Statistical analysis of Socio behavioral study to determine the effect of Teslar watch in healthy volunteers with non restorative sleep
Comments
The analysis was conducted in two phases. First, statistical analyses were conducted
to see if differences exist between the baseline and termination phases of the study’s
four outcomes: (1) Sleep Phase, (2) Sleep Efficiency, (3) Epworth Scale (sleepiness)
and (4) SF-36 physical and mental functioning. Second, standardized effect sizes and
confidence intervals were calculated to provide a measure of treatment effect. Effect
size (ES) is a name given to a family of indices that measure the magnitude of a
treatment effect. Unlike significance tests, these indices are independent of sample
size. ES measures are the common currency of all clinical studies that summarize the
findings from a specific area of research.
Generally speaking the treatment appears to have a positive effect on all four outcomes.
Statistical significance was uncovered for the measures (1) Stage-of-Sleep, (2) Epworth
Sleepiness Scale, and (3) SF-36 Mental Health Subscale (p < .05). Borderline
statistical significance was found for Sleep Efficiency (p <.10), and no difference was
detected for SF-36 Physical Functioning subscale. Nevertheless, given the small
sample size (n=10) and the fact that there is no control group, it is important to remain
cautious of any interpretation.
Stage of Sleep
Statistical Analysis
A comparison was made for change in “Quality” sleep, defined as the percentage of
sleep in Stage 3 and REM, between the baseline and termination measurements. The
modified paired t-test (p < .01) and the Wilcoxon Sign-Rank (p < .01) tests were used
and both demonstrated statistical significance. A 13.5 percent (confidence interval of
7.6% <> 19.4%) improvement in “Quality” sleep was noted between the baseline and
termination phases of the study. All subjects reported a significant improvement in
“Quality” sleep. Using Cohen’s D the standardized effect size estimate is 0.86 which is
a large effect. Table 1 provides descriptive statistics for group changes in sleep stage.
Table 2 provides information about subject-specific changes in stage of sleep while
Graph 1 plots the changes in sleep stage by each subject.
Table 1. Descriptive
Statistics for Group Changes--Stages of Sleep
|
Time |
Statistic |
Stage 1 |
Stage 2 |
Stage 3 |
REM |
Stage III and REM |
|
Baseline |
Mean |
12.62% |
55.83% |
19.62% |
8.68% |
28.30% |
|
|
SD |
7.39% |
9.91% |
8.68% |
9.18% |
17.86% |
|
Termination |
Mean |
8.89% |
46.04% |
26.81% |
18.24% |
45.05% |
|
|
SD |
7.24% |
10.00% |
13.06% |
8.30% |
21.36% |
|
Difference |
Mean |
-3.73% |
-9.79% |
7.19% |
9.56% |
16.75% |
|
|
SD |
-0.15% |
0.09% |
4.38% |
-0.88% |
3.50% |
Sleep Efficiency
Statistical Analysis
Two analysis were conducted to look for differences between the baseline and
termination measures for sleep efficiency. Overall sleep efficiency appears to improve
by an average of 3.6 percent (confidence interval of -0.1%<>8.1%). A plot of the
observed changes in sleep efficiency by subject shows us that eight of-the-ten persons
improved. Nevertheless, both the modified paired t-test (p < .10) and the Wilcoxon
Sign-Rank (p < .12) tests show no statistically significant improvement in sleep
efficiency at the five percent level. If you increase the probability of a false positive
(Type 1 error) to ten percent then there is evidence of statistical significance. A
calculation of standardized effect (measure of treatment effect) using Cohen’s D is
0.33, which is defined as a medium effect. Table 1 provides descriptive statistics for the
change in sleep efficiency while Graph 1 shows this change with reference to each
person and overall.
Epworth Sleepiness Scale
Statistical Analysis
Epworth Sleepiness Scale scores range between between zero and 24. Scale Scores
from one to six indicate that patients are getting enough sleep (Zone 1). Scale scores
between seven and eight are average (Zone 2). Any score from nine and up indicates a
potential sleep disorder (Zone 3). To analyze this data the modified paired t-test (p < .
01) and the Wilcoxon Sign-Rank (p < .01) tests were used, and both results
demonstrated statistical significance. Nine of the ten patients reported an
improvement in sleepiness. Overall the intervention reduced sleepiness by an average
of 3.70 points (confidence interval of 1.50<>5.90), thus taking the subjects from the
“Seek Medical Assistance” zone to the “Rested” zone. Using Cohen’s D the
standardized effect size (ES) estimate is 0.97 which is a very large effect. An ES of
0.97 indicates that the mean at the termination point is at the 82nd percentile of the
baseline point. Table 1 provides descriptive statistics for the change in sleepiness
scores while Graph 1 shows this change with reference to the major cut-points.