|
Comparing the new strategies from 15 randomly sampled journals (R15) and from 2 top-yielding journals (T2) with the existing search strategies (ES) in the original large journal database |
|||||
| Ovid search strategy* |
Sensitivity (%) (95% CI) |
Specificity (%) (CI) |
Precision (%) (CI) |
Accuracy (%) (CI) |
|
|
|
|||||
| High sensitivity |
ES: clinical trial.mp, pt. OR random:.mp. OR tu.xs. |
99.2 (98.7, 99.8) |
70.4 (69.8, 70.9) |
9.9 (9.3, 10.5) |
71.3 (70.8, 71.8) |
| R15: clinical trial.mp, pt. OR random:.mp. OR between group:.tw. |
98.5 (97.7, 99.3) |
76.3‡ (75.8, 76.8) |
11.9‡ (11.2, 12.7) |
77.0‡ (76.5, 77.4) |
|
| T2: clinical trial.mp, pt. OR exp longitudinal studies. |
96.8† (95.6, 97.9) |
84.1‡ (83.7, 84.6) |
16.6‡ (15.6, 17.6) |
84.5‡ (84.1, 84.9) |
|
|
|
|||||
| High specificity |
ES: randomized controlled trial.mp, pt. |
93.1 (91.5, 94.8) |
97.5 (97.3, 97.6) |
54.4 (52.0, 56.8) |
97.3 (97.1, 97.5) |
| R15: double-blind.mp. OR random: assigned.tw. |
47.1† (43.9, 50.3) |
98.8‡ (98.7, 99.0) |
57.0 (53.5, 60.5) |
97.2 (97.0, 97.4) |
|
| T2: double-blind:.mp. OR random: assigned.tw. |
53.4† (50.2, 56.7) |
98.4‡ (98.3, 98.6) |
52.7 (49.5, 55.8) |
97.0† (96.8, 97.2) |
|
|
|
|||||
| Balanced combination of sensitivity |
ES: randomized controlled trial.pt. OR randomized.mp. OR placebo.mp. |
95.8 (94.5, 97.1) |
95.0 (94.8, 95.3) |
38.5 (36.5, 40.5) |
95.0 (94.8, 95.3) |
| R15: randomized controlled trial.pt. OR random: assigned.tw. OR exp research design. |
95.2 (93.8, 96.5) |
94.6† (94.3, 94.8) |
36.3 (34.4, 38.2) |
94.6† (94.3, 94.8) |
|
| T2: randomized controlled trial.mp, pt. OR random: assigned.tw. OR blind:.mp. |
95.4 (94.0, 96.7) |
96.1‡ (95.9, 96.3) |
44.3‡ (42.2, 46.5) |
96.1‡ (95.8,96.3) |
|
|
*mp = multiple postings, search term appears in title, abstract or subject heading; pt = publication type; : = truncation; tu = therapeutic use subheading; xs = exploded subheading; tw = textword; exp = explosion. † Statistically significant difference (p-value < 0.05) in favor of the existing search strategy. ‡ Statistically significant difference (p-value < 0.05) in favor of the new search strategy. | |||||
Yao et al. BMC Medical Informatics and Decision Making 2008 8:43 doi:10.1186/1472-6947-8-43 |
|||||