Journal of Hospital Infection xxx (xxxx) xxx
ww.sciencedirect.comAvailable online at wJournal of Hospital Infection
journal homepage: www.elsevier .com/locate/ jhinGenetic relationship between bacteria isolated from
intraoperative air samples and surgical site infections
at a major teaching hospital in Ghana
M.A. Stauning a,*, A. Bediako-Bowan b,c,d,e, S. Bjerrum f, L.P. Andersen a,
S. Andreu-Sánchez g, A-K. Labi h, j, J.A.L. Kurtzhals a,h,*, R.L. Marvig g,
J.A. Opintan k
aDepartment of Clinical Microbiology, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark
bDepartment of Surgery, School of Medicine and Dentistry, University of Ghana, Accra, Ghana
cDepartment of Surgery, Korle-Bu Teaching Hospital, Accra, Ghana
dDepartment of Veterinary and Animal Sciences, University of Copenhagen, Copenhagen, Denmark
e Statens Serum Institut, Copenhagen, Denmark
fGlobal Health Section, Department of Public Health, University of Copenhagen, Copenhagen, Denmark
gCentre for Genomic Medicine, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark
hCentre for Medical Parasitology, Department of Immunology and Microbiology, University of Copenhagen, Copenhagen,
Denmark
jDepartment of Microbiology, Korle-Bu Teaching Hospital, Accra, Ghana
kDepartment of Medical Microbiology, School of Biomedical and Allied Health Sciences, University of Ghana, Accra, GhanaA R T I C L E I N F O
Article history:
Received 7 July 2019
Accepted 11 November 2019
Available online xxx
Keywords:
Surgical site infections
Airborne bacteria
Low- and middle-income
countries
Staphylococcus aureus
Whole genome sequencing
Metagenomic* Corresponding authors. Address: Departme
E-mail addresses: marius.stauning@sund.k
https://doi.org/10.1016/j.jhin.2019.11.007
0195-6701/ª 2019 The Healthcare Infection S
Please cite this article as: Stauning MA et al
infections at a major teaching hospital in GhS U M M A R Y
Background: In low- and middle-income countries (LMICs) the rate of surgical site infec-
tions (SSI) is high, leading to negative patient outcomes and excess healthcare costs. A
causal relationship between airborne bacteria in the operating room and SSI has not been
established, at a molecular or genetic level. We studied the relationship between intra-
operative airborne bacteria and bacteria causing SSI in an LMIC.
Methods: Active air sampling using a portable impactor was performed during clean or
clean-contaminated elective surgical procedures. Active patient follow-up consisting of
phone calls and clinical examinations was performed 3, 14 and 30 days after surgery.
Bacterial isolates recovered from SSI and air samples were compared by matrix-assisted
laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF) identification,
ribotyping, whole genome sequencing (WGS), and metagenomic analysis.
Results: Of 128 included patients, 116 (91%) completed follow-up and 11 (9%) developed
SSI. Known pathogenic bacteria were isolated from intraoperative air samples in all cases
with SSI. A match between air and SSI isolates was found by MALDI-TOF in eight cases.
Matching ribotypes were found in six cases and in one case both WGS and metagenomic
analysis showed identity between air- and SSI-isolates.
Conclusion: The study showed high levels of intraoperative airborne bacteria, an SSI-
rate of 9% and a genetic link between intraoperative airborne bacteria and bacteriant of Clinical Microbiology, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark.
u.dk, bqw629@alumni.ku.dk (M.A. Stauning), Joergen.kurtzhals@regionh.dk (J.A.L. Kurtzhals).
ociety. Published by Elsevier Ltd. All rights reserved.
., Genetic relationship between bacteria isolated from intraoperative air samples and surgical site
ana, Journal of Hospital Infection, https://doi.org/10.1016/j.jhin.2019.11.007
2 M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxxPlease cite this article as: Stauning MA et al
infections at a major teaching hospital in Ghisolated from SSIs. This indicates the need for awareness of intraoperative air quality in
LMICs.
ª 2019 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.Introduction
Surgical site infections (SSI) are among the most common
surgical complications leading to negative patient outcomes
and excess costs [1,2]. Studies are needed to document the
burden and the causes of SSI in low- and middle-income
countries (LMICs), in order to guide preventive interventions
and prioritize scarce funds [3,4]. Previous studies have iden-
tified several risk factors for SSI such as overweight, smoking,
diabetes, health-status, and duration of procedure [1,2].
Intraoperative airborne bacteria represent a possible source of
SSI, but international recommendations in many cases do not
consider air quality important for reduction of SSI in LMICs
[1,2]. We have recently found high levels of airborne bacteria
in operating rooms at a major teaching hospital in Ghana [5].
Despite the established association between intraoperative
colony-forming units per cubic metre of air (cfu/m3) and SSI
risk, the aetiological role of these bacteria has not been for-
mally demonstrated [6e8]. In this study, we assessed pre-
dictors for SSI in an LMIC and used whole-genome sequencing
(WGS) and metagenomic analysis to assess the relationship
between intraoperative airborne bacteria and SSI. To our
knowledge, this is the first attempt to prove causality between
environmental intraoperative airborne bacteria and SSI with
the use of genetic methods.Methods
Setting
The study adds to our previously published work from the
General Surgery Unit, Korle-Bu Teaching Hospital [5]. Operat-
ing rooms were equipped with non-laminar ventilation with
highly efficient particulate air (HEPA) filters. Both air intake
and exhaust are located at the ceiling. Operating rooms are not
constructed to obtain positive pressure [5]. No routine main-
tenance services were in place at the time of study. No details
of air change per hour were available. Staff wore institutional
clothing, shoes, disposable hoods and facemasks [5]. The scrub
team further wore sterile gowns and gloves [5]. Incision sites
were disinfected three times before incision with disinfection
liquid consisting of 0.15% chlorhexidine, 1.5% cetrimide, 34%
alcohol (95% ethanol and 5% methanol), and water. Skin dis-
infectant was produced at the hospital’s central pharmacy. No
routine sterility test of the ingredients or final mixture was
made at the hospital pharmacy or at the surgical unit. Non-
disinfected skin was covered with sterile cotton sheets.
Instruments were sterilised by steam autoclaving for 55 min at
121
C 2 atm for plastic items, and 45 min at 134
C 3 atm for
remaining items. Correct sterilization was controlled with
heat-sensitive tape. As previously described a high level of
human activity was seen during surgery and both the number of
individuals and the number of door openings were predictive
factors of cfu counts in the operating room [5]. Prophylactic
antibiotics where administered prior to first incision. No local., Genetic relationship betwe
ana, Journal of Hospital Infeprotocol for antibiotic prophylaxis existed at the time of the
study. Between each surgery, surfaces, kickbuckets and floor
were cleaned from cleanest to dirtiest, and top to bottom with
a 10% chlorine solution. Operating rooms were equipped with
temperature control and ranged between 16 and 25
C.
Remaining parts of the hospital had no temperature control.Selection of patients
Patients undergoing surgery were included consecutively if:
age 18 years; American Society of Anaesthesiologists physical
classification score (ASA-score) III; elective non-implant
procedure; and wounds classified as clean or clean-
contaminated according to the Centre for Disease Control
(CDC) classification [9,10].Observation and patient characteristics
We piloted and used standardized questionnaires to record
patient age, height, weight, sex, smoking habits, diabetes
status, ASA-score, ICD-10-PCS procedure code, date and
duration of procedure, peri-operative antibiotics, number of
blood transfusions, and surgeon’s assessment of the CDCwound
classification [11].Air samples
Air samples were collected during surgery as part of our
previously published study [5]. In brief, air samples were
obtained on 5% blood agar using a portable impactor (MAS-100;
Merck, Darmstadt, Germany) operating 5 min at a flowrate of
100 L/min every 20 min from the time of the first incision to
final wound closure, shifting between a position 30e60 cm from
the wound and a position opposite the entrance 1.5 m from the
wall [5]. Plates were incubated at 37
C for 48 h and cfu were
thereafter counted [5]. After cfu count, each plate was scra-
ped with a sterile loop and biological material transferred to a
sterile broth-based medium with 10% glycerol, mixed and fro-
zen at 80
C before transfer to Denmark on dry ice.Follow-up
Patients were followed by surgeons with phone calls 3, 14
and 30 days after surgery, or until SSI diagnosis. Patients were
asked whether they had received wound treatment and inter-
viewed for SSI symptoms according to CDC criteria [12]. If SSI
were suspected, patients were invited for a clinical examina-
tion. If SSI was clinically confirmed, a sample was obtained
using E-Swabs (COPAN ITALIA S.P.A, Brescia, Italy). Sampling
was performed by trained surgeons. Samples were taken by
rotating a sterile swab in the infected wound carefully touching
only the wound. Swabs were frozen at 80
C in sterile broth-
based freezing medium with 10% glycerol until transport to
Denmark.en bacteria isolated from intraoperative air samples and surgical site
ction, https://doi.org/10.1016/j.jhin.2019.11.007
M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx 3Identification of bacterial isolates from air and wound
samples
A sample from each vial containing material from air sam-
ples obtained during procedures with subsequent SSI as well as
wound swabs were thawed and plated on 5% blood agar, blue
agar (a selective medium for large Gram-negative rods), and
7.5% NaCl plates (SSI Diagnostica, Hillerød, Denmark). After
incubation for 48 h at 37
C, morphologically different colonies
were sub-cultured on 5% blood agar, incubated 24 h and iden-
tified by matrix-assisted laser desorption/ionization time-of-
flight mass spectrometry (MALDI-TOF, Microflex, Bruker, Bre-
men, Germany).
WGS of SSI isolates
DNAwas prepared with DNeasy Blood and Tissue kit (Qiagen,
Hilden, Germany). Purification of DNA was conducted with
lysozyme treatment according to MALDI-TOF identification.
Illumina Miseq was used for sequencing, generating 250-base
pair paired-end reads using a multiplexed Nextera XT proto-
col (Illumina, Hilden, Germany). Sequence reads were error-
corrected using ALLPATHS-LG’s stand-alone error correction
tool and de-novo-assembled using de Bruijn graph based
assembler Velvet v.1.2.10 with Velvet Optimizer v. 2.2.5 [13].
Comparison of air and SSI isolates
Air and SSI isolates were paired with respect to operating
room and time of surgery. Paired isolates matching at species
level after MALDI-TOF identification were compared by ribo-
typing (RipoPrinter system, DuPont Qualicon, Geneva, Swit-
zerland) according to manufacturers’ instructions [14]. Air
isolates in the same ribogroup or with >85% similarity to a
paired SSI isolate were selected for WGS following the same
protocols as SSI isolates. We used BacDist with Snippy v.4.1.0 to
call SNPs in each of the isolates relative to a relevant National
Center for Biotechnology Information (NCBI) reference genome
and compare shared sequences [15,16]. Only SNPs at positions
covered by at least 10 reads in both isolates with one isolate
showing >80% non-reference reads were included. SNP dis-
tances were computed with snp-dists v.0.6.3 [17]. Approx-
imately maximum-likelihood phylogenetic trees were
constructed with Parsnp and visualized in iTOL-v3 [18,19].
Metagenomic sequencing of air samples
Metagenomic sequencing was performed on DNA extracted
directly from each vial containing material from air samples
obtained during procedures with subsequent SSI. DNA purifi-
cation was conducted with lysozyme treatment to extract DNA
from both Gram-positive and -negative bacteria. Metagenomic
taxonomic classification was performed with Kaiju in Greedy-1
mode with s ¼ 100 [20]. Only species with >5000 classified
reads/metagenome were included in the results.
Comparison of metagenomes from air samples and
whole genomes of SSI isolates
SNPs unique to each SSI isolate (ID-SNPs) were identified by
Parsnp comparison with complete NBCI genomes [19]. ToPlease cite this article as: Stauning MA et al., Genetic relationship betwe
infections at a major teaching hospital in Ghana, Journal of Hospital Infeccheck for the presence of ID-SNPs in the metagenomes, we
mapped reads of each metagenome to the genome of the SSI
isolate with the respective ID-SNPs using bwa-mem [21].
Metagenomic readcounts for each base at each ID-SNP posi-
tion were counted using bam-readcount to determine
whether the ID-SNP was present. Only mapped reads with a
minimal quality >30 were included in read counts to deter-
mine the number of reads that contained the respective ID-
SNP. To control for location-specific ID-SNPs not present in
the NCBI database, Staphylococcus aureus results were fur-
ther compared with 15 WGS results from two Ghanaian hos-
pitals obtained by Donker et al. [22]. The full details and
pipeline used for the metagenomic comparison can be found
at GitHub [23].
Statistics
Odds ratios were provided by logistic regression between
SSI, cfu/m3 and possible confounders. Variables were first
analysed by univariate regression and entered in a multi-
variable model if P<0.1. In the multivariable model back-
wards elimination was applied and variables were kept if
P<0.05. All non-significant variables were heretafter re-
entered one by one and kept for the final model if P<0.05.
Differences in cfu/m3 between surgeries with and without
subsequent SSI were assessed by Wilcoxon-rank-sum-test.
Analyses were performed with R v.3.4.1 and the epitools-
package [24,25]. Data is reported with 95% confidence
intervals (CIs). P-values <0.05 were considered statistically
significant.
Ethics
Informed consent was mandatory for inclusion. Sampling did
not alter the surgical procedure. For all SSI, an additional swab
was obtained and processed independently of the study. The
study was reviewed and granted approval from Korle-Bu
Teaching Hospital Institutional Review Board (ref. KBTH-IRB/
0004/2016), the Danish National Committee on Health
Research Ethics (ref. 1610254) and the Danish Data Protection
Agency (ref. 2012-58-0004).
Results
SSI and air sampling
A total of 214 patients were found eligible for inclusion, 128
were included and 116 completed follow-up (91%,
Supplementary Figure S1). SSI was suspected for 13 patients
and clinically confirmed for 11 patients (9.5%, 11/116). For SSI
rate according to type of surgery, see Table I. Most infections
were identified post-discharge (N ¼ 8, 73%). Air samples were
available for 124 of 128 cases, and for all cases with SSI
(Supplementary Figure S1). As previously described, mean cfu/
m3 in empty operating rooms were 39 cfu/m3 (95% CI: 36, 42;
range: 12, 81) after 1 h of ventilation [5].
Risk factors associated with SSI
There were significantly higher cfu/m3 during procedures on
patients that developed SSI than the remaining proceduresen bacteria isolated from intraoperative air samples and surgical site
tion, https://doi.org/10.1016/j.jhin.2019.11.007
4 M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx
Table I
Surgical environment and infection rate according to type of surgery
Number of procedures Completed follow-up (N) SSI detected (N) SSI rate
General anaesthesia
Thyroidectomy and parathyroidectomy 26 25 0 0%
Non-cosmetic mammary surgery 30 26 5 19%
Excision of lipomas or subcutaneous tissue 3 3 0 0%
Controlled abdominal surgery 8 7 4 57%
Local anaesthesia
Repair of Inguinal hernia 28 25 1 4%
Non-cosmetic mammary surgery 19 18 0 0%
Excision of lipomas or subcutaneous tissue 14 10 1 10%
For type of surgery, ICD-10-PCS procedure codes were collected, grouped according to similarity, and reported in our previous study [5]. Non-
cosmetic mammary surgery in general anaesthesia were mastectomies and wide local resections (ICD-10-PCS; 0HBT0ZX (N ¼ 6), 0HBT0ZZ (N ¼
1), 0HBU0ZX (N¼ 6), 0HBU0ZZ (N¼ 1), 0HTT0ZZ (N¼ 8), 0HTU0ZZ (N¼ 7) and 0WB80ZZ (N¼ 1)). Non-cosmetic mammary surgery in local anesthesia
were excision biopsies (ICD-10-PCS; 0HBT0ZX (N ¼ 5), 0HBU0ZX (N ¼ 13), and 0HTT0ZZ (N ¼ 1)). Controlled abdominal surgery was defined as
procedures with surgical entry through the peritoneum, with no leak of intestinal fluid (ICD-10-PCS; 0FT40ZZ (N¼ 5), 0GT20ZZ (N¼ 1), and 0WQF0ZZ
(N ¼ 2)). Excision of lipomas or subcutaneous tissue in general anesthesia include ICD-10-PCS; 07B60ZX (N ¼ 1), 0JBF0ZZ (N ¼ 1), and 0YB80ZZ (N ¼
1). Excision of lipomas or subcutaneous tissue in local anesthesia include ICD-10-PCS; 07B20ZX (N ¼ 1), 0JB10ZZ (N ¼ 1), 0JB50ZZ (N ¼ 2), 0JB70ZZ
(N ¼ 2), 0JBF0ZZ (N ¼ 1), 0JBL0ZZ (N ¼ 1), 0QBM0ZZ (N ¼ 1), 0VQ60ZZ (N ¼ 2), 0WBK0ZZ (N ¼ 1), 0XB30ZX (N ¼ 1), and 0YBG0ZZ (N ¼ 1). Thy-
roidectomies include ICD-10-PCS; 0GTK0ZZ (N¼24). Parathyroidectomies include ICD-10-PCS; 0GTQ0ZZ (N¼1) and 0CT90ZZ (N¼1). Repair of inguinal
hernia include ICD-10-PCS; 0YQ50ZZ (N ¼ 21) and 0YQ60ZZ (N ¼ 7). Surgical site infection (SSI) rate is the percentage of patients who completed
follow-up that developed SSI. For a detailed description on human activity and level of intraoperative air contamination see Ref. [5]. All controlled
abdominal cases were classified as clean contaminated wounds by the surgeons. All other cases were classified as clean.(P¼0.03). In eight of 11 SSI cases, mean cfu/m3 of air samples
taken during the individual procedures was >360 cfu/m3
(Figure 1, Table II). Surgeons classified wounds as clean-
contaminated in five cases and three of these developed SSI
(Table II). ASA-score >1, clean-contaminated wounds and
levels of airborne bacteria >360 cfu/m3 were significantly
associated with SSI in the multivariate logistic regression
(Table II).600
500
400
300
200
100
SSI detected SSI not detected
Figure 1. Average air contamination during surgery for each
procedure. The density of airborne bacteria was significantly
higher during procedures that were followed by a surgical site
infection (SSI, red circles) than those that were not (black circles),
P¼0.03. The filled red circle indicates the case where whole-
genome sequencing confirmed a match between bacteria iso-
lated from air and wound sample. The line represents the
Healthcare Infection Society’s recommended maximum level of
180 cfu/m3 of air [32]. The red line represents 360 cfu/m3 of air,
which is chosen as a level of gross contamination.
Please cite this article as: Stauning MA et al., Genetic relationship betwe
infections at a major teaching hospital in Ghana, Journal of Hospital Infe
cfu/m3Identification of bacteria in SSI and air isolates
Seventeen species of bacteria were identified in 11 SSI
samples (Table III). More than one species was isolated from the
majority of SSI. Thirty-nine different bacteria were identified
in air, sampled during cases with subsequent SSI. Most were
non-pathogenic environmental and skin bacteria, but also
pathogens and/or potential reservoirs of multidrug resistance
such as S. aureus, Klebsiella spp. and Acinetobacter spp. were
found (Supplementary Table S1).
Comparison of air and SSI isolates
In eight of 11 cases a match could be found at species level
by MALDI-TOF between one or more bacteria isolated from SSI
and the corresponding air isolates (Table III). Six pairs of SSI and
air isolates showed close similarity after ribotyping and were
successfully whole-genome sequenced. A median of 797,888
(range 241,932 to 1,569,838) read pairs were generated per
sequenced isolate. Median coverage depths of Velvet assem-
blies were 25e116 reads. In one case (ID 98, S. aureus) the
match was confirmed by WGS (Table III, Figure 2).
Metagenomic analysis
A median of 2,819,836 (range 294,597 to 4,721,524) read
pairs/sample were generated by metagenomic analysis. The
analysis identified 82 species with >5000 identified reads/
metagenome in air sampled during cases with SSI
(Supplementary Table S1). Most identified bacteria were non-
pathogenic skin and environmental bacteria, but also patho-
gens such as S. aureus, Klebsiella spp., Enterobacter spp. and
Acinetobacter spp. were found (Supplementary Table S1). For
bacteria with >50 complete NCBI-genomes available, analyses
showed high likelihood of identical bacteria in paired SSI and
air samples for ID 98 and clear negative results in the remainingen bacteria isolated from intraoperative air samples and surgical site
ction, https://doi.org/10.1016/j.jhin.2019.11.007
M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx 5
Please cite this article as: Stauning MA et al., Genetic relationship between bacteria isolated from intraoperative air samples and surgical site
infections at a major teaching hospital in Ghana, Journal of Hospital Infection, https://doi.org/10.1016/j.jhin.2019.11.007
Table II
Patient characteristics and factors associated with surgical site infections
Variable Overall (N ¼ 128) Completed Completed Loss to follow-up Unadjusted 95% CI P Adjusted 95% CI P
follow-up SSI, follow-up, (N ¼ 12) OR OR
not detected SSI detected
(N ¼ 105) (N ¼ 11)
Average mean air 318 cfu/m3 314 cfu/m3 405 cfu/m3 272 cfu/m3
contaminationa 95% CI: 296, 340 95% CI: (291, 337) 95% CI: 314, 495 95% CI: 175, 370
Range: 38, 659 Range: 83, 626 Range: 160, 615 Range: 38, 659
cfu/m3 >180 107 (84%) 89 (85%) 10 (90%) 8 (67%) 1.34 0.23, 25.8 0.79 0.45 0.05, 9.8 0.51
cfu/m3 >360 43 (34%) 34 (32%) 8 (73%) 1 (8%) 5.25 1.42, 25.2 0.02 4.68 1.1, 24.7 0.045
General anaesthesia 67 (52%) 53 (50%) 9 (82%) 5 (42%) 4.58 1.12, 31.0 0.06 1.1 1.51, 9.9 0.95
Preoperative antibiotics 9 (7%) 5 (5%) 3 (27%) 1 (8%) 7.5 1.35, 36.9 0.014 2.54 0.2, 21.0 0.41
ASA >1 47 (37%) 36 (34%) 9 (81%) 2 (16%) 8.63 2.09, 58.6 0.008 8.17 1.77, 0.02
61.1
Male sexb 46 (36%) 36 (34%) 4 (36%) 6 (50%) 1.10 0.27, 3.88 0.89 1.18 0.18, 6.1 0.85
Wound classified clean 5 (4%) 2 (2%) 3 (27%) 0 (0%) 19.3 2.83, 0.003 1.87 1.49, 371 0.03
contaminatedc 164.2
BMI >30 35 (27%) 25 (24%) 5 (45%) 5 (42%) 2.6 0.71, 960 0.13 1.18 0.22, 0.84
6.20
Duration of surgery>60 min 67 (52%) 54 (51%) 8 (73%) 5 (42%) 2.52 0.68, 12.0 0.19 1.35 0.25, 7.1 0.68
Diabetesd 9 (7%) 8 (8%) 1 (9%) 0 (0%) 1.21 0.06, 7.6 0.86 0.88 0.04, 0.91
7.51
Smokersd 1 (1%) 1 (1%) 0 (0%) 0 (0%) d d d d d d
Blood transfusion given 1 (1%) 0 (0%) 1 (9%) 0 (0%) d d d d d d
Unadjusted odds ratio (OR) for surgical site infection (SSI) (N ¼ 11) for each variable tested independently. Adjusted OR is controlled for cfu/m3 >360, wound classified as clean-contaminated
and American Society of Anaesthesiologists physical classification score (ASA) >1. Adjusted OR for >180 cfu/m3 is only controlled for ASA >1 and wound classified as clean-contaminated. BMI,
body mass index; CI, confidence interval.
P less than 0.05 is considered significant and highligthed in bold.
a For each surgery mean cfu/m3of the obtained air samples was calculated. Average air contamination is the average of these values for all surgeries in a given group. Values of cfu/m3 were
gathered as part of our previous study [5]. cfu/m3 is missing from four cases due to technical problems during sampling. These four cases completed follow-up and none developed SSI.
b Classified according to biological sex.
c Surgeon’s assessment at closure. Only wounds classified as clean- or clean-contaminated were included.
d Self-reported by patients.
6 M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx
Please cite this article as: Stauning MA et al., Genetic relationship between bacteria isolated from intraoperative air samples and surgical site
infections at a major teaching hospital in Ghana, Journal of Hospital Infection, https://doi.org/10.1016/j.jhin.2019.11.007
Table III
Comparison of bacteria isolated from surgical site infections and airborne bacteria during surgery
Patient and SSI data Comparison by MALDI-TOF, ribotyping and WGS Comparison by metagenomic analysis
Case Infection Type of SSI Species Match Match found NCBI reference Match found Number of No. of SNPs No. of Fraction of ID- Interpretation of
ID. type surgery Isolate determined found by genome used in by whole available in core positions with SNPs that were meta-genomic
by WGS at Riboprinting whole genome genome complete genome ID-SNPs that identified in at analysis
species comparison comparison reference that are are covered least one
level by genomes unique to by sequence read in
MALDI from NCBI SSI isolate metagenomic the matched
TOF (ID-SNPs) reads metagenome
4 Deep Non-cosmetic 4-S-O-b Serratia marcescens No No d No 32 3194 328 4,26% Low likelihood of
mammary match
surgery
4 Deep Non-cosmetic 4-S-aa-a Pseudomonas No No d No 145 547 177 7,34% Low likelihood of
mammary aeruginosa match
surgery
15 Deep Non-cosmetic 15-S-O-c1 Acinetobacter No No d No 113 266 43 13,95% Low likelihood of
mammary Baumani match
surgery
15 Deep Non-cosmetic 15-S-O-A1 Bacillus Yes Yes (89% d No (different 42 569 266 6,01% Low likelihood of
mammary thuringiensis similarity) species, 71% match
surgery shared genome)
15 Deep Non-cosmetic 15-S-O-d1 Corynebacterium No No d No 3 219 206 53,88% Low likelihood of
mammary jeikeium match
surgery
17 Super-ficial Controlled 17-S-O-A Staphylococcus Yes No d No 11 549 549 31% Low likelihood of
abdominal epidermidis match
surgery
17 Super-ficial Controlled 17-S-O-B Staphylococcus Yes No d No 5 860 860 77% Possible match, but
abdominal haemolyticus not clear due to few
surgery reference genomes
25 Deep Controlled S-25 Staphylococcus Yes No d No 2 31275 730 17,397% Low likelihood of
abdominal hominis match
surgery
33 Superficial Non-cosmetic S-33 Staphylococcus Yes Yes (87% NC_007795.1 No (12344 snp 342 131 55 10,9% Low likelihood of
mammary aureus similarity) difference) match
surgery
40 Superficial Controlled 40-S-O-A1 Staphylococcus Yes Yes NC_007168.1 No (2967 SNP 5 1870 1116 33,15% Low likelihood of
abdominal haemolyticus difference) match
surgery
40 Superficial Controlled 40-S-O-b Achromobactecr No No d No 7 25436 4440 19,52% Low likelihood of
abdominal xylosoxidans match
surgery
49 Deep Non-cosmetic S-49-A Enterobacter No No d No 43 34 3 66,66% Possible match, but
mammary cloacae not clear due to few
surgery reference genomes
and low cover
49 Deep Non-cosmetic S-49-b Pseudomonas No No d No 145 508 192 14% Low likelihood of
mammary aeruginosa match
surgery
63 Organ space Controlled S-63 Staphylococcus Yes Yes (same NC_004461.1 No (23232SNP 11 1291 1161 66,58% Possible match, but
abdominal epidermidis RiboGroup) difference) not clear due to few
surgery reference genomes
79 Deep Non-cosmetic S-79-A Staphylococcus Yes No d No 14 9441 9312 53,84% Low likelihood of
mammary epidermidis match
surgery
M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx 7
Please cite this article as: Stauning MA et al., Genetic relationship between bacteria isolated from intraoperative air samples and surgical site
infections at a major teaching hospital in Ghana, Journal of Hospital Infection, https://doi.org/10.1016/j.jhin.2019.11.007
79 Deep Non-cosmetic S-79-BD Staphylococcus Yes No d No 5 6196 6082 81% Possible match, but
mammary haemolyticus not clear due to few
surgery reference genomes
79 Deep Non-cosmetic S-79-C Staphylococcus Yes Yes (same NZ_ALWK01000001.1 No (7831 SNP 0 d d d Not analysed
mammary arlettae RiboGroup) difference)
surgery
79 Deep Non-cosmetic S-79-FH Staphylococcus Yes Yes NC_007350.1 No (2082 SNP 6 11973 9833 82,42% Possible match, but
mammary saprophyticus difference) not clear due to few
surgery reference genomes
79 Deep Non-cosmetic S-79-G Staphylococcus Yes No d No 3 111682 100544 66,00% Low likelihood of
mammary warneri match
surgery
98 Superficial Repair of S-98 Staphylococcus Yes Yes (same NC_007795.1 Yes (0 SNP 342 66 56 94,64% High likelihood of
Inguinal hernia aureus RiboGroup) difference) match
98 Superficial Repair of S-98-AF Corynebacterium No No d No 3 215 58 17,24% Low likelihood of
Inguinal hernia jeikeium match
98 Superficial Repair of S-98-r2 Pseudomonas No No d No 145 102 10 0% Low likelihood of
Inguinal hernia aeruginosa match
99 Deep Excision of S-99-a Corynebacterium No No d No 0 d d d Not analysed
lipomas or glaucum
subcutaneous
tissue
99 Deep Excision of S-99-b Neisseria sp. No No d No 3 906 38 0% Low likelihood of
lipomas or match
subcutaneous
tissue
Comparison of bacterial isolates from surgical site infection (SSI) and air samples andmatchedmetagenomes. Morphologically different colonies from pooled biomass of air and SSI samples were
subcultured before, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF) identification, ribotyping screening and whole-genome comparison. For whole-
genome comparison we used BacDist with Snippy v.4.1.0 to call single nucleotide polymorphisms (SNPs) in each of the isolates relative to a relevant National Center for Biotechnology
Information (NCBI) reference genome and compare all homologous sequences shared by isolates. Only SNPs at positions covered by at least 10 reads in both isolates with one isolate showing
>80% non-reference reads were included. Metagenomic methods require no subculture after air sampling and biological material is sampled directly from pooled homogenized biomass in air
samples obtained during surgeries with SSI.
*Both air and SSI isolates were initially identified as Bacillus cereus by MALDI-TOF and therefore included in the riboprint analysis. Whole genome sequencing (WGS) subsequently identified the
air sample as B. cereus and the SSI sample as B. thuringiensis. For type of surgery, see Table I.
8 M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx
Please cite this article as: Stauning MA et al., Genetic relationship between bacteria isolated from intraoperative air samples and surgical site
infections at a major teaching hospital in Ghana, Journal of Hospital Infection, https://doi.org/10.1016/j.jhin.2019.11.007
Figure 2. Whole genome-based phylogeny of Staphylococcus aureus. (a) Phylogenetic tree base on the core genome of wound isolates (S-98 and S-33), and air isolates (L-33-Salt-
AA and L-98-c), together with 342 available complete genomes in NCBI and 15 extra genomes collected from Ghana (Donkor et al. [22], green dots). Bacteria from the same surgery
are coloured in the same colours (red or blue). The branch lengths are not represented. (b) Second tree representation of the S-98’s clade. Branch lengths are represented. Colour
indicates surgery origin.
M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx 9
S-33; Staphylococcus aureus S-98; Staphylococcus aureus
28-09-06 04-11-16
100 100
% of major ID SNPs
75 75
100
75
50 50
50
25
25 25
0
0 0
15-S-O-C1; Acinetobacter baumannii 4-s-aa-a; Psedomonas aeruginosa
15-09-2016 07-09-16
100 100
75 75
50 50
25 25
0 0
S-49-b; Pseudomonas aeruginosa S-98-r2; Pseudomonas aeruginosa
05-10-16 04-11-16
100
80
75 60
50 40
25 20
0 0
Metagenome
Figure 3. Metagenomic comparison of bacteria with >50 reference NBCI-genomes available. Bars show fraction of SNPs unique to each
surgical site infection (SSI) isolate (ID-SNPs) that were identified in at least one sequence read in each of the metagenomes. Only positions
with mapped reads are used to determine the fraction, i.e. the presence or absence of ID-SNPs in positions with no mapped reads is not
included in the count. The colour of the bar indicates the percentage of covered positions in which the ID-SNP constitutes the majority of
mapped reads. Metagenomes are represented by date of surgery, and genome of SSI isolates is represented by sample name. Supple-
mental Ghanaian Staphylococcus aureus strains were obtained from Donkor et al. [22]. Surgical dates were: case 4 (07.09.16), case 15
(15.09.16), case 17 (16.09.16), case 25 (23.09.16), case 33 (28.09.16), case 40 (30.09.16), case 49 (05.10.16), case 63 (14.10.16), case 79
(26.10.16), case 98 (04.11.16) and case 99 (08.11.16).
Please cite this article as: Stauning MA et al., Genetic relationship between bacteria isolated from intraoperative air samples and surgical site
infections at a major teaching hospital in Ghana, Journal of Hospital Infection, https://doi.org/10.1016/j.jhin.2019.11.007
% of reads with ID SNP from covered position
04.11.16
05.10.16
07.09.16
08.11.16
14.10.16
15.09.16
04.11.16 04.11.16 16.09.16
23.09.16
05.10.16 05.10.16 26.10.16
28.09.16
07.09.16 07.09.16 30.09.16
Donker et. al. - A
08.11.16 08.11.16 Donker et. al. - B
14.10.16 14.10.16 Donker et. al. - CDonker et. al. - D
15.09.16 15-S-o-c1 Donker et. al. - EDonker et. al. - F
16.09.16 15.09.16 Donker et. al. - GDonker et. al. - H
23.09.16 16.09.16 Donker et. al. - I
Donker et. al. - J
26.10.16 23.09.16 Donker et. al. - K
Donker et. al. - L
28.09.16 26.10.16 Donker et. al. - M
Donker et. al. - N
30.09.16 28.09.16 Donker et. al. - O
L-33-Salt-AA
S-49-b 30.09.16 S-33
04.11.16
04.11.16 04.11.16 05.10.16
07.09.16
05.10.16 05.10.16 08.11.16
14.10.16
07.09.16 07.09.16 15.09.16
16.09.16
08.11.16 08.11.16 23.09.16
26.10.16
14.10.16 14.10.16 28.09.16
30.09.16
15.09.16 15.09.16 Donker et. al. - A
Donker et. al. - B
16.09.16 16.09.16 Donker et. al. - C
Donker et. al. - D
23.09.16 23.09.16 Donker et. al. - E
Donker et. al. - F
26.10.16 26.10.16 Donker et. al. - G
Donker et. al. - H
28.09.16 28.09.16 Donker et. al. - I
Donker et. al. - J
30.09.16 30.09.16 Donker et. al. - K
Donker et. al. - L
S-98-r2 4-S-aa-a Donker et. al. - M
Donker et. al. - N
Donker et. al. - O
L-33-Salt-AA
L-98
S-98
10 M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxxcases (Figure 3, Table III). For bacteria with <50 NCBI-genomes
a match could be suspected for ID 49 and 79 (Table III,
Supplementary Figure S2). Furthermore, results for ID 25, 79
and 99 suggested a match between SSI isolates and air samples
obtained at procedures prior to the date of surgery
(Supplementary Figure S2).Discussion
We found an SSI rate of 9.5 % after clean and clean-
contaminated surgery in this tertiary-level hospital in Ghana.
There was significantly increased risk of SSI when bacterial
counts in the air exceeded 360 cfu/m3 and, in one case (9%, 1/
11) both WGS and metagenomic analysis showed a match
between air- and SSI-isolates. The association between air
contamination and SSI risk remained significant when adjusting
for wound-class and ASA-score. To our knowledge, no previous
studies have examined the association between intraoperative
airborne bacteria and SSI in an LMIC.
No routine system for SSI monitoring is yet implemented in
Ghana. A recent multicentre point-prevalence survey at 10
Ghanaian Hospitals found SSI to be the most common hospital-
acquired infection (HAI) [26]. Our result is slightly lower than
the World Health Organization’s (WHO’s) estimated 14.1% (95%
CI: 11.6e16.8) pooled incidence of SSI in general surgery in
LMIC countries, but the WHO estimate is based on scarce data
and includes both elective and acute procedures [4].
A strength of our study was the active follow-up. With a
scattered population and no register to gather data from out-
patient clinics, the combination of phone calls and clinical
examinations gave a good balance between logistically feasible
and clinically reliable methods. All included patients could
provide one or more telephone numbers and 91% completed
follow-up. A recent review found similar follow-up rates and
feasibility of telephone follow-up to detect SSI in LMICs [27].
Studies testing sensitivity and specificity of phone calls for
detection of SSI in LMIC settings will be of high value.
We foundmore than one species of bacteria in most infected
wounds. Some of the organisms may have been skin con-
taminants. Conversely, mixed infections could arise due to
exposure to high bacterial loads and diversity in the environ-
ment both on admission and after discharge. We have not found
collected reports on the aetiologies of SSI in low- and middle-
income countries and further studies of this will be valuable.
Given the association between air contamination and SSI, we
aimed to genetically compare air- and SSI isolates. This approach
has previously been used to analyse the global outbreak of
Mycobacterium chimaera infections, associated with aerosols
formed in contaminated extra corporal circulation systems [28].
We could demonstrate a match at whole-genome resolution for
such paired isolates in one of 11 cases. Air sampling and wound
swabs are not able to capture all organisms, and individual
bacteria may die during transport or be missed when selecting
colonies for subculture. In particular, the use of only blood agar
for recovery of airborne organisms could miss fungi and bacteria
that require more complex media. Furthermore, the method-
ology used for detection of bacteria in the wounds may have
missed slow-growing and biofilm-associated organisms. Finally,
the study was restricted to a 3-month periodPlease cite this article as: Stauning MA et al., Genetic relationship betwe
infections at a major teaching hospital in Ghana, Journal of Hospital Infe(SeptembereNovember) with relatively average weather. We
did not observe any major changes in cfu counts over the period
[5].Wecannot ruleout that cfu countscould riseduring thedusty
Harmattan period around January. The proportion of SSI caused
by airborne bacteria should thus be considered a conservative
estimate. As a novel approach to find additional matches, we
used metagenomic analysis. This did not show additional
matches, but correctly identified and confirmed the result of
WGS comparison. An advantage ofmetagenomic analysis is high-
throughput taxa detection and classification. To reduce risk of
misclassification, Kaiju settings only allowed one mismatch at
amino acid level to the NCBI database and only species with
>5000 classified reads/metagenomewere included. This results
in a high specificity but reduced sensitivity [20]. Even so, meta-
genomic analysis identified 82 species compared to the 39
identified by sub-culture and MALDI-TOF (Supplementary
Table S1). For bacteria with <50 NCBI-genomes, we are less
able to identify true ID-SNP’s and suggested matches should
therefore be interpreted with caution. No generally accepted
method for comparing metagenomes is yet available. A
reference-free approach has been suggested by Brooks et al.,
basing comparison on de novo assembly of metagenomic reads
[29]. With the vast biomaterial in our air samples we generated
few reads/bacteria and did not find an assembly-based strategy
suitable. De novo assembly from metagenomes is a novel
method, lacking standardization and benchmarking [30,31].
Based on studies from high-income countries as well as
expert opinions, the Healthcare Infection Society recommends
a maximum of 180 cfu/m3 in the air during non-implant surgery
[32]. However no internationally accepted standard exists and,
among others, Scandinavian health authorities set a maximum
as low as 100 cfu/m3 [33,34]. Our results indicate that even
when such low levels of cfu/m3 cannot be reached, a reduction
to <360 cfu/m3 may lead to reduction in SSI risk. Especially in
high risk areas such as operating rooms a reduction of patho-
genic bioaerosols may be valuable. This emphasizes the
importance of considering intraoperative air quality in guide-
lines and interventions to prevent SSI in LMIC. We have pre-
viously suggested that air quality may be improved by reducing
human activity in operating rooms [5]. This challenges the
recently published comprehensive WHO Global Guidelines for
Prevention of Surgical Site Infections that lacks recom-
mendations on staff behaviour [1]. Improved ventilation,
mobile air filtering devices and improved staff clothing are
other strategies to consider [8,32,35]. Active air sampling is a
valuable tool to create awareness of air quality and can be
carried out with limited investment in equipment and basic
laboratory facilities. Interventional studies are needed to test
the clinical effects of improved air quality in LMICs.Acknowledgements
The authors would like to thank all involved staff and
patients at the Korle-Bu Teaching Hospital. Special thanks go to
Dr. Nii Armah Adu-Aryee, Professor Mercy J. Newman, Post doc
Enid Owusu and laboratory technician Amos Akumwena for
support during the study. Special thanks are further given to Dr.
Maame Araba Buadu for assistance with follow-up and Marlene
Høg for assistance with sample preparation and handling.en bacteria isolated from intraoperative air samples and surgical site
ction, https://doi.org/10.1016/j.jhin.2019.11.007
M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx 11Conflict of interest statement
The authors declare that there are no known conflicts of
interest associated with this publication and there has been
no significant financial support for this work that could have
influenced its outcome.
Funding sources
DANIDA (ref. A30025 and 16-P01-GHA), Augustinusfonden
(ref. 17-0408), Rigshospitalets Forskningsfond, University of
Copenhagen (ref. A5287), Dansk Tennisfond (ref.13.02.90),
Knud Højgaards fond (ref. 16-01-0991), Nordea Fonden (ref.
01-2016-001569) and Oticon Fonden (ref. 16-1716). R.L.M. is
supported by the Danish National Research Foundation
(grant number 126). The funders had no role in study design,
data collection and analysis, decision to publish, or prepa-
ration of the manuscript.Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jhin.2019.11.007.
WGS and Metagenome sequences are uploaded to the NCBI
Reference Sequence Database. Bioproject reference
RJNA592050 Release date 31/12/2019 https://www.ncbi.nlm.
nih.gov/refseq/References
[1] Allegranzi B, Bischoff P, Kubila Z, Zayed B, De-Jonge S, Abbas M,
et al. Global guidelines for the prevention of surgical site infec-
tions. 1st ed. Geneva: WHO Document Production Services; 2016.
[2] Brenner P, Nercelles P. Prevention of Surgical Site Infections. In:
Friedman Candace, Arbor A, editors. Basic concepts infect. Con-
trol. 3rd ed. Portadown, N Ireland: International federation of
Infection Control; 2016.
[3] World bank country and lending groups e World bank atlas
method 2017. Available at: https://datahelpdesk.worldbank.org/
knowledgebase/articles/906519-world-bank-country-and-lend-
ing-groups [last accessed September 2017].
[4] Allegranzi B, Bagheri Nejad S, Combescure C, Graafmans W,
Attar H, Donaldson L, et al. Burden of endemic health-care-
associated infection in developing countries: systematic review
and meta-analysis. Lancet 2011;377:228e41.
[5] Stauning MT, Bediako-Bowan A, Andersen LP, Opintan JA,
Labi AK, Kurtzhals JAL, et al. Traffic flow and microbial air con-
tamination in operating rooms at a major teaching hospital in
Ghana. J Hosp Infect 2018;99:263e70.
[6] Lidwell OM, Lowbury EJL, Whyte W, Blowers R, Stanley SJ,
Lowe D. Airborne contamination of wounds in joint replacement
operations: the relationship to sepsis rates. J Hosp Infect
1983;4:111e31.
[7] Lidwell OM, Lowbury EJ, Whyte W, Blowers R, Stanley SJ, Lowe D.
Effect of ultraclean air in operating rooms on deep sepsis in the
joint after total hip or knee replacement: a randomised study. Br
Med J (Clin Res Ed) 1982;285:10e4.
[8] Darouiche RO, Green DM, Harrington MA, Ehni BL, Kougias P,
Bechara CF, et al. Association of airborne microorganisms in the
operating room with implant infections: a randomized controlled
trial. Infect Control Hosp Epidemiol 2017;38:3e10.
[9] ASA House of Delegates. American society of anesthesiologists -
ASA physical status classification system. 2014.Please cite this article as: Stauning MA et al., Genetic relationship betwe
infections at a major teaching hospital in Ghana, Journal of Hospital Infec[10] Center of Disease Prevention and control C. Procedure associated
module: surgical site infection (SSI) event. Atlanta, USA. 2016.
[11] U.S. Department of Health & Human Services. Center for medi-
care and medicaid (CMS) and National Center for Health Statistics
(NCHS) e ICD-10 procedure coding system (ICD-10-PCS) (2016).
n.d. https://www.cms.gov/Medicare/Coding/ICD10/2016-ICD-
10-PCS-and-GEMs.html [last accessed September 2016]
[12] Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC
Definitions of Nosocomial Surgical Site Infections, 1992: A Mod-
ification of CDC Definitions of Surgical Wound Infections. Infect
Control Hosp Epidemiol 1992;13:606e8.
[13] Zerbino DR, Birney E. Velvet: Algorithms for de novo short read
assembly using de Bruijn graphs. Genome Res 2008;18:821e9.
[14] Qualicon Diagnostics LLC H. RiboPrinter system sample prepa-
ration ready reference. n.d. https://www.hygiena.com/
riboprinter.html#resources [last accessed March 2019].
[15] Gabrielaite M. BacDist: snakemake pipeline for bacterial SNP
distance and phylogeny analysis. n.d. https://github.com/
MigleSur/BacDist [last accessed May 2019].
[16] Seemann T. Snippy e rapid haploid variant calling and core
genome alignment. n.d. https://github.com/tseemann/snippy
[last accessed May 2019].
[17] Seemann T, Klötzl F, Page AJ. Pairwise SNP distance matrix from
a FASTA sequence alignment. 2018-10-27 n.d, https://github.
com/tseemann/snp-dists [last accessed May 2019].
[18] Letunic I, Bork P. Interactive tree of life (iTOL) v3: an online tool
for the display and annotation of phylogenetic and other trees.
Nucleic Acids Res 2016;44:W242e5.
[19] Treangen TJ, Ondov BD, Koren S, Phillippy AM. The Harvest
suite for rapid core-genome alignment and visualization of
thousands of intraspecific microbial genomes. Genome Biol
2014;15:524.
[20] Menzel P, Ng KL, Krogh A. Fast and sensitive taxonomic classi-
fication for metagenomics with Kaiju. Nat Commun 2016;7:11257.
[21] Li Heng. Aligning sequence reads, clone sequences and assembly
contigs with BWA-MEM, V2 arXiv.org. Cornell University; 26th May
2013. arXiv:1303.3997v2. https://arxiv.org/abs/1303.3997.
[22] Donkor ES, Jamrozy D, Mills RO, Dankwah T, Amoo PK, Egyir B,
et al. A genomic infection control study for Staphylococcus aur-
eus in two Ghanaian hospitals. Infect Drug Resist
2018;11:1757e65.
[23] Andreu-Sánchez S. Full technical description of the metagenomic
pipeline. SergioSEa/Stauning2019_IDSNP; 2019. https://github.
com/sergioSEa/Stauning2019_IDSNP [last accessed August 2019].
[24] R Core Team. R: a language and environment for statistical
computing. Version 3.4.1 “single candle” copyright (C) 2017. The
R Foundation for Statistical Computing; 2017.
[25] Aragon TJ. epitools: epidemiology Tools. R package version 0.5-
10. 2017.
[26] Labi A-K, Obeng-Nkrumah N, Owusu E, Bjerrum S, Bediako-
Bowan A, Sunkwa-Mills G, et al. Multi-centre point-prevalence
survey of hospital-acquired infections in Ghana. J Hosp Infect
2019;101:60e8.
[27] Sandberg CEJ, Knight SR, Qureshi AU, Pathak S. using tele-
medicine to diagnose surgical site infections in low- and middle-
income countries: systematic review. JMIR MHealth UHealth
2019;7:e13309.
[28] van Ingen J, Kohl TA, Kranzer K, Hasse B, Keller PM, Katarzyna
Szafran
ska A, et al. Global outbreak of severe Mycobacterium
chimaera disease after cardiac surgery: a molecular epidemio-
logical study. Lancet Infect Dis 2017;17:1033e41.
[29] Brooks B, Olm MR, Firek BA, Baker R, Thomas BC, Morowitz MJ,
et al. Strain-resolved analysis of hospital rooms and infants
reveals overlap between the human and room microbiome. Nat
Commun 2017;8:1814.en bacteria isolated from intraoperative air samples and surgical site
tion, https://doi.org/10.1016/j.jhin.2019.11.007
12 M.A. Stauning et al. / Journal of Hospital Infection xxx (xxxx) xxx[30] Sczyrba A, Hofmann P, Belmann P, Koslicki D, Janssen S, Dröge J,
et al. Critical Assessment of Metagenome Interpretationda
benchmark of metagenomics software. Nat Methods 2017;14:
1063e71.
[31] Sangwan N, Xia F, Gilbert JA. Recovering complete and draft
population genomes from metagenome datasets. Microbiome
2016;4:8.
[32] Hoffman PN, Williams J, Stacey A, Bennett AM, Ridgway GL,
Dobson C, et al. Microbiological commissioning and monitoring of
operating theatre suites. J Hosp Infect 2002;52:1e28.Please cite this article as: Stauning MA et al., Genetic relationship betwe
infections at a major teaching hospital in Ghana, Journal of Hospital Infe[33] Sweidish Standards Institute. Mikrobiologisk renhet i oper-
ationsrum - Förebyggande av luftburen smitta - Vägledning och
grundläggande krav. Stockholm. 2012.
[34] Danish Ministry of Health Statens Serum Institut. Nationale
infektionshygiejniske retningslinjer. Copenhagen. 2015.
[35] Dharan S, Pittet D. Environmental controls in operating theatres.
J Hosp Infect 2002;51:79e84.en bacteria isolated from intraoperative air samples and surgical site
ction, https://doi.org/10.1016/j.jhin.2019.11.007