Continuous vs. intermittent cefotaxime administration in patients with chronic obstructive pulmonary disease and respiratory tract infections: pharmacokinetics/pharmacodynamics, bacterial susceptibility and clinical efficacy


A. R. H. van Zanten MD, Internist-intensivist, Department of Intensive Care, Gelderse Vallei Hospital, PO Box 9025, 6710 HN Ede, the Netherlands.
Tel: + 31 3 1843 4115
Fax: + 31 3 1843 4116



To compare the pharmacokinetics/pharmacodynamics, antibiotic resistance and clinical efficacy of continuous (CA) vs. intermittent administration (IA) of cefotaxime in patients with obstructive pulmonary disease and respiratory infections.


A randomized controlled prospective nonblinded study was performed in 93 consecutive hospitalized patients requiring antibiotics for acute exacerbations of chronic obstructive pulmonary disease. Forty-seven patients received 2 g of cefotaxime intravenously over 24 h plus a loading dose of 1 g, and 46 patients were given the drug intermittently (1 g three times daily).


Similar pathogens were identified in both groups, being mostly Haemophilus influenzae (51%), Streptococcus pneumoniae (21%) and Moraxella catharralis (18%). Mean minimal inhibitory concentration (MIC) values were also similar before and after treatment in both groups. Clinical cure was achieved in 37/40 (93%) (CA) vs. 40/43 (93%) (IA) of patients (= 0.93). In microbiologically evaluable patients, criteria such as 70% of treatment time with antibiotic concentrations ≥ MIC (CA 100%vs. IA 60% of patients) and/or ≥ 5 × MIC (CA 100%vs. IA 55% of patients) were significantly better following continuous administration (< 0.01). Samples with suboptimal antibiotic concentrations were found in 0% of CA vs. 65% of IA patients (< 0.01).


Although clinical cure rates were comparable, continuous cefotaxime administration led to significantly greater proportions of concentrations > MIC and > 5 × MIC compared with intermittent dosing. Continuous administration of cefotaxime at a lower dose [2 g (CA) vs. 3 g (CI)] is equally effective pharmacodynamically and microbiologically, may be more cost-effective and offers at least the same clinical efficacy. Based on these observations, we recommend continuous administration of cefotaxime as the preferred mode of administration.


Based on current knowledge of the pharmacokinetics and pharmacodynamics of β-lactam antibiotics, continuous infusion of these drugs is usually recommended as the most effective form of treatment. The reason for this is that various studies have shown a close relationship between treatment efficacy and the duration of time that serum concentrations of the antibiotic exceed the minimal inhibitory concentration (MIC) for the causative microorganism [1, 2]. In contrast to other classes of antibiotics such as aminoglycosides and quinolones, β-lactam drugs do not have a significant ‘postantibiotic effect’. This implies that the fluctuations in serum antibiotic concentrations associated with intermittent administration could lead to concentrations below the MIC between scheduled doses. This could decrease the efficacy of the antibiotic strategy and increase the risk of development of resistance [3, 4]. In addition, maximum bacterial killing rates may be achieved only at concentrations four or five times the MIC of the causative pathogen [1]. According to the literature, bacterial killing rates by β-lactam antibiotics are largely determined by the percentage of time – at least 60–70% of the dosing interval – that antibiotic concentrations are four to five times the MIC, whereas bacterial resistance is related to the period when antibiotic concentrations are lower than the MIC [6, 7].

Recommendations regarding the use of β-lactam antibiotics are usually based on the above considerations. However, although these hypotheses have been extensively studied in vitro and in animals, they have not been thoroughly validated in the clinical setting, although some authors have called for such studies to be performed [5].

Accordingly, we performed a prospective randomized trial to compare the pharmacokinetics and pharmacodynamics of continuous vs. intermittent administration of the β-lactam antibiotic cefotaxime in patients with chronic obstructive pulmonary disease (COPD). We also assessed the clinical course of the disease, as well as evaluating antibiotic resistance.

Materials and methods

Study design

A randomized, controlled, prospective, open-label study was performed to compare continuous vs. intermittent antibiotic administration. Patients meeting our inclusion criteria were enrolled during a 1-year period and were randomized to receive either a continuous infusion of cefotaxime (2 g over 24 h) after an initial loading dose of 1 g given over 30 min (group 1), or cefotaxime 1 g intravenously in three dosages per day at intervals of 8 h infused over a period of 30 min (group 2).

The doses for continuous administration were chosen to reach target concentrations of five times the MIC for bacteria such as Streptococcus pneumoniae, Haemophilus influenzae and Moraxella catharralis, which are the most frequently cultured microorganisms in our COPD population. These bacteria had MICs < 0.5 mg l−1 in > 99% of our historical cultures. Therefore, to achieve the necessary pharmacodynamic effect, the historical MIC was multiplied by the desired 5 × MIC, demonstrating that a target steady-state concentration (Css) of 2.5 mg l−1 was required for optimal clinical and antibacterial effects. The dose for the continuous infusion was calculated from the equation: Dose (mg h−1) = Css (mg l−1) × Cl (l h−1). Based on previous data on cefotaxime [half-life (t1/2) of 0.9–1.4 h, total body clearance (Cl) of 15.6–26.4 l h−1, and volume of distribution (Vd) of 0.3–0.4 l kg−1], we estimated that a continuous infusion of 2 g cefotaxime over 24 h would generate steady-state concentrations of 3.2–5.3 mg l−1. Furthermore, it was calculated that a loading dose of 200 mg would be adequate to achieve concentrations ≥ 4 × MIC for patients weighing ≤ 100 kg [dose in mg = concentration at steady state (mg l−1) × Vd (l kg−1) × body weight (kg)]. However, for practical reasons a loading dose of 1 g was chosen.

The planned duration of cefotaxime administration was 7 days in both groups. This treatment period could have been extended on clinical grounds at the discretion of the attending physician. Clinical criteria for extending the treatment period were either ongoing production of purulent secretions, fever ≥ 38.5 °C, elevated white blood cell count, or high C-reactive protein (CRP > 20 mg l−1). The study was approved by the hospital ethics committee and prior informed consent was obtained from all patients prior to enrolment.

Study population

Ninety-three consecutive patients aged ≥ 18 years (range 34–76 years) requiring hospital admission and antibiotic treatment for moderate to severe acute exacerbations of COPD (GOLD classes 2–4) were enrolled. The requirement of antibiotic treatment was determined by a panel of four pulmonary physicians on the basis of clinical criteria (including fever ≥ 38.5 °C, sputum production, colour of the patients’ sputum, dyspnoea, tachypnoea and white blood cell count) and/or the presence of pulmonary infiltrate(s) on the chest radiograph. Exclusion criteria included suspected or proven resistance to cefotaxime, administration of antibiotics in the preceding 48 h, allergy to β-lactam antibiotics, bilirubin concentrations > 20 µmol l−1, serum creatinin concentrations > 120 µmol l−1 and whole blood count < 3.0 × 109 l−1.

Clinical assessment

Successful treatment was defined and scored as follows:

  • 1Clinical cure in combination with infiltrate resolution or improvement on chest X-ray; OR
  • 2Clinical improvement and resolution on chest X-ray; OR
  • 3No requirement for antibiotic treatment within 48 h after discontinuation of cefotaxime administration.

Treatment failure was defined as:

  • 1The necessity to change the antibiotic regimen during treatment; OR
  • 2The requirement of repeat antibiotic administration within 48 h after discontinuation of cefotaxime administration; OR
  • 3Clinical deterioration during treatment; OR
  • 4Death related to treatment failure.

Cases that did not fit this success/failure classification were classified as non-evaluable.

Microbiological analysis

Two sputum samples for culture were obtained from each patient, one before initiation of antibiotic treatment, and one 2 days following cessation of therapy. MIC values for cefotaxime were determined using the Epsilometer test (E-test®; AB Biodisc, Stockholm, Sweden).

Drug analysis

To determine serum concentrations of cefotaxime, blood samples were drawn 0, 30 min, 3 h and 7 h after the start of the loading dose on days 2, 4 and 6. For the continuous infusion, the start of the loading dose was used as the reference time and on days 2, 4 and 6 samples were drawn 24, 48 and 96 h later. Blood samples were collected in tubes without anticoagulant and centrifuged at 478 g for 10 min to obtain serum. Samples were frozen and stored at − 20 °C until analysis, which was performed within 2 weeks. Serum cefotaxime concentrations were determined using high-performance liquid chromatography (HPLC).

The equipment comprised a P2000, AS3000 and SpectraFOCUS HPLC (Thermo Separation Products, San Jose, CA, USA), and a CI8-column (Chromspher 5CI8/3 mm/10 cm; Varian Analytical Instruments, Bergen op Zoom, the Netherlands). The mobile phase consisted of phosphate buffer (6.8 g l−1, pH 5.5) and methanol in a 92 : 8 ratio (v/v) at a flow rate of 1.3 ml min−1. Cefazoline was used as the internal standard. Bovine calf serum (S6648; Sigma Chemical Co., St Louis, MO, USA) was used to prepare standards and controls as required. Serum protein was removed from each sample by solid-phase extraction (SepPak tC18; Waters, Milford, MA, USA). The assay was linear over a range of 0.2–125 mg l−1. The limit of detection was 0.07 mg l−1. The within-run mean accuracy and precision (CV%) at the lower limit of quantification (0.2 mg l−1) (n = 5) was 96.0% (12.4%). The mean within-run accuracy and precision (CV%) at 10 mg l−1 and 125 mg l−1 were 106.0% (2.3%) and 103.4% (1.87%), respectively. Between-run precision (n = 10) was 107.8% (2.71%) at 10 mg l−1 and 107.0% (4.57%) at 125 mg l−1.

Pharmacokinetic analysis

The elimination half-life (t1/2), the area under the curve extrapolated to infinity (AUC), total body clearance (CL) and volume of distribution (at steady state; Vss) were determined by one-compartment pharmacokinetic analysis for the intermittent administration group (MwPharm software version 3.15; Mediware, Heerenveen, the Netherlands).

AUC for the continuous administration group was determined from the mean serum drug concentrations of cefotaxime, CL from the total daily dose and Cmean using the expression 2000/(24 × Cmean).

Pharmacodynamic analysis

We determined whether predefined target concentrations of cefotaxime (0.5, 1.0 and 2.5 mg l−1) could be achieved at all times during the dosing interval in both treatment groups. Assessments were also made of antibiotic concentrations in relation to the MIC and 5 × MIC of the pathogen cultured from each patient (MICactual). Furthermore, numbers of patients with serum drug concentrations below MICactual and 5 × MICactual for more than 30% of the overall treatment time were determined. The average time during which antibiotic concentrations were < 5 × MICactual was calculated in patients with single cefotaxime concentrations < 5 × MIC for > 30% of the dosing interval of 8 h.

Statistical analysis

Results are presented as mean ± SD. Analysis of variance for repeated measurements (anova) was used for comparisons within patients. Student’s unpaired t-test was used to compare the two study groups and Welch corrections were made when appropriate. Clinical results were evaluated using the fourfold table of the χ2 test. Statistical significance was accepted for P < 0.05.


Of the 93 patients initially enrolled, 10 were excluded for the following reasons: death due to cardiac failure (n = 5); antibiotic treatment in the 48-h period before initiation of cefotaxime therapy (n = 2); final diagnosis of squamous cell carcinoma instead of infection (n = 1); and protocol violations (n = 2). These consisted of an unintended conversion from intermittent to continuous therapy (n = 1) and a switch to oral therapy after losing venous access (n = 1). Of the five patients who died after inclusion in the study, three died within 24 h of admission. The other two patients, one in group 1 and one in group 2, died on day 3 after admission. All patients died from cardiac causes [congestive heart failure (n = 4) or decompensated cor pulmonale (n = 1)]. None of these cases was thought to be related to a failure of antibiotic treatment.

The characteristics of all patients initially included in the study are shown in Table 1. The outcome data of the 83 patients who were evaluable are shown in Tables 2 and 3. Failure of therapy occurred in six patients, three in group 1 and three in group 2. The mean duration of therapy was 9.3 ± 2.6 days in the continuous group and 9.5 ± 1.5 days in the intermittent group (= 0.64). Both treatment regimens produced identical success rates: 93% (37/40) in the continuous therapy group and 93% (40/43) in the intermittent therapy group (= 0.93). Tolerance to cefotaxime in both treatment regimens was excellent and no specific adverse effects related to the study medication were noted.

Table 1. 
General and bacteriological characteristics of 93 chronic obstructive pulmonary disease patients with respiratory tract infections initially included in the study
 Continuous cefotaxime
(group 1)
Intermittent cefotaxime
(group 2)
  • *

    Mostly cardiac causes of death, none obviously linked to antibiotic treatment failure.

  • In some patients more than one sputum sample was obtained. The total number of patients with pathogens cultured from sputum was 25 in group 1 and 24 in group 2.

  • For analyses only pathogen with highest minimal inhibitory concentration (MIC) was used in case of polymicrobial flora.

  • §

    Many patients were unable to produce sputum.

  • Pseudomonas aeruginosa (= 3), Haemophilus influenzae (= 4), Moraxella catharralis (= 3), combination of H. influenzae and M. catharralis (= 1).

Patients (n)4746 
Age (range; median)65.3 ± 8.4 (34–76; 66.8)68.6 ± 5.3 (52–75; 69.4)0.03
Excluded patients*730.20
Creatinine (µmol l−1) (range; median)90 ± 24 (55–160; 87)86 ± 19 (46–132; 84)0.36
Creatinine clearance (Cockroft–Gould (ml min−1) (range; median)85.5 ± 25.2 (35.2–165.1; 83.1)80.7 ± 23.8 (36.6–138.4; 74.0)0.37
Number of evaluable patients40/47 (85.1%)43/46 (93.5%)0.19
Number of patients with pathogens isolated from sputum before antibiotic therapy25240.54
Patients with clinical signs of infection but no growth of pathogens, or no pathogens isolated15210.30
Number of patients in whom sputum samples could be obtained before antibiotic treatment38/47 (80.9%)31/46 (67.4%)0.14
Number of sputum cultures where pathogens could be isolated38290.06
Successful MIC determinations of isolated pathogens33270.26
Number of patients with pathogens in sputum and successful MIC determination24200.47
Number of patients in whom sputum samples could be obtained after antibiotic treatment§18/47 (38.3%)14/46 (30.4%)0.43
Number of patients with pathogens isolated from sputum after antibiotic therapy740.36
Number of sputum cultures where pathogens could be isolated840.23
Successful MIC determinations of isolated pathogens840.23
Cultures with no growth or no pathogens isolated (n)10100.96
Table 2. 
Pretreatment and post-treatment minimal inhibitory concentration (MIC) values for the most commonly isolated pathogens in 93 hospitalized chronic obstructive pulmonary disease patients with respiratory tract infections
mg l−1 (mean/range)
mg l−1 (mean/range)
  • *

    Including Pseudomonas aeruginosa.

  • Not applicable.

Haemophilus influenzae
 Total340.045 (0.016–0.23) 50.15 (0.016–0.47)
 Continuous20 0.04 (0.016–0.16) 40.07 (0.016–0.016)
 Intermittent14 0.05 (0.016–0.23) 10.47
Streptococcus pneumoniae
 Total14 0.11 (0.016–0.230) 0
 Continuous 8 0.17 (0.016–0.23) 0
 Intermittent 6 0.02 (0.016–0.23) 0
Moraxella catharralis
 Total12 0.29 (0.032–0.75) 40.35 (0.032–0.94)
 Continuous 6 0.28 (0.032–0.75) 30.45 (0.032–0.94)
 Intermittent 6 0.30 (0.032–0.5) 10.06
Other pathogens*
 Total 7  9.0 (6.0–12.0) 3 6.7 (3.0–32.0)
 Continuous 4  9.0 (6.0–12.0) 13.0
 Intermittent 3NA 222.0 (12.0–32.0)
Total isolates67 12 
Table 3. 
Efficacy and pharmacokinetic and pharmacodynamic evaluation of 2 g continuous vs. 3 g t.i.d. intermittent cefotaxime administration in evaluable chronic obstructive pulmonary disease patients
 Group I (continuous*)Group II (intermittent*)95% CIs of differencesP-value
  • *

    Values expressed as numbers of patients and percentages.

  • Vd and t1/2 not calculated during continuous infusion.

  • For analyses average time was calculated only for patients with single cefotaxime concentrations < 5 × MIC > 30% the dosing interval of 8 h.

Clinical evaluation
Evaluable patients (n)40/47 (85.1%)43/46 (93.5)  
Treatment success37/40 (92.5%)40/43 (93%)  
Treatment failure3/40 (7.5%)3/43 (7%)  
Mean duration of treatment (days) (range; median)9.3 ± 2.6 (1–12; 10)9.5 ± 1.5 (4–11; 10)  
Pharmacokinetic evaluation
Evaluable patients (n)4443  
Cmax (mg l−1) (range; median)8.5 ± 3.4 (4.1–20.1; 7.5)45.3 ± 8.6 (28.2–69.2; 43.8)34.0, 39.6 < 0.01
Cmean (mg l−1) (range; median)5.6 ± 1.9 (2.9–10.7; 5.2)8.6 ± 3.3 (4.2–25.2; 8.1)1.8, 4.1 < 0.01
Cmin (mg l−1) (range; median)4.4 ± 2.0 (1.3–11.1; 3.9)0.7 ± 1.8 (0.0–9.6; 0.3)2.8, 4.5 < 0.01
AUC (mg h−1 24 l−1) (range; median)135.1 ± 44.6 (70.3–255.6; 124.6)205.8 ± 79.2 (100.6–603.6; 194.6)43.4, 98.2 < 0.01
Vd (l) (range; median)NA22.6 ± 3.9 (14.5–30.0; 22.8)
t1/2 (h) (range; median)NA1.0 ± 0.3 (0.5–2.3; 1.0)
Cl (l h−1) (range; median)16.3 ± 4.9 (7.8–28.5; 16.1)16.1 ± 4.8 (5.0–29.9; 22.8)− 2.3, 1.90.86
Pharmacodynamic evaluation
Evaluable patients (n)4443  
Single cefotaxime concentrations < 2.5 mg l−15/44 (11%)42/43 (98%)  < 0.01
Single cefotaxime concentrations < 1.0 mg l−10/44 (0%)18/43 (42%)  < 0.01
Single cefotaxime concentrations < 0.5 mg l−10/44 (0%)12/43 (28%)  < 0.01
Evaluable patients (n)2420  
Cefotaxime concentrations < MICactual0/24 (0%)8/20 (40%)  < 0.01
Cefotaxime concentrations < 5 × MICactual0/24 (0%)13/20 (65%)  < 0.01
Cefotaxime concentrations0/24 (0%) dosing intervals8/20 (40%)  < 0.01
Cefotaxime concentrations0/24 (0%) < 5 × MIC > 30% of the dosing intervals9/20 (45%)  < 0.01
Time (h) cefotaxime concentrations < 5 × MIC > 30% dosing interval0.0 ± 0.03.7 ± 1.0  < 0.01

We were able to collect reliable pretreatment sputum samples in 69 patients (83%). In 49 of these (71%) pathogenic microorganisms were cultured from 67 samples (in some patients more than one sputum sample was obtained). The three most commonly isolated pathogens were H. influenzae (51%), S. pneumoniae (21%) and M. catharralis (18%). Only two Pseudomonas aeruginosa isolates were found. Because cefotaxime is not effective for P. aeruginosa infections, these patients were considered to be microbiologically non-evaluable and were subsequently treated with other antibiotics (but they were included in the pharmacokinetic evaluation).

MIC values were determined in 60/67 (90%) of the positive cultures. The results for the three most frequently cultured pathogens are shown in Table 2. There was no difference in mean pretreatment MIC between the continuous and intermittent groups. The mean pretreatment MIC value for all susceptible (non-P. aeruginosa) organisms was 0.10 ± 0.16 mg l−1 (range 0.016–1.0). MIC was < 0.5 mg l−1 in 59/60 (98%) of cases.

We were able to collect post-treatment sputum samples in only 32 of the 83 patients. This was mainly due to the absence of sputum production following antibiotic treatment. No pathogens were isolated from 20/32 (63%) of the post-treatment specimens. Of the positive samples, H. influenzae was found in five (42%) and M. catharralis in four (33%). No S. pneumoniae was isolated from any of the post-treatment specimens. Values were similar in groups 1 and 2.

Pseudomonas aeruginosa was cultured following cefotaxime administration in three cases, one in the continuous and two in the intermittent group. Mean post-treatment MIC values for all non-P. aeruginosa pathogens were 0.16 ± 0.17 mg l−1 (range 0.016–0.94). MIC values were < 0.5 mg l−1 in all pathogens cultured after antibiotic treatment except for two M. catharralis strains (one in each group) with MIC values of 0.94 mg l−1 and the three P. aeruginosa isolates. There were no differences in susceptibility between the continuous and intermittent groups either before or after cefotaxime treatment (Table 2).

Pharmacokinetic evaluation of serum cefotaxime concentrations was performed in 87 of the 93 patients included initially in the study. Serum concentrations were not measured in six patients, because three patients in group 1 died before blood samples could be drawn and technical errors, such as lost blood samples, occurred in three patients in group 2. A total of 926 samples in the remaining 87 patients were analysed. The results are shown in Table 3.

In patients treated with continuous cefotaxime infusion the mean (± SD) total serum drug concentration (Cmean) was 5.6 ± 1.9 mg l−1 (range 2.9–10.7) and was significantly lower than the value in the intermittent group [8.6 ± 3.3 mg l−1 (range 4.2–25.2), < 0.01]. This is probably explained by the higher dose given to the intermittent group.

In group 2 the mean peak antibiotic concentration (Cmax, measured at 0.5 h) was 45.3 ± 8.6 mg l−1 and the mean trough concentration (Cmin, measured immediately before administration of the next dose) was 0.7 ± 0.8 mg l−1. Although Cmax, Cmean and AUC were significantly higher in group 2 (P-values < 0.01), Cmin was significantly lower (< 0.01). Mean total drug clearance (Cl) did not differ between the two groups (= 0.86).

Drug concentrations exceeded MIC in all samples in group 1 at all times (100%). Serum concentrations exceeding 5 × MIC for the causative organisms (indicating optimum bacterial killing ability [6–8]) were achieved at all times in all evaluable patients (n = 24) with continuous cefotaxime infusion (100%). Eight out 20 (40%) serum concentrations < 5 × MIC were observed in group 2 patients, and the mean time at which drug concentrations exceeded 1 mg l−1 was 6.6 ± 1.5 h (range 3.0–8.0) per dosing interval of 8 h, which is 19.8 per 24 h. The period during which antibiotic concentrations were > 0.5 mg l−1 was 6.7 ± 1.5 h (range 3.5–8.0). Serum drug concentrations < 1 mg l−1 for > 30% of the treatment time occurred in no patients in group 1 and in 18/43 (42%) patients in group 2 (< 0.01). Concentrations < 0.5 mg l−1 for > 30% of treatment time were observed in no patients in group 1 and in 12/43 (28%) patients in group 2 (< 0.01) (Table 3).

Samples with antibiotic concentrations < MICactual occurred in no patients in group 1 vs. 8/20 (40%) in group 2 (< 0.01). In addition, samples with MICs below the optimum killing concentration of 5 × MICactual were found in no patients in group 1 compared with 13/20 (65%) in group 2 (< 0.01). Serum drug concentrations < MICactual for more than 30% of the overall treatment time occurred in no patients in group 1 compared with 8/20 (40%) patients in group 2 (< 0.01). Drug concentrations < 5 × MICactual occurred in no patients in group 1 compared with 9/20 (45%) patients in group 2 (< 0.01). The average time during which antibiotic concentrations were < 5 × MICactual in these nine patients was 3.7 ± 1.0 per 8 h (46% of the dosing interval), which is 11.1 per 24 h.


Our results suggest that continuous administration of the β-lactam antibiotic cefotaxime produces similar clinical efficacy to intermittent administration in patients with chronic pulmonary disease and community-acquired airway infections. Furthermore, from a pharmacodynamic perspective, continuous administration appeared to be at least as effective. Antibiotic concentrations exceeded MIC at all times in all patients when cefotaxime was administered continuously. This was expected, based on knowledge of the pharmacokinetics/pharmacodynamics of continuous antibiotics administration. However, in our present study concentrations exceeding MIC were achieved despite the total daily dose of cefotaxime being 1 g lower than in patients receiving intermittent doses of cefotaxime. We did not observe differences in cefotaxime clearance comparing continuous infusion and intermittent administration.

Although Cmax, Cmean and AUC were lower in the continuous group, a more optimal pharmacodynamic profile was observed, particularly near the end of the dosing intervals. Concentrations below MIC were observed much more frequently in patients receiving cefotaxime intermittently. Antibiotic concentrations > MIC for at least 70% of treatment time are thought to be required for microbiological efficacy and to avoid induction of antibiotic resistance [6–8]. This implies that intermittent dosage is associated with at least a theoretical risk of antibiotic resistance in a substantial proportion of our patients. In the present study this did not lead to clinical problems and no significant drug resistance had developed in either group during this brief course of antibiotic treatment. Only patients with community-acquired infections were included in our study. These are caused by microorganisms that (at least in the Netherlands) are usually highly susceptible to most first-line antibiotics. Antimicrobial efficacy may be radically different in patients who develop infections caused by intrinsically more resistant bacteria, as is often the case in nosocomial infections. The same may hold true for environments where the general level of antibiotic resistance is higher than in the Netherlands, such as in the USA and many European countries such as France, the UK and Spain [9–15]. In these settings the differences in pharmacological profiles observed in our study are far more likely to lead to differences in clinical outcome.

The major determinant of bacterial killing is thought to be the period during which antibiotic concentrations are > 5 × MIC for 70% of the treatment time. Mouton et al.[8] determined the extent of in vitro killing for P. aeruginosa strains in a model reproducing pharmacokinetic profiles. These authors concluded that in settings where antibiotics are the only agents that actively and efficiently combat infections, as is the case in patients with severe neutropenia, even sustained antibiotic concentrations at or slightly above MIC are inadequate to combat infections effectively. Only strains exposed to sustained antibiotic concentrations of at least 4 × MIC showed sustained growth inhibition and bacterial death. This indicates that, at least in immunocompromised patients and for this particular microorganism, only concentrations ≥ 4 × MIC are effective. Although these parameters are likely to vary depending on the causative microorganism, the underlying diagnosis and clinical condition, and various other factors, in clinical practice the antibiotic target concentration is often set at 5 × MIC for the target organism. In the present study, this target was achieved in 100% of patients in group 1 compared with 55% in group 2. Thus, although mean concentrations ≥ 5 × MIC for > 70% of the time were achieved in both groups, in many individual patients in group 2 concentrations below this cut-off value were present for prolonged periods of time. Thus, the theoretical risk of treatment failure was significantly greater in group 2, although in this study of patients infected with bacteria relatively susceptible to antimicrobial therapy, no differences in clinical outcome were seen.

Studies on the continuous infusion of ceftazidime have been performed in other patient groups. Lipman and coworkers demonstrated that in 18 critically ill patients who had received ceftazidime 6 g day−1 by continuous infusion or bolus doses (2 g 8 hourly), each with a loading dose of 12 mg kg−1 ceftazidime, plasma ceftazidime concentrations in all continuous infusion patients remained above 40 mg l−1 for 40 h vs. 20–30% of bolus patients. They suggested that more studies on continuous infusion were warranted [16].

Our results are in accordance with findings by Lubasch and coworkers, who treated patients with acute exacerbation of severe chronic bronchitis with either ceftazidime 3 × 2 g intermittently or 2 × 2 g infused over two 7-h periods. They reported that the clinical and bacteriological efficacy of the two treatment modes was comparable, with cure rates of 90% in both groups [17]. In contrast to our findings, microbiological susceptibility after antibiotic treatment was not determined. However, the authors performed pharmacodynamic measurements in a small subgroup of 21 patients and observed a more favourable profile in patients receiving a continuous infusion of ceftazidime [17]. To our knowledge, our study and that of Lubasch et al. are the only two investigations in COPD patients that have looked at pharmacokinetic and pharmacodynamic aspects of continuous infusion of β-lactam antibiotics in a clinical setting.

In selecting dosing regimens, cost effectiveness and workload of the physician and nursing staff should be considered. In the present study drug costs were slightly lower for continuous therapy, because the required dosage was 2 g day−1 compared with 3 g in patients treated with intermittent doses. It is often assumed that continuous infusion using a syringe pump will be significantly more expensive than intermittent administration. However, this assumption may not always be correct.

Previously, we have performed a cost-effectiveness study comparing continuous and intermittent use of five different antibiotics for pneumonia or intra-abdominal infections in the intensive care unit and general ward [18]. Antibiotic treatment was associated with significant hidden costs, mostly determined by the mode of administration. Perhaps unexpectedly, when all cost drivers were taken into account, the cheapest method of administering intravenous antibiotics was through use of a continuous syringe pump [18]. Direct intravenous administration requires more time expenditure through the need for relatively highly trained (and paid) nurses or physicians. Moreover, the procedure often requires slow infusion of the antibiotic, which inevitably takes longer [18]. An additional factor is that increased staff workload is associated with adverse effects on outcome, increased rates of complications, increased length of stay and even with higher mortality [19–21]. Shortening the amount of time spent on administering intravenous antibiotics would allow healthcare workers to devote more time to other aspects of patient care, which in itself might lead to improvements in outcome.

A theoretical objection against continuous infusion might be that drug equilibration can take longer than intermittent (bolus) administration. In severe infections this delay of onset of antibacterial activity could be important. However, this problem can be easily circumvented by administration of an initial loading dose prior to continuous infusion, as was done in the present study, thus ensuring the rapid onset of antibacterial activity [22]. Another disadvantage of using continuous infusion might be that when unexpectedly more resistant pathogens are encountered, all antibiotic concentrations could be lower. In intermittent infusion, during at least some period of the dosing interval concentrations could be above the MIC. This problem may be circumvented by increasing knowledge on MIC values and common pathogens most frequently involved in community-acquired and nosocomial infections on the hospital ward. Initiation of (empirical) antibiotic treatment based on these data as well as individual patient characteristics will increase the likelihood of optimizing therapy.

One limitation of our study was that serum concentrations of desacetylcefotaxime, a metabolite of cefotaxime with bactericidal activity similar to that of the parent compound, were not determined. Following administration of cefotaxime, the half-life of this desacetyl metabolite in healthy subjects is 1.6 h [23]. However, Vallee et al. have reported that the contribution of desacetylcefotaxime to the serum bactericidal activity of cefotaxime is negligible [24]. Therefore, not considering the effects of desacetylcefotaxime would at most slightly underestimate the bactericidal activity of cefotaxime. In addition, there are likely to be no important differences in serum desacetylcefotaxime concentrations between the two treatment groups, because the rate of cefotaxime elimination was comparable.

Another potential limitation of the present study was the duration of antibiotic treatment, which may have been too short to induce bacterial resistance. Clearly, more prolonged observations of the effects of different administration strategies on microbiological susceptibility are needed. However, the finding that continuous administration mode was associated with 5 × MIC concentrations for 100% of the time (vs. 70% in the intermittent group) strongly suggests that such long-term observation would either show no difference, or (more likely) would demonstrate superiority of the continuous mode over the intermittent mode of administration due to emerging resistance during the latter regimen. Our results suggest that it is extremely unlikely that continuous administration would be associated with any additional risks of decreased efficacy or the emergence of resistance, and is highly likely to decrease these risks.

Therefore, based on the results of our present study and the arguments outlined above, we question whether the current routine approach, which is to administer intravenous antibiotics intermittently, is the most efficacious and cost-effective therapy for hospitalized patients, at least in the case of β-lactam antibiotics. Whereas the clinical efficacy and pharmacodynamics of cefotaxime in our study were similar for continuous administration, drug costs were slightly lower and hidden costs are likely to be substantially less [18]. Our view is that the same is likely to hold true for other antibiotics whose bacterial killing potential depends on the time above MIC (in contrast to antibiotics that have a significant postantibiotic effect, such as aminoglycosides).

In conclusion, the present work demonstrates that continuous administration of 2 g cefotaxime preceded by a 1-g loading dose is equally effective compared with the current standard practice of intermittent bolus administration, and has potentially important pharmacological advantages. In addition, this mode of administration is likely to be more cost-effective and will help decrease staff workload. Therefore, we recommend the continuous administration of cefotaxime using a syringe pump as the preferred method of giving this drug and probably all antibiotics whose efficacy depends on time above MIC. Long-term effects on antibiotic resistance remain to be determined, but outcomes are likely to be favourable.

Competing interests: None declared.

We thank A. de Jong and W. de Waard for their assistance with the serum concentration assays, E. L. Swart PharmD and I. Van Geijlswijk PharmD for their advice on interpretation of the pharmacokinetic data, and the Department of Clinical Chemistry for blood sampling. The manufacturer of cefotaxime, Hoechst Marion Roussel, provided a restricted research grant for analysing serum cefotaxime concentrations and for assessing MIC values.