Establishment of models to predict factors influencing periodontitis in patients with type 2 diabetes mellitus

Table 1 General information

Factor	Total number (n = 300)	Periodontitis group (n = 224)	Non-periodontitis group (n = 76)	t/χ²/Z	P value
Sex, n (%)				4.050	0.044
Male	164 (54.67)	130 (58.04)	34 (44.74)
Female	136 (45.33)	94 (41.96)	42 (55.26)
Age (yr), (mean ± SD)	59.18 ± 11.40	60.91 ± 9.49	54.11 ± 14.69	-4.646	< 0.001
Education level, n (%)				-3.626	< 0.001
Below high school education	118 (39.33)	98 (43.75)	20 (26.31)
High school education	98 (32.67)	76 (33.93)	22 (28.95)
Above high school education	84 (28.00)	50 (22.32)	34 (44.74)
Monthly income, n (%)				-1.890	0.059
< 2000 yuan	65 (21.67)	54 (24.11)	11 (14.47)
2000-5000 yuan	133 (44.33)	99 (44.20)	34 (44.74)
> 5000 yuan	102 (34.00)	71 (31.70)	31 (40.79)

Table 2 Lifestyle factors

Factor	Total (n = 300)	Periodontitis group (n = 224)	Non-periodontitis group (n = 76)	χ²	P value
Smoking, n (%)				3.500	0.061
No	238 (79.33)	172 (76.79)	66 (86.84)
Yes	62 (20.67)	52 (23.21)	10 (13.16)
Alcohol consumption, n (%)				0.460	0.497
No	249 (83.00)	184 (82.14)	65 (85.53)
Yes	51 (17.00)	40 (17.86)	11 (14.47)
Regular exercise, n (%)
No	142 (47.33)	109 (48.66)	33 (43.42)	0.625	0.429
Yes	158 (52.67)	115 (51.34)	43 (56.58)
Brushing frequency, n (%)				29.248	< 0.001
≥ 2 times/d	137 (45.67)	82 (36.61)	55 (72.37)
< 2 times/d	163 (54.33)	142 (63.39)	21 (27.63)

Table 3 Physical health status

Factor	Total number (n = 300)	Periodontitis group (n = 224)	Non-periodontitis group (n = 76)	t/Z	P value
BMI (kg/m2), median (P25, P75)	24.96 (22.36, 26.81)	24.86 (22.13, 26.65)	25.61 (23.26, 27.23)	-1.223	0.221
TG (mmol/L)	1.51 ± 0.28	1.53 ± 0.28	1.44 ± 0.26	-2.678	0.008
< 1.7	226 (75.33)	163 (72.77)	63 (82.89)
≥ 1.7	74 (26.67)	61 (27.23)	13 (17.11)
TC (mmol/L)	4.72 ± 0.77	4.86 ± 0.78	4.33 ± 0.59	-5.361	< 0.001
< 5.18	212 (70.67)	144 (64.29)	68 (89.47)
≥ 5.18	88 (29.33)	80 (35.71)	8 (10.53)
HDL-C (mmol/L)	1.09 ± 0.19	1.08 ± 0.20	1.12 ± 0.15	1.777	0.077
< 1.04, n (%)	114 (38.00)	93 (41.52)	21 (27.63)
≥ 1.04, n (%)	186 (62.00)	131 (58.48)	55 (72.37)
LDL-C (mmol/L)	2.42 ± 0.29	2.44 ± 0.29	2.39 ± 0.30	-1.183	0.238
< 3.37, n (%)	300 (100.00)	224 (100.00)	76 (100.00)
≥ 3.37, n (%)	0 (0.00)	0 (0.00)	0 (0.00)
HbA1c (mmol/L)	7.86 ± 1.17	8.10 ± 1.16	7.14 ± 0.85	-6.596	< 0.001
< 6.5%, n (%)	37 (12.33)	17 (7.59)	20 (26.32)
6.5%-7.5%, n (%)	83 (27.66)	52 (23.21)	31 (40.79)
7.5%-8.5%, n (%)	90 (30.00)	71 (31.70)	19 (25.00)
≥ 8.5%, n (%)	90 (30.00)	84 (37.50)	6 (7.89)

BMI: Body mass index; TG: Triglyceride; TC: Total cholesterol; HDL-C: High-density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol; HbA1c: Glycosylated hemoglobin.

Table 4 Assignment and description of related factors affecting the occurrence of periodontitis in patients with type 2 diabetes mellitus

Factor	Assignment of value
Sex	Female = 0
	Male = 1
Age	Continuous variables
Brushing frequency	≥ 2 times/d = 0
	< 2 times/d = 1
Education level	Below high school education = 0
	High school education = 1
	Above high school education = 2
HbA1c	Continuous variables
TC	Continuous variables
TG	Continuous variables

HbA1c: Glycosylated hemoglobin; TC: Total cholesterol; TG: Triglyceride.

Table 5 Multivariate logistic regression analysis of occurrence of periodontitis in patients with type 2 diabetes mellitus

Factor	β	SE	Wald χ²	P value	OR (95%CI)
Age	0.046	0.015	9.813	0.002	1.047 (1.017-1.078)
Sex	0.622	0.346	3.230	0.072	1.863 (0.945-3.674)
Brushing frequency	1.459	0.353	17.073	< 0.001	4.303 (2.154-8.599)
Education level	-0.639	0.212	9.065	0.003	0.528 (0.348-0.800)
HbA1c	0.934	0.185	25.392	< 0.001	2.545 (1.770-3.661)
TC	1.055	0.260	16.451	< 0.001	2.872 (1.725-4.781)
TG	1.196	0.600	3.967	0.046	3.306 (1.019-10.723)

HbA1c: Glycosylated hemoglobin; TC: Total cholesterol; TG: Triglyceride; OR: Odds ratio; 95%CI: 95% confidence interval.

Table 6 Efficacy of the two models in predicting the occurrence of periodontitis in patients with type 2 diabetes mellitus in the training dataset

Model	Sensitivity	Specificity	Accuracy	Recall	Precision	AUC (95%CI)
Random forest	1.000	1.000	1.000	1.000	1.000	1.000 (1.000-1.000)
Logistic regression	0.569	0.919	0.830	0.569	0.707	0.851 (0.791-0.910)

AUC: Area under curve; CI: Confidence interval.

Table 7 Efficacy of the two models in predicting the occurrence of periodontitis in patients with type 2 diabetes mellitus in the validation dataset

Model	Sensitivity	Specificity	Accuracy	Recall	Precision	AUC (95%CI)
Random forest	0.520	0.960	0.850	0.520	0.813	0.946 (0.905-0.986)
Logistic regression	0.640	0.907	0.840	0.640	0.696	0.915 (0.857-0.973)

AUC: Area under curve; CI: Confidence interval.