Table 2.

Univariate and multivariate association of weight loss, behavioral, and metabolic variables on lower diabetes incidence in participants in the ILS group: DPP, 1996–2001

		Models assessing one characteristic at a time*			Multivariate model†
Variables (unit change)	Model	HR (95% CI)‡	P	R2 (%)§	HR (95% CI)‡	P	R2 (%)§
Baseline variables
Lower weight (10 kg)	1	0.81 (0.75–0.86)	<0.0001	3.46	0.88 (0.81–0.96)	0.003	0.82
Higher leisure physical activity (MAQ) (5 MET hours/week)	2	1.00 (0.96–1.05)	0.89	0.00	1.00 (0.96–1.04)	0.98	0.00
Higher recreational activity (LoPAR) (20 MET hours/week)	3	1.12 (1.01–1.23)	0.03	0.42	1.08 (0.97–1.20)	0.16	0.19
Lower percent of calories from fat (5%)	4	0.96 (0.84–1.10)	0.57	0.03	0.91 (0.79–1.05)	0.19	0.16
Follow-up variables
Weight loss (5 kg)	1	0.49 (0.43–0.57)	<0.0001	7.55	0.42 (0.35–0.50)	<0.0001	8.09
Higher leisure physical activity (MAQ) (5 MET hours/week)	2	0.95 (0.88–1.02)	0.17	0.17	0.97 (0.90–1.05)	0.50	0.04
Higher recreational activity (LoPAR) (5 MET hours/week)	3	0.99 (0.97–1.02)	0.66	0.01	0.99 (0.96–1.02)	0.45	0.05
Lower percent of calories from fat (5%)	4	0.75 (0.63–0.88)	0.0007	1.06	0.93 (0.77–1.12)	0.42	0.06

^*Baseline proportional hazards models (1–4) include baseline and follow-up (time-dependent) variables for each characteristic and are also adjusted for baseline demographic factors (age, sex, race/ethnic group, and weight). Separate models are shown with numbers.

^†Multivariate proportional hazards model includes all baseline and follow-up (time-dependent) variables shown, as well as age, sex, ethnic group, HbA_1c, and baseline fasting measures of insulin, glucose, and IGR.

^‡HRs for specified unit differences: <1 indicates less diabetes at follow-up; >1 indicates more diabetes.

^§Partial R² indicates the proportion of explained variance in predicting the development of diabetes at a specified time point. LoPAR, Low Level Physical Activity Recall questionnaire; MAQ, Modifiable Activity Questionnaire.