Table 2. Misclassification-Adjusted Aspirin, BP, and Smoking Performance Scores and Sensitivities and Specificitiesa.
| Variable | Misclassification-adjusted median performance score, % (95% BCI) | Prior distributions used in bayesian latent class models | ||
|---|---|---|---|---|
| Prior set 1b | Prior set 2c | Prior set 1b | Prior set 2c | |
| Aspirin | ||||
| Performance score | 74.9 (70.5-79.1) | 75.4 (70.8-79.7) | β (13.59, 9.73); normal (0.58, 0.01) | Uniform (0.5, 1.0) |
| MRA | ||||
| Sensitivityd | 99.6 (98.2-100.0) | 99.4 (97.9-100.0) | Uniform (0.5, 1.0) | Uniform (0.5, 1.0) |
| Specificityd | 98.2 (92.4-99.9) | 95.8 (89.1-99.8) | Uniform (0.8, 1.0) | Uniform (0.5, 1.0) |
| EHR | ||||
| Sensitivityd | 96.8 (94.2-98.5) | 98.8 (96.3-99.9) | β (9.57, 5.18); normal (0.65, 0.01) | Uniform (0.5, 1.0) |
| Specificityd | 95.8 (93.9. 97.3) | 98.2 (93.4-99.9) | β (417.49, 20.78); normal (0.95, <0.01) | Uniform (0.5, 1.0) |
| BP control | ||||
| Prevalence | 75.0 (66.6-82.5) | 77.9 (68.1-85.9) | β (14.71, 8.99); normal (0.62, 0.01) | Uniform (0.5, 1.0) |
| MRA | ||||
| Sensitivityd | 98.2 (93.6-99.1) | 98.2 (93.6-99.9) | Uniform (0.5, 1.0) | Uniform (0.5, 1.0) |
| Specificityd | 78.4 (58.3-97.7) | 82.2 (59.2-99.0) | Uniform (0.8, 1.0) | Uniform (0.5, 1.0) |
| EHR | ||||
| Sensitivityd | 96.0 (88.8-99.6) | 94.6 (86.6-99.7) | β (42.06, 2.77); normal (0.94, <0.01) | Uniform (0.5, 1.0) |
| Specificityd | 94.9 (85.8-99.2) | 93.2 (76.6-99.7) | β (43.76, 5.75) | Uniform (0.5, 1.0) |
| Normal (0.88, <0.01) | ||||
| Smoking cessation counseling and intervention | ||||
| Prevalence | 83.0 (74.4-89.8) | 82.3 (69.1-91.2) | β (11.16, 5.03); normal (0.69, 0.01) | Uniform (0.5, 1.0) |
| MRA | ||||
| Sensitivityd | 97.7 (92.3-99.9) | 97.4 (91.6-99.9) | Uniform (0.5, 1.0) | Uniform (0.5, 1.0) |
| Specificityd | 86.0 (80.2-98.7) | 70.7 (51.0-98.2) | Uniform (0.8, 1.0) | Uniform (0.5, 1.0) |
| EHR | ||||
| Sensitivityd | 89.6 (83.4-94.9) | 88.7 (78.8-98.5) | β (42.06, 2.77); normal (0.94, <0.01) | Uniform (0.5, 1.0) |
| Specificityd | 89.2 (79.5-95.6) | 86.8 (61.8-99.3) | β (43.76, 5.75); normal (0.88, <0.01) | Uniform (0.5, 1.0) |
Abbreviations: BCI, bayesian credible interval; BP, blood pressure; EHR, electronic health record; MRA, medical record abstraction.
a
Prior distributions used in bayesian latent class models are provided. Equivalent normal (μ, σ2) distributions are presented for the beta (α, β) distributions.
b
Prior set 1 is based on prior elicitation from experts.
c
Prior set 2 represents diffuse priors.
d
Indicates with respect to the true (unknown) status of whether a patient received recommended care (ie, performance met or not).