The SurCeGIDD surveillance was developed through a collaborative and iterative approach launched in 2015 (Figure 1). First, the stakeholders engaged in STI clinics, infectious diseases, sexual health promotion, epidemiological surveillance, regional and national health administration, and population representatives were identified through a national concertation meeting held by the French Ministry of Health. The software companies were not engaged in the working groups to prevent any conflict of interest, assuming that the CeGIDDs will contract the implementation of the specifications for SurCeGIDD afterward.

Among the stakeholders, volunteers—regional health agencies (n=3), former STI clinics and HIV-hepatitis testing centers (n=4), charities (n=2), regional response coordination consortium for HIV infection (n=1), SpF, the Health Care and Health Directorates of the Ministry of Health—defined the core principles for the SurCeGIDD surveillance:

Subsequently, the stakeholders mapped the information needs at the national, subnational, and local levels (Textbox 1 [28]). Finally, SpF used this mapping exercise to propose a detailed metadata set that was discussed, amended, and agreed upon by all the stakeholders.

Second, to assess the applicability of the SurCeGIDD principles, SpF conducted a nonsystematic review of scientific articles and the gray literature using the keywords (surveillance or individual or cases-based data extraction or data export or data workflow or data flow or interoperability and France or Europe or the names of comparable industrialized countries [eg, United States of America, United Kingdom, England, Germany]) on PubMed, Embase, Scopus, and Google Scholar. The websites of the national public health agencies, references of interesting articles, and the documents shared by the European networks for surveillances were also checked. After scanning the titles and abstracts for the most relevant ones, articles were analyzed using a form to gather information of interest (eg, framework, metadata set, stakeholder, data collection, data sharing, data integration). The use of health care reimbursement data (no clinical details), the mandatory notification system for HIV (electronic or paper form, such as e-DO in France), and the syndromic surveillance system based on algorithms analyzing the global activity of health care services (no biological details) were excluded from this document review, considering the principle of using clinical and biological records routinely collected during the consultations [23,29-32].

In France, the literature review shows that the surveillance of the colorectal cancer relies on a validated data set of clinical and biological information [33]. This surveillance covers 50 screening centers using 3 software to manage the patients’ records. In the absence of interoperability, an application was developed to recode, extract, and transfer anonymized data from the screening centers to SpF through a secured data-sharing platform [33]. An intermittent data workflow is generated yearly by the screening centers, by activating a specific command on their computer [33]. Then, incoming data are checked and merged using SAS (SAS Institute Inc.) [33]. Another approach is an automated data extraction from 3 multisite laboratories (3Labos surveillance) including tests and diagnoses for 11 pathogens [34]. The 3Labos surveillance relies on a daily/monthly transmission (depending on pathogens) of individual data (sex, age, residence, test, and diagnosis) to SpF through a web service (a protocol enabling the communication between 2 electronic devices via internet) [34]. The incoming data are checked and merged using a Stata (StataCorp) script [34].

The literature review also showed the advantages and limitations of surveillance systems based on data extraction in other countries as well as the development stages of these surveillance systems: GUMCAD (the Genitourinary Medicine Clinic Activity Dataset) in England; CIDR (Computerised Infectious Disease Reporting) in Ireland; SSuN (STD Surveillance Network) in the United States; MIBA (the Danish Microbiology Database) in Denmark; ACCESS (Australian Collaboration for Coordinated Enhanced Sentinel Surveillance) in Australia; eCR (Electronic Case Reporting) in Utah, USA; SOAP (Dutch STI Database) in the Netherlands; and MAVEN (Massachusetts Virtual Epidemiologic Network) in the United States [35-44]. The main points in the development of these electronic and automated surveillance systems were the precise data specifications, the stakeholder’s involvement, the effective information systems to manage patients’ records, and the development and the maintenance of an information platform by the surveillance programs to collect and integrate the data [35-39,44,45]. A long-term strategic framework for the sustainability and the evolution of the surveillance system is also an asset.

Third, to better interpret the information retrieved from the literature, SpF conducted a series of in-depth interviews with French experts (clinicians, biologists, software specialists, data managers, and epidemiologists), using an interview guide to complete the information on the French context. Then, SpF organized a concertation meeting with several clinicians, biologists, and public health specialists to share the findings of the literature review and the interviews, and to discuss the informatics capabilities of the CeGIDDs as well as the levers and obstacles that could impact the implementation of the SurCeGIDD principles in each subnational region.

Fourth, in May 2016, SpF conducted a feasibility survey, by sending an electronic questionnaire to all CeGIDDs. This questionnaire was used to collect information on the computerization of the CeGIDDs, the specifications of their software for the management of the patients records, and their capability to export or recode and transfer a data set in a predefined CSV format. The response rate was 57.2% (183/320). Among the respondents, most CeGIDDs (147/183, 80.3%) used informatic tools to manage their patients/users. A majority of them could export individual data (117/183, 63.9%). Less than half of the CeGIDDs (79/183, 43.2%) were capable of recoding a data set but only 21.9% (40/183) reported being able to recode and transfer data according to a template. The CeGIDDs used mainly 6 software and numerous “homemade” applications. Interestingly, 2 major software companies decided to apply the SurCeGIDD principles at this stage, and communicated their decision to use the surveillance specifications as soon as they would be published. Few months after the feasibility study, one of these companies covered half of the CeGIDDs.

Fifth, the findings of the feasibility assessment were used to adjust the SurCeGIDD metadata set, taking into account the day-to-day practices in the CeGIDDs and their informatic limitation. Then, the order of November 23, 2016, stated the mandatory format for the SurCeGIDD data as well as the procedures for data transfer [46]. The specifications for the surveillance were subsequently published on the SpF website [46].

Considering the SurCeGIDD specifications, the literature review, and the results of the feasibility survey, a web service was designed for the data transmission [42,47]. If any software provider developed an access to the web service using the specifications shared by SpF, the CeGIDD can automatically transfer its individual database to SpF (Figure 2). Otherwise, the CeGIDDs can export and recode their data in the expected CSV format before their transmission to SpF via a secured data-sharing platform.

Then, the structure of the data is automatically checked and validated before its integration in the SurCeGIDD repository or its return to the CeGIDDs if this control stage fails. During this stage, empty variables are accepted. Warning messages pop-up if the data specifications are not applied. In such a case, SpF helps the CeGIDDs to correct its data set.

A pilot study conducted by SpF among 12 voluntary CeGIDDs helped to test, correct, redesign, and implement the surveillance system between March and April 2017. During the pilot development, 6 CeGIDDs failed to extract their data. Among the 6 data sets that were transmitted to SpF, only 3 were extracted and transferred according to the SurCeGIDD specifications. This highlighted CeGIDDs’ difficulties in performing effectively the expected data management, and the need for friendly software requiring no manual tasks to prepare the data.

Preliminary and descriptive analyses of 2017-2018 data were performed using numbers and proportions to assess the data quality and estimate for the first time the positivity rates for STI in the CeGIDDs (number of positive tests divided by the total of tests), giving an overview of the SurCeGIDD surveillance system after its deployment. Stata 14.2 was used for statistical analyses.

The protocol was reviewed and approved by the data protection officer of SpF. Then, ethical approval was granted by the French Personal Data Protection Authority (CNIL authorization no. 2049450). Data are collected with patients’ consent. Only anonymized data are transmitted through the SurCeGIDD system. Subsequently, any potential identifying information is aggregated or suppressed after the elimination of duplicates to ensure patients confidentiality. Access to database is strictly restricted to the authorized agents of SpF.

Despite numerous challenges, SurCeGIDD surveillance was successfully implemented. The effectiveness of this system to collect automatically important volumes of data was demonstrated by 143/320 (44.7%) CeGIDDs that reported simultaneously the STI tests, diagnoses, clinical details, and patients’ socioeconomic characteristics for the first time in France. Although these data need further consolidation, the preliminary analysis confirmed that CeGIDDs’ attendees are particularly exposed to STI with high positivity rates: 0.37% for HIV, 1.31% for HBV, 7.16% for Ct, 2.83% for gonorrhea, 1.04% for syphilis, and 5.96% for M. genitalium. These results also gave the first insight into M. genitalium testing in the CeGIDDs. Further analyses will better describe their situation.

However, a selection of CeGIDDs with greater human, informatic, or financial resources compared with the nonparticipating structures cannot be ruled out. But the participation rate of 143/320 (44.7%) will probably increase in the next years, considering the growing interest of the CeGIDDs and software providers for SurCeGIDD data. Experiences in STI surveillance in Denmark, England, Ireland, Australia, and United Sates seem to confirm this expectation [35-38,44]. Moreover, the increasing computerization of the CeGIDDs and the adapted support provided by SpF in the use of the specifications might contribute to a better representativeness of the data in the medium term.

This support is also an opportunity for SpF to correct the metadata set, taking into account the operational feedback from the structures. Nevertheless, the time to develop or modify the informatic platforms was underestimated by SpF and stakeholders, delaying the use of the updated metadata set in data collection. This was also observed in other contexts, and it must be better anticipated in informatic development [35-38,44]. Moreover, difficulties to extract or recode data by CeGIDDs according to the specifications hampered the surveillance implementation. Indeed, some CeGIDDs selected a range of variables considering feasibility burden. Consequently, nonprioritized variables were mostly missing.

The proportion of missing information was also greater than 50% for important variables such as the characteristics of sexual partners (311,703/382,890, 81.41%), PrEP use (296,399/382,890, 77.41%), HBV vaccine (325,215/382,890, 84.94%), HAV (347,429/382,890, 90.74%), HPV vaccine (338,165/382,890, 88.32%), diagnoses of genital warts (269,492/382,890, 70.38%), diagnoses of herpes (269,489/382,890, 70.38%), and antimicrobial resistance of N. gonorrhoeae. Some patient’s records could not be recoded despite the CeGIDDs’ willingness to conform to the surveillance specifications. For example, some CeGIDDs only recorded trademarks of vaccines instead of the names and doses that enable the monitoring of vaccination coverage. Some adjustment might be necessary in the data format to better take CeGIDDs’ needs into account. This also underlines the need for appropriate human and financial resources to support CeGIDDs’ adherence to the specifications. Nevertheless, data transmission will require no extra labor burden once the specifications have been implemented, in contrast to the use of paper or electronic forms [35,41,42].

Because of the legal context in France, it was not possible to engage software companies in the SurCeGIDD working groups. In England, software providers contributed to the early stage of the surveillance project, applying GUMCAD specification in the development of their tools [36]. Moreover, the lack of specific funding for information systems in CeGIDDs might limit SurCeGIDD’s expansion and evolution because each structure has to pay for the implementation of the specification changes. Conversely, incentives and additional funding contributed to ensure the largest and long-term coverage in other contexts [35,37,41,42]. Mandatory use of national specifications by all software companies might be a solution. It might not only enable database transmission by all CeGIDDs, but can also facilitate the implementation of any change simultaneously in all CeGIDDs (additional information) [35,36].

Regarding SurCeGIDD coverage, the list of CeGIDDs including emails and telephone numbers was not updated since 2016, despite the creation of new structures and closures of other ones by regional health agencies [8]. Therefore, it was not possible to engage with noninformed CeGIDDs in the surveillance. SpF advocates for a systematic transmission of updated phone books by regional public health agencies each year, to ensure the best representativeness of the surveillance data.

Considering the volumes of data shared, SpF needs to upgrade its tools and procedures, using algorithms to automatize data monitoring, quality checking, and epidemiological reports at national and subnational levels. There is also a need to formalize rules and process for data sharing and analyses [35,48,49].

Despite its limitations, SurCeGIDD surveillance creates new opportunities for research collaboration on sexual health in France [35,41,50]. Availability of a wide range of individual characteristics (health insurance, occupation) in this database will be an opportunity to better understand the impact of social health inequalities on STI epidemics in France. SurCeGIDD will also help to fill the gap in our knowledge of M. genitalium, genital warts, herpes, contraception, and vaccination in France [8]. Generating a continuous data workflow instead of a yearly transmission of data might be the next improvement of SurCeGIDD, to timely detect and investigate STI events (multidrug resistance, uncommon anatomic location, upsurge, clusters) [41,44,51-53]. An anonymous linkage of patients within the SurCeGIDD database might improve the epidemiological surveillance [35,36,42]. The almost real-time transmission of routinely collected data to monitor COVID-19 diagnoses from laboratories in France demonstrated the feasibility of such evolutions [48]. Moreover, the expansion of this laboratory surveillance toward STI diagnoses might help to complete the picture of STI epidemics in France.

The automated SurCeGIDD surveillance is functional despite the challenges that hamper its coverage expansion. A long-term vision relying on sustained funding, including informatic developments for STI clinics, and the mandatory use of the specifications by software providers might help to improve the repressiveness of surveillance data to better guide and evaluate prevention strategies and services delivery. Indeed, the timeliness and efficacy of the COVID-19 monitoring systems demonstrate how electronic medical records could effectively replace passive surveillance systems, particularly sentinel surveillance systems commonly overwhelmed by important volumes of information. Moreover, exhaustive medical and laboratory records could help to compare more accurately countries for STI trends, drivers of epidemics, and sexual health services delivery.