Table 3.
Systematic reviews and meta-analyses in different fields of cancer (excluding trials and treatment)
| Search terms | All articles, N (all) | n (SR) | n (MA) | n (SR)/n (MA) | n (all)/n (MA) |
|---|---|---|---|---|---|
| Gene/genome/genetic | 268,597 | 1,999 | 920 | 2.2 | 291 |
| Epigenetic/methylation/mutation | 115,763 | 497 | 159 | 3.1 | 728 |
| Immune/allergy/asthma | 29,046 | 107 | 25 | 4.3 | 1162 |
| Hormone | 53,679 | 148 | 51 | 2.9 | 1,032 |
| Social/socioeconomic | 11,531 | 224 | 50 | 4.5 | 231 |
| Diet/dietary/nutrition/nutritional | 19,549 | 289 | 147 | 2.0 | 133 |
| Physical activity/exercise/obesity | 8,919 | 192 | 74 | 2.6 | 121 |
| Virus/bacteria/infection/infectious | 88,881 | 331 | 109 | 3.0 | 815 |
| Carcinogen | 30,286 | 201 | 88 | 2.3 | 344 |
| Radiation | 30,124 | 229 | 86 | 2.6 | 350 |
| Occupation/occupational | 16,839 | 344 | 177 | 1.9 | 95 |
| Smoking/smoke/tobacco | 20,660 | 413 | 232 | 1.8 | 89 |
| Alcohol | 17,921 | 170 | 83 | 2.0 | 215 |
| Biomarker | 72,709 | 373 | 97 | 3.8 | 750 |
NOTE: On the basis of PubMed searches conducted on August 29, 2012. The search resulted in 1,295,958 items overall, of which 6,763 were tagged by PubMed as systematic reviews and 2,332 as meta-analyses; when limited to studies tagged as human, there were 941,360 items overall, 5,780 systematic reviews, and 2,319 meta-analyses. All searches use as a backbone the search strategy "Cancer NOT (trial* OR treatment)" so as to avoid capturing articles on clinical trials and treatments. Some articles from case-control and cohort studies where treatment or treatment-related questions are discussed would be excluded by this strategy, but this latter group accounts generally for few meta-analysis, for example, a search with "Cancer AND (trial* OR treatment) AND (case-control or cohort) AND meta-analysis [type]" yields an additional 451 items (besides the 2,332 identified with "Cancer NOT (trial* OR treatment) AND meta-analysis [type]"); in a sample of 20 of the 451, only 10 are related to cancer epidemiology. Conversely, in detailed scrutiny of a sample of 127 items, among those identified with the search "Cancer NOT (trial* OR treatment) AND meta-analysis [type]", 114 are indeed meta-analyses of cancer epidemiology topics (positive predictive value 90%). Thus, the provided estimate of 2,332 metaanalysis articles in cancer epidemiology seems to be quite accurate, with <250 falsely identified articles and roughly equal number of falsely nonidentified articles. All searches in the table, with the exception of Gene/Genome/Genetic and Epigenetic/Methylation/Mutation, use the string "NOT (Gene OR Genome OR Genetic)", so as to exclude items that deal primarily with genetics, for example, immune response genes or hormone-related genes.
Abbreviations: N (all), total number of articles retrieved; n (SR), number of articles retrieved with type: systematic review; n (MA), number of articles retrieved with type: meta-analysis.