Of 128 participants, 55 (43%) were men and 73 (57%) were women. The mean age of participants was 40.98 ± 8.75 (24 to 58 years). Moreover, 15.6 percent of participants had a high school degree and the remaining, 84.4%, had an academic degree. Eighteen subjects (14%) reported blood pressure history. Demographic characteristics of participants in terms of study groups are given in Table 1. The comparison of participants’ characteristics revealed no statistically significant difference in the age (Mann-Whitney test, P=0.241), work experience (Mann-Whitney test, P=0.363), gender (Chi-squared test, P=0.858), blood pressure history (Chi-squared test, P=0.611), and their education levels (Chi-squared test, P=0.572) between the cases and controls.
Table I Demographic characteristics of case and control groups
Characteristic
|
Cases
|
Controls
|
P-value
|
Age (years)
|
|
|
|
Mean ± SD
|
41.94 ± 9.04
|
40 ± 8.38
|
0.241†
|
Work experience (years)
|
|
|
|
Mean ± SD
|
15.98 ± 8.99
|
14.41 ± 8.80
|
0.363†
|
Gender
|
|
|
|
Male
|
28 (43.8%)
|
27 (56.3%)
|
0.858⁑
|
Female
|
36 (42.2%)
|
37 (57.8%)
|
BP history
|
|
|
|
Yes
|
10 (15.6%)
|
8 (12.5%)
|
0.611⁑
|
No
|
54 (84.4%)
|
56 (87.5%)
|
Education
|
|
|
|
High School diploma
|
12 (18.8%)
|
8 (12.5%)
|
0.572⁑
|
Undergraduate degree
|
29 (45.3%)
|
29 (45.3%)
|
Postgraduate degree
|
23 (35.9%)
|
27 (42.2%)
|
† Mann-Whitney test was used,
⁑ Chi-squared test was used.
Since we used a researcher-made questionnaire to measure the knowledge, perceived susceptibility, perceived severity, self-efficacy, perceived benefits, perceived barriers, and cues to action of subjects, the reliability of questionnaire has to be reported. Cronbach's alpha, the most common measure of internal consistency, was used to obtain the reliability of the questionnaire in measuring aforementioned components (Table 2). For each component, Cronbach's alpha was calculated twice; i) using the baseline data (n=128) and ii) using the data available at the end of study. It turns out that reliability to acceptable at most dimensions except for perceived benefits/barriers measured at baseline where the Cronbach's alpha was well below 70% threshold.
Table 2 Cronbach's alpha for measuring internal consistency of the dimensions of the questionnaire using the participants’ information at the baseline and at the end of study
Construct (Component)
|
Cronbach's α
|
Baseline (n=128)
|
End of study (n=128)
|
Knowledge
|
0.92
|
0.72
|
Perceived Susceptibility
|
0.58
|
0.61
|
Perceived severity
|
0.72
|
0.68
|
Perceived barriers
|
0.53
|
0.60
|
Perceived benefits
|
0.50
|
0.61
|
Cues to action
|
0.84
|
0.78
|
Self-efficacy
|
0.70
|
0.60
|
Practice
|
0.70
|
0.80
|
For each group, the mean and standard deviation of the questionnaire (components) at the baseline and end of study are presented in Table 3. Since the normality assumption was violated, the Mann-Whitney test was used to compare the differences between groups. The results revealed no statistically significant difference in component levels between two groups at baseline. At the end of study, the differences were all significant except for the ‘cues to action’ component.
Table 3 The comparison of components between case and control group at baseline and end of study using Mann-Whitney test. No difference between participants at baseline, but difference were observed following intervention at all components than the cues to action
Component
|
Baseline
|
|
End of study
|
|
Case
|
Control
|
P-value
|
|
Case
|
Control
|
P-value
|
Knowledge
|
2.45±2.14
|
2.59±2.12
|
0.503
|
|
3.44±1.76
|
2.55±2.22
|
0.011
|
Perceived Susceptibility
|
11.88±2.42
|
11.52±2.38
|
0.451
|
|
13.27±1.65
|
11.56±2.30
|
< 0.001
|
Perceived severity
|
13.42±2.58
|
13.53±2.61
|
0.780
|
|
15.0±1.51
|
13.66±2.41
|
0.001
|
Perceived barriers
|
14.33±2.72
|
14.41±2.79
|
0.806
|
|
16.58±1.86
|
14.25±2.79
|
< 0.001
|
Perceived benefits
|
16.22±2.23
|
16.58±2.37
|
0.316
|
|
18.16±1.34
|
16.33±2.31
|
< 0.001
|
Cues to action
|
1.94±1.22
|
2.0±1.26
|
0.596
|
|
2.03±1.13
|
1.95±1.24
|
0.914
|
Self-efficacy
|
13.12±2.24
|
13.55±2.18
|
0.251
|
|
14.61±1.12
|
13.64±1.92
|
0.005
|
Practice
|
9.25±1.91
|
9.33±1.84
|
0.817
|
|
12.59±0.79
|
9.25±1.78
|
< 0.001
|
Although the Mann-Whiney test confirmed the effectiveness of the educational intervention, it can be misleading at it fails to control for the level of components at the baseline. Thus, ANCOVA was used to test for differences in component means among the groups by adjusting the effect of components at baseline. The assumption of homogeneity of regression slopes as well as the independence of baseline scores and study groups were tested using Mann-Whitney and interaction effects. Results showed that these assumptions only met for the ‘cues to action’ component and for other components the latter assumption was violated. Thus ordinary ANCOVA was used to interpret the results of ‘cues to action’ component and the robust ANCOVA [17] was used for other components. The results of performing ordinary ANCOVA showed that compared to control group, education intervention increased the ‘cues to action’ in case group (F (1,124) =0.59, P = 0.011).
In Robust ANCOVA, the trimmed means (20%) were compared between two groups at some design points (usually five points) where the relationship between pre and post values were the same in both groups. Comparisons between trimmed means of case and control groups, , were made by constructing 95% confidence intervals using bootstrapping method. Confidence intervals were adjusted for inflation type I error in multiple comparisons. Table 4 represents the results of performing robust ANCOVA for knowledge, perceived susceptibility, and perceived severity components. In this Table, at each design point, and denote the sample sizes used to obtain trimmed means at case and control group respectively. The significant results are displayed in bold face under 95% CI column (Table 4). It appears that the educational intervention raised the knowledge score of those participants who had lower initial score (0 or 3), but it didn’t affect the knowledge of participants with high initial scores. It also affect mostly the participants with lower perceived susceptibility (7 to 11) and perceived severity (< 13) at the baseline. In other words, the intervention was quite successful for participants who had low scores at the beginning of the study.
Table 4 The result of performing robust ANCOVA for knowledge, perceived susceptibility, and perceived severity. Significant results are highlighted in bold face intervals
95% CI
|
n2
|
n1
|
Design point
|
Variable
|
(0.87, 2.46)
|
1.67
|
30
|
30
|
0
|
Knowledge
|
(0.20, 2.59)
|
1.19
|
18
|
47
|
3
|
(0.01, 0.77)
|
0.38
|
33
|
33
|
5
|
(4.17, 2.50)
|
3.33
|
13
|
13
|
7
|
Perceived Susceptibility
|
(3.01, 1.07)
|
2.04
|
38
|
40
|
10
|
(2.59, 0.71)
|
1.65
|
42
|
37
|
11
|
(1.38, 0.53)
|
0.42
|
35
|
31
|
14
|
(1.86, 0.03)
|
0.92
|
18
|
18
|
16
|
(3.59, 1.72)
|
2.66
|
17
|
19
|
9
|
Perceived severity
|
(3.76, 1.62)
|
2.69
|
28
|
25
|
11
|
(1.62, 0.06)
|
0.79
|
40
|
39
|
14
|
(1.34, 0.06)
|
0.64
|
33
|
37
|
16
|
(0.97, 0.64)
|
0.17
|
15
|
18
|
18
|
To avoid a lengthy Table, for remaining components (perceived barriers, perceived benefits, self-efficacy, and practice), the results of performing robust ANCOVA were plotted and displayed by Fig 1. In this Figure, the vertical dashed lines represent non-significant comparisons of the trimmed means between two groups at the defined design point. For example, at panel A (Fig. 1) which displays the scatter plot of the perceived barriers scores at baseline and the end of study, the educational intervention increased significantly the perceived barrier scores of participants with low to moderate baseline scores (non-significant results appeared only for high baseline values, the vertical dashed line 19, which compromises 3% of participants). In other words, intervention was successful in increasing the perceived barriers scores of participants with initial score lower than 19 (that is for 97% of subjects).
The educational intervention was also quite effective in promoting the ‘perceived benefits’ (panel B, Fig. 1) and ‘practice’ score (panel D, Fig. 1) of individuals (scores were significantly higher at case group at all design points). For self-efficacy, the results showed that intervention was successful only for those participants who had pre-score lower than 15 (66% of participants, panel C, Fig. 1).