The CONVINT trial is the latest to compare CVVHF to IHD in the critical care population.
The CONVINT trial therefore should be forgiven for being monocentric. They conducted the trial over nearly 6 years, and needed 200 patients per arm. Unfortunately, they did not achieve their target either (128, 122). The reason given was “a major change in RRT equipment and procedures beyond the investigators control”, which included a shift towards haemodiafiltration. You can only imagine the mood in the meeting when that was decided! If nothing else the evidence in this area demonstrates what a thankless and impossible task research is.
There were no differences in any outcome between the groups. The outcome is given as an odds ratio, which it may be worth refreshing your memory about. The concept of odds can be confusing despite it being a term used freely in conversation.
The odds of an event are different to the risk of an event. Where risk is incidence / no of events, odds are incidence of event / incidence of non-event. Risk is dependent on sample size, which in a cohort study (such as this) is determined by the trial design and therefore can’t be used. For this study, the event is death at 14 days after RRT, which occurred in 60.5% of the IHD group and 56.1% of the CVVHF group. The odds of death in the IHD group are therefore 60.5/39.5 = 1.53 and in the CVVHF group 56.1/43.9 = 1.28. The odds ratio is the ratio of these numbers (which can be done either way round in this study because there isn’t really an intervention and control group). The ORR therefore is 1.53/1.28 = 1.20 or 1.28/1.53=0.84. To give a confidence interval for an OR there’s a formula, or a stats package will just churn it out. For this OR of 0.84 the 95% CI is 0.49 to 1.41. This means we are 95% sure that the true OR is between these numbers. Any CI for OR that includes 1 will be non-significant, so the p value of 0.50 is not surprising.
This study used intention to treat, which as you know means that whichever group you are randomised to is the group your results belong to. Intention to treat analysis is generally considered to be a good thing. In this study however it’s worth noting that 46% of the CVVHF group crossed over to IHD (what do you think the implications of this might be?). The most common reason was a desire to mobilise or clinical improvement (39%), followed by repeated filter clotting (27%). This highlights some of the reasons we like IHD; it frees the patient up from the machine, there’s no need to switch modality, and there’s not a filter clotting every 10 minutes. On the latter note, 88% of prescribed dose was delivered, which is pretty typical in studies involving CVVHF. While I’m on my soapbox, I also like that we have our RRT overseen by a renal medicine consultant so the patient gets early expertise and continuity of care if still requiring RRT at discharge – any effects of this on outcome will not be included in this study.
Another quite interesting point about this paper is that approx. a quarter of the patients were still needing / getting RRT 60 days after their first treatment. That is higher than the number I quote to families, and wonder if we should look to see what our numbers are – volunteers please!
As always I’d appectiate your comments. What do you think are the advantages and disadvantages to way RRT is delivered around the region? Should we use CVVHF like everyone else? Should everyone else use dialysis? Does it matter at all?
Addendum (6/1/15) - Andy Morrison has given me his presentation entitled "RRT in critical care" which goes a little bit more into some of the other work in this area - it can be found here.