PCOS, MetS and SF36 parameter recordings from four BMI groups (18-24, 25-29, 30-34 & >35 kg/m2) were statistically analyzed employing multifactorial analysis and resulted appreciably (P<0.001) different (Table 1). Mean ovarian volume (ml), cycle duration (days) and Ferriman- Gallways scores were significantly (P<0.05) increased in morbidly obese PCOS women compared to women with other BMI levels, normal weight, overweight and morbidly obese, while mean fasting blood glucose (mg/dL), LDL (mg/dL), waist to hip ratio, both systolic and diastolic blood pressures (mmHg) were also examined significantly (P<0.05) high among morbidly obese PCOS women compared to other PCOS women. Both obese and morbidly obese PCOS women were found more affected from their metabolic dysfunctions when their physical domain of HRQoL was surveyed and compared to normal weight and overweight PCOS women. Whereas, these both type of PCOS women had non-significant (P>0.05) lower scores of all variables from psychological domain than the other PCOS women.
Model fit indices of four constructs including PCOS, MetS, SF-36 physical and psychological domains were estimated for PCOS women from normal weight, overweight, obese and morbidly obese groups are presented in table 2 and 3. Four of five tabulated goodness of fit indices were confirming the appropriateness of the constructs. While standardized regression coefficient were also estimated for all loading factors separately in each preliminary construct and presented in table 3. In latent variable PCOS, increased right ovary volume (ml) remarkably (P<0.05) impacted the latent variable of all BMI groups. Moreover the drop in the number of reproductive cycles per year was reciprocally influencing the latent variable from all groups, normal weight (β=-0.208, P<0.05), overweight (β=-0.275, P<0.05), obese (β=-0.045, P>0.05) and morbidly obese (β=-0.253, P>0.05) PCOS women. Interestingly, however, increase in F-G score loading factor was negatively influencing the latent variable from obese and morbidly obese women while overweight women’s
F-G score was very parallel to the latent variable. In MetS latent variable, elevated diastolic pressures (mmHg) and BMI (kg/m2) among overweight (β=0.561, P<0.05 and β=0.895, P>0.05 respectively) and morbidly obese (β=0.894, P<0.05 and β=0.475, P<0.05 respectively) women were appreciably parallel with latent variable. While, increase in pain loading factor positively contributed to SF-36 physical domains latent variables of overweight (β=0.819, P<0.05) and morbidly obese women (β=0.768, P>0.05). Whereas fall in the score of role limitation due to emotional problem and emotional functions loading factors among morbidly obese women hit reciprocally (β=-0.300, P>0.05 and β=-0.037, P>0.05 respectively) to the SF-36 psychological latent variable. The convergent validities of CFA including average variance extracted (AVE) and composite reliability (CR) for all constructs were though not closed to the set criteria (Table4) however discriminant validities were strongly ascertained that each construct was built with suitable variables since the correlation coefficient values estimated for each pair of correlating latent variables were not greater than the difference between two estimated AVE (Supplementary table 1). While model fit statistics of correlation constructs established for each group were also stood appropriate on six indices which were according to required criteria.
Table 5 shows direct, though non-significant, effects between two latent variable constructs but the rationality of the hypothesis was ascertained. Such as PCOS women with normal weights showed an alternate association between MetS & PCOS, MetS&SF-36 physical and psychological domains, while a negative impact of PCOS on SF-36 physical domain was appreciable (P=0.05). Among morbidly obese PCOS women a highly appreciable (P=0.042) inverse relation was examined between latent variable PCOS and SF-36 physical domain, while overweight and morbidly obese PCOS women groups showed a remarkable (P=0.003, P=0.001 respectively) positive influence of MetS on PCOS latent variables. PCOS women from all BMI groups (18-24, 25-29, 30-34 & >35 kg/m2) had a positive association between SF-36 physical and psychological domains, whereas goodness of fit indices and chi-square results of all groups were ascertaining the association between theory and model (Table 5).
However, indirect effects of latent constructs were also depicting the rationality but significance was compromised (supplementary table 2). Among morbidly obese women, MetS was inversely impacted on physical and psychological domains of HRQOL through intermediary drift of PCOS. Similarly PCOS and MetS have negative influence on psychological wellness through intermediary impact of physical domain.
Four intricate model diagrams using all latent variables were developed for all PCOS women with their BMI groups (Figure 1 A-D). Exogenous variables, defining each latent variable, were estimated with their exact effects by factor loading strength and error variance while a model modification opportunity was also applied where it was inevitable. All models, representing the four groups; normal weight, overweight, obese and morbidly obese, with their model fit statistics, were very according to hypothesis, MetS and PCOS had positive association. Whereas, MetS and PCOS had negative impact on physical and psychological scores. Moreover both of these HRQoL domains had a parallel association. All these latent variables had a progressive increment with increase in BMI of obese and morbidly obese PCOS women.