Home The Book Dr Articles Products Message Boards Journal Articles Search Our Surveys Surgery ESWT Dr Messages Find Good Drs video

chances for TTS surgical success: 42%, 100%, pick a number!

Posted by elliott on 10/24/01 at 16:06 (063557)

The first abstract below claims surgical success in 85-90% of patients. The second (a full article in addition to its abstract) claims 42% success even after careful screening for suitable surgical candidates. The third says the literature gives surgical success rates of 44-100%, and its own authors experienced (amazingly, almost exactly the average of this range) 71%.

Note: It seems like they are generally defining success as being better off than before the surgery. Fair enough. But which of the stats are more realistic? You feeling lucky?

One possible explanation is that offered in the full article link, namely that surgical success rates may initially appear to be around 71%, but with appropriate followup done after a decent interval and based on the patient's assessment rather than doc's technical assessment (e.g. patient says he/she is in agony and it's unbearable and worse than pre-surgery vs. slightly improved post-surgery nerve conduction results), it may be more like 42%. Either that or they just don't know how to do TTS releases in the UK. :-)

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=11499170&dopt=Abstract

http://www.rcsed.ac.uk/journal/old/4310013.htm

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8714673&dopt=Abstract

--------

Re: chances for TTS surgical success: 42%, 100%, pick a number!

cindyp on 10/24/01 at 16:41 (063558)

come to america

Re: chances for TTS surgical success: 42%, 100%, pick a number!

elliott on 10/24/01 at 18:29 (063567)

I'm already here. Lotta good it did me.

Re: Some reasons for variability

Ed Davis, DPM on 10/24/01 at 22:53 (063583)

Elliot:

Note that the last citation, the retrospective study, only included 40 patients. That is a fairly small sample size. Smaller sample sizes lead to increased variability.

Other things that create variability include the means by which improvement or satisfaction is measured (as you have pointed out) and mixtures of causes of TTS. Some cases of TTS have a specific structure causing compression on the nerve and some do not. Ideally, studies should separate those two groups. Another factor is the amount of time that TTS was present. I do not think that a 'point of no return' has ever been established but it seems reasonable to expect that nerves that have been damaged longer would have a poorer chance of recovery---at least this variable should be considered by the studies. I have generally found that due to the relative rarity of the problem, that cases of TTS tend to get diagnosed late.
Ed

Re: actually...

elliott on 10/26/01 at 11:28 (063622)

40 is plenty big enough if there really is some a priori probability of success, for then we have a binomial distribution, whose std dev, SQRT(npq), is rather small compared to its mean. Furthermore, I think I recall it takes a sample size of only around 30 for the binomial to be approximated very well by a normal distribution (central limit theorem and related results), which is often convenient for calculations and journal purposes. Assuming the real probability of success is .8 as per one study, the probability of seeing at most 16 successes out of 40 (the 40% as per the other study) is so close to 0 using the normal approximation to the binomial that it goes off the printed table charts. Using brute force on the binomial (assuming I made no mistake) gives 0.0000000352, or about 4 in a hundred million. And supposedly, the 40% study was limited to a set of patients supposedly more suitable for surgery, scary when you think about it.

Of course, I don't deny that ceteris paribus, a larger study is better, nor that all the other things you mentioned (differing levels of followup, definitions of success, netc.) all play a huge role in altering it from being a pure binomial, probability-of-success type of problem. But if conditions were reasonably similar--and they can be--after a few consistent studies they should be able to glean info into such things as when or when not surgery is recommended. I wish the medical field were more rigorous in this regard, rather than come up with all kinds of stats all over the place.

By the way, kudos to you for changing the post title from its default! First time I can remember. I wish everyone would do that more often, as it helps to visualize the flow of the thread, as well as to see who added what posts when.

Re: actually...

Ed Davis, DPM on 10/28/01 at 14:34 (063734)

I agree that 40 is a good sample size if we use the assumption that etiologies are the same or similar. The problem that I see is the need to divide the 40 based on the differing etiologies of TTS (and maybe on duration of the problem). The reason for this is that I feel that TTS caused by different etiologies respond differently to treatment. At best, I would like to see a minimum of two broad etiologic divisions: TTS due to an anatomically identifiable structure (abnormality) placing pressure on the nerve and those TTS cases where such a structure is not present. It is my experience that the success rate in the first group would be substantially higher than in the second group. I think the numbers would support this modification of the study. Once complete, the issue of duration needs to be looked at. That is a bit trickier because of the late diagnosis that often occurs although some leeway could be factored in by permitting the investigator to increase the duration time by a fixed percentage across the group.

You are definitely correct in your crticism of how the studies are performed. Internally, PhDs at med schools are quite critical of many of the papers written by surgeons, in particular.
Regards,
Ed

Re: back to the old default, eh? :-)

elliott on 10/28/01 at 15:47 (063736)

Looks like we're in agreement. I'll just add, though, that asertaining pre-surgery whether there is in fact an anatomically identifiable structure is difficult, what with an MRI being far from perfect (especially when dealing with TTS). How far? Well, it depends, circularly, on contradictory studies, of course. Check these links out. The first did a study on your other pet point, longer-term sufferers, and look at its MRI experience. Others look much more rosy.

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9345220&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=1672132&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7676025&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7584240&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8491431&dopt=Abstract

Re: citations

Ed Davis, DPM on 10/29/01 at 19:48 (063779)

Elliott:

All good citations. Definitely support use of MRI. I would assume that Frey in the two citations from USC has access to an extremity MRI or a device that can yield similar results. MRI technology is moving forward but there are still a lot of machines out there that are providing useless pictures for small detail in the extremities. There is also support for the concept of definable masses in the tarsal(as seen by MRI) being major culprits.

NCVs probably can be avoided if a mass can be identified in the tarsal tunnel. Sometimes NCV data is very useful when there is a question of other sources of neuropathic symptoms--neuropathy, nerve root irritation/lesions, that is, to try to sort out overlapping etiologies. A common scenario that I encounter is a diabetic with low back pain, diabetic neuropathy and possibly TTS---3 potential causes of symptoms which can mimick each other at times. The other issue with NCV is the skill of the electromyographer. Dr. Saeed, a local electromyographer with a subspecialty in TTS, will be lecturing to our residents next month. His results are so dependable that I can view his results with a high level of confidence. I will ask him what human factors are leading to the variability in such NCVs and also to see if he is networking with other individuals with his level of skill so that I can get some idea who to recommend for NCVs outside my area.
Ed

Re: what did I do now? :-)

elliott on 10/30/01 at 09:17 (063815)

Yes, then, of course there's the accuracy of the nerve conduction test, where one has to worry about both false positives and false negatives, with the first of even greater concern to the profession. The abstract below suggests it does rather poorly regarding false negatives:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8957166&dopt=Abstract

Here's one suggesting some of the difficulties encountered in testing:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8198776&dopt=Abstract

Here's another one on poor surgical results, and also reasonable NCV results in surgically confirmed cases (and so good in avoiding false negatives), but no correlation between positive NCV results and surgical outcome (when you look closely at some of these studies, the sample sizes of the subgroups start getting mighty small:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8056802&dopt=Abstract

The next one is on results of re-release:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7951970&dopt=Abstract

Here's another on MRI:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8491431&dopt=Abstract

This next abstract talks about and supports just about everything you say (BTW, do you have access to a good medical library? And nothing better to do?). Of note is that it considers fibrosis around the nerve as the single most important factor in determining surgical success (is that for re-release-indicating, perhaps, surgeon's technique and scar tissue, or do they mean the first time around?).

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9195255&dopt=Abstract

Since you sound 'tight' with Saeed (is that MA Saeed?), I was wondering if you could you ask him his opinion as to the reliability of other testing devices, such as PSSD and SSEP? Really curious. Thanks.

Re: oh, one more thing

elliott on 10/30/01 at 11:08 (063822)

You often talk about extremity MRIs, those that can see veins, etc. (BTW, I think one of my links on MRI in the above post was duplicate to a previous one.) The MRI I had a few days ago and on which I am awaiting results is an expensive high-resolution whatever, but is also used for the rest of the body (I had my lumbar spine MRI in the same machine). I sure hope you're not saying this was the useless type of MRI. To me, the quality of the pictures (I got a copy) seems unbelievable; if only I knew how to read them myself. (Novice self-reading is both fun and scary: looks like tendon ruptures and suspicious bullet holes all over the place.)

Re: Mohammed Saeed, MD

Ed Davis, DPM on 11/01/01 at 16:13 (063952)

Elliot:

Yes, you will see his name on a number of papers as MA Saeed. Thanks for the citations. I will be attending his lecture on the 13th of November and will pose the questions to him.
Ed

Re: oh, one more thing

Ed Davis, DPM on 11/02/01 at 20:02 (064030)

Extremity MRIs include machines with flexible coils that are wrapped around body parts such as ankles as opposed to the large units with fixed walls. There are a number of MRIs, that are not specifically extremity MRIs, that give good pictures too.
Ed

Re: chances for TTS surgical success: 42%, 100%, pick a number!

cindyp on 10/24/01 at 16:41 (063558)

come to america

Re: chances for TTS surgical success: 42%, 100%, pick a number!

elliott on 10/24/01 at 18:29 (063567)

I'm already here. Lotta good it did me.

Re: Some reasons for variability

Ed Davis, DPM on 10/24/01 at 22:53 (063583)

Elliot:

Note that the last citation, the retrospective study, only included 40 patients. That is a fairly small sample size. Smaller sample sizes lead to increased variability.

Other things that create variability include the means by which improvement or satisfaction is measured (as you have pointed out) and mixtures of causes of TTS. Some cases of TTS have a specific structure causing compression on the nerve and some do not. Ideally, studies should separate those two groups. Another factor is the amount of time that TTS was present. I do not think that a 'point of no return' has ever been established but it seems reasonable to expect that nerves that have been damaged longer would have a poorer chance of recovery---at least this variable should be considered by the studies. I have generally found that due to the relative rarity of the problem, that cases of TTS tend to get diagnosed late.
Ed

Re: actually...

elliott on 10/26/01 at 11:28 (063622)

40 is plenty big enough if there really is some a priori probability of success, for then we have a binomial distribution, whose std dev, SQRT(npq), is rather small compared to its mean. Furthermore, I think I recall it takes a sample size of only around 30 for the binomial to be approximated very well by a normal distribution (central limit theorem and related results), which is often convenient for calculations and journal purposes. Assuming the real probability of success is .8 as per one study, the probability of seeing at most 16 successes out of 40 (the 40% as per the other study) is so close to 0 using the normal approximation to the binomial that it goes off the printed table charts. Using brute force on the binomial (assuming I made no mistake) gives 0.0000000352, or about 4 in a hundred million. And supposedly, the 40% study was limited to a set of patients supposedly more suitable for surgery, scary when you think about it.

Of course, I don't deny that ceteris paribus, a larger study is better, nor that all the other things you mentioned (differing levels of followup, definitions of success, netc.) all play a huge role in altering it from being a pure binomial, probability-of-success type of problem. But if conditions were reasonably similar--and they can be--after a few consistent studies they should be able to glean info into such things as when or when not surgery is recommended. I wish the medical field were more rigorous in this regard, rather than come up with all kinds of stats all over the place.

By the way, kudos to you for changing the post title from its default! First time I can remember. I wish everyone would do that more often, as it helps to visualize the flow of the thread, as well as to see who added what posts when.

Re: actually...

Ed Davis, DPM on 10/28/01 at 14:34 (063734)

I agree that 40 is a good sample size if we use the assumption that etiologies are the same or similar. The problem that I see is the need to divide the 40 based on the differing etiologies of TTS (and maybe on duration of the problem). The reason for this is that I feel that TTS caused by different etiologies respond differently to treatment. At best, I would like to see a minimum of two broad etiologic divisions: TTS due to an anatomically identifiable structure (abnormality) placing pressure on the nerve and those TTS cases where such a structure is not present. It is my experience that the success rate in the first group would be substantially higher than in the second group. I think the numbers would support this modification of the study. Once complete, the issue of duration needs to be looked at. That is a bit trickier because of the late diagnosis that often occurs although some leeway could be factored in by permitting the investigator to increase the duration time by a fixed percentage across the group.

You are definitely correct in your crticism of how the studies are performed. Internally, PhDs at med schools are quite critical of many of the papers written by surgeons, in particular.
Regards,
Ed

Re: back to the old default, eh? :-)

elliott on 10/28/01 at 15:47 (063736)

Looks like we're in agreement. I'll just add, though, that asertaining pre-surgery whether there is in fact an anatomically identifiable structure is difficult, what with an MRI being far from perfect (especially when dealing with TTS). How far? Well, it depends, circularly, on contradictory studies, of course. Check these links out. The first did a study on your other pet point, longer-term sufferers, and look at its MRI experience. Others look much more rosy.

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9345220&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=1672132&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7676025&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7584240&dopt=Abstract

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8491431&dopt=Abstract

Re: citations

Ed Davis, DPM on 10/29/01 at 19:48 (063779)

Elliott:

All good citations. Definitely support use of MRI. I would assume that Frey in the two citations from USC has access to an extremity MRI or a device that can yield similar results. MRI technology is moving forward but there are still a lot of machines out there that are providing useless pictures for small detail in the extremities. There is also support for the concept of definable masses in the tarsal(as seen by MRI) being major culprits.

NCVs probably can be avoided if a mass can be identified in the tarsal tunnel. Sometimes NCV data is very useful when there is a question of other sources of neuropathic symptoms--neuropathy, nerve root irritation/lesions, that is, to try to sort out overlapping etiologies. A common scenario that I encounter is a diabetic with low back pain, diabetic neuropathy and possibly TTS---3 potential causes of symptoms which can mimick each other at times. The other issue with NCV is the skill of the electromyographer. Dr. Saeed, a local electromyographer with a subspecialty in TTS, will be lecturing to our residents next month. His results are so dependable that I can view his results with a high level of confidence. I will ask him what human factors are leading to the variability in such NCVs and also to see if he is networking with other individuals with his level of skill so that I can get some idea who to recommend for NCVs outside my area.
Ed

Re: what did I do now? :-)

elliott on 10/30/01 at 09:17 (063815)

Yes, then, of course there's the accuracy of the nerve conduction test, where one has to worry about both false positives and false negatives, with the first of even greater concern to the profession. The abstract below suggests it does rather poorly regarding false negatives:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8957166&dopt=Abstract

Here's one suggesting some of the difficulties encountered in testing:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8198776&dopt=Abstract

Here's another one on poor surgical results, and also reasonable NCV results in surgically confirmed cases (and so good in avoiding false negatives), but no correlation between positive NCV results and surgical outcome (when you look closely at some of these studies, the sample sizes of the subgroups start getting mighty small:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8056802&dopt=Abstract

The next one is on results of re-release:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=7951970&dopt=Abstract

Here's another on MRI:

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=8491431&dopt=Abstract

This next abstract talks about and supports just about everything you say (BTW, do you have access to a good medical library? And nothing better to do?). Of note is that it considers fibrosis around the nerve as the single most important factor in determining surgical success (is that for re-release-indicating, perhaps, surgeon's technique and scar tissue, or do they mean the first time around?).

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9195255&dopt=Abstract

Since you sound 'tight' with Saeed (is that MA Saeed?), I was wondering if you could you ask him his opinion as to the reliability of other testing devices, such as PSSD and SSEP? Really curious. Thanks.

Re: oh, one more thing

elliott on 10/30/01 at 11:08 (063822)

You often talk about extremity MRIs, those that can see veins, etc. (BTW, I think one of my links on MRI in the above post was duplicate to a previous one.) The MRI I had a few days ago and on which I am awaiting results is an expensive high-resolution whatever, but is also used for the rest of the body (I had my lumbar spine MRI in the same machine). I sure hope you're not saying this was the useless type of MRI. To me, the quality of the pictures (I got a copy) seems unbelievable; if only I knew how to read them myself. (Novice self-reading is both fun and scary: looks like tendon ruptures and suspicious bullet holes all over the place.)

Re: Mohammed Saeed, MD

Ed Davis, DPM on 11/01/01 at 16:13 (063952)

Elliot:

Yes, you will see his name on a number of papers as MA Saeed. Thanks for the citations. I will be attending his lecture on the 13th of November and will pose the questions to him.
Ed

Re: oh, one more thing

Ed Davis, DPM on 11/02/01 at 20:02 (064030)

Extremity MRIs include machines with flexible coils that are wrapped around body parts such as ankles as opposed to the large units with fixed walls. There are a number of MRIs, that are not specifically extremity MRIs, that give good pictures too.
Ed