Table 2.

PAM Level Change as a Predictor of Costs in the Post Intervention Period: A Linear Regression Model. Models are for total cost (not restricted to hospital or ED)

Primary model
	Estimate	Std. error	Multiplicative factor (Bootstrap 95% CI)	Bootstrap p value
Intercept	2.897	0.190
Baseline PAM level	− 0.027	0.014	0.94 (0.89, 0.98)	0.004
Change in PAM level	− 0.038	0.016	0.92 (0.87, 0.98)	0.006
FU.period = months 7 to 12	− 0.175	0.022	0.67 (0.55, 0.70)	< 0.001
Interaction (change in PAM level * FU.period = months 7 to 12)	0.003	0.021	1.01 (0.88, 1.11)	0.79
log10 of charges in 6 months before intervention	0.194	0.017	1.56 (1.47, 1.67)	< 0.001
Male sex	0.017	0.024	1.04 (0.96, 1.13)	0.37
Approximate age	− 0.005	0.002	0.99 (0.98, 1.00)	0.002
Age < 65	− 0.010	0.066	0.98 (0.78, 1.23)	0.88
Age ≥ 90	0.055	0.040	1.13 (1.00, 1.28)	0.064
Approximate income (per $10,000)	0.003	0.006	1.01 (0.99, 1.03)	0.59
log10 of baseline MARA risk score	0.661	0.034	4.59 (4.10, 5.17)	< 0.001

A linear model on a logarithmic scale becomes a multiplicative model on the untransformed values. The multiplicative factor is the factor by which follow-up cost is estimated to be multiplied for either a 1 unit increase in a numeric predictor or for a yes value. Hence, e.g., follow-up cost is estimated to be multiplied by 0.917, i.e., decreased by 8.3% for each increase in PAM level