Tip #32: Non-Latin Characters in Ovid MEDLINE

A recent search project on "17β-hydroxysteroid dehydrogenase deficiency" inspired me to investigate how non-Latin characters are handled across various databases. This week we will look at how Ovid MEDLINE handles non-Latin and special characters using the following examples:

1. 17β-HSD (17 beta-HSD)
2. cáncer (Spanish for Cancer)
3. سرطان (Arabic for Cancer)
   

The help documentation wasn't in-depth enough for my needs so I reached out to Ruth Bernstein, a Tech Support team member at Wolters Kluwer, for further assistance. This post is based on our email exchange. Thanks, Ruth!

If you attempt to run any of the above searches, you will get the following error message:

17β-HSD

As indicated by the error message, special characters (such as β) are not searchable on Ovid. You will need to transliterate the Greek character and include it with the acronym spelled out, the full name of the substance, and the substance as a MeSH term. 

1. 17 beta HSD.mp. 277
2. 17-Hydroxysteroid Dehydrogenases/ 2223
3. 17 beta-hydroxysteroid dehydrogenase.mp. 1485
4. 1 or 2 or 3 2620
5. 1 not (2 or 3) 53

Worth noting: Most but NOT all 277 records with “17 Beta-HSD “ also include the MeSH term or the full name, but you get 2300+ more records with those additional approaches.

The acronym search (line 1) that has 277 results total, finds 53 unique records for that term that are NOT found with Line 2 or 3, see (line 5 that shows what is found with line 1 but not line 2 or 3.)
 

The three search statements should be OR’d together for a more comprehensive search.

Cáncer and سرطان

The Spanish and Arabic forms of the word ‘cancer’ are a metadata and translation feature issue. MEDLINE does not supply metadata in languages that are not English, and there is not a translation/transliteration feature for Ovid search nor does Ovid enrich data, at least for MeSH in other languages.

Further, Ovid does not have diacritical marks that are searchable in this way, or special characters like the Greek letter Beta in the 17β-HSD example.

However, if the article is in another language, and the record has the original title in that language, then you can search for that word (without diacritics) in the Original Title (OT) field. The Original Title includes any non-English titles in the original language. If the original title was in a non-Latin alphabet, then the OT is transliterated.

Search Suggestions for Spanish:

When the non-English term is spelled EXACTLY like the English term, then ALSO AND to the search the (LG) Language field and the language needed. Use the “search fields” feature to check this, as in the example below:

1 cancer.ot. 25812

2 spanish.lg. 373996

3 1 and 2 7699

 

Search Suggestions for Arabic:

While you can’t search with Arabic characters for cancer, you can transliterate the term and use that to search as a keyword. In the example below, we used the noun: “saratan” and broadened it a bit with the adjective  “sarataniyy.”

1 arabic.lg. 312

2 (saratan or sarataniyy).ot. 1

3 (saratan or sarataniyy).ot,ab. 3

4 (saratan or sarataniyy).mp. 3

5 3 not 4 0

Please note that line 1 of the search for language of publication in Arabic was included show that in MEDLINE there are only 312 articles total in the Arabic language, so the likelihood of finding even one is very small. Thus Line 2 finds only 1 result.

However, searching ONLY the OT field will likely return more precise results. You may get false drops by expanding the non-English term to the abstract field (AB) or even the (MP) multipurpose field. In this example, you get the same three results with (AB) and (MP) and the additional records are likely false drops.