Table 2.

Association of Predictors with Patients’ Experience of Financial Stress^a

Predictor	Valid nd	Single-Predictor Modelsb		pe	Valid nd	Multi-Predictor Modelc		pe
		β	95% CI			β	95% CI
Female	442/175	0.519	0.191, 0.848	0.002	405/161	0.478	0.110, 0.846	0.011
Racial/ethnic minority	436/173	0.306	−0.044, 0.656	0.086		0.168	−0.233, 0.568	0.412
Age in years, baseline	442/175	−0.015	−0.028, −0.001	0.029		0.006	−0.014, 0.026	0.546
Marital status, baseline	442/175			0.059				0.459
Currently married		0.000				0.000
Previously married		0.260	−0.122, 0.641			0.269	−0.154, 0.693
Never Married		0.488	0.079, 0.897			0.118	−0.350, 0.586
Insurance, baselinef	442/175			0.091				0.065
Medicare		0.000				0.000
Medicaid		0.480	−0.031, 0.991			0.404	−0.192, 0.999
Commercial		0.179	−0.207, 0.565			0.443	−0.057, 0.943
None		0.664	0.041, 1.288			0.848	0.182, 1.515
Cancer as chronic comorbidity	442/175	−0.563	−1.027, −0.098	0.018		−0.478	−1.122, 0.165	0.145
Financially comfortable, baselineg	441/174	−0.309	−0.458, −0.161	<0.001		−0.235	−0.408, −0.061	0.008
Any children under age 10 living at home, baseline	442/175	0.606	0.100, 1.113	0.019		0.707	0.106, 1.309	0.021
Decreased employment from baseline to follow-uph	414/164	0.285	−0.008, 0.577	0.056		0.227	−0.121, 0.575	0.200
Social support, follow-upi	440/175	−0.204	−0.337, −0.071	0.003		−0.130	−0.296, 0.035	0.123
Emotional health, follow-upj	441/175	−0.175	−0.297, −0.053	0.005		−0.021	−0.172, 0.130	0.785
Quality of life, follow-upk	441/175	−0.183	−0.331, −0.034	0.016		−0.092	−0.270, 0.086	0.309

^aAssociations are based on clustered probit regression models (three post-discharge assessment points clustered under patient), estimated with weighted mean- and variance adjusted least squares (WLSMV). The outcome was coded 1 if the patient reported financial stress as one of the top three stressors at the point of the assessment; and 0 if financial stress was not one of the three top stressors or if the patient reported experiencing no stress of any kind.

^bExcept where noted, financial stress was regressed only on the variable (or set of dummy indicators) indicated in the row heading.

^cFinancial stress was regressed simultaneously on all predictors, with adjustment for site.

^dThe valid n is expressed as the number of assessments with valid responses / number of patients providing some valid data at one or more assessment points.

^eP-values for unordered categorical predictors were based on Wald’s test of parameter constraints.

^fThe “single-predictor” model was adjusted for site.

^gHow the patient felt at the end of an average month, when paying bills: 1=short on money and needed more for bills and basic needs, 2=barely enough for bills and basic needs, 3=enough for just a few extra things, 4=completely comfortable.

^hDecreased employment: 0=change to employment expected to provide higher pay (e.g. from part-time at baseline to full-time at follow-up, or from not employed for pay at baseline to employed for pay at follow-up); 1=no change (in the same category – FT/PT/no pay – at both time points); 2=change to employment expected to provide lower pay.

ⁱFrequency of having someone to confide in or talk to about problems: 1=never, 2=rarely, 3=sometimes, 4=usually, 5=always

^jFrequency in past 7 days of being bothered by emotional problems such as feeling anxious, depressed, or irritable: 1=always … 3 = sometimes … 5=never

^kSelf-assessed quality-of-life rating: 1=poor, 3=good, 5=excellent