Early detection of critical illness

A session blog from Day 1 of State of the Art, London December 2015. See the full list

Blogger Jack Wong

Chair: Gary Masterson

How many are we missing? Lessons from the All-Wales size of sepsis study

Tamas Szakmany, Aneurin Bevan University Health Board Cardiff University

Summary: How many are we missing? Loads.

Background

  • 2015 feasibility pilot in 4 hospital – Sepsis screening and delivery was lacking
  • You can’t drive improvement if you don’t know the size of the problem!

2015 study

  • CURES NHS Cardiff University
  • 1 day point prevalence study for 24 hours
  • Changed the data collection to electronic. Open source revolution

Methods

  • Medical student as data collector
  • Tablets for screening tool, guidelines, educational resources
  • Online and face to face training of data collectors
  • Real time monitoring of device and troubleshooting via Whatsapp groups
  • Standardised data, almost immediately available for analysis using E-CRF
  • 14 out of 16 hosptal
  • Adult patients with NEWS > 3 in A&E and general wards
  • 184 data collectors with 56 devices

Results

  • 24 hours – 1198 data collection forms
  • Focus on severe sepsis patient group – 101 patients on the wards
  • Only 42 severely septic patient trigger NEWS out of 101

Summary

  • 212 patients with sepsis (4%)
  • 101 severe sepsis
  • >50% have NEWS <6
  • Less than 20% of sepsis patients were screend
  • Sepsis Six completion was 15%

Discussion

  • Change of sepsis definition is coming
  • Life threatening organ dysfunction due to a dysregulated host response to infection
  • Future work:
    • To test if the new criteria is applicable to the wards
    • To see if any change in incidence in a years time
    • To see if detection is better and Sepsis 6 is more universally applied
    • We will give you an independent reality check

Take home message

  • 1 in 25 patients on the ward can have sepsis
  • ¼ unwell have sepsis
  • Only 20% got picked up
  • If you don’t spot it, you can’t treat it
  • Be prepare to find out something you might not like

Screen Shot 2015-12-08 at 22.55.52

Early warning and decision-support in Birmingham

Nandan Gautam, Queen Elizabeth Birmingham

Summary: To BEWS or not to BEWS?

Key messages

A solution for previous topic?

  • NICE July 2007 – Acutely ill patient – Physiological track and trigger system is required
  • Should include at least: HR, RR, SBP, GCS, Sats, T with graded response strategy
  • Problem with prediction scores – If you are well, it is not very discriminatory, if you are very unwell, you don’t need it!
  • Ideal score should be valid, objective, reproducible, low inter-rater variability, low intra-rater variability
  • Scores are just a crude descriptor of current physiological state, doesn’t predict the following – Mortality, Morbidity, LOS, identify cardiac arrest, anticipated organ failure.
  • Does MEWS mean anything at all? Does it only trigger but does it track?

 

There are many scoring systems

  • Over 30 scores identified, subtle variation in weighting, different thresholds, different settings, questionable validation, impossible comparisons
  • NEWS – good at predicting death within 24 hours and up to 48 hours
  • Doesn’t predict morbidity / critical illness
  • Good at predicting cardiac arrest – NEWS AUROC 0.86
  • Cardiac arrest – Reduction of calls
  • LOS – No difference
  • Need for organ support – No different in using critical care services / reduced length of stay

 

What are they good for?

  • Makes organisations think about safer system,
  • Empowers staff to call for assistance
  • Set a threshold at which an intervention is considered
  • Makes you think about futility

 

QIP Project in Queen Elizabeth Hospital

Objective

  • Review use of SEWS score
  • Improve care and timeliness of intervention
  • Improve specificity
  • Resource allocation

Demographic

  • 100k pt per year
  • 30k emergency
  • 1100 beds
  • 65 level 3 critical care
  • 24×7 outreach
  • 3 cardiac arrest zones

Statistic

  • 500-650 cardiac arrest calls
  • <200 true loss of cardiac output
  • 700 inpatient emergency admissions to critical care.
  • SEWS = 50% specificity, not valid for patient group, no reproducible

 

A bespoke score

    • Unique but comparable, dynamic, learning, relevant to the resources available locally, reflects changing health care

 

  • Birmingham Predictive Scores (BEWS / BIPS)

 

    • Electronic observation charts
    • Labs database that can be interrogated
    • Demographics database
    • Recorded outcomes
    • Large patient records
    • Emergency and elective patient cohort

 

Outcome and comparison

It is a better predictor – So what?

  • Linear so it means something
  • Dynamic and responsive
  • Better pick up rate (vs NEWS)
  • Resource allocation – No increased resources but BIPS showed better pick up before patient deteriorate hence resources can be allocated more meaningfully.

Predicting poor outcome

  • Change therapy or recognise futility.

The truth about lactate

Rinaldo Bellomo, ANZIC RC Melbourne, Australia

Summary: Lactate = stress ≠ tissue hypoxia

Key messages

  • Hyperlactataemia predicts death.
  • Dynamic lactate indices as predictor of outcome in critically ill patient.
  • Lactate level predicts mortality as compared to glucose
  • At ED presentation, raised lactate signifies increased risk. – ARISE trial
  • Bigger predictor than refractory hypotension – but why?
  • Belief = lactate is biomarker of tissue hypoxia and anaerobic glycolysis.

The golden rules of biomarkers

  • The biomarker link to the condition must be biologically plausible and independent
  • Must be tested in the population of interest
  • Accuracy of the biomarker must be assessed with AUROC eg. BNP, NGAL

Is lactate plausible?

  • Lactate level changes with training – but tissue hypoxia is tissue hypoxia, how can training change it?
  • Relatively unchanged lactate level with low PaO2 level in Mount Everest climber.
  • Lactate shuttle between organs and cells as shown in magnetic resonance spectroscopy data in animal model and laser microscopy.
  • Lactate as a hormone to mitochondria and nucleus
  • Renal lactate uptake in endotoxic shock
  • Lactate is released in lung in relation to CO2 production

Clinical plausibility

  • Lactate increase in adrenergic state, adrenaline infusion, liver failure, hyperdynamic state WITH NO EVIDENCE OF CO-EXISTING TISSUE HYPOXIA.

AUCROC for tissue hypoxia / anaerobic glycolysis

  • No such analysis ever done
  • The reason is that no one can confirm or refute tissue hypoxia – of what? the whole body? the organ? specific cells?

Conclusion

  • Lactate is not an accurate or reliable or robust marker of hypoxia
  • Its link with hypoxia is biologically flawed
  • More likely to be marker of physiological stress
  • It is now clear that lactate is a major mitochondrial fuel
  • Rapidly utilized in cell to cell and intracell shuttles
  • Taken up by mitochondria to optimize bioenergetics
  • It acts like a hormone with powerful effects on protein synthesis
  • It is associated with increased mortality, so is fever and hypotension
  • Our scientific journey toward real understanding of lactate has only just begun.