Does ketorolac increase the risk of bleeding?

Welcome back to Depth of Anesthesia. I'm David Hao and I'm an anesthesia resident. at the Massachusetts General Hospital. My guest today is Dr. Jamie Sparling. Dr. Sparling is a recent graduate of our anesthesia critical care fellowship at MGH, and she's a new faculty member here. Dr. Sparling, welcome to the show.

Thanks, David. I'm really excited you're doing this podcast and looking forward to examining the evidence together today. Awesome. So our case today is that of a 45-year-old lady who is presenting for an elective orthopedic case. The resident suggests giving Toradol, also known as ketorolac. for pain control, but the attending's a bit apprehensive and actually recommends against it because of the surgeon's concerns about operative blood loss.

Dr. Sparling, what do you think are some of the claims here in this case? Well David, I think there's both a primary claim as well as some supporting claims. The primary claim in this case is that Ketorolac increases bleeding and should be avoided in cases where we're concerned about operative blood loss. With that claim, there's a couple of subclaims. I think the first one is that administering ketorolac increases the bleeding time due to platelet inhibition.

The second one is claiming that this increased bleeding time translates to a higher rate of surgical bleeding. Also, that the magnitude of this bleeding propensity attributable to Ketorolac is actually clinically relevant, and lastly, that the risks of Ketorolac administration outweigh the potential benefits. So to our listeners, what's the level of your agreement with the claim that administering ketorolac increases bleeding and should be avoided in surgeries for which there is concern? And what is the level of evidence for what you believe? So we actually conducted a Twitter poll a couple days ago on our Depth of Anesthesia Twitter account.

And I asked the question, do you avoid Ketorolac if concerned about operative blood loss? We had over 70 votes and 42% said yes. 29% no, and 29% said only in specific surgeries. Dr. Sparling, do you think that's consistent with the clinical practice you've seen in some of your colleagues? I think overall, I think it's a really interesting discussion and something that probably varies, at least in our institution, between the anesthesia colleagues and the surgical colleagues I think that's a really fair point and definitely something that people will eventually bring up. We're ultimately sometimes not even the ones responsible for giving it because we'll suggest it and we'll get kind of denied by the surgeon who has a lot of concerns about bleeding.

Yeah, I think that's a really good point. So to start off, I think we should start by discussing the FDA black box warning. Dr. Sparling, could you talk a little bit about the black box warning for Ketorolac? Absolutely. So the black box label on Ketorolac is widely available for those of you who haven't seen it before.

You can find this either by looking on the FDA's website or also looking at the package insert for Ketorolac. The actual warning that's related to Ketorolac is, and I quote In post-marketing experience, postoperative hematomas and other signs of wound bleeding have been reported in association with the perioperative use of IV or IM dosing of Ketorolac. Therefore, perioperative use of tortol should be avoided and postoperative use should be undertaken with caution when hemostasis is critical. That's a fairly strong statement and I think coming from the FDA, it definitely is something that doctors and prescribers take pretty seriously. The evidence on which these statements were made was from a post-marketing surveillance cohort study of Toradol use, right?

Ketorolac use that was done back in 1996 by Strauman colleagues. That's correct. So we'll talk a little bit about this study. It was published in JAMA in 1996 and was a retrospective cohort study, including patients from 35 hospitals in the Philadelphia area. They included almost 10,000 patients.

And what they did is they looked at a comparison group of patients who had received parental opioids, that is, without Ketorolac. and they matched those two patients who had received Ketorolac, either with or without opiates. They were matched on the basis of the admitting hospital, the admitting service being either medical or surgical and the date of initiation of therapy. And that's one of the key points in this study, correct? Is that medical and surgical patients were included. That's correct.

So in terms of their methods, they looked at GI bleeding as well as operative site bleeding in the two different groups. And what they found was the operative site bleeding occurred in 39.6% of the Ketorolac courses. and 38.6% of the unexposed courses. So this was an odds ratio of 1.02. That was non-significant, with a confidence interval from 0.95 to 1.1 That seems super high, right? How did they decide what was considered operative site bleeding?

So it was pretty broad. The abstractors of this data reported as bleeding any mention in the patient's record of bleeding from a surgical site. So this could even include some blood spotting on the dressing. So did they do a subsequent analysis where they kind of pin that down a little bit into more serious or clinically relevant bleeding? They did.

They had a definition of what they called clinically serious outcomes, and those were bleeding scenarios that caused death. disability, prolonged hospitalization, or was considered to be life-threatening. For those groups, the odds ratio for the operative site bleeding uh was one point zero four. Again it was non-significant. And what do they find with the uh gastrointestinal bleeding So for GI bleeding, the odds ratio overall was 1.3, and that was significant with an increased risk among the ketorolac group It was also significant when they looked only at those clinically serious outcomes, and it was actually higher. It was an odds ratio of 1.44 So both the magnitude is larger and it is significant as compared to the operative site bleeding outcome.

Now did the authors do any sort of subset analysis on the types of surgery and in which kinds of surgery bleeding could actually be an issue? They did. They looked at particular surgical areas where there can be a heightened concern for operative bleeding Those included ophthalmologic surgery, neurosurgery, and plastic surgery. In each case there was no increased risk apparent from ketorlac versus the opioids only. And how about with age They did look at age as well.

In particular, they looked at the cutoff of greater than 75 years of age Interestingly, there was a market increase in both the Ketorolac as well as the opiate-only group, but this was more so in the Ketorolac group What were the actual odds ratios when they stratified based on age? So again for the GI bleeding, for those patients that were older than 75, the odds ratios was actually 1.7 uh for the Ketorolac group. And again that was significant. When I look at the operative site bleeding, those patients older than seventy five, there was an increased magnitude in the odds ratio. It was one point two seven.

But again, this still wasn't significant. Alright, so punchline for this study, large retrospective cohort, almost 10,000 patients with matching of patients who got toward Doll versus not. Seems like the use of IV Toradol contributed to more GI bleeding, especially in older patients, but this doesn't necessarily seem to translate to operative site bleeding. Is that correct? I'd agree with that.

And what were some of the strengths and limitations of this study? Well, I think the obvious strength is just that it's a huge study. It's almost 10,000 patients. But there are a couple of drawbacks which we brought up along the way. It was a cohort study, so you have some biases associated with that. um selection bias, information bias, and and misclassification bias.

One of the things we talked about as we went through the study is that their definition of operative site bleeding was rather generous. They included any patient with even blood spotting on the dressing, which you and I both know was probably most postoperative patients So I'd say that's definitely a drawback. I also think that looking at both medical and surgical patients together in the same cohort maybe makes it less interpretable in our population. And one of the other interesting things that I found in the course of reading through this study is that the drug labeling at that time of the study was had a recommendation of initial dose of sixty milligrams and then thirty milligrams every six hours after, which is you know a lot larger of a dose than we typically see nowadays.

That's that's right. So now that we've talked a little bit about where these concerns originated, let's cover that study that I know you really like, which is Uh the one addressing the claim that giving intraoperative tortill increases bleeding time Yeah, so I'm looking at a study by Singer and colleagues in the American Journal of Emergency Medicine from 2003. They actually looked at the effect of IM Ketorolac on some healthy volunteers. Interestingly, their healthy volunteers were EM residents. uh probably uh inspired to participate based on that helpful uh incentive money. Poor residents.

I know. It just doesn't seem fair, right? So their protocol, they gave each of these residents a single intramuscular dose of 60 milligrams of Ketorolac, and then they measured both the baseline and the four-hour bleeding times. They did a prospective paired unblinded interventional study in these subjects, and the number of patients was just 20. They had powered the study uh to detect an effect size of 0.5 And they were successful in recruiting their 20 patients.

The mean age, not surprisingly, was about 31 They did see an increased bleeding time with the Ketorolac administration. So the mean baseline was about three and a half minutes. And the four hour post-injection bleeding time was five minutes and twenty seconds. So that's a pretty significant increase. The the mean prolongation was one minute and forty-six seconds uh which seems like it it might be clinically relevant.

They didn't report any adverse events from the study. And the authors there had concluded that While Ketorolac administration certainly seemed to be associated with a prolongation of the bleeding time, it wasn't apparent how this would be clinically relevant or not. So punchline for this study, poor EM resins subjected to testing. It seems that bleeding times were prolonged after 60 milligram intramuscular administration of Toradol However, the clinical significance of this is unclear, but we do seem to have evidence that it does something. That's right.

So that's for healthy volunteers and what about in surgical patients, particularly where, for instance, a small increase in bleeding might be particularly clinically relevant or concerning for surgeons? Yeah, I think that's a great question. I want to look at a couple studies that look at those specific surgical subtypes. The first one is for lumbar decompression surgery. So I have a study here from Spine in 2008, and the author is Casanelli and Colleagues.

They looked at Ketorolac use for postoperative pain management following lumbar decompression. This was a prospective randomized double-blind and placebo-controlled study. It was small, only 25 patients, and they were randomized based on their age. So patients that were younger than 65 received uh either 30 milligrams of ketorolac or a m uh ml of saline, and those that were older than sixty-five either received fifteen milligrams of Ketorolac or or half a milliliter of saline. And there were some exclusion criteria, correct?

That's right. Importantly, they excluded all patients that were currently using anticoagulants or antiplatelet agents. They also excluded patients that had a history of gastric ulcers and those that had renal dysfunction. And how did they give the medication? The protocol was as follows.

The first dose was given at the time of surgical wound closure. They give a second dose, um the same dosage, at six hours post-op, and a third dose at twelve hours post-op. And what did they find? Yeah, their primary outcome that they looked at was actually a efficacy outcome. So they looked at the amount of IV morphine requirements required for each patient's entire postoperative course.

They also looked at postoperative pain levels and the rate of any complications, including bleeding. So what did they find in terms of the primary efficacy outcomes? They found that the Kachorlac group um had lower morphine requirements at each of their predetermined time points. So six, twelve, and twenty-four hours. as well as their total morphine consumption for the hospitalization. More interestingly for us, they looked at bleeding complications and the amount of postoperative blood transfusions They saw only three blood transfusions that were necessary, and these were actually all in the placebo group, not the catalog group They also had one major complication, which required that they broke the protocol and they unlighted the patient.

That patient had developed an epidural hematoma Interestingly, they also weren't the placebo group, not cachorlak. And what about drain output? Yeah, I thought this was a really um interesting way to look at the bleeding risk. Um they measured the drain output at 8, 16, and 24 hours post-op. In again, um maybe ironically, they they found a trend towards decreased output in the Ketorolac group at each of those time points.

Gotcha, so punchline for this, 25 patients going for lumbar decompression. It seems like there's no identifiable post -op complications seen with Ketorolac use. in this particular surgery, but again, it's relatively small and might be underpowered. I'd agree with that. What was the other surgical subset that we were gonna take a look at? Yeah, we were going to take a look at tonsilectomy patients.

And we have a randomized double-blind study. This is from the laryngeoscope in 1997 and its Bailey and colleagues. They looked at the risk of ketorolac and hemorrhages and tonsilectomy. So this was a prospective randomized double-blind study They randomized 95 adult patients. Of those, 37 were randomized into the Ketorolac group, 43 into a group receiving Maparidi.

And the other 15 were protocol violations, either because they didn't receive the intraoperative dose or because they didn't request any postoperative analgesia They excluded all patients that had some sort of bleeding diathesis that were hypersensitive to any of the protocol medications. Had a history of peptic ulcers, any history of visceral bleeding, any act of anticoagulants or recent NSATs as well as those that had an operative time longer than two hours or had estimated blood loss greater than three hundred. And I like what they did in this study in terms of standardizing the technique by which they would get hemostasis They actually clarified that they would do electrocautery technique with suture ligation, and that kind of gets to the issue of different surgical techniques causing more or less uh hemostatic complications.

Yeah, I think it was really important that they standardize that. So in addition to standardizing the surgical technique, they also standardized the anesthetic technique. So these patients were all intubated with propofol for induction and for maintenance. They received alfentanyl, a bolus up front, as well as a maintenance dose and they had nitrous oxide going in a 66% mixture with oxygen. They added isoflurane to that when needed and they didn't give it patients any pre-medication, either in terms of analgesia or otherwise.

And what about the exclusionary criteria of an EBL greater than 300? How should we think about that? I think in patients that already had an EBL greater than 300 CCs Obviously there's some other either surgical complication or underlying comorbidity that's causing that. What were the outcomes that this study looked at? So the primary outcome was postoperative hemorrhage, and they defined this in a pretty clinically relevant way.

They included cases where they required a return to the OR for hemostasis Were they required an inpatient admission or for those that were treated with an intervention as an outpatient? The second outcome they looked at was the bleeding time, and that was available on a subset of their patients And what do they find with this study? In terms of the incidence of hemorrhage, uh they did see a trend toward increase in the Ketorlac group. Uh the rate was 18.9 %. or seven out of the thirty-seven patients in that group. In the Baparidine group it was only seven percent, so only three incident of hemorrhage.

And although there was this trend, it was not statistically significant with a p-value of 0.1 Now in terms of the bleeding time, this data was available on 48 of those 80 patients. Overall, they saw a significant increase in the bleeding time with the ketorolac group Wherein the the bleeding time increased from 5 minutes and 19 seconds to 6 minutes and 27 seconds, this increase wasn't present in the Maparidine group. Alright, so take away from this study, we have a randomized double-blind study of Toradol and hemorrhage and tonsilectomy bleeding rates seemed to be slightly higher in the Toradol group, but that was not significant. And ultimately even though there was a increase that was significant in bleeding time that again does not seem to translate into a clinically significant increase in hemorrhage.

That's right. I think this is similar to some of the studies we looked at already, where there is an objective increase in terms of bleeding time, but the clinical relevance of that isn't necessarily clear What do you think about this study's conclusion, which I think is interesting because it seems to contradict what they actually found in terms of the actual analysis? Yeah, I think the authors were concerned about this trends towards increased bleeding in this particular group. Um, and certainly when you look at the magnitude of the increase, uh that's concerning uh 7% versus about 19%. I think that certainly it was a small study, only 80 patients, and There were many that did not adhere to the protocol.

Um so I think certainly looking at larger studies uh to draw the conclusion in terms of the safety of Ketorolac administration in tonsilectomy is appropriate. And that's awesome because I think we do have a meta-analysis in which we talk about tonsilectomies as a surgical uh subset. But first I think we'll cover the meta-analysis in which we hit across a wide range of surgical populations. We'll talk about that and then we'll go into the tonsilectomy paper. That sounds great.

Awesome. So this meta-analysis really doesn't hide at all its conclusion. It's literally named Ketorolac does not increase perioperative bleeding. But I think the nice thing is is this this is essentially a meta-analysis only of randomized controlled trials and includes the two papers we just discussed That's right. It actually includes 27 different studies, so pretty significant both depth and breadth of trials that are out there.

They included a wide range of operations, as you mentioned, including neurosurgical, abdominal surgery, urologic surgery, and orthopedic surgery. And what they looked at was the odds ratio of clinically significant postoperative bleeding. They defined that as requiring postoperative transfusion, readmission, or reoperation. And then they also did a subgroup analysis for high versus low dose ketorolac, which they defined as greater than or less than 30 milligrams. And what do they find with this meta-analysis?

Yeah, so overall they found a rate of postoperative bleeding of around 2.5 % in the Ketorolac group versus 2.1% in the control group. But this odds ratio, which was 1.1, uh, was not significant for the overall meta-analysis. They also looked at, as I mentioned, the low and high dose, and it's interesting when I'm looking at the forest plot here Every single study, um, in both the 30 milligrams or less versus the high dose, every single one crosses an Oz ratio of one. So none of them were significant. And how about adverse events in general?

They did comment on a pretty heterogeneous range of side effects like nausea vomiting, respiratory depression. What do they find when they actually looked at those as well as bleeding? So when they looked at the composite rate of adverse events, being those three that you just mentioned, so nausea, vomiting, respiratory depression, and bleeding. They actually saw that the low-dose Ketorolac group was significantly lower than controls, and that there was no difference or no statistically significant difference for the high-dose Ketorolac group. The strength of this meta-analysis obviously is that only double-blind randomized controlled studies were included.

What are some of the drawbacks or things that we should think about when applying something like this? Well, I think the first limitation of the study is inherent in any meta-analysis, and that's just the heterogeneity. So you look at the list of studies that are included There's a wide variety in terms of the surgical procedures, the patient populations, and also the protocols that were used in terms of Ketorolac administration. I also think that with a meta-analysis, it's pretty difficult to sufficiently power your study to detect an increased or decreased risk of the postoperative bleeding of the other adverse events uh with the use of Ketorolac. And then the last thing I'll say is that I'd acknowledge that In the higher dose ketorolac group in particular, there was a trend towards higher rates of bleeding.

It just wasn't significant. And I think that just speaks to the care we should take in general with extrapolating results that may have certain limitations and applying that broadly across clinical practice. I think that's exactly right. Alright, so uh you want to give us the punchline for this one? So in this meta-analysis, using a wide variety of surgeries and different patient populations.

They saw no increased risk of bleeding in either the high or low-dose ketorolac group. Further, they found a reduced risk of adverse events in the low-dose ketorolac group uh the low dose ketorolac group as compared with controls. Awesome. And one last thing that we forgot to mention, what about elderly patients? Yeah, in this particular meta-analysis, they weren't able to determine whether elderly patients had an increased risk because many of the studies in this review didn't stratify based on age.

They also thinking back to that original post-marketing study that informed the FDA black box warning, they had looked at catalog use greater than five days And none of the protocols included in this meta -analysis actually had that long of a duration. I think that's a def that's definitely a helpful point to make. So heading on to our last paper, this is a meta-analysis looking specifically at tonsilectomies, which is seems to be a surgical subset in which there's a trend as reflected by the previous study that we just talked about. I think that's fair, and I don't know about you, but as an anesthesiologist, the thought of a bloody airway gets my catecholamines going. So I think it's appropriate we end on this one.

This was a meta-analysis that was also published in the Ryngoscope in 2014 by Chan and Pereka. They looked only at studies that involved tonsilectomy patients. They looked at 10 studies. Seven of those included only children. Two included both children and adults, and one was exclusively adults.

Of those studies, they define postoperative bleeding differently. That ranged anything from noticing any bleeding by the patient to bleeding requiring a return to the OR for control. And it's interesting to think about this compared with a 2010 Cochran review where they had looked at NSAID use in children undergoing tonsilectomy That Cochrane review had reported quite a high odds ratio for the NSAD group of 3.1, however, it was not significant. And interestingly, for that Cochrane review, they had only included a hemorrhage if it required a return to the operating room So our definition of post op lead here is uh a more generous or a more inclusive one. One of the other differences I observed is that the Cochran review included only randomized controlled trials, whereas here it's a mixture of retrospective and prospective studies, right?

That's correct. So when they took these ten studies as a whole, they did see an increased relative risk of postoperative bleeding for the uh overall group including both adults and children, a relative risk of about two. Now, interestingly, when they looked at adults versus children They saw a high relative risk of 5.6 for adults. So they were five times more likely to have a post-tonsilectomy hemorrhage compared with those patients who didn't receive Gatorlac. Contrast that with children, for which there was a relative risk of 1.39, but this was not significant with a p-value of 0.2.

It was consistent both looking at the retrospective and the prospective studies in these subgroups. Were there any particular studies that seemed to drive the analysis? You know there were. When I'm looking through the list of trials that were included here, I'm seeing the uh Judkins trial from 1996 I had the largest group both of pediatric patients as well as a relatively large group of adults. So 200 pediatric patients and 53 adults When you look at the relative risk from that particular trial, it reports a relative risk of almost 15 for the adult groups And with being a larger patient cohort, um that seems to really be driving some of the um results of the pulled analysis.

There are also a couple confounders within this meta-analysis that the authors readily identify. What were some of those? Yeah, some of the uh limitations here uh were that you know the blinding wasn't always sufficient. So if you think through this, only seven of the trials were prospective Three of them were retrospective case control trials, and so I think this limits our validity and introduces some bias for sure. The other thing I saw is that in terms of follow-up there was some heterogeneity and with the prospective studies in particular It seemed that follow-up was only for one or two days, so there's a potential for hemorrhage after those days, and that's not being picked up in these studies.

What about the issue of power? So when I look at the number of children versus adults included here Um the power for the pediatric analysis is certainly better, so they included 544 children in the pooled analysis compared with only a hundred and twenty-five in the adult. Um I think this introduces the possibility um not only that we are underpowered for the adults, but also potentially for the children. It it's certainly theoretically possible that there is a significant increased risk of bleed with Ketorolac, but because we're looking at a meta-analysis that this may have been under powered for the pediatric. Awesome.

Doctor Sparling, you want to give us a summary or a punchline for this study? From this meta-analysis, the risk of bleeding with gator lac for adults is five-fold increase. That risk of bleeding is not demonstrated for ketorlac in children, but maybe that words further study. And this is in post-tonsilectomy patients specifically? That's correct So I think we covered all the literature that we had planned for today.

So for our listeners out there, just to go back to our original claim. The overall claim here was that administering ketorolac, also known as tortol, increases bleeding and should be avoided in surgeries for which there is a concern for bleeding. So for those of you listening out there, what's the level of your agreement with that claim? And what is the level of evidence for what you believe? Dr. Sperling, do you want to summarize some of the key points and studies that we've gone through today So thinking back to that first post-marketing surveillance study, they saw an increase in the bleeding events with Ketorolac administration, but that was largely driven by GI bleeding, not postoperative bleeding.

And furthermore, the magnitude of the increase was very small and not clinically relevant. That being said When we looked at the bleeding time, there was an objective and empirical increase with administration of Ketorolac even in healthy volunteers. I think the question is how do we apply that to our surgical population? When you look at these meta-analyses that examine patients across different types of surgeries, there really is no statistically significant increased rate of bleeding. uh at least not in the protocols and in the populations that we studied here. The one exclusion to that I think is what we just discussed The risk of bleeding after tonsylectomy, that's something that we all care about because none of us want to encounter a bloody airway in the Pacio In this meta-analysis that we just discussed, they saw both a statistically significant and a clinically relevant increase in bleeding in adults that had received ketorolac during their tonsilectomy operation.

That wasn't true for children, but again, maybe that warrants further examination. Dr. Sparling, I think that was an excellent summary of the literature that we just reviewed today, and thanks so much for joining us on the show Thanks for having me, David. And just as a teaser, I know you're going to be back in hopefully a couple weeks where we're going to do peripheral norepinephrine. I'm excited for that one and we're going to bring Katarina Rusik along for that as well Awesome. So our listeners, thanks for joining us on the show this week.

As always, you can reach us at Depth Anesthesia on Twitter where we conduct polls and share interesting studies. And thank you to everyone who's been emailing me with recommendations on future episodes. Hopefully we'll be able to get around to covering one of those topics. As always stay hungry and keep asking questions.