Volume 66, Issue 6 p. 866-874
Free Access

The influence of labour on the pharmacokinetics of intravenously administered amoxicillin in pregnant women

Anouk E. Muller

Anouk E. Muller

Erasmus MC, University Medical Centre Rotterdam, Department of Medical Microbiology and Infectious Diseases, 's-Gravendijkwal 320, 3015 CE Rotterdam, the Netherlands,

Search for more papers by this author
P. Joep Dörr

P. Joep Dörr

Medical Centre Haaglanden (MCH), Department of Obstetrics and Gynecology, Lijnbaan 32, 2512 VA the Hague, the Netherlands,

Search for more papers by this author
Johan W. Mouton

Johan W. Mouton

Canisius Wilhelmina Hospital, Department of Clinical Microbiology and Infectious Diseases, Weg door Jonkerbos 100, 6532 SZ Nijmegen, the Netherlands,

Search for more papers by this author
Joost De Jongh

Joost De Jongh

LAP&P Consultants BV, Archimedesweg 31, 2333 CM Leiden, the Netherlands,

Leiden-Amsterdam Center for Drug Research, Division of Pharmacology, Leiden University, P.O. Box 9502, 2300 RA Leiden, the Netherlands,

Search for more papers by this author
Paul M. Oostvogel

Paul M. Oostvogel

Medical Centre Haaglanden (MCH), Department of Clinical Microbiology, Lijnbaan 32, 2512 VA the Hague, the Netherlands and

Search for more papers by this author
Eric A. P. Steegers

Eric A. P. Steegers

Erasmus MC, University Medical Centre Rotterdam, Department of Obstetrics and Gynecology, Division of Obstetrics and Prenatal Medicine, 's-Gravendijkwal 320, 3015 CE Rotterdam, the Netherlands

Search for more papers by this author
Rob A. Voskuyl

Rob A. Voskuyl

Leiden-Amsterdam Center for Drug Research, Division of Pharmacology, Leiden University, P.O. Box 9502, 2300 RA Leiden, the Netherlands,

Search for more papers by this author
Meindert Danhof

Meindert Danhof

Leiden-Amsterdam Center for Drug Research, Division of Pharmacology, Leiden University, P.O. Box 9502, 2300 RA Leiden, the Netherlands,

Search for more papers by this author
First published: 20 November 2008
Citations: 22
Anouk E. Muller, MD, Erasmus MC, University Medical Centre Rotterdam, Department of Medical Microbiology and Infectious Diseases, Rotterdam, the Netherlands
Tel.: +31-10-703-3510
Fax: +31-10-703-3875
E-mail: [email protected]

Abstract

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Examples exist that pharmacokinetics of drugs in pregnant women can differ from that in non-pregnant individuals.

• In pregnant women before the onset of labour, the pharmacokinetics of amoxicillin is similar to that in non-pregnant individuals, but for women during labour this is unknown.

WHAT THIS STUDY ADDS

• Labour influences the pharmacokinetics of amoxicillin.

• During labour and even more in the immediate postpartum period, the peripheral volume of distribution was decreased compared with pregnant women before the onset of labour.

• The volume of distribution increases with an increasing amount of oedema.

AIMS

Many physiological changes take place during pregnancy and labour. These might change the pharmacokinetics of amoxicillin, necessitating adjustment of the dose for prevention of neonatal infections. We investigated the influence of labour on the pharmacokinetics of amoxicillin.

METHODS

Pregnant women before and during labour were recruited and treated with amoxicillin intravenously. A postpartum dose was offered. Blood samples were obtained and amoxicillin concentrations were determined using high-pressure liquid chromatography. The pharmacokinetics were characterized by nonlinear mixed-effects modelling using NONMEM.

RESULTS

The pharmacokinetics of amoxicillin in 34 patients was best described by a three-compartment model. Moderate interindividual variability was identified in CL, central and peripheral volumes of distribution. The volume of distribution (V) increased with an increasing amount of oedema. Labour influenced the parameter estimate of peripheral volume of distribution (V2). V2 was decreased during labour, and even more in the immediate postpartum period. For all patients the population estimates (mean ± SE) for CL and V were 21.1 ± 4.1 l h−1 (CL), 8.7 ± 6.6 l (V1), 11.8 ± 7.7 l (V2) and 20.5 ± 15.4 l (V3) respectively.

CONCLUSIONS

The peripheral distribution volume of amoxicillin in pregnant women during labour and immediately postpartum is decreased. However, these changes are not clinically relevant and do not warrant deviations from the recommended dosing regimen for amoxicillin during labour in healthy pregnant patients.

Introduction

The use of antibiotics during labour for prevention of neonatal infections has increased substantially in the last decades. Current guidelines from the Centers of Disease Control and Prevention (CDC) to prevent neonatal group B streptococcal (GBS) disease recommend the use of antibiotics during labour in all pregnant women carrying GBS [1]. The prevalence of GBS carriage varies according to the geographical region from 10% to 35% of all pregnant women, resulting in antibiotic use for this purpose in up to one of every three women [2–5]. Amoxicillin, a penicillin derivative, is one of the antibiotics used in the prevention of GBS-disease. Amoxicillin is widely used in Europe, while ampicillin is more commonly used in the US. Amoxicillin is a bactericidal antibiotic and is primarily excreted unchanged in the urine both by glomerular filtration and by tubular secretion in the kidneys.

Adequate dosing of antibiotics to the mother is essential to prevent the neonate from GBS-disease. Antibiotics reach the foetus after transplacental passage. An adequate concentration–time profile in maternal serum is therefore a prerequisite for an adequate concentration–time profile in the foetus. Current dosing regimens are similar for non-pregnant individuals, pregnant women before the onset of labour and for pregnant women during labour. Many physiological changes take place during pregnancy, which may modify the concentration–time profile of specific drugs, such as an increase in plasma volume, presence of the foetus and changes in elimination rate or metabolism [6]. In a previous study we found no differences in the pharmacokinetics (PK) of amoxicillin between pregnant women with preterm premature rupture of the membranes (PPROM) and values reported in the literature for non-pregnant individuals [7]. During labour additional physiological changes occur of which many are expected to change the PK behaviour of drugs [8]. Uterine contractions, mechanical compression of blood vessels by the gravid uterus as well as hyperventilation, might all have their separate circulatory influences affecting blood flow through the body and especially to the eliminating organs. The consequential change in concentration–time profile of the antibiotic for both mother and neonate are unpredictable.

Despite the regular use of amoxicillin during labour, the influence of labour on the PK has not been adequately studied. PK studies during labour face considerable ethical and practical difficulties, limiting the collection of blood samples. The number of blood samples collected in women during labour will therefore be smaller compared with pregnant women before the onset of labour. Unbalanced study groups harbour statistical problems. Non-Linear Mixed Effects Modelling (NONMEM) [9] allows weighted analysis of observations from unbalanced study designs and incorporation of patients with small or incomplete datasets [10, 11]. Studying the whole population as one group, the influence of specific circumstances, such as the presence of labour, on the individual PK parameters can be assessed using covariate analysis [12, 13]. A more detailed background of population modelling can be found elsewhere [14, 15]. The objective of this study is to investigate the influences of labour on the PK of intravenously administered amoxicillin.

Methods

Patients

In the period between February 7, 2005 and February 28, 2007, all women with gestational age of minimally 26 weeks who needed antibiotic treatment with amoxicillin were eligible for the study. From a subset of patients part of the data was used in a previous study [7]. These patients were all diagnosed with preterm premature rupture of the membranes and therefore before the onset of labour (in total 416 samples). Following the local guidelines, all women with proven or unknown Streptococcus agalactiae carriage were treated with antibiotics when pregnancy was complicated by one of the following factors: preterm premature rupture of the membranes, rupture of the membranes for >18 h, prematurity, fever (>37.8°C), bacteriuria in current pregnancy and a previous child with invasive GBS-disease. The choice of the antibiotic for this study was dictated by the local guidelines, which recommend amoxicillin as first choice. Women were admitted to the hospital and monitored for foetal condition, onset of labour and signs of infection. When intra-amniotic infection was suspected, delivery was induced. The study was approved by the Medical Ethics Committee of the Medical Centre Haaglanden, the Hague, the Netherlands. Written informed consent was obtained from all patients. Women were excluded from the study when i) they had been treated with oral or intramuscular antibiotics within 2 days before starting therapy, ii) were unwilling to comply with the requirements of the study, iii) were known to be allergic to amoxicillin or other penicillins, or iv) were receiving co-medication that exhibits interaction with amoxicillin. All patients were at least 18 year of age.

Both pregnant patients before the onset of labour and patients in labour were included in the study. Patients included before the onset of labour or during labour, who agreed to receive an additional dose of amoxicillin after their delivery for study purposes only were kept in the study until maximally 28 h after delivery. Being in labour was defined as the presence of uterine contractions resulting in progressive cervical dilatation. The vaginal examinations were performed by the physician responsible for the delivery. When labour started during the study period, the time of the onset of labour was recorded.

All patients received a standard work-up that included a medical history, biochemical and haematological examination. Furthermore blood pressure, pulse, oral temperature, and body weight were recorded at the onset of the study. Temperature and pulse were recorded to register the possibility of intra-amniotic infection, whereas the blood pressure and the amount of oedema were recorded to account for differences in distribution volumes. The amount of oedema was scored semiquantitatively from 0 (no oedema) to 3 (above the knee).

Drug administration and blood sampling

Before the administration of the amoxicillin, an intravenous catheter was placed in each arm. Amoxicillin was administered according to local guidelines in the hospital. Treatment started with an intravenous infusion of 2 g amoxicillin (50 mg ml−1) administered over 30 min, followed by an infusion of 1 g amoxicillin over 15 min every 4 h until delivery. For the prevention of GBS, antibiotics were administered until delivery. For the purpose of this study only, a single additional dose of 1 g amoxicillin was administered after delivery 4 h after the last dose before child birth. Blood samples of 2 ml were collected from the second catheter in the contralateral arm at timed intervals beginning at 1 min after the start of the infusion and, at 7 and 15 min (1 g infusion) or 15 and 30 min (2 g infusion). After the infusion, sampling was scheduled at 3, 6, 10, 16 and 36 min, and afterwards every 30 min until the next antibiotic dosage. Blood samples were collected when possible, taking into consideration the physical and emotional inconvenience to the woman. The exact sampling times were recorded.

Blood samples were placed immediately on ice, allowed to clot and processed within 1 h after collection. The samples were centrifuged at 1200 g for approximately 10 min. The supernatants were transferred into plastic storage tubes and frozen at −70°C until analysis.

Amoxicillin HPLC assay

Amoxicillin concentrations were determined by an isocratic high-pressure liquid chromatography (HPLC) (Shimadzu, Den Bosch, The Netherlands (NL)) method, using an ODS Gemini column (Bester, Amstelveen, NL) with 0.066 m KH2PO4 solution containing 10% methanol as a mobile phase. A perchloric acid solution of 0.1 ml was added to the sample in an equal volume and after vortexing, added to 0.56 ml 0.028 m citric acid containing cefadroxil (Sigma, Zwijndrecht, NL) as an internal standard. The assay was linear over the concentration range measured. Controls were included in every run. The lower limit of detection was 0.2 mg l−1 and the between run CV < 4%.

Pharmacokinetic analysis

Pharmacokinetic parameters were estimated by means of Non-Linear Mixed Effect Modelling (NONMEM). The model was implemented in the NONMEM ADVAN5 subroutine and the analysis was performed using the FOCE with INTERACTION method. All fitting procedures were performed with the use of the Compaq Visual FORTRAN standard edition 6.6 (Compaq Computer Cooperation, Euston, Texas, USA) and NONMEM® software package (version VI, release 1.2, ICON Development Solutions, Ellicott City, Maryland, USA).

To determine the basic structural pharmacokinetic parameters various two- and three-compartment models were tested. Model selection and identification of variability was based on evaluation of the mean objective function value (OFV), pharmacokinetic parameter point estimates, and their respective confidence intervals, and goodness-of-fit plots. For differences between two structural models, the OFV with a pre-specified level of significance of P < 0.001 was used (corresponding to a difference in OFV of 10 points). NONMEM minimizes an objective function in performing nonlinear regression analysis. To detect systematic deviations in the model fits, the goodness-of-fit plots were visually inspected. The data of individual observations vs individual or population predictions should be randomly distributed around the line of identity. The weighted residuals vs time or population predictions should be randomly distributed around zero. Population values were estimated for the parameters clearance (CL), the volumes of distribution (V) and the intercompartmental clearances (Q).

Individual estimates for pharmacokinetic parameters were assumed to follow a log-normal distribution. Therefore an exponential distribution model was used to account for interindividual variability. Possible correlation between interindividual variability coefficients on parameters was estimated and if present accounted for in the stochastic model (NONMEM Omega block option).

Selection of an appropriate residual error model was based on the likelihood ratio test and inspection of the goodness-of-fit plots. A proportional error model, additive error model and a combined proportional-additive error model were tested to describe the residual variability between the observed concentrations and those predicted by the model. The residual error term contains all the error terms which cannot be explained and refers to, for example, measurement and experimental error and structural model misspecification.

To refine the model covariate analysis was also performed. The estimated pharmacokinetic parameters, on which a random effect has been identified, were plotted against the covariates bodyweight, body mass index, gestational age, blood pressure, pulse, oral temperature, and the amount of oedema, single or twin pregnancy, creatinine, the renal function calculated with the Cockcroft-Gault (CG) and the modified Modification of Diet in Renal Disease (MDRD) equation to determine whether this influenced the pharmacokinetics [16, 17]. Covariate analysis was performed by forward addition of each candidate covariate into the model structure until no further improvement of goodness of fit was observed. A significance level of 0.05 was selected (corresponding to difference in OFV 3.84 points). A further criterion for acceptance of covariate effects was that the estimated 95% confidence interval of the covariate effect did not overlap with zero. Contribution of each covariate to the final model was the confirmed by backward elimination of each covariate from the model to account for possible interaction between covariates. The residual intra- and interindividual variability were visually evaluated. The volume of distribution at steady state (Vss) and terminal half-life (t1/2) were calculated following standard procedures [18].

Finally, the effect of the presence of labour was investigated on structural PK parameters. The state of being pregnant but before the onset of labour, being in labour and being in the immediate postpartum period were implemented in the model as covariate in the entire group of patients. A significance level of 0.05 was selected (corresponding to difference in OFV of 3.84 points). A further criterion for acceptance of the influence of labour was that the estimated 95% confidence interval of its effect did not overlap with zero. Contribution of an effect of labour to the final model was the confirmed by backward deletion of both the effect of labour and the continuous covariates from the model to account for possible interaction between covariates and the effect of labour.

The accuracy of the final population model for the entire population was established using a bootstrap method in NONMEM, consisting of repeated random sampling with replacement from the original data. This resampling was repeated 100 times. The estimated parameters from the bootstrap analysis were compared to the estimates from the original data.

Results

In total, 34 patients were included. From eight patients blood samples were taken both before the onset of labour and during labour. From 17 patients blood samples were taken only before the onset of labour and from nine patients only during labour. Eight patients agreed to receive a postpartum dose of amoxicillin as well. The postpartum doses of amoxicillin were administered between 1.5–3.8 h after child birth. The study population consisted of 31 singleton and three twin pregnancies. All 17 patients participating in the study during labour, delivered vaginally. The gestational age at the time of the study ranged from 30.0 to 40.6 weeks. The characteristics of the study patients are presented in Table 1.

Table 1.
Baseline patients' demographic data of 34 pregnant women
Data Mean SD Range Number of patients
Maternal age (years) 29.0 5.5 20–38 34
Gestational age (weeks) 35.9 2.3 30–41 34
Body mass index (kg m−2) 28.8 5.0 18–38 33
Weight (kg) 79.0 13.6 53–107 33
Leucocytes (×109 l−1) 12.5 5.2 6–28 34
Creatinine (µmol l−1) 45.1 8.2 30–74 34
Nulliparity 15
Oedema (none/around the ankle/up to the knee) 21/10/2
  • SD, Standard deviation.

Peak concentrations were comparable in patients during labour and in patients before the onset of labour. Peak concentrations after the 2 g infusion were 97.4 ± 20.7 mg l−1 in patients before the onset of labour and 92.3 ± 16.6 mg l−1 in patients during labour (mean ± SD). Peak concentrations after the 1 g infusion were 71.8 ± 14.8 mg l−1, 62.8 ± 7.9 mg l−1, and 65.7 ± 15.5 mg l−1, respectively, in patients before labour, during labour and postpartum. The terminal half-lives for the three stages of labour were not significantly different (1.1 ± 0.3 h during labour, 1.2 ± 0.2 h before labour and 1.2 ± 0.2 h postpartum). Vss was 40.4 l.

A total of 898 blood samples were collected in this study. Of these samples 550 were taken before the onset of labour, 187 during labour and 161 in the immediate postpartum period. For patients included before the onset of labour between seven and 34 samples were obtained per patient, during labour between five and 24 samples and in the postpartum period between 12 and 25 samples. A three compartment open model best described the data. The residual error was found to be proportional to the blood concentrations. Interindividual variability was mainly observed in CL (CV 19.8%), V1 (CV 23.1%) and V2 (CV 31.6%). Correlations between the random parameters for interindividual variability were found and implemented in the model. For the selected continuous covariates, there was a significant effect of oedema on the total volume of distribution. The volume of distribution increased with an increasing amount of oedema ( Figure 1b). Furthermore, the effect of renal function on CL was found inconsequential. Using the serum creatinine concentration and the estimated creatinin clearance calculated with the CG-formula, no influence on the CL of amoxicillin was seen. However, renal clearance was found to have a small, but significant effect on CL when calculated using the modified MDRD formula. CL was inversely correlated with an increased MDRD (Figure 1c). In Figure 2 the observed concentrations are plotted vs the model-based population predicted concentrations, illustrating the unbiased model-fit.

Details are in the caption following the image


Plots of values for V2 for the three stages of labour (A), values of V1vs the amount of oedema (B) and values of MDRD vs CL (C)

Details are in the caption following the image


Scatter plot of the population predicted vs observed concentrations of amoxicillin for 34 patients. The figure shows the individual data points for the entire population and the line of identity (x = y) with linear scale (A) and logarithmic scale (B)

Labour status in patients was found to have a small but significant effect on peripheral distribution volume (V2). V2 was larger in women before the onset of labour compared with women during labour and in the postpartum period. Compared with women before the onset of labour, V2 was decreased with 13.7% during labour and 29.5% in the immediate postpartum period. Figure 1a shows the different values for V2 for the three stages of labour. Figure 3a shows the observed concentrations of patients before and after the onset of labour for the first 4 h after the 2 g infusion, whereas Figure 3b shows the observed concentration for all three stages of labour after a 1 g dose. The estimates of the pharmacokinetic parameters of the final model and the relative standard errors derived from the bootstrap analysis are presented in Table 2.

Details are in the caption following the image


(A) shows the observed amoxicillin concentrations after a 2 g dose and (B) after a 1 g dose. Time of infusion is indicated by black bars. The open squares represent all data points of the patients before the onset of labour; the filled dots data points of patients during labour and data points of patients in the postpartum period are indicated by the open triangles. Because there was variation in the start time of the second infusion, in (B) the data were adapted assuming that the 1 g infusions started at t = 0 for all patients

Table 2.
Population model parameter values with standard error of 34 women presented with SE% or CV% and 95% confidential interval as derived from the bootstrap analysis
Structural model parameters Estimates of all patients
Mean (model) SE% (bootstrap) 95% CI (bootstrap)
CL (l h−1) 21.1 4.1 19.6, 23.0
V 1 (l) 8.7 6.6 7.5, 9.8
V 2 (l) 11.8 7.7 9.9, 13.4
V 3 (l) 20.5 15.4 14.3, 26.7
Q1 (l h−1) 21.9 16.9 15.0, 29.8
Q2 (l h−1) 1.5 41.3 0.28, 2.69
Variance model parameters Mean (×10−2) (model) CV% (bootstrap) 95% CI (×10−2) (bootstrap)
IIV in CL 4.2 19.8 1.6, 6.1
IIV in V1 5.1 23.1 −0.16, 10.6
IIV in V2 9.7 31.6 1.1, 17.9
Residual variability 4.6 21.5 3.3, 5.7
  • SE, standard error of the estimate; SE%, relative SE (%); 95% CI, 95% confidence interval; CL, clearance; V1, volume of distribution of the central compartment; V2, volume of distribution of the first peripheral compartment; V3, volume of distribution of the second peripheral compartment; Q1, intercompartmental clearance between V1 and V2; Q2, intercompartmental clearance between V1 and V3; IIV, Interindividual variability; CV%, relative coefficient of variation (%).

The bootstrap validation of the model of the entire population was performed with 100 runs. The mean parameter estimates of the runs obtained from the bootstrap analysis did not differ significantly from the predicted values from the NONMEM PK analysis (data not shown). The bootstrap validation was successful for 95 runs. The standard errors obtained from the bootstrap analysis were also comparable with those estimated by the model.

Discussion

The PK of intravenously administered amoxicillin of our entire study population was best described by a three-compartment open model. Volume of distribution increased with an increasing amount of oedema. An effect of labour was seen on the peripheral volume of distribution (V2). When compared with pregnant women before the onset of labour, V2 was decreased during labour and even more decreased in the immediate postpartum period. For our patients dose adjustments were not indicated, but it cannot be excluded from this study that the current dosing schedule is adequate for patients with pregnancy-related disorders.

Estimation of glomerular filtration rate (GFR) in pregnant women is difficult. Serum creatinine concentrations are often used to estimate GFR, especially in conjunction with either the CG or MDRD formulae [16, 17]. Both the CG and MDRD formulae are based on cohort studies of patients with mild to moderate renal insufficiency, and none of the subjects was pregnant [16, 17]. Unfortunately, estimation of GFR in the pregnant population using either the CG or the MDRD formula has not been validated. There have been several reports demonstrating that the MDRD formula tends to underestimate GFR in subjects with normal or near-normal renal function [19, 20], whereas the CG formula is weight based, and weight gain in pregnancy will obviously exaggerate estimated GFR. Alper et al. investigated the GFR in pre-eclamptic patients [21]. Neither the CG or the MDRD formulae were very accurate in predicting GFR in this group of patients, although the MDRD formula performed modestly better [21]. In our analysis, both serum creatinine values and estimated GFR using the CG formula did not have a significant effect on CL, but when GFR was estimated with the MDRD formula the CL decreased with an increasing GFR. This is unlikely, and therefore illustrates the importance of the use of validated formulae in special patient groups. Pharmacokinetic parameter estimates did not change after the implementation of GFR calculated with the MDRD formula. Based on the absence of an effect of serum creatinine values and the GFR estimated using the CG-formula, we concluded that CL was not influenced by renal function.

Other studies on the influence of labour on the PK of amoxicillin are not available. One study has investigated the PK of oral amoxicillin after a single dose both in the second and third trimester of pregnancy as well as postpartum [22]. It was found that during pregnancy the renal clearance was increased compared with the postpartum period [22]. However, in the study of Andrew et al. [22] patients were included 3 months after delivery, while our patients were measured within the first 27 h after delivery. This probably explains the difference between the two studies. The PK of ampicillin, an antibiotic closely related to amoxicillin, has been studied both during pregnancy and during labour [23–28]. Differences in PK of ampicillin between pregnant and non-pregnant individuals have been described, such as a shorter half-life and higher plasma clearance during pregnancy [25, 26, 28]. Furthermore, an effect of labour on the terminal half-life of ampicillin has been described [24]. The terminal half-life in patients in labour was increased, compared with pregnant women before the onset of labour from 39.2 ± 4.27 min to 58.3 ± 4.98 min [24]. Unlike other studies [25, 26, 28], these investigators also could not demonstrate differences in the PK of ampicillin between non-pregnant individuals and pregnant women before the onset of labour [24]. In contrast to this study, we did not find differences in terminal half-life of amoxicillin between women before and after the onset of labour. Differences in study design might explain the discrepancy between the results. In the ampicillin study four blood samples were collected from each patient, whereas in our study an average of 18 blood samples were obtained for each patient per stage of labour. More intensive sampling will result in more reliable estimates of the interindividual variability. The pharmacokinetic description of studies with a high sampling density are expected to be more accurate and harbour the possibility of detecting also small differences in pharmacokinetic parameters between various patients. Whether differences in methodology or true differences between the study populations underlie the different conclusions of the studies remains to be determined. Alternatively, differences in chemical structure or features of the two drugs might also explain differences in the results. However, since the pharmacokinetics of ampicillin and amoxicillin after intravenous administration in non-pregnant individuals has been shown to be similar in previous studies [29, 30], this is unlikely.

Clinically relevant pharmacokinetic changes during pregnancy have also been demonstrated for other drugs. For example, the plasma concentration of the anti-epileptic drug lamotrigine has been shown to decrease during pregnancy. Subsequently, in the immediate postpartum period, the plasma concentration increases rapidly, resulting in a risk for toxicity [31]. This indicates that changes of specific drugs during pregnancy and labour cannot be readily extrapolated on the basis of results obtained with other drugs.

The interindividual variability in amoxicillin pharmacokinetics between the patients was remarkably moderate. In Figure 2 the observed concentrations are plotted vs the population predicted concentrations. The majority of the data points are located near the line of identity. However, for some observed concentrations the predicted concentrations are overestimated. These blood samples were taken during the antibiotic infusion. The antibiotic concentration increases very quickly during the infusion. We recorded the sampling times with a precision of 0.5–1 min. Therefore, the model will predict concentrations during the antibiotic infusion less accurately compared with concentrations in blood samples collected after the antibiotic infusion.

An important question for clinical practice is whether the recommended amoxicillin dosing regimen is adequate for the prevention of GBS-disease. The efficacy of amoxicillin is determined by the time the concentration exceeds the minimum inhibitory concentration (t>MIC) and, in general, t>MIC for 40–50 % of the dosing interval is required for efficacy [32–34]. The MIC value of an antibiotic is the highest MIC value of different microbial strains that results in a high probability of cure. MIC value of amoxicillin for GBS as indicated by the EUCAST is 0.25 mg l−1[35]. All concentrations remained well above the MIC of GBS for a sufficient percentage of the dosing interval, even when taking into account the plasma protein binding of approximately 18–20 % [36, 37].

After delivery, the pregnancy-induced physiological adaptations will change back to normal. This process starts immediately postpartum, but will continue for several weeks [38, 39]. In the first days after delivery, this includes an increase in blood volume and cardiac output. We found only minor differences in pharmacokinetic behaviour of amoxicillin in the immediate postpartum period when compared with pregnant women. This supports our earlier study that demonstrated a similar PK of amoxicillin in pregnant women with PPROM compared with values reported for non-pregnant individuals.

In our study, toxic or sub-therapeutic concentration profiles were reached in none of the patients. For this reason the differences in PK in the three stages of labour were considered not clinically relevant. This finding supports the current practice that the dosing regimen is not adjusted during the course of labour. However, it should be noted that we included only healthy patients and that only eight patients were included both before and during labour. Our patient group may be considered to be small, but taking into account the situation in which the patients had to be studied this is a relatively large population. Antibiotics in the prevention of GBS are also used to protect the foetus. Therefore, adequate foetal concentrations are imperative. Since uterine contractions might influence the blood flow in the umbilical cord, studies investigating the transplacental transfer of amoxicillin during vaginal deliveries are needed to describe the PK in umbilical cord serum and ultimately in the foetus.

Source of financial support: Supported by a grant from the Stichting Nuts Ohra (SNO-T-06-31).