This was an observational trial of the effect of implementing a protocol for lower GI hemorrhage involving prompt CT angiography for all urgent and emergent cases (below).  The protocol was implemented in 2009.  Data was extracted from all patients undergoing invasive angiography for four years before and after protocol implementation. 

161 invasive angiographic procedures were performed, 78 before and 83 after implementation of the protocol.  Following protocol implementation the use of CT angiography increased from 4% to 57% and the use of tagged RBC scanning decreased from 83% to 51%, revealing incomplete protocol adherence. 

There was little difference in average outcomes following protocol implementation.  There was no difference in the ability to detect the source of bleeding at invasive angiography, success of embolization, the average minimum hemoglobin level, or the number of patients requiring surgery. 

CT angiography was superior at localizing the hemorrhage compared to tagged RBC scan, using invasive angiography as the gold standard (figure below).  This may explain the lower fluoroscopy time when invasive angiography was preceded by CT angiography compared to tagged RBC scan (18 minutes vs. 28 minutes, p=0.002).  Compared to invasive angiography following tagged RBC scan, invasive angiography following CT angiography was associated with greater of identification of the bleeding source (46% vs. 26%, p=0.05) and embolization (40% vs. 23%, p=0.07). 

Compared to patients who underwent tagged RBC scan, CT angiography did result in the use of more intravenous contrast (220 ml vs. 130 ml, p<0.001).  However, this did not lead to any deterioration in renal function.  There was actually a trend towards improvedrenal function among patients receiving CT angiography (average peak creatinine was 200% of baseline following tagged RBC scan vs. 160% of baseline following CT angiography, p=0.09).  This is difficult to interpret.  It is possible that CT angiography led to faster hemostasis and better renal perfusion.  However, it is also possible that this correlation could be the result of confounding due to clinicians selecting tagged RBC scans in patients with elevated creatinine.  As previously discussed on this blog, it is unclear whether newer contrast dyes are truly nephrotoxic.

Most evidence on this topic is with regard to either CT angiography or tagged RBC scan alone.  The primary strength of this study is that it provides a pragmatic comparison of the two approaches at a single medical center.

One weakness of the study may be that it selected only patients taken for invasive angiography, rather than all patients admitted with lower GI hemorrhage.  This may provide a skewed perspective of this disease process.  For example, if a patient was admitted, had a negative CT angiogram, and subsequently exsanguinated and died prior to invasive angiography this event would not be captured in the current study. 

Another weakness is that this is a before-and-after observational study following implementation of a new protocol.  This study design is subject to confounding factors associated with implementation of a new protocol, such as increased awareness of the disease process and increased enthusiasm for its treatment.  It is also possible that technological improvements in invasive angiography during the study period (2005-2012) could have been a confounding factor. 

Finally, there appears to have been incomplete adherence with the protocol.  Even after adoption of the new protocol, 51% of patients received at least one tagged RBC scan and only 57% of patients had a CT angiogram.  Poor adherence may have reduced differences in average outcomes between the two time periods.  In attempts to overcome this problem, subset analysis was utilized to compare patients who received a tagged RBC scan versus CT angiography prior to invasive angiography.  Unfortunately, this retrospective subset analysis introduces additional confounding factors.

It would be difficult to perform a prospective RCT comparing CT angiography to tagged RBC scan for evaluation of lower GI bleeding.  Such a study has never been done, and is unlikely to ever be performed.  In the absence of a definitive RCT, we are forced to rely on less direct comparisons.  The University of Pennsylvania experience provides useful information about what a transition to early CT angiography might look like. 

There was no benefit to average patient outcomes following implementation of the protocol utilizing CT angiography.  This may relate to poor adherence to the new protocol.  Compared to tagged RBC scans, CT angiography had greater accuracy of identifying the bleeding source and led to a greater likelihood of finding the source of bleeding during invasive angiography.  Although CT angiography did increase the volume of intravenous contrast dye administered, there was no evidence that this caused renal injury.

This study was performed at a large medical center with the availability of nuclear medicine to perform a tagged RBC scan 24 hours a day, seven days a week.  Many hospitals with fewer resources lack this capability.  At a hospital with only intermittent ability to perform tagged RBC scans, a CT angiography strategy would offer greater advantages.

Overall this study supports a CT angiography-based strategy as a legitimate and evidence-based approach to lower GI bleeding.  With ongoing improvements in multi-detector helical CT scanners and safer intravenous contrast, we expect that the pendulum will continue to swing towards CT angiography as an immediate and definitive approach to evaluate a patient with critical lower GI hemorrhage. 

Related posts: