In his first meta-analysis in the Proceedings
of the National Academy
of Sciences, Linus Pauling analyzed the findings of 4
placebo-controlled trials in which at least 0.1 g/day of vitamin C was
administered regularly to the study group (1971a).
In his second
meta-analysis in the American
Journal of Clinical Nutrition, Pauling
focused on the best 2 of the 4 trials (1971b;
Cowan et al. 1942;
Ritzel
1961;
Table 3 in Background).
Among the 4 trials included in Pauling’s meta-analysis
(1971a;
see Pauling 1972), the largest dose was used by Ritzel
(1961),
and Pauling based his quantitative estimations on this trial.
Ritzel found that the common cold symptoms in the vitamin C group were
31% shorter and the number of colds 45% lower in the vitamin C group.
Pauling also calculated the combination of duration and incidence,
‘integrated morbidity’ referring to the total
sickness days per person during the trial, and this was reduced by 61%
in the Ritzel trial (Table 3 in Background).
Pauling (1971a)
then modeled the
dose-dependency of vitamin C effect with exponential formulas for which
he took constants from the Ritzel trial. Pauling assumed that the main
problem in his estimation was inaccuracy caused by
‘experimental error,’ although he did note that
"The values are, of course, expected to depend somewhat on the nature
of the population and environment." However, even with these explicit
reservations he was far too optimistic. He could not imagine how great
the variations in the results would be in the forthcoming trials.
Neither did he consider the possibility that the effects observed by
Ritzel may have been caused at least in part by low dietary vitamin C
intake, in which case a smaller dose might have produced a similar
benefit, and in such a case modeling the vitamin C effect as a function
of the supplementary dose would be completely erroneous. Pauling
attributed the difference between the study groups entirely to the
large dose given to the treatment group. Furthermore, Ritzel carried
out his trial with schoolchildren in a skiing school in the Swiss Alps,
children who are not a good representative selection of the general
population even though technically the trial was good as it was
randomized, double-blind and placebo-controlled. Thus, when Pauling
extrapolated the results of Ritzel to all people (i.e., to children at
school and adults), he took a bold step and went wrong (Hemilä
1997b).
Pauling (1971a,
1971b)
put much weight on the ‘integrated
morbidity’ outcome and summarized the findings of trials by
this outcome in his later texts as well (1976a,
1976b,
1976c,
1986a). This
measure led Pauling to adhere strongly to the idea of regular
supplementation. However, this is not a good combined measure, since
the effects on incidence and duration/severity have quite different
patterns (Fig. 3 in Hemilä 2006), and it is thus more to the
point to analyze these
two outcomes separately. Thus, Pauling was qualitatively correct in his
conclusion that vitamin C does affect the duration and severity of
colds, and probably the incidence of colds in certain specific
conditions, but he was greatly over-optimistic.
It is worth noting that the Ritzel trial (1961)
falls to the group of 6
trials with participants under heavy acute physical and/or cold stress
that consistently found reduction in common cold incidence
(Hemilä 1996b;
p 48 in
Hemilä 2006; Douglas & Hemilä 2005;
Hemilä et al. 2007).
Thus, it was not
a misjudgment by Pauling to put the greatest weight on this trial, but
his error was to extrapolate the findings to the general population.
The other trial on which Pauling put great weight was the Cowan et al.
trial (1942;
Table 3 in Background)
which was carried out with schoolchildren during
the war years and probably the dietary vitamin C intake was low and in
this respect the benefit may be explained by the correction of marginal
deficiency as in the UK studies with schoolboys and male students
(Hemilä 1997;
pp
46-7 in Hemilä 2006).
As regards the errors in Pauling’s quantitative conclusions,
it should obviously be taken into account that essentially all of the
trials available today were carried out after Pauling worked on the
topic and, even more importantly, were carried out precisely because
Pauling popularized the topic (Fig. 2 in Background).
Without bold conjectures, progress in science is slow or non-existent,
and in this respect the accuracy of Pauling’s extrapolation
from the single placebo-controlled trial using regular 1 g/day
supplementation available to him in the early 1970s is of secondary
concern. Furthermore, Pauling’s own view of science was that
an occasional mistake, even when published, was not as bad as lowering
one’s sights to less challenging research (Lipscomb 1994).
Hemilä H, Chalker EB, Treacy B, Douglas RM (2007) Vitamin C
for preventing and treating the common cold [systematic review]. Cochrane
Database Syst Revs (4): CD000980