The WashU group in St. Louis compared their initial CMP III series performed by J. Cox with the ablation assisted CMP IV performed by R. Damiano (all stand alone AF). The rate of freedom from AF did not differ significantly while procedure time and perioperative morbidity was decreased in the latter. FU was done according to the guidelines, while more than half of the CMP III pts even received an EP study after surgery.

Overall, this data showed that the CMP IV using Cryo and bipolar RF energy did not compromise in success rates compared to cut&sew if performed by dedicated surgeons.

You find this paper in my list.

It is difficult to compare CMP III and CMP IV studies because important definitions (e.g. "paroxysmal" AF) and methods (how do you determine whether a patient is AF free) have changed in time and generally become stricter (i.e. they make the results of newer trials look worse).   When comparing study results, it is important to also look at the patient populations, since some characteristics (eg. AF type distribution, duration, LA size, and patient age) have a strong effect on the outcome.

As Timo Weimar wrote above, papers from the WashU group suggest that freedom from AF does not differ significantly if the CMP IV is performed by dedicated surgeons; if surgeons are not dedicated, the CMP IV may fail more frequently because both RF and cryoablation have pitfalls that can lead to incomplete ablation lines (which are difficult to detect at surgery time).

A recent paper that discusses this topic is the manuscript on the ABLATE trial (see my publications).

As already mentioned by Timo Weimar and Christian Zemlin, times have changed since the Initial phase of the MAZE III. Controls have become more strict, especially since the introduction of the event recorders and - not every MAZE IV is really performed completely, as described by J. Cox with the complete lesion set on the left and right side!

With a complete lesion set as well as a complete transmurality, the results of the two procedures should be comparable. 

I do thank all the answers so far. They were extremely thoughtful and helpful to clarify some important issues. To be honest, I am writing a manuscript showing my personal experience with the Cox Maze III procedure which I still perform as described by James Cox. The reason not to move to Cox Maze IV was due to economic restraints imposed by the Brazilian government. There are some concerns I would like to share:

Safety: The greater complexity and presumed higher risk associated with Cox Maze III was the main reason for not being more frequently performed. The Cox Maze IV brought a more friendly and quicker surgical approach, which allowed an incremental number of procedures. However, the Cox Maze IV still keep a similar concern among the cardiac surgery community since more than 40% of patients in need are not treated in North America. Safety is a major issue. It relates to patient severity profiles, and to the procedure itself. It is concerning that, both at the NEJM and the ABLATE trials, the rate of pacemaker implantation and MAE appear to be greater than historic Cox Maze III controls (despite of not being comparable groups). In addition, cardiopulmonary bypass and cross clamping times with CMIV seem to be excessive (giving the simplicity of the ablation as compared to the cut and sew technique), because they are close to the CMIII group when associated with other cardiac procedures. What is the reason for that?

Efficacy:  I understand that we have today a more precise and strict rhythm assessment. Were Cox Maze III reported outcomes not as good as it gets? The development of a definition of treatment success allowed the definition of late recurrence predictors. The issue of transmurality lesion set is of paramount importance in order to get an adequate result. This is never an issue with Cox Maze III. The Mayo Clinic group have analyzed their experience over two decades using different approaches (Stulak et al. JTCVS 2014) and they found that Cox Maze III was associated with less late recurrence than any other surgical approach using multivariate logistic regression. I do agree with Dr Zemlin"s comment in regard to the level of dedication and the final result. However, I assume that for most of the trials it is supposed to have listed the most committed and gifted surgeons. Results seem to be less than expected. That's the reason it is worth asking whether the Cox Maze III should still be performed in selected subset of patients (younger, or higher risk of recurrence) in selected centers of expertise with the technique.