Will 10 Million People Die a Year due to Antimicrobial Resistance by 2050? Prof. Stephan Harbarth Infection Control Program Geneva, Switzerland

Transcription

1 Will 10 Million People Die a Year due to Antimicrobial Resistance by 2050? Prof. Stephan Harbarth Infection Control Program Geneva, Switzerland Thanks for material provided by Marlieke de Kraker & Andrew Stewardson!

2 Antibiotic resistance: global public health concern Tackling drug-resistant infections globally (O Neill report) - Mai 2016

3

4 What are clinical implications of antimicrobial resistance? Treatment failure due to wrong choice Increased morbidity and mortality Use of more toxic, more expensive and less efficacious therapeutic alternatives Added burden of nosocomial infections Risk of explosive outbreaks

5 Main Conclusions: - Antibiotic resistance significantly impacts on illness burden in the community. - Patients with laboratory-confirmed antibiotic-resistant urinary and respiratory-tract infections are more likely to experience delays in clinical recovery after treatment with antibiotics.

6 Deadly MRSA Infection C. H. (71) first woman elected lieutenant governor in South Dakota. She had suffered a spinal fracture and 3 broken ribs Oct. 8 while sailing the Adriatic Sea. She underwent surgery in Zagreb, Croatia on Oct. 10, then was hospitalized Oct. 19 during a stop in Switzerland on her way back to the US. She suffered pneumonia, a bacterial blood infection, and a series of strokes, which claimed her life in Lausanne, Switzerland on October 25, 2007.

7 Acinetobacter Outbreak, Lausanne Index patient Severe burn injuries, transfer from Bali (Oct 2002) Multi-R Acinetobacter at admission Outbreak Spread to 2 patients 6 months later: 6 new cases Closure of the burn unit Environnement Widespread contamination: 16/161 (10%) positive swabs Patients Environnement Environmental cleaning & disinfection Complete replacement of all disposable material Zanetti G et al. Infect Control Hosp Epidemiol 2007; 28:

8

9 Economic burden of MDROs Increased direct costs of providing care to MDRO-infected patients; Indirect costs to patients, caregivers, & diminished quality of life; Infrastructure and productivity costs of surveillance, screening and isolation; Antibiotic treatment costs for therapy or empiric coverage of MDRO

10

11

12 Projected impact of antimicrobialresistant neonatal sepsis in India

13 Apocalypse soon?

14

15 Add Raoult, CID headline AMR burden

16

17 Public health burden of drug resistance C.E. Phelps, Med Care 1989; 27: The estimates of the burden caused by bacterial resistance depend heavily on unknown parameters. US General Accounting Office, 1999 Report to U.S. Congress Data are insufficient to determine full extent of public health burden associated with antibacterial resistance.

18 Affected people / deaths R. Rappuoli. Nat Med From Pasteur to genomics: progress and challenges in infectious diseases

19 Threat level: URGENT

20 Threat level: SERIOUS

21

22 Tackling drug-resistant infections globally (O Neill report) - Mai 2016

23

24 Usability/Business case, O Neill report General objective for the AMR review group: Defining the steps needed to avoid the AMR crisis Objective of this report: Determine the health and macro-economic consequences for the world, especially in emerging economies if antimicrobial resistance is not tackled Tackling drug-resistant infections globally (O Neill report) - Mai 2016

25 De Kraker N, Stewardson A, Harbarth S. PLoS Med 2016; 13: e

26 Caveats Accurate, with and without random error Internal validity? External validity? Assumptions? Peer review? In-accurate, with and without random error De Kraker N, Stewardson A, Harbarth S. PLoS Med 2016; 13: e

27 Estimating the Burden of Disease (BoD) Related to Antimicrobial Resistance Number of infections (I) Resistance proportions (R) Burden measure: Attributable mortality proportion (M) BoD= I*R*M Future scenarios Determine coefficient for change (c) Future BoD = I*c1 * R*c2 *M*c3 De Kraker N, Stewardson A, Harbarth S. PLoS Med 2016; 13: e

28 Number of infections (BoD= I*R*M) EARS-net data: Representativeness? Mainly tertiary care hospitals Few community/paediatric/ltcf isolates ECDC, EARS-Net Antimicrobial resistance surveillance report 2013/2014

29 Number of sets/1,000 pds Resistance proportions (BoD= I*R*M) EARS-net and WHO data: Representativeness? Highly variable blood culture rates ECDC, EARS-Net Antimicrobial resistance surveillance report 2012/2011*/2009**/2008***/2006

30 Extrapolation from bloodstream infections to infections at other sites Fifteen Brooklyn hospitals ESBL+ K. pneumoniae isolates 12 BSIs, 4 SSIs, 14 UTIs, 14 LRTIs: Infect Control Hosp Epidemiol 2002; 23: : 0.33 : 1.2 : bed hospital in Spain MRSA isolates 4 BSIs, 17 SSIs, 3 UTIs, 5 LRTIs: 1 : 4.3 : 0.8 : 1.3 Infect Control Hosp Epidemiol 2006; 27:

31 Attributable mortality (BoD= I*R*M) ECDC, The bacterial challenge: Time to react 2009

32 Methodological challenges -- Why is it so difficult to estimate the attributable mortality of AMRrelated infection?

33 Problem 1: - Severity of illness and underlying disease

34 PROBLEM High crude mortality in patients with infections caused by multidrug-resistant bacteria Carriers of multiresistant bacteria who die in the hospital may die either with simple asymptomatic carriage of resistant bacteria with infection by resistant bacteria or because of infection (primary cause of death)

35 Problem 2: - Appropriateness of antimicrobial therapy

36 Causal pathways & challenges Severity of underlying illness Resistant infection Death Appropriateness of antibiotics

37 Mortality prediction in nosocomial bacteremia Severity of underlying illness: RR = Bacteremia Death Inappropriate antibiotic therapy RR = McCabe & Jackson. Arch Intern Med. 1962;110: Freeman J et al. Rev Infect Dis 1988; 10:

38 Problem 3: - Timing of events and time-varying exposures

39 The importance of correct measurement When did the antibiotic-resistant infection occur? Non-infected Discharge Infected Admission infection Death

40 Cohort study, countries, 537 ICUs, pts Sophisticated statistical analyses adjusted for the timing of events and competing outcomes (multistate modelling) Lambert et al. Lancet Infect Dis 2011

41 Main findings High excess mortality associated with bacteremia and pneumonia acquired in the intensive care unit Substantially increased excess length of stay for pneumonia, but not for bloodstream infection Pseudomonas aeruginosa: greatest burden (not MRSA) AMR: only a small contribution to the overall burden of ICUacquired infections Lambert et al. Lancet Infect Dis 2011

42 Multicenter study (TIMBER) Population Patients with bloodstream infection (BSI) caused by S. aureus or Enterobacteriaceae Main exposure of interest Methicillin resistance or third-generation cephalosporin resistance Main comparison group Patients with infections by susceptible strains Main outcomes Excess length of stay (LoS) and in-hospital mortality = Extended-spectrum betalactamase-producing Enterobacteriaceae (ESBL-E) (e.g. E.coli, Klebsiella spp) Stewardson A,., Harbarth S. EuroSurveillance 2016; 21: 33

43 Methods Design: Multicentre retrospective cohort study 10 European hospitals Population: All acute inpatient admissions January 2010 December 2011 Data collection: Demographic, clinical, microbiologic & administrative data were extracted electronically One investigator from each site trained in standardized data collection during a workshop Stewardson A,., Harbarth S. EuroSurveillance 2016; 21: 33

44 Statistical methods Cox proportional hazards analysis Compute hazards of inpatient mortality Multivariable models Baseline covariates: for age, sex, elective versus emergent admission, previous hospitalisation, 17 comorbidities Time-varying covariates: bloodstream infection, ICU admission or surgery Multistate modeling Compute excess hospital LoS (days) attributable to each type of BSI Accounting for competing risks (discharge vs death) Stewardson A,., Harbarth S. EuroSurveillance 2016; 21: 33

45 Outcomes (unadjusted) S. aureus analysis Group N Incidence proportion (events/100 admissions) Total length of stay Median (IQR) Mortality Count (%) MRSA BSI (16 45) 36 (22.1%) MSSA BSI (13 39) 149 (16.8%) Non-infected (2 7) (1.7%) Enterobacteriaceae analysis Group n Incidence proportion (events/100 admissions) Total length of stay Median (IQR) Mortality Count (%) 3GCR-E BSI ( ) 58 (16.1%) 3GCS-E BSI (7 28) 212 (10.1%) Non-infected (2 7) (1.7%) 62 Stewardson A,., Harbarth S. EuroSurveillance 2016; 21: 33

46 In-hospital mortality Adjusted proportional hazards analysis Interpretation: Risk of death after bloodstream infection (BSI) Adjusted for age, sex, emergent/elective admission, comorbidities, nights hospitalised in previous 12 months, plus ICU-admission and surgical procedures as time-dependent covariates 63 Stewardson A,., Harbarth S. EuroSurveillance 2016; 21: 33

47 Excess length-of-stay Multistate model 64 Stewardson A,., Harbarth S. EuroSurveillance 2016; 21: 33

48

49 Assumptions for future scenarios KPMG report, the global economic impact of anti-microbial resistance 2014

50 Number of death certificates in England & Wales mentioning MRSA Office for National Statistics online 2010

51 The way forward Comprehensive, population-based antimicrobial resistance surveillance Paediatric & geriatric infections Community-acquired infections Low-, middle-, and high-income countries Different types of infections Morbidity and mortality data Accurate & valid analyses

52

53 Summary Preventing antimicrobial resistance is desirable by patients and society Consistency of data regarding the impact of antimicrobial resistance on clinical outcomes and risk of treatment failure However: Paucity of data regarding the overall impact of antimicrobial resistance on health services and societal burden, especially in LMIC Due to methodological limitations, we may have overestimated the attributable mortality and excess costs of antimicrobial resistance 83

54 Situation in people dying due to antibiotic resistance? Methodological challenges & flaws of these projections: To predict total number of infections To predict the proportion of resistance To predict the attributable mortality Lack of robust data note of caution: broad brush estimates, not certain forecasts De Kraker N, Stewardson A, Harbarth S. PLoS Med 2016; 13: e

55 What will happen in 2050? The first Rhino will be born on the North pole Geneva will be renamed Genève-sur-Mer Ivanka Trump, the 1 st female US president, will be re-elected for the 4 th time in a row people will die from AMR Courtesy: Marc Bonten (Utrecht)

56 Thanks for your attention!

57 Basic Copyright Notice & Disclaimer 2018 This presentation is copyright protected. All rights reserved. You may download or print out a hard copy for your private or internal use. You are not permitted to create any modifications or derivatives of this presentation without the prior written permission of the copyright owner. This presentation is for information purposes only and contains non-binding indications. Any opinions or views expressed are of the author and do not necessarily represent those of Swiss Re. Swiss Re makes no warranties or representations as to the accuracy, comprehensiveness, timeliness or suitability of this presentation for a particular purpose. Anyone shall at its own risk interpret and employ this presentation without relying on it in isolation. In no event will Swiss Re be liable for any loss or damages of any kind, including any direct, indirect or consequential damages, arising out of or in connection with the use of this presentation.