Wikidata:Property proposal/study type

study type edit

   Done: study type (P8363) (Talk and documentation)
Descriptionclassification system for clinical trial (Q30612), experiment (Q101965), research (Q42240)
Representsstudy type (Q78088984)
Data typeItem
Template parameternot currently implemented, but see en:PARAMOUNT trial for infobox example where this would go
Allowed valuesexpected use is with ClinicalTrials.gov (Q5133746) and ClinicalTrials.gov ID (P3098), which in 98% of cases take take 1 of 3 types, interventional study (Q78089383), observational study (Q78089804), or expanded access (Q407954). ClinicalTrials.gov strictly applies this property in this way, but this property could could apply to other research classification systems in other fields of study.
Example 1PARAMOUNT trial (Q17148583)interventional study (Q78089383)
Example 2RV 144 (Q7278028)interventional study (Q78089383)
Example 3Medical Cannabis Registry and Pharmacology (Q63597344)observational study (Q78089804)
Example 4Expanded Access Protocol to Provide Brincidofovir for the Treatment of Serious Adenovirus Infection or Disease (Q64790189)expanded access type (Q78093526)
SourceClinicalTrials.gov (Q5133746), Data elements
Planned useitems with ClinicalTrials.gov ID (P3098)
Number of IDs in source~320k for ClinicalTrials.gov ID (P3098); anyone else can bring in other research databases
Expected completenesseventually complete (Q21873974)
See alsoWikidata:Property proposal/research status, research intervention (P4844), clinical trial phase (P6099), research site (P6153), medical condition treated (P2175)

Motivation edit

  Notified participants of WikiProject Medicine

A group of Wikidata:WikiProject Medicine participants have been modeling out clinical trials at Wikidata:WikiProject Medicine/Data models/Trials. Related, I recently also proposed Wikidata:Property proposal/research status. See the data model at the above linked "trials" project page. We have been building out Wikidata content over time, first working with the property for study intervention and what Wikidata already had, then adapting other properties, and now this one should be the last property proposed before a catch up point at which time we finish some uploads and present some useful queries.

In clinical trials, the "study type" is a broad classification of research which notes whether an experiment puts experimental drugs into humans for drug development, or does postmarket / safety surveillance, or treats anyone with "expanded access" as a last resort when no other treatment works. This culture is well established in the United States and anywhere else connected to the United States pharmaceutical industry.

Here are some interesting queries which we can do with this:

  1. What is the research profile of a given university in terms of the sorts of studies its people conduct?
  2. What is the count of interventional studies per city?
  3. What fields of medicine tend to offer expanded access?
  4. What trends can we detect for observational studies after a major medical lawsuit?

Blue Rasberry (talk) 22:50, 10 December 2019 (UTC)[reply]

TiagoLubiana 01:35, 16 March 2020 Daniel Mietchen 01:42, 16 March 2020 (UTC)[reply]
Jodi.a.schneider 02:45, 16 March 2020 (UTC)[reply]
Chchowmein 02:45, 16 March 2020 (UTC)[reply]
Dhx1 03:38, 16 March 2020 (UTC)[reply]
Konrad Foerstner 06:02, 16 March 2020 (UTC)[reply]
Netha Hussain 06:19, 16 March 2020 (UTC)[reply]
Bodhisattwa 06:56, 16 March 2020 (UTC)[reply]
Neo-Jay 07:04, 16 March 2020 (UTC)[reply]
John Samuel 07:31, 16 March 2020 (UTC)[reply]
KlaudiuMihaila 07:53, 16 March 2020 (UTC)[reply]
Salgo60 09:11, 16 March 2020 (UTC)[reply]
Andrawaag 10:12, 16 March 2020 (UTC)[reply]
Whidou 10:16, 16 March 2020 (UTC)[reply]
Blue Rasberry 15:07, 16 March 2020 (UTC)[reply]
TJMSmith 16:15, 16 March 2020 (UTC)[reply]
Egon Willighagen 16:49, 16 March 2020 (UTC)[reply]
Nehaoua 20:32, 16 March 2020 (UTC)[reply]
Andy Mabbett (UTC)
Peter Murray-Rust 00:00, 17 March 2020 (UTC)[reply]
Kasyap 02:45, 17 March 2020 (UTC)[reply]
Denny 16:21, 17 March 2020 (UTC)[reply]
Kwj2772 16:56, 17 March 2020 (UTC)[reply]
Joalpe 22:47, 17 March 2020 (UTC)[reply]
Finn Årup Nielsen fnielsen) 10:59, 18 March 2020 (UTC)[reply]
Skim 11:45, 18 March 2020 (UTC)[reply]
SCIdude 15:15, 18 March 2020 (UTC)[reply]
Evolution and evolvability 01:23, 20 March 2020 (UTC)[reply]
Susanna Ånäs (Susannaanas) 07:05, 20 March 2020 (UTC)[reply]
Mlemusrojas 15:30, 20 March 2020 (UTC)[reply]
Yupik 20:23, 20 March 2020 (UTC)[reply]
Csisc 23:05, 20 March 2020 (UTC)[reply]
OAnick 10:26, 21 March 2020 (UTC)[reply]
Gnoeee 12:28, 21 March 2020 (UTC)[reply]
Jjkoehorst 14:27, 21 March 2020 (UTC)[reply]
So9q 08:58, 22 March 2020 (UTC)[reply]
Nandana 14:58, 23 March 2020 (UTC)[reply]
Addshore 15:56, 23 March 2020 (UTC)[reply]
Librarian lena 18:19, 24 March 2020 (UTC)[reply]
Jelabra 19:19, 24 March 2020 (UTC)[reply]
AlexanderPico 23:34, 27 March 2020 (UTC)[reply]
Higa4 02:51, 29 March 2020 (UTC)[reply]
JoranL 19:56, 29 March 2020 (UTC)[reply]
Alejgh 11:04, 1 April 2020 (UTC)[reply]
Will (Wiki Ed)) 17:36, 1 April 2020 (UTC)[reply]
Ranjithsiji 04:47, 2 April 2020 (UTC)[reply]
AntoineLogean 07:35, 2 April 2020 (UTC)[reply]
Hannolans 17:22, 2 April 2020 (UTC)[reply]
Farmbrough 21:15, 3 April 2020 (UTC)[reply]
Ecritures 21:26, 3 April 2020 (UTC)[reply]

  Notified participants of WikiProject COVID-19

Discussion edit

@TiagoLubiana: Thanks for asking. The disadvantage is that many observational studies are a subclass of many types of studies, and not only clinical trials. Items probably should have a property and value indicating both "clinical trial" and "observational study". Right now Wikidata is only beginning to index clinical trials, but if for example we had items for all sorts of research, then observational studies include business and market research, user experience in design, sociology or psychology studies which are not clinical trials, wildlife tracking, agriculture, and statistics of all kinds. If someone queries for clinical trials probably they want to know about human subject research, but if they query for observational studies without also saying they are looking for clinical trials, then they might get anything. To take that cannabis example further, the interventional cannabis studies are going to be about medical effects and probably only interesting to medical researchers. The observational clinical trials about cannabis might of interest to psychologists, business researchers, anthropologists, and education researchers.
The same is true for intervention study in that many fields do these. "Expanded access" is different probably unique to clinical trials, such that if a study is "expanded access", then it must be a clinical trial. Still, these are established designations from ClinicalTrials.gov. I think that it is ideal to have a channel for importing their values. Thanks. Blue Rasberry (talk) 16:21, 9 April 2020 (UTC)[reply]
@TiagoLubiana: In order to see the value of having something other than instance of (P31) to describe such entities, let's consider humans, where we have a generic statement of P31 = human (Q5) for every Wikidata item about a human who actually lives or lived, and all the differences between humans are expressed using other properties like occupation (P106) or position held (P39). The proposed property would lead us onto a similar path for clinical trials. Using P31 for expressing the differences would be akin to saying things like Marie Curie is/ was and "instance of" a chemist, physicist, Nobel laureate or some such, which would add lots of confusion compared to the status quo. --Daniel Mietchen (talk) 21:45, 9 April 2020 (UTC)[reply]
@Daniel Mietchen, Bluerasberry: Thank you very much for the answers. Now I can see clearly the benefits of spreading the info in specialized properties. It sure does make the data clearer, and reduces confusion. Once again, useful modeling. I believe that with a bit more modeling, it will be a great property. Thank you for the proposal. TiagoLubiana (talk) 22:31, 9 April 2020 (UTC)[reply]
  •   Comment I think it is useful to capture whether a particular study is observational or interventional in nature or something else or something more specific. I am not sure "study type" is the best way to frame and name it. I am leaning more towards "study design" or perhaps "study design type", but upon checking what we have on the matter, I noticed that there is design of experiments (Q2334061), which confuses the method with the scientific subdiscipline, so I feel a bit more cleanup is needed around this proposed property before it can be created. Otherwise, we might have to clean up lots of statements later if they involve this property, once it is created. --Daniel Mietchen (talk) 21:35, 9 April 2020 (UTC)[reply]
Right, I see that too now, better to sort now with a few dozens of items than to do edits at scale then have to fix tens of thousands. Let's model this a bit more in the next week then reconsider. I also agree that the name is ambiguous. It is the name mentioned in ClinicalTrials.gov but they do not define it as a technical term, I think. Let's check to see if we can find sources on the right name. Blue Rasberry (talk) 13:33, 10 April 2020 (UTC)[reply]
There is clinical study design (Q17007018) which also has an English Wikipedia article describing the concept. Part of the challenge is separating this concept as a method versus a field of practice. The product of work in the field of "study design" is a "study design". This item and Wikipedia article have the same problem of conflating these concepts. Let me think more about this. Blue Rasberry (talk) 21:05, 17 April 2020 (UTC)[reply]
about study type

Daniel asked how established "study type" was as a term, and considered whether "study design" or some other term could be better. Previously we knew that the United States' ClinicalTrials.gov registry used this term. Now I have new evidence that this is a global term. I found the WHO list of "primary registries", which is a term that they give to clinical trial registries which meet certain standards and which they aggregate into their own system. I made Wikidata items for each registry on that list and I checked to see if I could find out if and how they note the concept for study type. I found the following evidence that "study type" is the term used in this space:

  • China Study type statistics in Global - great breakdown of possible study types, "interventional" and "observational" are still almost all trials but other classifications also
  • Austrailia / New Zealand - trial search - study type, everything is either "interventional" or "observational"
  • India - trial search - called "type of trial", options are interventional, observational, post-marketing surveillance, and Bioavailability/Bioequivalence studies
  • Germany trial search - study type, everything is either "interventional" or "observational"
  • Japan single example - uses field "study type", this example is interventional

All of this use does not prove that "study type" is the best term, but I think it does demonstrate that this is a broadly established concept in global clinical research. I also think that even though interventional and observational are established as two most common study types, there is precedent for using other labels. I am not sure what precedent exists for anyone using this field outside of clinical research, but I think this concept applies, as do these common labels. Blue Rasberry (talk) 22:18, 10 June 2020 (UTC)[reply]

about study design

The English Wikipedia article describes "study design" as a different concept. See en:Clinical study design or design of experiments (Q2334061). I think this could be yet another Wikidata property for describing research, with possible valuesd being descriptions such as randomized controlled trial (Q1436668), blind experiment (Q608510), ecological study (Q295136), or cohort study (Q1778788). I recognize that calling one of these kinds of descriptions "study type" and the other as "study design" is arbitrary, as either could be called "type" or "design", but I think this precedent is in place. "Study type" seems more established for use in its particular way. Any of these designs could have either common study type, "interventional" or "observational", so I see no conflict in having a later "study design" property. Blue Rasberry (talk) 22:18, 10 June 2020 (UTC)[reply]


  •   Support Property is quite useful, nicely presented and mature. The discussion about the correct name is important, but not crucial for use, I think. Maybe the property could be created with aliases and the discussion continue (if needed) in the property talk page. Here they annotate studies with a similar idea using the headers study "method" and "detail_method", for example: COVID-19 SOLES (--> Quick Downloads --> Download Categorised Dataset). TiagoLubiana (talk) 12:27, 19 June 2020 (UTC)[reply]
@Bluerasberry, ديفيد عادل وهبة خليل 2, Tibbs001, Daniel Mietchen, TiagoLubiana: study type (P8363) has been created. Pamputt (talk) 05:19, 20 June 2020 (UTC)[reply]