Table 1

The relationship between dietary thiamine intake and cognitive decline using two-piecewise linear regression*

Inflection point, mg/day	n	Mean (SD)	β (95% CI)	P value	R²
Estimate (95% CI)	n	Mean (SD)	β (95% CI)	P value	R²
Decline rate in global cognitive scores, points/5 years
Model 1
0.67 (0.51 to 0.82)					0.270
<0.67	570	2.89 (11.13)	−7.43 (−18.79 to 3.93)	0.200
≥0.67	2536	2.09 (11.12)	3.10 (1.35 to 4.85)	<0.001
Model 2
0.68 (0.56 to 0.80)					0.312
<0.68	617	2.80 (10.95)	−7.13 (−18.77 to 4.50)	0.230
≥0.68	2489	2.09 (11.16)	4.24 (2.22 to 6.27)	<0.001
Decline rate in composite cognitive scores, SU/5 years
Model 1
0.66 (0.52 to 0.81)					0.256
<0.66	527	0.43 (1.47)	−0.90 (−2.50 to 0.70)	0.271
≥0.66	2579	0.29 (1.43)	0.33 (0.10 to 0.55)	0.005
Model 2
0.68 (0.56 to 0.80)					0.297
<0.68	617	0.40 (1.46)	−1.00 (−2.56 to 0.56)	0.208
≥0.68	2489	0.29 (1.44)	0.49 (0.23 to 0.76)	<0.001

p values less than 0.05 are shown in bold.
*Model 1 was adjusted for age, sex, global score, as well as the intakes of carbohydrate, protein and fat. Model 2 was further adjusted for smoking, alcohol consumption, BMI, SBP, DBP, education level, occupation, region, urban or rural residency, self-reported diabetes, antihypertensive medicine, physical activity, as well as the intakes of fibre, sodium and potassium.
BMI, body mass index; CI, confidence interval; DBP, diastolic blood pressure; SBP, systolic blood pressure; SD, standard deviation; SU, standard units by averaging z scores.