3.1. General characteristics of participants
Ages and body mass index (BMI) of the participants ranged from 18 to 44 years (mean= 28.1, SD= 5.21), and 18 to 40 (mean= 28.3, SD= 4.36), respectively. Besides, %38.9 of the participants were in the third trimester, %40.6 had one pregnancy in total, %40.9 were graduates of high school and %63.1 had income exceeding monthly expenses (Table 1).
3.2. Content validity
Using Davis method, expert opinion was obtained from 11 experts in the fields of statistics, linguistics, obstetrics and gynecology nursing, psychiatric nursing, and measurement and evaluation. I-CVI scores ranged from 0.91 to 1.00 whereas S-CVI was 0.98.
3.3. Item reduction
Items with an item-total correlation coefficient ≤ 0.30 were re-evaluated. Item-total correlation coefficients of the 34 items ranged from 0.30 to 0.52.
3.4. Exploratory factor analysis
KMO coefficient (0.84) and Bartlett’s test of sphericity (χ2 = 2961.62, df = 325, p < 0.001) indicated that data were fit for factor analysis. Scree plot and eigenvalue over 1 criterion were used to determine the number of factors. Scree plot illustrated that eigenvalues of six factors were higher than 1 and the eigenvalues decreased significantly after the sixth factor. The findings indicated that the PREG-QOL was a six-factor model.
8 items that were cross-loaded or had a factor loading lower than 0.30 and eigenvalue lower than 0.30 were removed with EFA to obtain a strong factor structure. EFA, which was recomputed after the removal of these items, revealed a six-factor PREG-QOL with 26 items. Factor loadings of the PREG-QOL ranged from 0.41 to 0.90. The first factor was composed of 10 items (10,5,6,7,9,11,7,8,22,29,1), the second factor had 4 items (18,16,15,17), the third factor 4 items (24,26,32,23), the fourth 3 items (3,2,20), the fifth 2 items (27,28), and the sixth factor had 3 items (13,14,34), respectively. These six factors explained %56.17 of total variance. Variances explained by each factor were %24.24, %10.38, %6.77, %5.56, %4.98 and %4.23 respectively (Table 2).
3.5. Confirmatory factor analysis
Six factors and 26 items suggested by the EFA were tested using CFA. Modification indexes were evaluated to improve model fit. Binary correlations, which were analyzed for binary items, did not exceed 0.90. Tukey’s test for non-additivity was performed to evaluate additivity of the scale. A non-additive p value below 0.50 (p < .50) indicates that the scale is non-additive. Since the PREG-QOL had a non-additive structure (F=56.74; p < .001), we performed first level CFA, which revealed that factors 3 and 5 constructed similar subjects on themes. Due to this reason, these two factors were combined under the factor of physical domain and the CFA of a five-factor model was performed (Figure 2). Given the content of the items, this combination was theoretically and logically appropriate. In the next step, factors were labelled according to the content of the items. These factors were perception of general satisfaction, emotional domain, physical domain, health support systems and social domain, respectively. A CFA model was developed based on the five-factor structure. Fit indices of the final model were GFI= 0.822, CFI= 0.872, CMIN= 455.692, DF= 286, CMIN/DF= 1.593, RMSEA= 0.063, and SRMR=0.072 (Table 3).
3.6. Parallel test method
Using parallel test method, we calculated the correlation between PREG-QOL and SF-36 scale. The analysis showed that most of the correlation coefficients were statistically significant (p<.05), positive but very weak or weak (Table 4).
3.7. Reliability analysis
Cronbach’s α coefficients of the six factors were 0.883, 0.654, 0.727, 0.705, 0.827, and 0.622, respectively. Cronbach’s α of the PREG-QOL was 0.885. Test-retest method was used to evaluate the stability of the PREG-QOL over time. The PREG-QOL was administered twice on 40 pregnant women with a 15-day interval. The ICC was computed to compare the scores obtained from test and retest. The ICC scores for the perception of general satisfaction, emotional domain, physical domain, health support systems and social domain were 0.98 (%95 GA = 0.96-0.99, p< .001), 0.97 (%95 GA = 0.94-0.98, p< .001), 0.98 (%95 GA = 0.96-0.98, p< .001), 0.98 (%95 GA = 0.96-0.98, p< .001), and 0.97 (%95 GA = 0.95-0.98, p< .001), respectively (Table 5).
3.8. Final instrument
The PREG-QOL was composed of 26 items in 5 factors, namely perception of general satisfaction (10 items), emotional domain (4 items), physical domain (6 items), health support systems (3 items) and social domain (3 items). The instrument was developed to evaluate quality of life during pregnancy and items were self-scored on a 5-point Likert scale. Scoring system was based on the calculation of mean scores for each factor, which ranged from 1 to 5. Items 11,12,13,14,15,16,19,20,21,22,23,25 and 26 were reverse-scored. Total score was not calculated. Higher scores obtained from the factors indicated a higher quality of life during pregnancy.