1. Stabilization in prevalence of ESBLs & ciprofloxacin non-susceptibility in Enterobacteriaceae from blood in the UK and Ireland
Rosy Reynolds, Russell Hope on behalf of the BSAC Working Party on Resistance Surveillance
The results in this presentation are mainly from the BSAC Bacteraemia Resistance Surveillance Programme...
O135, 20th ECCMID, Vienna, April 2010

2. BSAC Bacteraemia Resistance Surveillance
2008
25 laboratories / year
20 E. coli (10 up to 2007), 10 Klebsiella, Enterobacter, Proteeae, Others
per laboratory per year
Central testing -
HPA Centre for Infections, London.
BSAC agar dilution MICs
ESBL by clavulanate synergy; PCR for CTX-M
Data on website:
...which has been running continuously in the UK and Ireland since 2001.
BSAC=British Society for Antimicrobial Chemotherapy.

3. HPA Centre for Infections, Colindale Collecting Laboratories
ACKNOWLEDGEMENTS
British Society for Antimicrobial Chemotherapy (BSAC)
Bacteraemia Resistance Surveillance Programme
HPA Centre for Infections, Colindale
Russell Hope
David Livermore
Marina Warner
Rachael Adkin
Aiysha Chaudhry
Dorothy James
Geraldine Brick
Melissa Coleman
Nicola Potz
Caroline Henwood
Sponsors
Astellas
AstraZeneca
Cubist
Johnson & Johnson
Merck / MSD
Novartis (Chiron)
Pfizer (Wyeth)
Theravance
Wyeth
Collecting Laboratories

4. Distribution of genera in Enterobacteriaceae
All bacteraemia
Hospital bacteraemia ( 48 hours)
The focus is on E. coli, Klebsiella and Enterobacter as the big 3 - between them they account for 85% of bacteraemias involving Enterobacteriaceae. (Enterobacter has similar numbers to Proteeae, but is much more troublesome.)
(E. coli is dominant, even in hospital-acquired bacteraemia, and overwhelmingly dominant at 73% in community bacteraemia)
The distribution estimated for bacteraemia is based on the numbers of isolates reported through LabBase in 2008, and the proportion of those isolates that are H48 (calculated from BSAC bacteraemia , but for Citrobacter and non-Proteeae ‘others’).
Estimated from:
BSAC Bacteraemia ; HPA voluntary bacteraemia surveillance 2008

5. % non-susceptibility and ESBLs
Minor species
% non-susceptibility and ESBLs N
ESBL %
ciprofloxacin
non-susceptible %
Proteeae ( )
1685
0.2 10
Serratia ( )
637 29
Citrobacter ( )
198 2
others ( ) 56 25
‘Others’ are of 8 genera, but predominantly Raoultella (24) and Pantoea (20).
The others’ ESBLs are almost all from Raoultella (11/24=46%), with just 2 (of 20) Pantoea and 1 (of 2) Kluyvera.
No obvious time trend in cipNS in Serratia and no point looking for trends in other combinations - prevalence and/or numbers are just too low.
These data are combined for all available years, as there are no obvious trends.
BSAC Bacteraemia

6. % non-susceptibility and ESBLs
Enterobacter
% non-susceptibility and ESBLs
Good agreement between BSAC and HPA voluntary.
CTX-NS is not a good marker of ESBL in Enterobacter - it is higher than ESBL because of AmpC hyperproducers.
Little clear change over time.
BSAC Bacteraemia ; HPA voluntary bacteraemia surveillance 2008

7. % non-susceptibility and ESBLs
Klebsiella
% non-susceptibility and ESBLs
Good agreement between BSAC and HPA voluntary.
CTX-NS in Klebsiella is mostly caused by ESBL, so the CTX-NS and ESBL lines are closely parallel.
CIP-NS is associated with ESBL because the ESBL-producers tend to be multiresistant.
BSAC Bacteraemia ; HPA voluntary bacteraemia surveillance 2008

8. segmented logistic regression
Klebsiella
segmented logistic regression
Q1 2004
p <0.0005
OR = 1.60
p = 0.33
OR = 0.94
There are more points in the regression than on the graph because we have date of isolate collection, not just year. It would not be reasonable to fit a model like this with only 8 data points!
We assumed two time periods, and looked for the changepoint (knot) that gives the best-fitting model. It was in early 2004.
There is a significant rise in ESBL prevalence in the first period and a statistically insignificant fall in the second period.
BSAC Bacteraemia

9. % non-susceptibility and ESBLs
E. coli
% non-susceptibility and ESBLs
Good agreement between BSAC and HPA voluntary.
CTX-NS and ESBL coincide because ESBL production is the cause of CTX-NS in E. coli. 90% of ESBLs were CTX-M types.
CIP-NS is higher than ESBL because, as well as ESBL-producers commonly being multiresistant, a fair proportion of ESBL non-producers are CIP-NS.
BSAC Bacteraemia ; HPA voluntary bacteraemia surveillance 2008

10. E. coli - ESBL prevalence
segmented logistic regression
Q3 2005
p <0.0005
OR = 1.76
p = 0.15
OR = 0.86
There are more points in the regression than on the graph because we have date of isolate collection, not just year. It would not be reasonable to fit a model like this with only 8 data points!
We assumed two time periods, and looked for the changepoint (knot) that gives the best-fitting model. It was in the second half of 2005
There is a significant rise in ESBL prevalence in the first period. The apparent fall in the second period is statistically insignificant but it is a short period and the addition of data from 2009 will be important to clarify the picture.
BSAC Bacteraemia

11. ESBLs in hospital and community E. coli
(<48 vs. ≥ 48 hours in hospital)
Rates of resistance vary between groups of patients.
This graph shows that ESBLs are about twice as prevalent in hospital E. coli (>=48 hours) than ‘community’ E. coli (<48 hours in hospital). ESBL prevalence has increased with similar time-course in both hospital and community-acquired bacteraemia. It will be interesting to see whether this pattern continues, or whether the rates may converge. (2009 data will help.)
Does this support the ‘revolving door’ hypothesis? Is the ‘community’ group simply a diluted version of the ‘hospital’ group?
Enterobacter and Klebsiella also show similar pattern with any increase following similar time-course in both.
Enterobacter x2 as prevalent in hospital as community.
Klebsiella x3 as prevalent in hospital as community.
BSAC Bacteraemia

12. Ciprofloxacin resistance in ESBL-negative E. coli
CIP-NS is NOT all to do with the spread of multi-resistant CTX-M ESBL producers.
This rise in CIP-NS in ESBL non-producers may possibly have lasted slightly longer than the rise in prevalence of ESBL-producers.
2009 data will help to clarify whether the rise in CIP-NS among ESBL non-producers really has flattened out or reversed.
BSAC Bacteraemia

13. % non-susceptibility and ESBLs
Summary
% non-susceptibility and ESBLs N
ESBL %
ciprofloxacin
non-susceptible %
E. coli
467 9 19
Klebsiella
206 12 15
Enterobacter
157 13
Non-susceptibility to imipenem or meropenem or doripenem:
0/467 E. coli 1/205 Klebsiella 3/157 Enterobacter
(all ESBL-negative)
ESBL producers and CIP-NS isolates are still a minority of isolates (and possibly a shrinking minority), but they are still prevalent enough that they must be considered.
ESBL producers are commonly multi-resistant, leaving carbapenems as the treatment of choice. We have not seen carbapenem resistance amongst them.
However, there are carbapenem resistance mechanisms in existence in the UK, unrelated to ESBLs, and still uncommon. There is no obvious reason why these mechanisms should not be carried in ESBL producing isolates. This is an area of anxiety and we will be looking closely at the 2009 results.
BSAC Bacteraemia 2008

14. Conclusions
The prevalence of ESBLs and ciprofloxacin non-susceptibility in Klebsiella and E. coli has stabilised after a sharp rise.
Ciprofloxacin non-susceptibility also increased in ESBL-negative E. coli.
Carbapenem resistance was still rare in 2008.
Resistance varies between groups e.g. patients with hospital- vs. community-acquired bacteraemia.
Continued vigilance and care in the selection of empirical treatment is still needed.