In this part, we focus on the influence of the initial clinical findings on the effective failure rate, that is the rate of lost eyes.

Let’s start with the PVR data. This is the RDs distribution by PVR stages. Stage O was understood as cases without any PVR and without RP cells in the vitreous.

As we could expect, the more developed the PVR stage, the higher the failure rate declared by the surgeon as well as the number of eyes with remaining silicone on June 2011. As far as effective failure rate is concerned, there is no significant difference neither between stages O and A nor stages A and B but the difference is significant between 0 and B. The rate of success after only one surgery reaches 83% for stage 0 and drops down to 63% for stage C1.

These are the aggravating factors which were observed among the 7678 RDD.

The presence of aggravating factors seems to worsen the final failure rate. For instance, in comparison with the absence of any aggravating factor, the effective failure rate is three times higher for eyes with hypotony. However, these differences might come from the PVR distributions of the aggravating factors.

It is indeed obvious that the cases with vitreous hemorrhage, hypotony and choroidal detachment developed higher PVR stages than the cases without. More than a half of choroidal detachments are PVR stages B or C1 versus less than 30% of these cases. Thus, in order to show a better intrinsic impact of the presence of aggravating factors.

We compared the failure rates at each PVR stage. This bivariate analysis shows that the presence of vitreous hemorrhage, the second column, does not seem to worsen the failure rate and cannot be considered as a major independent explanatory variable of the failure rate. Besides, for each PVR stage the presence of either hypotony or choroidal detachment, column 3 and 4, increases the effective failure rate. This trend is especially visible for choroidal detachment which is linked to an effective failure rate 20 times higher in stage 0 and around 10 times higher in stage A and B.

When grouping together the hypotony and choroidal detachment, the trend remains the same for each PVR stage; This is true for effective failure as well as for remaining silicone. Except for stage A, the p values show that the differences are small enough to consider that these differences are significant.

The presence of hypotony or choroidal detachment is thus considered a major independent explanatory variable of the failure rate. This was confirmed by the multivariate analysis.

If we now go back to PVR considering the failure rate without the impact of choroidal detachment and hypotony, we see that stage C1 is a major explanatory variable of the failure rates and therefore should be studied separately. In addition, since the difference between stage B and Stage O is significant.

We can confirm that PVR stage can be considered as a main independent explanatory variable of failure rate. The multivariate analysis shows that only PVR stage C1 is a major independent variable of the effective failure rate.

Here is the distribution of the RRD by lens status. There is a majority of phakic cases and only a few aphakic cases.

There is no visible difference between the 3 lens status concerning the type of retinal breaks and the mean number of breaks.

Aphakic eyes have far more detached quadrants than pseudophakic eyes which have themselves a little bit more detached quadrants than phakic eyes.

When looking at the failure rates we can notice that there is no significant difference between phakic, pseudophakic and aphakic eyes for true failure rates. However, there is a significant difference concerning remaining silicone rates.

Now if we take into consideration the PVR stages they appear to be significantly more severe in aphakic cases than in pseudophakic cases, themselves significantly more severe than phakic cases.

And if we exclude the aggravating factors.

We can see that there is no significant difference between phakic and pseudophakic eyes concerning the true failure rates. The difference for remaining silicone rates between phakic and pseudophakic eyes is still significant for a given PVR stage and can be explained by the fact.

That more vitrectomies are performed in pseudophakic eyes, then more silicone oil is injected and therefore more cases have Remaining silicone.

With these univariate and bivariate analyses, the lens status appears therefore not to be an independent explanatory variable of the failure rate. As it will be shown at the end of this presentation, this result was confirmed by logistic regression (multivariate analysis).

This is the distribution of RRD by the number of detached quadrants.

The higher the number of detached quadrants, the higher the number of breaks.

The failure rates increase with the number of detached quadrants. These differences are significant at a level of 5% except between 1 and 2 quadrants, and 2 and 3 quadrants for true failures. In order to see if choroidal detachment and hypotony or the PVR stages could explain these differences, the distribution of these parameter for the number of detached quadrants were studied.

There are more aggravating factors if 3 quadrants are detached and even more if 4 quadrants are detached.

The difference between the number of detached quadrants is even more obvious when correlating it with the PVR stage distribution. When more quadrants are detached the easy stages 0 and A become less frequent and the more severe cases B and C1 are found more frequently.

So that if we analyze the failure rate according to the PVR stage after excluding the aggravating factors the influence of the number of detached quadrants appears less evident.

The final multivariate analysis confirmed the 4 detached quadrants as a major independent explanatory variable of the failure rate.

This is the distribution of RRD by the type of the worst retinal break. Most of them are normal size tears.

We had some doubts on the accuracy on the retinal break description the surgeons made. For instance, it is not normal to see so many holes (which normally are atrophic holes) with retraction and even more with stage B as atrophic holes do not have rolled up edges. This is why we studied only the tears, small large and giant.

The failure rates are significantly different between normal, large and giant tears, except for the true failure rate between Large and Giant tears with a p value of 0.63. This is why, and to face the fact that there were only a few number of giant tears, it has been decided to group together Large and Giant tears for the rest of the study.

At constant PVR stage, for stages B and C1, the effective failure rates are significantly different between normal size and big size.

So we can say that tear size is a major independent explanatory variable of the failure rate. This result was confirmed by the multivariate analysis.

Let’s finish with the number of breaks.

The distribution of the PVR among the number of breaks does not provide any useful information.

There is no obvious link between the number of breaks and the failure rate.

In conclusion the number of tears is not considered as an independent explanatory variable of the failure rate.

The goal of the multivariate analysis is to isolate independent variables that have an influence on the true failure rate and on the rate of remaining silicone. This was performed with a logistic regression, based on a step by step approach. The variables initially considered were all stages of PVR, the different lens status, the number of detached quadrants, the size of the break and the presence of aggravating factors.

The study shows that 3 variables remain independently linked to the true failure rate : PVR stage C1 which increases the effective failure rate. So does the presence of 4 detached quadrants and the presence of aggravating factors CD or H. The odds ratio is a rough estimate of the relative risk : an operation on an eye with choroidal detachment or hypothony has twice as much risk to be a failure than without CD or H.

Concerning the remaining silicone rate, the study shows that 7 variables remain independently linked to this failure rate : Lens state Ps, PVR stages B, C1, 3 , 4 detached quadrants and the presence of choroidal detachment or hypotony increase the risk of remaining silicone. On the contrary, a small break decreases the risk of remaining silicone.

So these are the major independent variables explaining the failure rate. There are a lot of variables that we did not checked as the pupillary dilatation, the vitreous turbidity, the posterior hyaloid statute etc. The problem is that asking more details to be answered, we would have decreased the numbers of sent cases.