Table 4.

Mediation analysis for the effect of symptom burden on the association between polypharmacy and quality of life

	Without covariates	With covariates*
	β [95% CI] (p value) (N = 239)	β [95% CI] (p value) (N = 238)
Direct effect (c’)†	− .035 [− .071, 0.000] (0.051)	− .031 [− .064, 0.003] (0.071)
Indirect effect (ab)†	− .030 [− .052, − .008] (0.007)	− .028 [− .048, − .007] (0.008)
Total effect (c)†	− .065 [− .099, − .031] (< .001)	− .059 [− .094, − .023] (0.001)
Proportion of total effect mediated (ab/c)	0.461	0.473
Ratio of indirect to direct effect (ab/c’)	0.855	0.897
Ratio of total to direct effect (c/c’)	1.855	1.897

Normal-based bootstrapped 95% CI and p value with –vce- option in Stata 15 (no. of bootstrap replications = 200)

*Covariates in the model are gender, age, primary DX, hospice, AKPS (Australia-Modified Karnofsky Performance Status), and CCI (Charlson Comorbidity Index)

†The direct effect (c’) is equal to the association between polypharmacy and quality of life when symptom burden is included in the model. The total effect (c) is equal to the association between polypharmacy and quality of life when symptom burden is not included in the model (c’ + ab). Without mediation, c should equal c’. Because there is mediation, c and c’ are different