··
1543 (21·0%)
293 (7·5%)
178 (13·9%)
2014 (16·1%)
··
107 (1·5%)
220 (5·6%)
7 (0·5%)
334 (2·7%)
··
Sex
ASA
Unknown
Missing
0
0
1 (0·1%)
1 (<0·1%)
··
HIV
No
6773 (92·3%)
3573 (91·2%)
Yes
13 (0·2%)
39 (1·0%)
5 (0·4%)
553 (7·5%)
306 (7·8%)
179 (14·0%)
Unknown
Missing
0
0
1097 (85·6%)
1 (0·1%)
11 443 (91·3%)
<0·001
57 (0·5%)
··
1038 (8·3%)
··
1 (<0·1%)
··
Malaria
No
7243 (98·7%)
Yes
Unknown
Missing
3845 (98·1%)
1157 (90·2%)
12 245 (97·7%)
<0·001
8 (0·1%)
3 (0·1%)
10 (0·8%)
21 (0·2%)
··
87 (1·2%)
69 (1·8%)
115 (9·0%)
271 (2·2%)
··
1 (<0·1%)
1 (<0·1%)
0
2 (<0·1%)
··
No
6484 (88·3%)
3564 (91·0%)
1184 (92·4%)
11 232 (89·6%)
<0·001
Yes
745 (10·2%)
309 (7·9%)
73 (5·7%)
1127 (9·0%)
··
Unknown
110 (1·5%)
44 (1·1%)
25 (2·0%)
179 (1·4%)
··
Diabetes
Missing
0
1 (<0·1%)
0
1 (<0·1%)
··
Immunosuppressive medication
No
6893 (93·9%)
3789 (96·7%)
1243 (97·0%)
Yes
446 (6·1%)
129 (3·3%)
39 (3·0%)
11 925 (95·1%)
614 (4·9%)
<0·001
··
No
6190 (84·3%)
3353 (85·6%)
1170 (91·3%)
10 713 (85·4%)
<0·001
Yes
1149 (15·7%)
565 (14·4%)
112 (8·7%)
1826 (14·6%)
··
Current smoker
Pathology
Appendicitis
2061 (28·1%)
1516 (38·7%)
502 (39·2%)
4079 (32·5%)
<0·001
Gallstone disease
2505 (34·1%)
1493 (38·1%)
290 (22·6%)
4288 (34·2%)
··
Malignancy
1510 (20·6%)
287 (7·3%)
104 (8·1%)
1901 (15·2%)
··
Benign foregut
446 (6·1%)
220 (5·6%)
49 (3·8%)
715 (5·7%)
··
Benign midgut or
hindgut
570 (7·8%)
150 (3·8%)
121 (9·4%)
841 (6·7%)
··
Infection
46 (0·6%)
41 (1·0%)
63 (4·9%)
150 (1·2%)
··
Congenital
47 (0·6%)
49 (1·3%)
85 (6·6%)
181 (1·4%)
··
Trauma or injury
18 (0·2%)
47 (1·2%)
45 (3·5%)
110 (0·9%)
··
Complication of previous
procedure
67 (0·9%)
23 (0·6%)
14 (1·1%)
104 (0·8%)
··
Other
33 (0·4%)
9 (0·2%)
6 (0·5%)
48 (0·4%)
··
No disease
36 (0·5%)
80 (2·0%)
3 (0·2%)
119 (0·9%)
··
Missing
0
3 (0·1%)
0
3 (<0·1%)
··
Procedure start-time
0800 h to 1759 h
5788 (78·9%)
2753 (70·3%)
865 (67·5%)
9406 (75·0%)
<0·001
1800 h to 2159 h
724 (9·9%)
381 (9·7%)
180 (14·0%)
1285 (10·2%)
··
2200 h to 0759 h
821 (11·2%)
782 (20·0%)
237 (18·5%)
1840 (14·7%)
··
6 (0·1%)
2 (0·1%)
0 (0·0%)
Missing
8 (0·1%)
··
(Table 1 continues on next page)
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
5
Articles
High HDI (n=7339)*
Middle HDI (n=3918)†
Low HDI (n=1282)‡
Total (n=12 539)§
p value
(Continued from previous page)
Admission to procedure time, h
<6
2291 (31·2%)
1052 (26·9%)
308 (24·0%)
3651 (29·1%)
<0·001
6–11
722 (9·8%)
366 (9·3%)
128 (10·0%)
1216 (9·7%)
··
12–23
1364 (18·6%)
675 (17·2%)
217 (16·9%)
2256 (18·0%)
··
24–47
1330 (18·1%)
552 (14·1%)
230 (17·9%)
2112 (16·8%)
··
≥48
1358 (18·5%)
1033 (26·4%)
338 (26·4%)
2729 (21·8%)
··
274 (3·7%)
240 (6·1%)
61 (4·8%)
575 (4·6%)
··
Elective
3941 (53·7%)
1997 (51·0%)
483 (37·7%)
6421 (51·2%)
<0·001
Emergency
3397 (46·3%)
1921 (49·0%)
799 (62·3%)
6117 (48·8%)
··
1 (<0·1%)
··
Missing
Urgency
Missing
1 (<0·1%)
0
0
Operative approach
Open
2679 (36·5%)
2153 (55·0%)
1055 (82·3%)
5887 (46·9%)
<0·001
Laparoscopic
4660 (63·5%)
1765 (45·0%)
227 (17·7%)
6652 (53·1%)
··
No
6554 (89·3%)
3666 (93·6%)
1230 (95·9%)
11 450 (91·3%)
<0·001
Yes
646 (8·8%)
210 (5·4%)
48 (3·7%)
904 (7·2%)
··
Unknown
139 (1·9%)
42 (1·1%)
4 (0·3%)
185 (1·5%)
··
Epidural
Antibiotic: pre-procedural or prophylactic
No
848 (11·6%)
472 (12·0%)
50 (3·9%)
1370 (10·9%)
<0·001
Yes
6446 (87·8%)
3392 (86·6%)
1224 (95·5%)
11062 (88·2%)
··
Missing
45 (0·6%)
54 (1·4%)
8 (0·6%)
107 (0·9%)
··
Intraoperative contamination
Clean–contaminated
5918 (80·6%)
3126 (79·8%)
878 (68·5%)
9922 (79·1%)
<0·001
Contaminated
779 (10·6%)
542 (13·8%)
219 (17·1%)
1540 (12·3%)
··
Dirty
574 (7·8%)
236 (6·0%)
181 (14·1%)
991 (7·9%)
··
68 (0·9%)
14 (0·4%)
4 (0·3%)
86 (0·7%)
··
No, not available
837 (11·4%)
1114 (28·4%)
308 (24·0%)
2259 (18·0%)
<0·001
No, but available
238 (3·2%)
690 (17·6%)
363 (28·3%)
1291 (10·3%)
··
6194 (84·4%)
2049 (52·3%)
600 (46·8%)
8843 (70·5%)
··
69 (0·9%)
65 (1·7%)
11 (0·9%)
145 (1·2%)
··
Missing
Safety checklist used
Yes
Unknown
Missing
1 (<0·1%)
0
0
1 (<0·1%)
··
Numbers are n (%), unless otherwise indicated. All tests are Pearson’s χ² test, except for the comparison of mean age, where a Kruskall-Wallis test has been applied.
ASA=American Society of Anesthesiologists classification grade. *Included 30 countries and 193 hospitals. †Included 18 countries and 82 hospitals. ‡Included 18 countries
and 68 hospitals. §Included 66 countries and 343 hospitals.
Table 1: Patient and operative characteristics by human development index (HDI) rank
operations were contaminated, and 991 (7·9%) operations
were dirty.
1538 (12·3%) patients had SSI within 30 days of
surgery, and 842 (6·7%) had SSI before discharge from
hospital (appendix p 5). The unadjusted SSI incidence
varied between countries with high HDI (691 [9·4%] of
7339 patients), middle HDI (549 [14·0%] of 3918 patients)
and low HDI (298 [23·2%] of 1282 patients).
Intraoperative contamination was more likely to be
classed as dirty in countries with low HDI (181 [14·1%] of
1282 patients) than in countries with middle HDI
(236 [6·0%] of 3918 patients) or high HDI (574 [7·8%] of
7339; table 1). SSI rates increased significantly with dirty
surgery compared with clean–contaminated surgery;
6
however, there was no significant interaction for SSI
between HDI and intraoperative contamination
(appendix pp 3, 4). After multivariable adjustment for
confounders (including contamination), a significantly
higher SSI rate was seen in countries with low HDI
(adjusted OR 1·60, 95% CI 1·05–2·37; p=0·030) but not
in middle-HDI settings (1·12, 0·77–1·61; p=0·539)
compared with high-HDI countries (figure 2;
appendix p 4). When adjusted for patient and hospital
factors, SSI increased markedly at the threshold between
countries with middle and low HDI (rank 100; figure 3).
The increase was observed for both clean–contaminated
and dirty surgery because there was no significant
interaction between contamination and HDI, suggesting
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
Articles
that HDI is an independent risk factor for SSI,
irrespective of intraoperative contamination.
235 (1·9%) patients died within 30 days of surgery, but
mortality varied between countries with high HDI
(110 [1·5%] of 7339 patients), middle HDI (64 [1·6%] of
3918 patients), and low HDI (61 [4·8%] of 1282 patients;
appendix p 5). Patients with SSI were more likely than
patients without SSI to die, to have a reintervention, to
have an organ space infection, or to have another healthcare-associated infection (table 2). The median length of
hospital stay was three times longer for patients with an
SSI than for patients without (median 7·0 days [IQR 11·0]
vs 2·0 days [4·0]; p<0·001).
Patients in LMICs were more likely to receive presurgery
antibiotic courses than patients in high-HDI settings
(appendix p 6). Prophylactic antibiotic administration was
generally high (10 225 [81·5%] of 12 539 patients), with slight
variation between HDI groups. Overall, administration of
preoperative or prophylactic antibiotics, or both, was higher
in groups with low HDI (1224 [95·5%] of 1282) than in
countries with middle HDI (3392 [86·6%] of 3918 patients)
and high HDI (6446 of 87·8%] of 7339 patients; p<0·001).
Patients in LMICs were more likely to receive
postoperative antibiotics than those in high-HDI countries
(3376 [46·0%] of 7339 patients in high-HDI countries vs
3135 [80·0%] of 3918 patients in middle-HDI countries vs
1098 [85·6%] of 1282 patients in low-HDI countries;
p<0·001; appendix p 7). The increased tendency to use
antibiotics after surgery in low-HDI countries persisted
despite adjustment for confounding factors (adjusted
OR 4·37, 95% CI 1·65–11·85, p=0·002), including
contamination of surgery (appendix pp 8, 9). Courses of
postoperative antibiotics were longer in patients in LMICs
than in high-income countries, with the number of patients
receiving antibiotics for 5 days or more increasing from
countries with high HDI to low HDI (1830 [24·9%] of
7339 patients in high-HDI countries vs 1837 [46·9%]
of 3918 patients in middle-HDI countries vs 650 [50·7%] of
1282 patients in low-HDI countries; p<0·001; appendix p 7).
A microbiological wound culture was available for
610 (39·7%) of 1538 patients with an SSI (table 3). A
summary and full lists of causative organisms are available
in the appendix (pp 10, 11). 301 [63·8%] of 472 patients had
bowel-derived infections, 97 (20·6%) patients had skinderived infections, and 53 (11·2%) patients had infections
of mixed origin. Organisms with resistance to the actual
prophylactic antibiotic used were isolated from 132 (21·6%)
of the 610 patients with SSI who had a wound culture
(table 3). The prevalence of resistance varied between
countries with high, middle, and low HDI (table 3).
Patients were identified for inclusion predominately
using theatre logbooks or computer systems (5771 [78·6%]
of 7339 patients in high-HDI countries; 2349 [60·0%] of
3918 patients in middle-HDI countries; 759 [59·2%]
of 1282 patients in low-HDI countries) and operating
lists (1318 [18·0%] of 7339 patients in high-HDI countries;
823 [21·0%] of 3918 patients in middle-HDI countries;
OR (95% CI)
p value
High
··
··
Middle
1·12 (0·77−1·61)
p=0·539
Low
1·60 (1·05−2·37)
p=0·030
1
··
··
2
1·44 (1·22−1·70)
p<0·001
3+
1·63 (1·32−2·02)
p<0·001
Unknown
1·43 (0·95−2·12)
p=0·082
No
··
··
Yes
1·40 (1·15−1·70)
p=0·001
No
··
··
Yes
1·12 (0·87−1·44)
p=0·381
No, not available
··
··
No, but available
1·49 (1·15−1·91)
p=0·002
Yes
1·01 (0·79−1·28)
p=0·970
Unknown
1·25 (0·69−2·21) p=0·440
HDI tertile
ASA
Diabetes
Immunosuppressive medication
Checklist
Operative approach
Open
··
Laparoscopic
0·36 (0·30−0·42) p<0·001
··
Antibiotics: pre− or prophylactic
No
··
Yes
0·81 (0·66−1·01) p=0·065
··
Intraoperative contamination
Clean−contaminated
··
··
Contaminated
2·13 (1·81−2·51)
p<0·001
Dirty
2·24 (1·83−2·73)
p<0·001
0·5
1·0
OR log scale
1·5
2·0
2·5 3·0
Figure 2: Multilevel model for factors associated with surgical site infection
Full model includes HDI tertile, age, American Society of Anesthesiologists (ASA) classification grade, diabetes
status, immunosuppressive medication treatment, current smoker, pathology, operative approach, antibiotic use
before surgery, intraoperative contamination, and WHO checklist used. Error bars are 95% credible interval.
Full data are in the appendix (p 4). OR=odds ratio. HDI=Human Development Index. CI=credible interval.
278 [21·7%] of 1282 patients in low-HDI countries; appendix
p 16). Many patients across HDI strata were followed
up by telephone (2708 [36·9%] of 7339 patients in
high-HDI countries; 2582 [65·9%] of 3918 patients in
middle-HDI countries; 483 [37·7%] of 1282 patients
in low-HDI countries; appendix p 17). Validators
identified 1476 cases that fulfilled inclusion criteria, and
1378 (93%) cases were ascertained (appendix p 18).
Accuracy was high for the validated continuous predictor
(Pearson’s correlation coefficient 0·99; appendix p 20),
categorical predictors (Cohen’s κ coefficients >0·90;
appendix p 21), and mortality (κ 0·91). The agreement for
30-day reintervention was lower (κ 0·65).
Discussion
We identified both the burden and clinical effect of SSI
on patients undergoing gastrointestinal surgery in many
parts of the world. SSI affected 12·3% of patients
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
7
Articles
Probability of surgical site infection (%)
Clean–contaminated
Contaminated or dirty
60
40
20
0
50
100
Human Development Index rank
150
Figure 3: Probability of surgical site infection (SSI) by human development index (HDI) rank
Adjusted predicted probability of SSI across HDI rank by intraoperative contamination. In the most developed
countries (rank 1), patients had a low probability of SSI. At rank 100, the probability of SSI increases linearly
through the least developed countries. This absolute difference between clean–contaminated and contaminated or
dirty surgery is shown, with no interaction between HDI and intraoperative contamination found. Shaded area is
the credible interval.
No SSI (n=11 001) SSI (n=1538)
p value
<0·001
30-day mortality
Alive
10 665 (96·9%)
1438 (93·5%)
Dead
162 (1·5%)
73 (4·7%)
··
Missing
174 (1·6%)
27 (1·8%)
··
<0·001
30-day reintervention
No
10 674 (97·0%)
1202 (78·2%)
Yes
235 (2·1%)
316 (20·5%)
··
92 (0·8%)
20 (1·3%)
··
<0·001
Missing
Organ space infection (abscess)
No
10 759 (97·8%)
1229 (79·9%)
Yes
146 (1·3%)
276 (17·9%)
··
96 (0·9%)
33 (2·1%)
··
<0·001
Missing
Other health-care-associated infection
No
10 546 (95·9%)
1292 (84·0%)
Yes
388 (3·5%)
214 (13·9%)
··
32 (2·1%)
··
Missing
67 (0·6%)
Median length of stay (IQR)
2·0 (4·0)
7·0 (11·0)
<0·001*
Numbers are n (%), unless otherwise indicated. All tests are χ² tests, except when indicated by*, where a Kruskall-Wallis
test has been applied.
Table 2: Associations between surgical site infection (SSI) and other outcomes
High HDI
(n=295)
Middle HDI
(n=187)
Low HDI
(n=128)
Total (n=610)
p value
<0·001
Antibiotic not used
27 (9·2%)
6 (3·2%)
0 (0·0%)
33 (5·4%)
Sensitive to antibiotic
92 (31·2%)
56 (29·9%)
40 (31·2%)
188 (30·8%)
··
Resistant to antibiotic
49 (16·6%)
37 (19·8%)
46 (35·9%)
132 (21·6%)
··
127 (43·1%)
88 (47·1%)
42 (32·8%)
257 (42·1%)
··
Sensitivity not available
Numbers are n (%), unless otherwise indicated. All tests are χ² tests.
Table 3: Sensitivity of organism by Human Development Index (HDI) from patients with a surgical site
infection who had a wound swab taken
8
worldwide, and the incidence increased across HDI
groups, reaching 39·8% of patients undergoing dirty
surgery in low-HDI settings. The incidence of SSI
remained higher in low-HDI countries than in
middle-HDI or high-HDI countries, despite adjustment
for factors describing patients, diseases (including
contamination), procedures, safety, and hospitals. Length
of hospital stay was three times longer for patients
affected by SSI than for patients with no SSI. Delayed
return to work or school carries a societal burden, which
is likely to be greater in LMICs.
These findings begin to characterise the relationship
between SSI and global antimicrobial resistance. Where
microbiological cultures were available, SSIs were more
likely to be caused by bowel-derived organisms. Large
amounts of antibiotics were consumed to prevent and
treat SSI, yet in 21·6% of cases with a positive culture,
the causative microorganism was resistant to the
prophylactic antibiotics that had been administered. The
prevalence of antimicrobial resistance increased to one
of three isolates in low-HDI countries. Postoperative
courses of antibiotics were longest for patients in
low-HDI countries, and this was not explained by
casemix. Although there is randomised evidence that
short postoperative antibiotic courses are as safe as long
antibiotic courses, this evidence was not derived in
LMICs, and caution is needed before changing practice.19
The high prevalence of SSIs that were resistant to the
initial prophylactic antibiotic illustrates a potentially
important area for improvement worldwide. Complete
microbiological analysis of all SSIs was not possible
within this observational study, so the problem might be
even larger that estimated here.
The focus in global surgery to date has been directed
towards mortality. The 30-day mortality in this study was
similar to that in the GlobalSurg 1 study (1·9% and
1·6% respectively).13 This generally low mortality
highlights the importance of studying more common
outcomes such as SSI across health systems, given the
impact on patients. We found an association between
SSI and death, with a three-fold increase from 1·5% in
patients without SSI to 4·7% in patients with SSI within
this study. This is an association, and no causal link can
be made with these data; it is likely that patients died
with an SSI rather than from an SSI. Since SSI was also
associated with deep organ space infection and other
health-care-associated infections, this supports its use as
a severity marker of illness.
Interest in the use of surgical safety checklists has
increased in the past 5 years, and they are now part of
clinical routine in many surgical units. In this study, the
failure to use an available surgical safety checklist was
associated with a high SSI rate. This association was not
explained by an omission of prophylactic antibiotics, nor
was it particular to emergency surgery, when haste might
improperly trump safety measures. The scientific
literature describing checklists and SSI is contrasting
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
Articles
and includes a recent systematic review of 14 studies.14
The data in this systematic review showed a decrease in
SSI with checklist use (range within individual studies
from 3·2% to 10·2% absolute risk reduction). The
GlobalSurg studies provide novel checklist data from
LMIC settings. The explanation for the observed effect is
unclear but probably describes a broader attitude to
safety in hospital systems that require further
investigation.
A major strength of this study is its provision of
prospective patient-level SSI data from a wide breadth of
settings around the world. In particular, outcome
assessment was standardised and training provided
through our online tool. Several small and generally
single-centre studies have been done in the past 20 years
in attempts to characterise SSI in LMICs. These were
systematically reviewed in a 2010 study1 that included
57 reports focusing on SSI. General methodological
quality was low and heterogeneity was high, with
reported SSI rates varying from 0·4% to 30·9%. Since
then, SSI outcomes from several single-centre and
national multicentre studies in LMICs have been
published.20–24 The lower than expected rates emphasise
the difficulty in robustly determining SSI, which,
together with the between-study variability, make
international comparisons difficult. The present study
contributes to closing this knowledge gap and allows
meaningful comparison from multiple income settings
with accurate casemix adjustment and standardised
training in outcome assessment. Reliability was
increased through the vetting of incomplete records and
was demonstrated in a parallel validation study.
A major limitation of this study was the inability to
follow up every patient 30 days after surgery. SSI
detection within randomised trials is higher when
proactively followed up as a primary endpoint than when
followed up as a secondary outcome.25 Within our study,
collaborators were trained and encouraged to directly
determine 30-day outcomes whenever possible. Overall,
this was successful; however, complete, in-person, 30-day
follow-up for thousands of patients would not have been
possible, particularly in resource-limited settings.
Nevertheless, we did assess SSI as a primary endpoint,
used a mandatory training package, and did a sensitivity
analysis using in-hospital SSI rates. The variation in
incidence of SSI before discharge from hospital and
within 30 days was similar between countries of high,
middle, and low HDI. Since these incidence data are
already comparable to those from high-quality
randomised trials, this provides some measure of
validity.26 Other limitations apply. First, with respect to
microbiological analysis, we did not standardise
specimen collection, laboratory assessment, techniques,
or definitions. A pragmatic view was taken to use local
protocols and techniques for collecting and processing
specimens and for determining antimicrobial resistance.
These measures were therefore recognised in advance as
being an exploratory analysis to describe the prevalence
of organisms with antimicrobial resistance against the
particular prophylactic antibiotic administered. Second,
although we did validation, there is still the potential for
missed cases or inaccurate data.13,27,28 The large number of
patients, a prospective protocol, and the use of local
coordinators might have minimised the potential bias.
Reducing SSI will contribute to ensuring safe and
essential surgery around the world.29 Costs to patients in
LMICs in terms of expenditure and time off work have
not been measured but are probably considerable.
The costs of preventive measures might be offset
by the realised cost-savings. WHO has published
recommendations to help reduce the incidence of SSI
that include global perspectives relevant to LMICs.9
Despite inclusion of strongly graded recommendations,
none of these could be based on high-quality evidence,
which is lacking in support of most interventions.
Virtually none of the existing evidence is derived from
LMICs, leading to uncertainty about future performance
of these measures.8 SSI research is complex, and bundles
of measures have been seen to paradoxically increase
SSI incidence.30 Implementation therefore necessitates
careful consideration and meticulous attention to
longer-term evaluation. In resource-limited settings, the
development of robust policy will remain difficult
without high-quality evidence. Our findings provide the
rationale to plan, fund, and perform high-quality surgical
research that can effect change in health policy. There are
no multicentre, multi-country randomised trials on SSI
prevention in LMICs at a time when efforts to combat
SSI should be informed by high-quality research derived
in these settings.8
Contributors
Collaborating members are listed together with their roles in the
appendix. The writing group contributed to study design, data analysis,
data interpretation, and the writing and review of the final report:
Aneel Bhangu, Adesoji O Ademuyiwa, Maria Lorena Aguilera,
Philip Alexander, Sara W Al-Saqqa, Giuliano Borda-Luque,
Ainhoa Costas-Chavarri, Thomas M Drake, Faustin Ntirenganya,
J Edward Fitzgerald, Stuart J Fergusson, James Glasbey, J C Allen Ingabire,
Lawani Ismaïl, Hosni Khairy Salem, Anyomih Theophilus Teddy Kojo,
Marie Carmela Lapitan, Richard Lilford, Andre L Mihaljevic, Dion Morton,
Alphonse Zeta Mutabazi, Dmitri Nepogodiev, Adewale O Adisa, Riinu Ots,
Francesco Pata, Thomas Pinkney, Tomas Poškus, Ahmad Uzair Qureshi,
Antonio Ramos-De la Medina, Sarah Rayne, Catherine A Shaw,
Sebastian Shu, Richard Spence, Neil Smart, Stephen Tabiri,
Ewen M Harrison. Annel Bhangu and Ewen M Harrison are study
guarantors.
Declaration of interests
This study was funded by a DFID-MRC-Wellcome Trust Joint Global
Health Trial Development Grant and a National Institute of Health
Research Global Health Research Unit Grant. JEF reports personal fees
from KPMG Global Healthcare Practice outside the submitted work.
All other authors declare no competing interests.
Acknowledgments
Funded by DFID-MRC-Wellcome Trust Joint Global Health Trial
Development Grant (MR/N022114/1) and a National Institute of Health
Research (NIHR) Global Health Research Unit Grant (NIHR 17-0799).
The views expressed are those of the authors and not necessarily those
of the National Health Service, the NIHR, or the UK Department of
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4
9
Articles
Health and Social Care. Organisations assisting in dissemination or
translation, or both (alphabetical): Asian Medical Students’ Association,
Association of Surgeons in Training, College of Surgeons of East,
Central & Southern Africa, Cutting Edge Manipal, Egyptian Medical
Student Research Association, International Collaboration For Essential
Surgery, International Federation of Medical Student Associations,
Italian Society of Colorectal Surgery, Lifebox Foundation, School of
Surgery, Student Audit and Research in Surgery, The Electives Network,
United Kingdom National Research Collaborative, World Society of
Emergency Surgery, World Surgical Association.
References
1
Allegranzi B, Bagheri Nejad S, Combescure C, et al. Burden of
endemic health-care-associated infection in developing countries:
systematic review and meta-analysis. Lancet 2011; 377: 228–41.
2
Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance
definition of health care-associated infection and criteria for specific
types of infections in the acute care setting. Am J Infect Control 2008;
36: 309–32.
3
Tanner J, Khan D, Aplin C, Ball J, Thomas M, Bankart J.
Post-discharge surveillance to identify colorectal surgical site
infection rates and related costs. J Hosp Infec 2009; 72: 243–50.
4
Leaper DJ, van Goor H, Reilly J, et al. Surgical site infection—
a European perspective of incidence and economic burden.
Int Wound J 2004; 1: 247–73.
5
Bagheri Nejad S, Allegranzi B, Syed SB, Ellis B, Pittet D.
Health-care-associated infection in Africa: a systematic review.
Bull World Health Organ 2011; 89: 757–65.
6
Allegranzi B, Zayed B, Bischoff P, et al. New WHO recommendations
on intraoperative and postoperative measures for surgical site
infection prevention: an evidence-based global perspective.
Lancet Infect Dis 2016; 16: e288–303.
7
Rosenthal VD, Richtmann R, Singh S, et al. Surgical site infections,
International Nosocomial Infection Control Consortium (INICC)
report, data summary of 30 countries, 2005–2010.
Infect Control Hosp Epidemiol 2013; 34: 597–604.
8
Aiken AM, Karuri DM, Wanyoro AK, Macleod J. Interventional
studies for preventing surgical site infections in sub-Saharan Africa:
a systematic review. Int J Surg 2012; 10: 242–49.
9
WHO. Global guidelines on the prevention of surgical site
infection. Geneva: World Health Organization, 2016.
10 Allegranzi B, Bischoff P, de Jonge S, et al. New WHO
recommendations on preoperative measures for surgical site
infection prevention: an evidence-based global perspective.
Lancet Infect Dis 2016; 16: e276–87.
11 GlobalSurg Collaborative. Determining the worldwide epidemiology
of surgical site infections after gastrointestinal resection surgery:
protocol for a multicentre, international, prospective cohort study
(GlobalSurg 2). BMJ Open 2017; 7: e012150.
12 Bhangu A, Kolias AG, Pinkney T, Hall NJ, Fitzgerald JE. Surgical
research collaboratives in the UK. Lancet 2013; 382: 1091–92.
13 Mortality of emergency abdominal surgery in high-, middle- and
low-income countries. Br J Surg 2016; 103: 971–88.
14 de Jager E, McKenna C, Bartlett L, Gunnarsson R, Ho YH.
Postoperative adverse events inconsistently improved by the World
Health Organization surgical safety checklist: a systematic literature
review of 25 studies. World J Surg 2016; 40: 1842–58.
10
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Berard F, Gandon J. Postoperative wound infections: the influence
of ultraviolet irradiation of the operating room and of various other
factors. Ann Surg 1964; 160 (suppl 2): 1–192.
Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG.
Research electronic data capture (REDCap)—a metadata-driven
methodology and workflow process for providing translational
research informatics support. J Biomed Inform 2009; 42: 377–81.
Watters DA, Hollands MJ, Gruen RL, et al. Perioperative mortality
rate (POMR): a global indicator of access to safe surgery and
anaesthesia. World J Surg 2015; 39: 856–64.
Albert V, Mndolo S, Harrison EM, O’Sullivan E, Wilson IH,
Walker IA. Lifebox pulse oximeter implementation in Malawi:
evaluation of educational outcomes and impact on oxygen
desaturation episodes during anaesthesia. Anaesthesia 2017;
72: 686–93.
Sawyer RG, Claridge JA, Nathens AB, et al. Trial of short-course
antimicrobial therapy for intraabdominal infection. N Engl J Med 2015;
372: 1996–2005.
Rosenthal VD. International Nosocomial Infection Control
Consortium (INICC) resources: INICC multidimensional approach
and INICC surveillance online system. Am J Infect Control 2016;
44: e81–90.
Petroze RT, Byiringiro JC, Kyamanywa P, Ntakiyiruta G, Calland JF,
Sawyer RG. Infectious outcomes assessment for health system
strengthening in low-resource settings: the novel use of a trauma
registry in Rwanda. Surg Infect (Larchmt) 2014; 15: 382–86.
Alvarez-Moreno C, Perez-Fernandez AM, Rosenthal VD, et al.
Surgical site infection rates in 4 cities in Colombia: findings of the
International Nosocomial Infection Control Consortium (INICC).
Am J Infect Control 2014; 42: 1089–92.
Ahoyo TA, Bankole HS, Adeoti FM, et al. Prevalence of nosocomial
infections and anti-infective therapy in Benin: results of the first
nationwide survey in 2012. Antimicrob Resist Infect Control 2014; 3: 17.
Ramirez-Wong FM, Atencio-Espinoza T, Rosenthal VD, et al.
Surgical site infections rates in more than 13 000 surgical procedures
in three cities in Peru: findings of the International Nosocomial
Infection Control Consortium. Surg Infect (Larchmt) 2015;
16: 572–76.
Matthews JH, Bhanderi S, Chapman SJ, Nepogodiev D, Pinkney T,
Bhangu A. Underreporting of secondary endpoints in randomized
trials: cross-sectional, observational study. Ann Surg 2016; 264: 982–86.
Pinkney TD, Calvert M, Bartlett DC, et al. Impact of wound edge
protection devices on surgical site infection after laparotomy:
multicentre randomised controlled trial (ROSSINI Trial). BMJ 2013;
347: f4305.
European Society of Coloproctology. The relationship between method
of anastomosis and anastomotic failure after right hemicolectomy and
ileo-caecal resection: an international snapshot audit. Colorectal Dis
2017; published online March 6. DOI:10.1111/codi.13646.
STARSurg Collaborative. Multicentre prospective cohort study of
body mass index and postoperative complications following
gastrointestinal surgery. Br J Surg 2016; 103: 1157–72.
Meara JG, Hagander L, Leather AJ. Surgery and global health:
a Lancet Commission. Lancet 2014; 383: 12–13.
Anthony T, Murray BW, Sum-Ping JT, et al. Evaluating an
evidence-based bundle for preventing surgical site infection:
a randomized trial. Arch Surg 2011; 146: 263–69.
www.thelancet.com/infection Published online February 13, 2018 http://dx.doi.org/10.1016/S1473-3099(18)30101-4