BFDG-FCC April 2006

1
BFDG-FCC April 2006 Symposium Individual
differences in eating behaviour can they explain
variation in adiposity? Getting evidence for
most and least fattening local eating customs
from individuals reports in their cultures
terms David Booth, Psychology, U Bham,
England Louise Thibault, Nutrition, McGill U,
Quebec Caroline Chesneau, INA, Paris,
France Seolhyang Baek, Nursing, Dongguk U, S Korea
2
Theory 1. A subcultures consensus descriptions
of customs objectify communicable patterns of
activity. 2. Autobiographical memory can time
episodes of an activity as accurately as a
diary record can. 3. A sustained change in
frequency of a custom of eating or moving
about will change body weight towards
asymptote over a few weeks. 4. Change to a
frequency that is sufficiently prevalent in a
culture will be maintained indefinitely, along
with the asymptotic change in body fat content.
3
Method (with some short-cuts for this pilot
study) 1. Consensus descriptions elicited in one
culture (UK) were localised to another
culture (Quebec French) by investigators as
participant observers, checked by
back-translation to Englands language and
environs. 2. Exact current frequency is given by
the reciprocal of the time difference between
the last two occasions. 3. Effect on weight of
maintained change in frequency of a custom
after the first week (or 2 weeks) was
measured by correlation between the changes, 4.
Effects of initial and changed custom frequencies
on behaviour change maintenance (and weight
change maintenance for an effective custom)
were expressed as dose- response slopes.
4
Design Convenience sample of French-speakers in
Montreal 21 recruited 14 completed the planned
6 weeks of the study 12 women and 2 men, ages
18-46 years. At the same time of day on the
same day of the week, on a questionnaire in
French, each reported body weight (weighed that
day on the same scales) and recalled the last two
timings of carrying out each of 26 customs of
eating and/or drinking and 6 customs of moving
about.
5
Correlations of changes in behaviour and
weight Results are expressed here as r x 100,
where r is Pearsons product-moment correlation
coefficient. Where data are seriously skewed,
Spearmans correlation of ranks, rho, should be
used instead. If only one pattern of energy
intake or expenditure changes in frequency in a
given week, correlation with weight change can in
theory be perfect (r 1.0). With such a small
sample, the observed r values are highly
unreliable. When N is adequate for good
estimates of effect size, the 95 confidence
limits of r should be given.
6
Correlations of changes in behaviour and
weight Positive values indicate that the activity
is fattening. Negative values indicate an
effective slimming habit. Test of the causal
hypothesis Frequency change correlates with
later Weight change. This is a rather
severe test, discounting weight lost in 1st week
or two (though 1st week is mainly
water). Cross-lagged control Weight change
does not correlate in the same direction
with later Frequency change. Correct the
r-value measuring Frequency on Weight by
subtracting r-value for Weight on
Frequency. Effect of belief on
behaviour Weight change correlated in the
opposite direction with later Frequency
change indicates action on a correct belief
as to what can produce that change in weight.
7
Evidence for eating that slims in Francophone
Canada correlation of amount of one change in
first week of that change with amount of the
other change in subsequent two weeks r
x 100 r x 100 rF,W - rW,F
rW,F/rF,W Activity (described Freq 0-1
Wt 0-1 Cross-lag Strength in French
) Wt 1-3 Freq 1-3 corrected
of belief smaller meal a -63 -50
-13 - vegetables in meal -85
54 (-85) 0.6 cheese, dairy cream b
-93 38 (-93) 0.4 food with added sugar
-25 58 (-25) 2.3 sucreries en
surplus -30 45 (-30) 1.5 a
smaller than normal. b des matières grasses
animales sic sous forme de ...
8
Customs that slim in Montreal (described in
French) Means of correlation values (r x 100)
available from up to four lags
behaviour-Frequency change over one week (F0-1)
or two weeks (F0-2) and Weight change over the
next week (W1-2,W2-3) or two (W1-3,W2-4).
Mean Mean Mean Mean Pattern
of behaviour Freq Wt on
cross-lag strength on Wt
Freq corrected of belief smaller meal
-62 -43 -60 0.9 vegetables in
meal -28 -9 -14
3.0 cheese, dairy cream -70 17
-48 0.2 food with added sugar -6
45 -6 4.3 sucreries en surplus
-10 16 1 - take exercise
until tired -34 -28 -15 2.4
(mean) difference from two of the four (or one
of the three) lags available
9
Customs that fatten in Montreal (described in
French) Frequency change over one week (F0-1) or
two weeks (F0-2) and Weight change over the next
week (W1-2,W2-3) or two (W1-3,W2-4) - mean
values Mean Mean Mean
Mean Pattern of behaviour
Freq Wt on cross-lag strength
on Wt Freq corrected
of belief use fat in cooking/prepn 59 20
57 - drink alcohol 53 -48
(40) 1.1 choose fibre-rich foods
55 11 31 - eat fruit or salad
73 7 65 - aerobics/un club
sportiv 75 18 62 -
10
Other ways to test the causal hypothesis e.g.,
correlation (unlagged) of a persisting change in
frequency of an activity with the amount of
weight lost Less food than normal in a meal
Increase in frequency with greatest decrease in
weight
Each persons biggest change in frequency during
a period of consistently up or down change
Decrease in frequency (briefly) with (almost) no
change in body weight
Change in weight (kg t1 t2)) over whole period
of change
11
Does maintenance of changed frequency of a
practice depend on how frequent the custom was or
is? Less food than normal in a meal
Number of weeks
Number of weeks
down up
Starting frequency (times per week)
Frequency change in first week of change
Change in frequency increase of eating smaller
meals is maintained for longer if frequency
starts higher.
No sign of effect of size of increase in
frequency. (Sharp decreases were not well
sustained.)
12

• Implications
• The psychobiological long haul
• study of the mechanisms of appetite and its
  sating,
• and of vigour and its fatigue
  
  
  .
  
  physically energetic
• - by itself, such work cant show how to control
  weight
• yet the mechanisms of choice and its
  satiety/fatigue
• must be measured, in order to work out how to
  support these psychosocially
  identified maintainable least fattening
  customs,
• e.g. through supply of foods, transport,
  leisure, etc.

13
Implications Meanwhile Personally tailor
locally valid evidence to inform a succession of
changes each 1-3 weeks, - especially of
eating habits when a more sedentary
lifestyle begins. Track changes in behaviour
components - the only way to rescue obesity
RCTs from unusability and
invalidity. END