Table V.

Factors related to family physician awareness of the Rapid Palliative Radiotherapy Program ^a

Factor	Unaware (%) (n=141)	Aware (%) (n=31)	Independent predictors [or (95% ci)]
Previously sought advice b from radiation oncologist?
Yes	52	81	4.60 (1.78–11.96)
No	48	19
Provides palliative care c for their patients.
Yes	77	94	4.26 (0.96–18.80)
No	23	6
Previous referral of patients for palliative radiotherapy?
Yes	50	81	4.12 (1.59–10.62)
No	50	19

	Logit (aware ofrprp) = β0 + β1 + β2
Variable	Regression coefficient (β)	or (95% ci)	Wald p(z)
0. Intercept	−2.982
1. Advice	1.142	3.13 (1.15–8.53)	0.026
2. Palliative care	1.228	3.42 (1.32–8.86)	0.012

^aMultivariate logistic regression model estimating the probability of a family physician being “aware” of the Rapid Palliative Radiotherapy Program. A significant association (Wald p(z) < 0.25 by chi-square) was necessary for a factor to be eligible for entry into logistic model building. Final model:

^bTest for model validity: classification = Hosmer and Lemeshow “goodness of fit” (χ² = 2.924, df =6, p = 0.82).

^cTest for model validity: discrimination = area under the roc curve c = 0.69.

or = odds ratio (odds of the outcome occurring for every unit increase in an individual independent variable, controlling for the other variables in the model); ci = confidence interval; intercept = a mathematical constant (no clinical interpretation); roc = receiver operator curve (test for ability of the model to discriminate between variables).