Critical Appraisal
Louis Muller
October 2009
EBM – Hierarchy of Evidence
Prospective Cohort study
• Definition:
Prospective – a study that collects data after
the study begins
Cohort – a group of similar individuals
A study that follows over time a group of similar
individuals who differ with respect to certain factors
under study. In order to determine how these factors
affect rates of a certain outcome
Prospective Cohort study
Prospective Cohort Study
Outcomes after Intravenous
Opioids in Emergency patients:
A Prospective Cohort Analysis
Acad emerg med – June 2009
Vol 16, No 6
OBJECTIVES
• To measure the outcomes following
administration of IV opioids
• To identify clinical factors that may
predict poor analgesic outcomes in
these patients
METHODS
Methods: In this prospective cohort study, emergency patients were
enrolled if they were prescribed IV
morphine or hydromorphone (the most commonly used IV opioids in the
study hospital) as their initial
analgesic. Patients were surveyed at the time of opioid administration and
1 to 2 hours after the initial
opioid dosage. They scored their pain using a verbal 0–10 pain scale. The
following binary analgesic
variables were primarily used to identify patients with poor analgesic
outcomes:
1) a pain score reduction
of less than 50%,
2) a postanalgesic pain score of 7 or greater (using the 0–10 numeric
rating scale),
3) the development of opioid-related side effects.
Logistic regression analyses were used to study
the effects of demographic, clinical, and treatment covariates on the
outcome variables
RESULTS
• Results: A total of 2,414 were approached for enrollment, of whom 1,312
were ineligible (658 were identified more than 2 hours after IV opioid was
administered and 341 received another analgesic before or with the IV
opioid) and 369 declined to consent.
• A total of 691 patients with a median baseline pain score of 9 were
included in the final analyses.
• Following treatment, 57% of the cohort failed to achieve a 50% pain score
reduction, 36% had a pain score of 7 or greater, 48% wanted additional
analgesics, and 23%developed opioid-related side effects.
• In the logistic regression analyses, the factors associated with poor
analgesia (both <50% pain score reduction and postanalgesic pain score of
‡7) were
• the use of longacting opioids at home, administration of additional
analgesics, provider concern for drug-seeking behavior, and older age.
• An initial pain score of 10 was also strongly associated with a
postanalgesicpain score of ‡7.
• African American patients who were not taking opioids at home were
less likely to achieve a 50% pain score reduction than other patients,
despite receiving similar initial and total equianalgesic dosages.
CONCLUSIONS
• Conclusions: Poor analgesic outcomes were common
in this cohort of ED patients prescribed IV opioids.
• Patients taking long-acting opioids, those thought to
be drug-seeking, older patients, those with an
• initial pain score of 10, and possibly African
American patients are at especially high risk of poor
analgesiafollowing IV opioid administration
Why was study done?
• Over 40% of ED visits related to pain – US
• Inadequate pain relief continues to be common
• Many studies – documented failings in the processes
of ED pain Mx
• In Multicenter sample – NY only 60% of pts with mod
to severe pain received any analgesia and median
delay was 90min.
Why was study done?
• Use of IV opioids = the provider ID the
presence of significant pain and is attempting
to reduce pain in the MOST AGGRESSIVE
MANNER POSSIBLE
• HOWEVER – little is known about the clinical
outcomes resulting from the admin of opioids
in diverse ED pt population – OBJECTIVE 1
Why was study done?
• Several risk factors for poor analgesic
prescribing practices have been identified in
ED pts. (older age, race, female sex, ED
overcrowding)
• However little is known about which clinical
factors might predict clinical outcomes
following IV opioids. – OBJECTIVE 2
Was a control group used?
• Is it possible, in this specific study?
– Opioid vs placebo ? (unethical)
– Opioid vs NSAID ?
Changes the objective ? Different design!
• Home-opioid vs no home-opioid use
– Only retrospective finding
Design
• Are the aims clearly stated? Yes
• Is the design appropriate to the stated aims?
– Could these aims have been studied by a different design?
Should this study have been done by RCT?
– Unethical / impossible / impractical
– Paracetamol vs opioid (would change the objective of
study)
– Also too many variables in patient population ( Renal colic
vs fractured limb)
– In RCT – strictly speaking – should the groups be identical
apart from the intervention
Design
• Was study ethical?
– Yes – doing the study did not change the normal
patient management in ED.
– Informed consent
Design/Method
• Was the sample size justified?
– Study conducted in ED with 93 000 annual visits.
– Study over 29 month period (232 500)
– 2414 pts approached (0.01%)
– 691 in final analysis ( 0.002%)
– How many pts receive opioids/yr ?
– Missed many patients ! (diagram)
Design/Method
•
How were subject recruited? Inclusion and exclusion
criteria?
– Enrollment occurred between July 2004 and November 2006
– Trained ED enrollers present in ED 8 am to midnight ( screened
triage diagnoses, alert by nurses and providers)
– Eligible for inclusion: >18yrs and received IV opioid within 2hrs of
enrollment as initial analgesia.
– Exclusion criteria: prior other analgesia received in ED or via EMS,
cognitive fx imp, unstable vital signs, unable to communicate in English
Methods
• What outcomes were measured and how?
– Analgesic outcome
• Pain score before and after analgesics (1-2hrs post dose)
• Standard verbal 0-10 numeric rating scale
• Poor outcome – (3)
– Development of opioid-related s/effects
• Documented in chart/orders for naloxone/vital Sx
• Reported by the subject or nurse
Methods
• Are the measurements likely to be reliable and
valid? Was the intervention accurately measured?
– verbal pain scoring system – validated/accepted but not as
good as written/diagram
– pain perception vs personalities, but trend not affected?
– timing of post analgesia survey will not be exactly the
same for every patient ( within 1-2hrs)
– Did use 7 as a cut off.
– Probably reliable and valid
Design / Analysis
• Are the statistical methods described?
– Kruskal-Wallis test for continuous variables
– the Fisher exact test was used for categorical
variables.
• Were the basic data adequately described
and do the numbers add up?
– confusing/don’t think so (table 1)
Results
• Only 29 % of patients approached, included in
final analysis
–
–
–
–
median age = 40
initial pain score = 9
60% female
32% taken opioids < 48hrs
Results
• Opioid dosage
– Mean initial was 4.0 Mmg or 0.055Mmg/kg when adjusted for weight
– Hydromorphone – higher equianalgesic dose of 6.7Mmg or 0.098
Mmg/kg
• Postanalgesic pain score
– mean of 87 min after initial dose
– 57% of pt’s provided preanalgesic pain score
AFTER initial dose (comparison revealed no difference)
Results
• Analgesic outcomes
– 57% failed to achieve a 50% pain score reduction
– 36% pain score of 7 or greater
– 23% developed at least 1 s/effect
– 48% wanted additional analgesics
Results
•
Demographic, clinical and treatment characteristics of the cohort
according to analgesic outcomes are presented in table 1.
–
–
•
difficult to interpret
Population not broken down very well?
Characteristics substantially associated with poor analgesic outcome.
1.
2.
3.
4.
5.
6.
African American race
Initial pain score of 10
Home opioid use
History of chronic medical disease
Provider concern for drug seeking
Admin of additional analgesics
Results
• Comparison of outcomes - based on home
opioid status (table 2)
 pt’s taking long acting opioids received
substantially higher doses
0.09 Mmg vs 0.06 Mmg
Despite this – poor analgesic outcomes
80% - less than 50% PS reduction
62% - postanalgesic PS more 7
What do the main findings mean?
• Most pt’s in this cohort failed to achieve good
analgesia, after receiving IV opioids
• Several clinical variables were associated with
failure to achieve good analgesia.
– Should we have a more aggressive approach to
patients with these high risk markers?
– Needs further study.
Limitations
• Dosing –inadequate? Morphine - 0.1mg/kg Further studies to identify
optimal opioid dosing strategy for ED patients is needed.
• Sample size – 29% ( final analysis) – good representation?
• Pain scoring system very subjective. (different thresholds and personalities
etc.) Is there any other way to evaluate pain?
• Should they have done another follow up on pain score?
• Can results be generalized? Do ED patients really differ from centre to
centre?
• Results biased by patient selection factors. (female vs male, type of pain
etc.)
• Practically very challenging study to do.
– enrollers missed many patients
– maybe study on specific injuries/type of pain could be considered. Less variables. More
focused?
Implications for our practice?
• Different patient population – HIV/TB vs rest
• Do HIV and TB affect bodies response to opioids?
• Ritonavir decrease morphine effect.
• Opioids – the gold standard. More aggressive use?
• Opioids vs combinations?
• We probably have a higher rate of analgesia failure.
(1ste world study vs our overcrowded, short staffed
ED)