Application of search strategies to three pathologies: numbers of citations retrieved, proportions of potentially pertinent articles and overall NNR values
PubMed query | Meningioma (n=9460) | Pancreatitis (n=24497) | Atrial fibrillation (n=20313) | Overall (n=54247) | ||||||||
Retrieved, n | Pertinent, n (%) | NNR | Retrieved, n | Pertinent, n (%) | NNR | Retrieved, n | Pertinent, n (%) | NNR | Retrieved, n | Pertinent, n (%) | NNR | |
‘First string’ (more specific strategy) | 20 | 16 (80) | 1.2 | 19 | 10 (53) | 1.9 | 20 | 11 (55) | 1.8 | 59 | 37 (63) | 1.6 |
‘First string’+narrow PubMed clinical query for aetiology9 | 5 | 4 (80) | 1.2 | 5 | 1 (20) | 5.0 | 2 | 1 (50) | 2.0 | 12 | 6 (50) | 2.0 |
‘First string’+broad PubMed clinical query for aetiology 9 | 17 | 14 (82) | 1.2 | 14 | 6 (43) | 2.3 | 6 | 2 (33) | 3.0 | 37 | 22 (59) | 1.7 |
‘Second string’ (more sensitive strategy) | 80 | 18 (23) | 4.4 | 195 | 22 (11) | 8.9 | 232 | 22 (9) | 10.5 | 507 | 62 (12) | 8.2 |
‘Second string’ NOT ‘first string’ (incremental contribution of the second string) | 60 | 2 (3) | 30.0 | 176 | 12 (7) | 14.7 | 212 | 11 (5) | 19.3 | 448 | 25 (6) | 17.9 |
String proposed by Schaafsma et al4† | 20 | 13 (65) | 1.5 | 36 | 13 (36) | 2.8 | 18 | 7 (39) | 2.6 | 74 | 33 (45) | 2.2 |
String used by the Cochrane Occupational Health Field3‡ | 51 | 17 (33) | 3.0 | 108 | 18 (17) | 6.0 | 60 | 15 (25) | 4.0 | 220 | 50 (23) | 4.4 |