Table 3.
Best multiple-term strategies maximising precision. Values are percentages (95% confidence intervals)
| Search strategy in Ovid format | Sensitivity* | Specificity** | Precision† |
|---|---|---|---|
| Top precision performer‡ | |||
| Medline.tw. or systematic review.tw. or meta-analysis.pt. | |||
| Development | 75.2 (67.0 to 82.3) | 99.4 (99.2 to 99.5) | 60.2 (52.4 to 67.7) |
| Validation without CDSR | 74.4 (69.7 to 79.1) | 98.6 (98.5 to 98.7) | 26.3 (23.5 to 29.1) |
| Validation | 71.2 (68.0 to 74.4) | 99.2 (99.1 to 99.3) | 57.1 (53.9 to 60.3) |
|
Combining most precise term with most sensitive terms (seetable 1)§
|
|
||
| Cochrane database of systematic reviews.jn. or search.tw. or meta-analysis.pt. or Medline.tw. or systematic review.tw. | 90.2 (88.1 to 92.3) | 98.4 (98.3 to 98.5) | 46.5 (43.9 to 49.0) |
| Cochrane database of systematic reviews.jn. or search:.tw. | 79.0 (76.1 to 81.9) | 98.8 (98.8 to 98.9) | 51.8 (48.9 to 54.7) |
| Cochrane database of systematic reviews.jn. or Medline.tw. | 74.4 (71.3 to 77.5) | 99.5 (99.4 to 99.5) | 68.8 (65.6 to 72.0) |
CDSR=Cochrane Database of Systematic Reviews.
*
Development dataset (n=133); validation dataset without CDSR (n=332); validation dataset (n=753).
**
Development dataset (n=10 313); validation dataset without CDSR (n=48 258); validation dataset (n=48 275).
†
Numbers vary by row.
‡
Keeping sensitivity ≥75%.
§
Reported for the complete validation dataset. The term for the Cochrane Database of Systematic Reviews (Cochrane database of systematic reviews.jn.) does not apply to the development or to the validation dataset without CDSR since it does not include records from that source.