Prospectively randomized adult patients with blood culture positive for S. aureus from 5 university hospitals and 7 tertiary care hospitals in Finland were taken into the trial from January 1999 to May 1999 and from January 2000 to August 2002. Originally, the trial was designed for evaluation on the effect of fluoroquinolone (trovafloxacin or levofloxacin) given in addition to most effective standard treatment in SAB 20 [see Additional file 1]. Clinical, pathological, and echocardiographic data of patients with definite or possible IE according to modified Duke criteria were collected 21, and these cases were included into the analysis. The protocol was approved by the ethics committees of all study sites, and written informed consent was obtained from all patients or their representatives. The trial was conducted in accordance with the Declaration of Helsinki.

The exclusion criteria were age younger than 18 years, imprisonment, proven or suspected pregnancy, breastfeeding, epilepsy, another bacteremia during the previous 28 days, polymicrobial bacteremia (≥ 3 microbes), history of allergy to any quinolone antibiotic, previous tendinitis during fluoroquinolone therapy, prior fluoroquinolone use for more than 5 days before randomization, positive culture for S. aureus only from a central intravenous catheter, meningitis, neutropenia (< 0.5 × 10⁹/L), or failure to supply an informed consent. Patients with bacteremia due to methicillin-resistant S. aureus (MRSA) and a S. aureus strain resistant to any fluoroquinolone were also excluded.

IDUs were defined as patients who had injected drugs within the past 6 months before randomization. Routine bacteriological methods were used to detect S. aureus growth in blood 20 [see Additional file 1]. Bacteremia was hospital-acquired if the first positive blood culture was obtained ≥ 48 hours after admission, or the patient was a resident in a long-term care facility or attended hemodialysis within the preceding 2 months. Prognosis or severity of underlying diseases were classified as healthy, nonfatal, ultimately, or rapidly fatal according to the criteria of McCabe and Jackson 22. Severe sepsis at the time of first blood culture positive for S. aureus was defined as an infectious process leading to organ dysfunction or signs of hypoperfusion or hypotension 23.

The infection focus was documented by clinical, bacteriological, radiological, or pathological investigations. Intravenous catheter-associated bacteremia was defined using the guidelines of the Infectious Diseases Society of America 24. Extracardiac deep infection was defined as pneumonia, deep-seated abscess, osteomyelitis, septic arthritis, meningitis, septic thrombophlebitis, mediastinitis, urinary tract infection, infection of any extracardiac prosthetic device, or recurrent SAB. Clinical findings of IE such as documentation of a new valvular regurgitant murmur, or evidence of any vascular phenomena such as infarction and embolic event or intracranial hemorrhage were detected. Endocarditis of prosthetic valve was classified as early when occurring ≤ 60 days after valve replacement and as late > 60 days after valve replacement. Leucocytosis was determined if white blood cell count was over 12 × 10⁹/L and leucopenia less than 4 × 10⁹/L.

Two-dimensional imaging from multiple tomographic planes and spectral Doppler and color flow imaging were used in all study sites. Transthoracic echocardiography (TTE) and/or transesophageal echocardiography (TEE) as clinically indicated were performed at the discretion of the treating physician and by experienced echocardiographers. The presence of cardiac vegetations, oscillation, paravalvular or intracardiac abscess, new valvular regurgitation, prosthetic valve dehiscence, and valve perforation were recorded 21.

The detailed information of the antimicrobial agents, doses, indications and duration of treatment have been described in our former article 20 [see Additional file 1]. All patients with endocarditis were assigned to receive initially cloxacillin or dicloxacillin (2 g q4 h) for 4 to 6 weeks intravenously. Alternatively, cefuroxime (1.5 g q6 h) or vancomycin (1 g bid) were allowed if a contraindication against the use of semisynthetic penicillins was noted. Aminoglycoside (either tobramycin or netilmicin at 1 mg/kg of body weight q8h) was added to the drug therapy described above for the first 7 days. Rifampicin (450 mg once daily for patients under 50 kg and 600 mg once daily for patients over 50 kg in weight, orally or intravenously) was recommended for at least 4 weeks when an endocarditis or an extracardiac deep infection was suspected or confirmed. Half of the patients were randomized to receive a fluoroquinolone either trovafloxacin or levofloxacin in addition to therapy above. Fluoroquinolone treatment was not observed to have an impact on mortality in SAB or in patients with IE 20 [see Additional file 1]. In cases of renal dysfunction the antibiotic doses were adjusted as recommended by the manufacturers.

All patients were prospectively followed for 3 months after the first positive blood culture for S. aureus. Case fatality rates were recorded at 7 days and 28 days, and at 3 months. Other outcome measures were site of valvular involvement, evidence of extracardiac deep infections or thromboembolic events, duration of fever (recorded in days until axillary temperature was < 37.5 C°), duration of hospitalization and need for cardiac surgery. Laboratory tests were conducted on the day of the first positive blood culture for S. aureus.

Statistical analyses were performed with SPSS^® version 12.0. (SPSS Inc., Chicago, IL, USA). Univariate analyses for categorical variables were calculated with chi-square test or Fisher's exact test as appropriate, and for continuous variables the Mann-Whitney test was used. Odds ratios (OR) with 95% confidence intervals (CI) were as association measures. Cox regression analysis was used to compare survival of IDUs with nonaddicts, and results were given as hazard ratios (HR) with 95% CI. Survival was calculated from the day of the first blood culture positive for S. aureus until 3 months. All tests were 2-tailed and P values of < 0.05 were considered significant.

During the study period, 430 patients with S. aureus bacteremia were included into this trial. The included patients represented 31% of all SAB patients in study hospitals. Endocarditis was observed in 74 of 430 patients (17%), of whom 20 were IDUs and 54 were nonaddicts. Thus, IE was observed in 46% of all 44 IDUs and in 14% of all 386 nonaddicts with SAB (OR, 5.12; 95% CI, 2.65–9.91; P < 0.001). Patients with endocarditis differed from those with SAB only by having significantly more often a predisposing heart disease or a pre-existing liver disease (chronic hepatitis C infection in 19 patients and hepatic cirrhosis in 5 patients), but less preceding trauma (Table 1). Other predisposing characteristics and underlying diseases of the patients with IE did not differ from those with SAB.

Among patients with endocarditis, IDUs as a group were younger and had less predisposing heart conditions, coronary artery disease or diabetes than nonaddicts (Table 2). When the underlying diseases were grouped by the predicted prognoses (McCabe's classification), none among the IDUs had a rapidly fatal or ultimately fatal disease but they were found in 20 of 54 nonaddicts (37%) (P = 0.001). However, there were three IDUs with HIV infection.

Only one drug abuser had a hospital-acquired bacteremia and none associated with the use of the central intravenous catheter (Table 2). Severe sepsis was observed in 9 of 20 IDUs (45%) and in 28 of 54 nonaddicts (52%) without significant difference between the groups (P = 0.79).

According to the modified Duke criteria for endocarditis, 56 patients (76%) were confirmed as definite (10 by pathologic criteria and 46 by clinical criteria), and 18 patients (24%) as possible (Table 3). Left-sided involvement was observed in 93% of nonaddicts whereas in 60% of IDUs had right-sided endocarditis (tricuspid valve involvement in all patients). Among nonaddicts aortic valve was slightly more often involved (44%) than mitral valve (35%).

Prosthetic valve endocarditis, involving left side only, occurred in 17 of 74 patients (23%) and they all presented in nonaddicts. The most of these patients had an early onset of prosthetic valve IE (12 patients).

Echocardiography was performed in 263 of 430 patients (61%) with SAB. Addicts underwent echocardiography significantly more often (91%) compared with nonaddicts (58%) (OR, 7.31; 95% CI, 2.56–20.84; P < 0.001). In endocarditis, TTE was performed in 72 of 74 patients (97%) with no significant difference between IDUs and nonaddicts (Table 3). TEE was performed significantly more often in nonaddicts than in IDUs (74% vs 45%).

A vegetation was evident by echocardiography in 53% of cases and a new regurgitation in 71% of cases without significant differences between the groups (Table 3). Only 4 patients had an intracardial abscess or valve perforation.

There were no differences in the clinical manifestations between the groups except the tendency for more frequent occurrence of various vascular phenomena among IDUs (Table 4). The most common infection focus was skin or soft tissue infection in 46 of 74 patients (62%). An extracardiac deep infection was found in 85% of IDUs and in 89% of nonaddicts (P = 0.70).

Vascular complications, including arterial or venous thromboembolic events, were detected in 60% of IDUs but in only 35% of nonaddicts (P = 0.07) (Table 4). Especially, septic pulmonary embolism was observed only in IDUs. Whereas all coronary artery related diseases were among nonaddicts, all 3 parenchymal embolic events were observed in drug abusers. Congestive heart failure in acute phase was present in 11 of 74 patients (15%) with no difference between the groups. Two patients developed recurrent bacteremia during the 3 months follow-up period. There were no differences in median duration of fever (3 days) or median duration of hospitalization (32 days) between IDUs and nonaddicts (Table 5). Cardiac surgery was performed in 15% of IDUs but only in 7% of nonaddicts (OR, 2.21; 95% CI, 0.45–10.87; P = 0.38).

Case fatality rate of all patients with IE was 23% at 28 days, and 31% at 3 months (Table 5). Mortality was significantly higher in nonaddicts than in addicts at 28 days (OR, 8.00; 95% CI, 0.99–64.94; P = 0.03), and also at 3 months (OR, 5.73; 95% CI, 1.20–27.25; P = 0.02). Significant factors for lower mortality based on univariate analyses were injection drug abuse (HR, 0.22; 95% CI, 0.05–0.92; P = 0.04), age (HR, 1.03; 95% CI, 1.01–1.06; P = 0.006), and none or nonfatal underlying diseases by McCabe's classification (HR, 0.24; 95% CI, 0.10–0.54; P = 0.001). Statistically significant association for mortality were not found with the following: right-sided involvement (HR, 0.30; 95% CI, 0.07–1.30; P = 0.11), left-sided involvement (HR, 2.31; 95% CI, 0.69–7.78; P = 0.18), severe sepsis at the time of first positive blood culture for S. aureus (HR, 1.71; 95% CI, 0.74–3.95; P = 0.21), or arterial embolic events (HR, 2.15; 95% CI, 0.94–4.90; P = 0.07). After adjusting by age and underlying diseases, injection drug abuse was not significantly associated with lower mortality (HR, 0.74; 95% CI, 0.10–5.40; P = 0.77).

At the time of first positive blood culture for S. aureus, leucocytosis was significantly more common in nonaddicts than in IDUs (55% vs 15%, respectively; P = 0.003), and only 2 patients (4%) had leucopenia. The median of serum C-reactive protein on the day of the first positive blood culture for S. aureus was 198 mg/L (range, 60–413 mg/L) in addicts compared with 171 mg/L (range, 5–478 mg/L) in nonaddicts without significant difference between the groups. Only 2 nonaddicts and none of IDUs had alanine aminotransferase level 2-fold above normal limit.

Sixty-five of 74 patients (88%) were treated with cloxacillin or dicloxacillin with no significant difference between the groups. Seven patients (10%) received cefuroxime, and one received vancomycin and one ceftriaxone. The median duration of parenteral antibiotic therapy from the first positive blood culture for S. aureus was 30 days (interquartile range, 24–43 days) in IDUs and 26 days (interquartile range, 13–34 days) in nonaddicts (P = 0.12), respectively.

An aminoglycoside was significantly more often given to addicts (19 of 20 patients) than to nonaddicts (29 of 54 patients) (OR, 16.38; 95% CI, 2.05–131.21; P = 0.001). Eighteen of 20 IDUs (90%) received rifampicin compared with 48 of 54 nonaddicts (89%) with no significant difference between the groups.