Group
26
Presenter/s
Case 26#
Chris Barron, Rebecca Brown, Phil Castrovinci, Matthew Ho, Emma
Jones, Kathleen McFadden, Alex Podlaski, Victoria Psomiadis
Clinical Scenario
You are on call in your hospital intensive care unit when you receive a call from endoscopy suite on the 3 rd floor. Two hours ago they have completed screening colonoscopy on a 68 year old male. The procedure involved several colonic biopsies of fairly large but benign looking polyps. Two hours later the patient had very large volume of clots and bright red blood per rectum. Blood pressure, which was 138/82 before colonoscopy is now 92/50, heart rate increased from 76 to
115/min. The gastroenterologist has already called the on-call surgeon who is currently operating and cannot attend right away, and both physicians call you in the ICU for help. You start by making sure that patient blood samples have been sent to measure the hemoglobin and coagulation studies and to cross & type to prepare for transfusion. A bed in ICU is being prepared for urgent admission.
A student who is doing an elective rotation in your medical-surgical ICU mentions that during the previous week, while working in the trauma ICU, he noticed use of tranexamic acid in bleeding patients and actually looked for evidence supporting its use. You do not remember using this drug before but, pending transfer of the patient to ICU, you decide to look at the data.

Clinical Question:
In patients with procedural related bleeding, is tranexamic acid (TXA), compared to no treatment, useful in reducing blood loss?
Patient or problem – patients with procedural related bleeding
Intervention – use of tranexamic acid
Control – no treatment
Outcome – reduced bleeding in procedure-related bleeding
(T) Question Type (Therapy, Diagnosis, Harm, Prognosis, etc.)
Therapy
Your best answer before searching ( hint: matching the study design to the question)
The best answer would have been a review of randomized controlled trials that pointed to a clinically significant effect of TXA either for or against reduced bleeding in procedure-related bleeding.

Search
(Please provide the search strategy used.
hint: matching the study design to the question)
Search Strategy
We began with a broad search strategy whereby each member of the group searched relevant databases to assess the quality and quantity of research on
TXA in treatment of postoperative bleeding. Once individual members of the group identified relevant articles, the group as a whole collectively decided upon the article which would be appraised. Our search focused on articles that were broad enough in scope to cover all surgical patients (and not those exposed to one specific procedure) in order to best apply the study to our patient. Finally, given that our question focused on treatment, we focused on RCTs and SRs of
RCTs in order to best formulate a treatment strategy for the population in our question.
Database you searched, and why?
The Cochrane Library, because it provides a library of systematic reviews and since we had already decided to use a systematic review, it was a guaranteed reliable source for articles.
Which reference did you pick?
“Tranexamic acid for reducing mortality in emergency and urgent surgery.” (2013)
Why?
The systematic review directly addressed the clinical question and at first sight appeared to be a strong, mostly unbiased report on the therapeutic efficacy of
TXA.
The Article
Citation:
Perel P, Ker K, Morales Uribe CH, Roberts I. Tranexamic acid for reducing mortality in emergency and urgent surgery. Cochrane Database of Systematic
Reviews 2013, Issue 1. Art. No.: CD010245. DOI:
10.1002/14651858.CD010245.pub2.
Study design: Systematic Review of Randomized Controlled Trials.
Chose the appropriate critical appraisal form and complete the assignment (e.g. diagnosis, prognosis, treatment or harms).

Reviews and Meta-Analysis Checklist
1.0 Objectives and Hypothesis
1.1 Are the objectives of the study clearly stated?
Yes, the authors clearly stated that their goal was as follows:
“To assess the effects of tranexamic acid on mortality, blood transfusion and thromboembolic even ts in adults undergoing emergency or urgent surgery”.
2.0 Design
2.1 Is the study design suitable for the objectives?
Yes, study design looked at Randomized Controlled Trials, useful for determining treatment efficacy.
2.2 Were the search methods used to locate relevant studies comprehensive?
Yes, the search was done through electronic databases and covered unpublished studies through online trial registries. The trials found compared the use of TXA vs not using it or the use of a placebo in adults undergoing urgent or emergency surgery.
2.3 Was this the right sample to answer the objectives?
Yes, the samples were specifically adults undergoing urgent or emergency surgery.
2.4 Is the study large enough to achieve its objectives? Have sample size estimates been performed?
No, the review describes that,
“The small sample sizes and low event rates in the included trials prevent us from drawing reliable conclusions about the effect of tranexamic acid on any of the other outcomes considered in this review.” The largest sample size was 110 in the Zufferey (2010) study.
2.5 Were all subjects accounted for?
No, 3 out of the five studies did not have report Attrition Bias, making it unclear as to whether all subjects were accounted for. Zufferey (2010) and Pfizer (2011) reported an intention to treat analysis with Zufferey not reporting any loss to follow up while
Pfizer had reported a 17% loss to follow-up.

3.0 Measurement and Observation
3.1 Is it clear what was measured, how it was measured and what the outcomes were?
Mortality, probability of blood transfusion, deep vein thrombosis and stroke were measured in the five papers identified. These were the main measurements, although there were secondary outcomes. The mortality, as a dichotomous outcome, was measured by Risk Ratio, and continuous outcomes were measured and compared by mean difference and 95%CI.
3.2 Were explicit methods used to determine which studies to include in the review?
Studies were selected from a number of databases: “the Cochrane Injuries Group
Specialised Register (22 August 2012); Cochrane Central Register of Controlled
Trials (2012, Issue 8 of 12); MEDLINE (OvidSP) (1950 to week 2 August 2012);
PubMed (1 June 2012 to 22 August 2012); EMBASE (OvidSP) (1980 to Week 33
2012); ISI Web of Science: Conference Proceedings Citation Index-Science (CPCI-
S) (1990 to 22 August 2012); ISI Web of Science: Science Citation Index Expanded
(SCI-EXPANDED) (1970 to 22 August 2012). We also searched online trials registers on 22 August 2012 to identify unpublished studies.
”
After extraction of 12233 articles, there was a paring down process due to removal of duplicates, completeness (full text), screened for eligibility, and then found five studies for inclusion.
3.3 Was the selection of primary studies re-reproducible and free from bias?
The selection of studies is reproducible from the algorithm described above. Two authors independently selected from the extraction of studies, and a third author, blinded from the results of the studies checked for their inclusion of emergent/urgent surgery. The authors did not disagree about the inclusion of studies in the review.
3.4 Was the methodological quality of the primary studies assessed?
The Cochrane review included a bias risk summary for the five studies included.
There is a risk of selection bias, detection bias, attrition bias in at least 3 studies, and reporting bias in four studies. The authors describe ‘other sources’ of bias in all five of the studies selected.

The authors also created a risk of bias graph that averaged all the studies together:
The authors rate the risk of bias overall as low.
3.5 Are the measurements valid?
The measurements are valid as they are looking at both dichotomous and continuous data.
3.6 Are the measurements reliable?
The measurements are not reliable because many of the included studies are abstracts and the full text not available. The measurements might not be reliable.
3.7 Are the measurements reproducible?
According to the report, “None of the trial reports provided data on the number of units of blood transfused.” Furthermore, “two trials were reported as conference abstracts and although we contacted the authors, we were unable to obtain further details about their methods or results.”
From this analysis, I do not believe these measurements could be reproduced.
4.0 Presentation of Results
4.1 Are the basic data adequately described?
Yes, the basic data is adequately described. The Background information of the study provided relevant characteristics detailing the prevalence of emergency surgery and the need to control bleeding. The Data presented analysis, including descriptions of the confidence intervals, of the effects of TXA, specifically that TXA reduces the probability of cell transfusion by 39% and a non-significant reduction in risk of reoperation and mortality. Moreover, the basic data is clearly presented under a two separate tables: “Characteristics of included studies [ordered by study
ID]” and “Comparison 1. Tranexamic Acid versus control.” Data/statistical

methodology included Risk Ratios (M-H, Fixed, 95% Cl) and effect sizes. In addition, the figures section provides a seamless display of risk bias graphs, comparison of outcomes of TXA and other affects such as stroke, myocardial infarction, pulmonary embolisms, Deep Vein Thrombosis, Transfusion, and Mortality.
4.2 Were the differences between studies adequately described?
The study identified five trials evaluating the effects of TXA in emergency or urgent surgery, but only three of them provided pre-specified outcomes. Of those 3 studies, two of them were deemed “low risk of bias for allocation concealment and blinding. Moreover, the table “Characteristics of Included Studies” provided analysis of the methods, risk of bias (including author’s judgment and support for judgment), interventions, outcomes, notes, participants, and blinding of outcome assessments.
4.3 Are the results presented, objectively and in sufficient detail to enable readers to make their own judgment?
Yes, the data is absolutely presented in an objective format and sufficient detail to enable the reader to make their own judgment. The study, specifically the Tables and Figures section, provide an in-depth discussion and analysis of each of the analyzed articles. Moreover, the authors Perel, Ker, Uribe, and Roberts demonstrate bias in one of the three most relevant studies. This provides the reader with context and sufficient background knowledge to adequately judge the validity of the study.
To describe the “Assessment of risk of bias in included studies” the team used the Cochrane Collaboration’s tool described in a previous study (Higgins
2011). The risk of bias was assessed as “low, high or unclear” risk and evaluated under sequence of generation, allocation concealment, blinding, incomplete outcome data, and selective outcome reporting.
4.4 Are the results internally consistent, i.e do the numbers add up properly?
Yes, the results are internally consistent. Specifically, Analysis 1.2 Comparison 1
Tranexamic Acid versus control, Outcome 2 Transfusion clearly shows a Risk Ratio diagram between the Sadeghi 2007, Zufferey 2010, Pfizer 2011. The Risk-Ratio (M-
H, Fixed, 95% CI) plot clearly displays a positive relation for favors Tranexamic Acid in reduction of the use of Transfusion. Moreover, a second plot is consistent with the idea that TXA is not correlated to an increase in Myocardial Infarction (Analysis 1.5
Comparison 1 Tranexamic Acid vs Control, Outcome 5 Myocardial Infarction).
5.0 Analysis
5.1 Were the results of primary studies combined appropriately?
Yes, the study authors were specific in their inclusion criteria in order to allow for proper integration of the respective RCTs. Only studies that included some report of mortality, receipt of blood transfusion, units of blood transfused, reoperation, seizures and thromboembolic events (MI, stroke, DVT, and PE). By using studies that assessed similar parameters, the studies were able to be compared effectively.
There was also an assessment performed for heterogeneity and risk of bias in the

studies being compared. Finally, these data were compared using a fixed-effect model which treated each of the variables as if the quantities were non-random.
5.2 Has a sensitivity analysis been performed?
Yes, a sensitivity analysis was conducted for the risk of bias judgment for allocation concealment. The authors only performed this analysis for transfusion because it was the only outcome measure included in at least three studies. The authors concluded that “RR remained unchanged when using a random-effects model, and when [they] included only trials with low risk of bias for allocation concealment the
RR was 0.47 (95% CI 0.26 to 0.85).”
5.3 Were all the important outcomes considered?
The authors included both primary outcomes (mortality) and secondary outcomes
(myocardial infarction, stroke, deep vein thrombosis, pulmonary embolism, seizure, renal failure, reoperation, blood transfusion, and units of blood transfused). The range of outcomes assessed is thorough, but could also perhaps include readmission statistics. Of the five studies included in this systematic review, the authors were only able to obtain “mortality and transfusion data for three trials; data on deep venous thrombosis for two trials; and data on pulmonary embolism, myocardial infarction and stroke for one trial.”
5.4 Are the data suitable for analysis?
An assessment of heterogeneity amongst the trial participants, interventions, and outcomes was performed using I
2
statistic and Chi
2
test. The I
2 value was found to be
0% for all outcomes assessed in this analysis. Based on these results, the studies can be evaluated as entirely homogenous, providing the rationale for their comparison and analysis in this SR.
5.5 Are the methods appropriate to the data?
The fixed effects model seems to be appropriate for these data. During heterogeneity analysis, the study designers found an absence heterogeneity with respect to participants, interventions, and outcomes (I
2
=0% in all instances). These I
2 values suggested that the studies were entirely homogeneous, thus indicating the fixed-effect model (and thus the assumption that the explanatory variables were nonrandom) was appropriate. The fixed-effect model is also useful in controlling for unobserved heterogeneity, which would be proper given the lack of observed heterogeneity in this study. The other methods implemented (calculating risk ratios and 95% CIs) allowed for a valid assessment of each outcome being assessed.
While the authors were comparing independent studies which tested multiple variables, the utilization of RR/CI analysis provided for individual comparison of each relevant outcome/variable. Finally, a sensitivity analysis was also performed which indicated that the RR was unchanged when using a random-effects model as well.

5.6 Are any statistics correctly performed and interpreted?
First, the statistics were performed properly for several reasons. No subgroup analysis was performed, which adds validity to their study. The authors performed their “sensitivity analysis according to the risk of bias judgment for allocation concealment” and for the “fixed-effect versus random effects models for data synthesis.
”
These efforts provided for thorough statistical analysis taking into account the heterogeneity or lack thereof of those studies. The statistics were properly interpreted as follows:
 Mortality (RR 1.01, 95% CI 0.14-7.3) was ruled to be unimproved by TXA

 administration.
Necessity of transfusion following treatment (RR 0.70, 95% CI 0.52-0.94) was correctly described as being reduced
Deep Venous Thrombosis (DVT) Risk (RR 2.29, 95% CI 0.68 to 7.66) was
 correctly interpreted as being not significantly reduced
Other outcomes : PE, MI, and stroke were not evaluated due to lack of available data. Only one stroke was observed in the patient population in these studies (RR 2.79, 96% CI 0.12 to 67.10), but one would question the validity of this analysis given the lack of study participants who experienced this outcome.
6.0 DISCUSSION
6.1 Are the results discussed in relation to existing knowledge on the subject and study objectives?
Yes, because the authors reference a Cochrane systematic review of “the effects of antifibrinolytic agents for patients undergoing elective surgery” (Henry 2011) and believes that their results are consistent with those of the previously completed systematic review.
6.2 Is the discussion biased?
Yes, because during the assessment of bias risk the same two authors worked on every point. The discussion itself is fair but the authors admitted that while their source studies had low risk of bias, the systematic review had a low-level risk of bias because they “cannot exclude the possibility that some trials were missed” , and that there was no consensus on definitions of emergency and urgent surgery.
7.0 Interpretation
7.1 Are the authors’ conclusions justified by the data?
The authors’ conclusions were justified by the data given that the study revolved around analyzing data from multiple sources. The authors state, “Taking into account the evidence provided in this review and in the context of other evidence such as from the review by Henry 2011, we can conclude that there is evidence that

tranexamic acid reduces blood transfusion in patients undergoing emergency or urgent surgery.”
7.2 What level of evidence has the paper presented? (using CEBM levels)
1a (therapy)- this is an SR of RTCs with sufficient homogeneity between the studies.
7.3 Does this paper help me answer my problem?
Considering that this paper accomplished a systematic review of all plausible papers on the subject of using TXA post-operatively for urgent surgeries, I would feel comfortable using TXA in my case after having read this systematic review.
8.0 Implementation
8.1 Can any necessary change be implemented in practice?
The use of TXA can be added to the standard of care in urgent and emergent surgery based on the reduction in need for blood transfusions which come with a host of side effects associated with the use of this therapy. Before this could be implemented however, the correlation between the use of TXA and death, thrombosis, and other clinically relevant outcomes should be examined.
8.2 What aids to implementation exist?
The significant finding of this systematic review- a reduction in risk of transfusion during emergent or urgent surgery with the use of TXA, would appeal to health professionals performing and assisting with these surgeries as well as following up with the patients. These professionals could provide momentum for the implementation of this therapy.
8.3 What barriers to implementation exist?
The abovementioned deficiency in information regarding the correlation between the use of TXA and other clinically relevant outcomes would serve as a barrier to implementation because it would cause doubt as to the safety of this drug.