Heparin or citrate? When prescribing continuous renal replacement therapy (CRRT), this question is often encountered. In a recent article appearing in the American Journal of Kidney Diseases, Drs. Mei-Yi Wu (MW) and Ka-Wai Tam (KT) discuss their systemic review of heparin versus regional citrate anticoagulation for CRRT with eAJKD advisory board member Dr. Joel Topf. The review looks at 6 randomized controlled trials (RCTs) done on these 2 agents.

eAJKD: What motivated you to investigate this particular topic?

MW: We use heparin as the anticoagulant agent in our intensive care units, however this entails a significant risk of bleeding. Bleeding while anticoagulated for CRRT is reported in 5%-25% of patients. This question has bothered us for a long time. Reviewing the literature, we found several randomized controlled trials and prospective studies comparing citrate with heparin anticoagulation in critically ill patients. Most of these studies found citrate was associated with similar or prolonged dialysis “circuit life” and less bleeding with lower need of transfusions. We felt a meta analysis to better understand the literature and the use of this practice was worthwhile.

eAJKD: Do you remember a particular patient with a bleeding complication or heparin induced thrombocytopenia (HIT) that prompted your interested in this subject?

MW: Yes, there was one patient who developed an intracranial hemorrhage while on ECMO and CRRT and died due to multiple complications. We had to discontinue the heparin due to active bleeding, but clotting limited our ability to then provide adequate dialysis.

eAJKD: Has this topic been studied before?

KT: There are 191 studies looking at the efficacy and safety of heparin or citrate anticoagulation in CRRT. Eighty-eight studies had no relevant comparison and focused on the dosage of heparin, comparisons of prostaglandin with heparin, and various CRRT techniques. For our systematic review, we limited ourselves to randomized controlled trials that focused on the question of “heparin vs citrate.” We initially found 11 studies (RCTs and prospective studies), but in our final analysis decided to go with only 6 RCTs.

eAJKD: Were the conclusions any different with the 5 additional studies you didn’t include?

KT: The 5 excluded studies focused on circuit survival time. Major bleeding and metabolic alkalosis were not their main findings. In circuit survival time, citrate fared better than heparin. However when pooling those data in a forest plot, it was still not statistically significant.

MW: Randomized controlled trials in critically ill patients are difficult in clinical practice. Nevertheless, we think the 6 RCTs provide valuable clinical information to use in our practice.

eAJKD: I was underwhelmed with the quality of the studies. By limiting your analysis to RCTs you guaranteed a minimum Jaded score of 1, and yet the average score was only 2.3. No study had a score of 4 or 5. Any reason why?

KT: Jadad scores are not my favorite tools to assess study quality. I prefer CAT maker, CASP, and Cochrane risk of bias table. Given its wide use and the suggestion of a reviewer, we used the Jadad scores. In the new version of the Cochrane Handbook, they suggest that the main point in critical appraisal is judgment and not scoring.

eAJKD: In your judgment, how good do you believe the studies were?

KT: The methodological quality of the included studies is not very good. In these types of studies blinding is very difficult and usually abandoned, and the randomization is unclear. However, the main outcomes (circuit survival time, laboratory data, and the incidence of bleeding) are objective outcomes, making the lack of unblinding acceptable.

MW: In regards to randomization, this was especially problematic in 3 studies. I agree that there is a problem.

eAJKD: Speaking of the difficulty with ill ICU patients, how did the studies account for patients that expired with functioning filters? This is particularly relevant since at least 1 study demonstrated increased survival with citrate. Was the improve filter survival a function of improved patient survival?

MW: In the real world, patients die with functioning filters. In the 6 RCTs we reviewed it was unclear how the authors dealt with circuit survival at the time subjects died. It gets back to study quality. Some of the studies did provide the reason for circuit disconnection (eg, circuit failure, catheter failure, transportation, death).

eAJKD: Can you explain why one of the studies could not be included in the end-point analysis of filter life?

KT: It did not provide the standard deviation. It only mentioned mean time.

eAJKD: Wow, something that simple. Do you think journal editors need to enforce higher publication standards to ensure the feasibility of future meta-analysis? Especially in a field plagued by small, underpowered studies?

KT: That is why evidence based medicine is so important. If the editor makes an effort to control the quality of each study, we will be able to build a more robust body of evidence.

eAJKD: A number needed to treat (NNT) of less than 7 for major bleeding was noted. How do you interpret this?

MW: Bleeding with anticoagulation for CRRT is reported in 5%-25% of patients. The incidence of bleeding with heparin is:

37% in Betjes
0% in Fealy
15% in Hetzel
50% in Kutsiguannis
8.3% in Monchi

but we don’t have as much clinical experience with citrate. Although the bleeding incidence rate varies widely, this may be a function of the small sample size. Nevertheless, citrate decreased incidence of bleeding compared with heparin in each study. Despite exclusion of patients with high bleeding risk in several of the RCTs, citrate still demonstrated less bleeding compared with heparin. We believe in the marked benefit of citrate.

eAJKD: Did you consider looking at mortality differences?

MW: Irrespective of the advantages of citrate with regard to bleeding, we cannot confirm a beneficial effect of citrate anticoagulation on mortality. In many studies, sepsis was the predominant cause for death. There are a variety of pathophysiological mechanisms that lead to organ failure in septic patients. The complexity of the underlying disease brings to question whether any single procedure will ever have a clear-cut effect on mortality. Therefore, it is understandable that we did not see an effect of anticoagulation protocol on mortality.

KT: We didn’t use mortality as our primary or secondary outcome. This is due to the heterogeneity of the included patients among studies (illness severity). Also, the assessment of mortality among studies was different, eg, in-hospital mortality vs 3-month mortality. Mortality is mainly affected by the patient’s illness itself, not side issues like circuit time, bleeding, and hypocalcaemia.

eAJKD: Does this analysis, despite its short-comings, close the book on anticoagulation in CRRT? What is the state of equipoise regarding this question?

KT: Our study demonstrates the advantages of citrate as the anticoagulation agent in patients who require CRRT but are at high risk of bleeding. Nevertheless, clinical decisions always require a careful assessment of individual patients’ risks and comorbidities.

eAJKD: What type of study would you design to move the science on this question forward?

KT: A study that used a uniform protocol in critically ill patients with a more uniform diagnosis, for example, post-CABG patients or septic patients would clarify the benefits of citrate. A large-scale cohort study or a well-designed RCT can be performed using the same outcome analysis.

To view the article abstract or full-text (subscription required), please visit AJKD.org.