Critical appraisal of:
Glickman SW et al. Disease Progression in Hemodynamically Stable Patients Presenting to the Emergency Department With Sepsis. Acad Emerg Med. 2010 17:383-90
Interactive quiz on early goal-directed therapy, surviving sepsis guidelines and EBM topic of prognosis studies.
Journal club - Disease progression in hemodynamically stable patients presenting to the emergency department with sepsis
1. JOURNAL CLUB
Disease Progression in Hemodynamically Stable Patients Presenting
to the Emergency Department With Sepsis
Glickman SW, Cairns CB, Otero RM, Woods CW, Tsalik EL, Langley RJ, van Velkinburgh JC, Park LP,
Glickman LT, Fowler VG Jr, Kingsmore SF, Rivers EP.
Acad Emerg Med. 2010 17:383-90
Farooq Khan MDCM
PGY2 FRCP-EM
McGill University
Joseph Choi MD
PGY1 FRCP-EM
McGill University
April 6th 2011
2. Objectives
• Refresh knowledge on sepsis and EGDT
• Understand the author’s rationale for doing the study
• Understanding the elements of appraising a prognosis study
• Appraise the article using what was covered in the questions
4. Which of the following were eligibility
criteria in Rivers’ original EGDT study?
A. Lactate > 4mmol/L
B. sBP < 90mmHg despite fluids
C. O2 saturation < 92%
D. Heart rate > 100bpm
E. Respiratory rate > 20
F. PaCO2 < 32
1. A, B, C, D only
2. A, B, D, E only
3. A, B, D, E, F only
4. A and B only
5. All of the above
5. Which of the following were eligibility
criteria in Rivers’ original EGDT study?
A. Lactate > 4mmol/L
B. sBP < 90mmHg despite fluids
C. O2 saturation < 92%
D. Heart rate > 100bpm
E. Respiratory rate > 20
F. PaCO2 < 32
1. A, B, C, D only
2. A, B, D, E only
3. A, B, D, E, F only
4. A and B only
5. All of the above
7. Barriers to identifying septic patients in
the ED do not include:
A. Patients with sepsis have complex and varied presentations with clinical
signs and symptoms that often overlap with non-infectious disease
states
B. There is significant variation in the microbial etiology of sepsis
C. Lab tests are neither sensitive nor specific
D. SIRS criteria are specific but not sensitive
E. SIRS criteria are sensitive but not specific
1. A, B, and D
2. A, B, and C
3. C, D and E
4. D only
5. All of the above
6. None of the above
8. Barriers to identifying septic patients in
the ED do not include:
A. Patients with sepsis have complex and varied presentations with clinical
signs and symptoms that often overlap with non-infectious disease
states
B. There is significant variation in the microbial etiology of sepsis
C. Lab tests are neither sensitive nor specific
D. SIRS criteria are specific but not sensitive
E. SIRS criteria are sensitive but not specific
1. A, B, and D
2. A, B, and C
3. C, D and E
4. D only
5. All of the above
6. None of the above
10. Whichof the followingare barriersto cost-
effectivetreatmentof sepsisin the ED
A. EGDT is invasive, labour intensive and associated with significant
potential complications
B. EGDT studies were limited to the sickest patients and are not necessarily
generalizable to all septic patients presenting to the ED
C. Mortality benefit from EGDT is time sensitive
D. Mortality benefit from EGDT is statistically significant but not clinically
significant
E. It is difficult to identify which patients will deteriorate quickly and
benefit the most from aggressive protocolized treatment
1. A, B, C, and D
2. B, C, D, and E
3. A, B, C and E
4. D only
5. All of the above
6. None of the above
11. Whichof the followingare barriersto cost-
effectivetreatmentof sepsisin the ED
A. EGDT is invasive, labour intensive and associated with significant
potential complications
B. EGDT studies were limited to the sickest patients and are not necessarily
generalizable to all septic patients presenting to the ED
C. Mortality benefit from EGDT is time sensitive
D. Mortality benefit from EGDT is statistically significant but not clinically
significant
E. It is difficult to identify which patients will deteriorate quickly and
benefit the most from aggressive protocolized treatment
1. A, B, C, and D
2. B, C, D, and E
3. A, B, C and E
4. D only
5. All of the above
6. None of the above
12. EGDT
• Patients with septic shock experience a 15% ARR
with EGDT vs conventional therapy (Rivers 2001)
• EGDT costs $26,600/QALY
13. EGDT requires…
• Intensive nursing/physician care
• Central lines
• ScvO2 measurement
• Arterial lines
• Vasoactive agents/inotropes
14. Central Lines
• 15% complication rate
• Mechanical 5-19%
• Infection 5-26%
• Thrombosis 2-26%
McGee, DC, Gould, MK. Preventing Complications of Central Venous Catheterization. N Engl J Med 2003; 348:1123-1133.
15.
16. Progression of Sepsis
• Sepsis often worsens
quickly
• Median 1 day
• Mortality increases
proportionately
• SIRS 7%
• Sepsis 16%
• Severe sepsis 20%
• Septic shock 46%
Rangel-Frausto MS, Pittet D, Costigan M, Hwang T, Davis CS, Wenzel RP.
The natural history of the systemic inflammatory response syndrome
(SIRS). A prospective study. JAMA. 1995;273(2):117.
17. Time Sensitivity
• Retrospective, multi-centered,
2154 patients
• Antibiotic therapy in severe
sepsis/septic shock patients most
powerful predictor of survival
• 44.4% patients from ED to ICU
• 79.9% patients survive if
antibiotics given in 1st hour
• 7.2% decrease in survival per
hour delay of antibiotics in first 6
hours from onset of hypotension
Kumar, A, Roberts, D, Wood, KE, et. al. Critical Care Medicine. 34(6):1589-1596, June 2006.
18. Youaredesigningastudyontheprognosisofa
particulargroupofpatients,whichofthefollowing
couldsystematicallybiasyour results?
A. All patients are selected from tertiary referral centers
B. The study’s inclusion and exclusion criteria are clearly defined
C. Your sample includes a mixture of newly diagnosed patients and end-
stage patients
D. Your sample is divided into subgroups based on factors known to affect
outcome
E. Your definition of disease and outcome is very objective
1. A, B, and C
2. A and C
3. B and D
4. C only
5. All of the above
6. None of the above
19. Youaredesigningastudyontheprognosisofa
particulargroupofpatients,whichofthefollowing
couldsystematicallybiasyour results?
A. All patients are selected from tertiary referral centers
B. The study’s inclusion and exclusion criteria are clearly defined
C. Your sample includes a mixture of newly diagnosed patients and end-
stage patients
D. Your sample is divided into subgroups based on factors known to affect
outcome
E. Your definition of disease and outcome is very objective
1. A, B, and C
2. A and C
3. B and D
4. C only
5. All of the above
6. None of the above
20. Was the sample
representative?
• Were clear inclusion/exclusion criteria used?
• Was there selection bias?
• Was the sampling method clear and
appropriate?
• Were disease definitions objective?
21. Were patients sufficiently
homogeneous with respect to risk?
• Were they at a similar well-described point in
their disease process?
• Were subgroups identified that may have a
higher or lower risk than the overall group?
22. Nowthatyou haveawellselectedsampleofpatients
foryourstudy,whichofthefollowingcould
potentiallybiasyourinterpretationofthedata?
1. A, B and C
2. B, C, and D
3. C, D and E
4. D and E only
5. All of the above
6. None of the above
A. Factors known to affect outcome were analyzed in relation to one another
using sophisticated statistical techniques
B. Your follow-up time is longer than the expected natural history of the
outcome studied
C. Patients lost to follow-up were more likely to be associated with the
adverse outcome
D. Your observed event rate of the outcome was low (e.g. 1%) with an
average loss to follow up (e.g. 10%)
E. The risk of adverse outcomes is variable over time
23. Nowthatyou haveawellselectedsampleofpatients
foryourstudy,whichofthefollowingcould
potentiallybiasyourinterpretationofthedata?
1. A, B and C
2. B, C, and D
3. C, D and E
4. D and E only
5. All of the above
6. None of the above
A. Factors known to affect outcome were analyzed in relation to one another
using sophisticated statistical techniques
B. Your follow-up time is longer than the expected natural history of the
outcome studied
C. Patients lost to follow-up were more likely to be associated with the
adverse outcome
D. Your observed event rate of the outcome was low (e.g. 1%) with an
average loss to follow up (e.g. 10%)
E. The risk of adverse outcomes is variable over time
25. Follow up
• Was follow up sufficiently long?
• Was there significant loss to follow up?
• Is there a relationship between losses to follow-
up and adverse outcome of interest?
• Is the event rate low enough for small losses to
follow up to have significant impact?
26. Outcomes
• Were objective and unbiased outcome criteria
used?
• How likely are the outcomes over time?
• How precise are the estimates of likelihood?
29. The following could have an impact on
outcome in a prognosis study..
A. The selection of patients eligible for treatment varied across participating
centers
B. Patients were offered different treatment agents at different participating
centers
C. Patients were treated at different times in their disease process
D. New treatment guidelines were implemented over the course of the study
period
E. New technology was introduced in the diagnosis and treatment of the
disease within study period
1. A, B, and C
2. B and D
3. C, D, and E
4. A only
5. All of the above
6. None of the above
30. The following could have an impact on
outcome in a prognosis study..
A. The selection of patients eligible for treatment varied across participating
centers
B. Patients were offered different treatment agents at different participating
centers
C. Patients were treated at different times in their disease process
D. New treatment guidelines were implemented over the course of the study
period
E. New technology was introduced in the diagnosis and treatment of the
disease within study period
1. A, B, and C
2. B and D
3. C, D, and E
4. A only
5. All of the above
6. None of the above
31. Applicability
• Bias of treatment effect on prognosis
• Difference in practice patterns over region/time
• Evolving technology and data
• Affects your ability to apply the data to your patients
33. Was the sample
representative?
• What was the study population?
• Were clear inclusion/exclusion criteria used?
• Was the sampling method clear and appropriate?
• Were disease definitions objective?
• Was there selection bias?
35. Inclusion/exclusion criteria
– Inclusion
• Known or suspected infection
• 2 or more SIRS criteria
• Over 18
– Exclusions
• Imminently terminal comorbid condition
• Advanced AIDS (CD4 <50)
• On antibiotics
• Enrolled in another trial
– Not analyzed
• Patients in septic shock at presentation
• Patients later determined not to have had an infection
• Appropriate or inappropriate?
36. Sampling method
• Convenience sampling
• Authors say this was necessary given the need
for prompt study specimen handling for
metabolic studies
• Issues with this?
37. Disease definitions
• Sepsis defined as SIRS plus suspected infection
• Independent adjudicators of infection status (up to 3 with high
interrater agreement κ > 0.8)
• Severe sepsis/septic shock included multiple objective
criteria
• evidence of end organ dysfunction:
• Lactate > 1.5 times upper limit of normal
• Arterial pH < 7.3
• Platelets < 80 × 103
• Intubation or PaO2/FiO2 < 250
• Urine output < 0.5mg/kg/hr despite adequate fluid resuscitation
• sBP < 90 mm Hg or MAP < 65 mm Hg despite adequate fluid resuscitation
• Evidence of shock
• sBP < 90 mm Hg or MAP < 65 mm Hg despite initial fluid challenge
• Lactate > 4 mmol/L
• Issues with these definitions?
38. Selection bias
• Tertiary care centers
• Included more ethnic diversity
• Sampled more community patients with uncomplicated
disease
• Problems inherent with the definitions of sepsis
• Not necessarily representative of night-time disease. Is
this significant?
39. Were patients sufficientlyhomogeneous
with respect to prognostic risk?
• Were they at a similar well-described point in their disease
process?
• Were subgroups identified that may have a higher or lower
risk than the overall group?
• Can you think of factors that the investigators have neglected
that are likely to define subgroups with very different
prognoses?
• Were prognostic factors analyzed in relation to one another?
43. Factors investigators neglected that are
likely to define subgroups with very
different prognoses?
• Demographics
• Physiological markers
• Comorbidities
• Infection site
• Causative organisms
• Treatment!
• Any others?
44. Were prognostic factors analyzed in
relation to one another?
• Multiple variable logistic regression model
45. Follow-up, Outcome and
Applicability
• Was follow up sufficiently complete?
• Were objective and unbiased outcome criteria used?
• How likely are the outcomes over time?
• How precise are the estimates of likelihood?
• What does this study add? Can I apply it to my practice?
46. Was follow up sufficiently
complete?
• 100% at 72h
• ? Losses to follow-up at 30 days
47. Were objective and unbiased
outcome criteria used?
• Progression to septic shock at 72 hours
• Progression to severe sepsis and/or septic shock
at 72 hours?
• Death at 30 days
50. How precise are the estimates
of likelihood?
• Mortality with early progression
OR = 4.72, 95% CI = [2.01 to 11.1]
51. Prognostic risk factors
• Hyperthermia (per °C) OR 1.34 [1.06–1.68]
• Female Sex OR 2.57 [1.50–4.40]
• Vascular access site OR 5.06 [1.81–14.1]
• Age (per decade) OR 1.22 [1.05–1.42]
• Hematocrit (per %) OR 0.96 [0.92–1.00]
• Lung disease OR 2.30 [1.29–4.10]
• Issues?
52. What does this study add? Can
I apply it to my practice?
•The floor is open!
53. References
• Jaimes, F., J. Garcés, J. Cuervo, et al., The systemic inflammatory response syndrome (SIRS)
to identify infected patients in the emergency room. Intensive Care Medicine, 2003. 29(8):
p. 1368-1371.
• Bone, R.C., R.A. Balk, F.B. Cerra, et al., Definitions for sepsis and organ failure and guidelines
for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference
Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest,
1992. 101(6): p. 1644-1655.
• Dellinger, R.P., M.M. Levy, J.M. Carlet, et al., Surviving Sepsis Campaign: International
guidelines for management of severe sepsis and septic shock: 2008. Critical Care Medicine,
2008. 36(1): p. 296-327 10.1097/01.CCM.0000298158.12101.41.
• Rangel-Frausto, M.S., D. Pittet, M. Costigan, et al., The Natural History of the Systemic
Inflammatory Response Syndrome (SIRS) A Prospective Study. JAMA, 1995. 273(2): p. 117-
123.
• Rivers, E., B. Nguyen, S. Havstad, et al., Early Goal-Directed Therapy in the Treatment of
Severe Sepsis and Septic Shock. New England Journal of Medicine, 2001. 345(19): p. 1368-
1377.
• Jones, A.E., M.D. Brown, S. Trzeciak, et al., The effect of a quantitative resuscitation strategy
on mortality in patients with sepsis: A meta-analysis *. Critical Care Medicine, 2008. 36(10):
p. 2734-2739 10.1097/CCM.0b013e318186f839.
• Rivers, E.P., L. McIntyre, D.C. Morro, et al., Early and innovative interventions for severe
sepsis and septic shock: taking advantage of a window of opportunity. CMAJ, 2005. 173(9):
p. 1054-1065.
54. References
• Pollmächer, T., J. Mullington, C. Korth, et al., Diurnal Variations in the Human
Host Response to Endotoxin. The Journal of Infectious Diseases, 1996. 174(5): p.
1040-1045.
• Scheff, J.D., S.E. Calvano, S.F. Lowry, et al., Modeling the influence of circadian
rhythms on the acute inflammatory response. Journal of Theoretical Biology,
2010. 264(3): p. 1068-1076.
• Fromm, R.E., Jr., L.R. Gibbs, W.G. McCallum, et al., Critical care in the emergency
department: a time-based study. Crit Care Med, 1993. 21(7): p. 970-6.
• Magid, D.J., Y. Wang, J. Herrin, et al., Relationship Between Time of Day, Day of
Week, Timeliness of Reperfusion, and In-Hospital Mortality for Patients With
Acute ST-Segment Elevation Myocardial Infarction. JAMA, 2005. 294(7): p. 803-
812.
• Alberti, C., C. Brun-Buisson, S. Chevret, et al., Systemic Inflammatory Response
and Progression to Severe Sepsis in Critically Ill Infected Patients. Am. J. Respir.
Crit. Care Med., 2005. 171(5): p. 461-468.
• Martin, GS, Mannino, DM, Eaton, S, Moss, M. The epidemiology of sepsis in the
United States from 1979 through 2000. N Engl Med 2003; 348:1546.
• User’s Guide to the Medical Literature, JAMA, online pubs.ama-
ssn.org/misc/usersguides.dtl
Editor's Notes
Explanation for answer 1
Explanation for answer 3
Explanation for answer 3
Explanation for answer 3
Explanation for answer 3Picture of pneumothorax as a complication of CVL placement
Explanation for answer 3
Explanation for answer 3
If we go through the answersA is an example of a selection bias, one that prognosis studies often fall prey to.Clear definitions of inclusion and exclusion criteria lead to less potential bias since it decreases the risk of your sample being unrepresentative
So the take home points for appraising prognosis study are...Clear definitions of inclusion and exclusion criteria lead to less potential bias since it decreases the risk of your sample being unrepresentative
This requires an understanding of the biology of the condition being studied-- can you think of factors that the investigators have neglected that are likely to define subgroups with very different prognoses?Failure to identify these groups can lead to unrepresentative results. Where one subgroup biases the result towards a favourable estimate of prognosis, and the other towards and unfavourable one and the overall result is representative of neither group individually.
To eliminate confounding and identify which factors emerge as independent prognostic indicators, use multivariate analyses
Given what you know about the natural history of the outcome studiedIf this is 40% its a no brainer, most researchers will say that 10% is acceptable.Obviously if the patients were lost because they all died then there’s a problemAnd furthermore the degree to which the loss to follow up rate undermines your study’s validity depends on how rare your outcome isIf you measure a bad outcome in 20% percent of the patients in your study and your losses are 10% then worst case scenario your true event rate is 30% which isn’t that much different. But if you only measured it 1% of the time with the same loss then your true event rate is now 11% which is significantly different.
Just like disease definitions, outcomes should be easy to measure,objective and pre-definedThe less objective the outcome measure, the more blinded assessors should beRisk may change over time this is the reason why in prognosis studies results are reported in Kaplan-Meier curveFinally just as with any other result in all types of studies
Left is the survival after MI, notice it drops precipitously in the first few hoursAs opposed to the curve on the right which is for survival after hip replacement
Notice the confidence intervals widen as you go further along the survival curve to loss of power from deaths/dropouts/loss to follow up.
This is just to highlight the fact that treatment is one of the biggest factors that affect outcome in a prognosis study and can be a huge confounder if not dealt with properly.
Eliminates confoundingExcluding patients on antibiotics would decrease the number of false negative cultures and avoids the confounding effect of treatment on outcome in a prognosis study. Whether this would undermine the external validity of the study is unclear since the proportion of patients excluded is not reported
While physiological studies have shown circadian fluctuation of inflammatory mediators such as cytokines and glucocorticoids, the net effect of the time of day on the acute inflammatory response is unclear8, 9. It is known that a significant proportion of critically ill patients present to the ED during off-hours10, and that there may even be an increase in mortality due to a lack of resources or delays in diagnosis and treatment11
How much fluid is adequate? How many litres in a challenge?Did the authors have consistencies in the amount/type of fluids used? Time from initiation of fluids to enrolment and collection of data?
Measures taken to avoid selection bias
This is the group they derive the risk factors from
93 patients with severe sepsis at the time of enrolment were not excluded from the derivation of risk factors for progression to shock, even though they clearly have a different baseline prognosis and would already be candidates for EGDT by earlier guidelines. Although they were presented separately in the flowcharts, no analysis of their outcomes, i.e. progression to shock or mortality, is reported. Instead a composite outcome of progression to either septic shock and/or severe sepsis was reported as part of the subgroup analysis of patients with uncomplicated sepsis. This group of patients, who presented initially with uncomplicated sepsis and then deteriorated within 72 hours, represents the real population of interest according to the study objective, so it is unclear why the authors did not analyze the risk factors separately for these patients.
consistency of use, timing and choice of antibiotics and resuscitation96% of patients received appropriate antibiotics within 1st 24 hours not different between early progressors and non-progressors
In-hospital mortality based on medical records was determined at 30 days but no losses to follow-up were documented. A Kaplan-Meier graph was provided but this by definition censors dropoutsOut-of-hospital mortality was not assessed nor was transfer to another facility. In fact the authors did not appear to make any efforts to ascertain patients’ status post-discharge.
Previous sepsis studies have been carried out on ICU patients, with nosocomial infections, and a different natural history. Here the setting is in the ED, with community acquired infections and course of disease that is closer to what the average EP would encounter in his/her practice. While it is common knowledge that sepsis carries a grave prognosis for a significant number of patients, this study appears to be amongst the first to quantify that risk in the ED, and was for the most part appropriately designed and carried out with a rigorous methodology.More importantly, it suggests that it may be possible to select stable patients who would benefit from EGDT without wasting resources on those who would not. One caveat is that it is a single study which was insufficiently powered to determine risk factors with a great degree of certainty. The greater caveat is that the authors included patients with severe sepsis who are already known candidates for EGDT in the derivation of those risk factors. Nevertheless, this study may provide hypothesis generation for further studies, making its impact on the literature significant even if it is not yet practicpe-changing