Descriptive statistics
In our study, most subjects were men (61.1%, n=899) and the mean BMI was 26.80±3.64 kg/m2. Based on the cut of values of BMI for Chinese adults [33], there were 7 (0.5%) underweight participants, 284 (19.3%) normal weight participants, 683 (46.4%) overweight participants, and 499 (33.9%) obese participants. The other demographic characteristics of the study participants are shown in Table 1.
Table 1 : Sample characteristics (n=1472)
Characteristics
|
Total (N=1472)
N (%)/M±SD
|
t/F
|
P
|
Sex
|
|
1470
|
0.802
|
Man
|
899(61.1)
|
|
|
Woman
|
573(38.9)
|
|
|
Age
|
|
3.821
|
0.022
|
≤39
|
221(15.0)
|
|
|
40-59
|
695(47.2)
|
|
|
≥60
|
556(37.8)
|
|
|
Education level
|
|
20.692
|
0.000
|
Primary school and below
|
463(31.5)
|
|
|
Secondary school
|
467(31.7)
|
|
|
High school
|
281(19.1)
|
|
|
University or above
|
261(17.7)
|
|
|
Occupation
|
|
28.870
|
0.000
|
Self-employ
|
278(18.9)
|
|
|
Worker
|
623(42.3)
|
|
|
Farmer or other
|
571(38.8)
|
|
|
Exercise
|
|
2.766
|
0.063
|
no
|
349(23.7)
|
|
|
Yes,Once in a while
|
803(54.6)
|
|
|
Yes, less than once a week
|
320(21.7)
|
|
|
Income
|
|
17.515
|
0.000
|
<1000
|
184(12.5)
|
|
|
1000-1999
|
470(31.9)
|
|
|
2000-2999
|
208(14.1)
|
|
|
≥3000
|
610(41.4)
|
|
|
BMI
|
26.80±3.46
|
-
|
-
|
Reliability analysis
The CHLSalt-22 consists of 22 items. The KR-20 and split-half reliability of the CHLSalt-22 were 0.815 and 0.713, respectively. After one month, the test-retest ICC of the CHLSalt-22 was 0.982, which met the recommended criteria [41,42], representing good stability.
Construct validity analysis
The result of the Harman single-factor test analyzes Common Method Biases(CMB) was 26.228 < 40, and the contribution rate was 67.993%>60% [43]. The statistically significant results of Bartlett’s test of sphericity ( χ2(231)=3815.612, p<.001) and the Kaiser-Meyer-Olkin Measure of Sampling Adequacy > 0.80 (KMO=0.815) indicate that the data meet the conditions for using factor analysis. Therefore, the first principal component analysis (PCA) was performed to determine the likely number of factors. As a result, seven f actors that explained a total of 49.709% of the variance had initial eigenvalues >1 each. The scree plot further confirms the seven-factor structure. After varimax orthogonal rotation, these seven extracted factors explained 10.985, 21.925, 32.716, 42.813, 52.232, 61.396 and 67.993% of the variance, respectively. The scree plot is presented in Fig. 1.
Table 2 presents the factor loading of each item. All the correlation coefficients were more prominent than 0.40 and statistically significant at P<0.01. A CFA was performed on Sample 1 (n=469). The two models of the scale in this study were evaluated, and the results showed the fitting index of the seven-factor model (Table 3). The structural equation model and the standardized regression coefficients of the seven-factor model of the CHLSalt-22 are appeared in Fig. 2.
Table 2: Factor loadings of the CHLSalt-22 (n=469, salient factor loadings are indicated in italics.)
Item
|
Estimate
|
1.Which of the following statements best describes the relationship between salt and sodium?
|
0.699
|
2.What is the daily limit of salt intake (in grams)that is recommended by the World Health Organization for an adult?
|
0.895
|
3. Which type of biscuits would you choose if you wish to minimize salt intake?
|
0.746
|
4.How much the salt content of Lunch meat(100g)?
|
0.784
|
5.How much the salt content of Instant noodles with seasoning powder(100g)?
|
0.560
|
6.How much the salt content of Ketchup or Salad dressing (100g)?
|
0.888
|
7.How much the salt content of Oyster sauce(100g)?
|
0.770
|
8.Do you agree that the high blood pressure can be caused by high salt intake?
|
0.720
|
9.Do you agree that the cardiovascular disease can be caused by high salt intake?
|
0.626
|
10.Do you agree that the diabetes mellitus can be caused by high salt intake?
|
0.719
|
11.Sodium intake can be reduced by replacing salt with plenty of Chicken essence during cooking.
|
0.744
|
12.Most foods available at restaurants(e.g,Chinese restaurants,fast food restaurants)are high in salt.
|
0.738
|
13.Drinking more water can neutralize salt intake from my diet.
|
0.783
|
14.Most low salt foods taste bad.
|
0.433
|
15.I feel too much pressure to eat a healthy diet .
|
0.960
|
16.Limiting the amount of salt intake is important to my health.
|
0.896
|
17.Add salt or sauce or condiments at the table.
|
0.504
|
18.Consume canned foods.
|
0.700
|
19.Consume salted fish, salted vegetables, salted duck eggs.
|
0.761
|
20.Pay attention to whether the food is labeled as “No added salt” or “Low in salt”.
|
0.655
|
21.Read the sodium content stated on the nutrition labels on food packages.
|
0.707
|
22.Purchase foods according to the sodium content stated on the nutrition labels.
|
0.859
|
Table 3: Confirmatory factor analysis of the CHLSalt-22 with different models
Model
|
χ2
|
df
|
χ2/df
|
CFI
|
TLI
|
SRMR
|
RMSEA[90%CI]
|
1
|
354.170
|
181
|
1.957
|
0.956
|
0.944
|
0.041
|
0.045[0.038-0.052]
|
2
|
632.060
|
182
|
3.473
|
0.922
|
0.901
|
0.049
|
0.050[0.045-0.054]
|
χ2, chi-square, df, degrees of freedom, CFI, comparative ft index, TLI, Tucker-Lewis index, SRMR, standardized root mean residual, RMSEA, root mean square error of approximation, 90% CI, 90% confidence interval, M1, 469 samples structure model, M2, 1003 samples model
Convergent validity and discriminant validity
The path coefficient of CFA was to analyze the composite reliability (CR) and average variance extract (AVE). The critical ratios among the seven common factors in the CHLSalt-22 were over 0.7, and all of the average variance extraction values were over 0.45 [39,44]. The convergent validity and discriminant validity of the CHLSalt-22 are shown in table 4. The bold numbers on the diagonal lines of the table are the square root of the extraction amount of the mean variance of the corresponding dimensions, and the non-diagonal numbers are the inter-dimensional correlation coefficients.
Table 4: Convergent validity and discriminant validity of the CHLSalt-22
Factor
|
CR
|
AVE
|
F1
|
F2
|
F3
|
F4
|
F5
|
F6
|
F7
|
F1
|
0.826
|
0.615
|
0.784
|
|
|
|
|
|
|
F2
|
0.842
|
0.577
|
0.516**
|
0.760
|
|
|
|
|
|
F3
|
0.731
|
0.476
|
0.323**
|
0.540**
|
0.690
|
|
|
|
|
F4
|
0.799
|
0.570
|
0.396**
|
0.557**
|
0.790**
|
0.755
|
|
|
|
F5
|
0.828
|
0.637
|
0.420**
|
0.495**
|
0.418**
|
0.450**
|
0.798
|
|
|
F6
|
0.700
|
0.441
|
0.148**
|
0.289**
|
0.192**
|
0.173**
|
0.347**
|
0.664
|
|
F7
|
0.787
|
0.556
|
0.627**
|
0.542**
|
0.338**
|
0.372**
|
0.468**
|
0.276**
|
0.746
|
CR,composite reliability,AVE,average variance extract,**P<0.01,*:P<0.05
Content validity
The content validity of the CHLSalt-22 scale was evaluated by an expert evaluation [45]. The expert group is composed of seven experts, including two dieticians, three nursing experts proficient in both Chinese and English, and two psychiatrists. The content validity analysis result shows that the I-CVI of the CHLSalt-22 scale is 0.857-1, and the S-CVI /Ave is 0.968, which has good content validity.
Prediction validity
The ROC curve of the CHlsalt-22 was drawn in this study with hypertension as reference. According to the Youden’s Index (YI) maximization principle [46,47], 16.5>17 was the best cut off value for the CHLSalt-22 scale. The area under the ROC curve for the scale was 0.774 (p<.01), with a sensitivity of 0.843, specificity of 0.623, and 95%CI (0.741-0.807). The ROC curve for the CHLSalt-22 is shown in Figure 3.
Table 5: Pearson’s correlations between the CHLSalt-22 count and MCRSDH-SUST,BIPQ and BFS
|
1
|
2
|
3
|
4
|
1.CHLSalt-22
|
-
|
|
|
|
4.MCRSDH-SUST
|
0.437**
|
|
|
|
2.BIPQ
|
0.226**
|
0.484**
|
|
|
3.BFS
|
0.348**
|
0.513**
|
0.340**
|
-
|
CHLSalt-22: revised Chinese Health Literacy Scale For Low Salt Consumption, BIPQ:The Brief Illness Perception Questionnaire, BFS:Benefit Finding Scales, MCRSDH-SUST:The Continuous Behavior Change Sub-scale, ** p<0.01, * p<0.05
Table 6: Results of multiple linear regression models of factors influencing the CHLSalt-22 scores of subjects with different characteristics
Variate
|
β
|
SD
|
Β’
|
t
|
P
|
R2
|
DR2
|
F
|
P
|
Model 1
|
|
|
|
|
|
0.074
|
0.070
|
16.799
|
0.001
|
Sex
|
1.307
|
0.452
|
0.081
|
2.889
|
0.004
|
|
|
|
|
Age
|
1.330
|
0.576
|
0.084
|
2.307
|
0.021
|
|
|
|
|
BMI
|
-0.195
|
0.058
|
-0.086
|
-3.349
|
0.001
|
|
|
|
|
Income
|
2.317
|
0.630
|
0.103
|
3.676
|
0.000
|
|
|
|
|
education level
|
2.574
|
0.566
|
0.126
|
4.550
|
0.000
|
|
|
|
|
Occupation
|
1.461
|
0.493
|
0.092
|
2.966
|
0.003
|
|
|
|
|
Model 2
|
|
|
|
|
|
0.212
|
0.211
|
134.649
|
0.001
|
MCRSDH-SUST
|
0.412
|
0.038
|
0.351
|
10.737
|
0.000
|
|
|
|
|
BFS
|
0.116
|
0.022
|
0.168
|
5.145
|
0.000
|
|
|
|
|
BMI, body mass index, BFS:Benefit Finding Scales, MCRSDH-SUST:The Continuous Behavior Change Sub-scale, P, P-Value, Bold values correspond to statistically significant correlations (p < 0.05)
Health literacy related to salt consumption and sociodemographic differences
The CHLSalt-22-diagnosed health literacy related to salt consumption count was 17 in patients with hypertension (SD= 17.58, range= 0-44). According to the ROC of the CHLSalt-22 analysis, the higher the low salt health literacy score, the more likely they are to have high blood pressure. This proves that people with hypertension are aware of the prevention of hypertension and other diseases, and pay special attention to the salt content of diet in their daily lives. Many studies have shown that people with high blood pressure have better health literacy [48]. There were no significant differences in the CHLSalt-22 between men and women or between different exercise frequencies. However, there were statistically significant differences in age, education level, occupation, income, and BMI. Table 1 presents the results. There are differences in the level of health literacy, which is similar to the results of previous studies conducted at home and abroad [49,50]. However, a part of the result of the original scale was different [29]. This result is due to all kinds of cultures, customs, and demographic characteristics. Table 5 presents the factors associated with the CHLSalt-22 score: the CHLSalt-22 score was positively correlated with the BIPQ, BFS, and MCRSDH-SUST. The effects of different sociodemographic groups on the CHLSalt-22 score were assessed using bilinear regression. Sociodemographic groups were considered as categorical predictors (with one category being the reference group), and the CHLSalt-22 score represented the continuous outcome variable. The results of the multivariate regression analysis results are shown in Table 6. Sex, age, BMI, income, education level, occupation, BFS, and MCRSDH-SUST were the influencing factors of the CHLSalt-22. This reflects that health literacy is influenced by individual and environmental social determinants [51]. But, BMI was negatively correlated with the CHLSalt-22 and other characteristics were positively correlated with the CHLSalt-22. This suggests that the higher weight of hypertensive patients, the lower the health literacy of low salt intake, and the less attention to salt intake in the normal diet process. The CHLSalt-22 count score increased by 2.317 points for each one-unit increase in income, by 2.574 points for each one-unit increase in education level, by 1.461 points for each one-unit increase in occupation, and decreased by 0.195 points for each one-unit increase in BMI. This trend is consistent with the results of previous studies [52-55]. The degrees of BFS and MCRSDH-SUST were correlated with the CHLSalt-22 count scores. With every increase of one SD in the BFS, the CHLSalt-22 score increased by 0.168 SD. 0.351 SD increased the score increased by the CHLSalt-22 for each increase of one SD in the MCRSDH-SUST.