Download Lectures on Biostatistics (1971).
Corrected and searchable version of Google books edition

Download review of Lectures on Biostatistics (THES, 1973).

Latest Tweets

This post was one of the earliest on my old improbable science page (late 2002). It has been reproduced here because of the interest attached to Stephen Senn’s analysis of what appears to be a rather dubious trial, and because the subject purports to be regular medicine rather than alternative quackery. It is unlikely that all of the links will still be valid.

During research for a talk on quack medicines. I came across several web sites that were selling the common co-enzyme, NADH, for a whole variety of ailments. One of these sites, Wizards Gate, no longer sells NADH (“because the profit is too small”). but continues to carry the following advertisement for it.

A recent Georgetown University Medical Center clinical pilot study among CFS patients reports that participants taking ENADA® were four times more likely than those taking a placebo to experience a reduction in symptoms.

Please click on the links below for more information.

pilot study graph.jpg

What this does NOT mention is that the huge column on the left represents eight people out of 26, and the tiny one on the right represents two people (out of 26)!

The reader is then referred to ImmuneSupport.com, who describe NADH as follows (with my comments in italics).

  • Improves cellular energy [This, as it stands, is meaningless]

  • Elevates mental clarity [There is no evidence that this is true]

  • Promotes mental alertness and concentration [Again, there is no evidence that this is true]

The ImmuneSupport site goes on to say

“NADH is proven to trigger energy production through ATP generation [true].
A naturally occurring coenzyme found in all living cells, NADH helps supply cells with energy.[true].
Thus, as NADH levels rise in the body, the cells become more energized, helping the body feel stronger and more vitalized. [This sentence confuses energy in the technical sense that is used in physics, with the everyday use of the term to describe a mental and physical state, In fact there is no reason to think that NADH “helps the body feel stronger” to any greater extent than a placebo].

Other sites that promote NADH heavily are Enada.com, and the Menuco Corporation, of which more below

The clinical trial referred to in all of these advertisements is Forsyth, L. M., Preuss, H. G., MacDowell, A. L., Chiazze, L., Jr., Birkmayer, G. D., & Bellanti, J. A. (1999). Therapeutic effects of oral NADH on the symptoms of patients with chronic fatigue syndrome. Ann.Allergy Asthma Immunol. 82, 185-191.

It is very clear from the letter below that this trial is flawed, but it is impossible to see exactly how flawed it is because the authors, in breach of all guidelines on such matters, have refused, for over a year, to make available the raw data that would allow a correct analysis to be done.

Annals of Allergy, Asthma, & Immunology ©2000;84:639-640

To the Editor:

We should like to comment on the paper by Forsyth et al. 1 This paper claims, on the basis of a randomized double-blind clinical trial, that NADH is effective in the treatment of chronic fatigue syndrome. It seems that there are several problems with the work described in this paper.

  • On page 187, the ‘raw symptom score’ is defined as a score from a questionnaire normalized to lie between 0 and 1 (the questionnaire had 50 questions, each scored 1, 2, 3 or 4, so raw scores are in the range 50 to 200). This was normalized to lie between 0 and 1 by first subtracting 50 and then dividing by 150. However, patients with chronic fatigue syndrome are said to have scores ranging from 10.9 to 55.8, so the normalized score has presumably been expressed as a percentage, whereas in the case of normal patients the values of 0.01 to 0.13 suggest that proportions have been used. Furthermore, even using percentages rather than proportions, we cannot reproduce the ranges given for patients. For example, a raw score of 66 yields a normalized score of 10.7 whereas the next possible value 67 translates to 11.3.  How in that case could Forsyth et al get 10.9? Similarly 133 yields a score of 55.3 but 134 yields 56.  So what raw score can have produced 55.8?
  • It is stated that, “an individual was considered improved if s/he demonstrated at least a 10% improvement,” but then it is also stated that, “the degree of improvement for the results to show a positive effect was an improvement by one point in 10 questions or by 2 points in 5 questions” (p 187). This corresponds to a decrease in the raw score of 10 points, not a “10% improvement.” Table 5 gives the scores (presumably normalized score x 100) for each of the 26 patients on each of the tests. It is stated that 8/26 improve on NADH and 2/26 improve on placebo. This appears to be a change in normalized score (x 100) by 10, not to a change in raw score of 10.
  • The method of analysis given on page 187 is a test of difference between two independent estimates of a binomial proportion. This is entirely inappropriate for the analysis of a crossover trial. 2
  • Most importantly, nothing whatsoever is said about which of the 26 patients in Table 5 received which treatment. Without knowing which treatment was given in which period it is impossible for the reader to repeat the analysis using methods more appropriate for crossover trials. The simple addition of one column here, giving the sequence to which the patient was allocated, would have permitted others to have performed the necessary calculations.

Drs Forsythe and Bellanti were asked in June 1999 to provide the raw data, in particular the crucial information on sequences that would allow re-analysis of their results. Despite several further requests we have not had any response. This is made all the more disturbing in view of fact that their paper is being used to justify vigorous marketing 3 of NADH as a treatment that “Enhances energy naturally, elevates mental clarity, improves alertness, and concentration,” which treatment is also being claimed by some to be useful for everything from improving athletic performance to Alzheimer’s disease. In view of this it seems that the least that it is owed the readers of your journal is for the authors to produce the original data so that their claims can be verified.

David Colquhoun

Stephen Senn

Professor of Pharmacology, University College London, Hon, Director Wellcome Laboratory for Molecular Pharmacology, Dept of Pharmacology, University College London, London, United Kingdom and ,

Professor of Pharmaceutical and Health Statistics, Department of Epidemiology & Public Health,
and Department of Statistics, University College London, London, United Kingdom



1. Forsythe LM, Preuss HG, MacDowell AL, et al. Ann Allergy Asthma Immunol

2. Senn SJ. Cross-over trials in clinical research. Wiley, Chichester,

3. See, for example, http://www.enada.com/,http://www.wizardsgate.com/
and many others.

Professor Bellanti’s response:

We have reviewed the two “Letters to the Editor” from Professors Colquhoun and Senn and from Professor Lorden and wish to respond to both. Recognizing that there are various methods and points of view concerning statistical analyses, since this was a preliminary study and not a definitive one, we chose to analyze the data in the manner described. This was done to determine whether there was a basis for the therapeutic efficacy of NADH in chronic fatigue syndrome and, if so, then to proceed to a larger, more definitive study. We feel that we have accomplished this goal in obtaining sufficient evidence to proceed further. We are grateful for the constructive elements of the readers’ comments and we shall take these into consideration as we plan the definitive studies.

Joseph A Bellanti, MD

Professor of Pediatrics and Microbiology-Immunology

Georgetown University School of Medicine

Washington, DC

This reply from Professor Bellanti stands in contrast to his own statement in a recent report presented to Annual Meeting of American College of Allergy, Asthma and Immunology in which he says “We have recently demonstrated the clinical effectiveness of reduced nicotinamide adenine dinucleotide (NADH) in a group of 26 patients with CFS in a double-blinded, placebo-controlled, crossover study”.

There is no mention here (or in any of the advertisements) of the study being “preliminary” or “not definitive”. (The measurement of 5-HIAA urinary concentration as a possible predictive marker of disease activity and therapeutic efficay of NADH in the chronic fatigue syndrome. Joseph A Bellanti MD; Linda M Forsyth MD; Ana Luiza MacDowell-Carneiro MD; Dawn B. Wallerstedt, MSN, FNP; Harry G Preuss MD and Georg D Birkmayer MD. PhD ).

This reply gives no reason at all for refusing to disclose the raw data, to allow others to check the conclusions. It does say that “there are various methods and points of view concerning statistical analyses”, but does not (unsurprisingly) refer to anyone who thinks that a test of difference between two independent estimates of a binomial proportion is an appropriate way to analyse a cross over trial.

Financial interests

The editor of Annals of Allergy, Asthma, & Immunology, Dr Edward O’Connell, tells me that the journal requires authors to disclose any financial interest in the work, and that no such interests were declared in the case of Forsyth et al. This seems very odd, since the address of one of the authors, Dr G.D Birkmayer is given as Birkmayer Pharmaceuticals (Vienna), who market the ENADA brand of NADH, on which Dr Birkmayer holds patents. Birkmayer Pharmacuticals supported the work by a grant, and analysed the data

Latest news The case of the Bellanti paper was referred to the Georgetown University’s Research Integrity Committee for review, in February 2001. That committee has reported that “no violation within the purview of the Code has been committed”. In the light of the facts concerning (a) improperly analysed data, (b) non-disclosure of financial interest and (c) the refusal to allow independent analysis of the data, it seems the the University has a novel view of what constitutes research integrity!


Print Friendly

Leave a Reply