Skip to main content
. 2022 Jul 7;31(11):3283–3292. doi: 10.1007/s11136-022-03178-z

Table 4.

Regression estimates predicting change in EQ-5D-3L conditional on changes in BMI, including stratification by weight loss or weight gain

Model specification 4: linear relationship Model specification 5: asymmetric relationship Model specification 6: multi-morbidity score
Coefficient estimate Standard error Coefficient estimate Standard error Coefficient Estimate Standard error
Change in BMI − 0.0111 0.0019*** − 0.0146 0.0041*** − 0.0173 0.0052***
Change in BMI during weight loss 0.0059 0.0057 0.0089 0.0072
Baseline EQ-5D − 0.0878 0.0217*** − 0.0890 0.0218*** − 0.0840 0.0236***
12 week intervention 0.0052 0.0105 0.0049 0.0105 − 0.0059 0.0111
52 week intervention − 0.0080 0.0086 − 0.0083 0.0086 − 0.010 0.0097
New comorbidity − 0.0112 0.0138
Constant 0.0574 0.0196*** 0.0633 0.0205*** 0.0620 0.0222**
Observations 2616 2616 2093

This table reports the ordinary least squares regression output for three model specifications to estimate the association between change in BMI and change in EQ-5D-3L. The table reports analyses without imputation of missing values. Model specification 4 includes change BMI and baseline BMI as variables. Model specification 5 includes an interaction term for change in BMI during periods of weight loss. Model specification 6 include the effects of comorbidities. Statistical significance at three thresholds are indicated by asterisks *** < 0.001 ** < 0.01 * < 0.05