Appropriateness of empirical antibiotic therapy and added value of adjunctive gentamicin in patients with septic shock: a prospective cohort study in the ICU

Abstract

Objectives

To determine the appropriateness of empiric antibiotic therapy and the possible benefit of adding short-course gentamicin in septic shock patients with abdominal, urogenital, or an unknown focus. Secondary objectives were the effect of gentamicin addition on shock reversal and the incidence of a fungal infection.

Methods

Microbiological cultures, antibiotic treatment, and antibiotic resistance patterns of the cultured microorganisms were recorded during the first 5 days of admission. Inappropriate antibiotic therapy was defined as a prescription within the first 24 h that did not cover cultured bacteria during the first 5 days of admission and was determined in the overall group and in patients receiving adjunctive gentamicin (combination therapy) versus patients receiving monotherapy. Binomial logistic regression analysis was used to investigate the association of gentamicin addition with shock reversal.

Results

Of 203 septic shock patients, with abdominal (n = 143), urogenital (n = 27) or unknown (n = 33) focus, 115 patients received monotherapy, and 88 patients received combination therapy. Inappropriate therapy occurred in 29 patients (14%), more frequently in monotherapy (17%) versus combination therapy (10%). Combination therapy would have been effective in 55% of patients with inappropriate monotherapy. We found no association between gentamicin addition and shock reversal (p = .223). A fungal infection was present in 22 patients (11%).

Conclusion

Inappropriate empirical antibiotic therapy occurs in 17% of septic shock patients receiving monotherapy. In 55% of these patients, additional gentamicin would have resulted in appropriate therapy. When clinical course is unfavourable, lowering the threshold for administering adjunctive aminoglycoside and antifungal therapy should be considered.

Keywords:

• Sepsis
• septic shock
• antibiotic therapy
• gentamicin
• aminoglycoside
• intensive care

Introduction

Septic shock is a subcategory of sepsis with profound circulatory, cellular, and metabolic abnormalities [1], associated with high mortality [2]. One of the most important treatments of these critically ill patients is the swift initiation of appropriate broad-spectrum antibiotic therapy [3]. Failure to start appropriate antibiotic treatment in patients with septic shock, substantially increases morbidity and mortality [4–6].

Inappropriate antibiotic treatment can result from an infection with extended spectrum beta-lactamase-producing (ESBL) Gram-negative pathogens or fungal infection. ESBL faecal carriage is increasing worldwide, especially in the ICU [7]. The risk of infection with these pathogens is higher in patients with sepsis of suspected abdominal, urogenital, or unknown origin [8]. Several studies report inappropriate antibiotic treatment for intensive care unit (ICU) patients in up to 20–30% of cases [5,6,9].

The addition of an aminoglycoside broadens the empirical antibiotic spectrum, thereby reducing the risk of inappropriate treatment compared to monotherapy [10]. It also has the property of killing bacteria fast and concentration-dependent, working in synergy with beta-lactam antibiotics [11]. Although short courses (one or two doses) are advocated in antibiotic guidelines [12], controversy remains on the added value of aminoglycoside therapy. A meta-analysis did not show faster shock reversal or improved survival in sepsis patients treated with an adjunctive aminoglycoside. In this meta-analysis, an increased risk of nephrotoxicity and renal failure was observed [13], in contrast to other studies showing that the addition of an aminoglycoside seems to be safe regarding nephrotoxicity in sepsis patients [14–16].

Current Surviving Sepsis Guidelines recommend adding an agent targeting Gram-negative bacteria to the empiric regimen in critically ill patients at high risk of infection with multi-drug resistant pathogens to ensure at least one of the agents is effective [3]. However, the appropriateness of empirical antibiotic therapy and the potential of gentamicin, as the agent targeting Gram-negative bacteria, in such a highly selective group of septic shock patients with a high risk of suffering from either potentially resistant Gram-negative microorganisms, or fungal infection is largely unknown in the ICU.

The aim of this study was to investigate both the appropriateness of empiric antibiotic therapy and the possible benefit of adding short-course gentamicin in septic shock patients with abdominal, urogenital, or unknown focus. Secondary objectives were investigating the effect of gentamicin on shock reversal and assessing the incidence of fungal infection in this population.

Methods

Settings

This study was conducted at the ICU of the Maastricht University Medical Centre+ (MUMC+), a tertiary care, 715-bed university hospital in the Netherlands with 33 intensive care unit beds and approximately 2200 admissions annually. Our hospital is a tertiary referral centre for trauma, neuro-surgical, neurological, and extracorporeal life support (ECLS) patients.

Patient population

We systematically screened all patients admitted to the ICU for sepsis since 2012 and enrolled all patients admitted with sepsis in a prospectively recorded database. Admission with sepsis was defined as any ICU admission clinically coded as infection and at least one organ dysfunction, according to the Surviving Sepsis Campaign guidelines of 2012 [17]. For this study, we included the subset of patients with septic shock due to (suspected) abdominal, urogenital, and unknown focus of infection between 2012 and 2017. Septic shock was defined as sepsis with circulatory failure and lactate levels >2 mmol/L despite adequate fluid resuscitation and requiring vasopressor treatment to maintain adequate mean arterial pressure (MAP) of ≥65 mmHg, according to the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-)3 criteria [1].

Patients were treated according to Surviving Sepsis Campaign guidelines, with the prompt administration of broad-spectrum antibiotics, fluid resuscitation, and vasopressors. Intravascular volume replacement was guided by either clinical variables and/or pulse contour measurements (Picco®, Pulse index Contour Continuous Cardiac Output, Pulsion Medical Systems, Germany) and/or echocardiography. Norepinephrine was the vasopressor of choice to treat persistent hypotension after adequate fluid resuscitation. Empirical antibiotic treatment was started in consultation with the clinical microbiologist and was de-escalated when considered appropriate. In our centre, the first-choice antibiotic therapy for sepsis with abdominal, urogenital, or unknown focus is an extended-spectrum penicillin/beta-lactamase inhibitor combination (amoxicillin/clavulanic acid or piperacillin/tazobactam) with (combination therapy group) or without (monotherapy group) gentamicin (7 mg/kg dosage). According to local guidelines, patients with community-acquired sepsis without neutropenia were treated with amoxicillin/clavulanic acid and gentamicin whereas patients with nosocomial sepsis without neutropenia were treated with piperacillin/tazobactam [12]. In selective patients with suspicion of or cultured earlier with micro-organisms that were not susceptible to these antibiotics, an even broader antibiotic such as meropenem was administered. It was at the discretion of the treating physician to add a short course (no more than 3 dosages) of adjunctive aminoglycoside treatment (gentamicin 7 mg/kg). Patient management and antibiotic strategy were discussed daily in the multidisciplinary meeting in the presence of a clinical microbiologist. All mechanically ventilated patients received selective digestive decontamination (SDD) from the day of tracheal intubation until ICU discharge. The SDD suspension consisted of Tobramycin, Colistin and Amphotericin-B and was administered in the oral cavity and into the gut through a nasogastric tube. In patients not receiving broad-spectrum antibiotics, Cefotaxime intravenously was administered during the first four days to bridge the first period in which SDD is not yet fully functional.

Collection of data and study-design

Data were recorded regarding sex, age, source of infection, comorbidities, active malignancy, and severity of disease (APACHE-II score) for all included patients. Furthermore, the need for invasive mechanical ventilation, vasopressor need, serum lactate levels, renal function, and need for continuous veno-venous hemofiltration (CVVH) were also recorded. The study was approved by the local Medical Ethical Committee of the MUMC+ (reference number 2018-0689).

Microbiological data and antibiotic resistance

The administered antibiotics and relevant microbiological cultures were recorded by manually analysing the medical records of the included patients from the Patient Data Management Systems (PDMS): Systems, Applications, and Products in Data Processing

(SAP©, Walldorf, Germany) and Intellispace Critical Care and Anaesthesia (ICCA^© Philips Healthcare, Amsterdam, the Netherlands). Microbiological cultures, antibiotic treatment, and antibiotic resistance patterns of the cultured microorganisms were recorded during the first 5 days (120 h) of ICU admission. Blood cultures were registered, and results from other cultures were expressed either quantitatively by colony forming units (CFU) of the cultured micro-organism(s) or, in case of a liquid culture, semi-quantitatively by a four-point scale; sporadic, little, intermediate, or abundant growth. We included all positive blood cultures and other cultures with growth of bacteria when present at levels greater than 10⁴ CFU or labelled as abundantly present on the four-point scale. Bacterial resistance was defined as both intermediately susceptible as well as truly resistant.

Definitions

Appropriateness of antibiotic therapy

Inappropriate antibiotic therapy was defined as a prescription of antibiotics within the first 24 h that did not cover bacteria present in cultures obtained between day 0 and day 5 after admission to the ICU.

Shock reversal

Shock reversal definition was based on vasopressor requirement and was defined as a decrease of 25% per day in norepinephrine requirement or an absolute requirement of <0.15 mcg/kg/minute.

Outcome measures

The primary outcome measure was the occurrence of inappropriate antibiotic therapy. The primary outcome was determined both in the overall patient group as well as in patients treated with gentamicin (combination therapy) versus patients not treated with gentamicin (monotherapy).

Secondary outcomes were shock reversal during the first five days of ICU admission and the presence of (invasive) fungal infection.

Statistical analysis

Statistical analyses were performed using IBM SPSS version 25 (SPSS Inc., Chicago, USA). Descriptive statistics, independent samples t-tests, Chi-square tests, and one-way ANOVA were used to describe patient demographics and compare the treatment groups. Binomial logistic regression was performed to investigate parameters associated with shock reversal. Sex, age, comorbidity, severity of disease, and variables with p < .10 (norepinephrine dosage on day 1, highest serum lactate level in the first 24 h) and treatment with gentamicin were applied in the model. Goodness of fit for the logistic regression was assessed by the Hosmer-Lemeshow statistic. All tests were two-tailed and an α < 0.05 was considered statistically significant. Data are presented as mean (±standard deviation), median (±interquartile range), mode, or number (%) as appropriate.

Results

Study population

The flowchart of the study is depicted in Figure 1. We excluded 78 patients, 77 because of negative cultures or because no cultures were taken (22%, 3%, and 2% of patients with abdominal, urogenital, and unknown focus of septic shock, respectively), and one patient due to lack of consent. In all, 203 septic shock patients were included in the analysis, 115 (57%) patients with monotherapy, and 88 (43%) patients with combination therapy.

Figure 1. Flowchart of the study.

Baseline characteristics (Table 1), show a septic shock population with predominantly male (62%), older than 65 years of age (56%), invasively ventilated (79%), surgical (64%) patients, who were already admitted to hospital (47% of patients in the ward, 21% in the operating theatre versus 29% of patients admitted directly from the emergency department). Mean APACHE II score was 27 ± 8 with a mean norepinephrine dosage of 0.37 ± 0.35 mcg/kg/minute, and mean lactate levels were 6.1 ± 4.2 mmol/L. Both comorbidity and active malignancy were present in almost half of the included patients, and ICU mortality was 40%.

Table 1. Baseline characteristics of the study population.

	Overall group (n = 203)	Abdominal (n = 143)	Urogenital (n = 27)	Unknown (n = 33)	P Value^*
Sex					0.069
Male	126 (62%)	86 (60%)	14 (52%)	26 (79%)
Age					0.245
<44 yr	15 (7%)	10 (7%)	1 (4%)	4 (12%)
45–54 yr	25 (12%)	19 (13%)	1 (4%)	5 (16%)
55–64 yr	50 (25%)	36 (25%)	4 (15%)	10 (30%)
65–74 yr	63 (31%)	43 (30%)	10 (37%)	10 (30%)
>75 yr	50 (25%)	35 (25%)	11 (40%)	4 (12%)
Invasive ventilation	160 (79%)	122 (85%)	13 (48%)	25 (76%)	<0.001
Severe comorbidy^a	94 (46%)	59 (41%)	12 (44%)	23 (70%)	0.012
Acute renal failure^b	84 (41%)	51 (36%)	11 (40%)	22 (67%)	0.005
Comatose at admission^c	13 (6%)	8 (6%)	0 (0%)	5 (15%)	0.045
Acidosis at admission^d	102 (50%)	69 (48%)	11 (40%)	22 (67%)	0.092
Leucocytes >40 or <1 mm^3	48 (24%)	24 (17%)	5 (19%)	19 (58%)	<0.001
Patient category					<0.001
Surgical	130 (64%)	112 (78%)	16 (59%)	2 (6%)
Medical	73 (36%)	31 (22%)	11 (41%)	31 (94%)
Patient admitted from					<0.001
Emergency department	59 (29%)	31 (22%)	16 (59%)	12 (36%)
Ward	95 (47%)	70 (49%)	6 (22%)	19 (58%)
Other ICU	7 (3%)	4 (3%)	1 (4%)	2 (6%)
Operating room	42 (21%)	38 (26%)	4 (15%)	0 (0%)
Norepinephrine					0.001
mcg/kg/min ± SD	0.37 ± 0.35	0.35 ± 0.32	0.24 ± 0.21	0.56 ± 0.49
Lactate^f mmol/L ± SD	6.1 ± 4.2	5.8 ± 4.0	4.4 ± 1.6	8.7 ± 5.5	<0.001
Thrombocytes 10^9/L ± SD	214 ± 149	244 ± 146	153 ± 84	131 ± 156	<0.001
Active malignancy n(%)	98 (48%)	69 (48%)	9 (33%)	20 (61%)	0.110
APACHE-II score	27 ± 8	25 ± 8	26 ± 4	34 ± 6	<0.001
ICU mortality	82 (40%)	57 (40%)	3 (11%)	22 (67%)	<0.001
Empirical therapy
Piperacillin/tazobactam	100 (49%)	75 (53%)	7 (26%)	18 (55%)	0.338
Amoxicillin/clavulanic acid	66 (33%)	43 (30%)	15 (54%)	8 (24%)
Meropenem	15 (7%)	12 (8%)	1 (5%)	2 (6%)
Others	22 (11%)	13(9%)	4 (15%)	5 (15%)

Data are n (%) unless otherwise specified.

^aNYHA IV cardiac failure; chronic restrictive or obstructive respiratory failure with functional impairment, hepatic cirrhosis, immunosuppression; ^bCreatinine >175 µmol/L; ^cGlasgow Coma Scale ≤8 without sedation; ^dpH <7.25; ^eHighest dosage during the first 24 h after admission; ^fHighest value during the first 24 h after admission. *p-value derived from one-way ANOVA or Chi-square test as appropriate.

Septic shock was predominantly caused by an abdominal focus (143 patients, 70%). Baseline characteristics differed significantly between the three distinct infectious foci, showing that patients with a urogenital focus had a lower incidence of invasive ventilation, lower need for vasopressors, lower lactate (4.4 ± 1.6), and lower mortality (11%) than the two other groups. The APACHE II score (34 ± 6), vasopressor need (0.56 ± 0.49 mcg/kg/minute), lactate (8.7 ± 5.5), comorbidity (70%), presence of an active malignancy (61%), and mortality (67%) were highest in the patients with an unknown focus of infection.

Comparison of patients receiving combination therapy and monotherapy

Patients receiving gentamicin had a higher need for vasopressor treatment than those treated with monotherapy (0.45 ± 0.39 vs. 0.31 ± 0.31 µg/kg/min, p = .006). Furthermore, patients receiving gentamicin were more often admitted from the emergency department (35 patients (40%) versus 24 patients (21%), p = .018). There was also a difference in antibiotic therapy as the patients in the combination group were treated more often with amoxicillin clavulanic acid (44 vs 24%, p = .002). Treatment with piperacillin/tazobactam and meropenem did not significantly differ between both groups. Only 2 patients (both in the monotherapy group) received vancomycin and 4 patients (also all in the monotherapy group) received ciprofloxacin. We found no differences between the monotherapy and the combination group in the other characteristics, including renal function and number of patients requiring renal replacement therapy (Table 2).

Table 2. Patient characteristics per antibiotic treatment group.

	No gentamicin addition N = 115	Gentamicin addition N = 88	p Value^*
Sex			0.291
Male	75 (65%)	51 (58%)
Age			0.242
≤44 yr	7 (6%)	8 (9%)
45–54 yr	18 (15%)	7 (8%)
55–64 yr	25 (22%)	25 (28%)
65–74 yr	33 (29%)	30 (34%)
≥75 yr	32 (28%)	18 (21%)
Invasive ventilation	92 (80%)	68 (77%)	0.637
Severe comorbidity^a	51 (44%)	43 (49%)	0.523
Acute renal failure^b	47 (41%)	37 (42%)	0.866
Comatose at admission^c	8 (7%)	5 (6%)	0.713
Acidosis at admission^d	54 (47%)	48 (55%)	0.284
Leucocytes >40 or <1 mm^3	26 (23%)	22 (25%)	0.691
Patient category			0.199
Surgical	78 (68%)	52 (59%)
Medical	37 (32%)	36 (41%)
Patient admitted from			0.018
Emergency department	24 (21%)	35 (40%)
Ward	60 (52%)	35 (40%)
Other ICU	3 (3%)	4 (4%)
Operating room	28 (24%)	14 (16%)
Norepinephrine ± SD,*			0.006
µg/kg/min	0.31 ± 0.31	0.45 ± 0.39
Lactate ± SD, mmol/L	5.6 ± 3.9	6.7 ± 4.6	0.069
Thrombocytes 10^9/L ± SD	220 ± 153	205 ± 142	0.475
Active malignancy (%)	54 (47%)	44 (50%)	0.667
APACHE-II score	26 ± 8	28 ± 8	0.207
ICU mortality	42 (37%)	40 (46%)	0.199
Shock reversibility	90 (79%)	56 (63%)	0.016
Empirical therapy			0.037
Piperacillin/tazobactam	63 (55%)	37 (42%)	0.072
Amoxicillin/clavulanic acid	27 (24%)	39 (44%)	0.002
Meropenem
Ciprofloxacin	11 (10%)	4 (5%)	0.175
Cefotaxim	4 (4%)	0 (0%)	0.135
Ceftriaxon	2 (2%)	0 (0%)	0.51
Vancomycin	2 (2%)	3 (3%)	0.65
Cefuroxim	2 (2%)	0 (0%)	0.51
Amoxicillin	1 (1%)	1 (1%)	1.0
Tobramycin	1 (1%)	0 (0%)	1.0
Others/missing	1 (1%)	0 (0%)	1.0
	1 (1%)	4 (5%)
Sepsis focus			0.449
Abdominal	85 (74%)	58 (66%)
Urogenital	13 (11%)	14 (16%)
Unknown	17 (15%)	16 (18%)
Creatinine µmol/L
Day 1	174 ± 123	166 ± 102	0.636
Day 2	172 ± 114	179 ± 119	0.661
Day 3	164 ± 103	173 ± 107	0.582
Day 4	155 ± 106	164 ± 107	0.610
Day 5	138 ± 92	156 ± 107	0.330
CVVH^e	25 (22%)	13 (15%)	0.722

Data are n (%) unless otherwise specified.

^aNYHA IV cardiac failure; chronic restrictive or obstructive respiratory failure with functional impairment, hepatic cirrhosis, immunosuppression; ^bCreatinine >175 µmol/L; ^cGlasgow Coma Scale ≤8 without sedation; ^dpH <7.25; ^eContinuous Venovenous Hemofiltration needed on day 5 after ICU admission.*p-values derived from either independent samples T-test, Chi-square or Fishers’ exact test as appropriate.

Occurrence of (in)appropriate antibiotic therapy

In the overall group, inappropriate antibiotic therapy was found in 29 (14%) patients. In 20 out of 115 (17%) patients treated with monotherapy, therapy was inadequate. If an aminoglycoside would have been added, appropriate antibacterial therapy would increase from 83% (95/115 patients) to 92% (106/115 patients. Inappropriate therapy was most frequently encountered in patients receiving amoxicillin/clavulanic acid (30%) or piperacillin/tazobactam (16%) monotherapy.

Inappropriate treatment occurred in 9 out of 88 (10%) patients receiving combination therapy. In 17 out of 88 (19%) patients in the combination group, gentamicin was the only effective component of the antibiotic therapy because the pathogen showed resistance against the primary agent (amoxicillin/clavulanic acid or piperacillin/tazobactam) (Table 3). We found inadequate therapy in 14% of piperacillin/tazobactam/gentamicin prescriptions and 10% of amoxicillin/clavulanic acid/gentamicin prescriptions. Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa and Enterobacter cloacae were the most frequent cultured Gram-negative micro-organisms resistant to the prescribed antibiotic (Table S1).

Table 3. Antibiotic prescriptions, number of inappropriate antibiotic treatments and incidence of fungal infections.

	#prescriptions	Inappropriate therapy	Gentamicin would have been effective	Fungal infection
No Gentamicin added
Meropenem	11	1	–	2
Piperacillin/tazobactam	63	10	6	6
Amoxicillin/clavulanic acid	27	8	3	7
Other	14	1	2	2
Total	115	20	11	17
Gentamicin added			Gentamicin is the only effective component of combination therapy
Meropenem & Gentamicin	4	–	–	–
Piperacillin/tazobactam & Gentamicin	37	5	7	3
Amoxicillin/clavulanic acid & Gentamicin	39	4	10	2
Other	8	–	–	–
Total	88	9	17	5

Incidence of fungal infection

A fungal infection was present in 11% (22/203 patients,17 in the group without gentamicin and 5 in the group treated with gentamicin). Fungi were cultured from abscesses in 21 patients, in 5 patients fungi were sampled from the blood cultures. In total, 6 different Candida species were found in the cultures, most often Candida albicans (18 cases), occasionally patients had positive cultures with more than one Candida species (Table S2).

Effect of antibiotic therapy on shock reversal

Mortality was 46% in the combination group vs. 37% in the monotherapy group (p = .199). Shock reversibility was significantly higher in the monotherapy group (90 out of 115 patients, 79%) when compared to the combination group (56 out of 88 patients, 63%) (p = .016) (Table 2). Logistic regression analysis was performed to identify possible predictors for shock reversal. Hosmer-Lemeshow goodness of fit test indicated no evidence of poor fit (Chi-square 5.9, p = .658). Binomial logistic regression analysis pointed out that the addition of gentamicin was not associated with shock reversal (B 1.578 (0.758 − 3.286), p = .223). Norepinephrine dosage at day 1, APACHE II score, and highest lactate in the first 24 h were associated with shock reversal (Figure 2).

Figure 2. Forest plot of binomial logistic regression analysis for possible predictors of shock reversal.

Discussion

Inappropriate empirical antibiotic therapy was prescribed in 14% (29/203) patients admitted to the ICU with septic shock with an abdominal, urogenital, or unknown focus. Inappropriate antibiotic therapy occurred more often in the monotherapy group when compared to the combination therapy group with adjunctive gentamicin (17 vs 10%). In 55% of patients treated with inadequate monotherapy, gentamicin addition would have resulted in appropriate antibiotic therapy. In 11% of septic shock patients a fungal infection was present. We found no difference in ICU mortality between the monotherapy and the combination group. Although shock reversal occurred significantly more in the monotherapy group, we found no association between shock reversal and administration of gentamicin.

Several studies have assessed the added value of aminoglycoside therapy in the non-ICU [18] and ICU population [10, 14, 19] with sepsis. A non-ICU study, including 626 sepsis patients with Gram-negative bloodstream infection, did not show clinical benefit of short-course aminoglycoside addition regarding mortality, although the risk of inappropriate therapy was reduced eight-fold with combination therapy. In this study, the influence of clinical severity of disease on whether or not to add aminoglycoside was addressed. The association between severity of disease and adding an aminoglycoside could confound by indication and seems to be the case in our study. Gentamicin was prescribed to more severely ill patients who needed significantly higher vasopressor dosage and tended to have a higher APACHE II score and lactate levels.

In a prospective ICU trial, including 648 severe sepsis or septic shock patients (245 of them receiving gentamicin), short-course gentamicin therapy was not associated with faster reversal of shock or improved survival. Still, it did increase the incidence of renal failure. Two retrospective ICU studies did not associate short-course aminoglycoside addition with nephrotoxicity, nor did they show a clinical benefit for gentamicin. The study results in our centre extend these findings to a septic shock population at high risk (abdominal, urogenital, or unknown focus) for extended- spectrum-beta-lactamase Gram-negative microorganisms and fungal infection, despite the higher number of appropriate therapy courses in the gentamicin group (90 vs 83%).

The number of inappropriate therapies in our centre (17 vs 10% in the monotherapy and combination therapy, respectively) falls in the upper range of other studies reporting appropriateness of mono- or combination- antibiotic therapy, ranging from 6–30% [6,9,20]. This can be explained by the fact that included patients in our study were severely ill and at relatively high risk for resistant Gram-negative microorganisms and fungal infections, due to the aetiology of the infection. Thus, the clinician should still take in account the possibility of inadequate therapy, even in this severely ill patient group with broad antibiotic coverage.

Patients receiving gentamicin were more often treated with amoxicillin/clavulanic acid than patients receiving monotherapy in this study (44 vs 24%, p = .002). We believe this can be explained by the fact that in the gentamicin group, more patients are admitted directly from the emergency department (40 vs 21%). Amoxicillin/clavulanic acid is the first choice antibiotic therapy for patients admitted from out-of-hospital according to the antibiotic treatment protocol in our centre. Although more patients in the monotherapy group received meropenem, this difference was not statistically different between both groups (10 vs 5%, p = .175). The fact that less patients in the combination group received broad spectrum antibiotics, could also be explained as an advantage of gentamicin combination therapies.

Fungi were cultured in 11% of included patients. Two-thirds of the study population were surgical patients, and 78% of abdominal infections were surgical ICU patients. Recent major surgery is a known risk factor for fungal infection next to total parental nutrition, immunocompromised state, and comorbidities like chronic liver and renal failure [21,22]. Strikingly, in a recent prospective matched case-control study including 192 patients with Candida bloodstream infection and 411 control patients, exposure to aminoglycoside treatment was associated with candidemia [23]. In our study, Candida infection occurred more in the monotherapy group (15%) than in the gentamicin treated group (6%).

Our study has several strengths. First of all, the study population comprises a representative septic shock population with very ill patients and high mortality. Appropriateness of antibiotic therapy is investigated in a relevant group with a high risk of putatively resistant microorganisms, not earlier explored in detail, to the best of our knowledge. Furthermore, randomised controlled trials are not to be expected in this complex setting. All included patients were treated according to the same local protocol in a setting with a low prevalence of antibiotic resistance.

There are several limitations to this study. First of all, this was a single-centre study in a hospital with secondary and tertiary care, so the generalisability and extrapolation of the results can be debated. Although the results apply strongly to our centre and environment and can be different in other settings, we believe that they give valuable insights into the pathogens involved and the appropriateness of therapy in this severely ill patient group. The limited sample size of this highly selected population might influence power in this study. Moreover,there were 76 patients excluded from the analysis because there were no cultures taken (for blood cultures ranging from 3% in urogenital infections to 22% in abdominal infections) or cultures were all negative. We further analysed this finding, and the absence of blood cultures in patients with abdominal sepsis could partially be explained by the fact that more than half of these patients died within 24 h. Apparently, these patients were in end-stage septic shock, and other more urgent treatments had priority. Furthermore, confounding by indication might be reflected by the fact that the treating physician decided to administrate gentamicin, and patients receiving gentamicin tended to be more severely ill. Nevertheless, the only significant difference between the treatment groups was the vasopressor dosage, which was higher in the gentamicin group. Data on dosage and drug monitoring are not presented in the present study; however, the intended administered dosage was 7 mg/kg gentamicin in all treated patients during the study period.

In conclusion, inappropriate empirical antibiotic therapy is frequently encountered in severely ill septic shock patients with an abdominal, urogenital, or unknown aetiology.

Empirical monotherapy was inappropriate in almost one in five patients, and in more than half of these cases, adding gentamicin would have resulted in appropriate coverage. In 11% of all patients, a fungal infection was present. In case of unfavourable clinical course under antibiotic monotherapy lowering the threshold for administering adjunctive aminoglycoside and antifungal therapy should be considered.

Author contributions

RS and DB designed the study, RD has written the manuscript and performed the statistical analyses, RG and JK collected the data, GO, DP and IH critically reviewed the manuscript.

Disclosure statement

All authors report no conflict of interest relevant to this article.