Duplicates in ACE tables CSTS, FBSTC, FCSTC

FAQ Health Data Hub
,
1

Hi,

I am currently using the PMSI outpatient consultations and services at public hospitals (ACE), listed in the table CSTS with cost details in FBSTC, FCSTC, and VALOACE for post-2013. Having read the fact sheet on duplicate entries in the PMSI ACE tables, I have already filtered out eta_num entries referring to AP-HM, HCL, and AP-HP.

After that, I would like to merge value tables with the main CSTC table but using eta_num + seq_num + year of a respective row (I am using the full sample period), but I still encounter a lot of duplicates.

As I am working with the EDP Santé, I can see that many times the same eta_num + seq_num + date of service (not just year) combination is being used for different EDP IDs, hence, different persons. Hence, I am wondering (a) does this refer to patients being accompanied by parents or guardians, or is there another reason? And (b) what is the cleanest way to avoid this issue (besides dropping all duplicate groups)?

Thank you very much in advance!

2

Bonjour,

Merci d’avoir posé votre question sur le forum SNDS.

Après vérification sur le portail SNDS de la CNAM, les tables CSTC des années 2021 à 2024 ne contiennent pas de doublons : le couple (ETA_NUM, SEQ_NUM) constitue bien la clé primaire, chaque couple étant associé à un unique NIR_ANO_17.

Pour les tables annuelles FBSTC et FCSTC, le triplet (ETA_NUM, SEQ_NUM, date des soins) ne suffit pas à constituer une clé primaire. L’unicité d’un enregistrement dépend également d’autres variables, telles que la catégorie d’acte (ACT_COD), le coefficient de l’acte, la date de réalisation de l’acte, etc. En revanche, chaque couple (ETA_NUM, SEQ_NUM) est associé à un unique NIR_ANO_17. Ne connaissant pas les “EDP IDs”, ceci est-il équivalent à l’identifiant patient ?

J’espère que ces éléments répondent à votre question.

Bien à vous,

Fei GAO – Health Data Hub

Hello,

Thank you for asking your question on the SNDS forum.

After checking the CNAM SNDS portal, the CSTC tables for the years 2021 to 2024 do not contain any duplicates: the pair (ETA_NUM, SEQ_NUM) effectively constitutes the primary key, with each pair being associated with a unique NIR_ANO_17.

For the annual FBSTC and FCSTC tables, the triplet (ETA_NUM, SEQ_NUM, care date) is not sufficient to constitute a primary key. The uniqueness of a record also depends on other variables, such as the procedure category (ACT_COD), the procedure coefficient, the date of the act, etc. However, each pair (ETA_NUM, SEQ_NUM) is associated with a unique NIR_ANO_17. Since I am not familiar with « EDP IDs », is this equivalent to the patient identifier? I hope this information answers your question.

Kind regards,

Fei GAO – Health Data Hub