The primary HRPWBI was developed with 138 items and then, its psychometric properties were evaluated.
Psychometric Evaluation
Face and content validity evaluation
In face validity evaluation, no item was omitted and six items were revised. In content validity evaluation, 44 items were omitted because their CVR values were less than 0.49 and two items were omitted because their I-CVI values were less than 0.70. Experts also noted that fourteen items were overlapping and hence, they were either omitted or combined. Accordingly, the number of items reached 78 and the S-CVI of the 78-item HRPWBI was 0.91. The Cohen’s kappa coefficients of the items were 0.76–1, denoting acceptable content validity (35).
Item analysis
In the item analysis of the 78-item HRPWBI, six items had inter-item correlation coefficients more than 0.7, denoting their similarity or conceptual overlap. These six items were either omitted or combined. Moreover, the item-scale correlation coefficients of fourteen items were less than 0.2. Finally, seventeen items were omitted in this step and the 61-item HRPWBI was evaluated for its construct validity.
Construct validity evaluation
Participants’ characteristics
Participants were 376 women with HRP with an age mean of 30.48±6.7 years in the range of 16–47. Thirty participants (8.1%) were in their first trimester, 109 participants (29.4%) were in their second trimester, and 237 participants (62.5%) were in their third trimester. Moreover, 103 participants (27.4%) were in their first pregnancy, 171 participants (45.5%) were in their second or third pregnancy, and 102 participants (27.1%) were in their fourth pregnancy or more. The most common causes of HRP were diabetes mellitus (n = 71; 18.88%), cardiovascular disease (n = 50; 13.3%), and hypertension (n = 48; 12.9%).
Normality testing
The distribution of the scores of 57 items (93.44%) was normal, while the mean scores of items 14, 44, and 59 were not normal based on their skewness and the mean scores of items 14, 44, 58, and 59 were not normal based on their kurtosis. The non-normal distribution of these items was controlled through assessing floor and ceiling effects (36).
Missing values
Missing values in each item were less than 1% and were replaced with the mean score of the item.
Correlation
Items 1, 3, 4, and 15 had no correlation with any other item at a coefficient of more than 0.3 and hence, were omitted(38). None of the items had a strong correlation with other items at a coefficient of more than 0.7.
Factorability
The Keyser-Meyer-Olkin statistic was 0.806 and the Bartlett’s test was statistically significant, confirming sampling adequacy and appropriate factor analysis model.
Communalities of items
The table of communalities was assessed after performing principal component analysis and loading the items on their relevant factors. Items with the lowest communalities (i.e., outlier variables) were omitted. After omitting each item with the lowest communality, factor analysis was re-performed and finally, six items (i.e., items 7, 16, 17, 26, 27, and 35) were omitted and items with communalities more than 0.4 were accepted and kept (38).
Primary factor extraction
Nine items had eigenvalues equal to 1 or more. Scree plot also revealed seven factors for HRPWBI. Each of these seven factors explained more than 5% of the variance and the cumulative variance was 53.777 (Table 1).
Factors were extracted through the principal component analysis. Items with high correlation with each other were grouped into a factor and the minimum acceptable factor loading was considered to be 0.4. Items which were not loaded on any factor were omitted and principal component analysis was re-performed. Accordingly, seventeen items (i.e., items 2, 6, 14, 22, 28, 29, 32, 33, 39, 41, 43, 44, 45, 56, 60, and 61) were omitted because they were not loaded on any factor.
Rotation of the extracted factors
Varimax rotation provided the best factor loading with the lowest cross-loading and the best interpretation. Item 25 (“I am concerned with the long-term effects of HRP on the child”) was loaded on both the first and the fifth factors with factor loading values more than 0.4 and factor loading difference of less than 0.2, denoting cross-loading. Therefore, this item was omitted. Finally, the 33 remaining items were loaded on seven factors. Each factor explained more than 5% of the variance and all factors explained 53.771% of the total variance. Extracted factors had acceptable comprehensibility and interpretability and adequate number of items (three or more) and their factor loading values were high (more than 0.4) (Table 1).
Labeling of the factors
Factors were labeled based on their items, particularly items with the highest factor loading values (Table 1). The seven extracted factors of HRPWBI were subjective well-being, marital well-being, self-efficacy and independence, social well-being, perceived well-being about healthcare services, health anxiety, and spiritual well-being.
Reliability evaluation
The Cronbach’s alpha of HRPWBI was 0.837 and the theta coefficients of HRPWBI and all its factors were more than 0.7 (Table 2), confirming acceptable internal consistency. The ICC and SEM of HRPWBI were 0.974 and 2.92 (Table 2), which confirm the acceptable relative and absolute stability of the index, respectively.
Table 2
Internal consistency and relative and absolute stability of HRPWBI
No.
|
Factor
|
Alpha
|
Theta
|
ICC
|
95%CI for ICC
|
SEM
|
P value
|
1
|
Subjective well-being
|
0.866
|
0.814
|
0.965
|
0.935–0.981
|
1.773
|
< 0.001
|
2
|
Marital well-being
|
0.817
|
0.875
|
0.95
|
0.907-0.973
|
0.776
|
< 0.001
|
3
|
Self-efficacy and independence
|
0.865
|
0.825
|
0.971
|
0.946-0.984
|
0.878
|
< 0.001
|
4
|
Social well-being
|
0.702
|
0.758
|
0.919
|
0.85-0.956
|
1.19
|
< 0.001
|
5
|
Perceived well-being about healthcare services
|
0.644
|
0.788
|
0.909
|
0.833-0.951
|
0.8
|
< 0.001
|
6
|
Health anxiety
|
0.675
|
0.799
|
0.948
|
0.903-0.972
|
0.77
|
< 0.001
|
7
|
Spiritual well-being
|
0.873
|
0.799
|
0.974
|
0.952-0.986
|
0.42
|
< 0.001
|
8
|
Total
|
0.837
|
|
0.974
|
0.951-0.986
|
2.92
|
< 0.001
|
Responsiveness and interpretability
The MDC, MDC%, and SEM of HRPWBI were respectively 8.09, 6.63, and 2.92 and MIC was less than MDC (Table 3), confirming the acceptable responsiveness of the index.
Table 3
SEM, MDC, MDC%, MIC, and Kappa values to assess the test-retest absolute stability and agreement of HRPWBI
No.
|
Factors
|
Score range
|
Mean±SD
|
SEM
|
MDC
|
MIC
|
Kappa
|
MDC%
|
Result
|
1
|
Subjective well-being
|
10–50
|
31.44±9.85
|
1.773
|
4.9
|
1.82
|
+
|
15.58
|
Acceptable
|
2
|
Marital well-being
|
25-5
5-25
|
22.18±4.54
|
0.776
|
2.14
|
0.77
|
+
|
10.36
|
Acceptable
|
3
|
Self-efficacy and independence
|
5-25
|
22.75±4.36
|
0.878
|
2.43
|
0.86
|
+
|
10.68
|
Acceptable
|
4
|
Social well-being
|
4-20
|
15.58±4.22
|
1.19
|
3.29
|
1.14
|
+
|
21.11
|
Acceptable
|
5
|
Perceived well-being about healthcare services
|
3-15
|
12.38±2.54
|
0.8
|
2.21
|
1.14
|
+
|
17.85
|
Acceptable
|
6
|
Health anxiety
|
3-15
|
7.63±3.53
|
0.77
|
2.13
|
0.79
|
+
|
27.91
|
Acceptable
|
7
|
Spiritual well-being
|
3-15
|
13.8±1.45
|
0.42
|
1.21
|
0.41
|
+
|
8.76
|
Excellent
|
8
|
Total
|
33-165
|
121.85±16.62
|
2.92
|
8.09
|
2.92
|
+
|
6.63
|
Excellent
|
Roof and ceiling effects
None of the participants obtained the lowest possible score of HRP (i.e., 33) and only one participant (0.3%) obtained its highest possible score (i.e., 165). Therefore, the roof effect was zero and the ceiling effect was 0.3%. Floor and ceiling effects less than 15% are acceptable.
Easy applicability
The easy applicability of HRPWBI was assessed through measuring the amount of time needed for its answering which was 6–18 minutes with a mean of 11.1±1.96. Moreover, missing values of each item were less than 1%. Therefore, the easy applicability of HRPWBI is confirmed.
Scoring
Items are scored 1–5 and hence, the possible total score of the 33-item HRPWBI is 33–165.