bcc4: delaney on stats and trials "stuff"
DESCRIPTION
Delaney shares insights into the mysterious world of statistics and trials. This 12 minute podcast is particularly useful for Registrars preparing for their exams and was recorded at BCC4. For similar podcasts and audio; head to www.intensivecarenetwork.com and to rego for BCC5 in Cairns, check out www.bedsidecriticalcare.comTRANSCRIPT
STATS AND TRIALS STUFF
Anthony Delaney MBBS MSc FACEM FCICM
Staff Specialist Malcolm Fisher Department of Intensive Care Medicine
Disclaimer
I ain’t a statistician More of an enthusiastic amateur
So…..
Difference between mortality and survival? How to interpret a “negative” trial result?
Mortality or survival?
Mortality: Number of deaths/number at risk at the end of a period
of time 28 day mortality Rate
Survival: Time to event analysis How long it takes for the event to happen If you have survived for x time, what are your chances of
dying in x+1 time Hazard
Population: >18 yo Source of infection Temperature >38.3oC or <35.6oC Heart rate > 90bpm SBP <90 mmHg for 1 hour if adequate fluids and some pressors Urine output <0.5 ml/kg/hr for > 1 hr or PaO2/FiO2 <280 Lactate >2 mmol/L Ventilated Excluded:
pregnant, contra/indication to steroids, advanced cancer, AMI, PE, AIDS,
Intervention: Hydrocortisone mg q6h ivi Fludrocortisone 50mg po daily For 7 days
Comparison: Placebo For 7 days
Outcome: The primary endpoint was the 28-day survival
distribution from randomisation in non-responders to the short corticotropin test
Point one Post-randomisation sub groups are dubious
Is the subgroup variable a characteristic measured at baseline or after randomisation?
“The credibility of subgroup hypotheses based on post-randomisation characteristics is severely compromised, and can be rejected simply on this criterion”
Subdivision of patients in ISIS-2 with respect to birth signs
Gemini and Libra shows an adverse effect on mortality
Results: 300 participants In non-responders
Placebo 73/115 (63%) Steroids 60/114 (53%) Hazard ratio 0.67 95% CI 0.47-0.95; P=0.02
Conclusion: Treatment with hydrocortisone and
fludrocortisone significantly reduced the risk of death in patients with septic shock and adrenal insufficiency
Date of download: 9/11/2013Copyright © 2012 American Medical Association.
All rights reserved.
From: Effect of Treatment With Low Doses of Hydrocortisone and Fludrocortisone on Mortality in Patients With Septic Shock
JAMA. 2002;288(7):862-871. doi:10.1001/jama.288.7.862
Results are according to the response to the short corticotropintest. In nonresponders, the median time to death was 12 days in the placeboand 24 days in the corticosteroid groups; in responders, 14 days in the placeboand 16.5 days in the corticosteroid groups; and in all patients, 13 days inthe placebo and 19.5 in the corticosteroid groups.
Figure Legend:
In nonresponders, the median time to death was 12 days in the placebo and 24 days in the corticosteroid groups;
in responders, 14 days in the placebo and 16.5 days in the corticosteroid groups;
and in all patients, 13 days in the placebo and 19.5 in the corticosteroid groups.
Mortality or Survival
Time (days)28
Survival
1.0
0.5
i
How big a difference in mortality do you think putting a tracheostomy in at Day 4 compared to Day 10 would make on 30 day mortality?
50% RRR (15% ARR) 25% RRR (7.5% ARR) 10% RRR (3% ARR) 5% RRR (1.5% ARR)
“Negative trials”
n
Population: Mechanically ventilated adults Had been ventilated for 4 days and thought to
require at least 7 more days of ventilation Excluded:
Those requiring a tracheostomy, contraindication to tracheostomy, respiratory failure due to chronic neurological disease
Intervention: Trachesotomy by Day 4
Comparison: Tracheostomy after Day 10 if still required
Outcome: All cause mortality 30 days from randomisation
Sample Size Calculation: Baseline mortality of 30% Absolute risk reduction 6.3% (21% RRR) Power 80% Alpha 5% 4% loss to follow up
N=1692
Due to study fatigue and exhaustion of funding
N=899
“Tracheostomy within 4 days of critical care admission was not associated with an improvement in 30 day mortality”
We are 95% certain that early tracheostomy might be between
5.4% worse to 6.7% better in absolute risk About 20% better or worse in terms of relative
risk
6.3% of patients had a complication of tracheostomy
53% of patients who were randomised to delayed trache didn’t need one
2 year mortality was 52.3% Only 5 lost to follow up
Conclusions: Unable to rule out a clinically important difference
between early and late trache It probably doesn’t make a big difference to
mortality Unknown about patient perspective
Useful information about the patient cohort
Not really a “negative trial”
QUESTIONS??