Calculating PVS recovery statistics from the pool of those available to recover
This is a follow-up on my earlier article about recovery rates of PVS:
Here's an excerpt from an article in the New England Journal of Medicine: Late Improvement in Consciousness after Post-Traumatic Vegetative State by Nancy L. Childs, M.D., and Walt N. Mercer, Ph.D.
The American Academy of Neurology (7) and the Multi-Society Task Force on Persistent Vegetative State (5,6) consider the vegetative state permanent if it lasts for 12 months or more after traumatic injury. Irreversibility is established when the risk of prognostic error is "exceedingly small." (5) The acceptable risk of prognostic error was defined as 0.1 percent by the American Medical Association's Council on Scientific Affairs.4 The Multi-Society Task Force on Persistent Vegetative State (6) empirically estimated the risk of prognostic error at 1.6 percent (7 cases of known recovery after 12 months divided by 434 cases of vegetative state 1 month after injury).
More relevant is the risk of prognostic error in patients in a persistent vegetative state who survive for 12 months. The available data are insufficient to provide a trustworthy estimate of the incidence of late improvement, because of erratic follow-up, incomplete reporting, and uncertain diagnosis. Only the Traumatic Coma Data Bank (12) reliably defined persistent vegetative state and reported follow-up on patients after 12 months; 6 of 25 patients recovered consciousness 1 to 3 years after injury. However, the Multi-Society Task Force believed the condition of three of these six patients had improved before one year (6) (and Ashwal S: personal communication). If we assume that there was no improvement in the condition of patients who were lost to follow-up, a conservative estimate of the incidence of improvement after permanent post-traumatic vegetative state is therefore 14 percent (3 of 22 patients), which is substantially larger than 1.6 percent.
[In examining it further, I think this is a completely different study than that quoted in the Task Force article saying 7 patients recovered.]
One very interesting point about this: If three of those six patients had improved before one year, than there must by definition have been a period of time when they were diagnosed as being in a PVS when in reality they had emerged from it.
Here's the Coma Data Bank article:
Vegetative state after closed-head injury. A Traumatic Coma Data Bank Report
H. S. Levin, C. Saydjari, H. M. Eisenberg, M. Foulkes, L. F. Marshall, R. M. Ruff, J. A. Jane and A. Marmarou
Division of Neurosurgery, University of Texas Medical Branch, Galveston 77550.
To elucidate the clinical course of the vegetative state after severe closed-head injury, the Traumatic Coma Data Bank was analyzed for outcome at the time of discharge from the hospital and after follow-up intervals ranging up to 3 years after injury. Of 650 patients with closed-head injury available for analysis, 93 (14%) were discharged in a vegetative state. In comparison with conscious survivors, patients in a vegetative state sustained more severe closed-head injury as reflected by the Glasgow Coma Scale scores and pupillary findings and more frequently had diffuse injury complicated by swelling or shift in midline structures. Of 84 patients in a vegetative state who provided follow-up data, 41% became conscious by 6 months, 52% regained consciousness by 1 year, and 58% recovered consciousness within the 3-year follow-up interval. A logistic regression failed to identify predictors of recovery from the vegetative state.
Even if not a single patient died between 12 months and 3 years, that means at least 11-12% of the PVS patients who were in a PVS at 12 months recovered after 12 months. (That's 6% of the original pool including those who had already recovered by 12 months.) If you take into account the ones that died, it would be higher.
I found a copy of the task force article "Medical Aspects of the Persistent Vegetative State--Second of Two Parts" which includes a chart with the numbers of patients who recovered (in this case, "recovered" just means they attained a higher state of consciousness than PVS--the amount of recovery varied) and died during each time period. I guess none of the patients regressed into a coma or lower state of consciousness than a PVS.
Here is the information taken directly from the tables in that article:
For adults with traumatic injury (434):
At 3 months, 15% had died, 33% had recovered consciousness, and 52% were diagnosed as being PVS.
At 6 months, 24% had died, 46% had recovered consciousness, and 30% were diagnosed as being PVS.
At 12 months, 33% had died, 52% had recovered consciousness, and 15% were diagnosed as being PVS.
For adults with non-traumatic injury (169):
At 3 months, 24% had died, 11% had recovered consciousness, and 65% were diagnosed as being PVS.
At 6 months, 40% had died, 15% had recovered consciousness, and 45% were diagnosed as being PVS.
At 12 months, 53% had died, 15% had recovered consciousness, and 32% were diagnosed as being PVS.
Children with traumatic injury (106):
At 3 months, 4% had died, 24% had recovered consciousness, and 72% were diagnosed as being PVS.
At 6 months, 9% had died, 51% had recovered consciousness, and 40% were diagnosed as being PVS.
At 12 months, 9% had died, 62% had recovered consciousness, and 29% were diagnosed as being PVS.
Children with non-traumatic injury (45):
At 3 months, 20% had died, 11% had recovered consciousness, and 69% were diagnosed as being PVS.
At 6 months, 22% had died, 11% had recovered consciousness, and 67% were diagnosed as being PVS.
At 12 months, 22% had died, 13% had recovered consciousness, and 65% were diagnosed as being PVS.
(end information from tables)
It is important to note that the pools of all but adults with traumatic injury are very small, especially after the first three months--really too small to get an accurate statistical analysis. Even that one is pretty small, but it's significantly larger than the others.
For example, none of the non-traumatic head injury patients in this pool recovered after 12 months. But we need to remember that we're talking about a pool of only 54 patients at that point--hardly enough to calculate an accurate statistic from. In children with non-traumatic injury the pool at 12 months after injury was only 29 kids. So just because 0 out of 29 or even out of 54 recovered after 12 months doesn't mean that several out of a pool of 100 or 500 wouldn't have recovered. We do know that some patients do recover from non-traumatic injury PVS after 12 months--there are at least 6-10 documented cases of this happening in recent years. In all of these pools, the best we can get is a general idea of the probability of emergence from a PVS.
It is interesting to note that according to those numbers, a child with a non-traumatic injury has a BETTER chance of recovering between 6 and 12 months after an injury than between 3 and 6 months. I think that's a good illustration of how the small size of the pool probably messes up the results.
Anyway, just for the sake of getting a slightly more accurate prognosis, let's calculate percentages from the available pool that recovered consciousness between 3 and 6 months, 6-12 months, and 12 months-3 years. This is the number you would want if you were looking at a living PVS patient and wanting to know what their chances of recovery were at a given point, compared to other patients in a PVS the same length of time after their injury.
Now, math is not my strong suit so if you find an error in my calculations please let me know.
Traumatic injury in adults:
Starting pool 434 patients.
0-3 months: 65 died (15%), 143 recovered (33%). Remaining pool of living PVS patients 226 (52% of original pool)
3-6 months: 39 died (9% of original pool, 17% of current pool), 56 recovered (13% of original pool, 25% of current pool). Remaining pool 130 (58% of last pool)
6-12 months: 39 died (9% of original pool, 30% of current pool), 26 recovered (6% of original pool, 20% of current pool). Remaining pool 65 (50% of last pool)
12 months-3 years: We don't know how many died, but we know that 7 recovered. That's only 1.6% of the original pool, but almost 11% of the current pool. Huge difference.
According to the statistics in this study:
For Adult Traumatic head injury:
In the first 3 months a living patient in PVS has about a 54% chance of recovering by 3 years after injury.
A patient who has been in a PVS for 3 months has about a 39% chance of recovering by 3 years after injury.
A patient who has been in a PVS for 6 months has about a 25% chance of recovering by 3 years after injury.
A patient who has been in a PVS for 12 months has about an 11% chance of recovering sometime in the next 2 years. [Note that the study quoted above suggests a 14% or higher chance of recovery at this point.]
In a non-traumatic adult injury, the chances of recovery between 3 and 6 months were just 4% of the original pool, but 6.3% of the actual pool available to recover at the end of 3 months.
See what a huge difference it makes if you calculate the percentage from those in a PVS but alive rather than including all the dead and recovered patients in the pool?
Think of it this way: In a game where kids are picking teams, you can't calculate each player's chances of getting chosen each round based on the size of the original pool. If you started out with 10 kids and 5 have been chosen, than each unchosen kid has double the chances of being chosen now than they did at the beginning. It only makes sense to calculate it that way. When you have only 2 kids left, each one has a 50% chance of being chosen next, not a 10% chance which is what you'd get by calculating their chances from the original pool.
You just can't do it by calculating the percentage of the original pool and get an accurate picture of someone's chances compared to others in the same situation.
Here's an excerpt from an article in the New England Journal of Medicine: Late Improvement in Consciousness after Post-Traumatic Vegetative State by Nancy L. Childs, M.D., and Walt N. Mercer, Ph.D.
The American Academy of Neurology (7) and the Multi-Society Task Force on Persistent Vegetative State (5,6) consider the vegetative state permanent if it lasts for 12 months or more after traumatic injury. Irreversibility is established when the risk of prognostic error is "exceedingly small." (5) The acceptable risk of prognostic error was defined as 0.1 percent by the American Medical Association's Council on Scientific Affairs.4 The Multi-Society Task Force on Persistent Vegetative State (6) empirically estimated the risk of prognostic error at 1.6 percent (7 cases of known recovery after 12 months divided by 434 cases of vegetative state 1 month after injury).
More relevant is the risk of prognostic error in patients in a persistent vegetative state who survive for 12 months. The available data are insufficient to provide a trustworthy estimate of the incidence of late improvement, because of erratic follow-up, incomplete reporting, and uncertain diagnosis. Only the Traumatic Coma Data Bank (12) reliably defined persistent vegetative state and reported follow-up on patients after 12 months; 6 of 25 patients recovered consciousness 1 to 3 years after injury. However, the Multi-Society Task Force believed the condition of three of these six patients had improved before one year (6) (and Ashwal S: personal communication). If we assume that there was no improvement in the condition of patients who were lost to follow-up, a conservative estimate of the incidence of improvement after permanent post-traumatic vegetative state is therefore 14 percent (3 of 22 patients), which is substantially larger than 1.6 percent.
[In examining it further, I think this is a completely different study than that quoted in the Task Force article saying 7 patients recovered.]
One very interesting point about this: If three of those six patients had improved before one year, than there must by definition have been a period of time when they were diagnosed as being in a PVS when in reality they had emerged from it.
Here's the Coma Data Bank article:
Vegetative state after closed-head injury. A Traumatic Coma Data Bank Report
H. S. Levin, C. Saydjari, H. M. Eisenberg, M. Foulkes, L. F. Marshall, R. M. Ruff, J. A. Jane and A. Marmarou
Division of Neurosurgery, University of Texas Medical Branch, Galveston 77550.
To elucidate the clinical course of the vegetative state after severe closed-head injury, the Traumatic Coma Data Bank was analyzed for outcome at the time of discharge from the hospital and after follow-up intervals ranging up to 3 years after injury. Of 650 patients with closed-head injury available for analysis, 93 (14%) were discharged in a vegetative state. In comparison with conscious survivors, patients in a vegetative state sustained more severe closed-head injury as reflected by the Glasgow Coma Scale scores and pupillary findings and more frequently had diffuse injury complicated by swelling or shift in midline structures. Of 84 patients in a vegetative state who provided follow-up data, 41% became conscious by 6 months, 52% regained consciousness by 1 year, and 58% recovered consciousness within the 3-year follow-up interval. A logistic regression failed to identify predictors of recovery from the vegetative state.
Even if not a single patient died between 12 months and 3 years, that means at least 11-12% of the PVS patients who were in a PVS at 12 months recovered after 12 months. (That's 6% of the original pool including those who had already recovered by 12 months.) If you take into account the ones that died, it would be higher.
I found a copy of the task force article "Medical Aspects of the Persistent Vegetative State--Second of Two Parts" which includes a chart with the numbers of patients who recovered (in this case, "recovered" just means they attained a higher state of consciousness than PVS--the amount of recovery varied) and died during each time period. I guess none of the patients regressed into a coma or lower state of consciousness than a PVS.
Here is the information taken directly from the tables in that article:
For adults with traumatic injury (434):
At 3 months, 15% had died, 33% had recovered consciousness, and 52% were diagnosed as being PVS.
At 6 months, 24% had died, 46% had recovered consciousness, and 30% were diagnosed as being PVS.
At 12 months, 33% had died, 52% had recovered consciousness, and 15% were diagnosed as being PVS.
For adults with non-traumatic injury (169):
At 3 months, 24% had died, 11% had recovered consciousness, and 65% were diagnosed as being PVS.
At 6 months, 40% had died, 15% had recovered consciousness, and 45% were diagnosed as being PVS.
At 12 months, 53% had died, 15% had recovered consciousness, and 32% were diagnosed as being PVS.
Children with traumatic injury (106):
At 3 months, 4% had died, 24% had recovered consciousness, and 72% were diagnosed as being PVS.
At 6 months, 9% had died, 51% had recovered consciousness, and 40% were diagnosed as being PVS.
At 12 months, 9% had died, 62% had recovered consciousness, and 29% were diagnosed as being PVS.
Children with non-traumatic injury (45):
At 3 months, 20% had died, 11% had recovered consciousness, and 69% were diagnosed as being PVS.
At 6 months, 22% had died, 11% had recovered consciousness, and 67% were diagnosed as being PVS.
At 12 months, 22% had died, 13% had recovered consciousness, and 65% were diagnosed as being PVS.
(end information from tables)
It is important to note that the pools of all but adults with traumatic injury are very small, especially after the first three months--really too small to get an accurate statistical analysis. Even that one is pretty small, but it's significantly larger than the others.
For example, none of the non-traumatic head injury patients in this pool recovered after 12 months. But we need to remember that we're talking about a pool of only 54 patients at that point--hardly enough to calculate an accurate statistic from. In children with non-traumatic injury the pool at 12 months after injury was only 29 kids. So just because 0 out of 29 or even out of 54 recovered after 12 months doesn't mean that several out of a pool of 100 or 500 wouldn't have recovered. We do know that some patients do recover from non-traumatic injury PVS after 12 months--there are at least 6-10 documented cases of this happening in recent years. In all of these pools, the best we can get is a general idea of the probability of emergence from a PVS.
It is interesting to note that according to those numbers, a child with a non-traumatic injury has a BETTER chance of recovering between 6 and 12 months after an injury than between 3 and 6 months. I think that's a good illustration of how the small size of the pool probably messes up the results.
Anyway, just for the sake of getting a slightly more accurate prognosis, let's calculate percentages from the available pool that recovered consciousness between 3 and 6 months, 6-12 months, and 12 months-3 years. This is the number you would want if you were looking at a living PVS patient and wanting to know what their chances of recovery were at a given point, compared to other patients in a PVS the same length of time after their injury.
Now, math is not my strong suit so if you find an error in my calculations please let me know.
Traumatic injury in adults:
Starting pool 434 patients.
0-3 months: 65 died (15%), 143 recovered (33%). Remaining pool of living PVS patients 226 (52% of original pool)
3-6 months: 39 died (9% of original pool, 17% of current pool), 56 recovered (13% of original pool, 25% of current pool). Remaining pool 130 (58% of last pool)
6-12 months: 39 died (9% of original pool, 30% of current pool), 26 recovered (6% of original pool, 20% of current pool). Remaining pool 65 (50% of last pool)
12 months-3 years: We don't know how many died, but we know that 7 recovered. That's only 1.6% of the original pool, but almost 11% of the current pool. Huge difference.
According to the statistics in this study:
For Adult Traumatic head injury:
In the first 3 months a living patient in PVS has about a 54% chance of recovering by 3 years after injury.
A patient who has been in a PVS for 3 months has about a 39% chance of recovering by 3 years after injury.
A patient who has been in a PVS for 6 months has about a 25% chance of recovering by 3 years after injury.
A patient who has been in a PVS for 12 months has about an 11% chance of recovering sometime in the next 2 years. [Note that the study quoted above suggests a 14% or higher chance of recovery at this point.]
In a non-traumatic adult injury, the chances of recovery between 3 and 6 months were just 4% of the original pool, but 6.3% of the actual pool available to recover at the end of 3 months.
See what a huge difference it makes if you calculate the percentage from those in a PVS but alive rather than including all the dead and recovered patients in the pool?
Think of it this way: In a game where kids are picking teams, you can't calculate each player's chances of getting chosen each round based on the size of the original pool. If you started out with 10 kids and 5 have been chosen, than each unchosen kid has double the chances of being chosen now than they did at the beginning. It only makes sense to calculate it that way. When you have only 2 kids left, each one has a 50% chance of being chosen next, not a 10% chance which is what you'd get by calculating their chances from the original pool.
You just can't do it by calculating the percentage of the original pool and get an accurate picture of someone's chances compared to others in the same situation.
Copyright 2009. All rights reserved.
Posted by purple_kangaroo at 12:19 AM[Permanent link]
23 Comments:
Hi, Robert. Thanks for your comments.
I haven't studied statistical analysis in depth, so my methodology may be flawed. Please feel free to point out exactly in what way it would be flawed and what would be a better way to approach the problem.
What I am trying to figure out is this: Imagine a doctor is talking to a loved one of someone who has been in a traumatic head injury PVS for 3 months. What would be the most accurate way to describe statistics relevant to that person's recovery in the current situation, moving forward from where they are now?
Obviously, describing the statistics of the number of people known to have emerged from a PVS which includes those who improved before the 3-month point isn't relevant.
But to draw the statistic from a pool including people who have already recovered or died isn't relevant either, because both those categories have a 0% chance of coming out of a PVS at 3 months post-injury.
What I'm interested in is discovering the percentage of people who were still alive and still in a PVS at the 3-month point that recovered. I think I did the math correctly to find that number, didn't I?
I feel that the way the numbers were calculated in the original study misrepresented the situation of those still alive at particular points.
It seems misleading because a person's chances of dying increase as their time in a PVS increases, and their chance of recovery does decrease, but perhaps not quite as much as the numbers in that study would lead us to believe.
Calculating a percentage of people who recovered from a PVS between 6 and 12 months, for example, while including in that pool the people who died or recovered at 3 months, doesn't at all give an accurate picture of what someone's chances look like if they have survived that long.
Let me put it this way: Imagine you have a pool of 100 people, and you take 50 of them off life-sustaining treatment so that they die at 3 months (I'm just using this as an example; I know that many of them probably died of other causes), and then half the remaining patients recover by 6 months. If you calculated it from the original pool, a 25% recovery rate would not be accurate in that case. From this data it appears likely that half the dead patients would have recovered if they had not had treatment removed.
The current pool also seems more relevant because the greater the brain injury and general health of the person, the higher their chances of death and the lower their chances of recovery. So we can extrapolate that a person who lived for 6 months probably had less brain injury and/or was in better general health than someone who died before then. Realistically, they probably have a somewhat higher chance of recovery than did the person who died.
So comparing patients alive at a given point only to patients who also survived for that length of time would have the same basic effect as separating out the traumatic vs. non-traumatic brain injury. It makes sense because their prognosis is significantly different.
In my next post on this topic, I was specifically trying to find the rate of misdiagnosis of permanency. I explained my reasoning a little more in that post; did you see it?
As I'm considering this, though, I am thinking I may have made a mistake in my wording, by using the term "chance of recovery" rather than "percentage of living PVS patients who recovered at that point." I think that's especially true of the sentence you quoted.
I'm not sure about that, though, because the chances of surviving for that long are so small that it seems the very fact that the patient has survived that long would suggest they might have a higher chance of recovery. You know--how a doctor might say something about a critically ill patient like, "It looks like she will die, but if she makes it through the next week than she is likely to survive."
I didn't make up the method, BTW . . . I think it's pretty standard. If you have 10 items and take away three, then the chances of the 7 remaining for whatever you're figuring (if you want to calculate their chances from that point forward) have to be calcuated as part of a group of 7, not a group of 10. That's just basic math.
If I'm 3 months past my injury, I'm a lot more interested in knowing what my chances are NOW (given that I've survived this long) rather than comparing them to what they were 3 months ago or to those of the people who already died, KWIM?
As the article I quoted at the beginning of this post (from the New England Journal of Medicine) stated, rather than figuring the prognosis of patients alive at 12 months post-injury based on statistics using the original pool, "More relevant is the risk of prognostic error in patients in a persistent vegetative state who survive for 12 months."
Robert: Which sample group you use depends upon which question you are trying to answer. If you just want an overall statistic of how many of the original pool recover/die/whatever, you use the original pool.
But if you want to know, for instance, the rate of misdiagnosis of permanency, you have to work from only the pool of those who were diagnosed permanent. In order to calculate how many PVS patients were misdiagnosed as permanent, the math problem is very simple.
You find out how many patients were diagnosed as permanent (the number alive and in a PVS at the "point of permanency"), find out how many recovered, and calculate what percent of those diagnosed as permanent actually recovered. That is the only way to calculate the rate of misdiagnosis.
Calculating it from the original pool makes no sense because most of those recovered/died before there was ever a diagnosis of permanency to make.
Say you have a pool of 1000 patients who are diagnosed as PVS 3 months after a non-traumatic injury, and you want to know how many of those will recover. How would you calculate it to get the most accurate number possible?
Obviously the smaller the pool the less accurate the numbers are going to be (thus a rate of 50% recovery if you only have two patients and one recovers, which we know is almost certainly not an accurate statistic). But if you are trying to answer the question of how many people in a group of x number in a given condition recovered, you have to calculate from the current pool rather than the original pool.
I don't understand what you mean by "If this is true, why did you calculate the probability of recovery for patients within 3 months and 6 months of injury? Obviously, many of those patients aren't going to be available to recover after 12 months in a PVS."
At each stage, I only included the patients currently alive and in a PVS in calculating the statistics. The statistic about patients at 12 months doesn't include the patients who recovered or died at 3 or 6 months.
As for whether there's enough information about post-12-month recovery to draw any conclusions, both the Coma Data Bank and the Multi-Society Task Force on PVS used those statistics to draw conclusions from, as did the peer-reviewed article in the New England Journal of Medicine.
I think you must have been referring to the tenth paragraph of the NEJM article, of which the next sentence after the one saying there isn't much available data goes on to say that "the Traumatic Coma Data Bank reliably defined persistent vegetative state and reported follow-up on patients after 12 months" and goes on to conclude that "a conservative estimate of the incidence of improvement after permanent post-traumatic vegetative state is therefore 14 percent."
The Task Force article didn't include the late recoveries in the particular table you mentioned, but the same article included a whole separate table dealing with several documented cases of late recovery (after 3 months for non-traumatic and after 12 months for traumatic injury). I suppose they felt it was redundant to include that information in both tables. I'm going to have to go back and look at it again now.
I started doing these calculations because I was interested in figuring out how many patients recover at the different stages.
My line of thinking was this: If there are x number of patients who have been in a PVS for x amount of months, how many of them are likely to recover? Like I said before, if you have 1000 patients in a PVS 3 months after a non-traumatic injury, how many of those 1000 can we reasonably expect to regain consciousness?
I started out calculating it just for myself, because I wanted to know what Maria Korp's chances of recovery were. Then I got curious about how many patients that are diagnosed as being in a permanent PVS actually recover.
I'm the type of person that once I start a task I feel driven to complete it, which is why I calculated the statistics for all available time periods given in the study.
I found the statistical analysis interesting, and I know a lot of my readers are interested in PVS, so I posted it.
Either way you calculate it--as a percentage of the original pool or as a percentage of the current pool--you should come up with the same actual number of patients who recover in this particular group.
Again, if you find any specific errors in my mathematics or methodology, please let me know. But I do think the way I approached it was valid for the questions I was trying to answer.
Looking at it now, though, I think my numbers concluding how many of the 10,000-25,000 PVS patients in the USA may not be accurate. I'm going to have to double-check whether I did that wrong or not.
I will double-check those numbers and explain a bit more tomorrow.
For now, I'll leave it at this: I think I calculated the possibility of Maria Korp recovering at this point to be between 6 and 8 percent, perhaps more, depending on what study you're using. I think the exact number I quoted earlier was 6.3 percent for a non-traumatic injury after 3 months.
Several of the numbers in your last post don't make sense. For instance, none of the pools had only 7 survivors.
None of these studies give us enough information for certainty with the size of the pools and the methodology they used, which is why I wonder at the statistics being used to decide when a case is hopeless enough to give up on. Really all they can do is allow us to make a somewhat educated guess.
As I mentioned before, I need to go back and clarify in the other article I wrote. A pool of two isn't enough to make a statistic out of no matter what you're testing (whether the two are the original pool or the current pool), so if one out of two patients recovers after 15 years all that tells us is that recovery at that point is possible, not how likely it is.
I've had a busy weekend, so I will work on this in a day or so when I have time.
I don't blog full-time, so sometimes other things have to take priority. :)
What are you referring to as "table 2"? There is no "table 2" in any of the articles I linked to in this post. Could you link to the table you're talking about, please?
I think you may be confused because at some points we're calculating the number of patients at the beginning of the time period, and at other points we're calculating the number at the end. The number at the beginning minus those who recovered and died during that time period gives us the number at the end of that time period, which is the number I calculate from to get statistics for the next time period.
So the number of patients alive and in a PVS at the end of the 0-3 month period is the number I'm calculating percentages of during the 3-6 month period. Does that help?
Wow... what a discussion. Just a brief comment about the statistical analysis.
First, as the data set gets smaller, certainty obviously gets smaller as well. So, in the later datasets, the statistical analyses provide far less certainty. You pointed this out, robert, and you were right... but you implied that the later numbers were therefore worthless, which isn't true. They're just less certain.
With regard to whether or not it is valid to re-align the dataset at various points, let me refer to the smoking example that robert gave at one point. robert, you wrote that the smoking industry would be amiss to state that only 1% of smokers die after 40 years. The wording there is deceptive. But, it would be perfectly appropriate for them to say, "Smokers who survive for 40 years have only a 1% chance of dying of lung cancer". This would be accurate, non-deceptive, and actually somewhat useful, as it would be solid evidence that some people had some type of genetic protection against lung cancer. I don't believe that purple_kangaroo has deceptively stated her statistics.
"I know you mean well, but no matter what you do with the data, you're not going to change the number of patients who've recovered, or their true probability of recovery."
Obviously not. We can never know whether an individual will live or die, and even our statistics are weak tools because there is so much we don't know about the causes of PVS, or even how to diagnose it accurately. However, statistics are often cited in these cases, as purple_kangaroo has cited, to help doctors and guardians decide when the chance of survival is so small as to be negligible.
Consider, for a moment, the possibility that some people (as with your smoking example) have a genetic predisposition that gives them a greater likelihood of surviving PVS and eventually recovering. In this case, it might be quite useful to say that, of those patients that survive in a PVS for 12 months, there is a greater chance of eventual recovery. Purple_kangaroo appears to me to be looking at the evidence to see if some such pattern exists.
"no honest person would want this type of recovery for theirself, a friend, or a loved one."
I'm pretty sure I'm honest, and I most certainly would want that level of recovery for myself, a friend, or a loved one. You can certainly speak for yourself, but you cannot impose your views as the moral high ground, any more than I can impose mine on you in that way.
Mark
I don't understand how you can unequivocally say that nobody could wish for a recovery less than moderate or good. I'm not saying that anyone would wish for only a poor recovery--I think everyone would want as good a recovery as possible for any patient--but poor recovery is certainly in my mind better than no recovery at all. I'm sure that many honest people feel that way, and your implication that anyone who thinks so is not honest is not fair.
What would you wish for instead? That they died? Or that they remained in a PVS? Do you think severely disabled patients should be euthanized? What would be the alternative?
Here is the definition of severe disability from the Multi-Society Task Force: "Patients with severe disability are no longer capable of engaging in most previous personal, social, and work activities. Such patients have limited communication skills and abnormal behavioral and emotional responses. They are partially or totally dependent on assistance from others in performing the activities of daily living."
That would describe an awful lot of people, recovered from a PVS or not. I know a 4-year-old boy with cerebral palsy who would fit that description perfectly. He can't talk or walk, but he can smile and coo and he definitely experiences emotions and responds to his environment. He seems to be quite happy most of the time despite the fact that he can't do much or communicate very well. He's a beautiful child who is very much loved and valued by his family and responds very well to their love.
The NEJM article on Late Improvement in Consciousness after Post-Traumatic Vegetative State went into some detail about this. It made several comments regarding a patient who emerged from a PVS into what I'm guessing from the description was probably a minimally conscious state gradually between 15-17 months and a few years after traumatic injury:
Three years after injury the patient was communicating using eye blinks for yes or no. She accepted limited oral feedings inconsistently. She had no volitional movements in her legs or trunk, and spasticity and weakness limited movement in her arms.
Five years after injury, the patient could follow conversations and was communicating by mouthing words and short phrases. She enjoyed pampering, and her mood was usually euphoric. Her attention span was limited to 15 minutes, with consistent orientation to person only. Her nutritional needs were supplied by oral feeding (weight and nutritional status were good throughout her course). She remained wheelchair-bound and totally dependent for all care. Complications during the course of her illness included hypertension, heterotrophic ossification, recurrent sinusitis, a femoral fracture, and deep venous thrombosis. She was sent home from a long-term care facility 5.2 years after injury. . . .
Our patient remained severely disabled and totally dependent. She had no behavioral evidence of depression or despondency over her deficits. She enjoyed humor, making jokes and teasing her caretakers. Andrews (13) described a series of 11 patients in a persistent vegetative state who recovered awareness after four or more months. these patients, "even the most profoundly disabled, were able to take pleasure in their surroundings" and "showed no obvious distress at their condition."
I just don't think we can place any value or judgement about quality on someone else's life, or that we should. It's not our responsiblity or right to decide whether someone else's life is worth living or not.
Besides, there are a few cases where people have shown moderate to good recovery after the point of permanency. Sergeant Mack is one example.
The Task force article says, "The Traumatic Coma Data Bank study reported that 6 of 93 adult patients in a vegetative state recovered consciousness one to three years after injury. Four of these six patients had severe disability, and one had moderate disability; the status of the sixth patient could not be determined."
Robert said, "it really isn't fair to use the 7 patients who recovered after 12 months, to calculate the overall probability of recovery for patients in a PVS from 1 to 12 months."
I didn't do that.
R--"So instead of calculating the probability of any recovery, why don't you calculate the probability of moderate or good recovery?
I doubt that the probabilities will be any better than those found in "Table 2", And maybe that's why the NEJM didn't bother to calculate the probability of recovery after 12 months in a PVS ... "
Again, the NEJM article by the Task Force DID include information dealing with late recovery.
Medical Aspects of the Persistent Vegetative State— Second of Two Parts
Table 5. Verified Reports of Five Patients with a Late Recovery from a Persistent Vegetative State (PVS).
Of the 5 patients in the chart, 3 had severe disability and two (one of which had been in a PVS for 36 months before recovering) had moderate disability.
I got my figures from table 3 in the Multi-Society Task Force report. So the AAN numbers are probably more accurate. I don't think I realized there were charts at the end of that article until you pointed it out, which is why I was confused.
I was calculating it by figuring the percentages given in the task force article. For instance, the Task Force chart said that 52% of the original pool of 434 traumatic adult PVS patients were still in a PVS at 3 months. So I calculated 52% of 434 to come up with the pool for the 3-6 month time period.
robert,
You wrote:
As to deceptive statistics, it may not be intentional, but I think it is, in fact, deceptive to say that "a patient who has been in a PVS for 12 months has about an 11% chance of recovery sometime in the next 2 years".
Very well. You're disputing a different part of the quote now than you were initially. How about this statement:
"a patient who has been in a PVS for 12 months as about an XX% chance of regaining some level of consciousness within the next 2 years."
I think purple_kangaroo, who wasn't boiling down the information to a soundbite but presenting it in meticulous detail, communicated all of that in her initial post. If we boil it down to that, do you accept it as a fair and reasonable statement, or does it still seem "deceptive" to you?
Mark
Mark, you are correct that I did not communicate recovery to be anything more or less than emerging from a PVS. I said quite clearly, "In this case, "recovered" just means they attained a higher state of consciousness than PVS--the amount of recovery varied . . . I guess none of the patients regressed into a coma or lower state of consciousness than a PVS."
I think I will revise the post to be a little more clear though. For instance, changing "A patient who has been in a PVS for 3 months has about a 39% chance of recovering by 3 years after injury" to something more like this: In this study, approximately 39% of the patients who had been in a PVS for 3 months emerged from a PVS by 3 years after injury."
Would that be more clear, Robert?
Robert, three of the articles I quoted drew statistics and conclusions based on the entire 3-year study period.
The study accumulated data for three years; it would be dishonest to leave out the last two years' worth of data. The AAN, the National Coma Data Bank, and the PVS task force all used the data, so I included it.
Nothing I wrote even remotely suggested that a person's chances of recovery at 3 years are even close to the same as their chances at 3 months.
Most of what I wrote was simply summaries of the articles and reprinting the data contained in them, but if you haven't read them I guess you wouldn't understand that. I did link to sources where they are reprinted on other sites a number of times, and you don't have to have a paid subscription to read them on the NEJM site.
I was trying to be comprehensive, so I included all the available data. I quoted recovery rates (calculated by the people who wrote the articles, not by me) based on the entire pool for each time period between 0 and 12 months separately.
In my first few articles I dealt only with the likelihood of recovery within 12 months of the injury and didn't include any of the data after 12 months at all.
The only group I included statistics about recovery after a year in was the Traumatic Head Injury group. And I made it very clear that only 7 adults and 0 children recovered between 12 months and 3 years, which was the duration of the study.
The adult traumatic head injury group is the group in which 7 of the remaining 65 or so patients emerged from a PVS between 1 and 3 years after their injury. That is the only group in which I calculated the probability of recovery including a time period past 12 months. And may I remind you that the numbers I came up with were significantly lower than the numbers suggested by the National Coma Data Bank study?
Exactly what harm is it going to do families to know that between 6% and 14% of patients judged to be in a permanent PVS actually emerged from the PVS?
And exactly how is it misleading to say that in traumatic head injury adults, 1.6% of the original pool and 11% of those alive and in a PVS at the time recovered? Because that's what I said.
Or that (again direct quote from what I wrote in my original post), "in a non-traumatic adult injury, the chances of recovery between 3 and 6 months were just 4% of the original pool, but 6.3% of the actual pool available to recover at the end of 3 months."
Again, you're accusing me of saying and doing things I haven't said or done, taking my remarks completely out of context, and not reading the entirety of what I actually did say.
Robert said, "The NEJM feels that after 3 months in a PVS, patients with a traumatic injury have a 35% probabilty of recovery at 12 months.
But other than to make it seem as though the rate of recovery between 3 months and 3 years is constant, I can't see any reason for saying that 39% of patients PVS for 3 months have emerged from a PVS within three years of injury."
I guess the National Coma Data bank was trying to make it seem as though the rate of recovery was constant then. Here's the quote from that article again:
"Of 84 patients in a vegetative state who provided follow-up data, 41% became conscious by 6 months, 52% regained consciousness by 1 year, and 58% recovered consciousness within the 3-year follow-up interval."
It's pretty obvious in both of those examples, including yours, that only 4-6% of the patients recovered after 12 months. I also broke it out and gave the percentage that recovered between 12 and 36 months. How is that in any way misleading?
And the Intersociety Task Force on PVS wrote this in their summary of the prognosis of PVS: "For patients in a vegetative state as a result of traumatic brain injury, the prognosis for recovery remains unfavorable. Recovery of consciousness and function was determined by reviewing data from previously described series of patients rather than individual case reports.
Data were available on 434 patients in a vegetative state one month after a severe head injury (Figure 1 and Table 3). Recovery of consciousness varied with time. Three months after injury, 33 percent of the patients had recovered consciousness; 67 percent had died or remained in a vegetative state. Recovery had occurred in 46 percent of the patients at 6 months and in 52 percent at 12 months. Recovery after 12 months was reported in only 7 of the 434 patients. One patient recovered consciousness 30 months after injury and remained severely disabled. The Traumatic Coma Data Bank study reported that 6 of 93 adult patients in a vegetative state recovered consciousness one to three years after injury."
Robert, I think we're getting several different articles and studies confused. I made that same error at first when I thought the 7 out of 434 patients in the study was the same study as the National Coma Data Bank study. Actually they were two different studies.
My conjecture that there may have been 7 instead of 6 patients who recovered in the National Coma Data Bank article was stated clearly as a guess that I was not sure about, with the uncertainty very clear in the way I had written it. I said, "I'm not sure, but I think it may have been the same study as the one quoted in the task force, and IF so it would be 18%"--and I did not use that figure in any of my calculations. It was a very small "aside comment" in parenthesis which I made extremely clear that I was not sure about. It has already been removed as it was in error.
There were two different studies. Actually, more than that but right now we're talking about two in particular.
Study one: National Coma Data Bank. The original pool was 93 adult patients. By 12 months there were 25 still alive and in a PVS. Those 25 were diagnosed as being in a permanent vegetative state.
6 of those 25 either recovered between 1 and 3 years after their injury or were misdiagnosed as to their being in a PVS at 12 months in the first place (according to the task force evaluation, three of them were not in a PVS at all at 12 months--even though they were diagnosed in a permanent vegetative state at a year post-injury).
6 of 25 is 24%. Almost a quarter of the patients diagnosed as being in a permanent vegetative state in this study were misdiagnosed, either because they weren't actually in a PVS or because they recovered.
Doesn't that seem just a bit disturbing to you?
The way they came up with the 14% rate of recovery is by subtracting the three patients who the Task Force judged were not actually in a PVS at 12 months (misdiagnosed completely) from the 25, giving us 3 out of 22 that recovered after being diagnosed in a permanent vegetative state. 3 of 22 is 13.64 percent, which they rounded correctly to 14%. I did not come up with the 14% number, I simply copied it out of the article.
Again, the actual rate of misdiagnosis was 24% if you don't take out the three who may have been misdiagnosed at 12 months rather than recovering after 12 months.
Study 2: Different study, different group of patients.
Starting pool of 434 adult patients with traumatic head injury. The table in the Task Force article says that at 12 months, 33% had died, 52% had recovered consciousness, and 15% were diagnosed as being PVS.
So the number of patients alive and in a PVS at 12 months was 15% of 434, or 65.1 patients. Since you can't have a tenth of a person, we can safely assume it was actually 65 patients.
So, in the second study, 65 patients were alive and in a PVS at 12 months. Between 12 months and 3 years 7 of those 65 recovered. 7 is 10.77 percent of 65, which I rounded to 11% (all of the studies rounded to the whole percentage for any number over 2%).
Study 3: The third chart dealt with still a third group. We don't know what the original pool was in this case (the Task Force article didn't give that information), but we know that there were 5 patients judged to be permanent who had verified recovery after the point of permamency. Out of those 5, 3 had severe disability and 2 had moderate disability. I'll transcribe the whole chart for you when I get a chance--it includes type of injury and number of months before recovery.
One of the two patients with moderate disability recovered 8 months after anoxia (non-traumatic injury) and the other recovered 36 months after subarachnoid hemorrhage.
Does that help?
Also, don't forget that multiple studies have shown a 35-45% rate of misdiagnosis of PVS in the first place.
Since many of these were not discovered until many moths or sometimes years after the diagnosis, you can't convincingly make the case that accuracy in diagnosis is significantly higher after a period of time, although there should theoretically be at least a slightly higher degree of accuracy with time and more doctors. Some of these patients weren't correctly diagnosed until 2 years or more after their injury, and one only after almost a year of researchers' working very hard to find evidence of consciousness.
In most of the cases there was no chage in the patient or their responses--the only variable that changed was the method of examination and the amount of time spent examining them.
So unless we develop and maintain better ways of diagnosing PVS, it is likely that somewhere between a third and half of the 10,000 to 25,000 patients in a PVS have been misdiagnosed.
This is a completely separate issue from the likelihood of recovery.
Let me explain my reasons for bringing up these issues publicly.
My goal in bringing these issues to wider attention is to hopefully encourage people to hold the medical community accountable in several areas:
First, to study this issue more. There is much uncertainty, and many things we thought were certainties have been disproven. We obviously need more research in the area of consciousness, PVS and minimally responsive patients.
Secondly, to encourage a policy that living PVS patients continue to receive therapy and new treatments whenever possible rather than being relegated to a corner where they receive nothing other than simply being kept alive. In other words, treat every PVS patient as though they have at least some chance of recovery, even if small, rather than as though their chance is 0.
Research clearly shows that there is at least a small chance of recovery at the very least up to 36 months after injury. So don't treat living patients that have a chance of recovery as though they're already dead.
(I am choosing not to deal with the issue of when to justify the use or removal of specific treatments here, just saying don't treat them as 100% hopeless when they're not.)
Third, be honest with the families. Even a 1.6% chance of recovery is many times the acceptable 0.1% margin of error in judging someone completely unable to recover. Doctors should tell families the true chances and that there IS a chance, even if very small, of recovery.
Instead, many try to push families into removing life-sustaining treatment and donating organs by telling them the patient has absolutely no chance or almost no chance of recovery.
The difference between 0 or 0.01% chance and 1.6% or even a 11-14% chance is huge to some people. Even a 1.6% chance is significant. It won't seem like a big difference to some, but to others it will be hugely significant.
There is no reason not to give accurate statistics based on BOTH the original pool and the pool of patients with the same type and duration of injury as the patient currently has. People are smart enough to deal with that information and take both statistics into account when making their decisions.
Fourth, to encourage the development and use of more effective ways of allowing severely disabled patients to communicate and demonstrate awareness.
Along with this goes higher standards in the amount and type of testing and observation before someone can be diagnosed as PVS in the first place. NOBODY should ever be diagnosed as in a permanent vegetative state by a doctor who has spent an hour or less with the patient, and certainly not only a month or three months after their injury. Yet this happens routinely.
Finally, I hope to encourage people to reconsider the standards used in removing life-sustaining treatment, and the judgements of valuation of life. I'll deal with this and some of your earlier related questions in a separate post. I'm working on that one and will post it as a new article.
I posted the charts and some other additional information in my two new posts, here and here.
Again, I don't know why you keep insisting we can't know the percentage of recovery between 12 and 36 months in the study I calculated it from. We know that in the group of 434 adult PVS traumatic head injury patients, at the end of 12 months there were 65 patients alive and in a PVS, and that 7 of those recovered. Exactly how is it impossible to figure out a statistic from those figures? We can't make a judgement about the other groups, but we certainly have enough information to calculate a statistic about this particular group.
As for the national Coma Data Bank study, since the International Task Force on PVS specifically said that was the most reliable study available, I hardly think you can assume there was an agenda or that the study was poorly done. Yes, a pool of 93 patients to start out with (which left 25 alive and in a PVS at the end of a year) is a small pool.
But the only information we have about PVS is ALL calculated from very small pools. We need more research in this area. If the data is unreliable for determining likelihood of recovery, it is just as unreliable for determining when the chances are small enough to justify removing life support. You can't have it both ways.
I agree that the pools in all of these studies are small, and that this makes the statistics less reliable. But when you have several different studies with different patient groups that all show a significantly higher chance of recovery after the point of "permanence", then my argument is that the definition of permanence and the assumption that a PVS is permanent at that point needs to be reevaluated.
You said, "Now, let's talk about "outcome". If it takes numerous exams over a period of days or weeks for a neurologist to determine a patient has "recovered", I don't consider that to be a recovery."
Again, we're not talking about recovery at all in that state. We're talking about someone who was believed to be completely unaware of their environment and unable to communicate, who was in actuality aware and able to communicate with close to 100% consistency.
Do you see "no awareness" and "aware and able to communicate" as an important distinction, regardless of physical state? Or do you think patients who are in a locked-in state or minimally conscious state should be treated exactly the same as patients in a coma?
One huge reason it's important to me is that a person's quality of life can be significantly higher if they are treated as though they are aware.
Imagine being aware and able to respond, but being treated like a vegetable. Now that would be a difficult quality of life.
If we can identify those patients who are in fact aware and maximize their ability to communicate as well as make sure they have some stimulation and things they enjoy, that will make a huge difference in their quality of life as compared to leaving them in a corner with no stimulation and nothing other than basic life-sustaining care.
Do you see the difference that could make for someone who is actually aware but has been diagnosed as having absolutely no awareness?
That's a big part of why I feel that every patient should be interacted with as though they are aware--because we don't know who is and who isn't. I mean simple things like don't call them a vegetable in front of them and don't leave them in a dark room with no stimulation. This is totally separate from the issue of sustaining or removing life support.
Of course I expect that a good doctor would explain to the family what "recovery" means and the various types of recovery that are possible, just as I did in my posts (did you read the two recent ones and look at the charts?).
But I highly doubt anyone would feel betrayed if a doctor told them honestly what the various statistics were and what recovery may or may not include rather than just telling them briefly that the person was going to be a vegetable and had no chance of recovery.
Yes, I do feel that any living person should be treated as though they might be aware. What harm is it going to do to turn on the TV and talk to them occasionally, etc. if they're not aware?
It's undeniable that occasionally people are aware when we think they're not, so it makes sense to treat anyone who is alive with respect and as though they are a human being. There is a vast amount yet unknown about human consciousness.
Again, I'm not dealing with witholding or giving any specific treatment or with life support issues here.
But what would be the problem with treating someone with respect and interacting with them as though they can hear you when you really don't know for sure whether they can or not?
As for Terri Schiavo, there is ample evidence she was at least somewhat aware and responsive. But that's not what I'm dealing with here.
Oh, and about the 11% rate of recovery after a year: that is ONLY in the group of traumatic head injury patients. And the paper most certainly gives enough information to calculate that. Look at the charts I posted.
Oh, also that statistic is including the patients for which there is no follow-up data after a year and assuming that they did not recover.
Again, we know the number of adult head trauma patients in a PVS at the end of a year. And we know that at least 7 of those recovered in the next two years. That's enough information to calculate a statistic from.
If assuming that the patients for whom there is no followup adds any error, it would be in the direction of making the statistic for recovery higher, not lower.
OK, I see what you mean.
In the Traumatic Coma Data Bank study, the starting pool was 93 patients in a PVS out of 650 with closed head injuries.
Only a small percentage (14% in this case) of patients with head injuries end up in a PVS in the first place. Part of the purpose of the study was to compare the extent of head injuries in those who did and did not end up in a PVS.
So there were never 650 patients in a PVS in that study. There were 93 patients in a PVS at the beginning of the study.
From the Childs et. al paper on the National Coma Data Bank Study (and I'm assuming that author had a lot more information about the study than we do), we know that assuming that none of the 9 patients lost to followup died or recovered, we have a pool of at most 25 patients in a PVS at the 12 month point--22 if we subtract the 3 who supposedly weren't actually in a PVS at 12 months.
Of course, we don't have the full text of the study to look at so I would depend more on an article by someone who actually reviewed the whole study than on statistics we would come up with just from seeing a summary.
But the Childs et. al number of 22-25 out of 93 patients in a PVS after closed (traumatic) head injury at the end of a year is fairly consistent with the numbers in the other studies on the subject. That's assuming that none of the 9 lost to follow-up died or recovered.
I checked my numbers again, and you're right that they were wrong. I was assuming that none of the original pool died, and just subracting the number that recovered by 12 months from the original pool in order to come up with my starting pool at 12 months.
We know from the Childs et. al paper that at least 25 of them did die before 12 months, so my numbers in that particular calculation aren't accurate given all the information we have.
Thanks for pointing that out.
Robert, I dealt with all or most of these questions in my next post, as well as fixing some of the problems you've pointed out in previous comments. Hopefully you saw that.
As for the TCDB report, I'm certain there are probably other articles and a more detailed report on this study. This just happens to be the one I could find easily publicly available on the internet.
You're right in the point I think you're making that we can't really draw any rock-solid conclusions from such a brief summary lacking many important details. Other articles I've read quote other information from that particular study, so I'm sure there is more than a paragraph summary available. If you come across it, please let me know.
However, since the Task Force and most other major studies and articles on PVS mention the Traumatic Coma Data Bank study as a reliable and important piece of information on PVS, I don't think we should leave it out altogether.
Post a Comment
<< Purple Puzzle Place Home