Secondary Postpartum Hemorrhage

1. Justification of the non-interventional nature of the study: Severe late PPH is a rare disorder that can be difficult to diagnose. It is a source of severe maternal morbidity; its management is not well defined; and the lack of knowledge of its risk factors means that groups of women at risk of this disorder cannot be identified. Because of the lack of scientific data, it is impossible to set up monitoring appropriate to each woman's risk or to provide women who have just given birth with appropriate information. The practices studied will be those usually used in the healthcare facilities in Auvergne that manage PPH. The data useful for this study will be those collected routinely in medical files, but will be collected prospectively for women who are hospitalized after a severe late PPH. Investigators conducted this type of data collection for the Hera study (national PHRC), which covered immediate postpartum hemorrhages, for which management is well codified through the national good clinical practice guidelines, updated in 2014 .The section on the management of secondary hemorrhages shows the sparseness of the relevant scientific data on this subject. The guidelines are thus based mainly on expert opinion. Accordingly, the value of this study is that it will begin to develop an evidence base for the management of this disorder. The guidelines specify that "the management of late PPH depends on their cause and their severity [antibiotic therapy (Grade A) and uterotonic agents (professional consensus)]. This treatment must follow the hospital's own protocols (professional consensus). Most often, women are admitted for this management (professional consensus). In the case of persistence of the PPH and retention of placental fragments, aspiration-curettage under ultrasound control or operative hysteroscopy is recommended (professional consensus). In cases of vascular abnormalities, selective embolization is the treatment of choice (professional consensus)." This study will not direct professionals to use any specific rules or practices for the management of these late hemorrhages, other than the national guidelines cited above. Because of the great autonomy it offers for this management, it probably results in substantial heterogeneity in care, which this study will observe and examine. At the end of the study, investigator hope to have highlighted the most relevant management practices by cause of these severe late PPH.
2. Procedure for early withdrawal from the study: the woman's refusal to participate in the study.
3. Methods of recruitment: Women with a severe late PPH will be recruited prospectively after verification of the inclusion criteria by a gynecologist-obstetrician or the investigating midwives, who will provide the patients information about the study and include them unless they object.
4. Steps taken to reduce and avoid bias: Because the objective is to determine the incidence of severe late/secondary PPH, investigator will have several investigators at each site to be able to optimize the prospective identification of women meeting the inclusion criteria. Data entry will be performed by a research midwife who will have access to all of the medical data sources at each site (regional computerized pregnancy and delivery records, paper medical records, local computerized file for surgical, laboratory, pathology and imaging reports, hospital computerized discharge summary) to limit missing data. The aim is to obtain an exhaustive list of severe late PPH in the general population, to be ensured by the consultation of multiple data sources. Because these data sources are generated by different institutions and agencies, the selection bias associated by collection from a specific population (e.g., a single center) will be avoided.
5. Collection of data for women with a severe late PPH Patient inclusion and data collection will take place continuously. Every case of severe late PPH among women who give birth in the region of Auvergne will be included. The women will be informed about the planned study and procedures by posters in the maternity units (lounges and waiting rooms, the web site of the Auvergne perinatal network) and by an information sheet. At the time the case is identified, several data items will be collected immediately by one of the local investigators (data about the late PPH episode); the other data, as well as those concerning antenatal care during pregnancy will be collected by the research midwife from the regional computerized records and other portions of the woman's file, in each institution (laboratory, pathology, and surgery reports).

• Passive data collection: It will be based on the spontaneous reporting of cases by the correspondents in each maternity unit. These lists of cases, on a specific form, will be reported to the principal investigator monthly. *Active data collection: A research midwife will complete the files already underway. The investigator will visit the site to consult the woman's paper or computerized records and complete the information then missing from the study file. Auvergne has a regional computerized perinatal database. These site visits will be planned by the investigator, who will telephone the local correspondent to make an appointment and request from him or her a list of files to be made available. The frequency of site visits will vary according to the number of patients reported by each site. The different modes of active data collection:
• Data collection by telephone Will cover only the minimum data necessary to open the file or the small quantities of missing data.
• Data collection by mail The physician serving as the study director will contact the study correspondents, if necessary, once a quarter to review the number of women with late PPH at each facility. Investigators will check the data for each case and open new files when applicable.
• Collection from discharge database summaries (PMSI) Each year, the physician serving as the study director will ask the Medical Information Department of each healthcare facility to furnish a list of all women who were rehospitalized in the 42 days after giving birth or who had a surgical, or medical or interventional radiology procedure during their postpartum stay. All of these records will be studied manually by the investigator to check for cases of severe late PPH that were not previously reported (quality control for case ascertainment).

1. Procedures to validate the cases of late PPH All reports of eligible women (that is, who had a severe late PPH) will be checked at arrival by the investigator, who will verify if this is the woman's first inclusion in the study. The investigator will verify the woman's identity and date of birth during site visits, to avoid duplicate inclusions. If it is a first inclusion, the investigator will validate the inclusion criteria before submitting the file to one of the study directors, who will validate the etiological diagnosis. The investigator will verify the quality of the data collection and file completion: in the case of missing data, he or she will take the steps required to have the data completed at the reporting department. The investigator will code the administrative items and drugs by using specific thesauruses. The investigator will also enter the data for each record after coding, using RedCap® software, available at the Clermont-Ferrand UHC. A local index for each center, including the file number, if a file was opened, or information about its status if it was not opened, the woman's first and last name, and the reporting source will be maintained and kept up-to-date by the local investigator for each late PPH diagnosis included.
2. Data analysis methods The α risk (of a type I error) will be set at 5%. The statistical tests will be bilateral. A descriptive analysis of the entire dataset will be performed. The incidence of severe late PPH will be calculated with its 95% confidence interval. A Cox model (with severe late PPH the variable to be explained) will be used to adjust for the confounding factors or the clinically relevant prognostic factors for the onset of severe late PPH. The point of origin will be the delivery. The model will make it possible to calculate adjusted hazard ratios.
3. Patient information Patients will receive complete and fair information, expressed in understandable terms, about the study objectives, the nature of the information collected, and their right to object to at any point to the use of their data. The investigator must also inform the patients of the Ethics Committee decision. Women will be asked whether or not they object to participating in the study before their data are collected.

All of this information will be included on an information sheet given to the woman to keep. Her non-opposition will be collected and documented by the investigator.