Skip to main content
. 2012 Sep 28;2(5):e001384. doi: 10.1136/bmjopen-2012-001384

Table 3.

Patient and physician level predictors of alert generation

Number of visits=8931 Descriptive statistics
Multivariate logistic regression analysis*
No alert (n=3863)
Alert (n=5068)
OR 95% CI p Value
Patient
 Demographic characteristics n % n %
  Female 2730 70.7 3547 70 0.85 0.75 to 0.98 0.02
  Male 1133 29.3 1521 30 Reference Reference Reference
Mean SD Mean SD
Age (per year) 75.7 6.5 75.9 6.6 0.99 0.98 to 1 0.17
Risk factors for fall-related injuries n % n %
Fall-related injury 302 7.8 580 11.4 1.44 1.03 to 2.01 0.04
Gait and balance problems 938 24.3 1447 28.6 1.15 0.99 to 1.33 0.06
Lower-extremity weakness 317 8.2 435 8.6 0.95 0.75 to 1.21 0.68
Cognitive impairment 193 5 208 4.1 0.91 0.64 to 1.29 0.59
Mean SD Mean SD
Number of ambulatory care visits (year prior to visit) (OR per 5 visit increase) 11.2 7.7 12.9 10 1.08 1.05 to 1.12 <0.01
Number of active medications (at visit) 8.3 3.8 10.3 4.4 1.12 1.10 to 1.14 <0.01
Physician
 Demographic characteristics n % n %
  Female 1411 36.5 1462 28.8 0.61 0.42 to 0.91 0.01
  Male 2452 63.5 3606 71.2 Reference Reference Reference
Mean SD Mean SD
  Practice experience (OR per 5 year increase) 26.2 5.4 26.5 5.7 0.89 0.76 to 1.03 0.12
 Practice characteristics
  Daily practice volume (OR per 5 patient increase) 23.1 6.7 21.7 7 0.89 0.79 to 0.99 0.03
  Percent of patients 65 years and older (OR per 10% increase) 25.7 10.8 25.5 11.4 1 0.89 to 1.11 0.96
 Experience and skills related to electronic prescribing
  Electronic prescription speed (minutes per 3 e-RXs) (per minute increase) 5.3 1.6 5.6 1.6 1.01 0.93 to 1.10 0.78
  Electronic prescription rate (e-RXs per 100 visits) (OR per 10 e-RX increase) 25.7 13 27.6 12.3 1.06 0.96 to 1.17 0.28
 Alert level setting
  View severe alerts only 3143 41 4527 59 1.99 1.04 to 3.81 0.04
  View serious and severe alerts only, or view all alerts 885 44.9 1085 55.1 Reference Reference Reference

*The dataset comprised 8931 visits clustered within 3413 patients clustered within 61 physicians. Alternating logistic regression was used to account for clustering (visits within patients and patients within providers).