Virginia Kraus - Fraud, Osteoarthritis, and Vitamin C

SUBJECT: Virginia Kraus paper on Vitamin C and Osteoarthritis - a critical review
"Ascorbic Acid Increases the Severity of Spontaneous Knee Osteoarthritis in a Guinea Pig Model" in ARTHRITIS & RHEUMATISM Vol. 50, No. 6, June 2004, pp 1822–1831
Other researchers involved in this: Janet L. Huebner, Thomas Stabler, Charlene M. Flahiff, Lori A. Setton, Christian Fink, Vladimir Vilim, Amy G. Clark, Tom Tlusty, Greg Samsa, Jeroen DeGroot, Connie Ba, Nicole Verzijl, and Johan TeKoppele

Did Virginia Kraus twist scientific data to make news headlines by falsely claiming vitamin C causes Osteoarthritis?  The researchers restricted vitamin C in the diet of the low-dose group of guinea pigs' to levels that are about 1/10th of what is considered healthy, apparently to the point of causing 3 of the 15 to die from scurvy.  These 3 were excluded from consideration in the osteoarthritis scores which biased the results.

Virginia Kraus makes statements in the Duke University news release that appear contrary to the published research.  The difference in the press release and the research is biased in a way that sensationalizes the research rather than accurately reporting the results.  The research article itself appears sneaky in the way it reports the results.

1. The text under figure 5 implies the two images it compares are typical examples from the high and low dose groups.  But the body of the article states that the low dose example is of the guinea pig with the least amount of osteoarthritis in that group.  There is no indication of how the other example was chosen.  So the images are comparing the best example from the low dose group and possibly the worst example from the high dose group to make it appear as if there was a big difference between the low and high dose group.
2. Contrary to the press release, if you understand statistics and see the p=0.06 in the top left of figure 4, you will see that there was no statistically significant difference in total osteoarthritis scores based on the amount of vitamin C ingested.  The article never comes out and says this and obscures this information by using the phrases "low", "medium", and "high" when referring to blood concentration (which did show a slight difference IF the ones that died from scurvy are ignored) to make the reader think they're referring to low, medium, and high DOSE groups.  The whole point of the study was to find a correlation between vitamin C in the diet and osteoarhtritis.  No statistically significant difference was found, but the average reader of the article would not know this.
3. The jumbled data in figure 7 shows the osteoarthritis scores of the guinea pigs.  As shown in the text plot further below, the trend is not very convincing.
4. The low dose group weighed 10% less than the others, indicating ill health and possibly reducing stress on the knee joints.  The lack of tissue-building capabilities in the low-dose group may have also directly biased the histological results.

Fraud: In order to be fraud, there must be the intent to deceive for personal gain at the expense of those being deceived.  Did Virginia Kraus, et al knowingly lie in the research and news release at the expense of the public while in persuit of public funds?  She and Duke Universtiy answered my first set of emails, but thereafter stopped replying when they realized I discovered what happened and made them aware of it.  I will be glad to change this page if they wish to correct any of this information.

Virginia Kraus and Duke University did not wish to correct, clarify, or admit to the severity of the mis-information.  Looking at data in different ways AFTER the data is collected to "find" something is not statistically or scientifically correct unless the statistics are adjusted to reflect the tumbling around of the data.  They could have just as easily tumbled the data around to find the opposite conclusion.  For example, they could have counted the ones that died as something as bad as arthritis instead of ignoring them.  In order for statistics to be applicable, you must specify what you intend to measure for statistical significance before the measurement is made.  Otherwise, you can look at meaningless data 20 different ways, and statistics tells us there is a 50% chance that one of those interpretations will reach 95% confidence (1 in 20 occurs by chance alone).  Therefore a researcher seeking to prove something can look at data 100 different ways until she finds something she likes that might lead to more grant money.  In this case, the dose of vitamin C was not statistically correlated with osteoarthritis, but there was a slight correlation with blood plasma levels IF the dead guinea pigs were excluded.

The overlap (i.e., meaninglessness) between the medium and high dose groups' overall osteoarthritis scores is shown below (adapted from figure 7). Each x represents an individual guinea pig's osteoarthritis score.  The rest of the article very carefully obscures this data, but this data is the only relevant information.

Osteoarthritis Score:
high dose group:  xxxx x xxxxxxxxx   xx
medium dose:      xxx xxxxxxxxxxx  x
low dose:      xxxx xxxx xxxx      (and xxx died)

==============================================

To the news release contact about the Virginia Kraus paper and news release:

Becky,

Does Duke University review the accuracy of its news releases before release? Is there a retraction or re-release mechanism? You are probably aware the media disseminates them verbatim without looking at the research. The media assumes the information is accurate and ubiased. After reviewing the paper on vitamin C and osteoarthritis and comparing it to the news release, i found several egregious errors as described below. The lead author, Virginia Kraus, appears to be a primary author of the news release, but has not responded to my complaints with an explanation as to why the news release is not supported by the research. The errors in the news release are as follows:

1) "Because this study indicates potential drawbacks to long-term use of high-dose vitamin C supplements, adults should not supplement their dietary vitamin C levels above the recommended dietary allowance (RDA), Kraus said."

In no way, shape, or form did the research justify this statement. The research was an 8-month study on 46 guinea pigs (3 of which who were not taking the high doses died). There is an enormous amount of research in humans over long periods of time that indicate supplemental vitamin C pills are very beneficial to the health of the many people who take them.

2) "The high dose was 150 milligrams per day, an amount shown to protect against surgically-induced osteoarthritis in a short-term guinea pig study. The equivalent human dose is 1,500 to 2,500 milligrams per day."

Figure 3 of the paper clearly indicates 2,500 mg causes less of an effect on humans than 75% of the guinea pigs taking 150 mg. If you extrapolate figure 3, you can easily see humans would have to take approximately 10,000 mg for it to be equivalent to 150 mg vitamin C in a guinea pig. If one were to use the simpler, reliable, and standard method of making the comparison based on body weight, humans would have to take 65 times [150 pound person divided by 2.3 pound guinea pig] more than guinea pigs which also comes to 10,000 mg [to be comparable to guines pigs taking 150 mg].

3) "The researchers found that high-dose guinea pigs developed more cartilage damage and had more bony spurs form in their knee joints than did the medium- and low-dose groups." Also the title "Vitamin C Worsens Knee Osteoarthritis in Animal Study"

Looking at figure 4, you can see the high dose guinea pigs' overall osteoarthritis score did NOT show a statistically significant difference from the medium dose and high dose groups. Blood concentration of vitamin C did show a small but statistically significant difference, but that's mixing and matching the different groups and it is NOT based on the amount of vitamin C that was ingested. In figure 7 you can look at the data directly and let common sense be your guide rather than trusting or being misled by statistics. There is an ENORMOUS amount of overlap between the two groups.

Frankly, the title of the news release and the egregious errors appear to be strongly biased and downright dishonest in an attempt to excite the media and make headlines. This does not engender my respect for Duke University.

In consideration of the facts above and the severity by which the news release tries to falsely alarm the public (and indeed HARM the poublic) by trying to convince people vitamin C should not be supplemented, I am requesting a retraction in the form of a news release (that attempts to make just as many headlines) to try to undo some of the damage that has been done by issuing a false alarm. I do not think Duke University will do this because its news releases appear to be for self-aggrandizing rather than to honestly inform the public. An investigation into the standards used to guide news releases would be nice. Is there any check on the accuracy of news releases before they are printed? Do Duke University Medical Center researchers that issue press releases receive funding from entities that have an anti-nutiritional supplement bias that may be affecting the news releases?

I do not have a financial or career-oriented interest in this, nor do I have any ties to anyone that may be harmed by the news release other than family and friends who may choose to stop taking vitamin C supplements based on Duke University news releases rather than based on Duke University research. I am just a messenger. Is Duke University just a messenger, or something more?

Scott Roberts

=====================

excerpt from 1st email to Virginia Kraus:

My first complaint is about the Duke University news release stating 150 mg in guinea pigs is comparable to 1,5000 to 2,500 in humans. Animals average around 150 mg vitamin C per kg per day without much deviation with the exception of animals who do not make their own vitamin C and in which case they are grouped around 40 mg/kg/day with the exception of modern humans that no longer follow a paleolithic diet (which provided 2,000 to 3,000 mg (or approx 40 mg/kg)). So, 150 mg for a 1.5 pound guinea pig would correspond to 15,000 mg in a 150 pound human. The RDA equivalents for guinea pigs and primates is based on observed optimum health and lifespan in controlled and measurable environments whereas the RDAs for humans are based on the minimum amounts needed to prevent an adverse health effect. 30 mg in a guinea pig should be about 3,000 mg in a human, as it is with other primates (based on body weight) as evidenced by their RDA equivalents. I strongly advocate that humans should not be deprived of the nutrition that is given to primates in captivity. Those with the most experience and research into vitamin C advocate 2,000 to 4,000 mg a day. According to Patricia Hausman, the kidney stone problem has not been reported from doses less than 4,000 mg/day. Pauling believe 2,300 to 10,000 mg/day would provide the optimum health. I have not been able to find consistent info on the possibility of iron absorption at high doses.

===================

2nd email to Virginia Kraus:

Hi,

It's possible that blood concentration is an excellent way to compare, but i would have to see hourly concentration data. If the 30 mg/d group has the same blood concentration as 200 mg/d in humans then where does all the guinea pigs' vitamin C go? That's only 6.5 times more vitamin C for humans who have 65 times more blood and body weight. It's possible humans are 10 times more efficient at retaining their vitamin C at these dosages via their special kidney vitamin C pumps. This efficiency indicates either our ancestors got 10 times less vitamin c in the diet compared to guinea pigs or that 200 mg/d is much more important to human survival than 30 mg/d is to guinea pigs. The body weight comparision is a strong argument because vitamin C primary bulk function is to support the same extracellular matrix for all animals. Going by your figure 3 it may take as much as 2,500 three times a day for humans to have as high a blood concetration as the 150 mg/d in guinea pigs, so i do not understand the news release that equates 150 mg/d in guinea pigs to 1,500 to 2,500 in people.

I have had more time to closely review your paper. It seems to be a very minor difference between the 30 mg/d and 150 mg/d groups after plotting the individual OA scores shown in Figure 7. There is an extraordinary amount of overlap. I am surprised the news release says there was a difference between the two groups when you state in Fig 4 that there was not a statistically significant difference in the overall OA scores based on dose. The osteophyte and blood concentrations showed a statistically significant but minor difference, but if you have 20 different ways to look back on the data rather than sticking to the one way that should have been specified in a prospective protocol, then there is a ~25% chance you will find two p<0.05 "statistically significant" observations that are really just random fluctuations that are to be expected. For example, we can look at the study and claim the high dose really decreases the chances of OA if we say the deaths of the 3 in the low and medium dose groups were equivalent to bad cases of OA (afterall, the joints no longer move on their own). At the very least we can say the overall health of the high dose group was better and alert the press that everyone should therefore take 10,000 mg/d. The most surprising thing is the press release that claims the research indicates people should not take any vitamin C supplements. It was a non-prospective review of data on 43 guinea pigs over 8 months. There are enormous studies over many years in humans that show people live 5 to 10 years longer when they consume 1 to 2 g/d. This is my 5th way of trying to to show that 30 mg/d in guinea pigs is comparable to 2,000 mg/d in humans. 1) paleolithic diet was at least 2,000 mg/day. 2) vitamin C per body weight is consistent across species 3) primate RDA equivalents are set to optimum health around 2,000 mg per 70 kg while human RDA is by definition not set for optimum health 4) evolution theory noted above 5) large human studies. If people believe the press release and stop taking their 1,000 mg/d it may cost them on average 5 years of their health and life.

Figure 5 would have been much more interesting and fair if it had also shown the worst case OA in the lose dose group because one of them (as indicated in Fig 7) was almost as bad as the worst 2 high dose members.