Clinical Evaluation - state of the art literature search

Reg1000

Registered
Hello,
I am writing a clinical evaluation report under the MDR. This will be reviewed by a notified body. I have been reading around the topic and would welcome advice:

1) Should there be 2 separate literature searches performed? The first for the state of the art section and the second for clinical data for the device under evaluation. Is this what notified bodies are expecting to see under the MDR?
2) Do NB expect to see PICO used for SoA?
3) What sort of criteria have you found NB are looking for re: the inclusion/exclusion criteria for SoA articles? For the device I am working on I expect there would be 1000's of hits to screen.
4) Apart from meddev, is there any other guidance/industry training that you can recommend that gives practical advice on conducting and reporting the SoA search?

Thanks!
 

Raisin picker

Quite Involved in Discussions
Hello and welcome,

The notified bodies I know don't look too much in formalities regarding SotA, rather that you do define the parameters for clinical benefit from SotA (see MDR Annex XIV (1.a)).
1) That's up to you. You can do one search and state for each hit if it's applicable for SotA, for clinical data or not at all. Or you do two separate searches, one for the CEP (SotA), one for the CER (pivotal data/clinical data).
2) There are now about 25 NBs around. I guess at least one does expect PICO, but I also guess that several others don't ;-)
3) That's up to you and the device type. Try to find parameters to reduce the number of hits to a reasonable number (e.g. reviews, HTAs, patient numbers, age, ...)
4) The only other guidance I'm aware of is IMDRF MDCE WG/N56FINAL:2019 (formerly GHTF/SG5/N2R8:2007). But MDR and MEDDEV 2.7/1 and the MDCG documents should be sufficient, and they reference each other (MDR >> MDCG >> MEDDEV).
 

MDRexpert

Involved In Discussions
Hi Everyone.

Can someone clarify - if we run 2 separate literature searches - one for the state-of-the-art section and the second for clinical data for the device under evaluation. The only difference between the two run is the competitor devices. Most of the articles found in these are similar. How do you summarize them? Do you list all the articles screened, even if they are the same or you combine the list in one and then include/exclude them with a justification?

Thanks!
 

Vetty007

Involved In Discussions
Hi,

in my Opinion, it doesn't make sense to do two different runs and it would even confuse me, if this is needed. Usually your clinical data will give you information about the state of the art. Possibly medical guidelines will not be be found in the literature search, depending on your search strategy, but independend on the search on medical databases you should check the relevant associations/societies regarding a relevant guideline.

The most important is a sound search strategy incl. filters - if its one of the named strategies perfect, but if not and your search strategy is sound, that shouldn't be a problem. But often one of these well known search strategies are used as they really make sense, sometimes its just not recognized that its used or people only often struggle with the definition of a useful search term concept and instead search wildly for relevant terms or just for the product without using a meaningful context. This will bring many results, but often don't help either and that also costs more time than thinking about a good search strategy.
Of course duplicates needs to be removed and you also need to define, how you make sure to delete them (also includes e.g. the same data, but given but with different titels). I import all data of my literature search in one list, read the abstracts in it and state in the follwing columne, if I read it in detail or if the publication is not relevant incl. explanation (e.g. not relevant due to other intention of use). For all relevant literature I state if they give information about performance, safety (will be also used as input for risk assessment) or state of the art. In the following columns I rate the publications according to some quality criteria incl. evidence, and based on this stating, if the publication is used for the CER (only good quality, which is easier to define if you have fixed criteria). Its pretty much work in the beginning, but its worth - e.g. I had approx. 300 search results (without proper filters it was about 3.000), but approx. 230 could be easily excluded by reading the abstract and for CER only 15 needed to be taken into account and additional 3 for risk analysis (due to poor quality only used for this purpose).

Hope this helps.
 
Hi Everyone.

Can someone clarify - if we run 2 separate literature searches - one for the state-of-the-art section and the second for clinical data for the device under evaluation. The only difference between the two run is the competitor devices. Most of the articles found in these are similar. How do you summarize them? Do you list all the articles screened, even if they are the same or you combine the list in one and then include/exclude them with a justification?

Thanks!

Yes - it's best to run two separate literature searches - one for clinical background/SOTA, one for evidence relating to the subject device. When planning the searches, it's worth thinking carefully about the objective of each: clinical background/SOTA is, from a clinical evaluation perspective, about the determination of safety and performance benchmarks - how are other, similar devices performing? The subject device search is then about determining whether the subject device is non-inferior to those benchmarks. From this perspective it's great that the articles are similar, because they are likely to report outcome measures that are comparable. In terms of summarising them, this is a big topic but in short form: write a short summary of each paper setting out the salient points along with an appraisal against pre-determined criteria. You do need to include all relevant articles and maintain a record of all excluded sources against each search term - given that there can be many hundreds of results, this requires an efficient process. Realistically, there shouldn't be any sources that are included in both clinical background/SOTA and in the section for the subject device - if an article includes the subject device, it should be in the latter section, not the former. For this reason, one exclusion criteria should be 'duplicate article'.
 

Raisin picker

Quite Involved in Discussions
To be more specific, one search includes (list not necessarily complete):
- device type/ generic device group
- indications (to account for alternative therapies)
- inclusion criteria
- product technology
- ...

The other search includes (list not necessarily complete):
- your device
- predecessor devices
- equivalent devices

And if you combine both, you have to separate the results again later.

MDCG 2020-13 suggests (on page 16) that you might want to use "systematic search and review methods", and lists some examples. By the way, the whole section D there (p. 15-18) seems to be interesting for your topic, including references to MEDDEV 2.7/1.
 
Top Bottom