Osteoporosis Screening Intervention Effectiveness According to Four Types of Analyses
Effects | Conventional Analyses | Hierarchical Bayesian Analyses | ||||||
---|---|---|---|---|---|---|---|---|
Logistic | Hierarchical | Without Collateral Data | With Collateral Data | |||||
OR | P* | OR | P* | OR | P† | OR | P† | |
Regular menstruation | 0.44 | .027 | 0.44 | .028 | 0.42 | .026 | 0.43 | .030 |
Total medications | 0.78 | .004 | 0.78 | .004 | 0.76 | .002 | 0.77 | .002 |
Osteoporosis history | 2.56 | .007 | 2.56 | .008 | 2.69 | .005 | 2.68 | .006 |
Had BMD during past 14 months | 0.19 | .001 | 0.19 | .001 | 0.17 | <.001 | 0.17 | <.001 |
Average BMD rate within each clinic | 1.06 | .059 | 1.06 | NA | 1.07 | .101 | 1.07 | .052 |
Chart reminder | 2.27 | .163 | 2.27 | NA | 2.58 | .172 | 2.37 | .156 |
Chart reminder + patient education | 4.99 | .032 | 4.99 | NA | 5.94 | .060 | 5.47 | .029 |
* Nominal one-sided P based on the t-distribution.
† Bayesian analog of one-sided P: the probability that the intervention did not increase BMD testing.
BMD, bone mineral density; OR, odds ratio.