The sample comprised 201 adults with T2DM (57.2% female and 42.8% male), between 28 and 87 years of age (\(\overline} }\)= 65.12, SD = ± 11.617), with a 13.4 year average suffering from T2DM (SD = ± 8.088, Minimum = 1, Maximum = 35) and an average 158.84 mg/dL (DE = ± 61.913, Minimum = 89, Maximum = 380) fasting plasma glucose. Along with the T2DM diagnosis, 61.7% suffered from high blood pressure, while 24.9% from obesity; 50.7% of the participants had at least two chronic-degenerative diseases including T2DM. Table 1 shows the frequency and percentage of the participants’ sociodemographic and clinical characteristics.

The association between the three main dependent variables was carried out (e.g. TA, QoL and DRQoL), obtaining three significant weak to moderate correlations between QoL and DRQoL (r = -0.538, p< 0.01), TA and QoL (r = 0.222, p< 0.01) and in TA and DRQoL (r = -0.218, p< 0.01).

When analyzing the association between the sociodemographic and clinical variables, three statistically significant correlations were identified: 1) age and years of T2DM diagnosis (r = .440, p< 0.01), 2) fasting plasma glucose and age (r = -0.211, p< 0.05) and 3) years suffering from T2DM and TA (r = 0.188, p< 0.01).

On the other hand, when performing the association between the dimensions of the psychometric scales WHOQOL-BREF-16, TAS-15, and QoLI-27, it was found that the three highest coefficients were obtained in the Environmental dimension of the WHOQOL-16 and the Social Impact dimension of the QoLI-27 (r = -.547, p< 0.01), Sexual Function of the QoLI-17 and Physical Health of the WHOQOL-16 (r = -.547, p< 0.01) and in the Energy-Mobility dimension of the QoLI-27 and the Environmental dimension of the WHOQOL-16 (r = -.546, p< 0.01) as can be observed in Table 2.

Marital status and occupation were the sociodemographic variables in which significant differences were found. There were significant TA differences (Z = 11.081, p = 0.026) according to the participants’ marital status, with a large size effect (r = 0.78); when conducting post-hoc analysis it was observed that single people (Me = 53) perform more TA behaviors than people living with a partner (Me = 49). As far as occupation is concerned, significant QoL differences were found (Z = 12.054, p = 0.034), with a large size effect (r = 0.85), the post hoc analysis shows that homemakers (Mdn = 44) have a better QoL than pensioners (Mdn = 40), with a large size effect (r = 1.04). Occupation also showed significant differences in DRQoL (Z = 12.445, p = 0.029), with a large size effect (r = 0.88), nevertheless, when performing a post hoc analysis, the significance was adjusted with the Bonferroni correction was not significant.

The state of T2DM (controlled or uncontrolled) showed significant differences in QoL (Z = 2.423, p = 0.015), with a small size effect (r = 0.17); patients with an uncontrolled T2DM perceived a higher QoL (Mdn = 43), as opposed to those that have a controlled T2DM (Mdn = 40). In the DRQoL the state of diabetes also showed significant differences (Z = -4.678, p = 0.000), with a moderate size effect (r = 0.33); those who have a controlled T2DM perceived a lower DRQoL (Mdn = 150), in contrast to people with an uncontrolled T2DM (Mdn = 94.5).

In Table 3 all the comparisons between the state of T2DM and the three psychometric scales and their dimensions can be observed. In the total score and the five dimensions of the QoLI-27 it was found that people with a controlled T2DM have a lower DRQoL (p< 0.05), with a moderate size effect. Whereas, in dimension three the WHOQOL-BREF-16 and of the TAS-15 significant differences were found, with a small to moderate size effect.

The tolerance coefficients and the VIF indicated an absence of collinearity between the model variables, ranging between .708 and 1.412. When analyzing the multivariate assumptions, we found that the distance (leverage, Mahalanobis, and Cook's D) and influence (DfBeta and DfAdjustment) statistics did not identify the presence of outliers, therefore, the data had multivariate normality. Homoscedasticity was found and the scatter plot showed that, although the cases were not clustered near the line of best fit, they did show a linear trend in all variables. In addition, the independence assumption was met by obtaining a d=1.812.

After multivariate assumptions were tested, all sociodemographic (gender, age, education, and marital status), clinical variables (years suffering from T2DM, number of comorbidities, and state of glucose) and the total scores obtained in the psychometric scales (QoLI-27 and TAS-15) were entered into the linear regression model using the standard method. In this regard, education was a positive effect on QoL β = 0.163 (CI 95%: 0.429─3.415, p = 0.012), whereas the score obtained on the QoLI-27 scale had a negative effect β = -0.546 (CI 95%: -0.127─-0.080, p = 0.001). This model yielded a correlation of .584 and an explained variance of 31%. The analysis of variance showed that the model was significant F (9, 191) = 10.989, p = 001, making it generalizable to the population. (Table 4). The graphical representation of the model is shown in Fig. 1.

Model of variables that have an effect on quality of life in patients with T2DM. Note: The factor that most explains the quality of life in the patients in this sample is diabetes-specific quality of life followed by level of education; 31% of the quality of life experienced by a patient is explained by the variables presented in this model