LETTERS 4. Shirlow MI, Mathers CD. Caffeine consumption lesterol levels. Int I Epidemiol 1984; 13:422-7. S. van Dusseldorp M, Katan MB, Demacker PNM.
and
TO
serum
Effect
cho-
of decaf-
Reply Dear
I would
like
to respond
to the
comments
of Pearson
for the
authors. First, our has no effect
randomized trial on the lipoprotein
point
was
Second, is the
trial
involved
number
emphasized
Pearson
sign
of any
results
was
of randomization
by
classic
group
assignment.
of subjects
in their suspicion of assignment to this group. was developed over 2 y and underwent intense
in the decaffeinated
(NIH)
group group
Questionnaires
that
ofHealth
cal-
as to the
or decaffeinated-coffee
vealed
Institutes
The
power
correct design
the National
Our
were blinded
caffeinated-
de-
participants.
subjects placed in the no-coffee assignment and the decaffeinated
of their
60%
Trial
investigation.
many
and subjects into
those group
issues.
clinical with
determined
The investigators
suspicious
good
coffee study.
section.
trial-design
design
of participants
that caffeinated measured in our
discussion
several
a randomized
groups. Certainly, were aware oftheir
indicates variables
in the
addresses
foundation
culations.
was
Third,
the
was
issue
of compliance
determined
Coffee Dear
DS. Coffee
and
article
coffee
Study
by the
is an important difference
lipoprotein
one.
by
Superko
diet
records.
and ingestion There
groups. There Fourth, the
NIH-funded
between
groups.
clinical leading variables
returned
was
was
no
study This
at each
determined
Compli-
between
the
to test
is the purpose
ofhaving
calculations the main
were performed to design hypothesis. We believe that
et al (1)
on
coffee
and
regular
could
take
of coffee drinking led to a decline 2) Under Laboratoryprocedures C concentration was formula for calculating LDL-C
terol,
C is cholesterol, and
TG
In Results observed
coffee
After
as before.
total HDL
With
the position
that
in LDL-C. it is clearly
stated
Berkeley,
C - (HDL-C
+
they
the no-
the cessation that
LDL-
practice. The is as follows:
TG/5)
is high-density-lipoprotein
choles-
is triglyceride. it is stated
for changes
that
Superko
Center
Berkeley Laboratory of California CA
94720
Reference I . Superko HR, Bachorik cholesterol measurements.
“No
in concentration
3)
In Table are
PS, Wood PD. High-density JAMA l986;256:27 14-7.
significant of plasma
differences triglycerides,
1 (Baseline
definite
lipoprotein
with
the
in the
other
small, 4
237
but
)
intakes
are
two
of the
groups.
Their
Admittedly, ofthe
6)
and
two measurements completion of the
three
TGs
the
differences
Cups per day are described as follows: mL of fluid coffee” as opposed to the
Only upon
calculated groups)
decaffeinated-coffee
group are
higher,
values are Also, their
differences
that
are
is surprising.
“each cup equal to generally accepted
definition of a cup constituting 150 mL. 5) Statistical analyses were applied to group tween baseline and end-of-study measurements lowing each subject to be his own control.
were
the
are higher, their HDL-C B values are higher.
lower.
it is the consistency
or apo-Asubjects.” If ofthe equa-
could
characteristics
differences
their total C concentrations lower, and their apolipoprotein carbohydrate
calculated. This is the usual the LDL-C concentration =
in
lipoprotein
H Robert Research
as compared
one
groups
subtle
total cholesterol, HDL cholesterol, HDL2 cholesterol, I for the decaffeinated-coffee group vs the control there were no differences in any ofthe components
lipoprotein
showed
(LDL-C).
to drink
of change
that could answer with smaller pop-
ulations may lack the statistical power to reveal differences induced by components of coffee.
all,
drinkers
in low-density-lipoprotein-cholesterol at least
addressed
cholesterol
1, the caffeinated-coffee
continued
was
control
a trial studies
there
group
between
for significance
a decline
in Figure
and
in compliance
between groups. of the study that
was
Compliance
trials. Testing for a change from baseline can be misbecause it does not take into account the multitude of that can alter lipoprotein measurements (1). The power
Lawrence University
by
visit.
by questionnaire
difference
was no crossover statistical design
in this
were
We believe
in weight
no coffee
The trial review by
group
Section.
dispensed
with
why,
where
Br
coffee
tion for the calculation of LDL-C, how value have shown a significant increase?
merely
cholesterol.
SF, Arnesen E, et al. The Tromso Heart Study: and serum lipid concentrations in men with a randomized intervention study. Br Med I
cholesterol raises many questions requiring further explanation before the authors’ conclusions can be accepted with confidence. 1) From the data presented it is impossible to understand
coffee
and serum
10. Forde OH, Knutsen coffee consumption hypercholestcrolaemia: 198S:290:893-S.
Cholesterol
Sir:
The
9. Arnesen E, Forde OH, Thelle Med I 1984;288:1960.
re-
our investigation is the only US coffee investigation funded the federal government and not by commercial interests. ance
8. Fried RE, Pearson TA, Levine DM, et al. The effect offiltered coffee consumption on plasma lipids: results ofa randomized clinical trial. JAMA 1992;267:81 1-S.
to TA Pearson
Sir:
This
605
EDITOR
differences instead
of lipids were obtained: study. Thus, one does
not
beof alat entry know
Downloaded from https://academic.oup.com/ajcn/article-abstract/56/3/605/4715450 by East Carolina University user on 15 January 2019
feinated coffee versus regular coffee on serum lipoproteins. Am I Epidemiol 1990; 132:33-40. 6. Zock PL, Katan MB, Merkus MP, et al. Effect ofa lipid-rich fraction from boiled coffee on serum cholesterol. Lancet 1990;335:l23S-7. 7. Bun- ML, Gallacher JEI, Butland BK, et al. Coffee, blood pressure and plasma lipids: a randomized controlled trial. Eur I Clin Nutr l989;43:477-83.
THE