A TEST-RETEST RELIABILITY OF 0.99 FOR MEASURING GRADUAL COGNITIVE CHANGE: THE DANBURY MS BLUEBERRY STUDY.
C. Pappas, Danbury MSSM Support Group, Danbury, CT 06812, K.E.
Gerstmann, NY, NY 10014, J.A. Joseph, USDA Human Nutrition Research
Center on Aging at Tufts University, Boston, MA 02111, R. Martin*, MMT
Corp., Sherman, CT 06784, A. Pruchnicki, Mount Sinai Medical Center,
NY, NY 10029, B. Sachs, HR Herbs, Sherman, CT 06784, R. Schnoll,
Brooklyn College, Brooklyn, NY 11210, and A. Wetherell, DERA Chemical
and Biological Defence Sector, UK. Contact email:
DrRMartin@Netscape.net.
C. Pappas, Danbury MSSM Support Group, Danbury, CT 06812, K.E. Gerstmann,
NY, NY 10014, J.A. Joseph, USDA Human Nutrition Research Center on Aging at
Tufts University, Boston, MA 02111, R. Martin*, MMT Corp., Sherman, CT
06784, A. Pruchnicki, Mount Sinai Medical Center, NY, NY 10029, B. Sachs, HR
Herbs, Sherman, CT 06784, R. Schnoll, Brooklyn College, Brooklyn, NY 11210,
and A. Wetherell, DERA Chemical and Biological Defence Sector, UK. Contact
email: DrRMartin@Netscape.net.
High measurement precision is critical for studies of gradual change,
since statistical significance can be lost if measurement error far
exceeds the change under investigation. Precision is especially
critical for aging studies, where change over one year may be just 1%
and treatment effects 0.5% even if half the anticipated decline is
prevented.
Test-retest reliabilities for 77 performance tests listed in recent
reviews averaged only 0.63, indicating that the average participant may
score 20% higher or lower when re-taking a typical performance test.
This high measurement error is enough to prevent detection at 95%
confidence of performance changes as large as 5% even if 100 people are
included in a study, and is therefore large enough to jeopardize
measurements during relatively large clinical trials spanning 5 years.
In this report, we describe methods for response time measurement that
produced an overall test-retest reliability of 0.991 during a study to
determine whether blueberries could improve cognitive performance and
overall health in people with multiple sclerosis. Our basic approach
was to gather large amounts of data during twice-weekly at-home
sessions throughout the ten-week study, and thereby increase our
measurement precision approximately ten-fold. Data were collected
during home visits and via Internet transmission. Reliability for
results received over the Internet was 0.994.
The study was a five step, open-label, crossover study of the effects
of flash-frozen blueberries on choice reaction time and overall health,
using the same varieties of blueberries reported by Joseph et al. (29th
Ann. Mtg. Amer. Aging Assoc. June, 2000, abstr. 17) to partly reverse
performance decline in aging lab animals. After a practice period
(step 1) and baseline measurement period (step 2), participants
consumed 1 cup of uncooked berries daily for 3 weeks (step 3) and then
two cups daily for 3 weeks (step 4) before discontinuing blueberries
and obtaining follow-up data (step 5). Eighteen subjects started the
trial, 13 finished, and 11 provided useful performance data. No
significant side effects were reported. Self-reported overall health,
mood, energy, sleep quality, sharpness, stress level and social life
all improved slightly when berries were consumed, for 38%, 54%, 46%,
31%, 39%, 39% and 15% of the 13 participants, respectively (two-tailed
t-test p = 0.03). MS symptoms improved in three but worsened in two
particiants. Four participants reported unusually vivid dreams.
Choice reaction times scores improved by 7.9% when 1 cup of berries was
consumed daily, and another 6.2% when 2 cups were consumed.
Improvement was greatest in those who reported more energy (p = 0.06),
however this improvement may be due largely to higher expectations and
delayed practice effects. The improvement indicates only that a
larger, rigorously-controlled follow-up trial appears justified.
We thank the Michigan Blueberry Growers Association and Wild Blueberry
Association of North America for providing each participant with 50
pounds of blueberries, Stew Leonard's "Farm Fresh Foods" in Danbury, CT
for distributing this mountain of berries from their freezer, and Lee
Martin for computer services. This study was implemented by the
Danbury MSSM Support Group, Catherine Pappas, coordinator.
Key words:
retest, reliability, blueberry, reaction, time
Problems or questions regarding this site should be directed to
webmaster@americanaging.org