On the Etiological Relevance of Escherichia coli and Staphylococcus aureus in Superficial and Deep Infections - A Hypothesis-Forming, Retrospective Assessment.

INTRODUCTION: Escherichia coli and Staphylococcus aureus are important causes of severe diseases like blood stream infections. This study comparatively assessed potential differences in their impact on disease severity in local and systemic infections.
METHODS: Over a 5-year interval, patients in whom either E. coli or S. aureus was detected in superficial or primary sterile compartments were assessed for the primary endpoint death during hospital stay and the secondary endpoints duration of hospital stay and infectious disease as the main diagnosis.
RESULTS: Significance was achieved for the impacts as follows: Superficial infection with S. aureus was associated with an odds ratio of 0.27 regarding the risk of death and of 1.42 regarding infectious disease as main diagnosis. Superficial infection with E. coli was associated with a reduced duration of hospital stay by -2.46 days and a reduced odds ratio of infectious diseases as main diagnosis of 0.04. The hospital stay of patients with E. coli was increased due to third-generation cephalosporin and ciprofloxacin resistance, and in the case of patients with S. aureus due to tetracycline and fusidic acid resistance.
CONCLUSIONS: Reduced disease severity of superficial infections due to both E. coli and S. aureus and resistance-driven prolonged stays in hospital were confirmed, while other outcome parameters were comparable.

1. Introduction

Staphylococcus aureus and Escherichia coli frequently cause superficial and systemic infections [1-10]. Further, both species are among the most frequent causes of bacteremia and sepsis in Western industrialized countries [11-16], with observed proportions of 16.3% to 21.6% for S. aureus and 5.6% to 24.2% for E. coli among all causes of sepsis, associated with considerable morbidity and mortality [17, 18]. Prolonged antibiotic therapy is recommended for S. aureus-associated bacteremia due to high risk of secondary foci of infection and particularly high mortality [19, 20], both for methicillin-susceptible and for methicillin-resistant strains [17]. Along with systemic infections, both species can play a role in superficial infections such as wound infections [21-23] or in urinary tract infections [24, 25].

While the etiological relevance of both of these facultatively pathogenic species under the conditions described can be considered as well documented, the pathogenic potential of individual strains depends on the presence or absence of pathogenic factors and toxins [26, 27] and may vary. The pathogenic factors of E. coli comprise adhesins (fimbrial as well as afimbrial ones and outer membrane proteins), curli, flagella, fimbriae, invasins, iron acquisition factors (siderophores), lipopolysaccharides, pili, polysaccharide capsules, secreted serine proteases, and metalloproteases, and toxins like oligopeptides, AB (alpha and beta subunit)-toxins, and RTX (repeats in toxin) pore-forming toxins [25, 26, 28–32]. For S. aureus, agglutinins, coagulases and staphylokinases, exoenzymes like nucleases and proteases, secreted toxins such as host protease modulators, pore-forming toxins, and superantigens, as well as the ability of forming biofilms due to cell surface-associated proteins, are among the described pathogenic factors [33-41]. As recently discussed elsewhere [42, 43], resistance against bactericidal first-line drugs like beta-lactam antibiotics, caused by enzymes like, e.g., extended spectrum beta-lactamases (ESBL) or carbapenems for E. coli, as well as by penicillin-binding proteins for methicillin-resistant S. aureus (MRSA), can become a severe problem for medical care.

However, previous studies have suggested a trade-off between resistance and pathogenicity [44, 45], assuming that expressing resistance determinants may cost additional energy and could thus decrease the fitness and competitiveness of the strain. Highly varying resistance rates have been observed in S. aureus and E. coli strains in previous assessments [2, 6, 7].

This study was conducted to provide hypotheses to be proven by future prospective studies, focusing on two questions. Firstly, we have performed a comparative head-to-head assessment of various systemic or superficial infections exclusively associated either with S. aureus or with E. coli to get hints on potentially differing disease severity as measured by the outcome parameters, namely, death during hospital stay, duration of hospital stay, and infectious disease as the main diagnosis. Secondly, associations of resistance with the clinical course of documented infections have been assessed to discern hints supporting the above-mentioned fitness cost hypothesis.

2. Patients and Methods

2.1. Study Design

The assessment was conducted as a single-center retrospective observational study over 5 years at a German university hospital. Inclusion criteria will be discussed in detail later under the respective heading. Data were obtained from a laboratory information system (LIS) of the DIN EN ISO 15189-accredited Institute for Medical Microbiology, Virology, and Hygiene of the University Medicine Rostock, Germany. In detail, cases were identified by screening for the search terms: Staphylococcus aureus and Escherichia coli.

For the identification of the bacterial isolates assessed in this study, VITEK 2 identification cards (bioMérieux, Marcyl’Étoile, France) or matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) with a Shimadzu/Kratos “AXIMA Assurance” MALDI-TOF mass spectrometer (Shimadzu Germany Ltd., Duisburg, Germany) were used. As described by the manufacturer (bioMérieux), alpha-cyano-4-hydroxycinnamic acid preparation was carried out for all strains prior to MALDI-TOF assessment. The spectral fingerprints obtained were interpreted using the Vitek MSID IVD-mode database version 3.2.0.-6. (bioMérieux). The equivalence of these methods regarding their diagnostic reliability has repeatedly been confirmed in the literature [46-49]. Antibiotic resistance was analyzed applying Clinical and Laboratory Standards Institute ((CLSI), CLSI M100-S17/M2-A9, M7-A7 January 2007; CLSI M100-S19, M2-A8, M7-A8 January 2009) and European Committee on Antimicrobial

Susceptibility Testing ((EUCAST), Clinical Break Point Version 2 January 2012 and Version 3 January 2013) clinical breakpoints using the appropriate VITEK 2 AST cards in the course of the study. No adjustment for breakpoint changes was performed. Assessed antimicrobials comprised penicillins (ampicillin, ampicillin-sulbactam, oxacillin, and piperacillin–tazobactam), cephalosporins (3rd and 4th generation cephalosporins, as well as cefuroxime and cefoxitin), the carbapenem imipenem, fluoroquinolones (norfloxacin, ciprofloxacin, levofloxacin, and moxifloxacin), aminoglycosides (gentamicin and tobramycin), glycopeptides (vancomycin and teicoplanin), the macrolide erythromycin, the lincosamide clindamycin, the streptogramin quinupristin, tetracycline, the glycylcycline tigecycline, the drug combination co-trimoxazole, fosfomycin, fusidic acid, rifampicin, the oxazolidinone linezolid, the lipopeptide daptomycin, and mupirocin.

After the removal of copy strains, except for the first isolate of each case, inpatient cases were assessed anonymously. Each patient was counted only once. Data were anonymously extracted from the patients’ case files and collected in Microsoft Excel worksheets for statistical analysis.

2.2. Outcome Parameters

Primary and secondary outcome parameters were defined, with death during hospital stay as the primary outcome parameter and the duration of hospital stay, as well as presence or absence of an infectious disease as main diagnosis (localized or systemic infections) versus a non-infectious disease-related main diagnosis (from the fields of cardiovascular diseases, endocrinology, gastroenterology, neurology, orthopedics and trauma surgery, pneumology, rheumatology and tumors, urology, and others) as secondary outcome parameters.

The outcome parameters were assessed for patients with infections due to either S. aureus or E. coli with additional focus on superficial and systemic infections. Superficial infections assessed comprised skin and urinary tract infections, while lower respiratory tract infections, bacteremia, and infections of primary sterile compartments were defined as systemic infections.

Association of antibiotic resistance with the assessed outcome-parameter was also investigated.

2.3. Factors Potentially Affecting the Outcome

In addition to the outcome parameters described above, a number of factors were documented to assess potential effects on the outcome parameters. These variables comprised continuous parameters, such as age, and also noncontinuous parameters, such as gender; isolation site of the strain; non-surgical ward vs. surgical and intensive care ward; main diagnoses from the fields of cardiovascular disease, endocrinology/metabolic disorders, gastroenterology, local infections or systemic infections, neurology, orthopedics/traumatology, pulmonology, rheumatology, neoplasia, urology and others; peak values of leukocytes; procalcitonin (PCT) and C-reactive protein (CRP); and the presence or absence of antibiotic treatment at the time of hospital admission. Leukocytes, CRP, and PCT were semiquantitatively categorized as shown in Table 1. Sample materials in which S. aureus or E. coli were identified were grouped as abscess materials, ascites, aspirates, biopsies and invasive foreign material, blood cultures, bronchial lavage, respiratory secretions, wound swabs, and urine. Antibiotic susceptibility or resistance as defined by EUCAST was recorded for comparison within the species.

Table 1.

Algorithm of semi-quantification of leukocyte counts as well as CRP and PCT concentrations

Parameter	Unit	Reference value	Category 1	Category 2	Category 3	Category 4
Leukocyte count	109/L	4–9	Reduced (<4)	Normal (4–9)Slightly increased	Increased (>9)Moderately increased
CRP	mg/L	<5	Normal (<5)	(5–50)Slightly increased	(>50–100)Moderately increased	Severely increased(>100)
PCT	ng/L	<0.06	Normal (<0.06)	(>0.06–10)	(>10–100)	Severely increased(>100)

The distribution of both the outcome parameters and potentially confounding variables is presented in Table 2 for both S. aureus and E. coli.

Table 2.

Overview on the distribution of individually assessed parameters in patients with S. aureus or E. coli infections

Parameter	S. aureusn = 1040	E. colin = 975	Overalln = 2015
Death n (%)	128 (12.4%)	91 (9.3%)	219 (10.9%)
Gender
Male	648 (62.3%)	423 (43.4%)	1071 (53.2%)
Female	392 (37.7%)	552 (56.6%)	944 (46.9%)
Patient group
Systemic infection	405 (38.9%)	353 (36.2%)	758 (37.6%)
Superficial infection	557 (53.6%)	482 (49.4%)	1039 (51.6%)
Combined superficial and systemic infection	78 (5.5%)	140 (14.4%)	218 (10.8%)
Sample materials
Ascites	3 (0.3%)	10 (1.0%)	13 (0.7%)
Biopsies and invasive foreign material	47 (4.5%)	428 (43.9%)	475 (23.6%)
Blood cultures	234 (22.5%)	245 (25.1%)	479 (23.8%)
Bronchial lavage	17 (1.6%)	19 (2.0%)	36 (1.8%)
Respiratory secretions	127 (12.2%)	37 (3.8%)	164 (8.1%)
Aspirates	14 (1.4%)	35 (3.6%)	49 (2.4%)
Abscess materials	70 (6.7%)	73 (7.5%)	143 (7.1%)
Wound swabs	501 (48.2%)	27 (2.8%)	528 (26.2%)
Urine	27 (2.6%)	101 (10.4%)	128 (6.4%)
Main diagnoses
Endocrinology/metabolic disorders	16 (1.5%)	4 (0.4%)	20 (1.0%)
Gastroenterology	56 (5.4%)	175 (17.8%)	231 (11.5%)
Cardiovascular disease	99 (9.5%)	82 (8.4%)	181 (9.0%)
Local infection	333 (32.0%)	35 (3.6%)	368 (18.3%)
Systemic infection	103 (9.9%)	183 (18.8%)	286 (14.2%)
Rheumatology and neoplasia	108 (10.4%)	131 (13.4%)	239 (11.7%)
Neurological disorder	116 (11.2%)	24 (2.5%)	140 (7.0%)
Orthopedics/Traumatology	79 (7.6%)	24 (2.4%)	103 (5.1%)
Pulmonary disease	60 (5.8%)	80 (8.2%)	140 (7.0%)
Other diseases	37 (3.6%)	50 (5.1%)	87 (4.3%)
Urologic disease	33 (3.2%)	187 (19.2%)	220 (10.9%)
Ward
Surgical	422 (40.6%)	394 (40.4%)	816 (40.5%)
Medical	618 (59.4%)	581 (59.6%)	1199 (59.5%)
Previous antibiotic therapy
Not documented	594 (57.1%)	665 (68.2%)	1259 (62.5%)
Documented	446 (42.9%)	310 (31.8%)	756 (37.5%)
Ampicillin
Susceptible	40 (37.7%)	462 (47.7%)	502 (46.7%)
Resistant	66 (62.3%)	507 (52.3%)	573 (53.3%)
Levofloxacin
Susceptible	841 (85.8%)	n.a.	841 (85.8%)
Resistant	139 (14.2%)	n.a.	139 (14.2%)
Norfloxacin
Susceptible	220 (85.9%)	n.a.	220 (85.9%)
Resistant	36 (14.1%)	n.a.	36 (14.1%)
Ciprofloxacin
Susceptible	221 (86.0%)	728 (75.2%)	949 (77.5%)
Resistant	36 (14.0%)	240 (24.8%)	276 (22.5%)
Moxifloxacin
Susceptible	860 (87.8%)	n.a.	860 (87.8%)
Resistant	119 (12.2%)	n.a.	119 (12.2%)
Erythromycin
Susceptible	880 (89.8%)	n.a.	880 (89.8%)
Resistant	100 (10.2%)	n.a.	100 (10.2%)
Clindamycin
Susceptible	885 (90.3%)	n.a.	885 (90.3%)
Resistant	95 (9.7%)	n.a.	95 (9.7%)
Quinupristin
Susceptible	256 (100%)	n.a.	256 (100%)
Resistant	0	n.a.	0
Gentamicin
Susceptible	957 (97.8%)	915 (94.3%)	1872 (96.1%)
Resistant	22 (2.3%)	55 (5.7%)	77 (4.0%)
Tetracycline
Susceptible	905 (92.4%)	390 (61.2%)	1295 (80.1%)
Resistant	74 (7.6%)	247 (38.8%)	321 (19.9%)
Tobramycin
Susceptible	822 (97.4%)	n.a.	822 (97.4%)
Resistant	22 (2.6%)	n.a.	22 (2.6%)
Tigecycline
Susceptible	722 (100%)	426 (100%)	1148 (100%)
Resistant	0	0	0
Co-trimoxazole
Susceptible	972 (99.3%)	686 (70.4%)	1658 (84.9%)
Resistant	7 (0.7%)	289 (29.6%)	296 (15.2%)
Fosfomycin
Susceptible	972 (99.4%)	334 (99.4%)	1306 (99.4%)
Resistant	6 (0.6%)	2 (0.6%)	8 (0.6%)
Fusidic acid
Susceptible	950 (97.5%)	n.a.	950 (97.5%)
Resistant	24 (2.5%)	n.a.	24 (2.5%)
Rifampicin
Susceptible	976 (99.7%)	n.a.	976 (99.7%)
Resistant	3 (0.3%)	n.a.	3 (0.3%)
Oxacillin
Susceptible	1035 (100%)	n.a.	1035 (100%)
Resistant	0	n.a.	0
Ampicillin–Sulbactam
Susceptible	948 (100%)	580 (65.2%)	1528 (83.1%)
Resistant	0	310 (84.8%)	310 (16.9%)
Cefoxitin
Susceptible	964 (100%)	n.a.	964 (100%)
Resistant	0	n.a.	0
Cefuroxime
Susceptible	948 (100%)	n.a.	948 (100%)
Resistant	0	n.a.	0
Imipenem
Susceptible	948 (100%)	974 (100%)	1922 (100%)
Resistant	0	0	0
Teicoplanin
Susceptible	978 (100%)	n.a.	978 (100%)
Resistant	0	n.a.	0
Vancomycin
Susceptible	978 (100%)	n.a.	978 (100%)
Resistant	0	n.a.	0
Linezolid
Susceptible	980 (100%)	n.a.	980 (100%)
Resistant	0	n.a.	0
Daptomycin
Susceptible	137 (99.3%)	n.a.	137 (99.3%)
Resistant	1 (0.7%)	n.a.	1 (0.7%)
Mupirocin
Susceptible	963 (99.8%)	n.a.	963 (99.8%)
Resistant	2 (0.2%)	n.a.	2 (0.2%)
Third-generation cephalosporins
Susceptible	n.a.	835 (88.6%)	835 (88.6%)
Resistant	n.a.	108 (11.5%)	108 (11.5%)
Fourth-generation cephalosporins
Susceptible	n.a.	901 (92.6%)	901 (92.6%)
Resistant	n.a.	72 (7.4%)	72 (7.4%)
Piperacillin–Tazobactam
Susceptible	n.a.	747 (85.4%)	747 (85.4%)
Resistant	n.a.	128 (14.6%)	128 (14.6%)
Days in hospital	1440	974	2014
(available for (n))
Mean (SD)	19.6 (19.2)	15.3 (16.3)	17.6 (18.0)
Median	14.0	10.0	12.0
Days to sample	1040	975	2015
acquisition
(available for (n))
Mean (SD)	4.7 (10.4)	3.8 (7.5)	4.3 (9.1)
Median	1.0	1.0	1.0
Age (available for (n))	1040	975	2015
Mean (SD)	58.3 (22.0)	69.3 (15.1)	63.6 (19.7)
Median	63.0	72.0	69.0
CRP (available for (n))	1009	899	1908
Mean (SD)	1.9(1.1)	2.1 (1.0)	2.0 (1.0)
Median	2.0	2.0	2.0
PCT (available for (n))	242	157	399
Mean (SD)	1.2 (0.5)	1.3 (0.2)	1.3 (0.5)
Median	1.0	1.0	1.0
Leukocytes	998	975	1973
(available for (n))
Mean (SD)	2.5 (0.6)	2.5 (0.6)	2.5 (0.6)
Median	3.0	3.0	3.0

2.4. Inclusion and Exclusion Criteria

Patients were included if either S. aureus or E. coli was identified in the microbiological laboratory in any clinical sample material and if clinical information from the case files was available.

Incompleteness of the assessable dataset alone was not considered as exclusion criterion, although it led to a reduction in the number of interpretable cases.

2.5. Statistical Assessment

Statistical assessment was done using STATA 15.1 (StataCorp, USA) with an exploratory aim. Binary logistic regression was used for the binary endpoint parameters, namely, death and infectious disease as main diagnosis. Linear regression was used for the endpoint parameter duration of hospital stay. All models were implemented as backward selection models with a significance level of 0.1 to exclude parameters from the model. Modeling was performed for the whole dataset (global modeling), as well as for patients with either S. aureus or E. coli (local modeling). Comparisons between disjoint subpopulations generally refer to their complement. As there have been no defined references in this study, for practical reasons, the category with the lowest score or the first appearance in each group has been used as a reference for the calculations. Parameters with insufficient numbers for regression analysis were excluded from the modeling. For the endpoint-parameter infectious disease as a main diagnosis, the parameters of main diagnoses were not included into the model due to lack of independence.

2.6. Ethics

Ethical clearance for the assessment was obtained from the Ethics Committee of the University Medicine Rostock (Registration number A 2014–0054). The study was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Do to the retrospective design of the study, the assessment was allowed by the ethics committee in an anonymous way without consent to participate.

3. Results

3.1. Study Population

After removal of copy strains, the study population comprised 2015 cases: 1040 isolations of S. aureus and 975 isolations of E. coli. These isolations were associated with systemic infections in 758 cases, with superficial infections in 1039 cases, and with combined systemic and superficial infections in 218 cases.

3.2. Assessed Parameters

Detailed information on the distribution of assessed patient and strain characteristics is provided in Table 2. Thereby, parameters are presented as distributed by either species of the bacterial isolates (S. aureus or E. coli), as well as for the whole assessed population. Based on those data, global modeling for the whole dataset, as well as local modeling for patients with either S. aureus or E. coli with a focus on the study endpoints, was performed; the results are presented in the following sections.

3.3. Assessment of the Study Endpoints

Neither isolation of S. aureus nor that of E. coli was associated with any of the endpoint parameters in the global modeling. However, superficial infection with S. aureus was negatively associated with the primary outcome parameter death (P < 0.001), but it was positively associated with infectious disease as the main diagnosis (P = 0.013). Superficial infection with E. coli, in contrast, was associated with shorter duration of hospital stay (P < 0.001) and negatively associated with infectious disease as main diagnosis (P < 0.001). Several variables potentially affected the primary and secondary outcome parameters as shown in Tables 3–5.

3.4. Associations between Resistance and Superficial or Systemic Infections

Enhanced resistance, i.e., resistance to 3 or more antibiotic substance classes, was not generally positively or negatively associated with invasive infections.

Resistance against tetracycline (P = 0.027) and fusidic acid (P < 0.001) in patients with S. aureus and against third-generation cephalosporins (P < 0.001) and ciprofloxacin (P = 0.097) in patients with E. coli was associated with increased duration of hospital stay. Also, there was a tendency for an increased risk of death in patients with clindamycinresistant S. aureus (P = 0.083) (Tables 3 and 4).

Table 3.

Explorative binary logistic regression model with backward selection for the global model with patients with S. aureus and/or E. coli infections. Association with the outcome parameter death

Association with the outcome parameter death	Global modeling with both patients with S. aureus and patients with E. coli		Modeling with patients with S. aureus		Modeling with patients with E. coli
	Odds ratio with0.95 CI	P value	Odds ratio with0.95 CI	P value	Odds ratio with0.95 CI	P value
Isolation from biopsies and invasive foreign material	0.52 (0.32, 0.85)	0.009	n.a.	n.a.	0.09 (0.04, 0.29)	<0.001
Isolation from blood cultures	0.71 (0.47, 1.06)	0.091	n.a.	n.a.	0.13 (0.06, 0.29)	<0.001
Isolation from respiratory secretions	n.a.	n.a.	n.a.	n.a.	0.31 (0.09, 1.11)	0.073
Isolation from aspirates	n.a.	n.a.	n.a.	n.a.	0.26 (0.07, 1.00)	0.050
Isolation from abscess materials	2.34 (1.47, 3.72)	<0.001	n.a.	n.a.	n.a.	n.a.
Isolation from urine	n.a.	n.a.	n.a.	n.a.	0.21 (0.08, 0.56)	0.002
Gastroenterology	0.60 (0.36, 1.00)	0.051	n.a.	n.a.	n.a.	n.a.
Cardiovascular disease	n.a.	n.a.	n.a.	n.a.	2.33 (1.00, 5.47)	0.050
Local infection	0.61 (0.40, 0.91)	0.016	n.a.	n.a.	n.a.	n.a.
Rheumatology and neoplasia	0.56 (0.32, 0.97)	0.039	n.a.	n.a.	n.a.	n.a.
Pulmonary disease	n.a.	n.a.	n.a.	n.a.	2.28 (1.04, 4.97)	0.039
Urologic disease	0.04 (0.01, 0.30)	0.002	n.a.	n.a.	0.12 (0.02, 0.90)	0.039
Documented previous antibiotic therapy	1.39 (1.02, 1.89)	0.037	n.a.	n.a.	1.78 (0.99, 3.20)	0.055
Leukocyte count	0.79 (0.62, 1.01)	0.063	n.a.	n.a.	n.a.	n.a.
Age	n.a.	n.a.	n.a.	n.a.	1.03 (1.00, 1.05)	0.019
Superficial infections	n.a.	n.a.	0.27 (0.17, 0.41)	<0.001	n.a.	n.a.
Clindamycin resistance	n.a.	n.a.	1.73 (0.93, 3.22)	0.083	n.a.	n.a.
	N 1883, Pseudo R2 0.0684		N 851, Pseudo R2 0.0604		N 745, Pseudo R2 0.1901

Odds ratios >1 indicate a risk association with the outcome “death,” but odds ratios <1 indicate a protective association.

Table 4.

Explorative linear regression model with backward selection for the global model with patients with S. aureus and/or E. coli infections. Association with the outcome parameter duration of hospital stay

Association with the outcome parameter duration of hospital stay	Global modeling with both patients with S. aureus and patients with E. coli		Modeling with patients with S. aureus		Modeling with patients with E. coli
	Odds ratio with0.95 CI	P value	Odds ratio with0.95 CI	P value	Odds ratio with0.95 CI	P value
Biopsies and invasive foreign material	–1.36 (–2.23, –0.49)	0.002	n.a.	n.a.	–2.00 (–3.14, –0.88)	0.001
Blood cultures	n.a.	n.a.	1.40 (0.08, 2.71)	0.037	–3.15 (–4.49, –1.81)	<0.001
Bronchial lavage	n.a.	n.a.	n.a.	n.a.	–5.51 (–8.45, –2.56)	<0.001
Respiratory secretions	n.a.	n.a.	2.16 (0.52, 3.80)	0.010	–2.89 (–5.48, –0.29)	0.029
Aspirates	n.a.	n.a.	n.a.	n.a.	–2.55 (–4.66, –0.44)	0.018
Wound swabs	–1.16 (–1.98, –0.34)	0.005	n.a.	n.a.	n.a.	n.a.
Gastroenterology	3.18 (2.11, 4.25)	<0.001	5.18 (2.84, 7.52)	<0.001	2.57 (1.36, 3.79)	<0.001
Cardiovascular disease	1.68 (0.50, 2.85)	0.005	n.a.	n.a.	2.26 (0.70, 3.81)	0.005
Rheumatology and neoplasia	2.90 (1.79, 4.01)	<0.001	2.98 (1.01, 4.06)	0.003	2.86 (1.54, 4.17)	<0.001
Pulmonary disease	2.01 (0.72, 3.30)	0.002	n.a.	n.a.	2.83 (1.27, 4.38)	<0.001
Other diseases	2.79 (1.14, 4.43)	0.001	3.15 (0.19, 6.19)	0.037	2.39 (0.45, 4.32)	0.016
Urologic disease	1.99 (0.85, 3.13)	0.001	n.a.	n.a.	2.18 (0.99, 3.38)	<0.001
Previous antibiotic therapy	1.46 (0.79, 2.13)	<0.001	2.10 (1.01, 3.20)	<0.001	n.a.	n.a.
CRP	–0.47 (–0.80, –0.14)	0.005	–0.73 (–1.28, –0.18)	0.010	–0.44 (–0.83, –0.04)	0.032
Age	–0.036 (–0.05, –0.02)	<0.001	–0.04 (–0.06, –0.01)	0.002	n.a.	n.a.
Third-generation cephalosporin resistance	n.a.	n.a.	n.a.	n.a.	2.59 (1.38, 3.81)	<0.001
Tetracycline resistance	n.a.	n.a.	2.22 (0.25, 4.18)	0.027	n.a.	n.a.
Fusidic acid resistance	n.a.	n.a.	5.67 (2.26, 9.08)	0.001	n.a.	n.a.
Ciprofloxacin resistance	n.a.	n.a.	n.a.	n.a.	0.81 (–0.15, 1.76)	0.097
Superficial infection	n.a.	n.a.	n.a.	n.a.	–2.46 (–3.71, –1.21)	<0.001
	N 1883, adjusted R2 0.3987		N 918, adjusted R2 0.3863		N 774, adjusted R2 0.5340

Coefficients >0 indicate a prolonging association with the endpoint days to discharge (parameter extends on average by the shown number of days), coefficients <0 indicate a shortening association (parameter reduces on average by the shown number of days).

4. Discussion

The study was conducted to analyze any differences in the etiological relevance of S. aureus and E. coli over a study period of 5 years with inpatients at a German university hospital with superficial or systemic infections. The focus was on the primary endpoint death during hospital stay, and the two secondary endpoints, duration of hospital stay and a main diagnosis of infectious disease. Potential associations between resistance and severity of the infectious diseases were also assessed.

The results of the assessment differed for the different cases. First of all, neither S. aureus nor E. coli as individual species alone were associated with any of the outcome parameters. As expected, superficial infections with S. aureus were associated with reduced risk of death compared with systemic infections; most interestingly, this association was not seen for superficial E. coli infections. The high relevance of E. coli for urinary tract infections [30] (which were classed in the superficial infection group in this study) and their specific complications is a likely reason for this. Pathogenic factors associated with uropathogenic E. coli comprise fimbriae, curli, pili, capsules, iron scavenger receptors, flagella, toxins, and lipopolysaccharides [25, 26, 30]. In contrast, superficial E. coli infections were associated with a decrease of duration of hospital stay and the likelihood of infectious disease as the main diagnosis compared with systemic infections. Also in contrast, the prominence of S. aureus-associated skin and soft-tissue infections [50] allowed an association of superficial S. aureus infections and infectious disease as main diagnosis. Clumping factor B, Panton–Valentine leukocidin, and bi-component pore-forming toxins are prominent virulence factors that have been associated with skin and soft tissue infections due to S. aureus [51-54]. Admittedly, molecular screening for virulence factors was beyond the scope of this study.

Antibiotic drug resistance was only weakly associated with disease severity as measured by the chosen outcome parameters. Potential resistance factors associated with the observed disease severity-associated resistance patterns comprise ribosomal methylases or efflux pumps causing clindamycin resistance [55], tetracycline resistance genes of the tet gene family associated with tetracycline resistance [56], or fusidic acid resistance genes of the fus gen family associated with fusidic acid resistance [57] in S. aureus, as well as beta-lactamases associated with 3rd generation cephalosporin resistance in E. coli [55]. In all observed cases, higher levels of resistance were associated with increased disease severity. Accordingly, this hypothesis-forming assessment provided no indications supporting the trade-off theory of fitness cost and antibiotic resistance, as suggested in the literature [44, 45]. Admittedly, the study was only an explorative assessment, and pathogenic factors of the isolates were not assessed, an undeniable limitation of the study. Further, applied interpretation standards of resistance testing have been changed in the course of the study, which interferes with interpretability, although major discrepancies are nevertheless unlikely.

The study has a number of further limitations. Since the study is a retrospective exploration, conclusions regarding associative relationships can only provide hypotheses. In addition, especially in the global model of both groups with either S. aureus or E. coli and in the local model for patients with S. aureus, the models account for only a small part of the total dispersion, so it must be assumed that a considerable amount of information that could have been explanatory was not collected. However, the hypotheses derived from this study can be used to guide future prospective studies.

5. Conclusion

In conclusion, the study did not show specific association of disease severity, as defined by the endpoint parameters with the isolation of either S. aureus or E. coli, while, as expected, superficial infections were generally associated with milder diseases in comparison to systemic infections, as suggested by the outcome parameters. Interestingly, a reduced risk of death was shown for superficial infections due to S. aureus but not for those due to E. coli, in comparison to systemic infections. As far as can be determined despite the limitations of the study, resistance was associated with increased disease severity as defined by the endpoint parameters, so the phenomenon of trade-off between resistance and pathogenicity was not supported.

Table 5.

Explorative binary logistic regression model with backward selection for the global model with patients with S. aureus and/or E. coli infections. Association with the outcome parameter infectious disease as main diagnosis

Association with the outcome parameter infectious disease as main diagnosis	Global modeling with both patients with S. aureus and patients with E. coli		Modeling with patients with S. aureus		Modeling with patients with E. coli
	Odds ratio with0.95 CI	P value	Odds ratio with0.95 CI	P value	Odds ratio with0.95 CI	P value
Biopsies and invasive foreign material	n.a.	n.a.	n.a.	n.a.	4.26 (1.78, 10.19)	0.001
Blood cultures	5.53 (4.15, 7.38)	<0.001	n.a.	n.a.	5.69 (2.78, 11.68)	<0.001
Bronchial lavage	n.a.	n.a.	0.23 (0.05, 1.07)	0.061	n.a.	n.a.
Respiratory secretions	n.a.	n.a.	0.20 (0.11, 0.37)	<0.001	n.a.	n.a.
Aspirates	5.40 (2.92, 10.01)	<0.001	n.a.	n.a.	14.19 (5.24, 38.40)	<0.001
Abscess materials	5.52 (3.61, 8.46)	<0.001	4.88 (2.58, 9.25)	<0.001	n.a.	n.a.
Wound swabs	8.08 (6.06, 10.77)	<0.001	1.83 (1.33, 2.51)	<0.001	651.55 (157.87, 2689.04)	<0.001
Urine	n.a.	n.a.	0.13 (0.03, 0.60)	0.008	18.99 (6.41, 56.22)	<0.001
Age	0.99 (0.98, 0.99)	<0.001	0.99 (0.98, 0.99)	0.001	1.01 (1.00, 1.03)	0.083
Gender	1.25 (1.00, 1.54)	0.045	n.a.	n.a.	n.a.	n.a.
Leukocyte count	1.47 (1.22, 1.75)	<0.001	1.35 (1.06, 1.73)	0.016	1.71 (1.22, 2.41)	0.002
CRP	n.a.	n.a.	0.86 (0.74, 0.99)	0.034	1.43 (1.12, 1.81)	0.004
Superficial infections	n.a.	n.a.	1.42 (1.08, 1.87)	0.013	0.04 (0.02, 0.08)	<0.001
Surgical and intensive care wards	n.a.	n.a.	n.a.	n.a.	0.18 (0.11, 0.30)	<0.001
	N 1885, Pseudo R2 0.1385		N 986, Pseudo R2 0.1292		N 899, Pseudo R2 0.3473,

Odds ratios >1 indicate a risk association with the outcome infectious disease as main diagnosis, whereas odds ratios <1 indicate a protective association.