Reinterpretation of the mechanism of type 2 diabetes mellitus based on a psychosomatic medicine framework: a real-world study | BMC psychiatry

Reinterpretation of the mechanism of type 2 diabetes mellitus based on a psychosomatic medicine framework: a real-world study |  BMC psychiatry

Subjects were divided into five groups based on comorbidities: 77 patients in the pre-DM group, 162 in the DM group, 134 in the BD group, 54 in the BD + pre-DM group, and 14 in the BD + group. DM . We refer to the DM and pre-DM groups as the ‘abnormal glucose metabolism group’ (239 patients) and the BD + pre-DM and BD + DM groups as the ‘comorbidities group’ (68 patients). The comorbidity rate of abnormal glucose metabolism in TB patients was 33.7% in this study.

Comparisons of clinical data between the five groups

clinical data

There were significant differences in sex, age at onset of DM, preDM and BD, marital status, educational level, family history, hypertensive comorbidity rate and stable income among the five groups (corrected for p<0.01, Supplementary Table 1).

NEO-FFI scores

Scores for all five factors of the NEO-FFI were also significantly different between the five groups (p< 0.001, Supplementary Table 2). The conscientiousness factor score in the DM group was significantly higher than the norm, while it was significantly lower in the pre-DM group.

Agreeableness factor scores decreased from the BD group (46.28 ± 4.141) to the abnormal glucose metabolism group (44.78 ± 7.853) and then to the comorbidities group (41.30 ± 4.945) (p<0.001). The awareness factor score in the comorbidities group (38.81 ± 6.873) was significantly lower than that of the BD (42.64 ± 5.885) and abnormal glucose metabolism (43.80 ± 10.363) groups (corrected) . p= 0.01 for the first, 0.00 for the second).

neuroendocrine axes

There were significant differences in the levels of the seven hormones detected among the five groups (TSH: p= 0.019, TT3: p= 0.002, and p< 0.001 for the others; Supplementary Table 3). In addition to TSH (p= 0.521), the levels of the other six hormones were significantly different between the control and case groups (p< 0.001), where the levels of FT3 and FT4 were reduced, while those of ACTH and COR increased in the case groups. The incidences of abnormal HPT or HPA axis and biax among the five groups were also significantly different (p<0.001) and all were taller than those in the control group (Supplementary Fig. 1). The incidences of abnormal HPT and HPA axis in the BD and DM groups were significantly lower than those in the pre-DM group (corrected p<0.001).

Logistic regression and cluster analysis for the five groups

We constructed a multiple logistic regression where the five groups were taken as dependent variables, the classification variables with significant differences between the five groups as independent variables, and the numerical variables as covariates. the p-likelihood ratio test values ​​for 11 indicators, including the scores of the five factors of the NEO-FFI and the levels of TT3, TT4, FT3, FT4, ACTH and COR, were < 0,01, sin colinealidad mediante la prueba de multicolinealidad ( valor de tolerancia > 0.1, variance inflation factor (VIF) < 10).

Cluster analysis, including the above 11 indicators, showed that the BD, BD + DM and DM groups were classified as a single group (referred to as the “disease group”), the BD + pre-DM group was classified as a single group (referred to as “disease group”). mixed-period group’), and the pre-DM group as a group (referred to as the ‘pre-disease group’) (Fig. 1A).

Figure 1
Figure 1

The cluster tree and the regression model. A The cluster tree of five groups. B. Classification of the regression model

Comparisons of the three groups after pooling

clinical data

The comorbidity of hypertension among the three groups was significantly different (p< 0.001), being significantly higher in the disease group than in the pre-disease and mixed period groups (corrected p= 0.04 and 0.000, respectively). Furthermore, patients in the disease group showed a significantly higher rate of positive family history of DM and BD (p< 0.001), while married marital status was found to be significantly higher in the pre-illness group than in the mixed and ill groups (corrected for p= 0.000 and 0.01, respectively) (Table 1).

Table 1 Comparisons of demographic data between the three groups after pooling

NEO-FFI scores

The scores of the five factors of the NEO-FFI were significantly different between the three groups (p< 0.01, Table 2), with the awareness factor scores in the illness group higher than the norm, while those in the pre-illness group were lower than the norm. In addition, neurotic factor scores in the pre-illness and mixed-period groups were significantly higher than those in the (corrected) illness group. p<0.001).

Table 2 Comparisons of NEO-FFI Five-Factor Scores Between the Three Groups After Pooling

neuroendocrine axes

There were significant differences in the levels of the seven hormones detected among the three groups (TSH, p= 0.028; FT3, p= 0.002; ACTH, p= 0.042; Y p< 0.001 for the others; Table 3). Compared to the control group, FT3 and FT4 levels in all three groups decreased, while ACTH and COR levels increased.

Table 3 Comparisons of hormone levels and abnormal rates among the three groups after pooling

Table 3 shows that the incidences of abnormal PTH and HPA axis and biax among the three groups were also significantly different (p<0.001). Furthermore, all incidences of abnormal neuroendocrine axes in the case groups were higher than in the control group.

From the pre-disease group to the mixed-period group and then to the disease group, the incidence of abnormal HPA axis gradually decreased, while the incidences of abnormal HPT axis and biax increased slightly at first and then decreased significantly (Fig. 2). .

Figure 2
Figure 2

Comparisons of Neuroendocrine Axis Abnormal Rates Among the Three Groups After Pooling

Logistic regression for the three groups after pooling

In multiple logistic regression, where the three groups were taken as dependent variables, the classification variables with significant differences between the three groups were taken as independent variables, and the numerical variables as covariates, the significance level of fit of the regression model was 0.000 and the maximum pseudo-RThe square value of the model was 0.833, which suggests a good degree of fit. Eleven indicators, including NEO-FFI neuroticism, agreeableness, and conscientiousness factor scores; TT3, TT4, ACTH and COR levels; and the anomaly of the TT4, FT4, COR and HPA axes, were finally introduced in the regression, with p-likelihood ratio test values ​​<0.05 and no multicollinearity.

Taking the ‘illness group’ as a control, the regression model of the ‘pre-illness group’ and the ‘mixed period group’ were, respectively, as follows:

  • G1 = LOG (P(pre-disease) / P(disease)]= 6.039 + 0.08*neuroticism score – 0.344*agreeableness score – 0.078 × Conscientiousness score + 0.444*TT3 level – 0.007*TT4 level + 0 .03*ACTH level + 0.006*COR level + 2.985*TT4 abnormality (0: normal; 1: abnormal) – 3.224*FT4 abnormality + 1.136*COR abnormality + 0.346*HPA axis abnormality.

  • G2 = REGISTER[P(mixed period) / P(disease)] = -0.191 + 0.084 × neuroticism score – 0.253*sympathy score – 0.173*conscientiousness score + 2.951*TT3 level – 0.08*TT4 level + 0.07*ACTH level + 0.004* COR level + 3.892*TT4 abnormality – 4.405*FT4 anomaly – 20.312 *COR anomaly + 22.368*HPA axis anomaly.

The predictions for each stage of the above models are shown in Fig. 1B. Prediction rates for pre-disease, mixed, and disease stages were 81.6%, 83.3%, and 97.1%, respectively. The overall prediction rate of the model was 92.7%.

Leave a Comment