. Author manuscript; available in PMC: 2013 Aug 1.

Published in final edited form as: Am J Geriatr Pharmacother. 2012 Jun 9;10(4):230–241. doi: 10.1016/j.amjopharm.2012.05.001

Table 4.

Discrete Changes in Predicted Probabilities

	Discrete Change in Predicted Probability of CR PIM use^*	Bootstrapped 95% CI
Medication use
Caregiver (CG)
0 CG PIMs → CG taking 1 PIM	+9.9	0.5, 17.7
Care recipient (CR)
Number of medications currently taking
0-3 medications → 4-8 medications	+21.1	12.8, 29.0
0-3 medications → 9+ medications	+41.5	27.0, 53.3
Other factors
CG age (1 SD increase)	-8.7	-13.9, -2.7
CG relationship to care-recipient
Non-spouse → Spouse caregiver	+25.9	13.5, 34.6
CG race/ethnicity
White non-Hispanic → Hispanic	+21.6	-2.4, 50.3^†
CG years living in US (1 SD increase)	+11.0	-0.1, 25.5^†
CR sex
Female → Male	-15.6	-26.1, -4.6

Discrete change is the change in predicted probability (multiplied by 100) going from 0 to 1 for binary independent variables with all other covariates set to their mean. For continuous variables, discrete change represents the change in predicted probability of care-recipient PIM associated with a 1 standard deviation change at the mean.

^†

While the bootstrapped 95% CI of the discrete change in predicted probability includes the value of zero, the logistic regression model indicated statistical significance (see Table 3).

PIM, Potentially Inappropriate Medication; CG, Caregiver; CR, Care-recipient; CI, Confidence Interval ; SD, Standard Deviation.