Gram negative bacteraemia

Gram negative bacteraemia David Enoch Consultant Medical Microbiologist PHE Cambridge Cambridge University Hospitals NHS FT

Overview Gram negative bacteraemia Changing epidemiology in England Epidemiology in Peterborough E. coli bacteraemia as a surveillance tool Explore some of the reasons for the changes in epidemiology

Gram negative bacteraemia Wilson et al (2011) Clin Microbiol Infect 17: 451.

Gram negative bacteraemia Wilson et al (2011) Clin Microbiol Infect 17: 451.

Gram negative bacteraemia 2004 2008 % change p-value E. coli 12579 16663 +32.5 <0.001 CNS 9086 10165 +11.9 0.003 S. aureus 11370 8660-23.8 <0.001 Enterococcus 4980 4531-9.0 <0.001 Klebsiella spp. 3739 4274 +14.3 0.004 Pneumococcus 3616 3415-5.6 <0.001 Pseudomonas spp. 2290 2829 +23.5 <0.001 Proteeae 1859 2101 +13.0 0.095 Enterobacter spp. 1812 1653-8.8 <0.001 Group B streptococcus 879 1135 +29.1 <0.001

Gram negative bacteraemia E. coli There have been a year-on-year increases since 2008 Now 32% of all bacteraemias (25% in 2008) All Gram negative bacteraemias Significant morbidity and mortality (Peterborough data) 24 patients (12%) died within 7 days 45 (22%) died within 30 days PHE data: http://www.hpa.org.uk/webc/hpawebfile/hpaweb_c/1317138991349 Enoch DA et al (2011) Quarterly Journal of Medicine 104: 411 19.

2008 2012: 26% rise 2011 2012: 1.6% rise PHE data: http://www.hpa.org.uk/webc/hpawebfile/hpaweb_c/1317138991349

Susceptibility data

Peterborough data; poster data (224) Four fold increase in Emergency Department Gram negative bacteraemias from 2008 2012 Table 2: Demographic, resistance profiles and outcome data for pre and post intervention time periods Co-morbidities July 2008 June 2009: 45 in one year Charlson co-morbidity score Range = 0 11 Median = 3 Mean = 3 Outcome 2008/9 Dead: at discharge 6 (13.3%) at 1 month 7 (15.5%) at 1 week 2 (4.4%) Nov / Dec 2012 Jan 2013: 45 in 3 months Charlson co-morbidity score Range = 0 11 Median = 5 Mean = 4.89 2012/3 Dead: at discharge 10 (22.2%) at 1 month 10 (22.2%) at 1 week 8 (17.8%)

Something needs to be done

E. coli bacteraemia mandatory surveillance Mandatory surveillance introduced for MRSA bacteraemia C. difficile infections Nothing (positive) happened until targets were introduced

E. coli bacteraemia mandatory surveillance

E. coli mandatory surveillance

Is E. coli bacteraemia the best surveillance marker? Prospective study: ALL Gram negative bacteraemias January 1 st to June 30 th 2011 Preventability scale (1 6) Preventable certainly probably possibly Not preventable possibly probably certainly 1. One microbiologist discussed with ward (medical and nursing) staff 2. discussed with the other microbiologist 3. 20 medical notes were randomly selected and retrieved six months later and reviewed independently and blind of the initial result Bonnal C et al (2010) BMJ Quality & Safety 19: e30.

Peterborough Beautiful new District General Hospital Approximately 600 beds 75,000 admissions per year All Gram negative bacteraemias Hospital (>48h) HCAI Community 28 day rule re: recurrence

Case 1 33M Diabetic (type I); hypoglycaemic episode ITU Intubated, CVC, PVC, urinary catheter Slow, partial recovery but discharged from ITU to the rehabilitation ward Difficult patient on the ward PEG Urinary catheter

Case 1 Becomes more agitated Temperature of 38.5 C, pulse 100, BP 120/80mmHg Suprapubic tenderness Blood cultures and CSU taken Given co-amoxiclav and gentamicin and urinary catheter removed (it turns out it wasn t needed)

Case 1 Blood cultures and CSU grow Gram negative rods Identified as Proteus mirabilis the following day Changed to amoxicillin to complete 10 days Discharged 63 days later

Question 1 Is this case: Preventable certainly probably possibly Not preventable possibly probably certainly

Case 2 66M Diabetes, IHD, CCF, HT Enlarged prostate and raised PSA Admitted for elective prostate biopsy 3 days prior to admission Admitted with temperature of 40 C, BP 90/40mmHg Blood cultures and urine sent Commenced on co-amoxiclav & gentamicin Requires ITU for 4 days, including inotropes

Case 2 Blood cultures grow E. coli 4 days of co-amoxiclav and then 10 days of ciprofloxacin Goes home well

Question 2 Is this case: Preventable certainly probably possibly Not preventable possibly probably certainly

Peterborough data Prospective data (6 months) 141 bacteraemias 122 episodes (28 day period) 118 patients (4 recurrences) Age 0-98 (median 75; IQR 59-82) 54 (45.8%) female

Specialty distribution Medicine 80 Surgery 20 Cancer 14 Paediatrics 6 Gynaecology 1 Emergency 1

Onset HCAI 63 Community 35 Hospital 24

Source Urine 61 Biliary tract 19 Pneumonia 12 Bowel 6 Line 4 Abscess 2 SSTI 1 Pancreatitis 1 Unknown 15

Peterborough data Underlying conditions Number (%) Non-fatal condition 44 (36.1) Ultimately fatal condition 65 (53.3) Rapidly fatal condition 13 (10.7) Need for ITU 12 (9.8) Need for surgery 11 (9) Clinical condition Number (%) Temperature 38 C 86 (73.5) Blood pressure <100mmHg systolic 53 (45.3)

Bacteriology 80 60 40 20 0 E. coli 75 Klebsiella spp 21 Proteus spp 9 P. aeruginosa 8 Serratia spp 4 Enterobacter spp 3 Citrobacter spp 2 S. maltophilia 2 Salmonella spp 1

Peterborough data Antibiotic Resistance rate (%) Amoxicillin 82 (67.2) Co-amoxiclav 39 (32) Ciprofloxacin 13 (10.7) Cefuroxime 22 (18) Gentamicin 6 (4.9) Piperacillin-tazobactam 4 (3.3) Meropenem 2 (1.6) ESBL 1 (0.8) S. maltophilia x2, K. pneumoniae x2 S. maltophilia x2

Outcome 16 patients (13.1%) died within 7 days 25 (20.5%) died within 30 days

Peterborough data Definitely preventable (1) 0 Probably preventable (2) 3 2 urinary catheter-related (H) E. coli, P. mirabilis 1 diabetic ulcer (HCAI) P. mirabilis

Peterborough data 21 (17.2%) were thought to be possibly preventable (Lickert scale 3) Klebsiella spp. 8 E. coli 6 Proteus spp. 5 P. aeruginosa 2 One each of Providencia spp., S. marcescens and C. koseri Four of the 21 episodes were hospital onset Urinary catheter use 3 CVC 1 Fifteen were HCAI-onset Urinary catheter use 10 CVC 3 Cholangitis 1 Unknown 1

Peterborough data 98 episodes (80.3%) were considered to be not-preventable (Lickert scale 4 6) E. coli 68 Klebsiella spp. 12 Pseudomonas aeruginosa 6 Serratia spp., 3 Enterobacter spp. 3 Possibly not preventable 17 Hospital 7 HCAI 9 Community 1 Probably not preventable 56 Definitely not preventable 25

Preventability by organism Preventable (1 3) N = 24 Not preventable (4 6) N = 98 Univariate analysis P value E. coli 7 68 <0.0001 Klebsiella spp. 6 12 NS Proteus spp. 7 2 <0.0001 P. aeruginosa 2 6 NS

Preventability by source Preventable (1 3) N = 24 Not preventable (4 6) N = 98 Univariate analysis P value Urine 17 44 0.023 Biliary tract 1 18 0.071 Bowel 0 6 0.260 Pneumonia 0 12 0.063 NS for: unknown, abscess, SSTI, pancreatitis

Preventability by risk factors Preventable (1 3) N = 24 Not preventable (4 6) N = 98 Univariate analysis P value Urinary catheter 17 19 <0.0001 District nurse 9 13 0.006 Dependent functional state 12 20 0.003 CVC 5 4 0.014 Chronic wound / ulcer 4 1 0.005 Neurological disease 8 11 0.007

Multivariate analysis Adjusted odds ratio P value Urinary catheter 23.85 0.009 District nurse 1.84 NS Dependent functional state 112.85 0.03 CVC 142.27 0.01 Chronic wound / ulcer 5.47 NS Neurological disease 1.66 NS

Preventability by organism E. coli Preventable Not preventable Total Sensitivity % Specificity % PPV % NPV % E. coli 7 68 75 29.2 30.6 9.3 63.8 Non-E. coli 17 30 47 Total 24 98 122 P. aeruginosa Preventable Not preventable Total Sensitivity % Specificity % PPV % NPV % P. aeruginosa 2 6 8 8.3 93.9 25 80.7 Non-P. aeruginosa 22 92 114 Total 24 98 122

Preventability by onset Hospital cases Preventable Not preventable Total Sensitivity % Specificity % Hospital 6 18 24 25 71.4 25 71.4 Non-hospital 18 70 98 Total 24 98 122 PPV % NPV % Hospital and HCAI Preventable Not preventable Total Sensitivity % Specificity % PPV % NPV % Hospital & HCAI 21 66 87 87.5 32.7 24.1 91.4 Community 3 32 35 Total 24 98 122

Preventability by risk factor Urinary catheter Urinary catheter No urinary catheter Preventable Not preventable Total Sensitivity % Specificity % PPV % NPV % 17 19 36 70.8 80.6 47.2 88.8 7 79 86 Total 24 98 122

Gram negative bacteraemia Common High mortality rate Difficult to treat 24 (19.7%) thought to be preventable Particular concern around urinary catheters Enoch DA et al., (2013) Gram negative bacteraemia; how many are preventable and what will surveillance of E. coli add? J Infect Prevent. 14: 54-9.

E. coli bacteraemia mandatory surveillance Surveillance (and targets) has its role Dramatic reductions in MRSA bacteraemia However, with regard to E. coli bacteraemia: Insensitive: missed 2 of 3 cases probably preventable missed 15 of 21 cases possibly preventable Non-specific: 58 cases were not preventable Can t concentrate just on hospital cases Insensitive: missed 6 of 24 preventable cases Hospital and HCAI cases: 87 of 122 cases Should we not just target urinary catheter usage? Would cover 17 of 24 cases Reasonably sensitive Relatively low number of cases to review

Limitations Single centre Subjectivity / bias One microbiologist Previous studies suggest this method is moderately reproducible Tried to standardise / control for this Not a formal root cause analysis (RCA) Incomplete information / documentation, particularly with community and HCAI cases Time consuming 10 15 minutes per episode for data collection Longer to discuss with the clinicians

Other studies Underwood J, et al., (2011) J Hosp Infect. 79: 364 5. E. coli bacteraemia 19% were preventable Most were due to urinary catheter use

Other studies Melzer M & Welch C (2012) Lancet 12: 103 1372 E. coli bacteraemias (Sept 2007 November 2010) 20% of CA-UTI are preventable then 38 of 1372 (2.8%) infections may be preventable Plus CVC-associated bacteraemias (65 of 1372; 4.7%)

The problem All homes said the majority of their catheterised clients came from hospital: the majority of people we send to hospital come back with a catheter sometimes you can understand the reasoning if they ve had a fractured hip and they ve got a wound and they re very incontinent I have actually asked the wards why is this person catheterised? Because they re incontinent. Which I don t think is a good enough reason. "I always get the impression that it s easy to put the catheter in, because then they don t have the toilet rounds like we do sadly, do they? and it s very different I suppose where they re so short-staffed at the hospital, they re using them because its more convenient. usually what happens: they catheterise in hospital.. they come back to us and normally the catheters taken out unless there s a reason. McNulty et al (2006)

Other reasons for the rise in E. coli / Gram negative bacteraemia Increased awareness Sepsis care bundle More blood cultures taken (Peterborough) Increase in resistance Seasonality

Increase in resistance Oxford group Increase also seen in urinary E. coli isolates Schlackow et al., 2012. J Antimicrob Chemother; 67: 1514 1524

Oxford group An increase in mean weekly air temperature of 5 C was associated with a 5% increase in E. coli incidence. The effect size was similar in both susceptible and resistant subgroups.

Seasonality Wilson J et al (2011) Clin Microbiol Infect. 17: 451-8.

Seasonality E. coli Al-Hasan et al 2009 7% increase in monthly incidence per 5 C rise Schlackow et al 2012 5% increase per 5 C rise Eber et al 2011 12% fluctuation between summer and winter Acinetobacter spp. Eber et al 2011 51.8% fluctuation Pseudomonas spp. Eber et al 2011 28.1% fluctuation Klebsiella pneumoniae Anderson et al 2008 41% fluctuation Eber et al 2011 28.6% fluctuation

Potential reasons for seasonality Al-Hasan et al (2009) Most E. coli bacteraemias were UTI (80%) Most were community onset (59%) Primarily UTI-related Human behaviour-related? Mild dehydration UTI (Beetz, 2003) Changes in diet Travel Recreational water exposure Diet / food preparation practices Sexual activity Bacterial factors Environmental changes

Summary Gram negative bacteraemias are increasing in frequency Associated with significant mortality Surveillance is useful targets for E. coli aren t Urinary catheters are associated with significant mortality and morbidity Risk factor for ESBL They re not the easy option Use alternatives when possible We need a combined hospital and community approach CVCs remain a risk factor for Gram negative bacteraemia Target antimicrobial resistance Further work is required to look into the effect of seasonality and other risk factors such as dehydration

Acknowledgements Peterborough Dennis Mlangeni, Andrew Sismey, Emily Jarvis, Gill Clark Ward staff at PCH Cambridge Sani Aliyu, Andreas Karas Nashville (statistics) James Ekundayo, Muktar Aliyu