Results
Study population consists of 195 women with a history of at least one delivery. PM women constituted 8.2% (n=16), MP women constituted 37.4% (n=73), GMP women constituted 23.6% (n=46) and GGMP women constituted 30.8% (n=60) of the study population. The mean age was 50.6±16.3 and mean parity was 6.5±4.2. The characteristics of the study population were given in Table I.
The E velocity (p=0.017), A velocity (p=0.000), lateral e’ velocity (p=0.000), lateral s’ (p=0.027), septal e’ (p=0.000), septal s’ (p=0.000) and EF (p=0.000) values were significantly different among all parity groups. The results were shown in Table II. Binary comparison of the study groups evaluating the echocardiographic parameters can be seen in Table III.
Diastolic dysfunction classification was done according to the echocardiographic parameters. For the PM group, 87.5% (n=14) had normal diastolic function, 6.25% (n=1) had grade 1 diastolic dysfunction and 6.25% (n=1) had grade 2 diastolic dysfunction. For the MP women; 71.2% (n=52) had normal diastolic function, 12.4% (n=9) had grade 1 diastolic dysfunction and 16.4% (n=12) had grade 2 diastolic dysfunction. For the GMP women; 56.5% (n=26) had normal diastolic function, 10.9% (n=5) had grade 1 diastolic dysfunction and 32.6% (n=15) had grade 2 diastolic dysfunction. For the GGMP women; 28.6% (n=17) had normal diastolic function, 33.2% (n=20) had grade 1 diastolic dysfunction and 38.2% (n=23) had grade 2 diastolic dysfunction (Table IV). There were no women with grade 3 diastolic dysfunction among the study population.
Spearman correlation analysis showed that diastolic dysfunction has positive correlations with parity, age, hypertension, and diabetes mellitus (Table V).
Table VI and Table VII report the findings of the binary and multinomial logistic regressions. Explanatory variables in both models are age, number of parity, hypertension and diabetes mellitus. The difference among the models stems from how the dependent variable is handled. In the binary logistic regression, dependent variables are grouped into two categories: diastolic dysfunction existence or the patient has normal diastolic function. On the other hand, multinomial logistic regression in this study separates the patients into three groups: patients without diastolic dysfunction, patients with grade 1 and grade 2 diastolic dysfunction. Both models show that only number of parity and age are statistically significant.
ROC analysis showed that the best cut-off value of the parity number for predicting left ventricular diastolic dysfunction was 6.5, with 66.3% sensitivity and 66.7% specificity (Figure 1).