1. INTERMITTENT IV VANCOMYCIN WHAT DOSE SHOULD WE START WITH?
Clare Nash
Sheffield Children’s NHS Foundation Trust
November 2014 1

2. Vancomycin Glycopeptide antibiotic in use since 1950s
Useful vs Staph including MRSA
Renal toxicity
Monitor trough concentrations
Time dependent killing
Small post antibiotic effect

3. 15mg/kg tds 5-10mg/l (10-15mg/l for less sensitive organisms)
Pre 2007
2007 onwards
5-10mg/l (10-15mg/l for less sensitive organisms)
10-15mg/l (15-20mg/l)

4. Why were target troughs increased?
Increasing resistance
Long use shown less renal toxicity than originally thought
AUC24/MIC >400mg.h/l

6. Method Prospective data collection Seven paediatric centres
Record every dose, time dose given, level and time levels were taken.
Collect demographic data, indication, renal function, co-morbidities likely to affect vancomycin kinetics.
Descriptive Statistics/PK analysis
Basing our work around this study we decided to do something similar that would be applicable to UK practice.

7. Descriptive Statistics
243 patients
1017 vancomycin levels
First level
in range 24%
Low 69%
High 7%
Patients NEVER in range 33%

8. Trough levels by Creatinine Clearance
CRCL (ml/min/1.73m2)
Number %
Mean Trough levels (mg/L)
Min
Max
Stand. Dev.
[0-50] 41
7.77%
17.19
0.15
22.3
6.81
>50
487
92.23%
11.60
0.1 24
6.47
Total
528
100.00%

9. Mean Trough levels (mg/L)
Trough levels by age
Age
Number %
Mean Trough levels (mg/L)
Min
Max
Stand. Dev.
1-6months
101
19.17%
13.46
3.1 28
5.55
6months-2years
127
24.10%
10.85
2.4 40
6.13
2years to 12 years
215
40.80%
11.99
2.1 46
7.47
>12 years 84
15.94%
12.21
7.59
Total
527
100.00%

10. Results of Pharmacokinetic Modelling and Simulation
Number of patients = 226
Patients on renal/extracorporeal support excluded
Number of observations (plasma levels) = 912
>85% trough levels, < 5 % peak levels
All data pooled into a single analysis dataset
Pharmacokinetic parameters (clearance, volume of distribution) estimated using ‘Non-linear mixed effects modelling’
1-compartment PK model best fit the data

11. Results
Clearance (CL) found to be dependent on actual Body Weight and Creatinine Clearance
Volume of distribution (Vd) found to be dependent on Age
Significant variability in found in both parameters : CL (46 % CV), Vd (72% CV)
Final PK model then used to simulate BNFC and optimum doses
Doses that achieve AUC0-24 > 400 identified
Model thoroughly evaluated using standard statistical techniques

12. Simulations from the model
BNF-C Dose (15mg/kg every 8 hours)
New doses to achieve target AUC>400

15. Simulated Steady State AUC
Age Category
BNF C
mg/kg
8 hourly
AUC0-24,ss
Model
Dose
6 hourly
Model Dose
Model Dose mg/kg
1 mnth 15
544 10
485
6 mnth
328
437
1 yr
259
346
17.5
404 20
461
3 yr
278
371
433
494
6 yr
297
397
463
529
12 yr
329
439
16 yr
353
470

16. Simulated steady state trough levels
Age Category
BNF C
mg/kg
8 hourly
Troughss
Model
Dose
6 hourly
Model Dose
Model Dose mg/kg
1 mnth 15
15.1 10
15.2
6 mnth
10.4
1 yr
8.6
12.3
17.5
14.3 20
16.38
3 yr
9.8
13.8
16.1
18.36
6 yr
14.6
17.1
19.50
12 yr
10.7
15.4
16 yr
11.1
16.5

17. Conclusions (1)
Robust 1-compartment model developed from prospectively collected TDM data
Clearance influenced by actual body weight and creatinine clearance
Volume influenced by age
Current BNF-C dose (15mg/kg 8 hourly) does not achieve target AUC>400 in the typical patient (except infants 1-6 months)

18. Conclusions(2) Infant 1 month: 10mg/kg 6 hourly
To achieve target AUC>400, model optimised doses proposed:
Infant 1 month: 10mg/kg 6 hourly
Infant 6 month: 15mg/kg 6 hourly
Child 1 – 6 year: 17.5mg/kg 6 hourly
Child year: 15mg/kg 6 hourly
These doses equate to trough levels mg/L

19. Some caveats…..
Modelling excluded patients with severe renal failure and renal/extracorporeal support
Proposed doses are for the ‘typical’ patient in the population and:
are initial starting doses
subsequent doses should be ‘individualised’ using TDM
Proposed doses assume ‘normal’ creatinine clearance for age
doses will need to be adjusted with changes in renal function
Proposed doses need to be evaluated in a prospective study

20. Thank you to all those collecting data
All my pharmacy colleagues at Sheffield CH
Nigel Gooding Addenbrookes, Cambridge
Caroline Cole University Hospital, Southampton
Tricia Cummings Our Lady’s Children’s Hospital, Dublin
Abimbola Olusoga, LGI,Leeds
Jenny Haylor, Bristol Children’s Hospital
David Sharpe, Alder Hey Children’s Hospital Liverpool

21. Project Team Clare Nash: Sheffield Children’s Hospital
Nigel Gooding: Lead Pharmacist Paediatrics, Addenbrooks
Hussain Mulla: Senior Research Pharmacist – Paediatric Clinical Pharmacology, University Hospitals of Leicester (Pharmacokineticist)
Bradley Manktelow: Senior Research Fellow, Department of Health Sciences, University of Leicester (Statistician)

22. Funded by the Neonatal and Paediatric Pharmacists Group