The paracetamol concentration-effect relation in neonates



Karel Allegaert, MD, PhD, Neonatal Intensive Care Unit, University Hospital, Herestraat 49, 3000 Leuven, Belgium




We suggested a loading dose (20 mg·kg−1) followed by 10 mg·kg−1 q6h of intravenous (IV) paracetamol to achieve an effect compartment concentration of 11 mg·l−1 in neonates. Since there are no pharmacodynamic data to support such an effect compartment concentration, pain scores collected in neonates treated with an IV paracetamol loading dose (20 mg·kg−1) were used to validate this effect compartment concentration.


Pain scores (Leuven Neonatal Pain Score, LNPS, 0–14) before and 0.5, 1, 2, 3, 4, 5, and 6 h after IV paracetamol loading dose administration in neonates to whom IV paracetamol was administered as single analgesic (PARANEO,, NCT00969176) were collected. Trends were analyzed using repeated measures anova. An Emax model with a delayed response compartment was fitted to data using population modeling.


Nineteen of 60 neonates included in the PARANEO study received monotherapy with IV paracetamol to treat mild to moderate pain (e.g., alprostadil administration, delivery related trauma). Using repeated measures anova, there was a trend (= 0.02) for lower pain scores within 30 min after administration, with a slight increase in pain scores from 5 to 6 h. An Emax model had a maximum effect of 4.15 pain units, an EC50 of 2.07 mg·l−1. Equilibration halftime (T1/2keo) was 1.58 h.


Intravenous paracetamol is effective for moderate pain. An effect compartment concentration of 10 mg·l−1 (loading dose of 20 mg·kg−1) is associated with a pain score reduction of 3.4 LNPS units. This analysis suggests a similar paracetamol effect compartment concentration in neonates compared to children.


Improved understanding of developmental pharmacokinetics (PK) and pharmacodynamics (PD) contributes to safer and more effective analgesic dosing regimens for neonates [1]. One of the commonly administered compounds is paracetamol (N-acetyl p-aminophenol, acetaminophen) to treat mild to moderate pain or fever [2-9]. An intravenous (IV) formulation can be used to improve effect predictability or when an enteral route is not available [2-9]. Aspects of PK, metabolism, and safety (hepatic tolerance, hemodynamics, normothermia) of IV paracetamol in neonates have been previously published [10-13].

A loading dose of intravenous (IV) paracetamol 20 mg·kg−1 followed by 10 mg·kg−1 6 hourly to achieve an effect compartment concentration of approximately 11 mg·l−1 in neonates has been suggested, but that dosing schedule was based on paracetamol analgesic pharmacodynamics documented in children; there are no data supporting the same effect compartment concentration in neonates [10, 14].

Effective analgesia in neonates is still in part hampered by paucity of pharmacodynamic data. In an attempt to provide evidence for analgesic effectiveness of IV paracetamol (loading dose, 20 mg·kg−1) in neonates and for (dis)similarity in effect compartment between children and neonates, pain scores after an IV paracetamol loading dose (, NCT00969176) were analyzed using both repeated measures anova and an Emax model with a delayed response compartment.


Study registration and ethics

The PARANEO study was a prospective, single center, open label study on PK (primary outcome), PD (pain), and safety (hepatic tolerance, secondary outcomes) of IV paracetamol in neonates [10]. Following study registration (PARANEO,, NCR 00969176) and approval by the ethics committee (EC Leuven, Eudract nr 2009-01143-39), 60 neonates were included following informed written parental consent. The PK of repeated dose administration of IV paracetamol was the primary outcome, and these observations were recently published using a pooled PK approach [10].

Clinical characteristics, dosing regimen, and blood sampling strategy

Sixty neonates (i.e., <29 days of postnatal age), either preterm (<37 weeks gestational age) or term, were included. The decision to prescribe paracetamol was made by the attending physician and in line with the current standing orders in our neonatal unit, that is, either as monotherapy for mild to moderate pain, independent of the cause of pain (postoperative, traumatic, or medical) or as part of multimodal analgesia [10, 15]. The dosing regimen of IV paracetamol administration in neonates was based on a loading dose (20 mg·kg−1), followed by an postmenstrual age-dependent maintenance dose of 20–40 mg·kg−1 24 h−1, divided in four doses (5–10 mg·kg−1 6 hourly) [10]. In 19 of 60 neonates, IV paracetamol was the only analgesic administered.

The absence of an arterial access was a contra-indication (blood sampling for PK aspect of study, primary outcome) for inclusion. Consequently, an overrepresentation of neonates in whom multimodal analgesia was applied for severe, postoperative pain was anticipated. In 60 neonates, 445 blood samples were collected in the first 24 h, with specific emphasis on the pharmacokinetics following the loading dose with sampling both shortly after (<2 h, peak levels) and just before the next administration (5–6 h, through levels). Blood samples were centrifuged, and plasma was analyzed using high-performance liquid chromatography. Variation coefficients in intra- and inter-day accuracy and precision were all below 15%, the lower limit of quantification was 0.08 μg·ml−1, and linearity of the calibration curve in plasma was found in the range between 0.078 and 40 μg·ml−1 [10, 16].

There were 43 blood samples collected in the 6-h time interval after IV paracetamol (20 mg·kg−1) loading dose administration available from 19 neonates with only exposition to IV paracetamol (Figure 1).

Figure 1.

Paracetamol plasma measurements (= 43) collected in 19 neonates to whom paracetamol was administered as single analgesic during the PARANEO study following loading dose administration (20 mg·kg−1) [X-axis = time (hours); Y-axis = paracetamol plasma concentration (mg·l−1).

Pain assessment

In all 60 neonates, pain scores were prospectively collected (Leuven Neonatal Pain Score, LNPS) just before, and at 0.5, 1, 2, 3, 4, 5, and 6 h after IV paracetamol loading dose administration and during maintenance dosing [15, 17]. In the current secondary outcome analysis, we only report on observations in cases to whom IV paracetamol was administered as monotherapy to treat pain and up to 6 h after the loading dose was administered (until just before, the first maintenance dose was provided). LNPS pain scores were collected by one of the authors (KA, SV) of the publication.

The LNPS score is a multidimensional pain scale, based on seven items (sleep, facial appearance, crying, heart rate, motor tone, movement, and consolability). Each variable is graded 0, 1, or 2 points. Validation was based on paired observations and on the link between the LNPS and catecholamine release. Six hundred twenty-eight observations in neonates admitted in our neonatal unit were scored simultaneously by two caregivers blinded for each other's observations. A strong inter-observer correlation (r = 0.88) in LNPS scores was observed. In addition, a strong correlation (r = 0.8) between an increase in pain score and an increase in catecholamines was demonstrated during endotracheal aspiration [15, 17].


Clinical characteristics were reported by median and range or incidence. LNPS pain score were reported as median and 95% CI. LNPS trends were analyzed using repeated measures anova. A P-value of <0.05 was significant (MedCalc®, Mariakerke, Belgium).

PK-PD model

Leuven Neonatal Pain Scores were added to the existing PK data and the PK-PD data analyzed simultaneously using non-linear mixed effects models (NONMEM VII, Icon Development Solutions, Ellicott City, MD, USA). The PK modeling was the same as previously described [10]. The ADVAN six subroutine was used to add a delayed effect compartment. This delayed effect compartment was given a very small volume by scaling the first-order rate constant for transfer from the central to the effect compartment to 1/1000 of that of the elimination rate constant from the central compartment, so as not to influence the PK of the central compartment [18].

The population parameter variability is modelled in terms of random effect (η) variables. Each of these variables is assumed to have mean 0 and a variance denoted by ω2, which is estimated. We report the estimate of ω for each variability component expressed as a percentage because these quantities are approximate coefficients of variation for a log normal distribution.

A modified Emax model was used to describe drug effect:

display math(1)

Where E0 is the pain score before paracetamol, Emax is the maximum effect, Ce (mg·l−1) is the paracetamol concentration in the effect compartment, and EC50 (mg·l−1) is the concentration producing 50% Emax. Initial analyses revealed that Emax approached 0, so the model was simplified:

display math(2)

The population geometric mean parameters between subject variance and additive residual variance were estimated using the first-order conditional interaction estimation method. Convergence criterion was to three significant digits [18].

In any model, the quality of the individual parameter estimate will depend heavily on the observed data available. Data were sparse in this current model, and this can result in reduced variance (ω2) of parameter estimates and distortions of the distribution shape. For example, if no data are available on a particular individual, the individual's estimate will be equal to the population value; the variance is shrinking toward zero as the quantity of information at the individual level diminishes, a phenomenon defined as η-shrinkage (shε). This was calculated:

display math(3)

Where SD approximates the standard deviation. When there is no shrinkage, the model is correct and individual data are sufficiently abundant for individual parameter estimation. Data contain virtually no information about these parameters when shrinkage is 100%, and the individual parameter values approach the typical parameter value.


There were 60 neonates in the PARANEO study. Nineteen received monotherapy with IV paracetamol to treat mild to moderate pain. These provided 142 pain scores. The clinical characteristics and the indications to initiate treatment in these 19 cases are reported in Table 1, and individual paracetamol concentrations as quantified in these 19 neonates are shown in Figure 1.

Table 1. Clinical characteristics and indications in neonates included in the PARANEO study and to whom IV paracetamol was administered as single analgesic. Data provided by median and range or absolute numbers
Clinical characteristics
Gestational age (weeks)38 (26–41)
Postnatal age (days)2 (1–17)
Postmenstrual age (weeks)38 (27–42)
Preterm (<37 weeks at birth)5/19
Birth weight (kg)2.95 (0.9–4.02)
Weight at inclusion (kg)2.84 (0.9–3.96)
Main indication
Medical conditions
Subarachnoidal bleeding2/19
Necrotizing enterocolitis1/19
Alprostadil administration7/19
Traumatic/postoperative pain9/19

Before administration, median (95% CI) LNPS scores were 4 (2.8–6). Thirty minutes, 1 h, 2 h, 3 h, 4 h, 5 h, and 6 h after administration, LNPS scores were 1 (0–2), 1 (0–2), 0 (0–1), 0 (0–1), 0 (0–1), 0 (0–2), and 0 (0–3), respectively. Using repeated measures anova, there was a significant trend (= 0.02) for lower pain scores within 30 min after administration, with a non-significant increase in pain scores from 5 to 6 h. LNPS trends are provided in Figure 2. Pain scores approached zero and simplification of the model equation (1) implies that Emax is equivalent to E0.

Figure 2.

Leuven Neonates Pain Scale scores before and in the first 6 h after loading dose administration (20 mg·kg−1 paracetamol) are provided in 19 neonates (Box Whisker plots), who received paracetamol as monotherapy [X-axis = time (hours); Y-axis = Leuven Neonatal Pain Scale (0–14)].

Paracetamol PD estimates are shown in Table 2. Emax (equivalent to E0) was 4.15 pain units. An effect compartment concentration of 10 mg·l−1 is associated with a pain score reduction of 3.4 pain units (Figure 3). A visual predictive check – a modeling tool that estimates the pain score prediction intervals and graphically superimposes these intervals on observed pain scores – was used to evaluate how well the model predicted the distribution of observed pain scores (Figure 4) [16].

Table 2. Pharmacodynamic parameter estimates.%CV is the coefficient of variation of the parameter estimate,%SE is the standard error of the mean of the structural estimate, and% Shrinkage is a measure of the information contained in the data
ParameterEstimate% CV% SE% Shrinkage
  1. a

    The T1/2keo is that for a 70 kg person. A 3 kg neonate would have a T1/2keo of 1 h 12 min.

  2. display math(4)
E0 (pain units)4.1512316.771.7
EC50 (mg·l−1)2.0716142.939.5
aT1/2keo ha1.5820830.556.8
Additive error (pain units)0.928.79.17
Figure 3.

The relationship between paracetamol concentration in the effect-site and pain score is shown in this figure. Individual observations are shown as open diamonds while the relationship described by the Emax model is shown as a solid line.

Figure 4.

Visual predictive check for the paracetamol PD model. All plots show median and 90% intervals (solid and dashed lines). Left hand plot shows all observed pain scores. Right hand plot shows prediction percentiles (10%, 50%, and 90%) for observations (lines with symbols) and predictions (lines) with 95% confidence intervals for prediction percentiles (gray shaded areas).


We report on the analgesic efficacy of an IV loading dose of paracetamol in neonates determined as a secondary analysis of the PARANEO study. Modeling these pain scores estimated an Emax of 4.15, similar to the Emax of 5.17 determined in children using tonsillectomy as a pain model [14]. Pain relief at an effect-site concentration of 10 mg·l−1 was also similar (3.4 vs 2.6 pain units) while the equilibration half time was greater in the current analysis (71 min vs 24 min in a 3 kg neonate) [14]. The two studies differ considerably because age, pain insult, and pain assessment are all dissimilar. In addition, pain insults within the current study were not uniform while the earlier analysis included a placebo effect. Despite these differences, the current data justify the use of a similar target concentration in neonates compared to children in future PD studies (e.g., opioid sparing effect of paracetamol in neonates) [10, 14].

Dosing of analgesics in infants should be based on PK and PD considerations and adverse effects profiles [1, 19, 20]. While an effect-site paracetamol concentration of 10 mg·l−1 has been proposed for analgesia after tonsillectomy, lower concentrations may be satisfactory for children in less pain. van der Marel showed that a lower concentration achieved adequate analgesia in infants following craniofacial surgery [4, 8]. Speed of analgesic onset is also an important consideration, and this is dependent on the route of administration [20]. Intravenous paracetamol proved to be more effective compared to rectal paracetamol in the craniofacial surgery model in infants [8]. This is because the target concentration is reached sooner compared to rectal administration [8, 20] that is associated with slow and variable absorption [14, 19].

There are observations in children that the paracetamol cerebrospinal fluid (CSF) concentrations are more closely related to pain scores than blood concentrations and that the delay in analgesia in part reflects the time needed for paracetamol to appear in the CSF [21]. However, the time to peak paracetamol concentration in CSF of children seems to be similar when compared to newborns. Consequently, it is less likely that PK-related maturational changes in CSF permeability explain the differences in equilibration halftimes observed [21, 22]. Besides maturational changes in paracetamol PK, maturational changes in paracetamol PD should also be considered [9]. The mechanisms of action of paracetamol relate in part to prostaglandin inhibition, cannabinoid receptor, or serotonin receptor interactions [9]. These neural pathways also display maturational changes that may explain differences in both onset of action, the level of effective analgesia, and its duration. Our observations suggest that the overall phenotypic level of analgesia is similar in neonates compared to children, although more data are needed to make such a claim robust.

There was also a slight decrease in effective analgesia from 5 to 6 h, reflected by the equilibration halftime. This observation is consistent with the observations of Capici et al. who documented that the proportion of children without need for rescue analgesia following IV paracetamol (15 mg·kg−1) decreased significantly from 6 h onwards following adenotonsillectomy [23]. However, our current data did not reveal a return to the initial pain scores. This may in part be explained by either residual paracetamol effects, pain resolution or other coping mechanisms.

We are very much aware that the current report and the study design have limitations. Effective analgesia was only a secondary outcome variable of the PARANEO study, IV paracetamol was administered as an open label drug without any blinding procedure, and only cases with monotherapy for mild to moderate pain syndromes and after the first loading dose administration were considered in this analysis. Moreover, there were limited data from which to determine a population analysis. Shrinkage estimates are greater than 30% (Table 2). When data are very informative, shrinkage is zero and it moves toward 100% when data are less informative [18]. Bias can result with only 20–30% shrinkage. The current high shrinkage estimates combined with high standard errors associated with the structural parameters informs us that the current analysis is rudimentary [18]. More robust data are required to characterize this PD relationship with greater certainty in neonates. Despite these limitations, the current analysis suggests that IV paracetamol does result in effective analgesia for moderate pain.


Karel Allegaert is supported by the Fund for Scientific Research, Flanders (Belgium) (F.W.O. Vlaanderen) by a Fundamental Clinical Investigatorship (1800209N).