Secondary Logo

Journal Logo

Pediatric Anesthesiology: Research Report

Measured Versus Predicted Blood Propofol Concentrations in Children During Scoliosis Surgery

Panchatsharam, Selvakumar FRCA*; Callaghan, Michael FFARCSI*; Day, Rachel*; Sury, Michael R. J. FRCA, PhD*†

Author Information
doi: 10.1213/ANE.0000000000000413
  • Free

Propofol total IV anesthesia is the preferred anesthetic technique for scoliosis surgery for 2 main reasons. First, scoliosis surgery requires intraoperative monitoring of evoked potentials to detect spinal cord ischemia,1,2 and inhaled anesthetic drugs suppress both sensory-3–5 and motor-evoked6–8 potentials more than propofol. Second, scoliotic children may have a myopathy, for example, Duchenne or Becker muscular dystrophy, for which inhaled drugs are contraindicated because they may cause rhabdomyolysis and cardiac failure.9–11 In our practice, we have observed prolonged recovery and cardiovascular depression with target-controlled infusion (TCI) of propofol, and we have been concerned that patients may have been exposed to excessive blood levels of propofol.

Propofol is commonly delivered by TCI, which involves computerized automatic adjustment of the infusion rate to achieve a desired target blood level. The computer algorithm is developed from a pharmacokinetic (PK) model created with data from a sample of patients whose age and body weight (and other potentially important characteristics) are used as covariates.12–15 However, because of interpatient PK variability within the sample, the ability of a model to predict the true blood level in any individual within that sample will be in doubt. Furthermore, the application of a model to individuals who do not share the characteristics of the original sample may worsen the prediction of the true propofol blood level (i.e., the performance of the model).16

Several PK models have been developed for children.15 The Paedfusor model has been shown to be predictive in children aged 1 to 15 years undergoing cardiac procedures,17 but other models including those developed by Marsh et al.18 and Kataria et al.19 did not perform well in children aged 6 to 12 years.20 Sepúlveda et al.21 tested 8 TCI models in infants and small children (≤3 years old) and found that there was wide variation in performance, although 6 were acceptable. The Paedfusor model may be the most suitable for children and young people having scoliosis surgery, but its performance in these patients is unknown. PK variation in scoliosis patients might be related to several factors—for example, the size of distribution compartments (such as reduced muscle bulk in myopathies) or altered propofol metabolism.

Scoliosis operations can last up to 6 hours, and we have observed that some children have had hemodynamic depression and delayed recovery. These problems could have been caused by excessively high propofol blood concentrations, and the presence of a cardiomyopathy could be particularly hazardous in these circumstances. Conversely, if there is major blood loss and large volumes of replacement IV fluids are needed, blood propofol concentrations may decrease and potentially risk awareness. In summary, the true blood level achieved during TCI in children having scoliosis surgery is unknown and could be important.

Measurement of blood propofol concentration for clinical purposes is not usually undertaken because of its expense and the delay involved with off-site analysis. A new device, the Pelorus 1500 (Sphere Medical, Cambridge, United Kingdom), has become available that makes point-of-care blood propofol analysis feasible.22 In this pilot observational study, we used this device to compare measured (Cm) with predicted concentrations (Cp) of propofol in children having TCI to determine if the differences were large enough to justify concern.


The study was approved by the National Research Ethics Service, London-Bloomsbury Committee (12/Lo/1847), authorized by the United Kingdom Medicines and Healthcare Regulatory Agency (EudraCT number 2012-003790-25) and registered with database (NCT01932424) on August 29, 2013. Written informed consent was obtained from parents and if possible from the patients. Children between the ages 5 and 18 years were recruited who were scheduled to have major spinal surgery lasting more than 3 hours. Exclusion criteria were major hepatic or renal disease.

Anesthetic management was unchanged by this study. Induction of anesthesia was either by TCI propofol or by inhaled sevoflurane according to patient preference. After inhaled inductions, TCI propofol began once IV access was established, and sevoflurane administration reduced gradually and was terminated by 15 minutes. Anesthesia was maintained by TCI in all patients using 2% propofol (Diprivan 2%, Astra Zeneca, Cambridge, United Kingdom) delivered by an Alaris Asena PK syringe infusion pump (Cardinal Health, Basingstoke, United Kingdom) using either the Paedfusor or Marsh TCI. On this device, the Paedfusor TCI is limited for body weight <61 kg, and the Marsh TCI age limit is ≥16 years. Hence, children who were above the weight limit for Paedfusor model and younger than 16 years were given Paedfusor or Marsh TCI according to the anesthesiologist’s preference. Remifentanil was infused (0.1–0.5 µg·kg−1·min−1) for intraoperative analgesia. An arterial cannula was inserted for intraoperative monitoring. After positioning the patient for surgery, arterial blood samples were taken in heparinized syringes every 20 to 30 minutes during the maintenance phase of anesthesia (not during emergence). Blood concentrations of propofol, lactate, glucose, and hemoglobin (Hb) were measured from these samples in addition to acid-base status. The maximum number of blood propofol measurements per patient was limited to 10. Blood samples were taken at least 5 minutes after any alteration of the target concentration. Blood propofol concentration was measured using Pelorus 1500 (Sphere Medical).22 This point-of-care device enabled blood propofol measurements from 0.5-mL samples of blood within 3 minutes. The Pelorus 1500 uses solid phase extraction and colorimetric detection. Calibration and quality control tests were performed before each case as per manufacturer’s recommendation. Compared with high performance liquid chromatography, the Pelorus propofol estimation in blood has a bias of 0.13 µg·mL−1 and a precision 0.16 to 0.42 µg·mL−1.23

The blood propofol Cps from the TCI pump and physiologic variables from a Philips IntelliVue MP60 were downloaded onto a PC using RugloopII© software (Demed, Belgium). At each measurement of Cm, pH and concentrations of Hb, glucose, and lactate were measured. The volumes of blood and all other IV fluids used between Cm measurements were recorded.

The Cm and Cp blood propofol concentrations in each patient were plotted against time. The distribution of all Cm–Cp was described for the patient group. The performance of a TCI model was evaluated using methods by Varvel et al.16 The following descriptors are explained in Appendix 1. Performance error (PE) and absolute PE (APE) were calculated for each pair of Cm and Cp. For each patient, the median PE (MDPE) and the median APE (MDAPE) were calculated and represented bias and precision, respectively. The steadiness of PE was described by the wobble (median of the difference between PE and MDPE). The divergence described the linear relationship of APE to time. The effect of blood loss was examined by correlation of the cumulative IV fluid volume with PE and also the relationship between the change in PE and the increment of IV volume infused between each blood propofol measurement.

Descriptive statistics were performed on Microsoft Excel and R (version 2.15.1). In this pilot study, a sample size of 20 patients was expected to provide sufficient blood samples (5–10 per child) to determine whether there was an appreciable difference between the Cm and Cp values.17,18,20


In the 20 children, the ages ranged from 9 to 17 years (mean, 15 years), body weight 24.5 to 95 kg (mean, 48.4 kg), the sex ratio was 10/10, and 14 had scoliosis related to neuromuscular disease of a muscular dystrophy (Appendix 2 lists diseases and syndromes). The Paedfusor model was used in 16 children, of whom 13 weighed <61 kg and were ≤18 years; 3 others weighed 95, 74, and 66 kg and were 12, 14, and 17 years old, respectively. The Marsh model was used in 4 children aged 14, 14, 16, and 16 years who weighed 69, 73, 60, and 56 kg, respectively.

Surgery lasted >2 hours in all but 1 patient and >4 hours in 9 patients (Table 1). Blood losses and IV fluid volumes were higher, and the lowest Hb concentrations were lower in the Paedfusor group; the smallest patients tended to lose more blood and have more IV fluids in proportion to their body weight. In any patient, the lowest pH, base excess, and blood glucose level was 7.21, −3.2, and 3.9 mmol·L−1, respectively, and the highest blood lactate concentration was 4.31 mmol·L−1.

Table 1
Table 1:
Duration of Anesthesia, Blood Loss, Hemoglobin, and Acid-Base Status

Anesthesia was induced with sevoflurane in 8 children. Target propofol concentration was set between 3 and 7 µg·mL−1 during surgery. There were 154 blood propofol measurements. The numbers of blood samples collected in children were as follows: 6 samples in 5 children, 7 in 3, 8 in 7, 9 in 3, and 10 in 2. Technical problems delayed measurements in 1 patient for 185 minutes.

Cm–Cp and PE were normally distributed (Shapiro-Wilk 0.865 and 0.9194, respectively). Figure 1 is a scatter plot of Cm–Cp versus time and shows that Cm was usually higher than Cp and that there was a trend for Cm–Cp to decrease over time. However, this plot does not show the Cm–Cp in relation to Cp. Figure 2 separates the data in Figure 1 into 5 smaller scatter plots of Cm–Cp versus time at various Cps. It can be seen that there were an appreciable number of negative Cm–Cp values when the Cp was between 3 and 4 mcg/mL.

Figure 1
Figure 1:
Difference between measured concentrations (Cm) and predicted concentrations (Cp) of propofol during maintenance phase of target-controlled infusion.
Figure 2
Figure 2:
Difference between measured concentrations (Cm) and predicted concentrations (Cp) against time at various predicted concentrations during the maintenance phase of target-controlled infusion.

Using all data points in all children, mean Cm–Cp was 1.5 µg·mL−1 (LOA, −1.4 to 4.5 µg·mL−1),a and mean PE was 44.7% (LOA, −40.1% to 130.2%).24 For the whole patient sample, there was a trend for PE to decrease over time (Pearson correlation coefficient = −0.47, P < 0.0001, r2 = 0.22).

Figure 3 shows the trend in PE for individual patients over time. Cm was almost always more than Cp, and therefore MDPE and MDAPE were equal except for 2 patients. The median of MDPEs for the patient group was 39.8%: the MDPE was >50% in 8 children (Table 2). The MDPE was not associated with body weight (Spearman correlation ρ = 0.36, P = 0.23). Two children had consistently lower Cm than Cp. Their lowest Cms were 1.74 and 1.96 µg·mL−1 when the Cp was 3 µg·mL−1. Both had the Paedfusor model and their body weights were 28 and 33 kg.

Table 2
Table 2:
Target-Controlled Infusion Performance Measures
Figure 3
Figure 3:
Progression of performance error with time.

The median (and range) of the wobble was 13.6% (4.0%–42.3%). All but 1 patient had a negative divergence (Table 3). Two children had wobble >30%; they also had the Paedfusor model and their body weights were 24.5 and 55.9 kg. The measures of performance tended to be highest (i.e., worse) in children who had the Paedfusor model (Table 3).

Table 3
Table 3:
Target-Controlled Infusion Performance Measures

The estimated blood loss was not accurate enough to justify analysis. The effect of IV fluid infusion on performance was examined. Total IV fluid volume infused was not associated with MDPE (Spearman correlation ρ = −0.25, P = 0.29) (Fig. 4). Combining all data, PE decreased with time and also with the cumulative IV fluid volume replacement (Pearson correlation coefficients 0.476 and 0.51, respectively); combining both time and IV volume into a regression model did not improve the correlation over the effect of IV volume alone. There was no association between increment of IV volume infused and the change in PE (Pearson correlation R = −0.051, P = 0.56) (Fig. 5).

Figure 4
Figure 4:
Median performance error (MDPE) versus total IV fluid volume infused. The linear regression line for all patients is shown (Spearman correlation ρ = −0.25,P = 0.29).
Figure 5
Figure 5:
Change in performance error versus IV volume increment for all patients. Linear regression line is shown (Pearson correlation R = −0.051,P = 0.56).


There was often a major difference between Cm and Cp during scoliosis surgery and, usually, the Cm was higher than Cp. There was no evidence of accumulation. Divergences were negative except in 1 patient in whom the highest Cm at the end of surgery was 5.64 µg·mL−1 (Cp = 3.5 µg·mL−1). Two children, however, had Cm appreciably lower than Cp. During Cp 3 µg·mL−1, their lowest Cm(s) were 1.74 and 1.96 µg·mL−1, and at these levels, there may have been a risk of intraoperative awareness. The limits of acceptable TCI performance, proposed by Schüttler et al.,25 are that the mean PE should be <20% and the mean |PE| should be <30%. The MDPE exceeded 20% in all but 1 of our patients.

Other studies have reported better performances of TCI models. A model designed for children (aged <11 years) by Marsh et al.18 had a mean PE of 0.9% and mean |PE| of 20.1%. This “Marsh” model was tested by Short et al.26 in Chinese children aged 4 to 10 years and found to have a mean PE of −18.5% and a mean |PE| of 24.8%. With some adjustment to the PK variables, the mean |PE| was improved to 11.9%. Currently median, rather than mean, PE and |PE| are preferred. Absalom et al. showed that the Paedfusor model, in children aged 1 to 15 years, had a MDPE of 4.1% and MDAPE of 9.7%.17 In a study of infants and young children aged 3 to 26 months, 8 TCI models were compared.21 The Paedfusor model performed favorably (MDPE 5%, MDAPE 28.3%), but the Short model was best (MDPE 5.4%, MDAPE 18.2%).26 In adults, Coetzee et al.27 found that the Marsh model performed well: MDPE −7% and MDAPE 18%.

Wide differences between Cm and Cp may be foreseen given the potential variation in the physical and physiological characteristics of children. Many of the children in this study had scoliosis associated with syndromes for which the PK of propofol is unknown. For children with uncommon diseases, it may be impractical to study enough cases to calculate a PK model specific to that patient group. The influence of other anesthesia drugs, especially remifentanil, on the kinetics of propofol may also be important. There was a dilemma in deciding which model to use for children ≥61 kg and <16 years. Our data are not sufficient to support the preference of either Marsh of Paedfusor models, but the performance variables of the Paedfusor TCI model tended to be worse in younger and smaller children.

Blood loss tended to be higher in the smaller patients and the influence of blood loss and IV fluids may have been important. Nevertheless, there was no relationship between the volume of IV fluids administered and change in PE, that is, the PE increased or decreased, equally as often, irrespective of IV fluid volume infused. Blood loss may reduce Cm because propofol is lost with the blood, but this may be offset by a reduction in the central compartment volume, which might increase Cm if Cp is maintained.

Unintentional and unpredictable high Cm may be an important problem especially when the effects of excessive propofol are hazardous (e.g., cardiac suppression in cardiomyopathy). The high Cm compared to Cp suggests that the central compartment is overestimated by the TCI models. Yet this was not found in all children, and therefore, if an adjustment of the TCI model was considered, it would not be suitable for all children.

Continuous monitoring of the concentration of propofol in blood would be a major step forward, but it is not yet available. Exhaled breath propofol measurement is an attractive method,28 but current published data show that it is not as reliable as blood propofol assay.29 Now that reliable, albeit intermittent, point-of-care blood propofol assay is feasible, the performance of TCI can be checked and an adjustment made to the TCI model for the individual patient. This has been proposed by Cowley and Clutton-Brock,b who showed that the performance of the Marsh TCI model can be appreciably improved in individual patients by recalibrating the model using a proportional correction factor made on a single measurement of Cm after 30 minutes of TCI (Correction factor = Cm at 30 minutes/Cp at 30 minutes; Cp(corrected) = Cp × correction factor). We have applied this process to our data. First, a correction factor for each patient was calculated based on the first Cm taken >30 minutes after starting TCI (the first Cm could be much higher than Cp and was therefore ignored). The correction factor was applied to all subsequent Cps and corrected PEs were calculated. Performance was appreciably improved by this adjustment. The means of uncorrected and corrected MDPEs were 41.5% and −9%: mean difference 50.5%, 95% CI, 36.4 to 64.6. The means of uncorrected and corrected MDAPEs were 45.6% and 16.8%: mean difference 28.9%, 95% CI, 10.9 to 46.8. A plot of corrected PEs against time is shown in Figure 6.

Figure 6
Figure 6:
Progression of corrected performance error with time.

We studied a small sample of heterogeneous children who were larger than the children enrolled into other studies. Our results may not reflect the performance of TCI in children generally or in other specific situations. Nevertheless, wide and variable differences between Cm and Cp during major prolonged surgery may be found elsewhere, and this report may serve as a warning to clinicians if they encounter unexpected clinical signs of either excessive or inadequate propofol anesthesia at what they perceive as a safe and effective dose.


Performance Error = PE ij (%) = Cmij – Cpij × 100 Cpij

(Median = MDPE)

Absolute Performance Error = APEij (%) = | PEij% |

(Median = MDAPE)

Wobble i (%) = Median of | PEij – MDPEi |

Divergence i (%/minute) = regression coefficient of | PEij% | against time ij

(Cm = Concentration Measured; Cp = Concentration Predicted; i represents the individual; j represents the sample of measurement)

Appendix 2
Appendix 2:
Patients with Conditions Associated with Scoliosis


Name: Selvakumar Panchatsharam, FRCA.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Attestation: Selvakumar Panchatsharam has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Michael Callaghan, FFARCSI.

Contribution: This author helped conduct the study and analyze the data.

Attestation: Michael Callaghan has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Rachel Day, medical student.

Contribution: This author helped analyze the data.

Attestation: Rachel Day has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.

Name: Michael R. J. Sury, FRCA, PhD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Attestation: Michael R. J. Sury has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.

This manuscript was handled by: James A. DiNardo, MD.


a LOA = 95% limit of agreement. LOAs, being wide, were estimated as mean ± SD × 2, and calculated for repeated measures, taking into account covariance within subjects.24
Cited Here

b Cowley N, Clutton-Brock T. Use of a device to measure blood propofol levels to improve inter-patient bias of propofol target-controlled infusion. Annual Scientific Meeting: Society for Intravenous Anaesthesia UK (SIVA UK); Nottingham, 2012.
Cited Here


1. Lieberman JA, Lyon R, Feiner J, Hu SS, Berven SH. The efficacy of motor evoked potentials in fixed sagittal imbalance deformity correction surgery. Spine (Phila Pa 1976). 2008;33:E414–24
2. Li F, Gorji R, Allott G, Modes K, Lunn R, Yang ZJ. The usefulness of intraoperative neurophysiological monitoring in cervical spine surgery: a retrospective analysis of 200 consecutive patients. J Neurosurg Anesthesiol. 2012;24:185–90
3. Boisseau N, Madany M, Staccini P, Armando G, Martin F, Grimaud D, Raucoules-Aimé M. Comparison of the effects of sevoflurane and propofol on cortical somatosensory evoked potentials. Br J Anaesth. 2002;88:785–9
4. Liu EH, Wong HK, Chia CP, Lim HJ, Chen ZY, Lee TL. Effects of isoflurane and propofol on cortical somatosensory evoked potentials during comparable depth of anaesthesia as guided by bispectral index. Br J Anaesth. 2005;94:193–7
5. Fung NY, Hu Y, Irwin MG, Chow BE, Yuen MY. Comparison between sevoflurane/remifentanil and propofol/remifentanil anaesthesia in providing conditions for somatosensory evoked potential monitoring during scoliosis corrective surgery. Anaesth Intensive Care. 2008;36:779–85
6. Lotto ML, Banoub M, Schubert A. Effects of anesthetic agents and physiologic changes on intraoperative motor evoked potentials. J Neurosurg Anesthesiol. 2004;16:32–42
7. Sloan TB, Heyer EJ. Anesthesia for intraoperative neurophysiologic monitoring of the spinal cord. J Clin Neurophysiol. 2002;19:430–43
8. Wang AC, Than KD, Etame AB, La Marca F, Park P. Impact of anesthesia on transcranial electric motor evoked potential monitoring during spine surgery: a review of the literature. Neurosurg Focus. 2009;27:E7
9. Gurnaney H, Brown A, Litman RS. Malignant hyperthermia and muscular dystrophies. Anesth Analg. 2009;109:1043–8
10. Takahashi H, Shimokawa M, Sha K, Sakamoto T, Kawaguchi M, Kitaguchi K, Furuya H. Sevoflurane can induce rhabdomyolysis in Duchenne’s muscular dystrophy [in Japanese]. Masui. 2002;51:190–2
11. Schmidt GN, Burmeister MA, Lilje C, Wappler F, Bischoff P. Acute heart failure during spinal surgery in a boy with Duchenne muscular dystrophy. Br J Anaesth. 2003;90:800–4
12. White M, Kenny GN. Intravenous propofol anaesthesia using a computerised infusion system. Anaesthesia. 1990;45:204–9
13. Schüttler J, Ihmsen H. Population pharmacokinetics of propofol: a multicenter study. Anesthesiology. 2000;92:727–38
14. Anderson BJ. Pediatric models for adult target-controlled infusion pumps. Paediatr Anaesth. 2010;20:223–32
15. Mani V, Morton NS. Overview of total intravenous anesthesia in children. Paediatr Anaesth. 2010;20:211–22
16. Varvel JR, Donoho DL, Shafer SL. Measuring the predictive performance of computer-controlled infusion pumps. J Pharmacokinet Biopharm. 1992;20:63–94
17. Absalom A, Amutike D, Lal A, White M, Kenny GN. Accuracy of the “Paedfusor” in children undergoing cardiac surgery or catheterization. Br J Anaesth. 2003;91:507–13
18. Marsh B, White M, Morton N, Kenny GN. Pharmacokinetic model driven infusion of propofol in children. Br J Anaesth. 1991;67:41–8
19. Kataria BK, Ved SA, Nicodemus HF, Hoy GR, Lea D, Dubois MY, Mandema JW, Shafer SL. The pharmacokinetics of propofol in children using three different data analysis approaches. Anesthesiology. 1994;80:104–22
20. Rigouzzo A, Servin F, Constant I. Pharmacokinetic-pharmacodynamic modeling of propofol in children. Anesthesiology. 2010;113:343–52
21. Sepúlveda P, Cortínez LI, Sáez C, Penna A, Solari S, Guerra I, Absalom AR. Performance evaluation of paediatric propofol pharmacokinetic models in healthy young children. Br J Anaesth. 2011;107:593–600
22. Liu B, Pettigrew DM, Bates S, Laitenberger PG, Troughton G. Performance evaluation of a whole blood propofol analyser. J Clin Monit Comput. 2012;26:29–36
23. Cowley NJ, Laitenberger P, Liu B, Jarvis J, Clutton-Brock TH. Evaluation of a new analyser for rapid measurement of blood propofol concentration during cardiac surgery. Anaesthesia. 2012;67:870–4
24. Bland JM, Altman DG. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat. 2007;17:571–82
25. Schüttler J, Kloos S, Schwilden H, Stoeckel H. Total intravenous anaesthesia with propofol and alfentanil by computer-assisted infusion. Anaesthesia. 1988;43 Suppl:2–7
26. Short TG, Aun CS, Tan P, Wong J, Tam YH, Oh TE. A prospective evaluation of pharmacokinetic model controlled infusion of propofol in paediatric patients. Br J Anaesth. 1994;72:302–6
27. Coetzee JF, Glen JB, Wium CA, Boshoff L. Pharmacokinetic model selection for target controlled infusions of propofol. Assessment of three parameter sets. Anesthesiology. 1995;82:1328–45
28. Harrison GR, Critchley AD, Mayhew CA, Thompson JM. Real-time breath monitoring of propofol and its volatile metabolites during surgery using a novel mass spectrometric technique: a feasibility study. Br J Anaesth. 2003;91:797–9
29. Perl T, Carstens E, Hirn A, Quintel M, Vautz W, Nolte J, Jünger M. Determination of serum propofol concentrations by breath analysis using ion mobility spectrometry. Br J Anaesth. 2009;103:822–7
© 2014 International Anesthesia Research Society