Table 7. Confounder adjusted association between costs due to TB care (diagnosis and treatment) and various predictor variables using linear regression models in the study of Bao'an district, Shenzhen City, China, 2013 (n = 514).

Predictors in the model	Beta coefficient	95% CI	P value
Whether in hospital due to TB diagnosis
Yes	-0.26	-0.38,-0.15	< 0.001
No	Ref	Ref	Ref
Household registration
Native patients	Ref	Ref	Ref
Migrant patients	0.13	0.01,0.26	0.038
Number of times visiting health-care facilities
< = 2	-0.15	-0.26,-0.03	0.011
2~6	-0.1	-0.22,0.02	0.094
> = 7	Ref	Ref	Ref

Linear regression was done after conversion to log scale.

Only significant variables were presented.

Ref-reference.