What Killed Libby Zion?

A Death in the Hospital /What Killed Libby Zion?

“The True Administration of Justice Is The Firmest Pillar Of Good Government” are words chiseled in granite above the New York State Supreme Court at Foley Square in Manhattan. The reader who explores “A Death in the Hospital” will come away with the understanding that the principle intention of the New York State Court System in their consideration of the case of Speken V. Columbia was not Justice but rather the imposition of silence.

Nine years before Seth died, another young New Yorker, Libby Zion, died at a premier New York City hospital. As was the case with Seth, she died due to the actions of medical trainees. Inexperienced trainees were given the responsibility, unsupervised, of caring for the sickest of patients. Over the course of the next eleven years, there was much legal wrangling about the issues of the case. The Hospital mounted a vigorous public defense that they were not responsible for Libby’s death.

For several months in the winter of 1994-1995, the Nation was transfixed as the Zion case was shown live on Court TV. The Hospital moved ‘Heaven and Hell’ in order to convince the Jury that Libby was responsible for her own death because of her cocaine use. If there was any blame, it could only be to a ‘system’ of medical education that overworked the trainees into a state of sleep deprivation.

But it was not so much that the students weren’t getting enough sleep, as their supervisors getting too much. The reader of “What Killed Libby Zion? A Reexamination of the Evidence” will learn that the cocaine issue was bogus. As with Seth, Libby’s death resulted from the actions of the trainees.

From the start, the Hospital and State Authorities knew this. Seth Speken would die nine years later as a result of the same illegal 5 point restraining that killed Libby Zion. Were the Government to have been protecting the Public, the true implications of Libby Zion’s death would have been recognized and changes in medical trainee supervision imposed. Seth would be alive today. The Zion and Speken cases represent yet again the manner in which Government can fail to protect.

Justice was denied Libby Zion, Seth Speken – and how many others?

 

Dedicated to Libby Zion and Her Family

 

What Killed Libby Zion? A Reexamination of the Evidence

By Stephanie Zoe Speken, M.S. (Nutrition)
Columbia University, School of Public Health, 1969

When most people remember Libby Zion today, if they remember her at all, they think of the unfortunate young woman whose cocaine use sadly contributed to her death in March of 1984. After all, she tested positive for cocaine, didn’t she? The Medical Examiner’s Office of the City of New York in its official toxicology report of May 8, 1984, stated that cocaine had been found in her body. It was signed by the Chief Toxicologist, the late Dr. Milton Bastos. He said of her premortem blood, “cocaine detected by radioimmunoassay (RIA) and swabs – traces of cocaine detected by RI[A].” Libby Zion was just 18 years old.

Fredric Rieders, Ph.D., a renowned forensic toxicologist and the defense expert witness, told the Jury that the RIA tests performed on Libby Zion were all positive for cocaine, even those that the Medical Examiner had called questionable. As Dr. Rieders was testifying, Court T.V. flashed a message across the screen – “RIA Test Positive For Cocaine.” This was for the entire Nation to see.

Robert Morgenthau, the District Attorney for New York County, said, “she lied about her drug use.” ( Robins, Natalie. The Girl Who Died Twice: The Libby Zion Case And The Hidden Hazards Of Hospitals. New York, N.Y.: Delacorte Press, 1995, p. 204.) Dr. Gregg Stone, a first year resident from New York Hospital who saw Libby Zion on that fateful night in March, 1984, testified, “I think cocaine is what killed this poor girl.” (Robins, p. 209)

Dr. Harold Osborne, the Zion family expert witness, did not altogether throw out cocaine. “He admitted that Libby’s urine is a negative for the purposes of the report,” that is, (according to Robins) “calling it a negative was an administrative decision by the Medical Examiner’s Office and did not mean the absence of cocaine.” (Robins, p. 269)

The Jury believed that cocaine was implicated in Libby Zion’s death. But, this was a tragic mistake. Let’s see why.

First we must start slowly, from the ground up, so that everything will be made perfectly clear. This will involve learning some very elementary statistics – statistics being a way of telling us what data means.

So, let’s start by looking at the word Mean. This is just a word statisticians use for the Average. The Average or Mean is the total of all the numbers reported divided by how many numbers there were in the first place. The Mean is just one way statisticians describe the ‘middle of things.’ For example, two groups of things can have the same Mean and yet they are different in the spread of their values. The following example is adapted from David Voelker and Peter Orton in Quick Review of Statistics Cliff Notes, 1993, p. 34. The only things I’ve changed are that I call Employee A, ‘Edith’ and I call Employee B, ‘Archie.’ Edith and Archie are both employed. We will take a look at the earnings of Edith and Archie for 8 days.

Edith’s Earnings For 8 Days Edith’s Earnings For 8 Days
$200 $200
$210 $20
$190 $400
$201 $0
$199 $390
$195 $10
$205 $200
$200 $380

Both Edith’s and Archie’s earnings for that 8 day period add up to $1600 each. They made the same amount. And they also have the same Mean amount:

$1600÷8 days = $200 Mean (Average) per day, which is the same amount for both of them.

But, Edith’s earnings are much closer to the Mean than Archie’s earnings. Remember,
the Mean is just another way of saying the ‘middle of things.’ Edith’s earnings show less jumping around the average value of $200 than Archie’s. Archie’s earnings are much less predictable. You have less confidence in them. One day they are $400 and the next day they are $0. In other words, Archie’s earnings have a much greater Deviation – his earnings are farther away from the Mean even though the Mean is the same for both Edith and Archie for that 8 day period.

How do we figure out Edith’s Deviation or distance from the Mean? We just subtract each of her daily earnings from the Mean of $200. The following is adapted from Quick Review of Statistics, (p. 35):

Day Edith’s Earnings

Deviation – How far away does Edith’s Earnings Get From the Mean ($200)?

Squared Deviation Of Edith’s Earnings per day

1 $200 $200 – $200 = 0 0
2 $210 $200 – $210 = -10 100
3 $190 $200 – $190 = +10 100
4 $201 $200 – $201 = -1 1
5 $199 $200 – $199 = +1 1
6 $195 $200 – $195 = +5 25
7 $205 $200 – $205 = -5 25
8 $200 $200 – $200 = 0 0

Notice the third column. It makes the Deviation more correct to multiply each Deviation by itself (‘squaring’) once. This way we get rid of the sign, positive or negative, and that’s what we want. Now, after this is done we add up all of Edith’s Squared Deviations from the Mean. In other words, we add up: 0, 100, 100, 1, 1, 25, 25, 0. We get a total of 252. This is a very important number. It is the Sum of Edith’s Squared Deviations.

Now for Archie’s Deviations.

Day Edith’s Earnings

Deviation – How far away does Edith’s Earnings Get From the Mean ($200)?

Squared Deviation Of Edith’s Earnings per day

1 $200 $200 – $200 = 0 0
2 $20 $200 – $20 = +180 32,400
3 $400 $200 – $400 = -200 40,000
4 $0 $200 – $0 = +200 40,000
5 $390 $200 – $390 = -190 36,100
6 $10 $200 – $10 = +190 36,100
7 $200 $200 – $200 = 0 0
8 $380 $200 – $380 = -180 32,400

Now, adding up all the Squared Deviations that have no sign we get the number 217,000. This is again a very important number for Archie. It is the sum of Archie’s Squared Deviations.

Standard Deviation from the Mean

Notice, that the total of Edith’s Squared Deviations is 252, a much smaller number than the total of Archie’s Squared Deviations of 217,000. Statisticians say that Archie’s earnings show a greater ‘Deviation from the Mean’ than Edith’s earnings. In other words, Archie’s earnings jump all over the place. Edith’s earnings are much more predictable.

In order to see quickly and by how much a jump is from the Mean, statisticians use the word Deviation. A big number means a larger jump and a smaller number means a smaller jump. Now, statisticians use a new word they get from the concept of the Deviation. They call it the Variance. All that it is necessary for you to know about the statistics of Variance is that you take each Squared Deviation and add them up as we did, divide that number by the number of all the readings ( in this case, days) – less 1 reading. We call the number of all readings minus one, ‘n-1.’

Here is the definition of Variance: Variance is the sum of the Squared

Deviations ÷ n-1. Now we’re almost there. Statisticians discovered a much more useful way to use Variance. This discovery gives us an actual picture of just how much the data jumps around the Mean. First, we need to introduce another concept and it’s called the Standard Deviation. This is simple. The Standard Deviation is just the positive square root of the Variance. Let’s work this out for both Edith and Archie.

Remember, for Edith, the total of her Squared Deviations from the Mean is 252. Next, we divide by the number of days minus one, in other words, eight days minus 1 day equals 7 days. So, 252 ÷ 7 = 36. This is Edith’s Variance.

Now, we take the positive square root of Edith’s Variance to get Edith’s Standard Deviation from the Mean. So, the positive square root of 36 (√36) equals 6. This 6, a small number, is Edith’s Standard Deviation from the Mean.

Now, for Archie. Remember, the total of his Squared Deviations is 217,000. We divide that number by 7 (n-1) and we get 31,000. This is Archie’s Variance. Now, taking the positive square root (√31,000) we get approximately 176. This is Archie’s Standard Deviation from the Mean, a much larger number than Edith’s.

The Standard Deviation is a mathematical way of looking at reliability. For example, if Archie came home one day with a pay of $20, this amount would not be very reliable as to how much he would ‘average’ after 8 days of work. After 8 days, he would actually average $200 daily. The high Standard Deviation of 176 in Archie’s case tips us off that looking at any one day’s earnings is not a reliable indication of how much he will average in earnings over 8 days.

Not so with Edith. We know in advance, looking at the small Standard Deviation in her salary, that any one day’s earnings will be pretty close to what the average of all the days will be.

This concept of Standard Deviation becomes crucial when we look at whether cocaine was implicated in the death of Libby Zion.

Standard Deviation, Radioimmunoassay, and Cocaine

Radioimmunoassay (RIA) was developed in the 1960’s using radioactive isotopes to detect very minute amounts of substances. RIA can detect substances in the nanogram (ng) amount. A nanogram is 1 billionth of a gram or 10-9 grams. In simple arithmetic, it is written as 0.000, 000, 001 grams. Remember, 454 grams is equal to 1 pound, and this is a billionth of only one gram.

The RIA tests done on Libby Zion were looking, not for cocaine directly, but for benzoylecgonine. Benzoylecgonine, abbreviated BZE, is the main breakdown product of cocaine in the body. Pure cocaine does not remain in the body unchanged for very long. It quickly becomes BZE and other breakdown products.

Briefly, the test is as follows: you have a precisely known amount of BZE that has a radioactive label on it (iodine125), and you have an unknown amount of BZE, which you suspect is in the patient’s sample and which is not radioactive. It is unlabeled. The radioactive BZE from the manufacturer and the unlabeled BZE from the patient’s sample – if it is there at all – compete with one another for a limited number of docking sites (receptor sites) on an antibody from the manufacturer’s test kit. The antibody is specific for BZE, since this is a test for BZE cocaine breakdown product. The amount of attachment to the antibody of BZE from the labeled BZE and the unlabeled patient BZE is proportional to the concentration of each in the reaction test tube. When the reaction is finished, you will have an antibody attached with patient BZE and with manufacturer supplied radioactive BZE. The mixture is separated and the amount of radioactive BZE in the test tube will be inversely (that is, oppositely) proportional to the amount of BZE in the patient’s sample. Radioactive BZE will then be counted in a gamma counter. The higher the counts, the lower the amount of BZE, and the lower the counts, the higher the amount or concentration of BZE in the patient’s sample.

Here comes a very important point to remember. The less the actual amount of cocaine and therefore BZE in the patient’s sample, the less reliable the RIA test becomes. That is, the less certain it is that any reading actually means cocaine.

To understand why, I will have to explain one final simple tool statisticians use that comes from the Standard Deviation. But, first I will present the data regarding the Radioimmunoassay of Libby’s blood, bile, and nasal swab as it was presented to the Jury during late 1994-1995. I will then be able to show how the new tool describes just what the data was telling us.

The following are the results of Libby Zion’s Radioimmunoasssay (RIA) tests. Her date of death was March 5, 1984:

Results of Libby’s blood tests, before and after death:

Sample Date Test Was Done Negative Control
In Radioactive Counts
Positive Control (cutoff) in
Radioactive Counts
Libby’s Readings In Radioactive Counts Average of Libby’s Readings in Radioactive Counts
Blood Taken Before Death (Premortem blood) 3/7/1984 28,738
22, 228

3338
3728

10,069
8,498

9,284
Same Premortem Blood Retested 5/3/1984

26,420
25,004

(no reading given to jury)

24,416
25,660
26,080

25,385
Blood Taken
After Libby’s Death (postmortem blood)
5/3/1984

20,192
18,944

(no reading given to jury)

21,992
20,541

21,267

Results of Libby’s Nasal Swab:

Sample Date Test Was Done Negative Control
In Radioactive Counts
Positive Control (cutoff) in
Radioactive Counts
Libby’s Readings In Radioactive Counts
Nasal Swab Postmortem

Date Not Given

Assumed to be 3/7/1984 (same controls as Premortem blood)

28,228
28,738

3338
3728

17,668

Results of Libby’s Bile:

Sample Date Test Was Done Negative Control
In Radioactive Counts
Positive Control (cutoff) in
Radioactive Counts
Libby’s Readings In Radioactive Counts Average of Libby’s Readings in Radioactive Counts
Bile Postmortem

April 30, 1984

20,672
19,424

4308
3980

17,304
17,164

17,234

Coefficient of Variation

I will now explain the final statistical tool we’ll need to understand just what this data was telling us. It is called the Coefficient of Variation.

The Coefficient of Variation is just the Standard Deviation divided by the Mean x (multiplied) by 100:

S.D./Mean x 100 = Coefficient of Variation (read as a percent)

We know that the Standard Deviation is a measure of how much the data spreads out around the Average (Mean). Remember Edith’s very small Standard Deviation of 6? It tells us that her earnings cluster closely around the average of $200. A large Standard Deviation like Archie’s of 176 tells us that his earnings are much more scattered around the average. Statisticians have found that using the Coefficient of Variation (CV) provides a much more useful way of understanding what the data is and is not telling us.

The CV shows what percentage of the Mean is represented by the Standard Deviation.

A low percentage implies that any one reading is close to the Mean (Average) and therefore reliable. A high percentage tells us that any reading can be far from the Mean and much less reliable.

For example, let’s return to Edith and Archie. Edith’s CV is 6÷200 x 100= 3%. Archie’s CV is 176÷200 x 100= 88%. We have already seen that any day’s earnings in the case of Edith is pretty close to her weekly average earnings. Relying on any one day of Archie’s earnings is risky at best if you are interested in his true weekly average earnings. All this can be seen in advance when we know that the Coefficient of Variation in Archie’s data is high. Once again, this high CV tells us that we can’t rely on any one day’s earnings if we want to know what Archie’s average earnings will be over 8 days.

Now, armed with this concept of Coefficient of Variation, we are in a position to look at what the data for Libby Zion really was telling us.

The following graph is constructed by the developer of a test kit for benzoylecgonine (BZE) by Radioimmunassay (Diagnostic Products Corporation, 5700 West 96th Street, Los Angeles, California 90045). It is similar to the test used on Libby Zion’s specimens. It is made by first preparing samples of known amounts of benzoylecgonine and then seeing what radioactive counts are obtained by the RIA test.

This is a graph of the reliability of the RIA test for finding BZE.

What do we learn from it?

  1. Notice that the concentration of BZE in ng/ml increases from 0 ng/ml to 5400 ng/ml. 0 ng/ml is also called the ‘negative control.’ It has no BZE in it. It is from the manufacturer and will give you a certain number of radioactive counts that correspond to that negative control.
  2. The 300 ng/ml control is called the ‘positive cutoff control.’ It has exactly 300 ng/ml of BZE in it and is also from the manufacturer’s test kit. It will give a certain number of radioactive counts that correspond to that positive control.
  3. The CV (Coefficient of Variation) expressed as a percentage is pretty much stable when the BZE concentration in ng/ml is between 300 ng/ml to 5400 ng/ml, or approximately 4-5%. And this CV is as low as it gets. The researchers who developed the RIA test kit for cocaine (BZE) have found that higher CV values mean the test data is much less reliable. That is, when you have a reading that looks positive, you don’t know whether this actually means you’ve found cocaine (BZE).
  4. Below the concentration of 300 ng/ml, the less there is of BZE, and the CV starts to climb. At 300ng/ml it is greater than 4%. At 150 ng/ml it is already greater than 7.5%. Remember, that the bigger the CV, the less reliable and less trustworthy the test becomes. If we go back to the example of Archie’s earnings, his large Standard Deviation means that his earnings jumped more around the average and were less predictable and reliable. In other words, less precise. This is exactly what happens at concentrations of BZE less than 300 ng/ml. The lower the concentration of BZE below300 ng/ml, the less reliable the test becomes and the less we can count on the results we find.
  5. Notice that near the negative control of 0 ng/ml of BZE (from the manufacturer), the CV approaches 15% (most unreliable.)
  6. The 300 ng/ml cutoff, or the positive control, is the amount of BZE where false readings are at a minimum. Below 300 ng/ml false positives are more likely to occur.

Understanding Libby Zion’s RIA Test Results

The RIA test kit shows how many radioactive counts a known amount of 300 ng/ml of BZE will give you. It also shows how many counts you’ll get when there is absolutely no BZE in the solution you’re testing. The 300 ng/ml is used, as we’ve seen, because lesser amounts of BZE (cocaine breakdown product) give you increasingly unreliable readings.

The few specimens from Libby Zion were taken before and after she died (premortem and postmortem).

In light of what we’ve learned, let’s look at the blood taken from Libby just before she died and tested by RIA 2 days after she died.

The test was repeated, so we’ll look at the average readings:

  1. A 300 ng/ml known amount gave an average of 3,533 radioactive counts (3338 + 3728 ÷ 2). This was the ‘positive control.’
  2. No BZE (cocaine breakdown product) in the solution gave you an average of 25, 483 counts (28,738 + 22, 228 ÷ 2). This was the ‘negative control.’
  3. The RIA test done on Libby’s premortem blood gave an average of 9,284 counts.

Now, what did this 9,284 reading for Libby mean? Remember that a low number of counts may mean cocaine (BZE) and a high number may mean no cocaine. Counts and cocaine breakdown product (BZE) are inversely related to each other.

For her specimen to have been considered positive, using the 300 ng/ml cutoff, you needed 3,533 radioactive counts or less (meaning that the sample had 300 ng/ml of BZE or more.) And to have been negative, it needed to have 25, 483 counts (or more counts.)

So this meant that Libby Zion’s blood tested between the negative and the positive. These results for Libby were in what I’ll call a ‘gray area.’ They weren’t definitely negative readings, but they weren’t definitely positive either.

What did her readings mean? Did she have cocaine in her or didn’t she? Whenever a reading is in the ‘gray area’ there are two possibilities:

  1. Other substances or conditions may be contaminating the test results. For example, RIA testing following the administration of topical dental anesthetics containing cocaine is known to give readings that appear suspiciously positive for several days. Also, the RIA test is very sensitive and even minor things – such as reagent instability, incorrect calibration of the equipment, or contamination from previous tests – can affect the results.
  2. And the second possibility was that there was cocaine (BZE) in the sample even though less than 300 ng/ml.

Here is where our knowledge about the Coefficient of Variation really helps us. The reliability of finding cocaine (BZE) in a specimen such as Libby Zion’s premortem blood by RIA was more uncertain when the amount was less than 300 ng/ml. We know this from the CV curve described above!

The fact was, by looking at Libby’s reading of 9,284 counts we had no way of knowing whether she did or did not have cocaine in her. And that’s the point. Readings in the ‘gray area,’ that is, between the negative and the positive, all must be thrown out unless we have another way of looking at what’s going on.

In fact, there are two other ways that are much more precise than the RIA for detecting cocaine (BZE). These are called Gas Chromatography and Mass Spectrometry (‘GC and MS’). These are much better tests for finding cocaine breakdown product (BZE) in a sample. The scientists who study such matters state that the RIA is only a “screening test” for cocaine ( Burtis, CA & Ashwood, ER, Eds. Teitz Textbook of Clinical Chemistry, 3rd Edition. Philadelphia: W.B. Saunders & Co., 1999, p. 941.) That is, if you find a sample that you suspect of having cocaine, you must ‘confirm’ it with either MS or GC. And that’s even when you get RIA readings that are reported as positive at the 300 ng/ml level or above.

So in short, Libby Zion’s premortem BZE reading should have been discarded because the reading was unconfirmed. The same argument applies to the readings for bile and postmortem nasal swab. They should also have been discarded as being unconfirmed ‘gray area’ data. Finally the same premortem blood and a sample of postmortem blood were tested by RIA 2 months later. These tests were clearly negative so that confirmation was unnecessary.

To repeat the point, since the RIA is a screening test only, whenever you get radioactive counts at a level that suggests cocaine, you must confirm the results with a better test. If you don’t, the science of the matter states you must throw out the unconfirmed results because there is no way of knowing what they mean!

Fredric Rieders, Ph.D

The trial over the cause of Libby Zion’s death finally took place in late 1994 and a verdict was reached in February of 1995. A record of the trial was made on videotape by Court T.V. Of all the testimony presented to the Jury, that of Dr. Fredric Rieders, the defense expert, was the most damaging to the plaintiff’s case. Dr. Rieders was able to inject into the minds of most Jurors that there was strong evidence that Libby had cocaine in her. While he presented his position as scientific fact, in fact, it was just the opposite.

Paraphrasing Dr. Rieders’ most persuasive (and incorrect) statement to the Jury: if a reading is not negative, it’s positive (in his ‘opinion.’)

For example, from the table of data, Libby’s premortem blood gave readings of 10,068 and 8,498 radioactive counts. Dr. Rieders’ opinion was “…they are positive for cocaine because they are much lower than the negative 28,000 and higher than the low standard.” By “low standard,” Dr. Rieders meant the positive cutoff of 300 ng/ml that gave radioactive counts of 3338 and 3728. “So they are positive, not as positive as the low control,” he claimed. He then repeated the same reasoning for the number of counts obtained from nasal swab and bile (see chart above.)

The observant reader of this document, in which the significance of Coefficient of Variation is now clear, will immediately ask – how could Dr. Rieders have been so certain? After all, any reading between the negative ( 0 ng/ml) and positive cutoff (300 ng/ml) was clearly in a region with an increased Coefficient of Variation and was therefore uncertain. It was possible cocaine was there and just as possible that it wasn’t. It was a 50/50 proposition. You just don’t know whether it was there or not.

But, Dr. Rieders presented his opinion to the Jury as though it were an absolute fact that Libby had cocaine in her. He did so by calling into question the meaning and significance of the positive cutoff ( 300 ng/ml). He told the Jury that this was just an “administrative” value and compared the situation to the test for blood alcohol (as in “Driving Under the Influence.”)

In the various blood alcohol tests, which incidentally are measuring much higher amounts of the alcohol molecule than the billionth of a gram (nanogram) levels of benzyolecgonine that the RIA is looking for, there is indeed an “administrative cutoff” below which an individual is said not be intoxicated. Though not intoxicated, there can still be alcohol molecules present in the blood. It’s just that investigators have learned by studying large numbers of individuals that there is a certain cutoff level above which most people will be obviously drunk.

Dr. Rieders was successful in convincing the Jury that the same situation applied to the positive cutoff of 300 ng/ml of BZE. He convinced them that any cutoff level is arbitrary and that extremely low levels of cocaine in Libby’s blood – much less than the 300 ng/ml cutoff of the RIA test – could and did give readings like Libby’s and were consistent with a true positive.

Now, what was wrong with this?

First of all, the observant reader will note from the graph of the Coefficient of Variation that the 300 ng/ml level is the lowest amount of BZE where the Coefficient of Variation is stable at approximately the 4-5 % level (and it remains that way to the highest amount of BZE at around 5400 ng/ml.)

In other words, this level of BZE (300 ng/ml) is not “administrative” in the sense of being arbitrary but rather statistically acceptable as the lowest concentration of BZE where false positives will be minimized (see Jones, R.T. Pharmacokinetics of Cocaine: Concentration When Assessing Cocaine Use By Urinalysis. National Institute of Drug Abuse, Monograph 175, p.11.) Since we have an ‘acceptable’ Coefficient of Variation at this level, we have some certainty that any reading on the RIA test may well be true when confirmed by more precise testing.

Now, there is a situation where an ‘administrative’ decision has been made. The Department of Defense, acting administratively, will use as their positive cutoff, the BZE value of 150 ng/ml. (Teitz: p. 941) That is, they go below the statistically optimum cutoff of 300 ng/ml.

This was done by the Department of Defense for a particular reason. They know that the lower positive cutoff of 150 ng/ml will lead to many more false positives than the 300 ng/ml cutoff. This is important to them since it will lead to finding more recruits who are cocaine abusers.

But, and this a major ‘but’ and the core of the false testimony given by Dr. Rieders, the Department of Defense will accept no reading as positive unless it is ‘confirmed.’ The observant reader of this document knows that the RIA is considered a “screening test.” (Teitz, p. 941) This means that more precise testing by either Gas Chromatography or Mass Spectrometry or both is needed before it is ‘certain’ that the person actually has cocaine (BZE) in the body.

In fact, according to Teitz, and the manufacturer RIA test kit, ‘confirmation’ is needed on all RIA test results for benzoylecgonine (Teitz, p. 41). This applies even when positive appearing readings are obtained using the usual 300 ng/ml cutoff or above.

In his testimony, Rieders told none of this to the Jury. He simply convinced them that: if a reading is not negative, it’s positive. He even dismissed as unnecessary the scientifically accepted teaching that confirmation was ever needed at all. When questioned by Luke Pitoni, a defense attorney, about the necessity for confirmation of RIA results, Dr. Rieders said, “It depends on what the issue is. If you have the means and the willingness and if it is crucial.” He went on to indicate that it was not necessary in this case (“in my opinion.”)

This was totally wrong and distorted and should never been told to the Jury. In fact, Dr. Rieders himself knew it was wrong. From a paper he co-authored in 1992 (Rieders, Frederic, et al., “Cocaine and Some of Its Products in Hair by RIA and GC/MS.” Journal of Analytical Toxicology, Vol. 16, March/April 1992, p. 114) Dr. Rieders stated,

“By convention, for purposes of forensic acceptability, congruence of results from analyses by two physicochemically independent methods are required for substance identification…To perform independent chemical analysis, one needs to know the limitations of both methods and the cutoff that will best eliminate false erroneous positives and minimize false negatives.”

In other words, Dr. Rieders was well aware that confirmation of Libby Zion’s RIA test results was needed by either Gas Chromatography or Mass Spectrometry (that is, ‘congruence by two different methods.’) He was also well aware that the 300 ng/ml cutoff was not “administrative” but rather a level that minimized false results. In fact, his own toxicological testing company (which among other things does forensic testing), National Medical Services, clearly states in its test catalog (www.nmslab.com), that the immunoassay test is for screening purposes only and confirmation is done by GC/MS whenever a suspected positive reading is found.

What was left in the mind of the Jury – and the Nation after the publication of the book, The Girl Who Died Twice, – was that Libby Zion had cocaine in her. Sidney Zion’s lawyers, Thomas A. Moore and Judith Livingston, were blindsided by Dr. Rieders’ testimony. They had no way of knowing that Dr. Rieders had given masterfully scripted and false testimony to the Jury. This testimony should be remembered as one of the most glaring examples of how Juries can be persuaded by the clever misuse of science.

It was reported in Robins’ book that the late Judge Elliott Wilk was going to hold a formal hearing over the cocaine evidence and RIA test prior to the trial. The Hospital objected and the hearing was never held.

This was a tragedy. Were there to have been a hearing with a complete statistical analysis of what Libby’s data was really saying about the cocaine, it would have quickly become obvious that nothing in this data could be used as evidence before the Jury. There were no confirmatory tests done at all and therefore the data was meaningless. To have been consistent with himself, Dr. Rieders should have said that the RIA test results did not meet “forensic evidence acceptability.”

The Case For New York Hospital, Its Doctors, and Medical Trainees

In the almost 11 years between Libby’s death and the trial, there was a concerted and very public campaign on the part of many interested parties to create a believable case exonerating New York Hospital. The following is a critical analysis of the key elements of this campaign contrasted with the scientific evidence:

A. New York County Medical Examiner

On May 8, 1984, 3 days after Libby Zion’s death (as documented by Natalie Robins in her book, The Girl Who Died Twice (p. 121), the Medical Examiner’s Office “reported a positive reading for cocaine, considerably stronger than a borderline one…” This was in reference to the ‘gray area’ reading of the premortem blood.

The reader of this document now knows that this “positive reading” was nothing of the sort. It may have arisen out of a misunderstanding in the New York County Medical Examiner’s Office about how to interpret the RIA test. Alternatively, given the expense of doing the required confirmation tests (GC and/or MS), the Medical Examiner’s Office may have had a policy of calling all such readings – that is, readings between the negative and positive cutoff – as positive. This ‘administrative’ policy may have been needed to save New York City money.

But, whatever the reason, this scientifically unsupportable conclusion became the basis of the entire defense of the Hospital, its Physicians, and its Medical Trainees.

Lorenzo Galante, M.D., “Assistant Chief Toxicologist” in the Medical Examiner’s Office, is reported to have said (Robins, p. 123), “The RIA test is one of the most highly sensitive tests available in his office. Dr. Galante said that the Gas Chromatography test was used as a general screen for basic drugs rather than a confirmatory test for benzoylecgonine, and that is why it was not used to confirm the finding in Libby’s blood sample. The better RIA was proof enough.”

Once again, we do not know why the “Assistant Chief Toxicologist” said this. It was either a direct lie or an unintentional mistake. In any case, this opinion was exactly wrong as the observant reader now knows. The RIA test is the screening test requiring confirmation by Gas Chromatography or Mass Spectrometry (see Teitz, above, p. 341.)

On page 135, Robins reports an unnamed toxicologist as having said, “scientifically there was cocaine in her bile and that, in fact, the bile is tested only when there is a prior indication of cocaine either in the blood or urine. This toxicologist also believed that Libby had ingested up to 100 mg. (milligrams) of cocaine, anywhere from just before she died to 12-16 hours before her death.”

This “unnamed toxicologist” was also exactly wrong. The bile readings were between the negative and positive cutoff and therefore required confirmation before they reached the level of acceptable forensic evidence (see Rieders, above.)

B. New York Hospital’s Public Position

Robins reports a press release from the Hospital given on January 14, 1987. (p. 206) The release states, “perhaps the tragic death of the young person involved here will teach all concerned of the danger of illicit drugs, such as cocaine and, indeed, of the risks in the use and concealment of usage of a vast range of drugs, licit and illicit.”

Subsequent to the scientifically invalid reports coming from the New York Medical Examiner’s Office, New York Hospital – whenever the subject was raised – tried to take the ‘high road.’ Their position was that Libby Zion was a cocaine abuser who did not tell the medical housestaff that she had taken a large amount of ‘coke’ in the few hours before coming to the Hospital.

The Hospital’s defense was that the doctors and trainees did not ‘know’ what the cause of Libby’s symptoms were, because they claimed she denied taking cocaine. The Hospital claimed that were they to have known of her cocaine use they could have saved her life.

And so, as Libby’s father Sydney Zion was trying to find answers, the Hospital’s statements were all designed to blame Libby Zion. The high minded press release served to inject a note of both compassion and warning so that others would not repeat the same ‘mistake’ made by Libby Zion.

C. The District Attorney of New York County – Robert Morgenthau

Sidney Zion wanted criminal charges to be brought against the staff whom, he believed, was responsible for his daughter’s death.

The well-known New York County District Attorney, Robert Morgenthau, then ‘went public’ regarding his position. His statement further bolstered the Hospital’s defense. The District Attorney’s comments were particularly harsh.

On page 201 of The Girl Who Died Twice, it was reported that the New York Times quoted the District Attorney as saying, “there was insufficient evidence, especially regarding the cause of death, upon which to base an indictment.”

In fact, the District Attorney was wrong regarding the case for criminal charges. As will be developed below, one of the clear facts in the case was that Libby Zion died because of the restraints. The use of restraints was in violation of the well-established New York State Law on Restraints (New York State Mental Hygiene Law On Restraints §33.04). The fact that Libby died in violation of the Law should have changed her case from one of simple negligence to that of criminal negligence (criminally negligent homicide). That is, while unintentional, Libby’s death was due to the breaking of a Law and therefore criminal in nature.

The District Attorney, whose responsibility it was to do so, did not acknowledge the issue of restraints. This was a tragedy. Untold numbers of patients have been killed in the Nation’s hospitals and nursing homes over the years by the illegal application of restraints.

On page 204, Robins reports Robert Morgenthau as having said, ” there’s no doubt about it being a tragedy, but she lied about her drug use. With drug users, reactions are unpredictable and often fatal.”

Here the District Attorney ‘convicts’ Libby Zion of lying and he does so on the basis of false evidence. In short, he was saying that her death was her own fault.

On page 207, Robins quotes Robert Morgenthau as describing Sidney Zion’s denial of his daughter’s drug use, “he’s living in his own world. He is unrealistic. He’s got to have a devil. Sydney Zion had a tremendous sense of guilt.”

Here, the District Attorney’s attack against Sidney Zion was cruel. Although he did not have the scientific background to know it – and in fact no one from the scientific community came forward to correct the incorrect science behind the Hospital’s defense – Libby’s father Sidney Zion was totally correct. Zion’s former lawyer, Ted Friedman, alleged that there were experts who would have rebutted the defense testimony but who were intimidated into silence by their colleagues (Robins, p. 249.) There was no forensic evidence at all that Libby died from cocaine.

D. The Jury

The Jury delivered its verdict on February 6, 1995, almost 11 years following Libby Zion’s death. The verdict was a finding that Libby was 50% responsible for her own death because she did not tell the staff about her cocaine use.

On page 284 (Robins), Juror Michelle Winfeld reflected, “I believe the verdict was just because I went on the evidence. Thank God a verdict is final.”

On page 276 (Robins), the foreperson of the Jury, Janet Dubin said, “There wasn’t enough evidence to dispute the fact that she had done cocaine.”

Dr. Rieders, and others, had succeeded in convincing Jury members that they had enough understanding to make decisions about scientific data. They had no way of knowing that they had been presented false explanations that withheld crucial information. Dr. Rieders was well aware that the ‘evidence’ he was presenting needed to be discarded because it was obtained from a “screening test” that required confirmation by more reliable methods.

What Killed Libby Zion – A Reexamination Of The Evidence

Robins reports (p. 132) that the pathologist who performed the autopsy on Libby Zion, Dr. Jon Pearl, wrote to the Department of Health’s Division of Vital Records, and included a complete amendment to Libby’s Certificate of Death. The amendment said, “acute pneumonitis 4 days following dental extraction and in the course of treatment with erythromycin. Hyperpyrexia (high fever) and sudden collapse shortly following injection of meperidine (Demerol) and haloperidol (Haldol) while in restraints for toxic agitation (emphasis added.) History of therapeutic phenelzine (Nardil) injection. Unclassified.”

Libby Zion died due to the illegal application of restraints. The restraints were applied as a substitute for treatment of a patient who had a rapidly rising temperature. Rather than sending her to an Intensive Care Unit where she should have received intravenous antibiotics, vigorous hydration, and close monitoring of her vital signs – and where a cooling blanket should have been put over her at the first sign of a dangerous rise in her fever – she was placed on an open ward, essentially abandoned and alone. She was put in 5-point physical restraints and further given an injection of Haldol, a ‘chemical restraint.’ There was only an unlicensed trainee Intern with several floors of patients to take care of. In this situation, Libby’s condition couldn’t possibly be monitored. In short, Libby ‘burned up’ while in illegal restraints.

Why was Libby Zion’s fever rising so precipitously? It is only possible to speculate about this because she received neither appropriate diagnosis nor treatment. Following are several possibilities as to why her temperature was rising:

  1. Cocaine.
    It was possible to speculate that cocaine was the cause of the fever. But it was still necessary to prove that cocaine was in her body. Since there was no confirmatory test, and since all her readings were in the ‘gray area,’ the cocaine issue was no more than a speculation.
  2. Drug Interaction
    Libby was given an injection of meperidine (Demerol) to sedate her 3 hours prior to death. This was dangerous since the combination of Demerol and the Nardil (phenelzine) she had been taking for depression can cause a precipitous rise in temperature. This was the position of the plaintiff, Sydney Zion, but it was also only a speculation.
  3. Extreme Agitation
    It was possible to speculate that Libby Zion’s extreme agitation was due to an interaction between phenelzine (Nardil) and the ephedrine in the decongestant Actifed she had been taking. Once again, only a speculation.
  4. Sepsis
    Sepsis is defined as the presence of an overwhelming infection in the blood stream caused by bacteria, resulting in systemic signs and symptoms.

Libby Zion had pneumonia at autopsy. She had an infected molar tooth extracted less than 4 days prior to her death; streptococcus and other bacterial strains were found in her lungs at autopsy. It was possible that the rapid rise in the temperature was due to a developing infection in her body. She had small hemorrhages under her skin (petechiae) consistent with developing infection, as well as being generally very flushed. She repeatedly told the housestaff that she had a temperature of 105 degrees Fahrenheit that day. Robins reports Libby as saying that she was “burning up inside.” (p. 16)

Libby’s signs and symptoms at New York Hospital were consistent with sepsis. She had decreased urine output, low blood pressure, violent shaking, confusion, delirium and a rapid heart rate. Her high white blood cell count showed increased types of cells that are consistent with an infection.

Whatever the cause of the lethal rise in her temperature, the fact is that the unsupervised medical trainees did not recognize that Libby Zion desperately needed to be cooled down quickly. The intern, Dr. Luise Weinstein, rejected the suggestion that a cooling blanket be used. (Robins, p. 111) Instead, the housestaff saw her as a ‘hysteric’ who needed restraining. Libby Zion came into New York Hospital with a temperature of 103.5 degrees F. Her temperature near death was around 106 degrees F., Sidney Zion was told; the death temperature “…was 108, in which case Libby’s internal temperature would have been 110 if the reading had been taken rectally.” (Robins, p.117) Discovering this, the nurses frantically attempted to pack her in ice but it was too late.

Dr. Gregg Stone, the First Year Resident caring for Libby, said of Sidney Zion (Robins, p.234), “The case wasn’t about malpractice, it was about a political and personal agenda that one man had.” Robins said that Dr. Stone told her that “he (referring to Sidney Zion) manipulated the system, to create an emotional and political juggernaut that turned into the Libby Zion case.” She further reports that Dr. Stone told her, “Sidney Zion should look within and try to live with what was going on in Libby’s life. Whether or not he knew of it, I don’t know.”

New York Hospital, its Physicians and Medical Trainees realized it was important to have a defense that deflected the Jury’s attention away from the negligent care Libby Zion received. The cocaine defense served New York Hospital well. Predictably, the Jury turned on Libby Zion, even in her death. However, were Dr. Reiders to have been scientifically truthful, the entire cocaine defense should have been thrown out as being forensically unacceptable.

Dr. Stone’s tone in describing Sidney Zion mirrored that of the District Attorney’s character assassination. Sidney Zion was not “manipulating” the system, but rather attempting to tell the “system” that it was at fault in the death of his beloved daughter.

In The Girl Who Died Twice (p., xiv), the former Dean of the New York Cornell College of Medicine, Dr. Robert Michels, stated, “I can see the headlines. New York Hospital Cornell Medical Center that killed Libby Zion discovers cure for cancer.” Ms. Robins stated he was “positively haunted by the Zion case.”

Yes, Dr. Michels was correct. His Hospital did kill Libby Zion. We should all be “haunted” by the case and what it tells us about our culture, and our political, legal, and medical system.

Stephanie Zoe Speken, M.S.
March 14, 2004

Copyright © 2004 by Stephanie Zoe Speken