For years we’ve touted the benefits of searching MEDLINE on the Dialog® platform. Aside from our intuitive interface and thesauri, Dialog lets you search across multiple databases at the same time. You can also ensure you stay up to date with the latest developments by creating alerts; and can export your results in multiple formats. You know what we can do, but the question is: how does Dialog compare to PubMed for accessing MEDLINE?
In 2011, PubMed’s efficiency as a search platform was compared to the performance of Ovid-MEDLINE – take a look here1. The systematic review search – composed of key terms for clinical trials, rheumatoid arthritis, and methotrexate – found that PubMed retrieved more articles and was (slightly) more sensitive for relevant articles than Ovid-MEDLINE. Round one to PubMed – but, this got us thinking – how would Dialog MEDLINE perform under similar circumstances?
We decided to find out, and this is how we did it.
Results
First, we created a table of the individual components of the search strategy to help us compare Dialog to PubMed.
| Term | Dialog | PubMed |
| Arthritis, Rheumatoid[MeSH Terms:noexp] | 97688 | |
| MESH.EXACT(“Arthritis, Rheumatoid”) | 97688 | |
| Caplan Syndrome[Mesh:noexp] | 176 | |
| MESH.EXACT(“Caplan Syndrome”) | 176 | |
| Felty Syndrome[Mesh:noexp] | 716 | |
| MESH.EXACT(“Felty Syndrome”) | 716 | |
| Rheumatoid Nodule[Mesh:noexp] | 929 | |
| MESH.EXACT(“Rheumatoid Nodule”) | 929 | |
| Still’s Disease, Adult-Onset [Mesh:noexp] | 1301 | |
| MESH.EXACT(“Still’s Disease, Adult-Onset”) | 1301 | |
| Rheumatic Diseases [Mesh:noexp] | 22637 | |
| MESH.EXACT(“Rheumatic Diseases”) | 22637 | |
| Methotrexate[Mesh:noexp] | 37377 | |
| MESH.EXACT(“Methotrexate”) | 37377 | |
| clinical trial[Publication Type] | 849589 | |
| dtype(“CLINICAL TRIAL*” or “randomized controlled trial” or “equivalence trial”) | 852432** | |
| Clinical Trials as Topic[Mesh] | 189896 | |
| MESH.EXACT.EXPLODE(“Clinical Trials as Topic”) | 189896 |
Please note: **On PubMed, clinical trial [publication type] pulls from the Clinical Trial publication type tree. To appropriately compare the two platforms, we needed to add additional terms to the Dialog search.
Next, we created consolidated search strategies – limited to children aged 18 or younger, and to a publication date range of Jan 1, 2012 to Feb 3, 2020.
Here’s the Dialog Search Strategy:
| Set# | Searched for | Results |
| S1 | ((MESH.EXACT(“Arthritis, Rheumatoid”))) and (pd(20120101-20200203)) | 20915* |
| S2 | ((MESH.EXACT(“Caplan Syndrome”))) and (pd(20120101-20200203)) | 4° |
| S3 | ((MESH.EXACT(“Felty Syndrome”))) and (pd(20120101-20200203)) | 34° |
| S4 | ((MESH.EXACT(“Rheumatoid Nodule”))) and (pd(20120101-20200203)) | 96° |
| S5 | ((MESH.EXACT(“Still’s Disease, Adult-Onset”))) and (pd(20120101-20200203)) | 507° |
| S6 | ((MESH.EXACT(“Rheumatic Diseases”))) and (pd(20120101-20200203)) | 3053° |
| S7 | (MESH.EXACT(“Methotrexate”)) and (pd(20120101-20200203)) | 7653* |
| S8 | ((dtype(“CLINICAL TRIAL*” or “randomized controlled trial” or “equivalence trial”))) and (pd(20120101-20200203)) | 253466* |
| S9 | (MESH.EXACT.Explode(“Clinical Trials as Topic”)) and (pd(20120101-20200203)) | 90785* |
| S10 | (S6 OR S5 OR S3 OR S2 OR S1) and (pd(20120101-20200203)) | 24286* |
| S11 | (S9 OR S8) and (pd(20120101-20200203)) | 340274* |
| S12 | (S7 and S10 and S11) and (pd(20120101-20200203)) | 623° |
| S13 | (S12 and (mesh.exact(“adolescent” OR infant OR “infant,newborn” OR “child” OR “child,preschool”))) and (pd(20120101-20200203)) | 46° |
Please note: °Duplicates were automatically removed from the search and from the result count
Here’s the PubMed Search Strategy:
Conclusions: How does Dialog compare to PubMed?
We found that Dialog’s retrieval and sensitivity is comparable to PubMed. However, one issue we ran across was that PubMed explodes terms ‘behind the scenes’. We saw this happen on PubMed with clinical trials[publication type] and RHEUMATISM. When we checked the Search details box, this is what PubMed was searching:
“rheumatic diseases”[MeSH Terms] OR (“rheumatic”[All Fields] AND “diseases”[All Fields]) OR “rheumatic diseases”[All Fields] OR “rheumatism”[All Fields]
For our search, we decided to limit to Rheumatic Diseases [Mesh:noexp]. This searches the unexploded MeSH term.
Although the results of our test show that both PubMed and Dialog MEDLINE offer commensurate results, using a single database is not enough. For example, under new EU regulations for medical devices due to come into effect over the next three years2, one of the main requirements is the need for robust systematic searches – run within multiple databases to capture all relevant articles. This is where we have an advantage over PubMed – in Dialog you can run your search, not just in MEDLINE, but in Embase, Biosis, SciSearch and many other databases simultaneously. We automatically remove duplicate articles to save you time and increase efficiency – so you can rest assured that you have not only the most appropriate results, but the widest range of results too!
If you’d like to find out more about Dialog Solutions, and the wide-ranging ways we simplify the research process, please get in touch.
1. Katchamart, W., Faulkner, A., Feldman, B., Tomlinson, G., & Bombardier, C. (2011). PubMed had a higher sensitivity than Ovid-MEDLINE in the search for systematic reviews. Journal of Clinical Epidemiology, 64(7), 805-7. https://www.jclinepi.com/article/S0895-4356(10)00228-3/fulltext
2. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32017R0745
